Index: projects/largeSMP/Makefile.inc1 =================================================================== --- projects/largeSMP/Makefile.inc1 (revision 221494) +++ projects/largeSMP/Makefile.inc1 (revision 221495) @@ -1,1597 +1,1597 @@ # # $FreeBSD$ # # Make command line options: # -DNO_CLEANDIR run ${MAKE} clean, instead of ${MAKE} cleandir # -DNO_CLEAN do not clean at all # -DNO_SHARE do not go into share subdir # -DKERNFAST define NO_KERNEL{CONFIG,CLEAN,DEPEND,OBJ} # -DNO_KERNELCONFIG do not run config in ${MAKE} buildkernel # -DNO_KERNELCLEAN do not run ${MAKE} clean in ${MAKE} buildkernel # -DNO_KERNELDEPEND do not run ${MAKE} depend in ${MAKE} buildkernel # -DNO_KERNELOBJ do not run ${MAKE} obj in ${MAKE} buildkernel # -DNO_PORTSUPDATE do not update ports in ${MAKE} update # -DNO_DOCUPDATE do not update doc in ${MAKE} update # -DNO_CTF do not run the DTrace CTF conversion tools on built objects # LOCAL_DIRS="list of dirs" to add additional dirs to the SUBDIR list # TARGET="machine" to crossbuild world for a different machine type # TARGET_ARCH= may be required when a TARGET supports multiple endians # # The intended user-driven targets are: # buildworld - rebuild *everything*, including glue to help do upgrades # installworld- install everything built by "buildworld" # doxygen - build API documentation of the kernel # update - convenient way to update your source tree (eg: cvsup/cvs) # # Standard targets (not defined here) are documented in the makefiles in # /usr/share/mk. These include: # obj depend all install clean cleandepend cleanobj # You are supposed to define both of these when calling Makefile.inc1 # directly. However, some old scripts don't. Cope for the moment, but # issue a new warning for a transition period. .if defined(TARGET) && !defined(TARGET_ARCH) .warning "You must pass both TARGET and TARGET_ARCH to Makefile.inc1. Setting TARGET_ARCH=${TARGET}." TARGET_ARCH=${TARGET} .endif .if !defined(TARGET) || !defined(TARGET_ARCH) .error "Both TARGET and TARGET_ARCH must be defined." .endif .include .include # We must do share/info early so that installation of info `dir' # entries works correctly. Do it first since it is less likely to # grow dependencies on include and lib than vice versa. # # We must do lib/ and libexec/ before bin/, because if installworld # installs a new /bin/sh, the 'make' command will *immediately* # use that new version. And the new (dynamically-linked) /bin/sh # will expect to find appropriate libraries in /lib and /libexec. # SUBDIR= share/info lib libexec SUBDIR+=bin .if ${MK_GAMES} != "no" SUBDIR+=games .endif .if ${MK_CDDL} != "no" SUBDIR+=cddl .else NO_CTF=1 .endif SUBDIR+=gnu include .if ${MK_KERBEROS} != "no" SUBDIR+=kerberos5 .endif .if ${MK_RESCUE} != "no" SUBDIR+=rescue .endif SUBDIR+=sbin .if ${MK_CRYPT} != "no" SUBDIR+=secure .endif .if !defined(NO_SHARE) SUBDIR+=share .endif SUBDIR+=sys usr.bin usr.sbin .if ${MK_OFED} != "no" SUBDIR+=contrib/ofed .endif # # We must do etc/ last for install/distribute to work. # SUBDIR+=etc # These are last, since it is nice to at least get the base system # rebuilt before you do them. .for _DIR in ${LOCAL_DIRS} .if exists(${.CURDIR}/${_DIR}/Makefile) SUBDIR+= ${_DIR} .endif .endfor .if defined(SUBDIR_OVERRIDE) SUBDIR= ${SUBDIR_OVERRIDE} .endif .if defined(NOCLEAN) NO_CLEAN= ${NOCLEAN} .endif .if defined(NO_CLEANDIR) CLEANDIR= clean cleandepend .else CLEANDIR= cleandir .endif CVS?= cvs CVSFLAGS?= -A -P -d -I! SVN?= svn SVNFLAGS?= -r HEAD SUP?= /usr/bin/csup SUPFLAGS?= -g -L 2 .if defined(SUPHOST) SUPFLAGS+= -h ${SUPHOST} .endif MAKEOBJDIRPREFIX?= /usr/obj .if !defined(OSRELDATE) .if exists(/usr/include/osreldate.h) OSRELDATE!= awk '/^\#define[[:space:]]*__FreeBSD_version/ { print $$3 }' \ /usr/include/osreldate.h .else OSRELDATE= 0 .endif .endif .if !defined(VERSION) VERSION!= uname -srp VERSION+= ${OSRELDATE} .endif KNOWN_ARCHES?= amd64 arm armeb/arm i386 i386/pc98 ia64 mipsel/mips mipseb/mips mips64el/mips mips64eb/mips mipsn32el/mips mipsn32eb/mips powerpc powerpc64/powerpc sparc64 sparc64/sun4v .if ${TARGET} == ${TARGET_ARCH} _t= ${TARGET} .else _t= ${TARGET_ARCH}/${TARGET} .endif .for _t in ${_t} .if empty(KNOWN_ARCHES:M${_t}) .error Unknown target ${TARGET_ARCH}:${TARGET}. .endif .endfor .if ${TARGET} == ${MACHINE} TARGET_CPUTYPE?=${CPUTYPE} .else TARGET_CPUTYPE?= .endif .if !empty(TARGET_CPUTYPE) _TARGET_CPUTYPE=${TARGET_CPUTYPE} .else _TARGET_CPUTYPE=dummy .endif _CPUTYPE!= MAKEFLAGS= CPUTYPE=${_TARGET_CPUTYPE} ${MAKE} \ -f /dev/null -m ${.CURDIR}/share/mk -V CPUTYPE .if ${_CPUTYPE} != ${_TARGET_CPUTYPE} .error CPUTYPE global should be set with ?=. .endif .if make(buildworld) BUILD_ARCH!= uname -p .if ${MACHINE_ARCH} != ${BUILD_ARCH} .error To cross-build, set TARGET_ARCH. .endif .endif .if ${MACHINE} == ${TARGET} && ${MACHINE_ARCH} == ${TARGET_ARCH} && !defined(CROSS_BUILD_TESTING) OBJTREE= ${MAKEOBJDIRPREFIX} .else OBJTREE= ${MAKEOBJDIRPREFIX}/${TARGET}.${TARGET_ARCH} .endif WORLDTMP= ${OBJTREE}${.CURDIR}/tmp # /usr/games added for fortune which depend on strfile BPATH= ${WORLDTMP}/legacy/usr/sbin:${WORLDTMP}/legacy/usr/bin:${WORLDTMP}/legacy/usr/games XPATH= ${WORLDTMP}/usr/sbin:${WORLDTMP}/usr/bin:${WORLDTMP}/usr/games STRICTTMPPATH= ${BPATH}:${XPATH} TMPPATH= ${STRICTTMPPATH}:${PATH} # # Avoid running mktemp(1) unless actually needed. # It may not be functional, e.g., due to new ABI # when in the middle of installing over this system. # .if make(distributeworld) || make(installworld) INSTALLTMP!= /usr/bin/mktemp -d -u -t install .endif # # Building a world goes through the following stages # # 1. legacy stage [BMAKE] # This stage is responsible for creating compatibility # shims that are needed by the bootstrap-tools, # build-tools and cross-tools stages. # 1. bootstrap-tools stage [BMAKE] # This stage is responsible for creating programs that # are needed for backward compatibility reasons. They # are not built as cross-tools. # 2. build-tools stage [TMAKE] # This stage is responsible for creating the object # tree and building any tools that are needed during # the build process. # 3. cross-tools stage [XMAKE] # This stage is responsible for creating any tools that # are needed for cross-builds. A cross-compiler is one # of them. # 4. world stage [WMAKE] # This stage actually builds the world. # 5. install stage (optional) [IMAKE] # This stage installs a previously built world. # BOOTSTRAPPING?= 0 # Common environment for world related stages CROSSENV= MAKEOBJDIRPREFIX=${OBJTREE} \ MACHINE_ARCH=${TARGET_ARCH} \ MACHINE=${TARGET} \ CPUTYPE=${TARGET_CPUTYPE} .if ${OSRELDATE} < 700044 CROSSENV+= AR=gnu-ar RANLIB=gnu-ranlib .endif .if ${MK_GROFF} != "no" CROSSENV+= GROFF_BIN_PATH=${WORLDTMP}/legacy/usr/bin \ GROFF_FONT_PATH=${WORLDTMP}/legacy/usr/share/groff_font \ GROFF_TMAC_PATH=${WORLDTMP}/legacy/usr/share/tmac .endif # bootstrap-tools stage BMAKEENV= INSTALL="sh ${.CURDIR}/tools/install.sh" \ PATH=${BPATH}:${PATH} \ WORLDTMP=${WORLDTMP} \ VERSION="${VERSION}" \ MAKEFLAGS="-m ${.CURDIR}/tools/build/mk ${.MAKEFLAGS}" BMAKE= MAKEOBJDIRPREFIX=${WORLDTMP} \ ${BMAKEENV} ${MAKE} -f Makefile.inc1 \ DESTDIR= \ BOOTSTRAPPING=${OSRELDATE} \ SSP_CFLAGS= \ -DWITHOUT_HTML -DWITHOUT_INFO -DNO_LINT -DWITHOUT_MAN \ -DNO_PIC -DWITHOUT_PROFILE -DNO_SHARED \ -DNO_CPU_CFLAGS -DNO_WARNS -DNO_CTF # build-tools stage TMAKE= MAKEOBJDIRPREFIX=${OBJTREE} \ ${BMAKEENV} ${MAKE} -f Makefile.inc1 \ TARGET=${TARGET} TARGET_ARCH=${TARGET_ARCH} \ DESTDIR= \ SSP_CFLAGS= \ BOOTSTRAPPING=${OSRELDATE} -DNO_LINT -DNO_CPU_CFLAGS \ -DNO_WARNS -DNO_CTF # cross-tools stage XMAKE= TOOLS_PREFIX=${WORLDTMP} ${BMAKE} \ TARGET=${TARGET} TARGET_ARCH=${TARGET_ARCH} \ -DWITHOUT_GDB # world stage WMAKEENV= ${CROSSENV} \ _SHLIBDIRPREFIX=${WORLDTMP} \ VERSION="${VERSION}" \ INSTALL="sh ${.CURDIR}/tools/install.sh" \ PATH=${TMPPATH} .if ${MK_CDDL} == "no" WMAKEENV+= NO_CTF=1 .endif WMAKE= ${WMAKEENV} ${MAKE} -f Makefile.inc1 DESTDIR=${WORLDTMP} .if ${TARGET_ARCH} == "amd64" || ${TARGET_ARCH} == "powerpc64" # 32 bit world LIB32TMP= ${OBJTREE}${.CURDIR}/lib32 .if ${TARGET_ARCH} == "amd64" .if empty(TARGET_CPUTYPE) LIB32CPUFLAGS= -march=i686 -mmmx -msse -msse2 .else LIB32CPUFLAGS= -march=${TARGET_CPUTYPE} .endif LIB32CPUFLAGS+= -mfancy-math-387 LIB32WMAKEENV= MACHINE=i386 MACHINE_ARCH=i386 \ MACHINE_CPU="i686 mmx sse sse2" \ LD="${LD} -m elf_i386_fbsd -Y P,${LIB32TMP}/usr/lib32" \ AS="${AS} --32" .elif ${TARGET_ARCH} == "powerpc64" .if empty(TARGET_CPUTYPE) LIB32CPUFLAGS= -mcpu=powerpc .else LIB32CPUFLAGS= -mcpu=${TARGET_CPUTYPE} .endif LIB32WMAKEENV= MACHINE=powerpc MACHINE_ARCH=powerpc \ LD="${LD} -m elf32ppc" .endif LIB32FLAGS= -m32 ${LIB32CPUFLAGS} -DCOMPAT_32BIT \ -isystem ${LIB32TMP}/usr/include/ \ -L${LIB32TMP}/usr/lib32 \ -B${LIB32TMP}/usr/lib32 # Yes, the flags are redundant. LIB32WMAKEENV+= MAKEOBJDIRPREFIX=${OBJTREE}/lib32 \ _SHLIBDIRPREFIX=${LIB32TMP} \ VERSION="${VERSION}" \ INSTALL="sh ${.CURDIR}/tools/install.sh" \ PATH=${TMPPATH} \ CC="${CC} ${LIB32FLAGS}" \ CXX="${CXX} ${LIB32FLAGS}" \ LIBDIR=/usr/lib32 \ SHLIBDIR=/usr/lib32 LIB32WMAKE= ${LIB32WMAKEENV} ${MAKE} -DNO_CPU_CFLAGS -DCOMPAT_32BIT \ -DWITHOUT_BIND -DWITHOUT_MAN -DWITHOUT_INFO \ -DWITHOUT_HTML -DNO_CTF -DNO_LINT DESTDIR=${LIB32TMP} LIB32IMAKE= ${LIB32WMAKE:NINSTALL=*:NDESTDIR=*} -DNO_INCS .endif # install stage IMAKEENV= ${CROSSENV} IMAKE= ${IMAKEENV} ${MAKE} -f Makefile.inc1 .if empty(.MAKEFLAGS:M-n) IMAKEENV+= PATH=${STRICTTMPPATH}:${INSTALLTMP} \ LD_LIBRARY_PATH=${INSTALLTMP} \ PATH_LOCALE=${INSTALLTMP}/locale IMAKE+= __MAKE_SHELL=${INSTALLTMP}/sh .else IMAKEENV+= PATH=${TMPPATH}:${INSTALLTMP} .endif # kernel stage KMAKEENV= ${WMAKEENV} KMAKE= ${KMAKEENV} ${MAKE} KERNEL=${INSTKERNNAME} # # buildworld # # Attempt to rebuild the entire system, with reasonable chance of # success, regardless of how old your existing system is. # _worldtmp: .if ${.CURDIR:C/[^,]//g} != "" # The m4 build of sendmail files doesn't like it if ',' is used # anywhere in the path of it's files. @echo @echo "*** Error: path to source tree contains a comma ','" @echo false .endif @echo @echo "--------------------------------------------------------------" @echo ">>> Rebuilding the temporary build tree" @echo "--------------------------------------------------------------" .if !defined(NO_CLEAN) rm -rf ${WORLDTMP} .if defined(LIB32TMP) rm -rf ${LIB32TMP} .endif .else rm -rf ${WORLDTMP}/legacy/usr/include # XXX - These three can depend on any header file. rm -f ${OBJTREE}${.CURDIR}/usr.bin/kdump/ioctl.c rm -f ${OBJTREE}${.CURDIR}/usr.bin/kdump/kdump_subr.c rm -f ${OBJTREE}${.CURDIR}/usr.bin/truss/ioctl.c .endif .for _dir in \ lib usr legacy/usr mkdir -p ${WORLDTMP}/${_dir} .endfor mtree -deU -f ${.CURDIR}/etc/mtree/BSD.usr.dist \ -p ${WORLDTMP}/legacy/usr >/dev/null .if ${MK_GROFF} != "no" mtree -deU -f ${.CURDIR}/etc/mtree/BSD.groff.dist \ -p ${WORLDTMP}/legacy/usr >/dev/null .endif mtree -deU -f ${.CURDIR}/etc/mtree/BSD.usr.dist \ -p ${WORLDTMP}/usr >/dev/null mtree -deU -f ${.CURDIR}/etc/mtree/BSD.include.dist \ -p ${WORLDTMP}/usr/include >/dev/null ln -sf ${.CURDIR}/sys ${WORLDTMP} .if ${MK_BIND_LIBS} != "no" mtree -deU -f ${.CURDIR}/etc/mtree/BIND.include.dist \ -p ${WORLDTMP}/usr/include >/dev/null .endif _legacy: @echo @echo "--------------------------------------------------------------" @echo ">>> stage 1.1: legacy release compatibility shims" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${BMAKE} legacy _bootstrap-tools: @echo @echo "--------------------------------------------------------------" @echo ">>> stage 1.2: bootstrap tools" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${BMAKE} bootstrap-tools _cleanobj: .if !defined(NO_CLEAN) @echo @echo "--------------------------------------------------------------" @echo ">>> stage 2.1: cleaning up the object tree" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${WMAKE} ${CLEANDIR:S/^/par-/} .if defined(LIB32TMP) ${_+_}cd ${.CURDIR}; ${LIB32WMAKE} -f Makefile.inc1 ${CLEANDIR:S/^/par-/} .endif .endif _obj: @echo @echo "--------------------------------------------------------------" @echo ">>> stage 2.2: rebuilding the object tree" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${WMAKE} par-obj _build-tools: @echo @echo "--------------------------------------------------------------" @echo ">>> stage 2.3: build tools" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${TMAKE} build-tools _cross-tools: @echo @echo "--------------------------------------------------------------" @echo ">>> stage 3: cross tools" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${XMAKE} cross-tools _includes: @echo @echo "--------------------------------------------------------------" @echo ">>> stage 4.1: building includes" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${WMAKE} SHARED=symlinks par-includes _libraries: @echo @echo "--------------------------------------------------------------" @echo ">>> stage 4.2: building libraries" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; \ ${WMAKE} -DNO_FSCHG -DWITHOUT_HTML -DWITHOUT_INFO -DNO_LINT \ -DWITHOUT_MAN -DWITHOUT_PROFILE libraries _depend: @echo @echo "--------------------------------------------------------------" @echo ">>> stage 4.3: make dependencies" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${WMAKE} par-depend everything: @echo @echo "--------------------------------------------------------------" @echo ">>> stage 4.4: building everything" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${WMAKE} par-all .if defined(LIB32TMP) build32: @echo @echo "--------------------------------------------------------------" @echo ">>> stage 5.1: building 32 bit shim libraries" @echo "--------------------------------------------------------------" mkdir -p ${LIB32TMP}/usr/include mtree -deU -f ${.CURDIR}/etc/mtree/BSD.usr.dist \ -p ${LIB32TMP}/usr >/dev/null mtree -deU -f ${.CURDIR}/etc/mtree/BSD.include.dist \ -p ${LIB32TMP}/usr/include >/dev/null mkdir -p ${WORLDTMP} ln -sf ${.CURDIR}/sys ${WORLDTMP} .if ${MK_KERBEROS} != "no" .for _t in obj depend all cd ${.CURDIR}/kerberos5/tools; \ MAKEOBJDIRPREFIX=${OBJTREE}/lib32 ${MAKE} SSP_CFLAGS= DESTDIR= \ DIRPRFX=kerberos5/tools/ ${_t} .endfor .endif .for _t in obj includes cd ${.CURDIR}/include; ${LIB32WMAKE} DIRPRFX=include/ ${_t} cd ${.CURDIR}/lib; ${LIB32WMAKE} DIRPRFX=lib/ ${_t} .if ${MK_CDDL} != "no" cd ${.CURDIR}/cddl/lib; ${LIB32WMAKE} DIRPRFX=cddl/lib/ ${_t} .endif cd ${.CURDIR}/gnu/lib; ${LIB32WMAKE} DIRPRFX=gnu/lib/ ${_t} .if ${MK_CRYPT} != "no" cd ${.CURDIR}/secure/lib; ${LIB32WMAKE} DIRPRFX=secure/lib/ ${_t} .endif .if ${MK_KERBEROS} != "no" cd ${.CURDIR}/kerberos5/lib; ${LIB32WMAKE} DIRPRFX=kerberos5/lib ${_t} .endif .endfor .for _dir in usr.bin/lex/lib cd ${.CURDIR}/${_dir}; ${LIB32WMAKE} DIRPRFX=${_dir}/ obj .endfor .for _dir in lib/ncurses/ncurses lib/ncurses/ncursesw lib/libmagic cd ${.CURDIR}/${_dir}; \ MAKEOBJDIRPREFIX=${OBJTREE}/lib32 ${MAKE} SSP_CFLAGS= DESTDIR= \ DIRPRFX=${_dir}/ build-tools .endfor cd ${.CURDIR}; \ ${LIB32WMAKE} -f Makefile.inc1 libraries .for _t in obj depend all cd ${.CURDIR}/libexec/rtld-elf; PROG=ld-elf32.so.1 ${LIB32WMAKE} \ DIRPRFX=libexec/rtld-elf/ ${_t} cd ${.CURDIR}/usr.bin/ldd; PROG=ldd32 ${LIB32WMAKE} \ DIRPRFX=usr.bin/ldd ${_t} .endfor distribute32 install32: cd ${.CURDIR}/lib; ${LIB32IMAKE} ${.TARGET:S/32$//} .if ${MK_CDDL} != "no" cd ${.CURDIR}/cddl/lib; ${LIB32IMAKE} ${.TARGET:S/32$//} .endif cd ${.CURDIR}/gnu/lib; ${LIB32IMAKE} ${.TARGET:S/32$//} .if ${MK_CRYPT} != "no" cd ${.CURDIR}/secure/lib; ${LIB32IMAKE} ${.TARGET:S/32$//} .endif .if ${MK_KERBEROS} != "no" cd ${.CURDIR}/kerberos5/lib; ${LIB32IMAKE} ${.TARGET:S/32$//} .endif cd ${.CURDIR}/libexec/rtld-elf; \ PROG=ld-elf32.so.1 ${LIB32IMAKE} ${.TARGET:S/32$//} cd ${.CURDIR}/usr.bin/ldd; PROG=ldd32 ${LIB32IMAKE} ${.TARGET:S/32$//} .endif WMAKE_TGTS= .if !defined(SUBDIR_OVERRIDE) WMAKE_TGTS+= _worldtmp _legacy _bootstrap-tools .endif WMAKE_TGTS+= _cleanobj _obj _build-tools .if !defined(SUBDIR_OVERRIDE) WMAKE_TGTS+= _cross-tools .endif WMAKE_TGTS+= _includes _libraries _depend everything .if defined(LIB32TMP) && ${MK_LIB32} != "no" WMAKE_TGTS+= build32 .endif buildworld: buildworld_prologue ${WMAKE_TGTS} buildworld_epilogue .ORDER: buildworld_prologue ${WMAKE_TGTS} buildworld_epilogue buildworld_prologue: @echo "--------------------------------------------------------------" @echo ">>> World build started on `LC_ALL=C date`" @echo "--------------------------------------------------------------" buildworld_epilogue: @echo @echo "--------------------------------------------------------------" @echo ">>> World build completed on `LC_ALL=C date`" @echo "--------------------------------------------------------------" # # We need to have this as a target because the indirection between Makefile # and Makefile.inc1 causes the correct PATH to be used, rather than a # modification of the current environment's PATH. In addition, we need # to quote multiword values. # buildenvvars: @echo ${WMAKEENV:Q} buildenv: @echo Entering world for ${TARGET_ARCH}:${TARGET} @cd ${.CURDIR} && env ${WMAKEENV} sh || true TOOLCHAIN_TGTS= ${WMAKE_TGTS:N_depend:Neverything:Nbuild32} toolchain: ${TOOLCHAIN_TGTS} kernel-toolchain: ${TOOLCHAIN_TGTS:N_includes:N_libraries} # # installcheck # # Checks to be sure system is ready for installworld/installkernel. # installcheck: # # Require DESTDIR to be set if installing for a different architecture. # .if ${TARGET_ARCH} != ${MACHINE_ARCH} || ${TARGET} != ${MACHINE} .if !make(distributeworld) installcheck: installcheck_DESTDIR installcheck_DESTDIR: .if !defined(DESTDIR) || empty(DESTDIR) @echo "ERROR: Please set DESTDIR!"; \ false .endif .endif .endif # # Check for missing UIDs/GIDs. # CHECK_UIDS= CHECK_GIDS= audit .if ${MK_SENDMAIL} != "no" CHECK_UIDS+= smmsp CHECK_GIDS+= smmsp .endif .if ${MK_PF} != "no" CHECK_UIDS+= proxy CHECK_GIDS+= proxy authpf .endif installcheck: installcheck_UGID installcheck_UGID: .for uid in ${CHECK_UIDS} @if ! `id -u ${uid} >/dev/null 2>&1`; then \ echo "ERROR: Required ${uid} user is missing, see /usr/src/UPDATING."; \ false; \ fi .endfor .for gid in ${CHECK_GIDS} @if ! `find / -prune -group ${gid} >/dev/null 2>&1`; then \ echo "ERROR: Required ${gid} group is missing, see /usr/src/UPDATING."; \ false; \ fi .endfor # # Required install tools to be saved in a scratch dir for safety. # .if ${MK_INFO} != "no" _install-info= install-info .endif .if ${MK_ZONEINFO} != "no" _zoneinfo= zic tzsetup .endif ITOOLS= [ awk cap_mkdb cat chflags chmod chown \ date echo egrep find grep ${_install-info} \ ln lockf make mkdir mtree mv pwd_mkdb rm sed sh sysctl \ test true uname wc ${_zoneinfo} # # distributeworld # # Distributes everything compiled by a `buildworld'. # # installworld # # Installs everything compiled by a 'buildworld'. # # Non-base distributions produced by the base system EXTRA_DISTRIBUTIONS= doc games .if defined(LIB32TMP) && ${MK_LIB32} != "no" EXTRA_DISTRIBUTIONS+= lib32 .endif distributeworld installworld: installcheck mkdir -p ${INSTALLTMP} progs=$$(for prog in ${ITOOLS}; do \ if progpath=`which $$prog`; then \ echo $$progpath; \ else \ echo "Required tool $$prog not found in PATH." >&2; \ exit 1; \ fi; \ done); \ libs=$$(ldd -f "%o %p\n" -f "%o %p\n" $$progs 2>/dev/null | sort -u | \ while read line; do \ set -- $$line; \ if [ "$$2 $$3" != "not found" ]; then \ echo $$2; \ else \ echo "Required library $$1 not found." >&2; \ exit 1; \ fi; \ done); \ cp $$libs $$progs ${INSTALLTMP} cp -R $${PATH_LOCALE:-"/usr/share/locale"} ${INSTALLTMP}/locale .if make(distributeworld) .for dist in ${EXTRA_DISTRIBUTIONS} -mkdir ${DESTDIR}/${DISTDIR}/${dist} mtree -deU -f ${.CURDIR}/etc/mtree/BSD.root.dist \ -p ${DESTDIR}/${DISTDIR}/${dist} >/dev/null mtree -deU -f ${.CURDIR}/etc/mtree/BSD.usr.dist \ -p ${DESTDIR}/${DISTDIR}/${dist}/usr >/dev/null mtree -deU -f ${.CURDIR}/etc/mtree/BSD.include.dist \ -p ${DESTDIR}/${DISTDIR}/${dist}/usr/include >/dev/null .endfor -mkdir ${DESTDIR}/${DISTDIR}/base ${_+_}cd ${.CURDIR}; ${IMAKE} distrib-dirs \ DESTDIR=${DESTDIR}/${DISTDIR}/base .endif ${_+_}cd ${.CURDIR}; ${IMAKE} re${.TARGET:S/world$//}; \ ${IMAKEENV} rm -rf ${INSTALLTMP} .if make(distributeworld) .for dist in ${EXTRA_DISTRIBUTIONS} find ${DESTDIR}/${DISTDIR}/${dist} -empty -delete .endfor .endif packageworld: .for dist in base ${EXTRA_DISTRIBUTIONS} ${_+_}cd ${DESTDIR}/${DISTDIR}/${dist}; \ tar cvJf ${DESTDIR}/${DISTDIR}/${dist}.txz . .endfor # # reinstall # # If you have a build server, you can NFS mount the source and obj directories # and do a 'make reinstall' on the *client* to install new binaries from the # most recent server build. # reinstall: @echo "--------------------------------------------------------------" @echo ">>> Making hierarchy" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${MAKE} -f Makefile.inc1 hierarchy @echo @echo "--------------------------------------------------------------" @echo ">>> Installing everything" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${MAKE} -f Makefile.inc1 install .if defined(LIB32TMP) && ${MK_LIB32} != "no" ${_+_}cd ${.CURDIR}; ${MAKE} -f Makefile.inc1 install32 .endif redistribute: @echo "--------------------------------------------------------------" @echo ">>> Distributing everything" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${MAKE} -f Makefile.inc1 distribute .if defined(LIB32TMP) && ${MK_LIB32} != "no" ${_+_}cd ${.CURDIR}; ${MAKE} -f Makefile.inc1 distribute32 \ DISTRIBUTION=lib32 .endif distrib-dirs distribution: cd ${.CURDIR}/etc; ${CROSSENV} PATH=${TMPPATH} ${MAKE} ${.TARGET} # # buildkernel and installkernel # # Which kernels to build and/or install is specified by setting # KERNCONF. If not defined a GENERIC kernel is built/installed. # Only the existing (depending TARGET) config files are used # for building kernels and only the first of these is designated # as the one being installed. # # Note that we have to use TARGET instead of TARGET_ARCH when # we're in kernel-land. Since only TARGET_ARCH is (expected) to # be set to cross-build, we have to make sure TARGET is set # properly. .if defined(KERNFAST) NO_KERNELCLEAN= t NO_KERNELCONFIG= t NO_KERNELDEPEND= t NO_KERNELOBJ= t # Shortcut for KERNCONF=Blah -DKERNFAST is now KERNFAST=Blah .if !defined(KERNCONF) && ${KERNFAST} != "1" KERNCONF=${KERNFAST} .endif .endif .if !defined(KERNCONF) && defined(KERNEL) KERNCONF= ${KERNEL} KERNWARN= .else .if ${TARGET_ARCH} == "powerpc64" KERNCONF?= GENERIC64 .else KERNCONF?= GENERIC .endif .endif INSTKERNNAME?= kernel KERNSRCDIR?= ${.CURDIR}/sys KRNLCONFDIR= ${KERNSRCDIR}/${TARGET}/conf KRNLOBJDIR= ${OBJTREE}${KERNSRCDIR} KERNCONFDIR?= ${KRNLCONFDIR} BUILDKERNELS= INSTALLKERNEL= .for _kernel in ${KERNCONF} .if exists(${KERNCONFDIR}/${_kernel}) BUILDKERNELS+= ${_kernel} .if empty(INSTALLKERNEL) INSTALLKERNEL= ${_kernel} .endif .endif .endfor # # buildkernel # # Builds all kernels defined by BUILDKERNELS. # buildkernel: .if empty(BUILDKERNELS) @echo "ERROR: Missing kernel configuration file(s) (${KERNCONF})."; \ false .endif .if defined(KERNWARN) @echo "--------------------------------------------------------------" @echo ">>> WARNING: KERNEL= setting should be changed to KERNCONF=" @echo "--------------------------------------------------------------" @sleep 3 .endif @echo .for _kernel in ${BUILDKERNELS} @echo "--------------------------------------------------------------" @echo ">>> Kernel build for ${_kernel} started on `LC_ALL=C date`" @echo "--------------------------------------------------------------" @echo "===> ${_kernel}" mkdir -p ${KRNLOBJDIR} .if !defined(NO_KERNELCONFIG) @echo @echo "--------------------------------------------------------------" @echo ">>> stage 1: configuring the kernel" @echo "--------------------------------------------------------------" cd ${KRNLCONFDIR}; \ PATH=${TMPPATH} \ config ${CONFIGARGS} -d ${KRNLOBJDIR}/${_kernel} \ ${KERNCONFDIR}/${_kernel} .endif .if !defined(NO_CLEAN) && !defined(NO_KERNELCLEAN) @echo @echo "--------------------------------------------------------------" @echo ">>> stage 2.1: cleaning up the object tree" @echo "--------------------------------------------------------------" cd ${KRNLOBJDIR}/${_kernel}; ${KMAKE} ${CLEANDIR} .endif .if !defined(NO_KERNELOBJ) @echo @echo "--------------------------------------------------------------" @echo ">>> stage 2.2: rebuilding the object tree" @echo "--------------------------------------------------------------" cd ${KRNLOBJDIR}/${_kernel}; ${KMAKE} obj .endif @echo @echo "--------------------------------------------------------------" @echo ">>> stage 2.3: build tools" @echo "--------------------------------------------------------------" cd ${KRNLOBJDIR}/${_kernel}; \ MAKESRCPATH=${KERNSRCDIR}/dev/aic7xxx/aicasm \ ${MAKE} SSP_CFLAGS= -DNO_CPU_CFLAGS \ -f ${KERNSRCDIR}/dev/aic7xxx/aicasm/Makefile # XXX - Gratuitously builds aicasm in the ``makeoptions NO_MODULES'' case. .if !defined(MODULES_WITH_WORLD) && !defined(NO_MODULES) && exists(${KERNSRCDIR}/modules) .for target in obj depend all cd ${KERNSRCDIR}/modules/aic7xxx/aicasm; \ MAKEOBJDIRPREFIX=${KRNLOBJDIR}/${_kernel}/modules \ ${MAKE} SSP_CFLAGS= -DNO_CPU_CFLAGS -DNO_CTF ${target} .endfor .endif .if !defined(NO_KERNELDEPEND) @echo @echo "--------------------------------------------------------------" @echo ">>> stage 3.1: making dependencies" @echo "--------------------------------------------------------------" cd ${KRNLOBJDIR}/${_kernel}; ${KMAKE} depend -DNO_MODULES_OBJ .endif @echo @echo "--------------------------------------------------------------" @echo ">>> stage 3.2: building everything" @echo "--------------------------------------------------------------" cd ${KRNLOBJDIR}/${_kernel}; ${KMAKE} all -DNO_MODULES_OBJ @echo "--------------------------------------------------------------" @echo ">>> Kernel build for ${_kernel} completed on `LC_ALL=C date`" @echo "--------------------------------------------------------------" .endfor # # installkernel, etc. # # Install the kernel defined by INSTALLKERNEL # installkernel installkernel.debug \ reinstallkernel reinstallkernel.debug: installcheck .if empty(INSTALLKERNEL) @echo "ERROR: No kernel \"${KERNCONF}\" to install."; \ false .endif @echo "--------------------------------------------------------------" - @echo ">>> Installing kernel ${KERNCONF}" + @echo ">>> Installing kernel ${INSTALLKERNEL}" @echo "--------------------------------------------------------------" cd ${KRNLOBJDIR}/${INSTALLKERNEL}; \ ${CROSSENV} PATH=${TMPPATH} \ ${MAKE} KERNEL=${INSTKERNNAME} ${.TARGET:S/kernel//} distributekernel distributekernel.debug: .if empty(INSTALLKERNEL) @echo "ERROR: No kernel \"${KERNCONF}\" to install."; \ false .endif cd ${KRNLOBJDIR}/${INSTALLKERNEL}; \ ${CROSSENV} PATH=${TMPPATH} ${MAKE} KERNEL=${INSTKERNNAME} \ DESTDIR=${DESTDIR}/${DISTDIR}/kernel \ ${.TARGET:S/distributekernel/install/} packagekernel: ${_+_}cd ${DESTDIR}/${DISTDIR}/kernel; \ tar cvJf ${DESTDIR}/${DISTDIR}/kernel.txz . # # doxygen # # Build the API documentation with doxygen # doxygen: @if [ ! -x `/usr/bin/which doxygen` ]; then \ echo "You need doxygen (devel/doxygen) to generate the API documentation of the kernel." | /usr/bin/fmt; \ exit 1; \ fi cd ${.CURDIR}/tools/kerneldoc/subsys && ${MAKE} obj all # # update # # Update the source tree, by running cvsup and/or running cvs to update to the # latest copy. # update: .if defined(SUP_UPDATE) @echo "--------------------------------------------------------------" @echo ">>> Running ${SUP}" @echo "--------------------------------------------------------------" .if defined(SUPFILE) @${SUP} ${SUPFLAGS} ${SUPFILE} .endif .if defined(SUPFILE1) @${SUP} ${SUPFLAGS} ${SUPFILE1} .endif .if defined(SUPFILE2) @${SUP} ${SUPFLAGS} ${SUPFILE2} .endif .if defined(PORTSSUPFILE) && !defined(NO_PORTSUPDATE) @${SUP} ${SUPFLAGS} ${PORTSSUPFILE} .endif .if defined(DOCSUPFILE) && !defined(NO_DOCUPDATE) @${SUP} ${SUPFLAGS} ${DOCSUPFILE} .endif .endif .if defined(CVS_UPDATE) @cd ${.CURDIR} ; \ if [ -d CVS ] ; then \ echo "--------------------------------------------------------------" ; \ echo ">>> Updating ${.CURDIR} from CVS repository" ${CVSROOT} ; \ echo "--------------------------------------------------------------" ; \ echo ${CVS} -R -q update ${CVSFLAGS} ; \ ${CVS} -R -q update ${CVSFLAGS} ; \ fi .endif .if defined(SVN_UPDATE) @cd ${.CURDIR} ; \ if [ -d .svn ] ; then \ echo "--------------------------------------------------------------" ; \ echo ">>> Updating ${.CURDIR} using Subversion" ; \ echo "--------------------------------------------------------------" ; \ echo ${SVN} update ${SVNFLAGS} ; \ ${SVN} update ${SVNFLAGS} ; \ fi .endif # # ------------------------------------------------------------------------ # # From here onwards are utility targets used by the 'make world' and # related targets. If your 'world' breaks, you may like to try to fix # the problem and manually run the following targets to attempt to # complete the build. Beware, this is *not* guaranteed to work, you # need to have a pretty good grip on the current state of the system # to attempt to manually finish it. If in doubt, 'make world' again. # # # legacy: Build compatibility shims for the next three targets # legacy: .if ${BOOTSTRAPPING} < 600034 && ${BOOTSTRAPPING} != 0 @echo "ERROR: Source upgrades from versions prior to 6.0 not supported."; \ false .endif .for _tool in tools/build ${_+_}@${ECHODIR} "===> ${_tool} (obj,includes,depend,all,install)"; \ cd ${.CURDIR}/${_tool}; \ ${MAKE} DIRPRFX=${_tool}/ obj; \ ${MAKE} DIRPRFX=${_tool}/ DESTDIR=${MAKEOBJDIRPREFIX}/legacy includes; \ ${MAKE} DIRPRFX=${_tool}/ depend; \ ${MAKE} DIRPRFX=${_tool}/ all; \ ${MAKE} DIRPRFX=${_tool}/ DESTDIR=${MAKEOBJDIRPREFIX}/legacy install .endfor # # bootstrap-tools: Build tools needed for compatibility # .if ${MK_GAMES} != "no" _strfile= games/fortune/strfile .endif .if ${MK_CXX} != "no" _gperf= gnu/usr.bin/gperf .endif .if ${MK_GROFF} != "no" _groff= gnu/usr.bin/groff .endif .if ${BOOTSTRAPPING} >= 700044 && ${BOOTSTRAPPING} < 800022 _ar= usr.bin/ar .endif .if ${BOOTSTRAPPING} < 800013 _mklocale= usr.bin/mklocale .endif .if ${BOOTSTRAPPING} < 900002 _sed= usr.bin/sed .endif .if ${BOOTSTRAPPING} < 900006 _lex= usr.bin/lex _yacc= usr.bin/yacc .endif .if ${BOOTSTRAPPING} < 700018 _gensnmptree= usr.sbin/bsnmpd/gensnmptree .endif .if ${MK_RESCUE} != "no" && \ ${BOOTSTRAPPING} < 700026 _crunchgen= usr.sbin/crunch/crunchgen .endif .if ${MK_CLANG} != "no" _clang_tblgen= \ lib/clang/libllvmsupport \ usr.bin/clang/tblgen .endif .if ${MK_CDDL} != "no" _dtrace_tools= cddl/usr.bin/sgsmsg cddl/lib/libctf lib/libelf \ lib/libdwarf cddl/usr.bin/ctfconvert cddl/usr.bin/ctfmerge .endif .if ${MK_FDT} != "no" _dtc= gnu/usr.bin/dtc .endif bootstrap-tools: .for _tool in \ ${_clang_tblgen} \ ${_dtrace_tools} \ ${_strfile} \ ${_gperf} \ ${_groff} \ ${_ar} \ ${_dtc} \ usr.bin/lorder \ usr.bin/makewhatis \ ${_mklocale} \ usr.bin/rpcgen \ ${_sed} \ ${_lex} \ ${_yacc} \ usr.bin/xinstall \ ${_gensnmptree} \ usr.sbin/config \ ${_crunchgen} ${_+_}@${ECHODIR} "===> ${_tool} (obj,depend,all,install)"; \ cd ${.CURDIR}/${_tool}; \ ${MAKE} DIRPRFX=${_tool}/ obj; \ ${MAKE} DIRPRFX=${_tool}/ depend; \ ${MAKE} DIRPRFX=${_tool}/ all; \ ${MAKE} DIRPRFX=${_tool}/ DESTDIR=${MAKEOBJDIRPREFIX}/legacy install .endfor # # build-tools: Build special purpose build tools # .if defined(MODULES_WITH_WORLD) && exists(${KERNSRCDIR}/modules) _aicasm= sys/modules/aic7xxx/aicasm .endif .if !defined(NO_SHARE) _share= share/syscons/scrnmaps .endif .if ${MK_KERBEROS} != "no" _kerberos5_tools= kerberos5/tools .endif .if ${MK_RESCUE} != "no" _rescue= rescue/rescue .endif build-tools: .for _tool in \ bin/csh \ bin/sh \ ${_rescue} \ lib/ncurses/ncurses \ lib/ncurses/ncursesw \ ${_share} \ ${_aicasm} \ usr.bin/awk \ lib/libmagic \ usr.sbin/sysinstall \ usr.bin/mkesdb_static \ usr.bin/mkcsmapper_static ${_+_}@${ECHODIR} "===> ${_tool} (obj,build-tools)"; \ cd ${.CURDIR}/${_tool}; \ ${MAKE} DIRPRFX=${_tool}/ obj; \ ${MAKE} DIRPRFX=${_tool}/ build-tools .endfor .for _tool in \ gnu/usr.bin/cc/cc_tools \ ${_kerberos5_tools} ${_+_}@${ECHODIR} "===> ${_tool} (obj,depend,all)"; \ cd ${.CURDIR}/${_tool}; \ ${MAKE} DIRPRFX=${_tool}/ obj; \ ${MAKE} DIRPRFX=${_tool}/ depend; \ ${MAKE} DIRPRFX=${_tool}/ all .endfor # # cross-tools: Build cross-building tools # .if ${TARGET_ARCH} != ${MACHINE_ARCH} || ${BOOTSTRAPPING} < 800035 .if ${TARGET_ARCH} == "amd64" || ${TARGET_ARCH} == "i386" _btxld= usr.sbin/btxld .endif .endif .if ${TARGET_ARCH} != ${MACHINE_ARCH} .if ${MK_RESCUE} != "no" || defined(RELEASEDIR) _crunchide= usr.sbin/crunch/crunchide .endif .if ${TARGET_ARCH} == "i386" && defined(RELEASEDIR) _kgzip= usr.sbin/kgzip .endif .endif .if ${MK_CLANG} != "no" .if ${CC:T:Mclang} == "clang" _clang= usr.bin/clang _clang_libs= lib/clang .endif .endif cross-tools: .for _tool in \ ${_clang_libs} \ ${_clang} \ gnu/usr.bin/binutils \ gnu/usr.bin/cc \ usr.bin/xlint/lint1 usr.bin/xlint/lint2 usr.bin/xlint/xlint \ ${_btxld} \ ${_crunchide} \ ${_kgzip} ${_+_}@${ECHODIR} "===> ${_tool} (obj,depend,all,install)"; \ cd ${.CURDIR}/${_tool}; \ ${MAKE} DIRPRFX=${_tool}/ obj; \ ${MAKE} DIRPRFX=${_tool}/ depend; \ ${MAKE} DIRPRFX=${_tool}/ all; \ ${MAKE} DIRPRFX=${_tool}/ DESTDIR=${MAKEOBJDIRPREFIX} install .endfor # # hierarchy - ensure that all the needed directories are present # hierarchy: cd ${.CURDIR}/etc; ${MAKE} distrib-dirs # # libraries - build all libraries, and install them under ${DESTDIR}. # # The list of libraries with dependents (${_prebuild_libs}) and their # interdependencies (__L) are built automatically by the # ${.CURDIR}/tools/make_libdeps.sh script. # libraries: cd ${.CURDIR}; \ ${MAKE} -f Makefile.inc1 _prereq_libs; \ ${MAKE} -f Makefile.inc1 _startup_libs; \ ${MAKE} -f Makefile.inc1 _prebuild_libs; \ ${MAKE} -f Makefile.inc1 _generic_libs; # # static libgcc.a prerequisite for shared libc # _prereq_libs= gnu/lib/libssp/libssp_nonshared gnu/lib/libgcc lib/libcompiler_rt # These dependencies are not automatically generated: # # gnu/lib/csu, gnu/lib/libgcc, lib/csu and lib/libc must be built before # all shared libraries for ELF. # _startup_libs= gnu/lib/csu .if exists(${.CURDIR}/lib/csu/${MACHINE_ARCH}-elf) _startup_libs+= lib/csu/${MACHINE_ARCH}-elf .elif exists(${.CURDIR}/lib/csu/${MACHINE_ARCH}) _startup_libs+= lib/csu/${MACHINE_ARCH} .else _startup_libs+= lib/csu/${MACHINE_CPUARCH} .endif _startup_libs+= gnu/lib/libgcc _startup_libs+= lib/libcompiler_rt _startup_libs+= lib/libc gnu/lib/libgcc__L: lib/libc__L _prebuild_libs= ${_kerberos5_lib_libasn1} ${_kerberos5_lib_libheimntlm} \ ${_kerberos5_lib_libhx509} ${_kerberos5_lib_libkrb5} \ ${_kerberos5_lib_libroken} \ lib/libbz2 lib/libcom_err lib/libcrypt \ lib/libexpat lib/libfetch \ ${_lib_libgssapi} ${_lib_libipx} \ lib/libkiconv lib/libkvm lib/liblzma lib/libmd \ lib/ncurses/ncurses lib/ncurses/ncursesw \ lib/libopie lib/libpam ${_lib_libthr} \ lib/libradius lib/libsbuf lib/libtacplus \ ${_cddl_lib_libumem} \ lib/libutil ${_lib_libypclnt} lib/libz lib/msun \ ${_secure_lib_libcrypto} ${_secure_lib_libssh} \ ${_secure_lib_libssl} .if ${MK_LIBTHR} != "no" _lib_libthr= lib/libthr .endif .if ${MK_OFED} != "no" _ofed_lib= contrib/ofed/usr.lib/ .endif _generic_libs= ${_cddl_lib} gnu/lib ${_kerberos5_lib} lib ${_secure_lib} usr.bin/lex/lib ${_ofed_lib} lib/libopie__L lib/libtacplus__L: lib/libmd__L .if ${MK_CDDL} != "no" _cddl_lib_libumem= cddl/lib/libumem _cddl_lib= cddl/lib .endif .if ${MK_CRYPT} != "no" .if ${MK_OPENSSL} != "no" _secure_lib_libcrypto= secure/lib/libcrypto _secure_lib_libssl= secure/lib/libssl lib/libradius__L secure/lib/libssl__L: secure/lib/libcrypto__L lib/libfetch__L: secure/lib/libcrypto__L secure/lib/libssl__L lib/libmd__L .if ${MK_OPENSSH} != "no" _secure_lib_libssh= secure/lib/libssh secure/lib/libssh__L: lib/libz__L secure/lib/libcrypto__L lib/libcrypt__L .if ${MK_KERBEROS_SUPPORT} != "no" secure/lib/libssh__L: lib/libgssapi__L kerberos5/lib/libkrb5__L \ kerberos5/lib/libhx509__L kerberos5/lib/libasn1__L lib/libcom_err__L \ lib/libmd__L kerberos5/lib/libroken__L .endif .endif .endif _secure_lib= secure/lib .endif .if ${MK_GSSAPI} != "no" _lib_libgssapi= lib/libgssapi .endif .if ${MK_IPX} != "no" _lib_libipx= lib/libipx .endif .if ${MK_KERBEROS} != "no" _kerberos5_lib= kerberos5/lib _kerberos5_lib_libasn1= kerberos5/lib/libasn1 _kerberos5_lib_libkrb5= kerberos5/lib/libkrb5 _kerberos5_lib_libhx509= kerberos5/lib/libhx509 _kerberos5_lib_libroken= kerberos5/lib/libroken _kerberos5_lib_libheimntlm= kerberos5/lib/libheimntlm .endif .if ${MK_NIS} != "no" _lib_libypclnt= lib/libypclnt .endif .if ${MK_OPENSSL} == "no" lib/libfetch__L lib/libradius__L: lib/libmd__L .endif .for _lib in ${_prereq_libs} ${_lib}__PL: .PHONY .if exists(${.CURDIR}/${_lib}) ${_+_}@${ECHODIR} "===> ${_lib} (obj,depend,all,install)"; \ cd ${.CURDIR}/${_lib}; \ ${MAKE} DIRPRFX=${_lib}/ obj; \ ${MAKE} DIRPRFX=${_lib}/ depend; \ ${MAKE} -DWITHOUT_PROFILE -DNO_PIC DIRPRFX=${_lib}/ all; \ ${MAKE} -DWITHOUT_PROFILE -DNO_PIC DIRPRFX=${_lib}/ install .endif .endfor .for _lib in ${_startup_libs} ${_prebuild_libs:Nlib/libpam} ${_generic_libs} ${_lib}__L: .PHONY .if exists(${.CURDIR}/${_lib}) ${_+_}@${ECHODIR} "===> ${_lib} (obj,depend,all,install)"; \ cd ${.CURDIR}/${_lib}; \ ${MAKE} DIRPRFX=${_lib}/ obj; \ ${MAKE} DIRPRFX=${_lib}/ depend; \ ${MAKE} DIRPRFX=${_lib}/ all; \ ${MAKE} DIRPRFX=${_lib}/ install .endif .endfor # libpam is special: we need to build static PAM modules before # static PAM library, and dynamic PAM library before dynamic PAM # modules. lib/libpam__L: .PHONY ${_+_}@${ECHODIR} "===> lib/libpam (obj,depend,all,install)"; \ cd ${.CURDIR}/lib/libpam; \ ${MAKE} DIRPRFX=lib/libpam/ obj; \ ${MAKE} DIRPRFX=lib/libpam/ depend; \ ${MAKE} DIRPRFX=lib/libpam/ -D_NO_LIBPAM_SO_YET all; \ ${MAKE} DIRPRFX=lib/libpam/ -D_NO_LIBPAM_SO_YET install _prereq_libs: ${_prereq_libs:S/$/__PL/} _startup_libs: ${_startup_libs:S/$/__L/} _prebuild_libs: ${_prebuild_libs:S/$/__L/} _generic_libs: ${_generic_libs:S/$/__L/} .for __target in all clean cleandepend cleandir depend includes obj .for entry in ${SUBDIR} ${entry}.${__target}__D: .PHONY ${_+_}@if test -d ${.CURDIR}/${entry}.${MACHINE_ARCH}; then \ ${ECHODIR} "===> ${DIRPRFX}${entry}.${MACHINE_ARCH} (${__target})"; \ edir=${entry}.${MACHINE_ARCH}; \ cd ${.CURDIR}/$${edir}; \ else \ ${ECHODIR} "===> ${DIRPRFX}${entry} (${__target})"; \ edir=${entry}; \ cd ${.CURDIR}/$${edir}; \ fi; \ ${MAKE} ${__target} DIRPRFX=${DIRPRFX}$${edir}/ .endfor par-${__target}: ${SUBDIR:S/$/.${__target}__D/} .endfor .include .if make(check-old) || make(check-old-dirs) || \ make(check-old-files) || make(check-old-libs) || \ make(delete-old) || make(delete-old-dirs) || \ make(delete-old-files) || make(delete-old-libs) # # check for / delete old files section # .include "ObsoleteFiles.inc" OLD_LIBS_MESSAGE="Please be sure no application still uses those libraries, \ else you can not start such an application. Consult UPDATING for more \ information regarding how to cope with the removal/revision bump of a \ specific library." .if !defined(BATCH_DELETE_OLD_FILES) RM_I=-i .else RM_I=-v .endif delete-old-files: @echo ">>> Removing old files (only deletes safe to delete libs)" # Ask for every old file if the user really wants to remove it. # It's annoying, but better safe than sorry. # NB: We cannot pass the list of OLD_FILES as a parameter because the # argument list will get too long. Using .for/.endfor make "loops" will make # the Makefile parser segfault. @exec 3<&0; \ ${MAKE} -f ${.CURDIR}/Makefile.inc1 ${.MAKEFLAGS} ${.TARGET} \ -V OLD_FILES -V "OLD_FILES:Musr/share/*.gz:R" | xargs -n1 | \ while read file; do \ if [ -f "${DESTDIR}/$${file}" -o -L "${DESTDIR}/$${file}" ]; then \ chflags noschg "${DESTDIR}/$${file}" 2>/dev/null || true; \ rm ${RM_I} "${DESTDIR}/$${file}" <&3; \ fi; \ done # Remove catpages without corresponding manpages. @exec 3<&0; \ find ${DESTDIR}/usr/share/man/cat* ! -type d | \ sed -ep -e's:${DESTDIR}/usr/share/man/cat:${DESTDIR}/usr/share/man/man:' | \ while read catpage; do \ read manpage; \ if [ ! -e "$${manpage}" ]; then \ rm ${RM_I} $${catpage} <&3; \ fi; \ done @echo ">>> Old files removed" check-old-files: @echo ">>> Checking for old files" @${MAKE} -f ${.CURDIR}/Makefile.inc1 ${.MAKEFLAGS} ${.TARGET} \ -V OLD_FILES -V "OLD_FILES:Musr/share/*.gz:R" | xargs -n1 | \ while read file; do \ if [ -f "${DESTDIR}/$${file}" -o -L "${DESTDIR}/$${file}" ]; then \ echo "${DESTDIR}/$${file}"; \ fi; \ done # Check for catpages without corresponding manpages. @find ${DESTDIR}/usr/share/man/cat* ! -type d | \ sed -ep -e's:${DESTDIR}/usr/share/man/cat:${DESTDIR}/usr/share/man/man:' | \ while read catpage; do \ read manpage; \ if [ ! -e "$${manpage}" ]; then \ echo $${catpage}; \ fi; \ done delete-old-libs: @echo ">>> Removing old libraries" @echo "${OLD_LIBS_MESSAGE}" | fmt @exec 3<&0; \ ${MAKE} -f ${.CURDIR}/Makefile.inc1 ${.MAKEFLAGS} ${.TARGET} \ -V OLD_LIBS | xargs -n1 | \ while read file; do \ if [ -f "${DESTDIR}/$${file}" -o -L "${DESTDIR}/$${file}" ]; then \ chflags noschg "${DESTDIR}/$${file}" 2>/dev/null || true; \ rm ${RM_I} "${DESTDIR}/$${file}" <&3; \ fi; \ done @echo ">>> Old libraries removed" check-old-libs: @echo ">>> Checking for old libraries" @${MAKE} -f ${.CURDIR}/Makefile.inc1 ${.MAKEFLAGS} ${.TARGET} \ -V OLD_LIBS | xargs -n1 | \ while read file; do \ if [ -f "${DESTDIR}/$${file}" -o -L "${DESTDIR}/$${file}" ]; then \ echo "${DESTDIR}/$${file}"; \ fi; \ done delete-old-dirs: @echo ">>> Removing old directories" @${MAKE} -f ${.CURDIR}/Makefile.inc1 ${.MAKEFLAGS} ${.TARGET} \ -V OLD_DIRS | xargs -n1 | \ while read dir; do \ if [ -d "${DESTDIR}/$${dir}" ]; then \ rmdir -v "${DESTDIR}/$${dir}" || true; \ elif [ -L "${DESTDIR}/$${dir}" ]; then \ echo "${DESTDIR}/$${dir} is a link, please remove everything manually."; \ fi; \ done @echo ">>> Old directories removed" check-old-dirs: @echo ">>> Checking for old directories" @${MAKE} -f ${.CURDIR}/Makefile.inc1 ${.MAKEFLAGS} ${.TARGET} \ -V OLD_DIRS | xargs -n1 | \ while read dir; do \ if [ -d "${DESTDIR}/$${dir}" ]; then \ echo "${DESTDIR}/$${dir}"; \ elif [ -L "${DESTDIR}/$${dir}" ]; then \ echo "${DESTDIR}/$${dir} is a link, please remove everything manually."; \ fi; \ done delete-old: delete-old-files delete-old-dirs @echo "To remove old libraries run '${MAKE} delete-old-libs'." check-old: check-old-files check-old-libs check-old-dirs @echo "To remove old files and directories run '${MAKE} delete-old'." @echo "To remove old libraries run '${MAKE} delete-old-libs'." .endif # # showconfig - show build configuration. # showconfig: @${MAKE} -n -f bsd.own.mk -V dummy -dg1 | grep ^MK_ | sort .if !empty(KRNLOBJDIR) && !empty(KERNCONF) DTBOUTPUTPATH= ${KRNLOBJDIR}/${KERNCONF}/ .if !defined(FDT_DTS_FILE) || empty(FDT_DTS_FILE) .if exists(${KERNCONFDIR}/${KERNCONF}) FDT_DTS_FILE!= awk 'BEGIN {FS="="} /^makeoptions[[:space:]]+FDT_DTS_FILE/ {print $$2}' \ ${KERNCONFDIR}/${KERNCONF} .endif .endif .endif .if !defined(DTBOUTPUTPATH) || !exists(${DTBOUTPUTPATH}) DTBOUTPUTPATH= ${.CURDIR} .endif # # Build 'standalone' Device Tree Blob # builddtb: @if [ "${FDT_DTS_FILE}" = "" ]; then \ echo "ERROR: FDT_DTS_FILE must be specified!"; \ exit 1; \ fi; \ if [ ! -f ${.CURDIR}/sys/boot/fdt/dts/${FDT_DTS_FILE} ]; then \ echo "ERROR: Specified DTS file (${FDT_DTS_FILE}) does not \ exist!"; \ exit 1; \ fi; \ if [ "${DTBOUTPUTPATH}" = "${.CURDIR}" ]; then \ echo "WARNING: DTB will be placed in the current working \ directory"; \ fi @PATH=${TMPPATH} \ dtc -O dtb -o \ ${DTBOUTPUTPATH}/`echo ${FDT_DTS_FILE} | cut -d. -f1`.dtb -b 0 \ -p 1024 ${.CURDIR}/sys/boot/fdt/dts/${FDT_DTS_FILE} ############### .if defined(XDEV) && defined(XDEV_ARCH) .if ${XDEV} == ${MACHINE} && ${XDEV_ARCH} == ${MACHINE_ARCH} XDEV_CPUTYPE?=${CPUTYPE} .else XDEV_CPUTYPE?=${TARGET_CPUTYPE} .endif NOFUN=-DNO_FSCHG -DWITHOUT_HTML -DWITHOUT_INFO -DNO_LINT \ -DWITHOUT_MAN -DWITHOUT_NLS -DWITHOUT_PROFILE \ -DWITHOUT_KERBEROS -DWITHOUT_RESCUE -DNO_WARNS \ TARGET=${XDEV} TARGET_ARCH=${XDEV_ARCH} \ CPUTYPE=${XDEV_CPUTYPE} XDDIR=${XDEV_ARCH}-freebsd XDTP=/usr/${XDDIR} CDBENV=MAKEOBJDIRPREFIX=${MAKEOBJDIRPREFIX}/${XDDIR} CDENV= ${CDBENV} \ _SHLIBDIRPREFIX=${XDTP} \ TOOLS_PREFIX=${XDTP} CD2ENV=${CDENV} \ MACHINE=${XDEV} MACHINE_ARCH=${XDEV_ARCH} CDTMP= ${MAKEOBJDIRPREFIX}/${XDEV}/${.CURDIR}/tmp CDMAKE=${CDENV} ${MAKE} ${NOFUN} CD2MAKE=${CD2ENV} PATH=${CDTMP}/usr/bin:${XDTP}/usr/bin:${PATH} ${MAKE} ${NOFUN} XDDESTDIR=${DESTDIR}${XDTP} .if !defined(OSREL) OSREL!= uname -r | sed -e 's/[-(].*//' .endif .ORDER: xdev-build xdev-install xdev: xdev-build xdev-install .ORDER: _xb-build-tools _xb-cross-tools xdev-build: _xb-build-tools _xb-cross-tools _xb-build-tools: ${_+_}@cd ${.CURDIR}; \ ${CDBENV} ${MAKE} -f Makefile.inc1 ${NOFUN} build-tools _xb-cross-tools: .for _tool in \ gnu/usr.bin/binutils \ gnu/usr.bin/cc ${_+_}@${ECHODIR} "===> xdev ${_tool} (obj,depend,all)"; \ cd ${.CURDIR}/${_tool}; \ ${CDMAKE} DIRPRFX=${_tool}/ obj; \ ${CDMAKE} DIRPRFX=${_tool}/ depend; \ ${CDMAKE} DIRPRFX=${_tool}/ all .endfor _xi-mtree: ${_+_}@${ECHODIR} "mtree populating ${XDDESTDIR}" mkdir -p ${XDDESTDIR} mtree -deU -f ${.CURDIR}/etc/mtree/BSD.root.dist \ -p ${XDDESTDIR} >/dev/null mtree -deU -f ${.CURDIR}/etc/mtree/BSD.usr.dist \ -p ${XDDESTDIR}/usr >/dev/null mtree -deU -f ${.CURDIR}/etc/mtree/BSD.include.dist \ -p ${XDDESTDIR}/usr/include >/dev/null .ORDER: _xi-mtree _xi-cross-tools _xi-includes _xi-libraries _xi-links xdev-install: _xi-mtree _xi-cross-tools _xi-includes _xi-libraries _xi-links _xi-cross-tools: @echo "_xi-cross-tools" .for _tool in \ gnu/usr.bin/binutils \ gnu/usr.bin/cc ${_+_}@${ECHODIR} "===> xdev ${_tool} (install)"; \ cd ${.CURDIR}/${_tool}; \ ${CDMAKE} DIRPRFX=${_tool}/ install DESTDIR=${XDDESTDIR} .endfor _xi-includes: ${_+_}cd ${.CURDIR}; ${CD2MAKE} -f Makefile.inc1 par-includes \ DESTDIR=${XDDESTDIR} _xi-libraries: ${_+_}cd ${.CURDIR}; ${CD2MAKE} -f Makefile.inc1 libraries \ DESTDIR=${XDDESTDIR} _xi-links: ${_+_}cd ${XDDESTDIR}/usr/bin; \ for i in *; do \ ln -sf ../../${XDTP}/usr/bin/$$i \ ../../../../usr/bin/${XDDIR}-$$i; \ ln -sf ../../${XDTP}/usr/bin/$$i \ ../../../../usr/bin/${XDDIR}${OSREL}-$$i; \ done .endif Index: projects/largeSMP/bin/sh/arith_yacc.c =================================================================== --- projects/largeSMP/bin/sh/arith_yacc.c (revision 221494) +++ projects/largeSMP/bin/sh/arith_yacc.c (revision 221495) @@ -1,378 +1,380 @@ /*- * Copyright (c) 1993 * The Regents of the University of California. All rights reserved. * Copyright (c) 2007 * Herbert Xu . All rights reserved. * * This code is derived from software contributed to Berkeley by * Kenneth Almquist. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include "arith.h" #include "arith_yacc.h" #include "expand.h" #include "shell.h" #include "error.h" #include "memalloc.h" #include "output.h" #include "options.h" #include "var.h" #if ARITH_BOR + 11 != ARITH_BORASS || ARITH_ASS + 11 != ARITH_EQ #error Arithmetic tokens are out of order. #endif static const char *arith_startbuf; const char *arith_buf; union yystype yylval; static int last_token; #define ARITH_PRECEDENCE(op, prec) [op - ARITH_BINOP_MIN] = prec static const char prec[ARITH_BINOP_MAX - ARITH_BINOP_MIN] = { ARITH_PRECEDENCE(ARITH_MUL, 0), ARITH_PRECEDENCE(ARITH_DIV, 0), ARITH_PRECEDENCE(ARITH_REM, 0), ARITH_PRECEDENCE(ARITH_ADD, 1), ARITH_PRECEDENCE(ARITH_SUB, 1), ARITH_PRECEDENCE(ARITH_LSHIFT, 2), ARITH_PRECEDENCE(ARITH_RSHIFT, 2), ARITH_PRECEDENCE(ARITH_LT, 3), ARITH_PRECEDENCE(ARITH_LE, 3), ARITH_PRECEDENCE(ARITH_GT, 3), ARITH_PRECEDENCE(ARITH_GE, 3), ARITH_PRECEDENCE(ARITH_EQ, 4), ARITH_PRECEDENCE(ARITH_NE, 4), ARITH_PRECEDENCE(ARITH_BAND, 5), ARITH_PRECEDENCE(ARITH_BXOR, 6), ARITH_PRECEDENCE(ARITH_BOR, 7), }; #define ARITH_MAX_PREC 8 static __dead2 void yyerror(const char *s) { error("arithmetic expression: %s: \"%s\"", s, arith_startbuf); /* NOTREACHED */ } static arith_t arith_lookupvarint(char *varname) { const char *str; char *p; arith_t result; str = lookupvar(varname); + if (uflag && str == NULL) + yyerror("variable not set"); if (str == NULL || *str == '\0') str = "0"; errno = 0; result = strtoarith_t(str, &p, 0); if (errno != 0 || *p != '\0') yyerror("variable conversion error"); return result; } static inline int arith_prec(int op) { return prec[op - ARITH_BINOP_MIN]; } static inline int higher_prec(int op1, int op2) { return arith_prec(op1) < arith_prec(op2); } static arith_t do_binop(int op, arith_t a, arith_t b) { switch (op) { default: case ARITH_REM: case ARITH_DIV: if (!b) yyerror("division by zero"); if (a == ARITH_MIN && b == -1) yyerror("divide error"); return op == ARITH_REM ? a % b : a / b; case ARITH_MUL: return a * b; case ARITH_ADD: return a + b; case ARITH_SUB: return a - b; case ARITH_LSHIFT: return a << b; case ARITH_RSHIFT: return a >> b; case ARITH_LT: return a < b; case ARITH_LE: return a <= b; case ARITH_GT: return a > b; case ARITH_GE: return a >= b; case ARITH_EQ: return a == b; case ARITH_NE: return a != b; case ARITH_BAND: return a & b; case ARITH_BXOR: return a ^ b; case ARITH_BOR: return a | b; } } static arith_t assignment(int var, int noeval); static arith_t primary(int token, union yystype *val, int op, int noeval) { arith_t result; again: switch (token) { case ARITH_LPAREN: result = assignment(op, noeval); if (last_token != ARITH_RPAREN) yyerror("expecting ')'"); last_token = yylex(); return result; case ARITH_NUM: last_token = op; return val->val; case ARITH_VAR: last_token = op; return noeval ? val->val : arith_lookupvarint(val->name); case ARITH_ADD: token = op; *val = yylval; op = yylex(); goto again; case ARITH_SUB: *val = yylval; return -primary(op, val, yylex(), noeval); case ARITH_NOT: *val = yylval; return !primary(op, val, yylex(), noeval); case ARITH_BNOT: *val = yylval; return ~primary(op, val, yylex(), noeval); default: yyerror("expecting primary"); } } static arith_t binop2(arith_t a, int op, int precedence, int noeval) { for (;;) { union yystype val; arith_t b; int op2; int token; token = yylex(); val = yylval; b = primary(token, &val, yylex(), noeval); op2 = last_token; if (op2 >= ARITH_BINOP_MIN && op2 < ARITH_BINOP_MAX && higher_prec(op2, op)) { b = binop2(b, op2, arith_prec(op), noeval); op2 = last_token; } a = noeval ? b : do_binop(op, a, b); if (op2 < ARITH_BINOP_MIN || op2 >= ARITH_BINOP_MAX || arith_prec(op2) >= precedence) return a; op = op2; } } static arith_t binop(int token, union yystype *val, int op, int noeval) { arith_t a = primary(token, val, op, noeval); op = last_token; if (op < ARITH_BINOP_MIN || op >= ARITH_BINOP_MAX) return a; return binop2(a, op, ARITH_MAX_PREC, noeval); } static arith_t and(int token, union yystype *val, int op, int noeval) { arith_t a = binop(token, val, op, noeval); arith_t b; op = last_token; if (op != ARITH_AND) return a; token = yylex(); *val = yylval; b = and(token, val, yylex(), noeval | !a); return a && b; } static arith_t or(int token, union yystype *val, int op, int noeval) { arith_t a = and(token, val, op, noeval); arith_t b; op = last_token; if (op != ARITH_OR) return a; token = yylex(); *val = yylval; b = or(token, val, yylex(), noeval | !!a); return a || b; } static arith_t cond(int token, union yystype *val, int op, int noeval) { arith_t a = or(token, val, op, noeval); arith_t b; arith_t c; if (last_token != ARITH_QMARK) return a; b = assignment(yylex(), noeval | !a); if (last_token != ARITH_COLON) yyerror("expecting ':'"); token = yylex(); *val = yylval; c = cond(token, val, yylex(), noeval | !!a); return a ? b : c; } static arith_t assignment(int var, int noeval) { union yystype val = yylval; int op = yylex(); arith_t result; char sresult[DIGITS(result) + 1]; if (var != ARITH_VAR) return cond(var, &val, op, noeval); if (op != ARITH_ASS && (op < ARITH_ASS_MIN || op >= ARITH_ASS_MAX)) return cond(var, &val, op, noeval); result = assignment(yylex(), noeval); if (noeval) return result; if (op != ARITH_ASS) result = do_binop(op - 11, arith_lookupvarint(val.name), result); snprintf(sresult, sizeof(sresult), ARITH_FORMAT_STR, result); setvar(val.name, sresult, 0); return result; } arith_t arith(const char *s) { struct stackmark smark; arith_t result; setstackmark(&smark); arith_buf = arith_startbuf = s; result = assignment(yylex(), 0); if (last_token) yyerror("expecting EOF"); popstackmark(&smark); return result; } /* * The exp(1) builtin. */ int expcmd(int argc, char **argv) { const char *p; char *concat; char **ap; arith_t i; if (argc > 1) { p = argv[1]; if (argc > 2) { /* * Concatenate arguments. */ STARTSTACKSTR(concat); ap = argv + 2; for (;;) { while (*p) STPUTC(*p++, concat); if ((p = *ap++) == NULL) break; STPUTC(' ', concat); } STPUTC('\0', concat); p = grabstackstr(concat); } } else p = ""; i = arith(p); out1fmt(ARITH_FORMAT_STR "\n", i); return !i; } Index: projects/largeSMP/bin/sh/parser.c =================================================================== --- projects/largeSMP/bin/sh/parser.c (revision 221494) +++ projects/largeSMP/bin/sh/parser.c (revision 221495) @@ -1,1849 +1,1851 @@ /*- * Copyright (c) 1991, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Kenneth Almquist. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef lint #if 0 static char sccsid[] = "@(#)parser.c 8.7 (Berkeley) 5/16/95"; #endif #endif /* not lint */ #include __FBSDID("$FreeBSD$"); #include #include #include #include "shell.h" #include "parser.h" #include "nodes.h" #include "expand.h" /* defines rmescapes() */ #include "syntax.h" #include "options.h" #include "input.h" #include "output.h" #include "var.h" #include "error.h" #include "memalloc.h" #include "mystring.h" #include "alias.h" #include "show.h" #include "eval.h" #include "exec.h" /* to check for special builtins */ #ifndef NO_HISTORY #include "myhistedit.h" #endif /* * Shell command parser. */ #define EOFMARKLEN 79 #define PROMPTLEN 128 /* values of checkkwd variable */ #define CHKALIAS 0x1 #define CHKKWD 0x2 #define CHKNL 0x4 /* values returned by readtoken */ #include "token.h" struct heredoc { struct heredoc *next; /* next here document in list */ union node *here; /* redirection node */ char *eofmark; /* string indicating end of input */ int striptabs; /* if set, strip leading tabs */ }; struct parser_temp { struct parser_temp *next; void *data; }; static struct heredoc *heredoclist; /* list of here documents to read */ static int doprompt; /* if set, prompt the user */ static int needprompt; /* true if interactive and at start of line */ static int lasttoken; /* last token read */ MKINIT int tokpushback; /* last token pushed back */ static char *wordtext; /* text of last word returned by readtoken */ MKINIT int checkkwd; /* 1 == check for kwds, 2 == also eat newlines */ static struct nodelist *backquotelist; static union node *redirnode; static struct heredoc *heredoc; static int quoteflag; /* set if (part of) last token was quoted */ static int startlinno; /* line # where last token started */ static int funclinno; /* line # where the current function started */ static struct parser_temp *parser_temp; static union node *list(int, int); static union node *andor(void); static union node *pipeline(void); static union node *command(void); static union node *simplecmd(union node **, union node *); static union node *makename(void); static void parsefname(void); static void parseheredoc(void); static int peektoken(void); static int readtoken(void); static int xxreadtoken(void); static int readtoken1(int, char const *, char *, int); static int noexpand(char *); static void synexpect(int) __dead2; static void synerror(const char *) __dead2; static void setprompt(int); static void * parser_temp_alloc(size_t len) { struct parser_temp *t; INTOFF; t = ckmalloc(sizeof(*t)); t->data = NULL; t->next = parser_temp; parser_temp = t; t->data = ckmalloc(len); INTON; return t->data; } static void * parser_temp_realloc(void *ptr, size_t len) { struct parser_temp *t; INTOFF; t = parser_temp; if (ptr != t->data) error("bug: parser_temp_realloc misused"); t->data = ckrealloc(t->data, len); INTON; return t->data; } static void parser_temp_free_upto(void *ptr) { struct parser_temp *t; int done = 0; INTOFF; while (parser_temp != NULL && !done) { t = parser_temp; parser_temp = t->next; done = t->data == ptr; ckfree(t->data); ckfree(t); } INTON; if (!done) error("bug: parser_temp_free_upto misused"); } static void parser_temp_free_all(void) { struct parser_temp *t; INTOFF; while (parser_temp != NULL) { t = parser_temp; parser_temp = t->next; ckfree(t->data); ckfree(t); } INTON; } /* * Read and parse a command. Returns NEOF on end of file. (NULL is a * valid parse tree indicating a blank line.) */ union node * parsecmd(int interact) { int t; /* This assumes the parser is not re-entered, * which could happen if we add command substitution on PS1/PS2. */ parser_temp_free_all(); heredoclist = NULL; tokpushback = 0; doprompt = interact; if (doprompt) setprompt(1); else setprompt(0); needprompt = 0; t = readtoken(); if (t == TEOF) return NEOF; if (t == TNL) return NULL; tokpushback++; return list(1, 1); } static union node * list(int nlflag, int erflag) { union node *ntop, *n1, *n2, *n3; int tok; checkkwd = CHKNL | CHKKWD | CHKALIAS; if (!nlflag && !erflag && tokendlist[peektoken()]) return NULL; ntop = n1 = NULL; for (;;) { n2 = andor(); tok = readtoken(); if (tok == TBACKGND) { if (n2->type == NCMD || n2->type == NPIPE) { n2->ncmd.backgnd = 1; } else if (n2->type == NREDIR) { n2->type = NBACKGND; } else { n3 = (union node *)stalloc(sizeof (struct nredir)); n3->type = NBACKGND; n3->nredir.n = n2; n3->nredir.redirect = NULL; n2 = n3; } } if (ntop == NULL) ntop = n2; else if (n1 == NULL) { n1 = (union node *)stalloc(sizeof (struct nbinary)); n1->type = NSEMI; n1->nbinary.ch1 = ntop; n1->nbinary.ch2 = n2; ntop = n1; } else { n3 = (union node *)stalloc(sizeof (struct nbinary)); n3->type = NSEMI; n3->nbinary.ch1 = n1->nbinary.ch2; n3->nbinary.ch2 = n2; n1->nbinary.ch2 = n3; n1 = n3; } switch (tok) { case TBACKGND: case TSEMI: tok = readtoken(); /* FALLTHROUGH */ case TNL: if (tok == TNL) { parseheredoc(); if (nlflag) return ntop; } else if (tok == TEOF && nlflag) { parseheredoc(); return ntop; } else { tokpushback++; } checkkwd = CHKNL | CHKKWD | CHKALIAS; if (!nlflag && !erflag && tokendlist[peektoken()]) return ntop; break; case TEOF: if (heredoclist) parseheredoc(); else pungetc(); /* push back EOF on input */ return ntop; default: if (nlflag || erflag) synexpect(-1); tokpushback++; return ntop; } } } static union node * andor(void) { union node *n1, *n2, *n3; int t; n1 = pipeline(); for (;;) { if ((t = readtoken()) == TAND) { t = NAND; } else if (t == TOR) { t = NOR; } else { tokpushback++; return n1; } n2 = pipeline(); n3 = (union node *)stalloc(sizeof (struct nbinary)); n3->type = t; n3->nbinary.ch1 = n1; n3->nbinary.ch2 = n2; n1 = n3; } } static union node * pipeline(void) { union node *n1, *n2, *pipenode; struct nodelist *lp, *prev; int negate, t; negate = 0; checkkwd = CHKNL | CHKKWD | CHKALIAS; TRACE(("pipeline: entered\n")); while (readtoken() == TNOT) negate = !negate; tokpushback++; n1 = command(); if (readtoken() == TPIPE) { pipenode = (union node *)stalloc(sizeof (struct npipe)); pipenode->type = NPIPE; pipenode->npipe.backgnd = 0; lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); pipenode->npipe.cmdlist = lp; lp->n = n1; do { prev = lp; lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); checkkwd = CHKNL | CHKKWD | CHKALIAS; t = readtoken(); tokpushback++; if (t == TNOT) lp->n = pipeline(); else lp->n = command(); prev->next = lp; } while (readtoken() == TPIPE); lp->next = NULL; n1 = pipenode; } tokpushback++; if (negate) { n2 = (union node *)stalloc(sizeof (struct nnot)); n2->type = NNOT; n2->nnot.com = n1; return n2; } else return n1; } static union node * command(void) { union node *n1, *n2; union node *ap, **app; union node *cp, **cpp; union node *redir, **rpp; int t; int is_subshell; checkkwd = CHKNL | CHKKWD | CHKALIAS; is_subshell = 0; redir = NULL; n1 = NULL; rpp = &redir; /* Check for redirection which may precede command */ while (readtoken() == TREDIR) { *rpp = n2 = redirnode; rpp = &n2->nfile.next; parsefname(); } tokpushback++; switch (readtoken()) { case TIF: n1 = (union node *)stalloc(sizeof (struct nif)); n1->type = NIF; if ((n1->nif.test = list(0, 0)) == NULL) synexpect(-1); if (readtoken() != TTHEN) synexpect(TTHEN); n1->nif.ifpart = list(0, 0); n2 = n1; while (readtoken() == TELIF) { n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif)); n2 = n2->nif.elsepart; n2->type = NIF; if ((n2->nif.test = list(0, 0)) == NULL) synexpect(-1); if (readtoken() != TTHEN) synexpect(TTHEN); n2->nif.ifpart = list(0, 0); } if (lasttoken == TELSE) n2->nif.elsepart = list(0, 0); else { n2->nif.elsepart = NULL; tokpushback++; } if (readtoken() != TFI) synexpect(TFI); checkkwd = CHKKWD | CHKALIAS; break; case TWHILE: case TUNTIL: { int got; n1 = (union node *)stalloc(sizeof (struct nbinary)); n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL; if ((n1->nbinary.ch1 = list(0, 0)) == NULL) synexpect(-1); if ((got=readtoken()) != TDO) { TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : "")); synexpect(TDO); } n1->nbinary.ch2 = list(0, 0); if (readtoken() != TDONE) synexpect(TDONE); checkkwd = CHKKWD | CHKALIAS; break; } case TFOR: if (readtoken() != TWORD || quoteflag || ! goodname(wordtext)) synerror("Bad for loop variable"); n1 = (union node *)stalloc(sizeof (struct nfor)); n1->type = NFOR; n1->nfor.var = wordtext; while (readtoken() == TNL) ; if (lasttoken == TWORD && ! quoteflag && equal(wordtext, "in")) { app = ≈ while (readtoken() == TWORD) { n2 = (union node *)stalloc(sizeof (struct narg)); n2->type = NARG; n2->narg.text = wordtext; n2->narg.backquote = backquotelist; *app = n2; app = &n2->narg.next; } *app = NULL; n1->nfor.args = ap; if (lasttoken != TNL && lasttoken != TSEMI) synexpect(-1); } else { static char argvars[5] = { CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0' }; n2 = (union node *)stalloc(sizeof (struct narg)); n2->type = NARG; n2->narg.text = argvars; n2->narg.backquote = NULL; n2->narg.next = NULL; n1->nfor.args = n2; /* * Newline or semicolon here is optional (but note * that the original Bourne shell only allowed NL). */ if (lasttoken != TNL && lasttoken != TSEMI) tokpushback++; } checkkwd = CHKNL | CHKKWD | CHKALIAS; if ((t = readtoken()) == TDO) t = TDONE; else if (t == TBEGIN) t = TEND; else synexpect(-1); n1->nfor.body = list(0, 0); if (readtoken() != t) synexpect(t); checkkwd = CHKKWD | CHKALIAS; break; case TCASE: n1 = (union node *)stalloc(sizeof (struct ncase)); n1->type = NCASE; if (readtoken() != TWORD) synexpect(TWORD); n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg)); n2->type = NARG; n2->narg.text = wordtext; n2->narg.backquote = backquotelist; n2->narg.next = NULL; while (readtoken() == TNL); if (lasttoken != TWORD || ! equal(wordtext, "in")) synerror("expecting \"in\""); cpp = &n1->ncase.cases; checkkwd = CHKNL | CHKKWD, readtoken(); while (lasttoken != TESAC) { *cpp = cp = (union node *)stalloc(sizeof (struct nclist)); cp->type = NCLIST; app = &cp->nclist.pattern; if (lasttoken == TLP) readtoken(); for (;;) { *app = ap = (union node *)stalloc(sizeof (struct narg)); ap->type = NARG; ap->narg.text = wordtext; ap->narg.backquote = backquotelist; checkkwd = CHKNL | CHKKWD; if (readtoken() != TPIPE) break; app = &ap->narg.next; readtoken(); } ap->narg.next = NULL; if (lasttoken != TRP) synexpect(TRP); cp->nclist.body = list(0, 0); checkkwd = CHKNL | CHKKWD | CHKALIAS; if ((t = readtoken()) != TESAC) { if (t != TENDCASE) synexpect(TENDCASE); else checkkwd = CHKNL | CHKKWD, readtoken(); } cpp = &cp->nclist.next; } *cpp = NULL; checkkwd = CHKKWD | CHKALIAS; break; case TLP: n1 = (union node *)stalloc(sizeof (struct nredir)); n1->type = NSUBSHELL; n1->nredir.n = list(0, 0); n1->nredir.redirect = NULL; if (readtoken() != TRP) synexpect(TRP); checkkwd = CHKKWD | CHKALIAS; is_subshell = 1; break; case TBEGIN: n1 = list(0, 0); if (readtoken() != TEND) synexpect(TEND); checkkwd = CHKKWD | CHKALIAS; break; /* Handle an empty command like other simple commands. */ case TBACKGND: case TSEMI: case TAND: case TOR: /* * An empty command before a ; doesn't make much sense, and * should certainly be disallowed in the case of `if ;'. */ if (!redir) synexpect(-1); case TNL: case TEOF: case TWORD: case TRP: tokpushback++; n1 = simplecmd(rpp, redir); return n1; default: synexpect(-1); } /* Now check for redirection which may follow command */ while (readtoken() == TREDIR) { *rpp = n2 = redirnode; rpp = &n2->nfile.next; parsefname(); } tokpushback++; *rpp = NULL; if (redir) { if (!is_subshell) { n2 = (union node *)stalloc(sizeof (struct nredir)); n2->type = NREDIR; n2->nredir.n = n1; n1 = n2; } n1->nredir.redirect = redir; } return n1; } static union node * simplecmd(union node **rpp, union node *redir) { union node *args, **app; union node **orig_rpp = rpp; union node *n = NULL; int special; /* If we don't have any redirections already, then we must reset */ /* rpp to be the address of the local redir variable. */ if (redir == 0) rpp = &redir; args = NULL; app = &args; /* * We save the incoming value, because we need this for shell * functions. There can not be a redirect or an argument between * the function name and the open parenthesis. */ orig_rpp = rpp; for (;;) { if (readtoken() == TWORD) { n = (union node *)stalloc(sizeof (struct narg)); n->type = NARG; n->narg.text = wordtext; n->narg.backquote = backquotelist; *app = n; app = &n->narg.next; } else if (lasttoken == TREDIR) { *rpp = n = redirnode; rpp = &n->nfile.next; parsefname(); /* read name of redirection file */ } else if (lasttoken == TLP && app == &args->narg.next && rpp == orig_rpp) { /* We have a function */ if (readtoken() != TRP) synexpect(TRP); funclinno = plinno; /* * - Require plain text. * - Functions with '/' cannot be called. * - Reject name=(). * - Reject ksh extended glob patterns. */ if (!noexpand(n->narg.text) || quoteflag || strchr(n->narg.text, '/') || strchr("!%*+-=?@}~", n->narg.text[strlen(n->narg.text) - 1])) synerror("Bad function name"); rmescapes(n->narg.text); if (find_builtin(n->narg.text, &special) >= 0 && special) synerror("Cannot override a special builtin with a function"); n->type = NDEFUN; n->narg.next = command(); funclinno = 0; return n; } else { tokpushback++; break; } } *app = NULL; *rpp = NULL; n = (union node *)stalloc(sizeof (struct ncmd)); n->type = NCMD; n->ncmd.backgnd = 0; n->ncmd.args = args; n->ncmd.redirect = redir; return n; } static union node * makename(void) { union node *n; n = (union node *)stalloc(sizeof (struct narg)); n->type = NARG; n->narg.next = NULL; n->narg.text = wordtext; n->narg.backquote = backquotelist; return n; } void fixredir(union node *n, const char *text, int err) { TRACE(("Fix redir %s %d\n", text, err)); if (!err) n->ndup.vname = NULL; if (is_digit(text[0]) && text[1] == '\0') n->ndup.dupfd = digit_val(text[0]); else if (text[0] == '-' && text[1] == '\0') n->ndup.dupfd = -1; else { if (err) synerror("Bad fd number"); else n->ndup.vname = makename(); } } static void parsefname(void) { union node *n = redirnode; if (readtoken() != TWORD) synexpect(-1); if (n->type == NHERE) { struct heredoc *here = heredoc; struct heredoc *p; int i; if (quoteflag == 0) n->type = NXHERE; TRACE(("Here document %d\n", n->type)); if (here->striptabs) { while (*wordtext == '\t') wordtext++; } if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN) synerror("Illegal eof marker for << redirection"); rmescapes(wordtext); here->eofmark = wordtext; here->next = NULL; if (heredoclist == NULL) heredoclist = here; else { for (p = heredoclist ; p->next ; p = p->next); p->next = here; } } else if (n->type == NTOFD || n->type == NFROMFD) { fixredir(n, wordtext, 0); } else { n->nfile.fname = makename(); } } /* * Input any here documents. */ static void parseheredoc(void) { struct heredoc *here; union node *n; while (heredoclist) { here = heredoclist; heredoclist = here->next; if (needprompt) { setprompt(2); needprompt = 0; } readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX, here->eofmark, here->striptabs); n = (union node *)stalloc(sizeof (struct narg)); n->narg.type = NARG; n->narg.next = NULL; n->narg.text = wordtext; n->narg.backquote = backquotelist; here->here->nhere.doc = n; } } static int peektoken(void) { int t; t = readtoken(); tokpushback++; return (t); } static int readtoken(void) { int t; struct alias *ap; #ifdef DEBUG int alreadyseen = tokpushback; #endif top: t = xxreadtoken(); /* * eat newlines */ if (checkkwd & CHKNL) { while (t == TNL) { parseheredoc(); t = xxreadtoken(); } } /* * check for keywords and aliases */ if (t == TWORD && !quoteflag) { const char * const *pp; if (checkkwd & CHKKWD) for (pp = parsekwd; *pp; pp++) { if (**pp == *wordtext && equal(*pp, wordtext)) { lasttoken = t = pp - parsekwd + KWDOFFSET; TRACE(("keyword %s recognized\n", tokname[t])); goto out; } } if (checkkwd & CHKALIAS && (ap = lookupalias(wordtext, 1)) != NULL) { pushstring(ap->val, strlen(ap->val), ap); goto top; } } out: if (t != TNOT) checkkwd = 0; #ifdef DEBUG if (!alreadyseen) TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); else TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); #endif return (t); } /* * Read the next input token. * If the token is a word, we set backquotelist to the list of cmds in * backquotes. We set quoteflag to true if any part of the word was * quoted. * If the token is TREDIR, then we set redirnode to a structure containing * the redirection. * In all cases, the variable startlinno is set to the number of the line * on which the token starts. * * [Change comment: here documents and internal procedures] * [Readtoken shouldn't have any arguments. Perhaps we should make the * word parsing code into a separate routine. In this case, readtoken * doesn't need to have any internal procedures, but parseword does. * We could also make parseoperator in essence the main routine, and * have parseword (readtoken1?) handle both words and redirection.] */ #define RETURN(token) return lasttoken = token static int xxreadtoken(void) { int c; if (tokpushback) { tokpushback = 0; return lasttoken; } if (needprompt) { setprompt(2); needprompt = 0; } startlinno = plinno; for (;;) { /* until token or start of word found */ c = pgetc_macro(); switch (c) { case ' ': case '\t': continue; case '#': while ((c = pgetc()) != '\n' && c != PEOF); pungetc(); continue; case '\\': if (pgetc() == '\n') { startlinno = ++plinno; if (doprompt) setprompt(2); else setprompt(0); continue; } pungetc(); goto breakloop; case '\n': plinno++; needprompt = doprompt; RETURN(TNL); case PEOF: RETURN(TEOF); case '&': if (pgetc() == '&') RETURN(TAND); pungetc(); RETURN(TBACKGND); case '|': if (pgetc() == '|') RETURN(TOR); pungetc(); RETURN(TPIPE); case ';': if (pgetc() == ';') RETURN(TENDCASE); pungetc(); RETURN(TSEMI); case '(': RETURN(TLP); case ')': RETURN(TRP); default: goto breakloop; } } breakloop: return readtoken1(c, BASESYNTAX, (char *)NULL, 0); #undef RETURN } #define MAXNEST_static 8 struct tokenstate { const char *syntax; /* *SYNTAX */ int parenlevel; /* levels of parentheses in arithmetic */ enum tokenstate_category { TSTATE_TOP, TSTATE_VAR_OLD, /* ${var+-=?}, inherits dquotes */ TSTATE_VAR_NEW, /* other ${var...}, own dquote state */ TSTATE_ARITH } category; }; /* * Called to parse command substitutions. */ static char * parsebackq(char *out, struct nodelist **pbqlist, int oldstyle, int dblquote, int quoted) { struct nodelist **nlpp; union node *n; char *volatile str; struct jmploc jmploc; struct jmploc *const savehandler = handler; int savelen; int saveprompt; const int bq_startlinno = plinno; char *volatile ostr = NULL; struct parsefile *const savetopfile = getcurrentfile(); struct heredoc *const saveheredoclist = heredoclist; struct heredoc *here; str = NULL; if (setjmp(jmploc.loc)) { popfilesupto(savetopfile); if (str) ckfree(str); if (ostr) ckfree(ostr); heredoclist = saveheredoclist; handler = savehandler; if (exception == EXERROR) { startlinno = bq_startlinno; synerror("Error in command substitution"); } longjmp(handler->loc, 1); } INTOFF; savelen = out - stackblock(); if (savelen > 0) { str = ckmalloc(savelen); memcpy(str, stackblock(), savelen); } handler = &jmploc; heredoclist = NULL; INTON; if (oldstyle) { /* We must read until the closing backquote, giving special treatment to some slashes, and then push the string and reread it as input, interpreting it normally. */ char *oout; int c; int olen; STARTSTACKSTR(oout); for (;;) { if (needprompt) { setprompt(2); needprompt = 0; } CHECKSTRSPACE(2, oout); switch (c = pgetc()) { case '`': goto done; case '\\': if ((c = pgetc()) == '\n') { plinno++; if (doprompt) setprompt(2); else setprompt(0); /* * If eating a newline, avoid putting * the newline into the new character * stream (via the USTPUTC after the * switch). */ continue; } if (c != '\\' && c != '`' && c != '$' && (!dblquote || c != '"')) USTPUTC('\\', oout); break; case '\n': plinno++; needprompt = doprompt; break; case PEOF: startlinno = plinno; synerror("EOF in backquote substitution"); break; default: break; } USTPUTC(c, oout); } done: USTPUTC('\0', oout); olen = oout - stackblock(); INTOFF; ostr = ckmalloc(olen); memcpy(ostr, stackblock(), olen); setinputstring(ostr, 1); INTON; } nlpp = pbqlist; while (*nlpp) nlpp = &(*nlpp)->next; *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist)); (*nlpp)->next = NULL; if (oldstyle) { saveprompt = doprompt; doprompt = 0; } n = list(0, oldstyle); if (oldstyle) doprompt = saveprompt; else { if (readtoken() != TRP) synexpect(TRP); } (*nlpp)->n = n; if (oldstyle) { /* * Start reading from old file again, ignoring any pushed back * tokens left from the backquote parsing */ popfile(); tokpushback = 0; } STARTSTACKSTR(out); CHECKSTRSPACE(savelen + 1, out); INTOFF; if (str) { memcpy(out, str, savelen); STADJUST(savelen, out); ckfree(str); str = NULL; } if (ostr) { ckfree(ostr); ostr = NULL; } here = saveheredoclist; if (here != NULL) { while (here->next != NULL) here = here->next; here->next = heredoclist; heredoclist = saveheredoclist; } handler = savehandler; INTON; if (quoted) USTPUTC(CTLBACKQ | CTLQUOTE, out); else USTPUTC(CTLBACKQ, out); return out; } /* * If eofmark is NULL, read a word or a redirection symbol. If eofmark * is not NULL, read a here document. In the latter case, eofmark is the * word which marks the end of the document and striptabs is true if * leading tabs should be stripped from the document. The argument firstc * is the first character of the input token or document. * * Because C does not have internal subroutines, I have simulated them * using goto's to implement the subroutine linkage. The following macros * will run code that appears at the end of readtoken1. */ #define CHECKEND() {goto checkend; checkend_return:;} #define PARSEREDIR() {goto parseredir; parseredir_return:;} #define PARSESUB() {goto parsesub; parsesub_return:;} #define PARSEARITH() {goto parsearith; parsearith_return:;} static int readtoken1(int firstc, char const *initialsyntax, char *eofmark, int striptabs) { int c = firstc; char *out; int len; char line[EOFMARKLEN + 1]; struct nodelist *bqlist; int quotef; int newvarnest; int level; int synentry; struct tokenstate state_static[MAXNEST_static]; int maxnest = MAXNEST_static; struct tokenstate *state = state_static; startlinno = plinno; quotef = 0; bqlist = NULL; newvarnest = 0; level = 0; state[level].syntax = initialsyntax; state[level].parenlevel = 0; state[level].category = TSTATE_TOP; STARTSTACKSTR(out); loop: { /* for each line, until end of word */ CHECKEND(); /* set c to PEOF if at end of here document */ for (;;) { /* until end of line or end of word */ CHECKSTRSPACE(4, out); /* permit 4 calls to USTPUTC */ synentry = state[level].syntax[c]; switch(synentry) { case CNL: /* '\n' */ if (state[level].syntax == BASESYNTAX) goto endword; /* exit outer loop */ USTPUTC(c, out); plinno++; if (doprompt) setprompt(2); else setprompt(0); c = pgetc(); goto loop; /* continue outer loop */ case CWORD: USTPUTC(c, out); break; case CCTL: if (eofmark == NULL || initialsyntax != SQSYNTAX) USTPUTC(CTLESC, out); USTPUTC(c, out); break; case CBACK: /* backslash */ c = pgetc(); if (c == PEOF) { USTPUTC('\\', out); pungetc(); } else if (c == '\n') { plinno++; if (doprompt) setprompt(2); else setprompt(0); } else { if (state[level].syntax == DQSYNTAX && c != '\\' && c != '`' && c != '$' && (c != '"' || (eofmark != NULL && newvarnest == 0)) && (c != '}' || state[level].category != TSTATE_VAR_OLD)) USTPUTC('\\', out); if ((eofmark == NULL || newvarnest > 0) && state[level].syntax == BASESYNTAX) USTPUTC(CTLQUOTEMARK, out); if (SQSYNTAX[c] == CCTL) USTPUTC(CTLESC, out); USTPUTC(c, out); if ((eofmark == NULL || newvarnest > 0) && state[level].syntax == BASESYNTAX && state[level].category == TSTATE_VAR_OLD) USTPUTC(CTLQUOTEEND, out); quotef++; } break; case CSQUOTE: USTPUTC(CTLQUOTEMARK, out); state[level].syntax = SQSYNTAX; break; case CDQUOTE: USTPUTC(CTLQUOTEMARK, out); state[level].syntax = DQSYNTAX; break; case CENDQUOTE: if (eofmark != NULL && newvarnest == 0) USTPUTC(c, out); else { if (state[level].category == TSTATE_VAR_OLD) USTPUTC(CTLQUOTEEND, out); state[level].syntax = BASESYNTAX; quotef++; } break; case CVAR: /* '$' */ PARSESUB(); /* parse substitution */ break; case CENDVAR: /* '}' */ if (level > 0 && ((state[level].category == TSTATE_VAR_OLD && state[level].syntax == state[level - 1].syntax) || (state[level].category == TSTATE_VAR_NEW && state[level].syntax == BASESYNTAX))) { if (state[level].category == TSTATE_VAR_NEW) newvarnest--; level--; USTPUTC(CTLENDVAR, out); } else { USTPUTC(c, out); } break; case CLP: /* '(' in arithmetic */ state[level].parenlevel++; USTPUTC(c, out); break; case CRP: /* ')' in arithmetic */ if (state[level].parenlevel > 0) { USTPUTC(c, out); --state[level].parenlevel; } else { if (pgetc() == ')') { if (level > 0 && state[level].category == TSTATE_ARITH) { level--; USTPUTC(CTLENDARI, out); } else USTPUTC(')', out); } else { /* * unbalanced parens * (don't 2nd guess - no error) */ pungetc(); USTPUTC(')', out); } } break; case CBQUOTE: /* '`' */ out = parsebackq(out, &bqlist, 1, state[level].syntax == DQSYNTAX && (eofmark == NULL || newvarnest > 0), state[level].syntax == DQSYNTAX || state[level].syntax == ARISYNTAX); break; case CEOF: goto endword; /* exit outer loop */ case CIGN: break; default: if (level == 0) goto endword; /* exit outer loop */ USTPUTC(c, out); } c = pgetc_macro(); } } endword: if (state[level].syntax == ARISYNTAX) synerror("Missing '))'"); if (state[level].syntax != BASESYNTAX && eofmark == NULL) synerror("Unterminated quoted string"); if (state[level].category == TSTATE_VAR_OLD || state[level].category == TSTATE_VAR_NEW) { startlinno = plinno; synerror("Missing '}'"); } if (state != state_static) parser_temp_free_upto(state); USTPUTC('\0', out); len = out - stackblock(); out = stackblock(); if (eofmark == NULL) { if ((c == '>' || c == '<') && quotef == 0 && len <= 2 && (*out == '\0' || is_digit(*out))) { PARSEREDIR(); return lasttoken = TREDIR; } else { pungetc(); } } quoteflag = quotef; backquotelist = bqlist; grabstackblock(len); wordtext = out; return lasttoken = TWORD; /* end of readtoken routine */ /* * Check to see whether we are at the end of the here document. When this * is called, c is set to the first character of the next input line. If * we are at the end of the here document, this routine sets the c to PEOF. */ checkend: { if (eofmark) { if (striptabs) { while (c == '\t') c = pgetc(); } if (c == *eofmark) { if (pfgets(line, sizeof line) != NULL) { char *p, *q; p = line; for (q = eofmark + 1 ; *q && *p == *q ; p++, q++); if (*p == '\n' && *q == '\0') { c = PEOF; plinno++; needprompt = doprompt; } else { pushstring(line, strlen(line), NULL); } } } } goto checkend_return; } /* * Parse a redirection operator. The variable "out" points to a string * specifying the fd to be redirected. The variable "c" contains the * first character of the redirection operator. */ parseredir: { char fd = *out; union node *np; np = (union node *)stalloc(sizeof (struct nfile)); if (c == '>') { np->nfile.fd = 1; c = pgetc(); if (c == '>') np->type = NAPPEND; else if (c == '&') np->type = NTOFD; else if (c == '|') np->type = NCLOBBER; else { np->type = NTO; pungetc(); } } else { /* c == '<' */ np->nfile.fd = 0; c = pgetc(); if (c == '<') { if (sizeof (struct nfile) != sizeof (struct nhere)) { np = (union node *)stalloc(sizeof (struct nhere)); np->nfile.fd = 0; } np->type = NHERE; heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc)); heredoc->here = np; if ((c = pgetc()) == '-') { heredoc->striptabs = 1; } else { heredoc->striptabs = 0; pungetc(); } } else if (c == '&') np->type = NFROMFD; else if (c == '>') np->type = NFROMTO; else { np->type = NFROM; pungetc(); } } if (fd != '\0') np->nfile.fd = digit_val(fd); redirnode = np; goto parseredir_return; } /* * Parse a substitution. At this point, we have read the dollar sign * and nothing else. */ parsesub: { char buf[10]; int subtype; int typeloc; int flags; char *p; static const char types[] = "}-+?="; int bracketed_name = 0; /* used to handle ${[0-9]*} variables */ int linno; int length; int c1; c = pgetc(); if (c != '(' && c != '{' && (is_eof(c) || !is_name(c)) && !is_special(c)) { USTPUTC('$', out); pungetc(); } else if (c == '(') { /* $(command) or $((arith)) */ if (pgetc() == '(') { PARSEARITH(); } else { pungetc(); out = parsebackq(out, &bqlist, 0, state[level].syntax == DQSYNTAX && (eofmark == NULL || newvarnest > 0), state[level].syntax == DQSYNTAX || state[level].syntax == ARISYNTAX); } } else { USTPUTC(CTLVAR, out); typeloc = out - stackblock(); USTPUTC(VSNORMAL, out); subtype = VSNORMAL; flags = 0; if (c == '{') { bracketed_name = 1; c = pgetc(); subtype = 0; } varname: if (!is_eof(c) && is_name(c)) { length = 0; do { STPUTC(c, out); c = pgetc(); length++; } while (!is_eof(c) && is_in_name(c)); if (length == 6 && strncmp(out - length, "LINENO", length) == 0) { /* Replace the variable name with the * current line number. */ linno = plinno; if (funclinno != 0) linno -= funclinno - 1; snprintf(buf, sizeof(buf), "%d", linno); STADJUST(-6, out); STPUTS(buf, out); flags |= VSLINENO; } } else if (is_digit(c)) { if (bracketed_name) { do { STPUTC(c, out); c = pgetc(); } while (is_digit(c)); } else { STPUTC(c, out); c = pgetc(); } } else if (is_special(c)) { c1 = c; c = pgetc(); if (subtype == 0 && c1 == '#') { subtype = VSLENGTH; if (strchr(types, c) == NULL && c != ':' && c != '#' && c != '%') goto varname; c1 = c; c = pgetc(); if (c1 != '}' && c == '}') { pungetc(); c = c1; goto varname; } pungetc(); c = c1; c1 = '#'; subtype = 0; } USTPUTC(c1, out); } else { subtype = VSERROR; if (c == '}') pungetc(); else if (c == '\n' || c == PEOF) synerror("Unexpected end of line in substitution"); else USTPUTC(c, out); } if (subtype == 0) { switch (c) { case ':': flags |= VSNUL; c = pgetc(); /*FALLTHROUGH*/ default: p = strchr(types, c); if (p == NULL) { if (c == '\n' || c == PEOF) synerror("Unexpected end of line in substitution"); if (flags == VSNUL) STPUTC(':', out); STPUTC(c, out); subtype = VSERROR; } else subtype = p - types + VSNORMAL; break; case '%': case '#': { int cc = c; subtype = c == '#' ? VSTRIMLEFT : VSTRIMRIGHT; c = pgetc(); if (c == cc) subtype++; else pungetc(); break; } } } else if (subtype != VSERROR) { + if (subtype == VSLENGTH && c != '}') + subtype = VSERROR; pungetc(); } STPUTC('=', out); if (state[level].syntax == DQSYNTAX || state[level].syntax == ARISYNTAX) flags |= VSQUOTE; *(stackblock() + typeloc) = subtype | flags; if (subtype != VSNORMAL) { if (level + 1 >= maxnest) { maxnest *= 2; if (state == state_static) { state = parser_temp_alloc( maxnest * sizeof(*state)); memcpy(state, state_static, MAXNEST_static * sizeof(*state)); } else state = parser_temp_realloc(state, maxnest * sizeof(*state)); } level++; state[level].parenlevel = 0; if (subtype == VSMINUS || subtype == VSPLUS || subtype == VSQUESTION || subtype == VSASSIGN) { /* * For operators that were in the Bourne shell, * inherit the double-quote state. */ state[level].syntax = state[level - 1].syntax; state[level].category = TSTATE_VAR_OLD; } else { /* * The other operators take a pattern, * so go to BASESYNTAX. * Also, ' and " are now special, even * in here documents. */ state[level].syntax = BASESYNTAX; state[level].category = TSTATE_VAR_NEW; newvarnest++; } } } goto parsesub_return; } /* * Parse an arithmetic expansion (indicate start of one and set state) */ parsearith: { if (level + 1 >= maxnest) { maxnest *= 2; if (state == state_static) { state = parser_temp_alloc( maxnest * sizeof(*state)); memcpy(state, state_static, MAXNEST_static * sizeof(*state)); } else state = parser_temp_realloc(state, maxnest * sizeof(*state)); } level++; state[level].syntax = ARISYNTAX; state[level].parenlevel = 0; state[level].category = TSTATE_ARITH; USTPUTC(CTLARI, out); if (state[level - 1].syntax == DQSYNTAX) USTPUTC('"',out); else USTPUTC(' ',out); goto parsearith_return; } } /* end of readtoken */ #ifdef mkinit RESET { tokpushback = 0; checkkwd = 0; } #endif /* * Returns true if the text contains nothing to expand (no dollar signs * or backquotes). */ static int noexpand(char *text) { char *p; char c; p = text; while ((c = *p++) != '\0') { if ( c == CTLQUOTEMARK) continue; if (c == CTLESC) p++; else if (BASESYNTAX[(int)c] == CCTL) return 0; } return 1; } /* * Return true if the argument is a legal variable name (a letter or * underscore followed by zero or more letters, underscores, and digits). */ int goodname(const char *name) { const char *p; p = name; if (! is_name(*p)) return 0; while (*++p) { if (! is_in_name(*p)) return 0; } return 1; } /* * Called when an unexpected token is read during the parse. The argument * is the token that is expected, or -1 if more than one type of token can * occur at this point. */ static void synexpect(int token) { char msg[64]; if (token >= 0) { fmtstr(msg, 64, "%s unexpected (expecting %s)", tokname[lasttoken], tokname[token]); } else { fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]); } synerror(msg); } static void synerror(const char *msg) { if (commandname) outfmt(out2, "%s: %d: ", commandname, startlinno); outfmt(out2, "Syntax error: %s\n", msg); error((char *)NULL); } static void setprompt(int which) { whichprompt = which; #ifndef NO_HISTORY if (!el) #endif { out2str(getprompt(NULL)); flushout(out2); } } /* * called by editline -- any expansions to the prompt * should be added here. */ char * getprompt(void *unused __unused) { static char ps[PROMPTLEN]; char *fmt; const char *pwd; int i, trim; static char internal_error[] = "??"; /* * Select prompt format. */ switch (whichprompt) { case 0: fmt = nullstr; break; case 1: fmt = ps1val(); break; case 2: fmt = ps2val(); break; default: return internal_error; } /* * Format prompt string. */ for (i = 0; (i < 127) && (*fmt != '\0'); i++, fmt++) if (*fmt == '\\') switch (*++fmt) { /* * Hostname. * * \h specifies just the local hostname, * \H specifies fully-qualified hostname. */ case 'h': case 'H': ps[i] = '\0'; gethostname(&ps[i], PROMPTLEN - i); /* Skip to end of hostname. */ trim = (*fmt == 'h') ? '.' : '\0'; while ((ps[i+1] != '\0') && (ps[i+1] != trim)) i++; break; /* * Working directory. * * \W specifies just the final component, * \w specifies the entire path. */ case 'W': case 'w': pwd = lookupvar("PWD"); if (pwd == NULL) pwd = "?"; if (*fmt == 'W' && *pwd == '/' && pwd[1] != '\0') strlcpy(&ps[i], strrchr(pwd, '/') + 1, PROMPTLEN - i); else strlcpy(&ps[i], pwd, PROMPTLEN - i); /* Skip to end of path. */ while (ps[i + 1] != '\0') i++; break; /* * Superuser status. * * '$' for normal users, '#' for root. */ case '$': ps[i] = (geteuid() != 0) ? '$' : '#'; break; /* * A literal \. */ case '\\': ps[i] = '\\'; break; /* * Emit unrecognized formats verbatim. */ default: ps[i++] = '\\'; ps[i] = *fmt; break; } else ps[i] = *fmt; ps[i] = '\0'; return (ps); } Index: projects/largeSMP/cddl/contrib/opensolaris =================================================================== --- projects/largeSMP/cddl/contrib/opensolaris (revision 221494) +++ projects/largeSMP/cddl/contrib/opensolaris (revision 221495) Property changes on: projects/largeSMP/cddl/contrib/opensolaris ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/cddl/contrib/opensolaris:r221444-221493 Index: projects/largeSMP/contrib/bind9 =================================================================== --- projects/largeSMP/contrib/bind9 (revision 221494) +++ projects/largeSMP/contrib/bind9 (revision 221495) Property changes on: projects/largeSMP/contrib/bind9 ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/bind9:r221444-221493 Index: projects/largeSMP/contrib/binutils =================================================================== --- projects/largeSMP/contrib/binutils (revision 221494) +++ projects/largeSMP/contrib/binutils (revision 221495) Property changes on: projects/largeSMP/contrib/binutils ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/binutils:r221444-221493 Index: projects/largeSMP/contrib/bzip2 =================================================================== --- projects/largeSMP/contrib/bzip2 (revision 221494) +++ projects/largeSMP/contrib/bzip2 (revision 221495) Property changes on: projects/largeSMP/contrib/bzip2 ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/bzip2:r221444-221493 Index: projects/largeSMP/contrib/dialog =================================================================== --- projects/largeSMP/contrib/dialog (revision 221494) +++ projects/largeSMP/contrib/dialog (revision 221495) Property changes on: projects/largeSMP/contrib/dialog ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/dialog:r221444-221493 Index: projects/largeSMP/contrib/ee =================================================================== --- projects/largeSMP/contrib/ee (revision 221494) +++ projects/largeSMP/contrib/ee (revision 221495) Property changes on: projects/largeSMP/contrib/ee ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/ee:r221444-221493 Index: projects/largeSMP/contrib/expat =================================================================== --- projects/largeSMP/contrib/expat (revision 221494) +++ projects/largeSMP/contrib/expat (revision 221495) Property changes on: projects/largeSMP/contrib/expat ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/expat:r221444-221493 Index: projects/largeSMP/contrib/file =================================================================== --- projects/largeSMP/contrib/file (revision 221494) +++ projects/largeSMP/contrib/file (revision 221495) Property changes on: projects/largeSMP/contrib/file ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/file:r221444-221493 Index: projects/largeSMP/contrib/gcc =================================================================== --- projects/largeSMP/contrib/gcc (revision 221494) +++ projects/largeSMP/contrib/gcc (revision 221495) Property changes on: projects/largeSMP/contrib/gcc ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/gcc:r221444-221493 Index: projects/largeSMP/contrib/gdb =================================================================== --- projects/largeSMP/contrib/gdb (revision 221494) +++ projects/largeSMP/contrib/gdb (revision 221495) Property changes on: projects/largeSMP/contrib/gdb ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/gdb:r221444-221493 Index: projects/largeSMP/contrib/gdtoa =================================================================== --- projects/largeSMP/contrib/gdtoa (revision 221494) +++ projects/largeSMP/contrib/gdtoa (revision 221495) Property changes on: projects/largeSMP/contrib/gdtoa ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/gdtoa:r221444-221493 Index: projects/largeSMP/contrib/gnu-sort =================================================================== --- projects/largeSMP/contrib/gnu-sort (revision 221494) +++ projects/largeSMP/contrib/gnu-sort (revision 221495) Property changes on: projects/largeSMP/contrib/gnu-sort ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/gnu-sort:r221444-221493 Index: projects/largeSMP/contrib/groff =================================================================== --- projects/largeSMP/contrib/groff (revision 221494) +++ projects/largeSMP/contrib/groff (revision 221495) Property changes on: projects/largeSMP/contrib/groff ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/groff:r221444-221493 Index: projects/largeSMP/contrib/less =================================================================== --- projects/largeSMP/contrib/less (revision 221494) +++ projects/largeSMP/contrib/less (revision 221495) Property changes on: projects/largeSMP/contrib/less ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/less:r221444-221493 Index: projects/largeSMP/contrib/libpcap =================================================================== --- projects/largeSMP/contrib/libpcap (revision 221494) +++ projects/largeSMP/contrib/libpcap (revision 221495) Property changes on: projects/largeSMP/contrib/libpcap ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/libpcap:r221444-221493 Index: projects/largeSMP/contrib/libstdc++ =================================================================== --- projects/largeSMP/contrib/libstdc++ (revision 221494) +++ projects/largeSMP/contrib/libstdc++ (revision 221495) Property changes on: projects/largeSMP/contrib/libstdc++ ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/libstdc++:r221444-221493 Index: projects/largeSMP/contrib/llvm/tools/clang =================================================================== --- projects/largeSMP/contrib/llvm/tools/clang (revision 221494) +++ projects/largeSMP/contrib/llvm/tools/clang (revision 221495) Property changes on: projects/largeSMP/contrib/llvm/tools/clang ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/llvm/tools/clang:r221444-221493 Index: projects/largeSMP/contrib/llvm =================================================================== --- projects/largeSMP/contrib/llvm (revision 221494) +++ projects/largeSMP/contrib/llvm (revision 221495) Property changes on: projects/largeSMP/contrib/llvm ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/llvm:r221444-221493 Index: projects/largeSMP/contrib/ncurses =================================================================== --- projects/largeSMP/contrib/ncurses (revision 221494) +++ projects/largeSMP/contrib/ncurses (revision 221495) Property changes on: projects/largeSMP/contrib/ncurses ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/ncurses:r221444-221493 Index: projects/largeSMP/contrib/netcat =================================================================== --- projects/largeSMP/contrib/netcat (revision 221494) +++ projects/largeSMP/contrib/netcat (revision 221495) Property changes on: projects/largeSMP/contrib/netcat ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/netcat:r221444-221493 Index: projects/largeSMP/contrib/ntp =================================================================== --- projects/largeSMP/contrib/ntp (revision 221494) +++ projects/largeSMP/contrib/ntp (revision 221495) Property changes on: projects/largeSMP/contrib/ntp ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/ntp:r221444-221493 Index: projects/largeSMP/contrib/one-true-awk =================================================================== --- projects/largeSMP/contrib/one-true-awk (revision 221494) +++ projects/largeSMP/contrib/one-true-awk (revision 221495) Property changes on: projects/largeSMP/contrib/one-true-awk ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/one-true-awk:r221444-221493 Index: projects/largeSMP/contrib/openbsm =================================================================== --- projects/largeSMP/contrib/openbsm (revision 221494) +++ projects/largeSMP/contrib/openbsm (revision 221495) Property changes on: projects/largeSMP/contrib/openbsm ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/openbsm:r221444-221493 Index: projects/largeSMP/contrib/openpam =================================================================== --- projects/largeSMP/contrib/openpam (revision 221494) +++ projects/largeSMP/contrib/openpam (revision 221495) Property changes on: projects/largeSMP/contrib/openpam ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/openpam:r221444-221493 Index: projects/largeSMP/contrib/pf =================================================================== --- projects/largeSMP/contrib/pf (revision 221494) +++ projects/largeSMP/contrib/pf (revision 221495) Property changes on: projects/largeSMP/contrib/pf ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/pf:r221444-221493 Index: projects/largeSMP/contrib/sendmail =================================================================== --- projects/largeSMP/contrib/sendmail (revision 221494) +++ projects/largeSMP/contrib/sendmail (revision 221495) Property changes on: projects/largeSMP/contrib/sendmail ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/sendmail:r221444-221493 Index: projects/largeSMP/contrib/tcpdump =================================================================== --- projects/largeSMP/contrib/tcpdump (revision 221494) +++ projects/largeSMP/contrib/tcpdump (revision 221495) Property changes on: projects/largeSMP/contrib/tcpdump ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/tcpdump:r221444-221493 Index: projects/largeSMP/contrib/tcsh =================================================================== --- projects/largeSMP/contrib/tcsh (revision 221494) +++ projects/largeSMP/contrib/tcsh (revision 221495) Property changes on: projects/largeSMP/contrib/tcsh ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/tcsh:r221444-221493 Index: projects/largeSMP/contrib/top/install-sh =================================================================== --- projects/largeSMP/contrib/top/install-sh (revision 221494) +++ projects/largeSMP/contrib/top/install-sh (revision 221495) Property changes on: projects/largeSMP/contrib/top/install-sh ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/top/install-sh:r221444-221493 Index: projects/largeSMP/contrib/top =================================================================== --- projects/largeSMP/contrib/top (revision 221494) +++ projects/largeSMP/contrib/top (revision 221495) Property changes on: projects/largeSMP/contrib/top ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/top:r221444-221493 Index: projects/largeSMP/contrib/tzcode/stdtime =================================================================== --- projects/largeSMP/contrib/tzcode/stdtime (revision 221494) +++ projects/largeSMP/contrib/tzcode/stdtime (revision 221495) Property changes on: projects/largeSMP/contrib/tzcode/stdtime ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/tzcode/stdtime:r221444-221493 Index: projects/largeSMP/contrib/tzcode/zic =================================================================== --- projects/largeSMP/contrib/tzcode/zic (revision 221494) +++ projects/largeSMP/contrib/tzcode/zic (revision 221495) Property changes on: projects/largeSMP/contrib/tzcode/zic ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/tzcode/zic:r221444-221493 Index: projects/largeSMP/contrib/tzdata =================================================================== --- projects/largeSMP/contrib/tzdata (revision 221494) +++ projects/largeSMP/contrib/tzdata (revision 221495) Property changes on: projects/largeSMP/contrib/tzdata ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/tzdata:r221444-221493 Index: projects/largeSMP/contrib/wpa =================================================================== --- projects/largeSMP/contrib/wpa (revision 221494) +++ projects/largeSMP/contrib/wpa (revision 221495) Property changes on: projects/largeSMP/contrib/wpa ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/wpa:r221444-221493 Index: projects/largeSMP/contrib/xz =================================================================== --- projects/largeSMP/contrib/xz (revision 221494) +++ projects/largeSMP/contrib/xz (revision 221495) Property changes on: projects/largeSMP/contrib/xz ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/xz:r221444-221493 Index: projects/largeSMP/crypto/openssh/ChangeLog =================================================================== --- projects/largeSMP/crypto/openssh/ChangeLog (revision 221494) +++ projects/largeSMP/crypto/openssh/ChangeLog (revision 221495) @@ -1,1258 +1,1259 @@ 20110403 - (djm) [README contrib/caldera/openssh.spec contrib/redhat/openssh.spec] [contrib/suse/openssh.spec] Prepare for 5.8p2 release. + - (djm) [version.h] crank version - Release 5.8p2 20110329 - (djm) [entropy.c] closefrom() before running ssh-rand-helper; leftover fds noticed by tmraz AT redhat.com 20110221 - (dtucker) [contrib/cygwin/ssh-host-config] From Corinna: revamp of the Cygwin-specific service installer script ssh-host-config. The actual functionality is the same, the revisited version is just more exact when it comes to check for problems which disallow to run certain aspects of the script. So, part of this script and the also rearranged service helper script library "csih" is to check if all the tools required to run the script are available on the system. The new script also is more thorough to inform the user why the script failed. Patch from vinschen at redhat com. 20110206 - (dtucker) [openbsd-compat/port-linux.c] Bug #1851: fix syntax error in selinux code. Patch from Leonardo Chiquitto - (dtucker) [contrib/cygwin/ssh-{host,user}-config] Add ECDSA key generation and simplify. Patch from Corinna Vinschen. 20110204 - OpenBSD CVS Sync - djm@cvs.openbsd.org 2011/01/31 21:42:15 [PROTOCOL.mux] cut'n'pasto; from bert.wesarg AT googlemail.com - djm@cvs.openbsd.org 2011/02/04 00:44:21 [key.c] fix uninitialised nonce variable; reported by Mateusz Kocielski - djm@cvs.openbsd.org 2011/02/04 00:44:43 [version.h] openssh-5.8 - (djm) [README contrib/caldera/openssh.spec contrib/redhat/openssh.spec] [contrib/suse/openssh.spec] update versions in docs and spec files. - Release OpenSSH 5.8p1 20110128 - (djm) [openbsd-compat/port-linux.c] Check whether SELinux is enabled before attempting setfscreatecon(). Check whether matchpathcon() succeeded before using its result. Patch from cjwatson AT debian.org; bz#1851 20110125 - (djm) [configure.ac Makefile.in ssh.c openbsd-compat/port-linux.c openbsd-compat/port-linux.h] Move SELinux-specific code from ssh.c to port-linux.c to avoid compilation errors. Add -lselinux to ssh when building with SELinux support to avoid linking failure; report from amk AT spamfence.net; ok dtucker 20110122 - (dtucker) [configure.ac openbsd-compat/openssl-compat.{c,h}] Add RSA_get_default_method() for the benefit of openssl versions that don't have it (at least openssl-engine-0.9.6b). Found and tested by Kevin Brott, ok djm@. - OpenBSD CVS Sync - djm@cvs.openbsd.org 2011/01/22 09:18:53 [version.h] crank to OpenSSH-5.7 - (djm) [README contrib/caldera/openssh.spec contrib/redhat/openssh.spec] [contrib/suse/openssh.spec] update versions in docs and spec files. - (djm) Release 5.7p1 20110119 - (tim) [contrib/caldera/openssh.spec] Use CFLAGS from Makefile instead of RPM so build completes. Signatures were changed to .asc since 4.1p1. - (djm) [configure.ac] Disable ECC on OpenSSL <0.9.8g. Releases prior to 0.9.8 lacked it, and 0.9.8a through 0.9.8d have proven buggy in pre- release testing (random crashes and failure to load ECC keys). ok dtucker@ 20110117 - (djm) [regress/Makefile] use $TEST_SSH_KEYGEN instead of the one in $PATH, fix cleanup of droppings; reported by openssh AT roumenpetrov.info; ok dtucker@ - (djm) [regress/agent-ptrace.sh] Fix false failure on OS X by adding its unique snowflake of a gdb error to the ones we look for. - (djm) [regress/agent-getpeereid.sh] leave stdout attached when running ssh-add to avoid $SUDO failures on Linux - (dtucker) [openbsd-compat/port-linux.c] Bug #1838: Add support for the new Linux OOM-killer magic values that changed in 2.6.36 kernels, with fallback to the old values. Feedback from vapier at gentoo org and djm, ok djm. - (djm) [configure.ac regress/agent-getpeereid.sh regress/multiplex.sh] [regress/sftp-glob.sh regress/test-exec.sh] Rework how feature tests are disabled on platforms that do not support them; add a "config_defined()" shell function that greps for defines in config.h and use them to decide on feature tests. Convert a couple of existing grep's over config.h to use the new function Add a define "FILESYSTEM_NO_BACKSLASH" for filesystem that can't represent backslash characters in filenames, enable it for Cygwin and use it to turn of tests for quotes backslashes in sftp-glob.sh. based on discussion with vinschen AT redhat.com and dtucker@; ok dtucker@ - (tim) [regress/agent-getpeereid.sh] shell portability fix. - (dtucker) [openbsd-compat/port-linux.c] Fix minor bug caught by -Werror on the tinderbox. - (dtucker) [LICENCE Makefile.in audit-bsm.c audit-linux.c audit.c audit.h configure.ac defines.h loginrec.c] Bug #1402: add linux audit subsystem support, based on patches from Tomas Mraz and jchadima at redhat. 20110116 - (dtucker) [Makefile.in configure.ac regress/kextype.sh] Skip sha256-based on configurations that don't have it. - OpenBSD CVS Sync - djm@cvs.openbsd.org 2011/01/16 11:50:05 [clientloop.c] Use atomicio when flushing protocol 1 std{out,err} buffers at session close. This was a latent bug exposed by setting a SIGCHLD handler and spotted by kevin.brott AT gmail.com; ok dtucker@ - djm@cvs.openbsd.org 2011/01/16 11:50:36 [sshconnect.c] reset the SIGPIPE handler when forking to execute child processes; ok dtucker@ - djm@cvs.openbsd.org 2011/01/16 12:05:59 [clientloop.c] a couple more tweaks to the post-close protocol 1 stderr/stdout flush: now that we use atomicio(), convert them from while loops to if statements add test and cast to compile cleanly with -Wsigned 20110114 - OpenBSD CVS Sync - djm@cvs.openbsd.org 2011/01/13 21:54:53 [mux.c] correct error messages; patch from bert.wesarg AT googlemail.com - djm@cvs.openbsd.org 2011/01/13 21:55:25 [PROTOCOL.mux] correct protocol names and add a couple of missing protocol number defines; patch from bert.wesarg AT googlemail.com - (djm) [Makefile.in] Use shell test to disable ecdsa key generating in host-key-force target rather than a substitution that is replaced with a comment so that the Makefile.in is still a syntactically valid Makefile (useful to run the distprep target) - (tim) [regress/cert-hostkey.sh] Typo. Missing $ on variable name. - (tim) [regress/cert-hostkey.sh] Add missing TEST_SSH_ECC guard around some ecdsa bits. 20110113 - (djm) [misc.c] include time.h for nanosleep() prototype - (tim) [Makefile.in] test the ECC bits if we have the capability. ok djm - (tim) [Makefile.in configure.ac opensshd.init.in] Add support for generating ecdsa keys. ok djm. - (djm) [entropy.c] cast OPENSSL_VERSION_NUMBER to u_long to avoid gcc warning on platforms where it defaults to int - (djm) [regress/Makefile] add a few more generated files to the clean target - (djm) [myproposal.h] Fix reversed OPENSSL_VERSION_NUMBER test and bad #define that was causing diffie-hellman-group-exchange-sha256 to be incorrectly disabled - (djm) [regress/kextype.sh] Testing diffie-hellman-group-exchange-sha256 should not depend on ECC support 20110112 - OpenBSD CVS Sync - nicm@cvs.openbsd.org 2010/10/08 21:48:42 [openbsd-compat/glob.c] Extend GLOB_LIMIT to cover readdir and stat and bump the malloc limit from ARG_MAX to 64K. Fixes glob-using programs (notably ftp) able to be triggered to hit resource limits. Idea from a similar NetBSD change, original problem reported by jasper@. ok millert tedu jasper - djm@cvs.openbsd.org 2011/01/12 01:53:14 avoid some integer overflows mostly with GLOB_APPEND and GLOB_DOOFFS and sanity check arguments (these will be unnecessary when we switch struct glob members from being type into to size_t in the future); "looks ok" tedu@ feedback guenther@ - (djm) [configure.ac] Turn on -Wno-unused-result for gcc >= 4.4 to avoid silly warnings on write() calls we don't care succeed or not. - (djm) [configure.ac] Fix broken test for gcc >= 4.4 with per-compiler flag tests that don't depend on gcc version at all; suggested by and ok dtucker@ 20110111 - (tim) [regress/host-expand.sh] Fix for building outside of read only source tree. - (djm) [platform.c] Some missing includes that show up under -Werror - OpenBSD CVS Sync - djm@cvs.openbsd.org 2011/01/08 10:51:51 [clientloop.c] use host and not options.hostname, as the latter may have unescaped substitution characters - djm@cvs.openbsd.org 2011/01/11 06:06:09 [sshlogin.c] fd leak on error paths; from zinovik@ NB. Id sync only; we use loginrec.c that was also audited and fixed recently - djm@cvs.openbsd.org 2011/01/11 06:13:10 [clientloop.c ssh-keygen.c sshd.c] some unsigned long long casts that make things a bit easier for portable without resorting to dropping PRIu64 formats everywhere 20110109 - (djm) [Makefile.in] list ssh_host_ecdsa key in PATHSUBS; spotted by openssh AT roumenpetrov.info 20110108 - (djm) [regress/keytype.sh] s/echo -n/echon/ to repair failing regress test on OSX and others. Reported by imorgan AT nas.nasa.gov 20110107 - (djm) [regress/cert-hostkey.sh regress/cert-userkey.sh] fix shell test for no-ECC case. Patch from cristian.ionescu-idbohrn AT axis.com - djm@cvs.openbsd.org 2011/01/06 22:23:53 [ssh.c] unbreak %n expansion in LocalCommand; patch from bert.wesarg AT googlemail.com; ok markus@ - djm@cvs.openbsd.org 2011/01/06 22:23:02 [clientloop.c] when exiting due to ServerAliveTimeout, mention the hostname that caused it (useful with backgrounded controlmaster) - djm@cvs.openbsd.org 2011/01/06 22:46:21 [regress/Makefile regress/host-expand.sh] regress test for LocalCommand %n expansion from bert.wesarg AT googlemail.com; ok markus@ - djm@cvs.openbsd.org 2011/01/06 23:01:35 [sshconnect.c] reset SIGCHLD handler to SIG_DFL when execuring LocalCommand; ok markus@ 20110106 - (djm) OpenBSD CVS Sync - markus@cvs.openbsd.org 2010/12/08 22:46:03 [scp.1 scp.c] add a new -3 option to scp: Copies between two remote hosts are transferred through the local host. Without this option the data is copied directly between the two remote hosts. ok djm@ (bugzilla #1837) - jmc@cvs.openbsd.org 2010/12/09 14:13:33 [scp.1 scp.c] scp.1: grammer fix scp.c: add -3 to usage() - markus@cvs.openbsd.org 2010/12/14 11:59:06 [sshconnect.c] don't mention key type in key-changed-warning, since we also print this warning if a new key type appears. ok djm@ - djm@cvs.openbsd.org 2010/12/15 00:49:27 [readpass.c] fix ControlMaster=ask regression reset SIGCHLD handler before fork (and restore it after) so we don't miss the the askpass child's exit status. Correct test for exit status/signal to account for waitpid() failure; with claudio@ ok claudio@ markus@ - djm@cvs.openbsd.org 2010/12/24 21:41:48 [auth-options.c] don't send the actual forced command in a debug message; ok markus deraadt - otto@cvs.openbsd.org 2011/01/04 20:44:13 [ssh-keyscan.c] handle ecdsa-sha2 with various key lengths; hint and ok djm@ 20110104 - (djm) [configure.ac Makefile.in] Use mandoc as preferred manpage formatter if it is present, followed by nroff and groff respectively. Fixes distprep target on OpenBSD (which has bumped groff/nroff to ports in favour of mandoc). feedback and ok tim 20110103 - (djm) [Makefile.in] revert local hack I didn't intend to commit 20110102 - (djm) [loginrec.c] Fix some fd leaks on error paths. ok dtucker - (djm) [configure.ac] Check whether libdes is needed when building with Heimdal krb5 support. On OpenBSD this library no longer exists, so linking it unconditionally causes a build failure; ok dtucker 20101226 - (dtucker) OpenBSD CVS Sync - djm@cvs.openbsd.org 2010/12/08 04:02:47 [ssh_config.5 sshd_config.5] explain that IPQoS arguments are separated by whitespace; iirc requested by jmc@ a while back 20101205 - (dtucker) openbsd-compat/openssl-compat.c] remove sleep leftover from debugging. Spotted by djm. - (dtucker) OpenBSD CVS Sync - djm@cvs.openbsd.org 2010/12/03 23:49:26 [schnorr.c] check that g^x^q === 1 mod p; recommended by JPAKE author Feng Hao (this code is still disabled, but apprently people are treating it as a reference implementation) - djm@cvs.openbsd.org 2010/12/03 23:55:27 [auth-rsa.c] move check for revoked keys to run earlier (in auth_rsa_key_allowed) bz#1829; patch from ldv AT altlinux.org; ok markus@ - djm@cvs.openbsd.org 2010/12/04 00:18:01 [sftp-server.c sftp.1 sftp-client.h sftp.c PROTOCOL sftp-client.c] add a protocol extension to support a hard link operation. It is available through the "ln" command in the client. The old "ln" behaviour of creating a symlink is available using its "-s" option or through the preexisting "symlink" command; based on a patch from miklos AT szeredi.hu in bz#1555; ok markus@ - djm@cvs.openbsd.org 2010/12/04 13:31:37 [hostfile.c] fix fd leak; spotted and ok dtucker - djm@cvs.openbsd.org 2010/12/04 00:21:19 [regress/sftp-cmds.sh] adjust for hard-link support - (dtucker) [regress/Makefile] Id sync. 20101204 - (djm) [openbsd-compat/bindresvport.c] Use arc4random_uniform(range) instead of (arc4random() % range) - (dtucker) [configure.ac moduli.c openbsd-compat/openssl-compat.{c,h}] Add shims for the new, non-deprecated OpenSSL key generation functions for platforms that don't have the new interfaces. 20101201 - OpenBSD CVS Sync - deraadt@cvs.openbsd.org 2010/11/20 05:12:38 [auth2-pubkey.c] clean up cases of ;; - djm@cvs.openbsd.org 2010/11/21 01:01:13 [clientloop.c misc.c misc.h ssh-agent.1 ssh-agent.c] honour $TMPDIR for client xauth and ssh-agent temporary directories; feedback and ok markus@ - djm@cvs.openbsd.org 2010/11/21 10:57:07 [authfile.c] Refactor internals of private key loading and saving to work on memory buffers rather than directly on files. This will make a few things easier to do in the future; ok markus@ - djm@cvs.openbsd.org 2010/11/23 02:35:50 [auth.c] use strict_modes already passed as function argument over referencing global options.strict_modes - djm@cvs.openbsd.org 2010/11/23 23:57:24 [clientloop.c] avoid NULL deref on receiving a channel request on an unknown or invalid channel; report bz#1842 from jchadima AT redhat.com; ok dtucker@ - djm@cvs.openbsd.org 2010/11/24 01:24:14 [channels.c] remove a debug() that pollutes stderr on client connecting to a server in debug mode (channel_close_fds is called transitively from the session code post-fork); bz#1719, ok dtucker - djm@cvs.openbsd.org 2010/11/25 04:10:09 [session.c] replace close() loop for fds 3->64 with closefrom(); ok markus deraadt dtucker - djm@cvs.openbsd.org 2010/11/26 05:52:49 [scp.c] Pass through ssh command-line flags and options when doing remote-remote transfers, e.g. to enable agent forwarding which is particularly useful in this case; bz#1837 ok dtucker@ - markus@cvs.openbsd.org 2010/11/29 18:57:04 [authfile.c] correctly load comment for encrypted rsa1 keys; report/fix Joachim Schipper; ok djm@ - djm@cvs.openbsd.org 2010/11/29 23:45:51 [auth.c hostfile.c hostfile.h ssh.c ssh_config.5 sshconnect.c] [sshconnect.h sshconnect2.c] automatically order the hostkeys requested by the client based on which hostkeys are already recorded in known_hosts. This avoids hostkey warnings when connecting to servers with new ECDSA keys that are preferred by default; with markus@ 20101124 - (dtucker) [platform.c session.c] Move the getluid call out of session.c and into the platform-specific code Only affects SCO, tested by and ok tim@. - (djm) [loginrec.c] Relax permission requirement on btmp logs to allow group read/write. ok dtucker@ - (dtucker) [packet.c] Remove redundant local declaration of "int tos". - (djm) [defines.h] Add IP DSCP defines 20101122 - (dtucker) Bug #1840: fix warning when configuring --with-ssl-engine, patch from vapier at gentoo org. 20101120 - OpenBSD CVS Sync - djm@cvs.openbsd.org 2010/11/05 02:46:47 [packet.c] whitespace KNF - djm@cvs.openbsd.org 2010/11/10 01:33:07 [kexdhc.c kexdhs.c kexgexc.c kexgexs.c key.c moduli.c] use only libcrypto APIs that are retained with OPENSSL_NO_DEPRECATED. these have been around for years by this time. ok markus - djm@cvs.openbsd.org 2010/11/13 23:27:51 [clientloop.c misc.c misc.h packet.c packet.h readconf.c readconf.h] [servconf.c servconf.h session.c ssh.c ssh_config.5 sshd_config.5] allow ssh and sshd to set arbitrary TOS/DSCP/QoS values instead of hardcoding lowdelay/throughput. bz#1733 patch from philipp AT redfish-solutions.com; ok markus@ deraadt@ - jmc@cvs.openbsd.org 2010/11/15 07:40:14 [ssh_config.5] libary -> library; - jmc@cvs.openbsd.org 2010/11/18 15:01:00 [scp.1 sftp.1 ssh.1 sshd_config.5] add IPQoS to the various -o lists, and zap some trailing whitespace; 20101111 - (djm) [servconf.c ssh-add.c ssh-keygen.c] don't look for ECDSA keys on platforms that don't support ECC. Fixes some spurious warnings reported by tim@ 20101109 - (tim) [regress/kextype.sh] Not all platforms have time in /usr/bin. Feedback from dtucker@ - (tim) [configure.ac openbsd-compat/bsd-misc.h openbsd-compat/bsd-misc.c] Add support for platforms missing isblank(). ok djm@ 20101108 - (tim) [regress/Makefile] Fixes to allow building/testing outside source tree. - (tim) [regress/kextype.sh] Shell portability fix. 20101107 - (dtucker) [platform.c] includes.h instead of defines.h so that we get the correct typedefs. 20101105 - (djm) [loginrec.c loginrec.h] Use correct uid_t/pid_t types instead of int. Should fix bz#1817 cleanly; ok dtucker@ - OpenBSD CVS Sync - djm@cvs.openbsd.org 2010/09/22 12:26:05 [regress/Makefile regress/kextype.sh] regress test for each of the key exchange algorithms that we support - djm@cvs.openbsd.org 2010/10/28 11:22:09 [authfile.c key.c key.h ssh-keygen.c] fix a possible NULL deref on loading a corrupt ECDH key store ECDH group information in private keys files as "named groups" rather than as a set of explicit group parameters (by setting the OPENSSL_EC_NAMED_CURVE flag). This makes for shorter key files and retrieves the group's OpenSSL NID that we need for various things. - jmc@cvs.openbsd.org 2010/10/28 18:33:28 [scp.1 ssh-add.1 ssh-keygen.1 ssh.1 ssh_config.5 sshd.8 sshd_config.5] knock out some "-*- nroff -*-" lines; - djm@cvs.openbsd.org 2010/11/04 02:45:34 [sftp-server.c] umask should be parsed as octal. reported by candland AT xmission.com; ok markus@ - (dtucker) [configure.ac platform.{c,h} session.c openbsd-compat/port-solaris.{c,h}] Bug #1824: Add Solaris Project support. Patch from cory.erickson at csu mnscu edu with a bit of rework from me. ok djm@ - (dtucker) [platform.c platform.h session.c] Add a platform hook to run after the user's groups are established and move the selinux calls into it. - (dtucker) [platform.c session.c] Move the AIX setpcred+chroot hack into platform.c - (dtucker) [platform.c session.c] Move the BSDI setpgrp into platform.c. - (dtucker) [platform.c] Only call setpgrp on BSDI if running as root to retain previous behavior. - (dtucker) [platform.c session.c] Move the PAM credential establishment for the LOGIN_CAP case into platform.c. - (dtucker) platform.c session.c] Move the USE_LIBIAF fragment into platform.c - (dtucker) [platform.c session.c] Move aix_usrinfo frament into platform.c. - (dtucker) [platform.c session.c] Move irix setusercontext fragment into platform.c. - (dtucker) [platform.c session.c] Move PAM credential establishment for the non-LOGIN_CAP case into platform.c. - (dtucker) [platform.c platform.h session.c] Move the Cygwin special-case check into platform.c - (dtucker) [regress/keytype.sh] Import new test. - (dtucker) [Makefile configure.ac regress/Makefile regress/keytype.sh] Import recent changes to regress/Makefile, pass a flag to enable ECC tests from configure through to regress/Makefile and use it in the tests. - (dtucker) [regress/kextype.sh] Add missing "test". - (dtucker) [regress/kextype.sh] Make sha256 test depend on ECC. This is not strictly correct since while ECC requires sha256 the reverse is not true however it does prevent spurious test failures. - (dtucker) [platform.c] Need servconf.h and extern options. 20101025 - (tim) [openbsd-compat/glob.h] Remove sys/cdefs.h include that came with 1.12 to unbreak Solaris build. ok djm@ - (dtucker) [defines.h] Use SIZE_T_MAX for SIZE_MAX for platforms that have a native one. 20101024 - (dtucker) [includes.h] Add missing ifdef GLOB_HAS_GL_STATV to fix build. - (dtucker) [regress/cert-hostkey.sh] Disable ECC-based tests on platforms which don't have ECC support in libcrypto. - (dtucker) [regress/cert-userkey.sh] Disable ECC-based tests on platforms which don't have ECC support in libcrypto. - (dtucker) [defines.h] Add SIZE_MAX for the benefit of platforms that don't have it. - (dtucker) OpenBSD CVS Sync - sthen@cvs.openbsd.org 2010/10/23 22:06:12 [sftp.c] escape '[' in filename tab-completion; fix a type while there. ok djm@ 20101021 - OpenBSD CVS Sync - dtucker@cvs.openbsd.org 2010/10/12 02:22:24 [mux.c] Typo in confirmation message. bz#1827, patch from imorgan at nas nasa gov - djm@cvs.openbsd.org 2010/08/31 12:24:09 [regress/cert-hostkey.sh regress/cert-userkey.sh] tests for ECDSA certificates 20101011 - (djm) [canohost.c] Zero a4 instead of addr to better match type. bz#1825, reported by foo AT mailinator.com - (djm) [sshconnect.c] Need signal.h for prototype for kill(2) 20101011 - (djm) [configure.ac] Use = instead of == in shell tests. Patch from dr AT vasco.com 20101007 - (djm) [ssh-agent.c] Fix type for curve name. - (djm) OpenBSD CVS Sync - matthew@cvs.openbsd.org 2010/09/24 13:33:00 [misc.c misc.h configure.ac openbsd-compat/openbsd-compat.h] [openbsd-compat/timingsafe_bcmp.c] Add timingsafe_bcmp(3) to libc, mention that it's already in the kernel in kern(9), and remove it from OpenSSH. ok deraadt@, djm@ NB. re-added under openbsd-compat/ for portable OpenSSH - djm@cvs.openbsd.org 2010/09/25 09:30:16 [sftp.c configure.ac openbsd-compat/glob.c openbsd-compat/glob.h] make use of new glob(3) GLOB_KEEPSTAT extension to save extra server rountrips to fetch per-file stat(2) information. NB. update openbsd-compat/ glob(3) implementation from OpenBSD libc to match. - djm@cvs.openbsd.org 2010/09/26 22:26:33 [sftp.c] when performing an "ls" in columnated (short) mode, only call ioctl(TIOCGWINSZ) once to get the window width instead of per- filename - djm@cvs.openbsd.org 2010/09/30 11:04:51 [servconf.c] prevent free() of string in .rodata when overriding AuthorizedKeys in a Match block; patch from rein AT basefarm.no - djm@cvs.openbsd.org 2010/10/01 23:05:32 [cipher-3des1.c cipher-bf1.c cipher-ctr.c openbsd-compat/openssl-compat.h] adapt to API changes in openssl-1.0.0a NB. contains compat code to select correct API for older OpenSSL - djm@cvs.openbsd.org 2010/10/05 05:13:18 [sftp.c sshconnect.c] use default shell /bin/sh if $SHELL is ""; ok markus@ - djm@cvs.openbsd.org 2010/10/06 06:39:28 [clientloop.c ssh.c sshconnect.c sshconnect.h] kill proxy command on fatal() (we already kill it on clean exit); ok markus@ - djm@cvs.openbsd.org 2010/10/06 21:10:21 [sshconnect.c] swapped args to kill(2) - (djm) [openbsd-compat/glob.c] restore ARG_MAX compat code. - (djm) [cipher-acss.c] Add missing header. - (djm) [openbsd-compat/Makefile.in] Actually link timingsafe_bcmp 20100924 - (djm) OpenBSD CVS Sync - naddy@cvs.openbsd.org 2010/09/10 15:19:29 [ssh-keygen.1] * mention ECDSA in more places * less repetition in FILES section * SSHv1 keys are still encrypted with 3DES help and ok jmc@ - djm@cvs.openbsd.org 2010/09/11 21:44:20 [ssh.1] mention RFC 5656 for ECC stuff - jmc@cvs.openbsd.org 2010/09/19 21:30:05 [sftp.1] more wacky macro fixing; - djm@cvs.openbsd.org 2010/09/20 04:41:47 [ssh.c] install a SIGCHLD handler to reap expiried child process; ok markus@ - djm@cvs.openbsd.org 2010/09/20 04:50:53 [jpake.c schnorr.c] check that received values are smaller than the group size in the disabled and unfinished J-PAKE code. avoids catastrophic security failure found by Sebastien Martini - djm@cvs.openbsd.org 2010/09/20 04:54:07 [jpake.c] missing #include - djm@cvs.openbsd.org 2010/09/20 07:19:27 [mux.c] "atomically" create the listening mux socket by binding it on a temorary name and then linking it into position after listen() has succeeded. this allows the mux clients to determine that the server socket is either ready or stale without races. stale server sockets are now automatically removed ok deraadt - djm@cvs.openbsd.org 2010/09/22 05:01:30 [kex.c kex.h kexecdh.c kexecdhc.c kexecdhs.c readconf.c readconf.h] [servconf.c servconf.h ssh_config.5 sshconnect2.c sshd.c sshd_config.5] add a KexAlgorithms knob to the client and server configuration to allow selection of which key exchange methods are used by ssh(1) and sshd(8) and their order of preference. ok markus@ - jmc@cvs.openbsd.org 2010/09/22 08:30:08 [ssh.1 ssh_config.5] ssh.1: add kexalgorithms to the -o list ssh_config.5: format the kexalgorithms in a more consistent (prettier!) way ok djm - djm@cvs.openbsd.org 2010/09/22 22:58:51 [atomicio.c atomicio.h misc.c misc.h scp.c sftp-client.c] [sftp-client.h sftp.1 sftp.c] add an option per-read/write callback to atomicio factor out bandwidth limiting code from scp(1) into a generic bandwidth limiter that can be attached using the atomicio callback mechanism add a bandwidth limit option to sftp(1) using the above "very nice" markus@ - jmc@cvs.openbsd.org 2010/09/23 13:34:43 [sftp.c] add [-l limit] to usage(); - jmc@cvs.openbsd.org 2010/09/23 13:36:46 [scp.1 sftp.1] add KexAlgorithms to the -o list; 20100910 - (dtucker) [openbsd-compat/port-linux.c] Check is_selinux_enabled for exact return code since it can apparently return -1 under some conditions. From openssh bugs werbittewas de, ok djm@ - OpenBSD CVS Sync - djm@cvs.openbsd.org 2010/08/31 12:33:38 [ssh-add.c ssh-agent.c ssh-keygen.c ssh-keysign.c ssh.c sshd.c] reintroduce commit from tedu@, which I pulled out for release engineering: OpenSSL_add_all_algorithms is the name of the function we have a man page for, so use that. ok djm - jmc@cvs.openbsd.org 2010/08/31 17:40:54 [ssh-agent.1] fix some macro abuse; - jmc@cvs.openbsd.org 2010/08/31 21:14:58 [ssh.1] small text tweak to accommodate previous; - naddy@cvs.openbsd.org 2010/09/01 15:21:35 [servconf.c] pick up ECDSA host key by default; ok djm@ - markus@cvs.openbsd.org 2010/09/02 16:07:25 [ssh-keygen.c] permit -b 256, 384 or 521 as key size for ECDSA; ok djm@ - markus@cvs.openbsd.org 2010/09/02 16:08:39 [ssh.c] unbreak ControlPersist=yes for ControlMaster=yes; ok djm@ - naddy@cvs.openbsd.org 2010/09/02 17:21:50 [ssh-keygen.c] Switch ECDSA default key size to 256 bits, which according to RFC5656 should still be better than our current RSA-2048 default. ok djm@, markus@ - jmc@cvs.openbsd.org 2010/09/03 11:09:29 [scp.1] add an EXIT STATUS section for /usr/bin; - jmc@cvs.openbsd.org 2010/09/04 09:38:34 [ssh-add.1 ssh.1] two more EXIT STATUS sections; - naddy@cvs.openbsd.org 2010/09/06 17:10:19 [sshd_config] add ssh_host_ecdsa_key to /etc; from Mattieu Baptiste ok deraadt@ - djm@cvs.openbsd.org 2010/09/08 03:54:36 [authfile.c] typo - deraadt@cvs.openbsd.org 2010/09/08 04:13:31 [compress.c] work around name-space collisions some buggy compilers (looking at you gcc, at least in earlier versions, but this does not forgive your current transgressions) seen between zlib and openssl ok djm - djm@cvs.openbsd.org 2010/09/09 10:45:45 [kex.c kex.h kexecdh.c key.c key.h monitor.c ssh-ecdsa.c] ECDH/ECDSA compliance fix: these methods vary the hash function they use (SHA256/384/512) depending on the length of the curve in use. The previous code incorrectly used SHA256 in all cases. This fix will cause authentication failure when using 384 or 521-bit curve keys if one peer hasn't been upgraded and the other has. (256-bit curve keys work ok). In particular you may need to specify HostkeyAlgorithms when connecting to a server that has not been upgraded from an upgraded client. ok naddy@ - (djm) [authfd.c authfile.c bufec.c buffer.h configure.ac kex.h kexecdh.c] [kexecdhc.c kexecdhs.c key.c key.h myproposal.h packet.c readconf.c] [ssh-agent.c ssh-ecdsa.c ssh-keygen.c ssh.c] Disable ECDH and ECDSA on platforms that don't have the requisite OpenSSL support. ok dtucker@ - (dtucker) [kex.h key.c packet.h ssh-agent.c ssh.c] A few more ECC ifdefs for missing headers and compiler warnings. 20100831 - OpenBSD CVS Sync - jmc@cvs.openbsd.org 2010/08/08 19:36:30 [ssh-keysign.8 ssh.1 sshd.8] use the same template for all FILES sections; i.e. -compact/.Pp where we have multiple items, and .Pa for path names; - tedu@cvs.openbsd.org 2010/08/12 23:34:39 [ssh-add.c ssh-agent.c ssh-keygen.c ssh-keysign.c ssh.c sshd.c] OpenSSL_add_all_algorithms is the name of the function we have a man page for, so use that. ok djm - djm@cvs.openbsd.org 2010/08/16 04:06:06 [ssh-add.c ssh-agent.c ssh-keygen.c ssh-keysign.c ssh.c sshd.c] backout previous temporarily; discussed with deraadt@ - djm@cvs.openbsd.org 2010/08/31 09:58:37 [auth-options.c auth1.c auth2.c bufaux.c buffer.h kex.c key.c packet.c] [packet.h ssh-dss.c ssh-rsa.c] Add buffer_get_cstring() and related functions that verify that the string extracted from the buffer contains no embedded \0 characters* This prevents random (possibly malicious) crap from being appended to strings where it would not be noticed if the string is used with a string(3) function. Use the new API in a few sensitive places. * actually, we allow a single one at the end of the string for now because we don't know how many deployed implementations get this wrong, but don't count on this to remain indefinitely. - djm@cvs.openbsd.org 2010/08/31 11:54:45 [PROTOCOL PROTOCOL.agent PROTOCOL.certkeys auth2-jpake.c authfd.c] [authfile.c buffer.h dns.c kex.c kex.h key.c key.h monitor.c] [monitor_wrap.c myproposal.h packet.c packet.h pathnames.h readconf.c] [ssh-add.1 ssh-add.c ssh-agent.1 ssh-agent.c ssh-keygen.1 ssh-keygen.c] [ssh-keyscan.1 ssh-keyscan.c ssh-keysign.8 ssh.1 ssh.c ssh2.h] [ssh_config.5 sshconnect.c sshconnect2.c sshd.8 sshd.c sshd_config.5] [uuencode.c uuencode.h bufec.c kexecdh.c kexecdhc.c kexecdhs.c ssh-ecdsa.c] Implement Elliptic Curve Cryptography modes for key exchange (ECDH) and host/user keys (ECDSA) as specified by RFC5656. ECDH and ECDSA offer better performance than plain DH and DSA at the same equivalent symmetric key length, as well as much shorter keys. Only the mandatory sections of RFC5656 are implemented, specifically the three REQUIRED curves nistp256, nistp384 and nistp521 and only ECDH and ECDSA. Point compression (optional in RFC5656 is NOT implemented). Certificate host and user keys using the new ECDSA key types are supported. Note that this code has not been tested for interoperability and may be subject to change. feedback and ok markus@ - (djm) [Makefile.in] Add new ECC files - (djm) [bufec.c kexecdh.c kexecdhc.c kexecdhs.c ssh-ecdsa.c] include includes.h 20100827 - (dtucker) [contrib/redhat/sshd.init] Bug #1810: initlog is deprecated, remove. Patch from martynas at venck us 20100823 - (djm) Release OpenSSH-5.6p1 20100816 - (dtucker) [configure.ac openbsd-compat/Makefile.in openbsd-compat/openbsd-compat.h openbsd-compat/strptime.c] Add strptime to the compat library which helps on platforms like old IRIX. Based on work by djm, tested by Tom Christensen. - OpenBSD CVS Sync - djm@cvs.openbsd.org 2010/08/12 21:49:44 [ssh.c] close any extra file descriptors inherited from parent at start and reopen stdin/stdout to /dev/null when forking for ControlPersist. prevents tools that fork and run a captive ssh for communication from failing to exit when the ssh completes while they wait for these fds to close. The inherited fds may persist arbitrarily long if a background mux master has been started by ControlPersist. cvs and scp were effected by this. "please commit" markus@ - (djm) [regress/README.regress] typo 20100812 - (tim) [regress/login-timeout.sh regress/reconfigure.sh regress/reexec.sh regress/test-exec.sh] Under certain conditions when testing with sudo tests would fail because the pidfile could not be read by a regular user. "cat: cannot open ...../regress/pidfile: Permission denied (error 13)" Make sure cat is run by $SUDO. no objection from me. djm@ - (tim) [auth.c] add cast to quiet compiler. Change only affects SVR5 systems. 20100809 - (djm) bz#1561: don't bother setting IFF_UP on tun(4) device if it is already set. Makes FreeBSD user openable tunnels useful; patch from richard.burakowski+ossh AT mrburak.net, ok dtucker@ - (dtucker) bug #1530: strip trailing ":" from hostname in ssh-copy-id. based in part on a patch from Colin Watson, ok djm@ 20100809 - OpenBSD CVS Sync - djm@cvs.openbsd.org 2010/08/08 16:26:42 [version.h] crank to 5.6 - (djm) [README contrib/caldera/openssh.spec contrib/redhat/openssh.spec] [contrib/suse/openssh.spec] Crank version numbers 20100805 - OpenBSD CVS Sync - djm@cvs.openbsd.org 2010/08/04 05:37:01 [ssh.1 ssh_config.5 sshd.8] Remove mentions of weird "addr/port" alternate address format for IPv6 addresses combinations. It hasn't worked for ages and we have supported the more commen "[addr]:port" format for a long time. ok jmc@ markus@ - djm@cvs.openbsd.org 2010/08/04 05:40:39 [PROTOCOL.certkeys ssh-keygen.c] tighten the rules for certificate encoding by requiring that options appear in lexical order and make our ssh-keygen comply. ok markus@ - djm@cvs.openbsd.org 2010/08/04 05:42:47 [auth.c auth2-hostbased.c authfile.c authfile.h ssh-keysign.8] [ssh-keysign.c ssh.c] enable certificates for hostbased authentication, from Iain Morgan; "looks ok" markus@ - djm@cvs.openbsd.org 2010/08/04 05:49:22 [authfile.c] commited the wrong version of the hostbased certificate diff; this version replaces some strlc{py,at} verbosity with xasprintf() at the request of markus@ - djm@cvs.openbsd.org 2010/08/04 06:07:11 [ssh-keygen.1 ssh-keygen.c] Support CA keys in PKCS#11 tokens; feedback and ok markus@ - djm@cvs.openbsd.org 2010/08/04 06:08:40 [ssh-keysign.c] clean for -Wuninitialized (Id sync only; portable had this change) - djm@cvs.openbsd.org 2010/08/05 13:08:42 [channels.c] Fix a trio of bugs in the local/remote window calculation for datagram data channels (i.e. TunnelForward): Calculate local_consumed correctly in channel_handle_wfd() by measuring the delta to buffer_len(c->output) from when we start to when we finish. The proximal problem here is that the output_filter we use in portable modified the length of the dequeued datagram (to futz with the headers for !OpenBSD). In channel_output_poll(), don't enqueue datagrams that won't fit in the peer's advertised packet size (highly unlikely to ever occur) or which won't fit in the peer's remaining window (more likely). In channel_input_data(), account for the 4-byte string header in datagram packets that we accept from the peer and enqueue in c->output. report, analysis and testing 2/3 cases from wierbows AT us.ibm.com; "looks good" markus@ 20100803 - (dtucker) [monitor.c] Bug #1795: Initialize the values to be returned from PAM to sane values in case the PAM method doesn't write to them. Spotted by Bitman Zhou, ok djm@. - OpenBSD CVS Sync - djm@cvs.openbsd.org 2010/07/16 04:45:30 [ssh-keygen.c] avoid bogus compiler warning - djm@cvs.openbsd.org 2010/07/16 14:07:35 [ssh-rsa.c] more timing paranoia - compare all parts of the expected decrypted data before returning. AFAIK not exploitable in the SSH protocol. "groovy" deraadt@ - djm@cvs.openbsd.org 2010/07/19 03:16:33 [sftp-client.c] bz#1797: fix swapped args in upload_dir_internal(), breaking recursive upload depth checks and causing verbose printing of transfers to always be turned on; patch from imorgan AT nas.nasa.gov - djm@cvs.openbsd.org 2010/07/19 09:15:12 [clientloop.c readconf.c readconf.h ssh.c ssh_config.5] add a "ControlPersist" option that automatically starts a background ssh(1) multiplex master when connecting. This connection can stay alive indefinitely, or can be set to automatically close after a user-specified duration of inactivity. bz#1330 - patch by dwmw2 AT infradead.org, but further hacked on by wmertens AT cisco.com, apb AT cequrux.com, martin-mindrot-bugzilla AT earth.li and myself; "looks ok" markus@ - djm@cvs.openbsd.org 2010/07/21 02:10:58 [misc.c] sync timingsafe_bcmp() with the one dempsky@ committed to sys/lib/libkern - dtucker@cvs.openbsd.org 2010/07/23 08:49:25 [ssh.1] Ciphers is documented in ssh_config(5) these days 20100819 - (dtucker) [contrib/ssh-copy-ud.1] Bug #1786: update ssh-copy-id.1 with more details about its behaviour WRT existing directories. Patch from asguthrie at gmail com, ok djm. 20100716 - (djm) OpenBSD CVS Sync - djm@cvs.openbsd.org 2010/07/02 04:32:44 [misc.c] unbreak strdelim() skipping past quoted strings, e.g. AllowUsers "blah blah" blah was broken; report and fix in bz#1757 from bitman.zhou AT centrify.com ok dtucker; - djm@cvs.openbsd.org 2010/07/12 22:38:52 [ssh.c] Make ExitOnForwardFailure work with fork-after-authentication ("ssh -f") for protocol 2. ok markus@ - djm@cvs.openbsd.org 2010/07/12 22:41:13 [ssh.c ssh_config.5] expand %h to the hostname in ssh_config Hostname options. While this sounds useless, it is actually handy for working with unqualified hostnames: Host *.* Hostname %h Host * Hostname %h.example.org "I like it" markus@ - djm@cvs.openbsd.org 2010/07/13 11:52:06 [auth-rsa.c channels.c jpake.c key.c misc.c misc.h monitor.c] [packet.c ssh-rsa.c] implement a timing_safe_cmp() function to compare memory without leaking timing information by short-circuiting like memcmp() and use it for some of the more sensitive comparisons (though nothing high-value was readily attackable anyway); "looks ok" markus@ - djm@cvs.openbsd.org 2010/07/13 23:13:16 [auth-rsa.c channels.c jpake.c key.c misc.c misc.h monitor.c packet.c] [ssh-rsa.c] s/timing_safe_cmp/timingsafe_bcmp/g - jmc@cvs.openbsd.org 2010/07/14 17:06:58 [ssh.1] finally ssh synopsis looks nice again! this commit just removes a ton of hacks we had in place to make it work with old groff; - schwarze@cvs.openbsd.org 2010/07/15 21:20:38 [ssh-keygen.1] repair incorrect block nesting, which screwed up indentation; problem reported and fix OK by jmc@ 20100714 - (tim) [contrib/redhat/openssh.spec] Bug 1796: Test for skip_x11_askpass (line 77) should have been for no_x11_askpass. 20100702 - (djm) OpenBSD CVS Sync - jmc@cvs.openbsd.org 2010/06/26 00:57:07 [ssh_config.5] tweak previous; - djm@cvs.openbsd.org 2010/06/26 23:04:04 [ssh.c] oops, forgot to #include ; spotted and patch from chl@ - djm@cvs.openbsd.org 2010/06/29 23:15:30 [ssh-keygen.1 ssh-keygen.c] allow import (-i) and export (-e) of PEM and PKCS#8 encoded keys; bz#1749; ok markus@ - djm@cvs.openbsd.org 2010/06/29 23:16:46 [auth2-pubkey.c sshd_config.5] allow key options (command="..." and friends) in AuthorizedPrincipals; ok markus@ - jmc@cvs.openbsd.org 2010/06/30 07:24:25 [ssh-keygen.1] tweak previous; - jmc@cvs.openbsd.org 2010/06/30 07:26:03 [ssh-keygen.c] sort usage(); - jmc@cvs.openbsd.org 2010/06/30 07:28:34 [sshd_config.5] tweak previous; - millert@cvs.openbsd.org 2010/07/01 13:06:59 [scp.c] Fix a longstanding problem where if you suspend scp at the password/passphrase prompt the terminal mode is not restored. OK djm@ - phessler@cvs.openbsd.org 2010/06/27 19:19:56 [regress/Makefile] fix how we run the tests so we can successfully use SUDO='sudo -E' in our env - djm@cvs.openbsd.org 2010/06/29 23:59:54 [cert-userkey.sh] regress tests for key options in AuthorizedPrincipals 20100627 - (tim) [openbsd-compat/port-uw.c] Reorder includes. auth-options.h now needs key.h. 20100626 - (djm) OpenBSD CVS Sync - djm@cvs.openbsd.org 2010/05/21 05:00:36 [misc.c] colon() returns char*, so s/return (0)/return NULL/ - markus@cvs.openbsd.org 2010/06/08 21:32:19 [ssh-pkcs11.c] check length of value returned C_GetAttributValue for != 0 from mdrtbugzilla@codefive.co.uk; bugzilla #1773; ok dtucker@ - djm@cvs.openbsd.org 2010/06/17 07:07:30 [mux.c] Correct sizing of object to be allocated by calloc(), replacing sizeof(state) with sizeof(*state). This worked by accident since the struct contained a single int at present, but could have broken in the future. patch from hyc AT symas.com - djm@cvs.openbsd.org 2010/06/18 00:58:39 [sftp.c] unbreak ls in working directories that contains globbing characters in their pathnames. bz#1655 reported by vgiffin AT apple.com - djm@cvs.openbsd.org 2010/06/18 03:16:03 [session.c] Missing check for chroot_director == "none" (we already checked against NULL); bz#1564 from Jan.Pechanec AT Sun.COM - djm@cvs.openbsd.org 2010/06/18 04:43:08 [sftp-client.c] fix memory leak in do_realpath() error path; bz#1771, patch from anicka AT suse.cz - djm@cvs.openbsd.org 2010/06/22 04:22:59 [servconf.c sshd_config.5] expose some more sshd_config options inside Match blocks: AuthorizedKeysFile AuthorizedPrincipalsFile HostbasedUsesNameFromPacketOnly PermitTunnel bz#1764; feedback from imorgan AT nas.nasa.gov; ok dtucker@ - djm@cvs.openbsd.org 2010/06/22 04:32:06 [ssh-keygen.c] standardise error messages when attempting to open private key files to include "progname: filename: error reason" bz#1783; ok dtucker@ - djm@cvs.openbsd.org 2010/06/22 04:49:47 [auth.c] queue auth debug messages for bad ownership or permissions on the user's keyfiles. These messages will be sent after the user has successfully authenticated (where our client will display them with LogLevel=debug). bz#1554; ok dtucker@ - djm@cvs.openbsd.org 2010/06/22 04:54:30 [ssh-keyscan.c] replace verbose and overflow-prone Linebuf code with read_keyfile_line() based on patch from joachim AT joachimschipper.nl; bz#1565; ok dtucker@ - djm@cvs.openbsd.org 2010/06/22 04:59:12 [session.c] include the user name on "subsystem request for ..." log messages; bz#1571; ok dtucker@ - djm@cvs.openbsd.org 2010/06/23 02:59:02 [ssh-keygen.c] fix printing of extensions in v01 certificates that I broke in r1.190 - djm@cvs.openbsd.org 2010/06/25 07:14:46 [channels.c mux.c readconf.c readconf.h ssh.h] bz#1327: remove hardcoded limit of 100 permitopen clauses and port forwards per direction; ok markus@ stevesk@ - djm@cvs.openbsd.org 2010/06/25 07:20:04 [channels.c session.c] bz#1750: fix requirement for /dev/null inside ChrootDirectory for internal-sftp accidentally introduced in r1.253 by removing the code that opens and dup /dev/null to stderr and modifying the channels code to read stderr but discard it instead; ok markus@ - djm@cvs.openbsd.org 2010/06/25 08:46:17 [auth1.c auth2-none.c] skip the initial check for access with an empty password when PermitEmptyPasswords=no; bz#1638; ok markus@ - djm@cvs.openbsd.org 2010/06/25 23:10:30 [ssh.c] log the hostname and address that we connected to at LogLevel=verbose after authentication is successful to mitigate "phishing" attacks by servers with trusted keys that accept authentication silently and automatically before presenting fake password/passphrase prompts; "nice!" markus@ - djm@cvs.openbsd.org 2010/06/25 23:10:30 [ssh.c] log the hostname and address that we connected to at LogLevel=verbose after authentication is successful to mitigate "phishing" attacks by servers with trusted keys that accept authentication silently and automatically before presenting fake password/passphrase prompts; "nice!" markus@ 20100622 - (djm) [loginrec.c] crank LINFO_NAMESIZE (username length) to 512 bz#1579; ok dtucker 20100618 - (djm) [contrib/ssh-copy-id] Update key file explicitly under ~ rather than assuming that $CWD == $HOME. bz#1500, patch from timothy AT gelter.com 20100617 - (tim) [contrib/cygwin/README] Remove a reference to the obsolete minires-devel package, and to add the reference to the libedit-devel package since CYgwin now provides libedit. Patch from Corinna Vinschen. 20100521 - (djm) OpenBSD CVS Sync - djm@cvs.openbsd.org 2010/05/07 11:31:26 [regress/Makefile regress/cert-userkey.sh] regress tests for AuthorizedPrincipalsFile and "principals=" key option. feedback and ok markus@ - djm@cvs.openbsd.org 2010/05/11 02:58:04 [auth-rsa.c] don't accept certificates marked as "cert-authority" here; ok markus@ - djm@cvs.openbsd.org 2010/05/14 00:47:22 [ssh-add.c] check that the certificate matches the corresponding private key before grafting it on - djm@cvs.openbsd.org 2010/05/14 23:29:23 [channels.c channels.h mux.c ssh.c] Pause the mux channel while waiting for reply from aynch callbacks. Prevents misordering of replies if new requests arrive while waiting. Extend channel open confirm callback to allow signalling failure conditions as well as success. Use this to 1) fix a memory leak, 2) start using the above pause mechanism and 3) delay sending a success/ failure message on mux slave session open until we receive a reply from the server. motivated by and with feedback from markus@ - markus@cvs.openbsd.org 2010/05/16 12:55:51 [PROTOCOL.mux clientloop.h mux.c readconf.c readconf.h ssh.1 ssh.c] mux support for remote forwarding with dynamic port allocation, use with LPORT=`ssh -S muxsocket -R0:localhost:25 -O forward somehost` feedback and ok djm@ - djm@cvs.openbsd.org 2010/05/20 11:25:26 [auth2-pubkey.c] fix logspam when key options (from="..." especially) deny non-matching keys; reported by henning@ also bz#1765; ok markus@ dtucker@ - djm@cvs.openbsd.org 2010/05/20 23:46:02 [PROTOCOL.certkeys auth-options.c ssh-keygen.c] Move the permit-* options to the non-critical "extensions" field for v01 certificates. The logic is that if another implementation fails to implement them then the connection just loses features rather than fails outright. ok markus@ 20100511 - (dtucker) [Makefile.in] Bug #1770: Link libopenbsd-compat twice to solve circular dependency problem on old or odd platforms. From Tom Lane, ok djm@. - (djm) [openbsd-compat/openssl-compat.h] Fix build breakage on older libcrypto by defining OPENSSL_[DR]SA_MAX_MODULUS_BITS if they aren't already. ok dtucker@ 20100510 - OpenBSD CVS Sync - djm@cvs.openbsd.org 2010/04/23 01:47:41 [ssh-keygen.c] bz#1740: display a more helpful error message when $HOME is inaccessible while trying to create .ssh directory. Based on patch from jchadima AT redhat.com; ok dtucker@ - djm@cvs.openbsd.org 2010/04/23 22:27:38 [mux.c] set "detach_close" flag when registering channel cleanup callbacks. This causes the channel to close normally when its fds close and hangs when terminating a mux slave using ~. bz#1758; ok markus@ - djm@cvs.openbsd.org 2010/04/23 22:42:05 [session.c] set stderr to /dev/null for subsystems rather than just closing it. avoids hangs if a subsystem or shell initialisation writes to stderr. bz#1750; ok markus@ - djm@cvs.openbsd.org 2010/04/23 22:48:31 [ssh-keygen.c] refuse to generate keys longer than OPENSSL_[RD]SA_MAX_MODULUS_BITS, since we would refuse to use them anyway. bz#1516; ok dtucker@ - djm@cvs.openbsd.org 2010/04/26 22:28:24 [sshconnect2.c] bz#1502: authctxt.success is declared as an int, but passed by reference to function that accepts sig_atomic_t*. Convert it to the latter; ok markus@ dtucker@ - djm@cvs.openbsd.org 2010/05/01 02:50:50 [PROTOCOL.certkeys] typo; jmeltzer@ - dtucker@cvs.openbsd.org 2010/05/05 04:22:09 [sftp.c] restore mput and mget which got lost in the tab-completion changes. found by Kenneth Whitaker, ok djm@ - djm@cvs.openbsd.org 2010/05/07 11:30:30 [auth-options.c auth-options.h auth.c auth.h auth2-pubkey.c] [key.c servconf.c servconf.h sshd.8 sshd_config.5] add some optional indirection to matching of principal names listed in certificates. Currently, a certificate must include the a user's name to be accepted for authentication. This change adds the ability to specify a list of certificate principal names that are acceptable. When authenticating using a CA trusted through ~/.ssh/authorized_keys, this adds a new principals="name1[,name2,...]" key option. For CAs listed through sshd_config's TrustedCAKeys option, a new config option "AuthorizedPrincipalsFile" specifies a per-user file containing the list of acceptable names. If either option is absent, the current behaviour of requiring the username to appear in principals continues to apply. These options are useful for role accounts, disjoint account namespaces and "user@realm"-style naming policies in certificates. feedback and ok markus@ - jmc@cvs.openbsd.org 2010/05/07 12:49:17 [sshd_config.5] tweak previous; 20100423 - (dtucker) [configure.ac] Bug #1756: Check for the existence of a lib64 dir in the openssl install directory (some newer openssl versions do this on at least some amd64 platforms). 20100418 - OpenBSD CVS Sync - jmc@cvs.openbsd.org 2010/04/16 06:45:01 [ssh_config.5] tweak previous; ok djm - jmc@cvs.openbsd.org 2010/04/16 06:47:04 [ssh-keygen.1 ssh-keygen.c] tweak previous; ok djm - djm@cvs.openbsd.org 2010/04/16 21:14:27 [sshconnect.c] oops, %r => remote username, not %u - djm@cvs.openbsd.org 2010/04/16 01:58:45 [regress/cert-hostkey.sh regress/cert-userkey.sh] regression tests for v01 certificate format includes interop tests for v00 certs - (dtucker) [contrib/aix/buildbff.sh] Fix creation of ssh_prng_cmds.default file. 20100416 - (djm) Release openssh-5.5p1 - OpenBSD CVS Sync - djm@cvs.openbsd.org 2010/03/26 03:13:17 [bufaux.c] allow buffer_get_int_ret/buffer_get_int64_ret to take a NULL pointer argument to allow skipping past values in a buffer - jmc@cvs.openbsd.org 2010/03/26 06:54:36 [ssh.1] tweak previous; - jmc@cvs.openbsd.org 2010/03/27 14:26:55 [ssh_config.5] tweak previous; ok dtucker - djm@cvs.openbsd.org 2010/04/10 00:00:16 [ssh.c] bz#1746 - suppress spurious tty warning when using -O and stdin is not a tty; ok dtucker@ markus@ - djm@cvs.openbsd.org 2010/04/10 00:04:30 [sshconnect.c] fix terminology: we didn't find a certificate in known_hosts, we found a CA key - djm@cvs.openbsd.org 2010/04/10 02:08:44 [clientloop.c] bz#1698: kill channel when pty allocation requests fail. Fixed stuck client if the server refuses pty allocation. ok dtucker@ "think so" markus@ - djm@cvs.openbsd.org 2010/04/10 02:10:56 [sshconnect2.c] show the key type that we are offering in debug(), helps distinguish between certs and plain keys as the path to the private key is usually the same. - djm@cvs.openbsd.org 2010/04/10 05:48:16 [mux.c] fix NULL dereference; from matthew.haub AT alumni.adelaide.edu.au - djm@cvs.openbsd.org 2010/04/14 22:27:42 [ssh_config.5 sshconnect.c] expand %r => remote username in ssh_config:ProxyCommand; ok deraadt markus - markus@cvs.openbsd.org 2010/04/15 20:32:55 [ssh-pkcs11.c] retry lookup for private key if there's no matching key with CKA_SIGN attribute enabled; this fixes fixes MuscleCard support (bugzilla #1736) ok djm@ - djm@cvs.openbsd.org 2010/04/16 01:47:26 [PROTOCOL.certkeys auth-options.c auth-options.h auth-rsa.c] [auth2-pubkey.c authfd.c key.c key.h myproposal.h ssh-add.c] [ssh-agent.c ssh-dss.c ssh-keygen.1 ssh-keygen.c ssh-rsa.c] [sshconnect.c sshconnect2.c sshd.c] revised certificate format ssh-{dss,rsa}-cert-v01@openssh.com with the following changes: move the nonce field to the beginning of the certificate where it can better protect against chosen-prefix attacks on the signature hash Rename "constraints" field to "critical options" Add a new non-critical "extensions" field Add a serial number The older format is still support for authentication and cert generation (use "ssh-keygen -t v00 -s ca_key ..." to generate a v00 certificate) ok markus@ Index: projects/largeSMP/crypto/openssh/defines.h =================================================================== --- projects/largeSMP/crypto/openssh/defines.h (revision 221494) +++ projects/largeSMP/crypto/openssh/defines.h (revision 221495) @@ -1,810 +1,810 @@ /* * Copyright (c) 1999-2003 Damien Miller. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #ifndef _DEFINES_H #define _DEFINES_H -/* $Id: defines.h,v 1.164 2011/01/17 10:15:31 dtucker Exp $ */ +/* $Id: defines.h,v 1.165 2011/05/05 01:19:15 djm Exp $ */ /* Constants */ #if defined(HAVE_DECL_SHUT_RD) && HAVE_DECL_SHUT_RD == 0 enum { SHUT_RD = 0, /* No more receptions. */ SHUT_WR, /* No more transmissions. */ SHUT_RDWR /* No more receptions or transmissions. */ }; # define SHUT_RD SHUT_RD # define SHUT_WR SHUT_WR # define SHUT_RDWR SHUT_RDWR #endif /* * Definitions for IP type of service (ip_tos) */ +#include #include #ifndef IPTOS_LOWDELAY # define IPTOS_LOWDELAY 0x10 # define IPTOS_THROUGHPUT 0x08 # define IPTOS_RELIABILITY 0x04 # define IPTOS_LOWCOST 0x02 # define IPTOS_MINCOST IPTOS_LOWCOST #endif /* IPTOS_LOWDELAY */ /* * Definitions for DiffServ Codepoints as per RFC2474 */ -#include #ifndef IPTOS_DSCP_AF11 # define IPTOS_DSCP_AF11 0x28 # define IPTOS_DSCP_AF12 0x30 # define IPTOS_DSCP_AF13 0x38 # define IPTOS_DSCP_AF21 0x48 # define IPTOS_DSCP_AF22 0x50 # define IPTOS_DSCP_AF23 0x58 # define IPTOS_DSCP_AF31 0x68 # define IPTOS_DSCP_AF32 0x70 # define IPTOS_DSCP_AF33 0x78 # define IPTOS_DSCP_AF41 0x88 # define IPTOS_DSCP_AF42 0x90 # define IPTOS_DSCP_AF43 0x98 # define IPTOS_DSCP_EF 0xb8 #endif /* IPTOS_DSCP_AF11 */ #ifndef IPTOS_DSCP_CS0 # define IPTOS_DSCP_CS0 0x00 # define IPTOS_DSCP_CS1 0x20 # define IPTOS_DSCP_CS2 0x40 # define IPTOS_DSCP_CS3 0x60 # define IPTOS_DSCP_CS4 0x80 # define IPTOS_DSCP_CS5 0xa0 # define IPTOS_DSCP_CS6 0xc0 # define IPTOS_DSCP_CS7 0xe0 #endif /* IPTOS_DSCP_CS0 */ #ifndef IPTOS_DSCP_EF # define IPTOS_DSCP_EF 0xb8 #endif /* IPTOS_DSCP_EF */ #ifndef MAXPATHLEN # ifdef PATH_MAX # define MAXPATHLEN PATH_MAX # else /* PATH_MAX */ # define MAXPATHLEN 64 /* realpath uses a fixed buffer of size MAXPATHLEN, so force use of ours */ # ifndef BROKEN_REALPATH # define BROKEN_REALPATH 1 # endif /* BROKEN_REALPATH */ # endif /* PATH_MAX */ #endif /* MAXPATHLEN */ #ifndef PATH_MAX # ifdef _POSIX_PATH_MAX # define PATH_MAX _POSIX_PATH_MAX # endif #endif #if defined(HAVE_DECL_MAXSYMLINKS) && HAVE_DECL_MAXSYMLINKS == 0 # define MAXSYMLINKS 5 #endif #ifndef STDIN_FILENO # define STDIN_FILENO 0 #endif #ifndef STDOUT_FILENO # define STDOUT_FILENO 1 #endif #ifndef STDERR_FILENO # define STDERR_FILENO 2 #endif #ifndef NGROUPS_MAX /* Disable groupaccess if NGROUP_MAX is not set */ #ifdef NGROUPS #define NGROUPS_MAX NGROUPS #else #define NGROUPS_MAX 0 #endif #endif #if defined(HAVE_DECL_O_NONBLOCK) && HAVE_DECL_O_NONBLOCK == 0 # define O_NONBLOCK 00004 /* Non Blocking Open */ #endif #ifndef S_ISDIR # define S_ISDIR(mode) (((mode) & (_S_IFMT)) == (_S_IFDIR)) #endif /* S_ISDIR */ #ifndef S_ISREG # define S_ISREG(mode) (((mode) & (_S_IFMT)) == (_S_IFREG)) #endif /* S_ISREG */ #ifndef S_ISLNK # define S_ISLNK(mode) (((mode) & S_IFMT) == S_IFLNK) #endif /* S_ISLNK */ #ifndef S_IXUSR # define S_IXUSR 0000100 /* execute/search permission, */ # define S_IXGRP 0000010 /* execute/search permission, */ # define S_IXOTH 0000001 /* execute/search permission, */ # define _S_IWUSR 0000200 /* write permission, */ # define S_IWUSR _S_IWUSR /* write permission, owner */ # define S_IWGRP 0000020 /* write permission, group */ # define S_IWOTH 0000002 /* write permission, other */ # define S_IRUSR 0000400 /* read permission, owner */ # define S_IRGRP 0000040 /* read permission, group */ # define S_IROTH 0000004 /* read permission, other */ # define S_IRWXU 0000700 /* read, write, execute */ # define S_IRWXG 0000070 /* read, write, execute */ # define S_IRWXO 0000007 /* read, write, execute */ #endif /* S_IXUSR */ #if !defined(MAP_ANON) && defined(MAP_ANONYMOUS) #define MAP_ANON MAP_ANONYMOUS #endif #ifndef MAP_FAILED # define MAP_FAILED ((void *)-1) #endif /* *-*-nto-qnx doesn't define this constant in the system headers */ #ifdef MISSING_NFDBITS # define NFDBITS (8 * sizeof(unsigned long)) #endif /* SCO Open Server 3 has INADDR_LOOPBACK defined in rpc/rpc.h but including rpc/rpc.h breaks Solaris 6 */ #ifndef INADDR_LOOPBACK #define INADDR_LOOPBACK ((u_long)0x7f000001) #endif /* Types */ /* If sys/types.h does not supply intXX_t, supply them ourselves */ /* (or die trying) */ #ifndef HAVE_U_INT typedef unsigned int u_int; #endif #ifndef HAVE_INTXX_T # if (SIZEOF_CHAR == 1) typedef char int8_t; # else # error "8 bit int type not found." # endif # if (SIZEOF_SHORT_INT == 2) typedef short int int16_t; # else # ifdef _UNICOS # if (SIZEOF_SHORT_INT == 4) typedef short int16_t; # else typedef long int16_t; # endif # else # error "16 bit int type not found." # endif /* _UNICOS */ # endif # if (SIZEOF_INT == 4) typedef int int32_t; # else # ifdef _UNICOS typedef long int32_t; # else # error "32 bit int type not found." # endif /* _UNICOS */ # endif #endif /* If sys/types.h does not supply u_intXX_t, supply them ourselves */ #ifndef HAVE_U_INTXX_T # ifdef HAVE_UINTXX_T typedef uint8_t u_int8_t; typedef uint16_t u_int16_t; typedef uint32_t u_int32_t; # define HAVE_U_INTXX_T 1 # else # if (SIZEOF_CHAR == 1) typedef unsigned char u_int8_t; # else # error "8 bit int type not found." # endif # if (SIZEOF_SHORT_INT == 2) typedef unsigned short int u_int16_t; # else # ifdef _UNICOS # if (SIZEOF_SHORT_INT == 4) typedef unsigned short u_int16_t; # else typedef unsigned long u_int16_t; # endif # else # error "16 bit int type not found." # endif # endif # if (SIZEOF_INT == 4) typedef unsigned int u_int32_t; # else # ifdef _UNICOS typedef unsigned long u_int32_t; # else # error "32 bit int type not found." # endif # endif # endif #define __BIT_TYPES_DEFINED__ #endif /* 64-bit types */ #ifndef HAVE_INT64_T # if (SIZEOF_LONG_INT == 8) typedef long int int64_t; # else # if (SIZEOF_LONG_LONG_INT == 8) typedef long long int int64_t; # endif # endif #endif #ifndef HAVE_U_INT64_T # if (SIZEOF_LONG_INT == 8) typedef unsigned long int u_int64_t; # else # if (SIZEOF_LONG_LONG_INT == 8) typedef unsigned long long int u_int64_t; # endif # endif #endif #ifndef HAVE_U_CHAR typedef unsigned char u_char; # define HAVE_U_CHAR #endif /* HAVE_U_CHAR */ #ifndef SIZE_T_MAX #define SIZE_T_MAX ULONG_MAX #endif /* SIZE_T_MAX */ #ifndef HAVE_SIZE_T typedef unsigned int size_t; # define HAVE_SIZE_T # define SIZE_T_MAX UINT_MAX #endif /* HAVE_SIZE_T */ #ifndef SIZE_MAX #define SIZE_MAX SIZE_T_MAX #endif #ifndef HAVE_SSIZE_T typedef int ssize_t; # define HAVE_SSIZE_T #endif /* HAVE_SSIZE_T */ #ifndef HAVE_CLOCK_T typedef long clock_t; # define HAVE_CLOCK_T #endif /* HAVE_CLOCK_T */ #ifndef HAVE_SA_FAMILY_T typedef int sa_family_t; # define HAVE_SA_FAMILY_T #endif /* HAVE_SA_FAMILY_T */ #ifndef HAVE_PID_T typedef int pid_t; # define HAVE_PID_T #endif /* HAVE_PID_T */ #ifndef HAVE_SIG_ATOMIC_T typedef int sig_atomic_t; # define HAVE_SIG_ATOMIC_T #endif /* HAVE_SIG_ATOMIC_T */ #ifndef HAVE_MODE_T typedef int mode_t; # define HAVE_MODE_T #endif /* HAVE_MODE_T */ #if !defined(HAVE_SS_FAMILY_IN_SS) && defined(HAVE___SS_FAMILY_IN_SS) # define ss_family __ss_family #endif /* !defined(HAVE_SS_FAMILY_IN_SS) && defined(HAVE_SA_FAMILY_IN_SS) */ #ifndef HAVE_SYS_UN_H struct sockaddr_un { short sun_family; /* AF_UNIX */ char sun_path[108]; /* path name (gag) */ }; #endif /* HAVE_SYS_UN_H */ #ifndef HAVE_IN_ADDR_T typedef u_int32_t in_addr_t; #endif #ifndef HAVE_IN_PORT_T typedef u_int16_t in_port_t; #endif #if defined(BROKEN_SYS_TERMIO_H) && !defined(_STRUCT_WINSIZE) #define _STRUCT_WINSIZE struct winsize { unsigned short ws_row; /* rows, in characters */ unsigned short ws_col; /* columns, in character */ unsigned short ws_xpixel; /* horizontal size, pixels */ unsigned short ws_ypixel; /* vertical size, pixels */ }; #endif /* *-*-nto-qnx does not define this type in the system headers */ #ifdef MISSING_FD_MASK typedef unsigned long int fd_mask; #endif /* Paths */ #ifndef _PATH_BSHELL # define _PATH_BSHELL "/bin/sh" #endif #ifdef USER_PATH # ifdef _PATH_STDPATH # undef _PATH_STDPATH # endif # define _PATH_STDPATH USER_PATH #endif #ifndef _PATH_STDPATH # define _PATH_STDPATH "/usr/bin:/bin:/usr/sbin:/sbin" #endif #ifndef SUPERUSER_PATH # define SUPERUSER_PATH _PATH_STDPATH #endif #ifndef _PATH_DEVNULL # define _PATH_DEVNULL "/dev/null" #endif #ifndef MAIL_DIRECTORY # define MAIL_DIRECTORY "/var/spool/mail" #endif #ifndef MAILDIR # define MAILDIR MAIL_DIRECTORY #endif #if !defined(_PATH_MAILDIR) && defined(MAILDIR) # define _PATH_MAILDIR MAILDIR #endif /* !defined(_PATH_MAILDIR) && defined(MAILDIR) */ #ifndef _PATH_NOLOGIN # define _PATH_NOLOGIN "/etc/nologin" #endif /* Define this to be the path of the xauth program. */ #ifdef XAUTH_PATH #define _PATH_XAUTH XAUTH_PATH #endif /* XAUTH_PATH */ /* derived from XF4/xc/lib/dps/Xlibnet.h */ #ifndef X_UNIX_PATH # ifdef __hpux # define X_UNIX_PATH "/var/spool/sockets/X11/%u" # else # define X_UNIX_PATH "/tmp/.X11-unix/X%u" # endif #endif /* X_UNIX_PATH */ #define _PATH_UNIX_X X_UNIX_PATH #ifndef _PATH_TTY # define _PATH_TTY "/dev/tty" #endif /* Macros */ #if defined(HAVE_LOGIN_GETCAPBOOL) && defined(HAVE_LOGIN_CAP_H) # define HAVE_LOGIN_CAP #endif #ifndef MAX # define MAX(a,b) (((a)>(b))?(a):(b)) # define MIN(a,b) (((a)<(b))?(a):(b)) #endif #ifndef roundup # define roundup(x, y) ((((x)+((y)-1))/(y))*(y)) #endif #ifndef timersub #define timersub(a, b, result) \ do { \ (result)->tv_sec = (a)->tv_sec - (b)->tv_sec; \ (result)->tv_usec = (a)->tv_usec - (b)->tv_usec; \ if ((result)->tv_usec < 0) { \ --(result)->tv_sec; \ (result)->tv_usec += 1000000; \ } \ } while (0) #endif #ifndef TIMEVAL_TO_TIMESPEC #define TIMEVAL_TO_TIMESPEC(tv, ts) { \ (ts)->tv_sec = (tv)->tv_sec; \ (ts)->tv_nsec = (tv)->tv_usec * 1000; \ } #endif #ifndef TIMESPEC_TO_TIMEVAL #define TIMESPEC_TO_TIMEVAL(tv, ts) { \ (tv)->tv_sec = (ts)->tv_sec; \ (tv)->tv_usec = (ts)->tv_nsec / 1000; \ } #endif #ifndef __P # define __P(x) x #endif #if !defined(IN6_IS_ADDR_V4MAPPED) # define IN6_IS_ADDR_V4MAPPED(a) \ ((((u_int32_t *) (a))[0] == 0) && (((u_int32_t *) (a))[1] == 0) && \ (((u_int32_t *) (a))[2] == htonl (0xffff))) #endif /* !defined(IN6_IS_ADDR_V4MAPPED) */ #if !defined(__GNUC__) || (__GNUC__ < 2) # define __attribute__(x) #endif /* !defined(__GNUC__) || (__GNUC__ < 2) */ #if !defined(HAVE_ATTRIBUTE__SENTINEL__) && !defined(__sentinel__) # define __sentinel__ #endif #if !defined(HAVE_ATTRIBUTE__BOUNDED__) && !defined(__bounded__) # define __bounded__(x, y, z) #endif #if !defined(HAVE_ATTRIBUTE__NONNULL__) && !defined(__nonnull__) # define __nonnull__(x) #endif /* *-*-nto-qnx doesn't define this macro in the system headers */ #ifdef MISSING_HOWMANY # define howmany(x,y) (((x)+((y)-1))/(y)) #endif #ifndef OSSH_ALIGNBYTES #define OSSH_ALIGNBYTES (sizeof(int) - 1) #endif #ifndef __CMSG_ALIGN #define __CMSG_ALIGN(p) (((u_int)(p) + OSSH_ALIGNBYTES) &~ OSSH_ALIGNBYTES) #endif /* Length of the contents of a control message of length len */ #ifndef CMSG_LEN #define CMSG_LEN(len) (__CMSG_ALIGN(sizeof(struct cmsghdr)) + (len)) #endif /* Length of the space taken up by a padded control message of length len */ #ifndef CMSG_SPACE #define CMSG_SPACE(len) (__CMSG_ALIGN(sizeof(struct cmsghdr)) + __CMSG_ALIGN(len)) #endif /* given pointer to struct cmsghdr, return pointer to data */ #ifndef CMSG_DATA #define CMSG_DATA(cmsg) ((u_char *)(cmsg) + __CMSG_ALIGN(sizeof(struct cmsghdr))) #endif /* CMSG_DATA */ /* * RFC 2292 requires to check msg_controllen, in case that the kernel returns * an empty list for some reasons. */ #ifndef CMSG_FIRSTHDR #define CMSG_FIRSTHDR(mhdr) \ ((mhdr)->msg_controllen >= sizeof(struct cmsghdr) ? \ (struct cmsghdr *)(mhdr)->msg_control : \ (struct cmsghdr *)NULL) #endif /* CMSG_FIRSTHDR */ #if defined(HAVE_DECL_OFFSETOF) && HAVE_DECL_OFFSETOF == 0 # define offsetof(type, member) ((size_t) &((type *)0)->member) #endif /* Set up BSD-style BYTE_ORDER definition if it isn't there already */ /* XXX: doesn't try to cope with strange byte orders (PDP_ENDIAN) */ #ifndef BYTE_ORDER # ifndef LITTLE_ENDIAN # define LITTLE_ENDIAN 1234 # endif /* LITTLE_ENDIAN */ # ifndef BIG_ENDIAN # define BIG_ENDIAN 4321 # endif /* BIG_ENDIAN */ # ifdef WORDS_BIGENDIAN # define BYTE_ORDER BIG_ENDIAN # else /* WORDS_BIGENDIAN */ # define BYTE_ORDER LITTLE_ENDIAN # endif /* WORDS_BIGENDIAN */ #endif /* BYTE_ORDER */ /* Function replacement / compatibility hacks */ #if !defined(HAVE_GETADDRINFO) && (defined(HAVE_OGETADDRINFO) || defined(HAVE_NGETADDRINFO)) # define HAVE_GETADDRINFO #endif #ifndef HAVE_GETOPT_OPTRESET # undef getopt # undef opterr # undef optind # undef optopt # undef optreset # undef optarg # define getopt(ac, av, o) BSDgetopt(ac, av, o) # define opterr BSDopterr # define optind BSDoptind # define optopt BSDoptopt # define optreset BSDoptreset # define optarg BSDoptarg #endif #if defined(BROKEN_GETADDRINFO) && defined(HAVE_GETADDRINFO) # undef HAVE_GETADDRINFO #endif #if defined(BROKEN_GETADDRINFO) && defined(HAVE_FREEADDRINFO) # undef HAVE_FREEADDRINFO #endif #if defined(BROKEN_GETADDRINFO) && defined(HAVE_GAI_STRERROR) # undef HAVE_GAI_STRERROR #endif #if defined(BROKEN_UPDWTMPX) && defined(HAVE_UPDWTMPX) # undef HAVE_UPDWTMPX #endif #if defined(BROKEN_SHADOW_EXPIRE) && defined(HAS_SHADOW_EXPIRE) # undef HAS_SHADOW_EXPIRE #endif #if defined(HAVE_OPENLOG_R) && defined(SYSLOG_DATA_INIT) && \ defined(SYSLOG_R_SAFE_IN_SIGHAND) # define DO_LOG_SAFE_IN_SIGHAND #endif #if !defined(HAVE_MEMMOVE) && defined(HAVE_BCOPY) # define memmove(s1, s2, n) bcopy((s2), (s1), (n)) #endif /* !defined(HAVE_MEMMOVE) && defined(HAVE_BCOPY) */ #if defined(HAVE_VHANGUP) && !defined(HAVE_DEV_PTMX) # define USE_VHANGUP #endif /* defined(HAVE_VHANGUP) && !defined(HAVE_DEV_PTMX) */ #ifndef GETPGRP_VOID # include # define getpgrp() getpgrp(0) #endif #ifdef USE_BSM_AUDIT # define SSH_AUDIT_EVENTS # define CUSTOM_SSH_AUDIT_EVENTS #endif #ifdef USE_LINUX_AUDIT # define SSH_AUDIT_EVENTS # define CUSTOM_SSH_AUDIT_EVENTS #endif #if !defined(HAVE___func__) && defined(HAVE___FUNCTION__) # define __func__ __FUNCTION__ #elif !defined(HAVE___func__) # define __func__ "" #endif #if defined(KRB5) && !defined(HEIMDAL) # define krb5_get_err_text(context,code) error_message(code) #endif #if defined(SKEYCHALLENGE_4ARG) # define _compat_skeychallenge(a,b,c,d) skeychallenge(a,b,c,d) #else # define _compat_skeychallenge(a,b,c,d) skeychallenge(a,b,c) #endif /* Maximum number of file descriptors available */ #ifdef HAVE_SYSCONF # define SSH_SYSFDMAX sysconf(_SC_OPEN_MAX) #else # define SSH_SYSFDMAX 10000 #endif #ifdef FSID_HAS_VAL /* encode f_fsid into a 64 bit value */ #define FSID_TO_ULONG(f) \ ((((u_int64_t)(f).val[0] & 0xffffffffUL) << 32) | \ ((f).val[1] & 0xffffffffUL)) #elif defined(FSID_HAS___VAL) #define FSID_TO_ULONG(f) \ ((((u_int64_t)(f).__val[0] & 0xffffffffUL) << 32) | \ ((f).__val[1] & 0xffffffffUL)) #else # define FSID_TO_ULONG(f) ((f)) #endif #if defined(__Lynx__) /* * LynxOS defines these in param.h which we do not want to include since * it will also pull in a bunch of kernel definitions. */ # define ALIGNBYTES (sizeof(int) - 1) # define ALIGN(p) (((unsigned)p + ALIGNBYTES) & ~ALIGNBYTES) /* Missing prototypes on LynxOS */ int snprintf (char *, size_t, const char *, ...); int mkstemp (char *); char *crypt (const char *, const char *); int seteuid (uid_t); int setegid (gid_t); char *mkdtemp (char *); int rresvport_af (int *, sa_family_t); int innetgr (const char *, const char *, const char *, const char *); #endif /* * Define this to use pipes instead of socketpairs for communicating with the * client program. Socketpairs do not seem to work on all systems. * * configure.ac sets this for a few OS's which are known to have problems * but you may need to set it yourself */ /* #define USE_PIPES 1 */ /** ** login recorder definitions **/ /* FIXME: put default paths back in */ #ifndef UTMP_FILE # ifdef _PATH_UTMP # define UTMP_FILE _PATH_UTMP # else # ifdef CONF_UTMP_FILE # define UTMP_FILE CONF_UTMP_FILE # endif # endif #endif #ifndef WTMP_FILE # ifdef _PATH_WTMP # define WTMP_FILE _PATH_WTMP # else # ifdef CONF_WTMP_FILE # define WTMP_FILE CONF_WTMP_FILE # endif # endif #endif /* pick up the user's location for lastlog if given */ #ifndef LASTLOG_FILE # ifdef _PATH_LASTLOG # define LASTLOG_FILE _PATH_LASTLOG # else # ifdef CONF_LASTLOG_FILE # define LASTLOG_FILE CONF_LASTLOG_FILE # endif # endif #endif #if defined(HAVE_SHADOW_H) && !defined(DISABLE_SHADOW) # define USE_SHADOW #endif /* The login() library function in libutil is first choice */ #if defined(HAVE_LOGIN) && !defined(DISABLE_LOGIN) # define USE_LOGIN #else /* Simply select your favourite login types. */ /* Can't do if-else because some systems use several... */ # if !defined(DISABLE_UTMPX) # define USE_UTMPX # endif # if defined(UTMP_FILE) && !defined(DISABLE_UTMP) # define USE_UTMP # endif # if defined(WTMPX_FILE) && !defined(DISABLE_WTMPX) # define USE_WTMPX # endif # if defined(WTMP_FILE) && !defined(DISABLE_WTMP) # define USE_WTMP # endif #endif #ifndef UT_LINESIZE # define UT_LINESIZE 8 #endif /* I hope that the presence of LASTLOG_FILE is enough to detect this */ #if defined(LASTLOG_FILE) && !defined(DISABLE_LASTLOG) # define USE_LASTLOG #endif #ifdef HAVE_OSF_SIA # ifdef USE_SHADOW # undef USE_SHADOW # endif # define CUSTOM_SYS_AUTH_PASSWD 1 #endif #if defined(HAVE_LIBIAF) && defined(HAVE_SET_ID) && !defined(HAVE_SECUREWARE) # define CUSTOM_SYS_AUTH_PASSWD 1 #endif #if defined(HAVE_LIBIAF) && defined(HAVE_SET_ID) && !defined(BROKEN_LIBIAF) # define USE_LIBIAF #endif /* HP-UX 11.11 */ #ifdef BTMP_FILE # define _PATH_BTMP BTMP_FILE #endif #if defined(USE_BTMP) && defined(_PATH_BTMP) # define CUSTOM_FAILED_LOGIN #endif /** end of login recorder definitions */ #ifdef BROKEN_GETGROUPS # define getgroups(a,b) ((a)==0 && (b)==NULL ? NGROUPS_MAX : getgroups((a),(b))) #endif #if defined(HAVE_MMAP) && defined(BROKEN_MMAP) # undef HAVE_MMAP #endif #ifndef IOV_MAX # if defined(_XOPEN_IOV_MAX) # define IOV_MAX _XOPEN_IOV_MAX # elif defined(DEF_IOV_MAX) # define IOV_MAX DEF_IOV_MAX # else # define IOV_MAX 16 # endif #endif #ifndef EWOULDBLOCK # define EWOULDBLOCK EAGAIN #endif #ifndef INET6_ADDRSTRLEN /* for non IPv6 machines */ #define INET6_ADDRSTRLEN 46 #endif #ifndef SSH_IOBUFSZ # define SSH_IOBUFSZ 8192 #endif #ifndef _NSIG # ifdef NSIG # define _NSIG NSIG # else # define _NSIG 128 # endif #endif #endif /* _DEFINES_H */ Index: projects/largeSMP/crypto/openssh =================================================================== --- projects/largeSMP/crypto/openssh (revision 221494) +++ projects/largeSMP/crypto/openssh (revision 221495) Property changes on: projects/largeSMP/crypto/openssh ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,2 ## Merged /head/crypto/openssh:r221444-221493 Merged /vendor-crypto/openssh/dist:r221395-221486 Index: projects/largeSMP/crypto/openssl =================================================================== --- projects/largeSMP/crypto/openssl (revision 221494) +++ projects/largeSMP/crypto/openssl (revision 221495) Property changes on: projects/largeSMP/crypto/openssl ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/crypto/openssl:r221444-221493 Index: projects/largeSMP/etc/periodic/daily/220.backup-pkgdb =================================================================== --- projects/largeSMP/etc/periodic/daily/220.backup-pkgdb (revision 221494) +++ projects/largeSMP/etc/periodic/daily/220.backup-pkgdb (revision 221495) @@ -1,50 +1,51 @@ #!/bin/sh # # $FreeBSD$ # # If there is a global system configuration file, suck it in. # if [ -r /etc/defaults/periodic.conf ] then . /etc/defaults/periodic.conf source_periodic_confs fi rc=0 case "$daily_backup_pkgdb_enable" in [Yy][Ee][Ss]) bak="${daily_backup_pkgdb_dir:-/var/backups}" bak_file="${bak}/pkgdb.bak.tbz" - pkg_dbdir=`make -f/usr/share/mk/bsd.port.mk -V PKG_DBDIR 2>/dev/null` + pkg_dbdir=`make -f/usr/share/mk/bsd.port.mk -V PKG_DBDIR 2>/dev/null` || + pkg_dbdir=/var/db/pkg if [ ! -d "$bak" ] then install -d -o root -g wheel -m 750 $bak || { echo '$daily_backup_pkgdb_enable is enabled but' \ "$daily_backup_pkgdb_dbdir doesn't exist" ; exit 2 ; } fi echo '' echo 'Backing up package db directory:' new_bak_file=`mktemp ${bak_file}-XXXXX` - if tar -cjf "${new_bak_file}" "$pkg_dbdir"; then + if tar -cjHf "${new_bak_file}" "$pkg_dbdir" 2>/dev/null; then chmod 644 "${new_bak_file}" if [ -e "${bak_file}.2" -a -e "${bak_file}" ]; then unlink "${bak_file}.2" mv "${bak_file}" "${bak_file}.2" fi [ -e "${bak_file}" ] && mv "${bak_file}" "${bak_file}.2" mv "${new_bak_file}" "${bak_file}" else rc=3 fi ;; esac exit $rc Index: projects/largeSMP/gnu/lib =================================================================== --- projects/largeSMP/gnu/lib (revision 221494) +++ projects/largeSMP/gnu/lib (revision 221495) Property changes on: projects/largeSMP/gnu/lib ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/gnu/lib:r221444-221493 Index: projects/largeSMP/gnu/usr.bin/binutils =================================================================== --- projects/largeSMP/gnu/usr.bin/binutils (revision 221494) +++ projects/largeSMP/gnu/usr.bin/binutils (revision 221495) Property changes on: projects/largeSMP/gnu/usr.bin/binutils ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/gnu/usr.bin/binutils:r221444-221493 Index: projects/largeSMP/gnu/usr.bin/cc/cc_tools =================================================================== --- projects/largeSMP/gnu/usr.bin/cc/cc_tools (revision 221494) +++ projects/largeSMP/gnu/usr.bin/cc/cc_tools (revision 221495) Property changes on: projects/largeSMP/gnu/usr.bin/cc/cc_tools ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/gnu/usr.bin/cc/cc_tools:r221444-221493 Index: projects/largeSMP/gnu/usr.bin/gdb =================================================================== --- projects/largeSMP/gnu/usr.bin/gdb (revision 221494) +++ projects/largeSMP/gnu/usr.bin/gdb (revision 221495) Property changes on: projects/largeSMP/gnu/usr.bin/gdb ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/gnu/usr.bin/gdb:r221444-221493 Index: projects/largeSMP/lib/libarchive/Makefile =================================================================== --- projects/largeSMP/lib/libarchive/Makefile (revision 221494) +++ projects/largeSMP/lib/libarchive/Makefile (revision 221495) @@ -1,272 +1,275 @@ # $FreeBSD$ .include LIB= archive -DPADD= ${LIBZ} ${LIBMD} -LDADD= -lz -lmd +DPADD= ${LIBZ} +LDADD= -lz DPADD+= ${LIBBZ2} LDADD+= -lbz2 CFLAGS+= -DHAVE_BZLIB_H=1 DPADD+= ${LIBLZMA} LDADD+= -llzma CFLAGS+= -DHAVE_LIBLZMA=1 -DHAVE_LZMA_H=1 # FreeBSD SHLIB_MAJOR value is managed as part of the FreeBSD system. # It has no real relation to the libarchive version number. SHLIB_MAJOR= 5 CFLAGS+= -DPLATFORM_CONFIG_H=\"config_freebsd.h\" CFLAGS+= -I${.OBJDIR} .if ${MK_OPENSSL} != "no" CFLAGS+= -DWITH_OPENSSL DPADD+= ${LIBCRYPTO} LDADD+= -lcrypto +.else +DPADD+= ${LIBMD} +LDADD+= -lmd .endif # Headers to be installed in /usr/include INCS= archive.h archive_entry.h # Sources to be compiled. SRCS= archive_check_magic.c \ archive_entry.c \ archive_entry_copy_stat.c \ archive_entry_stat.c \ archive_entry_strmode.c \ archive_entry_link_resolver.c \ archive_entry_xattr.c \ archive_read.c \ archive_read_data_into_fd.c \ archive_read_disk.c \ archive_read_disk_entry_from_file.c \ archive_read_disk_set_standard_lookup.c \ archive_read_extract.c \ archive_read_open_fd.c \ archive_read_open_file.c \ archive_read_open_filename.c \ archive_read_open_memory.c \ archive_read_support_compression_all.c \ archive_read_support_compression_bzip2.c \ archive_read_support_compression_compress.c \ archive_read_support_compression_gzip.c \ archive_read_support_compression_none.c \ archive_read_support_compression_program.c \ archive_read_support_compression_uu.c \ archive_read_support_compression_xz.c \ archive_read_support_format_all.c \ archive_read_support_format_ar.c \ archive_read_support_format_cpio.c \ archive_read_support_format_empty.c \ archive_read_support_format_iso9660.c \ archive_read_support_format_mtree.c \ archive_read_support_format_raw.c \ archive_read_support_format_tar.c \ archive_read_support_format_zip.c \ archive_string.c \ archive_string_sprintf.c \ archive_util.c \ archive_virtual.c \ archive_write.c \ archive_write_disk.c \ archive_write_disk_set_standard_lookup.c \ archive_write_open_fd.c \ archive_write_open_file.c \ archive_write_open_filename.c \ archive_write_open_memory.c \ archive_write_set_compression_bzip2.c \ archive_write_set_compression_compress.c \ archive_write_set_compression_gzip.c \ archive_write_set_compression_none.c \ archive_write_set_compression_program.c \ archive_write_set_compression_xz.c \ archive_write_set_format.c \ archive_write_set_format_ar.c \ archive_write_set_format_by_name.c \ archive_write_set_format_cpio.c \ archive_write_set_format_cpio_newc.c \ archive_write_set_format_mtree.c \ archive_write_set_format_pax.c \ archive_write_set_format_shar.c \ archive_write_set_format_ustar.c \ archive_write_set_format_zip.c \ filter_fork.c # Man pages to be installed. MAN= archive_entry.3 \ archive_read.3 \ archive_read_disk.3 \ archive_util.3 \ archive_write.3 \ archive_write_disk.3 \ cpio.5 \ libarchive.3 \ libarchive-formats.5 \ tar.5 # Symlink the man pages under each function name. MLINKS+= archive_entry.3 archive_entry_acl_add_entry.3 MLINKS+= archive_entry.3 archive_entry_acl_add_entry_w.3 MLINKS+= archive_entry.3 archive_entry_acl_clear.3 MLINKS+= archive_entry.3 archive_entry_acl_count.3 MLINKS+= archive_entry.3 archive_entry_acl_next.3 MLINKS+= archive_entry.3 archive_entry_acl_next_w.3 MLINKS+= archive_entry.3 archive_entry_acl_reset.3 MLINKS+= archive_entry.3 archive_entry_acl_text_w.3 MLINKS+= archive_entry.3 archive_entry_clear.3 MLINKS+= archive_entry.3 archive_entry_clone.3 MLINKS+= archive_entry.3 archive_entry_copy_fflags_text.3 MLINKS+= archive_entry.3 archive_entry_copy_fflags_text_w.3 MLINKS+= archive_entry.3 archive_entry_copy_gname.3 MLINKS+= archive_entry.3 archive_entry_copy_gname_w.3 MLINKS+= archive_entry.3 archive_entry_copy_hardlink_w.3 MLINKS+= archive_entry.3 archive_entry_copy_link.3 MLINKS+= archive_entry.3 archive_entry_copy_link_w.3 MLINKS+= archive_entry.3 archive_entry_copy_pathname_w.3 MLINKS+= archive_entry.3 archive_entry_copy_stat.3 MLINKS+= archive_entry.3 archive_entry_copy_symlink_w.3 MLINKS+= archive_entry.3 archive_entry_copy_uname.3 MLINKS+= archive_entry.3 archive_entry_copy_uname_w.3 MLINKS+= archive_entry.3 archive_entry_dev.3 MLINKS+= archive_entry.3 archive_entry_devmajor.3 MLINKS+= archive_entry.3 archive_entry_devminor.3 MLINKS+= archive_entry.3 archive_entry_filetype.3 MLINKS+= archive_entry.3 archive_entry_fflags.3 MLINKS+= archive_entry.3 archive_entry_fflags_text.3 MLINKS+= archive_entry.3 archive_entry_free.3 MLINKS+= archive_entry.3 archive_entry_gid.3 MLINKS+= archive_entry.3 archive_entry_gname.3 MLINKS+= archive_entry.3 archive_entry_gname_w.3 MLINKS+= archive_entry.3 archive_entry_hardlink.3 MLINKS+= archive_entry.3 archive_entry_ino.3 MLINKS+= archive_entry.3 archive_entry_mode.3 MLINKS+= archive_entry.3 archive_entry_mtime.3 MLINKS+= archive_entry.3 archive_entry_mtime_nsec.3 MLINKS+= archive_entry.3 archive_entry_nlink.3 MLINKS+= archive_entry.3 archive_entry_new.3 MLINKS+= archive_entry.3 archive_entry_pathname.3 MLINKS+= archive_entry.3 archive_entry_pathname_w.3 MLINKS+= archive_entry.3 archive_entry_rdev.3 MLINKS+= archive_entry.3 archive_entry_rdevmajor.3 MLINKS+= archive_entry.3 archive_entry_rdevminor.3 MLINKS+= archive_entry.3 archive_entry_set_atime.3 MLINKS+= archive_entry.3 archive_entry_set_ctime.3 MLINKS+= archive_entry.3 archive_entry_set_dev.3 MLINKS+= archive_entry.3 archive_entry_set_devmajor.3 MLINKS+= archive_entry.3 archive_entry_set_devminor.3 MLINKS+= archive_entry.3 archive_entry_set_fflags.3 MLINKS+= archive_entry.3 archive_entry_set_gid.3 MLINKS+= archive_entry.3 archive_entry_set_gname.3 MLINKS+= archive_entry.3 archive_entry_set_hardlink.3 MLINKS+= archive_entry.3 archive_entry_set_link.3 MLINKS+= archive_entry.3 archive_entry_set_mode.3 MLINKS+= archive_entry.3 archive_entry_set_mtime.3 MLINKS+= archive_entry.3 archive_entry_set_nlink.3 MLINKS+= archive_entry.3 archive_entry_set_pathname.3 MLINKS+= archive_entry.3 archive_entry_set_rdev.3 MLINKS+= archive_entry.3 archive_entry_set_rdevmajor.3 MLINKS+= archive_entry.3 archive_entry_set_rdevminor.3 MLINKS+= archive_entry.3 archive_entry_set_size.3 MLINKS+= archive_entry.3 archive_entry_set_symlink.3 MLINKS+= archive_entry.3 archive_entry_set_uid.3 MLINKS+= archive_entry.3 archive_entry_set_uname.3 MLINKS+= archive_entry.3 archive_entry_size.3 MLINKS+= archive_entry.3 archive_entry_stat.3 MLINKS+= archive_entry.3 archive_entry_symlink.3 MLINKS+= archive_entry.3 archive_entry_uid.3 MLINKS+= archive_entry.3 archive_entry_uname.3 MLINKS+= archive_entry.3 archive_entry_uname_w.3 MLINKS+= archive_read.3 archive_read_data.3 MLINKS+= archive_read.3 archive_read_data_block.3 MLINKS+= archive_read.3 archive_read_data_into_buffer.3 MLINKS+= archive_read.3 archive_read_data_into_fd.3 MLINKS+= archive_read.3 archive_read_data_skip.3 MLINKS+= archive_read.3 archive_read_extract.3 MLINKS+= archive_read.3 archive_read_extract_set_progress_callback.3 MLINKS+= archive_read.3 archive_read_extract_set_skip_file.3 MLINKS+= archive_read.3 archive_read_finish.3 MLINKS+= archive_read.3 archive_read_new.3 MLINKS+= archive_read.3 archive_read_next_header.3 MLINKS+= archive_read.3 archive_read_next_header2.3 MLINKS+= archive_read.3 archive_read_open.3 MLINKS+= archive_read.3 archive_read_open2.3 MLINKS+= archive_read.3 archive_read_open_FILE.3 MLINKS+= archive_read.3 archive_read_open_fd.3 MLINKS+= archive_read.3 archive_read_open_file.3 MLINKS+= archive_read.3 archive_read_open_filename.3 MLINKS+= archive_read.3 archive_read_open_memory.3 MLINKS+= archive_read.3 archive_read_support_compression_all.3 MLINKS+= archive_read.3 archive_read_support_compression_bzip2.3 MLINKS+= archive_read.3 archive_read_support_compression_compress.3 MLINKS+= archive_read.3 archive_read_support_compression_gzip.3 MLINKS+= archive_read.3 archive_read_support_compression_lzma.3 MLINKS+= archive_read.3 archive_read_support_compression_none.3 MLINKS+= archive_read.3 archive_read_support_compression_program.3 MLINKS+= archive_read.3 archive_read_support_compression_program_signature.3 MLINKS+= archive_read.3 archive_read_support_compression_xz.3 MLINKS+= archive_read.3 archive_read_support_format_all.3 MLINKS+= archive_read.3 archive_read_support_format_ar.3 MLINKS+= archive_read.3 archive_read_support_format_cpio.3 MLINKS+= archive_read.3 archive_read_support_format_empty.3 MLINKS+= archive_read.3 archive_read_support_format_iso9660.3 MLINKS+= archive_read.3 archive_read_support_format_raw.3 MLINKS+= archive_read.3 archive_read_support_format_tar.3 MLINKS+= archive_read.3 archive_read_support_format_zip.3 MLINKS+= archive_read_disk.3 archive_read_disk_entry_from_file.3 MLINKS+= archive_read_disk.3 archive_read_disk_gname.3 MLINKS+= archive_read_disk.3 archive_read_disk_new.3 MLINKS+= archive_read_disk.3 archive_read_disk_set_gname_lookup.3 MLINKS+= archive_read_disk.3 archive_read_disk_set_standard_lookup.3 MLINKS+= archive_read_disk.3 archive_read_disk_set_symlink_hybrid.3 MLINKS+= archive_read_disk.3 archive_read_disk_set_symlink_logical.3 MLINKS+= archive_read_disk.3 archive_read_disk_set_symlink_physical.3 MLINKS+= archive_read_disk.3 archive_read_disk_set_uname_lookup.3 MLINKS+= archive_read_disk.3 archive_read_disk_uname.3 MLINKS+= archive_util.3 archive_clear_error.3 MLINKS+= archive_util.3 archive_compression.3 MLINKS+= archive_util.3 archive_compression_name.3 MLINKS+= archive_util.3 archive_errno.3 MLINKS+= archive_util.3 archive_error_string.3 MLINKS+= archive_util.3 archive_file_count.3 MLINKS+= archive_util.3 archive_format.3 MLINKS+= archive_util.3 archive_format_name.3 MLINKS+= archive_util.3 archive_set_error.3 MLINKS+= archive_write.3 archive_write_close.3 MLINKS+= archive_write.3 archive_write_data.3 MLINKS+= archive_write.3 archive_write_finish.3 MLINKS+= archive_write.3 archive_write_finish_entry.3 MLINKS+= archive_write.3 archive_write_get_bytes_in_last_block.3 MLINKS+= archive_write.3 archive_write_get_bytes_per_block.3 MLINKS+= archive_write.3 archive_write_header.3 MLINKS+= archive_write.3 archive_write_new.3 MLINKS+= archive_write.3 archive_write_open.3 MLINKS+= archive_write.3 archive_write_open_FILE.3 MLINKS+= archive_write.3 archive_write_open_fd.3 MLINKS+= archive_write.3 archive_write_open_file.3 MLINKS+= archive_write.3 archive_write_open_filename.3 MLINKS+= archive_write.3 archive_write_open_memory.3 MLINKS+= archive_write.3 archive_write_set_bytes_in_last_block.3 MLINKS+= archive_write.3 archive_write_set_bytes_per_block.3 MLINKS+= archive_write.3 archive_write_set_callbacks.3 MLINKS+= archive_write.3 archive_write_set_compression_bzip2.3 MLINKS+= archive_write.3 archive_write_set_compression_compress.3 MLINKS+= archive_write.3 archive_write_set_compression_gzip.3 MLINKS+= archive_write.3 archive_write_set_compression_none.3 MLINKS+= archive_write.3 archive_write_set_compression_program.3 MLINKS+= archive_write.3 archive_write_set_format_pax.3 MLINKS+= archive_write.3 archive_write_set_format_shar.3 MLINKS+= archive_write.3 archive_write_set_format_ustar.3 MLINKS+= archive_write_disk.3 archive_write_disk_new.3 MLINKS+= archive_write_disk.3 archive_write_disk_set_group_lookup.3 MLINKS+= archive_write_disk.3 archive_write_disk_set_options.3 MLINKS+= archive_write_disk.3 archive_write_disk_set_skip_file.3 MLINKS+= archive_write_disk.3 archive_write_disk_set_standard_lookup.3 MLINKS+= archive_write_disk.3 archive_write_disk_set_user_lookup.3 MLINKS+= libarchive.3 archive.3 .PHONY: check test check test: cd ${.CURDIR}/test && make test .include Index: projects/largeSMP/lib/libarchive/archive_hash.h =================================================================== --- projects/largeSMP/lib/libarchive/archive_hash.h (revision 221494) +++ projects/largeSMP/lib/libarchive/archive_hash.h (revision 221495) @@ -1,196 +1,196 @@ /*- * Copyright (c) 2009 Joerg Sonnenberger * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * $FreeBSD$ */ #ifndef __LIBARCHIVE_BUILD #error This header is only to be used internally to libarchive. #endif /* * Hash function support in various Operating Systems: * * NetBSD: * - MD5 and SHA1 in libc: without _ after algorithm name * - SHA2 in libc: with _ after algorithm name * * OpenBSD: * - MD5, SHA1 and SHA2 in libc: without _ after algorithm name * - OpenBSD 4.4 and earlier have SHA2 in libc with _ after algorithm name * * DragonFly and FreeBSD (XXX not used yet): - * - MD5 and SHA1 in libmd: without _ after algorithm name - * - SHA256: with _ after algorithm name + * - MD5 in libmd: without _ after algorithm name + * - SHA{1,256,512} in libmd: with _ after algorithm name (conflicts w/OpenSSL!) * * OpenSSL: - * - MD5, SHA1 and SHA2 in libcrypto: with _ after algorithm name + * - MD5, SHA1, SHA2, SHA{256,384,512} in libcrypto: with _ after algorithm name */ #if defined(HAVE_MD5_H) && defined(HAVE_MD5INIT) # include # define ARCHIVE_HAS_MD5 typedef MD5_CTX archive_md5_ctx; # define archive_md5_init(ctx) MD5Init(ctx) # define archive_md5_final(ctx, buf) MD5Final(buf, ctx) # define archive_md5_update(ctx, buf, n) MD5Update(ctx, buf, n) #elif defined(HAVE_OPENSSL_MD5_H) # include # define ARCHIVE_HAS_MD5 typedef MD5_CTX archive_md5_ctx; # define archive_md5_init(ctx) MD5_Init(ctx) # define archive_md5_final(ctx, buf) MD5_Final(buf, ctx) # define archive_md5_update(ctx, buf, n) MD5_Update(ctx, buf, n) #elif defined(_WIN32) && !defined(__CYGWIN__) && defined(CALG_MD5) # define ARCHIVE_HAS_MD5 typedef MD5_CTX archive_md5_ctx; # define archive_md5_init(ctx) MD5_Init(ctx) # define archive_md5_final(ctx, buf) MD5_Final(buf, ctx) # define archive_md5_update(ctx, buf, n) MD5_Update(ctx, buf, n) #endif #if defined(HAVE_RMD160_H) && defined(HAVE_RMD160INIT) # include # define ARCHIVE_HAS_RMD160 typedef RMD160_CTX archive_rmd160_ctx; # define archive_rmd160_init(ctx) RMD160Init(ctx) # define archive_rmd160_final(ctx, buf) RMD160Final(buf, ctx) # define archive_rmd160_update(ctx, buf, n) RMD160Update(ctx, buf, n) #elif defined(HAVE_OPENSSL_RIPEMD_H) # include # define ARCHIVE_HAS_RMD160 typedef RIPEMD160_CTX archive_rmd160_ctx; # define archive_rmd160_init(ctx) RIPEMD160_Init(ctx) # define archive_rmd160_final(ctx, buf) RIPEMD160_Final(buf, ctx) # define archive_rmd160_update(ctx, buf, n) RIPEMD160_Update(ctx, buf, n) #endif #if defined(HAVE_SHA1_H) && defined(HAVE_SHA1INIT) # include # define ARCHIVE_HAS_SHA1 typedef SHA1_CTX archive_sha1_ctx; # define archive_sha1_init(ctx) SHA1Init(ctx) # define archive_sha1_final(ctx, buf) SHA1Final(buf, ctx) # define archive_sha1_update(ctx, buf, n) SHA1Update(ctx, buf, n) #elif defined(HAVE_OPENSSL_SHA_H) # include # define ARCHIVE_HAS_SHA1 typedef SHA_CTX archive_sha1_ctx; # define archive_sha1_init(ctx) SHA1_Init(ctx) # define archive_sha1_final(ctx, buf) SHA1_Final(buf, ctx) # define archive_sha1_update(ctx, buf, n) SHA1_Update(ctx, buf, n) #elif defined(_WIN32) && !defined(__CYGWIN__) && defined(CALG_SHA1) # define ARCHIVE_HAS_SHA1 typedef SHA1_CTX archive_sha1_ctx; # define archive_sha1_init(ctx) SHA1_Init(ctx) # define archive_sha1_final(ctx, buf) SHA1_Final(buf, ctx) # define archive_sha1_update(ctx, buf, n) SHA1_Update(ctx, buf, n) #endif #if defined(HAVE_SHA2_H) && defined(HAVE_SHA256_INIT) # include # define ARCHIVE_HAS_SHA256 typedef SHA256_CTX archive_sha256_ctx; # define archive_sha256_init(ctx) SHA256_Init(ctx) # define archive_sha256_final(ctx, buf) SHA256_Final(buf, ctx) # define archive_sha256_update(ctx, buf, n) SHA256_Update(ctx, buf, n) #elif defined(HAVE_SHA2_H) && defined(HAVE_SHA256INIT) # include # define ARCHIVE_HAS_SHA256 typedef SHA256_CTX archive_sha256_ctx; # define archive_sha256_init(ctx) SHA256Init(ctx) # define archive_sha256_final(ctx, buf) SHA256Final(buf, ctx) # define archive_sha256_update(ctx, buf, n) SHA256Update(ctx, buf, n) #elif defined(HAVE_OPENSSL_SHA_H) && defined(HAVE_OPENSSL_SHA256_INIT) # include # define ARCHIVE_HAS_SHA256 typedef SHA256_CTX archive_sha256_ctx; # define archive_sha256_init(ctx) SHA256_Init(ctx) # define archive_sha256_final(ctx, buf) SHA256_Final(buf, ctx) # define archive_sha256_update(ctx, buf, n) SHA256_Update(ctx, buf, n) #elif defined(_WIN32) && !defined(__CYGWIN__) && defined(CALG_SHA_256) # define ARCHIVE_HAS_SHA256 typedef SHA256_CTX archive_sha256_ctx; # define archive_sha256_init(ctx) SHA256_Init(ctx) # define archive_sha256_final(ctx, buf) SHA256_Final(buf, ctx) # define archive_sha256_update(ctx, buf, n) SHA256_Update(ctx, buf, n) #endif #if defined(HAVE_SHA2_H) && defined(HAVE_SHA384_INIT) # include # define ARCHIVE_HAS_SHA384 typedef SHA384_CTX archive_sha384_ctx; # define archive_sha384_init(ctx) SHA384_Init(ctx) # define archive_sha384_final(ctx, buf) SHA384_Final(buf, ctx) # define archive_sha384_update(ctx, buf, n) SHA384_Update(ctx, buf, n) #elif defined(HAVE_SHA2_H) && defined(HAVE_SHA384INIT) # include # define ARCHIVE_HAS_SHA384 typedef SHA384_CTX archive_sha384_ctx; # define archive_sha384_init(ctx) SHA384Init(ctx) # define archive_sha384_final(ctx, buf) SHA384Final(buf, ctx) # define archive_sha384_update(ctx, buf, n) SHA384Update(ctx, buf, n) #elif defined(HAVE_OPENSSL_SHA_H) && defined(HAVE_OPENSSL_SHA384_INIT) # include # define ARCHIVE_HAS_SHA384 typedef SHA512_CTX archive_sha384_ctx; # define archive_sha384_init(ctx) SHA384_Init(ctx) # define archive_sha384_final(ctx, buf) SHA384_Final(buf, ctx) # define archive_sha384_update(ctx, buf, n) SHA384_Update(ctx, buf, n) #elif defined(_WIN32) && !defined(__CYGWIN__) && defined(CALG_SHA_384) # define ARCHIVE_HAS_SHA384 typedef SHA512_CTX archive_sha384_ctx; # define archive_sha384_init(ctx) SHA384_Init(ctx) # define archive_sha384_final(ctx, buf) SHA384_Final(buf, ctx) # define archive_sha384_update(ctx, buf, n) SHA384_Update(ctx, buf, n) #endif #if defined(HAVE_SHA2_H) && defined(HAVE_SHA512_INIT) # include # define ARCHIVE_HAS_SHA512 typedef SHA512_CTX archive_sha512_ctx; # define archive_sha512_init(ctx) SHA512_Init(ctx) # define archive_sha512_final(ctx, buf) SHA512_Final(buf, ctx) # define archive_sha512_update(ctx, buf, n) SHA512_Update(ctx, buf, n) #elif defined(HAVE_SHA2_H) && defined(HAVE_SHA512INIT) # include # define ARCHIVE_HAS_SHA512 typedef SHA512_CTX archive_sha512_ctx; # define archive_sha512_init(ctx) SHA512Init(ctx) # define archive_sha512_final(ctx, buf) SHA512Final(buf, ctx) # define archive_sha512_update(ctx, buf, n) SHA512Update(ctx, buf, n) #elif defined(HAVE_OPENSSL_SHA_H) && defined(HAVE_OPENSSL_SHA512_INIT) # include # define ARCHIVE_HAS_SHA512 typedef SHA512_CTX archive_sha512_ctx; # define archive_sha512_init(ctx) SHA512_Init(ctx) # define archive_sha512_final(ctx, buf) SHA512_Final(buf, ctx) # define archive_sha512_update(ctx, buf, n) SHA512_Update(ctx, buf, n) #elif defined(_WIN32) && !defined(__CYGWIN__) && defined(CALG_SHA_512) # define ARCHIVE_HAS_SHA512 typedef SHA512_CTX archive_sha512_ctx; # define archive_sha512_init(ctx) SHA512_Init(ctx) # define archive_sha512_final(ctx, buf) SHA512_Final(buf, ctx) # define archive_sha512_update(ctx, buf, n) SHA512_Update(ctx, buf, n) #endif Index: projects/largeSMP/lib/libarchive/config_freebsd.h =================================================================== --- projects/largeSMP/lib/libarchive/config_freebsd.h (revision 221494) +++ projects/largeSMP/lib/libarchive/config_freebsd.h (revision 221495) @@ -1,153 +1,164 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * $FreeBSD$ */ /* FreeBSD 5.0 and later have ACL and extattr support. */ #if __FreeBSD__ > 4 #define HAVE_ACL_CREATE_ENTRY 1 #define HAVE_ACL_GET_LINK_NP 1 #define HAVE_ACL_GET_PERM_NP 1 #define HAVE_ACL_INIT 1 #define HAVE_ACL_SET_FD 1 #define HAVE_ACL_SET_FD_NP 1 #define HAVE_ACL_SET_FILE 1 #define HAVE_ACL_USER 1 #define HAVE_EXTATTR_GET_FILE 1 #define HAVE_EXTATTR_LIST_FILE 1 #define HAVE_EXTATTR_SET_FD 1 #define HAVE_EXTATTR_SET_FILE 1 #define HAVE_SYS_ACL_H 1 #define HAVE_SYS_EXTATTR_H 1 #endif #ifdef WITH_OPENSSL #define HAVE_OPENSSL_MD5_H 1 #define HAVE_OPENSSL_RIPEMD_H 1 #define HAVE_OPENSSL_SHA_H 1 +#define HAVE_OPENSSL_SHA256_INIT 1 +#define HAVE_OPENSSL_SHA384_INIT 1 +#define HAVE_OPENSSL_SHA512_INIT 1 +#define HAVE_SHA256 1 #define HAVE_SHA384 1 #define HAVE_SHA512 1 +#else +#define HAVE_MD5_H 1 +#define HAVE_MD5INIT 1 +#define HAVE_SHA_H 1 +#define HAVE_SHA1 1 +#define HAVE_SHA1_INIT 1 +#define HAVE_SHA256 1 +#define HAVE_SHA256_H 1 +#define HAVE_SHA256_INIT 1 +#define HAVE_SHA512 1 +#define HAVE_SHA512_H 1 +#define HAVE_SHA512_INIT 1 #endif #define HAVE_BSDXML_H 1 #define HAVE_CHFLAGS 1 #define HAVE_CHOWN 1 #define HAVE_DECL_INT64_MAX 1 #define HAVE_DECL_INT64_MIN 1 #define HAVE_DECL_SIZE_MAX 1 #define HAVE_DECL_SSIZE_MAX 1 #define HAVE_DECL_STRERROR_R 1 #define HAVE_DECL_UINT32_MAX 1 #define HAVE_DECL_UINT64_MAX 1 #define HAVE_DIRENT_H 1 #define HAVE_EFTYPE 1 #define HAVE_EILSEQ 1 #define HAVE_ERRNO_H 1 #define HAVE_FCHDIR 1 #define HAVE_FCHFLAGS 1 #define HAVE_FCHMOD 1 #define HAVE_FCHOWN 1 #define HAVE_FCNTL 1 #define HAVE_FCNTL_H 1 #define HAVE_FSEEKO 1 #define HAVE_FSTAT 1 #define HAVE_FTRUNCATE 1 #define HAVE_FUTIMES 1 #define HAVE_GETEUID 1 #define HAVE_GETPID 1 #define HAVE_GRP_H 1 #define HAVE_INTTYPES_H 1 #define HAVE_LCHFLAGS 1 #define HAVE_LCHMOD 1 #define HAVE_LCHOWN 1 #define HAVE_LIMITS_H 1 #define HAVE_LINK 1 #define HAVE_LSTAT 1 #define HAVE_LUTIMES 1 #define HAVE_MALLOC 1 #define HAVE_MD5 1 -#define HAVE_MD5_H 1 #define HAVE_MEMMOVE 1 #define HAVE_MKDIR 1 #define HAVE_MKFIFO 1 #define HAVE_MKNOD 1 #define HAVE_PIPE 1 #define HAVE_POLL 1 #define HAVE_POLL_H 1 #define HAVE_PWD_H 1 #define HAVE_READLINK 1 #define HAVE_RMD160 1 #define HAVE_SELECT 1 #define HAVE_SETENV 1 -#define HAVE_SHA_H 1 -#define HAVE_SHA1 1 -#define HAVE_SHA256 1 -#define HAVE_SHA256_H 1 #define HAVE_SIGNAL_H 1 #define HAVE_STDINT_H 1 #define HAVE_STDLIB_H 1 #define HAVE_STRCHR 1 #define HAVE_STRDUP 1 #define HAVE_STRERROR 1 #define HAVE_STRERROR_R 1 #define HAVE_STRINGS_H 1 #define HAVE_STRING_H 1 #define HAVE_STRRCHR 1 #define HAVE_STRUCT_STAT_ST_BLKSIZE 1 #define HAVE_STRUCT_STAT_ST_BIRTHTIME 1 #define HAVE_STRUCT_STAT_ST_BIRTHTIMESPEC_TV_NSEC 1 #define HAVE_STRUCT_STAT_ST_FLAGS 1 #define HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC 1 #define HAVE_SYMLINK 1 #define HAVE_SYS_CDEFS_H 1 #define HAVE_SYS_IOCTL_H 1 #define HAVE_SYS_SELECT_H 1 #define HAVE_SYS_STAT_H 1 #define HAVE_SYS_TIME_H 1 #define HAVE_SYS_TYPES_H 1 #undef HAVE_SYS_UTIME_H #define HAVE_SYS_WAIT_H 1 #define HAVE_TIMEGM 1 #define HAVE_TZSET 1 #define HAVE_UNISTD_H 1 #define HAVE_UNSETENV 1 #define HAVE_UTIME 1 #define HAVE_UTIMES 1 #define HAVE_UTIME_H 1 #define HAVE_VFORK 1 #define HAVE_WCHAR_H 1 #define HAVE_WCSCPY 1 #define HAVE_WCSLEN 1 #define HAVE_WCTOMB 1 #define HAVE_WMEMCMP 1 #define HAVE_WMEMCPY 1 #define HAVE_ZLIB_H 1 #define TIME_WITH_SYS_TIME 1 /* FreeBSD 4 and earlier lack intmax_t/uintmax_t */ #if __FreeBSD__ < 5 #define intmax_t int64_t #define uintmax_t uint64_t #endif Index: projects/largeSMP/lib/libc/stdtime =================================================================== --- projects/largeSMP/lib/libc/stdtime (revision 221494) +++ projects/largeSMP/lib/libc/stdtime (revision 221495) Property changes on: projects/largeSMP/lib/libc/stdtime ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/lib/libc/stdtime:r221444-221493 Index: projects/largeSMP/lib/libc =================================================================== --- projects/largeSMP/lib/libc (revision 221494) +++ projects/largeSMP/lib/libc (revision 221495) Property changes on: projects/largeSMP/lib/libc ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/lib/libc:r221444-221493 Index: projects/largeSMP/lib/libcrypt/crypt-sha256.c =================================================================== --- projects/largeSMP/lib/libcrypt/crypt-sha256.c (revision 221494) +++ projects/largeSMP/lib/libcrypt/crypt-sha256.c (revision 221495) @@ -1,477 +1,477 @@ /* * Copyright (c) 2011 The FreeBSD Project. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ /* Based on: * SHA256-based Unix crypt implementation. Released into the Public Domain by * Ulrich Drepper . */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include "crypt.h" /* Define our magic string to mark salt for SHA256 "encryption" replacement. */ static const char sha256_salt_prefix[] = "$5$"; /* Prefix for optional rounds specification. */ static const char sha256_rounds_prefix[] = "rounds="; /* Maximum salt string length. */ #define SALT_LEN_MAX 16 /* Default number of rounds if not explicitly specified. */ #define ROUNDS_DEFAULT 5000 /* Minimum number of rounds. */ #define ROUNDS_MIN 1000 /* Maximum number of rounds. */ #define ROUNDS_MAX 999999999 static char * -sha256_crypt_r(const char *key, const char *salt, char *buffer, int buflen) +crypt_sha256_r(const char *key, const char *salt, char *buffer, int buflen) { u_long srounds; int n; uint8_t alt_result[32], temp_result[32]; SHA256_CTX ctx, alt_ctx; size_t salt_len, key_len, cnt, rounds; char *cp, *copied_key, *copied_salt, *p_bytes, *s_bytes, *endp; const char *num; bool rounds_custom; copied_key = NULL; copied_salt = NULL; /* Default number of rounds. */ rounds = ROUNDS_DEFAULT; rounds_custom = false; /* Find beginning of salt string. The prefix should normally always * be present. Just in case it is not. */ if (strncmp(sha256_salt_prefix, salt, sizeof(sha256_salt_prefix) - 1) == 0) /* Skip salt prefix. */ salt += sizeof(sha256_salt_prefix) - 1; if (strncmp(salt, sha256_rounds_prefix, sizeof(sha256_rounds_prefix) - 1) == 0) { num = salt + sizeof(sha256_rounds_prefix) - 1; srounds = strtoul(num, &endp, 10); if (*endp == '$') { salt = endp + 1; rounds = MAX(ROUNDS_MIN, MIN(srounds, ROUNDS_MAX)); rounds_custom = true; } } salt_len = MIN(strcspn(salt, "$"), SALT_LEN_MAX); key_len = strlen(key); /* Prepare for the real work. */ SHA256_Init(&ctx); /* Add the key string. */ SHA256_Update(&ctx, key, key_len); /* The last part is the salt string. This must be at most 8 * characters and it ends at the first `$' character (for * compatibility with existing implementations). */ SHA256_Update(&ctx, salt, salt_len); /* Compute alternate SHA256 sum with input KEY, SALT, and KEY. The * final result will be added to the first context. */ SHA256_Init(&alt_ctx); /* Add key. */ SHA256_Update(&alt_ctx, key, key_len); /* Add salt. */ SHA256_Update(&alt_ctx, salt, salt_len); /* Add key again. */ SHA256_Update(&alt_ctx, key, key_len); /* Now get result of this (32 bytes) and add it to the other context. */ SHA256_Final(alt_result, &alt_ctx); /* Add for any character in the key one byte of the alternate sum. */ for (cnt = key_len; cnt > 32; cnt -= 32) SHA256_Update(&ctx, alt_result, 32); SHA256_Update(&ctx, alt_result, cnt); /* Take the binary representation of the length of the key and for * every 1 add the alternate sum, for every 0 the key. */ for (cnt = key_len; cnt > 0; cnt >>= 1) if ((cnt & 1) != 0) SHA256_Update(&ctx, alt_result, 32); else SHA256_Update(&ctx, key, key_len); /* Create intermediate result. */ SHA256_Final(alt_result, &ctx); /* Start computation of P byte sequence. */ SHA256_Init(&alt_ctx); /* For every character in the password add the entire password. */ for (cnt = 0; cnt < key_len; ++cnt) SHA256_Update(&alt_ctx, key, key_len); /* Finish the digest. */ SHA256_Final(temp_result, &alt_ctx); /* Create byte sequence P. */ cp = p_bytes = alloca(key_len); for (cnt = key_len; cnt >= 32; cnt -= 32) { memcpy(cp, temp_result, 32); cp += 32; } memcpy(cp, temp_result, cnt); /* Start computation of S byte sequence. */ SHA256_Init(&alt_ctx); /* For every character in the password add the entire password. */ for (cnt = 0; cnt < 16 + alt_result[0]; ++cnt) SHA256_Update(&alt_ctx, salt, salt_len); /* Finish the digest. */ SHA256_Final(temp_result, &alt_ctx); /* Create byte sequence S. */ cp = s_bytes = alloca(salt_len); for (cnt = salt_len; cnt >= 32; cnt -= 32) { memcpy(cp, temp_result, 32); cp += 32; } memcpy(cp, temp_result, cnt); /* Repeatedly run the collected hash value through SHA256 to burn CPU * cycles. */ for (cnt = 0; cnt < rounds; ++cnt) { /* New context. */ SHA256_Init(&ctx); /* Add key or last result. */ if ((cnt & 1) != 0) SHA256_Update(&ctx, p_bytes, key_len); else SHA256_Update(&ctx, alt_result, 32); /* Add salt for numbers not divisible by 3. */ if (cnt % 3 != 0) SHA256_Update(&ctx, s_bytes, salt_len); /* Add key for numbers not divisible by 7. */ if (cnt % 7 != 0) SHA256_Update(&ctx, p_bytes, key_len); /* Add key or last result. */ if ((cnt & 1) != 0) SHA256_Update(&ctx, alt_result, 32); else SHA256_Update(&ctx, p_bytes, key_len); /* Create intermediate result. */ SHA256_Final(alt_result, &ctx); } /* Now we can construct the result string. It consists of three * parts. */ cp = stpncpy(buffer, sha256_salt_prefix, MAX(0, buflen)); buflen -= sizeof(sha256_salt_prefix) - 1; if (rounds_custom) { n = snprintf(cp, MAX(0, buflen), "%s%zu$", sha256_rounds_prefix, rounds); cp += n; buflen -= n; } cp = stpncpy(cp, salt, MIN((size_t)MAX(0, buflen), salt_len)); buflen -= MIN((size_t)MAX(0, buflen), salt_len); if (buflen > 0) { *cp++ = '$'; --buflen; } b64_from_24bit(alt_result[0], alt_result[10], alt_result[20], 4, &buflen, &cp); b64_from_24bit(alt_result[21], alt_result[1], alt_result[11], 4, &buflen, &cp); b64_from_24bit(alt_result[12], alt_result[22], alt_result[2], 4, &buflen, &cp); b64_from_24bit(alt_result[3], alt_result[13], alt_result[23], 4, &buflen, &cp); b64_from_24bit(alt_result[24], alt_result[4], alt_result[14], 4, &buflen, &cp); b64_from_24bit(alt_result[15], alt_result[25], alt_result[5], 4, &buflen, &cp); b64_from_24bit(alt_result[6], alt_result[16], alt_result[26], 4, &buflen, &cp); b64_from_24bit(alt_result[27], alt_result[7], alt_result[17], 4, &buflen, &cp); b64_from_24bit(alt_result[18], alt_result[28], alt_result[8], 4, &buflen, &cp); b64_from_24bit(alt_result[9], alt_result[19], alt_result[29], 4, &buflen, &cp); b64_from_24bit(0, alt_result[31], alt_result[30], 3, &buflen, &cp); if (buflen <= 0) { errno = ERANGE; buffer = NULL; } else *cp = '\0'; /* Terminate the string. */ /* Clear the buffer for the intermediate result so that people * attaching to processes or reading core dumps cannot get any * information. We do it in this way to clear correct_words[] inside * the SHA256 implementation as well. */ SHA256_Init(&ctx); SHA256_Final(alt_result, &ctx); memset(temp_result, '\0', sizeof(temp_result)); memset(p_bytes, '\0', key_len); memset(s_bytes, '\0', salt_len); memset(&ctx, '\0', sizeof(ctx)); memset(&alt_ctx, '\0', sizeof(alt_ctx)); if (copied_key != NULL) memset(copied_key, '\0', key_len); if (copied_salt != NULL) memset(copied_salt, '\0', salt_len); return buffer; } /* This entry point is equivalent to crypt(3). */ char * -sha256_crypt(const char *key, const char *salt) +crypt_sha256(const char *key, const char *salt) { /* We don't want to have an arbitrary limit in the size of the * password. We can compute an upper bound for the size of the * result in advance and so we can prepare the buffer we pass to - * `sha256_crypt_r'. */ + * `crypt_sha256_r'. */ static char *buffer; static int buflen; int needed; char *new_buffer; needed = (sizeof(sha256_salt_prefix) - 1 + sizeof(sha256_rounds_prefix) + 9 + 1 + strlen(salt) + 1 + 43 + 1); if (buflen < needed) { new_buffer = (char *)realloc(buffer, needed); if (new_buffer == NULL) return NULL; buffer = new_buffer; buflen = needed; } - return sha256_crypt_r(key, salt, buffer, buflen); + return crypt_sha256_r(key, salt, buffer, buflen); } #ifdef TEST static const struct { const char *input; const char result[32]; } tests[] = { /* Test vectors from FIPS 180-2: appendix B.1. */ { "abc", "\xba\x78\x16\xbf\x8f\x01\xcf\xea\x41\x41\x40\xde\x5d\xae\x22\x23" "\xb0\x03\x61\xa3\x96\x17\x7a\x9c\xb4\x10\xff\x61\xf2\x00\x15\xad" }, /* Test vectors from FIPS 180-2: appendix B.2. */ { "abcdbcdecdefdefgefghfghighijhijkijkljklmklmnlmnomnopnopq", "\x24\x8d\x6a\x61\xd2\x06\x38\xb8\xe5\xc0\x26\x93\x0c\x3e\x60\x39" "\xa3\x3c\xe4\x59\x64\xff\x21\x67\xf6\xec\xed\xd4\x19\xdb\x06\xc1" }, /* Test vectors from the NESSIE project. */ { "", "\xe3\xb0\xc4\x42\x98\xfc\x1c\x14\x9a\xfb\xf4\xc8\x99\x6f\xb9\x24" "\x27\xae\x41\xe4\x64\x9b\x93\x4c\xa4\x95\x99\x1b\x78\x52\xb8\x55" }, { "a", "\xca\x97\x81\x12\xca\x1b\xbd\xca\xfa\xc2\x31\xb3\x9a\x23\xdc\x4d" "\xa7\x86\xef\xf8\x14\x7c\x4e\x72\xb9\x80\x77\x85\xaf\xee\x48\xbb" }, { "message digest", "\xf7\x84\x6f\x55\xcf\x23\xe1\x4e\xeb\xea\xb5\xb4\xe1\x55\x0c\xad" "\x5b\x50\x9e\x33\x48\xfb\xc4\xef\xa3\xa1\x41\x3d\x39\x3c\xb6\x50" }, { "abcdefghijklmnopqrstuvwxyz", "\x71\xc4\x80\xdf\x93\xd6\xae\x2f\x1e\xfa\xd1\x44\x7c\x66\xc9\x52" "\x5e\x31\x62\x18\xcf\x51\xfc\x8d\x9e\xd8\x32\xf2\xda\xf1\x8b\x73" }, { "abcdbcdecdefdefgefghfghighijhijkijkljklmklmnlmnomnopnopq", "\x24\x8d\x6a\x61\xd2\x06\x38\xb8\xe5\xc0\x26\x93\x0c\x3e\x60\x39" "\xa3\x3c\xe4\x59\x64\xff\x21\x67\xf6\xec\xed\xd4\x19\xdb\x06\xc1" }, { "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789", "\xdb\x4b\xfc\xbd\x4d\xa0\xcd\x85\xa6\x0c\x3c\x37\xd3\xfb\xd8\x80" "\x5c\x77\xf1\x5f\xc6\xb1\xfd\xfe\x61\x4e\xe0\xa7\xc8\xfd\xb4\xc0" }, { "123456789012345678901234567890123456789012345678901234567890" "12345678901234567890", "\xf3\x71\xbc\x4a\x31\x1f\x2b\x00\x9e\xef\x95\x2d\xd8\x3c\xa8\x0e" "\x2b\x60\x02\x6c\x8e\x93\x55\x92\xd0\xf9\xc3\x08\x45\x3c\x81\x3e" } }; #define ntests (sizeof (tests) / sizeof (tests[0])) static const struct { const char *salt; const char *input; const char *expected; } tests2[] = { { "$5$saltstring", "Hello world!", "$5$saltstring$5B8vYYiY.CVt1RlTTf8KbXBH3hsxY/GNooZaBBGWEc5" }, { "$5$rounds=10000$saltstringsaltstring", "Hello world!", "$5$rounds=10000$saltstringsaltst$3xv.VbSHBb41AL9AvLeujZkZRBAwqFMz2." "opqey6IcA" }, { "$5$rounds=5000$toolongsaltstring", "This is just a test", "$5$rounds=5000$toolongsaltstrin$Un/5jzAHMgOGZ5.mWJpuVolil07guHPvOW8" "mGRcvxa5" }, { "$5$rounds=1400$anotherlongsaltstring", "a very much longer text to encrypt. This one even stretches over more" "than one line.", "$5$rounds=1400$anotherlongsalts$Rx.j8H.h8HjEDGomFU8bDkXm3XIUnzyxf12" "oP84Bnq1" }, { "$5$rounds=77777$short", "we have a short salt string but not a short password", "$5$rounds=77777$short$JiO1O3ZpDAxGJeaDIuqCoEFysAe1mZNJRs3pw0KQRd/" }, { "$5$rounds=123456$asaltof16chars..", "a short string", "$5$rounds=123456$asaltof16chars..$gP3VQ/6X7UUEW3HkBn2w1/Ptq2jxPyzV/" "cZKmF/wJvD" }, { "$5$rounds=10$roundstoolow", "the minimum number is still observed", "$5$rounds=1000$roundstoolow$yfvwcWrQ8l/K0DAWyuPMDNHpIVlTQebY9l/gL97" "2bIC" }, }; #define ntests2 (sizeof (tests2) / sizeof (tests2[0])) int main(void) { SHA256_CTX ctx; uint8_t sum[32]; int result = 0; int i, cnt; for (cnt = 0; cnt < (int)ntests; ++cnt) { SHA256_Init(&ctx); SHA256_Update(&ctx, tests[cnt].input, strlen(tests[cnt].input)); SHA256_Final(sum, &ctx); if (memcmp(tests[cnt].result, sum, 32) != 0) { for (i = 0; i < 32; i++) printf("%02X", tests[cnt].result[i]); printf("\n"); for (i = 0; i < 32; i++) printf("%02X", sum[i]); printf("\n"); printf("test %d run %d failed\n", cnt, 1); result = 1; } SHA256_Init(&ctx); for (i = 0; tests[cnt].input[i] != '\0'; ++i) SHA256_Update(&ctx, &tests[cnt].input[i], 1); SHA256_Final(sum, &ctx); if (memcmp(tests[cnt].result, sum, 32) != 0) { for (i = 0; i < 32; i++) printf("%02X", tests[cnt].result[i]); printf("\n"); for (i = 0; i < 32; i++) printf("%02X", sum[i]); printf("\n"); printf("test %d run %d failed\n", cnt, 2); result = 1; } } /* Test vector from FIPS 180-2: appendix B.3. */ char buf[1000]; memset(buf, 'a', sizeof(buf)); SHA256_Init(&ctx); for (i = 0; i < 1000; ++i) SHA256_Update(&ctx, buf, sizeof(buf)); SHA256_Final(sum, &ctx); static const char expected[32] = "\xcd\xc7\x6e\x5c\x99\x14\xfb\x92\x81\xa1\xc7\xe2\x84\xd7\x3e\x67" "\xf1\x80\x9a\x48\xa4\x97\x20\x0e\x04\x6d\x39\xcc\xc7\x11\x2c\xd0"; if (memcmp(expected, sum, 32) != 0) { printf("test %d failed\n", cnt); result = 1; } for (cnt = 0; cnt < ntests2; ++cnt) { - char *cp = sha256_crypt(tests2[cnt].input, tests2[cnt].salt); + char *cp = crypt_sha256(tests2[cnt].input, tests2[cnt].salt); if (strcmp(cp, tests2[cnt].expected) != 0) { printf("test %d: expected \"%s\", got \"%s\"\n", cnt, tests2[cnt].expected, cp); result = 1; } } if (result == 0) puts("all tests OK"); return result; } #endif /* TEST */ Index: projects/largeSMP/lib/libcrypt/crypt-sha512.c =================================================================== --- projects/largeSMP/lib/libcrypt/crypt-sha512.c (revision 221494) +++ projects/largeSMP/lib/libcrypt/crypt-sha512.c (revision 221495) @@ -1,500 +1,500 @@ /* * Copyright (c) 2011 The FreeBSD Project. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ /* Based on: * SHA512-based Unix crypt implementation. Released into the Public Domain by * Ulrich Drepper . */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include "crypt.h" /* Define our magic string to mark salt for SHA512 "encryption" replacement. */ static const char sha512_salt_prefix[] = "$6$"; /* Prefix for optional rounds specification. */ static const char sha512_rounds_prefix[] = "rounds="; /* Maximum salt string length. */ #define SALT_LEN_MAX 16 /* Default number of rounds if not explicitly specified. */ #define ROUNDS_DEFAULT 5000 /* Minimum number of rounds. */ #define ROUNDS_MIN 1000 /* Maximum number of rounds. */ #define ROUNDS_MAX 999999999 static char * -sha512_crypt_r(const char *key, const char *salt, char *buffer, int buflen) +crypt_sha512_r(const char *key, const char *salt, char *buffer, int buflen) { u_long srounds; int n; uint8_t alt_result[64], temp_result[64]; SHA512_CTX ctx, alt_ctx; size_t salt_len, key_len, cnt, rounds; char *cp, *copied_key, *copied_salt, *p_bytes, *s_bytes, *endp; const char *num; bool rounds_custom; copied_key = NULL; copied_salt = NULL; /* Default number of rounds. */ rounds = ROUNDS_DEFAULT; rounds_custom = false; /* Find beginning of salt string. The prefix should normally always * be present. Just in case it is not. */ if (strncmp(sha512_salt_prefix, salt, sizeof(sha512_salt_prefix) - 1) == 0) /* Skip salt prefix. */ salt += sizeof(sha512_salt_prefix) - 1; if (strncmp(salt, sha512_rounds_prefix, sizeof(sha512_rounds_prefix) - 1) == 0) { num = salt + sizeof(sha512_rounds_prefix) - 1; srounds = strtoul(num, &endp, 10); if (*endp == '$') { salt = endp + 1; rounds = MAX(ROUNDS_MIN, MIN(srounds, ROUNDS_MAX)); rounds_custom = true; } } salt_len = MIN(strcspn(salt, "$"), SALT_LEN_MAX); key_len = strlen(key); /* Prepare for the real work. */ SHA512_Init(&ctx); /* Add the key string. */ SHA512_Update(&ctx, key, key_len); /* The last part is the salt string. This must be at most 8 * characters and it ends at the first `$' character (for * compatibility with existing implementations). */ SHA512_Update(&ctx, salt, salt_len); /* Compute alternate SHA512 sum with input KEY, SALT, and KEY. The * final result will be added to the first context. */ SHA512_Init(&alt_ctx); /* Add key. */ SHA512_Update(&alt_ctx, key, key_len); /* Add salt. */ SHA512_Update(&alt_ctx, salt, salt_len); /* Add key again. */ SHA512_Update(&alt_ctx, key, key_len); /* Now get result of this (64 bytes) and add it to the other context. */ SHA512_Final(alt_result, &alt_ctx); /* Add for any character in the key one byte of the alternate sum. */ for (cnt = key_len; cnt > 64; cnt -= 64) SHA512_Update(&ctx, alt_result, 64); SHA512_Update(&ctx, alt_result, cnt); /* Take the binary representation of the length of the key and for * every 1 add the alternate sum, for every 0 the key. */ for (cnt = key_len; cnt > 0; cnt >>= 1) if ((cnt & 1) != 0) SHA512_Update(&ctx, alt_result, 64); else SHA512_Update(&ctx, key, key_len); /* Create intermediate result. */ SHA512_Final(alt_result, &ctx); /* Start computation of P byte sequence. */ SHA512_Init(&alt_ctx); /* For every character in the password add the entire password. */ for (cnt = 0; cnt < key_len; ++cnt) SHA512_Update(&alt_ctx, key, key_len); /* Finish the digest. */ SHA512_Final(temp_result, &alt_ctx); /* Create byte sequence P. */ cp = p_bytes = alloca(key_len); for (cnt = key_len; cnt >= 64; cnt -= 64) { memcpy(cp, temp_result, 64); cp += 64; } memcpy(cp, temp_result, cnt); /* Start computation of S byte sequence. */ SHA512_Init(&alt_ctx); /* For every character in the password add the entire password. */ for (cnt = 0; cnt < 16 + alt_result[0]; ++cnt) SHA512_Update(&alt_ctx, salt, salt_len); /* Finish the digest. */ SHA512_Final(temp_result, &alt_ctx); /* Create byte sequence S. */ cp = s_bytes = alloca(salt_len); for (cnt = salt_len; cnt >= 64; cnt -= 64) { memcpy(cp, temp_result, 64); cp += 64; } memcpy(cp, temp_result, cnt); /* Repeatedly run the collected hash value through SHA512 to burn CPU * cycles. */ for (cnt = 0; cnt < rounds; ++cnt) { /* New context. */ SHA512_Init(&ctx); /* Add key or last result. */ if ((cnt & 1) != 0) SHA512_Update(&ctx, p_bytes, key_len); else SHA512_Update(&ctx, alt_result, 64); /* Add salt for numbers not divisible by 3. */ if (cnt % 3 != 0) SHA512_Update(&ctx, s_bytes, salt_len); /* Add key for numbers not divisible by 7. */ if (cnt % 7 != 0) SHA512_Update(&ctx, p_bytes, key_len); /* Add key or last result. */ if ((cnt & 1) != 0) SHA512_Update(&ctx, alt_result, 64); else SHA512_Update(&ctx, p_bytes, key_len); /* Create intermediate result. */ SHA512_Final(alt_result, &ctx); } /* Now we can construct the result string. It consists of three * parts. */ cp = stpncpy(buffer, sha512_salt_prefix, MAX(0, buflen)); buflen -= sizeof(sha512_salt_prefix) - 1; if (rounds_custom) { n = snprintf(cp, MAX(0, buflen), "%s%zu$", sha512_rounds_prefix, rounds); cp += n; buflen -= n; } cp = stpncpy(cp, salt, MIN((size_t)MAX(0, buflen), salt_len)); buflen -= MIN((size_t)MAX(0, buflen), salt_len); if (buflen > 0) { *cp++ = '$'; --buflen; } b64_from_24bit(alt_result[0], alt_result[21], alt_result[42], 4, &buflen, &cp); b64_from_24bit(alt_result[22], alt_result[43], alt_result[1], 4, &buflen, &cp); b64_from_24bit(alt_result[44], alt_result[2], alt_result[23], 4, &buflen, &cp); b64_from_24bit(alt_result[3], alt_result[24], alt_result[45], 4, &buflen, &cp); b64_from_24bit(alt_result[25], alt_result[46], alt_result[4], 4, &buflen, &cp); b64_from_24bit(alt_result[47], alt_result[5], alt_result[26], 4, &buflen, &cp); b64_from_24bit(alt_result[6], alt_result[27], alt_result[48], 4, &buflen, &cp); b64_from_24bit(alt_result[28], alt_result[49], alt_result[7], 4, &buflen, &cp); b64_from_24bit(alt_result[50], alt_result[8], alt_result[29], 4, &buflen, &cp); b64_from_24bit(alt_result[9], alt_result[30], alt_result[51], 4, &buflen, &cp); b64_from_24bit(alt_result[31], alt_result[52], alt_result[10], 4, &buflen, &cp); b64_from_24bit(alt_result[53], alt_result[11], alt_result[32], 4, &buflen, &cp); b64_from_24bit(alt_result[12], alt_result[33], alt_result[54], 4, &buflen, &cp); b64_from_24bit(alt_result[34], alt_result[55], alt_result[13], 4, &buflen, &cp); b64_from_24bit(alt_result[56], alt_result[14], alt_result[35], 4, &buflen, &cp); b64_from_24bit(alt_result[15], alt_result[36], alt_result[57], 4, &buflen, &cp); b64_from_24bit(alt_result[37], alt_result[58], alt_result[16], 4, &buflen, &cp); b64_from_24bit(alt_result[59], alt_result[17], alt_result[38], 4, &buflen, &cp); b64_from_24bit(alt_result[18], alt_result[39], alt_result[60], 4, &buflen, &cp); b64_from_24bit(alt_result[40], alt_result[61], alt_result[19], 4, &buflen, &cp); b64_from_24bit(alt_result[62], alt_result[20], alt_result[41], 4, &buflen, &cp); b64_from_24bit(0, 0, alt_result[63], 2, &buflen, &cp); if (buflen <= 0) { errno = ERANGE; buffer = NULL; } else *cp = '\0'; /* Terminate the string. */ /* Clear the buffer for the intermediate result so that people * attaching to processes or reading core dumps cannot get any * information. We do it in this way to clear correct_words[] inside * the SHA512 implementation as well. */ SHA512_Init(&ctx); SHA512_Final(alt_result, &ctx); memset(temp_result, '\0', sizeof(temp_result)); memset(p_bytes, '\0', key_len); memset(s_bytes, '\0', salt_len); memset(&ctx, '\0', sizeof(ctx)); memset(&alt_ctx, '\0', sizeof(alt_ctx)); if (copied_key != NULL) memset(copied_key, '\0', key_len); if (copied_salt != NULL) memset(copied_salt, '\0', salt_len); return buffer; } /* This entry point is equivalent to crypt(3). */ char * -sha512_crypt(const char *key, const char *salt) +crypt_sha512(const char *key, const char *salt) { /* We don't want to have an arbitrary limit in the size of the * password. We can compute an upper bound for the size of the * result in advance and so we can prepare the buffer we pass to - * `sha512_crypt_r'. */ + * `crypt_sha512_r'. */ static char *buffer; static int buflen; int needed; char *new_buffer; needed = (sizeof(sha512_salt_prefix) - 1 + sizeof(sha512_rounds_prefix) + 9 + 1 + strlen(salt) + 1 + 86 + 1); if (buflen < needed) { new_buffer = (char *)realloc(buffer, needed); if (new_buffer == NULL) return NULL; buffer = new_buffer; buflen = needed; } - return sha512_crypt_r(key, salt, buffer, buflen); + return crypt_sha512_r(key, salt, buffer, buflen); } #ifdef TEST static const struct { const char *input; const char result[64]; } tests[] = { /* Test vectors from FIPS 180-2: appendix C.1. */ { "abc", "\xdd\xaf\x35\xa1\x93\x61\x7a\xba\xcc\x41\x73\x49\xae\x20\x41\x31" "\x12\xe6\xfa\x4e\x89\xa9\x7e\xa2\x0a\x9e\xee\xe6\x4b\x55\xd3\x9a" "\x21\x92\x99\x2a\x27\x4f\xc1\xa8\x36\xba\x3c\x23\xa3\xfe\xeb\xbd" "\x45\x4d\x44\x23\x64\x3c\xe8\x0e\x2a\x9a\xc9\x4f\xa5\x4c\xa4\x9f" }, /* Test vectors from FIPS 180-2: appendix C.2. */ { "abcdefghbcdefghicdefghijdefghijkefghijklfghijklmghijklmn" "hijklmnoijklmnopjklmnopqklmnopqrlmnopqrsmnopqrstnopqrstu", "\x8e\x95\x9b\x75\xda\xe3\x13\xda\x8c\xf4\xf7\x28\x14\xfc\x14\x3f" "\x8f\x77\x79\xc6\xeb\x9f\x7f\xa1\x72\x99\xae\xad\xb6\x88\x90\x18" "\x50\x1d\x28\x9e\x49\x00\xf7\xe4\x33\x1b\x99\xde\xc4\xb5\x43\x3a" "\xc7\xd3\x29\xee\xb6\xdd\x26\x54\x5e\x96\xe5\x5b\x87\x4b\xe9\x09" }, /* Test vectors from the NESSIE project. */ { "", "\xcf\x83\xe1\x35\x7e\xef\xb8\xbd\xf1\x54\x28\x50\xd6\x6d\x80\x07" "\xd6\x20\xe4\x05\x0b\x57\x15\xdc\x83\xf4\xa9\x21\xd3\x6c\xe9\xce" "\x47\xd0\xd1\x3c\x5d\x85\xf2\xb0\xff\x83\x18\xd2\x87\x7e\xec\x2f" "\x63\xb9\x31\xbd\x47\x41\x7a\x81\xa5\x38\x32\x7a\xf9\x27\xda\x3e" }, { "a", "\x1f\x40\xfc\x92\xda\x24\x16\x94\x75\x09\x79\xee\x6c\xf5\x82\xf2" "\xd5\xd7\xd2\x8e\x18\x33\x5d\xe0\x5a\xbc\x54\xd0\x56\x0e\x0f\x53" "\x02\x86\x0c\x65\x2b\xf0\x8d\x56\x02\x52\xaa\x5e\x74\x21\x05\x46" "\xf3\x69\xfb\xbb\xce\x8c\x12\xcf\xc7\x95\x7b\x26\x52\xfe\x9a\x75" }, { "message digest", "\x10\x7d\xbf\x38\x9d\x9e\x9f\x71\xa3\xa9\x5f\x6c\x05\x5b\x92\x51" "\xbc\x52\x68\xc2\xbe\x16\xd6\xc1\x34\x92\xea\x45\xb0\x19\x9f\x33" "\x09\xe1\x64\x55\xab\x1e\x96\x11\x8e\x8a\x90\x5d\x55\x97\xb7\x20" "\x38\xdd\xb3\x72\xa8\x98\x26\x04\x6d\xe6\x66\x87\xbb\x42\x0e\x7c" }, { "abcdefghijklmnopqrstuvwxyz", "\x4d\xbf\xf8\x6c\xc2\xca\x1b\xae\x1e\x16\x46\x8a\x05\xcb\x98\x81" "\xc9\x7f\x17\x53\xbc\xe3\x61\x90\x34\x89\x8f\xaa\x1a\xab\xe4\x29" "\x95\x5a\x1b\xf8\xec\x48\x3d\x74\x21\xfe\x3c\x16\x46\x61\x3a\x59" "\xed\x54\x41\xfb\x0f\x32\x13\x89\xf7\x7f\x48\xa8\x79\xc7\xb1\xf1" }, { "abcdbcdecdefdefgefghfghighijhijkijkljklmklmnlmnomnopnopq", "\x20\x4a\x8f\xc6\xdd\xa8\x2f\x0a\x0c\xed\x7b\xeb\x8e\x08\xa4\x16" "\x57\xc1\x6e\xf4\x68\xb2\x28\xa8\x27\x9b\xe3\x31\xa7\x03\xc3\x35" "\x96\xfd\x15\xc1\x3b\x1b\x07\xf9\xaa\x1d\x3b\xea\x57\x78\x9c\xa0" "\x31\xad\x85\xc7\xa7\x1d\xd7\x03\x54\xec\x63\x12\x38\xca\x34\x45" }, { "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789", "\x1e\x07\xbe\x23\xc2\x6a\x86\xea\x37\xea\x81\x0c\x8e\xc7\x80\x93" "\x52\x51\x5a\x97\x0e\x92\x53\xc2\x6f\x53\x6c\xfc\x7a\x99\x96\xc4" "\x5c\x83\x70\x58\x3e\x0a\x78\xfa\x4a\x90\x04\x1d\x71\xa4\xce\xab" "\x74\x23\xf1\x9c\x71\xb9\xd5\xa3\xe0\x12\x49\xf0\xbe\xbd\x58\x94" }, { "123456789012345678901234567890123456789012345678901234567890" "12345678901234567890", "\x72\xec\x1e\xf1\x12\x4a\x45\xb0\x47\xe8\xb7\xc7\x5a\x93\x21\x95" "\x13\x5b\xb6\x1d\xe2\x4e\xc0\xd1\x91\x40\x42\x24\x6e\x0a\xec\x3a" "\x23\x54\xe0\x93\xd7\x6f\x30\x48\xb4\x56\x76\x43\x46\x90\x0c\xb1" "\x30\xd2\xa4\xfd\x5d\xd1\x6a\xbb\x5e\x30\xbc\xb8\x50\xde\xe8\x43" } }; #define ntests (sizeof (tests) / sizeof (tests[0])) static const struct { const char *salt; const char *input; const char *expected; } tests2[] = { { "$6$saltstring", "Hello world!", "$6$saltstring$svn8UoSVapNtMuq1ukKS4tPQd8iKwSMHWjl/O817G3uBnIFNjnQJu" "esI68u4OTLiBFdcbYEdFCoEOfaS35inz1" }, { "$6$rounds=10000$saltstringsaltstring", "Hello world!", "$6$rounds=10000$saltstringsaltst$OW1/O6BYHV6BcXZu8QVeXbDWra3Oeqh0sb" "HbbMCVNSnCM/UrjmM0Dp8vOuZeHBy/YTBmSK6H9qs/y3RnOaw5v." }, { "$6$rounds=5000$toolongsaltstring", "This is just a test", "$6$rounds=5000$toolongsaltstrin$lQ8jolhgVRVhY4b5pZKaysCLi0QBxGoNeKQ" "zQ3glMhwllF7oGDZxUhx1yxdYcz/e1JSbq3y6JMxxl8audkUEm0" }, { "$6$rounds=1400$anotherlongsaltstring", "a very much longer text to encrypt. This one even stretches over more" "than one line.", "$6$rounds=1400$anotherlongsalts$POfYwTEok97VWcjxIiSOjiykti.o/pQs.wP" "vMxQ6Fm7I6IoYN3CmLs66x9t0oSwbtEW7o7UmJEiDwGqd8p4ur1" }, { "$6$rounds=77777$short", "we have a short salt string but not a short password", "$6$rounds=77777$short$WuQyW2YR.hBNpjjRhpYD/ifIw05xdfeEyQoMxIXbkvr0g" "ge1a1x3yRULJ5CCaUeOxFmtlcGZelFl5CxtgfiAc0" }, { "$6$rounds=123456$asaltof16chars..", "a short string", "$6$rounds=123456$asaltof16chars..$BtCwjqMJGx5hrJhZywWvt0RLE8uZ4oPwc" "elCjmw2kSYu.Ec6ycULevoBK25fs2xXgMNrCzIMVcgEJAstJeonj1" }, { "$6$rounds=10$roundstoolow", "the minimum number is still observed", "$6$rounds=1000$roundstoolow$kUMsbe306n21p9R.FRkW3IGn.S9NPN0x50YhH1x" "hLsPuWGsUSklZt58jaTfF4ZEQpyUNGc0dqbpBYYBaHHrsX." }, }; #define ntests2 (sizeof (tests2) / sizeof (tests2[0])) int main(void) { SHA512_CTX ctx; uint8_t sum[64]; int result = 0; int i, cnt; for (cnt = 0; cnt < (int)ntests; ++cnt) { SHA512_Init(&ctx); SHA512_Update(&ctx, tests[cnt].input, strlen(tests[cnt].input)); SHA512_Final(sum, &ctx); if (memcmp(tests[cnt].result, sum, 64) != 0) { printf("test %d run %d failed\n", cnt, 1); result = 1; } SHA512_Init(&ctx); for (i = 0; tests[cnt].input[i] != '\0'; ++i) SHA512_Update(&ctx, &tests[cnt].input[i], 1); SHA512_Final(sum, &ctx); if (memcmp(tests[cnt].result, sum, 64) != 0) { printf("test %d run %d failed\n", cnt, 2); result = 1; } } /* Test vector from FIPS 180-2: appendix C.3. */ char buf[1000]; memset(buf, 'a', sizeof(buf)); SHA512_Init(&ctx); for (i = 0; i < 1000; ++i) SHA512_Update(&ctx, buf, sizeof(buf)); SHA512_Final(sum, &ctx); static const char expected[64] = "\xe7\x18\x48\x3d\x0c\xe7\x69\x64\x4e\x2e\x42\xc7\xbc\x15\xb4\x63" "\x8e\x1f\x98\xb1\x3b\x20\x44\x28\x56\x32\xa8\x03\xaf\xa9\x73\xeb" "\xde\x0f\xf2\x44\x87\x7e\xa6\x0a\x4c\xb0\x43\x2c\xe5\x77\xc3\x1b" "\xeb\x00\x9c\x5c\x2c\x49\xaa\x2e\x4e\xad\xb2\x17\xad\x8c\xc0\x9b"; if (memcmp(expected, sum, 64) != 0) { printf("test %d failed\n", cnt); result = 1; } for (cnt = 0; cnt < ntests2; ++cnt) { - char *cp = sha512_crypt(tests2[cnt].input, tests2[cnt].salt); + char *cp = crypt_sha512(tests2[cnt].input, tests2[cnt].salt); if (strcmp(cp, tests2[cnt].expected) != 0) { printf("test %d: expected \"%s\", got \"%s\"\n", cnt, tests2[cnt].expected, cp); result = 1; } } if (result == 0) puts("all tests OK"); return result; } #endif /* TEST */ Index: projects/largeSMP/lib/libcrypt/crypt.c =================================================================== --- projects/largeSMP/lib/libcrypt/crypt.c (revision 221494) +++ projects/largeSMP/lib/libcrypt/crypt.c (revision 221495) @@ -1,141 +1,141 @@ /* * Copyright (c) 1999 * Mark Murray. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY MARK MURRAY AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL MARK MURRAY OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include "crypt.h" static const struct { const char *const name; char *(*const func)(const char *, const char *); const char *const magic; } crypt_types[] = { #ifdef HAS_DES { "des", crypt_des, NULL }, #endif { "md5", crypt_md5, "$1$" }, #ifdef HAS_BLOWFISH { "blf", crypt_blowfish, "$2" }, #endif { "nth", crypt_nthash, "$3$" }, { "sha256", - sha256_crypt, + crypt_sha256, "$5$" }, { "sha512", - sha512_crypt, + crypt_sha512, "$6$" }, { NULL, NULL, NULL } }; static int crypt_type = -1; static void crypt_setdefault(void) { char *def; size_t i; if (crypt_type != -1) return; def = auth_getval("crypt_default"); if (def == NULL) { crypt_type = 0; return; } for (i = 0; i < sizeof(crypt_types) / sizeof(crypt_types[0]) - 1; i++) { if (strcmp(def, crypt_types[i].name) == 0) { crypt_type = (int)i; return; } } crypt_type = 0; } const char * crypt_get_format(void) { crypt_setdefault(); return (crypt_types[crypt_type].name); } int crypt_set_format(const char *type) { size_t i; crypt_setdefault(); for (i = 0; i < sizeof(crypt_types) / sizeof(crypt_types[0]) - 1; i++) { if (strcmp(type, crypt_types[i].name) == 0) { crypt_type = (int)i; return (1); } } return (0); } char * crypt(const char *passwd, const char *salt) { size_t i; crypt_setdefault(); for (i = 0; i < sizeof(crypt_types) / sizeof(crypt_types[0]) - 1; i++) { if (crypt_types[i].magic != NULL && strncmp(salt, crypt_types[i].magic, strlen(crypt_types[i].magic)) == 0) return (crypt_types[i].func(passwd, salt)); } return (crypt_types[crypt_type].func(passwd, salt)); } Index: projects/largeSMP/lib/libcrypt/crypt.h =================================================================== --- projects/largeSMP/lib/libcrypt/crypt.h (revision 221494) +++ projects/largeSMP/lib/libcrypt/crypt.h (revision 221495) @@ -1,43 +1,43 @@ /* LINTLIBRARY */ /* * Copyright (c) 1999 * Mark Murray. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY MARK MURRAY AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL MARK MURRAY OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ * */ /* magic sizes */ #define MD4_SIZE 16 #define MD5_SIZE 16 char *crypt_des(const char *pw, const char *salt); char *crypt_md5(const char *pw, const char *salt); char *crypt_nthash(const char *pw, const char *salt); char *crypt_blowfish(const char *pw, const char *salt); -char *sha256_crypt (const char *pw, const char *salt); -char *sha512_crypt (const char *pw, const char *salt); +char *crypt_sha256 (const char *pw, const char *salt); +char *crypt_sha512 (const char *pw, const char *salt); extern void _crypt_to64(char *s, u_long v, int n); extern void b64_from_24bit(uint8_t B2, uint8_t B1, uint8_t B0, int n, int *buflen, char **cp); Index: projects/largeSMP/lib/libutil =================================================================== --- projects/largeSMP/lib/libutil (revision 221494) +++ projects/largeSMP/lib/libutil (revision 221495) Property changes on: projects/largeSMP/lib/libutil ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/lib/libutil:r221444-221493 Index: projects/largeSMP/lib/libz =================================================================== --- projects/largeSMP/lib/libz (revision 221494) +++ projects/largeSMP/lib/libz (revision 221495) Property changes on: projects/largeSMP/lib/libz ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/lib/libz:r221444-221493 Index: projects/largeSMP/release/Makefile.sysinstall =================================================================== --- projects/largeSMP/release/Makefile.sysinstall (revision 221494) +++ projects/largeSMP/release/Makefile.sysinstall (revision 221495) @@ -1,1442 +1,1442 @@ # $FreeBSD$ # # make release [BUILDNAME=somename] CHROOTDIR=/some/dir CVSROOT=/cvs/dir \ # [RELEASETAG=tag] [SVNROOT=svn://svn.freebsd.org/base] \ # [SVNBRANCH=some/branch] # # Where "/some/dir" is the pathname of a directory on a some filesystem with # at least 1000MB of free space, "somename" is what you want the release to # call itself, "/cvs/dir" is where our source repo resides and, optionally, # which CVS "tag" name should be used when checking out the sources to build # the release (default is HEAD). # # Please note the support for building from SVN is preliminary and there # are still questions about things like how to handle updates of # /usr/src on production systems (csup(1) replacement). It is a work # in progress and may change as the other issues get worked out. # # Please note: the md(4) driver must be present in the kernel # (either by being compiled in or available as a kld(4) module), # otherwise the target 'release.8' and possibly others will fail. # # Note: If you add options to this file, please keep release(7) updated! # # Set these, release builder! # # Fixed version: #BUILDNAME=9.0-CURRENT # # Automatic SNAP versioning: DATE != date +%Y%m%d BASE = 9.0 BUILDNAME?=${BASE}-${DATE}-SNAP # #CHROOTDIR=/junk/release # If this is a -stable snapshot, then set #RELEASETAG=RELENG_8 # # To test a release with a source tree containing patches and # other work. This tree will get copied instead of getting the # src/ tree from a CVS checkout. For "rerelease", this will NOT # be copied; cvs update will be used instead. #EXTSRCDIR=/usr/src # # To use a checked-out ports collection directory instead of # checking out from a local CVS repository, set this option. #EXTPORTSDIR=/usr/ports # # To use a checked-out doc/ directory instead of # checking out from a local CVS repository, set this option. #EXTDOCDIR=/usr/doc # # To add other options to the CVS subcommands (co,up), set #CVSCMDARGS="-D '01/01/2002 00:00:00 UTC'" # # To add other options to the CVS command, set #CVSARGS="-lfq" # # To prefix the CVS command #CVSPREFIX="/usr/bin/time" # # Where the CVS repository is #CVSROOT="/home/ncvs" # # To add other options to the Subversion subcommands (co,up), set #SVNCMDARGS="-r '{ 01/01/2002 00:00:00 UTC }'" # # To prefix the Subversion command #SVNPREFIX="/usr/bin/time" # # Where the Subversion repository is #SVNROOT=svn://svn.freebsd.org/base # # Subversion branch to build for src. If this is not set then it is # automatically computed from RELEASETAG. #SVNBRANCH=stable/8 # # Non-zero if ${RELEASETAG} is in the form "RELENG_ver_RELEASE"; we # are building an official release. Otherwise, we are building for # a branch. .if defined(RELEASETAG) ISRELEASE!= expr "${RELEASETAG}" : '^RELENG_.*_RELEASE$$' || true .if ${ISRELEASE} != 0 # Convert "RELENG_ver_RELEASE" to "RELEASE_ver" for ports and doc trees. AUXRELEASETAG!= echo ${RELEASETAG} | sed -e 's/^RELENG_/RELEASE_/' -e 's/_RELEASE$$//' DOCRELEASETAG?= ${AUXRELEASETAG} PORTSRELEASETAG?= ${AUXRELEASETAG} .endif .endif # Determine the Subversion source branch that corresponds to the requested # RELEASETAG. .if !defined(SVNBRANCH) .if defined(RELEASETAG) SVNBRANCH!= echo "${RELEASETAG}" | awk -f ${.CURDIR}/svnbranch.awk .else SVNBRANCH= head .endif .endif # If you want to pass flags to the world build such as -j X, use # WORLD_FLAGS. Similarly, you can specify make flags for kernel # builds via KERNEL_FLAGS. #WORLD_FLAGS=-j4 #KERNEL_FLAGS=-j4 TARGET_ARCH?= ${MACHINE_ARCH} .if ${TARGET_ARCH} == ${MACHINE_ARCH} TARGET?= ${MACHINE} TARGET_CPUARCH?= ${MACHINE_CPUARCH} .else TARGET?= ${TARGET_ARCH} TARGET_CPUARCH?= ${TARGET_ARCH} .endif CROSSENV= TARGET_ARCH=${TARGET_ARCH} TARGET=${TARGET} CROSSMAKE= ${MAKE} ${CROSSENV} NATIVEMAKE= ${MAKE} TARGET_ARCH=${MACHINE_ARCH} TARGET=${MACHINE} # If you are using a local CVS repository with components stored in # non-standard modules, override these on the make commandline or # in the environment. RELEASESRCMODULE?= src RELEASEDOCMODULE?= doc RELEASEPORTSMODULE?= ports # Uncomment this to disable the doc.1 target. Docs normally require # the ports tree, so NOPORTS can be set together with NODOC in order # to have neither ports or docs. If only NOPORTS is set to YES, but # docs are still desired, the DOMINIMALDOCPORTS logic below will only # install the ports that are minimally required for the release note documentation. This is # intended as a compromise, less disk space is required than for using # the entire ports collection (and much less time due to the huge number # of directories it would create), but still quite a bit as well as some # CPU cycles (some of the programs are C++, and things like ghostscript # belong to the required ports nevertheless). # #NODOC= YES #NOPORTS= YES # When retrieving ports using the pkg_add -r method, set your proxies to these. # src/release/${arch}/mkisoimages.sh can use these. #HTTP_PROXY?= www.example.com #FTP_PROXY?= ftp.example.com # When creating ISO images, point ${CD_PACKAGE_TREE} to a directory containing # the package split by an earlier invocation of the 'package-split' target. #CD_PACKAGE_TREE= /path/to/pkg # Extra source tarballs; each argument is a pair of source dir and # distribution name. The dist name should not exceed 7 characters # (another "s" for "source" will be prepended). #EXTRA_SRC= games/fortune fortune # Modify this definition if you want the release notes # and other release documentation in a language other than English. RELNOTES_LANG?= en_US.ISO8859-1 # As an alternative to installing the entire ports collection (which # can take a huge amount of time, in particular on slower disks), # setting ${MINIMALDOCPORTS} allows to install and build just those # ports that are really required for getting the docs up & running. .if defined(NOPORTS) && !defined(NODOC) DOMINIMALDOCPORTS= YES .include "Makefile.inc.docports" RELEASEPORTSMODULE= ${MINIMALDOCPORTS} ports/sysutils/cdrtools .endif # Helper variable .if defined(NOPORTS) .if !defined(DOMINIMALDOCPORTS) || ${DOMINIMALDOCPORTS} != "YES" NOPORTSATALL= YES .endif .endif # # Doing 'make index' in /usr/ports requires Perl. MAKEINDEXPORTS= lang/perl5.10 DOCPORTS= textproc/docproj # Set this to wherever the distfiles required by release procedures. .if defined(DOCDISTFILES) # Respect DOCDISTFILES which is used before. RELEASEDISTFILES?= ${DOCDISTFILES} .else RELEASEDISTFILES?= ${.CURDIR}/../../ports/distfiles .endif # Set this to 1 if you want -P to be used for automatic keyboard detection # on the boot floppy. WARNING: Breaks on some Athlon (K7) motherboards. AUTO_KEYBOARD_DETECT?= 0 .if !defined(NODOC) DIST_DOCS_ARCH_INDEP= hardware readme relnotes errata DIST_DOCS_ARCH_DEP= .endif # Things which without too much trouble can be considered variables # BASE_DISTS are special in that they get full /etc installation sets. # OTHER_DISTS?= catpages manpages proflibs dict info BASE_DISTS?= base doc games ${ARCH_DISTS} .if ${TARGET_ARCH} == "amd64" || ${TARGET_ARCH} == "powerpc64" ARCH_DISTS?= lib32 .endif DISTRIBUTIONS?= ${BASE_DISTS} ${OTHER_DISTS} # # Build and package both GENERIC and SMP kernels if the target # has both configuration files. Otherwise only GENERIC is done. # .if ${TARGET_ARCH} == "powerpc64" KERN_GENERIC?= GENERIC64 .else KERN_GENERIC?= GENERIC .endif .if exists(${.CURDIR}/../sys/${TARGET}/conf/SMP) KERNELS_BASE?= ${KERN_GENERIC} SMP .else KERNELS_BASE?= ${KERN_GENERIC} .endif # mountpoint for filesystems. MNT= /mnt # Various floppy image parameters. # .undef MAKE_FLOPPIES .if ${TARGET_ARCH} == "i386" MAKE_DVD= SEPARATE_LIVEFS= SPLIT_MFSROOT= .if ${TARGET} == "pc98" MAKE_FLOPPIES= true SMALLFLOPPYSIZE= 1200 SMALLFLOPPYSPLITSIZE= 1152 SMALLFLOPPYLABEL= fd1200 SPLIT_MFSROOT= .endif FLOPPYSIZE= 1440 FLOPPYSPLITSIZE= 1392 FLOPPYINODE= 40000 FLOPPYLABEL= fd1440 BOOTINODE= 80000 MFSSIZE= 4320 MFSINODE= 8000 MFSLABEL= minimum3 .elif ${TARGET_ARCH} == "sparc64" DISKLABEL= sunlabel MAKE_DVD= MFSSIZE= 4096 MFSINODE= 8192 MFSLABEL= auto MINIROOT= SEPARATE_LIVEFS= .elif ${TARGET_ARCH} == "ia64" DISKLABEL= "" MAKE_DVD= MFSSIZE= 8192 MFSINODE= 8192 MFSLABEL= auto SEPARATE_LIVEFS= .elif ${TARGET_ARCH} == "amd64" MAKE_DVD= FLOPPYSIZE= 1440 FLOPPYSPLITSIZE= 1392 FLOPPYINODE= 40000 FLOPPYLABEL= fd1440 BOOTINODE= 80000 MFSSIZE= 4096 MFSINODE= 8192 MFSLABEL= auto SEPARATE_LIVEFS= SPLIT_MFSROOT= .elif ${TARGET_ARCH} == "powerpc" DISKLABEL= "" MFSSIZE= 4096 MFSINODE= 8192 MFSLABEL= auto SEPARATE_LIVEFS= .elif ${TARGET_ARCH} == "powerpc64" DISKLABEL= "" MFSSIZE= 8192 MFSINODE= 8192 MFSLABEL= auto SEPARATE_LIVEFS= .endif .if defined(NO_FLOPPIES) .undef MAKE_FLOPPIES .endif .if exists(/sbin/bsdlabel) DISKLABEL?= bsdlabel .else DISKLABEL?= disklabel .endif ZIPPER= gzip -9 --no-name ZIPNSPLIT= ${ZIPPER} -c | split -b 1392k - # Things which may get you into trouble if you change them MTREEFILES= ${.CURDIR}/../etc/mtree _R?= /R RD= ${_R}/stage RND= ${RD}/release.doc FD= ${_R}/ftp CD= ${_R}/cdrom CD_BOOT= ${CD}/bootonly CD_DISC1= ${CD}/disc1 CD_DISC2= ${CD}/disc2 .if defined(MAKE_DVD) CD_DVD1= ${CD}/dvd1 .endif .if !defined(NODOC) CD_DOCS= ${CD}/docs .endif .if defined(SEPARATE_LIVEFS) CD_LIVEFS= ${CD}/livefs .else CD_LIVEFS= ${CD_DISC1} .endif _MK?= ${CHROOTDIR}/mk # Where the bootstrap ports (see DOCPORTS) get installed. LOCALDIR= /usr/local/bin .if ${TARGET} != ${MACHINE} && ${DISKLABEL} == "bsdlabel" DOFS_SH= ${.CURDIR}/scripts/doFS.sh ${DISKLABEL} ${TARGET} .else DOFS_SH= ${.CURDIR}/scripts/doFS.sh ${DISKLABEL} "" .endif CRUNCH_TARGETS= boot .if ${TARGET_ARCH} == "i386" CRUNCH_TARGETS+=fixit .if ${TARGET} == "pc98" CRUNCH_TARGETS+=fixit-small .endif .endif .if defined(MAKE_FLOPPIES) EXTRAS= floppies.1 .if ${TARGET_ARCH} == "i386" EXTRAS+= floppies.2 .endif EXTRAS+= floppies.3 .endif EXTRAS+= ftp.1 .if !defined(NOCDROM) EXTRAS+= cdrom.1 cdrom.2 cdrom.3 .if defined(MAKE_ISOS) EXTRAS+= iso.1 .endif .if ${TARGET} == "pc98" BOOTABLE="-G" .else BOOTABLE="-b" .endif .endif .if !defined(NODOC) DOCREL= doc.1 doc.2 .endif .if !make(release) && !make(rerelease) && !make(package-split) BINMAKE!= cd ${.CURDIR}/..; ${MAKE} -V BINMAKE WMAKEENV!= cd ${.CURDIR}/..; \ ${BINMAKE} ${CROSSENV} -f Makefile.inc1 -V WMAKEENV WMAKE= ${WMAKEENV} ${BINMAKE} .endif CVS_SRCARGS= -P .if defined(RELEASETAG) CVS_SRCARGS+= -r ${RELEASETAG} .endif CVS_DOCARGS= -P .if defined(DOCRELEASETAG) CVS_DOCARGS+= -r ${DOCRELEASETAG} .endif CVS_PORTSARGS= -P .if defined(PORTSRELEASETAG) CVS_PORTSARGS+= -r ${PORTSRELEASETAG} .endif WORLDDIR?= ${.CURDIR}/.. release rerelease: .if !defined(CHROOTDIR) || !defined(BUILDNAME) @echo "To make a release you must set CHROOTDIR and BUILDNAME" && false .endif .if !defined(NOPORTSATALL) && !defined(EXTPORTSDIR) && !defined(CVSROOT) @echo "Building ports requires CVSROOT or EXTPORTSDIR" && false .endif .if !defined(NODOC) && !defined(EXTDOCDIR) && !defined(CVSROOT) @echo "Building docs requires CVSROOT or EXTDOCDIR" && false .endif .if !defined(EXTSRCDIR) && !defined(CVSROOT) && !defined(SVNROOT) @echo "The source tree requires SVNROOT, CVSROOT, or EXTSRCDIR" && false .endif .if defined(NOPORTSATALL) && !defined(NODOC) @echo "Ports are required for building the release docs. Either set NODOC or" @echo "unset NOPORTS, or set at least DOMINIMALDOCPORTS to YES!" @exit 1 .endif .if defined(LOCAL_PATCHES) && !empty(LOCAL_PATCHES) .for p in ${LOCAL_PATCHES} .if !exists(${p}) @echo "The patch file ${p} does not exist!" @exit 1 .endif .endfor .endif .if defined(LOCAL_SCRIPT) && !exists(${LOCAL_SCRIPT}) @echo "The local script ${LOCAL_SCRIPT} does not exist!" @exit 1 .endif .if make(release) .if exists(${CHROOTDIR}) # The first command will fail on a handful of files that have their schg # flags set. But it greatly speeds up the next two commands. # NB: clear any vestigial devfs mount, just in case -umount ${CHROOTDIR}/dev > /dev/null 2>&1 -rm -rf ${CHROOTDIR} 2>/dev/null -chflags -R 0 ${CHROOTDIR}/. -rm -rf ${CHROOTDIR} .endif mkdir -p ${CHROOTDIR} @echo ">>> make release for ${TARGET} started on `LC_ALL=C TZ=GMT date`" cd ${WORLDDIR} && ${NATIVEMAKE} -DWITHOUT_GAMES -DWITHOUT_HTML -DWITHOUT_LIB32 \ -DWITHOUT_MAN -DWITHOUT_NLS -DWITHOUT_PROFILE installworld DESTDIR=${CHROOTDIR} cd ${WORLDDIR} && ${NATIVEMAKE} distribution DESTDIR=${CHROOTDIR} if [ -f /etc/resolv.conf ]; then \ cp -p /etc/resolv.conf ${CHROOTDIR}/etc; \ fi .if defined(EXTLOCALDIR) rm -rf ${CHROOTDIR}/usr/local cd ${CHROOTDIR}/usr && cp -R -H ${EXTLOCALDIR} local .endif rm -rf ${CHROOTDIR}/usr/src .if defined(EXTSRCDIR) cd ${CHROOTDIR}/usr && \ cp -R -H ${EXTSRCDIR} src .elif defined(SVNROOT) cd ${CHROOTDIR}/usr && \ ${SVNPREFIX} svn co ${SVNCMDARGS} ${SVNROOT}/${SVNBRANCH} \ ${RELEASESRCMODULE} .else cd ${CHROOTDIR}/usr && \ ${CVSPREFIX} cvs -R ${CVSARGS} -d ${CVSROOT} \ co ${CVSCMDARGS} ${CVS_SRCARGS} ${RELEASESRCMODULE} .endif .if defined(LOCAL_PATCHES) && !empty(LOCAL_PATCHES) .for p in ${LOCAL_PATCHES} patch -d ${CHROOTDIR}/usr/${RELEASESRCMODULE} ${PATCH_FLAGS} < ${p} .endfor .endif .if defined(LOCAL_SCRIPT) cd ${CHROOTDIR} && env CHROOTDIR=${CHROOTDIR} BUILDNAME=${BUILDNAME} \ RELEASETAG=${RELEASETAG} ${LOCAL_SCRIPT} .endif rm -rf ${CHROOTDIR}/usr/ports .if !defined(NOPORTSATALL) .if defined(EXTPORTSDIR) cd ${CHROOTDIR}/usr && cp -R -H ${EXTPORTSDIR} ports # If there are distfiles downloaded removing them rm -rf ports/distfiles/* .else cd ${CHROOTDIR}/usr && ${CVSPREFIX} cvs -R ${CVSARGS} -d ${CVSROOT} \ co ${CVSCMDARGS} ${CVS_PORTSARGS} ${RELEASEPORTSMODULE} .endif .endif .if !defined(NODOC) rm -rf ${CHROOTDIR}/usr/doc .if defined(EXTDOCDIR) cd ${CHROOTDIR}/usr && cp -R -H ${EXTDOCDIR} doc .else cd ${CHROOTDIR}/usr && ${CVSPREFIX} cvs -R ${CVSARGS} -d ${CVSROOT} \ co ${CVSCMDARGS} ${CVS_DOCARGS} ${RELEASEDOCMODULE} .endif if [ -d ${RELEASEDISTFILES}/ ]; then \ cp -rp ${RELEASEDISTFILES} ${CHROOTDIR}/usr/ports/distfiles; \ else \ mkdir -p ${CHROOTDIR}/usr/ports/distfiles; \ fi .if !defined(NO_PREFETCHDISTFILES) @cd ${.CURDIR} && ${MAKE} fetch-distfiles .endif .endif .endif .if make(rerelease) .if !defined(RELEASENOUPDATE) && !defined(EXTSRCDIR) .if defined(SVNROOT) cd ${CHROOTDIR}/usr/src && ${SVNPREFIX} svn switch ${SVNCMDARGS} \ ${SVNROOT}/${SVNBRANCH} .elif !defined(RELEASETAG) cd ${CHROOTDIR}/usr/src && ${CVSPREFIX} cvs -R ${CVSARGS} -q \ update ${CVSCMDARGS} -P -d -A .else cd ${CHROOTDIR}/usr/src && ${CVSPREFIX} cvs -R ${CVSARGS} -q \ update ${CVSCMDARGS} -P -d -r ${RELEASETAG} .endif rm -f ${CHROOTDIR}/tmp/.world_done .if !defined(NOPORTS) && !defined(EXTPORTSDIR) cd ${CHROOTDIR}/usr/ports && ${CVSPREFIX} cvs -R ${CVSARGS} -q \ update ${CVSCMDARGS} -P -d rm -f ${CHROOTDIR}/tmp/.skip_ports_index .endif .if defined(DOMINIMALDOCPORTS) && ${DOMINIMALDOCPORTS} == "YES" for i in ${MINIMALDOCPORTS}; do \ ( cd ${CHROOTDIR}/usr/$$i && ${CVSPREFIX} cvs -R ${CVSARGS} -q \ update ${CVSCMDARGS} -P -d ) ; \ done .endif .if !defined(NODOC) cd ${CHROOTDIR}/usr/doc && ${CVSPREFIX} cvs -R ${CVSARGS} -q \ update ${CVSCMDARGS} -P -d .endif .endif .endif # Add version information to those things that need it. if [ ! -f ${CHROOTDIR}/tmp/.world_done ]; then \ cd ${CHROOTDIR}/usr/src/sys/conf && \ mv newvers.sh foo && \ sed "s/^RELEASE=.*/RELEASE=${BUILDNAME}/" foo > newvers.sh && \ rm foo; \ fi -test -f install.cfg && cp install.cfg ${CHROOTDIR}/usr/src/release echo "#!/bin/sh" > ${_MK} echo "set -ex" >> ${_MK} echo "trap 'umount /dev || true' 0" >> ${_MK} echo "_RELTARGET=\$${1:-doRELEASE}" >> ${_MK} .for var in \ AUTO_KEYBOARD_DETECT \ BUILDNAME \ CD_PACKAGE_TREE \ DISTRIBUTIONS \ DOMINIMALDOCPORTS \ EXTRA_SRC \ FTP_PASSIVE_MODE \ FTP_PROXY \ HTTP_PROXY \ KERNELS \ KERNELS_BASE \ KERNEL_FLAGS \ MAKE_DVD \ MAKE_FLOPPIES \ MAKE_ISOS \ NOCDROM \ NODOC \ NOPORTS \ NO_SHARED \ NOSRC \ NO_CPU_CFLAGS \ NO_CPU_COPTFLAGS \ NO_FLOPPIES \ RELEASETAG \ RELNOTES_LANG \ SEPARATE_LIVEFS \ TARGET \ TARGET_ARCH \ TARGET_CPUARCH \ WORLD_FLAGS .if defined(${var}) echo "export ${var}=\"${${var}}\"" >> ${_MK} .endif .endfor # Don't remove this, or the build will fall over! echo "export RELEASEDIR=${_R}" >> ${_MK} echo "export PATH=/bin:/usr/bin:/sbin:/usr/sbin:${LOCALDIR}" >> ${_MK} echo "export MANBUILDCAT=YES" >> ${_MK} # NB: these may fail if the host is running w/o devfs echo "umount /dev >/dev/null 2>&1 || true" >> ${_MK} echo "mount -t devfs devfs /dev >/dev/null 2>&1 || true" >> ${_MK} echo "if [ ! -c /dev/null ]; then" >> ${_MK} echo " echo /dev/null is not a device!" >> ${_MK} echo " exit 1" >> ${_MK} echo "fi" >> ${_MK} echo "if [ -x /etc/rc.d/ldconfig ]; then" >> ${_MK} echo " /etc/rc.d/ldconfig start" >> ${_MK} echo "else" >> ${_MK} echo " ldconfig /lib /usr/lib /usr/local/lib || true" >> ${_MK} echo "fi" >> ${_MK} echo "if [ ! -f /tmp/.world_done ]; then" >> ${_MK} echo " cd /usr/src" >> ${_MK} echo " ${CROSSMAKE} ${WORLD_FLAGS} -DNO_CLEAN buildworld && \\" >> ${_MK} echo " touch /tmp/.world_done || exit 1" >> ${_MK} echo "fi" >> ${_MK} echo "if [ ! -f /tmp/.skip_ports_index ]; then" >> ${_MK} echo " echo \">>> make index started on \`LC_ALL=C TZ=GMT date\`\"" >> ${_MK} echo " for i in ${MAKEINDEXPORTS}" >> ${_MK} echo " do" >> ${_MK} echo " cd /usr/ports/\$${i}" >> ${_MK} echo " env -i HTTP_PROXY=$${HTTP_PROXY} FTP_PROXY=$${FTP_PROXY} FTP_PASSIVE_MODE=$${FTP_PASSIVE_MODE:-no} PATH=$${PATH} \\" >> ${_MK} echo " make all install clean BATCH=yes FORCE_PKG_REGISTER=yes" >> ${_MK} echo " done" >> ${_MK} echo " cd /usr/ports" >> ${_MK} echo " rm -f INDEX*" >> ${_MK} echo " make index -DINDEX_PRISTINE" >> ${_MK} echo " rm -f INDEX*.tmp" >> ${_MK} echo " touch /tmp/.skip_ports_index" >> ${_MK} echo " echo \">>> make index finished on \`LC_ALL=C TZ=GMT date\`\"" >> ${_MK} echo "fi" >> ${_MK} echo "cd /usr/src/release" >> ${_MK} echo "make obj" >> ${_MK} echo "make -f Makefile.sysinstall \$${_RELTARGET}" >> ${_MK} echo "echo \">>> make ${.TARGET} for ${TARGET} finished on \`LC_ALL=C TZ=GMT date\`\"" >> ${_MK} chmod 755 ${_MK} .if defined(NOPORTS) touch ${CHROOTDIR}/tmp/.skip_ports_index .endif # Ensure md.ko is loaded if md(4) is not statically compiled into # the kernel -mdconfig 2>/dev/null env -i /usr/sbin/chroot `dirname ${_MK}` /`basename ${_MK}` clean: rm -rf ${CRUNCH_TARGETS:S/$/_crunch/} release.[0-8] ${EXTRAS} fetch-distfiles: @for i in ${MAKEINDEXPORTS}; do \ cd ${CHROOTDIR}/usr/ports/$$i && \ make PORTSDIR=${CHROOTDIR}/usr/ports BATCH=yes \ checksum-recursive ; \ done @for i in ${DOCPORTS}; do \ cd ${CHROOTDIR}/usr/ports/$$i && \ make PORTSDIR=${CHROOTDIR}/usr/ports BATCH=yes \ WITHOUT_X11=yes JADETEX=no WITHOUT_PYTHON=yes \ checksum-recursive ; \ done # Clean out ${_R} and make the directory structure. release.1: mkdir -p ${_R} -rm -rf ${_R}/* 2> /dev/null -chflags -R noschg ${_R}/. rm -rf ${_R}/* mkdir ${RD} .if defined(MAKE_FLOPPIES) mkdir ${RD}/floppies .endif mkdir ${RD}/trees for i in ${DISTRIBUTIONS}; do \ mkdir ${RD}/trees/$$i && \ mtree -deU -f ${MTREEFILES}/BSD.root.dist \ -p ${RD}/trees/$$i > /dev/null && \ mtree -deU -f ${MTREEFILES}/BSD.usr.dist \ -p ${RD}/trees/$$i/usr > /dev/null && \ mtree -deU -f ${MTREEFILES}/BSD.include.dist \ -p ${RD}/trees/$$i/usr/include > /dev/null; \ done mkdir ${RD}/kernels for i in ${KERNELS_BASE} ${KERNELS}; do \ mkdir -p ${RD}/kernels/$${i}; \ done touch ${.TARGET} # Install the system into the various distributions. release.2: cd ${.CURDIR}/.. && ${CROSSMAKE} distrib-dirs DESTDIR=${RD}/trees/base cd ${.CURDIR}/.. && ${CROSSMAKE} ${WORLD_FLAGS} distributeworld \ DISTDIR=${RD}/trees sh ${.CURDIR}/scripts/mm-mtree.sh -F "${CROSSENV}" -D "${RD}/trees/base" touch ${.TARGET} # Make and install the generic kernel(s). release.3: .for kernel in ${KERNELS_BASE} ${KERNELS} cd ${.CURDIR}/..; \ ${CROSSMAKE} ${KERNEL_FLAGS} \ KERNCONF=${kernel} kernel \ DESTDIR=${RD}/kernels KODIR=/${kernel} .endfor touch ${.TARGET} # Make and install the three crunched binaries which live on the floppies. # You are not supposed to like this :-) release.4: rm -rf ${RD}/crunch mkdir -p ${RD}/crunch .for j in ${CRUNCH_TARGETS} .if exists(${.CURDIR}/${TARGET}/${j}_crunch.conf) rm -rf ${j}_crunch mkdir ${j}_crunch cd ${j}_crunch; ${WMAKEENV} MAKEFLAGS="-m ${.CURDIR}/../share/mk" \ NO_WERROR= crunchgen -o \ ${.CURDIR}/${TARGET}/${j}_crunch.conf cd ${j}_crunch; ${WMAKE} -f ${j}_crunch.mk subclean cd ${.CURDIR}/..; ${BINMAKE} -f Makefile.inc1 _build-tools cd ${j}_crunch; CFLAGS="-Os -pipe" ${WMAKE} -f ${j}_crunch.mk \ -DNO_CPU_CFLAGS all ${WMAKEENV} strip -R .comment ${j}_crunch/${j}_crunch mv ${j}_crunch/${j}_crunch ${RD}/crunch/${j} .endif .endfor touch ${.TARGET} # # --==## Fix up the distributions. ##==-- # release.5: # Create any "synthetic dists" now. @for i in ${DISTRIBUTIONS}; do \ if [ -f ${.CURDIR}/scripts/$${i}-make.sh ]; then \ echo -n "Running $$i dist creation script... "; \ env RD=${RD} sh ${.CURDIR}/scripts/$${i}-make.sh || echo "$$i distribution script returned bad status."; \ echo "Done."; \ fi \ done \ # Remove all the directories we don't need. -cd ${RD}/trees && \ (find ${OTHER_DISTS} -path '*/var/empty' | xargs chflags noschg; \ find ${OTHER_DISTS} -depth -type d -empty -delete) touch ${.TARGET} # # --==## Package up the tarballs from assembled trees ##==-- # release.6: rm -rf ${RD}/dists mkdir -p ${RD}/dists @for i in ${DISTRIBUTIONS} ; \ do \ if [ -d ${RD}/trees/$${i} ] ; then \ cd ${.CURDIR} && $(MAKE) doTARBALL \ SD=${RD}/trees/$${i} \ TN=$$i TD=$$i ARG="." && \ echo "$${i} distribution is finished."; \ fi ; \ done @for i in ${KERNELS_BASE} ${KERNELS} ; \ do \ if [ -d ${RD}/kernels/$${i} ] ; then \ cd ${.CURDIR} && $(MAKE) doTARBALL \ SD=${RD}/kernels \ TN=$$i TD=kernels ARG="$$i" && \ echo "$${i} distribution is finished."; \ fi ; \ done .if !defined(NOPORTS) # XXX: Inline stripped version of doTARBALL @rm -rf ${RD}/dists/ports/ports* @mkdir -p ${RD}/dists/ports @echo rolling ports/ports tarball @tar --exclude CVS --exclude .svn --exclude 'ports/distfiles/*' \ -czf ${RD}/dists/ports/ports.tgz -C /usr ports @cp ${.CURDIR}/scripts/ports-install.sh ${RD}/dists/ports/install.sh @(cd ${RD}/dists/ports; \ rm -f CHECKSUM.MD5 CHECKSUM.SHA256; \ md5 * > .CHECKSUM.MD5; \ sha256 * > .CHECKSUM.SHA256; \ mv .CHECKSUM.MD5 CHECKSUM.MD5; \ mv .CHECKSUM.SHA256 CHECKSUM.SHA256) @echo "ports distribution is finished." .endif touch ${.TARGET} # # --==## Make source dists ##==-- # release.7: .if !defined(NOSRC) @cd ${.CURDIR} && $(MAKE) doTARBALL SD=/usr/src \ TD=src TN=sbase ARG="[A-Z]*" @for i in `cd /usr/src && echo [a-z]*` ; do \ if [ -d /usr/src/$$i ] ; then \ cd ${.CURDIR} && $(MAKE) doTARBALL \ TN=`echo s$$i | tr -d '.' | \ sed -e 's/usr/u/' \ -e 's/kerberos5/krb5/'` \ SD=/usr/src TD=src ARG="$$i" ; \ fi ; \ done .if defined(EXTRA_SRC) @set ${EXTRA_SRC} && \ while [ $$# -ge 2 ] ; do \ if [ -d /usr/src/$$1 ] ; then \ cd ${.CURDIR} && $(MAKE) doTARBALL \ SD=/usr/src TN="s$$2" TD=src ARG="$$1" ; \ fi && shift && shift ; \ done .endif (cd ${RD}/dists/src; \ rm -f CHECKSUM.MD5 CHECKSUM.SHA256; \ md5 * > .CHECKSUM.MD5; \ sha256 * > .CHECKSUM.SHA256; \ mv .CHECKSUM.MD5 CHECKSUM.MD5; \ mv .CHECKSUM.SHA256 CHECKSUM.SHA256) @echo "src distribution is finished." .endif touch ${.TARGET} # Build the memory root filesystem. release.8: cp ${RD}/trees/base/etc/disktab /etc rm -rf ${RD}/mfsfd mkdir ${RD}/mfsfd cd ${RD}/mfsfd && \ mkdir -p etc/defaults dev mnt stand/etc/defaults stand/help \ var/empty @cd ${.CURDIR} && $(MAKE) installCRUNCH CRUNCH=boot \ DIR=${RD}/mfsfd/stand ZIP=false ( cd ${RD}/mfsfd && \ for dir in bin sbin ; do \ ln -sf /stand $$dir; \ done ) cp ${RD}/trees/base/sbin/dhclient-script ${RD}/mfsfd/stand cp ${.CURDIR}/../etc/master.passwd ${RD}/mfsfd/etc/master.passwd cp ${RD}/trees/base/etc/*pwd.db ${RD}/mfsfd/etc/ ( for F in defaults/rc.conf netconfig protocols ; do \ sed -e '/^#.*$$/d' -e 's/[:space:]*#.*$$//g' \ ${RD}/trees/base/etc/$$F > ${RD}/mfsfd/stand/etc/$$F ; \ done ) grep -E '^(ftp|nameserver|domain|sunrpc|cmd|nfsd)[^-\w]' \ ${RD}/trees/base/etc/services | \ sed -e '/^#.*$$/d' -e 's/[:space:]*#.*$$//g' \ > ${RD}/mfsfd/stand/etc/services grep 'operator' ${RD}/trees/base/etc/group \ > ${RD}/mfsfd/stand/etc/group ln ${RD}/mfsfd/stand/etc/services ${RD}/mfsfd/etc/services ln ${RD}/mfsfd/stand/etc/group ${RD}/mfsfd/etc/group ln ${RD}/mfsfd/stand/etc/netconfig ${RD}/mfsfd/etc/netconfig cp ${RD}/trees/base/COPYRIGHT ${RD}/mfsfd/stand/help/COPYRIGHT.hlp .if !defined(NODOC) @for i in ${DIST_DOCS_ARCH_INDEP}; do \ cp ${RND}/${RELNOTES_LANG}/$$i/article.txt \ ${RD}/mfsfd/stand/help/`echo $${i} | tr 'a-z' 'A-Z'`.TXT; \ done @for i in ${DIST_DOCS_ARCH_DEP}; do \ cp ${RND}/${RELNOTES_LANG}/$$i/${TARGET}/article.txt \ ${RD}/mfsfd/stand/help/`echo $${i} | tr 'a-z' 'A-Z'`.TXT; \ done .endif -test -f ${.CURDIR}/install.cfg \ && cp ${.CURDIR}/install.cfg ${RD}/mfsfd @mkdir -p ${RD}/mfsfd/boot .if ${TARGET_ARCH} != "ia64" && ${TARGET_CPUARCH} != "powerpc" @cp ${RD}/trees/base/boot/boot* ${RD}/mfsfd/boot .endif .if ${TARGET} == "i386" || ${TARGET_ARCH} == "amd64" @cp ${RD}/trees/base/boot/mbr ${RD}/mfsfd/boot .endif @tar --exclude CVS --exclude .svn -cf - \ -C ${.CURDIR}/../usr.sbin/sysinstall help | \ tar xf - -C ${RD}/mfsfd/stand @mkdir -p ${RD}/mfsroot sh -e ${DOFS_SH} ${RD}/mfsroot/mfsroot ${RD} ${MNT} \ ${MFSSIZE} ${RD}/mfsfd ${MFSINODE} ${MFSLABEL} @${ZIPPER} -fv ${RD}/mfsroot/mfsroot touch ${.TARGET} KERNFLOPPYSET= ${RD}/floppyset/kern/kernel.gz .if defined(SMALLFLOPPYSIZE) SMALLKERNFLOPPYSET= ${RD}/floppyset/kern-small/kernel.gz .endif .if defined(SPLIT_MFSROOT) MFSROOTFLOPPYSET= ${RD}/floppyset/mfsroot/mfsroot.gz .if defined(SMALLFLOPPYSIZE) SMALLMFSROOTFLOPPYSET= ${RD}/floppyset/mfsroot-small/mfsroot.gz .endif .endif # Build boot and install floppies. floppies.1: @${ZIPPER} -c ${RD}/kernels/${KERN_GENERIC}/kernel > ${RD}/kernels/kernel.gz @echo "Making the kernel boot floppies..." @cd ${.CURDIR} && ${MAKE} makeFloppySet FLOPPYBASE=kern \ FLOPPYDESC="Kernel" SPLITFILE=${RD}/kernels/kernel.gz .if defined(SMALLFLOPPYSIZE) @echo "Making the small kernel boot floppies..." @cd ${.CURDIR} && ${MAKE} makeFloppySet FLOPPYBASE=kern-small \ FLOPPYDESC="Kernel" SPLITFILE=${RD}/kernels/kernel.gz \ FDSIZE="SMALL" .endif .if defined(SPLIT_MFSROOT) @echo "Making the mfsroot boot floppies..." @cd ${.CURDIR} && ${MAKE} makeFloppySet FLOPPYBASE=mfsroot \ FLOPPYDESC="Memory Filesystem" SPLITFILE=${RD}/mfsroot/mfsroot.gz .if defined(SMALLFLOPPYSIZE) @echo "Making the small mfsroot boot floppies..." @cd ${.CURDIR} && ${MAKE} makeFloppySet FLOPPYBASE=mfsroot-small \ FLOPPYDESC="Memory Filesystem" SPLITFILE=${RD}/mfsroot/mfsroot.gz \ FDSIZE="SMALL" .endif @cd ${.CURDIR} && ${MAKE} buildBootFloppy FSIMAGE="boot" \ KERNFILE="${KERNFLOPPYSET}.split ${KERNFLOPPYSET}.boot" \ MFSROOTFILE="${MFSROOTFLOPPYSET}.split ${MFSROOTFLOPPYSET}.boot" .if defined(SMALLFLOPPYSIZE) @cd ${.CURDIR} && ${MAKE} buildBootFloppy FSIMAGE="boot-small" \ KERNFILE="${SMALLKERNFLOPPYSET}.split ${SMALLKERNFLOPPYSET}.boot" \ MFSROOTFILE="${SMALLMFSROOTFLOPPYSET}.split ${SMALLMFSROOTFLOPPYSET}.boot" \ FDSIZE="SMALL" .endif .else # !SPLIT_MFSROOT @cd ${.CURDIR} && ${MAKE} buildBootFloppy FSIMAGE="boot" \ KERNFILE="${KERNFLOPPYSET}.split ${KERNFLOPPYSET}.boot" \ MFSROOTFILE=${RD}/mfsroot/mfsroot.gz .if defined(SMALLFLOPPYSIZE) @cd ${.CURDIR} && ${MAKE} buildBootFloppy FSIMAGE="boot-small" \ KERNFILE="${SMALLKERNFLOPPYSET}.split ${SMALLKERNFLOPPYSET}.boot" \ MFSROOTFILE=${RD}/mfsroot/mfsroot.gz \ FDSIZE="SMALL" .endif .endif # SPLIT_MFSROOT touch ${.TARGET} # Build fixit floppy. floppies.2: @echo "Making fixit floppy." @rm -rf ${RD}/fixitfd @mkdir ${RD}/fixitfd @cd ${RD}/fixitfd && \ mkdir -p dev stand bin sbin etc mnt mnt1 mnt2 mnt3 mnt4 tmp \ usr/share/misc @cp ${RD}/trees/base/etc/spwd.db ${RD}/trees/base/etc/group \ ${RD}/fixitfd/etc @sed -e 's/#.*//' ${RD}/trees/base/etc/protocols \ > ${RD}/fixitfd/etc/protocols @sed -e 's/#.*//' ${RD}/trees/base/usr/share/misc/scsi_modes \ > ${RD}/fixitfd/usr/share/misc/scsi_modes @cp ${.CURDIR}/fixit.profile ${RD}/fixitfd/.profile @cp ${.CURDIR}/fixit.services ${RD}/fixitfd/etc/services @cp ${.CURDIR}/scripts/tar.sh ${RD}/fixitfd/stand/tar @chmod 555 ${RD}/fixitfd/stand/tar .if defined(SMALLFLOPPYSIZE) @cd ${.CURDIR} && ${MAKE} installCRUNCH CRUNCH=fixit-small \ DIR=${RD}/fixitfd/stand ZIP=false @sh -e ${DOFS_SH} ${RD}/floppies/fixit-small.flp ${RD} ${MNT} \ ${SMALLFLOPPYSIZE} ${RD}/fixitfd ${FLOPPYINODE} ${SMALLFLOPPYLABEL} @rm -rf ${RD}/fixitfd/stand @mkdir ${RD}/fixitfd/stand .endif @cd ${.CURDIR} && ${MAKE} installCRUNCH CRUNCH=fixit \ DIR=${RD}/fixitfd/stand ZIP=false @sh -e ${DOFS_SH} ${RD}/floppies/fixit.flp ${RD} ${MNT} \ ${FLOPPYSIZE} ${RD}/fixitfd ${FLOPPYINODE} ${FLOPPYLABEL} touch ${.TARGET} # Do our last minute floppies directory setup floppies.3: @(cd ${RD}/floppies; md5 *.flp > CHECKSUM.MD5) @(cd ${RD}/floppies; sha256 *.flp > CHECKSUM.SHA256) touch ${.TARGET} # # --==## Setup a suitable ftp-area ##==-- # ftp.1: @echo "Setting up FTP distribution area" @mkdir -p ${FD} -@ln -s . ${FD}/${BUILDNAME} .if defined(MAKE_FLOPPIES) @cd ${RD} && find floppies -print | cpio -dumpl ${FD} .endif @cd ${RD}/dists && find . -print | cpio -dumpl ${FD} .if !defined(NODOC) @for i in ${DIST_DOCS_ARCH_INDEP}; do \ cp ${RND}/${RELNOTES_LANG}/$$i/article.txt \ ${FD}/`echo $${i} | tr 'a-z' 'A-Z'`.TXT; \ cp ${RND}/${RELNOTES_LANG}/$$i/article.html \ ${FD}/`echo $${i} | tr 'a-z' 'A-Z'`.HTM; \ done @for i in ${DIST_DOCS_ARCH_DEP}; do \ cp ${RND}/${RELNOTES_LANG}/$$i/${TARGET}/article.txt \ ${FD}/`echo $${i} | tr 'a-z' 'A-Z'`.TXT; \ cp ${RND}/${RELNOTES_LANG}/$$i/${TARGET}/article.html \ ${FD}/`echo $${i} | tr 'a-z' 'A-Z'`.HTM; \ done @cp ${RND}/${RELNOTES_LANG}/readme/docbook.css ${FD} .endif @echo "CD_VERSION = ${BUILDNAME}" > ${FD}/cdrom.inf touch ${.TARGET} # Build a live filesystem cdrom image cdrom.1: @echo "Building CDROM live filesystem image" @mkdir -p ${CD_LIVEFS} @for i in ${DISTRIBUTIONS} ; \ do \ if [ -d ${RD}/trees/$${i} ] ; then \ chflags -R noschg ${RD}/trees/$${i} || true ; \ ( cd ${RD}/trees/$${i} && \ find . -depth -print | cpio -dumpl ${CD_LIVEFS} ) ; \ fi \ done .if defined(MAKE_DVD) @echo "Building DVD filesystem image as well as CDROM" @mkdir -p ${CD_DVD1}/${BUILDNAME} @for i in ${DISTRIBUTIONS} ; \ do \ if [ -d ${RD}/trees/$${i} ] ; then \ chflags -R noschg ${RD}/trees/$${i} || true ; \ ( cd ${RD}/trees/$${i} && \ find . -depth -print | cpio -dumpl ${CD_DVD1} ) ; \ fi \ done .endif @echo "Copy GENERIC kernel to boot area" @cp -Rp ${RD}/kernels/${KERN_GENERIC}/ ${CD_LIVEFS}/boot/kernel @rm -f ${CD_LIVEFS}/boot/kernel/*.symbols @rm -f ${CD_LIVEFS}/.profile @cp ${.CURDIR}/fixit.profile ${CD_LIVEFS}/.profile @ln -sf /rescue ${CD_LIVEFS}/stand @echo "Setting up CDROM boot area" @rm -f ${CD_LIVEFS}/boot/loader.conf @cp ${RD}/mfsroot/mfsroot.gz ${CD_LIVEFS}/boot/mfsroot.gz @echo 'mfsroot_load="YES"' > ${CD_LIVEFS}/boot/loader.conf @echo 'mfsroot_type="mfs_root"' >> ${CD_LIVEFS}/boot/loader.conf @echo 'mfsroot_name="/boot/mfsroot"' >> ${CD_LIVEFS}/boot/loader.conf .if exists(${RD}/trees/base/boot/device.hints) # Break the link to device.hints so we can modify it @rm -f ${CD_LIVEFS}/boot/device.hints @cp ${RD}/trees/base/boot/device.hints ${CD_LIVEFS}/boot/device.hints .endif @echo "CD_VERSION = ${BUILDNAME}" > ${CD_LIVEFS}/cdrom.inf .if defined(MAKE_DVD) @cp -Rp ${RD}/kernels/${KERN_GENERIC}/ ${CD_DVD1}/boot/kernel @rm -f ${CD_DVD1}/boot/kernel/*.symbols @rm -f ${CD_DVD1}/.profile @cp ${.CURDIR}/fixit.profile ${CD_DVD1}/.profile @ln -sf /rescue ${CD_DVD1}/stand @echo "CD_VERSION = ${BUILDNAME}" > ${CD_DVD1}/cdrom.inf @rm -f ${CD_DVD1}/boot/loader.conf @cp ${RD}/mfsroot/mfsroot.gz ${CD_DVD1}/boot/mfsroot.gz @echo 'mfsroot_load="YES"' > ${CD_DVD1}/boot/loader.conf @echo 'mfsroot_type="mfs_root"' >> ${CD_DVD1}/boot/loader.conf @echo 'mfsroot_name="/boot/mfsroot"' >> ${CD_DVD1}/boot/loader.conf .if exists(${RD}/trees/base/boot/device.hints) @rm -f ${CD_DVD1}/boot/device.hints @cp ${RD}/trees/base/boot/device.hints ${CD_DVD1}/boot/device.hints .endif .endif touch ${.TARGET} # Build disc1 and disc2 cdrom images cdrom.2: @echo "Building CDROM disc1 filesystem image" @mkdir -p ${CD_DISC1}/${BUILDNAME} .if defined(MAKE_FLOPPIES) @cd ${RD} && find floppies -print | cpio -dumpl ${CD_DISC1} .endif @cd ${RD}/dists && find . -print | cpio -dumpl ${CD_DISC1}/${BUILDNAME} .if !defined(NODOC) @for i in ${DIST_DOCS_ARCH_INDEP}; do \ cp ${RND}/${RELNOTES_LANG}/$$i/article.txt \ ${CD_DISC1}/`echo $${i} | tr 'a-z' 'A-Z'`.TXT; \ cp ${RND}/${RELNOTES_LANG}/$$i/article.html \ ${CD_DISC1}/`echo $${i} | tr 'a-z' 'A-Z'`.HTM; \ done @for i in ${DIST_DOCS_ARCH_DEP}; do \ cp ${RND}/${RELNOTES_LANG}/$$i/${TARGET}/article.txt \ ${CD_DISC1}/`echo $${i} | tr 'a-z' 'A-Z'`.TXT; \ cp ${RND}/${RELNOTES_LANG}/$$i/${TARGET}/article.html \ ${CD_DISC1}/`echo $${i} | tr 'a-z' 'A-Z'`.HTM; \ done @cp ${RND}/${RELNOTES_LANG}/readme/docbook.css ${CD_DISC1} .endif .if defined(SEPARATE_LIVEFS) @cp -Rp ${CD_LIVEFS}/boot ${CD_DISC1} @echo "CD_VERSION = ${BUILDNAME}" > ${CD_DISC1}/cdrom.inf .endif @echo "CD_VOLUME = 1" >> ${CD_DISC1}/cdrom.inf @echo "Building CDROM disc2 filesystem image" @mkdir -p ${CD_DISC2} @echo "CD_VERSION = ${BUILDNAME}" > ${CD_DISC2}/cdrom.inf @echo "CD_VOLUME = 2" >> ${CD_DISC2}/cdrom.inf .if defined(MAKE_DVD) .if defined(MAKE_FLOPPIES) @cd ${RD} && find floppies -print | cpio -dumpl ${CD_DVD1} .endif @cd ${RD}/dists && find . -print | cpio -dumpl ${CD_DVD1}/${BUILDNAME} .if !defined(NODOC) @for i in ${DIST_DOCS_ARCH_INDEP}; do \ cp ${RND}/${RELNOTES_LANG}/$$i/article.txt \ ${CD_DVD1}/`echo $${i} | tr 'a-z' 'A-Z'`.TXT; \ cp ${RND}/${RELNOTES_LANG}/$$i/article.html \ ${CD_DVD1}/`echo $${i} | tr 'a-z' 'A-Z'`.HTM; \ done @for i in ${DIST_DOCS_ARCH_DEP}; do \ cp ${RND}/${RELNOTES_LANG}/$$i/${TARGET}/article.txt \ ${CD_DVD1}/`echo $${i} | tr 'a-z' 'A-Z'`.TXT; \ cp ${RND}/${RELNOTES_LANG}/$$i/${TARGET}/article.html \ ${CD_DVD1}/`echo $${i} | tr 'a-z' 'A-Z'`.HTM; \ done @cp ${RND}/${RELNOTES_LANG}/readme/docbook.css ${CD_DVD1} .endif @echo "CD_VERSION = ${BUILDNAME}" > ${CD_DVD1}/cdrom.inf @echo "CD_VOLUME = 1" >> ${CD_DVD1}/cdrom.inf .endif .if !defined(NODOC) @echo "Building CDROM docs filesystem image" @mkdir -p ${CD_DOCS} @echo "CD_VERSION = ${BUILDNAME}" > ${CD_DOCS}/cdrom.inf @echo "CD_VOLUME = 3" >> ${CD_DOCS}/cdrom.inf .endif touch ${.TARGET} # # --==## Setup a suitable cdrom-area ##==-- # cdrom.3: .if defined(CD_BOOT) @echo "Building bootonly CDROM filesystem image" @mkdir -p ${CD_BOOT} @cp -Rp ${CD_LIVEFS}/boot ${CD_BOOT} @echo "CD_VERSION = ${BUILDNAME}" > ${CD_BOOT}/cdrom.inf .if defined(MINIROOT) @echo "Building bootonly UFS filesystem image" @mkdir -p ${FD}/miniroot @sh -e ${DOFS_SH} ${FD}/miniroot/miniroot.ufs \ ${RD} ${MNT} 0 ${CD_BOOT} 8192 auto @${ZIPPER} -v ${FD}/miniroot/miniroot.ufs .endif .endif touch ${.TARGET} .if make(iso.1) .if defined(CD_PACKAGE_TREE) .if exists(${CD_PACKAGE_TREE}/disc1) CD_DISC1_PKGS= ${CD_PACKAGE_TREE}/disc1 .endif .if exists(${CD_PACKAGE_TREE}/disc2) CD_DISC2_PKGS= ${CD_PACKAGE_TREE}/disc2 .endif # scripts/package-trees.sh names all discs according to the "discX" # scheme where X is the number of the disc .if exists(${CD_PACKAGE_TREE}/disc3) CD_DOCS_PKGS= ${CD_PACKAGE_TREE}/disc3 .endif .if exists(${CD_PACKAGE_TREE}/dvd1) CD_DVD1_PKGS= ${CD_PACKAGE_TREE}/dvd1 .endif .endif .endif iso.1: -.if exists(${.CURDIR}/${TARGET_CPUARCH}/mkisoimages.sh) +.if exists(${.CURDIR}/${TARGET}/mkisoimages.sh) @echo "Creating ISO images..." .if defined(CD_BOOT) - @sh ${.CURDIR}/${TARGET_CPUARCH}/mkisoimages.sh ${BOOTABLE} \ + @sh ${.CURDIR}/${TARGET}/mkisoimages.sh ${BOOTABLE} \ FreeBSD_bootonly \ ${CD}/FreeBSD-${BUILDNAME}-${TARGET}-bootonly.iso ${CD_BOOT} .endif - @sh ${.CURDIR}/${TARGET_CPUARCH}/mkisoimages.sh ${BOOTABLE} \ + @sh ${.CURDIR}/${TARGET}/mkisoimages.sh ${BOOTABLE} \ FreeBSD_Install \ ${CD}/FreeBSD-${BUILDNAME}-${TARGET}-disc1.iso ${CD_DISC1} \ ${CD_DISC1_PKGS} - @sh ${.CURDIR}/${TARGET_CPUARCH}/mkisoimages.sh \ + @sh ${.CURDIR}/${TARGET}/mkisoimages.sh \ FreeBSD_Packages \ ${CD}/FreeBSD-${BUILDNAME}-${TARGET}-disc2.iso ${CD_DISC2} \ ${CD_DISC2_PKGS} .if defined(MAKE_DVD) - @sh ${.CURDIR}/${TARGET_CPUARCH}/mkisoimages.sh ${BOOTABLE} \ + @sh ${.CURDIR}/${TARGET}/mkisoimages.sh ${BOOTABLE} \ FreeBSD_Install \ ${CD}/FreeBSD-${BUILDNAME}-${TARGET}-dvd1.iso ${CD_DVD1} \ ${CD_DVD1_PKGS} .endif .if !defined(NODOC) - @sh ${.CURDIR}/${TARGET_CPUARCH}/mkisoimages.sh \ + @sh ${.CURDIR}/${TARGET}/mkisoimages.sh \ FreeBSD_Documentation \ ${CD}/FreeBSD-${BUILDNAME}-${TARGET}-disc3.iso ${CD_DOCS} \ ${CD_DOCS_PKGS} .endif .if defined(SEPARATE_LIVEFS) - @sh ${.CURDIR}/${TARGET_CPUARCH}/mkisoimages.sh ${BOOTABLE} \ + @sh ${.CURDIR}/${TARGET}/mkisoimages.sh ${BOOTABLE} \ FreeBSD_LiveFS \ ${CD}/FreeBSD-${BUILDNAME}-${TARGET}-livefs.iso ${CD_LIVEFS} .endif @echo "Generating MD5 and SHA256 sums..." @(cd ${CD} && md5 *.iso > FreeBSD-${BUILDNAME}-${TARGET}-iso.CHECKSUM.MD5) @(cd ${CD} && sha256 *.iso > FreeBSD-${BUILDNAME}-${TARGET}-iso.CHECKSUM.SHA256) touch ${.TARGET} .else - @echo "Do not know how to create an ISO for ${TARGET_CPUARCH}." + @echo "Do not know how to create an ISO for ${TARGET}." .endif # # --==## Documentation Project tools required to build the release notes ##==-- # doc.1: @echo "Making docproj tools..." @for i in ${DOCPORTS}; do \ cd /usr/ports/$$i && \ env -i FTP_PASSIVE_MODE=$${FTP_PASSIVE_MODE:-no} PATH=$${PATH} \ make all install clean BATCH=yes WITHOUT_X11=yes JADETEX=no \ WITHOUT_PYTHON=yes FORCE_PKG_REGISTER=yes; \ done touch ${.TARGET} # # --==## RELNOTESng: Next-generation replacements for *.TXT files ##==-- # doc.2: @echo "Making release documentation..." @cd ${.CURDIR}/doc && make all install clean 'FORMATS=html txt' \ INSTALL_COMPRESSED='' URLS_ABSOLUTE=YES DOCDIR=${RND} touch ${.TARGET} # Various "subroutine" and other supporting targets. # RD= # SD= # TD= # ARG= doTARBALL: .if !defined(SD) @echo "SD undefined in doTARBALL" && exit 1 .endif .if !defined(TD) @echo "TD undefined in doTARBALL" && exit 1 .endif .if !defined(ARG) @echo "ARG undefined in doTARBALL" && exit 1 .endif @rm -rf ${RD}/dists/${TD}/${TN}* @mkdir -p ${RD}/dists/${TD} @( cd ${SD} && \ tn=`echo ${TN} | tr 'A-Z' 'a-z'` && \ echo rolling ${TD}/$$tn tarball &&\ tar --exclude CVS --exclude .svn --exclude obj --exclude BOOTMFS -cf - ${ARG} | \ ${ZIPNSPLIT} ${RD}/dists/${TD}/$$tn. && \ sh ${.CURDIR}/scripts/info.sh ${RD}/dists/${TD}/$$tn \ > ${RD}/dists/${TD}/$$tn.inf && \ if [ -f ${.CURDIR}/scripts/$${TD}-install.sh ]; then \ cp -p ${.CURDIR}/scripts/$${TD}-install.sh \ ${RD}/dists/${TD}/install.sh && \ chmod +x ${RD}/dists/${TD}/install.sh; \ fi && \ if [ "${SD}" != "/usr/src" ]; then \ mtree -c -i -p ${SD}/${ARG} \ -k gname,md5digest,mode,nlink,uname,size,link,type \ > ${RD}/dists/${TD}/$$tn.mtree ; \ else \ true; \ fi; \ ( cd ${RD}/dists/${TD}; \ rm -f CHECKSUM.MD5 CHECKSUM.SHA256; \ md5 * > .CHECKSUM.MD5; \ sha256 * > .CHECKSUM.SHA256; \ mv .CHECKSUM.MD5 CHECKSUM.MD5; \ mv .CHECKSUM.SHA256 CHECKSUM.SHA256) \ ) doRELEASE: release.1 release.2 ${DOCREL} release.3 release.4 \ release.5 release.6 release.7 release.8 ${EXTRAS} @echo "Release done" floppies: @rm -f release.4 release.8 floppies.[123] @cd ${.CURDIR} && ${MAKE} release.4 release.8 floppies.1 floppies.2 \ floppies.3 @cd ${RD} && find floppies -print | cpio -dumpl ${FD} installCRUNCH: .if !defined(CRUNCH) @echo "CRUNCH undefined in installCRUNCH" && exit 1 .endif .if !defined(DIR) @echo "DIR undefined in installCRUNCH" && exit 1 .endif .if !defined(ZIP) @echo "ZIP undefined in installCRUNCH" && exit 1 .endif @if ${ZIP} ; then \ ${ZIPPER} < ${RD}/crunch/${CRUNCH} > ${DIR}/${CRUNCH}_crunch ; \ else \ ln -f ${RD}/crunch/${CRUNCH} ${DIR}/${CRUNCH}_crunch ; \ fi @chmod 555 ${DIR}/${CRUNCH}_crunch @if [ -f ${.CURDIR}/${TARGET}/${CRUNCH}_crunch.conf ] ; then \ for i in `crunchgen -l ${.CURDIR}/${TARGET}/${CRUNCH}_crunch.conf` ; do \ ln -f ${DIR}/${CRUNCH}_crunch ${DIR}/$$i ; \ done \ else \ for i in `crunchgen -l ${.CURDIR}/${CRUNCH}_crunch.conf` ; do \ ln -f ${DIR}/${CRUNCH}_crunch ${DIR}/$$i ; \ done \ fi # # --==## Build a floppy set for a splitfs file ##==-- # # FLOPPYBASE - basename of floppy image files # FLOPPYDESC - description of floppy set # SPLITFILE - filename of the file to split # FDSIZE - if specified and "small", small floppy is created .if make(makeFloppySet) SPLITDIR= ${RD}/floppyset/${FLOPPYBASE} .if defined(FDSIZE) && ${FDSIZE} == "SMALL" FLPSPLITSIZE= ${SMALLFLOPPYSPLITSIZE} FLPSIZE= ${SMALLFLOPPYSIZE} FLPLABEL= ${SMALLFLOPPYLABEL} .else FLPSPLITSIZE= ${FLOPPYSPLITSIZE} FLPSIZE= ${FLOPPYSIZE} FLPLABEL= ${FLOPPYLABEL} .endif .endif makeFloppySet: .if !defined(FLOPPYBASE) @echo "FLOPPYBASE undefined in ${.TARGET}" && exit 1 .endif .if !defined(FLOPPYDESC) @echo "FLOPPYDESC undefined in ${.TARGET}" && exit 1 .endif .if !defined(SPLITFILE) @echo "SPLITFILE undefined in ${.TARGET}" && exit 1 .endif sh ${.CURDIR}/scripts/split-file.sh ${SPLITFILE} \ ${RD}/floppyset/${FLOPPYBASE} ${FLPSPLITSIZE} "${FLOPPYDESC}" ( splitfile=${SPLITDIR}/`basename ${SPLITFILE}`.split ; \ lines=`cat $${splitfile} | wc -l`; \ lines=$$(($$lines - 1)) ; \ for line in `jot $$lines`; do \ file=`head -n $$(($${line} + 1)) $${splitfile} | tail -1 | cut -f 1 -d ' '` ; \ sh -e ${DOFS_SH} ${RD}/floppies/${FLOPPYBASE}$${line}.flp \ ${RD} ${MNT} ${FLPSIZE} ${SPLITDIR}/$${file} \ ${BOOTINODE} ${FLPLABEL}; \ done ) # # --==## Build a boot floppy ##==-- # # FSIMAGE - base floppy image name # FDSIZE - if specified and "small", small floppy is created # KERNFILE - path to kernel split file # MFSROOTFILE - path to mfsroot split file .if make(buildBootFloppy) IMAGEDIR= ${RD}/image.${FSIMAGE} BOOTDIR= ${RD}/trees/base/boot HINTSFILE= ${BOOTDIR}/device.hints ACPI_KO= ${RD}/kernels/${KERN_GENERIC}/acpi.ko IMAGEFILE= ${RD}/floppies/${FSIMAGE}.flp .if defined(FDSIZE) && ${FDSIZE} == "SMALL" FLPSIZE= ${SMALLFLOPPYSIZE} FLPLABEL= ${SMALLFLOPPYLABEL} .else FLPSIZE= ${FLOPPYSIZE} FLPLABEL= ${FLOPPYLABEL} .endif .endif buildBootFloppy: .if !defined(FSIMAGE) @echo "FSIMAGE undefined in ${.TARGET}" && exit 1 .endif .if !defined(KERNFILE) @echo "KERNFILE undefined in ${.TARGET}" && exit 1 .endif .if !defined(MFSROOTFILE) @echo "MFSROOTFILE undefined in ${.TARGET}" && exit 1 .endif @echo "Running ${.TARGET} for ${FSIMAGE}" @rm -rf ${IMAGEDIR} @mkdir ${IMAGEDIR} @echo "Setting up /boot directory for ${FSIMAGE} floppy" @mkdir -p ${IMAGEDIR}/boot .if ${TARGET_ARCH} == "i386" @${WMAKEENV} kgzip -v -l ${RD}/trees/base/usr/lib/kgzldr.o -o \ ${IMAGEDIR}/boot/loader ${BOOTDIR}/loader .else @cp ${BOOTDIR}/loader ${IMAGEDIR}/boot .endif @cp -Rp ${BOOTDIR}/*.4th ${BOOTDIR}/defaults ${BOOTDIR}/loader.help \ ${BOOTDIR}/loader.rc ${IMAGEDIR}/boot .if exists(${HINTSFILE}) @cp ${HINTSFILE} ${IMAGEDIR}/boot/device.hints @${ZIPPER} ${IMAGEDIR}/boot/device.hints .endif @${ZIPPER} ${IMAGEDIR}/boot/*.4th ${IMAGEDIR}/boot/loader.help \ ${IMAGEDIR}/boot/defaults/loader.conf @echo 'bootfile="/kernel"' > ${IMAGEDIR}/boot/loader.conf .if exists(${ACPI_KO}) @${ZIPPER} -c ${ACPI_KO} > ${IMAGEDIR}/acpi.ko.gz @echo 'acpi_load="YES"' >> ${IMAGEDIR}/boot/loader.conf @echo 'acpi_name="/acpi.ko"' >> ${IMAGEDIR}/boot/loader.conf @echo 'acpi_before="read -p \"Insert boot floppy and press Enter\""' >> ${IMAGEDIR}/boot/loader.conf .endif @echo 'mfsroot_load="YES"' >> ${IMAGEDIR}/boot/loader.conf @echo 'mfsroot_type="mfs_root"' >> ${IMAGEDIR}/boot/loader.conf @echo 'mfsroot_name="/mfsroot"' >> ${IMAGEDIR}/boot/loader.conf .if !exists(${ACPI_KO}) @echo 'mfsroot_before="read -p \"Insert boot floppy and press Enter\""' >> ${IMAGEDIR}/boot/loader.conf .endif .if defined(SPLIT_MFSROOT) @echo 'mfsroot_after="read -p \"Insert boot floppy and press Enter\""' >> ${IMAGEDIR}/boot/loader.conf .endif .if ${TARGET_ARCH} == "i386" || ${TARGET_ARCH} == "amd64" .if ${AUTO_KEYBOARD_DETECT} @echo "-P" >> ${IMAGEDIR}/boot.config .endif .endif @rm -f ${IMAGEFILE} @cp ${KERNFILE} ${MFSROOTFILE} ${IMAGEDIR} sh -e ${DOFS_SH} ${IMAGEFILE} ${RD} ${MNT} ${FLPSIZE} ${IMAGEDIR} \ ${BOOTINODE} ${FLPLABEL} @echo "Created ${RD}/floppies/${FSIMAGE}.flp" .if make(package-split) # Targets related to making a package split # # PKG_COPY instructs the script to copy the actual package files rather than # make hard links # PKG_TREE is the path to the package tree to be split # PKG_DEST is the path to the destination tree to create the split in # PKG_VERBOSE asks for verbose output of the layout process # PKG_INDEX is the path to the INDEX file. By default ${PKG_TREE}/INDEX. .if defined(PKG_COPY) PKG_DO_COPY= cp .else PKG_DO_COPY= ln .endif PKG_WRKIDX= ${PKG_DEST}/INDEX.master PKG_ENV?= .if defined(TARGET_ARCH) PKG_ENV+= PKG_ARCH=${TARGET_ARCH} .endif .if defined(PKG_VERBOSE) PKG_ENV+= PKG_VERBOSE=1 .endif PKG_INDEX?= ${PKG_TREE}/INDEX package-split: .if !defined(PKG_TREE) @echo "PKG_TREE must be defined" && exit 1 .endif .if !defined(PKG_DEST) @echo "PKG_DEST must be defined" && exit 1 .endif @env ${PKG_ENV} python ${.CURDIR}/scripts/package-split.py \ ${PKG_INDEX} ${PKG_WRKIDX} @env ${PKG_ENV} sh ${.CURDIR}/scripts/package-trees.sh ${PKG_DO_COPY} \ ${PKG_WRKIDX} ${PKG_TREE} ${PKG_DEST} .endif .include Index: projects/largeSMP/release/amd64/mkisoimages.sh =================================================================== --- projects/largeSMP/release/amd64/mkisoimages.sh (revision 221494) +++ projects/largeSMP/release/amd64/mkisoimages.sh (revision 221495) @@ -1,60 +1,45 @@ #!/bin/sh # # Module: mkisoimages.sh # Author: Jordan K Hubbard # Date: 22 June 2001 # # $FreeBSD$ # # This script is used by release/Makefile to build the (optional) ISO images # for a FreeBSD release. It is considered architecture dependent since each # platform has a slightly unique way of making bootable CDs. This script # is also allowed to generate any number of images since that is more of # publishing decision than anything else. # # Usage: # # mkisoimages.sh [-b] image-label image-name base-bits-dir [extra-bits-dir] # # Where -b is passed if the ISO image should be made "bootable" by # whatever standards this architecture supports (may be unsupported), # image-label is the ISO image label, image-name is the filename of the # resulting ISO image, base-bits-dir contains the image contents and # extra-bits-dir, if provided, contains additional files to be merged # into base-bits-dir as part of making the image. publisher="The FreeBSD Project. http://www.freebsd.org/" if [ "x$1" = "x-b" ]; then # This is highly x86-centric and will be used directly below. - bootable="-b boot/cdboot -no-emul-boot" + bootable="-o bootimage=i386;$4/boot/cdboot -o no-emul-boot" shift else bootable="" fi if [ $# -lt 3 ]; then echo Usage: $0 '[-b] image-label image-name base-bits-dir [extra-bits-dir]' exit 1 fi -type mkisofs 2>&1 | grep " is " >/dev/null -if [ $? -ne 0 ]; then - echo The cdrtools port is not installed. Trying to get it now. - if [ -f /usr/ports/sysutils/cdrtools/Makefile ]; then - cd /usr/ports/sysutils/cdrtools && make install BATCH=yes && make clean - else - if ! pkg_add -r cdrtools; then - echo "Could not get it via pkg_add - please go install this" - echo "from the ports collection and run this script again." - exit 2 - fi - fi -fi - LABEL=$1; shift NAME=$1; shift -echo "/dev/iso9660/$LABEL / cd9660 ro 0 0" > $1/etc/fstab -mkisofs $bootable -r -J -V $LABEL -publisher "$publisher" -o $NAME $* +echo "/dev/iso9660/`echo $LABEL | tr '[:lower:]' '[:upper:]'` / cd9660 ro 0 0" > $1/etc/fstab +makefs -t cd9660 $bootable -o rockridge -o label=$LABEL $NAME $* rm $1/etc/fstab - Index: projects/largeSMP/release/i386/mkisoimages.sh =================================================================== --- projects/largeSMP/release/i386/mkisoimages.sh (revision 221494) +++ projects/largeSMP/release/i386/mkisoimages.sh (revision 221495) @@ -1,63 +1,45 @@ #!/bin/sh # # Module: mkisoimages.sh # Author: Jordan K Hubbard # Date: 22 June 2001 # # $FreeBSD$ # # This script is used by release/Makefile to build the (optional) ISO images # for a FreeBSD release. It is considered architecture dependent since each # platform has a slightly unique way of making bootable CDs. This script # is also allowed to generate any number of images since that is more of # publishing decision than anything else. # # Usage: # # mkisoimages.sh [-b] image-label image-name base-bits-dir [extra-bits-dir] # # Where -b is passed if the ISO image should be made "bootable" by # whatever standards this architecture supports (may be unsupported), # image-label is the ISO image label, image-name is the filename of the # resulting ISO image, base-bits-dir contains the image contents and # extra-bits-dir, if provided, contains additional files to be merged # into base-bits-dir as part of making the image. publisher="The FreeBSD Project. http://www.freebsd.org/" - if [ "x$1" = "x-b" ]; then - bootable="-b boot/cdboot -no-emul-boot" + # This is highly x86-centric and will be used directly below. + bootable="-o bootimage=i386;$4/boot/cdboot -o no-emul-boot" shift -elif [ "x$1" = "x-G" ]; then - bootable="-G /R/cdrom/bootonly/boot/cdboot" - shift else bootable="" fi if [ $# -lt 3 ]; then - echo Usage: $0 '[-bG] image-label image-name base-bits-dir [extra-bits-dir]' + echo Usage: $0 '[-b] image-label image-name base-bits-dir [extra-bits-dir]' exit 1 fi -type mkisofs 2>&1 | grep " is " >/dev/null -if [ $? -ne 0 ]; then - echo The cdrtools port is not installed. Trying to get it now. - if [ -f /usr/ports/sysutils/cdrtools/Makefile ]; then - cd /usr/ports/sysutils/cdrtools && make install BATCH=yes && make clean - else - if ! pkg_add -r cdrtools; then - echo "Could not get it via pkg_add - please go install this" - echo "from the ports collection and run this script again." - exit 2 - fi - fi -fi - LABEL=$1; shift NAME=$1; shift -echo "/dev/iso9660/$LABEL / cd9660 ro 0 0" > $1/etc/fstab -mkisofs $bootable -r -J -V $LABEL -publisher "$publisher" -o $NAME $* +echo "/dev/iso9660/`echo $LABEL | tr '[:lower:]' '[:upper:]'` / cd9660 ro 0 0" > $1/etc/fstab +makefs -t cd9660 $bootable -o rockridge -o label=$LABEL $NAME $* rm $1/etc/fstab - Index: projects/largeSMP/release/pc98/mkisoimages.sh =================================================================== --- projects/largeSMP/release/pc98/mkisoimages.sh (nonexistent) +++ projects/largeSMP/release/pc98/mkisoimages.sh (revision 221495) @@ -0,0 +1,45 @@ +#!/bin/sh +# +# Module: mkisoimages.sh +# Author: Jordan K Hubbard +# Date: 22 June 2001 +# +# $FreeBSD$ +# +# This script is used by release/Makefile to build the (optional) ISO images +# for a FreeBSD release. It is considered architecture dependent since each +# platform has a slightly unique way of making bootable CDs. This script +# is also allowed to generate any number of images since that is more of +# publishing decision than anything else. +# +# Usage: +# +# mkisoimages.sh [-b] image-label image-name base-bits-dir [extra-bits-dir] +# +# Where -b is passed if the ISO image should be made "bootable" by +# whatever standards this architecture supports (may be unsupported), +# image-label is the ISO image label, image-name is the filename of the +# resulting ISO image, base-bits-dir contains the image contents and +# extra-bits-dir, if provided, contains additional files to be merged +# into base-bits-dir as part of making the image. + +publisher="The FreeBSD Project. http://www.freebsd.org/" +if [ "x$1" = "x-b" ]; then + # This is highly x86-centric and will be used directly below. + bootable="-o generic-bootimage=$4/boot/cdboot" + shift +else + bootable="" +fi + +if [ $# -lt 3 ]; then + echo Usage: $0 '[-b] image-label image-name base-bits-dir [extra-bits-dir]' + exit 1 +fi + +LABEL=$1; shift +NAME=$1; shift + +echo "/dev/iso9660/`echo $LABEL | tr '[:lower:]' '[:upper:]'` / cd9660 ro 0 0" > $1/etc/fstab +makefs -t cd9660 $bootable -o rockridge -o label=$LABEL $NAME $* +rm $1/etc/fstab Property changes on: projects/largeSMP/release/pc98/mkisoimages.sh ___________________________________________________________________ Added: svn:keywords ## -0,0 +1 ## +FreeBSD=%H \ No newline at end of property Index: projects/largeSMP/release/powerpc/hfs.map =================================================================== --- projects/largeSMP/release/powerpc/hfs.map (revision 221494) +++ projects/largeSMP/release/powerpc/hfs.map (nonexistent) @@ -1,3 +0,0 @@ -# $FreeBSD$ -.tbxi - 'chrp' 'tbxi' "bootstrap" -* - 'fbsd' 'TEXT' "FreeBSD file" Property changes on: projects/largeSMP/release/powerpc/hfs.map ___________________________________________________________________ Deleted: svn:keywords ## -1 +0,0 ## -FreeBSD=%H \ No newline at end of property Index: projects/largeSMP/release/powerpc/generate-hfs.sh =================================================================== --- projects/largeSMP/release/powerpc/generate-hfs.sh (nonexistent) +++ projects/largeSMP/release/powerpc/generate-hfs.sh (revision 221495) @@ -0,0 +1,64 @@ +#!/bin/sh + +# This script generates the dummy HFS filesystem used for the PowerPC boot +# blocks. It uses hfsutils (emulators/hfsutils) to generate a template +# filesystem with the relevant interesting files. These are then found by +# grep, and the offsets written to a Makefile snippet. +# +# Because of licensing concerns, and because it is overkill, we do not +# distribute hfsutils as a build tool. If you need to regenerate the HFS +# template (e.g. because the boot block or the CHRP script have grown), +# you must install it from ports. + +# $FreeBSD$ + +HFS_SIZE=400 #Size in 2048-byte blocks of the produced image +LOADER_SIZE=300k + +# Generate 800K HFS image +OUTPUT_FILE=hfs-boot + +dd if=/dev/zero of=$OUTPUT_FILE bs=2048 count=$HFS_SIZE +hformat -l "FreeBSD Install" $OUTPUT_FILE +hmount $OUTPUT_FILE + +# Create and bless a directory for the boot loader +hmkdir ppc +hattrib -b ppc +hcd ppc + +# Make the CHRP boot script, which gets loader from the ISO9660 partition +cat > bootinfo.txt << EOF + +FreeBSD/powerpc bootloader +FreeBSD + $FreeBSD: head/sys/boot/powerpc/boot1.chrp/bootinfo.txt 184490 2008-10 +-31 00:52:31Z nwhitehorn $ + + +MacRISC MacRISC3 MacRISC4 + + +" screen" output +boot &device;:,\ppc\loader &device;:0 + + +EOF +echo 'Loader START' | dd of=loader.tmp cbs=$LOADER_SIZE count=1 conv=block + +hcopy bootinfo.txt :bootinfo.txt +hcopy loader.tmp :loader +hattrib -c chrp -t tbxi bootinfo.txt +humount + +rm bootinfo.txt +rm loader.tmp + +bzip2 $OUTPUT_FILE +echo 'HFS boot filesystem created by generate-hfs.sh' > $OUTPUT_FILE.bz2.uu +echo 'DO NOT EDIT' >> $OUTPUT_FILE.bz2.uu +echo '$FreeBSD$' >> $OUTPUT_FILE.bz2.uu + +uuencode $OUTPUT_FILE.bz2 $OUTPUT_FILE.bz2 >> $OUTPUT_FILE.bz2.uu +rm $OUTPUT_FILE.bz2 + Property changes on: projects/largeSMP/release/powerpc/generate-hfs.sh ___________________________________________________________________ Added: svn:executable ## -0,0 +1 ## +* \ No newline at end of property Added: svn:keywords ## -0,0 +1 ## +FreeBSD=%H \ No newline at end of property Index: projects/largeSMP/release/powerpc/hfs-boot.bz2.uu =================================================================== --- projects/largeSMP/release/powerpc/hfs-boot.bz2.uu (nonexistent) +++ projects/largeSMP/release/powerpc/hfs-boot.bz2.uu (revision 221495) @@ -0,0 +1,23 @@ +HFS boot filesystem created by generate-hfs.sh +DO NOT EDIT +$FreeBSD$ +begin 644 hfs-boot.bz2 +M0EIH.3%!62936=#$Y.(``"___?_O_G)7!_Y]OW??5#]U_^!`0`,@1`!!``!@ +M(0!`3,`"L"4.2U"2(H9%/:*>TFDVU3RGZID'J&C3U,@T-&@`/1!H>IZC1D#( +M`D24TTRFC1D]*;$@!IZ@Q!IH-````#0&33":?J@<:&AH:`:`Q`T!D``&F@`: +M`9````PDI1J>0Q,H/1J&@```!D-`````T-``#SZHHR(+Y*Y9J*Z:4KK=T]W@ +M4RV\LG/;0XDPC7^`9DO3/3&*].462T,NO#AA9C,O\0!)^Q@@S6";2`!);?8(B6BR1(-7Q*8?>>HPQSBQE>M$A9K6FFW<:#6[D931J%U.F8*`0 +MD(,M80A(0;]K'X'SI]G(C)Y).AK_/0)D(2$$]JB$]]F').M4RF8O%[IHK[0O +MMQ?'>O[U8EM;>U",KY^*%FO9C05_435?.4*F[6AIL%&9C<9W&& +M[ZMP(TG!'*4Z>@6TM)0BS.D._O,WR9OJ728V48I;DD=8QGBE7J-&L^?=0JEF +MV2`B/)/D\)83IS@32(;!P,152U()0VGQH>2F$UC:M!D5`F#1W$\:KKDR[TQB +MN;N5B1;P!7:AG#1BP2%-Q$7I6QE,Y?NHIK_LR+HA]22HW0.8(^G4/X`Z@!Z6 +8#PQ(0))/`A(0.'VUG_\7&1 | grep " is " >/dev/null -if [ $? -ne 0 ]; then - echo The cdrtools port is not installed. Trying to get it now. - if [ -f /usr/ports/sysutils/cdrtools/Makefile ]; then - cd /usr/ports/sysutils/cdrtools && make install BATCH=yes && make clean - else - if ! pkg_add -r cdrtools; then - echo "Could not get it via pkg_add - please go install this" - echo "from the ports collection and run this script again." - exit 2 - fi - fi -fi - LABEL=$1; shift NAME=$1; shift -echo "/dev/iso9660/$LABEL / cd9660 ro 0 0" > $1/etc/fstab -mkisofs $bootable -l -r -part -no-desktop -V $LABEL -o $NAME $* +echo "/dev/iso9660/`echo $LABEL | tr '[:lower:]' '[:upper:]'` / cd9660 ro 0 0" > $1/etc/fstab +makefs -t cd9660 $bootable -o rockridge -o label=$LABEL $NAME $* rm $1/etc/fstab - +rm /tmp/hfs-boot-block Index: projects/largeSMP/release/sun4v/mkisoimages.sh =================================================================== --- projects/largeSMP/release/sun4v/mkisoimages.sh (nonexistent) +++ projects/largeSMP/release/sun4v/mkisoimages.sh (revision 221495) @@ -0,0 +1,72 @@ +#!/bin/sh +# +# Module: mkisoimages.sh +# Author: Jordan K Hubbard +# Date: 22 June 2001 +# +# $FreeBSD$ +# +# This script is used by release/Makefile to build the (optional) ISO images +# for a FreeBSD release. It is considered architecture dependent since each +# platform has a slightly unique way of making bootable CDs. This script +# is also allowed to generate any number of images since that is more of +# publishing decision than anything else. +# +# Usage: +# +# mkisoimages.sh [-b] image-label image-name base-bits-dir [extra-bits-dir] +# +# Where -b is passed if the ISO image should be made "bootable" by +# whatever standards this architecture supports (may be unsupported), +# image-label is the ISO image label, image-name is the filename of the +# resulting ISO image, base-bits-dir contains the image contents and +# extra-bits-dir, if provided, contains additional files to be merged +# into base-bits-dir as part of making the image. + +publisher="The FreeBSD Project. http://www.freebsd.org/" +IMG=/tmp/bootfs +MNT=/mnt + +if [ "x$1" = "x-b" ]; then + dd if=/dev/zero of=${IMG} bs=512 count=1024 + MD=`mdconfig -a -t vnode -f ${IMG}` + sunlabel -w -B -b $4/boot/boot1 ${MD} auto + newfs -O1 -o space -m 0 /dev/${MD} + mount /dev/${MD} ${MNT} + mkdir ${MNT}/boot + cp $4/boot/loader ${MNT}/boot + umount ${MNT} + mdconfig -d -u ${MD#md} + bootable="-B ,,,,${IMG}" + shift +else + bootable="" +fi + +if [ $# -lt 3 ]; then + echo Usage: $0 '[-b] image-label image-name base-bits-dir [extra-bits-dir]' + rm -f ${IMG} + exit 1 +fi + +type mkisofs 2>&1 | grep " is " >/dev/null +if [ $? -ne 0 ]; then + echo The cdrtools port is not installed. Trying to get it now. + if [ -f /usr/ports/sysutils/cdrtools/Makefile ]; then + cd /usr/ports/sysutils/cdrtools && make install BATCH=yes && make clean + else + if ! pkg_add -r cdrtools; then + echo "Could not get it via pkg_add - please go install this" + echo "from the ports collection and run this script again." + exit 2 + fi + fi +fi + +LABEL=$1; shift +NAME=$1; shift + +echo "/dev/iso9660/$LABEL / cd9660 ro 0 0" > $1/etc/fstab +mkisofs $bootable -r -J -V $LABEL -publisher "$publisher" -o $NAME $* +rm $1/etc/fstab +rm -f ${IMG} Property changes on: projects/largeSMP/release/sun4v/mkisoimages.sh ___________________________________________________________________ Added: svn:keywords ## -0,0 +1 ## +FreeBSD=%H \ No newline at end of property Index: projects/largeSMP/sbin/ipfw =================================================================== --- projects/largeSMP/sbin/ipfw (revision 221494) +++ projects/largeSMP/sbin/ipfw (revision 221495) Property changes on: projects/largeSMP/sbin/ipfw ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sbin/ipfw:r221444-221493 Index: projects/largeSMP/sbin =================================================================== --- projects/largeSMP/sbin (revision 221494) +++ projects/largeSMP/sbin (revision 221495) Property changes on: projects/largeSMP/sbin ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sbin:r221444-221493 Index: projects/largeSMP/share/man/man9/bus_adjust_resource.9 =================================================================== --- projects/largeSMP/share/man/man9/bus_adjust_resource.9 (revision 221494) +++ projects/largeSMP/share/man/man9/bus_adjust_resource.9 (revision 221495) @@ -1,101 +1,101 @@ .\" -*- nroff -*- .\" .\" Copyright (c) 2011 Advanced Computing Technologies LLC .\" Written by: John H. Baldwin .\" All rights reserved. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" .\" $FreeBSD$ .\" .Dd April 29, 2011 .Dt BUS_ADJUST_RESOURCE 9 .Os .Sh NAME .Nm bus_adjust_resource .Nd adjust resource allocated from a parent bus .Sh SYNOPSIS .In sys/param.h .In sys/bus.h .Pp .In machine/bus.h .In sys/rman.h .In machine/resource.h .Ft int .Fn bus_adjust_resource "device_t dev" "int type" "struct resource *r" "u_long start" "u_long end" .Sh DESCRIPTION This function is used to ask the parent bus to adjust the resource range assigned to an allocated resource. The resource .Fa r should have been allocated by a previous call to .Xr bus_alloc_resource 9 . The new resource range must overlap the existing range of .Fa r . The .Fa type argument should match the .Fa type argument passed to .Xr bus_alloc_resource 9 when the resource was initially allocated. .Pp Note that none of the constraints of the original allocation request such as alignment or boundary restrictions are checked by .Fn bus_adjust_resource . It is the caller's responsibility to enforce any such requirements. .Sh RETURN VALUES The .Fn bus_adjust_resource method returns zero on success or an error code on failure. .Sh ERRORS .Fn bus_adjust_resource will fail if: .Bl -tag -width Er .It Bq Er EINVAL The .Fa dev device does not have a parent device. .It Bq Er EINVAL The .Fa r resource is a shared resource. .It Bq Er EINVAL The new address range does not overlap with the existing address range of .Fa r . .It Bq Er EBUSY The new address range conflicts with another allocated resource. .El .Sh EXAMPLES Grow an existing memory resource by 4096 bytes. .Bd -literal struct resource *res; int error; error = bus_adjust_resource(dev, SYS_RES_MEMORY, res, rman_get_start(res), rman_get_end(res) + 0x1000); -.Ed +.Ed .Sh SEE ALSO .Xr bus_alloc_resource 9 , .Xr bus_release_resource 9 , .Xr device 9 , .Xr driver 9 Index: projects/largeSMP/share/man/man9/make_dev.9 =================================================================== --- projects/largeSMP/share/man/man9/make_dev.9 (revision 221494) +++ projects/largeSMP/share/man/man9/make_dev.9 (revision 221495) @@ -1,400 +1,402 @@ .\" Copyright (c) 1999 Chris Costello .\" All rights reserved. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" .\" $FreeBSD$ .\" -.Dd May 03, 2011 +.Dd May 3, 2011 .Dt MAKE_DEV 9 .Os .Sh NAME .Nm make_dev , .Nm make_dev_cred , .Nm make_dev_credf , .Nm make_dev_p , .Nm make_dev_alias , .Nm make_dev_alias_p , .Nm destroy_dev , .Nm destroy_dev_sched , .Nm destroy_dev_sched_cb , .Nm destroy_dev_drain , .Nm dev_depends .Nd manage .Vt cdev Ns 's and DEVFS registration for devices .Sh SYNOPSIS .In sys/param.h .In sys/conf.h .Ft struct cdev * .Fn make_dev "struct cdevsw *cdevsw" "int unit" "uid_t uid" "gid_t gid" "int perms" "const char *fmt" ... .Ft struct cdev * .Fn make_dev_cred "struct cdevsw *cdevsw" "int unit" "struct ucred *cr" "uid_t uid" "gid_t gid" "int perms" "const char *fmt" ... .Ft struct cdev * .Fn make_dev_credf "int flags" "struct cdevsw *cdevsw" "int unit" "struct ucred *cr" "uid_t uid" "gid_t gid" "int perms" "const char *fmt" ... .Ft int .Fn make_dev_p "int flags" "struct cdev **cdev" "struct cdevsw *devsw" "struct ucred *cr" "uid_t uid" "gid_t gid" "int mode" "const char *fmt" ... .Ft struct cdev * .Fn make_dev_alias "struct cdev *pdev" "const char *fmt" ... .Ft int .Fn make_dev_alias_p "int flags" "struct cdev **cdev" "struct cdev *pdev" "const char *fmt" ... .Ft void .Fn destroy_dev "struct cdev *dev" .Ft void .Fn destroy_dev_sched "struct cdev *dev" .Ft void .Fn destroy_dev_sched_cb "struct cdev *dev" "void (*cb)(void *)" "void *arg" .Ft void .Fn destroy_dev_drain "struct cdevsw *csw" .Ft void .Fn dev_depends "struct cdev *pdev" "struct cdev *cdev" .Sh DESCRIPTION The .Fn make_dev_credf function creates a .Fa cdev structure for a new device. It also notifies .Xr devfs 5 of the presence of the new device, that causes corresponding nodes to be created. Besides this, a .Xr devctl 4 notification is sent. The device will be owned by .Va uid , with the group ownership as .Va gid . -The name is the expansion of -.Va fmt +The name is the expansion of +.Va fmt and following arguments as .Xr printf 9 would print it. The name determines its path under .Pa /dev -or other -.Xr devfs 5 +or other +.Xr devfs 5 mount point and may contain slash .Ql / characters to denote subdirectories. The permissions of the file specified in .Va perms are defined in .In sys/stat.h : .Pp .Bd -literal -offset indent -compact #define S_IRWXU 0000700 /* RWX mask for owner */ #define S_IRUSR 0000400 /* R for owner */ #define S_IWUSR 0000200 /* W for owner */ #define S_IXUSR 0000100 /* X for owner */ #define S_IRWXG 0000070 /* RWX mask for group */ #define S_IRGRP 0000040 /* R for group */ #define S_IWGRP 0000020 /* W for group */ #define S_IXGRP 0000010 /* X for group */ #define S_IRWXO 0000007 /* RWX mask for other */ #define S_IROTH 0000004 /* R for other */ #define S_IWOTH 0000002 /* W for other */ #define S_IXOTH 0000001 /* X for other */ #define S_ISUID 0004000 /* set user id on execution */ #define S_ISGID 0002000 /* set group id on execution */ #define S_ISVTX 0001000 /* sticky bit */ #ifndef _POSIX_SOURCE #define S_ISTXT 0001000 #endif .Ed .Pp The .Va cr argument specifies credentials that will be stored in the .Fa si_cred member of the initialized .Fa struct cdev . The .Va flags argument alters the operation of .Fn make_dev_credf or .Fn make_dev_p . The following values are currently accepted: .Pp .Bl -tag -width "MAKEDEV_CHECKNAME" -compact -offset indent .It MAKEDEV_REF reference the created device .It MAKEDEV_NOWAIT do not sleep, the call may fail .It MAKEDEV_WAITOK allow the function to sleep to satisfy malloc .It MAKEDEV_ETERNAL created device will be never destroyed .It MAKEDEV_CHECKNAME return an error if the device name is invalid or already exists .El .Pp -Only +Only .Dv MAKEDEV_NOWAIT , -.Dv MAKEDEV_WAITOK +.Dv MAKEDEV_WAITOK and .Dv MAKEDEV_CHECKNAME values are accepted for the .Fn make_dev_alias_p function. .Pp The .Dv MAKEDEV_WAITOK flag is assumed if none of .Dv MAKEDEV_WAITOK , .Dv MAKEDEV_NOWAIT is specified. .Pp The .Xr dev_clone 9 event handler shall specify .Dv MAKEDEV_REF flag when creating a device in response to lookup, to avoid race where the device created is destroyed immediately after .Xr devfs_lookup 9 drops his reference to cdev. .Pp The .Dv MAKEDEV_ETERNAL flag allows the kernel to not acquire some locks when translating system calls into the cdevsw methods calls. It is responsibility of the driver author to make sure that .Fn destroy_dev is never called on the returned cdev. For the convenience, use the .Dv MAKEDEV_ETERNAL_KLD flag for the code that can be compiled into kernel or loaded (and unloaded) as loadable module. .Pp A panic will occur if the MAKEDEV_CHECKNAME flag is not specified and the device name is invalid or already exists. .Pp The .Fn make_dev_cred function is equivalent to the call .Bd -literal -offset indent make_dev_credf(0, cdevsw, unit, cr, uid, gid, perms, fmt, ...); .Ed .Pp The .Fn make_dev function call is the same as .Bd -literal -offset indent make_dev_credf(0, cdevsw, unit, NULL, uid, gid, perms, fmt, ...); .Ed .Pp The .Fn make_dev_p function is similar to .Fn make_dev_credf but it may return an error number and takes a pointer to the resulting .Ft *cdev as an argument. .Pp The .Fn make_dev_alias function takes the returned .Ft cdev from .Fn make_dev and makes another (aliased) name for this device. It is an error to call .Fn make_dev_alias prior to calling .Fn make_dev . .Pp .Fn make_dev_alias_p function is similar to .Fn make_dev_alias but it takes a pointer to the resulting .Ft *cdev as an argument and may return an error. .Pp The .Fa cdev returned by .Fn make_dev and .Fn make_dev_alias has two fields, .Fa si_drv1 and .Fa si_drv2 , that are available to store state. Both fields are of type .Ft void * . These are designed to replace the .Fa unit argument to .Fn make_dev , which can be obtained with .Fn dev2unit . .Pp The .Fn destroy_dev function takes the returned .Fa cdev from .Fn make_dev and destroys the registration for that device. The notification is sent to .Xr devctl 4 about the destruction event. Do not call .Fn destroy_dev on devices that were created with .Fn make_dev_alias . .Pp The .Fn dev_depends function establishes a parent-child relationship between two devices. The net effect is that a .Fn destroy_dev of the parent device will also result in the destruction of the child device(s), if any exist. A device may simultaneously be a parent and a child, so it is possible to build a complete hierarchy. .Pp The .Fn destroy_dev_sched_cb function schedules execution of the .Fn destroy_dev for the specified .Fa cdev in the safe context. After .Fn destroy_dev is finished, and if the supplied .Fa cb is not .Dv NULL , the callback .Fa cb is called, with argument .Fa arg . The .Fn destroy_dev_sched function is the same as .Bd -literal -offset indent destroy_dev_sched_cb(cdev, NULL, NULL); .Ed .Pp The .Fn d_close driver method cannot call .Fn destroy_dev -directly. Doing so causes deadlock when +directly. +Doing so causes deadlock when .Fn destroy_dev waits for all threads to leave the driver methods. Also, because .Fn destroy_dev sleeps, no non-sleepable locks may be held over the call. The .Fn destroy_dev_sched family of functions overcome these issues. .Pp The device driver may call the .Fn destroy_dev_drain function to wait until all devices that have supplied .Fa csw -as cdevsw, are destroyed. This is useful when driver knows that +as cdevsw, are destroyed. +This is useful when driver knows that .Fn destroy_dev_sched is called for all instantiated devices, but need to postpone module unload until .Fn destroy_dev is actually finished for all of them. .Sh RETURN VALUES If successful, .Fn make_dev_p will return 0, otherwise it will return an error. If successful, .Fn make_dev_credf will return a valid .Fa cdev pointer, otherwise it will return .Dv NULL . .Sh ERRORS The .Fn make_dev_p and .Fn make_dev_alias_p call will fail and the device will be not registered if: .Bl -tag -width Er .It Bq Er ENOMEM The .Dv MAKEDEV_NOWAIT flag was specified and a memory allocation request could not be satisfied. .It Bq Er ENAMETOOLONG The .Dv MAKEDEV_CHECKNAME flag was specified and the provided device name is longer than .Dv SPECNAMELEN . .It Bq Er EINVAL The .Dv MAKEDEV_CHECKNAME flag was specified and the provided device name is empty, contains a .Qq \&. or .Qq .. path component or ends with .Ql / . .It Bq Er EEXIST The .Dv MAKEDEV_CHECKNAME flag was specified and the provided device name already exists. .El .Pp .Sh SEE ALSO .Xr devctl 4 , +.Xr devfs 5 , .Xr destroy_dev_drain 9 , -.Xr dev_clone 9 , -.Xr devfs 5 +.Xr dev_clone 9 .Sh HISTORY The .Fn make_dev and .Fn destroy_dev functions first appeared in .Fx 4.0 . The function .Fn make_dev_alias first appeared in .Fx 4.1 . The function .Fn dev_depends first appeared in .Fx 5.0 . The functions .Fn make_dev_credf , .Fn destroy_dev_sched , .Fn destroy_dev_sched_cb first appeared in .Fx 7.0 . The function .Fn make_dev_p first appeared in .Fx 8.2 . Index: projects/largeSMP/share/mk/bsd.arch.inc.mk =================================================================== --- projects/largeSMP/share/mk/bsd.arch.inc.mk (revision 221494) +++ projects/largeSMP/share/mk/bsd.arch.inc.mk (revision 221495) Property changes on: projects/largeSMP/share/mk/bsd.arch.inc.mk ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/share/mk/bsd.arch.inc.mk:r221444-221493 Index: projects/largeSMP/share/zoneinfo =================================================================== --- projects/largeSMP/share/zoneinfo (revision 221494) +++ projects/largeSMP/share/zoneinfo (revision 221495) Property changes on: projects/largeSMP/share/zoneinfo ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/share/zoneinfo:r221444-221493 Index: projects/largeSMP/sys/amd64/include/xen =================================================================== --- projects/largeSMP/sys/amd64/include/xen (revision 221494) +++ projects/largeSMP/sys/amd64/include/xen (revision 221495) Property changes on: projects/largeSMP/sys/amd64/include/xen ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/amd64/include/xen:r221444-221493 Index: projects/largeSMP/sys/boot/i386/efi =================================================================== --- projects/largeSMP/sys/boot/i386/efi (revision 221494) +++ projects/largeSMP/sys/boot/i386/efi (revision 221495) Property changes on: projects/largeSMP/sys/boot/i386/efi ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/boot/i386/efi:r221444-221493 Index: projects/largeSMP/sys/boot/ia64/efi =================================================================== --- projects/largeSMP/sys/boot/ia64/efi (revision 221494) +++ projects/largeSMP/sys/boot/ia64/efi (revision 221495) Property changes on: projects/largeSMP/sys/boot/ia64/efi ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/boot/ia64/efi:r221444-221493 Index: projects/largeSMP/sys/boot/ia64/ski =================================================================== --- projects/largeSMP/sys/boot/ia64/ski (revision 221494) +++ projects/largeSMP/sys/boot/ia64/ski (revision 221495) Property changes on: projects/largeSMP/sys/boot/ia64/ski ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/boot/ia64/ski:r221444-221493 Index: projects/largeSMP/sys/boot/powerpc/boot1.chrp =================================================================== --- projects/largeSMP/sys/boot/powerpc/boot1.chrp (revision 221494) +++ projects/largeSMP/sys/boot/powerpc/boot1.chrp (revision 221495) Property changes on: projects/largeSMP/sys/boot/powerpc/boot1.chrp ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/boot/powerpc/boot1.chrp:r221444-221493 Index: projects/largeSMP/sys/boot/powerpc/ofw =================================================================== --- projects/largeSMP/sys/boot/powerpc/ofw (revision 221494) +++ projects/largeSMP/sys/boot/powerpc/ofw (revision 221495) Property changes on: projects/largeSMP/sys/boot/powerpc/ofw ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/boot/powerpc/ofw:r221444-221493 Index: projects/largeSMP/sys/boot =================================================================== --- projects/largeSMP/sys/boot (revision 221494) +++ projects/largeSMP/sys/boot (revision 221495) Property changes on: projects/largeSMP/sys/boot ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/boot:r221444-221493 Index: projects/largeSMP/sys/cddl/contrib/opensolaris =================================================================== --- projects/largeSMP/sys/cddl/contrib/opensolaris (revision 221494) +++ projects/largeSMP/sys/cddl/contrib/opensolaris (revision 221495) Property changes on: projects/largeSMP/sys/cddl/contrib/opensolaris ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/cddl/contrib/opensolaris:r221444-221493 Index: projects/largeSMP/sys/conf =================================================================== --- projects/largeSMP/sys/conf (revision 221494) +++ projects/largeSMP/sys/conf (revision 221495) Property changes on: projects/largeSMP/sys/conf ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/conf:r221444-221493 Index: projects/largeSMP/sys/contrib/dev/acpica =================================================================== --- projects/largeSMP/sys/contrib/dev/acpica (revision 221494) +++ projects/largeSMP/sys/contrib/dev/acpica (revision 221495) Property changes on: projects/largeSMP/sys/contrib/dev/acpica ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/contrib/dev/acpica:r221444-221493 Index: projects/largeSMP/sys/contrib/octeon-sdk =================================================================== --- projects/largeSMP/sys/contrib/octeon-sdk (revision 221494) +++ projects/largeSMP/sys/contrib/octeon-sdk (revision 221495) Property changes on: projects/largeSMP/sys/contrib/octeon-sdk ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/contrib/octeon-sdk:r221444-221493 Index: projects/largeSMP/sys/contrib/pf =================================================================== --- projects/largeSMP/sys/contrib/pf (revision 221494) +++ projects/largeSMP/sys/contrib/pf (revision 221495) Property changes on: projects/largeSMP/sys/contrib/pf ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/contrib/pf:r221444-221493 Index: projects/largeSMP/sys/contrib/x86emu =================================================================== --- projects/largeSMP/sys/contrib/x86emu (revision 221494) +++ projects/largeSMP/sys/contrib/x86emu (revision 221495) Property changes on: projects/largeSMP/sys/contrib/x86emu ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/contrib/x86emu:r221444-221493 Index: projects/largeSMP/sys/dev/ath/ath_hal/ar5416/ar2133.c =================================================================== --- projects/largeSMP/sys/dev/ath/ath_hal/ar5416/ar2133.c (revision 221494) +++ projects/largeSMP/sys/dev/ath/ath_hal/ar5416/ar2133.c (revision 221495) @@ -1,551 +1,551 @@ /* * Copyright (c) 2002-2009 Sam Leffler, Errno Consulting * Copyright (c) 2002-2008 Atheros Communications, Inc. * * Permission to use, copy, modify, and/or distribute this software for any * purpose with or without fee is hereby granted, provided that the above * copyright notice and this permission notice appear in all copies. * * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. * * $FreeBSD$ */ #include "opt_ah.h" #include "ah.h" #include "ah_internal.h" #include "ah_eeprom_v14.h" #include "ar5416/ar5416.h" #include "ar5416/ar5416reg.h" #include "ar5416/ar5416phy.h" #define N(a) (sizeof(a)/sizeof(a[0])) struct ar2133State { RF_HAL_FUNCS base; /* public state, must be first */ uint16_t pcdacTable[1]; uint32_t *Bank0Data; uint32_t *Bank1Data; uint32_t *Bank2Data; uint32_t *Bank3Data; uint32_t *Bank6Data; uint32_t *Bank7Data; /* NB: Bank*Data storage follows */ }; #define AR2133(ah) ((struct ar2133State *) AH5212(ah)->ah_rfHal) #define ar5416ModifyRfBuffer ar5212ModifyRfBuffer /*XXX*/ void ar5416ModifyRfBuffer(uint32_t *rfBuf, uint32_t reg32, uint32_t numBits, uint32_t firstBit, uint32_t column); static void ar2133WriteRegs(struct ath_hal *ah, u_int modesIndex, u_int freqIndex, int writes) { (void) ath_hal_ini_write(ah, &AH5416(ah)->ah_ini_bb_rfgain, freqIndex, writes); } /* * Fix on 2.4 GHz band for orientation sensitivity issue by increasing * rf_pwd_icsyndiv. * * Theoretical Rules: * if 2 GHz band * if forceBiasAuto * if synth_freq < 2412 * bias = 0 * else if 2412 <= synth_freq <= 2422 * bias = 1 * else // synth_freq > 2422 * bias = 2 * else if forceBias > 0 * bias = forceBias & 7 * else * no change, use value from ini file * else * no change, invalid band * * 1st Mod: * 2422 also uses value of 2 * * * 2nd Mod: * Less than 2412 uses value of 0, 2412 and above uses value of 2 */ static void ar2133ForceBias(struct ath_hal *ah, uint16_t synth_freq) { uint32_t tmp_reg; int reg_writes = 0; uint32_t new_bias = 0; struct ar2133State *priv = AR2133(ah); /* XXX this is a bit of a silly check for 2.4ghz channels -adrian */ if (synth_freq >= 3000) return; if (synth_freq < 2412) new_bias = 0; else if (synth_freq < 2422) new_bias = 1; else new_bias = 2; /* pre-reverse this field */ tmp_reg = ath_hal_reverseBits(new_bias, 3); HALDEBUG(ah, HAL_DEBUG_ANY, "%s: Force rf_pwd_icsyndiv to %1d on %4d\n", __func__, new_bias, synth_freq); /* swizzle rf_pwd_icsyndiv */ ar5416ModifyRfBuffer(priv->Bank6Data, tmp_reg, 3, 181, 3); /* write Bank 6 with new params */ ath_hal_ini_bank_write(ah, &AH5416(ah)->ah_ini_bank6, priv->Bank6Data, reg_writes); } /* * Take the MHz channel value and set the Channel value * * ASSUMES: Writes enabled to analog bus */ static HAL_BOOL ar2133SetChannel(struct ath_hal *ah, const struct ieee80211_channel *chan) { uint32_t channelSel = 0; uint32_t bModeSynth = 0; uint32_t aModeRefSel = 0; uint32_t reg32 = 0; uint16_t freq; CHAN_CENTERS centers; OS_MARK(ah, AH_MARK_SETCHANNEL, chan->ic_freq); ar5416GetChannelCenters(ah, chan, ¢ers); freq = centers.synth_center; if (freq < 4800) { uint32_t txctl; if (((freq - 2192) % 5) == 0) { channelSel = ((freq - 672) * 2 - 3040)/10; bModeSynth = 0; } else if (((freq - 2224) % 5) == 0) { channelSel = ((freq - 704) * 2 - 3040) / 10; bModeSynth = 1; } else { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: invalid channel %u MHz\n", __func__, freq); return AH_FALSE; } channelSel = (channelSel << 2) & 0xff; channelSel = ath_hal_reverseBits(channelSel, 8); txctl = OS_REG_READ(ah, AR_PHY_CCK_TX_CTRL); if (freq == 2484) { /* Enable channel spreading for channel 14 */ OS_REG_WRITE(ah, AR_PHY_CCK_TX_CTRL, txctl | AR_PHY_CCK_TX_CTRL_JAPAN); } else { OS_REG_WRITE(ah, AR_PHY_CCK_TX_CTRL, txctl &~ AR_PHY_CCK_TX_CTRL_JAPAN); } } else if ((freq % 20) == 0 && freq >= 5120) { channelSel = ath_hal_reverseBits(((freq - 4800) / 20 << 2), 8); - if (AR_SREV_SOWL_10_OR_LATER(ah)) + if (AR_SREV_HOWL(ah) || AR_SREV_SOWL_10_OR_LATER(ah)) aModeRefSel = ath_hal_reverseBits(3, 2); else aModeRefSel = ath_hal_reverseBits(1, 2); } else if ((freq % 10) == 0) { channelSel = ath_hal_reverseBits(((freq - 4800) / 10 << 1), 8); if (AR_SREV_HOWL(ah) || AR_SREV_SOWL_10_OR_LATER(ah)) aModeRefSel = ath_hal_reverseBits(2, 2); else aModeRefSel = ath_hal_reverseBits(1, 2); } else if ((freq % 5) == 0) { channelSel = ath_hal_reverseBits((freq - 4800) / 5, 8); aModeRefSel = ath_hal_reverseBits(1, 2); } else { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: invalid channel %u MHz\n", __func__, freq); return AH_FALSE; } /* Workaround for hw bug - AR5416 specific */ if (AR_SREV_OWL(ah) && ath_hal_ar5416_biasadj) ar2133ForceBias(ah, freq); reg32 = (channelSel << 8) | (aModeRefSel << 2) | (bModeSynth << 1) | (1 << 5) | 0x1; OS_REG_WRITE(ah, AR_PHY(0x37), reg32); AH_PRIVATE(ah)->ah_curchan = chan; return AH_TRUE; } /* * Return a reference to the requested RF Bank. */ static uint32_t * ar2133GetRfBank(struct ath_hal *ah, int bank) { struct ar2133State *priv = AR2133(ah); HALASSERT(priv != AH_NULL); switch (bank) { case 1: return priv->Bank1Data; case 2: return priv->Bank2Data; case 3: return priv->Bank3Data; case 6: return priv->Bank6Data; case 7: return priv->Bank7Data; } HALDEBUG(ah, HAL_DEBUG_ANY, "%s: unknown RF Bank %d requested\n", __func__, bank); return AH_NULL; } /* * Reads EEPROM header info from device structure and programs * all rf registers * * REQUIRES: Access to the analog rf device */ static HAL_BOOL ar2133SetRfRegs(struct ath_hal *ah, const struct ieee80211_channel *chan, uint16_t modesIndex, uint16_t *rfXpdGain) { struct ar2133State *priv = AR2133(ah); int writes; HALASSERT(priv); /* Setup Bank 0 Write */ ath_hal_ini_bank_setup(priv->Bank0Data, &AH5416(ah)->ah_ini_bank0, 1); /* Setup Bank 1 Write */ ath_hal_ini_bank_setup(priv->Bank1Data, &AH5416(ah)->ah_ini_bank1, 1); /* Setup Bank 2 Write */ ath_hal_ini_bank_setup(priv->Bank2Data, &AH5416(ah)->ah_ini_bank2, 1); /* Setup Bank 3 Write */ ath_hal_ini_bank_setup(priv->Bank3Data, &AH5416(ah)->ah_ini_bank3, modesIndex); /* Setup Bank 6 Write */ ath_hal_ini_bank_setup(priv->Bank6Data, &AH5416(ah)->ah_ini_bank6, modesIndex); /* Only the 5 or 2 GHz OB/DB need to be set for a mode */ if (IEEE80211_IS_CHAN_2GHZ(chan)) { HALDEBUG(ah, HAL_DEBUG_EEPROM, "%s: 2ghz: OB_2:%d, DB_2:%d\n", __func__, ath_hal_eepromGet(ah, AR_EEP_OB_2, AH_NULL), ath_hal_eepromGet(ah, AR_EEP_DB_2, AH_NULL)); ar5416ModifyRfBuffer(priv->Bank6Data, ath_hal_eepromGet(ah, AR_EEP_OB_2, AH_NULL), 3, 197, 0); ar5416ModifyRfBuffer(priv->Bank6Data, ath_hal_eepromGet(ah, AR_EEP_DB_2, AH_NULL), 3, 194, 0); } else { HALDEBUG(ah, HAL_DEBUG_EEPROM, "%s: 5ghz: OB_5:%d, DB_5:%d\n", __func__, ath_hal_eepromGet(ah, AR_EEP_OB_5, AH_NULL), ath_hal_eepromGet(ah, AR_EEP_DB_5, AH_NULL)); ar5416ModifyRfBuffer(priv->Bank6Data, ath_hal_eepromGet(ah, AR_EEP_OB_5, AH_NULL), 3, 203, 0); ar5416ModifyRfBuffer(priv->Bank6Data, ath_hal_eepromGet(ah, AR_EEP_DB_5, AH_NULL), 3, 200, 0); } /* Setup Bank 7 Setup */ ath_hal_ini_bank_setup(priv->Bank7Data, &AH5416(ah)->ah_ini_bank7, 1); /* Write Analog registers */ writes = ath_hal_ini_bank_write(ah, &AH5416(ah)->ah_ini_bank0, priv->Bank0Data, 0); writes = ath_hal_ini_bank_write(ah, &AH5416(ah)->ah_ini_bank1, priv->Bank1Data, writes); writes = ath_hal_ini_bank_write(ah, &AH5416(ah)->ah_ini_bank2, priv->Bank2Data, writes); writes = ath_hal_ini_bank_write(ah, &AH5416(ah)->ah_ini_bank3, priv->Bank3Data, writes); writes = ath_hal_ini_bank_write(ah, &AH5416(ah)->ah_ini_bank6, priv->Bank6Data, writes); (void) ath_hal_ini_bank_write(ah, &AH5416(ah)->ah_ini_bank7, priv->Bank7Data, writes); return AH_TRUE; #undef RF_BANK_SETUP } /* * Read the transmit power levels from the structures taken from EEPROM * Interpolate read transmit power values for this channel * Organize the transmit power values into a table for writing into the hardware */ static HAL_BOOL ar2133SetPowerTable(struct ath_hal *ah, int16_t *pPowerMin, int16_t *pPowerMax, const struct ieee80211_channel *chan, uint16_t *rfXpdGain) { return AH_TRUE; } #if 0 static int16_t ar2133GetMinPower(struct ath_hal *ah, EXPN_DATA_PER_CHANNEL_5112 *data) { int i, minIndex; int16_t minGain,minPwr,minPcdac,retVal; /* Assume NUM_POINTS_XPD0 > 0 */ minGain = data->pDataPerXPD[0].xpd_gain; for (minIndex=0,i=1; ipDataPerXPD[i].xpd_gain < minGain) { minIndex = i; minGain = data->pDataPerXPD[i].xpd_gain; } } minPwr = data->pDataPerXPD[minIndex].pwr_t4[0]; minPcdac = data->pDataPerXPD[minIndex].pcdac[0]; for (i=1; ipDataPerXPD[minIndex].pwr_t4[i] < minPwr) { minPwr = data->pDataPerXPD[minIndex].pwr_t4[i]; minPcdac = data->pDataPerXPD[minIndex].pcdac[i]; } } retVal = minPwr - (minPcdac*2); return(retVal); } #endif static HAL_BOOL ar2133GetChannelMaxMinPower(struct ath_hal *ah, const struct ieee80211_channel *chan, int16_t *maxPow, int16_t *minPow) { #if 0 struct ath_hal_5212 *ahp = AH5212(ah); int numChannels=0,i,last; int totalD, totalF,totalMin; EXPN_DATA_PER_CHANNEL_5112 *data=AH_NULL; EEPROM_POWER_EXPN_5112 *powerArray=AH_NULL; *maxPow = 0; if (IS_CHAN_A(chan)) { powerArray = ahp->ah_modePowerArray5112; data = powerArray[headerInfo11A].pDataPerChannel; numChannels = powerArray[headerInfo11A].numChannels; } else if (IS_CHAN_G(chan) || IS_CHAN_108G(chan)) { /* XXX - is this correct? Should we also use the same power for turbo G? */ powerArray = ahp->ah_modePowerArray5112; data = powerArray[headerInfo11G].pDataPerChannel; numChannels = powerArray[headerInfo11G].numChannels; } else if (IS_CHAN_B(chan)) { powerArray = ahp->ah_modePowerArray5112; data = powerArray[headerInfo11B].pDataPerChannel; numChannels = powerArray[headerInfo11B].numChannels; } else { return (AH_TRUE); } /* Make sure the channel is in the range of the TP values * (freq piers) */ if ((numChannels < 1) || (chan->channel < data[0].channelValue) || (chan->channel > data[numChannels-1].channelValue)) return(AH_FALSE); /* Linearly interpolate the power value now */ for (last=0,i=0; (ichannel > data[i].channelValue); last=i++); totalD = data[i].channelValue - data[last].channelValue; if (totalD > 0) { totalF = data[i].maxPower_t4 - data[last].maxPower_t4; *maxPow = (int8_t) ((totalF*(chan->channel-data[last].channelValue) + data[last].maxPower_t4*totalD)/totalD); totalMin = ar2133GetMinPower(ah,&data[i]) - ar2133GetMinPower(ah, &data[last]); *minPow = (int8_t) ((totalMin*(chan->channel-data[last].channelValue) + ar2133GetMinPower(ah, &data[last])*totalD)/totalD); return (AH_TRUE); } else { if (chan->channel == data[i].channelValue) { *maxPow = data[i].maxPower_t4; *minPow = ar2133GetMinPower(ah, &data[i]); return(AH_TRUE); } else return(AH_FALSE); } #else *maxPow = *minPow = 0; return AH_FALSE; #endif } /* * The ordering of nfarray is thus: * * nfarray[0]: Chain 0 ctl * nfarray[1]: Chain 1 ctl * nfarray[2]: Chain 2 ctl * nfarray[3]: Chain 0 ext * nfarray[4]: Chain 1 ext * nfarray[5]: Chain 2 ext */ static void ar2133GetNoiseFloor(struct ath_hal *ah, int16_t nfarray[]) { struct ath_hal_5416 *ahp = AH5416(ah); int16_t nf; /* * Blank nf array - some chips may only * have one or two RX chainmasks enabled. */ nfarray[0] = nfarray[1] = nfarray[2] = 0; nfarray[3] = nfarray[4] = nfarray[5] = 0; switch (ahp->ah_rx_chainmask) { case 0x7: nf = MS(OS_REG_READ(ah, AR_PHY_CH2_CCA), AR_PHY_CH2_MINCCA_PWR); if (nf & 0x100) nf = 0 - ((nf ^ 0x1ff) + 1); HALDEBUG(ah, HAL_DEBUG_NFCAL, "NF calibrated [ctl] [chain 2] is %d\n", nf); nfarray[2] = nf; nf = MS(OS_REG_READ(ah, AR_PHY_CH2_EXT_CCA), AR_PHY_CH2_EXT_MINCCA_PWR); if (nf & 0x100) nf = 0 - ((nf ^ 0x1ff) + 1); HALDEBUG(ah, HAL_DEBUG_NFCAL, "NF calibrated [ext] [chain 2] is %d\n", nf); nfarray[5] = nf; /* fall thru... */ case 0x3: case 0x5: nf = MS(OS_REG_READ(ah, AR_PHY_CH1_CCA), AR_PHY_CH1_MINCCA_PWR); if (nf & 0x100) nf = 0 - ((nf ^ 0x1ff) + 1); HALDEBUG(ah, HAL_DEBUG_NFCAL, "NF calibrated [ctl] [chain 1] is %d\n", nf); nfarray[1] = nf; nf = MS(OS_REG_READ(ah, AR_PHY_CH1_EXT_CCA), AR_PHY_CH1_EXT_MINCCA_PWR); if (nf & 0x100) nf = 0 - ((nf ^ 0x1ff) + 1); HALDEBUG(ah, HAL_DEBUG_NFCAL, "NF calibrated [ext] [chain 1] is %d\n", nf); nfarray[4] = nf; /* fall thru... */ case 0x1: nf = MS(OS_REG_READ(ah, AR_PHY_CCA), AR_PHY_MINCCA_PWR); if (nf & 0x100) nf = 0 - ((nf ^ 0x1ff) + 1); HALDEBUG(ah, HAL_DEBUG_NFCAL, "NF calibrated [ctl] [chain 0] is %d\n", nf); nfarray[0] = nf; nf = MS(OS_REG_READ(ah, AR_PHY_EXT_CCA), AR_PHY_EXT_MINCCA_PWR); if (nf & 0x100) nf = 0 - ((nf ^ 0x1ff) + 1); HALDEBUG(ah, HAL_DEBUG_NFCAL, "NF calibrated [ext] [chain 0] is %d\n", nf); nfarray[3] = nf; break; } } /* * Adjust NF based on statistical values for 5GHz frequencies. * Stubbed:Not used by Fowl */ static int16_t ar2133GetNfAdjust(struct ath_hal *ah, const HAL_CHANNEL_INTERNAL *c) { return 0; } /* * Free memory for analog bank scratch buffers */ static void ar2133RfDetach(struct ath_hal *ah) { struct ath_hal_5212 *ahp = AH5212(ah); HALASSERT(ahp->ah_rfHal != AH_NULL); ath_hal_free(ahp->ah_rfHal); ahp->ah_rfHal = AH_NULL; } /* * Allocate memory for analog bank scratch buffers * Scratch Buffer will be reinitialized every reset so no need to zero now */ HAL_BOOL ar2133RfAttach(struct ath_hal *ah, HAL_STATUS *status) { struct ath_hal_5212 *ahp = AH5212(ah); struct ar2133State *priv; uint32_t *bankData; HALDEBUG(ah, HAL_DEBUG_ATTACH, "%s: attach AR2133 radio\n", __func__); HALASSERT(ahp->ah_rfHal == AH_NULL); priv = ath_hal_malloc(sizeof(struct ar2133State) + AH5416(ah)->ah_ini_bank0.rows * sizeof(uint32_t) + AH5416(ah)->ah_ini_bank1.rows * sizeof(uint32_t) + AH5416(ah)->ah_ini_bank2.rows * sizeof(uint32_t) + AH5416(ah)->ah_ini_bank3.rows * sizeof(uint32_t) + AH5416(ah)->ah_ini_bank6.rows * sizeof(uint32_t) + AH5416(ah)->ah_ini_bank7.rows * sizeof(uint32_t) ); if (priv == AH_NULL) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: cannot allocate private state\n", __func__); *status = HAL_ENOMEM; /* XXX */ return AH_FALSE; } priv->base.rfDetach = ar2133RfDetach; priv->base.writeRegs = ar2133WriteRegs; priv->base.getRfBank = ar2133GetRfBank; priv->base.setChannel = ar2133SetChannel; priv->base.setRfRegs = ar2133SetRfRegs; priv->base.setPowerTable = ar2133SetPowerTable; priv->base.getChannelMaxMinPower = ar2133GetChannelMaxMinPower; priv->base.getNfAdjust = ar2133GetNfAdjust; bankData = (uint32_t *) &priv[1]; priv->Bank0Data = bankData, bankData += AH5416(ah)->ah_ini_bank0.rows; priv->Bank1Data = bankData, bankData += AH5416(ah)->ah_ini_bank1.rows; priv->Bank2Data = bankData, bankData += AH5416(ah)->ah_ini_bank2.rows; priv->Bank3Data = bankData, bankData += AH5416(ah)->ah_ini_bank3.rows; priv->Bank6Data = bankData, bankData += AH5416(ah)->ah_ini_bank6.rows; priv->Bank7Data = bankData, bankData += AH5416(ah)->ah_ini_bank7.rows; ahp->ah_pcdacTable = priv->pcdacTable; ahp->ah_pcdacTableSize = sizeof(priv->pcdacTable); ahp->ah_rfHal = &priv->base; /* * Set noise floor adjust method; we arrange a * direct call instead of thunking. */ AH_PRIVATE(ah)->ah_getNfAdjust = priv->base.getNfAdjust; AH_PRIVATE(ah)->ah_getNoiseFloor = ar2133GetNoiseFloor; return AH_TRUE; } Index: projects/largeSMP/sys/dev/ath/ath_hal/ar5416/ar5416_cal.c =================================================================== --- projects/largeSMP/sys/dev/ath/ath_hal/ar5416/ar5416_cal.c (revision 221494) +++ projects/largeSMP/sys/dev/ath/ath_hal/ar5416/ar5416_cal.c (revision 221495) @@ -1,821 +1,829 @@ /* * Copyright (c) 2002-2009 Sam Leffler, Errno Consulting * Copyright (c) 2002-2008 Atheros Communications, Inc. * * Permission to use, copy, modify, and/or distribute this software for any * purpose with or without fee is hereby granted, provided that the above * copyright notice and this permission notice appear in all copies. * * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. * * $FreeBSD$ */ #include "opt_ah.h" #include "ah.h" #include "ah_internal.h" #include "ah_devid.h" #include "ah_eeprom_v14.h" #include "ar5212/ar5212.h" /* for NF cal related declarations */ #include "ar5416/ar5416.h" #include "ar5416/ar5416reg.h" #include "ar5416/ar5416phy.h" /* Owl specific stuff */ #define NUM_NOISEFLOOR_READINGS 6 /* 3 chains * (ctl + ext) */ static void ar5416StartNFCal(struct ath_hal *ah); static void ar5416LoadNF(struct ath_hal *ah, const struct ieee80211_channel *); static int16_t ar5416GetNf(struct ath_hal *, struct ieee80211_channel *); static uint16_t ar5416GetDefaultNF(struct ath_hal *ah, const struct ieee80211_channel *chan); static void ar5416SanitizeNF(struct ath_hal *ah, int16_t *nf); /* * Determine if calibration is supported by device and channel flags */ /* * ADC GAIN/DC offset calibration is for calibrating two ADCs that * are acting as one by interleaving incoming symbols. This isn't * relevant for 2.4GHz 20MHz wide modes because, as far as I can tell, * the secondary ADC is never enabled. It is enabled however for * 5GHz modes. * * It hasn't been confirmed whether doing this calibration is needed * at all in the above modes and/or whether it's actually harmful. * So for now, let's leave it enabled and just remember to get * confirmation that it needs to be clarified. * * See US Patent No: US 7,541,952 B1: * " Method and Apparatus for Offset and Gain Compensation for * Analog-to-Digital Converters." */ static OS_INLINE HAL_BOOL ar5416IsCalSupp(struct ath_hal *ah, const struct ieee80211_channel *chan, HAL_CAL_TYPE calType) { struct ar5416PerCal *cal = &AH5416(ah)->ah_cal; switch (calType & cal->suppCals) { case IQ_MISMATCH_CAL: /* Run IQ Mismatch for non-CCK only */ return !IEEE80211_IS_CHAN_B(chan); case ADC_GAIN_CAL: case ADC_DC_CAL: /* Run ADC Gain Cal for either 5ghz any or 2ghz HT40 */ if (IEEE80211_IS_CHAN_5GHZ(chan)) return AH_TRUE; if (IEEE80211_IS_CHAN_HT40(chan)) return AH_TRUE; return AH_FALSE; } return AH_FALSE; } /* * Setup HW to collect samples used for current cal */ static void ar5416SetupMeasurement(struct ath_hal *ah, HAL_CAL_LIST *currCal) { /* Start calibration w/ 2^(INIT_IQCAL_LOG_COUNT_MAX+1) samples */ OS_REG_RMW_FIELD(ah, AR_PHY_TIMING_CTRL4, AR_PHY_TIMING_CTRL4_IQCAL_LOG_COUNT_MAX, currCal->calData->calCountMax); /* Select calibration to run */ switch (currCal->calData->calType) { case IQ_MISMATCH_CAL: OS_REG_WRITE(ah, AR_PHY_CALMODE, AR_PHY_CALMODE_IQ); HALDEBUG(ah, HAL_DEBUG_PERCAL, "%s: start IQ Mismatch calibration\n", __func__); break; case ADC_GAIN_CAL: OS_REG_WRITE(ah, AR_PHY_CALMODE, AR_PHY_CALMODE_ADC_GAIN); HALDEBUG(ah, HAL_DEBUG_PERCAL, "%s: start ADC Gain calibration\n", __func__); break; case ADC_DC_CAL: OS_REG_WRITE(ah, AR_PHY_CALMODE, AR_PHY_CALMODE_ADC_DC_PER); HALDEBUG(ah, HAL_DEBUG_PERCAL, "%s: start ADC DC calibration\n", __func__); break; case ADC_DC_INIT_CAL: OS_REG_WRITE(ah, AR_PHY_CALMODE, AR_PHY_CALMODE_ADC_DC_INIT); HALDEBUG(ah, HAL_DEBUG_PERCAL, "%s: start Init ADC DC calibration\n", __func__); break; } /* Kick-off cal */ OS_REG_SET_BIT(ah, AR_PHY_TIMING_CTRL4, AR_PHY_TIMING_CTRL4_DO_CAL); } /* * Initialize shared data structures and prepare a cal to be run. */ static void ar5416ResetMeasurement(struct ath_hal *ah, HAL_CAL_LIST *currCal) { struct ar5416PerCal *cal = &AH5416(ah)->ah_cal; /* Reset data structures shared between different calibrations */ OS_MEMZERO(cal->caldata, sizeof(cal->caldata)); cal->calSamples = 0; /* Setup HW for new calibration */ ar5416SetupMeasurement(ah, currCal); /* Change SW state to RUNNING for this calibration */ currCal->calState = CAL_RUNNING; } #if 0 /* * Run non-periodic calibrations. */ static HAL_BOOL ar5416RunInitCals(struct ath_hal *ah, int init_cal_count) { struct ath_hal_5416 *ahp = AH5416(ah); struct ar5416PerCal *cal = &AH5416(ah)->ah_cal; HAL_CHANNEL_INTERNAL ichan; /* XXX bogus */ HAL_CAL_LIST *curCal = ahp->ah_cal_curr; HAL_BOOL isCalDone; int i; if (curCal == AH_NULL) return AH_FALSE; ichan.calValid = 0; for (i = 0; i < init_cal_count; i++) { /* Reset this Cal */ ar5416ResetMeasurement(ah, curCal); /* Poll for offset calibration complete */ if (!ath_hal_wait(ah, AR_PHY_TIMING_CTRL4, AR_PHY_TIMING_CTRL4_DO_CAL, 0)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: Cal %d failed to finish in 100ms.\n", __func__, curCal->calData->calType); /* Re-initialize list pointers for periodic cals */ cal->cal_list = cal->cal_last = cal->cal_curr = AH_NULL; return AH_FALSE; } /* Run this cal */ ar5416DoCalibration(ah, &ichan, ahp->ah_rxchainmask, curCal, &isCalDone); if (!isCalDone) HALDEBUG(ah, HAL_DEBUG_ANY, "%s: init cal %d did not complete.\n", __func__, curCal->calData->calType); if (curCal->calNext != AH_NULL) curCal = curCal->calNext; } /* Re-initialize list pointers for periodic cals */ cal->cal_list = cal->cal_last = cal->cal_curr = AH_NULL; return AH_TRUE; } #endif HAL_BOOL ar5416InitCalHardware(struct ath_hal *ah, const struct ieee80211_channel *chan) { if (AR_SREV_MERLIN_10_OR_LATER(ah)) { /* Enable Rx Filter Cal */ OS_REG_CLR_BIT(ah, AR_PHY_ADC_CTL, AR_PHY_ADC_CTL_OFF_PWDADC); OS_REG_SET_BIT(ah, AR_PHY_AGC_CONTROL, AR_PHY_AGC_CONTROL_FLTR_CAL); /* Clear the carrier leak cal bit */ OS_REG_CLR_BIT(ah, AR_PHY_CL_CAL_CTL, AR_PHY_CL_CAL_ENABLE); /* kick off the cal */ OS_REG_SET_BIT(ah, AR_PHY_AGC_CONTROL, AR_PHY_AGC_CONTROL_CAL); /* Poll for offset calibration complete */ if (!ath_hal_wait(ah, AR_PHY_AGC_CONTROL, AR_PHY_AGC_CONTROL_CAL, 0)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: offset calibration failed to complete in 1ms; " "noisy environment?\n", __func__); return AH_FALSE; } /* Set the cl cal bit and rerun the cal a 2nd time */ /* Enable Rx Filter Cal */ OS_REG_CLR_BIT(ah, AR_PHY_ADC_CTL, AR_PHY_ADC_CTL_OFF_PWDADC); OS_REG_SET_BIT(ah, AR_PHY_AGC_CONTROL, AR_PHY_AGC_CONTROL_FLTR_CAL); OS_REG_SET_BIT(ah, AR_PHY_CL_CAL_CTL, AR_PHY_CL_CAL_ENABLE); } /* Calibrate the AGC */ OS_REG_SET_BIT(ah, AR_PHY_AGC_CONTROL, AR_PHY_AGC_CONTROL_CAL); /* Poll for offset calibration complete */ if (!ath_hal_wait(ah, AR_PHY_AGC_CONTROL, AR_PHY_AGC_CONTROL_CAL, 0)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: offset calibration did not complete in 1ms; " "noisy environment?\n", __func__); return AH_FALSE; } return AH_TRUE; } /* * Initialize Calibration infrastructure. */ #define MAX_CAL_CHECK 32 HAL_BOOL ar5416InitCal(struct ath_hal *ah, const struct ieee80211_channel *chan) { struct ar5416PerCal *cal = &AH5416(ah)->ah_cal; HAL_CHANNEL_INTERNAL *ichan; int i; ichan = ath_hal_checkchannel(ah, chan); HALASSERT(ichan != AH_NULL); /* Do initial chipset-specific calibration */ if (! AH5416(ah)->ah_cal_initcal(ah, chan)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: initial chipset calibration did " "not complete in time; noisy environment?\n", __func__); return AH_FALSE; } /* If there's PA Cal, do it */ if (AH5416(ah)->ah_cal_pacal) AH5416(ah)->ah_cal_pacal(ah, AH_TRUE); /* * Do NF calibration after DC offset and other CALs. * Per system engineers, noise floor value can sometimes be 20 dB * higher than normal value if DC offset and noise floor cal are * triggered at the same time. */ /* XXX this actually kicks off a NF calibration -adrian */ OS_REG_SET_BIT(ah, AR_PHY_AGC_CONTROL, AR_PHY_AGC_CONTROL_NF); /* * This sometimes takes a -lot- longer than it should. * Just give it a bit more time. */ for (i = 0; i < MAX_CAL_CHECK; i++) { if (ar5212WaitNFCalComplete(ah, 10000)) break; HALDEBUG(ah, HAL_DEBUG_ANY, "%s: initial NF calibration did " "not complete in time; noisy environment (pass %d)?\n", __func__, i); } /* * Although periodic and NF calibrations shouldn't run concurrently, * this was causing the radio to not be usable on the active * channel if the channel was busy. * * Instead, now simply print a warning and continue. That way if users * report "weird crap", they should get this warning. */ if (i >= MAX_CAL_CHECK) { ath_hal_printf(ah, "[ath] Warning - initial NF calibration did " "not complete in time, noisy environment?\n"); /* return AH_FALSE; */ } /* Initialize list pointers */ cal->cal_list = cal->cal_last = cal->cal_curr = AH_NULL; /* * Enable IQ, ADC Gain, ADC DC Offset Cals */ if (AR_SREV_SOWL_10_OR_LATER(ah)) { /* Setup all non-periodic, init time only calibrations */ /* XXX: Init DC Offset not working yet */ #if 0 if (ar5416IsCalSupp(ah, chan, ADC_DC_INIT_CAL)) { INIT_CAL(&cal->adcDcCalInitData); INSERT_CAL(cal, &cal->adcDcCalInitData); } /* Initialize current pointer to first element in list */ cal->cal_curr = cal->cal_list; if (cal->ah_cal_curr != AH_NULL && !ar5416RunInitCals(ah, 0)) return AH_FALSE; #endif } /* If Cals are supported, add them to list via INIT/INSERT_CAL */ if (ar5416IsCalSupp(ah, chan, ADC_GAIN_CAL)) { INIT_CAL(&cal->adcGainCalData); INSERT_CAL(cal, &cal->adcGainCalData); HALDEBUG(ah, HAL_DEBUG_PERCAL, "%s: enable ADC Gain Calibration.\n", __func__); } if (ar5416IsCalSupp(ah, chan, ADC_DC_CAL)) { INIT_CAL(&cal->adcDcCalData); INSERT_CAL(cal, &cal->adcDcCalData); HALDEBUG(ah, HAL_DEBUG_PERCAL, "%s: enable ADC DC Calibration.\n", __func__); } if (ar5416IsCalSupp(ah, chan, IQ_MISMATCH_CAL)) { INIT_CAL(&cal->iqCalData); INSERT_CAL(cal, &cal->iqCalData); HALDEBUG(ah, HAL_DEBUG_PERCAL, "%s: enable IQ Calibration.\n", __func__); } /* Initialize current pointer to first element in list */ cal->cal_curr = cal->cal_list; /* Kick off measurements for the first cal */ if (cal->cal_curr != AH_NULL) ar5416ResetMeasurement(ah, cal->cal_curr); /* Mark all calibrations on this channel as being invalid */ ichan->calValid = 0; return AH_TRUE; #undef MAX_CAL_CHECK } /* * Entry point for upper layers to restart current cal. * Reset the calibration valid bit in channel. */ HAL_BOOL ar5416ResetCalValid(struct ath_hal *ah, const struct ieee80211_channel *chan) { struct ar5416PerCal *cal = &AH5416(ah)->ah_cal; HAL_CHANNEL_INTERNAL *ichan = ath_hal_checkchannel(ah, chan); HAL_CAL_LIST *currCal = cal->cal_curr; if (!AR_SREV_SOWL_10_OR_LATER(ah)) return AH_FALSE; if (currCal == AH_NULL) return AH_FALSE; if (ichan == AH_NULL) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: invalid channel %u/0x%x; no mapping\n", __func__, chan->ic_freq, chan->ic_flags); return AH_FALSE; } /* * Expected that this calibration has run before, post-reset. * Current state should be done */ if (currCal->calState != CAL_DONE) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: Calibration state incorrect, %d\n", __func__, currCal->calState); return AH_FALSE; } /* Verify Cal is supported on this channel */ if (!ar5416IsCalSupp(ah, chan, currCal->calData->calType)) return AH_FALSE; HALDEBUG(ah, HAL_DEBUG_PERCAL, "%s: Resetting Cal %d state for channel %u/0x%x\n", __func__, currCal->calData->calType, chan->ic_freq, chan->ic_flags); /* Disable cal validity in channel */ ichan->calValid &= ~currCal->calData->calType; currCal->calState = CAL_WAITING; return AH_TRUE; } /* * Recalibrate the lower PHY chips to account for temperature/environment * changes. */ static void ar5416DoCalibration(struct ath_hal *ah, HAL_CHANNEL_INTERNAL *ichan, uint8_t rxchainmask, HAL_CAL_LIST *currCal, HAL_BOOL *isCalDone) { struct ar5416PerCal *cal = &AH5416(ah)->ah_cal; /* Cal is assumed not done until explicitly set below */ *isCalDone = AH_FALSE; HALDEBUG(ah, HAL_DEBUG_PERCAL, "%s: %s Calibration, state %d, calValid 0x%x\n", __func__, currCal->calData->calName, currCal->calState, ichan->calValid); /* Calibration in progress. */ if (currCal->calState == CAL_RUNNING) { /* Check to see if it has finished. */ if (!(OS_REG_READ(ah, AR_PHY_TIMING_CTRL4) & AR_PHY_TIMING_CTRL4_DO_CAL)) { HALDEBUG(ah, HAL_DEBUG_PERCAL, "%s: sample %d of %d finished\n", __func__, cal->calSamples, currCal->calData->calNumSamples); /* * Collect measurements for active chains. */ currCal->calData->calCollect(ah); if (++cal->calSamples >= currCal->calData->calNumSamples) { int i, numChains = 0; for (i = 0; i < AR5416_MAX_CHAINS; i++) { if (rxchainmask & (1 << i)) numChains++; } /* * Process accumulated data */ currCal->calData->calPostProc(ah, numChains); /* Calibration has finished. */ ichan->calValid |= currCal->calData->calType; currCal->calState = CAL_DONE; *isCalDone = AH_TRUE; } else { /* * Set-up to collect of another sub-sample. */ ar5416SetupMeasurement(ah, currCal); } } } else if (!(ichan->calValid & currCal->calData->calType)) { /* If current cal is marked invalid in channel, kick it off */ ar5416ResetMeasurement(ah, currCal); } } /* * Internal interface to schedule periodic calibration work. */ HAL_BOOL ar5416PerCalibrationN(struct ath_hal *ah, struct ieee80211_channel *chan, u_int rxchainmask, HAL_BOOL longcal, HAL_BOOL *isCalDone) { struct ar5416PerCal *cal = &AH5416(ah)->ah_cal; HAL_CAL_LIST *currCal = cal->cal_curr; HAL_CHANNEL_INTERNAL *ichan; OS_MARK(ah, AH_MARK_PERCAL, chan->ic_freq); *isCalDone = AH_TRUE; /* * Since ath_hal calls the PerCal method with rxchainmask=0x1; * override it with the current chainmask. The upper levels currently * doesn't know about the chainmask. */ rxchainmask = AH5416(ah)->ah_rx_chainmask; /* Invalid channel check */ ichan = ath_hal_checkchannel(ah, chan); if (ichan == AH_NULL) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: invalid channel %u/0x%x; no mapping\n", __func__, chan->ic_freq, chan->ic_flags); return AH_FALSE; } /* * For given calibration: * 1. Call generic cal routine * 2. When this cal is done (isCalDone) if we have more cals waiting * (eg after reset), mask this to upper layers by not propagating * isCalDone if it is set to TRUE. * Instead, change isCalDone to FALSE and setup the waiting cal(s) * to be run. */ if (currCal != AH_NULL && (currCal->calState == CAL_RUNNING || currCal->calState == CAL_WAITING)) { ar5416DoCalibration(ah, ichan, rxchainmask, currCal, isCalDone); if (*isCalDone == AH_TRUE) { cal->cal_curr = currCal = currCal->calNext; if (currCal->calState == CAL_WAITING) { *isCalDone = AH_FALSE; ar5416ResetMeasurement(ah, currCal); } } } /* Do NF cal only at longer intervals */ if (longcal) { /* Do PA calibration if the chipset supports */ if (AH5416(ah)->ah_cal_pacal) AH5416(ah)->ah_cal_pacal(ah, AH_FALSE); /* Do temperature compensation if the chipset needs it */ AH5416(ah)->ah_olcTempCompensation(ah); /* * Get the value from the previous NF cal * and update the history buffer. */ ar5416GetNf(ah, chan); /* * Load the NF from history buffer of the current channel. * NF is slow time-variant, so it is OK to use a * historical value. */ ar5416LoadNF(ah, AH_PRIVATE(ah)->ah_curchan); /* start NF calibration, without updating BB NF register*/ ar5416StartNFCal(ah); } return AH_TRUE; } /* * Recalibrate the lower PHY chips to account for temperature/environment * changes. */ HAL_BOOL ar5416PerCalibration(struct ath_hal *ah, struct ieee80211_channel *chan, HAL_BOOL *isIQdone) { struct ath_hal_5416 *ahp = AH5416(ah); struct ar5416PerCal *cal = &AH5416(ah)->ah_cal; HAL_CAL_LIST *curCal = cal->cal_curr; if (curCal != AH_NULL && curCal->calData->calType == IQ_MISMATCH_CAL) { return ar5416PerCalibrationN(ah, chan, ahp->ah_rx_chainmask, AH_TRUE, isIQdone); } else { HAL_BOOL isCalDone; *isIQdone = AH_FALSE; return ar5416PerCalibrationN(ah, chan, ahp->ah_rx_chainmask, AH_TRUE, &isCalDone); } } static HAL_BOOL ar5416GetEepromNoiseFloorThresh(struct ath_hal *ah, const struct ieee80211_channel *chan, int16_t *nft) { if (IEEE80211_IS_CHAN_5GHZ(chan)) { ath_hal_eepromGet(ah, AR_EEP_NFTHRESH_5, nft); return AH_TRUE; } if (IEEE80211_IS_CHAN_2GHZ(chan)) { ath_hal_eepromGet(ah, AR_EEP_NFTHRESH_2, nft); return AH_TRUE; } HALDEBUG(ah, HAL_DEBUG_ANY, "%s: invalid channel flags 0x%x\n", __func__, chan->ic_flags); return AH_FALSE; } static void ar5416StartNFCal(struct ath_hal *ah) { OS_REG_SET_BIT(ah, AR_PHY_AGC_CONTROL, AR_PHY_AGC_CONTROL_ENABLE_NF); OS_REG_SET_BIT(ah, AR_PHY_AGC_CONTROL, AR_PHY_AGC_CONTROL_NO_UPDATE_NF); OS_REG_SET_BIT(ah, AR_PHY_AGC_CONTROL, AR_PHY_AGC_CONTROL_NF); } static void ar5416LoadNF(struct ath_hal *ah, const struct ieee80211_channel *chan) { static const uint32_t ar5416_cca_regs[] = { AR_PHY_CCA, AR_PHY_CH1_CCA, AR_PHY_CH2_CCA, AR_PHY_EXT_CCA, AR_PHY_CH1_EXT_CCA, AR_PHY_CH2_EXT_CCA }; struct ar5212NfCalHist *h; int i; int32_t val; uint8_t chainmask; int16_t default_nf = ar5416GetDefaultNF(ah, chan); /* * Force NF calibration for all chains. */ if (AR_SREV_KITE(ah)) { /* Kite has only one chain */ chainmask = 0x9; } else if (AR_SREV_MERLIN(ah)) { /* Merlin has only two chains */ chainmask = 0x1B; } else { chainmask = 0x3F; } /* * Write filtered NF values into maxCCApwr register parameter * so we can load below. */ h = AH5416(ah)->ah_cal.nfCalHist; HALDEBUG(ah, HAL_DEBUG_NFCAL, "CCA: "); for (i = 0; i < AR5416_NUM_NF_READINGS; i ++) { - /* Don't write to EXT radio CCA registers */ + /* Don't write to EXT radio CCA registers unless in HT/40 mode */ /* XXX this check should really be cleaner! */ - if (i >= 3 && !IEEE80211_IS_CHAN_HT40(chan)) + if (i > 2 && !IEEE80211_IS_CHAN_HT40(chan)) continue; if (chainmask & (1 << i)) { int16_t nf_val; if (h) nf_val = h[i].privNF; else nf_val = default_nf; val = OS_REG_READ(ah, ar5416_cca_regs[i]); val &= 0xFFFFFE00; val |= (((uint32_t) nf_val << 1) & 0x1ff); HALDEBUG(ah, HAL_DEBUG_NFCAL, "[%d: %d]", i, nf_val); OS_REG_WRITE(ah, ar5416_cca_regs[i], val); } } HALDEBUG(ah, HAL_DEBUG_NFCAL, "\n"); /* Load software filtered NF value into baseband internal minCCApwr variable. */ OS_REG_CLR_BIT(ah, AR_PHY_AGC_CONTROL, AR_PHY_AGC_CONTROL_ENABLE_NF); OS_REG_CLR_BIT(ah, AR_PHY_AGC_CONTROL, AR_PHY_AGC_CONTROL_NO_UPDATE_NF); OS_REG_SET_BIT(ah, AR_PHY_AGC_CONTROL, AR_PHY_AGC_CONTROL_NF); /* Wait for load to complete, should be fast, a few 10s of us. */ if (! ar5212WaitNFCalComplete(ah, 1000)) { /* * We timed out waiting for the noisefloor to load, probably due to an * in-progress rx. Simply return here and allow the load plenty of time * to complete before the next calibration interval. We need to avoid * trying to load -50 (which happens below) while the previous load is * still in progress as this can cause rx deafness. Instead by returning * here, the baseband nf cal will just be capped by our present * noisefloor until the next calibration timer. */ HALDEBUG(ah, HAL_DEBUG_ANY, "Timeout while waiting for nf " "to load: AR_PHY_AGC_CONTROL=0x%x\n", OS_REG_READ(ah, AR_PHY_AGC_CONTROL)); return; } /* * Restore maxCCAPower register parameter again so that we're not capped * by the median we just loaded. This will be initial (and max) value * of next noise floor calibration the baseband does. */ for (i = 0; i < AR5416_NUM_NF_READINGS; i ++) + + /* Don't write to EXT radio CCA registers unless in HT/40 mode */ + /* XXX this check should really be cleaner! */ + if (i > 2 && !IEEE80211_IS_CHAN_HT40(chan)) + continue; + if (chainmask & (1 << i)) { val = OS_REG_READ(ah, ar5416_cca_regs[i]); val &= 0xFFFFFE00; val |= (((uint32_t)(-50) << 1) & 0x1ff); OS_REG_WRITE(ah, ar5416_cca_regs[i], val); } } /* * This just initialises the "good" values for AR5416 which * may not be right; it'lll be overridden by ar5416SanitizeNF() * to nominal values. */ void ar5416InitNfHistBuff(struct ar5212NfCalHist *h) { int i, j; for (i = 0; i < AR5416_NUM_NF_READINGS; i ++) { h[i].currIndex = 0; h[i].privNF = AR5416_CCA_MAX_GOOD_VALUE; h[i].invalidNFcount = AR512_NF_CAL_HIST_MAX; for (j = 0; j < AR512_NF_CAL_HIST_MAX; j ++) h[i].nfCalBuffer[j] = AR5416_CCA_MAX_GOOD_VALUE; } } /* * Update the noise floor buffer as a ring buffer */ static void -ar5416UpdateNFHistBuff(struct ar5212NfCalHist *h, int16_t *nfarray) +ar5416UpdateNFHistBuff(struct ath_hal *ah, struct ar5212NfCalHist *h, + int16_t *nfarray) { int i; + /* XXX TODO: don't record nfarray[] entries for inactive chains */ for (i = 0; i < AR5416_NUM_NF_READINGS; i ++) { h[i].nfCalBuffer[h[i].currIndex] = nfarray[i]; if (++h[i].currIndex >= AR512_NF_CAL_HIST_MAX) h[i].currIndex = 0; if (h[i].invalidNFcount > 0) { if (nfarray[i] < AR5416_CCA_MIN_BAD_VALUE || nfarray[i] > AR5416_CCA_MAX_HIGH_VALUE) { h[i].invalidNFcount = AR512_NF_CAL_HIST_MAX; } else { h[i].invalidNFcount--; h[i].privNF = nfarray[i]; } } else { h[i].privNF = ar5212GetNfHistMid(h[i].nfCalBuffer); } } } static uint16_t ar5416GetDefaultNF(struct ath_hal *ah, const struct ieee80211_channel *chan) { struct ar5416NfLimits *limit; if (!chan || IEEE80211_IS_CHAN_2GHZ(chan)) limit = &AH5416(ah)->nf_2g; else limit = &AH5416(ah)->nf_5g; return limit->nominal; } static void ar5416SanitizeNF(struct ath_hal *ah, int16_t *nf) { struct ar5416NfLimits *limit; int i; if (IEEE80211_IS_CHAN_2GHZ(AH_PRIVATE(ah)->ah_curchan)) limit = &AH5416(ah)->nf_2g; else limit = &AH5416(ah)->nf_5g; for (i = 0; i < AR5416_NUM_NF_READINGS; i++) { if (!nf[i]) continue; if (nf[i] > limit->max) { HALDEBUG(ah, HAL_DEBUG_NFCAL, "NF[%d] (%d) > MAX (%d), correcting to MAX\n", i, nf[i], limit->max); nf[i] = limit->max; } else if (nf[i] < limit->min) { HALDEBUG(ah, HAL_DEBUG_NFCAL, "NF[%d] (%d) < MIN (%d), correcting to NOM\n", i, nf[i], limit->min); nf[i] = limit->nominal; } } } /* * Read the NF and check it against the noise floor threshhold */ static int16_t ar5416GetNf(struct ath_hal *ah, struct ieee80211_channel *chan) { int16_t nf, nfThresh; int i; if (ar5212IsNFCalInProgress(ah)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: NF didn't complete in calibration window\n", __func__); nf = 0; } else { /* Finished NF cal, check against threshold */ int16_t nfarray[NUM_NOISEFLOOR_READINGS] = { 0 }; HAL_CHANNEL_INTERNAL *ichan = ath_hal_checkchannel(ah, chan); /* TODO - enhance for multiple chains and ext ch */ ath_hal_getNoiseFloor(ah, nfarray); nf = nfarray[0]; ar5416SanitizeNF(ah, nfarray); if (ar5416GetEepromNoiseFloorThresh(ah, chan, &nfThresh)) { if (nf > nfThresh) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: noise floor failed detected; " "detected %d, threshold %d\n", __func__, nf, nfThresh); /* * NB: Don't discriminate 2.4 vs 5Ghz, if this * happens it indicates a problem regardless * of the band. */ chan->ic_state |= IEEE80211_CHANSTATE_CWINT; nf = 0; } } else { nf = 0; } /* Update MIMO channel statistics, regardless of validity or not (for now) */ for (i = 0; i < 3; i++) { ichan->noiseFloorCtl[i] = nfarray[i]; ichan->noiseFloorExt[i] = nfarray[i + 3]; } ichan->privFlags |= CHANNEL_MIMO_NF_VALID; - ar5416UpdateNFHistBuff(AH5416(ah)->ah_cal.nfCalHist, nfarray); + ar5416UpdateNFHistBuff(ah, AH5416(ah)->ah_cal.nfCalHist, nfarray); ichan->rawNoiseFloor = nf; } return nf; } Index: projects/largeSMP/sys/dev/ath/ath_hal/ar5416/ar5416_reset.c =================================================================== --- projects/largeSMP/sys/dev/ath/ath_hal/ar5416/ar5416_reset.c (revision 221494) +++ projects/largeSMP/sys/dev/ath/ath_hal/ar5416/ar5416_reset.c (revision 221495) @@ -1,2584 +1,2606 @@ /* * Copyright (c) 2002-2009 Sam Leffler, Errno Consulting * Copyright (c) 2002-2008 Atheros Communications, Inc. * * Permission to use, copy, modify, and/or distribute this software for any * purpose with or without fee is hereby granted, provided that the above * copyright notice and this permission notice appear in all copies. * * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. * * $FreeBSD$ */ #include "opt_ah.h" #include "ah.h" #include "ah_internal.h" #include "ah_devid.h" #include "ah_eeprom_v14.h" #include "ar5416/ar5416.h" #include "ar5416/ar5416reg.h" #include "ar5416/ar5416phy.h" /* Eeprom versioning macros. Returns true if the version is equal or newer than the ver specified */ #define EEP_MINOR(_ah) \ (AH_PRIVATE(_ah)->ah_eeversion & AR5416_EEP_VER_MINOR_MASK) #define IS_EEP_MINOR_V2(_ah) (EEP_MINOR(_ah) >= AR5416_EEP_MINOR_VER_2) #define IS_EEP_MINOR_V3(_ah) (EEP_MINOR(_ah) >= AR5416_EEP_MINOR_VER_3) /* Additional Time delay to wait after activiting the Base band */ #define BASE_ACTIVATE_DELAY 100 /* 100 usec */ #define PLL_SETTLE_DELAY 300 /* 300 usec */ #define RTC_PLL_SETTLE_DELAY 1000 /* 1 ms */ static void ar5416InitDMA(struct ath_hal *ah); static void ar5416InitBB(struct ath_hal *ah, const struct ieee80211_channel *); static void ar5416InitIMR(struct ath_hal *ah, HAL_OPMODE opmode); static void ar5416InitQoS(struct ath_hal *ah); static void ar5416InitUserSettings(struct ath_hal *ah); static void ar5416UpdateChainMasks(struct ath_hal *ah, HAL_BOOL is_ht); static void ar5416OverrideIni(struct ath_hal *ah, const struct ieee80211_channel *); #if 0 static HAL_BOOL ar5416ChannelChange(struct ath_hal *, const struct ieee80211_channel *); #endif static void ar5416SetDeltaSlope(struct ath_hal *, const struct ieee80211_channel *); static HAL_BOOL ar5416SetResetPowerOn(struct ath_hal *ah); static HAL_BOOL ar5416SetReset(struct ath_hal *ah, int type); static HAL_BOOL ar5416SetPowerPerRateTable(struct ath_hal *ah, struct ar5416eeprom *pEepData, const struct ieee80211_channel *chan, int16_t *ratesArray, uint16_t cfgCtl, uint16_t AntennaReduction, uint16_t twiceMaxRegulatoryPower, uint16_t powerLimit); static void ar5416Set11nRegs(struct ath_hal *ah, const struct ieee80211_channel *chan); static void ar5416MarkPhyInactive(struct ath_hal *ah); /* * Places the device in and out of reset and then places sane * values in the registers based on EEPROM config, initialization * vectors (as determined by the mode), and station configuration * * bChannelChange is used to preserve DMA/PCU registers across * a HW Reset during channel change. */ HAL_BOOL ar5416Reset(struct ath_hal *ah, HAL_OPMODE opmode, struct ieee80211_channel *chan, HAL_BOOL bChannelChange, HAL_STATUS *status) { #define N(a) (sizeof (a) / sizeof (a[0])) #define FAIL(_code) do { ecode = _code; goto bad; } while (0) struct ath_hal_5212 *ahp = AH5212(ah); HAL_CHANNEL_INTERNAL *ichan; uint32_t saveDefAntenna, saveLedState; uint32_t macStaId1; uint16_t rfXpdGain[2]; HAL_STATUS ecode; uint32_t powerVal, rssiThrReg; uint32_t ackTpcPow, ctsTpcPow, chirpTpcPow; int i; uint64_t tsf = 0; OS_MARK(ah, AH_MARK_RESET, bChannelChange); /* Bring out of sleep mode */ if (!ar5416SetPowerMode(ah, HAL_PM_AWAKE, AH_TRUE)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: chip did not wakeup\n", __func__); FAIL(HAL_EIO); } /* * Map public channel to private. */ ichan = ath_hal_checkchannel(ah, chan); if (ichan == AH_NULL) FAIL(HAL_EINVAL); switch (opmode) { case HAL_M_STA: case HAL_M_IBSS: case HAL_M_HOSTAP: case HAL_M_MONITOR: break; default: HALDEBUG(ah, HAL_DEBUG_ANY, "%s: invalid operating mode %u\n", __func__, opmode); FAIL(HAL_EINVAL); break; } HALASSERT(AH_PRIVATE(ah)->ah_eeversion >= AR_EEPROM_VER14_1); /* XXX Turn on fast channel change for 5416 */ /* * Preserve the bmiss rssi threshold and count threshold * across resets */ rssiThrReg = OS_REG_READ(ah, AR_RSSI_THR); /* If reg is zero, first time thru set to default val */ if (rssiThrReg == 0) rssiThrReg = INIT_RSSI_THR; /* * Preserve the antenna on a channel change */ saveDefAntenna = OS_REG_READ(ah, AR_DEF_ANTENNA); if (saveDefAntenna == 0) /* XXX magic constants */ saveDefAntenna = 1; /* Save hardware flag before chip reset clears the register */ macStaId1 = OS_REG_READ(ah, AR_STA_ID1) & (AR_STA_ID1_BASE_RATE_11B | AR_STA_ID1_USE_DEFANT); /* Save led state from pci config register */ saveLedState = OS_REG_READ(ah, AR_MAC_LED) & (AR_MAC_LED_ASSOC | AR_MAC_LED_MODE | AR_MAC_LED_BLINK_THRESH_SEL | AR_MAC_LED_BLINK_SLOW); /* For chips on which the RTC reset is done, save TSF before it gets cleared */ if (AR_SREV_HOWL(ah) || (AR_SREV_MERLIN_20_OR_LATER(ah) && ath_hal_eepromGetFlag(ah, AR_EEP_OL_PWRCTRL))) tsf = ar5212GetTsf64(ah); /* Mark PHY as inactive; marked active in ar5416InitBB() */ ar5416MarkPhyInactive(ah); if (!ar5416ChipReset(ah, chan)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: chip reset failed\n", __func__); FAIL(HAL_EIO); } /* Restore bmiss rssi & count thresholds */ OS_REG_WRITE(ah, AR_RSSI_THR, rssiThrReg); /* Restore TSF */ if (tsf) ar5212SetTsf64(ah, tsf); OS_MARK(ah, AH_MARK_RESET_LINE, __LINE__); if (AR_SREV_MERLIN_10_OR_LATER(ah)) OS_REG_SET_BIT(ah, AR_GPIO_INPUT_EN_VAL, AR_GPIO_JTAG_DISABLE); AH5416(ah)->ah_writeIni(ah, chan); /* Override ini values (that can be overriden in this fashion) */ ar5416OverrideIni(ah, chan); /* Setup 11n MAC/Phy mode registers */ ar5416Set11nRegs(ah, chan); OS_MARK(ah, AH_MARK_RESET_LINE, __LINE__); /* * Some AR91xx SoC devices frequently fail to accept TSF writes * right after the chip reset. When that happens, write a new * value after the initvals have been applied, with an offset * based on measured time difference */ if (AR_SREV_HOWL(ah) && (ar5212GetTsf64(ah) < tsf)) { tsf += 1500; ar5212SetTsf64(ah, tsf); } HALDEBUG(ah, HAL_DEBUG_RESET, ">>>2 %s: AR_PHY_DAG_CTRLCCK=0x%x\n", __func__, OS_REG_READ(ah,AR_PHY_DAG_CTRLCCK)); HALDEBUG(ah, HAL_DEBUG_RESET, ">>>2 %s: AR_PHY_ADC_CTL=0x%x\n", __func__, OS_REG_READ(ah,AR_PHY_ADC_CTL)); /* * Setup ah_tx_chainmask / ah_rx_chainmask before we fiddle * with enabling the TX/RX radio chains. */ ar5416UpdateChainMasks(ah, IEEE80211_IS_CHAN_HT(chan)); /* * This routine swaps the analog chains - it should be done * before any radio register twiddling is done. */ ar5416InitChainMasks(ah); /* Setup the open-loop temperature compensation if required */ AH5416(ah)->ah_olcInit(ah); /* Setup the transmit power values. */ if (!ah->ah_setTxPower(ah, chan, rfXpdGain)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: error init'ing transmit power\n", __func__); FAIL(HAL_EIO); } /* Write the analog registers */ if (!ahp->ah_rfHal->setRfRegs(ah, chan, IEEE80211_IS_CHAN_2GHZ(chan) ? 2: 1, rfXpdGain)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: ar5212SetRfRegs failed\n", __func__); FAIL(HAL_EIO); } /* Write delta slope for OFDM enabled modes (A, G, Turbo) */ if (IEEE80211_IS_CHAN_OFDM(chan)|| IEEE80211_IS_CHAN_HT(chan)) ar5416SetDeltaSlope(ah, chan); AH5416(ah)->ah_spurMitigate(ah, chan); /* Setup board specific options for EEPROM version 3 */ if (!ah->ah_setBoardValues(ah, chan)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: error setting board options\n", __func__); FAIL(HAL_EIO); } OS_MARK(ah, AH_MARK_RESET_LINE, __LINE__); OS_REG_WRITE(ah, AR_STA_ID0, LE_READ_4(ahp->ah_macaddr)); OS_REG_WRITE(ah, AR_STA_ID1, LE_READ_2(ahp->ah_macaddr + 4) | macStaId1 | AR_STA_ID1_RTS_USE_DEF | ahp->ah_staId1Defaults ); ar5212SetOperatingMode(ah, opmode); /* Set Venice BSSID mask according to current state */ OS_REG_WRITE(ah, AR_BSSMSKL, LE_READ_4(ahp->ah_bssidmask)); OS_REG_WRITE(ah, AR_BSSMSKU, LE_READ_2(ahp->ah_bssidmask + 4)); /* Restore previous led state */ - OS_REG_WRITE(ah, AR_MAC_LED, OS_REG_READ(ah, AR_MAC_LED) | saveLedState); + if (AR_SREV_HOWL(ah)) + OS_REG_WRITE(ah, AR_MAC_LED, + AR_MAC_LED_ASSOC_ACTIVE | AR_CFG_SCLK_32KHZ); + else + OS_REG_WRITE(ah, AR_MAC_LED, OS_REG_READ(ah, AR_MAC_LED) | + saveLedState); /* Restore previous antenna */ OS_REG_WRITE(ah, AR_DEF_ANTENNA, saveDefAntenna); /* then our BSSID */ OS_REG_WRITE(ah, AR_BSS_ID0, LE_READ_4(ahp->ah_bssid)); OS_REG_WRITE(ah, AR_BSS_ID1, LE_READ_2(ahp->ah_bssid + 4)); /* Restore bmiss rssi & count thresholds */ OS_REG_WRITE(ah, AR_RSSI_THR, ahp->ah_rssiThr); OS_REG_WRITE(ah, AR_ISR, ~0); /* cleared on write */ if (!ar5212SetChannel(ah, chan)) FAIL(HAL_EIO); OS_MARK(ah, AH_MARK_RESET_LINE, __LINE__); /* Set 1:1 QCU to DCU mapping for all queues */ for (i = 0; i < AR_NUM_DCU; i++) OS_REG_WRITE(ah, AR_DQCUMASK(i), 1 << i); ahp->ah_intrTxqs = 0; for (i = 0; i < AH_PRIVATE(ah)->ah_caps.halTotalQueues; i++) ah->ah_resetTxQueue(ah, i); ar5416InitIMR(ah, opmode); ar5212SetCoverageClass(ah, AH_PRIVATE(ah)->ah_coverageClass, 1); ar5416InitQoS(ah); ar5416InitUserSettings(ah); /* * disable seq number generation in hw */ OS_REG_WRITE(ah, AR_STA_ID1, OS_REG_READ(ah, AR_STA_ID1) | AR_STA_ID1_PRESERVE_SEQNUM); ar5416InitDMA(ah); /* * program OBS bus to see MAC interrupts */ OS_REG_WRITE(ah, AR_OBS, 8); #ifdef AH_AR5416_INTERRUPT_MITIGATION OS_REG_WRITE(ah, AR_MIRT, 0); OS_REG_RMW_FIELD(ah, AR_RIMT, AR_RIMT_LAST, 500); OS_REG_RMW_FIELD(ah, AR_RIMT, AR_RIMT_FIRST, 2000); OS_REG_RMW_FIELD(ah, AR_TIMT, AR_TIMT_LAST, 300); OS_REG_RMW_FIELD(ah, AR_TIMT, AR_TIMT_FIRST, 750); #endif ar5416InitBB(ah, chan); /* Setup compression registers */ ar5212SetCompRegs(ah); /* XXX not needed? */ /* * 5416 baseband will check the per rate power table * and select the lower of the two */ ackTpcPow = 63; ctsTpcPow = 63; chirpTpcPow = 63; powerVal = SM(ackTpcPow, AR_TPC_ACK) | SM(ctsTpcPow, AR_TPC_CTS) | SM(chirpTpcPow, AR_TPC_CHIRP); OS_REG_WRITE(ah, AR_TPC, powerVal); if (!ar5416InitCal(ah, chan)) FAIL(HAL_ESELFTEST); ar5416RestoreChainMask(ah); AH_PRIVATE(ah)->ah_opmode = opmode; /* record operating mode */ if (bChannelChange && !IEEE80211_IS_CHAN_DFS(chan)) chan->ic_state &= ~IEEE80211_CHANSTATE_CWINT; + if (AR_SREV_HOWL(ah)) { + /* + * Enable the MBSSID block-ack fix for HOWL. + * This feature is only supported on Howl 1.4, but it is safe to + * set bit 22 of STA_ID1 on other Howl revisions (1.1, 1.2, 1.3), + * since bit 22 is unused in those Howl revisions. + */ + unsigned int reg; + reg = (OS_REG_READ(ah, AR_STA_ID1) | (1<<22)); + OS_REG_WRITE(ah,AR_STA_ID1, reg); + ath_hal_printf(ah, "MBSSID Set bit 22 of AR_STA_ID 0x%x\n", reg); + } + HALDEBUG(ah, HAL_DEBUG_RESET, "%s: done\n", __func__); OS_MARK(ah, AH_MARK_RESET_DONE, 0); return AH_TRUE; bad: OS_MARK(ah, AH_MARK_RESET_DONE, ecode); if (status != AH_NULL) *status = ecode; return AH_FALSE; #undef FAIL #undef N } #if 0 /* * This channel change evaluates whether the selected hardware can * perform a synthesizer-only channel change (no reset). If the * TX is not stopped, or the RFBus cannot be granted in the given * time, the function returns false as a reset is necessary */ HAL_BOOL ar5416ChannelChange(struct ath_hal *ah, const structu ieee80211_channel *chan) { uint32_t ulCount; uint32_t data, synthDelay, qnum; uint16_t rfXpdGain[4]; struct ath_hal_5212 *ahp = AH5212(ah); HAL_CHANNEL_INTERNAL *ichan; /* * Map public channel to private. */ ichan = ath_hal_checkchannel(ah, chan); /* TX must be stopped or RF Bus grant will not work */ for (qnum = 0; qnum < AH_PRIVATE(ah)->ah_caps.halTotalQueues; qnum++) { if (ar5212NumTxPending(ah, qnum)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: frames pending on queue %d\n", __func__, qnum); return AH_FALSE; } } /* * Kill last Baseband Rx Frame - Request analog bus grant */ OS_REG_WRITE(ah, AR_PHY_RFBUS_REQ, AR_PHY_RFBUS_REQ_REQUEST); if (!ath_hal_wait(ah, AR_PHY_RFBUS_GNT, AR_PHY_RFBUS_GRANT_EN, AR_PHY_RFBUS_GRANT_EN)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: could not kill baseband rx\n", __func__); return AH_FALSE; } ar5416Set11nRegs(ah, chan); /* NB: setup 5416-specific regs */ /* Change the synth */ if (!ar5212SetChannel(ah, chan)) return AH_FALSE; /* Setup the transmit power values. */ if (!ah->ah_setTxPower(ah, chan, rfXpdGain)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: error init'ing transmit power\n", __func__); return AH_FALSE; } /* * Wait for the frequency synth to settle (synth goes on * via PHY_ACTIVE_EN). Read the phy active delay register. * Value is in 100ns increments. */ data = OS_REG_READ(ah, AR_PHY_RX_DELAY) & AR_PHY_RX_DELAY_DELAY; if (IS_CHAN_CCK(ichan)) { synthDelay = (4 * data) / 22; } else { synthDelay = data / 10; } OS_DELAY(synthDelay + BASE_ACTIVATE_DELAY); /* Release the RFBus Grant */ OS_REG_WRITE(ah, AR_PHY_RFBUS_REQ, 0); /* Write delta slope for OFDM enabled modes (A, G, Turbo) */ if (IEEE80211_IS_CHAN_OFDM(ichan)|| IEEE80211_IS_CHAN_HT(chan)) { HALASSERT(AH_PRIVATE(ah)->ah_eeversion >= AR_EEPROM_VER5_3); ar5212SetSpurMitigation(ah, chan); ar5416SetDeltaSlope(ah, chan); } /* XXX spur mitigation for Melin */ if (!IEEE80211_IS_CHAN_DFS(chan)) chan->ic_state &= ~IEEE80211_CHANSTATE_CWINT; ichan->channel_time = 0; ichan->tsf_last = ar5212GetTsf64(ah); ar5212TxEnable(ah, AH_TRUE); return AH_TRUE; } #endif static void ar5416InitDMA(struct ath_hal *ah) { struct ath_hal_5212 *ahp = AH5212(ah); /* * set AHB_MODE not to do cacheline prefetches */ OS_REG_SET_BIT(ah, AR_AHB_MODE, AR_AHB_PREFETCH_RD_EN); /* * let mac dma reads be in 128 byte chunks */ OS_REG_WRITE(ah, AR_TXCFG, (OS_REG_READ(ah, AR_TXCFG) & ~AR_TXCFG_DMASZ_MASK) | AR_TXCFG_DMASZ_128B); /* * let mac dma writes be in 128 byte chunks */ OS_REG_WRITE(ah, AR_RXCFG, (OS_REG_READ(ah, AR_RXCFG) & ~AR_RXCFG_DMASZ_MASK) | AR_RXCFG_DMASZ_128B); /* restore TX trigger level */ OS_REG_WRITE(ah, AR_TXCFG, (OS_REG_READ(ah, AR_TXCFG) &~ AR_FTRIG) | SM(ahp->ah_txTrigLev, AR_FTRIG)); /* * Setup receive FIFO threshold to hold off TX activities */ OS_REG_WRITE(ah, AR_RXFIFO_CFG, 0x200); /* * reduce the number of usable entries in PCU TXBUF to avoid * wrap around. */ if (AR_SREV_KITE(ah)) /* * For AR9285 the number of Fifos are reduced to half. * So set the usable tx buf size also to half to * avoid data/delimiter underruns */ OS_REG_WRITE(ah, AR_PCU_TXBUF_CTRL, AR_9285_PCU_TXBUF_CTRL_USABLE_SIZE); else OS_REG_WRITE(ah, AR_PCU_TXBUF_CTRL, AR_PCU_TXBUF_CTRL_USABLE_SIZE); } static void ar5416InitBB(struct ath_hal *ah, const struct ieee80211_channel *chan) { uint32_t synthDelay; /* * Wait for the frequency synth to settle (synth goes on * via AR_PHY_ACTIVE_EN). Read the phy active delay register. * Value is in 100ns increments. */ synthDelay = OS_REG_READ(ah, AR_PHY_RX_DELAY) & AR_PHY_RX_DELAY_DELAY; if (IEEE80211_IS_CHAN_CCK(chan)) { synthDelay = (4 * synthDelay) / 22; } else { synthDelay /= 10; } /* Turn on PLL on 5416 */ HALDEBUG(ah, HAL_DEBUG_RESET, "%s %s channel\n", __func__, IEEE80211_IS_CHAN_5GHZ(chan) ? "5GHz" : "2GHz"); AH5416(ah)->ah_initPLL(ah, chan); /* Activate the PHY (includes baseband activate and synthesizer on) */ OS_REG_WRITE(ah, AR_PHY_ACTIVE, AR_PHY_ACTIVE_EN); /* * If the AP starts the calibration before the base band timeout * completes we could get rx_clear false triggering. Add an * extra BASE_ACTIVATE_DELAY usecs to ensure this condition * does not happen. */ if (IEEE80211_IS_CHAN_HALF(chan)) { OS_DELAY((synthDelay << 1) + BASE_ACTIVATE_DELAY); } else if (IEEE80211_IS_CHAN_QUARTER(chan)) { OS_DELAY((synthDelay << 2) + BASE_ACTIVATE_DELAY); } else { OS_DELAY(synthDelay + BASE_ACTIVATE_DELAY); } } static void ar5416InitIMR(struct ath_hal *ah, HAL_OPMODE opmode) { struct ath_hal_5212 *ahp = AH5212(ah); /* * Setup interrupt handling. Note that ar5212ResetTxQueue * manipulates the secondary IMR's as queues are enabled * and disabled. This is done with RMW ops to insure the * settings we make here are preserved. */ ahp->ah_maskReg = AR_IMR_TXERR | AR_IMR_TXURN | AR_IMR_RXERR | AR_IMR_RXORN | AR_IMR_BCNMISC; #ifdef AH_AR5416_INTERRUPT_MITIGATION ahp->ah_maskReg |= AR_IMR_TXINTM | AR_IMR_RXINTM | AR_IMR_TXMINTR | AR_IMR_RXMINTR; #else ahp->ah_maskReg |= AR_IMR_TXOK | AR_IMR_RXOK; #endif if (opmode == HAL_M_HOSTAP) ahp->ah_maskReg |= AR_IMR_MIB; OS_REG_WRITE(ah, AR_IMR, ahp->ah_maskReg); #ifdef ADRIAN_NOTYET /* This is straight from ath9k */ if (! AR_SREV_HOWL(ah)) { OS_REG_WRITE(ah, AR_INTR_SYNC_CAUSE, 0xFFFFFFFF); OS_REG_WRITE(ah, AR_INTR_SYNC_ENABLE, AR_INTR_SYNC_DEFAULT); OS_REG_WRITE(ah, AR_INTR_SYNC_MASK, 0); } #endif /* Enable bus errors that are OR'd to set the HIUERR bit */ #if 0 OS_REG_WRITE(ah, AR_IMR_S2, OS_REG_READ(ah, AR_IMR_S2) | AR_IMR_S2_GTT | AR_IMR_S2_CST); #endif } static void ar5416InitQoS(struct ath_hal *ah) { /* QoS support */ OS_REG_WRITE(ah, AR_QOS_CONTROL, 0x100aa); /* XXX magic */ OS_REG_WRITE(ah, AR_QOS_SELECT, 0x3210); /* XXX magic */ /* Turn on NOACK Support for QoS packets */ OS_REG_WRITE(ah, AR_NOACK, SM(2, AR_NOACK_2BIT_VALUE) | SM(5, AR_NOACK_BIT_OFFSET) | SM(0, AR_NOACK_BYTE_OFFSET)); /* * initialize TXOP for all TIDs */ OS_REG_WRITE(ah, AR_TXOP_X, AR_TXOP_X_VAL); OS_REG_WRITE(ah, AR_TXOP_0_3, 0xFFFFFFFF); OS_REG_WRITE(ah, AR_TXOP_4_7, 0xFFFFFFFF); OS_REG_WRITE(ah, AR_TXOP_8_11, 0xFFFFFFFF); OS_REG_WRITE(ah, AR_TXOP_12_15, 0xFFFFFFFF); } static void ar5416InitUserSettings(struct ath_hal *ah) { struct ath_hal_5212 *ahp = AH5212(ah); /* Restore user-specified settings */ if (ahp->ah_miscMode != 0) OS_REG_WRITE(ah, AR_MISC_MODE, OS_REG_READ(ah, AR_MISC_MODE) | ahp->ah_miscMode); if (ahp->ah_sifstime != (u_int) -1) ar5212SetSifsTime(ah, ahp->ah_sifstime); if (ahp->ah_slottime != (u_int) -1) ar5212SetSlotTime(ah, ahp->ah_slottime); if (ahp->ah_acktimeout != (u_int) -1) ar5212SetAckTimeout(ah, ahp->ah_acktimeout); if (ahp->ah_ctstimeout != (u_int) -1) ar5212SetCTSTimeout(ah, ahp->ah_ctstimeout); if (AH_PRIVATE(ah)->ah_diagreg != 0) OS_REG_WRITE(ah, AR_DIAG_SW, AH_PRIVATE(ah)->ah_diagreg); if (AH5416(ah)->ah_globaltxtimeout != (u_int) -1) ar5416SetGlobalTxTimeout(ah, AH5416(ah)->ah_globaltxtimeout); } static void ar5416SetRfMode(struct ath_hal *ah, const struct ieee80211_channel *chan) { uint32_t rfMode; if (chan == AH_NULL) return; /* treat channel B as channel G , no B mode suport in owl */ rfMode = IEEE80211_IS_CHAN_CCK(chan) ? AR_PHY_MODE_DYNAMIC : AR_PHY_MODE_OFDM; if (AR_SREV_MERLIN_20(ah) && IS_5GHZ_FAST_CLOCK_EN(ah, chan)) { /* phy mode bits for 5GHz channels require Fast Clock */ rfMode |= AR_PHY_MODE_DYNAMIC | AR_PHY_MODE_DYN_CCK_DISABLE; } else if (!AR_SREV_MERLIN_10_OR_LATER(ah)) { rfMode |= IEEE80211_IS_CHAN_5GHZ(chan) ? AR_PHY_MODE_RF5GHZ : AR_PHY_MODE_RF2GHZ; } OS_REG_WRITE(ah, AR_PHY_MODE, rfMode); } /* * Places the hardware into reset and then pulls it out of reset */ HAL_BOOL ar5416ChipReset(struct ath_hal *ah, const struct ieee80211_channel *chan) { OS_MARK(ah, AH_MARK_CHIPRESET, chan ? chan->ic_freq : 0); /* * Warm reset is optimistic. */ if (AR_SREV_MERLIN_20_OR_LATER(ah) && ath_hal_eepromGetFlag(ah, AR_EEP_OL_PWRCTRL)) { if (!ar5416SetResetReg(ah, HAL_RESET_POWER_ON)) return AH_FALSE; } else { if (!ar5416SetResetReg(ah, HAL_RESET_WARM)) return AH_FALSE; } /* Bring out of sleep mode (AGAIN) */ if (!ar5416SetPowerMode(ah, HAL_PM_AWAKE, AH_TRUE)) return AH_FALSE; AH5416(ah)->ah_initPLL(ah, chan); /* * Perform warm reset before the mode/PLL/turbo registers * are changed in order to deactivate the radio. Mode changes * with an active radio can result in corrupted shifts to the * radio device. */ if (chan != AH_NULL) ar5416SetRfMode(ah, chan); return AH_TRUE; } /* * Delta slope coefficient computation. * Required for OFDM operation. */ static void ar5416GetDeltaSlopeValues(struct ath_hal *ah, uint32_t coef_scaled, uint32_t *coef_mantissa, uint32_t *coef_exponent) { #define COEF_SCALE_S 24 uint32_t coef_exp, coef_man; /* * ALGO -> coef_exp = 14-floor(log2(coef)); * floor(log2(x)) is the highest set bit position */ for (coef_exp = 31; coef_exp > 0; coef_exp--) if ((coef_scaled >> coef_exp) & 0x1) break; /* A coef_exp of 0 is a legal bit position but an unexpected coef_exp */ HALASSERT(coef_exp); coef_exp = 14 - (coef_exp - COEF_SCALE_S); /* * ALGO -> coef_man = floor(coef* 2^coef_exp+0.5); * The coefficient is already shifted up for scaling */ coef_man = coef_scaled + (1 << (COEF_SCALE_S - coef_exp - 1)); *coef_mantissa = coef_man >> (COEF_SCALE_S - coef_exp); *coef_exponent = coef_exp - 16; #undef COEF_SCALE_S } void ar5416SetDeltaSlope(struct ath_hal *ah, const struct ieee80211_channel *chan) { #define INIT_CLOCKMHZSCALED 0x64000000 uint32_t coef_scaled, ds_coef_exp, ds_coef_man; uint32_t clockMhzScaled; CHAN_CENTERS centers; /* half and quarter rate can divide the scaled clock by 2 or 4 respectively */ /* scale for selected channel bandwidth */ clockMhzScaled = INIT_CLOCKMHZSCALED; if (IEEE80211_IS_CHAN_TURBO(chan)) clockMhzScaled <<= 1; else if (IEEE80211_IS_CHAN_HALF(chan)) clockMhzScaled >>= 1; else if (IEEE80211_IS_CHAN_QUARTER(chan)) clockMhzScaled >>= 2; /* * ALGO -> coef = 1e8/fcarrier*fclock/40; * scaled coef to provide precision for this floating calculation */ ar5416GetChannelCenters(ah, chan, ¢ers); coef_scaled = clockMhzScaled / centers.synth_center; ar5416GetDeltaSlopeValues(ah, coef_scaled, &ds_coef_man, &ds_coef_exp); OS_REG_RMW_FIELD(ah, AR_PHY_TIMING3, AR_PHY_TIMING3_DSC_MAN, ds_coef_man); OS_REG_RMW_FIELD(ah, AR_PHY_TIMING3, AR_PHY_TIMING3_DSC_EXP, ds_coef_exp); /* * For Short GI, * scaled coeff is 9/10 that of normal coeff */ coef_scaled = (9 * coef_scaled)/10; ar5416GetDeltaSlopeValues(ah, coef_scaled, &ds_coef_man, &ds_coef_exp); /* for short gi */ OS_REG_RMW_FIELD(ah, AR_PHY_HALFGI, AR_PHY_HALFGI_DSC_MAN, ds_coef_man); OS_REG_RMW_FIELD(ah, AR_PHY_HALFGI, AR_PHY_HALFGI_DSC_EXP, ds_coef_exp); #undef INIT_CLOCKMHZSCALED } /* * Set a limit on the overall output power. Used for dynamic * transmit power control and the like. * * NB: limit is in units of 0.5 dbM. */ HAL_BOOL ar5416SetTxPowerLimit(struct ath_hal *ah, uint32_t limit) { uint16_t dummyXpdGains[2]; AH_PRIVATE(ah)->ah_powerLimit = AH_MIN(limit, MAX_RATE_POWER); return ah->ah_setTxPower(ah, AH_PRIVATE(ah)->ah_curchan, dummyXpdGains); } HAL_BOOL ar5416GetChipPowerLimits(struct ath_hal *ah, struct ieee80211_channel *chan) { struct ath_hal_5212 *ahp = AH5212(ah); int16_t minPower, maxPower; /* * Get Pier table max and min powers. */ if (ahp->ah_rfHal->getChannelMaxMinPower(ah, chan, &maxPower, &minPower)) { /* NB: rf code returns 1/4 dBm units, convert */ chan->ic_maxpower = maxPower / 2; chan->ic_minpower = minPower / 2; } else { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: no min/max power for %u/0x%x\n", __func__, chan->ic_freq, chan->ic_flags); chan->ic_maxpower = AR5416_MAX_RATE_POWER; chan->ic_minpower = 0; } HALDEBUG(ah, HAL_DEBUG_RESET, "Chan %d: MaxPow = %d MinPow = %d\n", chan->ic_freq, chan->ic_maxpower, chan->ic_minpower); return AH_TRUE; } /************************************************************** * ar5416WriteTxPowerRateRegisters * * Write the TX power rate registers from the raw values given * in ratesArray[]. * * The CCK and HT40 rate registers are only written if needed. * HT20 and 11g/11a OFDM rate registers are always written. * * The values written are raw values which should be written * to the registers - so it's up to the caller to pre-adjust * them (eg CCK power offset value, or Merlin TX power offset, * etc.) */ void ar5416WriteTxPowerRateRegisters(struct ath_hal *ah, const struct ieee80211_channel *chan, const int16_t ratesArray[]) { #define POW_SM(_r, _s) (((_r) & 0x3f) << (_s)) /* Write the OFDM power per rate set */ OS_REG_WRITE(ah, AR_PHY_POWER_TX_RATE1, POW_SM(ratesArray[rate18mb], 24) | POW_SM(ratesArray[rate12mb], 16) | POW_SM(ratesArray[rate9mb], 8) | POW_SM(ratesArray[rate6mb], 0) ); OS_REG_WRITE(ah, AR_PHY_POWER_TX_RATE2, POW_SM(ratesArray[rate54mb], 24) | POW_SM(ratesArray[rate48mb], 16) | POW_SM(ratesArray[rate36mb], 8) | POW_SM(ratesArray[rate24mb], 0) ); if (IEEE80211_IS_CHAN_2GHZ(chan)) { /* Write the CCK power per rate set */ OS_REG_WRITE(ah, AR_PHY_POWER_TX_RATE3, POW_SM(ratesArray[rate2s], 24) | POW_SM(ratesArray[rate2l], 16) | POW_SM(ratesArray[rateXr], 8) /* XR target power */ | POW_SM(ratesArray[rate1l], 0) ); OS_REG_WRITE(ah, AR_PHY_POWER_TX_RATE4, POW_SM(ratesArray[rate11s], 24) | POW_SM(ratesArray[rate11l], 16) | POW_SM(ratesArray[rate5_5s], 8) | POW_SM(ratesArray[rate5_5l], 0) ); HALDEBUG(ah, HAL_DEBUG_RESET, "%s AR_PHY_POWER_TX_RATE3=0x%x AR_PHY_POWER_TX_RATE4=0x%x\n", __func__, OS_REG_READ(ah,AR_PHY_POWER_TX_RATE3), OS_REG_READ(ah,AR_PHY_POWER_TX_RATE4)); } /* Write the HT20 power per rate set */ OS_REG_WRITE(ah, AR_PHY_POWER_TX_RATE5, POW_SM(ratesArray[rateHt20_3], 24) | POW_SM(ratesArray[rateHt20_2], 16) | POW_SM(ratesArray[rateHt20_1], 8) | POW_SM(ratesArray[rateHt20_0], 0) ); OS_REG_WRITE(ah, AR_PHY_POWER_TX_RATE6, POW_SM(ratesArray[rateHt20_7], 24) | POW_SM(ratesArray[rateHt20_6], 16) | POW_SM(ratesArray[rateHt20_5], 8) | POW_SM(ratesArray[rateHt20_4], 0) ); if (IEEE80211_IS_CHAN_HT40(chan)) { /* Write the HT40 power per rate set */ OS_REG_WRITE(ah, AR_PHY_POWER_TX_RATE7, POW_SM(ratesArray[rateHt40_3], 24) | POW_SM(ratesArray[rateHt40_2], 16) | POW_SM(ratesArray[rateHt40_1], 8) | POW_SM(ratesArray[rateHt40_0], 0) ); OS_REG_WRITE(ah, AR_PHY_POWER_TX_RATE8, POW_SM(ratesArray[rateHt40_7], 24) | POW_SM(ratesArray[rateHt40_6], 16) | POW_SM(ratesArray[rateHt40_5], 8) | POW_SM(ratesArray[rateHt40_4], 0) ); /* Write the Dup/Ext 40 power per rate set */ OS_REG_WRITE(ah, AR_PHY_POWER_TX_RATE9, POW_SM(ratesArray[rateExtOfdm], 24) | POW_SM(ratesArray[rateExtCck], 16) | POW_SM(ratesArray[rateDupOfdm], 8) | POW_SM(ratesArray[rateDupCck], 0) ); } } /************************************************************** * ar5416SetTransmitPower * * Set the transmit power in the baseband for the given * operating channel and mode. */ HAL_BOOL ar5416SetTransmitPower(struct ath_hal *ah, const struct ieee80211_channel *chan, uint16_t *rfXpdGain) { #define N(a) (sizeof (a) / sizeof (a[0])) MODAL_EEP_HEADER *pModal; struct ath_hal_5212 *ahp = AH5212(ah); int16_t ratesArray[Ar5416RateSize]; int16_t txPowerIndexOffset = 0; uint8_t ht40PowerIncForPdadc = 2; int i; uint16_t cfgCtl; uint16_t powerLimit; uint16_t twiceAntennaReduction; uint16_t twiceMaxRegulatoryPower; int16_t maxPower; HAL_EEPROM_v14 *ee = AH_PRIVATE(ah)->ah_eeprom; struct ar5416eeprom *pEepData = &ee->ee_base; HALASSERT(AH_PRIVATE(ah)->ah_eeversion >= AR_EEPROM_VER14_1); /* Setup info for the actual eeprom */ OS_MEMZERO(ratesArray, sizeof(ratesArray)); cfgCtl = ath_hal_getctl(ah, chan); powerLimit = chan->ic_maxregpower * 2; twiceAntennaReduction = chan->ic_maxantgain; twiceMaxRegulatoryPower = AH_MIN(MAX_RATE_POWER, AH_PRIVATE(ah)->ah_powerLimit); pModal = &pEepData->modalHeader[IEEE80211_IS_CHAN_2GHZ(chan)]; HALDEBUG(ah, HAL_DEBUG_RESET, "%s Channel=%u CfgCtl=%u\n", __func__,chan->ic_freq, cfgCtl ); if (IS_EEP_MINOR_V2(ah)) { ht40PowerIncForPdadc = pModal->ht40PowerIncForPdadc; } if (!ar5416SetPowerPerRateTable(ah, pEepData, chan, &ratesArray[0],cfgCtl, twiceAntennaReduction, twiceMaxRegulatoryPower, powerLimit)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: unable to set tx power per rate table\n", __func__); return AH_FALSE; } if (!AH5416(ah)->ah_setPowerCalTable(ah, pEepData, chan, &txPowerIndexOffset)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: unable to set power table\n", __func__); return AH_FALSE; } maxPower = AH_MAX(ratesArray[rate6mb], ratesArray[rateHt20_0]); if (IEEE80211_IS_CHAN_2GHZ(chan)) { maxPower = AH_MAX(maxPower, ratesArray[rate1l]); } if (IEEE80211_IS_CHAN_HT40(chan)) { maxPower = AH_MAX(maxPower, ratesArray[rateHt40_0]); } ahp->ah_tx6PowerInHalfDbm = maxPower; AH_PRIVATE(ah)->ah_maxPowerLevel = maxPower; ahp->ah_txPowerIndexOffset = txPowerIndexOffset; /* * txPowerIndexOffset is set by the SetPowerTable() call - * adjust the rate table (0 offset if rates EEPROM not loaded) */ for (i = 0; i < N(ratesArray); i++) { ratesArray[i] = (int16_t)(txPowerIndexOffset + ratesArray[i]); if (ratesArray[i] > AR5416_MAX_RATE_POWER) ratesArray[i] = AR5416_MAX_RATE_POWER; } #ifdef AH_EEPROM_DUMP /* * Dump the rate array whilst it represents the intended dBm*2 * values versus what's being adjusted before being programmed * in. Keep this in mind if you code up this function and enable * this debugging; the values won't necessarily be what's being * programmed into the hardware. */ ar5416PrintPowerPerRate(ah, ratesArray); #endif /* * Merlin and later have a power offset, so subtract * pwr_table_offset * 2 from each value. The default * power offset is -5 dBm - ie, a register value of 0 * equates to a TX power of -5 dBm. */ if (AR_SREV_MERLIN_20_OR_LATER(ah)) { int8_t pwr_table_offset; (void) ath_hal_eepromGet(ah, AR_EEP_PWR_TABLE_OFFSET, &pwr_table_offset); /* Underflow power gets clamped at raw value 0 */ /* Overflow power gets camped at AR5416_MAX_RATE_POWER */ for (i = 0; i < N(ratesArray); i++) { /* * + pwr_table_offset is in dBm * + ratesArray is in 1/2 dBm */ ratesArray[i] -= (pwr_table_offset * 2); if (ratesArray[i] < 0) ratesArray[i] = 0; else if (ratesArray[i] > AR5416_MAX_RATE_POWER) ratesArray[i] = AR5416_MAX_RATE_POWER; } } /* * Adjust rates for OLC where needed * * The following CCK rates need adjusting when doing 2.4ghz * CCK transmission. * * + rate2s, rate2l, rate1l, rate11s, rate11l, rate5_5s, rate5_5l * + rateExtCck, rateDupCck * * They're adjusted here regardless. The hardware then gets * programmed as needed. 5GHz operation doesn't program in CCK * rates for legacy mode but they seem to be initialised for * HT40 regardless of channel type. */ if (AR_SREV_MERLIN_20_OR_LATER(ah) && ath_hal_eepromGetFlag(ah, AR_EEP_OL_PWRCTRL)) { int adj[] = { rate2s, rate2l, rate1l, rate11s, rate11l, rate5_5s, rate5_5l, rateExtCck, rateDupCck }; int cck_ofdm_delta = 2; int i; for (i = 0; i < N(adj); i++) { ratesArray[i] -= cck_ofdm_delta; if (ratesArray[i] < 0) ratesArray[i] = 0; } } /* * Adjust the HT40 power to meet the correct target TX power * for 40MHz mode, based on TX power curves that are established * for 20MHz mode. * * XXX handle overflow/too high power level? */ if (IEEE80211_IS_CHAN_HT40(chan)) { ratesArray[rateHt40_0] += ht40PowerIncForPdadc; ratesArray[rateHt40_1] += ht40PowerIncForPdadc; ratesArray[rateHt40_2] += ht40PowerIncForPdadc; ratesArray[rateHt40_3] += ht40PowerIncForPdadc; ratesArray[rateHt40_4] += ht40PowerIncForPdadc; ratesArray[rateHt40_5] += ht40PowerIncForPdadc; ratesArray[rateHt40_6] += ht40PowerIncForPdadc; ratesArray[rateHt40_7] += ht40PowerIncForPdadc; } /* Write the TX power rate registers */ ar5416WriteTxPowerRateRegisters(ah, chan, ratesArray); /* Write the Power subtraction for dynamic chain changing, for per-packet powertx */ OS_REG_WRITE(ah, AR_PHY_POWER_TX_SUB, POW_SM(pModal->pwrDecreaseFor3Chain, 6) | POW_SM(pModal->pwrDecreaseFor2Chain, 0) ); return AH_TRUE; #undef POW_SM #undef N } /* * Exported call to check for a recent gain reading and return * the current state of the thermal calibration gain engine. */ HAL_RFGAIN ar5416GetRfgain(struct ath_hal *ah) { return HAL_RFGAIN_INACTIVE; } /* * Places all of hardware into reset */ HAL_BOOL ar5416Disable(struct ath_hal *ah) { if (!ar5212SetPowerMode(ah, HAL_PM_AWAKE, AH_TRUE)) return AH_FALSE; return ar5416SetResetReg(ah, HAL_RESET_COLD); } /* * Places the PHY and Radio chips into reset. A full reset * must be called to leave this state. The PCI/MAC/PCU are * not placed into reset as we must receive interrupt to * re-enable the hardware. */ HAL_BOOL ar5416PhyDisable(struct ath_hal *ah) { return ar5416SetResetReg(ah, HAL_RESET_WARM); } /* * Write the given reset bit mask into the reset register */ HAL_BOOL ar5416SetResetReg(struct ath_hal *ah, uint32_t type) { switch (type) { case HAL_RESET_POWER_ON: return ar5416SetResetPowerOn(ah); case HAL_RESET_WARM: case HAL_RESET_COLD: return ar5416SetReset(ah, type); default: HALASSERT(AH_FALSE); return AH_FALSE; } } static HAL_BOOL ar5416SetResetPowerOn(struct ath_hal *ah) { /* Power On Reset (Hard Reset) */ /* * Set force wake * * If the MAC was running, previously calling * reset will wake up the MAC but it may go back to sleep * before we can start polling. * Set force wake stops that * This must be called before initiating a hard reset. */ OS_REG_WRITE(ah, AR_RTC_FORCE_WAKE, AR_RTC_FORCE_WAKE_EN | AR_RTC_FORCE_WAKE_ON_INT); /* * RTC reset and clear */ if (! AR_SREV_HOWL(ah)) OS_REG_WRITE(ah, AR_RC, AR_RC_AHB); OS_REG_WRITE(ah, AR_RTC_RESET, 0); OS_DELAY(20); if (! AR_SREV_HOWL(ah)) OS_REG_WRITE(ah, AR_RC, 0); OS_REG_WRITE(ah, AR_RTC_RESET, 1); /* * Poll till RTC is ON */ if (!ath_hal_wait(ah, AR_RTC_STATUS, AR_RTC_PM_STATUS_M, AR_RTC_STATUS_ON)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: RTC not waking up\n", __func__); return AH_FALSE; } return ar5416SetReset(ah, HAL_RESET_COLD); } static HAL_BOOL ar5416SetReset(struct ath_hal *ah, int type) { uint32_t tmpReg, mask; uint32_t rst_flags; #ifdef AH_SUPPORT_AR9130 /* Because of the AR9130 specific registers */ if (AR_SREV_HOWL(ah)) { HALDEBUG(ah, HAL_DEBUG_ANY, "[ath] HOWL: Fiddling with derived clk!\n"); uint32_t val = OS_REG_READ(ah, AR_RTC_DERIVED_CLK); val &= ~AR_RTC_DERIVED_CLK_PERIOD; val |= SM(1, AR_RTC_DERIVED_CLK_PERIOD); OS_REG_WRITE(ah, AR_RTC_DERIVED_CLK, val); (void) OS_REG_READ(ah, AR_RTC_DERIVED_CLK); } #endif /* AH_SUPPORT_AR9130 */ /* * Force wake */ OS_REG_WRITE(ah, AR_RTC_FORCE_WAKE, AR_RTC_FORCE_WAKE_EN | AR_RTC_FORCE_WAKE_ON_INT); #ifdef AH_SUPPORT_AR9130 if (AR_SREV_HOWL(ah)) { rst_flags = AR_RTC_RC_MAC_WARM | AR_RTC_RC_MAC_COLD | AR_RTC_RC_COLD_RESET | AR_RTC_RC_WARM_RESET; } else { #endif /* AH_SUPPORT_AR9130 */ /* * Reset AHB */ tmpReg = OS_REG_READ(ah, AR_INTR_SYNC_CAUSE); if (tmpReg & (AR_INTR_SYNC_LOCAL_TIMEOUT|AR_INTR_SYNC_RADM_CPL_TIMEOUT)) { OS_REG_WRITE(ah, AR_INTR_SYNC_ENABLE, 0); OS_REG_WRITE(ah, AR_RC, AR_RC_AHB|AR_RC_HOSTIF); } else { OS_REG_WRITE(ah, AR_RC, AR_RC_AHB); } rst_flags = AR_RTC_RC_MAC_WARM; if (type == HAL_RESET_COLD) rst_flags |= AR_RTC_RC_MAC_COLD; #ifdef AH_SUPPORT_AR9130 } #endif /* AH_SUPPORT_AR9130 */ OS_REG_WRITE(ah, AR_RTC_RC, rst_flags); - OS_DELAY(50); + + if (AR_SREV_HOWL(ah)) + OS_DELAY(10000); + else + OS_DELAY(100); /* * Clear resets and force wakeup */ OS_REG_WRITE(ah, AR_RTC_RC, 0); if (!ath_hal_wait(ah, AR_RTC_RC, AR_RTC_RC_M, 0)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: RTC stuck in MAC reset\n", __func__); return AH_FALSE; } /* Clear AHB reset */ if (! AR_SREV_HOWL(ah)) OS_REG_WRITE(ah, AR_RC, 0); if (AR_SREV_HOWL(ah)) OS_DELAY(50); if (AR_SREV_HOWL(ah)) { uint32_t mask; mask = OS_REG_READ(ah, AR_CFG); if (mask & (AR_CFG_SWRB | AR_CFG_SWTB | AR_CFG_SWRG)) { HALDEBUG(ah, HAL_DEBUG_RESET, "CFG Byte Swap Set 0x%x\n", mask); } else { mask = INIT_CONFIG_STATUS | AR_CFG_SWRB | AR_CFG_SWTB; OS_REG_WRITE(ah, AR_CFG, mask); HALDEBUG(ah, HAL_DEBUG_RESET, "Setting CFG 0x%x\n", OS_REG_READ(ah, AR_CFG)); } } else { if (type == HAL_RESET_COLD) { if (isBigEndian()) { /* * Set CFG, little-endian for register * and descriptor accesses. */ mask = INIT_CONFIG_STATUS | AR_CFG_SWRD | AR_CFG_SWRG; #ifndef AH_NEED_DESC_SWAP mask |= AR_CFG_SWTD; #endif HALDEBUG(ah, HAL_DEBUG_RESET, "%s Applying descriptor swap\n", __func__); OS_REG_WRITE(ah, AR_CFG, LE_READ_4(&mask)); } else OS_REG_WRITE(ah, AR_CFG, INIT_CONFIG_STATUS); } } AH5416(ah)->ah_initPLL(ah, AH_NULL); return AH_TRUE; } void ar5416InitChainMasks(struct ath_hal *ah) { if (AH5416(ah)->ah_rx_chainmask == 0x5 || AH5416(ah)->ah_tx_chainmask == 0x5) OS_REG_WRITE(ah, AR_PHY_ANALOG_SWAP, AR_PHY_SWAP_ALT_CHAIN); /* Setup Chain Masks */ OS_REG_WRITE(ah, AR_PHY_RX_CHAINMASK, AH5416(ah)->ah_rx_chainmask); OS_REG_WRITE(ah, AR_PHY_CAL_CHAINMASK, AH5416(ah)->ah_rx_chainmask); OS_REG_WRITE(ah, AR_SELFGEN_MASK, AH5416(ah)->ah_tx_chainmask); if (AR_SREV_HOWL(ah)) { OS_REG_WRITE(ah, AR_PHY_ANALOG_SWAP, OS_REG_READ(ah, AR_PHY_ANALOG_SWAP) | 0x00000001); } } void ar5416RestoreChainMask(struct ath_hal *ah) { int rx_chainmask = AH5416(ah)->ah_rx_chainmask; if ((rx_chainmask == 0x5) || (rx_chainmask == 0x3)) { OS_REG_WRITE(ah, AR_PHY_RX_CHAINMASK, rx_chainmask); OS_REG_WRITE(ah, AR_PHY_CAL_CHAINMASK, rx_chainmask); } } /* * Update the chainmask based on the current channel configuration. * * XXX ath9k checks bluetooth co-existence here * XXX ath9k checks whether the current state is "off-channel". * XXX ath9k sticks the hardware into 1x1 mode for legacy; * we're going to leave multi-RX on for multi-path cancellation. */ static void ar5416UpdateChainMasks(struct ath_hal *ah, HAL_BOOL is_ht) { struct ath_hal_private *ahpriv = AH_PRIVATE(ah); HAL_CAPABILITIES *pCap = &ahpriv->ah_caps; if (is_ht) { AH5416(ah)->ah_tx_chainmask = pCap->halTxChainMask; } else { AH5416(ah)->ah_tx_chainmask = 1; } AH5416(ah)->ah_rx_chainmask = pCap->halRxChainMask; HALDEBUG(ah, HAL_DEBUG_RESET, "TX chainmask: 0x%x; RX chainmask: 0x%x\n", AH5416(ah)->ah_tx_chainmask, AH5416(ah)->ah_rx_chainmask); } #ifndef IS_5GHZ_FAST_CLOCK_EN #define IS_5GHZ_FAST_CLOCK_EN(ah, chan) AH_FALSE #endif void ar5416InitPLL(struct ath_hal *ah, const struct ieee80211_channel *chan) { uint32_t pll; if (AR_SREV_MERLIN_20(ah) && chan != AH_NULL && IEEE80211_IS_CHAN_5GHZ(chan)) { /* * PLL WAR for Merlin 2.0/2.1 * When doing fast clock, set PLL to 0x142c * Else, set PLL to 0x2850 to prevent reset-to-reset variation */ pll = IS_5GHZ_FAST_CLOCK_EN(ah, chan) ? 0x142c : 0x2850; } else if (AR_SREV_MERLIN_10_OR_LATER(ah)) { pll = SM(0x5, AR_RTC_SOWL_PLL_REFDIV); if (chan != AH_NULL) { if (IEEE80211_IS_CHAN_HALF(chan)) pll |= SM(0x1, AR_RTC_SOWL_PLL_CLKSEL); else if (IEEE80211_IS_CHAN_QUARTER(chan)) pll |= SM(0x2, AR_RTC_SOWL_PLL_CLKSEL); else if (IEEE80211_IS_CHAN_5GHZ(chan)) pll |= SM(0x28, AR_RTC_SOWL_PLL_DIV); else pll |= SM(0x2c, AR_RTC_SOWL_PLL_DIV); } else pll |= SM(0x2c, AR_RTC_SOWL_PLL_DIV); } else if (AR_SREV_SOWL_10_OR_LATER(ah)) { pll = SM(0x5, AR_RTC_SOWL_PLL_REFDIV); if (chan != AH_NULL) { if (IEEE80211_IS_CHAN_HALF(chan)) pll |= SM(0x1, AR_RTC_SOWL_PLL_CLKSEL); else if (IEEE80211_IS_CHAN_QUARTER(chan)) pll |= SM(0x2, AR_RTC_SOWL_PLL_CLKSEL); else if (IEEE80211_IS_CHAN_5GHZ(chan)) pll |= SM(0x50, AR_RTC_SOWL_PLL_DIV); else pll |= SM(0x58, AR_RTC_SOWL_PLL_DIV); } else pll |= SM(0x58, AR_RTC_SOWL_PLL_DIV); } else { pll = AR_RTC_PLL_REFDIV_5 | AR_RTC_PLL_DIV2; if (chan != AH_NULL) { if (IEEE80211_IS_CHAN_HALF(chan)) pll |= SM(0x1, AR_RTC_PLL_CLKSEL); else if (IEEE80211_IS_CHAN_QUARTER(chan)) pll |= SM(0x2, AR_RTC_PLL_CLKSEL); else if (IEEE80211_IS_CHAN_5GHZ(chan)) pll |= SM(0xa, AR_RTC_PLL_DIV); else pll |= SM(0xb, AR_RTC_PLL_DIV); } else pll |= SM(0xb, AR_RTC_PLL_DIV); } OS_REG_WRITE(ah, AR_RTC_PLL_CONTROL, pll); /* TODO: * For multi-band owl, switch between bands by reiniting the PLL. */ OS_DELAY(RTC_PLL_SETTLE_DELAY); OS_REG_WRITE(ah, AR_RTC_SLEEP_CLK, AR_RTC_SLEEP_DERIVED_CLK); } static void ar5416SetDefGainValues(struct ath_hal *ah, const MODAL_EEP_HEADER *pModal, const struct ar5416eeprom *eep, uint8_t txRxAttenLocal, int regChainOffset, int i) { if (IS_EEP_MINOR_V3(ah)) { txRxAttenLocal = pModal->txRxAttenCh[i]; if (AR_SREV_MERLIN_20_OR_LATER(ah)) { OS_REG_RMW_FIELD(ah, AR_PHY_GAIN_2GHZ + regChainOffset, AR_PHY_GAIN_2GHZ_XATTEN1_MARGIN, pModal->bswMargin[i]); OS_REG_RMW_FIELD(ah, AR_PHY_GAIN_2GHZ + regChainOffset, AR_PHY_GAIN_2GHZ_XATTEN1_DB, pModal->bswAtten[i]); OS_REG_RMW_FIELD(ah, AR_PHY_GAIN_2GHZ + regChainOffset, AR_PHY_GAIN_2GHZ_XATTEN2_MARGIN, pModal->xatten2Margin[i]); OS_REG_RMW_FIELD(ah, AR_PHY_GAIN_2GHZ + regChainOffset, AR_PHY_GAIN_2GHZ_XATTEN2_DB, pModal->xatten2Db[i]); } else { OS_REG_RMW_FIELD(ah, AR_PHY_GAIN_2GHZ + regChainOffset, AR_PHY_GAIN_2GHZ_BSW_MARGIN, pModal->bswMargin[i]); OS_REG_RMW_FIELD(ah, AR_PHY_GAIN_2GHZ + regChainOffset, AR_PHY_GAIN_2GHZ_BSW_ATTEN, pModal->bswAtten[i]); } } if (AR_SREV_MERLIN_20_OR_LATER(ah)) { OS_REG_RMW_FIELD(ah, AR_PHY_RXGAIN + regChainOffset, AR9280_PHY_RXGAIN_TXRX_ATTEN, txRxAttenLocal); OS_REG_RMW_FIELD(ah, AR_PHY_RXGAIN + regChainOffset, AR9280_PHY_RXGAIN_TXRX_MARGIN, pModal->rxTxMarginCh[i]); } else { OS_REG_RMW_FIELD(ah, AR_PHY_RXGAIN + regChainOffset, AR_PHY_RXGAIN_TXRX_ATTEN, txRxAttenLocal); OS_REG_RMW_FIELD(ah, AR_PHY_GAIN_2GHZ + regChainOffset, AR_PHY_GAIN_2GHZ_RXTX_MARGIN, pModal->rxTxMarginCh[i]); } } /* * Get the register chain offset for the given chain. * * Take into account the register chain swapping with AR5416 v2.0. * * XXX make sure that the reg chain swapping is only done for * XXX AR5416 v2.0 or greater, and not later chips? */ int ar5416GetRegChainOffset(struct ath_hal *ah, int i) { int regChainOffset; if (AR_SREV_OWL_20_OR_LATER(ah) && (AH5416(ah)->ah_rx_chainmask == 0x5 || AH5416(ah)->ah_tx_chainmask == 0x5) && (i != 0)) { /* Regs are swapped from chain 2 to 1 for 5416 2_0 with * only chains 0 and 2 populated */ regChainOffset = (i == 1) ? 0x2000 : 0x1000; } else { regChainOffset = i * 0x1000; } return regChainOffset; } /* * Read EEPROM header info and program the device for correct operation * given the channel value. */ HAL_BOOL ar5416SetBoardValues(struct ath_hal *ah, const struct ieee80211_channel *chan) { const HAL_EEPROM_v14 *ee = AH_PRIVATE(ah)->ah_eeprom; const struct ar5416eeprom *eep = &ee->ee_base; const MODAL_EEP_HEADER *pModal; int i, regChainOffset; uint8_t txRxAttenLocal; /* workaround for eeprom versions <= 14.2 */ HALASSERT(AH_PRIVATE(ah)->ah_eeversion >= AR_EEPROM_VER14_1); pModal = &eep->modalHeader[IEEE80211_IS_CHAN_2GHZ(chan)]; /* NB: workaround for eeprom versions <= 14.2 */ txRxAttenLocal = IEEE80211_IS_CHAN_2GHZ(chan) ? 23 : 44; OS_REG_WRITE(ah, AR_PHY_SWITCH_COM, pModal->antCtrlCommon); for (i = 0; i < AR5416_MAX_CHAINS; i++) { if (AR_SREV_MERLIN(ah)) { if (i >= 2) break; } regChainOffset = ar5416GetRegChainOffset(ah, i); OS_REG_WRITE(ah, AR_PHY_SWITCH_CHAIN_0 + regChainOffset, pModal->antCtrlChain[i]); OS_REG_WRITE(ah, AR_PHY_TIMING_CTRL4 + regChainOffset, (OS_REG_READ(ah, AR_PHY_TIMING_CTRL4 + regChainOffset) & ~(AR_PHY_TIMING_CTRL4_IQCORR_Q_Q_COFF | AR_PHY_TIMING_CTRL4_IQCORR_Q_I_COFF)) | SM(pModal->iqCalICh[i], AR_PHY_TIMING_CTRL4_IQCORR_Q_I_COFF) | SM(pModal->iqCalQCh[i], AR_PHY_TIMING_CTRL4_IQCORR_Q_Q_COFF)); /* * Large signal upgrade. * XXX update */ if ((i == 0) || AR_SREV_OWL_20_OR_LATER(ah)) ar5416SetDefGainValues(ah, pModal, eep, txRxAttenLocal, regChainOffset, i); } if (AR_SREV_MERLIN_20_OR_LATER(ah)) { if (IEEE80211_IS_CHAN_2GHZ(chan)) { OS_A_REG_RMW_FIELD(ah, AR_AN_RF2G1_CH0, AR_AN_RF2G1_CH0_OB, pModal->ob); OS_A_REG_RMW_FIELD(ah, AR_AN_RF2G1_CH0, AR_AN_RF2G1_CH0_DB, pModal->db); OS_A_REG_RMW_FIELD(ah, AR_AN_RF2G1_CH1, AR_AN_RF2G1_CH1_OB, pModal->ob_ch1); OS_A_REG_RMW_FIELD(ah, AR_AN_RF2G1_CH1, AR_AN_RF2G1_CH1_DB, pModal->db_ch1); } else { OS_A_REG_RMW_FIELD(ah, AR_AN_RF5G1_CH0, AR_AN_RF5G1_CH0_OB5, pModal->ob); OS_A_REG_RMW_FIELD(ah, AR_AN_RF5G1_CH0, AR_AN_RF5G1_CH0_DB5, pModal->db); OS_A_REG_RMW_FIELD(ah, AR_AN_RF5G1_CH1, AR_AN_RF5G1_CH1_OB5, pModal->ob_ch1); OS_A_REG_RMW_FIELD(ah, AR_AN_RF5G1_CH1, AR_AN_RF5G1_CH1_DB5, pModal->db_ch1); } OS_A_REG_RMW_FIELD(ah, AR_AN_TOP2, AR_AN_TOP2_XPABIAS_LVL, pModal->xpaBiasLvl); OS_A_REG_RMW_FIELD(ah, AR_AN_TOP2, AR_AN_TOP2_LOCALBIAS, !!(pModal->flagBits & AR5416_EEP_FLAG_LOCALBIAS)); OS_A_REG_RMW_FIELD(ah, AR_PHY_XPA_CFG, AR_PHY_FORCE_XPA_CFG, !!(pModal->flagBits & AR5416_EEP_FLAG_FORCEXPAON)); } OS_REG_RMW_FIELD(ah, AR_PHY_SETTLING, AR_PHY_SETTLING_SWITCH, pModal->switchSettling); OS_REG_RMW_FIELD(ah, AR_PHY_DESIRED_SZ, AR_PHY_DESIRED_SZ_ADC, pModal->adcDesiredSize); if (! AR_SREV_MERLIN_20_OR_LATER(ah)) OS_REG_RMW_FIELD(ah, AR_PHY_DESIRED_SZ, AR_PHY_DESIRED_SZ_PGA, pModal->pgaDesiredSize); OS_REG_WRITE(ah, AR_PHY_RF_CTL4, SM(pModal->txEndToXpaOff, AR_PHY_RF_CTL4_TX_END_XPAA_OFF) | SM(pModal->txEndToXpaOff, AR_PHY_RF_CTL4_TX_END_XPAB_OFF) | SM(pModal->txFrameToXpaOn, AR_PHY_RF_CTL4_FRAME_XPAA_ON) | SM(pModal->txFrameToXpaOn, AR_PHY_RF_CTL4_FRAME_XPAB_ON)); OS_REG_RMW_FIELD(ah, AR_PHY_RF_CTL3, AR_PHY_TX_END_TO_A2_RX_ON, pModal->txEndToRxOn); if (AR_SREV_MERLIN_10_OR_LATER(ah)) { OS_REG_RMW_FIELD(ah, AR_PHY_CCA, AR9280_PHY_CCA_THRESH62, pModal->thresh62); OS_REG_RMW_FIELD(ah, AR_PHY_EXT_CCA0, AR_PHY_EXT_CCA0_THRESH62, pModal->thresh62); } else { OS_REG_RMW_FIELD(ah, AR_PHY_CCA, AR_PHY_CCA_THRESH62, pModal->thresh62); OS_REG_RMW_FIELD(ah, AR_PHY_EXT_CCA, AR_PHY_EXT_CCA_THRESH62, pModal->thresh62); } /* Minor Version Specific application */ if (IS_EEP_MINOR_V2(ah)) { OS_REG_RMW_FIELD(ah, AR_PHY_RF_CTL2, AR_PHY_TX_FRAME_TO_DATA_START, pModal->txFrameToDataStart); OS_REG_RMW_FIELD(ah, AR_PHY_RF_CTL2, AR_PHY_TX_FRAME_TO_PA_ON, pModal->txFrameToPaOn); } if (IS_EEP_MINOR_V3(ah) && IEEE80211_IS_CHAN_HT40(chan)) /* Overwrite switch settling with HT40 value */ OS_REG_RMW_FIELD(ah, AR_PHY_SETTLING, AR_PHY_SETTLING_SWITCH, pModal->swSettleHt40); if (AR_SREV_MERLIN_20_OR_LATER(ah) && EEP_MINOR(ah) >= AR5416_EEP_MINOR_VER_19) OS_REG_RMW_FIELD(ah, AR_PHY_CCK_TX_CTRL, AR_PHY_CCK_TX_CTRL_TX_DAC_SCALE_CCK, pModal->miscBits); if (AR_SREV_MERLIN_20(ah) && EEP_MINOR(ah) >= AR5416_EEP_MINOR_VER_20) { if (IEEE80211_IS_CHAN_2GHZ(chan)) OS_A_REG_RMW_FIELD(ah, AR_AN_TOP1, AR_AN_TOP1_DACIPMODE, eep->baseEepHeader.dacLpMode); else if (eep->baseEepHeader.dacHiPwrMode_5G) OS_A_REG_RMW_FIELD(ah, AR_AN_TOP1, AR_AN_TOP1_DACIPMODE, 0); else OS_A_REG_RMW_FIELD(ah, AR_AN_TOP1, AR_AN_TOP1_DACIPMODE, eep->baseEepHeader.dacLpMode); OS_REG_RMW_FIELD(ah, AR_PHY_FRAME_CTL, AR_PHY_FRAME_CTL_TX_CLIP, pModal->miscBits >> 2); OS_REG_RMW_FIELD(ah, AR_PHY_TX_PWRCTRL9, AR_PHY_TX_DESIRED_SCALE_CCK, eep->baseEepHeader.desiredScaleCCK); } return AH_TRUE; } /* * Helper functions common for AP/CB/XB */ /* * ar5416SetPowerPerRateTable * * Sets the transmit power in the baseband for the given * operating channel and mode. */ static HAL_BOOL ar5416SetPowerPerRateTable(struct ath_hal *ah, struct ar5416eeprom *pEepData, const struct ieee80211_channel *chan, int16_t *ratesArray, uint16_t cfgCtl, uint16_t AntennaReduction, uint16_t twiceMaxRegulatoryPower, uint16_t powerLimit) { #define N(a) (sizeof(a)/sizeof(a[0])) /* Local defines to distinguish between extension and control CTL's */ #define EXT_ADDITIVE (0x8000) #define CTL_11A_EXT (CTL_11A | EXT_ADDITIVE) #define CTL_11G_EXT (CTL_11G | EXT_ADDITIVE) #define CTL_11B_EXT (CTL_11B | EXT_ADDITIVE) uint16_t twiceMaxEdgePower = AR5416_MAX_RATE_POWER; int i; int16_t twiceLargestAntenna; CAL_CTL_DATA *rep; CAL_TARGET_POWER_LEG targetPowerOfdm, targetPowerCck = {0, {0, 0, 0, 0}}; CAL_TARGET_POWER_LEG targetPowerOfdmExt = {0, {0, 0, 0, 0}}, targetPowerCckExt = {0, {0, 0, 0, 0}}; CAL_TARGET_POWER_HT targetPowerHt20, targetPowerHt40 = {0, {0, 0, 0, 0}}; int16_t scaledPower, minCtlPower; #define SUB_NUM_CTL_MODES_AT_5G_40 2 /* excluding HT40, EXT-OFDM */ #define SUB_NUM_CTL_MODES_AT_2G_40 3 /* excluding HT40, EXT-OFDM, EXT-CCK */ static const uint16_t ctlModesFor11a[] = { CTL_11A, CTL_5GHT20, CTL_11A_EXT, CTL_5GHT40 }; static const uint16_t ctlModesFor11g[] = { CTL_11B, CTL_11G, CTL_2GHT20, CTL_11B_EXT, CTL_11G_EXT, CTL_2GHT40 }; const uint16_t *pCtlMode; uint16_t numCtlModes, ctlMode, freq; CHAN_CENTERS centers; ar5416GetChannelCenters(ah, chan, ¢ers); /* Compute TxPower reduction due to Antenna Gain */ twiceLargestAntenna = AH_MAX(AH_MAX( pEepData->modalHeader[IEEE80211_IS_CHAN_2GHZ(chan)].antennaGainCh[0], pEepData->modalHeader[IEEE80211_IS_CHAN_2GHZ(chan)].antennaGainCh[1]), pEepData->modalHeader[IEEE80211_IS_CHAN_2GHZ(chan)].antennaGainCh[2]); #if 0 /* Turn it back on if we need to calculate per chain antenna gain reduction */ /* Use only if the expected gain > 6dbi */ /* Chain 0 is always used */ twiceLargestAntenna = pEepData->modalHeader[IEEE80211_IS_CHAN_2GHZ(chan)].antennaGainCh[0]; /* Look at antenna gains of Chains 1 and 2 if the TX mask is set */ if (ahp->ah_tx_chainmask & 0x2) twiceLargestAntenna = AH_MAX(twiceLargestAntenna, pEepData->modalHeader[IEEE80211_IS_CHAN_2GHZ(chan)].antennaGainCh[1]); if (ahp->ah_tx_chainmask & 0x4) twiceLargestAntenna = AH_MAX(twiceLargestAntenna, pEepData->modalHeader[IEEE80211_IS_CHAN_2GHZ(chan)].antennaGainCh[2]); #endif twiceLargestAntenna = (int16_t)AH_MIN((AntennaReduction) - twiceLargestAntenna, 0); /* XXX setup for 5212 use (really used?) */ ath_hal_eepromSet(ah, IEEE80211_IS_CHAN_2GHZ(chan) ? AR_EEP_ANTGAINMAX_2 : AR_EEP_ANTGAINMAX_5, twiceLargestAntenna); /* * scaledPower is the minimum of the user input power level and * the regulatory allowed power level */ scaledPower = AH_MIN(powerLimit, twiceMaxRegulatoryPower + twiceLargestAntenna); /* Reduce scaled Power by number of chains active to get to per chain tx power level */ /* TODO: better value than these? */ switch (owl_get_ntxchains(AH5416(ah)->ah_tx_chainmask)) { case 1: break; case 2: scaledPower -= pEepData->modalHeader[IEEE80211_IS_CHAN_2GHZ(chan)].pwrDecreaseFor2Chain; break; case 3: scaledPower -= pEepData->modalHeader[IEEE80211_IS_CHAN_2GHZ(chan)].pwrDecreaseFor3Chain; break; default: return AH_FALSE; /* Unsupported number of chains */ } scaledPower = AH_MAX(0, scaledPower); /* Get target powers from EEPROM - our baseline for TX Power */ if (IEEE80211_IS_CHAN_2GHZ(chan)) { /* Setup for CTL modes */ numCtlModes = N(ctlModesFor11g) - SUB_NUM_CTL_MODES_AT_2G_40; /* CTL_11B, CTL_11G, CTL_2GHT20 */ pCtlMode = ctlModesFor11g; ar5416GetTargetPowersLeg(ah, chan, pEepData->calTargetPowerCck, AR5416_NUM_2G_CCK_TARGET_POWERS, &targetPowerCck, 4, AH_FALSE); ar5416GetTargetPowersLeg(ah, chan, pEepData->calTargetPower2G, AR5416_NUM_2G_20_TARGET_POWERS, &targetPowerOfdm, 4, AH_FALSE); ar5416GetTargetPowers(ah, chan, pEepData->calTargetPower2GHT20, AR5416_NUM_2G_20_TARGET_POWERS, &targetPowerHt20, 8, AH_FALSE); if (IEEE80211_IS_CHAN_HT40(chan)) { numCtlModes = N(ctlModesFor11g); /* All 2G CTL's */ ar5416GetTargetPowers(ah, chan, pEepData->calTargetPower2GHT40, AR5416_NUM_2G_40_TARGET_POWERS, &targetPowerHt40, 8, AH_TRUE); /* Get target powers for extension channels */ ar5416GetTargetPowersLeg(ah, chan, pEepData->calTargetPowerCck, AR5416_NUM_2G_CCK_TARGET_POWERS, &targetPowerCckExt, 4, AH_TRUE); ar5416GetTargetPowersLeg(ah, chan, pEepData->calTargetPower2G, AR5416_NUM_2G_20_TARGET_POWERS, &targetPowerOfdmExt, 4, AH_TRUE); } } else { /* Setup for CTL modes */ numCtlModes = N(ctlModesFor11a) - SUB_NUM_CTL_MODES_AT_5G_40; /* CTL_11A, CTL_5GHT20 */ pCtlMode = ctlModesFor11a; ar5416GetTargetPowersLeg(ah, chan, pEepData->calTargetPower5G, AR5416_NUM_5G_20_TARGET_POWERS, &targetPowerOfdm, 4, AH_FALSE); ar5416GetTargetPowers(ah, chan, pEepData->calTargetPower5GHT20, AR5416_NUM_5G_20_TARGET_POWERS, &targetPowerHt20, 8, AH_FALSE); if (IEEE80211_IS_CHAN_HT40(chan)) { numCtlModes = N(ctlModesFor11a); /* All 5G CTL's */ ar5416GetTargetPowers(ah, chan, pEepData->calTargetPower5GHT40, AR5416_NUM_5G_40_TARGET_POWERS, &targetPowerHt40, 8, AH_TRUE); ar5416GetTargetPowersLeg(ah, chan, pEepData->calTargetPower5G, AR5416_NUM_5G_20_TARGET_POWERS, &targetPowerOfdmExt, 4, AH_TRUE); } } /* * For MIMO, need to apply regulatory caps individually across dynamically * running modes: CCK, OFDM, HT20, HT40 * * The outer loop walks through each possible applicable runtime mode. * The inner loop walks through each ctlIndex entry in EEPROM. * The ctl value is encoded as [7:4] == test group, [3:0] == test mode. * */ for (ctlMode = 0; ctlMode < numCtlModes; ctlMode++) { HAL_BOOL isHt40CtlMode = (pCtlMode[ctlMode] == CTL_5GHT40) || (pCtlMode[ctlMode] == CTL_2GHT40); if (isHt40CtlMode) { freq = centers.ctl_center; } else if (pCtlMode[ctlMode] & EXT_ADDITIVE) { freq = centers.ext_center; } else { freq = centers.ctl_center; } /* walk through each CTL index stored in EEPROM */ for (i = 0; (i < AR5416_NUM_CTLS) && pEepData->ctlIndex[i]; i++) { uint16_t twiceMinEdgePower; /* compare test group from regulatory channel list with test mode from pCtlMode list */ if ((((cfgCtl & ~CTL_MODE_M) | (pCtlMode[ctlMode] & CTL_MODE_M)) == pEepData->ctlIndex[i]) || (((cfgCtl & ~CTL_MODE_M) | (pCtlMode[ctlMode] & CTL_MODE_M)) == ((pEepData->ctlIndex[i] & CTL_MODE_M) | SD_NO_CTL))) { rep = &(pEepData->ctlData[i]); twiceMinEdgePower = ar5416GetMaxEdgePower(freq, rep->ctlEdges[owl_get_ntxchains(AH5416(ah)->ah_tx_chainmask) - 1], IEEE80211_IS_CHAN_2GHZ(chan)); if ((cfgCtl & ~CTL_MODE_M) == SD_NO_CTL) { /* Find the minimum of all CTL edge powers that apply to this channel */ twiceMaxEdgePower = AH_MIN(twiceMaxEdgePower, twiceMinEdgePower); } else { /* specific */ twiceMaxEdgePower = twiceMinEdgePower; break; } } } minCtlPower = (uint8_t)AH_MIN(twiceMaxEdgePower, scaledPower); /* Apply ctl mode to correct target power set */ switch(pCtlMode[ctlMode]) { case CTL_11B: for (i = 0; i < N(targetPowerCck.tPow2x); i++) { targetPowerCck.tPow2x[i] = (uint8_t)AH_MIN(targetPowerCck.tPow2x[i], minCtlPower); } break; case CTL_11A: case CTL_11G: for (i = 0; i < N(targetPowerOfdm.tPow2x); i++) { targetPowerOfdm.tPow2x[i] = (uint8_t)AH_MIN(targetPowerOfdm.tPow2x[i], minCtlPower); } break; case CTL_5GHT20: case CTL_2GHT20: for (i = 0; i < N(targetPowerHt20.tPow2x); i++) { targetPowerHt20.tPow2x[i] = (uint8_t)AH_MIN(targetPowerHt20.tPow2x[i], minCtlPower); } break; case CTL_11B_EXT: targetPowerCckExt.tPow2x[0] = (uint8_t)AH_MIN(targetPowerCckExt.tPow2x[0], minCtlPower); break; case CTL_11A_EXT: case CTL_11G_EXT: targetPowerOfdmExt.tPow2x[0] = (uint8_t)AH_MIN(targetPowerOfdmExt.tPow2x[0], minCtlPower); break; case CTL_5GHT40: case CTL_2GHT40: for (i = 0; i < N(targetPowerHt40.tPow2x); i++) { targetPowerHt40.tPow2x[i] = (uint8_t)AH_MIN(targetPowerHt40.tPow2x[i], minCtlPower); } break; default: return AH_FALSE; break; } } /* end ctl mode checking */ /* Set rates Array from collected data */ ratesArray[rate6mb] = ratesArray[rate9mb] = ratesArray[rate12mb] = ratesArray[rate18mb] = ratesArray[rate24mb] = targetPowerOfdm.tPow2x[0]; ratesArray[rate36mb] = targetPowerOfdm.tPow2x[1]; ratesArray[rate48mb] = targetPowerOfdm.tPow2x[2]; ratesArray[rate54mb] = targetPowerOfdm.tPow2x[3]; ratesArray[rateXr] = targetPowerOfdm.tPow2x[0]; for (i = 0; i < N(targetPowerHt20.tPow2x); i++) { ratesArray[rateHt20_0 + i] = targetPowerHt20.tPow2x[i]; } if (IEEE80211_IS_CHAN_2GHZ(chan)) { ratesArray[rate1l] = targetPowerCck.tPow2x[0]; ratesArray[rate2s] = ratesArray[rate2l] = targetPowerCck.tPow2x[1]; ratesArray[rate5_5s] = ratesArray[rate5_5l] = targetPowerCck.tPow2x[2]; ratesArray[rate11s] = ratesArray[rate11l] = targetPowerCck.tPow2x[3]; } if (IEEE80211_IS_CHAN_HT40(chan)) { for (i = 0; i < N(targetPowerHt40.tPow2x); i++) { ratesArray[rateHt40_0 + i] = targetPowerHt40.tPow2x[i]; } ratesArray[rateDupOfdm] = targetPowerHt40.tPow2x[0]; ratesArray[rateDupCck] = targetPowerHt40.tPow2x[0]; ratesArray[rateExtOfdm] = targetPowerOfdmExt.tPow2x[0]; if (IEEE80211_IS_CHAN_2GHZ(chan)) { ratesArray[rateExtCck] = targetPowerCckExt.tPow2x[0]; } } return AH_TRUE; #undef EXT_ADDITIVE #undef CTL_11A_EXT #undef CTL_11G_EXT #undef CTL_11B_EXT #undef SUB_NUM_CTL_MODES_AT_5G_40 #undef SUB_NUM_CTL_MODES_AT_2G_40 #undef N } /************************************************************************** * fbin2freq * * Get channel value from binary representation held in eeprom * RETURNS: the frequency in MHz */ static uint16_t fbin2freq(uint8_t fbin, HAL_BOOL is2GHz) { /* * Reserved value 0xFF provides an empty definition both as * an fbin and as a frequency - do not convert */ if (fbin == AR5416_BCHAN_UNUSED) { return fbin; } return (uint16_t)((is2GHz) ? (2300 + fbin) : (4800 + 5 * fbin)); } /* * ar5416GetMaxEdgePower * * Find the maximum conformance test limit for the given channel and CTL info */ uint16_t ar5416GetMaxEdgePower(uint16_t freq, CAL_CTL_EDGES *pRdEdgesPower, HAL_BOOL is2GHz) { uint16_t twiceMaxEdgePower = AR5416_MAX_RATE_POWER; int i; /* Get the edge power */ for (i = 0; (i < AR5416_NUM_BAND_EDGES) && (pRdEdgesPower[i].bChannel != AR5416_BCHAN_UNUSED) ; i++) { /* * If there's an exact channel match or an inband flag set * on the lower channel use the given rdEdgePower */ if (freq == fbin2freq(pRdEdgesPower[i].bChannel, is2GHz)) { twiceMaxEdgePower = MS(pRdEdgesPower[i].tPowerFlag, CAL_CTL_EDGES_POWER); break; } else if ((i > 0) && (freq < fbin2freq(pRdEdgesPower[i].bChannel, is2GHz))) { if (fbin2freq(pRdEdgesPower[i - 1].bChannel, is2GHz) < freq && (pRdEdgesPower[i - 1].tPowerFlag & CAL_CTL_EDGES_FLAG) != 0) { twiceMaxEdgePower = MS(pRdEdgesPower[i - 1].tPowerFlag, CAL_CTL_EDGES_POWER); } /* Leave loop - no more affecting edges possible in this monotonic increasing list */ break; } } HALASSERT(twiceMaxEdgePower > 0); return twiceMaxEdgePower; } /************************************************************** * ar5416GetTargetPowers * * Return the rates of target power for the given target power table * channel, and number of channels */ void ar5416GetTargetPowers(struct ath_hal *ah, const struct ieee80211_channel *chan, CAL_TARGET_POWER_HT *powInfo, uint16_t numChannels, CAL_TARGET_POWER_HT *pNewPower, uint16_t numRates, HAL_BOOL isHt40Target) { uint16_t clo, chi; int i; int matchIndex = -1, lowIndex = -1; uint16_t freq; CHAN_CENTERS centers; ar5416GetChannelCenters(ah, chan, ¢ers); freq = isHt40Target ? centers.synth_center : centers.ctl_center; /* Copy the target powers into the temp channel list */ if (freq <= fbin2freq(powInfo[0].bChannel, IEEE80211_IS_CHAN_2GHZ(chan))) { matchIndex = 0; } else { for (i = 0; (i < numChannels) && (powInfo[i].bChannel != AR5416_BCHAN_UNUSED); i++) { if (freq == fbin2freq(powInfo[i].bChannel, IEEE80211_IS_CHAN_2GHZ(chan))) { matchIndex = i; break; } else if ((freq < fbin2freq(powInfo[i].bChannel, IEEE80211_IS_CHAN_2GHZ(chan))) && (freq > fbin2freq(powInfo[i - 1].bChannel, IEEE80211_IS_CHAN_2GHZ(chan)))) { lowIndex = i - 1; break; } } if ((matchIndex == -1) && (lowIndex == -1)) { HALASSERT(freq > fbin2freq(powInfo[i - 1].bChannel, IEEE80211_IS_CHAN_2GHZ(chan))); matchIndex = i - 1; } } if (matchIndex != -1) { OS_MEMCPY(pNewPower, &powInfo[matchIndex], sizeof(*pNewPower)); } else { HALASSERT(lowIndex != -1); /* * Get the lower and upper channels, target powers, * and interpolate between them. */ clo = fbin2freq(powInfo[lowIndex].bChannel, IEEE80211_IS_CHAN_2GHZ(chan)); chi = fbin2freq(powInfo[lowIndex + 1].bChannel, IEEE80211_IS_CHAN_2GHZ(chan)); for (i = 0; i < numRates; i++) { pNewPower->tPow2x[i] = (uint8_t)ath_ee_interpolate(freq, clo, chi, powInfo[lowIndex].tPow2x[i], powInfo[lowIndex + 1].tPow2x[i]); } } } /************************************************************** * ar5416GetTargetPowersLeg * * Return the four rates of target power for the given target power table * channel, and number of channels */ void ar5416GetTargetPowersLeg(struct ath_hal *ah, const struct ieee80211_channel *chan, CAL_TARGET_POWER_LEG *powInfo, uint16_t numChannels, CAL_TARGET_POWER_LEG *pNewPower, uint16_t numRates, HAL_BOOL isExtTarget) { uint16_t clo, chi; int i; int matchIndex = -1, lowIndex = -1; uint16_t freq; CHAN_CENTERS centers; ar5416GetChannelCenters(ah, chan, ¢ers); freq = (isExtTarget) ? centers.ext_center :centers.ctl_center; /* Copy the target powers into the temp channel list */ if (freq <= fbin2freq(powInfo[0].bChannel, IEEE80211_IS_CHAN_2GHZ(chan))) { matchIndex = 0; } else { for (i = 0; (i < numChannels) && (powInfo[i].bChannel != AR5416_BCHAN_UNUSED); i++) { if (freq == fbin2freq(powInfo[i].bChannel, IEEE80211_IS_CHAN_2GHZ(chan))) { matchIndex = i; break; } else if ((freq < fbin2freq(powInfo[i].bChannel, IEEE80211_IS_CHAN_2GHZ(chan))) && (freq > fbin2freq(powInfo[i - 1].bChannel, IEEE80211_IS_CHAN_2GHZ(chan)))) { lowIndex = i - 1; break; } } if ((matchIndex == -1) && (lowIndex == -1)) { HALASSERT(freq > fbin2freq(powInfo[i - 1].bChannel, IEEE80211_IS_CHAN_2GHZ(chan))); matchIndex = i - 1; } } if (matchIndex != -1) { OS_MEMCPY(pNewPower, &powInfo[matchIndex], sizeof(*pNewPower)); } else { HALASSERT(lowIndex != -1); /* * Get the lower and upper channels, target powers, * and interpolate between them. */ clo = fbin2freq(powInfo[lowIndex].bChannel, IEEE80211_IS_CHAN_2GHZ(chan)); chi = fbin2freq(powInfo[lowIndex + 1].bChannel, IEEE80211_IS_CHAN_2GHZ(chan)); for (i = 0; i < numRates; i++) { pNewPower->tPow2x[i] = (uint8_t)ath_ee_interpolate(freq, clo, chi, powInfo[lowIndex].tPow2x[i], powInfo[lowIndex + 1].tPow2x[i]); } } } /* * Set the gain boundaries for the given radio chain. * * The gain boundaries tell the hardware at what point in the * PDADC array to "switch over" from one PD gain setting * to another. There's also a gain overlap between two * PDADC array gain curves where there's valid PD values * for 2 gain settings. * * The hardware uses the gain overlap and gain boundaries * to determine which gain curve to use for the given * target TX power. */ void ar5416SetGainBoundariesClosedLoop(struct ath_hal *ah, int i, uint16_t pdGainOverlap_t2, uint16_t gainBoundaries[]) { int regChainOffset; regChainOffset = ar5416GetRegChainOffset(ah, i); HALDEBUG(ah, HAL_DEBUG_EEPROM, "%s: chain %d: gainOverlap_t2: %d," " gainBoundaries: %d, %d, %d, %d\n", __func__, i, pdGainOverlap_t2, gainBoundaries[0], gainBoundaries[1], gainBoundaries[2], gainBoundaries[3]); OS_REG_WRITE(ah, AR_PHY_TPCRG5 + regChainOffset, SM(pdGainOverlap_t2, AR_PHY_TPCRG5_PD_GAIN_OVERLAP) | SM(gainBoundaries[0], AR_PHY_TPCRG5_PD_GAIN_BOUNDARY_1) | SM(gainBoundaries[1], AR_PHY_TPCRG5_PD_GAIN_BOUNDARY_2) | SM(gainBoundaries[2], AR_PHY_TPCRG5_PD_GAIN_BOUNDARY_3) | SM(gainBoundaries[3], AR_PHY_TPCRG5_PD_GAIN_BOUNDARY_4)); } /* * Get the gain values and the number of gain levels given * in xpdMask. * * The EEPROM xpdMask determines which power detector gain * levels were used during calibration. Each of these mask * bits maps to a fixed gain level in hardware. */ uint16_t ar5416GetXpdGainValues(struct ath_hal *ah, uint16_t xpdMask, uint16_t xpdGainValues[]) { int i; uint16_t numXpdGain = 0; for (i = 1; i <= AR5416_PD_GAINS_IN_MASK; i++) { if ((xpdMask >> (AR5416_PD_GAINS_IN_MASK - i)) & 1) { if (numXpdGain >= AR5416_NUM_PD_GAINS) { HALASSERT(0); break; } xpdGainValues[numXpdGain] = (uint16_t)(AR5416_PD_GAINS_IN_MASK - i); numXpdGain++; } } return numXpdGain; } /* * Write the detector gain and biases. * * There are four power detector gain levels. The xpdMask in the EEPROM * determines which power detector gain levels have TX power calibration * data associated with them. This function writes the number of * PD gain levels and their values into the hardware. * * This is valid for all TX chains - the calibration data itself however * will likely differ per-chain. */ void ar5416WriteDetectorGainBiases(struct ath_hal *ah, uint16_t numXpdGain, uint16_t xpdGainValues[]) { HALDEBUG(ah, HAL_DEBUG_EEPROM, "%s: numXpdGain: %d," " xpdGainValues: %d, %d, %d\n", __func__, numXpdGain, xpdGainValues[0], xpdGainValues[1], xpdGainValues[2]); OS_REG_WRITE(ah, AR_PHY_TPCRG1, (OS_REG_READ(ah, AR_PHY_TPCRG1) & ~(AR_PHY_TPCRG1_NUM_PD_GAIN | AR_PHY_TPCRG1_PD_GAIN_1 | AR_PHY_TPCRG1_PD_GAIN_2 | AR_PHY_TPCRG1_PD_GAIN_3)) | SM(numXpdGain - 1, AR_PHY_TPCRG1_NUM_PD_GAIN) | SM(xpdGainValues[0], AR_PHY_TPCRG1_PD_GAIN_1 ) | SM(xpdGainValues[1], AR_PHY_TPCRG1_PD_GAIN_2) | SM(xpdGainValues[2], AR_PHY_TPCRG1_PD_GAIN_3)); } /* * Write the PDADC array to the given radio chain i. * * The 32 PDADC registers are written without any care about * their contents - so if various chips treat values as "special", * this routine will not care. */ void ar5416WritePdadcValues(struct ath_hal *ah, int i, uint8_t pdadcValues[]) { int regOffset, regChainOffset; int j; int reg32; regChainOffset = ar5416GetRegChainOffset(ah, i); regOffset = AR_PHY_BASE + (672 << 2) + regChainOffset; for (j = 0; j < 32; j++) { reg32 = ((pdadcValues[4*j + 0] & 0xFF) << 0) | ((pdadcValues[4*j + 1] & 0xFF) << 8) | ((pdadcValues[4*j + 2] & 0xFF) << 16) | ((pdadcValues[4*j + 3] & 0xFF) << 24) ; OS_REG_WRITE(ah, regOffset, reg32); HALDEBUG(ah, HAL_DEBUG_EEPROM, "PDADC: Chain %d |" " PDADC %3d Value %3d | PDADC %3d Value %3d | PDADC %3d" " Value %3d | PDADC %3d Value %3d |\n", i, 4*j, pdadcValues[4*j], 4*j+1, pdadcValues[4*j + 1], 4*j+2, pdadcValues[4*j + 2], 4*j+3, pdadcValues[4*j + 3]); regOffset += 4; } } /************************************************************** * ar5416SetPowerCalTable * * Pull the PDADC piers from cal data and interpolate them across the given * points as well as from the nearest pier(s) to get a power detector * linear voltage to power level table. */ HAL_BOOL ar5416SetPowerCalTable(struct ath_hal *ah, struct ar5416eeprom *pEepData, const struct ieee80211_channel *chan, int16_t *pTxPowerIndexOffset) { CAL_DATA_PER_FREQ *pRawDataset; uint8_t *pCalBChans = AH_NULL; uint16_t pdGainOverlap_t2; static uint8_t pdadcValues[AR5416_NUM_PDADC_VALUES]; uint16_t gainBoundaries[AR5416_PD_GAINS_IN_MASK]; uint16_t numPiers, i; int16_t tMinCalPower; uint16_t numXpdGain, xpdMask; uint16_t xpdGainValues[AR5416_NUM_PD_GAINS]; uint32_t regChainOffset; OS_MEMZERO(xpdGainValues, sizeof(xpdGainValues)); xpdMask = pEepData->modalHeader[IEEE80211_IS_CHAN_2GHZ(chan)].xpdGain; if (IS_EEP_MINOR_V2(ah)) { pdGainOverlap_t2 = pEepData->modalHeader[IEEE80211_IS_CHAN_2GHZ(chan)].pdGainOverlap; } else { pdGainOverlap_t2 = (uint16_t)(MS(OS_REG_READ(ah, AR_PHY_TPCRG5), AR_PHY_TPCRG5_PD_GAIN_OVERLAP)); } if (IEEE80211_IS_CHAN_2GHZ(chan)) { pCalBChans = pEepData->calFreqPier2G; numPiers = AR5416_NUM_2G_CAL_PIERS; } else { pCalBChans = pEepData->calFreqPier5G; numPiers = AR5416_NUM_5G_CAL_PIERS; } /* Calculate the value of xpdgains from the xpdGain Mask */ numXpdGain = ar5416GetXpdGainValues(ah, xpdMask, xpdGainValues); /* Write the detector gain biases and their number */ ar5416WriteDetectorGainBiases(ah, numXpdGain, xpdGainValues); for (i = 0; i < AR5416_MAX_CHAINS; i++) { regChainOffset = ar5416GetRegChainOffset(ah, i); if (pEepData->baseEepHeader.txMask & (1 << i)) { if (IEEE80211_IS_CHAN_2GHZ(chan)) { pRawDataset = pEepData->calPierData2G[i]; } else { pRawDataset = pEepData->calPierData5G[i]; } /* Fetch the gain boundaries and the PDADC values */ ar5416GetGainBoundariesAndPdadcs(ah, chan, pRawDataset, pCalBChans, numPiers, pdGainOverlap_t2, &tMinCalPower, gainBoundaries, pdadcValues, numXpdGain); if ((i == 0) || AR_SREV_OWL_20_OR_LATER(ah)) { ar5416SetGainBoundariesClosedLoop(ah, i, pdGainOverlap_t2, gainBoundaries); } /* Write the power values into the baseband power table */ ar5416WritePdadcValues(ah, i, pdadcValues); } } *pTxPowerIndexOffset = 0; return AH_TRUE; } /************************************************************** * ar5416GetGainBoundariesAndPdadcs * * Uses the data points read from EEPROM to reconstruct the pdadc power table * Called by ar5416SetPowerCalTable only. */ void ar5416GetGainBoundariesAndPdadcs(struct ath_hal *ah, const struct ieee80211_channel *chan, CAL_DATA_PER_FREQ *pRawDataSet, uint8_t * bChans, uint16_t availPiers, uint16_t tPdGainOverlap, int16_t *pMinCalPower, uint16_t * pPdGainBoundaries, uint8_t * pPDADCValues, uint16_t numXpdGains) { int i, j, k; int16_t ss; /* potentially -ve index for taking care of pdGainOverlap */ uint16_t idxL, idxR, numPiers; /* Pier indexes */ /* filled out Vpd table for all pdGains (chanL) */ static uint8_t vpdTableL[AR5416_NUM_PD_GAINS][AR5416_MAX_PWR_RANGE_IN_HALF_DB]; /* filled out Vpd table for all pdGains (chanR) */ static uint8_t vpdTableR[AR5416_NUM_PD_GAINS][AR5416_MAX_PWR_RANGE_IN_HALF_DB]; /* filled out Vpd table for all pdGains (interpolated) */ static uint8_t vpdTableI[AR5416_NUM_PD_GAINS][AR5416_MAX_PWR_RANGE_IN_HALF_DB]; uint8_t *pVpdL, *pVpdR, *pPwrL, *pPwrR; uint8_t minPwrT4[AR5416_NUM_PD_GAINS]; uint8_t maxPwrT4[AR5416_NUM_PD_GAINS]; int16_t vpdStep; int16_t tmpVal; uint16_t sizeCurrVpdTable, maxIndex, tgtIndex; HAL_BOOL match; int16_t minDelta = 0; CHAN_CENTERS centers; ar5416GetChannelCenters(ah, chan, ¢ers); /* Trim numPiers for the number of populated channel Piers */ for (numPiers = 0; numPiers < availPiers; numPiers++) { if (bChans[numPiers] == AR5416_BCHAN_UNUSED) { break; } } /* Find pier indexes around the current channel */ match = ath_ee_getLowerUpperIndex((uint8_t)FREQ2FBIN(centers.synth_center, IEEE80211_IS_CHAN_2GHZ(chan)), bChans, numPiers, &idxL, &idxR); if (match) { /* Directly fill both vpd tables from the matching index */ for (i = 0; i < numXpdGains; i++) { minPwrT4[i] = pRawDataSet[idxL].pwrPdg[i][0]; maxPwrT4[i] = pRawDataSet[idxL].pwrPdg[i][4]; ath_ee_FillVpdTable(minPwrT4[i], maxPwrT4[i], pRawDataSet[idxL].pwrPdg[i], pRawDataSet[idxL].vpdPdg[i], AR5416_PD_GAIN_ICEPTS, vpdTableI[i]); } } else { for (i = 0; i < numXpdGains; i++) { pVpdL = pRawDataSet[idxL].vpdPdg[i]; pPwrL = pRawDataSet[idxL].pwrPdg[i]; pVpdR = pRawDataSet[idxR].vpdPdg[i]; pPwrR = pRawDataSet[idxR].pwrPdg[i]; /* Start Vpd interpolation from the max of the minimum powers */ minPwrT4[i] = AH_MAX(pPwrL[0], pPwrR[0]); /* End Vpd interpolation from the min of the max powers */ maxPwrT4[i] = AH_MIN(pPwrL[AR5416_PD_GAIN_ICEPTS - 1], pPwrR[AR5416_PD_GAIN_ICEPTS - 1]); HALASSERT(maxPwrT4[i] > minPwrT4[i]); /* Fill pier Vpds */ ath_ee_FillVpdTable(minPwrT4[i], maxPwrT4[i], pPwrL, pVpdL, AR5416_PD_GAIN_ICEPTS, vpdTableL[i]); ath_ee_FillVpdTable(minPwrT4[i], maxPwrT4[i], pPwrR, pVpdR, AR5416_PD_GAIN_ICEPTS, vpdTableR[i]); /* Interpolate the final vpd */ for (j = 0; j <= (maxPwrT4[i] - minPwrT4[i]) / 2; j++) { vpdTableI[i][j] = (uint8_t)(ath_ee_interpolate((uint16_t)FREQ2FBIN(centers.synth_center, IEEE80211_IS_CHAN_2GHZ(chan)), bChans[idxL], bChans[idxR], vpdTableL[i][j], vpdTableR[i][j])); } } } *pMinCalPower = (int16_t)(minPwrT4[0] / 2); k = 0; /* index for the final table */ for (i = 0; i < numXpdGains; i++) { if (i == (numXpdGains - 1)) { pPdGainBoundaries[i] = (uint16_t)(maxPwrT4[i] / 2); } else { pPdGainBoundaries[i] = (uint16_t)((maxPwrT4[i] + minPwrT4[i+1]) / 4); } pPdGainBoundaries[i] = (uint16_t)AH_MIN(AR5416_MAX_RATE_POWER, pPdGainBoundaries[i]); /* NB: only applies to owl 1.0 */ if ((i == 0) && !AR_SREV_OWL_20_OR_LATER(ah) ) { /* * fix the gain delta, but get a delta that can be applied to min to * keep the upper power values accurate, don't think max needs to * be adjusted because should not be at that area of the table? */ minDelta = pPdGainBoundaries[0] - 23; pPdGainBoundaries[0] = 23; } else { minDelta = 0; } /* Find starting index for this pdGain */ if (i == 0) { if (AR_SREV_MERLIN_20_OR_LATER(ah)) ss = (int16_t)(0 - (minPwrT4[i] / 2)); else ss = 0; /* for the first pdGain, start from index 0 */ } else { /* need overlap entries extrapolated below. */ ss = (int16_t)((pPdGainBoundaries[i-1] - (minPwrT4[i] / 2)) - tPdGainOverlap + 1 + minDelta); } vpdStep = (int16_t)(vpdTableI[i][1] - vpdTableI[i][0]); vpdStep = (int16_t)((vpdStep < 1) ? 1 : vpdStep); /* *-ve ss indicates need to extrapolate data below for this pdGain */ while ((ss < 0) && (k < (AR5416_NUM_PDADC_VALUES - 1))) { tmpVal = (int16_t)(vpdTableI[i][0] + ss * vpdStep); pPDADCValues[k++] = (uint8_t)((tmpVal < 0) ? 0 : tmpVal); ss++; } sizeCurrVpdTable = (uint8_t)((maxPwrT4[i] - minPwrT4[i]) / 2 +1); tgtIndex = (uint8_t)(pPdGainBoundaries[i] + tPdGainOverlap - (minPwrT4[i] / 2)); maxIndex = (tgtIndex < sizeCurrVpdTable) ? tgtIndex : sizeCurrVpdTable; while ((ss < maxIndex) && (k < (AR5416_NUM_PDADC_VALUES - 1))) { pPDADCValues[k++] = vpdTableI[i][ss++]; } vpdStep = (int16_t)(vpdTableI[i][sizeCurrVpdTable - 1] - vpdTableI[i][sizeCurrVpdTable - 2]); vpdStep = (int16_t)((vpdStep < 1) ? 1 : vpdStep); /* * for last gain, pdGainBoundary == Pmax_t2, so will * have to extrapolate */ if (tgtIndex >= maxIndex) { /* need to extrapolate above */ while ((ss <= tgtIndex) && (k < (AR5416_NUM_PDADC_VALUES - 1))) { tmpVal = (int16_t)((vpdTableI[i][sizeCurrVpdTable - 1] + (ss - maxIndex +1) * vpdStep)); pPDADCValues[k++] = (uint8_t)((tmpVal > 255) ? 255 : tmpVal); ss++; } } /* extrapolated above */ } /* for all pdGainUsed */ /* Fill out pdGainBoundaries - only up to 2 allowed here, but hardware allows up to 4 */ while (i < AR5416_PD_GAINS_IN_MASK) { pPdGainBoundaries[i] = pPdGainBoundaries[i-1]; i++; } while (k < AR5416_NUM_PDADC_VALUES) { pPDADCValues[k] = pPDADCValues[k-1]; k++; } return; } /* * The linux ath9k driver and (from what I've been told) the reference * Atheros driver enables the 11n PHY by default whether or not it's * configured. */ static void ar5416Set11nRegs(struct ath_hal *ah, const struct ieee80211_channel *chan) { uint32_t phymode; uint32_t enableDacFifo = 0; HAL_HT_MACMODE macmode; /* MAC - 20/40 mode */ if (AR_SREV_KITE_10_OR_LATER(ah)) enableDacFifo = (OS_REG_READ(ah, AR_PHY_TURBO) & AR_PHY_FC_ENABLE_DAC_FIFO); /* Enable 11n HT, 20 MHz */ phymode = AR_PHY_FC_HT_EN | AR_PHY_FC_SHORT_GI_40 | AR_PHY_FC_SINGLE_HT_LTF1 | AR_PHY_FC_WALSH | enableDacFifo; /* Configure baseband for dynamic 20/40 operation */ if (IEEE80211_IS_CHAN_HT40(chan)) { phymode |= AR_PHY_FC_DYN2040_EN; /* Configure control (primary) channel at +-10MHz */ if (IEEE80211_IS_CHAN_HT40U(chan)) phymode |= AR_PHY_FC_DYN2040_PRI_CH; #if 0 /* Configure 20/25 spacing */ if (ht->ht_extprotspacing == HAL_HT_EXTPROTSPACING_25) phymode |= AR_PHY_FC_DYN2040_EXT_CH; #endif macmode = HAL_HT_MACMODE_2040; } else macmode = HAL_HT_MACMODE_20; OS_REG_WRITE(ah, AR_PHY_TURBO, phymode); /* Configure MAC for 20/40 operation */ ar5416Set11nMac2040(ah, macmode); /* global transmit timeout (25 TUs default)*/ /* XXX - put this elsewhere??? */ OS_REG_WRITE(ah, AR_GTXTO, 25 << AR_GTXTO_TIMEOUT_LIMIT_S) ; /* carrier sense timeout */ OS_REG_SET_BIT(ah, AR_GTTM, AR_GTTM_CST_USEC); OS_REG_WRITE(ah, AR_CST, 0xF << AR_CST_TIMEOUT_LIMIT_S); } void ar5416GetChannelCenters(struct ath_hal *ah, const struct ieee80211_channel *chan, CHAN_CENTERS *centers) { uint16_t freq = ath_hal_gethwchannel(ah, chan); centers->ctl_center = freq; centers->synth_center = freq; /* * In 20/40 phy mode, the center frequency is * "between" the control and extension channels. */ if (IEEE80211_IS_CHAN_HT40U(chan)) { centers->synth_center += HT40_CHANNEL_CENTER_SHIFT; centers->ext_center = centers->synth_center + HT40_CHANNEL_CENTER_SHIFT; } else if (IEEE80211_IS_CHAN_HT40D(chan)) { centers->synth_center -= HT40_CHANNEL_CENTER_SHIFT; centers->ext_center = centers->synth_center - HT40_CHANNEL_CENTER_SHIFT; } else { centers->ext_center = freq; } } /* * Override the INI vals being programmed. */ static void ar5416OverrideIni(struct ath_hal *ah, const struct ieee80211_channel *chan) { uint32_t val; /* * Set the RX_ABORT and RX_DIS and clear if off only after * RXE is set for MAC. This prevents frames with corrupted * descriptor status. */ OS_REG_SET_BIT(ah, AR_DIAG_SW, (AR_DIAG_RX_DIS | AR_DIAG_RX_ABORT)); if (AR_SREV_MERLIN_20_OR_LATER(ah)) { val = OS_REG_READ(ah, AR_PCU_MISC_MODE2); if (!AR_SREV_9271(ah)) val &= ~AR_PCU_MISC_MODE2_HWWAR1; if (AR_SREV_9287_11_OR_LATER(ah)) val = val & (~AR_PCU_MISC_MODE2_HWWAR2); OS_REG_WRITE(ah, AR_PCU_MISC_MODE2, val); } /* * The AR5416 initvals have this already set to 0x11; AR9160 has * the register set to 0x0. Figure out whether AR9130/AR9160 needs * this before moving forward with it. */ #if 0 /* Disable BB clock gating for AR5416v2, AR9130, AR9160 */ if (AR_SREV_OWL_20_OR_LATER(ah) || AR_SREV_HOWL(ah) || AR_SREV_SOWL(ah)) { /* * Disable BB clock gating * Necessary to avoid issues on AR5416 2.0 */ OS_REG_WRITE(ah, 0x9800 + (651 << 2), 0x11); } #endif /* * Disable RIFS search on some chips to avoid baseband * hang issues. */ if (AR_SREV_HOWL(ah) || AR_SREV_SOWL(ah)) { val = OS_REG_READ(ah, AR_PHY_HEAVY_CLIP_FACTOR_RIFS); val &= ~AR_PHY_RIFS_INIT_DELAY; OS_REG_WRITE(ah, AR_PHY_HEAVY_CLIP_FACTOR_RIFS, val); } } struct ini { uint32_t *data; /* NB: !const */ int rows, cols; }; /* * Override XPA bias level based on operating frequency. * This is a v14 EEPROM specific thing for the AR9160. */ void ar5416EepromSetAddac(struct ath_hal *ah, const struct ieee80211_channel *chan) { #define XPA_LVL_FREQ(cnt) (pModal->xpaBiasLvlFreq[cnt]) MODAL_EEP_HEADER *pModal; HAL_EEPROM_v14 *ee = AH_PRIVATE(ah)->ah_eeprom; struct ar5416eeprom *eep = &ee->ee_base; uint8_t biaslevel; if (! AR_SREV_SOWL(ah)) return; if (EEP_MINOR(ah) < AR5416_EEP_MINOR_VER_7) return; pModal = &(eep->modalHeader[IEEE80211_IS_CHAN_2GHZ(chan)]); if (pModal->xpaBiasLvl != 0xff) biaslevel = pModal->xpaBiasLvl; else { uint16_t resetFreqBin, freqBin, freqCount = 0; CHAN_CENTERS centers; ar5416GetChannelCenters(ah, chan, ¢ers); resetFreqBin = FREQ2FBIN(centers.synth_center, IEEE80211_IS_CHAN_2GHZ(chan)); freqBin = XPA_LVL_FREQ(0) & 0xff; biaslevel = (uint8_t) (XPA_LVL_FREQ(0) >> 14); freqCount++; while (freqCount < 3) { if (XPA_LVL_FREQ(freqCount) == 0x0) break; freqBin = XPA_LVL_FREQ(freqCount) & 0xff; if (resetFreqBin >= freqBin) biaslevel = (uint8_t)(XPA_LVL_FREQ(freqCount) >> 14); else break; freqCount++; } } HALDEBUG(ah, HAL_DEBUG_EEPROM, "%s: overriding XPA bias level = %d\n", __func__, biaslevel); /* * This is a dirty workaround for the const initval data, * which will upset multiple AR9160's on the same board. * * The HAL should likely just have a private copy of the addac * data per instance. */ if (IEEE80211_IS_CHAN_2GHZ(chan)) HAL_INI_VAL((struct ini *) &AH5416(ah)->ah_ini_addac, 7, 1) = (HAL_INI_VAL(&AH5416(ah)->ah_ini_addac, 7, 1) & (~0x18)) | biaslevel << 3; else HAL_INI_VAL((struct ini *) &AH5416(ah)->ah_ini_addac, 6, 1) = (HAL_INI_VAL(&AH5416(ah)->ah_ini_addac, 6, 1) & (~0xc0)) | biaslevel << 6; #undef XPA_LVL_FREQ } static void ar5416MarkPhyInactive(struct ath_hal *ah) { OS_REG_WRITE(ah, AR_PHY_ACTIVE, AR_PHY_ACTIVE_DIS); } Index: projects/largeSMP/sys/dev/ath/ath_hal/ar5416/ar5416reg.h =================================================================== --- projects/largeSMP/sys/dev/ath/ath_hal/ar5416/ar5416reg.h (revision 221494) +++ projects/largeSMP/sys/dev/ath/ath_hal/ar5416/ar5416reg.h (revision 221495) @@ -1,680 +1,713 @@ /* * Copyright (c) 2002-2008 Sam Leffler, Errno Consulting * Copyright (c) 2002-2008 Atheros Communications, Inc. * * Permission to use, copy, modify, and/or distribute this software for any * purpose with or without fee is hereby granted, provided that the above * copyright notice and this permission notice appear in all copies. * * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. * * $FreeBSD$ */ #ifndef _DEV_ATH_AR5416REG_H #define _DEV_ATH_AR5416REG_H #include /* * Register added starting with the AR5416 */ #define AR_MIRT 0x0020 /* interrupt rate threshold */ #define AR_TIMT 0x0028 /* Tx Interrupt mitigation threshold */ #define AR_RIMT 0x002C /* Rx Interrupt mitigation threshold */ #define AR_GTXTO 0x0064 /* global transmit timeout */ #define AR_GTTM 0x0068 /* global transmit timeout mode */ #define AR_CST 0x006C /* carrier sense timeout */ #define AR_MAC_LED 0x1f04 /* LED control */ #define AR_WA 0x4004 /* PCIE work-arounds */ #define AR_PCIE_PM_CTRL 0x4014 #define AR_AHB_MODE 0x4024 /* AHB mode for dma */ #define AR_INTR_SYNC_CAUSE_CLR 0x4028 /* clear interrupt */ #define AR_INTR_SYNC_CAUSE 0x4028 /* check pending interrupts */ #define AR_INTR_SYNC_ENABLE 0x402c /* enable interrupts */ #define AR_INTR_ASYNC_MASK 0x4030 /* asynchronous interrupt mask */ #define AR_INTR_SYNC_MASK 0x4034 /* synchronous interrupt mask */ #define AR_INTR_ASYNC_CAUSE 0x4038 /* check pending interrupts */ #define AR_INTR_ASYNC_ENABLE 0x403c /* enable interrupts */ #define AR5416_PCIE_SERDES 0x4040 #define AR5416_PCIE_SERDES2 0x4044 #define AR_GPIO_IN_OUT 0x4048 /* GPIO input/output register */ #define AR_GPIO_OE_OUT 0x404c /* GPIO output enable register */ #define AR_GPIO_INTR_POL 0x4050 /* GPIO interrupt polarity */ #define AR_GPIO_INPUT_EN_VAL 0x4054 /* GPIO input enable and value */ #define AR_GPIO_INPUT_MUX1 0x4058 #define AR_GPIO_INPUT_MUX2 0x405c #define AR_GPIO_OUTPUT_MUX1 0x4060 #define AR_GPIO_OUTPUT_MUX2 0x4064 #define AR_GPIO_OUTPUT_MUX3 0x4068 #define AR_EEPROM_STATUS_DATA 0x407c #define AR_OBS 0x4080 #ifdef AH_SUPPORT_AR9130 #define AR_RTC_BASE 0x20000 #else #define AR_RTC_BASE 0x7000 #endif /* AH_SUPPORT_AR9130 */ #define AR_RTC_RC AR_RTC_BASE + 0x00 /* reset control */ #define AR_RTC_PLL_CONTROL AR_RTC_BASE + 0x14 #define AR_RTC_RESET AR_RTC_BASE + 0x40 /* RTC reset register */ #define AR_RTC_STATUS AR_RTC_BASE + 0x44 /* system sleep status */ #define AR_RTC_SLEEP_CLK AR_RTC_BASE + 0x48 #define AR_RTC_FORCE_WAKE AR_RTC_BASE + 0x4c /* control MAC force wake */ #define AR_RTC_INTR_CAUSE AR_RTC_BASE + 0x50 /* RTC interrupt cause/clear */ #define AR_RTC_INTR_ENABLE AR_RTC_BASE + 0x54 /* RTC interrupt enable */ #define AR_RTC_INTR_MASK AR_RTC_BASE + 0x58 /* RTC interrupt mask */ #ifdef AH_SUPPORT_AR9130 /* RTC_DERIVED_* - only for AR9130 */ #define AR_RTC_DERIVED_CLK (AR_RTC_BASE + 0x0038) #define AR_RTC_DERIVED_CLK_PERIOD 0x0000fffe #define AR_RTC_DERIVED_CLK_PERIOD_S 1 #endif /* AH_SUPPORT_AR9130 */ /* AR9280: rf long shift registers */ #define AR_AN_RF2G1_CH0 0x7810 #define AR_AN_RF5G1_CH0 0x7818 #define AR_AN_RF2G1_CH1 0x7834 #define AR_AN_RF5G1_CH1 0x783C #define AR_AN_TOP2 0x7894 #define AR_AN_SYNTH9 0x7868 #define AR9285_AN_RF2G1 0x7820 #define AR9285_AN_RF2G2 0x7824 #define AR9285_AN_RF2G3 0x7828 #define AR9285_AN_RF2G4 0x782C #define AR9285_AN_RF2G6 0x7834 #define AR9285_AN_RF2G7 0x7838 #define AR9285_AN_RF2G8 0x783C #define AR9285_AN_RF2G9 0x7840 #define AR9285_AN_RXTXBB1 0x7854 #define AR9285_AN_TOP2 0x7868 #define AR9285_AN_TOP3 0x786c #define AR9285_AN_TOP4 0x7870 #define AR9285_AN_TOP4_DEFAULT 0x10142c00 #define AR_RESET_TSF 0x8020 #define AR_RXFIFO_CFG 0x8114 #define AR_PHY_ERR_1 0x812c #define AR_PHY_ERR_MASK_1 0x8130 /* mask for AR_PHY_ERR_1 */ #define AR_PHY_ERR_2 0x8134 #define AR_PHY_ERR_MASK_2 0x8138 /* mask for AR_PHY_ERR_2 */ #define AR_TSFOOR_THRESHOLD 0x813c #define AR_PHY_ERR_3 0x8168 #define AR_PHY_ERR_MASK_3 0x816c /* mask for AR_PHY_ERR_3 */ #define AR_TXOP_X 0x81ec /* txop for legacy non-qos */ #define AR_TXOP_0_3 0x81f0 /* txop for various tid's */ #define AR_TXOP_4_7 0x81f4 #define AR_TXOP_8_11 0x81f8 #define AR_TXOP_12_15 0x81fc /* generic timers based on tsf - all uS */ #define AR_NEXT_TBTT 0x8200 #define AR_NEXT_DBA 0x8204 #define AR_NEXT_SWBA 0x8208 #define AR_NEXT_CFP 0x8208 #define AR_NEXT_HCF 0x820C #define AR_NEXT_TIM 0x8210 #define AR_NEXT_DTIM 0x8214 #define AR_NEXT_QUIET 0x8218 #define AR_NEXT_NDP 0x821C #define AR5416_BEACON_PERIOD 0x8220 #define AR_DBA_PERIOD 0x8224 #define AR_SWBA_PERIOD 0x8228 #define AR_HCF_PERIOD 0x822C #define AR_TIM_PERIOD 0x8230 #define AR_DTIM_PERIOD 0x8234 #define AR_QUIET_PERIOD 0x8238 #define AR_NDP_PERIOD 0x823C #define AR_TIMER_MODE 0x8240 #define AR_SLP32_MODE 0x8244 #define AR_SLP32_WAKE 0x8248 #define AR_SLP32_INC 0x824c #define AR_SLP_CNT 0x8250 /* 32kHz cycles with mac asleep */ #define AR_SLP_CYCLE_CNT 0x8254 /* absolute number of 32kHz cycles */ #define AR_SLP_MIB_CTRL 0x8258 #define AR_2040_MODE 0x8318 #define AR_EXTRCCNT 0x8328 /* extension channel rx clear count */ #define AR_SELFGEN_MASK 0x832c /* rx and cal chain masks */ #define AR_PCU_TXBUF_CTRL 0x8340 #define AR_PCU_MISC_MODE2 0x8344 /* DMA & PCI Registers in PCI space (usable during sleep)*/ #define AR_RC_AHB 0x00000001 /* AHB reset */ #define AR_RC_APB 0x00000002 /* APB reset */ #define AR_RC_HOSTIF 0x00000100 /* host interface reset */ #define AR_MIRT_VAL 0x0000ffff /* in uS */ #define AR_MIRT_VAL_S 16 #define AR_TIMT_LAST 0x0000ffff /* Last packet threshold */ #define AR_TIMT_LAST_S 0 #define AR_TIMT_FIRST 0xffff0000 /* First packet threshold */ #define AR_TIMT_FIRST_S 16 #define AR_RIMT_LAST 0x0000ffff /* Last packet threshold */ #define AR_RIMT_LAST_S 0 #define AR_RIMT_FIRST 0xffff0000 /* First packet threshold */ #define AR_RIMT_FIRST_S 16 #define AR_GTXTO_TIMEOUT_COUNTER 0x0000FFFF // Mask for timeout counter (in TUs) #define AR_GTXTO_TIMEOUT_LIMIT 0xFFFF0000 // Mask for timeout limit (in TUs) #define AR_GTXTO_TIMEOUT_LIMIT_S 16 // Shift for timeout limit #define AR_GTTM_USEC 0x00000001 // usec strobe #define AR_GTTM_IGNORE_IDLE 0x00000002 // ignore channel idle #define AR_GTTM_RESET_IDLE 0x00000004 // reset counter on channel idle low #define AR_GTTM_CST_USEC 0x00000008 // CST usec strobe #define AR_CST_TIMEOUT_COUNTER 0x0000FFFF // Mask for timeout counter (in TUs) #define AR_CST_TIMEOUT_LIMIT 0xFFFF0000 // Mask for timeout limit (in TUs) #define AR_CST_TIMEOUT_LIMIT_S 16 // Shift for timeout limit /* MAC tx DMA size config */ #define AR_TXCFG_DMASZ_MASK 0x00000003 #define AR_TXCFG_DMASZ_4B 0 #define AR_TXCFG_DMASZ_8B 1 #define AR_TXCFG_DMASZ_16B 2 #define AR_TXCFG_DMASZ_32B 3 #define AR_TXCFG_DMASZ_64B 4 #define AR_TXCFG_DMASZ_128B 5 #define AR_TXCFG_DMASZ_256B 6 #define AR_TXCFG_DMASZ_512B 7 #define AR_TXCFG_ATIM_TXPOLICY 0x00000800 /* MAC rx DMA size config */ #define AR_RXCFG_DMASZ_MASK 0x00000007 #define AR_RXCFG_DMASZ_4B 0 #define AR_RXCFG_DMASZ_8B 1 #define AR_RXCFG_DMASZ_16B 2 #define AR_RXCFG_DMASZ_32B 3 #define AR_RXCFG_DMASZ_64B 4 #define AR_RXCFG_DMASZ_128B 5 #define AR_RXCFG_DMASZ_256B 6 #define AR_RXCFG_DMASZ_512B 7 /* MAC Led registers */ +#define AR_CFG_SCLK_RATE_IND 0x00000003 /* sleep clock indication */ +#define AR_CFG_SCLK_RATE_IND_S 0 +#define AR_CFG_SCLK_32MHZ 0x00000000 /* Sleep clock rate */ +#define AR_CFG_SCLK_4MHZ 0x00000001 /* Sleep clock rate */ +#define AR_CFG_SCLK_1MHZ 0x00000002 /* Sleep clock rate */ +#define AR_CFG_SCLK_32KHZ 0x00000003 /* Sleep clock rate */ #define AR_MAC_LED_BLINK_SLOW 0x00000008 /* LED slowest blink rate mode */ #define AR_MAC_LED_BLINK_THRESH_SEL 0x00000070 /* LED blink threshold select */ #define AR_MAC_LED_MODE 0x00000380 /* LED mode select */ #define AR_MAC_LED_MODE_S 7 #define AR_MAC_LED_MODE_PROP 0 /* Blink prop to filtered tx/rx */ #define AR_MAC_LED_MODE_RPROP 1 /* Blink prop to unfiltered tx/rx */ #define AR_MAC_LED_MODE_SPLIT 2 /* Blink power for tx/net for rx */ #define AR_MAC_LED_MODE_RAND 3 /* Blink randomly */ #define AR_MAC_LED_MODE_POWON 5 /* Power LED on (s/w control) */ #define AR_MAC_LED_MODE_NETON 6 /* Network LED on (s/w control) */ #define AR_MAC_LED_ASSOC 0x00000c00 #define AR_MAC_LED_ASSOC_NONE 0x00000000 /* STA is not associated or trying */ #define AR_MAC_LED_ASSOC_ACTIVE 0x00000400 /* STA is associated */ #define AR_MAC_LED_ASSOC_PEND 0x00000800 /* STA is trying to associate */ #define AR_MAC_LED_ASSOC_S 10 #define AR_WA_UNTIE_RESET_EN 0x00008000 /* ena PCI reset to POR */ #define AR_WA_RESET_EN 0x00040000 /* ena AR_WA_UNTIE_RESET_EN */ #define AR_WA_ANALOG_SHIFT 0x00100000 #define AR_WA_POR_SHORT 0x00200000 /* PCIE phy reset control */ #define AR_WA_DEFAULT 0x0000073f #define AR9280_WA_DEFAULT 0x0040073f #define AR9285_WA_DEFAULT 0x004a05cb #define AR_PCIE_PM_CTRL_ENA 0x00080000 #define AR_AHB_EXACT_WR_EN 0x00000000 /* write exact bytes */ #define AR_AHB_BUF_WR_EN 0x00000001 /* buffer write upto cacheline*/ #define AR_AHB_EXACT_RD_EN 0x00000000 /* read exact bytes */ #define AR_AHB_CACHELINE_RD_EN 0x00000002 /* read upto end of cacheline */ #define AR_AHB_PREFETCH_RD_EN 0x00000004 /* prefetch upto page boundary*/ #define AR_AHB_PAGE_SIZE_1K 0x00000000 /* set page-size as 1k */ #define AR_AHB_PAGE_SIZE_2K 0x00000008 /* set page-size as 2k */ #define AR_AHB_PAGE_SIZE_4K 0x00000010 /* set page-size as 4k */ /* MAC PCU Registers */ #define AR_STA_ID1_PRESERVE_SEQNUM 0x20000000 /* Don't replace seq num */ /* Extended PCU DIAG_SW control fields */ #define AR_DIAG_DUAL_CHAIN_INFO 0x01000000 /* dual chain channel info */ #define AR_DIAG_RX_ABORT 0x02000000 /* abort rx */ #define AR_DIAG_SATURATE_CCNT 0x04000000 /* sat. cycle cnts (no shift) */ #define AR_DIAG_OBS_PT_SEL2 0x08000000 /* observation point sel */ #define AR_DIAG_RXCLEAR_CTL_LOW 0x10000000 /* force rx_clear(ctl) low/busy */ #define AR_DIAG_RXCLEAR_EXT_LOW 0x20000000 /* force rx_clear(ext) low/busy */ #define AR_TXOP_X_VAL 0x000000FF #define AR_RESET_TSF_ONCE 0x01000000 /* reset tsf once; self-clears*/ /* Interrupts */ #define AR_ISR_TXMINTR 0x00080000 /* Maximum interrupt tx rate */ #define AR_ISR_RXMINTR 0x01000000 /* Maximum interrupt rx rate */ #define AR_ISR_TXINTM 0x40000000 /* Tx int after mitigation */ #define AR_ISR_RXINTM 0x80000000 /* Rx int after mitigation */ #define AR_ISR_S2_CST 0x00400000 /* Carrier sense timeout */ #define AR_ISR_S2_GTT 0x00800000 /* Global transmit timeout */ #define AR_ISR_S2_TSFOOR 0x40000000 /* RX TSF out of range */ #define AR_ISR_S5 0x0098 #define AR_ISR_S5_S 0x00d8 #define AR_ISR_S5_TIM_TIMER 0x00000010 #define AR_INTR_SPURIOUS 0xffffffff #define AR_INTR_RTC_IRQ 0x00000001 /* rtc in shutdown state */ #define AR_INTR_MAC_IRQ 0x00000002 /* pending mac interrupt */ #define AR_INTR_EEP_PROT_ACCESS 0x00000004 /* eeprom protected access */ #define AR_INTR_MAC_AWAKE 0x00020000 /* mac is awake */ #define AR_INTR_MAC_ASLEEP 0x00040000 /* mac is asleep */ /* Interrupt Mask Registers */ #define AR_IMR_TXMINTR 0x00080000 /* Maximum interrupt tx rate */ #define AR_IMR_RXMINTR 0x01000000 /* Maximum interrupt rx rate */ #define AR_IMR_TXINTM 0x40000000 /* Tx int after mitigation */ #define AR_IMR_RXINTM 0x80000000 /* Rx int after mitigation */ #define AR_IMR_S2_CST 0x00400000 /* Carrier sense timeout */ #define AR_IMR_S2_GTT 0x00800000 /* Global transmit timeout */ /* synchronous interrupt signals */ #define AR_INTR_SYNC_RTC_IRQ 0x00000001 #define AR_INTR_SYNC_MAC_IRQ 0x00000002 #define AR_INTR_SYNC_EEPROM_ILLEGAL_ACCESS 0x00000004 #define AR_INTR_SYNC_APB_TIMEOUT 0x00000008 #define AR_INTR_SYNC_PCI_MODE_CONFLICT 0x00000010 #define AR_INTR_SYNC_HOST1_FATAL 0x00000020 #define AR_INTR_SYNC_HOST1_PERR 0x00000040 #define AR_INTR_SYNC_TRCV_FIFO_PERR 0x00000080 #define AR_INTR_SYNC_RADM_CPL_EP 0x00000100 #define AR_INTR_SYNC_RADM_CPL_DLLP_ABORT 0x00000200 #define AR_INTR_SYNC_RADM_CPL_TLP_ABORT 0x00000400 #define AR_INTR_SYNC_RADM_CPL_ECRC_ERR 0x00000800 #define AR_INTR_SYNC_RADM_CPL_TIMEOUT 0x00001000 #define AR_INTR_SYNC_LOCAL_TIMEOUT 0x00002000 #define AR_INTR_SYNC_PM_ACCESS 0x00004000 #define AR_INTR_SYNC_MAC_AWAKE 0x00008000 #define AR_INTR_SYNC_MAC_ASLEEP 0x00010000 #define AR_INTR_SYNC_MAC_SLEEP_ACCESS 0x00020000 #define AR_INTR_SYNC_ALL 0x0003FFFF /* default synchronous interrupt signals enabled */ #define AR_INTR_SYNC_DEFAULT \ (AR_INTR_SYNC_HOST1_FATAL | AR_INTR_SYNC_HOST1_PERR | \ AR_INTR_SYNC_RADM_CPL_EP | AR_INTR_SYNC_RADM_CPL_DLLP_ABORT | \ AR_INTR_SYNC_RADM_CPL_TLP_ABORT | AR_INTR_SYNC_RADM_CPL_ECRC_ERR | \ AR_INTR_SYNC_RADM_CPL_TIMEOUT | AR_INTR_SYNC_LOCAL_TIMEOUT | \ AR_INTR_SYNC_MAC_SLEEP_ACCESS) #define AR_INTR_SYNC_MASK_GPIO 0xFFFC0000 #define AR_INTR_SYNC_MASK_GPIO_S 18 #define AR_INTR_SYNC_ENABLE_GPIO 0xFFFC0000 #define AR_INTR_SYNC_ENABLE_GPIO_S 18 #define AR_INTR_ASYNC_MASK_GPIO 0xFFFC0000 /* async int mask */ #define AR_INTR_ASYNC_MASK_GPIO_S 18 #define AR_INTR_ASYNC_CAUSE_GPIO 0xFFFC0000 /* GPIO interrupts */ #define AR_INTR_ASYNC_USED (AR_INTR_MAC_IRQ | AR_INTR_ASYNC_CAUSE_GPIO) #define AR_INTR_ASYNC_ENABLE_GPIO 0xFFFC0000 /* enable interrupts */ #define AR_INTR_ASYNC_ENABLE_GPIO_S 18 /* RTC registers */ #define AR_RTC_RC_M 0x00000003 #define AR_RTC_RC_MAC_WARM 0x00000001 #define AR_RTC_RC_MAC_COLD 0x00000002 #ifdef AH_SUPPORT_AR9130 #define AR_RTC_RC_COLD_RESET 0x00000004 #define AR_RTC_RC_WARM_RESET 0x00000008 #endif /* AH_SUPPORT_AR9130 */ #define AR_RTC_PLL_DIV 0x0000001f #define AR_RTC_PLL_DIV_S 0 #define AR_RTC_PLL_DIV2 0x00000020 #define AR_RTC_PLL_REFDIV_5 0x000000c0 #define AR_RTC_SOWL_PLL_DIV 0x000003ff #define AR_RTC_SOWL_PLL_DIV_S 0 #define AR_RTC_SOWL_PLL_REFDIV 0x00003C00 #define AR_RTC_SOWL_PLL_REFDIV_S 10 #define AR_RTC_SOWL_PLL_CLKSEL 0x0000C000 #define AR_RTC_SOWL_PLL_CLKSEL_S 14 #define AR_RTC_RESET_EN 0x00000001 /* Reset RTC bit */ #define AR_RTC_PM_STATUS_M 0x0000000f /* Pwr Mgmt Status */ #ifdef AH_SUPPORT_AR9130 #define AR_RTC_STATUS_M 0x0000000f /* RTC Status */ #else #define AR_RTC_STATUS_M 0x0000003f /* RTC Status */ #endif /* AH_SUPPORT_AR9130 */ #define AR_RTC_STATUS_SHUTDOWN 0x00000001 #define AR_RTC_STATUS_ON 0x00000002 #define AR_RTC_STATUS_SLEEP 0x00000004 #define AR_RTC_STATUS_WAKEUP 0x00000008 #define AR_RTC_STATUS_COLDRESET 0x00000010 /* Not currently used */ #define AR_RTC_STATUS_PLLCHANGE 0x00000020 /* Not currently used */ #define AR_RTC_SLEEP_DERIVED_CLK 0x2 #define AR_RTC_FORCE_WAKE_EN 0x00000001 /* enable force wake */ #define AR_RTC_FORCE_WAKE_ON_INT 0x00000002 /* auto-wake on MAC interrupt */ #define AR_RTC_PLL_CLKSEL 0x00000300 #define AR_RTC_PLL_CLKSEL_S 8 /* AR9280: rf long shift registers */ #define AR_AN_RF2G1_CH0_OB 0x03800000 #define AR_AN_RF2G1_CH0_OB_S 23 #define AR_AN_RF2G1_CH0_DB 0x1C000000 #define AR_AN_RF2G1_CH0_DB_S 26 #define AR_AN_RF5G1_CH0_OB5 0x00070000 #define AR_AN_RF5G1_CH0_OB5_S 16 #define AR_AN_RF5G1_CH0_DB5 0x00380000 #define AR_AN_RF5G1_CH0_DB5_S 19 #define AR_AN_RF2G1_CH1_OB 0x03800000 #define AR_AN_RF2G1_CH1_OB_S 23 #define AR_AN_RF2G1_CH1_DB 0x1C000000 #define AR_AN_RF2G1_CH1_DB_S 26 #define AR_AN_RF5G1_CH1_OB5 0x00070000 #define AR_AN_RF5G1_CH1_OB5_S 16 #define AR_AN_RF5G1_CH1_DB5 0x00380000 #define AR_AN_RF5G1_CH1_DB5_S 19 #define AR_AN_TOP1 0x7890 #define AR_AN_TOP1_DACIPMODE 0x00040000 #define AR_AN_TOP1_DACIPMODE_S 18 #define AR_AN_TOP2_XPABIAS_LVL 0xC0000000 #define AR_AN_TOP2_XPABIAS_LVL_S 30 #define AR_AN_TOP2_LOCALBIAS 0x00200000 #define AR_AN_TOP2_LOCALBIAS_S 21 #define AR_AN_TOP2_PWDCLKIND 0x00400000 #define AR_AN_TOP2_PWDCLKIND_S 22 #define AR_AN_SYNTH9_REFDIVA 0xf8000000 #define AR_AN_SYNTH9_REFDIVA_S 27 /* AR9285 Analog registers */ #define AR9285_AN_RF2G1_ENPACAL 0x00000800 #define AR9285_AN_RF2G1_ENPACAL_S 11 #define AR9285_AN_RF2G1_PDPADRV1 0x02000000 #define AR9285_AN_RF2G1_PDPADRV1_S 25 #define AR9285_AN_RF2G1_PDPADRV2 0x01000000 #define AR9285_AN_RF2G1_PDPADRV2_S 24 #define AR9285_AN_RF2G1_PDPAOUT 0x00800000 #define AR9285_AN_RF2G1_PDPAOUT_S 23 #define AR9285_AN_RF2G2_OFFCAL 0x00001000 #define AR9285_AN_RF2G2_OFFCAL_S 12 #define AR9285_AN_RF2G3_PDVCCOMP 0x02000000 #define AR9285_AN_RF2G3_PDVCCOMP_S 25 #define AR9285_AN_RF2G3_OB_0 0x00E00000 #define AR9285_AN_RF2G3_OB_0_S 21 #define AR9285_AN_RF2G3_OB_1 0x001C0000 #define AR9285_AN_RF2G3_OB_1_S 18 #define AR9285_AN_RF2G3_OB_2 0x00038000 #define AR9285_AN_RF2G3_OB_2_S 15 #define AR9285_AN_RF2G3_OB_3 0x00007000 #define AR9285_AN_RF2G3_OB_3_S 12 #define AR9285_AN_RF2G3_OB_4 0x00000E00 #define AR9285_AN_RF2G3_OB_4_S 9 #define AR9285_AN_RF2G3_DB1_0 0x000001C0 #define AR9285_AN_RF2G3_DB1_0_S 6 #define AR9285_AN_RF2G3_DB1_1 0x00000038 #define AR9285_AN_RF2G3_DB1_1_S 3 #define AR9285_AN_RF2G3_DB1_2 0x00000007 #define AR9285_AN_RF2G3_DB1_2_S 0 #define AR9285_AN_RF2G4_DB1_3 0xE0000000 #define AR9285_AN_RF2G4_DB1_3_S 29 #define AR9285_AN_RF2G4_DB1_4 0x1C000000 #define AR9285_AN_RF2G4_DB1_4_S 26 #define AR9285_AN_RF2G4_DB2_0 0x03800000 #define AR9285_AN_RF2G4_DB2_0_S 23 #define AR9285_AN_RF2G4_DB2_1 0x00700000 #define AR9285_AN_RF2G4_DB2_1_S 20 #define AR9285_AN_RF2G4_DB2_2 0x000E0000 #define AR9285_AN_RF2G4_DB2_2_S 17 #define AR9285_AN_RF2G4_DB2_3 0x0001C000 #define AR9285_AN_RF2G4_DB2_3_S 14 #define AR9285_AN_RF2G4_DB2_4 0x00003800 #define AR9285_AN_RF2G4_DB2_4_S 11 #define AR9285_AN_RF2G6_CCOMP 0x00007800 #define AR9285_AN_RF2G6_CCOMP_S 11 #define AR9285_AN_RF2G6_OFFS 0x03f00000 #define AR9285_AN_RF2G6_OFFS_S 20 #define AR9271_AN_RF2G6_OFFS 0x07f00000 #define AR9271_AN_RF2G6_OFFS_S 20 #define AR9285_AN_RF2G7_PWDDB 0x00000002 #define AR9285_AN_RF2G7_PWDDB_S 1 #define AR9285_AN_RF2G7_PADRVGN2TAB0 0xE0000000 #define AR9285_AN_RF2G7_PADRVGN2TAB0_S 29 #define AR9285_AN_RF2G8_PADRVGN2TAB0 0x0001C000 #define AR9285_AN_RF2G8_PADRVGN2TAB0_S 14 #define AR9285_AN_RXTXBB1_PDRXTXBB1 0x00000020 #define AR9285_AN_RXTXBB1_PDRXTXBB1_S 5 #define AR9285_AN_RXTXBB1_PDV2I 0x00000080 #define AR9285_AN_RXTXBB1_PDV2I_S 7 #define AR9285_AN_RXTXBB1_PDDACIF 0x00000100 #define AR9285_AN_RXTXBB1_PDDACIF_S 8 #define AR9285_AN_RXTXBB1_SPARE9 0x00000001 #define AR9285_AN_RXTXBB1_SPARE9_S 0 #define AR9285_AN_TOP3_XPABIAS_LVL 0x0000000C #define AR9285_AN_TOP3_XPABIAS_LVL_S 2 #define AR9285_AN_TOP3_PWDDAC 0x00800000 #define AR9285_AN_TOP3_PWDDAC_S 23 /* Sleep control */ #define AR5416_SLEEP1_CAB_TIMEOUT 0xFFE00000 /* Cab timeout (TU) */ #define AR5416_SLEEP1_CAB_TIMEOUT_S 22 #define AR5416_SLEEP2_BEACON_TIMEOUT 0xFFE00000 /* Beacon timeout (TU)*/ #define AR5416_SLEEP2_BEACON_TIMEOUT_S 22 /* Sleep Registers */ #define AR_SLP32_HALFCLK_LATENCY 0x000FFFFF /* rising <-> falling edge */ #define AR_SLP32_ENA 0x00100000 #define AR_SLP32_TSF_WRITE_STATUS 0x00200000 /* tsf update in progress */ #define AR_SLP32_WAKE_XTL_TIME 0x0000FFFF /* time to wake crystal */ #define AR_SLP32_TST_INC 0x000FFFFF #define AR_SLP_MIB_CLEAR 0x00000001 /* clear pending */ #define AR_SLP_MIB_PENDING 0x00000002 /* clear counters */ #define AR_TIMER_MODE_TBTT 0x00000001 #define AR_TIMER_MODE_DBA 0x00000002 #define AR_TIMER_MODE_SWBA 0x00000004 #define AR_TIMER_MODE_HCF 0x00000008 #define AR_TIMER_MODE_TIM 0x00000010 #define AR_TIMER_MODE_DTIM 0x00000020 #define AR_TIMER_MODE_QUIET 0x00000040 #define AR_TIMER_MODE_NDP 0x00000080 #define AR_TIMER_MODE_OVERFLOW_INDEX 0x00000700 #define AR_TIMER_MODE_OVERFLOW_INDEX_S 8 #define AR_TIMER_MODE_THRESH 0xFFFFF000 #define AR_TIMER_MODE_THRESH_S 12 /* PCU Misc modes */ #define AR_PCU_FORCE_BSSID_MATCH 0x00000001 /* force bssid to match */ #define AR_PCU_MIC_NEW_LOC_ENA 0x00000004 /* tx/rx mic keys together */ #define AR_PCU_TX_ADD_TSF 0x00000008 /* add tx_tsf + int_tsf */ #define AR_PCU_CCK_SIFS_MODE 0x00000010 /* assume 11b sifs */ #define AR_PCU_RX_ANT_UPDT 0x00000800 /* KC_RX_ANT_UPDATE */ #define AR_PCU_TXOP_TBTT_LIMIT_ENA 0x00001000 /* enforce txop / tbtt */ #define AR_PCU_MISS_BCN_IN_SLEEP 0x00004000 /* count bmiss's when sleeping */ #define AR_PCU_BUG_12306_FIX_ENA 0x00020000 /* use rx_clear to count sifs */ #define AR_PCU_FORCE_QUIET_COLL 0x00040000 /* kill xmit for channel change */ #define AR_PCU_TBTT_PROTECT 0x00200000 /* no xmit upto tbtt+20 uS */ #define AR_PCU_CLEAR_VMF 0x01000000 /* clear vmf mode (fast cc)*/ #define AR_PCU_CLEAR_BA_VALID 0x04000000 /* clear ba state */ #define AR_PCU_MISC_MODE2_MGMT_CRYPTO_ENABLE 0x00000002 #define AR_PCU_MISC_MODE2_NO_CRYPTO_FOR_NON_DATA_PKT 0x00000004 #define AR_PCU_MISC_MODE2_HWWAR1 0x00100000 #define AR_PCU_MISC_MODE2_HWWAR2 0x02000000 /* GPIO Interrupt */ #define AR_INTR_GPIO 0x3FF00000 /* gpio interrupted */ #define AR_INTR_GPIO_S 20 #define AR_GPIO_OUT_CTRL 0x000003FF /* 0 = out, 1 = in */ #define AR_GPIO_OUT_VAL 0x000FFC00 #define AR_GPIO_OUT_VAL_S 10 #define AR_GPIO_INTR_CTRL 0x3FF00000 #define AR_GPIO_INTR_CTRL_S 20 #define AR_GPIO_IN_VAL 0x0FFFC000 /* pre-9280 */ #define AR_GPIO_IN_VAL_S 14 #define AR928X_GPIO_IN_VAL 0x000FFC00 #define AR928X_GPIO_IN_VAL_S 10 #define AR9285_GPIO_IN_VAL 0x00FFF000 #define AR9285_GPIO_IN_VAL_S 12 #define AR_GPIO_OE_OUT_DRV 0x3 /* 2 bit mask shifted by 2*bitpos */ #define AR_GPIO_OE_OUT_DRV_NO 0x0 /* tristate */ #define AR_GPIO_OE_OUT_DRV_LOW 0x1 /* drive if low */ #define AR_GPIO_OE_OUT_DRV_HI 0x2 /* drive if high */ #define AR_GPIO_OE_OUT_DRV_ALL 0x3 /* drive always */ #define AR_GPIO_INTR_POL_VAL 0x1FFF #define AR_GPIO_INTR_POL_VAL_S 0 #define AR_GPIO_JTAG_DISABLE 0x00020000 #define AR_2040_JOINED_RX_CLEAR 0x00000001 /* use ctl + ext rx_clear for cca */ #define AR_PCU_TXBUF_CTRL_SIZE_MASK 0x7FF #define AR_PCU_TXBUF_CTRL_USABLE_SIZE 0x700 #define AR_9285_PCU_TXBUF_CTRL_USABLE_SIZE 0x380 /* Eeprom defines */ #define AR_EEPROM_STATUS_DATA_VAL 0x0000ffff #define AR_EEPROM_STATUS_DATA_VAL_S 0 #define AR_EEPROM_STATUS_DATA_BUSY 0x00010000 #define AR_EEPROM_STATUS_DATA_BUSY_ACCESS 0x00020000 #define AR_EEPROM_STATUS_DATA_PROT_ACCESS 0x00040000 #define AR_EEPROM_STATUS_DATA_ABSENT_ACCESS 0x00080000 #define AR_SREV_REVISION_OWL_10 0x08 #define AR_SREV_REVISION_OWL_20 0x09 #define AR_SREV_REVISION_OWL_22 0x0a #define AR_RAD5133_SREV_MAJOR 0xc0 /* Fowl: 2+5G/3x3 */ #define AR_RAD2133_SREV_MAJOR 0xd0 /* Fowl: 2G/3x3 */ #define AR_RAD5122_SREV_MAJOR 0xe0 /* Fowl: 5G/2x2 */ #define AR_RAD2122_SREV_MAJOR 0xf0 /* Fowl: 2+5G/2x2 */ /* Test macro for owl 1.0 */ #define IS_5416V1(_ah) ((_ah)->ah_macRev == AR_SREV_REVISION_OWL_10) #define IS_5416V2(_ah) ((_ah)->ah_macRev >= AR_SREV_REVISION_OWL_20) #define IS_5416V2_2(_ah) ((_ah)->ah_macRev == AR_SREV_REVISION_OWL_22) /* Expanded Mac Silicon Rev (16 bits starting with Sowl) */ #define AR_XSREV_ID 0xFFFFFFFF /* Chip ID */ #define AR_XSREV_ID_S 0 #define AR_XSREV_VERSION 0xFFFC0000 /* Chip version */ #define AR_XSREV_VERSION_S 18 #define AR_XSREV_TYPE 0x0003F000 /* Chip type */ #define AR_XSREV_TYPE_S 12 #define AR_XSREV_TYPE_CHAIN 0x00001000 /* Chain Mode (1:3 chains, * 0:2 chains) */ #define AR_XSREV_TYPE_HOST_MODE 0x00002000 /* Host Mode (1:PCI, 0:PCIe) */ #define AR_XSREV_REVISION 0x00000F00 #define AR_XSREV_REVISION_S 8 #define AR_XSREV_VERSION_OWL_PCI 0x0D #define AR_XSREV_VERSION_OWL_PCIE 0x0C #define AR_XSREV_REVISION_OWL_10 0 /* Owl 1.0 */ #define AR_XSREV_REVISION_OWL_20 1 /* Owl 2.0/2.1 */ #define AR_XSREV_REVISION_OWL_22 2 /* Owl 2.2 */ #define AR_XSREV_VERSION_HOWL 0x14 /* Howl (AR9130) */ #define AR_XSREV_VERSION_SOWL 0x40 /* Sowl (AR9160) */ #define AR_XSREV_REVISION_SOWL_10 0 /* Sowl 1.0 */ #define AR_XSREV_REVISION_SOWL_11 1 /* Sowl 1.1 */ #define AR_XSREV_VERSION_MERLIN 0x80 /* Merlin Version */ #define AR_XSREV_REVISION_MERLIN_10 0 /* Merlin 1.0 */ #define AR_XSREV_REVISION_MERLIN_20 1 /* Merlin 2.0 */ #define AR_XSREV_REVISION_MERLIN_21 2 /* Merlin 2.1 */ #define AR_XSREV_VERSION_KITE 0xC0 /* Kite Version */ #define AR_XSREV_REVISION_KITE_10 0 /* Kite 1.0 */ #define AR_XSREV_REVISION_KITE_11 1 /* Kite 1.1 */ #define AR_XSREV_REVISION_KITE_12 2 /* Kite 1.2 */ +/* Owl (AR5416) */ #define AR_SREV_OWL(_ah) \ ((AH_PRIVATE((_ah))->ah_macVersion == AR_XSREV_VERSION_OWL_PCI) || \ (AH_PRIVATE((_ah))->ah_macVersion == AR_XSREV_VERSION_OWL_PCIE)) #define AR_SREV_OWL_20_OR_LATER(_ah) \ - ((AR_SREV_OWL(_ah) && AH_PRIVATE((_ah))->ah_macRev >= AR_XSREV_REVISION_OWL_20) || \ - AH_PRIVATE((_ah))->ah_macVersion >= AR_XSREV_VERSION_HOWL) + ((AR_SREV_OWL(_ah) && \ + AH_PRIVATE((_ah))->ah_macRev >= AR_XSREV_REVISION_OWL_20) || \ + AH_PRIVATE((_ah))->ah_macVersion >= AR_XSREV_VERSION_HOWL) + #define AR_SREV_OWL_22_OR_LATER(_ah) \ - ((AR_SREV_OWL(_ah) && AH_PRIVATE((_ah))->ah_macRev >= AR_XSREV_REVISION_OWL_22) || \ - AH_PRIVATE((_ah))->ah_macVersion >= AR_XSREV_VERSION_HOWL) + ((AR_SREV_OWL(_ah) && \ + AH_PRIVATE((_ah))->ah_macRev >= AR_XSREV_REVISION_OWL_22) || \ + AH_PRIVATE((_ah))->ah_macVersion >= AR_XSREV_VERSION_HOWL) +/* Howl (AR9130) */ + #define AR_SREV_HOWL(_ah) \ (AH_PRIVATE((_ah))->ah_macVersion == AR_XSREV_VERSION_HOWL) + #define AR_SREV_9100(_ah) AR_SREV_HOWL(_ah) +/* Sowl (AR9160) */ + #define AR_SREV_SOWL(_ah) \ (AH_PRIVATE((_ah))->ah_macVersion == AR_XSREV_VERSION_SOWL) + #define AR_SREV_SOWL_10_OR_LATER(_ah) \ (AH_PRIVATE((_ah))->ah_macVersion >= AR_XSREV_VERSION_SOWL) + #define AR_SREV_SOWL_11(_ah) \ (AR_SREV_SOWL(_ah) && \ AH_PRIVATE((_ah))->ah_macRev == AR_XSREV_REVISION_SOWL_11) +/* Merlin (AR9280) */ + #define AR_SREV_MERLIN(_ah) \ (AH_PRIVATE((_ah))->ah_macVersion == AR_XSREV_VERSION_MERLIN) + #define AR_SREV_MERLIN_10_OR_LATER(_ah) \ (AH_PRIVATE((_ah))->ah_macVersion >= AR_XSREV_VERSION_MERLIN) + #define AR_SREV_MERLIN_20(_ah) \ (AR_SREV_MERLIN(_ah) && \ AH_PRIVATE((_ah))->ah_macRev == AR_XSREV_REVISION_MERLIN_20) + #define AR_SREV_MERLIN_20_OR_LATER(_ah) \ - (AR_SREV_MERLIN_20(_ah) || \ - AH_PRIVATE((_ah))->ah_macVersion >= AR_XSREV_VERSION_MERLIN) + ((AH_PRIVATE((_ah))->ah_macVersion > AR_XSREV_VERSION_MERLIN) || \ + (AR_SREV_MERLIN((_ah)) && \ + AH_PRIVATE((_ah))->ah_macRev >= AR_XSREV_REVISION_MERLIN_20)) +/* Kite (AR9285) */ + #define AR_SREV_KITE(_ah) \ (AH_PRIVATE((_ah))->ah_macVersion == AR_XSREV_VERSION_KITE) + #define AR_SREV_KITE_10_OR_LATER(_ah) \ (AH_PRIVATE((_ah))->ah_macVersion >= AR_XSREV_VERSION_KITE) + #define AR_SREV_KITE_11(_ah) \ (AR_SREV_KITE(ah) && \ AH_PRIVATE((_ah))->ah_macRev == AR_XSREV_REVISION_KITE_11) + #define AR_SREV_KITE_11_OR_LATER(_ah) \ - (AR_SREV_KITE_11(_ah) || \ - AH_PRIVATE((_ah))->ah_macRev >= AR_XSREV_REVISION_KITE_11) + ((AH_PRIVATE((_ah))->ah_macVersion > AR_XSREV_VERSION_KITE) || \ + (AR_SREV_KITE((_ah)) && \ + AH_PRIVATE((_ah))->ah_macRev >= AR_XSREV_REVISION_KITE_11)) + #define AR_SREV_KITE_12(_ah) \ (AR_SREV_KITE(ah) && \ AH_PRIVATE((_ah))->ah_macRev == AR_XSREV_REVISION_KITE_12) + #define AR_SREV_KITE_12_OR_LATER(_ah) \ - (AR_SREV_KITE_12(_ah) || \ - AH_PRIVATE((_ah))->ah_macRev >= AR_XSREV_REVISION_KITE_12) + ((AH_PRIVATE((_ah))->ah_macVersion > AR_XSREV_VERSION_KITE) || \ + (AR_SREV_KITE((_ah)) && \ + AH_PRIVATE((_ah))->ah_macRev >= AR_XSREV_REVISION_KITE_12)) + #define AR_SREV_9285E_20(_ah) \ (AR_SREV_KITE_12_OR_LATER(_ah) && \ ((OS_REG_READ(_ah, AR_AN_SYNTH9) & 0x7) == 0x1)) /* Not yet implemented chips */ #define AR_SREV_9271(_ah) 0 #define AR_SREV_9287_11_OR_LATER(_ah) 0 #endif /* _DEV_ATH_AR5416REG_H */ Index: projects/largeSMP/sys/dev/bge/if_bge.c =================================================================== --- projects/largeSMP/sys/dev/bge/if_bge.c (revision 221494) +++ projects/largeSMP/sys/dev/bge/if_bge.c (revision 221495) @@ -1,5863 +1,5904 @@ /*- * Copyright (c) 2001 Wind River Systems * Copyright (c) 1997, 1998, 1999, 2001 * Bill Paul . All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by Bill Paul. * 4. Neither the name of the author nor the names of any co-contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY Bill Paul AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL Bill Paul OR THE VOICES IN HIS HEAD * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF * THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); /* * Broadcom BCM570x family gigabit ethernet driver for FreeBSD. * * The Broadcom BCM5700 is based on technology originally developed by * Alteon Networks as part of the Tigon I and Tigon II gigabit ethernet * MAC chips. The BCM5700, sometimes referred to as the Tigon III, has * two on-board MIPS R4000 CPUs and can have as much as 16MB of external * SSRAM. The BCM5700 supports TCP, UDP and IP checksum offload, jumbo * frames, highly configurable RX filtering, and 16 RX and TX queues * (which, along with RX filter rules, can be used for QOS applications). * Other features, such as TCP segmentation, may be available as part * of value-added firmware updates. Unlike the Tigon I and Tigon II, * firmware images can be stored in hardware and need not be compiled * into the driver. * * The BCM5700 supports the PCI v2.2 and PCI-X v1.0 standards, and will * function in a 32-bit/64-bit 33/66Mhz bus, or a 64-bit/133Mhz bus. * * The BCM5701 is a single-chip solution incorporating both the BCM5700 * MAC and a BCM5401 10/100/1000 PHY. Unlike the BCM5700, the BCM5701 * does not support external SSRAM. * * Broadcom also produces a variation of the BCM5700 under the "Altima" * brand name, which is functionally similar but lacks PCI-X support. * * Without external SSRAM, you can only have at most 4 TX rings, * and the use of the mini RX ring is disabled. This seems to imply * that these features are simply not available on the BCM5701. As a * result, this driver does not implement any support for the mini RX * ring. */ #ifdef HAVE_KERNEL_OPTION_HEADERS #include "opt_device_polling.h" #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "miidevs.h" #include #ifdef __sparc64__ #include #include #include #include #endif #include #include #include #define BGE_CSUM_FEATURES (CSUM_IP | CSUM_TCP) #define ETHER_MIN_NOPAD (ETHER_MIN_LEN - ETHER_CRC_LEN) /* i.e., 60 */ MODULE_DEPEND(bge, pci, 1, 1, 1); MODULE_DEPEND(bge, ether, 1, 1, 1); MODULE_DEPEND(bge, miibus, 1, 1, 1); /* "device miibus" required. See GENERIC if you get errors here. */ #include "miibus_if.h" /* * Various supported device vendors/types and their names. Note: the * spec seems to indicate that the hardware still has Alteon's vendor * ID burned into it, though it will always be overriden by the vendor * ID in the EEPROM. Just to be safe, we cover all possibilities. */ static const struct bge_type { uint16_t bge_vid; uint16_t bge_did; } const bge_devs[] = { { ALTEON_VENDORID, ALTEON_DEVICEID_BCM5700 }, { ALTEON_VENDORID, ALTEON_DEVICEID_BCM5701 }, { ALTIMA_VENDORID, ALTIMA_DEVICE_AC1000 }, { ALTIMA_VENDORID, ALTIMA_DEVICE_AC1002 }, { ALTIMA_VENDORID, ALTIMA_DEVICE_AC9100 }, { APPLE_VENDORID, APPLE_DEVICE_BCM5701 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5700 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5701 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5702 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5702_ALT }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5702X }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5703 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5703_ALT }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5703X }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5704C }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5704S }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5704S_ALT }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5705 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5705F }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5705K }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5705M }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5705M_ALT }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5714C }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5714S }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5715 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5715S }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5717 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5718 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5720 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5721 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5722 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5723 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5750 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5750M }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5751 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5751F }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5751M }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5752 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5752M }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5753 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5753F }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5753M }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5754 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5754M }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5755 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5755M }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5756 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5761 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5761E }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5761S }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5761SE }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5764 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5780 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5780S }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5781 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5782 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5784 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5785F }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5785G }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5786 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5787 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5787F }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5787M }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5788 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5789 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5901 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5901A2 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5903M }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5906 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM5906M }, { BCOM_VENDORID, BCOM_DEVICEID_BCM57760 }, + { BCOM_VENDORID, BCOM_DEVICEID_BCM57761 }, + { BCOM_VENDORID, BCOM_DEVICEID_BCM57765 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM57780 }, + { BCOM_VENDORID, BCOM_DEVICEID_BCM57781 }, + { BCOM_VENDORID, BCOM_DEVICEID_BCM57785 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM57788 }, { BCOM_VENDORID, BCOM_DEVICEID_BCM57790 }, + { BCOM_VENDORID, BCOM_DEVICEID_BCM57791 }, + { BCOM_VENDORID, BCOM_DEVICEID_BCM57795 }, { SK_VENDORID, SK_DEVICEID_ALTIMA }, { TC_VENDORID, TC_DEVICEID_3C996 }, { FJTSU_VENDORID, FJTSU_DEVICEID_PW008GE4 }, { FJTSU_VENDORID, FJTSU_DEVICEID_PW008GE5 }, { FJTSU_VENDORID, FJTSU_DEVICEID_PP250450 }, { 0, 0 } }; static const struct bge_vendor { uint16_t v_id; const char *v_name; } const bge_vendors[] = { { ALTEON_VENDORID, "Alteon" }, { ALTIMA_VENDORID, "Altima" }, { APPLE_VENDORID, "Apple" }, { BCOM_VENDORID, "Broadcom" }, { SK_VENDORID, "SysKonnect" }, { TC_VENDORID, "3Com" }, { FJTSU_VENDORID, "Fujitsu" }, { 0, NULL } }; static const struct bge_revision { uint32_t br_chipid; const char *br_name; } const bge_revisions[] = { { BGE_CHIPID_BCM5700_A0, "BCM5700 A0" }, { BGE_CHIPID_BCM5700_A1, "BCM5700 A1" }, { BGE_CHIPID_BCM5700_B0, "BCM5700 B0" }, { BGE_CHIPID_BCM5700_B1, "BCM5700 B1" }, { BGE_CHIPID_BCM5700_B2, "BCM5700 B2" }, { BGE_CHIPID_BCM5700_B3, "BCM5700 B3" }, { BGE_CHIPID_BCM5700_ALTIMA, "BCM5700 Altima" }, { BGE_CHIPID_BCM5700_C0, "BCM5700 C0" }, { BGE_CHIPID_BCM5701_A0, "BCM5701 A0" }, { BGE_CHIPID_BCM5701_B0, "BCM5701 B0" }, { BGE_CHIPID_BCM5701_B2, "BCM5701 B2" }, { BGE_CHIPID_BCM5701_B5, "BCM5701 B5" }, { BGE_CHIPID_BCM5703_A0, "BCM5703 A0" }, { BGE_CHIPID_BCM5703_A1, "BCM5703 A1" }, { BGE_CHIPID_BCM5703_A2, "BCM5703 A2" }, { BGE_CHIPID_BCM5703_A3, "BCM5703 A3" }, { BGE_CHIPID_BCM5703_B0, "BCM5703 B0" }, { BGE_CHIPID_BCM5704_A0, "BCM5704 A0" }, { BGE_CHIPID_BCM5704_A1, "BCM5704 A1" }, { BGE_CHIPID_BCM5704_A2, "BCM5704 A2" }, { BGE_CHIPID_BCM5704_A3, "BCM5704 A3" }, { BGE_CHIPID_BCM5704_B0, "BCM5704 B0" }, { BGE_CHIPID_BCM5705_A0, "BCM5705 A0" }, { BGE_CHIPID_BCM5705_A1, "BCM5705 A1" }, { BGE_CHIPID_BCM5705_A2, "BCM5705 A2" }, { BGE_CHIPID_BCM5705_A3, "BCM5705 A3" }, { BGE_CHIPID_BCM5750_A0, "BCM5750 A0" }, { BGE_CHIPID_BCM5750_A1, "BCM5750 A1" }, { BGE_CHIPID_BCM5750_A3, "BCM5750 A3" }, { BGE_CHIPID_BCM5750_B0, "BCM5750 B0" }, { BGE_CHIPID_BCM5750_B1, "BCM5750 B1" }, { BGE_CHIPID_BCM5750_C0, "BCM5750 C0" }, { BGE_CHIPID_BCM5750_C1, "BCM5750 C1" }, { BGE_CHIPID_BCM5750_C2, "BCM5750 C2" }, { BGE_CHIPID_BCM5714_A0, "BCM5714 A0" }, { BGE_CHIPID_BCM5752_A0, "BCM5752 A0" }, { BGE_CHIPID_BCM5752_A1, "BCM5752 A1" }, { BGE_CHIPID_BCM5752_A2, "BCM5752 A2" }, { BGE_CHIPID_BCM5714_B0, "BCM5714 B0" }, { BGE_CHIPID_BCM5714_B3, "BCM5714 B3" }, { BGE_CHIPID_BCM5715_A0, "BCM5715 A0" }, { BGE_CHIPID_BCM5715_A1, "BCM5715 A1" }, { BGE_CHIPID_BCM5715_A3, "BCM5715 A3" }, { BGE_CHIPID_BCM5717_A0, "BCM5717 A0" }, { BGE_CHIPID_BCM5717_B0, "BCM5717 B0" }, { BGE_CHIPID_BCM5755_A0, "BCM5755 A0" }, { BGE_CHIPID_BCM5755_A1, "BCM5755 A1" }, { BGE_CHIPID_BCM5755_A2, "BCM5755 A2" }, { BGE_CHIPID_BCM5722_A0, "BCM5722 A0" }, { BGE_CHIPID_BCM5761_A0, "BCM5761 A0" }, { BGE_CHIPID_BCM5761_A1, "BCM5761 A1" }, { BGE_CHIPID_BCM5784_A0, "BCM5784 A0" }, { BGE_CHIPID_BCM5784_A1, "BCM5784 A1" }, /* 5754 and 5787 share the same ASIC ID */ { BGE_CHIPID_BCM5787_A0, "BCM5754/5787 A0" }, { BGE_CHIPID_BCM5787_A1, "BCM5754/5787 A1" }, { BGE_CHIPID_BCM5787_A2, "BCM5754/5787 A2" }, { BGE_CHIPID_BCM5906_A1, "BCM5906 A1" }, { BGE_CHIPID_BCM5906_A2, "BCM5906 A2" }, + { BGE_CHIPID_BCM57765_A0, "BCM57765 A0" }, + { BGE_CHIPID_BCM57765_B0, "BCM57765 B0" }, { BGE_CHIPID_BCM57780_A0, "BCM57780 A0" }, { BGE_CHIPID_BCM57780_A1, "BCM57780 A1" }, { 0, NULL } }; /* * Some defaults for major revisions, so that newer steppings * that we don't know about have a shot at working. */ static const struct bge_revision const bge_majorrevs[] = { { BGE_ASICREV_BCM5700, "unknown BCM5700" }, { BGE_ASICREV_BCM5701, "unknown BCM5701" }, { BGE_ASICREV_BCM5703, "unknown BCM5703" }, { BGE_ASICREV_BCM5704, "unknown BCM5704" }, { BGE_ASICREV_BCM5705, "unknown BCM5705" }, { BGE_ASICREV_BCM5750, "unknown BCM5750" }, { BGE_ASICREV_BCM5714_A0, "unknown BCM5714" }, { BGE_ASICREV_BCM5752, "unknown BCM5752" }, { BGE_ASICREV_BCM5780, "unknown BCM5780" }, { BGE_ASICREV_BCM5714, "unknown BCM5714" }, { BGE_ASICREV_BCM5755, "unknown BCM5755" }, { BGE_ASICREV_BCM5761, "unknown BCM5761" }, { BGE_ASICREV_BCM5784, "unknown BCM5784" }, { BGE_ASICREV_BCM5785, "unknown BCM5785" }, /* 5754 and 5787 share the same ASIC ID */ { BGE_ASICREV_BCM5787, "unknown BCM5754/5787" }, { BGE_ASICREV_BCM5906, "unknown BCM5906" }, + { BGE_ASICREV_BCM57765, "unknown BCM57765" }, { BGE_ASICREV_BCM57780, "unknown BCM57780" }, { BGE_ASICREV_BCM5717, "unknown BCM5717" }, { 0, NULL } }; #define BGE_IS_JUMBO_CAPABLE(sc) ((sc)->bge_flags & BGE_FLAG_JUMBO) #define BGE_IS_5700_FAMILY(sc) ((sc)->bge_flags & BGE_FLAG_5700_FAMILY) #define BGE_IS_5705_PLUS(sc) ((sc)->bge_flags & BGE_FLAG_5705_PLUS) #define BGE_IS_5714_FAMILY(sc) ((sc)->bge_flags & BGE_FLAG_5714_FAMILY) #define BGE_IS_575X_PLUS(sc) ((sc)->bge_flags & BGE_FLAG_575X_PLUS) #define BGE_IS_5755_PLUS(sc) ((sc)->bge_flags & BGE_FLAG_5755_PLUS) #define BGE_IS_5717_PLUS(sc) ((sc)->bge_flags & BGE_FLAG_5717_PLUS) const struct bge_revision * bge_lookup_rev(uint32_t); const struct bge_vendor * bge_lookup_vendor(uint16_t); typedef int (*bge_eaddr_fcn_t)(struct bge_softc *, uint8_t[]); static int bge_probe(device_t); static int bge_attach(device_t); static int bge_detach(device_t); static int bge_suspend(device_t); static int bge_resume(device_t); static void bge_release_resources(struct bge_softc *); static void bge_dma_map_addr(void *, bus_dma_segment_t *, int, int); static int bge_dma_alloc(struct bge_softc *); static void bge_dma_free(struct bge_softc *); static int bge_dma_ring_alloc(struct bge_softc *, bus_size_t, bus_size_t, bus_dma_tag_t *, uint8_t **, bus_dmamap_t *, bus_addr_t *, const char *); static int bge_get_eaddr_fw(struct bge_softc *sc, uint8_t ether_addr[]); static int bge_get_eaddr_mem(struct bge_softc *, uint8_t[]); static int bge_get_eaddr_nvram(struct bge_softc *, uint8_t[]); static int bge_get_eaddr_eeprom(struct bge_softc *, uint8_t[]); static int bge_get_eaddr(struct bge_softc *, uint8_t[]); static void bge_txeof(struct bge_softc *, uint16_t); static void bge_rxcsum(struct bge_softc *, struct bge_rx_bd *, struct mbuf *); static int bge_rxeof(struct bge_softc *, uint16_t, int); static void bge_asf_driver_up (struct bge_softc *); static void bge_tick(void *); static void bge_stats_clear_regs(struct bge_softc *); static void bge_stats_update(struct bge_softc *); static void bge_stats_update_regs(struct bge_softc *); static struct mbuf *bge_check_short_dma(struct mbuf *); static struct mbuf *bge_setup_tso(struct bge_softc *, struct mbuf *, uint16_t *, uint16_t *); static int bge_encap(struct bge_softc *, struct mbuf **, uint32_t *); static void bge_intr(void *); static int bge_msi_intr(void *); static void bge_intr_task(void *, int); static void bge_start_locked(struct ifnet *); static void bge_start(struct ifnet *); static int bge_ioctl(struct ifnet *, u_long, caddr_t); static void bge_init_locked(struct bge_softc *); static void bge_init(void *); static void bge_stop(struct bge_softc *); static void bge_watchdog(struct bge_softc *); static int bge_shutdown(device_t); static int bge_ifmedia_upd_locked(struct ifnet *); static int bge_ifmedia_upd(struct ifnet *); static void bge_ifmedia_sts(struct ifnet *, struct ifmediareq *); static uint8_t bge_nvram_getbyte(struct bge_softc *, int, uint8_t *); static int bge_read_nvram(struct bge_softc *, caddr_t, int, int); static uint8_t bge_eeprom_getbyte(struct bge_softc *, int, uint8_t *); static int bge_read_eeprom(struct bge_softc *, caddr_t, int, int); static void bge_setpromisc(struct bge_softc *); static void bge_setmulti(struct bge_softc *); static void bge_setvlan(struct bge_softc *); static __inline void bge_rxreuse_std(struct bge_softc *, int); static __inline void bge_rxreuse_jumbo(struct bge_softc *, int); static int bge_newbuf_std(struct bge_softc *, int); static int bge_newbuf_jumbo(struct bge_softc *, int); static int bge_init_rx_ring_std(struct bge_softc *); static void bge_free_rx_ring_std(struct bge_softc *); static int bge_init_rx_ring_jumbo(struct bge_softc *); static void bge_free_rx_ring_jumbo(struct bge_softc *); static void bge_free_tx_ring(struct bge_softc *); static int bge_init_tx_ring(struct bge_softc *); static int bge_chipinit(struct bge_softc *); static int bge_blockinit(struct bge_softc *); static int bge_has_eaddr(struct bge_softc *); static uint32_t bge_readmem_ind(struct bge_softc *, int); static void bge_writemem_ind(struct bge_softc *, int, int); static void bge_writembx(struct bge_softc *, int, int); #ifdef notdef static uint32_t bge_readreg_ind(struct bge_softc *, int); #endif static void bge_writemem_direct(struct bge_softc *, int, int); static void bge_writereg_ind(struct bge_softc *, int, int); static int bge_miibus_readreg(device_t, int, int); static int bge_miibus_writereg(device_t, int, int, int); static void bge_miibus_statchg(device_t); #ifdef DEVICE_POLLING static int bge_poll(struct ifnet *ifp, enum poll_cmd cmd, int count); #endif #define BGE_RESET_START 1 #define BGE_RESET_STOP 2 static void bge_sig_post_reset(struct bge_softc *, int); static void bge_sig_legacy(struct bge_softc *, int); static void bge_sig_pre_reset(struct bge_softc *, int); static void bge_stop_fw(struct bge_softc *); static int bge_reset(struct bge_softc *); static void bge_link_upd(struct bge_softc *); /* * The BGE_REGISTER_DEBUG option is only for low-level debugging. It may * leak information to untrusted users. It is also known to cause alignment * traps on certain architectures. */ #ifdef BGE_REGISTER_DEBUG static int bge_sysctl_debug_info(SYSCTL_HANDLER_ARGS); static int bge_sysctl_reg_read(SYSCTL_HANDLER_ARGS); static int bge_sysctl_mem_read(SYSCTL_HANDLER_ARGS); #endif static void bge_add_sysctls(struct bge_softc *); static void bge_add_sysctl_stats_regs(struct bge_softc *, struct sysctl_ctx_list *, struct sysctl_oid_list *); static void bge_add_sysctl_stats(struct bge_softc *, struct sysctl_ctx_list *, struct sysctl_oid_list *); static int bge_sysctl_stats(SYSCTL_HANDLER_ARGS); static device_method_t bge_methods[] = { /* Device interface */ DEVMETHOD(device_probe, bge_probe), DEVMETHOD(device_attach, bge_attach), DEVMETHOD(device_detach, bge_detach), DEVMETHOD(device_shutdown, bge_shutdown), DEVMETHOD(device_suspend, bge_suspend), DEVMETHOD(device_resume, bge_resume), /* bus interface */ DEVMETHOD(bus_print_child, bus_generic_print_child), DEVMETHOD(bus_driver_added, bus_generic_driver_added), /* MII interface */ DEVMETHOD(miibus_readreg, bge_miibus_readreg), DEVMETHOD(miibus_writereg, bge_miibus_writereg), DEVMETHOD(miibus_statchg, bge_miibus_statchg), { 0, 0 } }; static driver_t bge_driver = { "bge", bge_methods, sizeof(struct bge_softc) }; static devclass_t bge_devclass; DRIVER_MODULE(bge, pci, bge_driver, bge_devclass, 0, 0); DRIVER_MODULE(miibus, bge, miibus_driver, miibus_devclass, 0, 0); static int bge_allow_asf = 1; TUNABLE_INT("hw.bge.allow_asf", &bge_allow_asf); SYSCTL_NODE(_hw, OID_AUTO, bge, CTLFLAG_RD, 0, "BGE driver parameters"); SYSCTL_INT(_hw_bge, OID_AUTO, allow_asf, CTLFLAG_RD, &bge_allow_asf, 0, "Allow ASF mode if available"); #define SPARC64_BLADE_1500_MODEL "SUNW,Sun-Blade-1500" #define SPARC64_BLADE_1500_PATH_BGE "/pci@1f,700000/network@2" #define SPARC64_BLADE_2500_MODEL "SUNW,Sun-Blade-2500" #define SPARC64_BLADE_2500_PATH_BGE "/pci@1c,600000/network@3" #define SPARC64_OFW_SUBVENDOR "subsystem-vendor-id" static int bge_has_eaddr(struct bge_softc *sc) { #ifdef __sparc64__ char buf[sizeof(SPARC64_BLADE_1500_PATH_BGE)]; device_t dev; uint32_t subvendor; dev = sc->bge_dev; /* * The on-board BGEs found in sun4u machines aren't fitted with * an EEPROM which means that we have to obtain the MAC address * via OFW and that some tests will always fail. We distinguish * such BGEs by the subvendor ID, which also has to be obtained * from OFW instead of the PCI configuration space as the latter * indicates Broadcom as the subvendor of the netboot interface. * For early Blade 1500 and 2500 we even have to check the OFW * device path as the subvendor ID always defaults to Broadcom * there. */ if (OF_getprop(ofw_bus_get_node(dev), SPARC64_OFW_SUBVENDOR, &subvendor, sizeof(subvendor)) == sizeof(subvendor) && (subvendor == FJTSU_VENDORID || subvendor == SUN_VENDORID)) return (0); memset(buf, 0, sizeof(buf)); if (OF_package_to_path(ofw_bus_get_node(dev), buf, sizeof(buf)) > 0) { if (strcmp(sparc64_model, SPARC64_BLADE_1500_MODEL) == 0 && strcmp(buf, SPARC64_BLADE_1500_PATH_BGE) == 0) return (0); if (strcmp(sparc64_model, SPARC64_BLADE_2500_MODEL) == 0 && strcmp(buf, SPARC64_BLADE_2500_PATH_BGE) == 0) return (0); } #endif return (1); } static uint32_t bge_readmem_ind(struct bge_softc *sc, int off) { device_t dev; uint32_t val; if (sc->bge_asicrev == BGE_ASICREV_BCM5906 && off >= BGE_STATS_BLOCK && off < BGE_SEND_RING_1_TO_4) return (0); dev = sc->bge_dev; pci_write_config(dev, BGE_PCI_MEMWIN_BASEADDR, off, 4); val = pci_read_config(dev, BGE_PCI_MEMWIN_DATA, 4); pci_write_config(dev, BGE_PCI_MEMWIN_BASEADDR, 0, 4); return (val); } static void bge_writemem_ind(struct bge_softc *sc, int off, int val) { device_t dev; if (sc->bge_asicrev == BGE_ASICREV_BCM5906 && off >= BGE_STATS_BLOCK && off < BGE_SEND_RING_1_TO_4) return; dev = sc->bge_dev; pci_write_config(dev, BGE_PCI_MEMWIN_BASEADDR, off, 4); pci_write_config(dev, BGE_PCI_MEMWIN_DATA, val, 4); pci_write_config(dev, BGE_PCI_MEMWIN_BASEADDR, 0, 4); } #ifdef notdef static uint32_t bge_readreg_ind(struct bge_softc *sc, int off) { device_t dev; dev = sc->bge_dev; pci_write_config(dev, BGE_PCI_REG_BASEADDR, off, 4); return (pci_read_config(dev, BGE_PCI_REG_DATA, 4)); } #endif static void bge_writereg_ind(struct bge_softc *sc, int off, int val) { device_t dev; dev = sc->bge_dev; pci_write_config(dev, BGE_PCI_REG_BASEADDR, off, 4); pci_write_config(dev, BGE_PCI_REG_DATA, val, 4); } static void bge_writemem_direct(struct bge_softc *sc, int off, int val) { CSR_WRITE_4(sc, off, val); } static void bge_writembx(struct bge_softc *sc, int off, int val) { if (sc->bge_asicrev == BGE_ASICREV_BCM5906) off += BGE_LPMBX_IRQ0_HI - BGE_MBX_IRQ0_HI; CSR_WRITE_4(sc, off, val); } /* * Map a single buffer address. */ static void bge_dma_map_addr(void *arg, bus_dma_segment_t *segs, int nseg, int error) { struct bge_dmamap_arg *ctx; if (error) return; KASSERT(nseg == 1, ("%s: %d segments returned!", __func__, nseg)); ctx = arg; ctx->bge_busaddr = segs->ds_addr; } static uint8_t bge_nvram_getbyte(struct bge_softc *sc, int addr, uint8_t *dest) { uint32_t access, byte = 0; int i; /* Lock. */ CSR_WRITE_4(sc, BGE_NVRAM_SWARB, BGE_NVRAMSWARB_SET1); for (i = 0; i < 8000; i++) { if (CSR_READ_4(sc, BGE_NVRAM_SWARB) & BGE_NVRAMSWARB_GNT1) break; DELAY(20); } if (i == 8000) return (1); /* Enable access. */ access = CSR_READ_4(sc, BGE_NVRAM_ACCESS); CSR_WRITE_4(sc, BGE_NVRAM_ACCESS, access | BGE_NVRAMACC_ENABLE); CSR_WRITE_4(sc, BGE_NVRAM_ADDR, addr & 0xfffffffc); CSR_WRITE_4(sc, BGE_NVRAM_CMD, BGE_NVRAM_READCMD); for (i = 0; i < BGE_TIMEOUT * 10; i++) { DELAY(10); if (CSR_READ_4(sc, BGE_NVRAM_CMD) & BGE_NVRAMCMD_DONE) { DELAY(10); break; } } if (i == BGE_TIMEOUT * 10) { if_printf(sc->bge_ifp, "nvram read timed out\n"); return (1); } /* Get result. */ byte = CSR_READ_4(sc, BGE_NVRAM_RDDATA); *dest = (bswap32(byte) >> ((addr % 4) * 8)) & 0xFF; /* Disable access. */ CSR_WRITE_4(sc, BGE_NVRAM_ACCESS, access); /* Unlock. */ CSR_WRITE_4(sc, BGE_NVRAM_SWARB, BGE_NVRAMSWARB_CLR1); CSR_READ_4(sc, BGE_NVRAM_SWARB); return (0); } /* * Read a sequence of bytes from NVRAM. */ static int bge_read_nvram(struct bge_softc *sc, caddr_t dest, int off, int cnt) { int err = 0, i; uint8_t byte = 0; if (sc->bge_asicrev != BGE_ASICREV_BCM5906) return (1); for (i = 0; i < cnt; i++) { err = bge_nvram_getbyte(sc, off + i, &byte); if (err) break; *(dest + i) = byte; } return (err ? 1 : 0); } /* * Read a byte of data stored in the EEPROM at address 'addr.' The * BCM570x supports both the traditional bitbang interface and an * auto access interface for reading the EEPROM. We use the auto * access method. */ static uint8_t bge_eeprom_getbyte(struct bge_softc *sc, int addr, uint8_t *dest) { int i; uint32_t byte = 0; /* * Enable use of auto EEPROM access so we can avoid * having to use the bitbang method. */ BGE_SETBIT(sc, BGE_MISC_LOCAL_CTL, BGE_MLC_AUTO_EEPROM); /* Reset the EEPROM, load the clock period. */ CSR_WRITE_4(sc, BGE_EE_ADDR, BGE_EEADDR_RESET | BGE_EEHALFCLK(BGE_HALFCLK_384SCL)); DELAY(20); /* Issue the read EEPROM command. */ CSR_WRITE_4(sc, BGE_EE_ADDR, BGE_EE_READCMD | addr); /* Wait for completion */ for(i = 0; i < BGE_TIMEOUT * 10; i++) { DELAY(10); if (CSR_READ_4(sc, BGE_EE_ADDR) & BGE_EEADDR_DONE) break; } if (i == BGE_TIMEOUT * 10) { device_printf(sc->bge_dev, "EEPROM read timed out\n"); return (1); } /* Get result. */ byte = CSR_READ_4(sc, BGE_EE_DATA); *dest = (byte >> ((addr % 4) * 8)) & 0xFF; return (0); } /* * Read a sequence of bytes from the EEPROM. */ static int bge_read_eeprom(struct bge_softc *sc, caddr_t dest, int off, int cnt) { int i, error = 0; uint8_t byte = 0; for (i = 0; i < cnt; i++) { error = bge_eeprom_getbyte(sc, off + i, &byte); if (error) break; *(dest + i) = byte; } return (error ? 1 : 0); } static int bge_miibus_readreg(device_t dev, int phy, int reg) { struct bge_softc *sc; uint32_t val; int i; sc = device_get_softc(dev); /* Clear the autopoll bit if set, otherwise may trigger PCI errors. */ if ((sc->bge_mi_mode & BGE_MIMODE_AUTOPOLL) != 0) { CSR_WRITE_4(sc, BGE_MI_MODE, sc->bge_mi_mode & ~BGE_MIMODE_AUTOPOLL); DELAY(80); } CSR_WRITE_4(sc, BGE_MI_COMM, BGE_MICMD_READ | BGE_MICOMM_BUSY | BGE_MIPHY(phy) | BGE_MIREG(reg)); /* Poll for the PHY register access to complete. */ for (i = 0; i < BGE_TIMEOUT; i++) { DELAY(10); val = CSR_READ_4(sc, BGE_MI_COMM); if ((val & BGE_MICOMM_BUSY) == 0) { DELAY(5); val = CSR_READ_4(sc, BGE_MI_COMM); break; } } if (i == BGE_TIMEOUT) { device_printf(sc->bge_dev, "PHY read timed out (phy %d, reg %d, val 0x%08x)\n", phy, reg, val); val = 0; } /* Restore the autopoll bit if necessary. */ if ((sc->bge_mi_mode & BGE_MIMODE_AUTOPOLL) != 0) { CSR_WRITE_4(sc, BGE_MI_MODE, sc->bge_mi_mode); DELAY(80); } if (val & BGE_MICOMM_READFAIL) return (0); return (val & 0xFFFF); } static int bge_miibus_writereg(device_t dev, int phy, int reg, int val) { struct bge_softc *sc; int i; sc = device_get_softc(dev); if (sc->bge_asicrev == BGE_ASICREV_BCM5906 && (reg == BRGPHY_MII_1000CTL || reg == BRGPHY_MII_AUXCTL)) return (0); /* Clear the autopoll bit if set, otherwise may trigger PCI errors. */ if ((sc->bge_mi_mode & BGE_MIMODE_AUTOPOLL) != 0) { CSR_WRITE_4(sc, BGE_MI_MODE, sc->bge_mi_mode & ~BGE_MIMODE_AUTOPOLL); DELAY(80); } CSR_WRITE_4(sc, BGE_MI_COMM, BGE_MICMD_WRITE | BGE_MICOMM_BUSY | BGE_MIPHY(phy) | BGE_MIREG(reg) | val); for (i = 0; i < BGE_TIMEOUT; i++) { DELAY(10); if (!(CSR_READ_4(sc, BGE_MI_COMM) & BGE_MICOMM_BUSY)) { DELAY(5); CSR_READ_4(sc, BGE_MI_COMM); /* dummy read */ break; } } /* Restore the autopoll bit if necessary. */ if ((sc->bge_mi_mode & BGE_MIMODE_AUTOPOLL) != 0) { CSR_WRITE_4(sc, BGE_MI_MODE, sc->bge_mi_mode); DELAY(80); } if (i == BGE_TIMEOUT) device_printf(sc->bge_dev, "PHY write timed out (phy %d, reg %d, val %d)\n", phy, reg, val); return (0); } static void bge_miibus_statchg(device_t dev) { struct bge_softc *sc; struct mii_data *mii; sc = device_get_softc(dev); mii = device_get_softc(sc->bge_miibus); if ((mii->mii_media_status & (IFM_ACTIVE | IFM_AVALID)) == (IFM_ACTIVE | IFM_AVALID)) { switch (IFM_SUBTYPE(mii->mii_media_active)) { case IFM_10_T: case IFM_100_TX: sc->bge_link = 1; break; case IFM_1000_T: case IFM_1000_SX: case IFM_2500_SX: if (sc->bge_asicrev != BGE_ASICREV_BCM5906) sc->bge_link = 1; else sc->bge_link = 0; break; default: sc->bge_link = 0; break; } } else sc->bge_link = 0; if (sc->bge_link == 0) return; BGE_CLRBIT(sc, BGE_MAC_MODE, BGE_MACMODE_PORTMODE); if (IFM_SUBTYPE(mii->mii_media_active) == IFM_1000_T || IFM_SUBTYPE(mii->mii_media_active) == IFM_1000_SX) BGE_SETBIT(sc, BGE_MAC_MODE, BGE_PORTMODE_GMII); else BGE_SETBIT(sc, BGE_MAC_MODE, BGE_PORTMODE_MII); if (IFM_OPTIONS(mii->mii_media_active & IFM_FDX) != 0) { BGE_CLRBIT(sc, BGE_MAC_MODE, BGE_MACMODE_HALF_DUPLEX); if ((IFM_OPTIONS(mii->mii_media_active) & IFM_ETH_TXPAUSE) != 0) BGE_SETBIT(sc, BGE_TX_MODE, BGE_TXMODE_FLOWCTL_ENABLE); else BGE_CLRBIT(sc, BGE_TX_MODE, BGE_TXMODE_FLOWCTL_ENABLE); if ((IFM_OPTIONS(mii->mii_media_active) & IFM_ETH_RXPAUSE) != 0) BGE_SETBIT(sc, BGE_RX_MODE, BGE_RXMODE_FLOWCTL_ENABLE); else BGE_CLRBIT(sc, BGE_RX_MODE, BGE_RXMODE_FLOWCTL_ENABLE); } else { BGE_SETBIT(sc, BGE_MAC_MODE, BGE_MACMODE_HALF_DUPLEX); BGE_CLRBIT(sc, BGE_TX_MODE, BGE_TXMODE_FLOWCTL_ENABLE); BGE_CLRBIT(sc, BGE_RX_MODE, BGE_RXMODE_FLOWCTL_ENABLE); } } /* * Intialize a standard receive ring descriptor. */ static int bge_newbuf_std(struct bge_softc *sc, int i) { struct mbuf *m; struct bge_rx_bd *r; bus_dma_segment_t segs[1]; bus_dmamap_t map; int error, nsegs; if (sc->bge_flags & BGE_FLAG_JUMBO_STD && (sc->bge_ifp->if_mtu + ETHER_HDR_LEN + ETHER_CRC_LEN + ETHER_VLAN_ENCAP_LEN > (MCLBYTES - ETHER_ALIGN))) { m = m_getjcl(M_DONTWAIT, MT_DATA, M_PKTHDR, MJUM9BYTES); if (m == NULL) return (ENOBUFS); m->m_len = m->m_pkthdr.len = MJUM9BYTES; } else { m = m_getcl(M_DONTWAIT, MT_DATA, M_PKTHDR); if (m == NULL) return (ENOBUFS); m->m_len = m->m_pkthdr.len = MCLBYTES; } if ((sc->bge_flags & BGE_FLAG_RX_ALIGNBUG) == 0) m_adj(m, ETHER_ALIGN); error = bus_dmamap_load_mbuf_sg(sc->bge_cdata.bge_rx_mtag, sc->bge_cdata.bge_rx_std_sparemap, m, segs, &nsegs, 0); if (error != 0) { m_freem(m); return (error); } if (sc->bge_cdata.bge_rx_std_chain[i] != NULL) { bus_dmamap_sync(sc->bge_cdata.bge_rx_mtag, sc->bge_cdata.bge_rx_std_dmamap[i], BUS_DMASYNC_POSTREAD); bus_dmamap_unload(sc->bge_cdata.bge_rx_mtag, sc->bge_cdata.bge_rx_std_dmamap[i]); } map = sc->bge_cdata.bge_rx_std_dmamap[i]; sc->bge_cdata.bge_rx_std_dmamap[i] = sc->bge_cdata.bge_rx_std_sparemap; sc->bge_cdata.bge_rx_std_sparemap = map; sc->bge_cdata.bge_rx_std_chain[i] = m; sc->bge_cdata.bge_rx_std_seglen[i] = segs[0].ds_len; r = &sc->bge_ldata.bge_rx_std_ring[sc->bge_std]; r->bge_addr.bge_addr_lo = BGE_ADDR_LO(segs[0].ds_addr); r->bge_addr.bge_addr_hi = BGE_ADDR_HI(segs[0].ds_addr); r->bge_flags = BGE_RXBDFLAG_END; r->bge_len = segs[0].ds_len; r->bge_idx = i; bus_dmamap_sync(sc->bge_cdata.bge_rx_mtag, sc->bge_cdata.bge_rx_std_dmamap[i], BUS_DMASYNC_PREREAD); return (0); } /* * Initialize a jumbo receive ring descriptor. This allocates * a jumbo buffer from the pool managed internally by the driver. */ static int bge_newbuf_jumbo(struct bge_softc *sc, int i) { bus_dma_segment_t segs[BGE_NSEG_JUMBO]; bus_dmamap_t map; struct bge_extrx_bd *r; struct mbuf *m; int error, nsegs; MGETHDR(m, M_DONTWAIT, MT_DATA); if (m == NULL) return (ENOBUFS); m_cljget(m, M_DONTWAIT, MJUM9BYTES); if (!(m->m_flags & M_EXT)) { m_freem(m); return (ENOBUFS); } m->m_len = m->m_pkthdr.len = MJUM9BYTES; if ((sc->bge_flags & BGE_FLAG_RX_ALIGNBUG) == 0) m_adj(m, ETHER_ALIGN); error = bus_dmamap_load_mbuf_sg(sc->bge_cdata.bge_mtag_jumbo, sc->bge_cdata.bge_rx_jumbo_sparemap, m, segs, &nsegs, 0); if (error != 0) { m_freem(m); return (error); } if (sc->bge_cdata.bge_rx_jumbo_chain[i] != NULL) { bus_dmamap_sync(sc->bge_cdata.bge_mtag_jumbo, sc->bge_cdata.bge_rx_jumbo_dmamap[i], BUS_DMASYNC_POSTREAD); bus_dmamap_unload(sc->bge_cdata.bge_mtag_jumbo, sc->bge_cdata.bge_rx_jumbo_dmamap[i]); } map = sc->bge_cdata.bge_rx_jumbo_dmamap[i]; sc->bge_cdata.bge_rx_jumbo_dmamap[i] = sc->bge_cdata.bge_rx_jumbo_sparemap; sc->bge_cdata.bge_rx_jumbo_sparemap = map; sc->bge_cdata.bge_rx_jumbo_chain[i] = m; sc->bge_cdata.bge_rx_jumbo_seglen[i][0] = 0; sc->bge_cdata.bge_rx_jumbo_seglen[i][1] = 0; sc->bge_cdata.bge_rx_jumbo_seglen[i][2] = 0; sc->bge_cdata.bge_rx_jumbo_seglen[i][3] = 0; /* * Fill in the extended RX buffer descriptor. */ r = &sc->bge_ldata.bge_rx_jumbo_ring[sc->bge_jumbo]; r->bge_flags = BGE_RXBDFLAG_JUMBO_RING | BGE_RXBDFLAG_END; r->bge_idx = i; r->bge_len3 = r->bge_len2 = r->bge_len1 = 0; switch (nsegs) { case 4: r->bge_addr3.bge_addr_lo = BGE_ADDR_LO(segs[3].ds_addr); r->bge_addr3.bge_addr_hi = BGE_ADDR_HI(segs[3].ds_addr); r->bge_len3 = segs[3].ds_len; sc->bge_cdata.bge_rx_jumbo_seglen[i][3] = segs[3].ds_len; case 3: r->bge_addr2.bge_addr_lo = BGE_ADDR_LO(segs[2].ds_addr); r->bge_addr2.bge_addr_hi = BGE_ADDR_HI(segs[2].ds_addr); r->bge_len2 = segs[2].ds_len; sc->bge_cdata.bge_rx_jumbo_seglen[i][2] = segs[2].ds_len; case 2: r->bge_addr1.bge_addr_lo = BGE_ADDR_LO(segs[1].ds_addr); r->bge_addr1.bge_addr_hi = BGE_ADDR_HI(segs[1].ds_addr); r->bge_len1 = segs[1].ds_len; sc->bge_cdata.bge_rx_jumbo_seglen[i][1] = segs[1].ds_len; case 1: r->bge_addr0.bge_addr_lo = BGE_ADDR_LO(segs[0].ds_addr); r->bge_addr0.bge_addr_hi = BGE_ADDR_HI(segs[0].ds_addr); r->bge_len0 = segs[0].ds_len; sc->bge_cdata.bge_rx_jumbo_seglen[i][0] = segs[0].ds_len; break; default: panic("%s: %d segments\n", __func__, nsegs); } bus_dmamap_sync(sc->bge_cdata.bge_mtag_jumbo, sc->bge_cdata.bge_rx_jumbo_dmamap[i], BUS_DMASYNC_PREREAD); return (0); } static int bge_init_rx_ring_std(struct bge_softc *sc) { int error, i; bzero(sc->bge_ldata.bge_rx_std_ring, BGE_STD_RX_RING_SZ); sc->bge_std = 0; for (i = 0; i < BGE_STD_RX_RING_CNT; i++) { if ((error = bge_newbuf_std(sc, i)) != 0) return (error); BGE_INC(sc->bge_std, BGE_STD_RX_RING_CNT); } bus_dmamap_sync(sc->bge_cdata.bge_rx_std_ring_tag, sc->bge_cdata.bge_rx_std_ring_map, BUS_DMASYNC_PREWRITE); sc->bge_std = 0; bge_writembx(sc, BGE_MBX_RX_STD_PROD_LO, BGE_STD_RX_RING_CNT - 1); return (0); } static void bge_free_rx_ring_std(struct bge_softc *sc) { int i; for (i = 0; i < BGE_STD_RX_RING_CNT; i++) { if (sc->bge_cdata.bge_rx_std_chain[i] != NULL) { bus_dmamap_sync(sc->bge_cdata.bge_rx_mtag, sc->bge_cdata.bge_rx_std_dmamap[i], BUS_DMASYNC_POSTREAD); bus_dmamap_unload(sc->bge_cdata.bge_rx_mtag, sc->bge_cdata.bge_rx_std_dmamap[i]); m_freem(sc->bge_cdata.bge_rx_std_chain[i]); sc->bge_cdata.bge_rx_std_chain[i] = NULL; } bzero((char *)&sc->bge_ldata.bge_rx_std_ring[i], sizeof(struct bge_rx_bd)); } } static int bge_init_rx_ring_jumbo(struct bge_softc *sc) { struct bge_rcb *rcb; int error, i; bzero(sc->bge_ldata.bge_rx_jumbo_ring, BGE_JUMBO_RX_RING_SZ); sc->bge_jumbo = 0; for (i = 0; i < BGE_JUMBO_RX_RING_CNT; i++) { if ((error = bge_newbuf_jumbo(sc, i)) != 0) return (error); BGE_INC(sc->bge_jumbo, BGE_JUMBO_RX_RING_CNT); } bus_dmamap_sync(sc->bge_cdata.bge_rx_jumbo_ring_tag, sc->bge_cdata.bge_rx_jumbo_ring_map, BUS_DMASYNC_PREWRITE); sc->bge_jumbo = 0; /* Enable the jumbo receive producer ring. */ rcb = &sc->bge_ldata.bge_info.bge_jumbo_rx_rcb; rcb->bge_maxlen_flags = BGE_RCB_MAXLEN_FLAGS(0, BGE_RCB_FLAG_USE_EXT_RX_BD); CSR_WRITE_4(sc, BGE_RX_JUMBO_RCB_MAXLEN_FLAGS, rcb->bge_maxlen_flags); bge_writembx(sc, BGE_MBX_RX_JUMBO_PROD_LO, BGE_JUMBO_RX_RING_CNT - 1); return (0); } static void bge_free_rx_ring_jumbo(struct bge_softc *sc) { int i; for (i = 0; i < BGE_JUMBO_RX_RING_CNT; i++) { if (sc->bge_cdata.bge_rx_jumbo_chain[i] != NULL) { bus_dmamap_sync(sc->bge_cdata.bge_mtag_jumbo, sc->bge_cdata.bge_rx_jumbo_dmamap[i], BUS_DMASYNC_POSTREAD); bus_dmamap_unload(sc->bge_cdata.bge_mtag_jumbo, sc->bge_cdata.bge_rx_jumbo_dmamap[i]); m_freem(sc->bge_cdata.bge_rx_jumbo_chain[i]); sc->bge_cdata.bge_rx_jumbo_chain[i] = NULL; } bzero((char *)&sc->bge_ldata.bge_rx_jumbo_ring[i], sizeof(struct bge_extrx_bd)); } } static void bge_free_tx_ring(struct bge_softc *sc) { int i; if (sc->bge_ldata.bge_tx_ring == NULL) return; for (i = 0; i < BGE_TX_RING_CNT; i++) { if (sc->bge_cdata.bge_tx_chain[i] != NULL) { bus_dmamap_sync(sc->bge_cdata.bge_tx_mtag, sc->bge_cdata.bge_tx_dmamap[i], BUS_DMASYNC_POSTWRITE); bus_dmamap_unload(sc->bge_cdata.bge_tx_mtag, sc->bge_cdata.bge_tx_dmamap[i]); m_freem(sc->bge_cdata.bge_tx_chain[i]); sc->bge_cdata.bge_tx_chain[i] = NULL; } bzero((char *)&sc->bge_ldata.bge_tx_ring[i], sizeof(struct bge_tx_bd)); } } static int bge_init_tx_ring(struct bge_softc *sc) { sc->bge_txcnt = 0; sc->bge_tx_saved_considx = 0; bzero(sc->bge_ldata.bge_tx_ring, BGE_TX_RING_SZ); bus_dmamap_sync(sc->bge_cdata.bge_tx_ring_tag, sc->bge_cdata.bge_tx_ring_map, BUS_DMASYNC_PREWRITE); /* Initialize transmit producer index for host-memory send ring. */ sc->bge_tx_prodidx = 0; bge_writembx(sc, BGE_MBX_TX_HOST_PROD0_LO, sc->bge_tx_prodidx); /* 5700 b2 errata */ if (sc->bge_chiprev == BGE_CHIPREV_5700_BX) bge_writembx(sc, BGE_MBX_TX_HOST_PROD0_LO, sc->bge_tx_prodidx); /* NIC-memory send ring not used; initialize to zero. */ bge_writembx(sc, BGE_MBX_TX_NIC_PROD0_LO, 0); /* 5700 b2 errata */ if (sc->bge_chiprev == BGE_CHIPREV_5700_BX) bge_writembx(sc, BGE_MBX_TX_NIC_PROD0_LO, 0); return (0); } static void bge_setpromisc(struct bge_softc *sc) { struct ifnet *ifp; BGE_LOCK_ASSERT(sc); ifp = sc->bge_ifp; /* Enable or disable promiscuous mode as needed. */ if (ifp->if_flags & IFF_PROMISC) BGE_SETBIT(sc, BGE_RX_MODE, BGE_RXMODE_RX_PROMISC); else BGE_CLRBIT(sc, BGE_RX_MODE, BGE_RXMODE_RX_PROMISC); } static void bge_setmulti(struct bge_softc *sc) { struct ifnet *ifp; struct ifmultiaddr *ifma; uint32_t hashes[4] = { 0, 0, 0, 0 }; int h, i; BGE_LOCK_ASSERT(sc); ifp = sc->bge_ifp; if (ifp->if_flags & IFF_ALLMULTI || ifp->if_flags & IFF_PROMISC) { for (i = 0; i < 4; i++) CSR_WRITE_4(sc, BGE_MAR0 + (i * 4), 0xFFFFFFFF); return; } /* First, zot all the existing filters. */ for (i = 0; i < 4; i++) CSR_WRITE_4(sc, BGE_MAR0 + (i * 4), 0); /* Now program new ones. */ if_maddr_rlock(ifp); TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) { if (ifma->ifma_addr->sa_family != AF_LINK) continue; h = ether_crc32_le(LLADDR((struct sockaddr_dl *) ifma->ifma_addr), ETHER_ADDR_LEN) & 0x7F; hashes[(h & 0x60) >> 5] |= 1 << (h & 0x1F); } if_maddr_runlock(ifp); for (i = 0; i < 4; i++) CSR_WRITE_4(sc, BGE_MAR0 + (i * 4), hashes[i]); } static void bge_setvlan(struct bge_softc *sc) { struct ifnet *ifp; BGE_LOCK_ASSERT(sc); ifp = sc->bge_ifp; /* Enable or disable VLAN tag stripping as needed. */ if (ifp->if_capenable & IFCAP_VLAN_HWTAGGING) BGE_CLRBIT(sc, BGE_RX_MODE, BGE_RXMODE_RX_KEEP_VLAN_DIAG); else BGE_SETBIT(sc, BGE_RX_MODE, BGE_RXMODE_RX_KEEP_VLAN_DIAG); } static void bge_sig_pre_reset(struct bge_softc *sc, int type) { /* * Some chips don't like this so only do this if ASF is enabled */ if (sc->bge_asf_mode) bge_writemem_ind(sc, BGE_SOFTWARE_GENCOMM, BGE_MAGIC_NUMBER); if (sc->bge_asf_mode & ASF_NEW_HANDSHAKE) { switch (type) { case BGE_RESET_START: bge_writemem_ind(sc, BGE_SDI_STATUS, 0x1); /* START */ break; case BGE_RESET_STOP: bge_writemem_ind(sc, BGE_SDI_STATUS, 0x2); /* UNLOAD */ break; } } } static void bge_sig_post_reset(struct bge_softc *sc, int type) { if (sc->bge_asf_mode & ASF_NEW_HANDSHAKE) { switch (type) { case BGE_RESET_START: bge_writemem_ind(sc, BGE_SDI_STATUS, 0x80000001); /* START DONE */ break; case BGE_RESET_STOP: bge_writemem_ind(sc, BGE_SDI_STATUS, 0x80000002); break; } } } static void bge_sig_legacy(struct bge_softc *sc, int type) { if (sc->bge_asf_mode) { switch (type) { case BGE_RESET_START: bge_writemem_ind(sc, BGE_SDI_STATUS, 0x1); /* START */ break; case BGE_RESET_STOP: bge_writemem_ind(sc, BGE_SDI_STATUS, 0x2); /* UNLOAD */ break; } } } static void bge_stop_fw(struct bge_softc *sc) { int i; if (sc->bge_asf_mode) { bge_writemem_ind(sc, BGE_SOFTWARE_GENCOMM_FW, BGE_FW_PAUSE); CSR_WRITE_4(sc, BGE_CPU_EVENT, CSR_READ_4(sc, BGE_CPU_EVENT) | (1 << 14)); for (i = 0; i < 100; i++ ) { if (!(CSR_READ_4(sc, BGE_CPU_EVENT) & (1 << 14))) break; DELAY(10); } } } /* * Do endian, PCI and DMA initialization. */ static int bge_chipinit(struct bge_softc *sc) { uint32_t dma_rw_ctl, misc_ctl; uint16_t val; int i; /* Set endianness before we access any non-PCI registers. */ misc_ctl = BGE_INIT; if (sc->bge_flags & BGE_FLAG_TAGGED_STATUS) misc_ctl |= BGE_PCIMISCCTL_TAGGED_STATUS; pci_write_config(sc->bge_dev, BGE_PCI_MISC_CTL, misc_ctl, 4); /* Clear the MAC control register */ CSR_WRITE_4(sc, BGE_MAC_MODE, 0); /* * Clear the MAC statistics block in the NIC's * internal memory. */ for (i = BGE_STATS_BLOCK; i < BGE_STATS_BLOCK_END + 1; i += sizeof(uint32_t)) BGE_MEMWIN_WRITE(sc, i, 0); for (i = BGE_STATUS_BLOCK; i < BGE_STATUS_BLOCK_END + 1; i += sizeof(uint32_t)) BGE_MEMWIN_WRITE(sc, i, 0); if (sc->bge_chiprev == BGE_CHIPREV_5704_BX) { /* * Fix data corruption caused by non-qword write with WB. * Fix master abort in PCI mode. * Fix PCI latency timer. */ val = pci_read_config(sc->bge_dev, BGE_PCI_MSI_DATA + 2, 2); val |= (1 << 10) | (1 << 12) | (1 << 13); pci_write_config(sc->bge_dev, BGE_PCI_MSI_DATA + 2, val, 2); } /* * Set up the PCI DMA control register. */ dma_rw_ctl = BGE_PCIDMARWCTL_RD_CMD_SHIFT(6) | BGE_PCIDMARWCTL_WR_CMD_SHIFT(7); if (sc->bge_flags & BGE_FLAG_PCIE) { /* Read watermark not used, 128 bytes for write. */ dma_rw_ctl |= BGE_PCIDMARWCTL_WR_WAT_SHIFT(3); } else if (sc->bge_flags & BGE_FLAG_PCIX) { if (BGE_IS_5714_FAMILY(sc)) { /* 256 bytes for read and write. */ dma_rw_ctl |= BGE_PCIDMARWCTL_RD_WAT_SHIFT(2) | BGE_PCIDMARWCTL_WR_WAT_SHIFT(2); dma_rw_ctl |= (sc->bge_asicrev == BGE_ASICREV_BCM5780) ? BGE_PCIDMARWCTL_ONEDMA_ATONCE_GLOBAL : BGE_PCIDMARWCTL_ONEDMA_ATONCE_LOCAL; } else if (sc->bge_asicrev == BGE_ASICREV_BCM5703) { /* * In the BCM5703, the DMA read watermark should * be set to less than or equal to the maximum * memory read byte count of the PCI-X command * register. */ dma_rw_ctl |= BGE_PCIDMARWCTL_RD_WAT_SHIFT(4) | BGE_PCIDMARWCTL_WR_WAT_SHIFT(3); } else if (sc->bge_asicrev == BGE_ASICREV_BCM5704) { /* 1536 bytes for read, 384 bytes for write. */ dma_rw_ctl |= BGE_PCIDMARWCTL_RD_WAT_SHIFT(7) | BGE_PCIDMARWCTL_WR_WAT_SHIFT(3); } else { /* 384 bytes for read and write. */ dma_rw_ctl |= BGE_PCIDMARWCTL_RD_WAT_SHIFT(3) | BGE_PCIDMARWCTL_WR_WAT_SHIFT(3) | 0x0F; } if (sc->bge_asicrev == BGE_ASICREV_BCM5703 || sc->bge_asicrev == BGE_ASICREV_BCM5704) { uint32_t tmp; /* Set ONE_DMA_AT_ONCE for hardware workaround. */ tmp = CSR_READ_4(sc, BGE_PCI_CLKCTL) & 0x1F; if (tmp == 6 || tmp == 7) dma_rw_ctl |= BGE_PCIDMARWCTL_ONEDMA_ATONCE_GLOBAL; /* Set PCI-X DMA write workaround. */ dma_rw_ctl |= BGE_PCIDMARWCTL_ASRT_ALL_BE; } } else { /* Conventional PCI bus: 256 bytes for read and write. */ dma_rw_ctl |= BGE_PCIDMARWCTL_RD_WAT_SHIFT(7) | BGE_PCIDMARWCTL_WR_WAT_SHIFT(7); if (sc->bge_asicrev != BGE_ASICREV_BCM5705 && sc->bge_asicrev != BGE_ASICREV_BCM5750) dma_rw_ctl |= 0x0F; } if (sc->bge_asicrev == BGE_ASICREV_BCM5700 || sc->bge_asicrev == BGE_ASICREV_BCM5701) dma_rw_ctl |= BGE_PCIDMARWCTL_USE_MRM | BGE_PCIDMARWCTL_ASRT_ALL_BE; if (sc->bge_asicrev == BGE_ASICREV_BCM5703 || sc->bge_asicrev == BGE_ASICREV_BCM5704) dma_rw_ctl &= ~BGE_PCIDMARWCTL_MINDMA; - if (BGE_IS_5717_PLUS(sc)) + if (BGE_IS_5717_PLUS(sc)) { dma_rw_ctl &= ~BGE_PCIDMARWCTL_DIS_CACHE_ALIGNMENT; + if (sc->bge_chipid == BGE_CHIPID_BCM57765_A0) + dma_rw_ctl &= ~BGE_PCIDMARWCTL_CRDRDR_RDMA_MRRS_MSK; + } pci_write_config(sc->bge_dev, BGE_PCI_DMA_RW_CTL, dma_rw_ctl, 4); /* * Set up general mode register. */ CSR_WRITE_4(sc, BGE_MODE_CTL, BGE_DMA_SWAP_OPTIONS | BGE_MODECTL_MAC_ATTN_INTR | BGE_MODECTL_HOST_SEND_BDS | BGE_MODECTL_TX_NO_PHDR_CSUM); /* * BCM5701 B5 have a bug causing data corruption when using * 64-bit DMA reads, which can be terminated early and then * completed later as 32-bit accesses, in combination with * certain bridges. */ if (sc->bge_asicrev == BGE_ASICREV_BCM5701 && sc->bge_chipid == BGE_CHIPID_BCM5701_B5) BGE_SETBIT(sc, BGE_MODE_CTL, BGE_MODECTL_FORCE_PCI32); /* * Tell the firmware the driver is running */ if (sc->bge_asf_mode & ASF_STACKUP) BGE_SETBIT(sc, BGE_MODE_CTL, BGE_MODECTL_STACKUP); /* * Disable memory write invalidate. Apparently it is not supported * properly by these devices. Also ensure that INTx isn't disabled, * as these chips need it even when using MSI. */ PCI_CLRBIT(sc->bge_dev, BGE_PCI_CMD, PCIM_CMD_INTxDIS | PCIM_CMD_MWIEN, 4); /* Set the timer prescaler (always 66Mhz) */ CSR_WRITE_4(sc, BGE_MISC_CFG, BGE_32BITTIME_66MHZ); /* XXX: The Linux tg3 driver does this at the start of brgphy_reset. */ if (sc->bge_asicrev == BGE_ASICREV_BCM5906) { DELAY(40); /* XXX */ /* Put PHY into ready state */ BGE_CLRBIT(sc, BGE_MISC_CFG, BGE_MISCCFG_EPHY_IDDQ); CSR_READ_4(sc, BGE_MISC_CFG); /* Flush */ DELAY(40); } return (0); } static int bge_blockinit(struct bge_softc *sc) { struct bge_rcb *rcb; bus_size_t vrcb; bge_hostaddr taddr; uint32_t val; int i, limit; /* * Initialize the memory window pointer register so that * we can access the first 32K of internal NIC RAM. This will * allow us to set up the TX send ring RCBs and the RX return * ring RCBs, plus other things which live in NIC memory. */ CSR_WRITE_4(sc, BGE_PCI_MEMWIN_BASEADDR, 0); /* Note: the BCM5704 has a smaller mbuf space than other chips. */ if (!(BGE_IS_5705_PLUS(sc))) { /* Configure mbuf memory pool */ CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_BASEADDR, BGE_BUFFPOOL_1); if (sc->bge_asicrev == BGE_ASICREV_BCM5704) CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_LEN, 0x10000); else CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_LEN, 0x18000); /* Configure DMA resource pool */ CSR_WRITE_4(sc, BGE_BMAN_DMA_DESCPOOL_BASEADDR, BGE_DMA_DESCRIPTORS); CSR_WRITE_4(sc, BGE_BMAN_DMA_DESCPOOL_LEN, 0x2000); } /* Configure mbuf pool watermarks */ - if (sc->bge_asicrev == BGE_ASICREV_BCM5717) { + if (sc->bge_asicrev == BGE_ASICREV_BCM5717 || + sc->bge_asicrev == BGE_ASICREV_BCM57765) { CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_READDMA_LOWAT, 0x0); if (sc->bge_ifp->if_mtu > ETHERMTU) { CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_MACRX_LOWAT, 0x7e); CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_HIWAT, 0xea); } else { CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_MACRX_LOWAT, 0x2a); CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_HIWAT, 0xa0); } } else if (!BGE_IS_5705_PLUS(sc)) { CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_READDMA_LOWAT, 0x50); CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_MACRX_LOWAT, 0x20); CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_HIWAT, 0x60); } else if (sc->bge_asicrev == BGE_ASICREV_BCM5906) { CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_READDMA_LOWAT, 0x0); CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_MACRX_LOWAT, 0x04); CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_HIWAT, 0x10); } else { CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_READDMA_LOWAT, 0x0); CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_MACRX_LOWAT, 0x10); CSR_WRITE_4(sc, BGE_BMAN_MBUFPOOL_HIWAT, 0x60); } /* Configure DMA resource watermarks */ CSR_WRITE_4(sc, BGE_BMAN_DMA_DESCPOOL_LOWAT, 5); CSR_WRITE_4(sc, BGE_BMAN_DMA_DESCPOOL_HIWAT, 10); /* Enable buffer manager */ if (!(BGE_IS_5705_PLUS(sc))) { CSR_WRITE_4(sc, BGE_BMAN_MODE, BGE_BMANMODE_ENABLE | BGE_BMANMODE_LOMBUF_ATTN); /* Poll for buffer manager start indication */ for (i = 0; i < BGE_TIMEOUT; i++) { DELAY(10); if (CSR_READ_4(sc, BGE_BMAN_MODE) & BGE_BMANMODE_ENABLE) break; } if (i == BGE_TIMEOUT) { device_printf(sc->bge_dev, "buffer manager failed to start\n"); return (ENXIO); } } /* Enable flow-through queues */ CSR_WRITE_4(sc, BGE_FTQ_RESET, 0xFFFFFFFF); CSR_WRITE_4(sc, BGE_FTQ_RESET, 0); /* Wait until queue initialization is complete */ for (i = 0; i < BGE_TIMEOUT; i++) { DELAY(10); if (CSR_READ_4(sc, BGE_FTQ_RESET) == 0) break; } if (i == BGE_TIMEOUT) { device_printf(sc->bge_dev, "flow-through queue init failed\n"); return (ENXIO); } /* * Summary of rings supported by the controller: * * Standard Receive Producer Ring * - This ring is used to feed receive buffers for "standard" * sized frames (typically 1536 bytes) to the controller. * * Jumbo Receive Producer Ring * - This ring is used to feed receive buffers for jumbo sized * frames (i.e. anything bigger than the "standard" frames) * to the controller. * * Mini Receive Producer Ring * - This ring is used to feed receive buffers for "mini" * sized frames to the controller. * - This feature required external memory for the controller * but was never used in a production system. Should always * be disabled. * * Receive Return Ring * - After the controller has placed an incoming frame into a * receive buffer that buffer is moved into a receive return * ring. The driver is then responsible to passing the * buffer up to the stack. Many versions of the controller * support multiple RR rings. * * Send Ring * - This ring is used for outgoing frames. Many versions of * the controller support multiple send rings. */ /* Initialize the standard receive producer ring control block. */ rcb = &sc->bge_ldata.bge_info.bge_std_rx_rcb; rcb->bge_hostaddr.bge_addr_lo = BGE_ADDR_LO(sc->bge_ldata.bge_rx_std_ring_paddr); rcb->bge_hostaddr.bge_addr_hi = BGE_ADDR_HI(sc->bge_ldata.bge_rx_std_ring_paddr); bus_dmamap_sync(sc->bge_cdata.bge_rx_std_ring_tag, sc->bge_cdata.bge_rx_std_ring_map, BUS_DMASYNC_PREREAD); if (BGE_IS_5717_PLUS(sc)) { /* * Bits 31-16: Programmable ring size (2048, 1024, 512, .., 32) * Bits 15-2 : Maximum RX frame size * Bit 1 : 1 = Ring Disabled, 0 = Ring ENabled * Bit 0 : Reserved */ rcb->bge_maxlen_flags = BGE_RCB_MAXLEN_FLAGS(512, BGE_MAX_FRAMELEN << 2); } else if (BGE_IS_5705_PLUS(sc)) { /* * Bits 31-16: Programmable ring size (512, 256, 128, 64, 32) * Bits 15-2 : Reserved (should be 0) * Bit 1 : 1 = Ring Disabled, 0 = Ring Enabled * Bit 0 : Reserved */ rcb->bge_maxlen_flags = BGE_RCB_MAXLEN_FLAGS(512, 0); } else { /* * Ring size is always XXX entries * Bits 31-16: Maximum RX frame size * Bits 15-2 : Reserved (should be 0) * Bit 1 : 1 = Ring Disabled, 0 = Ring Enabled * Bit 0 : Reserved */ rcb->bge_maxlen_flags = BGE_RCB_MAXLEN_FLAGS(BGE_MAX_FRAMELEN, 0); } if (sc->bge_asicrev == BGE_ASICREV_BCM5717) rcb->bge_nicaddr = BGE_STD_RX_RINGS_5717; else rcb->bge_nicaddr = BGE_STD_RX_RINGS; /* Write the standard receive producer ring control block. */ CSR_WRITE_4(sc, BGE_RX_STD_RCB_HADDR_HI, rcb->bge_hostaddr.bge_addr_hi); CSR_WRITE_4(sc, BGE_RX_STD_RCB_HADDR_LO, rcb->bge_hostaddr.bge_addr_lo); CSR_WRITE_4(sc, BGE_RX_STD_RCB_MAXLEN_FLAGS, rcb->bge_maxlen_flags); CSR_WRITE_4(sc, BGE_RX_STD_RCB_NICADDR, rcb->bge_nicaddr); /* Reset the standard receive producer ring producer index. */ bge_writembx(sc, BGE_MBX_RX_STD_PROD_LO, 0); /* * Initialize the jumbo RX producer ring control * block. We set the 'ring disabled' bit in the * flags field until we're actually ready to start * using this ring (i.e. once we set the MTU * high enough to require it). */ if (BGE_IS_JUMBO_CAPABLE(sc)) { rcb = &sc->bge_ldata.bge_info.bge_jumbo_rx_rcb; /* Get the jumbo receive producer ring RCB parameters. */ rcb->bge_hostaddr.bge_addr_lo = BGE_ADDR_LO(sc->bge_ldata.bge_rx_jumbo_ring_paddr); rcb->bge_hostaddr.bge_addr_hi = BGE_ADDR_HI(sc->bge_ldata.bge_rx_jumbo_ring_paddr); bus_dmamap_sync(sc->bge_cdata.bge_rx_jumbo_ring_tag, sc->bge_cdata.bge_rx_jumbo_ring_map, BUS_DMASYNC_PREREAD); rcb->bge_maxlen_flags = BGE_RCB_MAXLEN_FLAGS(0, BGE_RCB_FLAG_USE_EXT_RX_BD | BGE_RCB_FLAG_RING_DISABLED); if (sc->bge_asicrev == BGE_ASICREV_BCM5717) rcb->bge_nicaddr = BGE_JUMBO_RX_RINGS_5717; else rcb->bge_nicaddr = BGE_JUMBO_RX_RINGS; CSR_WRITE_4(sc, BGE_RX_JUMBO_RCB_HADDR_HI, rcb->bge_hostaddr.bge_addr_hi); CSR_WRITE_4(sc, BGE_RX_JUMBO_RCB_HADDR_LO, rcb->bge_hostaddr.bge_addr_lo); /* Program the jumbo receive producer ring RCB parameters. */ CSR_WRITE_4(sc, BGE_RX_JUMBO_RCB_MAXLEN_FLAGS, rcb->bge_maxlen_flags); CSR_WRITE_4(sc, BGE_RX_JUMBO_RCB_NICADDR, rcb->bge_nicaddr); /* Reset the jumbo receive producer ring producer index. */ bge_writembx(sc, BGE_MBX_RX_JUMBO_PROD_LO, 0); } /* Disable the mini receive producer ring RCB. */ if (BGE_IS_5700_FAMILY(sc)) { rcb = &sc->bge_ldata.bge_info.bge_mini_rx_rcb; rcb->bge_maxlen_flags = BGE_RCB_MAXLEN_FLAGS(0, BGE_RCB_FLAG_RING_DISABLED); CSR_WRITE_4(sc, BGE_RX_MINI_RCB_MAXLEN_FLAGS, rcb->bge_maxlen_flags); /* Reset the mini receive producer ring producer index. */ bge_writembx(sc, BGE_MBX_RX_MINI_PROD_LO, 0); } /* Choose de-pipeline mode for BCM5906 A0, A1 and A2. */ if (sc->bge_asicrev == BGE_ASICREV_BCM5906) { if (sc->bge_chipid == BGE_CHIPID_BCM5906_A0 || sc->bge_chipid == BGE_CHIPID_BCM5906_A1 || sc->bge_chipid == BGE_CHIPID_BCM5906_A2) CSR_WRITE_4(sc, BGE_ISO_PKT_TX, (CSR_READ_4(sc, BGE_ISO_PKT_TX) & ~3) | 2); } /* * The BD ring replenish thresholds control how often the * hardware fetches new BD's from the producer rings in host * memory. Setting the value too low on a busy system can * starve the hardware and recue the throughpout. * * Set the BD ring replentish thresholds. The recommended * values are 1/8th the number of descriptors allocated to * each ring. * XXX The 5754 requires a lower threshold, so it might be a * requirement of all 575x family chips. The Linux driver sets * the lower threshold for all 5705 family chips as well, but there * are reports that it might not need to be so strict. * * XXX Linux does some extra fiddling here for the 5906 parts as * well. */ if (BGE_IS_5705_PLUS(sc)) val = 8; else val = BGE_STD_RX_RING_CNT / 8; CSR_WRITE_4(sc, BGE_RBDI_STD_REPL_THRESH, val); if (BGE_IS_JUMBO_CAPABLE(sc)) CSR_WRITE_4(sc, BGE_RBDI_JUMBO_REPL_THRESH, BGE_JUMBO_RX_RING_CNT/8); if (BGE_IS_5717_PLUS(sc)) { CSR_WRITE_4(sc, BGE_STD_REPLENISH_LWM, 32); CSR_WRITE_4(sc, BGE_JMB_REPLENISH_LWM, 16); } /* * Disable all send rings by setting the 'ring disabled' bit * in the flags field of all the TX send ring control blocks, * located in NIC memory. */ if (!BGE_IS_5705_PLUS(sc)) /* 5700 to 5704 had 16 send rings. */ limit = BGE_TX_RINGS_EXTSSRAM_MAX; else limit = 1; vrcb = BGE_MEMWIN_START + BGE_SEND_RING_RCB; for (i = 0; i < limit; i++) { RCB_WRITE_4(sc, vrcb, bge_maxlen_flags, BGE_RCB_MAXLEN_FLAGS(0, BGE_RCB_FLAG_RING_DISABLED)); RCB_WRITE_4(sc, vrcb, bge_nicaddr, 0); vrcb += sizeof(struct bge_rcb); } /* Configure send ring RCB 0 (we use only the first ring) */ vrcb = BGE_MEMWIN_START + BGE_SEND_RING_RCB; BGE_HOSTADDR(taddr, sc->bge_ldata.bge_tx_ring_paddr); RCB_WRITE_4(sc, vrcb, bge_hostaddr.bge_addr_hi, taddr.bge_addr_hi); RCB_WRITE_4(sc, vrcb, bge_hostaddr.bge_addr_lo, taddr.bge_addr_lo); if (sc->bge_asicrev == BGE_ASICREV_BCM5717) RCB_WRITE_4(sc, vrcb, bge_nicaddr, BGE_SEND_RING_5717); else RCB_WRITE_4(sc, vrcb, bge_nicaddr, BGE_NIC_TXRING_ADDR(0, BGE_TX_RING_CNT)); RCB_WRITE_4(sc, vrcb, bge_maxlen_flags, BGE_RCB_MAXLEN_FLAGS(BGE_TX_RING_CNT, 0)); /* * Disable all receive return rings by setting the * 'ring diabled' bit in the flags field of all the receive * return ring control blocks, located in NIC memory. */ if (sc->bge_asicrev == BGE_ASICREV_BCM5717) { /* Should be 17, use 16 until we get an SRAM map. */ limit = 16; } else if (!BGE_IS_5705_PLUS(sc)) limit = BGE_RX_RINGS_MAX; - else if (sc->bge_asicrev == BGE_ASICREV_BCM5755) + else if (sc->bge_asicrev == BGE_ASICREV_BCM5755 || + sc->bge_asicrev == BGE_ASICREV_BCM57765) limit = 4; else limit = 1; /* Disable all receive return rings. */ vrcb = BGE_MEMWIN_START + BGE_RX_RETURN_RING_RCB; for (i = 0; i < limit; i++) { RCB_WRITE_4(sc, vrcb, bge_hostaddr.bge_addr_hi, 0); RCB_WRITE_4(sc, vrcb, bge_hostaddr.bge_addr_lo, 0); RCB_WRITE_4(sc, vrcb, bge_maxlen_flags, BGE_RCB_FLAG_RING_DISABLED); RCB_WRITE_4(sc, vrcb, bge_nicaddr, 0); bge_writembx(sc, BGE_MBX_RX_CONS0_LO + (i * (sizeof(uint64_t))), 0); vrcb += sizeof(struct bge_rcb); } /* * Set up receive return ring 0. Note that the NIC address * for RX return rings is 0x0. The return rings live entirely * within the host, so the nicaddr field in the RCB isn't used. */ vrcb = BGE_MEMWIN_START + BGE_RX_RETURN_RING_RCB; BGE_HOSTADDR(taddr, sc->bge_ldata.bge_rx_return_ring_paddr); RCB_WRITE_4(sc, vrcb, bge_hostaddr.bge_addr_hi, taddr.bge_addr_hi); RCB_WRITE_4(sc, vrcb, bge_hostaddr.bge_addr_lo, taddr.bge_addr_lo); RCB_WRITE_4(sc, vrcb, bge_nicaddr, 0); RCB_WRITE_4(sc, vrcb, bge_maxlen_flags, BGE_RCB_MAXLEN_FLAGS(sc->bge_return_ring_cnt, 0)); /* Set random backoff seed for TX */ CSR_WRITE_4(sc, BGE_TX_RANDOM_BACKOFF, IF_LLADDR(sc->bge_ifp)[0] + IF_LLADDR(sc->bge_ifp)[1] + IF_LLADDR(sc->bge_ifp)[2] + IF_LLADDR(sc->bge_ifp)[3] + IF_LLADDR(sc->bge_ifp)[4] + IF_LLADDR(sc->bge_ifp)[5] + BGE_TX_BACKOFF_SEED_MASK); /* Set inter-packet gap */ CSR_WRITE_4(sc, BGE_TX_LENGTHS, 0x2620); /* * Specify which ring to use for packets that don't match * any RX rules. */ CSR_WRITE_4(sc, BGE_RX_RULES_CFG, 0x08); /* * Configure number of RX lists. One interrupt distribution * list, sixteen active lists, one bad frames class. */ CSR_WRITE_4(sc, BGE_RXLP_CFG, 0x181); /* Inialize RX list placement stats mask. */ CSR_WRITE_4(sc, BGE_RXLP_STATS_ENABLE_MASK, 0x007FFFFF); CSR_WRITE_4(sc, BGE_RXLP_STATS_CTL, 0x1); /* Disable host coalescing until we get it set up */ CSR_WRITE_4(sc, BGE_HCC_MODE, 0x00000000); /* Poll to make sure it's shut down. */ for (i = 0; i < BGE_TIMEOUT; i++) { DELAY(10); if (!(CSR_READ_4(sc, BGE_HCC_MODE) & BGE_HCCMODE_ENABLE)) break; } if (i == BGE_TIMEOUT) { device_printf(sc->bge_dev, "host coalescing engine failed to idle\n"); return (ENXIO); } /* Set up host coalescing defaults */ CSR_WRITE_4(sc, BGE_HCC_RX_COAL_TICKS, sc->bge_rx_coal_ticks); CSR_WRITE_4(sc, BGE_HCC_TX_COAL_TICKS, sc->bge_tx_coal_ticks); CSR_WRITE_4(sc, BGE_HCC_RX_MAX_COAL_BDS, sc->bge_rx_max_coal_bds); CSR_WRITE_4(sc, BGE_HCC_TX_MAX_COAL_BDS, sc->bge_tx_max_coal_bds); if (!(BGE_IS_5705_PLUS(sc))) { CSR_WRITE_4(sc, BGE_HCC_RX_COAL_TICKS_INT, 0); CSR_WRITE_4(sc, BGE_HCC_TX_COAL_TICKS_INT, 0); } CSR_WRITE_4(sc, BGE_HCC_RX_MAX_COAL_BDS_INT, 1); CSR_WRITE_4(sc, BGE_HCC_TX_MAX_COAL_BDS_INT, 1); /* Set up address of statistics block */ if (!(BGE_IS_5705_PLUS(sc))) { CSR_WRITE_4(sc, BGE_HCC_STATS_ADDR_HI, BGE_ADDR_HI(sc->bge_ldata.bge_stats_paddr)); CSR_WRITE_4(sc, BGE_HCC_STATS_ADDR_LO, BGE_ADDR_LO(sc->bge_ldata.bge_stats_paddr)); CSR_WRITE_4(sc, BGE_HCC_STATS_BASEADDR, BGE_STATS_BLOCK); CSR_WRITE_4(sc, BGE_HCC_STATUSBLK_BASEADDR, BGE_STATUS_BLOCK); CSR_WRITE_4(sc, BGE_HCC_STATS_TICKS, sc->bge_stat_ticks); } /* Set up address of status block */ CSR_WRITE_4(sc, BGE_HCC_STATUSBLK_ADDR_HI, BGE_ADDR_HI(sc->bge_ldata.bge_status_block_paddr)); CSR_WRITE_4(sc, BGE_HCC_STATUSBLK_ADDR_LO, BGE_ADDR_LO(sc->bge_ldata.bge_status_block_paddr)); /* Set up status block size. */ if (sc->bge_asicrev == BGE_ASICREV_BCM5700 && sc->bge_chipid != BGE_CHIPID_BCM5700_C0) { val = BGE_STATBLKSZ_FULL; bzero(sc->bge_ldata.bge_status_block, BGE_STATUS_BLK_SZ); } else { val = BGE_STATBLKSZ_32BYTE; bzero(sc->bge_ldata.bge_status_block, 32); } bus_dmamap_sync(sc->bge_cdata.bge_status_tag, sc->bge_cdata.bge_status_map, BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); /* Turn on host coalescing state machine */ CSR_WRITE_4(sc, BGE_HCC_MODE, val | BGE_HCCMODE_ENABLE); /* Turn on RX BD completion state machine and enable attentions */ CSR_WRITE_4(sc, BGE_RBDC_MODE, BGE_RBDCMODE_ENABLE | BGE_RBDCMODE_ATTN); /* Turn on RX list placement state machine */ CSR_WRITE_4(sc, BGE_RXLP_MODE, BGE_RXLPMODE_ENABLE); /* Turn on RX list selector state machine. */ if (!(BGE_IS_5705_PLUS(sc))) CSR_WRITE_4(sc, BGE_RXLS_MODE, BGE_RXLSMODE_ENABLE); val = BGE_MACMODE_TXDMA_ENB | BGE_MACMODE_RXDMA_ENB | BGE_MACMODE_RX_STATS_CLEAR | BGE_MACMODE_TX_STATS_CLEAR | BGE_MACMODE_RX_STATS_ENB | BGE_MACMODE_TX_STATS_ENB | BGE_MACMODE_FRMHDR_DMA_ENB; if (sc->bge_flags & BGE_FLAG_TBI) val |= BGE_PORTMODE_TBI; else if (sc->bge_flags & BGE_FLAG_MII_SERDES) val |= BGE_PORTMODE_GMII; else val |= BGE_PORTMODE_MII; /* Turn on DMA, clear stats */ CSR_WRITE_4(sc, BGE_MAC_MODE, val); /* Set misc. local control, enable interrupts on attentions */ CSR_WRITE_4(sc, BGE_MISC_LOCAL_CTL, BGE_MLC_INTR_ONATTN); #ifdef notdef /* Assert GPIO pins for PHY reset */ BGE_SETBIT(sc, BGE_MISC_LOCAL_CTL, BGE_MLC_MISCIO_OUT0 | BGE_MLC_MISCIO_OUT1 | BGE_MLC_MISCIO_OUT2); BGE_SETBIT(sc, BGE_MISC_LOCAL_CTL, BGE_MLC_MISCIO_OUTEN0 | BGE_MLC_MISCIO_OUTEN1 | BGE_MLC_MISCIO_OUTEN2); #endif /* Turn on DMA completion state machine */ if (!(BGE_IS_5705_PLUS(sc))) CSR_WRITE_4(sc, BGE_DMAC_MODE, BGE_DMACMODE_ENABLE); val = BGE_WDMAMODE_ENABLE | BGE_WDMAMODE_ALL_ATTNS; /* Enable host coalescing bug fix. */ if (BGE_IS_5755_PLUS(sc)) val |= BGE_WDMAMODE_STATUS_TAG_FIX; /* Request larger DMA burst size to get better performance. */ if (sc->bge_asicrev == BGE_ASICREV_BCM5785) val |= BGE_WDMAMODE_BURST_ALL_DATA; /* Turn on write DMA state machine */ CSR_WRITE_4(sc, BGE_WDMA_MODE, val); DELAY(40); /* Turn on read DMA state machine */ val = BGE_RDMAMODE_ENABLE | BGE_RDMAMODE_ALL_ATTNS; if (sc->bge_asicrev == BGE_ASICREV_BCM5717) val |= BGE_RDMAMODE_MULT_DMA_RD_DIS; if (sc->bge_asicrev == BGE_ASICREV_BCM5784 || sc->bge_asicrev == BGE_ASICREV_BCM5785 || sc->bge_asicrev == BGE_ASICREV_BCM57780) val |= BGE_RDMAMODE_BD_SBD_CRPT_ATTN | BGE_RDMAMODE_MBUF_RBD_CRPT_ATTN | BGE_RDMAMODE_MBUF_SBD_CRPT_ATTN; if (sc->bge_flags & BGE_FLAG_PCIE) val |= BGE_RDMAMODE_FIFO_LONG_BURST; if (sc->bge_flags & (BGE_FLAG_TSO | BGE_FLAG_TSO3)) { val |= BGE_RDMAMODE_TSO4_ENABLE; if (sc->bge_flags & BGE_FLAG_TSO3 || sc->bge_asicrev == BGE_ASICREV_BCM5785 || sc->bge_asicrev == BGE_ASICREV_BCM57780) val |= BGE_RDMAMODE_TSO6_ENABLE; } if (sc->bge_asicrev == BGE_ASICREV_BCM5761 || sc->bge_asicrev == BGE_ASICREV_BCM5784 || sc->bge_asicrev == BGE_ASICREV_BCM5785 || sc->bge_asicrev == BGE_ASICREV_BCM57780 || BGE_IS_5717_PLUS(sc)) { /* * Enable fix for read DMA FIFO overruns. * The fix is to limit the number of RX BDs * the hardware would fetch at a fime. */ CSR_WRITE_4(sc, BGE_RDMA_RSRVCTRL, CSR_READ_4(sc, BGE_RDMA_RSRVCTRL) | BGE_RDMA_RSRVCTRL_FIFO_OFLW_FIX); } CSR_WRITE_4(sc, BGE_RDMA_MODE, val); DELAY(40); /* Turn on RX data completion state machine */ CSR_WRITE_4(sc, BGE_RDC_MODE, BGE_RDCMODE_ENABLE); /* Turn on RX BD initiator state machine */ CSR_WRITE_4(sc, BGE_RBDI_MODE, BGE_RBDIMODE_ENABLE); /* Turn on RX data and RX BD initiator state machine */ CSR_WRITE_4(sc, BGE_RDBDI_MODE, BGE_RDBDIMODE_ENABLE); /* Turn on Mbuf cluster free state machine */ if (!(BGE_IS_5705_PLUS(sc))) CSR_WRITE_4(sc, BGE_MBCF_MODE, BGE_MBCFMODE_ENABLE); /* Turn on send BD completion state machine */ CSR_WRITE_4(sc, BGE_SBDC_MODE, BGE_SBDCMODE_ENABLE); /* Turn on send data completion state machine */ val = BGE_SDCMODE_ENABLE; if (sc->bge_asicrev == BGE_ASICREV_BCM5761) val |= BGE_SDCMODE_CDELAY; CSR_WRITE_4(sc, BGE_SDC_MODE, val); /* Turn on send data initiator state machine */ if (sc->bge_flags & (BGE_FLAG_TSO | BGE_FLAG_TSO3)) CSR_WRITE_4(sc, BGE_SDI_MODE, BGE_SDIMODE_ENABLE | BGE_SDIMODE_HW_LSO_PRE_DMA); else CSR_WRITE_4(sc, BGE_SDI_MODE, BGE_SDIMODE_ENABLE); /* Turn on send BD initiator state machine */ CSR_WRITE_4(sc, BGE_SBDI_MODE, BGE_SBDIMODE_ENABLE); /* Turn on send BD selector state machine */ CSR_WRITE_4(sc, BGE_SRS_MODE, BGE_SRSMODE_ENABLE); CSR_WRITE_4(sc, BGE_SDI_STATS_ENABLE_MASK, 0x007FFFFF); CSR_WRITE_4(sc, BGE_SDI_STATS_CTL, BGE_SDISTATSCTL_ENABLE | BGE_SDISTATSCTL_FASTER); /* ack/clear link change events */ CSR_WRITE_4(sc, BGE_MAC_STS, BGE_MACSTAT_SYNC_CHANGED | BGE_MACSTAT_CFG_CHANGED | BGE_MACSTAT_MI_COMPLETE | BGE_MACSTAT_LINK_CHANGED); CSR_WRITE_4(sc, BGE_MI_STS, 0); /* * Enable attention when the link has changed state for * devices that use auto polling. */ if (sc->bge_flags & BGE_FLAG_TBI) { CSR_WRITE_4(sc, BGE_MI_STS, BGE_MISTS_LINK); } else { if (sc->bge_mi_mode & BGE_MIMODE_AUTOPOLL) { CSR_WRITE_4(sc, BGE_MI_MODE, sc->bge_mi_mode); DELAY(80); } if (sc->bge_asicrev == BGE_ASICREV_BCM5700 && sc->bge_chipid != BGE_CHIPID_BCM5700_B2) CSR_WRITE_4(sc, BGE_MAC_EVT_ENB, BGE_EVTENB_MI_INTERRUPT); } /* * Clear any pending link state attention. * Otherwise some link state change events may be lost until attention * is cleared by bge_intr() -> bge_link_upd() sequence. * It's not necessary on newer BCM chips - perhaps enabling link * state change attentions implies clearing pending attention. */ CSR_WRITE_4(sc, BGE_MAC_STS, BGE_MACSTAT_SYNC_CHANGED | BGE_MACSTAT_CFG_CHANGED | BGE_MACSTAT_MI_COMPLETE | BGE_MACSTAT_LINK_CHANGED); /* Enable link state change attentions. */ BGE_SETBIT(sc, BGE_MAC_EVT_ENB, BGE_EVTENB_LINK_CHANGED); return (0); } const struct bge_revision * bge_lookup_rev(uint32_t chipid) { const struct bge_revision *br; for (br = bge_revisions; br->br_name != NULL; br++) { if (br->br_chipid == chipid) return (br); } for (br = bge_majorrevs; br->br_name != NULL; br++) { if (br->br_chipid == BGE_ASICREV(chipid)) return (br); } return (NULL); } const struct bge_vendor * bge_lookup_vendor(uint16_t vid) { const struct bge_vendor *v; for (v = bge_vendors; v->v_name != NULL; v++) if (v->v_id == vid) return (v); panic("%s: unknown vendor %d", __func__, vid); return (NULL); } /* * Probe for a Broadcom chip. Check the PCI vendor and device IDs * against our list and return its name if we find a match. * * Note that since the Broadcom controller contains VPD support, we * try to get the device name string from the controller itself instead * of the compiled-in string. It guarantees we'll always announce the * right product name. We fall back to the compiled-in string when * VPD is unavailable or corrupt. */ static int bge_probe(device_t dev) { char buf[96]; char model[64]; const struct bge_revision *br; const char *pname; struct bge_softc *sc = device_get_softc(dev); const struct bge_type *t = bge_devs; const struct bge_vendor *v; uint32_t id; uint16_t did, vid; sc->bge_dev = dev; vid = pci_get_vendor(dev); did = pci_get_device(dev); while(t->bge_vid != 0) { if ((vid == t->bge_vid) && (did == t->bge_did)) { id = pci_read_config(dev, BGE_PCI_MISC_CTL, 4) >> BGE_PCIMISCCTL_ASICREV_SHIFT; if (BGE_ASICREV(id) == BGE_ASICREV_USE_PRODID_REG) { /* * Find the ASCI revision. Different chips * use different registers. */ switch (pci_get_device(dev)) { case BCOM_DEVICEID_BCM5717: case BCOM_DEVICEID_BCM5718: id = pci_read_config(dev, BGE_PCI_GEN2_PRODID_ASICREV, 4); break; + case BCOM_DEVICEID_BCM57761: + case BCOM_DEVICEID_BCM57765: + case BCOM_DEVICEID_BCM57781: + case BCOM_DEVICEID_BCM57785: + case BCOM_DEVICEID_BCM57791: + case BCOM_DEVICEID_BCM57795: + id = pci_read_config(dev, + BGE_PCI_GEN15_PRODID_ASICREV, 4); + break; default: id = pci_read_config(dev, BGE_PCI_PRODID_ASICREV, 4); } } br = bge_lookup_rev(id); v = bge_lookup_vendor(vid); if (bge_has_eaddr(sc) && pci_get_vpd_ident(dev, &pname) == 0) snprintf(model, 64, "%s", pname); else snprintf(model, 64, "%s %s", v->v_name, br != NULL ? br->br_name : "NetXtreme Ethernet Controller"); snprintf(buf, 96, "%s, %sASIC rev. %#08x", model, br != NULL ? "" : "unknown ", id); device_set_desc_copy(dev, buf); return (0); } t++; } return (ENXIO); } static void bge_dma_free(struct bge_softc *sc) { int i; /* Destroy DMA maps for RX buffers. */ for (i = 0; i < BGE_STD_RX_RING_CNT; i++) { if (sc->bge_cdata.bge_rx_std_dmamap[i]) bus_dmamap_destroy(sc->bge_cdata.bge_rx_mtag, sc->bge_cdata.bge_rx_std_dmamap[i]); } if (sc->bge_cdata.bge_rx_std_sparemap) bus_dmamap_destroy(sc->bge_cdata.bge_rx_mtag, sc->bge_cdata.bge_rx_std_sparemap); /* Destroy DMA maps for jumbo RX buffers. */ for (i = 0; i < BGE_JUMBO_RX_RING_CNT; i++) { if (sc->bge_cdata.bge_rx_jumbo_dmamap[i]) bus_dmamap_destroy(sc->bge_cdata.bge_mtag_jumbo, sc->bge_cdata.bge_rx_jumbo_dmamap[i]); } if (sc->bge_cdata.bge_rx_jumbo_sparemap) bus_dmamap_destroy(sc->bge_cdata.bge_mtag_jumbo, sc->bge_cdata.bge_rx_jumbo_sparemap); /* Destroy DMA maps for TX buffers. */ for (i = 0; i < BGE_TX_RING_CNT; i++) { if (sc->bge_cdata.bge_tx_dmamap[i]) bus_dmamap_destroy(sc->bge_cdata.bge_tx_mtag, sc->bge_cdata.bge_tx_dmamap[i]); } if (sc->bge_cdata.bge_rx_mtag) bus_dma_tag_destroy(sc->bge_cdata.bge_rx_mtag); if (sc->bge_cdata.bge_tx_mtag) bus_dma_tag_destroy(sc->bge_cdata.bge_tx_mtag); /* Destroy standard RX ring. */ if (sc->bge_cdata.bge_rx_std_ring_map) bus_dmamap_unload(sc->bge_cdata.bge_rx_std_ring_tag, sc->bge_cdata.bge_rx_std_ring_map); if (sc->bge_cdata.bge_rx_std_ring_map && sc->bge_ldata.bge_rx_std_ring) bus_dmamem_free(sc->bge_cdata.bge_rx_std_ring_tag, sc->bge_ldata.bge_rx_std_ring, sc->bge_cdata.bge_rx_std_ring_map); if (sc->bge_cdata.bge_rx_std_ring_tag) bus_dma_tag_destroy(sc->bge_cdata.bge_rx_std_ring_tag); /* Destroy jumbo RX ring. */ if (sc->bge_cdata.bge_rx_jumbo_ring_map) bus_dmamap_unload(sc->bge_cdata.bge_rx_jumbo_ring_tag, sc->bge_cdata.bge_rx_jumbo_ring_map); if (sc->bge_cdata.bge_rx_jumbo_ring_map && sc->bge_ldata.bge_rx_jumbo_ring) bus_dmamem_free(sc->bge_cdata.bge_rx_jumbo_ring_tag, sc->bge_ldata.bge_rx_jumbo_ring, sc->bge_cdata.bge_rx_jumbo_ring_map); if (sc->bge_cdata.bge_rx_jumbo_ring_tag) bus_dma_tag_destroy(sc->bge_cdata.bge_rx_jumbo_ring_tag); /* Destroy RX return ring. */ if (sc->bge_cdata.bge_rx_return_ring_map) bus_dmamap_unload(sc->bge_cdata.bge_rx_return_ring_tag, sc->bge_cdata.bge_rx_return_ring_map); if (sc->bge_cdata.bge_rx_return_ring_map && sc->bge_ldata.bge_rx_return_ring) bus_dmamem_free(sc->bge_cdata.bge_rx_return_ring_tag, sc->bge_ldata.bge_rx_return_ring, sc->bge_cdata.bge_rx_return_ring_map); if (sc->bge_cdata.bge_rx_return_ring_tag) bus_dma_tag_destroy(sc->bge_cdata.bge_rx_return_ring_tag); /* Destroy TX ring. */ if (sc->bge_cdata.bge_tx_ring_map) bus_dmamap_unload(sc->bge_cdata.bge_tx_ring_tag, sc->bge_cdata.bge_tx_ring_map); if (sc->bge_cdata.bge_tx_ring_map && sc->bge_ldata.bge_tx_ring) bus_dmamem_free(sc->bge_cdata.bge_tx_ring_tag, sc->bge_ldata.bge_tx_ring, sc->bge_cdata.bge_tx_ring_map); if (sc->bge_cdata.bge_tx_ring_tag) bus_dma_tag_destroy(sc->bge_cdata.bge_tx_ring_tag); /* Destroy status block. */ if (sc->bge_cdata.bge_status_map) bus_dmamap_unload(sc->bge_cdata.bge_status_tag, sc->bge_cdata.bge_status_map); if (sc->bge_cdata.bge_status_map && sc->bge_ldata.bge_status_block) bus_dmamem_free(sc->bge_cdata.bge_status_tag, sc->bge_ldata.bge_status_block, sc->bge_cdata.bge_status_map); if (sc->bge_cdata.bge_status_tag) bus_dma_tag_destroy(sc->bge_cdata.bge_status_tag); /* Destroy statistics block. */ if (sc->bge_cdata.bge_stats_map) bus_dmamap_unload(sc->bge_cdata.bge_stats_tag, sc->bge_cdata.bge_stats_map); if (sc->bge_cdata.bge_stats_map && sc->bge_ldata.bge_stats) bus_dmamem_free(sc->bge_cdata.bge_stats_tag, sc->bge_ldata.bge_stats, sc->bge_cdata.bge_stats_map); if (sc->bge_cdata.bge_stats_tag) bus_dma_tag_destroy(sc->bge_cdata.bge_stats_tag); if (sc->bge_cdata.bge_buffer_tag) bus_dma_tag_destroy(sc->bge_cdata.bge_buffer_tag); /* Destroy the parent tag. */ if (sc->bge_cdata.bge_parent_tag) bus_dma_tag_destroy(sc->bge_cdata.bge_parent_tag); } static int bge_dma_ring_alloc(struct bge_softc *sc, bus_size_t alignment, bus_size_t maxsize, bus_dma_tag_t *tag, uint8_t **ring, bus_dmamap_t *map, bus_addr_t *paddr, const char *msg) { struct bge_dmamap_arg ctx; bus_addr_t lowaddr; bus_size_t ring_end; int error; lowaddr = BUS_SPACE_MAXADDR; again: error = bus_dma_tag_create(sc->bge_cdata.bge_parent_tag, alignment, 0, lowaddr, BUS_SPACE_MAXADDR, NULL, NULL, maxsize, 1, maxsize, 0, NULL, NULL, tag); if (error != 0) { device_printf(sc->bge_dev, "could not create %s dma tag\n", msg); return (ENOMEM); } /* Allocate DMA'able memory for ring. */ error = bus_dmamem_alloc(*tag, (void **)ring, BUS_DMA_NOWAIT | BUS_DMA_ZERO | BUS_DMA_COHERENT, map); if (error != 0) { device_printf(sc->bge_dev, "could not allocate DMA'able memory for %s\n", msg); return (ENOMEM); } /* Load the address of the ring. */ ctx.bge_busaddr = 0; error = bus_dmamap_load(*tag, *map, *ring, maxsize, bge_dma_map_addr, &ctx, BUS_DMA_NOWAIT); if (error != 0) { device_printf(sc->bge_dev, "could not load DMA'able memory for %s\n", msg); return (ENOMEM); } *paddr = ctx.bge_busaddr; ring_end = *paddr + maxsize; if ((sc->bge_flags & BGE_FLAG_4G_BNDRY_BUG) != 0 && BGE_ADDR_HI(*paddr) != BGE_ADDR_HI(ring_end)) { /* * 4GB boundary crossed. Limit maximum allowable DMA * address space to 32bit and try again. */ bus_dmamap_unload(*tag, *map); bus_dmamem_free(*tag, *ring, *map); bus_dma_tag_destroy(*tag); if (bootverbose) device_printf(sc->bge_dev, "4GB boundary crossed, " "limit DMA address space to 32bit for %s\n", msg); *ring = NULL; *tag = NULL; *map = NULL; lowaddr = BUS_SPACE_MAXADDR_32BIT; goto again; } return (0); } static int bge_dma_alloc(struct bge_softc *sc) { bus_addr_t lowaddr; bus_size_t boundary, sbsz, rxmaxsegsz, txsegsz, txmaxsegsz; int i, error; lowaddr = BUS_SPACE_MAXADDR; if ((sc->bge_flags & BGE_FLAG_40BIT_BUG) != 0) lowaddr = BGE_DMA_MAXADDR; /* * Allocate the parent bus DMA tag appropriate for PCI. */ error = bus_dma_tag_create(bus_get_dma_tag(sc->bge_dev), 1, 0, lowaddr, BUS_SPACE_MAXADDR, NULL, NULL, BUS_SPACE_MAXSIZE_32BIT, 0, BUS_SPACE_MAXSIZE_32BIT, 0, NULL, NULL, &sc->bge_cdata.bge_parent_tag); if (error != 0) { device_printf(sc->bge_dev, "could not allocate parent dma tag\n"); return (ENOMEM); } /* Create tag for standard RX ring. */ error = bge_dma_ring_alloc(sc, PAGE_SIZE, BGE_STD_RX_RING_SZ, &sc->bge_cdata.bge_rx_std_ring_tag, (uint8_t **)&sc->bge_ldata.bge_rx_std_ring, &sc->bge_cdata.bge_rx_std_ring_map, &sc->bge_ldata.bge_rx_std_ring_paddr, "RX ring"); if (error) return (error); /* Create tag for RX return ring. */ error = bge_dma_ring_alloc(sc, PAGE_SIZE, BGE_RX_RTN_RING_SZ(sc), &sc->bge_cdata.bge_rx_return_ring_tag, (uint8_t **)&sc->bge_ldata.bge_rx_return_ring, &sc->bge_cdata.bge_rx_return_ring_map, &sc->bge_ldata.bge_rx_return_ring_paddr, "RX return ring"); if (error) return (error); /* Create tag for TX ring. */ error = bge_dma_ring_alloc(sc, PAGE_SIZE, BGE_TX_RING_SZ, &sc->bge_cdata.bge_tx_ring_tag, (uint8_t **)&sc->bge_ldata.bge_tx_ring, &sc->bge_cdata.bge_tx_ring_map, &sc->bge_ldata.bge_tx_ring_paddr, "TX ring"); if (error) return (error); /* * Create tag for status block. * Because we only use single Tx/Rx/Rx return ring, use * minimum status block size except BCM5700 AX/BX which * seems to want to see full status block size regardless * of configured number of ring. */ if (sc->bge_asicrev == BGE_ASICREV_BCM5700 && sc->bge_chipid != BGE_CHIPID_BCM5700_C0) sbsz = BGE_STATUS_BLK_SZ; else sbsz = 32; error = bge_dma_ring_alloc(sc, PAGE_SIZE, sbsz, &sc->bge_cdata.bge_status_tag, (uint8_t **)&sc->bge_ldata.bge_status_block, &sc->bge_cdata.bge_status_map, &sc->bge_ldata.bge_status_block_paddr, "status block"); if (error) return (error); /* Create tag for statistics block. */ error = bge_dma_ring_alloc(sc, PAGE_SIZE, BGE_STATS_SZ, &sc->bge_cdata.bge_stats_tag, (uint8_t **)&sc->bge_ldata.bge_stats, &sc->bge_cdata.bge_stats_map, &sc->bge_ldata.bge_stats_paddr, "statistics block"); if (error) return (error); /* Create tag for jumbo RX ring. */ if (BGE_IS_JUMBO_CAPABLE(sc)) { error = bge_dma_ring_alloc(sc, PAGE_SIZE, BGE_JUMBO_RX_RING_SZ, &sc->bge_cdata.bge_rx_jumbo_ring_tag, (uint8_t **)&sc->bge_ldata.bge_rx_jumbo_ring, &sc->bge_cdata.bge_rx_jumbo_ring_map, &sc->bge_ldata.bge_rx_jumbo_ring_paddr, "jumbo RX ring"); if (error) return (error); } /* Create parent tag for buffers. */ boundary = 0; if ((sc->bge_flags & BGE_FLAG_4G_BNDRY_BUG) != 0) { boundary = BGE_DMA_BNDRY; /* * XXX * watchdog timeout issue was observed on BCM5704 which * lives behind PCI-X bridge(e.g AMD 8131 PCI-X bridge). * Limiting DMA address space to 32bits seems to address * it. */ if (sc->bge_flags & BGE_FLAG_PCIX) lowaddr = BUS_SPACE_MAXADDR_32BIT; } error = bus_dma_tag_create(bus_get_dma_tag(sc->bge_dev), 1, boundary, lowaddr, BUS_SPACE_MAXADDR, NULL, NULL, BUS_SPACE_MAXSIZE_32BIT, 0, BUS_SPACE_MAXSIZE_32BIT, 0, NULL, NULL, &sc->bge_cdata.bge_buffer_tag); if (error != 0) { device_printf(sc->bge_dev, "could not allocate buffer dma tag\n"); return (ENOMEM); } /* Create tag for Tx mbufs. */ if (sc->bge_flags & (BGE_FLAG_TSO | BGE_FLAG_TSO3)) { txsegsz = BGE_TSOSEG_SZ; txmaxsegsz = 65535 + sizeof(struct ether_vlan_header); } else { txsegsz = MCLBYTES; txmaxsegsz = MCLBYTES * BGE_NSEG_NEW; } error = bus_dma_tag_create(sc->bge_cdata.bge_buffer_tag, 1, 0, BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL, txmaxsegsz, BGE_NSEG_NEW, txsegsz, 0, NULL, NULL, &sc->bge_cdata.bge_tx_mtag); if (error) { device_printf(sc->bge_dev, "could not allocate TX dma tag\n"); return (ENOMEM); } /* Create tag for Rx mbufs. */ if (sc->bge_flags & BGE_FLAG_JUMBO_STD) rxmaxsegsz = MJUM9BYTES; else rxmaxsegsz = MCLBYTES; error = bus_dma_tag_create(sc->bge_cdata.bge_buffer_tag, 1, 0, BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL, rxmaxsegsz, 1, rxmaxsegsz, 0, NULL, NULL, &sc->bge_cdata.bge_rx_mtag); if (error) { device_printf(sc->bge_dev, "could not allocate RX dma tag\n"); return (ENOMEM); } /* Create DMA maps for RX buffers. */ error = bus_dmamap_create(sc->bge_cdata.bge_rx_mtag, 0, &sc->bge_cdata.bge_rx_std_sparemap); if (error) { device_printf(sc->bge_dev, "can't create spare DMA map for RX\n"); return (ENOMEM); } for (i = 0; i < BGE_STD_RX_RING_CNT; i++) { error = bus_dmamap_create(sc->bge_cdata.bge_rx_mtag, 0, &sc->bge_cdata.bge_rx_std_dmamap[i]); if (error) { device_printf(sc->bge_dev, "can't create DMA map for RX\n"); return (ENOMEM); } } /* Create DMA maps for TX buffers. */ for (i = 0; i < BGE_TX_RING_CNT; i++) { error = bus_dmamap_create(sc->bge_cdata.bge_tx_mtag, 0, &sc->bge_cdata.bge_tx_dmamap[i]); if (error) { device_printf(sc->bge_dev, "can't create DMA map for TX\n"); return (ENOMEM); } } /* Create tags for jumbo RX buffers. */ if (BGE_IS_JUMBO_CAPABLE(sc)) { error = bus_dma_tag_create(sc->bge_cdata.bge_buffer_tag, 1, 0, BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL, MJUM9BYTES, BGE_NSEG_JUMBO, PAGE_SIZE, 0, NULL, NULL, &sc->bge_cdata.bge_mtag_jumbo); if (error) { device_printf(sc->bge_dev, "could not allocate jumbo dma tag\n"); return (ENOMEM); } /* Create DMA maps for jumbo RX buffers. */ error = bus_dmamap_create(sc->bge_cdata.bge_mtag_jumbo, 0, &sc->bge_cdata.bge_rx_jumbo_sparemap); if (error) { device_printf(sc->bge_dev, "can't create spare DMA map for jumbo RX\n"); return (ENOMEM); } for (i = 0; i < BGE_JUMBO_RX_RING_CNT; i++) { error = bus_dmamap_create(sc->bge_cdata.bge_mtag_jumbo, 0, &sc->bge_cdata.bge_rx_jumbo_dmamap[i]); if (error) { device_printf(sc->bge_dev, "can't create DMA map for jumbo RX\n"); return (ENOMEM); } } } return (0); } /* * Return true if this device has more than one port. */ static int bge_has_multiple_ports(struct bge_softc *sc) { device_t dev = sc->bge_dev; u_int b, d, f, fscan, s; d = pci_get_domain(dev); b = pci_get_bus(dev); s = pci_get_slot(dev); f = pci_get_function(dev); for (fscan = 0; fscan <= PCI_FUNCMAX; fscan++) if (fscan != f && pci_find_dbsf(d, b, s, fscan) != NULL) return (1); return (0); } /* * Return true if MSI can be used with this device. */ static int bge_can_use_msi(struct bge_softc *sc) { int can_use_msi = 0; /* Disable MSI for polling(4). */ #ifdef DEVICE_POLLING return (0); #endif switch (sc->bge_asicrev) { case BGE_ASICREV_BCM5714_A0: case BGE_ASICREV_BCM5714: /* * Apparently, MSI doesn't work when these chips are * configured in single-port mode. */ if (bge_has_multiple_ports(sc)) can_use_msi = 1; break; case BGE_ASICREV_BCM5750: if (sc->bge_chiprev != BGE_CHIPREV_5750_AX && sc->bge_chiprev != BGE_CHIPREV_5750_BX) can_use_msi = 1; break; default: if (BGE_IS_575X_PLUS(sc)) can_use_msi = 1; } return (can_use_msi); } static int bge_attach(device_t dev) { struct ifnet *ifp; struct bge_softc *sc; uint32_t hwcfg = 0, misccfg; u_char eaddr[ETHER_ADDR_LEN]; int capmask, error, f, msicount, phy_addr, reg, rid, trys; sc = device_get_softc(dev); sc->bge_dev = dev; TASK_INIT(&sc->bge_intr_task, 0, bge_intr_task, sc); /* * Map control/status registers. */ pci_enable_busmaster(dev); rid = PCIR_BAR(0); sc->bge_res = bus_alloc_resource_any(dev, SYS_RES_MEMORY, &rid, RF_ACTIVE); if (sc->bge_res == NULL) { device_printf (sc->bge_dev, "couldn't map memory\n"); error = ENXIO; goto fail; } /* Save various chip information. */ sc->bge_chipid = pci_read_config(dev, BGE_PCI_MISC_CTL, 4) >> BGE_PCIMISCCTL_ASICREV_SHIFT; if (BGE_ASICREV(sc->bge_chipid) == BGE_ASICREV_USE_PRODID_REG) { /* * Find the ASCI revision. Different chips use different * registers. */ switch (pci_get_device(dev)) { case BCOM_DEVICEID_BCM5717: case BCOM_DEVICEID_BCM5718: sc->bge_chipid = pci_read_config(dev, BGE_PCI_GEN2_PRODID_ASICREV, 4); break; + case BCOM_DEVICEID_BCM57761: + case BCOM_DEVICEID_BCM57765: + case BCOM_DEVICEID_BCM57781: + case BCOM_DEVICEID_BCM57785: + case BCOM_DEVICEID_BCM57791: + case BCOM_DEVICEID_BCM57795: + sc->bge_chipid = pci_read_config(dev, + BGE_PCI_GEN15_PRODID_ASICREV, 4); + break; default: sc->bge_chipid = pci_read_config(dev, BGE_PCI_PRODID_ASICREV, 4); } } sc->bge_asicrev = BGE_ASICREV(sc->bge_chipid); sc->bge_chiprev = BGE_CHIPREV(sc->bge_chipid); /* Set default PHY address. */ phy_addr = 1; /* * PHY address mapping for various devices. * * | F0 Cu | F0 Sr | F1 Cu | F1 Sr | * ---------+-------+-------+-------+-------+ * BCM57XX | 1 | X | X | X | * BCM5704 | 1 | X | 1 | X | * BCM5717 | 1 | 8 | 2 | 9 | * * Other addresses may respond but they are not * IEEE compliant PHYs and should be ignored. */ if (sc->bge_asicrev == BGE_ASICREV_BCM5717) { f = pci_get_function(dev); if (sc->bge_chipid == BGE_CHIPID_BCM5717_A0) { if (CSR_READ_4(sc, BGE_SGDIG_STS) & BGE_SGDIGSTS_IS_SERDES) phy_addr = f + 8; else phy_addr = f + 1; } else if (sc->bge_chipid == BGE_CHIPID_BCM5717_B0) { if (CSR_READ_4(sc, BGE_CPMU_PHY_STRAP) & BGE_CPMU_PHY_STRAP_IS_SERDES) phy_addr = f + 8; else phy_addr = f + 1; } } /* * Don't enable Ethernet@WireSpeed for the 5700, 5906, or the * 5705 A0 and A1 chips. */ - if (sc->bge_asicrev != BGE_ASICREV_BCM5700 && - sc->bge_asicrev != BGE_ASICREV_BCM5906 && - sc->bge_chipid != BGE_CHIPID_BCM5705_A0 && - sc->bge_chipid != BGE_CHIPID_BCM5705_A1 && - !BGE_IS_5717_PLUS(sc)) - sc->bge_phy_flags |= BGE_PHY_WIRESPEED; + if (sc->bge_asicrev == BGE_ASICREV_BCM5700 || + (sc->bge_asicrev == BGE_ASICREV_BCM5705 && + (sc->bge_chipid != BGE_CHIPID_BCM5705_A0 && + sc->bge_chipid != BGE_CHIPID_BCM5705_A1)) || + sc->bge_asicrev == BGE_ASICREV_BCM5906) + sc->bge_phy_flags |= BGE_PHY_NO_WIRESPEED; if (bge_has_eaddr(sc)) sc->bge_flags |= BGE_FLAG_EADDR; /* Save chipset family. */ switch (sc->bge_asicrev) { case BGE_ASICREV_BCM5717: + sc->bge_flags |= BGE_FLAG_SHORT_DMA_BUG; + case BGE_ASICREV_BCM57765: sc->bge_flags |= BGE_FLAG_5717_PLUS | BGE_FLAG_5755_PLUS | BGE_FLAG_575X_PLUS | BGE_FLAG_5705_PLUS | BGE_FLAG_JUMBO | - BGE_FLAG_SHORT_DMA_BUG | BGE_FLAG_JUMBO_FRAME; + BGE_FLAG_JUMBO_FRAME; break; case BGE_ASICREV_BCM5755: case BGE_ASICREV_BCM5761: case BGE_ASICREV_BCM5784: case BGE_ASICREV_BCM5785: case BGE_ASICREV_BCM5787: case BGE_ASICREV_BCM57780: sc->bge_flags |= BGE_FLAG_5755_PLUS | BGE_FLAG_575X_PLUS | BGE_FLAG_5705_PLUS; break; case BGE_ASICREV_BCM5700: case BGE_ASICREV_BCM5701: case BGE_ASICREV_BCM5703: case BGE_ASICREV_BCM5704: sc->bge_flags |= BGE_FLAG_5700_FAMILY | BGE_FLAG_JUMBO; break; case BGE_ASICREV_BCM5714_A0: case BGE_ASICREV_BCM5780: case BGE_ASICREV_BCM5714: sc->bge_flags |= BGE_FLAG_5714_FAMILY | BGE_FLAG_JUMBO_STD; /* FALLTHROUGH */ case BGE_ASICREV_BCM5750: case BGE_ASICREV_BCM5752: case BGE_ASICREV_BCM5906: sc->bge_flags |= BGE_FLAG_575X_PLUS; if (sc->bge_asicrev == BGE_ASICREV_BCM5906) sc->bge_flags |= BGE_FLAG_SHORT_DMA_BUG; /* FALLTHROUGH */ case BGE_ASICREV_BCM5705: sc->bge_flags |= BGE_FLAG_5705_PLUS; break; } /* Set various PHY bug flags. */ if (sc->bge_chipid == BGE_CHIPID_BCM5701_A0 || sc->bge_chipid == BGE_CHIPID_BCM5701_B0) sc->bge_phy_flags |= BGE_PHY_CRC_BUG; if (sc->bge_chiprev == BGE_CHIPREV_5703_AX || sc->bge_chiprev == BGE_CHIPREV_5704_AX) sc->bge_phy_flags |= BGE_PHY_ADC_BUG; if (sc->bge_chipid == BGE_CHIPID_BCM5704_A0) sc->bge_phy_flags |= BGE_PHY_5704_A0_BUG; if (pci_get_subvendor(dev) == DELL_VENDORID) sc->bge_phy_flags |= BGE_PHY_NO_3LED; if ((BGE_IS_5705_PLUS(sc)) && sc->bge_asicrev != BGE_ASICREV_BCM5906 && sc->bge_asicrev != BGE_ASICREV_BCM5717 && sc->bge_asicrev != BGE_ASICREV_BCM5785 && + sc->bge_asicrev != BGE_ASICREV_BCM57765 && sc->bge_asicrev != BGE_ASICREV_BCM57780) { if (sc->bge_asicrev == BGE_ASICREV_BCM5755 || sc->bge_asicrev == BGE_ASICREV_BCM5761 || sc->bge_asicrev == BGE_ASICREV_BCM5784 || sc->bge_asicrev == BGE_ASICREV_BCM5787) { if (pci_get_device(dev) != BCOM_DEVICEID_BCM5722 && pci_get_device(dev) != BCOM_DEVICEID_BCM5756) sc->bge_phy_flags |= BGE_PHY_JITTER_BUG; if (pci_get_device(dev) == BCOM_DEVICEID_BCM5755M) sc->bge_phy_flags |= BGE_PHY_ADJUST_TRIM; } else sc->bge_phy_flags |= BGE_PHY_BER_BUG; } /* Identify the chips that use an CPMU. */ if (BGE_IS_5717_PLUS(sc) || sc->bge_asicrev == BGE_ASICREV_BCM5784 || sc->bge_asicrev == BGE_ASICREV_BCM5761 || sc->bge_asicrev == BGE_ASICREV_BCM5785 || sc->bge_asicrev == BGE_ASICREV_BCM57780) sc->bge_flags |= BGE_FLAG_CPMU_PRESENT; if ((sc->bge_flags & BGE_FLAG_CPMU_PRESENT) != 0) sc->bge_mi_mode = BGE_MIMODE_500KHZ_CONST; else sc->bge_mi_mode = BGE_MIMODE_BASE; /* Enable auto polling for BCM570[0-5]. */ if (BGE_IS_5700_FAMILY(sc) || sc->bge_asicrev == BGE_ASICREV_BCM5705) sc->bge_mi_mode |= BGE_MIMODE_AUTOPOLL; /* * All controllers that are not 5755 or higher have 4GB * boundary DMA bug. * Whenever an address crosses a multiple of the 4GB boundary * (including 4GB, 8Gb, 12Gb, etc.) and makes the transition * from 0xX_FFFF_FFFF to 0x(X+1)_0000_0000 an internal DMA * state machine will lockup and cause the device to hang. */ if (BGE_IS_5755_PLUS(sc) == 0) sc->bge_flags |= BGE_FLAG_4G_BNDRY_BUG; misccfg = CSR_READ_4(sc, BGE_MISC_CFG) & BGE_MISCCFG_BOARD_ID; if (sc->bge_asicrev == BGE_ASICREV_BCM5705) { if (misccfg == BGE_MISCCFG_BOARD_ID_5788 || misccfg == BGE_MISCCFG_BOARD_ID_5788M) sc->bge_flags |= BGE_FLAG_5788; } capmask = BMSR_DEFCAPMASK; if ((sc->bge_asicrev == BGE_ASICREV_BCM5703 && (misccfg == 0x4000 || misccfg == 0x8000)) || (sc->bge_asicrev == BGE_ASICREV_BCM5705 && pci_get_vendor(dev) == BCOM_VENDORID && (pci_get_device(dev) == BCOM_DEVICEID_BCM5901 || pci_get_device(dev) == BCOM_DEVICEID_BCM5901A2 || pci_get_device(dev) == BCOM_DEVICEID_BCM5705F)) || (pci_get_vendor(dev) == BCOM_VENDORID && (pci_get_device(dev) == BCOM_DEVICEID_BCM5751F || pci_get_device(dev) == BCOM_DEVICEID_BCM5753F || pci_get_device(dev) == BCOM_DEVICEID_BCM5787F)) || pci_get_device(dev) == BCOM_DEVICEID_BCM57790 || sc->bge_asicrev == BGE_ASICREV_BCM5906) { /* These chips are 10/100 only. */ capmask &= ~BMSR_EXTSTAT; } /* * Some controllers seem to require a special firmware to use * TSO. But the firmware is not available to FreeBSD and Linux * claims that the TSO performed by the firmware is slower than * hardware based TSO. Moreover the firmware based TSO has one * known bug which can't handle TSO if ethernet header + IP/TCP * header is greater than 80 bytes. The workaround for the TSO * bug exist but it seems it's too expensive than not using * TSO at all. Some hardwares also have the TSO bug so limit * the TSO to the controllers that are not affected TSO issues * (e.g. 5755 or higher). */ if (BGE_IS_5717_PLUS(sc)) { /* BCM5717 requires different TSO configuration. */ sc->bge_flags |= BGE_FLAG_TSO3; } else if (BGE_IS_5755_PLUS(sc)) { /* * BCM5754 and BCM5787 shares the same ASIC id so * explicit device id check is required. * Due to unknown reason TSO does not work on BCM5755M. */ if (pci_get_device(dev) != BCOM_DEVICEID_BCM5754 && pci_get_device(dev) != BCOM_DEVICEID_BCM5754M && pci_get_device(dev) != BCOM_DEVICEID_BCM5755M) sc->bge_flags |= BGE_FLAG_TSO; } /* * Check if this is a PCI-X or PCI Express device. */ if (pci_find_cap(dev, PCIY_EXPRESS, ®) == 0) { /* * Found a PCI Express capabilities register, this * must be a PCI Express device. */ sc->bge_flags |= BGE_FLAG_PCIE; sc->bge_expcap = reg; if (pci_get_max_read_req(dev) != 4096) pci_set_max_read_req(dev, 4096); } else { /* * Check if the device is in PCI-X Mode. * (This bit is not valid on PCI Express controllers.) */ if (pci_find_cap(dev, PCIY_PCIX, ®) == 0) sc->bge_pcixcap = reg; if ((pci_read_config(dev, BGE_PCI_PCISTATE, 4) & BGE_PCISTATE_PCI_BUSMODE) == 0) sc->bge_flags |= BGE_FLAG_PCIX; } /* * The 40bit DMA bug applies to the 5714/5715 controllers and is * not actually a MAC controller bug but an issue with the embedded * PCIe to PCI-X bridge in the device. Use 40bit DMA workaround. */ if (BGE_IS_5714_FAMILY(sc) && (sc->bge_flags & BGE_FLAG_PCIX)) sc->bge_flags |= BGE_FLAG_40BIT_BUG; /* * Allocate the interrupt, using MSI if possible. These devices * support 8 MSI messages, but only the first one is used in * normal operation. */ rid = 0; if (pci_find_cap(sc->bge_dev, PCIY_MSI, ®) == 0) { sc->bge_msicap = reg; if (bge_can_use_msi(sc)) { msicount = pci_msi_count(dev); if (msicount > 1) msicount = 1; } else msicount = 0; if (msicount == 1 && pci_alloc_msi(dev, &msicount) == 0) { rid = 1; sc->bge_flags |= BGE_FLAG_MSI; } } /* * All controllers except BCM5700 supports tagged status but * we use tagged status only for MSI case on BCM5717. Otherwise * MSI on BCM5717 does not work. */ #ifndef DEVICE_POLLING if (sc->bge_flags & BGE_FLAG_MSI && BGE_IS_5717_PLUS(sc)) sc->bge_flags |= BGE_FLAG_TAGGED_STATUS; #endif sc->bge_irq = bus_alloc_resource_any(dev, SYS_RES_IRQ, &rid, RF_SHAREABLE | RF_ACTIVE); if (sc->bge_irq == NULL) { device_printf(sc->bge_dev, "couldn't map interrupt\n"); error = ENXIO; goto fail; } device_printf(dev, "CHIP ID 0x%08x; ASIC REV 0x%02x; CHIP REV 0x%02x; %s\n", sc->bge_chipid, sc->bge_asicrev, sc->bge_chiprev, (sc->bge_flags & BGE_FLAG_PCIX) ? "PCI-X" : ((sc->bge_flags & BGE_FLAG_PCIE) ? "PCI-E" : "PCI")); BGE_LOCK_INIT(sc, device_get_nameunit(dev)); /* Try to reset the chip. */ if (bge_reset(sc)) { device_printf(sc->bge_dev, "chip reset failed\n"); error = ENXIO; goto fail; } sc->bge_asf_mode = 0; if (bge_allow_asf && (bge_readmem_ind(sc, BGE_SOFTWARE_GENCOMM_SIG) == BGE_MAGIC_NUMBER)) { if (bge_readmem_ind(sc, BGE_SOFTWARE_GENCOMM_NICCFG) & BGE_HWCFG_ASF) { sc->bge_asf_mode |= ASF_ENABLE; sc->bge_asf_mode |= ASF_STACKUP; if (BGE_IS_575X_PLUS(sc)) sc->bge_asf_mode |= ASF_NEW_HANDSHAKE; } } /* Try to reset the chip again the nice way. */ bge_stop_fw(sc); bge_sig_pre_reset(sc, BGE_RESET_STOP); if (bge_reset(sc)) { device_printf(sc->bge_dev, "chip reset failed\n"); error = ENXIO; goto fail; } bge_sig_legacy(sc, BGE_RESET_STOP); bge_sig_post_reset(sc, BGE_RESET_STOP); if (bge_chipinit(sc)) { device_printf(sc->bge_dev, "chip initialization failed\n"); error = ENXIO; goto fail; } error = bge_get_eaddr(sc, eaddr); if (error) { device_printf(sc->bge_dev, "failed to read station address\n"); error = ENXIO; goto fail; } /* 5705 limits RX return ring to 512 entries. */ if (BGE_IS_5717_PLUS(sc)) sc->bge_return_ring_cnt = BGE_RETURN_RING_CNT; else if (BGE_IS_5705_PLUS(sc)) sc->bge_return_ring_cnt = BGE_RETURN_RING_CNT_5705; else sc->bge_return_ring_cnt = BGE_RETURN_RING_CNT; if (bge_dma_alloc(sc)) { device_printf(sc->bge_dev, "failed to allocate DMA resources\n"); error = ENXIO; goto fail; } bge_add_sysctls(sc); /* Set default tuneable values. */ sc->bge_stat_ticks = BGE_TICKS_PER_SEC; sc->bge_rx_coal_ticks = 150; sc->bge_tx_coal_ticks = 150; sc->bge_rx_max_coal_bds = 10; sc->bge_tx_max_coal_bds = 10; /* Initialize checksum features to use. */ sc->bge_csum_features = BGE_CSUM_FEATURES; if (sc->bge_forced_udpcsum != 0) sc->bge_csum_features |= CSUM_UDP; /* Set up ifnet structure */ ifp = sc->bge_ifp = if_alloc(IFT_ETHER); if (ifp == NULL) { device_printf(sc->bge_dev, "failed to if_alloc()\n"); error = ENXIO; goto fail; } ifp->if_softc = sc; if_initname(ifp, device_get_name(dev), device_get_unit(dev)); ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST; ifp->if_ioctl = bge_ioctl; ifp->if_start = bge_start; ifp->if_init = bge_init; ifp->if_snd.ifq_drv_maxlen = BGE_TX_RING_CNT - 1; IFQ_SET_MAXLEN(&ifp->if_snd, ifp->if_snd.ifq_drv_maxlen); IFQ_SET_READY(&ifp->if_snd); ifp->if_hwassist = sc->bge_csum_features; ifp->if_capabilities = IFCAP_HWCSUM | IFCAP_VLAN_HWTAGGING | IFCAP_VLAN_MTU; if ((sc->bge_flags & (BGE_FLAG_TSO | BGE_FLAG_TSO3)) != 0) { ifp->if_hwassist |= CSUM_TSO; ifp->if_capabilities |= IFCAP_TSO4 | IFCAP_VLAN_HWTSO; } #ifdef IFCAP_VLAN_HWCSUM ifp->if_capabilities |= IFCAP_VLAN_HWCSUM; #endif ifp->if_capenable = ifp->if_capabilities; #ifdef DEVICE_POLLING ifp->if_capabilities |= IFCAP_POLLING; #endif /* * 5700 B0 chips do not support checksumming correctly due * to hardware bugs. */ if (sc->bge_chipid == BGE_CHIPID_BCM5700_B0) { ifp->if_capabilities &= ~IFCAP_HWCSUM; ifp->if_capenable &= ~IFCAP_HWCSUM; ifp->if_hwassist = 0; } /* * Figure out what sort of media we have by checking the * hardware config word in the first 32k of NIC internal memory, * or fall back to examining the EEPROM if necessary. * Note: on some BCM5700 cards, this value appears to be unset. * If that's the case, we have to rely on identifying the NIC * by its PCI subsystem ID, as we do below for the SysKonnect * SK-9D41. */ if (bge_readmem_ind(sc, BGE_SOFTWARE_GENCOMM_SIG) == BGE_MAGIC_NUMBER) hwcfg = bge_readmem_ind(sc, BGE_SOFTWARE_GENCOMM_NICCFG); else if ((sc->bge_flags & BGE_FLAG_EADDR) && (sc->bge_asicrev != BGE_ASICREV_BCM5906)) { if (bge_read_eeprom(sc, (caddr_t)&hwcfg, BGE_EE_HWCFG_OFFSET, sizeof(hwcfg))) { device_printf(sc->bge_dev, "failed to read EEPROM\n"); error = ENXIO; goto fail; } hwcfg = ntohl(hwcfg); } /* The SysKonnect SK-9D41 is a 1000baseSX card. */ if ((pci_read_config(dev, BGE_PCI_SUBSYS, 4) >> 16) == SK_SUBSYSID_9D41 || (hwcfg & BGE_HWCFG_MEDIA) == BGE_MEDIA_FIBER) { if (BGE_IS_5714_FAMILY(sc)) sc->bge_flags |= BGE_FLAG_MII_SERDES; else sc->bge_flags |= BGE_FLAG_TBI; } if (sc->bge_flags & BGE_FLAG_TBI) { ifmedia_init(&sc->bge_ifmedia, IFM_IMASK, bge_ifmedia_upd, bge_ifmedia_sts); ifmedia_add(&sc->bge_ifmedia, IFM_ETHER | IFM_1000_SX, 0, NULL); ifmedia_add(&sc->bge_ifmedia, IFM_ETHER | IFM_1000_SX | IFM_FDX, 0, NULL); ifmedia_add(&sc->bge_ifmedia, IFM_ETHER | IFM_AUTO, 0, NULL); ifmedia_set(&sc->bge_ifmedia, IFM_ETHER | IFM_AUTO); sc->bge_ifmedia.ifm_media = sc->bge_ifmedia.ifm_cur->ifm_media; } else { /* * Do transceiver setup and tell the firmware the * driver is down so we can try to get access the * probe if ASF is running. Retry a couple of times * if we get a conflict with the ASF firmware accessing * the PHY. */ trys = 0; BGE_CLRBIT(sc, BGE_MODE_CTL, BGE_MODECTL_STACKUP); again: bge_asf_driver_up(sc); error = mii_attach(dev, &sc->bge_miibus, ifp, bge_ifmedia_upd, bge_ifmedia_sts, capmask, phy_addr, MII_OFFSET_ANY, MIIF_DOPAUSE); if (error != 0) { if (trys++ < 4) { device_printf(sc->bge_dev, "Try again\n"); bge_miibus_writereg(sc->bge_dev, 1, MII_BMCR, BMCR_RESET); goto again; } device_printf(sc->bge_dev, "attaching PHYs failed\n"); goto fail; } /* * Now tell the firmware we are going up after probing the PHY */ if (sc->bge_asf_mode & ASF_STACKUP) BGE_SETBIT(sc, BGE_MODE_CTL, BGE_MODECTL_STACKUP); } /* * When using the BCM5701 in PCI-X mode, data corruption has * been observed in the first few bytes of some received packets. * Aligning the packet buffer in memory eliminates the corruption. * Unfortunately, this misaligns the packet payloads. On platforms * which do not support unaligned accesses, we will realign the * payloads by copying the received packets. */ if (sc->bge_asicrev == BGE_ASICREV_BCM5701 && sc->bge_flags & BGE_FLAG_PCIX) sc->bge_flags |= BGE_FLAG_RX_ALIGNBUG; /* * Call MI attach routine. */ ether_ifattach(ifp, eaddr); callout_init_mtx(&sc->bge_stat_ch, &sc->bge_mtx, 0); /* Tell upper layer we support long frames. */ ifp->if_data.ifi_hdrlen = sizeof(struct ether_vlan_header); /* * Hookup IRQ last. */ if (BGE_IS_5755_PLUS(sc) && sc->bge_flags & BGE_FLAG_MSI) { /* Take advantage of single-shot MSI. */ CSR_WRITE_4(sc, BGE_MSI_MODE, CSR_READ_4(sc, BGE_MSI_MODE) & ~BGE_MSIMODE_ONE_SHOT_DISABLE); sc->bge_tq = taskqueue_create_fast("bge_taskq", M_WAITOK, taskqueue_thread_enqueue, &sc->bge_tq); if (sc->bge_tq == NULL) { device_printf(dev, "could not create taskqueue.\n"); ether_ifdetach(ifp); error = ENXIO; goto fail; } taskqueue_start_threads(&sc->bge_tq, 1, PI_NET, "%s taskq", device_get_nameunit(sc->bge_dev)); error = bus_setup_intr(dev, sc->bge_irq, INTR_TYPE_NET | INTR_MPSAFE, bge_msi_intr, NULL, sc, &sc->bge_intrhand); if (error) ether_ifdetach(ifp); } else error = bus_setup_intr(dev, sc->bge_irq, INTR_TYPE_NET | INTR_MPSAFE, NULL, bge_intr, sc, &sc->bge_intrhand); if (error) { bge_detach(dev); device_printf(sc->bge_dev, "couldn't set up irq\n"); } return (0); fail: bge_release_resources(sc); return (error); } static int bge_detach(device_t dev) { struct bge_softc *sc; struct ifnet *ifp; sc = device_get_softc(dev); ifp = sc->bge_ifp; #ifdef DEVICE_POLLING if (ifp->if_capenable & IFCAP_POLLING) ether_poll_deregister(ifp); #endif BGE_LOCK(sc); bge_stop(sc); bge_reset(sc); BGE_UNLOCK(sc); callout_drain(&sc->bge_stat_ch); if (sc->bge_tq) taskqueue_drain(sc->bge_tq, &sc->bge_intr_task); ether_ifdetach(ifp); if (sc->bge_flags & BGE_FLAG_TBI) { ifmedia_removeall(&sc->bge_ifmedia); } else { bus_generic_detach(dev); device_delete_child(dev, sc->bge_miibus); } bge_release_resources(sc); return (0); } static void bge_release_resources(struct bge_softc *sc) { device_t dev; dev = sc->bge_dev; if (sc->bge_tq != NULL) taskqueue_free(sc->bge_tq); if (sc->bge_intrhand != NULL) bus_teardown_intr(dev, sc->bge_irq, sc->bge_intrhand); if (sc->bge_irq != NULL) bus_release_resource(dev, SYS_RES_IRQ, sc->bge_flags & BGE_FLAG_MSI ? 1 : 0, sc->bge_irq); if (sc->bge_flags & BGE_FLAG_MSI) pci_release_msi(dev); if (sc->bge_res != NULL) bus_release_resource(dev, SYS_RES_MEMORY, PCIR_BAR(0), sc->bge_res); if (sc->bge_ifp != NULL) if_free(sc->bge_ifp); bge_dma_free(sc); if (mtx_initialized(&sc->bge_mtx)) /* XXX */ BGE_LOCK_DESTROY(sc); } static int bge_reset(struct bge_softc *sc) { device_t dev; uint32_t cachesize, command, pcistate, reset, val; void (*write_op)(struct bge_softc *, int, int); uint16_t devctl; int i; dev = sc->bge_dev; if (BGE_IS_575X_PLUS(sc) && !BGE_IS_5714_FAMILY(sc) && (sc->bge_asicrev != BGE_ASICREV_BCM5906)) { if (sc->bge_flags & BGE_FLAG_PCIE) write_op = bge_writemem_direct; else write_op = bge_writemem_ind; } else write_op = bge_writereg_ind; /* Save some important PCI state. */ cachesize = pci_read_config(dev, BGE_PCI_CACHESZ, 4); command = pci_read_config(dev, BGE_PCI_CMD, 4); pcistate = pci_read_config(dev, BGE_PCI_PCISTATE, 4); pci_write_config(dev, BGE_PCI_MISC_CTL, BGE_PCIMISCCTL_INDIRECT_ACCESS | BGE_PCIMISCCTL_MASK_PCI_INTR | BGE_HIF_SWAP_OPTIONS | BGE_PCIMISCCTL_PCISTATE_RW, 4); /* Disable fastboot on controllers that support it. */ if (sc->bge_asicrev == BGE_ASICREV_BCM5752 || BGE_IS_5755_PLUS(sc)) { if (bootverbose) device_printf(dev, "Disabling fastboot\n"); CSR_WRITE_4(sc, BGE_FASTBOOT_PC, 0x0); } /* * Write the magic number to SRAM at offset 0xB50. * When firmware finishes its initialization it will * write ~BGE_MAGIC_NUMBER to the same location. */ bge_writemem_ind(sc, BGE_SOFTWARE_GENCOMM, BGE_MAGIC_NUMBER); reset = BGE_MISCCFG_RESET_CORE_CLOCKS | BGE_32BITTIME_66MHZ; /* XXX: Broadcom Linux driver. */ if (sc->bge_flags & BGE_FLAG_PCIE) { if (CSR_READ_4(sc, 0x7E2C) == 0x60) /* PCIE 1.0 */ CSR_WRITE_4(sc, 0x7E2C, 0x20); if (sc->bge_chipid != BGE_CHIPID_BCM5750_A0) { /* Prevent PCIE link training during global reset */ CSR_WRITE_4(sc, BGE_MISC_CFG, 1 << 29); reset |= 1 << 29; } } /* * Set GPHY Power Down Override to leave GPHY * powered up in D0 uninitialized. */ if (BGE_IS_5705_PLUS(sc)) reset |= BGE_MISCCFG_GPHY_PD_OVERRIDE; /* Issue global reset */ write_op(sc, BGE_MISC_CFG, reset); if (sc->bge_asicrev == BGE_ASICREV_BCM5906) { val = CSR_READ_4(sc, BGE_VCPU_STATUS); CSR_WRITE_4(sc, BGE_VCPU_STATUS, val | BGE_VCPU_STATUS_DRV_RESET); val = CSR_READ_4(sc, BGE_VCPU_EXT_CTRL); CSR_WRITE_4(sc, BGE_VCPU_EXT_CTRL, val & ~BGE_VCPU_EXT_CTRL_HALT_CPU); } DELAY(1000); /* XXX: Broadcom Linux driver. */ if (sc->bge_flags & BGE_FLAG_PCIE) { if (sc->bge_chipid == BGE_CHIPID_BCM5750_A0) { DELAY(500000); /* wait for link training to complete */ val = pci_read_config(dev, 0xC4, 4); pci_write_config(dev, 0xC4, val | (1 << 15), 4); } devctl = pci_read_config(dev, sc->bge_expcap + PCIR_EXPRESS_DEVICE_CTL, 2); /* Clear enable no snoop and disable relaxed ordering. */ devctl &= ~(PCIM_EXP_CTL_RELAXED_ORD_ENABLE | PCIM_EXP_CTL_NOSNOOP_ENABLE); /* Set PCIE max payload size to 128. */ devctl &= ~PCIM_EXP_CTL_MAX_PAYLOAD; pci_write_config(dev, sc->bge_expcap + PCIR_EXPRESS_DEVICE_CTL, devctl, 2); /* Clear error status. */ pci_write_config(dev, sc->bge_expcap + PCIR_EXPRESS_DEVICE_STA, PCIM_EXP_STA_CORRECTABLE_ERROR | PCIM_EXP_STA_NON_FATAL_ERROR | PCIM_EXP_STA_FATAL_ERROR | PCIM_EXP_STA_UNSUPPORTED_REQ, 2); } /* Reset some of the PCI state that got zapped by reset. */ pci_write_config(dev, BGE_PCI_MISC_CTL, BGE_PCIMISCCTL_INDIRECT_ACCESS | BGE_PCIMISCCTL_MASK_PCI_INTR | BGE_HIF_SWAP_OPTIONS | BGE_PCIMISCCTL_PCISTATE_RW, 4); pci_write_config(dev, BGE_PCI_CACHESZ, cachesize, 4); pci_write_config(dev, BGE_PCI_CMD, command, 4); write_op(sc, BGE_MISC_CFG, BGE_32BITTIME_66MHZ); /* * Disable PCI-X relaxed ordering to ensure status block update * comes first then packet buffer DMA. Otherwise driver may * read stale status block. */ if (sc->bge_flags & BGE_FLAG_PCIX) { devctl = pci_read_config(dev, sc->bge_pcixcap + PCIXR_COMMAND, 2); devctl &= ~PCIXM_COMMAND_ERO; if (sc->bge_asicrev == BGE_ASICREV_BCM5703) { devctl &= ~PCIXM_COMMAND_MAX_READ; devctl |= PCIXM_COMMAND_MAX_READ_2048; } else if (sc->bge_asicrev == BGE_ASICREV_BCM5704) { devctl &= ~(PCIXM_COMMAND_MAX_SPLITS | PCIXM_COMMAND_MAX_READ); devctl |= PCIXM_COMMAND_MAX_READ_2048; } pci_write_config(dev, sc->bge_pcixcap + PCIXR_COMMAND, devctl, 2); } /* Re-enable MSI, if necessary, and enable the memory arbiter. */ if (BGE_IS_5714_FAMILY(sc)) { /* This chip disables MSI on reset. */ if (sc->bge_flags & BGE_FLAG_MSI) { val = pci_read_config(dev, sc->bge_msicap + PCIR_MSI_CTRL, 2); pci_write_config(dev, sc->bge_msicap + PCIR_MSI_CTRL, val | PCIM_MSICTRL_MSI_ENABLE, 2); val = CSR_READ_4(sc, BGE_MSI_MODE); CSR_WRITE_4(sc, BGE_MSI_MODE, val | BGE_MSIMODE_ENABLE); } val = CSR_READ_4(sc, BGE_MARB_MODE); CSR_WRITE_4(sc, BGE_MARB_MODE, BGE_MARBMODE_ENABLE | val); } else CSR_WRITE_4(sc, BGE_MARB_MODE, BGE_MARBMODE_ENABLE); if (sc->bge_asicrev == BGE_ASICREV_BCM5906) { for (i = 0; i < BGE_TIMEOUT; i++) { val = CSR_READ_4(sc, BGE_VCPU_STATUS); if (val & BGE_VCPU_STATUS_INIT_DONE) break; DELAY(100); } if (i == BGE_TIMEOUT) { device_printf(dev, "reset timed out\n"); return (1); } } else { /* * Poll until we see the 1's complement of the magic number. * This indicates that the firmware initialization is complete. * We expect this to fail if no chip containing the Ethernet * address is fitted though. */ for (i = 0; i < BGE_TIMEOUT; i++) { DELAY(10); val = bge_readmem_ind(sc, BGE_SOFTWARE_GENCOMM); if (val == ~BGE_MAGIC_NUMBER) break; } if ((sc->bge_flags & BGE_FLAG_EADDR) && i == BGE_TIMEOUT) device_printf(dev, "firmware handshake timed out, found 0x%08x\n", val); + /* BCM57765 A0 needs additional time before accessing. */ + if (sc->bge_chipid == BGE_CHIPID_BCM57765_A0) + DELAY(10 * 1000); /* XXX */ } /* * XXX Wait for the value of the PCISTATE register to * return to its original pre-reset state. This is a * fairly good indicator of reset completion. If we don't * wait for the reset to fully complete, trying to read * from the device's non-PCI registers may yield garbage * results. */ for (i = 0; i < BGE_TIMEOUT; i++) { if (pci_read_config(dev, BGE_PCI_PCISTATE, 4) == pcistate) break; DELAY(10); } /* Fix up byte swapping. */ CSR_WRITE_4(sc, BGE_MODE_CTL, BGE_DMA_SWAP_OPTIONS | BGE_MODECTL_BYTESWAP_DATA); /* Tell the ASF firmware we are up */ if (sc->bge_asf_mode & ASF_STACKUP) BGE_SETBIT(sc, BGE_MODE_CTL, BGE_MODECTL_STACKUP); CSR_WRITE_4(sc, BGE_MAC_MODE, 0); /* * The 5704 in TBI mode apparently needs some special * adjustment to insure the SERDES drive level is set * to 1.2V. */ if (sc->bge_asicrev == BGE_ASICREV_BCM5704 && sc->bge_flags & BGE_FLAG_TBI) { val = CSR_READ_4(sc, BGE_SERDES_CFG); val = (val & ~0xFFF) | 0x880; CSR_WRITE_4(sc, BGE_SERDES_CFG, val); } /* XXX: Broadcom Linux driver. */ if (sc->bge_flags & BGE_FLAG_PCIE && - sc->bge_asicrev != BGE_ASICREV_BCM5717 && + !BGE_IS_5717_PLUS(sc) && sc->bge_chipid != BGE_CHIPID_BCM5750_A0 && sc->bge_asicrev != BGE_ASICREV_BCM5785) { /* Enable Data FIFO protection. */ val = CSR_READ_4(sc, 0x7C00); CSR_WRITE_4(sc, 0x7C00, val | (1 << 25)); } DELAY(10000); return (0); } static __inline void bge_rxreuse_std(struct bge_softc *sc, int i) { struct bge_rx_bd *r; r = &sc->bge_ldata.bge_rx_std_ring[sc->bge_std]; r->bge_flags = BGE_RXBDFLAG_END; r->bge_len = sc->bge_cdata.bge_rx_std_seglen[i]; r->bge_idx = i; BGE_INC(sc->bge_std, BGE_STD_RX_RING_CNT); } static __inline void bge_rxreuse_jumbo(struct bge_softc *sc, int i) { struct bge_extrx_bd *r; r = &sc->bge_ldata.bge_rx_jumbo_ring[sc->bge_jumbo]; r->bge_flags = BGE_RXBDFLAG_JUMBO_RING | BGE_RXBDFLAG_END; r->bge_len0 = sc->bge_cdata.bge_rx_jumbo_seglen[i][0]; r->bge_len1 = sc->bge_cdata.bge_rx_jumbo_seglen[i][1]; r->bge_len2 = sc->bge_cdata.bge_rx_jumbo_seglen[i][2]; r->bge_len3 = sc->bge_cdata.bge_rx_jumbo_seglen[i][3]; r->bge_idx = i; BGE_INC(sc->bge_jumbo, BGE_JUMBO_RX_RING_CNT); } /* * Frame reception handling. This is called if there's a frame * on the receive return list. * * Note: we have to be able to handle two possibilities here: * 1) the frame is from the jumbo receive ring * 2) the frame is from the standard receive ring */ static int bge_rxeof(struct bge_softc *sc, uint16_t rx_prod, int holdlck) { struct ifnet *ifp; int rx_npkts = 0, stdcnt = 0, jumbocnt = 0; uint16_t rx_cons; rx_cons = sc->bge_rx_saved_considx; /* Nothing to do. */ if (rx_cons == rx_prod) return (rx_npkts); ifp = sc->bge_ifp; bus_dmamap_sync(sc->bge_cdata.bge_rx_return_ring_tag, sc->bge_cdata.bge_rx_return_ring_map, BUS_DMASYNC_POSTREAD); bus_dmamap_sync(sc->bge_cdata.bge_rx_std_ring_tag, sc->bge_cdata.bge_rx_std_ring_map, BUS_DMASYNC_POSTWRITE); if (BGE_IS_JUMBO_CAPABLE(sc) && ifp->if_mtu + ETHER_HDR_LEN + ETHER_CRC_LEN + ETHER_VLAN_ENCAP_LEN > (MCLBYTES - ETHER_ALIGN)) bus_dmamap_sync(sc->bge_cdata.bge_rx_jumbo_ring_tag, sc->bge_cdata.bge_rx_jumbo_ring_map, BUS_DMASYNC_POSTWRITE); while (rx_cons != rx_prod) { struct bge_rx_bd *cur_rx; uint32_t rxidx; struct mbuf *m = NULL; uint16_t vlan_tag = 0; int have_tag = 0; #ifdef DEVICE_POLLING if (ifp->if_capenable & IFCAP_POLLING) { if (sc->rxcycles <= 0) break; sc->rxcycles--; } #endif cur_rx = &sc->bge_ldata.bge_rx_return_ring[rx_cons]; rxidx = cur_rx->bge_idx; BGE_INC(rx_cons, sc->bge_return_ring_cnt); if (ifp->if_capenable & IFCAP_VLAN_HWTAGGING && cur_rx->bge_flags & BGE_RXBDFLAG_VLAN_TAG) { have_tag = 1; vlan_tag = cur_rx->bge_vlan_tag; } if (cur_rx->bge_flags & BGE_RXBDFLAG_JUMBO_RING) { jumbocnt++; m = sc->bge_cdata.bge_rx_jumbo_chain[rxidx]; if (cur_rx->bge_flags & BGE_RXBDFLAG_ERROR) { bge_rxreuse_jumbo(sc, rxidx); continue; } if (bge_newbuf_jumbo(sc, rxidx) != 0) { bge_rxreuse_jumbo(sc, rxidx); ifp->if_iqdrops++; continue; } BGE_INC(sc->bge_jumbo, BGE_JUMBO_RX_RING_CNT); } else { stdcnt++; m = sc->bge_cdata.bge_rx_std_chain[rxidx]; if (cur_rx->bge_flags & BGE_RXBDFLAG_ERROR) { bge_rxreuse_std(sc, rxidx); continue; } if (bge_newbuf_std(sc, rxidx) != 0) { bge_rxreuse_std(sc, rxidx); ifp->if_iqdrops++; continue; } BGE_INC(sc->bge_std, BGE_STD_RX_RING_CNT); } ifp->if_ipackets++; #ifndef __NO_STRICT_ALIGNMENT /* * For architectures with strict alignment we must make sure * the payload is aligned. */ if (sc->bge_flags & BGE_FLAG_RX_ALIGNBUG) { bcopy(m->m_data, m->m_data + ETHER_ALIGN, cur_rx->bge_len); m->m_data += ETHER_ALIGN; } #endif m->m_pkthdr.len = m->m_len = cur_rx->bge_len - ETHER_CRC_LEN; m->m_pkthdr.rcvif = ifp; if (ifp->if_capenable & IFCAP_RXCSUM) bge_rxcsum(sc, cur_rx, m); /* * If we received a packet with a vlan tag, * attach that information to the packet. */ if (have_tag) { m->m_pkthdr.ether_vtag = vlan_tag; m->m_flags |= M_VLANTAG; } if (holdlck != 0) { BGE_UNLOCK(sc); (*ifp->if_input)(ifp, m); BGE_LOCK(sc); } else (*ifp->if_input)(ifp, m); rx_npkts++; if (!(ifp->if_drv_flags & IFF_DRV_RUNNING)) return (rx_npkts); } bus_dmamap_sync(sc->bge_cdata.bge_rx_return_ring_tag, sc->bge_cdata.bge_rx_return_ring_map, BUS_DMASYNC_PREREAD); if (stdcnt > 0) bus_dmamap_sync(sc->bge_cdata.bge_rx_std_ring_tag, sc->bge_cdata.bge_rx_std_ring_map, BUS_DMASYNC_PREWRITE); if (jumbocnt > 0) bus_dmamap_sync(sc->bge_cdata.bge_rx_jumbo_ring_tag, sc->bge_cdata.bge_rx_jumbo_ring_map, BUS_DMASYNC_PREWRITE); sc->bge_rx_saved_considx = rx_cons; bge_writembx(sc, BGE_MBX_RX_CONS0_LO, sc->bge_rx_saved_considx); if (stdcnt) bge_writembx(sc, BGE_MBX_RX_STD_PROD_LO, (sc->bge_std + BGE_STD_RX_RING_CNT - 1) % BGE_STD_RX_RING_CNT); if (jumbocnt) bge_writembx(sc, BGE_MBX_RX_JUMBO_PROD_LO, (sc->bge_jumbo + BGE_JUMBO_RX_RING_CNT - 1) % BGE_JUMBO_RX_RING_CNT); #ifdef notyet /* * This register wraps very quickly under heavy packet drops. * If you need correct statistics, you can enable this check. */ if (BGE_IS_5705_PLUS(sc)) ifp->if_ierrors += CSR_READ_4(sc, BGE_RXLP_LOCSTAT_IFIN_DROPS); #endif return (rx_npkts); } static void bge_rxcsum(struct bge_softc *sc, struct bge_rx_bd *cur_rx, struct mbuf *m) { if (BGE_IS_5717_PLUS(sc)) { if ((cur_rx->bge_flags & BGE_RXBDFLAG_IPV6) == 0) { if (cur_rx->bge_flags & BGE_RXBDFLAG_IP_CSUM) { m->m_pkthdr.csum_flags |= CSUM_IP_CHECKED; if ((cur_rx->bge_error_flag & BGE_RXERRFLAG_IP_CSUM_NOK) == 0) m->m_pkthdr.csum_flags |= CSUM_IP_VALID; } if (cur_rx->bge_flags & BGE_RXBDFLAG_TCP_UDP_CSUM) { m->m_pkthdr.csum_data = cur_rx->bge_tcp_udp_csum; m->m_pkthdr.csum_flags |= CSUM_DATA_VALID | CSUM_PSEUDO_HDR; } } } else { if (cur_rx->bge_flags & BGE_RXBDFLAG_IP_CSUM) { m->m_pkthdr.csum_flags |= CSUM_IP_CHECKED; if ((cur_rx->bge_ip_csum ^ 0xFFFF) == 0) m->m_pkthdr.csum_flags |= CSUM_IP_VALID; } if (cur_rx->bge_flags & BGE_RXBDFLAG_TCP_UDP_CSUM && m->m_pkthdr.len >= ETHER_MIN_NOPAD) { m->m_pkthdr.csum_data = cur_rx->bge_tcp_udp_csum; m->m_pkthdr.csum_flags |= CSUM_DATA_VALID | CSUM_PSEUDO_HDR; } } } static void bge_txeof(struct bge_softc *sc, uint16_t tx_cons) { struct bge_tx_bd *cur_tx; struct ifnet *ifp; BGE_LOCK_ASSERT(sc); /* Nothing to do. */ if (sc->bge_tx_saved_considx == tx_cons) return; ifp = sc->bge_ifp; bus_dmamap_sync(sc->bge_cdata.bge_tx_ring_tag, sc->bge_cdata.bge_tx_ring_map, BUS_DMASYNC_POSTWRITE); /* * Go through our tx ring and free mbufs for those * frames that have been sent. */ while (sc->bge_tx_saved_considx != tx_cons) { uint32_t idx; idx = sc->bge_tx_saved_considx; cur_tx = &sc->bge_ldata.bge_tx_ring[idx]; if (cur_tx->bge_flags & BGE_TXBDFLAG_END) ifp->if_opackets++; if (sc->bge_cdata.bge_tx_chain[idx] != NULL) { bus_dmamap_sync(sc->bge_cdata.bge_tx_mtag, sc->bge_cdata.bge_tx_dmamap[idx], BUS_DMASYNC_POSTWRITE); bus_dmamap_unload(sc->bge_cdata.bge_tx_mtag, sc->bge_cdata.bge_tx_dmamap[idx]); m_freem(sc->bge_cdata.bge_tx_chain[idx]); sc->bge_cdata.bge_tx_chain[idx] = NULL; } sc->bge_txcnt--; BGE_INC(sc->bge_tx_saved_considx, BGE_TX_RING_CNT); } ifp->if_drv_flags &= ~IFF_DRV_OACTIVE; if (sc->bge_txcnt == 0) sc->bge_timer = 0; } #ifdef DEVICE_POLLING static int bge_poll(struct ifnet *ifp, enum poll_cmd cmd, int count) { struct bge_softc *sc = ifp->if_softc; uint16_t rx_prod, tx_cons; uint32_t statusword; int rx_npkts = 0; BGE_LOCK(sc); if (!(ifp->if_drv_flags & IFF_DRV_RUNNING)) { BGE_UNLOCK(sc); return (rx_npkts); } bus_dmamap_sync(sc->bge_cdata.bge_status_tag, sc->bge_cdata.bge_status_map, BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); rx_prod = sc->bge_ldata.bge_status_block->bge_idx[0].bge_rx_prod_idx; tx_cons = sc->bge_ldata.bge_status_block->bge_idx[0].bge_tx_cons_idx; statusword = sc->bge_ldata.bge_status_block->bge_status; sc->bge_ldata.bge_status_block->bge_status = 0; bus_dmamap_sync(sc->bge_cdata.bge_status_tag, sc->bge_cdata.bge_status_map, BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); /* Note link event. It will be processed by POLL_AND_CHECK_STATUS. */ if (statusword & BGE_STATFLAG_LINKSTATE_CHANGED) sc->bge_link_evt++; if (cmd == POLL_AND_CHECK_STATUS) if ((sc->bge_asicrev == BGE_ASICREV_BCM5700 && sc->bge_chipid != BGE_CHIPID_BCM5700_B2) || sc->bge_link_evt || (sc->bge_flags & BGE_FLAG_TBI)) bge_link_upd(sc); sc->rxcycles = count; rx_npkts = bge_rxeof(sc, rx_prod, 1); if (!(ifp->if_drv_flags & IFF_DRV_RUNNING)) { BGE_UNLOCK(sc); return (rx_npkts); } bge_txeof(sc, tx_cons); if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd)) bge_start_locked(ifp); BGE_UNLOCK(sc); return (rx_npkts); } #endif /* DEVICE_POLLING */ static int bge_msi_intr(void *arg) { struct bge_softc *sc; sc = (struct bge_softc *)arg; /* * This interrupt is not shared and controller already * disabled further interrupt. */ taskqueue_enqueue(sc->bge_tq, &sc->bge_intr_task); return (FILTER_HANDLED); } static void bge_intr_task(void *arg, int pending) { struct bge_softc *sc; struct ifnet *ifp; uint32_t status, status_tag; uint16_t rx_prod, tx_cons; sc = (struct bge_softc *)arg; ifp = sc->bge_ifp; BGE_LOCK(sc); if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { BGE_UNLOCK(sc); return; } /* Get updated status block. */ bus_dmamap_sync(sc->bge_cdata.bge_status_tag, sc->bge_cdata.bge_status_map, BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); /* Save producer/consumer indexess. */ rx_prod = sc->bge_ldata.bge_status_block->bge_idx[0].bge_rx_prod_idx; tx_cons = sc->bge_ldata.bge_status_block->bge_idx[0].bge_tx_cons_idx; status = sc->bge_ldata.bge_status_block->bge_status; status_tag = sc->bge_ldata.bge_status_block->bge_status_tag << 24; sc->bge_ldata.bge_status_block->bge_status = 0; bus_dmamap_sync(sc->bge_cdata.bge_status_tag, sc->bge_cdata.bge_status_map, BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); if ((sc->bge_flags & BGE_FLAG_TAGGED_STATUS) == 0) status_tag = 0; if ((status & BGE_STATFLAG_LINKSTATE_CHANGED) != 0) bge_link_upd(sc); /* Let controller work. */ bge_writembx(sc, BGE_MBX_IRQ0_LO, status_tag); if (ifp->if_drv_flags & IFF_DRV_RUNNING && sc->bge_rx_saved_considx != rx_prod) { /* Check RX return ring producer/consumer. */ BGE_UNLOCK(sc); bge_rxeof(sc, rx_prod, 0); BGE_LOCK(sc); } if (ifp->if_drv_flags & IFF_DRV_RUNNING) { /* Check TX ring producer/consumer. */ bge_txeof(sc, tx_cons); if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd)) bge_start_locked(ifp); } BGE_UNLOCK(sc); } static void bge_intr(void *xsc) { struct bge_softc *sc; struct ifnet *ifp; uint32_t statusword; uint16_t rx_prod, tx_cons; sc = xsc; BGE_LOCK(sc); ifp = sc->bge_ifp; #ifdef DEVICE_POLLING if (ifp->if_capenable & IFCAP_POLLING) { BGE_UNLOCK(sc); return; } #endif /* * Ack the interrupt by writing something to BGE_MBX_IRQ0_LO. Don't * disable interrupts by writing nonzero like we used to, since with * our current organization this just gives complications and * pessimizations for re-enabling interrupts. We used to have races * instead of the necessary complications. Disabling interrupts * would just reduce the chance of a status update while we are * running (by switching to the interrupt-mode coalescence * parameters), but this chance is already very low so it is more * efficient to get another interrupt than prevent it. * * We do the ack first to ensure another interrupt if there is a * status update after the ack. We don't check for the status * changing later because it is more efficient to get another * interrupt than prevent it, not quite as above (not checking is * a smaller optimization than not toggling the interrupt enable, * since checking doesn't involve PCI accesses and toggling require * the status check). So toggling would probably be a pessimization * even with MSI. It would only be needed for using a task queue. */ bge_writembx(sc, BGE_MBX_IRQ0_LO, 0); /* * Do the mandatory PCI flush as well as get the link status. */ statusword = CSR_READ_4(sc, BGE_MAC_STS) & BGE_MACSTAT_LINK_CHANGED; /* Make sure the descriptor ring indexes are coherent. */ bus_dmamap_sync(sc->bge_cdata.bge_status_tag, sc->bge_cdata.bge_status_map, BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); rx_prod = sc->bge_ldata.bge_status_block->bge_idx[0].bge_rx_prod_idx; tx_cons = sc->bge_ldata.bge_status_block->bge_idx[0].bge_tx_cons_idx; sc->bge_ldata.bge_status_block->bge_status = 0; bus_dmamap_sync(sc->bge_cdata.bge_status_tag, sc->bge_cdata.bge_status_map, BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); if ((sc->bge_asicrev == BGE_ASICREV_BCM5700 && sc->bge_chipid != BGE_CHIPID_BCM5700_B2) || statusword || sc->bge_link_evt) bge_link_upd(sc); if (ifp->if_drv_flags & IFF_DRV_RUNNING) { /* Check RX return ring producer/consumer. */ bge_rxeof(sc, rx_prod, 1); } if (ifp->if_drv_flags & IFF_DRV_RUNNING) { /* Check TX ring producer/consumer. */ bge_txeof(sc, tx_cons); } if (ifp->if_drv_flags & IFF_DRV_RUNNING && !IFQ_DRV_IS_EMPTY(&ifp->if_snd)) bge_start_locked(ifp); BGE_UNLOCK(sc); } static void bge_asf_driver_up(struct bge_softc *sc) { if (sc->bge_asf_mode & ASF_STACKUP) { /* Send ASF heartbeat aprox. every 2s */ if (sc->bge_asf_count) sc->bge_asf_count --; else { sc->bge_asf_count = 2; bge_writemem_ind(sc, BGE_SOFTWARE_GENCOMM_FW, BGE_FW_DRV_ALIVE); bge_writemem_ind(sc, BGE_SOFTWARE_GENNCOMM_FW_LEN, 4); bge_writemem_ind(sc, BGE_SOFTWARE_GENNCOMM_FW_DATA, 3); CSR_WRITE_4(sc, BGE_CPU_EVENT, CSR_READ_4(sc, BGE_CPU_EVENT) | (1 << 14)); } } } static void bge_tick(void *xsc) { struct bge_softc *sc = xsc; struct mii_data *mii = NULL; BGE_LOCK_ASSERT(sc); /* Synchronize with possible callout reset/stop. */ if (callout_pending(&sc->bge_stat_ch) || !callout_active(&sc->bge_stat_ch)) return; if (BGE_IS_5705_PLUS(sc)) bge_stats_update_regs(sc); else bge_stats_update(sc); if ((sc->bge_flags & BGE_FLAG_TBI) == 0) { mii = device_get_softc(sc->bge_miibus); /* * Do not touch PHY if we have link up. This could break * IPMI/ASF mode or produce extra input errors * (extra errors was reported for bcm5701 & bcm5704). */ if (!sc->bge_link) mii_tick(mii); } else { /* * Since in TBI mode auto-polling can't be used we should poll * link status manually. Here we register pending link event * and trigger interrupt. */ #ifdef DEVICE_POLLING /* In polling mode we poll link state in bge_poll(). */ if (!(sc->bge_ifp->if_capenable & IFCAP_POLLING)) #endif { sc->bge_link_evt++; if (sc->bge_asicrev == BGE_ASICREV_BCM5700 || sc->bge_flags & BGE_FLAG_5788) BGE_SETBIT(sc, BGE_MISC_LOCAL_CTL, BGE_MLC_INTR_SET); else BGE_SETBIT(sc, BGE_HCC_MODE, BGE_HCCMODE_COAL_NOW); } } bge_asf_driver_up(sc); bge_watchdog(sc); callout_reset(&sc->bge_stat_ch, hz, bge_tick, sc); } static void bge_stats_update_regs(struct bge_softc *sc) { struct ifnet *ifp; struct bge_mac_stats *stats; ifp = sc->bge_ifp; stats = &sc->bge_mac_stats; stats->ifHCOutOctets += CSR_READ_4(sc, BGE_TX_MAC_STATS_OCTETS); stats->etherStatsCollisions += CSR_READ_4(sc, BGE_TX_MAC_STATS_COLLS); stats->outXonSent += CSR_READ_4(sc, BGE_TX_MAC_STATS_XON_SENT); stats->outXoffSent += CSR_READ_4(sc, BGE_TX_MAC_STATS_XOFF_SENT); stats->dot3StatsInternalMacTransmitErrors += CSR_READ_4(sc, BGE_TX_MAC_STATS_ERRORS); stats->dot3StatsSingleCollisionFrames += CSR_READ_4(sc, BGE_TX_MAC_STATS_SINGLE_COLL); stats->dot3StatsMultipleCollisionFrames += CSR_READ_4(sc, BGE_TX_MAC_STATS_MULTI_COLL); stats->dot3StatsDeferredTransmissions += CSR_READ_4(sc, BGE_TX_MAC_STATS_DEFERRED); stats->dot3StatsExcessiveCollisions += CSR_READ_4(sc, BGE_TX_MAC_STATS_EXCESS_COLL); stats->dot3StatsLateCollisions += CSR_READ_4(sc, BGE_TX_MAC_STATS_LATE_COLL); stats->ifHCOutUcastPkts += CSR_READ_4(sc, BGE_TX_MAC_STATS_UCAST); stats->ifHCOutMulticastPkts += CSR_READ_4(sc, BGE_TX_MAC_STATS_MCAST); stats->ifHCOutBroadcastPkts += CSR_READ_4(sc, BGE_TX_MAC_STATS_BCAST); stats->ifHCInOctets += CSR_READ_4(sc, BGE_RX_MAC_STATS_OCTESTS); stats->etherStatsFragments += CSR_READ_4(sc, BGE_RX_MAC_STATS_FRAGMENTS); stats->ifHCInUcastPkts += CSR_READ_4(sc, BGE_RX_MAC_STATS_UCAST); stats->ifHCInMulticastPkts += CSR_READ_4(sc, BGE_RX_MAC_STATS_MCAST); stats->ifHCInBroadcastPkts += CSR_READ_4(sc, BGE_RX_MAC_STATS_BCAST); stats->dot3StatsFCSErrors += CSR_READ_4(sc, BGE_RX_MAC_STATS_FCS_ERRORS); stats->dot3StatsAlignmentErrors += CSR_READ_4(sc, BGE_RX_MAC_STATS_ALGIN_ERRORS); stats->xonPauseFramesReceived += CSR_READ_4(sc, BGE_RX_MAC_STATS_XON_RCVD); stats->xoffPauseFramesReceived += CSR_READ_4(sc, BGE_RX_MAC_STATS_XOFF_RCVD); stats->macControlFramesReceived += CSR_READ_4(sc, BGE_RX_MAC_STATS_CTRL_RCVD); stats->xoffStateEntered += CSR_READ_4(sc, BGE_RX_MAC_STATS_XOFF_ENTERED); stats->dot3StatsFramesTooLong += CSR_READ_4(sc, BGE_RX_MAC_STATS_FRAME_TOO_LONG); stats->etherStatsJabbers += CSR_READ_4(sc, BGE_RX_MAC_STATS_JABBERS); stats->etherStatsUndersizePkts += CSR_READ_4(sc, BGE_RX_MAC_STATS_UNDERSIZE); stats->FramesDroppedDueToFilters += CSR_READ_4(sc, BGE_RXLP_LOCSTAT_FILTDROP); stats->DmaWriteQueueFull += CSR_READ_4(sc, BGE_RXLP_LOCSTAT_DMA_WRQ_FULL); stats->DmaWriteHighPriQueueFull += CSR_READ_4(sc, BGE_RXLP_LOCSTAT_DMA_HPWRQ_FULL); stats->NoMoreRxBDs += CSR_READ_4(sc, BGE_RXLP_LOCSTAT_OUT_OF_BDS); stats->InputDiscards += CSR_READ_4(sc, BGE_RXLP_LOCSTAT_IFIN_DROPS); stats->InputErrors += CSR_READ_4(sc, BGE_RXLP_LOCSTAT_IFIN_ERRORS); stats->RecvThresholdHit += CSR_READ_4(sc, BGE_RXLP_LOCSTAT_RXTHRESH_HIT); ifp->if_collisions = (u_long)stats->etherStatsCollisions; ifp->if_ierrors = (u_long)(stats->NoMoreRxBDs + stats->InputDiscards + stats->InputErrors); } static void bge_stats_clear_regs(struct bge_softc *sc) { CSR_READ_4(sc, BGE_TX_MAC_STATS_OCTETS); CSR_READ_4(sc, BGE_TX_MAC_STATS_COLLS); CSR_READ_4(sc, BGE_TX_MAC_STATS_XON_SENT); CSR_READ_4(sc, BGE_TX_MAC_STATS_XOFF_SENT); CSR_READ_4(sc, BGE_TX_MAC_STATS_ERRORS); CSR_READ_4(sc, BGE_TX_MAC_STATS_SINGLE_COLL); CSR_READ_4(sc, BGE_TX_MAC_STATS_MULTI_COLL); CSR_READ_4(sc, BGE_TX_MAC_STATS_DEFERRED); CSR_READ_4(sc, BGE_TX_MAC_STATS_EXCESS_COLL); CSR_READ_4(sc, BGE_TX_MAC_STATS_LATE_COLL); CSR_READ_4(sc, BGE_TX_MAC_STATS_UCAST); CSR_READ_4(sc, BGE_TX_MAC_STATS_MCAST); CSR_READ_4(sc, BGE_TX_MAC_STATS_BCAST); CSR_READ_4(sc, BGE_RX_MAC_STATS_OCTESTS); CSR_READ_4(sc, BGE_RX_MAC_STATS_FRAGMENTS); CSR_READ_4(sc, BGE_RX_MAC_STATS_UCAST); CSR_READ_4(sc, BGE_RX_MAC_STATS_MCAST); CSR_READ_4(sc, BGE_RX_MAC_STATS_BCAST); CSR_READ_4(sc, BGE_RX_MAC_STATS_FCS_ERRORS); CSR_READ_4(sc, BGE_RX_MAC_STATS_ALGIN_ERRORS); CSR_READ_4(sc, BGE_RX_MAC_STATS_XON_RCVD); CSR_READ_4(sc, BGE_RX_MAC_STATS_XOFF_RCVD); CSR_READ_4(sc, BGE_RX_MAC_STATS_CTRL_RCVD); CSR_READ_4(sc, BGE_RX_MAC_STATS_XOFF_ENTERED); CSR_READ_4(sc, BGE_RX_MAC_STATS_FRAME_TOO_LONG); CSR_READ_4(sc, BGE_RX_MAC_STATS_JABBERS); CSR_READ_4(sc, BGE_RX_MAC_STATS_UNDERSIZE); CSR_READ_4(sc, BGE_RXLP_LOCSTAT_FILTDROP); CSR_READ_4(sc, BGE_RXLP_LOCSTAT_DMA_WRQ_FULL); CSR_READ_4(sc, BGE_RXLP_LOCSTAT_DMA_HPWRQ_FULL); CSR_READ_4(sc, BGE_RXLP_LOCSTAT_OUT_OF_BDS); CSR_READ_4(sc, BGE_RXLP_LOCSTAT_IFIN_DROPS); CSR_READ_4(sc, BGE_RXLP_LOCSTAT_IFIN_ERRORS); CSR_READ_4(sc, BGE_RXLP_LOCSTAT_RXTHRESH_HIT); } static void bge_stats_update(struct bge_softc *sc) { struct ifnet *ifp; bus_size_t stats; uint32_t cnt; /* current register value */ ifp = sc->bge_ifp; stats = BGE_MEMWIN_START + BGE_STATS_BLOCK; #define READ_STAT(sc, stats, stat) \ CSR_READ_4(sc, stats + offsetof(struct bge_stats, stat)) cnt = READ_STAT(sc, stats, txstats.etherStatsCollisions.bge_addr_lo); ifp->if_collisions += (uint32_t)(cnt - sc->bge_tx_collisions); sc->bge_tx_collisions = cnt; cnt = READ_STAT(sc, stats, ifInDiscards.bge_addr_lo); ifp->if_ierrors += (uint32_t)(cnt - sc->bge_rx_discards); sc->bge_rx_discards = cnt; cnt = READ_STAT(sc, stats, txstats.ifOutDiscards.bge_addr_lo); ifp->if_oerrors += (uint32_t)(cnt - sc->bge_tx_discards); sc->bge_tx_discards = cnt; #undef READ_STAT } /* * Pad outbound frame to ETHER_MIN_NOPAD for an unusual reason. * The bge hardware will pad out Tx runts to ETHER_MIN_NOPAD, * but when such padded frames employ the bge IP/TCP checksum offload, * the hardware checksum assist gives incorrect results (possibly * from incorporating its own padding into the UDP/TCP checksum; who knows). * If we pad such runts with zeros, the onboard checksum comes out correct. */ static __inline int bge_cksum_pad(struct mbuf *m) { int padlen = ETHER_MIN_NOPAD - m->m_pkthdr.len; struct mbuf *last; /* If there's only the packet-header and we can pad there, use it. */ if (m->m_pkthdr.len == m->m_len && M_WRITABLE(m) && M_TRAILINGSPACE(m) >= padlen) { last = m; } else { /* * Walk packet chain to find last mbuf. We will either * pad there, or append a new mbuf and pad it. */ for (last = m; last->m_next != NULL; last = last->m_next); if (!(M_WRITABLE(last) && M_TRAILINGSPACE(last) >= padlen)) { /* Allocate new empty mbuf, pad it. Compact later. */ struct mbuf *n; MGET(n, M_DONTWAIT, MT_DATA); if (n == NULL) return (ENOBUFS); n->m_len = 0; last->m_next = n; last = n; } } /* Now zero the pad area, to avoid the bge cksum-assist bug. */ memset(mtod(last, caddr_t) + last->m_len, 0, padlen); last->m_len += padlen; m->m_pkthdr.len += padlen; return (0); } static struct mbuf * bge_check_short_dma(struct mbuf *m) { struct mbuf *n; int found; /* * If device receive two back-to-back send BDs with less than * or equal to 8 total bytes then the device may hang. The two * back-to-back send BDs must in the same frame for this failure * to occur. Scan mbuf chains and see whether two back-to-back * send BDs are there. If this is the case, allocate new mbuf * and copy the frame to workaround the silicon bug. */ for (n = m, found = 0; n != NULL; n = n->m_next) { if (n->m_len < 8) { found++; if (found > 1) break; continue; } found = 0; } if (found > 1) { n = m_defrag(m, M_DONTWAIT); if (n == NULL) m_freem(m); } else n = m; return (n); } static struct mbuf * bge_setup_tso(struct bge_softc *sc, struct mbuf *m, uint16_t *mss, uint16_t *flags) { struct ip *ip; struct tcphdr *tcp; struct mbuf *n; uint16_t hlen; uint32_t poff; if (M_WRITABLE(m) == 0) { /* Get a writable copy. */ n = m_dup(m, M_DONTWAIT); m_freem(m); if (n == NULL) return (NULL); m = n; } m = m_pullup(m, sizeof(struct ether_header) + sizeof(struct ip)); if (m == NULL) return (NULL); ip = (struct ip *)(mtod(m, char *) + sizeof(struct ether_header)); poff = sizeof(struct ether_header) + (ip->ip_hl << 2); m = m_pullup(m, poff + sizeof(struct tcphdr)); if (m == NULL) return (NULL); tcp = (struct tcphdr *)(mtod(m, char *) + poff); m = m_pullup(m, poff + (tcp->th_off << 2)); if (m == NULL) return (NULL); /* * It seems controller doesn't modify IP length and TCP pseudo * checksum. These checksum computed by upper stack should be 0. */ *mss = m->m_pkthdr.tso_segsz; ip = (struct ip *)(mtod(m, char *) + sizeof(struct ether_header)); ip->ip_sum = 0; ip->ip_len = htons(*mss + (ip->ip_hl << 2) + (tcp->th_off << 2)); /* Clear pseudo checksum computed by TCP stack. */ tcp = (struct tcphdr *)(mtod(m, char *) + poff); tcp->th_sum = 0; /* * Broadcom controllers uses different descriptor format for * TSO depending on ASIC revision. Due to TSO-capable firmware * license issue and lower performance of firmware based TSO * we only support hardware based TSO. */ /* Calculate header length, incl. TCP/IP options, in 32 bit units. */ hlen = ((ip->ip_hl << 2) + (tcp->th_off << 2)) >> 2; if (sc->bge_flags & BGE_FLAG_TSO3) { /* * For BCM5717 and newer controllers, hardware based TSO * uses the 14 lower bits of the bge_mss field to store the * MSS and the upper 2 bits to store the lowest 2 bits of * the IP/TCP header length. The upper 6 bits of the header * length are stored in the bge_flags[14:10,4] field. Jumbo * frames are supported. */ *mss |= ((hlen & 0x3) << 14); *flags |= ((hlen & 0xF8) << 7) | ((hlen & 0x4) << 2); } else { /* * For BCM5755 and newer controllers, hardware based TSO uses * the lower 11 bits to store the MSS and the upper 5 bits to * store the IP/TCP header length. Jumbo frames are not * supported. */ *mss |= (hlen << 11); } return (m); } /* * Encapsulate an mbuf chain in the tx ring by coupling the mbuf data * pointers to descriptors. */ static int bge_encap(struct bge_softc *sc, struct mbuf **m_head, uint32_t *txidx) { bus_dma_segment_t segs[BGE_NSEG_NEW]; bus_dmamap_t map; struct bge_tx_bd *d; struct mbuf *m = *m_head; uint32_t idx = *txidx; uint16_t csum_flags, mss, vlan_tag; int nsegs, i, error; csum_flags = 0; mss = 0; vlan_tag = 0; if ((sc->bge_flags & BGE_FLAG_SHORT_DMA_BUG) != 0 && m->m_next != NULL) { *m_head = bge_check_short_dma(m); if (*m_head == NULL) return (ENOBUFS); m = *m_head; } if ((m->m_pkthdr.csum_flags & CSUM_TSO) != 0) { *m_head = m = bge_setup_tso(sc, m, &mss, &csum_flags); if (*m_head == NULL) return (ENOBUFS); csum_flags |= BGE_TXBDFLAG_CPU_PRE_DMA | BGE_TXBDFLAG_CPU_POST_DMA; } else if ((m->m_pkthdr.csum_flags & sc->bge_csum_features) != 0) { if (m->m_pkthdr.csum_flags & CSUM_IP) csum_flags |= BGE_TXBDFLAG_IP_CSUM; if (m->m_pkthdr.csum_flags & (CSUM_TCP | CSUM_UDP)) { csum_flags |= BGE_TXBDFLAG_TCP_UDP_CSUM; if (m->m_pkthdr.len < ETHER_MIN_NOPAD && (error = bge_cksum_pad(m)) != 0) { m_freem(m); *m_head = NULL; return (error); } } if (m->m_flags & M_LASTFRAG) csum_flags |= BGE_TXBDFLAG_IP_FRAG_END; else if (m->m_flags & M_FRAG) csum_flags |= BGE_TXBDFLAG_IP_FRAG; } if ((m->m_pkthdr.csum_flags & CSUM_TSO) == 0) { if (sc->bge_flags & BGE_FLAG_JUMBO_FRAME && m->m_pkthdr.len > ETHER_MAX_LEN) csum_flags |= BGE_TXBDFLAG_JUMBO_FRAME; if (sc->bge_forced_collapse > 0 && (sc->bge_flags & BGE_FLAG_PCIE) != 0 && m->m_next != NULL) { /* * Forcedly collapse mbuf chains to overcome hardware * limitation which only support a single outstanding * DMA read operation. */ if (sc->bge_forced_collapse == 1) m = m_defrag(m, M_DONTWAIT); else m = m_collapse(m, M_DONTWAIT, sc->bge_forced_collapse); if (m == NULL) m = *m_head; *m_head = m; } } map = sc->bge_cdata.bge_tx_dmamap[idx]; error = bus_dmamap_load_mbuf_sg(sc->bge_cdata.bge_tx_mtag, map, m, segs, &nsegs, BUS_DMA_NOWAIT); if (error == EFBIG) { m = m_collapse(m, M_DONTWAIT, BGE_NSEG_NEW); if (m == NULL) { m_freem(*m_head); *m_head = NULL; return (ENOBUFS); } *m_head = m; error = bus_dmamap_load_mbuf_sg(sc->bge_cdata.bge_tx_mtag, map, m, segs, &nsegs, BUS_DMA_NOWAIT); if (error) { m_freem(m); *m_head = NULL; return (error); } } else if (error != 0) return (error); /* Check if we have enough free send BDs. */ if (sc->bge_txcnt + nsegs >= BGE_TX_RING_CNT) { bus_dmamap_unload(sc->bge_cdata.bge_tx_mtag, map); return (ENOBUFS); } bus_dmamap_sync(sc->bge_cdata.bge_tx_mtag, map, BUS_DMASYNC_PREWRITE); if (m->m_flags & M_VLANTAG) { csum_flags |= BGE_TXBDFLAG_VLAN_TAG; vlan_tag = m->m_pkthdr.ether_vtag; } for (i = 0; ; i++) { d = &sc->bge_ldata.bge_tx_ring[idx]; d->bge_addr.bge_addr_lo = BGE_ADDR_LO(segs[i].ds_addr); d->bge_addr.bge_addr_hi = BGE_ADDR_HI(segs[i].ds_addr); d->bge_len = segs[i].ds_len; d->bge_flags = csum_flags; d->bge_vlan_tag = vlan_tag; d->bge_mss = mss; if (i == nsegs - 1) break; BGE_INC(idx, BGE_TX_RING_CNT); } /* Mark the last segment as end of packet... */ d->bge_flags |= BGE_TXBDFLAG_END; /* * Insure that the map for this transmission * is placed at the array index of the last descriptor * in this chain. */ sc->bge_cdata.bge_tx_dmamap[*txidx] = sc->bge_cdata.bge_tx_dmamap[idx]; sc->bge_cdata.bge_tx_dmamap[idx] = map; sc->bge_cdata.bge_tx_chain[idx] = m; sc->bge_txcnt += nsegs; BGE_INC(idx, BGE_TX_RING_CNT); *txidx = idx; return (0); } /* * Main transmit routine. To avoid having to do mbuf copies, we put pointers * to the mbuf data regions directly in the transmit descriptors. */ static void bge_start_locked(struct ifnet *ifp) { struct bge_softc *sc; struct mbuf *m_head; uint32_t prodidx; int count; sc = ifp->if_softc; BGE_LOCK_ASSERT(sc); if (!sc->bge_link || (ifp->if_drv_flags & (IFF_DRV_RUNNING | IFF_DRV_OACTIVE)) != IFF_DRV_RUNNING) return; prodidx = sc->bge_tx_prodidx; for (count = 0; !IFQ_DRV_IS_EMPTY(&ifp->if_snd);) { if (sc->bge_txcnt > BGE_TX_RING_CNT - 16) { ifp->if_drv_flags |= IFF_DRV_OACTIVE; break; } IFQ_DRV_DEQUEUE(&ifp->if_snd, m_head); if (m_head == NULL) break; /* * XXX * The code inside the if() block is never reached since we * must mark CSUM_IP_FRAGS in our if_hwassist to start getting * requests to checksum TCP/UDP in a fragmented packet. * * XXX * safety overkill. If this is a fragmented packet chain * with delayed TCP/UDP checksums, then only encapsulate * it if we have enough descriptors to handle the entire * chain at once. * (paranoia -- may not actually be needed) */ if (m_head->m_flags & M_FIRSTFRAG && m_head->m_pkthdr.csum_flags & (CSUM_DELAY_DATA)) { if ((BGE_TX_RING_CNT - sc->bge_txcnt) < m_head->m_pkthdr.csum_data + 16) { IFQ_DRV_PREPEND(&ifp->if_snd, m_head); ifp->if_drv_flags |= IFF_DRV_OACTIVE; break; } } /* * Pack the data into the transmit ring. If we * don't have room, set the OACTIVE flag and wait * for the NIC to drain the ring. */ if (bge_encap(sc, &m_head, &prodidx)) { if (m_head == NULL) break; IFQ_DRV_PREPEND(&ifp->if_snd, m_head); ifp->if_drv_flags |= IFF_DRV_OACTIVE; break; } ++count; /* * If there's a BPF listener, bounce a copy of this frame * to him. */ #ifdef ETHER_BPF_MTAP ETHER_BPF_MTAP(ifp, m_head); #else BPF_MTAP(ifp, m_head); #endif } if (count > 0) { bus_dmamap_sync(sc->bge_cdata.bge_tx_ring_tag, sc->bge_cdata.bge_tx_ring_map, BUS_DMASYNC_PREWRITE); /* Transmit. */ bge_writembx(sc, BGE_MBX_TX_HOST_PROD0_LO, prodidx); /* 5700 b2 errata */ if (sc->bge_chiprev == BGE_CHIPREV_5700_BX) bge_writembx(sc, BGE_MBX_TX_HOST_PROD0_LO, prodidx); sc->bge_tx_prodidx = prodidx; /* * Set a timeout in case the chip goes out to lunch. */ sc->bge_timer = 5; } } /* * Main transmit routine. To avoid having to do mbuf copies, we put pointers * to the mbuf data regions directly in the transmit descriptors. */ static void bge_start(struct ifnet *ifp) { struct bge_softc *sc; sc = ifp->if_softc; BGE_LOCK(sc); bge_start_locked(ifp); BGE_UNLOCK(sc); } static void bge_init_locked(struct bge_softc *sc) { struct ifnet *ifp; uint16_t *m; uint32_t mode; BGE_LOCK_ASSERT(sc); ifp = sc->bge_ifp; if (ifp->if_drv_flags & IFF_DRV_RUNNING) return; /* Cancel pending I/O and flush buffers. */ bge_stop(sc); bge_stop_fw(sc); bge_sig_pre_reset(sc, BGE_RESET_START); bge_reset(sc); bge_sig_legacy(sc, BGE_RESET_START); bge_sig_post_reset(sc, BGE_RESET_START); bge_chipinit(sc); /* * Init the various state machines, ring * control blocks and firmware. */ if (bge_blockinit(sc)) { device_printf(sc->bge_dev, "initialization failure\n"); return; } ifp = sc->bge_ifp; /* Specify MTU. */ CSR_WRITE_4(sc, BGE_RX_MTU, ifp->if_mtu + ETHER_HDR_LEN + ETHER_CRC_LEN + (ifp->if_capenable & IFCAP_VLAN_MTU ? ETHER_VLAN_ENCAP_LEN : 0)); /* Load our MAC address. */ m = (uint16_t *)IF_LLADDR(sc->bge_ifp); CSR_WRITE_4(sc, BGE_MAC_ADDR1_LO, htons(m[0])); CSR_WRITE_4(sc, BGE_MAC_ADDR1_HI, (htons(m[1]) << 16) | htons(m[2])); /* Program promiscuous mode. */ bge_setpromisc(sc); /* Program multicast filter. */ bge_setmulti(sc); /* Program VLAN tag stripping. */ bge_setvlan(sc); /* Override UDP checksum offloading. */ if (sc->bge_forced_udpcsum == 0) sc->bge_csum_features &= ~CSUM_UDP; else sc->bge_csum_features |= CSUM_UDP; if (ifp->if_capabilities & IFCAP_TXCSUM && ifp->if_capenable & IFCAP_TXCSUM) { ifp->if_hwassist &= ~(BGE_CSUM_FEATURES | CSUM_UDP); ifp->if_hwassist |= sc->bge_csum_features; } /* Init RX ring. */ if (bge_init_rx_ring_std(sc) != 0) { device_printf(sc->bge_dev, "no memory for std Rx buffers.\n"); bge_stop(sc); return; } /* * Workaround for a bug in 5705 ASIC rev A0. Poll the NIC's * memory to insure that the chip has in fact read the first * entry of the ring. */ if (sc->bge_chipid == BGE_CHIPID_BCM5705_A0) { uint32_t v, i; for (i = 0; i < 10; i++) { DELAY(20); v = bge_readmem_ind(sc, BGE_STD_RX_RINGS + 8); if (v == (MCLBYTES - ETHER_ALIGN)) break; } if (i == 10) device_printf (sc->bge_dev, "5705 A0 chip failed to load RX ring\n"); } /* Init jumbo RX ring. */ if (BGE_IS_JUMBO_CAPABLE(sc) && ifp->if_mtu + ETHER_HDR_LEN + ETHER_CRC_LEN + ETHER_VLAN_ENCAP_LEN > (MCLBYTES - ETHER_ALIGN)) { if (bge_init_rx_ring_jumbo(sc) != 0) { device_printf(sc->bge_dev, "no memory for jumbo Rx buffers.\n"); bge_stop(sc); return; } } /* Init our RX return ring index. */ sc->bge_rx_saved_considx = 0; /* Init our RX/TX stat counters. */ sc->bge_rx_discards = sc->bge_tx_discards = sc->bge_tx_collisions = 0; /* Init TX ring. */ bge_init_tx_ring(sc); /* Enable TX MAC state machine lockup fix. */ mode = CSR_READ_4(sc, BGE_TX_MODE); if (BGE_IS_5755_PLUS(sc) || sc->bge_asicrev == BGE_ASICREV_BCM5906) mode |= BGE_TXMODE_MBUF_LOCKUP_FIX; /* Turn on transmitter. */ CSR_WRITE_4(sc, BGE_TX_MODE, mode | BGE_TXMODE_ENABLE); /* Turn on receiver. */ BGE_SETBIT(sc, BGE_RX_MODE, BGE_RXMODE_ENABLE); /* * Set the number of good frames to receive after RX MBUF * Low Watermark has been reached. After the RX MAC receives * this number of frames, it will drop subsequent incoming * frames until the MBUF High Watermark is reached. */ - CSR_WRITE_4(sc, BGE_MAX_RX_FRAME_LOWAT, 2); + if (sc->bge_asicrev == BGE_ASICREV_BCM57765) + CSR_WRITE_4(sc, BGE_MAX_RX_FRAME_LOWAT, 1); + else + CSR_WRITE_4(sc, BGE_MAX_RX_FRAME_LOWAT, 2); /* Clear MAC statistics. */ if (BGE_IS_5705_PLUS(sc)) bge_stats_clear_regs(sc); /* Tell firmware we're alive. */ BGE_SETBIT(sc, BGE_MODE_CTL, BGE_MODECTL_STACKUP); #ifdef DEVICE_POLLING /* Disable interrupts if we are polling. */ if (ifp->if_capenable & IFCAP_POLLING) { BGE_SETBIT(sc, BGE_PCI_MISC_CTL, BGE_PCIMISCCTL_MASK_PCI_INTR); bge_writembx(sc, BGE_MBX_IRQ0_LO, 1); } else #endif /* Enable host interrupts. */ { BGE_SETBIT(sc, BGE_PCI_MISC_CTL, BGE_PCIMISCCTL_CLEAR_INTA); BGE_CLRBIT(sc, BGE_PCI_MISC_CTL, BGE_PCIMISCCTL_MASK_PCI_INTR); bge_writembx(sc, BGE_MBX_IRQ0_LO, 0); } bge_ifmedia_upd_locked(ifp); ifp->if_drv_flags |= IFF_DRV_RUNNING; ifp->if_drv_flags &= ~IFF_DRV_OACTIVE; callout_reset(&sc->bge_stat_ch, hz, bge_tick, sc); } static void bge_init(void *xsc) { struct bge_softc *sc = xsc; BGE_LOCK(sc); bge_init_locked(sc); BGE_UNLOCK(sc); } /* * Set media options. */ static int bge_ifmedia_upd(struct ifnet *ifp) { struct bge_softc *sc = ifp->if_softc; int res; BGE_LOCK(sc); res = bge_ifmedia_upd_locked(ifp); BGE_UNLOCK(sc); return (res); } static int bge_ifmedia_upd_locked(struct ifnet *ifp) { struct bge_softc *sc = ifp->if_softc; struct mii_data *mii; struct mii_softc *miisc; struct ifmedia *ifm; BGE_LOCK_ASSERT(sc); ifm = &sc->bge_ifmedia; /* If this is a 1000baseX NIC, enable the TBI port. */ if (sc->bge_flags & BGE_FLAG_TBI) { if (IFM_TYPE(ifm->ifm_media) != IFM_ETHER) return (EINVAL); switch(IFM_SUBTYPE(ifm->ifm_media)) { case IFM_AUTO: /* * The BCM5704 ASIC appears to have a special * mechanism for programming the autoneg * advertisement registers in TBI mode. */ if (sc->bge_asicrev == BGE_ASICREV_BCM5704) { uint32_t sgdig; sgdig = CSR_READ_4(sc, BGE_SGDIG_STS); if (sgdig & BGE_SGDIGSTS_DONE) { CSR_WRITE_4(sc, BGE_TX_TBI_AUTONEG, 0); sgdig = CSR_READ_4(sc, BGE_SGDIG_CFG); sgdig |= BGE_SGDIGCFG_AUTO | BGE_SGDIGCFG_PAUSE_CAP | BGE_SGDIGCFG_ASYM_PAUSE; CSR_WRITE_4(sc, BGE_SGDIG_CFG, sgdig | BGE_SGDIGCFG_SEND); DELAY(5); CSR_WRITE_4(sc, BGE_SGDIG_CFG, sgdig); } } break; case IFM_1000_SX: if ((ifm->ifm_media & IFM_GMASK) == IFM_FDX) { BGE_CLRBIT(sc, BGE_MAC_MODE, BGE_MACMODE_HALF_DUPLEX); } else { BGE_SETBIT(sc, BGE_MAC_MODE, BGE_MACMODE_HALF_DUPLEX); } break; default: return (EINVAL); } return (0); } sc->bge_link_evt++; mii = device_get_softc(sc->bge_miibus); LIST_FOREACH(miisc, &mii->mii_phys, mii_list) PHY_RESET(miisc); mii_mediachg(mii); /* * Force an interrupt so that we will call bge_link_upd * if needed and clear any pending link state attention. * Without this we are not getting any further interrupts * for link state changes and thus will not UP the link and * not be able to send in bge_start_locked. The only * way to get things working was to receive a packet and * get an RX intr. * bge_tick should help for fiber cards and we might not * need to do this here if BGE_FLAG_TBI is set but as * we poll for fiber anyway it should not harm. */ if (sc->bge_asicrev == BGE_ASICREV_BCM5700 || sc->bge_flags & BGE_FLAG_5788) BGE_SETBIT(sc, BGE_MISC_LOCAL_CTL, BGE_MLC_INTR_SET); else BGE_SETBIT(sc, BGE_HCC_MODE, BGE_HCCMODE_COAL_NOW); return (0); } /* * Report current media status. */ static void bge_ifmedia_sts(struct ifnet *ifp, struct ifmediareq *ifmr) { struct bge_softc *sc = ifp->if_softc; struct mii_data *mii; BGE_LOCK(sc); if (sc->bge_flags & BGE_FLAG_TBI) { ifmr->ifm_status = IFM_AVALID; ifmr->ifm_active = IFM_ETHER; if (CSR_READ_4(sc, BGE_MAC_STS) & BGE_MACSTAT_TBI_PCS_SYNCHED) ifmr->ifm_status |= IFM_ACTIVE; else { ifmr->ifm_active |= IFM_NONE; BGE_UNLOCK(sc); return; } ifmr->ifm_active |= IFM_1000_SX; if (CSR_READ_4(sc, BGE_MAC_MODE) & BGE_MACMODE_HALF_DUPLEX) ifmr->ifm_active |= IFM_HDX; else ifmr->ifm_active |= IFM_FDX; BGE_UNLOCK(sc); return; } mii = device_get_softc(sc->bge_miibus); mii_pollstat(mii); ifmr->ifm_active = mii->mii_media_active; ifmr->ifm_status = mii->mii_media_status; BGE_UNLOCK(sc); } static int bge_ioctl(struct ifnet *ifp, u_long command, caddr_t data) { struct bge_softc *sc = ifp->if_softc; struct ifreq *ifr = (struct ifreq *) data; struct mii_data *mii; int flags, mask, error = 0; switch (command) { case SIOCSIFMTU: if (BGE_IS_JUMBO_CAPABLE(sc) || (sc->bge_flags & BGE_FLAG_JUMBO_STD)) { if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > BGE_JUMBO_MTU) { error = EINVAL; break; } } else if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > ETHERMTU) { error = EINVAL; break; } BGE_LOCK(sc); if (ifp->if_mtu != ifr->ifr_mtu) { ifp->if_mtu = ifr->ifr_mtu; if (ifp->if_drv_flags & IFF_DRV_RUNNING) { ifp->if_drv_flags &= ~IFF_DRV_RUNNING; bge_init_locked(sc); } } BGE_UNLOCK(sc); break; case SIOCSIFFLAGS: BGE_LOCK(sc); if (ifp->if_flags & IFF_UP) { /* * If only the state of the PROMISC flag changed, * then just use the 'set promisc mode' command * instead of reinitializing the entire NIC. Doing * a full re-init means reloading the firmware and * waiting for it to start up, which may take a * second or two. Similarly for ALLMULTI. */ if (ifp->if_drv_flags & IFF_DRV_RUNNING) { flags = ifp->if_flags ^ sc->bge_if_flags; if (flags & IFF_PROMISC) bge_setpromisc(sc); if (flags & IFF_ALLMULTI) bge_setmulti(sc); } else bge_init_locked(sc); } else { if (ifp->if_drv_flags & IFF_DRV_RUNNING) { bge_stop(sc); } } sc->bge_if_flags = ifp->if_flags; BGE_UNLOCK(sc); error = 0; break; case SIOCADDMULTI: case SIOCDELMULTI: if (ifp->if_drv_flags & IFF_DRV_RUNNING) { BGE_LOCK(sc); bge_setmulti(sc); BGE_UNLOCK(sc); error = 0; } break; case SIOCSIFMEDIA: case SIOCGIFMEDIA: if (sc->bge_flags & BGE_FLAG_TBI) { error = ifmedia_ioctl(ifp, ifr, &sc->bge_ifmedia, command); } else { mii = device_get_softc(sc->bge_miibus); error = ifmedia_ioctl(ifp, ifr, &mii->mii_media, command); } break; case SIOCSIFCAP: mask = ifr->ifr_reqcap ^ ifp->if_capenable; #ifdef DEVICE_POLLING if (mask & IFCAP_POLLING) { if (ifr->ifr_reqcap & IFCAP_POLLING) { error = ether_poll_register(bge_poll, ifp); if (error) return (error); BGE_LOCK(sc); BGE_SETBIT(sc, BGE_PCI_MISC_CTL, BGE_PCIMISCCTL_MASK_PCI_INTR); bge_writembx(sc, BGE_MBX_IRQ0_LO, 1); ifp->if_capenable |= IFCAP_POLLING; BGE_UNLOCK(sc); } else { error = ether_poll_deregister(ifp); /* Enable interrupt even in error case */ BGE_LOCK(sc); BGE_CLRBIT(sc, BGE_PCI_MISC_CTL, BGE_PCIMISCCTL_MASK_PCI_INTR); bge_writembx(sc, BGE_MBX_IRQ0_LO, 0); ifp->if_capenable &= ~IFCAP_POLLING; BGE_UNLOCK(sc); } } #endif if ((mask & IFCAP_TXCSUM) != 0 && (ifp->if_capabilities & IFCAP_TXCSUM) != 0) { ifp->if_capenable ^= IFCAP_TXCSUM; if ((ifp->if_capenable & IFCAP_TXCSUM) != 0) ifp->if_hwassist |= sc->bge_csum_features; else ifp->if_hwassist &= ~sc->bge_csum_features; } if ((mask & IFCAP_RXCSUM) != 0 && (ifp->if_capabilities & IFCAP_RXCSUM) != 0) ifp->if_capenable ^= IFCAP_RXCSUM; if ((mask & IFCAP_TSO4) != 0 && (ifp->if_capabilities & IFCAP_TSO4) != 0) { ifp->if_capenable ^= IFCAP_TSO4; if ((ifp->if_capenable & IFCAP_TSO4) != 0) ifp->if_hwassist |= CSUM_TSO; else ifp->if_hwassist &= ~CSUM_TSO; } if (mask & IFCAP_VLAN_MTU) { ifp->if_capenable ^= IFCAP_VLAN_MTU; ifp->if_drv_flags &= ~IFF_DRV_RUNNING; bge_init(sc); } if ((mask & IFCAP_VLAN_HWTSO) != 0 && (ifp->if_capabilities & IFCAP_VLAN_HWTSO) != 0) ifp->if_capenable ^= IFCAP_VLAN_HWTSO; if ((mask & IFCAP_VLAN_HWTAGGING) != 0 && (ifp->if_capabilities & IFCAP_VLAN_HWTAGGING) != 0) { ifp->if_capenable ^= IFCAP_VLAN_HWTAGGING; if ((ifp->if_capenable & IFCAP_VLAN_HWTAGGING) == 0) ifp->if_capenable &= ~IFCAP_VLAN_HWTSO; BGE_LOCK(sc); bge_setvlan(sc); BGE_UNLOCK(sc); } #ifdef VLAN_CAPABILITIES VLAN_CAPABILITIES(ifp); #endif break; default: error = ether_ioctl(ifp, command, data); break; } return (error); } static void bge_watchdog(struct bge_softc *sc) { struct ifnet *ifp; BGE_LOCK_ASSERT(sc); if (sc->bge_timer == 0 || --sc->bge_timer) return; ifp = sc->bge_ifp; if_printf(ifp, "watchdog timeout -- resetting\n"); ifp->if_drv_flags &= ~IFF_DRV_RUNNING; bge_init_locked(sc); ifp->if_oerrors++; } /* * Stop the adapter and free any mbufs allocated to the * RX and TX lists. */ static void bge_stop(struct bge_softc *sc) { struct ifnet *ifp; BGE_LOCK_ASSERT(sc); ifp = sc->bge_ifp; callout_stop(&sc->bge_stat_ch); /* Disable host interrupts. */ BGE_SETBIT(sc, BGE_PCI_MISC_CTL, BGE_PCIMISCCTL_MASK_PCI_INTR); bge_writembx(sc, BGE_MBX_IRQ0_LO, 1); /* * Tell firmware we're shutting down. */ bge_stop_fw(sc); bge_sig_pre_reset(sc, BGE_RESET_STOP); /* * Disable all of the receiver blocks. */ BGE_CLRBIT(sc, BGE_RX_MODE, BGE_RXMODE_ENABLE); BGE_CLRBIT(sc, BGE_RBDI_MODE, BGE_RBDIMODE_ENABLE); BGE_CLRBIT(sc, BGE_RXLP_MODE, BGE_RXLPMODE_ENABLE); if (!(BGE_IS_5705_PLUS(sc))) BGE_CLRBIT(sc, BGE_RXLS_MODE, BGE_RXLSMODE_ENABLE); BGE_CLRBIT(sc, BGE_RDBDI_MODE, BGE_RBDIMODE_ENABLE); BGE_CLRBIT(sc, BGE_RDC_MODE, BGE_RDCMODE_ENABLE); BGE_CLRBIT(sc, BGE_RBDC_MODE, BGE_RBDCMODE_ENABLE); /* * Disable all of the transmit blocks. */ BGE_CLRBIT(sc, BGE_SRS_MODE, BGE_SRSMODE_ENABLE); BGE_CLRBIT(sc, BGE_SBDI_MODE, BGE_SBDIMODE_ENABLE); BGE_CLRBIT(sc, BGE_SDI_MODE, BGE_SDIMODE_ENABLE); BGE_CLRBIT(sc, BGE_RDMA_MODE, BGE_RDMAMODE_ENABLE); BGE_CLRBIT(sc, BGE_SDC_MODE, BGE_SDCMODE_ENABLE); if (!(BGE_IS_5705_PLUS(sc))) BGE_CLRBIT(sc, BGE_DMAC_MODE, BGE_DMACMODE_ENABLE); BGE_CLRBIT(sc, BGE_SBDC_MODE, BGE_SBDCMODE_ENABLE); /* * Shut down all of the memory managers and related * state machines. */ BGE_CLRBIT(sc, BGE_HCC_MODE, BGE_HCCMODE_ENABLE); BGE_CLRBIT(sc, BGE_WDMA_MODE, BGE_WDMAMODE_ENABLE); if (!(BGE_IS_5705_PLUS(sc))) BGE_CLRBIT(sc, BGE_MBCF_MODE, BGE_MBCFMODE_ENABLE); CSR_WRITE_4(sc, BGE_FTQ_RESET, 0xFFFFFFFF); CSR_WRITE_4(sc, BGE_FTQ_RESET, 0); if (!(BGE_IS_5705_PLUS(sc))) { BGE_CLRBIT(sc, BGE_BMAN_MODE, BGE_BMANMODE_ENABLE); BGE_CLRBIT(sc, BGE_MARB_MODE, BGE_MARBMODE_ENABLE); } /* Update MAC statistics. */ if (BGE_IS_5705_PLUS(sc)) bge_stats_update_regs(sc); bge_reset(sc); bge_sig_legacy(sc, BGE_RESET_STOP); bge_sig_post_reset(sc, BGE_RESET_STOP); /* * Keep the ASF firmware running if up. */ if (sc->bge_asf_mode & ASF_STACKUP) BGE_SETBIT(sc, BGE_MODE_CTL, BGE_MODECTL_STACKUP); else BGE_CLRBIT(sc, BGE_MODE_CTL, BGE_MODECTL_STACKUP); /* Free the RX lists. */ bge_free_rx_ring_std(sc); /* Free jumbo RX list. */ if (BGE_IS_JUMBO_CAPABLE(sc)) bge_free_rx_ring_jumbo(sc); /* Free TX buffers. */ bge_free_tx_ring(sc); sc->bge_tx_saved_considx = BGE_TXCONS_UNSET; /* Clear MAC's link state (PHY may still have link UP). */ if (bootverbose && sc->bge_link) if_printf(sc->bge_ifp, "link DOWN\n"); sc->bge_link = 0; ifp->if_drv_flags &= ~(IFF_DRV_RUNNING | IFF_DRV_OACTIVE); } /* * Stop all chip I/O so that the kernel's probe routines don't * get confused by errant DMAs when rebooting. */ static int bge_shutdown(device_t dev) { struct bge_softc *sc; sc = device_get_softc(dev); BGE_LOCK(sc); bge_stop(sc); bge_reset(sc); BGE_UNLOCK(sc); return (0); } static int bge_suspend(device_t dev) { struct bge_softc *sc; sc = device_get_softc(dev); BGE_LOCK(sc); bge_stop(sc); BGE_UNLOCK(sc); return (0); } static int bge_resume(device_t dev) { struct bge_softc *sc; struct ifnet *ifp; sc = device_get_softc(dev); BGE_LOCK(sc); ifp = sc->bge_ifp; if (ifp->if_flags & IFF_UP) { bge_init_locked(sc); if (ifp->if_drv_flags & IFF_DRV_RUNNING) bge_start_locked(ifp); } BGE_UNLOCK(sc); return (0); } static void bge_link_upd(struct bge_softc *sc) { struct mii_data *mii; uint32_t link, status; BGE_LOCK_ASSERT(sc); /* Clear 'pending link event' flag. */ sc->bge_link_evt = 0; /* * Process link state changes. * Grrr. The link status word in the status block does * not work correctly on the BCM5700 rev AX and BX chips, * according to all available information. Hence, we have * to enable MII interrupts in order to properly obtain * async link changes. Unfortunately, this also means that * we have to read the MAC status register to detect link * changes, thereby adding an additional register access to * the interrupt handler. * * XXX: perhaps link state detection procedure used for * BGE_CHIPID_BCM5700_B2 can be used for others BCM5700 revisions. */ if (sc->bge_asicrev == BGE_ASICREV_BCM5700 && sc->bge_chipid != BGE_CHIPID_BCM5700_B2) { status = CSR_READ_4(sc, BGE_MAC_STS); if (status & BGE_MACSTAT_MI_INTERRUPT) { mii = device_get_softc(sc->bge_miibus); mii_pollstat(mii); if (!sc->bge_link && mii->mii_media_status & IFM_ACTIVE && IFM_SUBTYPE(mii->mii_media_active) != IFM_NONE) { sc->bge_link++; if (bootverbose) if_printf(sc->bge_ifp, "link UP\n"); } else if (sc->bge_link && (!(mii->mii_media_status & IFM_ACTIVE) || IFM_SUBTYPE(mii->mii_media_active) == IFM_NONE)) { sc->bge_link = 0; if (bootverbose) if_printf(sc->bge_ifp, "link DOWN\n"); } /* Clear the interrupt. */ CSR_WRITE_4(sc, BGE_MAC_EVT_ENB, BGE_EVTENB_MI_INTERRUPT); bge_miibus_readreg(sc->bge_dev, 1, BRGPHY_MII_ISR); bge_miibus_writereg(sc->bge_dev, 1, BRGPHY_MII_IMR, BRGPHY_INTRS); } return; } if (sc->bge_flags & BGE_FLAG_TBI) { status = CSR_READ_4(sc, BGE_MAC_STS); if (status & BGE_MACSTAT_TBI_PCS_SYNCHED) { if (!sc->bge_link) { sc->bge_link++; if (sc->bge_asicrev == BGE_ASICREV_BCM5704) BGE_CLRBIT(sc, BGE_MAC_MODE, BGE_MACMODE_TBI_SEND_CFGS); CSR_WRITE_4(sc, BGE_MAC_STS, 0xFFFFFFFF); if (bootverbose) if_printf(sc->bge_ifp, "link UP\n"); if_link_state_change(sc->bge_ifp, LINK_STATE_UP); } } else if (sc->bge_link) { sc->bge_link = 0; if (bootverbose) if_printf(sc->bge_ifp, "link DOWN\n"); if_link_state_change(sc->bge_ifp, LINK_STATE_DOWN); } } else if ((sc->bge_mi_mode & BGE_MIMODE_AUTOPOLL) != 0) { /* * Some broken BCM chips have BGE_STATFLAG_LINKSTATE_CHANGED bit * in status word always set. Workaround this bug by reading * PHY link status directly. */ link = (CSR_READ_4(sc, BGE_MI_STS) & BGE_MISTS_LINK) ? 1 : 0; if (link != sc->bge_link || sc->bge_asicrev == BGE_ASICREV_BCM5700) { mii = device_get_softc(sc->bge_miibus); mii_pollstat(mii); if (!sc->bge_link && mii->mii_media_status & IFM_ACTIVE && IFM_SUBTYPE(mii->mii_media_active) != IFM_NONE) { sc->bge_link++; if (bootverbose) if_printf(sc->bge_ifp, "link UP\n"); } else if (sc->bge_link && (!(mii->mii_media_status & IFM_ACTIVE) || IFM_SUBTYPE(mii->mii_media_active) == IFM_NONE)) { sc->bge_link = 0; if (bootverbose) if_printf(sc->bge_ifp, "link DOWN\n"); } } } else { /* * For controllers that call mii_tick, we have to poll * link status. */ mii = device_get_softc(sc->bge_miibus); mii_pollstat(mii); bge_miibus_statchg(sc->bge_dev); } /* Clear the attention. */ CSR_WRITE_4(sc, BGE_MAC_STS, BGE_MACSTAT_SYNC_CHANGED | BGE_MACSTAT_CFG_CHANGED | BGE_MACSTAT_MI_COMPLETE | BGE_MACSTAT_LINK_CHANGED); } static void bge_add_sysctls(struct bge_softc *sc) { struct sysctl_ctx_list *ctx; struct sysctl_oid_list *children; char tn[32]; int unit; ctx = device_get_sysctl_ctx(sc->bge_dev); children = SYSCTL_CHILDREN(device_get_sysctl_tree(sc->bge_dev)); #ifdef BGE_REGISTER_DEBUG SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "debug_info", CTLTYPE_INT | CTLFLAG_RW, sc, 0, bge_sysctl_debug_info, "I", "Debug Information"); SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "reg_read", CTLTYPE_INT | CTLFLAG_RW, sc, 0, bge_sysctl_reg_read, "I", "Register Read"); SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "mem_read", CTLTYPE_INT | CTLFLAG_RW, sc, 0, bge_sysctl_mem_read, "I", "Memory Read"); #endif unit = device_get_unit(sc->bge_dev); /* * A common design characteristic for many Broadcom client controllers * is that they only support a single outstanding DMA read operation * on the PCIe bus. This means that it will take twice as long to fetch * a TX frame that is split into header and payload buffers as it does * to fetch a single, contiguous TX frame (2 reads vs. 1 read). For * these controllers, coalescing buffers to reduce the number of memory * reads is effective way to get maximum performance(about 940Mbps). * Without collapsing TX buffers the maximum TCP bulk transfer * performance is about 850Mbps. However forcing coalescing mbufs * consumes a lot of CPU cycles, so leave it off by default. */ sc->bge_forced_collapse = 0; snprintf(tn, sizeof(tn), "dev.bge.%d.forced_collapse", unit); TUNABLE_INT_FETCH(tn, &sc->bge_forced_collapse); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "forced_collapse", CTLFLAG_RW, &sc->bge_forced_collapse, 0, "Number of fragmented TX buffers of a frame allowed before " "forced collapsing"); /* * It seems all Broadcom controllers have a bug that can generate UDP * datagrams with checksum value 0 when TX UDP checksum offloading is * enabled. Generating UDP checksum value 0 is RFC 768 violation. * Even though the probability of generating such UDP datagrams is * low, I don't want to see FreeBSD boxes to inject such datagrams * into network so disable UDP checksum offloading by default. Users * still override this behavior by setting a sysctl variable, * dev.bge.0.forced_udpcsum. */ sc->bge_forced_udpcsum = 0; snprintf(tn, sizeof(tn), "dev.bge.%d.bge_forced_udpcsum", unit); TUNABLE_INT_FETCH(tn, &sc->bge_forced_udpcsum); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "forced_udpcsum", CTLFLAG_RW, &sc->bge_forced_udpcsum, 0, "Enable UDP checksum offloading even if controller can " "generate UDP checksum value 0"); if (BGE_IS_5705_PLUS(sc)) bge_add_sysctl_stats_regs(sc, ctx, children); else bge_add_sysctl_stats(sc, ctx, children); } #define BGE_SYSCTL_STAT(sc, ctx, desc, parent, node, oid) \ SYSCTL_ADD_PROC(ctx, parent, OID_AUTO, oid, CTLTYPE_UINT|CTLFLAG_RD, \ sc, offsetof(struct bge_stats, node), bge_sysctl_stats, "IU", \ desc) static void bge_add_sysctl_stats(struct bge_softc *sc, struct sysctl_ctx_list *ctx, struct sysctl_oid_list *parent) { struct sysctl_oid *tree; struct sysctl_oid_list *children, *schildren; tree = SYSCTL_ADD_NODE(ctx, parent, OID_AUTO, "stats", CTLFLAG_RD, NULL, "BGE Statistics"); schildren = children = SYSCTL_CHILDREN(tree); BGE_SYSCTL_STAT(sc, ctx, "Frames Dropped Due To Filters", children, COSFramesDroppedDueToFilters, "FramesDroppedDueToFilters"); BGE_SYSCTL_STAT(sc, ctx, "NIC DMA Write Queue Full", children, nicDmaWriteQueueFull, "DmaWriteQueueFull"); BGE_SYSCTL_STAT(sc, ctx, "NIC DMA Write High Priority Queue Full", children, nicDmaWriteHighPriQueueFull, "DmaWriteHighPriQueueFull"); BGE_SYSCTL_STAT(sc, ctx, "NIC No More RX Buffer Descriptors", children, nicNoMoreRxBDs, "NoMoreRxBDs"); BGE_SYSCTL_STAT(sc, ctx, "Discarded Input Frames", children, ifInDiscards, "InputDiscards"); BGE_SYSCTL_STAT(sc, ctx, "Input Errors", children, ifInErrors, "InputErrors"); BGE_SYSCTL_STAT(sc, ctx, "NIC Recv Threshold Hit", children, nicRecvThresholdHit, "RecvThresholdHit"); BGE_SYSCTL_STAT(sc, ctx, "NIC DMA Read Queue Full", children, nicDmaReadQueueFull, "DmaReadQueueFull"); BGE_SYSCTL_STAT(sc, ctx, "NIC DMA Read High Priority Queue Full", children, nicDmaReadHighPriQueueFull, "DmaReadHighPriQueueFull"); BGE_SYSCTL_STAT(sc, ctx, "NIC Send Data Complete Queue Full", children, nicSendDataCompQueueFull, "SendDataCompQueueFull"); BGE_SYSCTL_STAT(sc, ctx, "NIC Ring Set Send Producer Index", children, nicRingSetSendProdIndex, "RingSetSendProdIndex"); BGE_SYSCTL_STAT(sc, ctx, "NIC Ring Status Update", children, nicRingStatusUpdate, "RingStatusUpdate"); BGE_SYSCTL_STAT(sc, ctx, "NIC Interrupts", children, nicInterrupts, "Interrupts"); BGE_SYSCTL_STAT(sc, ctx, "NIC Avoided Interrupts", children, nicAvoidedInterrupts, "AvoidedInterrupts"); BGE_SYSCTL_STAT(sc, ctx, "NIC Send Threshold Hit", children, nicSendThresholdHit, "SendThresholdHit"); tree = SYSCTL_ADD_NODE(ctx, schildren, OID_AUTO, "rx", CTLFLAG_RD, NULL, "BGE RX Statistics"); children = SYSCTL_CHILDREN(tree); BGE_SYSCTL_STAT(sc, ctx, "Inbound Octets", children, rxstats.ifHCInOctets, "ifHCInOctets"); BGE_SYSCTL_STAT(sc, ctx, "Fragments", children, rxstats.etherStatsFragments, "Fragments"); BGE_SYSCTL_STAT(sc, ctx, "Inbound Unicast Packets", children, rxstats.ifHCInUcastPkts, "UnicastPkts"); BGE_SYSCTL_STAT(sc, ctx, "Inbound Multicast Packets", children, rxstats.ifHCInMulticastPkts, "MulticastPkts"); BGE_SYSCTL_STAT(sc, ctx, "FCS Errors", children, rxstats.dot3StatsFCSErrors, "FCSErrors"); BGE_SYSCTL_STAT(sc, ctx, "Alignment Errors", children, rxstats.dot3StatsAlignmentErrors, "AlignmentErrors"); BGE_SYSCTL_STAT(sc, ctx, "XON Pause Frames Received", children, rxstats.xonPauseFramesReceived, "xonPauseFramesReceived"); BGE_SYSCTL_STAT(sc, ctx, "XOFF Pause Frames Received", children, rxstats.xoffPauseFramesReceived, "xoffPauseFramesReceived"); BGE_SYSCTL_STAT(sc, ctx, "MAC Control Frames Received", children, rxstats.macControlFramesReceived, "ControlFramesReceived"); BGE_SYSCTL_STAT(sc, ctx, "XOFF State Entered", children, rxstats.xoffStateEntered, "xoffStateEntered"); BGE_SYSCTL_STAT(sc, ctx, "Frames Too Long", children, rxstats.dot3StatsFramesTooLong, "FramesTooLong"); BGE_SYSCTL_STAT(sc, ctx, "Jabbers", children, rxstats.etherStatsJabbers, "Jabbers"); BGE_SYSCTL_STAT(sc, ctx, "Undersized Packets", children, rxstats.etherStatsUndersizePkts, "UndersizePkts"); BGE_SYSCTL_STAT(sc, ctx, "Inbound Range Length Errors", children, rxstats.inRangeLengthError, "inRangeLengthError"); BGE_SYSCTL_STAT(sc, ctx, "Outbound Range Length Errors", children, rxstats.outRangeLengthError, "outRangeLengthError"); tree = SYSCTL_ADD_NODE(ctx, schildren, OID_AUTO, "tx", CTLFLAG_RD, NULL, "BGE TX Statistics"); children = SYSCTL_CHILDREN(tree); BGE_SYSCTL_STAT(sc, ctx, "Outbound Octets", children, txstats.ifHCOutOctets, "ifHCOutOctets"); BGE_SYSCTL_STAT(sc, ctx, "TX Collisions", children, txstats.etherStatsCollisions, "Collisions"); BGE_SYSCTL_STAT(sc, ctx, "XON Sent", children, txstats.outXonSent, "XonSent"); BGE_SYSCTL_STAT(sc, ctx, "XOFF Sent", children, txstats.outXoffSent, "XoffSent"); BGE_SYSCTL_STAT(sc, ctx, "Flow Control Done", children, txstats.flowControlDone, "flowControlDone"); BGE_SYSCTL_STAT(sc, ctx, "Internal MAC TX errors", children, txstats.dot3StatsInternalMacTransmitErrors, "InternalMacTransmitErrors"); BGE_SYSCTL_STAT(sc, ctx, "Single Collision Frames", children, txstats.dot3StatsSingleCollisionFrames, "SingleCollisionFrames"); BGE_SYSCTL_STAT(sc, ctx, "Multiple Collision Frames", children, txstats.dot3StatsMultipleCollisionFrames, "MultipleCollisionFrames"); BGE_SYSCTL_STAT(sc, ctx, "Deferred Transmissions", children, txstats.dot3StatsDeferredTransmissions, "DeferredTransmissions"); BGE_SYSCTL_STAT(sc, ctx, "Excessive Collisions", children, txstats.dot3StatsExcessiveCollisions, "ExcessiveCollisions"); BGE_SYSCTL_STAT(sc, ctx, "Late Collisions", children, txstats.dot3StatsLateCollisions, "LateCollisions"); BGE_SYSCTL_STAT(sc, ctx, "Outbound Unicast Packets", children, txstats.ifHCOutUcastPkts, "UnicastPkts"); BGE_SYSCTL_STAT(sc, ctx, "Outbound Multicast Packets", children, txstats.ifHCOutMulticastPkts, "MulticastPkts"); BGE_SYSCTL_STAT(sc, ctx, "Outbound Broadcast Packets", children, txstats.ifHCOutBroadcastPkts, "BroadcastPkts"); BGE_SYSCTL_STAT(sc, ctx, "Carrier Sense Errors", children, txstats.dot3StatsCarrierSenseErrors, "CarrierSenseErrors"); BGE_SYSCTL_STAT(sc, ctx, "Outbound Discards", children, txstats.ifOutDiscards, "Discards"); BGE_SYSCTL_STAT(sc, ctx, "Outbound Errors", children, txstats.ifOutErrors, "Errors"); } #undef BGE_SYSCTL_STAT #define BGE_SYSCTL_STAT_ADD64(c, h, n, p, d) \ SYSCTL_ADD_UQUAD(c, h, OID_AUTO, n, CTLFLAG_RD, p, d) static void bge_add_sysctl_stats_regs(struct bge_softc *sc, struct sysctl_ctx_list *ctx, struct sysctl_oid_list *parent) { struct sysctl_oid *tree; struct sysctl_oid_list *child, *schild; struct bge_mac_stats *stats; stats = &sc->bge_mac_stats; tree = SYSCTL_ADD_NODE(ctx, parent, OID_AUTO, "stats", CTLFLAG_RD, NULL, "BGE Statistics"); schild = child = SYSCTL_CHILDREN(tree); BGE_SYSCTL_STAT_ADD64(ctx, child, "FramesDroppedDueToFilters", &stats->FramesDroppedDueToFilters, "Frames Dropped Due to Filters"); BGE_SYSCTL_STAT_ADD64(ctx, child, "DmaWriteQueueFull", &stats->DmaWriteQueueFull, "NIC DMA Write Queue Full"); BGE_SYSCTL_STAT_ADD64(ctx, child, "DmaWriteHighPriQueueFull", &stats->DmaWriteHighPriQueueFull, "NIC DMA Write High Priority Queue Full"); BGE_SYSCTL_STAT_ADD64(ctx, child, "NoMoreRxBDs", &stats->NoMoreRxBDs, "NIC No More RX Buffer Descriptors"); BGE_SYSCTL_STAT_ADD64(ctx, child, "InputDiscards", &stats->InputDiscards, "Discarded Input Frames"); BGE_SYSCTL_STAT_ADD64(ctx, child, "InputErrors", &stats->InputErrors, "Input Errors"); BGE_SYSCTL_STAT_ADD64(ctx, child, "RecvThresholdHit", &stats->RecvThresholdHit, "NIC Recv Threshold Hit"); tree = SYSCTL_ADD_NODE(ctx, schild, OID_AUTO, "rx", CTLFLAG_RD, NULL, "BGE RX Statistics"); child = SYSCTL_CHILDREN(tree); BGE_SYSCTL_STAT_ADD64(ctx, child, "ifHCInOctets", &stats->ifHCInOctets, "Inbound Octets"); BGE_SYSCTL_STAT_ADD64(ctx, child, "Fragments", &stats->etherStatsFragments, "Fragments"); BGE_SYSCTL_STAT_ADD64(ctx, child, "UnicastPkts", &stats->ifHCInUcastPkts, "Inbound Unicast Packets"); BGE_SYSCTL_STAT_ADD64(ctx, child, "MulticastPkts", &stats->ifHCInMulticastPkts, "Inbound Multicast Packets"); BGE_SYSCTL_STAT_ADD64(ctx, child, "BroadcastPkts", &stats->ifHCInBroadcastPkts, "Inbound Broadcast Packets"); BGE_SYSCTL_STAT_ADD64(ctx, child, "FCSErrors", &stats->dot3StatsFCSErrors, "FCS Errors"); BGE_SYSCTL_STAT_ADD64(ctx, child, "AlignmentErrors", &stats->dot3StatsAlignmentErrors, "Alignment Errors"); BGE_SYSCTL_STAT_ADD64(ctx, child, "xonPauseFramesReceived", &stats->xonPauseFramesReceived, "XON Pause Frames Received"); BGE_SYSCTL_STAT_ADD64(ctx, child, "xoffPauseFramesReceived", &stats->xoffPauseFramesReceived, "XOFF Pause Frames Received"); BGE_SYSCTL_STAT_ADD64(ctx, child, "ControlFramesReceived", &stats->macControlFramesReceived, "MAC Control Frames Received"); BGE_SYSCTL_STAT_ADD64(ctx, child, "xoffStateEntered", &stats->xoffStateEntered, "XOFF State Entered"); BGE_SYSCTL_STAT_ADD64(ctx, child, "FramesTooLong", &stats->dot3StatsFramesTooLong, "Frames Too Long"); BGE_SYSCTL_STAT_ADD64(ctx, child, "Jabbers", &stats->etherStatsJabbers, "Jabbers"); BGE_SYSCTL_STAT_ADD64(ctx, child, "UndersizePkts", &stats->etherStatsUndersizePkts, "Undersized Packets"); tree = SYSCTL_ADD_NODE(ctx, schild, OID_AUTO, "tx", CTLFLAG_RD, NULL, "BGE TX Statistics"); child = SYSCTL_CHILDREN(tree); BGE_SYSCTL_STAT_ADD64(ctx, child, "ifHCOutOctets", &stats->ifHCOutOctets, "Outbound Octets"); BGE_SYSCTL_STAT_ADD64(ctx, child, "Collisions", &stats->etherStatsCollisions, "TX Collisions"); BGE_SYSCTL_STAT_ADD64(ctx, child, "XonSent", &stats->outXonSent, "XON Sent"); BGE_SYSCTL_STAT_ADD64(ctx, child, "XoffSent", &stats->outXoffSent, "XOFF Sent"); BGE_SYSCTL_STAT_ADD64(ctx, child, "InternalMacTransmitErrors", &stats->dot3StatsInternalMacTransmitErrors, "Internal MAC TX Errors"); BGE_SYSCTL_STAT_ADD64(ctx, child, "SingleCollisionFrames", &stats->dot3StatsSingleCollisionFrames, "Single Collision Frames"); BGE_SYSCTL_STAT_ADD64(ctx, child, "MultipleCollisionFrames", &stats->dot3StatsMultipleCollisionFrames, "Multiple Collision Frames"); BGE_SYSCTL_STAT_ADD64(ctx, child, "DeferredTransmissions", &stats->dot3StatsDeferredTransmissions, "Deferred Transmissions"); BGE_SYSCTL_STAT_ADD64(ctx, child, "ExcessiveCollisions", &stats->dot3StatsExcessiveCollisions, "Excessive Collisions"); BGE_SYSCTL_STAT_ADD64(ctx, child, "LateCollisions", &stats->dot3StatsLateCollisions, "Late Collisions"); BGE_SYSCTL_STAT_ADD64(ctx, child, "UnicastPkts", &stats->ifHCOutUcastPkts, "Outbound Unicast Packets"); BGE_SYSCTL_STAT_ADD64(ctx, child, "MulticastPkts", &stats->ifHCOutMulticastPkts, "Outbound Multicast Packets"); BGE_SYSCTL_STAT_ADD64(ctx, child, "BroadcastPkts", &stats->ifHCOutBroadcastPkts, "Outbound Broadcast Packets"); } #undef BGE_SYSCTL_STAT_ADD64 static int bge_sysctl_stats(SYSCTL_HANDLER_ARGS) { struct bge_softc *sc; uint32_t result; int offset; sc = (struct bge_softc *)arg1; offset = arg2; result = CSR_READ_4(sc, BGE_MEMWIN_START + BGE_STATS_BLOCK + offset + offsetof(bge_hostaddr, bge_addr_lo)); return (sysctl_handle_int(oidp, &result, 0, req)); } #ifdef BGE_REGISTER_DEBUG static int bge_sysctl_debug_info(SYSCTL_HANDLER_ARGS) { struct bge_softc *sc; uint16_t *sbdata; int error; int result; int i, j; result = -1; error = sysctl_handle_int(oidp, &result, 0, req); if (error || (req->newptr == NULL)) return (error); if (result == 1) { sc = (struct bge_softc *)arg1; sbdata = (uint16_t *)sc->bge_ldata.bge_status_block; printf("Status Block:\n"); for (i = 0x0; i < (BGE_STATUS_BLK_SZ / 4); ) { printf("%06x:", i); for (j = 0; j < 8; j++) { printf(" %04x", sbdata[i]); i += 4; } printf("\n"); } printf("Registers:\n"); for (i = 0x800; i < 0xA00; ) { printf("%06x:", i); for (j = 0; j < 8; j++) { printf(" %08x", CSR_READ_4(sc, i)); i += 4; } printf("\n"); } printf("Hardware Flags:\n"); if (BGE_IS_5755_PLUS(sc)) printf(" - 5755 Plus\n"); if (BGE_IS_575X_PLUS(sc)) printf(" - 575X Plus\n"); if (BGE_IS_5705_PLUS(sc)) printf(" - 5705 Plus\n"); if (BGE_IS_5714_FAMILY(sc)) printf(" - 5714 Family\n"); if (BGE_IS_5700_FAMILY(sc)) printf(" - 5700 Family\n"); if (sc->bge_flags & BGE_FLAG_JUMBO) printf(" - Supports Jumbo Frames\n"); if (sc->bge_flags & BGE_FLAG_PCIX) printf(" - PCI-X Bus\n"); if (sc->bge_flags & BGE_FLAG_PCIE) printf(" - PCI Express Bus\n"); if (sc->bge_phy_flags & BGE_PHY_NO_3LED) printf(" - No 3 LEDs\n"); if (sc->bge_flags & BGE_FLAG_RX_ALIGNBUG) printf(" - RX Alignment Bug\n"); } return (error); } static int bge_sysctl_reg_read(SYSCTL_HANDLER_ARGS) { struct bge_softc *sc; int error; uint16_t result; uint32_t val; result = -1; error = sysctl_handle_int(oidp, &result, 0, req); if (error || (req->newptr == NULL)) return (error); if (result < 0x8000) { sc = (struct bge_softc *)arg1; val = CSR_READ_4(sc, result); printf("reg 0x%06X = 0x%08X\n", result, val); } return (error); } static int bge_sysctl_mem_read(SYSCTL_HANDLER_ARGS) { struct bge_softc *sc; int error; uint16_t result; uint32_t val; result = -1; error = sysctl_handle_int(oidp, &result, 0, req); if (error || (req->newptr == NULL)) return (error); if (result < 0x8000) { sc = (struct bge_softc *)arg1; val = bge_readmem_ind(sc, result); printf("mem 0x%06X = 0x%08X\n", result, val); } return (error); } #endif static int bge_get_eaddr_fw(struct bge_softc *sc, uint8_t ether_addr[]) { if (sc->bge_flags & BGE_FLAG_EADDR) return (1); #ifdef __sparc64__ OF_getetheraddr(sc->bge_dev, ether_addr); return (0); #endif return (1); } static int bge_get_eaddr_mem(struct bge_softc *sc, uint8_t ether_addr[]) { uint32_t mac_addr; mac_addr = bge_readmem_ind(sc, 0x0c14); if ((mac_addr >> 16) == 0x484b) { ether_addr[0] = (uint8_t)(mac_addr >> 8); ether_addr[1] = (uint8_t)mac_addr; mac_addr = bge_readmem_ind(sc, 0x0c18); ether_addr[2] = (uint8_t)(mac_addr >> 24); ether_addr[3] = (uint8_t)(mac_addr >> 16); ether_addr[4] = (uint8_t)(mac_addr >> 8); ether_addr[5] = (uint8_t)mac_addr; return (0); } return (1); } static int bge_get_eaddr_nvram(struct bge_softc *sc, uint8_t ether_addr[]) { int mac_offset = BGE_EE_MAC_OFFSET; if (sc->bge_asicrev == BGE_ASICREV_BCM5906) mac_offset = BGE_EE_MAC_OFFSET_5906; return (bge_read_nvram(sc, ether_addr, mac_offset + 2, ETHER_ADDR_LEN)); } static int bge_get_eaddr_eeprom(struct bge_softc *sc, uint8_t ether_addr[]) { if (sc->bge_asicrev == BGE_ASICREV_BCM5906) return (1); return (bge_read_eeprom(sc, ether_addr, BGE_EE_MAC_OFFSET + 2, ETHER_ADDR_LEN)); } static int bge_get_eaddr(struct bge_softc *sc, uint8_t eaddr[]) { static const bge_eaddr_fcn_t bge_eaddr_funcs[] = { /* NOTE: Order is critical */ bge_get_eaddr_fw, bge_get_eaddr_mem, bge_get_eaddr_nvram, bge_get_eaddr_eeprom, NULL }; const bge_eaddr_fcn_t *func; for (func = bge_eaddr_funcs; *func != NULL; ++func) { if ((*func)(sc, eaddr) == 0) break; } return (*func == NULL ? ENXIO : 0); } Index: projects/largeSMP/sys/dev/bge/if_bgereg.h =================================================================== --- projects/largeSMP/sys/dev/bge/if_bgereg.h (revision 221494) +++ projects/largeSMP/sys/dev/bge/if_bgereg.h (revision 221495) @@ -1,2827 +1,2839 @@ /*- * Copyright (c) 2001 Wind River Systems * Copyright (c) 1997, 1998, 1999, 2001 * Bill Paul . All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by Bill Paul. * 4. Neither the name of the author nor the names of any co-contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY Bill Paul AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL Bill Paul OR THE VOICES IN HIS HEAD * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF * THE POSSIBILITY OF SUCH DAMAGE. * * $FreeBSD$ */ /* * BCM570x memory map. The internal memory layout varies somewhat * depending on whether or not we have external SSRAM attached. * The BCM5700 can have up to 16MB of external memory. The BCM5701 * is apparently not designed to use external SSRAM. The mappings * up to the first 4 send rings are the same for both internal and * external memory configurations. Note that mini RX ring space is * only available with external SSRAM configurations, which means * the mini RX ring is not supported on the BCM5701. * * The NIC's memory can be accessed by the host in one of 3 ways: * * 1) Indirect register access. The MEMWIN_BASEADDR and MEMWIN_DATA * registers in PCI config space can be used to read any 32-bit * address within the NIC's memory. * * 2) Memory window access. The MEMWIN_BASEADDR register in PCI config * space can be used in conjunction with the memory window in the * device register space at offset 0x8000 to read any 32K chunk * of NIC memory. * * 3) Flat mode. If the 'flat mode' bit in the PCI state register is * set, the device I/O mapping consumes 32MB of host address space, * allowing all of the registers and internal NIC memory to be * accessed directly. NIC memory addresses are offset by 0x01000000. * Flat mode consumes so much host address space that it is not * recommended. */ #define BGE_PAGE_ZERO 0x00000000 #define BGE_PAGE_ZERO_END 0x000000FF #define BGE_SEND_RING_RCB 0x00000100 #define BGE_SEND_RING_RCB_END 0x000001FF #define BGE_RX_RETURN_RING_RCB 0x00000200 #define BGE_RX_RETURN_RING_RCB_END 0x000002FF #define BGE_STATS_BLOCK 0x00000300 #define BGE_STATS_BLOCK_END 0x00000AFF #define BGE_STATUS_BLOCK 0x00000B00 #define BGE_STATUS_BLOCK_END 0x00000B4F #define BGE_SOFTWARE_GENCOMM 0x00000B50 #define BGE_SOFTWARE_GENCOMM_SIG 0x00000B54 #define BGE_SOFTWARE_GENCOMM_NICCFG 0x00000B58 #define BGE_SOFTWARE_GENCOMM_FW 0x00000B78 #define BGE_SOFTWARE_GENNCOMM_FW_LEN 0x00000B7C #define BGE_SOFTWARE_GENNCOMM_FW_DATA 0x00000B80 #define BGE_SOFTWARE_GENCOMM_END 0x00000FFF #define BGE_UNMAPPED 0x00001000 #define BGE_UNMAPPED_END 0x00001FFF #define BGE_DMA_DESCRIPTORS 0x00002000 #define BGE_DMA_DESCRIPTORS_END 0x00003FFF #define BGE_SEND_RING_5717 0x00004000 #define BGE_SEND_RING_1_TO_4 0x00004000 #define BGE_SEND_RING_1_TO_4_END 0x00005FFF /* Firmware interface */ #define BGE_FW_DRV_ALIVE 0x00000001 #define BGE_FW_PAUSE 0x00000002 /* Mappings for internal memory configuration */ #define BGE_STD_RX_RINGS 0x00006000 #define BGE_STD_RX_RINGS_END 0x00006FFF #define BGE_JUMBO_RX_RINGS 0x00007000 #define BGE_JUMBO_RX_RINGS_END 0x00007FFF #define BGE_BUFFPOOL_1 0x00008000 #define BGE_BUFFPOOL_1_END 0x0000FFFF #define BGE_BUFFPOOL_2 0x00010000 /* or expansion ROM */ #define BGE_BUFFPOOL_2_END 0x00017FFF #define BGE_BUFFPOOL_3 0x00018000 /* or expansion ROM */ #define BGE_BUFFPOOL_3_END 0x0001FFFF #define BGE_STD_RX_RINGS_5717 0x00040000 #define BGE_JUMBO_RX_RINGS_5717 0x00044400 /* Mappings for external SSRAM configurations */ #define BGE_SEND_RING_5_TO_6 0x00006000 #define BGE_SEND_RING_5_TO_6_END 0x00006FFF #define BGE_SEND_RING_7_TO_8 0x00007000 #define BGE_SEND_RING_7_TO_8_END 0x00007FFF #define BGE_SEND_RING_9_TO_16 0x00008000 #define BGE_SEND_RING_9_TO_16_END 0x0000BFFF #define BGE_EXT_STD_RX_RINGS 0x0000C000 #define BGE_EXT_STD_RX_RINGS_END 0x0000CFFF #define BGE_EXT_JUMBO_RX_RINGS 0x0000D000 #define BGE_EXT_JUMBO_RX_RINGS_END 0x0000DFFF #define BGE_MINI_RX_RINGS 0x0000E000 #define BGE_MINI_RX_RINGS_END 0x0000FFFF #define BGE_AVAIL_REGION1 0x00010000 /* or expansion ROM */ #define BGE_AVAIL_REGION1_END 0x00017FFF #define BGE_AVAIL_REGION2 0x00018000 /* or expansion ROM */ #define BGE_AVAIL_REGION2_END 0x0001FFFF #define BGE_EXT_SSRAM 0x00020000 #define BGE_EXT_SSRAM_END 0x000FFFFF /* * BCM570x register offsets. These are memory mapped registers * which can be accessed with the CSR_READ_4()/CSR_WRITE_4() macros. * Each register must be accessed using 32 bit operations. * * All registers are accessed through a 32K shared memory block. * The first group of registers are actually copies of the PCI * configuration space registers. */ /* * PCI registers defined in the PCI 2.2 spec. */ #define BGE_PCI_VID 0x00 #define BGE_PCI_DID 0x02 #define BGE_PCI_CMD 0x04 #define BGE_PCI_STS 0x06 #define BGE_PCI_REV 0x08 #define BGE_PCI_CLASS 0x09 #define BGE_PCI_CACHESZ 0x0C #define BGE_PCI_LATTIMER 0x0D #define BGE_PCI_HDRTYPE 0x0E #define BGE_PCI_BIST 0x0F #define BGE_PCI_BAR0 0x10 #define BGE_PCI_BAR1 0x14 #define BGE_PCI_SUBSYS 0x2C #define BGE_PCI_SUBVID 0x2E #define BGE_PCI_ROMBASE 0x30 #define BGE_PCI_CAPPTR 0x34 #define BGE_PCI_INTLINE 0x3C #define BGE_PCI_INTPIN 0x3D #define BGE_PCI_MINGNT 0x3E #define BGE_PCI_MAXLAT 0x3F #define BGE_PCI_PCIXCAP 0x40 #define BGE_PCI_NEXTPTR_PM 0x41 #define BGE_PCI_PCIX_CMD 0x42 #define BGE_PCI_PCIX_STS 0x44 #define BGE_PCI_PWRMGMT_CAPID 0x48 #define BGE_PCI_NEXTPTR_VPD 0x49 #define BGE_PCI_PWRMGMT_CAPS 0x4A #define BGE_PCI_PWRMGMT_CMD 0x4C #define BGE_PCI_PWRMGMT_STS 0x4D #define BGE_PCI_PWRMGMT_DATA 0x4F #define BGE_PCI_VPD_CAPID 0x50 #define BGE_PCI_NEXTPTR_MSI 0x51 #define BGE_PCI_VPD_ADDR 0x52 #define BGE_PCI_VPD_DATA 0x54 #define BGE_PCI_MSI_CAPID 0x58 #define BGE_PCI_NEXTPTR_NONE 0x59 #define BGE_PCI_MSI_CTL 0x5A #define BGE_PCI_MSI_ADDR_HI 0x5C #define BGE_PCI_MSI_ADDR_LO 0x60 #define BGE_PCI_MSI_DATA 0x64 /* * PCI Express definitions * According to * PCI Express base specification, REV. 1.0a */ /* PCI Express device control, 16bits */ #define BGE_PCIE_DEVCTL 0x08 #define BGE_PCIE_DEVCTL_MAX_READRQ_MASK 0x7000 #define BGE_PCIE_DEVCTL_MAX_READRQ_128 0x0000 #define BGE_PCIE_DEVCTL_MAX_READRQ_256 0x1000 #define BGE_PCIE_DEVCTL_MAX_READRQ_512 0x2000 #define BGE_PCIE_DEVCTL_MAX_READRQ_1024 0x3000 #define BGE_PCIE_DEVCTL_MAX_READRQ_2048 0x4000 #define BGE_PCIE_DEVCTL_MAX_READRQ_4096 0x5000 /* PCI MSI. ??? */ #define BGE_PCIE_CAPID_REG 0xD0 #define BGE_PCIE_CAPID 0x10 /* * PCI registers specific to the BCM570x family. */ #define BGE_PCI_MISC_CTL 0x68 #define BGE_PCI_DMA_RW_CTL 0x6C #define BGE_PCI_PCISTATE 0x70 #define BGE_PCI_CLKCTL 0x74 #define BGE_PCI_REG_BASEADDR 0x78 #define BGE_PCI_MEMWIN_BASEADDR 0x7C #define BGE_PCI_REG_DATA 0x80 #define BGE_PCI_MEMWIN_DATA 0x84 #define BGE_PCI_MODECTL 0x88 #define BGE_PCI_MISC_CFG 0x8C #define BGE_PCI_MISC_LOCALCTL 0x90 #define BGE_PCI_UNDI_RX_STD_PRODIDX_HI 0x98 #define BGE_PCI_UNDI_RX_STD_PRODIDX_LO 0x9C #define BGE_PCI_UNDI_RX_RTN_CONSIDX_HI 0xA0 #define BGE_PCI_UNDI_RX_RTN_CONSIDX_LO 0xA4 #define BGE_PCI_UNDI_TX_BD_PRODIDX_HI 0xA8 #define BGE_PCI_UNDI_TX_BD_PRODIDX_LO 0xAC #define BGE_PCI_ISR_MBX_HI 0xB0 #define BGE_PCI_ISR_MBX_LO 0xB4 #define BGE_PCI_PRODID_ASICREV 0xBC #define BGE_PCI_GEN2_PRODID_ASICREV 0xF4 +#define BGE_PCI_GEN15_PRODID_ASICREV 0xFC /* PCI Misc. Host control register */ #define BGE_PCIMISCCTL_CLEAR_INTA 0x00000001 #define BGE_PCIMISCCTL_MASK_PCI_INTR 0x00000002 #define BGE_PCIMISCCTL_ENDIAN_BYTESWAP 0x00000004 #define BGE_PCIMISCCTL_ENDIAN_WORDSWAP 0x00000008 #define BGE_PCIMISCCTL_PCISTATE_RW 0x00000010 #define BGE_PCIMISCCTL_CLOCKCTL_RW 0x00000020 #define BGE_PCIMISCCTL_REG_WORDSWAP 0x00000040 #define BGE_PCIMISCCTL_INDIRECT_ACCESS 0x00000080 #define BGE_PCIMISCCTL_TAGGED_STATUS 0x00000200 #define BGE_PCIMISCCTL_ASICREV 0xFFFF0000 #define BGE_PCIMISCCTL_ASICREV_SHIFT 16 #define BGE_HIF_SWAP_OPTIONS (BGE_PCIMISCCTL_ENDIAN_WORDSWAP) #if BYTE_ORDER == LITTLE_ENDIAN #define BGE_DMA_SWAP_OPTIONS \ BGE_MODECTL_WORDSWAP_NONFRAME| \ BGE_MODECTL_BYTESWAP_DATA|BGE_MODECTL_WORDSWAP_DATA #else #define BGE_DMA_SWAP_OPTIONS \ BGE_MODECTL_WORDSWAP_NONFRAME|BGE_MODECTL_BYTESWAP_NONFRAME| \ BGE_MODECTL_BYTESWAP_DATA|BGE_MODECTL_WORDSWAP_DATA #endif #define BGE_INIT \ (BGE_HIF_SWAP_OPTIONS|BGE_PCIMISCCTL_CLEAR_INTA| \ BGE_PCIMISCCTL_MASK_PCI_INTR|BGE_PCIMISCCTL_INDIRECT_ACCESS) #define BGE_CHIPID_TIGON_I 0x4000 #define BGE_CHIPID_TIGON_II 0x6000 #define BGE_CHIPID_BCM5700_A0 0x7000 #define BGE_CHIPID_BCM5700_A1 0x7001 #define BGE_CHIPID_BCM5700_B0 0x7100 #define BGE_CHIPID_BCM5700_B1 0x7101 #define BGE_CHIPID_BCM5700_B2 0x7102 #define BGE_CHIPID_BCM5700_B3 0x7103 #define BGE_CHIPID_BCM5700_ALTIMA 0x7104 #define BGE_CHIPID_BCM5700_C0 0x7200 #define BGE_CHIPID_BCM5701_A0 0x0000 /* grrrr */ #define BGE_CHIPID_BCM5701_B0 0x0100 #define BGE_CHIPID_BCM5701_B2 0x0102 #define BGE_CHIPID_BCM5701_B5 0x0105 #define BGE_CHIPID_BCM5703_A0 0x1000 #define BGE_CHIPID_BCM5703_A1 0x1001 #define BGE_CHIPID_BCM5703_A2 0x1002 #define BGE_CHIPID_BCM5703_A3 0x1003 #define BGE_CHIPID_BCM5703_B0 0x1100 #define BGE_CHIPID_BCM5704_A0 0x2000 #define BGE_CHIPID_BCM5704_A1 0x2001 #define BGE_CHIPID_BCM5704_A2 0x2002 #define BGE_CHIPID_BCM5704_A3 0x2003 #define BGE_CHIPID_BCM5704_B0 0x2100 #define BGE_CHIPID_BCM5705_A0 0x3000 #define BGE_CHIPID_BCM5705_A1 0x3001 #define BGE_CHIPID_BCM5705_A2 0x3002 #define BGE_CHIPID_BCM5705_A3 0x3003 #define BGE_CHIPID_BCM5750_A0 0x4000 #define BGE_CHIPID_BCM5750_A1 0x4001 #define BGE_CHIPID_BCM5750_A3 0x4000 #define BGE_CHIPID_BCM5750_B0 0x4100 #define BGE_CHIPID_BCM5750_B1 0x4101 #define BGE_CHIPID_BCM5750_C0 0x4200 #define BGE_CHIPID_BCM5750_C1 0x4201 #define BGE_CHIPID_BCM5750_C2 0x4202 #define BGE_CHIPID_BCM5714_A0 0x5000 #define BGE_CHIPID_BCM5752_A0 0x6000 #define BGE_CHIPID_BCM5752_A1 0x6001 #define BGE_CHIPID_BCM5752_A2 0x6002 #define BGE_CHIPID_BCM5714_B0 0x8000 #define BGE_CHIPID_BCM5714_B3 0x8003 #define BGE_CHIPID_BCM5715_A0 0x9000 #define BGE_CHIPID_BCM5715_A1 0x9001 #define BGE_CHIPID_BCM5715_A3 0x9003 #define BGE_CHIPID_BCM5755_A0 0xa000 #define BGE_CHIPID_BCM5755_A1 0xa001 #define BGE_CHIPID_BCM5755_A2 0xa002 #define BGE_CHIPID_BCM5722_A0 0xa200 #define BGE_CHIPID_BCM5754_A0 0xb000 #define BGE_CHIPID_BCM5754_A1 0xb001 #define BGE_CHIPID_BCM5754_A2 0xb002 #define BGE_CHIPID_BCM5761_A0 0x5761000 #define BGE_CHIPID_BCM5761_A1 0x5761100 #define BGE_CHIPID_BCM5784_A0 0x5784000 #define BGE_CHIPID_BCM5784_A1 0x5784100 #define BGE_CHIPID_BCM5787_A0 0xb000 #define BGE_CHIPID_BCM5787_A1 0xb001 #define BGE_CHIPID_BCM5787_A2 0xb002 #define BGE_CHIPID_BCM5906_A0 0xc000 #define BGE_CHIPID_BCM5906_A1 0xc001 #define BGE_CHIPID_BCM5906_A2 0xc002 #define BGE_CHIPID_BCM57780_A0 0x57780000 #define BGE_CHIPID_BCM57780_A1 0x57780001 #define BGE_CHIPID_BCM5717_A0 0x05717000 #define BGE_CHIPID_BCM5717_B0 0x05717100 +#define BGE_CHIPID_BCM57765_A0 0x57785000 +#define BGE_CHIPID_BCM57765_B0 0x57785100 /* shorthand one */ #define BGE_ASICREV(x) ((x) >> 12) #define BGE_ASICREV_BCM5701 0x00 #define BGE_ASICREV_BCM5703 0x01 #define BGE_ASICREV_BCM5704 0x02 #define BGE_ASICREV_BCM5705 0x03 #define BGE_ASICREV_BCM5750 0x04 #define BGE_ASICREV_BCM5714_A0 0x05 #define BGE_ASICREV_BCM5752 0x06 #define BGE_ASICREV_BCM5700 0x07 #define BGE_ASICREV_BCM5780 0x08 #define BGE_ASICREV_BCM5714 0x09 #define BGE_ASICREV_BCM5755 0x0a #define BGE_ASICREV_BCM5754 0x0b #define BGE_ASICREV_BCM5787 0x0b #define BGE_ASICREV_BCM5906 0x0c /* Should consult BGE_PCI_PRODID_ASICREV for ChipID */ #define BGE_ASICREV_USE_PRODID_REG 0x0f /* BGE_PCI_PRODID_ASICREV ASIC rev. identifiers. */ #define BGE_ASICREV_BCM5717 0x5717 #define BGE_ASICREV_BCM5761 0x5761 #define BGE_ASICREV_BCM5784 0x5784 #define BGE_ASICREV_BCM5785 0x5785 +#define BGE_ASICREV_BCM57765 0x57785 #define BGE_ASICREV_BCM57780 0x57780 /* chip revisions */ #define BGE_CHIPREV(x) ((x) >> 8) #define BGE_CHIPREV_5700_AX 0x70 #define BGE_CHIPREV_5700_BX 0x71 #define BGE_CHIPREV_5700_CX 0x72 #define BGE_CHIPREV_5701_AX 0x00 #define BGE_CHIPREV_5703_AX 0x10 #define BGE_CHIPREV_5704_AX 0x20 #define BGE_CHIPREV_5704_BX 0x21 #define BGE_CHIPREV_5750_AX 0x40 #define BGE_CHIPREV_5750_BX 0x41 /* BGE_PCI_PRODID_ASICREV chip rev. identifiers. */ #define BGE_CHIPREV_5717_AX 0x57170 #define BGE_CHIPREV_5717_BX 0x57171 #define BGE_CHIPREV_5761_AX 0x57611 #define BGE_CHIPREV_5784_AX 0x57841 /* PCI DMA Read/Write Control register */ #define BGE_PCIDMARWCTL_MINDMA 0x000000FF #define BGE_PCIDMARWCTL_DIS_CACHE_ALIGNMENT 0x00000001 #define BGE_PCIDMARWCTL_RDADRR_BNDRY 0x00000700 #define BGE_PCIDMARWCTL_WRADDR_BNDRY 0x00003800 #define BGE_PCIDMARWCTL_ONEDMA_ATONCE 0x0000C000 #define BGE_PCIDMARWCTL_ONEDMA_ATONCE_GLOBAL 0x00004000 #define BGE_PCIDMARWCTL_ONEDMA_ATONCE_LOCAL 0x00008000 #define BGE_PCIDMARWCTL_RD_WAT 0x00070000 #define BGE_PCIDMARWCTL_WR_WAT 0x00380000 #define BGE_PCIDMARWCTL_USE_MRM 0x00400000 #define BGE_PCIDMARWCTL_ASRT_ALL_BE 0x00800000 #define BGE_PCIDMARWCTL_DFLT_PCI_RD_CMD 0x0F000000 #define BGE_PCIDMARWCTL_DFLT_PCI_WR_CMD 0xF0000000 #define BGE_PCIDMARWCTL_RD_WAT_SHIFT(x) ((x) << 16) #define BGE_PCIDMARWCTL_WR_WAT_SHIFT(x) ((x) << 19) #define BGE_PCIDMARWCTL_RD_CMD_SHIFT(x) ((x) << 24) #define BGE_PCIDMARWCTL_WR_CMD_SHIFT(x) ((x) << 28) +#define BGE_PCIDMARWCTL_CRDRDR_RDMA_MRRS_MSK 0x00000380 + #define BGE_PCI_READ_BNDRY_DISABLE 0x00000000 #define BGE_PCI_READ_BNDRY_16BYTES 0x00000100 #define BGE_PCI_READ_BNDRY_32BYTES 0x00000200 #define BGE_PCI_READ_BNDRY_64BYTES 0x00000300 #define BGE_PCI_READ_BNDRY_128BYTES 0x00000400 #define BGE_PCI_READ_BNDRY_256BYTES 0x00000500 #define BGE_PCI_READ_BNDRY_512BYTES 0x00000600 #define BGE_PCI_READ_BNDRY_1024BYTES 0x00000700 #define BGE_PCI_WRITE_BNDRY_DISABLE 0x00000000 #define BGE_PCI_WRITE_BNDRY_16BYTES 0x00000800 #define BGE_PCI_WRITE_BNDRY_32BYTES 0x00001000 #define BGE_PCI_WRITE_BNDRY_64BYTES 0x00001800 #define BGE_PCI_WRITE_BNDRY_128BYTES 0x00002000 #define BGE_PCI_WRITE_BNDRY_256BYTES 0x00002800 #define BGE_PCI_WRITE_BNDRY_512BYTES 0x00003000 #define BGE_PCI_WRITE_BNDRY_1024BYTES 0x00003800 /* * PCI state register -- note, this register is read only * unless the PCISTATE_WR bit of the PCI Misc. Host Control * register is set. */ #define BGE_PCISTATE_FORCE_RESET 0x00000001 #define BGE_PCISTATE_INTR_STATE 0x00000002 #define BGE_PCISTATE_PCI_BUSMODE 0x00000004 /* 1 = PCI, 0 = PCI-X */ #define BGE_PCISTATE_PCI_BUSSPEED 0x00000008 /* 1 = 66/133, 0 = 33/66 */ #define BGE_PCISTATE_32BIT_BUS 0x00000010 /* 1 = 32bit, 0 = 64bit */ #define BGE_PCISTATE_WANT_EXPROM 0x00000020 #define BGE_PCISTATE_EXPROM_RETRY 0x00000040 #define BGE_PCISTATE_FLATVIEW_MODE 0x00000100 #define BGE_PCISTATE_PCI_TGT_RETRY_MAX 0x00000E00 /* * PCI Clock Control register -- note, this register is read only * unless the CLOCKCTL_RW bit of the PCI Misc. Host Control * register is set. */ #define BGE_PCICLOCKCTL_DETECTED_SPEED 0x0000000F #define BGE_PCICLOCKCTL_M66EN 0x00000080 #define BGE_PCICLOCKCTL_LOWPWR_CLKMODE 0x00000200 #define BGE_PCICLOCKCTL_RXCPU_CLK_DIS 0x00000400 #define BGE_PCICLOCKCTL_TXCPU_CLK_DIS 0x00000800 #define BGE_PCICLOCKCTL_ALTCLK 0x00001000 #define BGE_PCICLOCKCTL_ALTCLK_SRC 0x00002000 #define BGE_PCICLOCKCTL_PCIPLL_DISABLE 0x00004000 #define BGE_PCICLOCKCTL_SYSPLL_DISABLE 0x00008000 #define BGE_PCICLOCKCTL_BIST_ENABLE 0x00010000 #ifndef PCIM_CMD_MWIEN #define PCIM_CMD_MWIEN 0x0010 #endif #ifndef PCIM_CMD_INTxDIS #define PCIM_CMD_INTxDIS 0x0400 #endif /* * High priority mailbox registers * Each mailbox is 64-bits wide, though we only use the * lower 32 bits. To write a 64-bit value, write the upper 32 bits * first. The NIC will load the mailbox after the lower 32 bit word * has been updated. */ #define BGE_MBX_IRQ0_HI 0x0200 #define BGE_MBX_IRQ0_LO 0x0204 #define BGE_MBX_IRQ1_HI 0x0208 #define BGE_MBX_IRQ1_LO 0x020C #define BGE_MBX_IRQ2_HI 0x0210 #define BGE_MBX_IRQ2_LO 0x0214 #define BGE_MBX_IRQ3_HI 0x0218 #define BGE_MBX_IRQ3_LO 0x021C #define BGE_MBX_GEN0_HI 0x0220 #define BGE_MBX_GEN0_LO 0x0224 #define BGE_MBX_GEN1_HI 0x0228 #define BGE_MBX_GEN1_LO 0x022C #define BGE_MBX_GEN2_HI 0x0230 #define BGE_MBX_GEN2_LO 0x0234 #define BGE_MBX_GEN3_HI 0x0228 #define BGE_MBX_GEN3_LO 0x022C #define BGE_MBX_GEN4_HI 0x0240 #define BGE_MBX_GEN4_LO 0x0244 #define BGE_MBX_GEN5_HI 0x0248 #define BGE_MBX_GEN5_LO 0x024C #define BGE_MBX_GEN6_HI 0x0250 #define BGE_MBX_GEN6_LO 0x0254 #define BGE_MBX_GEN7_HI 0x0258 #define BGE_MBX_GEN7_LO 0x025C #define BGE_MBX_RELOAD_STATS_HI 0x0260 #define BGE_MBX_RELOAD_STATS_LO 0x0264 #define BGE_MBX_RX_STD_PROD_HI 0x0268 #define BGE_MBX_RX_STD_PROD_LO 0x026C #define BGE_MBX_RX_JUMBO_PROD_HI 0x0270 #define BGE_MBX_RX_JUMBO_PROD_LO 0x0274 #define BGE_MBX_RX_MINI_PROD_HI 0x0278 #define BGE_MBX_RX_MINI_PROD_LO 0x027C #define BGE_MBX_RX_CONS0_HI 0x0280 #define BGE_MBX_RX_CONS0_LO 0x0284 #define BGE_MBX_RX_CONS1_HI 0x0288 #define BGE_MBX_RX_CONS1_LO 0x028C #define BGE_MBX_RX_CONS2_HI 0x0290 #define BGE_MBX_RX_CONS2_LO 0x0294 #define BGE_MBX_RX_CONS3_HI 0x0298 #define BGE_MBX_RX_CONS3_LO 0x029C #define BGE_MBX_RX_CONS4_HI 0x02A0 #define BGE_MBX_RX_CONS4_LO 0x02A4 #define BGE_MBX_RX_CONS5_HI 0x02A8 #define BGE_MBX_RX_CONS5_LO 0x02AC #define BGE_MBX_RX_CONS6_HI 0x02B0 #define BGE_MBX_RX_CONS6_LO 0x02B4 #define BGE_MBX_RX_CONS7_HI 0x02B8 #define BGE_MBX_RX_CONS7_LO 0x02BC #define BGE_MBX_RX_CONS8_HI 0x02C0 #define BGE_MBX_RX_CONS8_LO 0x02C4 #define BGE_MBX_RX_CONS9_HI 0x02C8 #define BGE_MBX_RX_CONS9_LO 0x02CC #define BGE_MBX_RX_CONS10_HI 0x02D0 #define BGE_MBX_RX_CONS10_LO 0x02D4 #define BGE_MBX_RX_CONS11_HI 0x02D8 #define BGE_MBX_RX_CONS11_LO 0x02DC #define BGE_MBX_RX_CONS12_HI 0x02E0 #define BGE_MBX_RX_CONS12_LO 0x02E4 #define BGE_MBX_RX_CONS13_HI 0x02E8 #define BGE_MBX_RX_CONS13_LO 0x02EC #define BGE_MBX_RX_CONS14_HI 0x02F0 #define BGE_MBX_RX_CONS14_LO 0x02F4 #define BGE_MBX_RX_CONS15_HI 0x02F8 #define BGE_MBX_RX_CONS15_LO 0x02FC #define BGE_MBX_TX_HOST_PROD0_HI 0x0300 #define BGE_MBX_TX_HOST_PROD0_LO 0x0304 #define BGE_MBX_TX_HOST_PROD1_HI 0x0308 #define BGE_MBX_TX_HOST_PROD1_LO 0x030C #define BGE_MBX_TX_HOST_PROD2_HI 0x0310 #define BGE_MBX_TX_HOST_PROD2_LO 0x0314 #define BGE_MBX_TX_HOST_PROD3_HI 0x0318 #define BGE_MBX_TX_HOST_PROD3_LO 0x031C #define BGE_MBX_TX_HOST_PROD4_HI 0x0320 #define BGE_MBX_TX_HOST_PROD4_LO 0x0324 #define BGE_MBX_TX_HOST_PROD5_HI 0x0328 #define BGE_MBX_TX_HOST_PROD5_LO 0x032C #define BGE_MBX_TX_HOST_PROD6_HI 0x0330 #define BGE_MBX_TX_HOST_PROD6_LO 0x0334 #define BGE_MBX_TX_HOST_PROD7_HI 0x0338 #define BGE_MBX_TX_HOST_PROD7_LO 0x033C #define BGE_MBX_TX_HOST_PROD8_HI 0x0340 #define BGE_MBX_TX_HOST_PROD8_LO 0x0344 #define BGE_MBX_TX_HOST_PROD9_HI 0x0348 #define BGE_MBX_TX_HOST_PROD9_LO 0x034C #define BGE_MBX_TX_HOST_PROD10_HI 0x0350 #define BGE_MBX_TX_HOST_PROD10_LO 0x0354 #define BGE_MBX_TX_HOST_PROD11_HI 0x0358 #define BGE_MBX_TX_HOST_PROD11_LO 0x035C #define BGE_MBX_TX_HOST_PROD12_HI 0x0360 #define BGE_MBX_TX_HOST_PROD12_LO 0x0364 #define BGE_MBX_TX_HOST_PROD13_HI 0x0368 #define BGE_MBX_TX_HOST_PROD13_LO 0x036C #define BGE_MBX_TX_HOST_PROD14_HI 0x0370 #define BGE_MBX_TX_HOST_PROD14_LO 0x0374 #define BGE_MBX_TX_HOST_PROD15_HI 0x0378 #define BGE_MBX_TX_HOST_PROD15_LO 0x037C #define BGE_MBX_TX_NIC_PROD0_HI 0x0380 #define BGE_MBX_TX_NIC_PROD0_LO 0x0384 #define BGE_MBX_TX_NIC_PROD1_HI 0x0388 #define BGE_MBX_TX_NIC_PROD1_LO 0x038C #define BGE_MBX_TX_NIC_PROD2_HI 0x0390 #define BGE_MBX_TX_NIC_PROD2_LO 0x0394 #define BGE_MBX_TX_NIC_PROD3_HI 0x0398 #define BGE_MBX_TX_NIC_PROD3_LO 0x039C #define BGE_MBX_TX_NIC_PROD4_HI 0x03A0 #define BGE_MBX_TX_NIC_PROD4_LO 0x03A4 #define BGE_MBX_TX_NIC_PROD5_HI 0x03A8 #define BGE_MBX_TX_NIC_PROD5_LO 0x03AC #define BGE_MBX_TX_NIC_PROD6_HI 0x03B0 #define BGE_MBX_TX_NIC_PROD6_LO 0x03B4 #define BGE_MBX_TX_NIC_PROD7_HI 0x03B8 #define BGE_MBX_TX_NIC_PROD7_LO 0x03BC #define BGE_MBX_TX_NIC_PROD8_HI 0x03C0 #define BGE_MBX_TX_NIC_PROD8_LO 0x03C4 #define BGE_MBX_TX_NIC_PROD9_HI 0x03C8 #define BGE_MBX_TX_NIC_PROD9_LO 0x03CC #define BGE_MBX_TX_NIC_PROD10_HI 0x03D0 #define BGE_MBX_TX_NIC_PROD10_LO 0x03D4 #define BGE_MBX_TX_NIC_PROD11_HI 0x03D8 #define BGE_MBX_TX_NIC_PROD11_LO 0x03DC #define BGE_MBX_TX_NIC_PROD12_HI 0x03E0 #define BGE_MBX_TX_NIC_PROD12_LO 0x03E4 #define BGE_MBX_TX_NIC_PROD13_HI 0x03E8 #define BGE_MBX_TX_NIC_PROD13_LO 0x03EC #define BGE_MBX_TX_NIC_PROD14_HI 0x03F0 #define BGE_MBX_TX_NIC_PROD14_LO 0x03F4 #define BGE_MBX_TX_NIC_PROD15_HI 0x03F8 #define BGE_MBX_TX_NIC_PROD15_LO 0x03FC #define BGE_TX_RINGS_MAX 4 #define BGE_TX_RINGS_EXTSSRAM_MAX 16 #define BGE_RX_RINGS_MAX 16 #define BGE_RX_RINGS_MAX_5717 17 /* Ethernet MAC control registers */ #define BGE_MAC_MODE 0x0400 #define BGE_MAC_STS 0x0404 #define BGE_MAC_EVT_ENB 0x0408 #define BGE_MAC_LED_CTL 0x040C #define BGE_MAC_ADDR1_LO 0x0410 #define BGE_MAC_ADDR1_HI 0x0414 #define BGE_MAC_ADDR2_LO 0x0418 #define BGE_MAC_ADDR2_HI 0x041C #define BGE_MAC_ADDR3_LO 0x0420 #define BGE_MAC_ADDR3_HI 0x0424 #define BGE_MAC_ADDR4_LO 0x0428 #define BGE_MAC_ADDR4_HI 0x042C #define BGE_WOL_PATPTR 0x0430 #define BGE_WOL_PATCFG 0x0434 #define BGE_TX_RANDOM_BACKOFF 0x0438 #define BGE_RX_MTU 0x043C #define BGE_GBIT_PCS_TEST 0x0440 #define BGE_TX_TBI_AUTONEG 0x0444 #define BGE_RX_TBI_AUTONEG 0x0448 #define BGE_MI_COMM 0x044C #define BGE_MI_STS 0x0450 #define BGE_MI_MODE 0x0454 #define BGE_AUTOPOLL_STS 0x0458 #define BGE_TX_MODE 0x045C #define BGE_TX_STS 0x0460 #define BGE_TX_LENGTHS 0x0464 #define BGE_RX_MODE 0x0468 #define BGE_RX_STS 0x046C #define BGE_MAR0 0x0470 #define BGE_MAR1 0x0474 #define BGE_MAR2 0x0478 #define BGE_MAR3 0x047C #define BGE_RX_BD_RULES_CTL0 0x0480 #define BGE_RX_BD_RULES_MASKVAL0 0x0484 #define BGE_RX_BD_RULES_CTL1 0x0488 #define BGE_RX_BD_RULES_MASKVAL1 0x048C #define BGE_RX_BD_RULES_CTL2 0x0490 #define BGE_RX_BD_RULES_MASKVAL2 0x0494 #define BGE_RX_BD_RULES_CTL3 0x0498 #define BGE_RX_BD_RULES_MASKVAL3 0x049C #define BGE_RX_BD_RULES_CTL4 0x04A0 #define BGE_RX_BD_RULES_MASKVAL4 0x04A4 #define BGE_RX_BD_RULES_CTL5 0x04A8 #define BGE_RX_BD_RULES_MASKVAL5 0x04AC #define BGE_RX_BD_RULES_CTL6 0x04B0 #define BGE_RX_BD_RULES_MASKVAL6 0x04B4 #define BGE_RX_BD_RULES_CTL7 0x04B8 #define BGE_RX_BD_RULES_MASKVAL7 0x04BC #define BGE_RX_BD_RULES_CTL8 0x04C0 #define BGE_RX_BD_RULES_MASKVAL8 0x04C4 #define BGE_RX_BD_RULES_CTL9 0x04C8 #define BGE_RX_BD_RULES_MASKVAL9 0x04CC #define BGE_RX_BD_RULES_CTL10 0x04D0 #define BGE_RX_BD_RULES_MASKVAL10 0x04D4 #define BGE_RX_BD_RULES_CTL11 0x04D8 #define BGE_RX_BD_RULES_MASKVAL11 0x04DC #define BGE_RX_BD_RULES_CTL12 0x04E0 #define BGE_RX_BD_RULES_MASKVAL12 0x04E4 #define BGE_RX_BD_RULES_CTL13 0x04E8 #define BGE_RX_BD_RULES_MASKVAL13 0x04EC #define BGE_RX_BD_RULES_CTL14 0x04F0 #define BGE_RX_BD_RULES_MASKVAL14 0x04F4 #define BGE_RX_BD_RULES_CTL15 0x04F8 #define BGE_RX_BD_RULES_MASKVAL15 0x04FC #define BGE_RX_RULES_CFG 0x0500 #define BGE_MAX_RX_FRAME_LOWAT 0x0504 #define BGE_SERDES_CFG 0x0590 #define BGE_SERDES_STS 0x0594 #define BGE_SGDIG_CFG 0x05B0 #define BGE_SGDIG_STS 0x05B4 #define BGE_TX_MAC_STATS_OCTETS 0x0800 #define BGE_TX_MAC_STATS_RESERVE_0 0x0804 #define BGE_TX_MAC_STATS_COLLS 0x0808 #define BGE_TX_MAC_STATS_XON_SENT 0x080C #define BGE_TX_MAC_STATS_XOFF_SENT 0x0810 #define BGE_TX_MAC_STATS_RESERVE_1 0x0814 #define BGE_TX_MAC_STATS_ERRORS 0x0818 #define BGE_TX_MAC_STATS_SINGLE_COLL 0x081C #define BGE_TX_MAC_STATS_MULTI_COLL 0x0820 #define BGE_TX_MAC_STATS_DEFERRED 0x0824 #define BGE_TX_MAC_STATS_RESERVE_2 0x0828 #define BGE_TX_MAC_STATS_EXCESS_COLL 0x082C #define BGE_TX_MAC_STATS_LATE_COLL 0x0830 #define BGE_TX_MAC_STATS_RESERVE_3 0x0834 #define BGE_TX_MAC_STATS_RESERVE_4 0x0838 #define BGE_TX_MAC_STATS_RESERVE_5 0x083C #define BGE_TX_MAC_STATS_RESERVE_6 0x0840 #define BGE_TX_MAC_STATS_RESERVE_7 0x0844 #define BGE_TX_MAC_STATS_RESERVE_8 0x0848 #define BGE_TX_MAC_STATS_RESERVE_9 0x084C #define BGE_TX_MAC_STATS_RESERVE_10 0x0850 #define BGE_TX_MAC_STATS_RESERVE_11 0x0854 #define BGE_TX_MAC_STATS_RESERVE_12 0x0858 #define BGE_TX_MAC_STATS_RESERVE_13 0x085C #define BGE_TX_MAC_STATS_RESERVE_14 0x0860 #define BGE_TX_MAC_STATS_RESERVE_15 0x0864 #define BGE_TX_MAC_STATS_RESERVE_16 0x0868 #define BGE_TX_MAC_STATS_UCAST 0x086C #define BGE_TX_MAC_STATS_MCAST 0x0870 #define BGE_TX_MAC_STATS_BCAST 0x0874 #define BGE_TX_MAC_STATS_RESERVE_17 0x0878 #define BGE_TX_MAC_STATS_RESERVE_18 0x087C #define BGE_RX_MAC_STATS_OCTESTS 0x0880 #define BGE_RX_MAC_STATS_RESERVE_0 0x0884 #define BGE_RX_MAC_STATS_FRAGMENTS 0x0888 #define BGE_RX_MAC_STATS_UCAST 0x088C #define BGE_RX_MAC_STATS_MCAST 0x0890 #define BGE_RX_MAC_STATS_BCAST 0x0894 #define BGE_RX_MAC_STATS_FCS_ERRORS 0x0898 #define BGE_RX_MAC_STATS_ALGIN_ERRORS 0x089C #define BGE_RX_MAC_STATS_XON_RCVD 0x08A0 #define BGE_RX_MAC_STATS_XOFF_RCVD 0x08A4 #define BGE_RX_MAC_STATS_CTRL_RCVD 0x08A8 #define BGE_RX_MAC_STATS_XOFF_ENTERED 0x08AC #define BGE_RX_MAC_STATS_FRAME_TOO_LONG 0x08B0 #define BGE_RX_MAC_STATS_JABBERS 0x08B4 #define BGE_RX_MAC_STATS_UNDERSIZE 0x08B8 /* Ethernet MAC Mode register */ #define BGE_MACMODE_RESET 0x00000001 #define BGE_MACMODE_HALF_DUPLEX 0x00000002 #define BGE_MACMODE_PORTMODE 0x0000000C #define BGE_MACMODE_LOOPBACK 0x00000010 #define BGE_MACMODE_RX_TAGGEDPKT 0x00000080 #define BGE_MACMODE_TX_BURST_ENB 0x00000100 #define BGE_MACMODE_MAX_DEFER 0x00000200 #define BGE_MACMODE_LINK_POLARITY 0x00000400 #define BGE_MACMODE_RX_STATS_ENB 0x00000800 #define BGE_MACMODE_RX_STATS_CLEAR 0x00001000 #define BGE_MACMODE_RX_STATS_FLUSH 0x00002000 #define BGE_MACMODE_TX_STATS_ENB 0x00004000 #define BGE_MACMODE_TX_STATS_CLEAR 0x00008000 #define BGE_MACMODE_TX_STATS_FLUSH 0x00010000 #define BGE_MACMODE_TBI_SEND_CFGS 0x00020000 #define BGE_MACMODE_MAGIC_PKT_ENB 0x00040000 #define BGE_MACMODE_ACPI_PWRON_ENB 0x00080000 #define BGE_MACMODE_MIP_ENB 0x00100000 #define BGE_MACMODE_TXDMA_ENB 0x00200000 #define BGE_MACMODE_RXDMA_ENB 0x00400000 #define BGE_MACMODE_FRMHDR_DMA_ENB 0x00800000 #define BGE_PORTMODE_NONE 0x00000000 #define BGE_PORTMODE_MII 0x00000004 #define BGE_PORTMODE_GMII 0x00000008 #define BGE_PORTMODE_TBI 0x0000000C /* MAC Status register */ #define BGE_MACSTAT_TBI_PCS_SYNCHED 0x00000001 #define BGE_MACSTAT_TBI_SIGNAL_DETECT 0x00000002 #define BGE_MACSTAT_RX_CFG 0x00000004 #define BGE_MACSTAT_CFG_CHANGED 0x00000008 #define BGE_MACSTAT_SYNC_CHANGED 0x00000010 #define BGE_MACSTAT_PORT_DECODE_ERROR 0x00000400 #define BGE_MACSTAT_LINK_CHANGED 0x00001000 #define BGE_MACSTAT_MI_COMPLETE 0x00400000 #define BGE_MACSTAT_MI_INTERRUPT 0x00800000 #define BGE_MACSTAT_AUTOPOLL_ERROR 0x01000000 #define BGE_MACSTAT_ODI_ERROR 0x02000000 #define BGE_MACSTAT_RXSTAT_OFLOW 0x04000000 #define BGE_MACSTAT_TXSTAT_OFLOW 0x08000000 /* MAC Event Enable Register */ #define BGE_EVTENB_PORT_DECODE_ERROR 0x00000400 #define BGE_EVTENB_LINK_CHANGED 0x00001000 #define BGE_EVTENB_MI_COMPLETE 0x00400000 #define BGE_EVTENB_MI_INTERRUPT 0x00800000 #define BGE_EVTENB_AUTOPOLL_ERROR 0x01000000 #define BGE_EVTENB_ODI_ERROR 0x02000000 #define BGE_EVTENB_RXSTAT_OFLOW 0x04000000 #define BGE_EVTENB_TXSTAT_OFLOW 0x08000000 /* LED Control Register */ #define BGE_LEDCTL_LINKLED_OVERRIDE 0x00000001 #define BGE_LEDCTL_1000MBPS_LED 0x00000002 #define BGE_LEDCTL_100MBPS_LED 0x00000004 #define BGE_LEDCTL_10MBPS_LED 0x00000008 #define BGE_LEDCTL_TRAFLED_OVERRIDE 0x00000010 #define BGE_LEDCTL_TRAFLED_BLINK 0x00000020 #define BGE_LEDCTL_TREFLED_BLINK_2 0x00000040 #define BGE_LEDCTL_1000MBPS_STS 0x00000080 #define BGE_LEDCTL_100MBPS_STS 0x00000100 #define BGE_LEDCTL_10MBPS_STS 0x00000200 #define BGE_LEDCTL_TRADLED_STS 0x00000400 #define BGE_LEDCTL_BLINKPERIOD 0x7FF80000 #define BGE_LEDCTL_BLINKPERIOD_OVERRIDE 0x80000000 /* TX backoff seed register */ #define BGE_TX_BACKOFF_SEED_MASK 0x3F /* Autopoll status register */ #define BGE_AUTOPOLLSTS_ERROR 0x00000001 /* Transmit MAC mode register */ #define BGE_TXMODE_RESET 0x00000001 #define BGE_TXMODE_ENABLE 0x00000002 #define BGE_TXMODE_FLOWCTL_ENABLE 0x00000010 #define BGE_TXMODE_BIGBACKOFF_ENABLE 0x00000020 #define BGE_TXMODE_LONGPAUSE_ENABLE 0x00000040 #define BGE_TXMODE_MBUF_LOCKUP_FIX 0x00000100 /* Transmit MAC status register */ #define BGE_TXSTAT_RX_XOFFED 0x00000001 #define BGE_TXSTAT_SENT_XOFF 0x00000002 #define BGE_TXSTAT_SENT_XON 0x00000004 #define BGE_TXSTAT_LINK_UP 0x00000008 #define BGE_TXSTAT_ODI_UFLOW 0x00000010 #define BGE_TXSTAT_ODI_OFLOW 0x00000020 /* Transmit MAC lengths register */ #define BGE_TXLEN_SLOTTIME 0x000000FF #define BGE_TXLEN_IPG 0x00000F00 #define BGE_TXLEN_CRS 0x00003000 /* Receive MAC mode register */ #define BGE_RXMODE_RESET 0x00000001 #define BGE_RXMODE_ENABLE 0x00000002 #define BGE_RXMODE_FLOWCTL_ENABLE 0x00000004 #define BGE_RXMODE_RX_GIANTS 0x00000020 #define BGE_RXMODE_RX_RUNTS 0x00000040 #define BGE_RXMODE_8022_LENCHECK 0x00000080 #define BGE_RXMODE_RX_PROMISC 0x00000100 #define BGE_RXMODE_RX_NO_CRC_CHECK 0x00000200 #define BGE_RXMODE_RX_KEEP_VLAN_DIAG 0x00000400 /* Receive MAC status register */ #define BGE_RXSTAT_REMOTE_XOFFED 0x00000001 #define BGE_RXSTAT_RCVD_XOFF 0x00000002 #define BGE_RXSTAT_RCVD_XON 0x00000004 /* Receive Rules Control register */ #define BGE_RXRULECTL_OFFSET 0x000000FF #define BGE_RXRULECTL_CLASS 0x00001F00 #define BGE_RXRULECTL_HDRTYPE 0x0000E000 #define BGE_RXRULECTL_COMPARE_OP 0x00030000 #define BGE_RXRULECTL_MAP 0x01000000 #define BGE_RXRULECTL_DISCARD 0x02000000 #define BGE_RXRULECTL_MASK 0x04000000 #define BGE_RXRULECTL_ACTIVATE_PROC3 0x08000000 #define BGE_RXRULECTL_ACTIVATE_PROC2 0x10000000 #define BGE_RXRULECTL_ACTIVATE_PROC1 0x20000000 #define BGE_RXRULECTL_ANDWITHNEXT 0x40000000 /* Receive Rules Mask register */ #define BGE_RXRULEMASK_VALUE 0x0000FFFF #define BGE_RXRULEMASK_MASKVAL 0xFFFF0000 /* SERDES configuration register */ #define BGE_SERDESCFG_RXR 0x00000007 /* phase interpolator */ #define BGE_SERDESCFG_RXG 0x00000018 /* rx gain setting */ #define BGE_SERDESCFG_RXEDGESEL 0x00000040 /* rising/falling egde */ #define BGE_SERDESCFG_TX_BIAS 0x00000380 /* TXDAC bias setting */ #define BGE_SERDESCFG_IBMAX 0x00000400 /* bias current +25% */ #define BGE_SERDESCFG_IBMIN 0x00000800 /* bias current -25% */ #define BGE_SERDESCFG_TXMODE 0x00001000 #define BGE_SERDESCFG_TXEDGESEL 0x00002000 /* rising/falling edge */ #define BGE_SERDESCFG_MODE 0x00004000 /* TXCP/TXCN disabled */ #define BGE_SERDESCFG_PLLTEST 0x00008000 /* PLL test mode */ #define BGE_SERDESCFG_CDET 0x00010000 /* comma detect enable */ #define BGE_SERDESCFG_TBILOOP 0x00020000 /* local loopback */ #define BGE_SERDESCFG_REMLOOP 0x00040000 /* remote loopback */ #define BGE_SERDESCFG_INVPHASE 0x00080000 /* Reverse 125Mhz clock */ #define BGE_SERDESCFG_12REGCTL 0x00300000 /* 1.2v regulator ctl */ #define BGE_SERDESCFG_REGCTL 0x00C00000 /* regulator ctl (2.5v) */ /* SERDES status register */ #define BGE_SERDESSTS_RXSTAT 0x0000000F /* receive status bits */ #define BGE_SERDESSTS_CDET 0x00000010 /* comma code detected */ /* SGDIG config (not documented) */ #define BGE_SGDIGCFG_PAUSE_CAP 0x00000800 #define BGE_SGDIGCFG_ASYM_PAUSE 0x00001000 #define BGE_SGDIGCFG_SEND 0x40000000 #define BGE_SGDIGCFG_AUTO 0x80000000 /* SGDIG status (not documented) */ #define BGE_SGDIGSTS_DONE 0x00000002 #define BGE_SGDIGSTS_IS_SERDES 0x00000100 #define BGE_SGDIGSTS_PAUSE_CAP 0x00080000 #define BGE_SGDIGSTS_ASYM_PAUSE 0x00100000 /* MI communication register */ #define BGE_MICOMM_DATA 0x0000FFFF #define BGE_MICOMM_REG 0x001F0000 #define BGE_MICOMM_PHY 0x03E00000 #define BGE_MICOMM_CMD 0x0C000000 #define BGE_MICOMM_READFAIL 0x10000000 #define BGE_MICOMM_BUSY 0x20000000 #define BGE_MIREG(x) ((x & 0x1F) << 16) #define BGE_MIPHY(x) ((x & 0x1F) << 21) #define BGE_MICMD_WRITE 0x04000000 #define BGE_MICMD_READ 0x08000000 /* MI status register */ #define BGE_MISTS_LINK 0x00000001 #define BGE_MISTS_10MBPS 0x00000002 #define BGE_MIMODE_CLK_10MHZ 0x00000001 #define BGE_MIMODE_SHORTPREAMBLE 0x00000002 #define BGE_MIMODE_AUTOPOLL 0x00000010 #define BGE_MIMODE_CLKCNT 0x001F0000 #define BGE_MIMODE_500KHZ_CONST 0x00008000 #define BGE_MIMODE_BASE 0x000C0000 /* * Send data initiator control registers. */ #define BGE_SDI_MODE 0x0C00 #define BGE_SDI_STATUS 0x0C04 #define BGE_SDI_STATS_CTL 0x0C08 #define BGE_SDI_STATS_ENABLE_MASK 0x0C0C #define BGE_SDI_STATS_INCREMENT_MASK 0x0C10 #define BGE_ISO_PKT_TX 0x0C20 #define BGE_LOCSTATS_COS0 0x0C80 #define BGE_LOCSTATS_COS1 0x0C84 #define BGE_LOCSTATS_COS2 0x0C88 #define BGE_LOCSTATS_COS3 0x0C8C #define BGE_LOCSTATS_COS4 0x0C90 #define BGE_LOCSTATS_COS5 0x0C84 #define BGE_LOCSTATS_COS6 0x0C98 #define BGE_LOCSTATS_COS7 0x0C9C #define BGE_LOCSTATS_COS8 0x0CA0 #define BGE_LOCSTATS_COS9 0x0CA4 #define BGE_LOCSTATS_COS10 0x0CA8 #define BGE_LOCSTATS_COS11 0x0CAC #define BGE_LOCSTATS_COS12 0x0CB0 #define BGE_LOCSTATS_COS13 0x0CB4 #define BGE_LOCSTATS_COS14 0x0CB8 #define BGE_LOCSTATS_COS15 0x0CBC #define BGE_LOCSTATS_DMA_RQ_FULL 0x0CC0 #define BGE_LOCSTATS_DMA_HIPRIO_RQ_FULL 0x0CC4 #define BGE_LOCSTATS_SDC_QUEUE_FULL 0x0CC8 #define BGE_LOCSTATS_NIC_SENDPROD_SET 0x0CCC #define BGE_LOCSTATS_STATS_UPDATED 0x0CD0 #define BGE_LOCSTATS_IRQS 0x0CD4 #define BGE_LOCSTATS_AVOIDED_IRQS 0x0CD8 #define BGE_LOCSTATS_TX_THRESH_HIT 0x0CDC /* Send Data Initiator mode register */ #define BGE_SDIMODE_RESET 0x00000001 #define BGE_SDIMODE_ENABLE 0x00000002 #define BGE_SDIMODE_STATS_OFLOW_ATTN 0x00000004 #define BGE_SDIMODE_HW_LSO_PRE_DMA 0x00000008 /* Send Data Initiator stats register */ #define BGE_SDISTAT_STATS_OFLOW_ATTN 0x00000004 /* Send Data Initiator stats control register */ #define BGE_SDISTATSCTL_ENABLE 0x00000001 #define BGE_SDISTATSCTL_FASTER 0x00000002 #define BGE_SDISTATSCTL_CLEAR 0x00000004 #define BGE_SDISTATSCTL_FORCEFLUSH 0x00000008 #define BGE_SDISTATSCTL_FORCEZERO 0x00000010 /* * Send Data Completion Control registers */ #define BGE_SDC_MODE 0x1000 #define BGE_SDC_STATUS 0x1004 /* Send Data completion mode register */ #define BGE_SDCMODE_RESET 0x00000001 #define BGE_SDCMODE_ENABLE 0x00000002 #define BGE_SDCMODE_ATTN 0x00000004 #define BGE_SDCMODE_CDELAY 0x00000010 /* Send Data completion status register */ #define BGE_SDCSTAT_ATTN 0x00000004 /* * Send BD Ring Selector Control registers */ #define BGE_SRS_MODE 0x1400 #define BGE_SRS_STATUS 0x1404 #define BGE_SRS_HWDIAG 0x1408 #define BGE_SRS_LOC_NIC_CONS0 0x1440 #define BGE_SRS_LOC_NIC_CONS1 0x1444 #define BGE_SRS_LOC_NIC_CONS2 0x1448 #define BGE_SRS_LOC_NIC_CONS3 0x144C #define BGE_SRS_LOC_NIC_CONS4 0x1450 #define BGE_SRS_LOC_NIC_CONS5 0x1454 #define BGE_SRS_LOC_NIC_CONS6 0x1458 #define BGE_SRS_LOC_NIC_CONS7 0x145C #define BGE_SRS_LOC_NIC_CONS8 0x1460 #define BGE_SRS_LOC_NIC_CONS9 0x1464 #define BGE_SRS_LOC_NIC_CONS10 0x1468 #define BGE_SRS_LOC_NIC_CONS11 0x146C #define BGE_SRS_LOC_NIC_CONS12 0x1470 #define BGE_SRS_LOC_NIC_CONS13 0x1474 #define BGE_SRS_LOC_NIC_CONS14 0x1478 #define BGE_SRS_LOC_NIC_CONS15 0x147C /* Send BD Ring Selector Mode register */ #define BGE_SRSMODE_RESET 0x00000001 #define BGE_SRSMODE_ENABLE 0x00000002 #define BGE_SRSMODE_ATTN 0x00000004 /* Send BD Ring Selector Status register */ #define BGE_SRSSTAT_ERROR 0x00000004 /* Send BD Ring Selector HW Diagnostics register */ #define BGE_SRSHWDIAG_STATE 0x0000000F #define BGE_SRSHWDIAG_CURRINGNUM 0x000000F0 #define BGE_SRSHWDIAG_STAGEDRINGNUM 0x00000F00 #define BGE_SRSHWDIAG_RINGNUM_IN_MBX 0x0000F000 /* * Send BD Initiator Selector Control registers */ #define BGE_SBDI_MODE 0x1800 #define BGE_SBDI_STATUS 0x1804 #define BGE_SBDI_LOC_NIC_PROD0 0x1808 #define BGE_SBDI_LOC_NIC_PROD1 0x180C #define BGE_SBDI_LOC_NIC_PROD2 0x1810 #define BGE_SBDI_LOC_NIC_PROD3 0x1814 #define BGE_SBDI_LOC_NIC_PROD4 0x1818 #define BGE_SBDI_LOC_NIC_PROD5 0x181C #define BGE_SBDI_LOC_NIC_PROD6 0x1820 #define BGE_SBDI_LOC_NIC_PROD7 0x1824 #define BGE_SBDI_LOC_NIC_PROD8 0x1828 #define BGE_SBDI_LOC_NIC_PROD9 0x182C #define BGE_SBDI_LOC_NIC_PROD10 0x1830 #define BGE_SBDI_LOC_NIC_PROD11 0x1834 #define BGE_SBDI_LOC_NIC_PROD12 0x1838 #define BGE_SBDI_LOC_NIC_PROD13 0x183C #define BGE_SBDI_LOC_NIC_PROD14 0x1840 #define BGE_SBDI_LOC_NIC_PROD15 0x1844 /* Send BD Initiator Mode register */ #define BGE_SBDIMODE_RESET 0x00000001 #define BGE_SBDIMODE_ENABLE 0x00000002 #define BGE_SBDIMODE_ATTN 0x00000004 /* Send BD Initiator Status register */ #define BGE_SBDISTAT_ERROR 0x00000004 /* * Send BD Completion Control registers */ #define BGE_SBDC_MODE 0x1C00 #define BGE_SBDC_STATUS 0x1C04 /* Send BD Completion Control Mode register */ #define BGE_SBDCMODE_RESET 0x00000001 #define BGE_SBDCMODE_ENABLE 0x00000002 #define BGE_SBDCMODE_ATTN 0x00000004 /* Send BD Completion Control Status register */ #define BGE_SBDCSTAT_ATTN 0x00000004 /* * Receive List Placement Control registers */ #define BGE_RXLP_MODE 0x2000 #define BGE_RXLP_STATUS 0x2004 #define BGE_RXLP_SEL_LIST_LOCK 0x2008 #define BGE_RXLP_SEL_NON_EMPTY_BITS 0x200C #define BGE_RXLP_CFG 0x2010 #define BGE_RXLP_STATS_CTL 0x2014 #define BGE_RXLP_STATS_ENABLE_MASK 0x2018 #define BGE_RXLP_STATS_INCREMENT_MASK 0x201C #define BGE_RXLP_HEAD0 0x2100 #define BGE_RXLP_TAIL0 0x2104 #define BGE_RXLP_COUNT0 0x2108 #define BGE_RXLP_HEAD1 0x2110 #define BGE_RXLP_TAIL1 0x2114 #define BGE_RXLP_COUNT1 0x2118 #define BGE_RXLP_HEAD2 0x2120 #define BGE_RXLP_TAIL2 0x2124 #define BGE_RXLP_COUNT2 0x2128 #define BGE_RXLP_HEAD3 0x2130 #define BGE_RXLP_TAIL3 0x2134 #define BGE_RXLP_COUNT3 0x2138 #define BGE_RXLP_HEAD4 0x2140 #define BGE_RXLP_TAIL4 0x2144 #define BGE_RXLP_COUNT4 0x2148 #define BGE_RXLP_HEAD5 0x2150 #define BGE_RXLP_TAIL5 0x2154 #define BGE_RXLP_COUNT5 0x2158 #define BGE_RXLP_HEAD6 0x2160 #define BGE_RXLP_TAIL6 0x2164 #define BGE_RXLP_COUNT6 0x2168 #define BGE_RXLP_HEAD7 0x2170 #define BGE_RXLP_TAIL7 0x2174 #define BGE_RXLP_COUNT7 0x2178 #define BGE_RXLP_HEAD8 0x2180 #define BGE_RXLP_TAIL8 0x2184 #define BGE_RXLP_COUNT8 0x2188 #define BGE_RXLP_HEAD9 0x2190 #define BGE_RXLP_TAIL9 0x2194 #define BGE_RXLP_COUNT9 0x2198 #define BGE_RXLP_HEAD10 0x21A0 #define BGE_RXLP_TAIL10 0x21A4 #define BGE_RXLP_COUNT10 0x21A8 #define BGE_RXLP_HEAD11 0x21B0 #define BGE_RXLP_TAIL11 0x21B4 #define BGE_RXLP_COUNT11 0x21B8 #define BGE_RXLP_HEAD12 0x21C0 #define BGE_RXLP_TAIL12 0x21C4 #define BGE_RXLP_COUNT12 0x21C8 #define BGE_RXLP_HEAD13 0x21D0 #define BGE_RXLP_TAIL13 0x21D4 #define BGE_RXLP_COUNT13 0x21D8 #define BGE_RXLP_HEAD14 0x21E0 #define BGE_RXLP_TAIL14 0x21E4 #define BGE_RXLP_COUNT14 0x21E8 #define BGE_RXLP_HEAD15 0x21F0 #define BGE_RXLP_TAIL15 0x21F4 #define BGE_RXLP_COUNT15 0x21F8 #define BGE_RXLP_LOCSTAT_COS0 0x2200 #define BGE_RXLP_LOCSTAT_COS1 0x2204 #define BGE_RXLP_LOCSTAT_COS2 0x2208 #define BGE_RXLP_LOCSTAT_COS3 0x220C #define BGE_RXLP_LOCSTAT_COS4 0x2210 #define BGE_RXLP_LOCSTAT_COS5 0x2214 #define BGE_RXLP_LOCSTAT_COS6 0x2218 #define BGE_RXLP_LOCSTAT_COS7 0x221C #define BGE_RXLP_LOCSTAT_COS8 0x2220 #define BGE_RXLP_LOCSTAT_COS9 0x2224 #define BGE_RXLP_LOCSTAT_COS10 0x2228 #define BGE_RXLP_LOCSTAT_COS11 0x222C #define BGE_RXLP_LOCSTAT_COS12 0x2230 #define BGE_RXLP_LOCSTAT_COS13 0x2234 #define BGE_RXLP_LOCSTAT_COS14 0x2238 #define BGE_RXLP_LOCSTAT_COS15 0x223C #define BGE_RXLP_LOCSTAT_FILTDROP 0x2240 #define BGE_RXLP_LOCSTAT_DMA_WRQ_FULL 0x2244 #define BGE_RXLP_LOCSTAT_DMA_HPWRQ_FULL 0x2248 #define BGE_RXLP_LOCSTAT_OUT_OF_BDS 0x224C #define BGE_RXLP_LOCSTAT_IFIN_DROPS 0x2250 #define BGE_RXLP_LOCSTAT_IFIN_ERRORS 0x2254 #define BGE_RXLP_LOCSTAT_RXTHRESH_HIT 0x2258 /* Receive List Placement mode register */ #define BGE_RXLPMODE_RESET 0x00000001 #define BGE_RXLPMODE_ENABLE 0x00000002 #define BGE_RXLPMODE_CLASS0_ATTN 0x00000004 #define BGE_RXLPMODE_MAPOUTRANGE_ATTN 0x00000008 #define BGE_RXLPMODE_STATSOFLOW_ATTN 0x00000010 /* Receive List Placement Status register */ #define BGE_RXLPSTAT_CLASS0_ATTN 0x00000004 #define BGE_RXLPSTAT_MAPOUTRANGE_ATTN 0x00000008 #define BGE_RXLPSTAT_STATSOFLOW_ATTN 0x00000010 /* * Receive Data and Receive BD Initiator Control Registers */ #define BGE_RDBDI_MODE 0x2400 #define BGE_RDBDI_STATUS 0x2404 #define BGE_RX_JUMBO_RCB_HADDR_HI 0x2440 #define BGE_RX_JUMBO_RCB_HADDR_LO 0x2444 #define BGE_RX_JUMBO_RCB_MAXLEN_FLAGS 0x2448 #define BGE_RX_JUMBO_RCB_NICADDR 0x244C #define BGE_RX_STD_RCB_HADDR_HI 0x2450 #define BGE_RX_STD_RCB_HADDR_LO 0x2454 #define BGE_RX_STD_RCB_MAXLEN_FLAGS 0x2458 #define BGE_RX_STD_RCB_NICADDR 0x245C #define BGE_RX_MINI_RCB_HADDR_HI 0x2460 #define BGE_RX_MINI_RCB_HADDR_LO 0x2464 #define BGE_RX_MINI_RCB_MAXLEN_FLAGS 0x2468 #define BGE_RX_MINI_RCB_NICADDR 0x246C #define BGE_RDBDI_JUMBO_RX_CONS 0x2470 #define BGE_RDBDI_STD_RX_CONS 0x2474 #define BGE_RDBDI_MINI_RX_CONS 0x2478 #define BGE_RDBDI_RETURN_PROD0 0x2480 #define BGE_RDBDI_RETURN_PROD1 0x2484 #define BGE_RDBDI_RETURN_PROD2 0x2488 #define BGE_RDBDI_RETURN_PROD3 0x248C #define BGE_RDBDI_RETURN_PROD4 0x2490 #define BGE_RDBDI_RETURN_PROD5 0x2494 #define BGE_RDBDI_RETURN_PROD6 0x2498 #define BGE_RDBDI_RETURN_PROD7 0x249C #define BGE_RDBDI_RETURN_PROD8 0x24A0 #define BGE_RDBDI_RETURN_PROD9 0x24A4 #define BGE_RDBDI_RETURN_PROD10 0x24A8 #define BGE_RDBDI_RETURN_PROD11 0x24AC #define BGE_RDBDI_RETURN_PROD12 0x24B0 #define BGE_RDBDI_RETURN_PROD13 0x24B4 #define BGE_RDBDI_RETURN_PROD14 0x24B8 #define BGE_RDBDI_RETURN_PROD15 0x24BC #define BGE_RDBDI_HWDIAG 0x24C0 /* Receive Data and Receive BD Initiator Mode register */ #define BGE_RDBDIMODE_RESET 0x00000001 #define BGE_RDBDIMODE_ENABLE 0x00000002 #define BGE_RDBDIMODE_JUMBO_ATTN 0x00000004 #define BGE_RDBDIMODE_GIANT_ATTN 0x00000008 #define BGE_RDBDIMODE_BADRINGSZ_ATTN 0x00000010 /* Receive Data and Receive BD Initiator Status register */ #define BGE_RDBDISTAT_JUMBO_ATTN 0x00000004 #define BGE_RDBDISTAT_GIANT_ATTN 0x00000008 #define BGE_RDBDISTAT_BADRINGSZ_ATTN 0x00000010 /* * Receive Data Completion Control registers */ #define BGE_RDC_MODE 0x2800 /* Receive Data Completion Mode register */ #define BGE_RDCMODE_RESET 0x00000001 #define BGE_RDCMODE_ENABLE 0x00000002 #define BGE_RDCMODE_ATTN 0x00000004 /* * Receive BD Initiator Control registers */ #define BGE_RBDI_MODE 0x2C00 #define BGE_RBDI_STATUS 0x2C04 #define BGE_RBDI_NIC_JUMBO_BD_PROD 0x2C08 #define BGE_RBDI_NIC_STD_BD_PROD 0x2C0C #define BGE_RBDI_NIC_MINI_BD_PROD 0x2C10 #define BGE_RBDI_MINI_REPL_THRESH 0x2C14 #define BGE_RBDI_STD_REPL_THRESH 0x2C18 #define BGE_RBDI_JUMBO_REPL_THRESH 0x2C1C #define BGE_STD_REPLENISH_LWM 0x2D00 #define BGE_JMB_REPLENISH_LWM 0x2D04 /* Receive BD Initiator Mode register */ #define BGE_RBDIMODE_RESET 0x00000001 #define BGE_RBDIMODE_ENABLE 0x00000002 #define BGE_RBDIMODE_ATTN 0x00000004 /* Receive BD Initiator Status register */ #define BGE_RBDISTAT_ATTN 0x00000004 /* * Receive BD Completion Control registers */ #define BGE_RBDC_MODE 0x3000 #define BGE_RBDC_STATUS 0x3004 #define BGE_RBDC_JUMBO_BD_PROD 0x3008 #define BGE_RBDC_STD_BD_PROD 0x300C #define BGE_RBDC_MINI_BD_PROD 0x3010 /* Receive BD completion mode register */ #define BGE_RBDCMODE_RESET 0x00000001 #define BGE_RBDCMODE_ENABLE 0x00000002 #define BGE_RBDCMODE_ATTN 0x00000004 /* Receive BD completion status register */ #define BGE_RBDCSTAT_ERROR 0x00000004 /* * Receive List Selector Control registers */ #define BGE_RXLS_MODE 0x3400 #define BGE_RXLS_STATUS 0x3404 /* Receive List Selector Mode register */ #define BGE_RXLSMODE_RESET 0x00000001 #define BGE_RXLSMODE_ENABLE 0x00000002 #define BGE_RXLSMODE_ATTN 0x00000004 /* Receive List Selector Status register */ #define BGE_RXLSSTAT_ERROR 0x00000004 #define BGE_CPMU_CTRL 0x3600 #define BGE_CPMU_LSPD_10MB_CLK 0x3604 #define BGE_CPMU_LSPD_1000MB_CLK 0x360C #define BGE_CPMU_LNK_AWARE_PWRMD 0x3610 #define BGE_CPMU_HST_ACC 0x361C #define BGE_CPMU_CLCK_STAT 0x3630 #define BGE_CPMU_MUTEX_REQ 0x365C #define BGE_CPMU_MUTEX_GNT 0x3660 #define BGE_CPMU_PHY_STRAP 0x3664 /* Central Power Management Unit (CPMU) register */ #define BGE_CPMU_CTRL_LINK_IDLE_MODE 0x00000200 #define BGE_CPMU_CTRL_LINK_AWARE_MODE 0x00000400 #define BGE_CPMU_CTRL_LINK_SPEED_MODE 0x00004000 #define BGE_CPMU_CTRL_GPHY_10MB_RXONLY 0x00010000 /* Link Speed 10MB/No Link Power Mode Clock Policy register */ #define BGE_CPMU_LSPD_10MB_MACCLK_MASK 0x001F0000 #define BGE_CPMU_LSPD_10MB_MACCLK_6_25 0x00130000 /* Link Speed 1000MB Power Mode Clock Policy register */ #define BGE_CPMU_LSPD_1000MB_MACCLK_62_5 0x00000000 #define BGE_CPMU_LSPD_1000MB_MACCLK_12_5 0x00110000 #define BGE_CPMU_LSPD_1000MB_MACCLK_MASK 0x001F0000 /* Link Aware Power Mode Clock Policy register */ #define BGE_CPMU_LNK_AWARE_MACCLK_MASK 0x001F0000 #define BGE_CPMU_LNK_AWARE_MACCLK_6_25 0x00130000 #define BGE_CPMU_HST_ACC_MACCLK_MASK 0x001F0000 #define BGE_CPMU_HST_ACC_MACCLK_6_25 0x00130000 /* CPMU Clock Status register */ #define BGE_CPMU_CLCK_STAT_MAC_CLCK_MASK 0x001F0000 #define BGE_CPMU_CLCK_STAT_MAC_CLCK_62_5 0x00000000 #define BGE_CPMU_CLCK_STAT_MAC_CLCK_12_5 0x00110000 #define BGE_CPMU_CLCK_STAT_MAC_CLCK_6_25 0x00130000 /* CPMU Mutex Request register */ #define BGE_CPMU_MUTEX_REQ_DRIVER 0x00001000 #define BGE_CPMU_MUTEX_GNT_DRIVER 0x00001000 /* CPMU GPHY Strap register */ #define BGE_CPMU_PHY_STRAP_IS_SERDES 0x00000020 /* * Mbuf Cluster Free registers (has nothing to do with BSD mbufs) */ #define BGE_MBCF_MODE 0x3800 #define BGE_MBCF_STATUS 0x3804 /* Mbuf Cluster Free mode register */ #define BGE_MBCFMODE_RESET 0x00000001 #define BGE_MBCFMODE_ENABLE 0x00000002 #define BGE_MBCFMODE_ATTN 0x00000004 /* Mbuf Cluster Free status register */ #define BGE_MBCFSTAT_ERROR 0x00000004 /* * Host Coalescing Control registers */ #define BGE_HCC_MODE 0x3C00 #define BGE_HCC_STATUS 0x3C04 #define BGE_HCC_RX_COAL_TICKS 0x3C08 #define BGE_HCC_TX_COAL_TICKS 0x3C0C #define BGE_HCC_RX_MAX_COAL_BDS 0x3C10 #define BGE_HCC_TX_MAX_COAL_BDS 0x3C14 #define BGE_HCC_RX_COAL_TICKS_INT 0x3C18 /* ticks during interrupt */ #define BGE_HCC_TX_COAL_TICKS_INT 0x3C1C /* ticks during interrupt */ #define BGE_HCC_RX_MAX_COAL_BDS_INT 0x3C20 /* BDs during interrupt */ #define BGE_HCC_TX_MAX_COAL_BDS_INT 0x3C24 /* BDs during interrupt */ #define BGE_HCC_STATS_TICKS 0x3C28 #define BGE_HCC_STATS_ADDR_HI 0x3C30 #define BGE_HCC_STATS_ADDR_LO 0x3C34 #define BGE_HCC_STATUSBLK_ADDR_HI 0x3C38 #define BGE_HCC_STATUSBLK_ADDR_LO 0x3C3C #define BGE_HCC_STATS_BASEADDR 0x3C40 /* address in NIC memory */ #define BGE_HCC_STATUSBLK_BASEADDR 0x3C44 /* address in NIC memory */ #define BGE_FLOW_ATTN 0x3C48 #define BGE_HCC_JUMBO_BD_CONS 0x3C50 #define BGE_HCC_STD_BD_CONS 0x3C54 #define BGE_HCC_MINI_BD_CONS 0x3C58 #define BGE_HCC_RX_RETURN_PROD0 0x3C80 #define BGE_HCC_RX_RETURN_PROD1 0x3C84 #define BGE_HCC_RX_RETURN_PROD2 0x3C88 #define BGE_HCC_RX_RETURN_PROD3 0x3C8C #define BGE_HCC_RX_RETURN_PROD4 0x3C90 #define BGE_HCC_RX_RETURN_PROD5 0x3C94 #define BGE_HCC_RX_RETURN_PROD6 0x3C98 #define BGE_HCC_RX_RETURN_PROD7 0x3C9C #define BGE_HCC_RX_RETURN_PROD8 0x3CA0 #define BGE_HCC_RX_RETURN_PROD9 0x3CA4 #define BGE_HCC_RX_RETURN_PROD10 0x3CA8 #define BGE_HCC_RX_RETURN_PROD11 0x3CAC #define BGE_HCC_RX_RETURN_PROD12 0x3CB0 #define BGE_HCC_RX_RETURN_PROD13 0x3CB4 #define BGE_HCC_RX_RETURN_PROD14 0x3CB8 #define BGE_HCC_RX_RETURN_PROD15 0x3CBC #define BGE_HCC_TX_BD_CONS0 0x3CC0 #define BGE_HCC_TX_BD_CONS1 0x3CC4 #define BGE_HCC_TX_BD_CONS2 0x3CC8 #define BGE_HCC_TX_BD_CONS3 0x3CCC #define BGE_HCC_TX_BD_CONS4 0x3CD0 #define BGE_HCC_TX_BD_CONS5 0x3CD4 #define BGE_HCC_TX_BD_CONS6 0x3CD8 #define BGE_HCC_TX_BD_CONS7 0x3CDC #define BGE_HCC_TX_BD_CONS8 0x3CE0 #define BGE_HCC_TX_BD_CONS9 0x3CE4 #define BGE_HCC_TX_BD_CONS10 0x3CE8 #define BGE_HCC_TX_BD_CONS11 0x3CEC #define BGE_HCC_TX_BD_CONS12 0x3CF0 #define BGE_HCC_TX_BD_CONS13 0x3CF4 #define BGE_HCC_TX_BD_CONS14 0x3CF8 #define BGE_HCC_TX_BD_CONS15 0x3CFC /* Host coalescing mode register */ #define BGE_HCCMODE_RESET 0x00000001 #define BGE_HCCMODE_ENABLE 0x00000002 #define BGE_HCCMODE_ATTN 0x00000004 #define BGE_HCCMODE_COAL_NOW 0x00000008 #define BGE_HCCMODE_MSI_BITS 0x00000070 #define BGE_HCCMODE_STATBLK_SIZE 0x00000180 #define BGE_STATBLKSZ_FULL 0x00000000 #define BGE_STATBLKSZ_64BYTE 0x00000080 #define BGE_STATBLKSZ_32BYTE 0x00000100 /* Host coalescing status register */ #define BGE_HCCSTAT_ERROR 0x00000004 /* Flow attention register */ #define BGE_FLOWATTN_MB_LOWAT 0x00000040 #define BGE_FLOWATTN_MEMARB 0x00000080 #define BGE_FLOWATTN_HOSTCOAL 0x00008000 #define BGE_FLOWATTN_DMADONE_DISCARD 0x00010000 #define BGE_FLOWATTN_RCB_INVAL 0x00020000 #define BGE_FLOWATTN_RXDATA_CORRUPT 0x00040000 #define BGE_FLOWATTN_RDBDI 0x00080000 #define BGE_FLOWATTN_RXLS 0x00100000 #define BGE_FLOWATTN_RXLP 0x00200000 #define BGE_FLOWATTN_RBDC 0x00400000 #define BGE_FLOWATTN_RBDI 0x00800000 #define BGE_FLOWATTN_SDC 0x08000000 #define BGE_FLOWATTN_SDI 0x10000000 #define BGE_FLOWATTN_SRS 0x20000000 #define BGE_FLOWATTN_SBDC 0x40000000 #define BGE_FLOWATTN_SBDI 0x80000000 /* * Memory arbiter registers */ #define BGE_MARB_MODE 0x4000 #define BGE_MARB_STATUS 0x4004 #define BGE_MARB_TRAPADDR_HI 0x4008 #define BGE_MARB_TRAPADDR_LO 0x400C /* Memory arbiter mode register */ #define BGE_MARBMODE_RESET 0x00000001 #define BGE_MARBMODE_ENABLE 0x00000002 #define BGE_MARBMODE_TX_ADDR_TRAP 0x00000004 #define BGE_MARBMODE_RX_ADDR_TRAP 0x00000008 #define BGE_MARBMODE_DMAW1_TRAP 0x00000010 #define BGE_MARBMODE_DMAR1_TRAP 0x00000020 #define BGE_MARBMODE_RXRISC_TRAP 0x00000040 #define BGE_MARBMODE_TXRISC_TRAP 0x00000080 #define BGE_MARBMODE_PCI_TRAP 0x00000100 #define BGE_MARBMODE_DMAR2_TRAP 0x00000200 #define BGE_MARBMODE_RXQ_TRAP 0x00000400 #define BGE_MARBMODE_RXDI1_TRAP 0x00000800 #define BGE_MARBMODE_RXDI2_TRAP 0x00001000 #define BGE_MARBMODE_DC_GRPMEM_TRAP 0x00002000 #define BGE_MARBMODE_HCOAL_TRAP 0x00004000 #define BGE_MARBMODE_MBUF_TRAP 0x00008000 #define BGE_MARBMODE_TXDI_TRAP 0x00010000 #define BGE_MARBMODE_SDC_DMAC_TRAP 0x00020000 #define BGE_MARBMODE_TXBD_TRAP 0x00040000 #define BGE_MARBMODE_BUFFMAN_TRAP 0x00080000 #define BGE_MARBMODE_DMAW2_TRAP 0x00100000 #define BGE_MARBMODE_XTSSRAM_ROFLO_TRAP 0x00200000 #define BGE_MARBMODE_XTSSRAM_RUFLO_TRAP 0x00400000 #define BGE_MARBMODE_XTSSRAM_WOFLO_TRAP 0x00800000 #define BGE_MARBMODE_XTSSRAM_WUFLO_TRAP 0x01000000 #define BGE_MARBMODE_XTSSRAM_PERR_TRAP 0x02000000 /* Memory arbiter status register */ #define BGE_MARBSTAT_TX_ADDR_TRAP 0x00000004 #define BGE_MARBSTAT_RX_ADDR_TRAP 0x00000008 #define BGE_MARBSTAT_DMAW1_TRAP 0x00000010 #define BGE_MARBSTAT_DMAR1_TRAP 0x00000020 #define BGE_MARBSTAT_RXRISC_TRAP 0x00000040 #define BGE_MARBSTAT_TXRISC_TRAP 0x00000080 #define BGE_MARBSTAT_PCI_TRAP 0x00000100 #define BGE_MARBSTAT_DMAR2_TRAP 0x00000200 #define BGE_MARBSTAT_RXQ_TRAP 0x00000400 #define BGE_MARBSTAT_RXDI1_TRAP 0x00000800 #define BGE_MARBSTAT_RXDI2_TRAP 0x00001000 #define BGE_MARBSTAT_DC_GRPMEM_TRAP 0x00002000 #define BGE_MARBSTAT_HCOAL_TRAP 0x00004000 #define BGE_MARBSTAT_MBUF_TRAP 0x00008000 #define BGE_MARBSTAT_TXDI_TRAP 0x00010000 #define BGE_MARBSTAT_SDC_DMAC_TRAP 0x00020000 #define BGE_MARBSTAT_TXBD_TRAP 0x00040000 #define BGE_MARBSTAT_BUFFMAN_TRAP 0x00080000 #define BGE_MARBSTAT_DMAW2_TRAP 0x00100000 #define BGE_MARBSTAT_XTSSRAM_ROFLO_TRAP 0x00200000 #define BGE_MARBSTAT_XTSSRAM_RUFLO_TRAP 0x00400000 #define BGE_MARBSTAT_XTSSRAM_WOFLO_TRAP 0x00800000 #define BGE_MARBSTAT_XTSSRAM_WUFLO_TRAP 0x01000000 #define BGE_MARBSTAT_XTSSRAM_PERR_TRAP 0x02000000 /* * Buffer manager control registers */ #define BGE_BMAN_MODE 0x4400 #define BGE_BMAN_STATUS 0x4404 #define BGE_BMAN_MBUFPOOL_BASEADDR 0x4408 #define BGE_BMAN_MBUFPOOL_LEN 0x440C #define BGE_BMAN_MBUFPOOL_READDMA_LOWAT 0x4410 #define BGE_BMAN_MBUFPOOL_MACRX_LOWAT 0x4414 #define BGE_BMAN_MBUFPOOL_HIWAT 0x4418 #define BGE_BMAN_RXCPU_MBALLOC_REQ 0x441C #define BGE_BMAN_RXCPU_MBALLOC_RESP 0x4420 #define BGE_BMAN_TXCPU_MBALLOC_REQ 0x4424 #define BGE_BMAN_TXCPU_MBALLOC_RESP 0x4428 #define BGE_BMAN_DMA_DESCPOOL_BASEADDR 0x442C #define BGE_BMAN_DMA_DESCPOOL_LEN 0x4430 #define BGE_BMAN_DMA_DESCPOOL_LOWAT 0x4434 #define BGE_BMAN_DMA_DESCPOOL_HIWAT 0x4438 #define BGE_BMAN_RXCPU_DMAALLOC_REQ 0x443C #define BGE_BMAN_RXCPU_DMAALLOC_RESP 0x4440 #define BGE_BMAN_TXCPU_DMAALLOC_REQ 0x4444 #define BGE_BMAN_TXCPU_DMALLLOC_RESP 0x4448 #define BGE_BMAN_HWDIAG_1 0x444C #define BGE_BMAN_HWDIAG_2 0x4450 #define BGE_BMAN_HWDIAG_3 0x4454 /* Buffer manager mode register */ #define BGE_BMANMODE_RESET 0x00000001 #define BGE_BMANMODE_ENABLE 0x00000002 #define BGE_BMANMODE_ATTN 0x00000004 #define BGE_BMANMODE_TESTMODE 0x00000008 #define BGE_BMANMODE_LOMBUF_ATTN 0x00000010 /* Buffer manager status register */ #define BGE_BMANSTAT_ERRO 0x00000004 #define BGE_BMANSTAT_LOWMBUF_ERROR 0x00000010 /* * Read DMA Control registers */ #define BGE_RDMA_MODE 0x4800 #define BGE_RDMA_STATUS 0x4804 #define BGE_RDMA_RSRVCTRL 0x4900 /* Read DMA mode register */ #define BGE_RDMAMODE_RESET 0x00000001 #define BGE_RDMAMODE_ENABLE 0x00000002 #define BGE_RDMAMODE_PCI_TGT_ABRT_ATTN 0x00000004 #define BGE_RDMAMODE_PCI_MSTR_ABRT_ATTN 0x00000008 #define BGE_RDMAMODE_PCI_PERR_ATTN 0x00000010 #define BGE_RDMAMODE_PCI_ADDROFLOW_ATTN 0x00000020 #define BGE_RDMAMODE_PCI_FIFOOFLOW_ATTN 0x00000040 #define BGE_RDMAMODE_PCI_FIFOUFLOW_ATTN 0x00000080 #define BGE_RDMAMODE_PCI_FIFOOREAD_ATTN 0x00000100 #define BGE_RDMAMODE_LOCWRITE_TOOBIG 0x00000200 #define BGE_RDMAMODE_ALL_ATTNS 0x000003FC #define BGE_RDMAMODE_BD_SBD_CRPT_ATTN 0x00000800 #define BGE_RDMAMODE_MBUF_RBD_CRPT_ATTN 0x00001000 #define BGE_RDMAMODE_MBUF_SBD_CRPT_ATTN 0x00002000 #define BGE_RDMAMODE_FIFO_SIZE_128 0x00020000 #define BGE_RDMAMODE_FIFO_LONG_BURST 0x00030000 #define BGE_RDMAMODE_MULT_DMA_RD_DIS 0x01000000 #define BGE_RDMAMODE_TSO4_ENABLE 0x08000000 #define BGE_RDMAMODE_TSO6_ENABLE 0x10000000 /* Read DMA status register */ #define BGE_RDMASTAT_PCI_TGT_ABRT_ATTN 0x00000004 #define BGE_RDMASTAT_PCI_MSTR_ABRT_ATTN 0x00000008 #define BGE_RDMASTAT_PCI_PERR_ATTN 0x00000010 #define BGE_RDMASTAT_PCI_ADDROFLOW_ATTN 0x00000020 #define BGE_RDMASTAT_PCI_FIFOOFLOW_ATTN 0x00000040 #define BGE_RDMASTAT_PCI_FIFOUFLOW_ATTN 0x00000080 #define BGE_RDMASTAT_PCI_FIFOOREAD_ATTN 0x00000100 #define BGE_RDMASTAT_LOCWRITE_TOOBIG 0x00000200 /* Read DMA Reserved Control register */ #define BGE_RDMA_RSRVCTRL_FIFO_OFLW_FIX 0x00000004 /* * Write DMA control registers */ #define BGE_WDMA_MODE 0x4C00 #define BGE_WDMA_STATUS 0x4C04 /* Write DMA mode register */ #define BGE_WDMAMODE_RESET 0x00000001 #define BGE_WDMAMODE_ENABLE 0x00000002 #define BGE_WDMAMODE_PCI_TGT_ABRT_ATTN 0x00000004 #define BGE_WDMAMODE_PCI_MSTR_ABRT_ATTN 0x00000008 #define BGE_WDMAMODE_PCI_PERR_ATTN 0x00000010 #define BGE_WDMAMODE_PCI_ADDROFLOW_ATTN 0x00000020 #define BGE_WDMAMODE_PCI_FIFOOFLOW_ATTN 0x00000040 #define BGE_WDMAMODE_PCI_FIFOUFLOW_ATTN 0x00000080 #define BGE_WDMAMODE_PCI_FIFOOREAD_ATTN 0x00000100 #define BGE_WDMAMODE_LOCREAD_TOOBIG 0x00000200 #define BGE_WDMAMODE_ALL_ATTNS 0x000003FC #define BGE_WDMAMODE_STATUS_TAG_FIX 0x20000000 #define BGE_WDMAMODE_BURST_ALL_DATA 0xC0000000 /* Write DMA status register */ #define BGE_WDMASTAT_PCI_TGT_ABRT_ATTN 0x00000004 #define BGE_WDMASTAT_PCI_MSTR_ABRT_ATTN 0x00000008 #define BGE_WDMASTAT_PCI_PERR_ATTN 0x00000010 #define BGE_WDMASTAT_PCI_ADDROFLOW_ATTN 0x00000020 #define BGE_WDMASTAT_PCI_FIFOOFLOW_ATTN 0x00000040 #define BGE_WDMASTAT_PCI_FIFOUFLOW_ATTN 0x00000080 #define BGE_WDMASTAT_PCI_FIFOOREAD_ATTN 0x00000100 #define BGE_WDMASTAT_LOCREAD_TOOBIG 0x00000200 /* * RX CPU registers */ #define BGE_RXCPU_MODE 0x5000 #define BGE_RXCPU_STATUS 0x5004 #define BGE_RXCPU_PC 0x501C /* RX CPU mode register */ #define BGE_RXCPUMODE_RESET 0x00000001 #define BGE_RXCPUMODE_SINGLESTEP 0x00000002 #define BGE_RXCPUMODE_P0_DATAHLT_ENB 0x00000004 #define BGE_RXCPUMODE_P0_INSTRHLT_ENB 0x00000008 #define BGE_RXCPUMODE_WR_POSTBUF_ENB 0x00000010 #define BGE_RXCPUMODE_DATACACHE_ENB 0x00000020 #define BGE_RXCPUMODE_ROMFAIL 0x00000040 #define BGE_RXCPUMODE_WATCHDOG_ENB 0x00000080 #define BGE_RXCPUMODE_INSTRCACHE_PRF 0x00000100 #define BGE_RXCPUMODE_INSTRCACHE_FLUSH 0x00000200 #define BGE_RXCPUMODE_HALTCPU 0x00000400 #define BGE_RXCPUMODE_INVDATAHLT_ENB 0x00000800 #define BGE_RXCPUMODE_MADDRTRAPHLT_ENB 0x00001000 #define BGE_RXCPUMODE_RADDRTRAPHLT_ENB 0x00002000 /* RX CPU status register */ #define BGE_RXCPUSTAT_HW_BREAKPOINT 0x00000001 #define BGE_RXCPUSTAT_HLTINSTR_EXECUTED 0x00000002 #define BGE_RXCPUSTAT_INVALID_INSTR 0x00000004 #define BGE_RXCPUSTAT_P0_DATAREF 0x00000008 #define BGE_RXCPUSTAT_P0_INSTRREF 0x00000010 #define BGE_RXCPUSTAT_INVALID_DATAACC 0x00000020 #define BGE_RXCPUSTAT_INVALID_INSTRFTCH 0x00000040 #define BGE_RXCPUSTAT_BAD_MEMALIGN 0x00000080 #define BGE_RXCPUSTAT_MADDR_TRAP 0x00000100 #define BGE_RXCPUSTAT_REGADDR_TRAP 0x00000200 #define BGE_RXCPUSTAT_DATAACC_STALL 0x00001000 #define BGE_RXCPUSTAT_INSTRFETCH_STALL 0x00002000 #define BGE_RXCPUSTAT_MA_WR_FIFOOFLOW 0x08000000 #define BGE_RXCPUSTAT_MA_RD_FIFOOFLOW 0x10000000 #define BGE_RXCPUSTAT_MA_DATAMASK_OFLOW 0x20000000 #define BGE_RXCPUSTAT_MA_REQ_FIFOOFLOW 0x40000000 #define BGE_RXCPUSTAT_BLOCKING_READ 0x80000000 /* * V? CPU registers */ #define BGE_VCPU_STATUS 0x5100 #define BGE_VCPU_EXT_CTRL 0x6890 #define BGE_VCPU_STATUS_INIT_DONE 0x04000000 #define BGE_VCPU_STATUS_DRV_RESET 0x08000000 #define BGE_VCPU_EXT_CTRL_HALT_CPU 0x00400000 #define BGE_VCPU_EXT_CTRL_DISABLE_WOL 0x20000000 /* * TX CPU registers */ #define BGE_TXCPU_MODE 0x5400 #define BGE_TXCPU_STATUS 0x5404 #define BGE_TXCPU_PC 0x541C /* TX CPU mode register */ #define BGE_TXCPUMODE_RESET 0x00000001 #define BGE_TXCPUMODE_SINGLESTEP 0x00000002 #define BGE_TXCPUMODE_P0_DATAHLT_ENB 0x00000004 #define BGE_TXCPUMODE_P0_INSTRHLT_ENB 0x00000008 #define BGE_TXCPUMODE_WR_POSTBUF_ENB 0x00000010 #define BGE_TXCPUMODE_DATACACHE_ENB 0x00000020 #define BGE_TXCPUMODE_ROMFAIL 0x00000040 #define BGE_TXCPUMODE_WATCHDOG_ENB 0x00000080 #define BGE_TXCPUMODE_INSTRCACHE_PRF 0x00000100 #define BGE_TXCPUMODE_INSTRCACHE_FLUSH 0x00000200 #define BGE_TXCPUMODE_HALTCPU 0x00000400 #define BGE_TXCPUMODE_INVDATAHLT_ENB 0x00000800 #define BGE_TXCPUMODE_MADDRTRAPHLT_ENB 0x00001000 /* TX CPU status register */ #define BGE_TXCPUSTAT_HW_BREAKPOINT 0x00000001 #define BGE_TXCPUSTAT_HLTINSTR_EXECUTED 0x00000002 #define BGE_TXCPUSTAT_INVALID_INSTR 0x00000004 #define BGE_TXCPUSTAT_P0_DATAREF 0x00000008 #define BGE_TXCPUSTAT_P0_INSTRREF 0x00000010 #define BGE_TXCPUSTAT_INVALID_DATAACC 0x00000020 #define BGE_TXCPUSTAT_INVALID_INSTRFTCH 0x00000040 #define BGE_TXCPUSTAT_BAD_MEMALIGN 0x00000080 #define BGE_TXCPUSTAT_MADDR_TRAP 0x00000100 #define BGE_TXCPUSTAT_REGADDR_TRAP 0x00000200 #define BGE_TXCPUSTAT_DATAACC_STALL 0x00001000 #define BGE_TXCPUSTAT_INSTRFETCH_STALL 0x00002000 #define BGE_TXCPUSTAT_MA_WR_FIFOOFLOW 0x08000000 #define BGE_TXCPUSTAT_MA_RD_FIFOOFLOW 0x10000000 #define BGE_TXCPUSTAT_MA_DATAMASK_OFLOW 0x20000000 #define BGE_TXCPUSTAT_MA_REQ_FIFOOFLOW 0x40000000 #define BGE_TXCPUSTAT_BLOCKING_READ 0x80000000 /* * Low priority mailbox registers */ #define BGE_LPMBX_IRQ0_HI 0x5800 #define BGE_LPMBX_IRQ0_LO 0x5804 #define BGE_LPMBX_IRQ1_HI 0x5808 #define BGE_LPMBX_IRQ1_LO 0x580C #define BGE_LPMBX_IRQ2_HI 0x5810 #define BGE_LPMBX_IRQ2_LO 0x5814 #define BGE_LPMBX_IRQ3_HI 0x5818 #define BGE_LPMBX_IRQ3_LO 0x581C #define BGE_LPMBX_GEN0_HI 0x5820 #define BGE_LPMBX_GEN0_LO 0x5824 #define BGE_LPMBX_GEN1_HI 0x5828 #define BGE_LPMBX_GEN1_LO 0x582C #define BGE_LPMBX_GEN2_HI 0x5830 #define BGE_LPMBX_GEN2_LO 0x5834 #define BGE_LPMBX_GEN3_HI 0x5828 #define BGE_LPMBX_GEN3_LO 0x582C #define BGE_LPMBX_GEN4_HI 0x5840 #define BGE_LPMBX_GEN4_LO 0x5844 #define BGE_LPMBX_GEN5_HI 0x5848 #define BGE_LPMBX_GEN5_LO 0x584C #define BGE_LPMBX_GEN6_HI 0x5850 #define BGE_LPMBX_GEN6_LO 0x5854 #define BGE_LPMBX_GEN7_HI 0x5858 #define BGE_LPMBX_GEN7_LO 0x585C #define BGE_LPMBX_RELOAD_STATS_HI 0x5860 #define BGE_LPMBX_RELOAD_STATS_LO 0x5864 #define BGE_LPMBX_RX_STD_PROD_HI 0x5868 #define BGE_LPMBX_RX_STD_PROD_LO 0x586C #define BGE_LPMBX_RX_JUMBO_PROD_HI 0x5870 #define BGE_LPMBX_RX_JUMBO_PROD_LO 0x5874 #define BGE_LPMBX_RX_MINI_PROD_HI 0x5878 #define BGE_LPMBX_RX_MINI_PROD_LO 0x587C #define BGE_LPMBX_RX_CONS0_HI 0x5880 #define BGE_LPMBX_RX_CONS0_LO 0x5884 #define BGE_LPMBX_RX_CONS1_HI 0x5888 #define BGE_LPMBX_RX_CONS1_LO 0x588C #define BGE_LPMBX_RX_CONS2_HI 0x5890 #define BGE_LPMBX_RX_CONS2_LO 0x5894 #define BGE_LPMBX_RX_CONS3_HI 0x5898 #define BGE_LPMBX_RX_CONS3_LO 0x589C #define BGE_LPMBX_RX_CONS4_HI 0x58A0 #define BGE_LPMBX_RX_CONS4_LO 0x58A4 #define BGE_LPMBX_RX_CONS5_HI 0x58A8 #define BGE_LPMBX_RX_CONS5_LO 0x58AC #define BGE_LPMBX_RX_CONS6_HI 0x58B0 #define BGE_LPMBX_RX_CONS6_LO 0x58B4 #define BGE_LPMBX_RX_CONS7_HI 0x58B8 #define BGE_LPMBX_RX_CONS7_LO 0x58BC #define BGE_LPMBX_RX_CONS8_HI 0x58C0 #define BGE_LPMBX_RX_CONS8_LO 0x58C4 #define BGE_LPMBX_RX_CONS9_HI 0x58C8 #define BGE_LPMBX_RX_CONS9_LO 0x58CC #define BGE_LPMBX_RX_CONS10_HI 0x58D0 #define BGE_LPMBX_RX_CONS10_LO 0x58D4 #define BGE_LPMBX_RX_CONS11_HI 0x58D8 #define BGE_LPMBX_RX_CONS11_LO 0x58DC #define BGE_LPMBX_RX_CONS12_HI 0x58E0 #define BGE_LPMBX_RX_CONS12_LO 0x58E4 #define BGE_LPMBX_RX_CONS13_HI 0x58E8 #define BGE_LPMBX_RX_CONS13_LO 0x58EC #define BGE_LPMBX_RX_CONS14_HI 0x58F0 #define BGE_LPMBX_RX_CONS14_LO 0x58F4 #define BGE_LPMBX_RX_CONS15_HI 0x58F8 #define BGE_LPMBX_RX_CONS15_LO 0x58FC #define BGE_LPMBX_TX_HOST_PROD0_HI 0x5900 #define BGE_LPMBX_TX_HOST_PROD0_LO 0x5904 #define BGE_LPMBX_TX_HOST_PROD1_HI 0x5908 #define BGE_LPMBX_TX_HOST_PROD1_LO 0x590C #define BGE_LPMBX_TX_HOST_PROD2_HI 0x5910 #define BGE_LPMBX_TX_HOST_PROD2_LO 0x5914 #define BGE_LPMBX_TX_HOST_PROD3_HI 0x5918 #define BGE_LPMBX_TX_HOST_PROD3_LO 0x591C #define BGE_LPMBX_TX_HOST_PROD4_HI 0x5920 #define BGE_LPMBX_TX_HOST_PROD4_LO 0x5924 #define BGE_LPMBX_TX_HOST_PROD5_HI 0x5928 #define BGE_LPMBX_TX_HOST_PROD5_LO 0x592C #define BGE_LPMBX_TX_HOST_PROD6_HI 0x5930 #define BGE_LPMBX_TX_HOST_PROD6_LO 0x5934 #define BGE_LPMBX_TX_HOST_PROD7_HI 0x5938 #define BGE_LPMBX_TX_HOST_PROD7_LO 0x593C #define BGE_LPMBX_TX_HOST_PROD8_HI 0x5940 #define BGE_LPMBX_TX_HOST_PROD8_LO 0x5944 #define BGE_LPMBX_TX_HOST_PROD9_HI 0x5948 #define BGE_LPMBX_TX_HOST_PROD9_LO 0x594C #define BGE_LPMBX_TX_HOST_PROD10_HI 0x5950 #define BGE_LPMBX_TX_HOST_PROD10_LO 0x5954 #define BGE_LPMBX_TX_HOST_PROD11_HI 0x5958 #define BGE_LPMBX_TX_HOST_PROD11_LO 0x595C #define BGE_LPMBX_TX_HOST_PROD12_HI 0x5960 #define BGE_LPMBX_TX_HOST_PROD12_LO 0x5964 #define BGE_LPMBX_TX_HOST_PROD13_HI 0x5968 #define BGE_LPMBX_TX_HOST_PROD13_LO 0x596C #define BGE_LPMBX_TX_HOST_PROD14_HI 0x5970 #define BGE_LPMBX_TX_HOST_PROD14_LO 0x5974 #define BGE_LPMBX_TX_HOST_PROD15_HI 0x5978 #define BGE_LPMBX_TX_HOST_PROD15_LO 0x597C #define BGE_LPMBX_TX_NIC_PROD0_HI 0x5980 #define BGE_LPMBX_TX_NIC_PROD0_LO 0x5984 #define BGE_LPMBX_TX_NIC_PROD1_HI 0x5988 #define BGE_LPMBX_TX_NIC_PROD1_LO 0x598C #define BGE_LPMBX_TX_NIC_PROD2_HI 0x5990 #define BGE_LPMBX_TX_NIC_PROD2_LO 0x5994 #define BGE_LPMBX_TX_NIC_PROD3_HI 0x5998 #define BGE_LPMBX_TX_NIC_PROD3_LO 0x599C #define BGE_LPMBX_TX_NIC_PROD4_HI 0x59A0 #define BGE_LPMBX_TX_NIC_PROD4_LO 0x59A4 #define BGE_LPMBX_TX_NIC_PROD5_HI 0x59A8 #define BGE_LPMBX_TX_NIC_PROD5_LO 0x59AC #define BGE_LPMBX_TX_NIC_PROD6_HI 0x59B0 #define BGE_LPMBX_TX_NIC_PROD6_LO 0x59B4 #define BGE_LPMBX_TX_NIC_PROD7_HI 0x59B8 #define BGE_LPMBX_TX_NIC_PROD7_LO 0x59BC #define BGE_LPMBX_TX_NIC_PROD8_HI 0x59C0 #define BGE_LPMBX_TX_NIC_PROD8_LO 0x59C4 #define BGE_LPMBX_TX_NIC_PROD9_HI 0x59C8 #define BGE_LPMBX_TX_NIC_PROD9_LO 0x59CC #define BGE_LPMBX_TX_NIC_PROD10_HI 0x59D0 #define BGE_LPMBX_TX_NIC_PROD10_LO 0x59D4 #define BGE_LPMBX_TX_NIC_PROD11_HI 0x59D8 #define BGE_LPMBX_TX_NIC_PROD11_LO 0x59DC #define BGE_LPMBX_TX_NIC_PROD12_HI 0x59E0 #define BGE_LPMBX_TX_NIC_PROD12_LO 0x59E4 #define BGE_LPMBX_TX_NIC_PROD13_HI 0x59E8 #define BGE_LPMBX_TX_NIC_PROD13_LO 0x59EC #define BGE_LPMBX_TX_NIC_PROD14_HI 0x59F0 #define BGE_LPMBX_TX_NIC_PROD14_LO 0x59F4 #define BGE_LPMBX_TX_NIC_PROD15_HI 0x59F8 #define BGE_LPMBX_TX_NIC_PROD15_LO 0x59FC /* * Flow throw Queue reset register */ #define BGE_FTQ_RESET 0x5C00 #define BGE_FTQRESET_DMAREAD 0x00000002 #define BGE_FTQRESET_DMAHIPRIO_RD 0x00000004 #define BGE_FTQRESET_DMADONE 0x00000010 #define BGE_FTQRESET_SBDC 0x00000020 #define BGE_FTQRESET_SDI 0x00000040 #define BGE_FTQRESET_WDMA 0x00000080 #define BGE_FTQRESET_DMAHIPRIO_WR 0x00000100 #define BGE_FTQRESET_TYPE1_SOFTWARE 0x00000200 #define BGE_FTQRESET_SDC 0x00000400 #define BGE_FTQRESET_HCC 0x00000800 #define BGE_FTQRESET_TXFIFO 0x00001000 #define BGE_FTQRESET_MBC 0x00002000 #define BGE_FTQRESET_RBDC 0x00004000 #define BGE_FTQRESET_RXLP 0x00008000 #define BGE_FTQRESET_RDBDI 0x00010000 #define BGE_FTQRESET_RDC 0x00020000 #define BGE_FTQRESET_TYPE2_SOFTWARE 0x00040000 /* * Message Signaled Interrupt registers */ #define BGE_MSI_MODE 0x6000 #define BGE_MSI_STATUS 0x6004 #define BGE_MSI_FIFOACCESS 0x6008 /* MSI mode register */ #define BGE_MSIMODE_RESET 0x00000001 #define BGE_MSIMODE_ENABLE 0x00000002 #define BGE_MSIMODE_ONE_SHOT_DISABLE 0x00000020 #define BGE_MSIMODE_MULTIVEC_ENABLE 0x00000080 /* MSI status register */ #define BGE_MSISTAT_PCI_TGT_ABRT_ATTN 0x00000004 #define BGE_MSISTAT_PCI_MSTR_ABRT_ATTN 0x00000008 #define BGE_MSISTAT_PCI_PERR_ATTN 0x00000010 #define BGE_MSISTAT_MSI_FIFOUFLOW_ATTN 0x00000020 #define BGE_MSISTAT_MSI_FIFOOFLOW_ATTN 0x00000040 /* * DMA Completion registers */ #define BGE_DMAC_MODE 0x6400 /* DMA Completion mode register */ #define BGE_DMACMODE_RESET 0x00000001 #define BGE_DMACMODE_ENABLE 0x00000002 /* * General control registers. */ #define BGE_MODE_CTL 0x6800 #define BGE_MISC_CFG 0x6804 #define BGE_MISC_LOCAL_CTL 0x6808 #define BGE_CPU_EVENT 0x6810 #define BGE_EE_ADDR 0x6838 #define BGE_EE_DATA 0x683C #define BGE_EE_CTL 0x6840 #define BGE_MDI_CTL 0x6844 #define BGE_EE_DELAY 0x6848 #define BGE_FASTBOOT_PC 0x6894 /* * NVRAM Control registers */ #define BGE_NVRAM_CMD 0x7000 #define BGE_NVRAM_STAT 0x7004 #define BGE_NVRAM_WRDATA 0x7008 #define BGE_NVRAM_ADDR 0x700c #define BGE_NVRAM_RDDATA 0x7010 #define BGE_NVRAM_CFG1 0x7014 #define BGE_NVRAM_CFG2 0x7018 #define BGE_NVRAM_CFG3 0x701c #define BGE_NVRAM_SWARB 0x7020 #define BGE_NVRAM_ACCESS 0x7024 #define BGE_NVRAM_WRITE1 0x7028 #define BGE_NVRAMCMD_RESET 0x00000001 #define BGE_NVRAMCMD_DONE 0x00000008 #define BGE_NVRAMCMD_START 0x00000010 #define BGE_NVRAMCMD_WR 0x00000020 /* 1 = wr, 0 = rd */ #define BGE_NVRAMCMD_ERASE 0x00000040 #define BGE_NVRAMCMD_FIRST 0x00000080 #define BGE_NVRAMCMD_LAST 0x00000100 #define BGE_NVRAM_READCMD \ (BGE_NVRAMCMD_FIRST|BGE_NVRAMCMD_LAST| \ BGE_NVRAMCMD_START|BGE_NVRAMCMD_DONE) #define BGE_NVRAM_WRITECMD \ (BGE_NVRAMCMD_FIRST|BGE_NVRAMCMD_LAST| \ BGE_NVRAMCMD_START|BGE_NVRAMCMD_DONE|BGE_NVRAMCMD_WR) #define BGE_NVRAMSWARB_SET0 0x00000001 #define BGE_NVRAMSWARB_SET1 0x00000002 #define BGE_NVRAMSWARB_SET2 0x00000003 #define BGE_NVRAMSWARB_SET3 0x00000004 #define BGE_NVRAMSWARB_CLR0 0x00000010 #define BGE_NVRAMSWARB_CLR1 0x00000020 #define BGE_NVRAMSWARB_CLR2 0x00000040 #define BGE_NVRAMSWARB_CLR3 0x00000080 #define BGE_NVRAMSWARB_GNT0 0x00000100 #define BGE_NVRAMSWARB_GNT1 0x00000200 #define BGE_NVRAMSWARB_GNT2 0x00000400 #define BGE_NVRAMSWARB_GNT3 0x00000800 #define BGE_NVRAMSWARB_REQ0 0x00001000 #define BGE_NVRAMSWARB_REQ1 0x00002000 #define BGE_NVRAMSWARB_REQ2 0x00004000 #define BGE_NVRAMSWARB_REQ3 0x00008000 #define BGE_NVRAMACC_ENABLE 0x00000001 #define BGE_NVRAMACC_WRENABLE 0x00000002 /* Mode control register */ #define BGE_MODECTL_INT_SNDCOAL_ONLY 0x00000001 #define BGE_MODECTL_BYTESWAP_NONFRAME 0x00000002 #define BGE_MODECTL_WORDSWAP_NONFRAME 0x00000004 #define BGE_MODECTL_BYTESWAP_DATA 0x00000010 #define BGE_MODECTL_WORDSWAP_DATA 0x00000020 #define BGE_MODECTL_NO_FRAME_CRACKING 0x00000200 #define BGE_MODECTL_NO_RX_CRC 0x00000400 #define BGE_MODECTL_RX_BADFRAMES 0x00000800 #define BGE_MODECTL_NO_TX_INTR 0x00002000 #define BGE_MODECTL_NO_RX_INTR 0x00004000 #define BGE_MODECTL_FORCE_PCI32 0x00008000 #define BGE_MODECTL_STACKUP 0x00010000 #define BGE_MODECTL_HOST_SEND_BDS 0x00020000 #define BGE_MODECTL_TX_NO_PHDR_CSUM 0x00100000 #define BGE_MODECTL_RX_NO_PHDR_CSUM 0x00800000 #define BGE_MODECTL_TX_ATTN_INTR 0x01000000 #define BGE_MODECTL_RX_ATTN_INTR 0x02000000 #define BGE_MODECTL_MAC_ATTN_INTR 0x04000000 #define BGE_MODECTL_DMA_ATTN_INTR 0x08000000 #define BGE_MODECTL_FLOWCTL_ATTN_INTR 0x10000000 #define BGE_MODECTL_4X_SENDRING_SZ 0x20000000 #define BGE_MODECTL_FW_PROCESS_MCASTS 0x40000000 /* Misc. config register */ #define BGE_MISCCFG_RESET_CORE_CLOCKS 0x00000001 #define BGE_MISCCFG_TIMER_PRESCALER 0x000000FE #define BGE_MISCCFG_BOARD_ID 0x0001E000 #define BGE_MISCCFG_BOARD_ID_5788 0x00010000 #define BGE_MISCCFG_BOARD_ID_5788M 0x00018000 #define BGE_MISCCFG_EPHY_IDDQ 0x00200000 #define BGE_MISCCFG_GPHY_PD_OVERRIDE 0x04000000 #define BGE_32BITTIME_66MHZ (0x41 << 1) /* Misc. Local Control */ #define BGE_MLC_INTR_STATE 0x00000001 #define BGE_MLC_INTR_CLR 0x00000002 #define BGE_MLC_INTR_SET 0x00000004 #define BGE_MLC_INTR_ONATTN 0x00000008 #define BGE_MLC_MISCIO_IN0 0x00000100 #define BGE_MLC_MISCIO_IN1 0x00000200 #define BGE_MLC_MISCIO_IN2 0x00000400 #define BGE_MLC_MISCIO_OUTEN0 0x00000800 #define BGE_MLC_MISCIO_OUTEN1 0x00001000 #define BGE_MLC_MISCIO_OUTEN2 0x00002000 #define BGE_MLC_MISCIO_OUT0 0x00004000 #define BGE_MLC_MISCIO_OUT1 0x00008000 #define BGE_MLC_MISCIO_OUT2 0x00010000 #define BGE_MLC_EXTRAM_ENB 0x00020000 #define BGE_MLC_SRAM_SIZE 0x001C0000 #define BGE_MLC_BANK_SEL 0x00200000 /* 0 = 2 banks, 1 == 1 */ #define BGE_MLC_SSRAM_TYPE 0x00400000 /* 1 = ZBT, 0 = standard */ #define BGE_MLC_SSRAM_CYC_DESEL 0x00800000 #define BGE_MLC_AUTO_EEPROM 0x01000000 #define BGE_SSRAMSIZE_256KB 0x00000000 #define BGE_SSRAMSIZE_512KB 0x00040000 #define BGE_SSRAMSIZE_1MB 0x00080000 #define BGE_SSRAMSIZE_2MB 0x000C0000 #define BGE_SSRAMSIZE_4MB 0x00100000 #define BGE_SSRAMSIZE_8MB 0x00140000 #define BGE_SSRAMSIZE_16M 0x00180000 /* EEPROM address register */ #define BGE_EEADDR_ADDRESS 0x0000FFFC #define BGE_EEADDR_HALFCLK 0x01FF0000 #define BGE_EEADDR_START 0x02000000 #define BGE_EEADDR_DEVID 0x1C000000 #define BGE_EEADDR_RESET 0x20000000 #define BGE_EEADDR_DONE 0x40000000 #define BGE_EEADDR_RW 0x80000000 /* 1 = rd, 0 = wr */ #define BGE_EEDEVID(x) ((x & 7) << 26) #define BGE_EEHALFCLK(x) ((x & 0x1FF) << 16) #define BGE_HALFCLK_384SCL 0x60 #define BGE_EE_READCMD \ (BGE_EEHALFCLK(BGE_HALFCLK_384SCL)|BGE_EEDEVID(0)| \ BGE_EEADDR_START|BGE_EEADDR_RW|BGE_EEADDR_DONE) #define BGE_EE_WRCMD \ (BGE_EEHALFCLK(BGE_HALFCLK_384SCL)|BGE_EEDEVID(0)| \ BGE_EEADDR_START|BGE_EEADDR_DONE) /* EEPROM Control register */ #define BGE_EECTL_CLKOUT_TRISTATE 0x00000001 #define BGE_EECTL_CLKOUT 0x00000002 #define BGE_EECTL_CLKIN 0x00000004 #define BGE_EECTL_DATAOUT_TRISTATE 0x00000008 #define BGE_EECTL_DATAOUT 0x00000010 #define BGE_EECTL_DATAIN 0x00000020 /* MDI (MII/GMII) access register */ #define BGE_MDI_DATA 0x00000001 #define BGE_MDI_DIR 0x00000002 #define BGE_MDI_SEL 0x00000004 #define BGE_MDI_CLK 0x00000008 #define BGE_MEMWIN_START 0x00008000 #define BGE_MEMWIN_END 0x0000FFFF #define BGE_MEMWIN_READ(sc, x, val) \ do { \ pci_write_config(sc->bge_dev, BGE_PCI_MEMWIN_BASEADDR, \ (0xFFFF0000 & x), 4); \ val = CSR_READ_4(sc, BGE_MEMWIN_START + (x & 0xFFFF)); \ } while(0) #define BGE_MEMWIN_WRITE(sc, x, val) \ do { \ pci_write_config(sc->bge_dev, BGE_PCI_MEMWIN_BASEADDR, \ (0xFFFF0000 & x), 4); \ CSR_WRITE_4(sc, BGE_MEMWIN_START + (x & 0xFFFF), val); \ } while(0) /* * This magic number is written to the firmware mailbox at 0xb50 * before a software reset is issued. After the internal firmware * has completed its initialization it will write the opposite of * this value, ~BGE_MAGIC_NUMBER, to the same location, allowing the * driver to synchronize with the firmware. */ #define BGE_MAGIC_NUMBER 0x4B657654 typedef struct { uint32_t bge_addr_hi; uint32_t bge_addr_lo; } bge_hostaddr; #define BGE_HOSTADDR(x, y) \ do { \ (x).bge_addr_lo = ((uint64_t) (y) & 0xffffffff); \ (x).bge_addr_hi = ((uint64_t) (y) >> 32); \ } while(0) #define BGE_ADDR_LO(y) \ ((uint64_t) (y) & 0xFFFFFFFF) #define BGE_ADDR_HI(y) \ ((uint64_t) (y) >> 32) /* Ring control block structure */ struct bge_rcb { bge_hostaddr bge_hostaddr; uint32_t bge_maxlen_flags; uint32_t bge_nicaddr; }; #define RCB_WRITE_4(sc, rcb, offset, val) \ bus_write_4(sc->bge_res, rcb + offsetof(struct bge_rcb, offset), val) #define BGE_RCB_MAXLEN_FLAGS(maxlen, flags) ((maxlen) << 16 | (flags)) #define BGE_RCB_FLAG_USE_EXT_RX_BD 0x0001 #define BGE_RCB_FLAG_RING_DISABLED 0x0002 struct bge_tx_bd { bge_hostaddr bge_addr; #if BYTE_ORDER == LITTLE_ENDIAN uint16_t bge_flags; uint16_t bge_len; uint16_t bge_vlan_tag; uint16_t bge_mss; #else uint16_t bge_len; uint16_t bge_flags; uint16_t bge_mss; uint16_t bge_vlan_tag; #endif }; #define BGE_TXBDFLAG_TCP_UDP_CSUM 0x0001 #define BGE_TXBDFLAG_IP_CSUM 0x0002 #define BGE_TXBDFLAG_END 0x0004 #define BGE_TXBDFLAG_IP_FRAG 0x0008 #define BGE_TXBDFLAG_JUMBO_FRAME 0x0008 /* 5717 */ #define BGE_TXBDFLAG_IP_FRAG_END 0x0010 #define BGE_TXBDFLAG_HDRLEN_BIT2 0x0010 /* 5717 */ #define BGE_TXBDFLAG_SNAP 0x0020 /* 5717 */ #define BGE_TXBDFLAG_VLAN_TAG 0x0040 #define BGE_TXBDFLAG_COAL_NOW 0x0080 #define BGE_TXBDFLAG_CPU_PRE_DMA 0x0100 #define BGE_TXBDFLAG_CPU_POST_DMA 0x0200 #define BGE_TXBDFLAG_HDRLEN_BIT3 0x0400 /* 5717 */ #define BGE_TXBDFLAG_HDRLEN_BIT4 0x0800 /* 5717 */ #define BGE_TXBDFLAG_INSERT_SRC_ADDR 0x1000 #define BGE_TXBDFLAG_HDRLEN_BIT5 0x1000 /* 5717 */ #define BGE_TXBDFLAG_HDRLEN_BIT6 0x2000 /* 5717 */ #define BGE_TXBDFLAG_HDRLEN_BIT7 0x4000 /* 5717 */ #define BGE_TXBDFLAG_CHOOSE_SRC_ADDR 0x6000 #define BGE_TXBDFLAG_NO_CRC 0x8000 #define BGE_TXBDFLAG_MSS_SIZE_MASK 0x3FFF /* 5717 */ /* Bits [1:0] of the MSS header length. */ #define BGE_TXBDFLAG_MSS_HDRLEN_MASK 0xC000 /* 5717 */ #define BGE_NIC_TXRING_ADDR(ringno, size) \ BGE_SEND_RING_1_TO_4 + \ ((ringno * sizeof(struct bge_tx_bd) * size) / 4) struct bge_rx_bd { bge_hostaddr bge_addr; #if BYTE_ORDER == LITTLE_ENDIAN uint16_t bge_len; uint16_t bge_idx; uint16_t bge_flags; uint16_t bge_type; uint16_t bge_tcp_udp_csum; uint16_t bge_ip_csum; uint16_t bge_vlan_tag; uint16_t bge_error_flag; #else uint16_t bge_idx; uint16_t bge_len; uint16_t bge_type; uint16_t bge_flags; uint16_t bge_ip_csum; uint16_t bge_tcp_udp_csum; uint16_t bge_error_flag; uint16_t bge_vlan_tag; #endif uint32_t bge_rsvd; uint32_t bge_opaque; }; struct bge_extrx_bd { bge_hostaddr bge_addr1; bge_hostaddr bge_addr2; bge_hostaddr bge_addr3; #if BYTE_ORDER == LITTLE_ENDIAN uint16_t bge_len2; uint16_t bge_len1; uint16_t bge_rsvd1; uint16_t bge_len3; #else uint16_t bge_len1; uint16_t bge_len2; uint16_t bge_len3; uint16_t bge_rsvd1; #endif bge_hostaddr bge_addr0; #if BYTE_ORDER == LITTLE_ENDIAN uint16_t bge_len0; uint16_t bge_idx; uint16_t bge_flags; uint16_t bge_type; uint16_t bge_tcp_udp_csum; uint16_t bge_ip_csum; uint16_t bge_vlan_tag; uint16_t bge_error_flag; #else uint16_t bge_idx; uint16_t bge_len0; uint16_t bge_type; uint16_t bge_flags; uint16_t bge_ip_csum; uint16_t bge_tcp_udp_csum; uint16_t bge_error_flag; uint16_t bge_vlan_tag; #endif uint32_t bge_rsvd0; uint32_t bge_opaque; }; #define BGE_RXBDFLAG_END 0x0004 #define BGE_RXBDFLAG_JUMBO_RING 0x0020 #define BGE_RXBDFLAG_VLAN_TAG 0x0040 #define BGE_RXBDFLAG_ERROR 0x0400 #define BGE_RXBDFLAG_MINI_RING 0x0800 #define BGE_RXBDFLAG_IP_CSUM 0x1000 #define BGE_RXBDFLAG_TCP_UDP_CSUM 0x2000 #define BGE_RXBDFLAG_TCP_UDP_IS_TCP 0x4000 #define BGE_RXBDFLAG_IPV6 0x8000 #define BGE_RXERRFLAG_BAD_CRC 0x0001 #define BGE_RXERRFLAG_COLL_DETECT 0x0002 #define BGE_RXERRFLAG_LINK_LOST 0x0004 #define BGE_RXERRFLAG_PHY_DECODE_ERR 0x0008 #define BGE_RXERRFLAG_MAC_ABORT 0x0010 #define BGE_RXERRFLAG_RUNT 0x0020 #define BGE_RXERRFLAG_TRUNC_NO_RSRCS 0x0040 #define BGE_RXERRFLAG_GIANT 0x0080 #define BGE_RXERRFLAG_IP_CSUM_NOK 0x1000 /* 5717 */ struct bge_sts_idx { #if BYTE_ORDER == LITTLE_ENDIAN uint16_t bge_rx_prod_idx; uint16_t bge_tx_cons_idx; #else uint16_t bge_tx_cons_idx; uint16_t bge_rx_prod_idx; #endif }; struct bge_status_block { uint32_t bge_status; uint32_t bge_status_tag; #if BYTE_ORDER == LITTLE_ENDIAN uint16_t bge_rx_jumbo_cons_idx; uint16_t bge_rx_std_cons_idx; uint16_t bge_rx_mini_cons_idx; uint16_t bge_rsvd1; #else uint16_t bge_rx_std_cons_idx; uint16_t bge_rx_jumbo_cons_idx; uint16_t bge_rsvd1; uint16_t bge_rx_mini_cons_idx; #endif struct bge_sts_idx bge_idx[16]; }; #define BGE_STATFLAG_UPDATED 0x00000001 #define BGE_STATFLAG_LINKSTATE_CHANGED 0x00000002 #define BGE_STATFLAG_ERROR 0x00000004 /* * Broadcom Vendor ID * (Note: the BCM570x still defaults to the Alteon PCI vendor ID * even though they're now manufactured by Broadcom) */ #define BCOM_VENDORID 0x14E4 #define BCOM_DEVICEID_BCM5700 0x1644 #define BCOM_DEVICEID_BCM5701 0x1645 #define BCOM_DEVICEID_BCM5702 0x1646 #define BCOM_DEVICEID_BCM5702X 0x16A6 #define BCOM_DEVICEID_BCM5702_ALT 0x16C6 #define BCOM_DEVICEID_BCM5703 0x1647 #define BCOM_DEVICEID_BCM5703X 0x16A7 #define BCOM_DEVICEID_BCM5703_ALT 0x16C7 #define BCOM_DEVICEID_BCM5704C 0x1648 #define BCOM_DEVICEID_BCM5704S 0x16A8 #define BCOM_DEVICEID_BCM5704S_ALT 0x1649 #define BCOM_DEVICEID_BCM5705 0x1653 #define BCOM_DEVICEID_BCM5705K 0x1654 #define BCOM_DEVICEID_BCM5705F 0x166E #define BCOM_DEVICEID_BCM5705M 0x165D #define BCOM_DEVICEID_BCM5705M_ALT 0x165E #define BCOM_DEVICEID_BCM5714C 0x1668 #define BCOM_DEVICEID_BCM5714S 0x1669 #define BCOM_DEVICEID_BCM5715 0x1678 #define BCOM_DEVICEID_BCM5715S 0x1679 #define BCOM_DEVICEID_BCM5717 0x1655 #define BCOM_DEVICEID_BCM5718 0x1656 #define BCOM_DEVICEID_BCM5720 0x1658 #define BCOM_DEVICEID_BCM5721 0x1659 #define BCOM_DEVICEID_BCM5722 0x165A #define BCOM_DEVICEID_BCM5723 0x165B #define BCOM_DEVICEID_BCM5750 0x1676 #define BCOM_DEVICEID_BCM5750M 0x167C #define BCOM_DEVICEID_BCM5751 0x1677 #define BCOM_DEVICEID_BCM5751F 0x167E #define BCOM_DEVICEID_BCM5751M 0x167D #define BCOM_DEVICEID_BCM5752 0x1600 #define BCOM_DEVICEID_BCM5752M 0x1601 #define BCOM_DEVICEID_BCM5753 0x16F7 #define BCOM_DEVICEID_BCM5753F 0x16FE #define BCOM_DEVICEID_BCM5753M 0x16FD #define BCOM_DEVICEID_BCM5754 0x167A #define BCOM_DEVICEID_BCM5754M 0x1672 #define BCOM_DEVICEID_BCM5755 0x167B #define BCOM_DEVICEID_BCM5755M 0x1673 #define BCOM_DEVICEID_BCM5756 0x1674 #define BCOM_DEVICEID_BCM5761 0x1681 #define BCOM_DEVICEID_BCM5761E 0x1680 #define BCOM_DEVICEID_BCM5761S 0x1688 #define BCOM_DEVICEID_BCM5761SE 0x1689 #define BCOM_DEVICEID_BCM5764 0x1684 #define BCOM_DEVICEID_BCM5780 0x166A #define BCOM_DEVICEID_BCM5780S 0x166B #define BCOM_DEVICEID_BCM5781 0x16DD #define BCOM_DEVICEID_BCM5782 0x1696 #define BCOM_DEVICEID_BCM5784 0x1698 #define BCOM_DEVICEID_BCM5785F 0x16a0 #define BCOM_DEVICEID_BCM5785G 0x1699 #define BCOM_DEVICEID_BCM5786 0x169A #define BCOM_DEVICEID_BCM5787 0x169B #define BCOM_DEVICEID_BCM5787M 0x1693 #define BCOM_DEVICEID_BCM5787F 0x167f #define BCOM_DEVICEID_BCM5788 0x169C #define BCOM_DEVICEID_BCM5789 0x169D #define BCOM_DEVICEID_BCM5901 0x170D #define BCOM_DEVICEID_BCM5901A2 0x170E #define BCOM_DEVICEID_BCM5903M 0x16FF #define BCOM_DEVICEID_BCM5906 0x1712 #define BCOM_DEVICEID_BCM5906M 0x1713 #define BCOM_DEVICEID_BCM57760 0x1690 +#define BCOM_DEVICEID_BCM57761 0x16B0 +#define BCOM_DEVICEID_BCM57765 0x16B4 #define BCOM_DEVICEID_BCM57780 0x1692 +#define BCOM_DEVICEID_BCM57781 0x16B1 +#define BCOM_DEVICEID_BCM57785 0x16B5 #define BCOM_DEVICEID_BCM57788 0x1691 #define BCOM_DEVICEID_BCM57790 0x1694 +#define BCOM_DEVICEID_BCM57791 0x16B2 +#define BCOM_DEVICEID_BCM57795 0x16B6 /* * Alteon AceNIC PCI vendor/device ID. */ #define ALTEON_VENDORID 0x12AE #define ALTEON_DEVICEID_ACENIC 0x0001 #define ALTEON_DEVICEID_ACENIC_COPPER 0x0002 #define ALTEON_DEVICEID_BCM5700 0x0003 #define ALTEON_DEVICEID_BCM5701 0x0004 /* * 3Com 3c996 PCI vendor/device ID. */ #define TC_VENDORID 0x10B7 #define TC_DEVICEID_3C996 0x0003 /* * SysKonnect PCI vendor ID */ #define SK_VENDORID 0x1148 #define SK_DEVICEID_ALTIMA 0x4400 #define SK_SUBSYSID_9D21 0x4421 #define SK_SUBSYSID_9D41 0x4441 /* * Altima PCI vendor/device ID. */ #define ALTIMA_VENDORID 0x173b #define ALTIMA_DEVICE_AC1000 0x03e8 #define ALTIMA_DEVICE_AC1002 0x03e9 #define ALTIMA_DEVICE_AC9100 0x03ea /* * Dell PCI vendor ID */ #define DELL_VENDORID 0x1028 /* * Apple PCI vendor ID. */ #define APPLE_VENDORID 0x106b #define APPLE_DEVICE_BCM5701 0x1645 /* * Sun PCI vendor ID */ #define SUN_VENDORID 0x108e /* * Fujitsu vendor/device IDs */ #define FJTSU_VENDORID 0x10cf #define FJTSU_DEVICEID_PW008GE5 0x11a1 #define FJTSU_DEVICEID_PW008GE4 0x11a2 #define FJTSU_DEVICEID_PP250450 0x11cc /* PRIMEPOWER250/450 LAN */ /* * Offset of MAC address inside EEPROM. */ #define BGE_EE_MAC_OFFSET 0x7C #define BGE_EE_MAC_OFFSET_5906 0x10 #define BGE_EE_HWCFG_OFFSET 0xC8 #define BGE_HWCFG_VOLTAGE 0x00000003 #define BGE_HWCFG_PHYLED_MODE 0x0000000C #define BGE_HWCFG_MEDIA 0x00000030 #define BGE_HWCFG_ASF 0x00000080 #define BGE_VOLTAGE_1POINT3 0x00000000 #define BGE_VOLTAGE_1POINT8 0x00000001 #define BGE_PHYLEDMODE_UNSPEC 0x00000000 #define BGE_PHYLEDMODE_TRIPLELED 0x00000004 #define BGE_PHYLEDMODE_SINGLELED 0x00000008 #define BGE_MEDIA_UNSPEC 0x00000000 #define BGE_MEDIA_COPPER 0x00000010 #define BGE_MEDIA_FIBER 0x00000020 #define BGE_TICKS_PER_SEC 1000000 /* * Ring size constants. */ #define BGE_EVENT_RING_CNT 256 #define BGE_CMD_RING_CNT 64 #define BGE_STD_RX_RING_CNT 512 #define BGE_JUMBO_RX_RING_CNT 256 #define BGE_MINI_RX_RING_CNT 1024 #define BGE_RETURN_RING_CNT 1024 /* 5705 has smaller return ring size */ #define BGE_RETURN_RING_CNT_5705 512 /* * Possible TX ring sizes. */ #define BGE_TX_RING_CNT_128 128 #define BGE_TX_RING_BASE_128 0x3800 #define BGE_TX_RING_CNT_256 256 #define BGE_TX_RING_BASE_256 0x3000 #define BGE_TX_RING_CNT_512 512 #define BGE_TX_RING_BASE_512 0x2000 #define BGE_TX_RING_CNT BGE_TX_RING_CNT_512 #define BGE_TX_RING_BASE BGE_TX_RING_BASE_512 /* * Tigon III statistics counters. */ /* Statistics maintained MAC Receive block. */ struct bge_rx_mac_stats { bge_hostaddr ifHCInOctets; bge_hostaddr Reserved1; bge_hostaddr etherStatsFragments; bge_hostaddr ifHCInUcastPkts; bge_hostaddr ifHCInMulticastPkts; bge_hostaddr ifHCInBroadcastPkts; bge_hostaddr dot3StatsFCSErrors; bge_hostaddr dot3StatsAlignmentErrors; bge_hostaddr xonPauseFramesReceived; bge_hostaddr xoffPauseFramesReceived; bge_hostaddr macControlFramesReceived; bge_hostaddr xoffStateEntered; bge_hostaddr dot3StatsFramesTooLong; bge_hostaddr etherStatsJabbers; bge_hostaddr etherStatsUndersizePkts; bge_hostaddr inRangeLengthError; bge_hostaddr outRangeLengthError; bge_hostaddr etherStatsPkts64Octets; bge_hostaddr etherStatsPkts65Octetsto127Octets; bge_hostaddr etherStatsPkts128Octetsto255Octets; bge_hostaddr etherStatsPkts256Octetsto511Octets; bge_hostaddr etherStatsPkts512Octetsto1023Octets; bge_hostaddr etherStatsPkts1024Octetsto1522Octets; bge_hostaddr etherStatsPkts1523Octetsto2047Octets; bge_hostaddr etherStatsPkts2048Octetsto4095Octets; bge_hostaddr etherStatsPkts4096Octetsto8191Octets; bge_hostaddr etherStatsPkts8192Octetsto9022Octets; }; /* Statistics maintained MAC Transmit block. */ struct bge_tx_mac_stats { bge_hostaddr ifHCOutOctets; bge_hostaddr Reserved2; bge_hostaddr etherStatsCollisions; bge_hostaddr outXonSent; bge_hostaddr outXoffSent; bge_hostaddr flowControlDone; bge_hostaddr dot3StatsInternalMacTransmitErrors; bge_hostaddr dot3StatsSingleCollisionFrames; bge_hostaddr dot3StatsMultipleCollisionFrames; bge_hostaddr dot3StatsDeferredTransmissions; bge_hostaddr Reserved3; bge_hostaddr dot3StatsExcessiveCollisions; bge_hostaddr dot3StatsLateCollisions; bge_hostaddr dot3Collided2Times; bge_hostaddr dot3Collided3Times; bge_hostaddr dot3Collided4Times; bge_hostaddr dot3Collided5Times; bge_hostaddr dot3Collided6Times; bge_hostaddr dot3Collided7Times; bge_hostaddr dot3Collided8Times; bge_hostaddr dot3Collided9Times; bge_hostaddr dot3Collided10Times; bge_hostaddr dot3Collided11Times; bge_hostaddr dot3Collided12Times; bge_hostaddr dot3Collided13Times; bge_hostaddr dot3Collided14Times; bge_hostaddr dot3Collided15Times; bge_hostaddr ifHCOutUcastPkts; bge_hostaddr ifHCOutMulticastPkts; bge_hostaddr ifHCOutBroadcastPkts; bge_hostaddr dot3StatsCarrierSenseErrors; bge_hostaddr ifOutDiscards; bge_hostaddr ifOutErrors; }; /* Stats counters access through registers */ struct bge_mac_stats { /* TX MAC statistics */ uint64_t ifHCOutOctets; uint64_t Reserved0; uint64_t etherStatsCollisions; uint64_t outXonSent; uint64_t outXoffSent; uint64_t Reserved1; uint64_t dot3StatsInternalMacTransmitErrors; uint64_t dot3StatsSingleCollisionFrames; uint64_t dot3StatsMultipleCollisionFrames; uint64_t dot3StatsDeferredTransmissions; uint64_t Reserved2; uint64_t dot3StatsExcessiveCollisions; uint64_t dot3StatsLateCollisions; uint64_t Reserved3[14]; uint64_t ifHCOutUcastPkts; uint64_t ifHCOutMulticastPkts; uint64_t ifHCOutBroadcastPkts; uint64_t Reserved4[2]; /* RX MAC statistics */ uint64_t ifHCInOctets; uint64_t Reserved5; uint64_t etherStatsFragments; uint64_t ifHCInUcastPkts; uint64_t ifHCInMulticastPkts; uint64_t ifHCInBroadcastPkts; uint64_t dot3StatsFCSErrors; uint64_t dot3StatsAlignmentErrors; uint64_t xonPauseFramesReceived; uint64_t xoffPauseFramesReceived; uint64_t macControlFramesReceived; uint64_t xoffStateEntered; uint64_t dot3StatsFramesTooLong; uint64_t etherStatsJabbers; uint64_t etherStatsUndersizePkts; /* Receive List Placement control */ uint64_t FramesDroppedDueToFilters; uint64_t DmaWriteQueueFull; uint64_t DmaWriteHighPriQueueFull; uint64_t NoMoreRxBDs; uint64_t InputDiscards; uint64_t InputErrors; uint64_t RecvThresholdHit; }; struct bge_stats { uint8_t Reserved0[256]; /* Statistics maintained by Receive MAC. */ struct bge_rx_mac_stats rxstats; bge_hostaddr Unused1[37]; /* Statistics maintained by Transmit MAC. */ struct bge_tx_mac_stats txstats; bge_hostaddr Unused2[31]; /* Statistics maintained by Receive List Placement. */ bge_hostaddr COSIfHCInPkts[16]; bge_hostaddr COSFramesDroppedDueToFilters; bge_hostaddr nicDmaWriteQueueFull; bge_hostaddr nicDmaWriteHighPriQueueFull; bge_hostaddr nicNoMoreRxBDs; bge_hostaddr ifInDiscards; bge_hostaddr ifInErrors; bge_hostaddr nicRecvThresholdHit; bge_hostaddr Unused3[9]; /* Statistics maintained by Send Data Initiator. */ bge_hostaddr COSIfHCOutPkts[16]; bge_hostaddr nicDmaReadQueueFull; bge_hostaddr nicDmaReadHighPriQueueFull; bge_hostaddr nicSendDataCompQueueFull; /* Statistics maintained by Host Coalescing. */ bge_hostaddr nicRingSetSendProdIndex; bge_hostaddr nicRingStatusUpdate; bge_hostaddr nicInterrupts; bge_hostaddr nicAvoidedInterrupts; bge_hostaddr nicSendThresholdHit; uint8_t Reserved4[320]; }; /* * Tigon general information block. This resides in host memory * and contains the status counters, ring control blocks and * producer pointers. */ struct bge_gib { struct bge_stats bge_stats; struct bge_rcb bge_tx_rcb[16]; struct bge_rcb bge_std_rx_rcb; struct bge_rcb bge_jumbo_rx_rcb; struct bge_rcb bge_mini_rx_rcb; struct bge_rcb bge_return_rcb; }; #define BGE_FRAMELEN 1518 #define BGE_MAX_FRAMELEN 1536 #define BGE_JUMBO_FRAMELEN 9018 #define BGE_JUMBO_MTU (BGE_JUMBO_FRAMELEN-ETHER_HDR_LEN-ETHER_CRC_LEN) #define BGE_MIN_FRAMELEN 60 /* * Other utility macros. */ #define BGE_INC(x, y) (x) = (x + 1) % y /* * Register access macros. The Tigon always uses memory mapped register * accesses and all registers must be accessed with 32 bit operations. */ #define CSR_WRITE_4(sc, reg, val) \ bus_write_4(sc->bge_res, reg, val) #define CSR_READ_4(sc, reg) \ bus_read_4(sc->bge_res, reg) #define BGE_SETBIT(sc, reg, x) \ CSR_WRITE_4(sc, reg, (CSR_READ_4(sc, reg) | (x))) #define BGE_CLRBIT(sc, reg, x) \ CSR_WRITE_4(sc, reg, (CSR_READ_4(sc, reg) & ~(x))) #define PCI_SETBIT(dev, reg, x, s) \ pci_write_config(dev, reg, (pci_read_config(dev, reg, s) | (x)), s) #define PCI_CLRBIT(dev, reg, x, s) \ pci_write_config(dev, reg, (pci_read_config(dev, reg, s) & ~(x)), s) /* * Memory management stuff. */ #define BGE_NSEG_JUMBO 4 #define BGE_NSEG_NEW 32 #define BGE_TSOSEG_SZ 4096 /* Maximum DMA address for controllers that have 40bit DMA address bug. */ #if (BUS_SPACE_MAXADDR < 0xFFFFFFFFFF) #define BGE_DMA_MAXADDR BUS_SPACE_MAXADDR #else #define BGE_DMA_MAXADDR 0xFFFFFFFFFF #endif #ifdef PAE #define BGE_DMA_BNDRY 0x80000000 #else #if (BUS_SPACE_MAXADDR > 0xFFFFFFFF) #define BGE_DMA_BNDRY 0x100000000 #else #define BGE_DMA_BNDRY 0 #endif #endif /* * Ring structures. Most of these reside in host memory and we tell * the NIC where they are via the ring control blocks. The exceptions * are the tx and command rings, which live in NIC memory and which * we access via the shared memory window. */ struct bge_ring_data { struct bge_rx_bd *bge_rx_std_ring; bus_addr_t bge_rx_std_ring_paddr; struct bge_extrx_bd *bge_rx_jumbo_ring; bus_addr_t bge_rx_jumbo_ring_paddr; struct bge_rx_bd *bge_rx_return_ring; bus_addr_t bge_rx_return_ring_paddr; struct bge_tx_bd *bge_tx_ring; bus_addr_t bge_tx_ring_paddr; struct bge_status_block *bge_status_block; bus_addr_t bge_status_block_paddr; struct bge_stats *bge_stats; bus_addr_t bge_stats_paddr; struct bge_gib bge_info; }; #define BGE_STD_RX_RING_SZ \ (sizeof(struct bge_rx_bd) * BGE_STD_RX_RING_CNT) #define BGE_JUMBO_RX_RING_SZ \ (sizeof(struct bge_extrx_bd) * BGE_JUMBO_RX_RING_CNT) #define BGE_TX_RING_SZ \ (sizeof(struct bge_tx_bd) * BGE_TX_RING_CNT) #define BGE_RX_RTN_RING_SZ(x) \ (sizeof(struct bge_rx_bd) * x->bge_return_ring_cnt) #define BGE_STATUS_BLK_SZ sizeof (struct bge_status_block) #define BGE_STATS_SZ sizeof (struct bge_stats) /* * Mbuf pointers. We need these to keep track of the virtual addresses * of our mbuf chains since we can only convert from physical to virtual, * not the other way around. */ struct bge_chain_data { bus_dma_tag_t bge_parent_tag; bus_dma_tag_t bge_buffer_tag; bus_dma_tag_t bge_rx_std_ring_tag; bus_dma_tag_t bge_rx_jumbo_ring_tag; bus_dma_tag_t bge_rx_return_ring_tag; bus_dma_tag_t bge_tx_ring_tag; bus_dma_tag_t bge_status_tag; bus_dma_tag_t bge_stats_tag; bus_dma_tag_t bge_rx_mtag; /* Rx mbuf mapping tag */ bus_dma_tag_t bge_tx_mtag; /* Tx mbuf mapping tag */ bus_dma_tag_t bge_mtag_jumbo; /* Jumbo mbuf mapping tag */ bus_dmamap_t bge_tx_dmamap[BGE_TX_RING_CNT]; bus_dmamap_t bge_rx_std_sparemap; bus_dmamap_t bge_rx_std_dmamap[BGE_STD_RX_RING_CNT]; bus_dmamap_t bge_rx_jumbo_sparemap; bus_dmamap_t bge_rx_jumbo_dmamap[BGE_JUMBO_RX_RING_CNT]; bus_dmamap_t bge_rx_std_ring_map; bus_dmamap_t bge_rx_jumbo_ring_map; bus_dmamap_t bge_tx_ring_map; bus_dmamap_t bge_rx_return_ring_map; bus_dmamap_t bge_status_map; bus_dmamap_t bge_stats_map; struct mbuf *bge_tx_chain[BGE_TX_RING_CNT]; struct mbuf *bge_rx_std_chain[BGE_STD_RX_RING_CNT]; struct mbuf *bge_rx_jumbo_chain[BGE_JUMBO_RX_RING_CNT]; int bge_rx_std_seglen[BGE_STD_RX_RING_CNT]; int bge_rx_jumbo_seglen[BGE_JUMBO_RX_RING_CNT][4]; }; struct bge_dmamap_arg { bus_addr_t bge_busaddr; }; #define BGE_HWREV_TIGON 0x01 #define BGE_HWREV_TIGON_II 0x02 #define BGE_TIMEOUT 100000 #define BGE_TXCONS_UNSET 0xFFFF /* impossible value */ struct bge_bcom_hack { int reg; int val; }; #define ASF_ENABLE 1 #define ASF_NEW_HANDSHAKE 2 #define ASF_STACKUP 4 struct bge_softc { struct ifnet *bge_ifp; /* interface info */ device_t bge_dev; struct mtx bge_mtx; device_t bge_miibus; void *bge_intrhand; struct resource *bge_irq; struct resource *bge_res; struct ifmedia bge_ifmedia; /* TBI media info */ int bge_expcap; int bge_msicap; int bge_pcixcap; uint32_t bge_flags; #define BGE_FLAG_TBI 0x00000001 #define BGE_FLAG_JUMBO 0x00000002 #define BGE_FLAG_JUMBO_STD 0x00000004 #define BGE_FLAG_EADDR 0x00000008 #define BGE_FLAG_MII_SERDES 0x00000010 #define BGE_FLAG_CPMU_PRESENT 0x00000020 #define BGE_FLAG_TAGGED_STATUS 0x00000040 #define BGE_FLAG_MSI 0x00000100 #define BGE_FLAG_PCIX 0x00000200 #define BGE_FLAG_PCIE 0x00000400 #define BGE_FLAG_TSO 0x00000800 #define BGE_FLAG_TSO3 0x00001000 #define BGE_FLAG_JUMBO_FRAME 0x00002000 #define BGE_FLAG_5700_FAMILY 0x00010000 #define BGE_FLAG_5705_PLUS 0x00020000 #define BGE_FLAG_5714_FAMILY 0x00040000 #define BGE_FLAG_575X_PLUS 0x00080000 #define BGE_FLAG_5755_PLUS 0x00100000 #define BGE_FLAG_5788 0x00200000 #define BGE_FLAG_5717_PLUS 0x00400000 #define BGE_FLAG_40BIT_BUG 0x01000000 #define BGE_FLAG_4G_BNDRY_BUG 0x02000000 #define BGE_FLAG_RX_ALIGNBUG 0x04000000 #define BGE_FLAG_SHORT_DMA_BUG 0x08000000 uint32_t bge_phy_flags; -#define BGE_PHY_WIRESPEED 0x00000001 +#define BGE_PHY_NO_WIRESPEED 0x00000001 #define BGE_PHY_ADC_BUG 0x00000002 #define BGE_PHY_5704_A0_BUG 0x00000004 #define BGE_PHY_JITTER_BUG 0x00000008 #define BGE_PHY_BER_BUG 0x00000010 #define BGE_PHY_ADJUST_TRIM 0x00000020 #define BGE_PHY_CRC_BUG 0x00000040 #define BGE_PHY_NO_3LED 0x00000080 uint32_t bge_chipid; uint32_t bge_asicrev; uint32_t bge_chiprev; uint8_t bge_asf_mode; uint8_t bge_asf_count; struct bge_ring_data bge_ldata; /* rings */ struct bge_chain_data bge_cdata; /* mbufs */ uint16_t bge_tx_saved_considx; uint16_t bge_rx_saved_considx; uint16_t bge_ev_saved_considx; uint16_t bge_return_ring_cnt; uint16_t bge_std; /* current std ring head */ uint16_t bge_jumbo; /* current jumo ring head */ uint32_t bge_stat_ticks; uint32_t bge_rx_coal_ticks; uint32_t bge_tx_coal_ticks; uint32_t bge_tx_prodidx; uint32_t bge_rx_max_coal_bds; uint32_t bge_tx_max_coal_bds; uint32_t bge_mi_mode; int bge_if_flags; int bge_txcnt; int bge_link; /* link state */ int bge_link_evt; /* pending link event */ int bge_timer; int bge_forced_collapse; int bge_forced_udpcsum; int bge_csum_features; struct callout bge_stat_ch; uint32_t bge_rx_discards; uint32_t bge_tx_discards; uint32_t bge_tx_collisions; #ifdef DEVICE_POLLING int rxcycles; #endif /* DEVICE_POLLING */ struct bge_mac_stats bge_mac_stats; struct task bge_intr_task; struct taskqueue *bge_tq; }; #define BGE_LOCK_INIT(_sc, _name) \ mtx_init(&(_sc)->bge_mtx, _name, MTX_NETWORK_LOCK, MTX_DEF) #define BGE_LOCK(_sc) mtx_lock(&(_sc)->bge_mtx) #define BGE_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->bge_mtx, MA_OWNED) #define BGE_UNLOCK(_sc) mtx_unlock(&(_sc)->bge_mtx) #define BGE_LOCK_DESTROY(_sc) mtx_destroy(&(_sc)->bge_mtx) Index: projects/largeSMP/sys/dev/cxgbe/offload.h =================================================================== --- projects/largeSMP/sys/dev/cxgbe/offload.h (revision 221494) +++ projects/largeSMP/sys/dev/cxgbe/offload.h (revision 221495) @@ -1,86 +1,88 @@ /*- * Copyright (c) 2010 Chelsio Communications, Inc. * All rights reserved. * Written by: Navdeep Parhar * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ * */ #ifndef __T4_OFFLOAD_H__ #define __T4_OFFLOAD_H__ /* * Max # of ATIDs. The absolute HW max is 16K but we keep it lower. */ #define MAX_ATIDS 8192U struct serv_entry { void *data; }; union aopen_entry { void *data; union aopen_entry *next; }; /* * Holds the size, base address, free list start, etc of the TID, server TID, * and active-open TID tables. The tables themselves are allocated dynamically. */ struct tid_info { void **tid_tab; unsigned int ntids; struct serv_entry *stid_tab; unsigned long *stid_bmap; unsigned int nstids; unsigned int stid_base; union aopen_entry *atid_tab; unsigned int natids; + struct filter_entry *ftid_tab; unsigned int nftids; unsigned int ftid_base; + unsigned int ftids_in_use; union aopen_entry *afree; unsigned int atids_in_use; unsigned int stids_in_use; }; struct t4_range { unsigned int start; unsigned int size; }; struct t4_virt_res { /* virtualized HW resources */ struct t4_range ddp; struct t4_range iscsi; struct t4_range stag; struct t4_range rq; struct t4_range pbl; }; #endif Index: projects/largeSMP/sys/dev/cxgbe/t4_ioctl.h =================================================================== --- projects/largeSMP/sys/dev/cxgbe/t4_ioctl.h (revision 221494) +++ projects/largeSMP/sys/dev/cxgbe/t4_ioctl.h (revision 221495) @@ -1,59 +1,193 @@ /*- * Copyright (c) 2011 Chelsio Communications, Inc. * All rights reserved. * Written by: Navdeep Parhar * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ * */ #ifndef __T4_IOCTL_H__ #define __T4_IOCTL_H__ +#include +#include + /* * Ioctl commands specific to this driver. */ enum { T4_GETREG = 0x40, /* read register */ T4_SETREG, /* write register */ T4_REGDUMP, /* dump of all registers */ + T4_GET_FILTER_MODE, /* get global filter mode */ + T4_SET_FILTER_MODE, /* set global filter mode */ + T4_GET_FILTER, /* get information about a filter */ + T4_SET_FILTER, /* program a filter */ + T4_DEL_FILTER, /* delete a filter */ }; struct t4_reg { uint32_t addr; uint32_t size; uint64_t val; }; #define T4_REGDUMP_SIZE (160 * 1024) struct t4_regdump { uint32_t version; uint32_t len; /* bytes */ uint32_t *data; }; +/* + * A hardware filter is some valid combination of these. + */ +#define T4_FILTER_IPv4 0x1 /* IPv4 packet */ +#define T4_FILTER_IPv6 0x2 /* IPv6 packet */ +#define T4_FILTER_IP_SADDR 0x4 /* Source IP address or network */ +#define T4_FILTER_IP_DADDR 0x8 /* Destination IP address or network */ +#define T4_FILTER_IP_SPORT 0x10 /* Source IP port */ +#define T4_FILTER_IP_DPORT 0x20 /* Destination IP port */ +#define T4_FILTER_FCoE 0x40 /* Fibre Channel over Ethernet packet */ +#define T4_FILTER_PORT 0x80 /* Physical ingress port */ +#define T4_FILTER_OVLAN 0x100 /* Outer VLAN ID */ +#define T4_FILTER_IVLAN 0x200 /* Inner VLAN ID */ +#define T4_FILTER_IP_TOS 0x400 /* IPv4 TOS/IPv6 Traffic Class */ +#define T4_FILTER_IP_PROTO 0x800 /* IP protocol */ +#define T4_FILTER_ETH_TYPE 0x1000 /* Ethernet Type */ +#define T4_FILTER_MAC_IDX 0x2000 /* MPS MAC address match index */ +#define T4_FILTER_MPS_HIT_TYPE 0x4000 /* MPS match type */ +#define T4_FILTER_IP_FRAGMENT 0x8000 /* IP fragment */ + +/* Filter action */ +enum { + FILTER_PASS = 0, /* default */ + FILTER_DROP, + FILTER_SWITCH +}; + +/* 802.1q manipulation on FILTER_SWITCH */ +enum { + VLAN_NOCHANGE = 0, /* default */ + VLAN_REMOVE, + VLAN_INSERT, + VLAN_REWRITE +}; + +/* MPS match type */ +enum { + UCAST_EXACT = 0, /* exact unicast match */ + UCAST_HASH = 1, /* inexact (hashed) unicast match */ + MCAST_EXACT = 2, /* exact multicast match */ + MCAST_HASH = 3, /* inexact (hashed) multicast match */ + PROMISC = 4, /* no match but port is promiscuous */ + HYPPROMISC = 5, /* port is hypervisor-promisuous + not bcast */ + BCAST = 6, /* broadcast packet */ +}; + +/* Rx steering */ +enum { + DST_MODE_QUEUE, /* queue is directly specified by filter */ + DST_MODE_RSS_QUEUE, /* filter specifies RSS entry containing queue */ + DST_MODE_RSS, /* queue selected by default RSS hash lookup */ + DST_MODE_FILT_RSS /* queue selected by hashing in filter-specified + RSS subtable */ +}; + +struct t4_filter_tuple { + /* + * These are always available. + */ + uint8_t sip[16]; /* source IP address (IPv4 in [3:0]) */ + uint8_t dip[16]; /* destinatin IP address (IPv4 in [3:0]) */ + uint16_t sport; /* source port */ + uint16_t dport; /* destination port */ + + /* + * A combination of these (upto 36 bits) is available. TP_VLAN_PRI_MAP + * is used to select the global mode and all filters are limited to the + * set of fields allowed by the global mode. + */ + uint16_t ovlan; /* outer VLAN */ + uint16_t ivlan; /* inner VLAN */ + uint16_t ethtype; /* Ethernet type */ + uint8_t tos; /* TOS/Traffic Type */ + uint8_t proto; /* protocol type */ + uint32_t fcoe:1; /* FCoE packet */ + uint32_t iport:3; /* ingress port */ + uint32_t matchtype:3; /* MPS match type */ + uint32_t frag:1; /* fragmentation extension header */ + uint32_t macidx:9; /* exact match MAC index */ + uint32_t ivlan_vld:1; /* inner VLAN valid */ + uint32_t ovlan_vld:1; /* outer VLAN valid */ +}; + +struct t4_filter_specification { + uint32_t hitcnts:1; /* count filter hits in TCB */ + uint32_t prio:1; /* filter has priority over active/server */ + uint32_t type:1; /* 0 => IPv4, 1 => IPv6 */ + uint32_t action:2; /* drop, pass, switch */ + uint32_t rpttid:1; /* report TID in RSS hash field */ + uint32_t dirsteer:1; /* 0 => RSS, 1 => steer to iq */ + uint32_t iq:10; /* ingress queue */ + uint32_t maskhash:1; /* dirsteer=0: store RSS hash in TCB */ + uint32_t dirsteerhash:1;/* dirsteer=1: 0 => TCB contains RSS hash */ + /* 1 => TCB contains IQ ID */ + + /* + * Switch proxy/rewrite fields. An ingress packet which matches a + * filter with "switch" set will be looped back out as an egress + * packet -- potentially with some Ethernet header rewriting. + */ + uint32_t eport:2; /* egress port to switch packet out */ + uint32_t newdmac:1; /* rewrite destination MAC address */ + uint32_t newsmac:1; /* rewrite source MAC address */ + uint32_t newvlan:2; /* rewrite VLAN Tag */ + uint8_t dmac[ETHER_ADDR_LEN]; /* new destination MAC address */ + uint8_t smac[ETHER_ADDR_LEN]; /* new source MAC address */ + uint16_t vlan; /* VLAN Tag to insert */ + + /* + * Filter rule value/mask pairs. + */ + struct t4_filter_tuple val; + struct t4_filter_tuple mask; +}; + +struct t4_filter { + uint32_t idx; + uint64_t hits; + struct t4_filter_specification fs; +}; + #define CHELSIO_T4_GETREG _IOWR('f', T4_GETREG, struct t4_reg) #define CHELSIO_T4_SETREG _IOW('f', T4_SETREG, struct t4_reg) #define CHELSIO_T4_REGDUMP _IOWR('f', T4_REGDUMP, struct t4_regdump) +#define CHELSIO_T4_GET_FILTER_MODE _IOWR('f', T4_GET_FILTER_MODE, uint32_t) +#define CHELSIO_T4_SET_FILTER_MODE _IOW('f', T4_SET_FILTER_MODE, uint32_t) +#define CHELSIO_T4_GET_FILTER _IOWR('f', T4_GET_FILTER, struct t4_filter) +#define CHELSIO_T4_SET_FILTER _IOW('f', T4_SET_FILTER, struct t4_filter) +#define CHELSIO_T4_DEL_FILTER _IOW('f', T4_DEL_FILTER, struct t4_filter) #endif Index: projects/largeSMP/sys/dev/cxgbe/t4_main.c =================================================================== --- projects/largeSMP/sys/dev/cxgbe/t4_main.c (revision 221494) +++ projects/largeSMP/sys/dev/cxgbe/t4_main.c (revision 221495) @@ -1,2900 +1,3427 @@ /*- * Copyright (c) 2011 Chelsio Communications, Inc. * All rights reserved. * Written by: Navdeep Parhar * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "common/t4_hw.h" #include "common/common.h" +#include "common/t4_msg.h" #include "common/t4_regs.h" #include "common/t4_regs_values.h" #include "common/t4fw_interface.h" #include "t4_ioctl.h" /* T4 bus driver interface */ static int t4_probe(device_t); static int t4_attach(device_t); static int t4_detach(device_t); static device_method_t t4_methods[] = { DEVMETHOD(device_probe, t4_probe), DEVMETHOD(device_attach, t4_attach), DEVMETHOD(device_detach, t4_detach), /* bus interface */ DEVMETHOD(bus_print_child, bus_generic_print_child), DEVMETHOD(bus_driver_added, bus_generic_driver_added), { 0, 0 } }; static driver_t t4_driver = { "t4nex", t4_methods, sizeof(struct adapter) }; /* T4 port (cxgbe) interface */ static int cxgbe_probe(device_t); static int cxgbe_attach(device_t); static int cxgbe_detach(device_t); static device_method_t cxgbe_methods[] = { DEVMETHOD(device_probe, cxgbe_probe), DEVMETHOD(device_attach, cxgbe_attach), DEVMETHOD(device_detach, cxgbe_detach), { 0, 0 } }; static driver_t cxgbe_driver = { "cxgbe", cxgbe_methods, sizeof(struct port_info) }; static d_ioctl_t t4_ioctl; static d_open_t t4_open; static d_close_t t4_close; static struct cdevsw t4_cdevsw = { .d_version = D_VERSION, .d_flags = 0, .d_open = t4_open, .d_close = t4_close, .d_ioctl = t4_ioctl, .d_name = "t4nex", }; /* ifnet + media interface */ static void cxgbe_init(void *); static int cxgbe_ioctl(struct ifnet *, unsigned long, caddr_t); static void cxgbe_start(struct ifnet *); static int cxgbe_transmit(struct ifnet *, struct mbuf *); static void cxgbe_qflush(struct ifnet *); static int cxgbe_media_change(struct ifnet *); static void cxgbe_media_status(struct ifnet *, struct ifmediareq *); MALLOC_DEFINE(M_CXGBE, "cxgbe", "Chelsio T4 Ethernet driver and services"); /* * Tunables. */ SYSCTL_NODE(_hw, OID_AUTO, cxgbe, CTLFLAG_RD, 0, "cxgbe driver parameters"); static int force_firmware_install = 0; TUNABLE_INT("hw.cxgbe.force_firmware_install", &force_firmware_install); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, force_firmware_install, CTLFLAG_RDTUN, &force_firmware_install, 0, "install firmware on every attach."); /* * Holdoff timer and packet counter values. */ static unsigned int intr_timer[SGE_NTIMERS] = {1, 5, 10, 50, 100, 200}; static unsigned int intr_pktcount[SGE_NCOUNTERS] = {1, 8, 16, 32}; /* 63 max */ /* * Max # of tx and rx queues to use for each 10G and 1G port. */ static unsigned int max_ntxq_10g = 8; TUNABLE_INT("hw.cxgbe.max_ntxq_10G_port", &max_ntxq_10g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, max_ntxq_10G_port, CTLFLAG_RDTUN, &max_ntxq_10g, 0, "maximum number of tx queues per 10G port."); static unsigned int max_nrxq_10g = 8; TUNABLE_INT("hw.cxgbe.max_nrxq_10G_port", &max_nrxq_10g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, max_nrxq_10G_port, CTLFLAG_RDTUN, &max_nrxq_10g, 0, "maximum number of rxq's (per 10G port)."); static unsigned int max_ntxq_1g = 2; TUNABLE_INT("hw.cxgbe.max_ntxq_1G_port", &max_ntxq_1g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, max_ntxq_1G_port, CTLFLAG_RDTUN, &max_ntxq_1g, 0, "maximum number of tx queues per 1G port."); static unsigned int max_nrxq_1g = 2; TUNABLE_INT("hw.cxgbe.max_nrxq_1G_port", &max_nrxq_1g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, max_nrxq_1G_port, CTLFLAG_RDTUN, &max_nrxq_1g, 0, "maximum number of rxq's (per 1G port)."); /* * Holdoff parameters for 10G and 1G ports. */ static unsigned int tmr_idx_10g = 1; TUNABLE_INT("hw.cxgbe.holdoff_timer_idx_10G", &tmr_idx_10g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, holdoff_timer_idx_10G, CTLFLAG_RDTUN, &tmr_idx_10g, 0, "default timer index for interrupt holdoff (10G ports)."); static int pktc_idx_10g = 2; TUNABLE_INT("hw.cxgbe.holdoff_pktc_idx_10G", &pktc_idx_10g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, holdoff_pktc_idx_10G, CTLFLAG_RDTUN, &pktc_idx_10g, 0, "default pkt counter index for interrupt holdoff (10G ports)."); static unsigned int tmr_idx_1g = 1; TUNABLE_INT("hw.cxgbe.holdoff_timer_idx_1G", &tmr_idx_1g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, holdoff_timer_idx_1G, CTLFLAG_RDTUN, &tmr_idx_1g, 0, "default timer index for interrupt holdoff (1G ports)."); static int pktc_idx_1g = 2; TUNABLE_INT("hw.cxgbe.holdoff_pktc_idx_1G", &pktc_idx_1g); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, holdoff_pktc_idx_1G, CTLFLAG_RDTUN, &pktc_idx_1g, 0, "default pkt counter index for interrupt holdoff (1G ports)."); /* * Size (# of entries) of each tx and rx queue. */ static unsigned int qsize_txq = TX_EQ_QSIZE; TUNABLE_INT("hw.cxgbe.qsize_txq", &qsize_txq); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, qsize_txq, CTLFLAG_RDTUN, &qsize_txq, 0, "default queue size of NIC tx queues."); static unsigned int qsize_rxq = RX_IQ_QSIZE; TUNABLE_INT("hw.cxgbe.qsize_rxq", &qsize_rxq); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, qsize_rxq, CTLFLAG_RDTUN, &qsize_rxq, 0, "default queue size of NIC rx queues."); /* * Interrupt types allowed. */ static int intr_types = INTR_MSIX | INTR_MSI | INTR_INTX; TUNABLE_INT("hw.cxgbe.interrupt_types", &intr_types); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, interrupt_types, CTLFLAG_RDTUN, &intr_types, 0, "interrupt types allowed (bits 0, 1, 2 = INTx, MSI, MSI-X respectively)"); /* * Force the driver to use interrupt forwarding. */ static int intr_fwd = 0; TUNABLE_INT("hw.cxgbe.interrupt_forwarding", &intr_fwd); SYSCTL_UINT(_hw_cxgbe, OID_AUTO, interrupt_forwarding, CTLFLAG_RDTUN, &intr_fwd, 0, "always use forwarded interrupts"); +static unsigned int filter_mode = HW_TPL_FR_MT_PR_IV_P_FC; +TUNABLE_INT("hw.cxgbe.filter_mode", &filter_mode); +SYSCTL_UINT(_hw_cxgbe, OID_AUTO, filter_mode, CTLFLAG_RDTUN, + &filter_mode, 0, "default global filter mode."); + struct intrs_and_queues { int intr_type; /* INTx, MSI, or MSI-X */ int nirq; /* Number of vectors */ int intr_fwd; /* Interrupts forwarded */ int ntxq10g; /* # of NIC txq's for each 10G port */ int nrxq10g; /* # of NIC rxq's for each 10G port */ int ntxq1g; /* # of NIC txq's for each 1G port */ int nrxq1g; /* # of NIC rxq's for each 1G port */ }; +struct filter_entry { + uint32_t valid:1; /* filter allocated and valid */ + uint32_t locked:1; /* filter is administratively locked */ + uint32_t pending:1; /* filter action is pending firmware reply */ + uint32_t smtidx:8; /* Source MAC Table index for smac */ + + struct t4_filter_specification fs; +}; + enum { MEMWIN0_APERTURE = 2048, MEMWIN0_BASE = 0x1b800, MEMWIN1_APERTURE = 32768, MEMWIN1_BASE = 0x28000, MEMWIN2_APERTURE = 65536, MEMWIN2_BASE = 0x30000, }; enum { XGMAC_MTU = (1 << 0), XGMAC_PROMISC = (1 << 1), XGMAC_ALLMULTI = (1 << 2), XGMAC_VLANEX = (1 << 3), XGMAC_UCADDR = (1 << 4), XGMAC_MCADDRS = (1 << 5), XGMAC_ALL = 0xffff }; static int map_bars(struct adapter *); static void setup_memwin(struct adapter *); static int cfg_itype_and_nqueues(struct adapter *, int, int, struct intrs_and_queues *); static int prep_firmware(struct adapter *); static int get_capabilities(struct adapter *, struct fw_caps_config_cmd *); static int get_params(struct adapter *, struct fw_caps_config_cmd *); static void t4_set_desc(struct adapter *); static void build_medialist(struct port_info *); static int update_mac_settings(struct port_info *, int); static int cxgbe_init_locked(struct port_info *); static int cxgbe_init_synchronized(struct port_info *); static int cxgbe_uninit_locked(struct port_info *); static int cxgbe_uninit_synchronized(struct port_info *); static int first_port_up(struct adapter *); static int last_port_down(struct adapter *); static int t4_alloc_irq(struct adapter *, struct irq *, int rid, iq_intr_handler_t *, void *, char *); static int t4_free_irq(struct adapter *, struct irq *); static void reg_block_dump(struct adapter *, uint8_t *, unsigned int, unsigned int); static void t4_get_regs(struct adapter *, struct t4_regdump *, uint8_t *); static void cxgbe_tick(void *); static int t4_sysctls(struct adapter *); static int cxgbe_sysctls(struct port_info *); static int sysctl_int_array(SYSCTL_HANDLER_ARGS); static int sysctl_holdoff_tmr_idx(SYSCTL_HANDLER_ARGS); static int sysctl_holdoff_pktc_idx(SYSCTL_HANDLER_ARGS); static int sysctl_qsize_rxq(SYSCTL_HANDLER_ARGS); static int sysctl_qsize_txq(SYSCTL_HANDLER_ARGS); static int sysctl_handle_t4_reg64(SYSCTL_HANDLER_ARGS); static inline void txq_start(struct ifnet *, struct sge_txq *); +static uint32_t fconf_to_mode(uint32_t); +static uint32_t mode_to_fconf(uint32_t); +static uint32_t fspec_to_fconf(struct t4_filter_specification *); +static int get_filter_mode(struct adapter *, uint32_t *); +static int set_filter_mode(struct adapter *, uint32_t); +static int get_filter(struct adapter *, struct t4_filter *); +static int set_filter(struct adapter *, struct t4_filter *); +static int del_filter(struct adapter *, struct t4_filter *); +static void clear_filter(struct adapter *, struct filter_entry *); +static int set_filter_wr(struct adapter *, int); +static int del_filter_wr(struct adapter *, int); +void filter_rpl(struct adapter *, const struct cpl_set_tcb_rpl *); static int t4_mod_event(module_t, int, void *); struct t4_pciids { uint16_t device; uint8_t mpf; char *desc; } t4_pciids[] = { {0xa000, 0, "Chelsio Terminator 4 FPGA"}, {0x4400, 4, "Chelsio T440-dbg"}, {0x4401, 4, "Chelsio T420-CR"}, {0x4402, 4, "Chelsio T422-CR"}, {0x4403, 4, "Chelsio T440-CR"}, {0x4404, 4, "Chelsio T420-BCH"}, {0x4405, 4, "Chelsio T440-BCH"}, {0x4406, 4, "Chelsio T440-CH"}, {0x4407, 4, "Chelsio T420-SO"}, {0x4408, 4, "Chelsio T420-CX"}, {0x4409, 4, "Chelsio T420-BT"}, {0x440a, 4, "Chelsio T404-BT"}, }; static int t4_probe(device_t dev) { int i; uint16_t v = pci_get_vendor(dev); uint16_t d = pci_get_device(dev); if (v != PCI_VENDOR_ID_CHELSIO) return (ENXIO); for (i = 0; i < ARRAY_SIZE(t4_pciids); i++) { if (d == t4_pciids[i].device && pci_get_function(dev) == t4_pciids[i].mpf) { device_set_desc(dev, t4_pciids[i].desc); return (BUS_PROBE_DEFAULT); } } return (ENXIO); } static int t4_attach(device_t dev) { struct adapter *sc; int rc = 0, i, n10g, n1g, rqidx, tqidx; struct fw_caps_config_cmd caps; uint32_t p, v; struct intrs_and_queues iaq; struct sge *s; sc = device_get_softc(dev); sc->dev = dev; sc->pf = pci_get_function(dev); sc->mbox = sc->pf; pci_enable_busmaster(dev); pci_set_max_read_req(dev, 4096); snprintf(sc->lockname, sizeof(sc->lockname), "%s", device_get_nameunit(dev)); mtx_init(&sc->sc_lock, sc->lockname, 0, MTX_DEF); rc = map_bars(sc); if (rc != 0) goto done; /* error message displayed already */ memset(sc->chan_map, 0xff, sizeof(sc->chan_map)); /* Prepare the adapter for operation */ rc = -t4_prep_adapter(sc); if (rc != 0) { device_printf(dev, "failed to prepare adapter: %d.\n", rc); goto done; } /* Do this really early */ sc->cdev = make_dev(&t4_cdevsw, device_get_unit(dev), UID_ROOT, GID_WHEEL, 0600, "%s", device_get_nameunit(dev)); sc->cdev->si_drv1 = sc; /* Prepare the firmware for operation */ rc = prep_firmware(sc); if (rc != 0) goto done; /* error message displayed already */ /* Get device capabilities and select which ones we'll use */ rc = get_capabilities(sc, &caps); if (rc != 0) { device_printf(dev, "failed to initialize adapter capabilities: %d.\n", rc); goto done; } /* Choose the global RSS mode. */ rc = -t4_config_glbl_rss(sc, sc->mbox, FW_RSS_GLB_CONFIG_CMD_MODE_BASICVIRTUAL, F_FW_RSS_GLB_CONFIG_CMD_TNLMAPEN | F_FW_RSS_GLB_CONFIG_CMD_HASHTOEPLITZ | F_FW_RSS_GLB_CONFIG_CMD_TNLALLLKP); if (rc != 0) { device_printf(dev, "failed to select global RSS mode: %d.\n", rc); goto done; } /* These are total (sum of all ports) limits for a bus driver */ rc = -t4_cfg_pfvf(sc, sc->mbox, sc->pf, 0, 64, /* max # of egress queues */ 64, /* max # of egress Ethernet or control queues */ 64, /* max # of ingress queues with fl/interrupt */ 0, /* max # of ingress queues without interrupt */ 0, /* PCIe traffic class */ 4, /* max # of virtual interfaces */ M_FW_PFVF_CMD_CMASK, M_FW_PFVF_CMD_PMASK, 16, FW_CMD_CAP_PF, FW_CMD_CAP_PF); if (rc != 0) { device_printf(dev, "failed to configure pf/vf resources: %d.\n", rc); goto done; } /* Need this before sge_init */ for (i = 0; i < SGE_NTIMERS; i++) sc->sge.timer_val[i] = min(intr_timer[i], 200U); for (i = 0; i < SGE_NCOUNTERS; i++) sc->sge.counter_val[i] = min(intr_pktcount[i], M_THRESHOLD_0); /* Also need the cooked value of cclk before sge_init */ p = (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_DEV) | V_FW_PARAMS_PARAM_X(FW_PARAMS_PARAM_DEV_CCLK)); rc = -t4_query_params(sc, sc->mbox, sc->pf, 0, 1, &p, &v); if (rc != 0) { device_printf(sc->dev, "failed to obtain core clock value: %d.\n", rc); goto done; } sc->params.vpd.cclk = v; t4_sge_init(sc); - /* - * XXX: This is the place to call t4_set_filter_mode() - */ + t4_set_filter_mode(sc, filter_mode); + t4_set_reg_field(sc, A_TP_GLOBAL_CONFIG, + V_FIVETUPLELOOKUP(M_FIVETUPLELOOKUP), + V_FIVETUPLELOOKUP(M_FIVETUPLELOOKUP)); + t4_tp_wr_bits_indirect(sc, A_TP_INGRESS_CONFIG, F_CSUM_HAS_PSEUDO_HDR, + F_LOOKUPEVERYPKT); /* get basic stuff going */ rc = -t4_early_init(sc, sc->mbox); if (rc != 0) { device_printf(dev, "early init failed: %d.\n", rc); goto done; } rc = get_params(sc, &caps); if (rc != 0) goto done; /* error message displayed already */ /* These are finalized by FW initialization, load their values now */ v = t4_read_reg(sc, A_TP_TIMER_RESOLUTION); sc->params.tp.tre = G_TIMERRESOLUTION(v); sc->params.tp.dack_re = G_DELAYEDACKRESOLUTION(v); t4_read_mtu_tbl(sc, sc->params.mtus, NULL); /* tweak some settings */ t4_write_reg(sc, A_TP_SHIFT_CNT, V_SYNSHIFTMAX(6) | V_RXTSHIFTMAXR1(4) | V_RXTSHIFTMAXR2(15) | V_PERSHIFTBACKOFFMAX(8) | V_PERSHIFTMAX(8) | V_KEEPALIVEMAXR1(4) | V_KEEPALIVEMAXR2(9)); t4_write_reg(sc, A_ULP_RX_TDDP_PSZ, V_HPZ0(PAGE_SHIFT - 12)); setup_memwin(sc); rc = t4_create_dma_tag(sc); if (rc != 0) goto done; /* error message displayed already */ /* * First pass over all the ports - allocate VIs and initialize some * basic parameters like mac address, port type, etc. We also figure * out whether a port is 10G or 1G and use that information when * calculating how many interrupts to attempt to allocate. */ n10g = n1g = 0; for_each_port(sc, i) { struct port_info *pi; pi = malloc(sizeof(*pi), M_CXGBE, M_ZERO | M_WAITOK); sc->port[i] = pi; /* These must be set before t4_port_init */ pi->adapter = sc; pi->port_id = i; /* Allocate the vi and initialize parameters like mac addr */ rc = -t4_port_init(pi, sc->mbox, sc->pf, 0); if (rc != 0) { device_printf(dev, "unable to initialize port %d: %d\n", i, rc); free(pi, M_CXGBE); sc->port[i] = NULL; /* indicates init failed */ continue; } snprintf(pi->lockname, sizeof(pi->lockname), "%sp%d", device_get_nameunit(dev), i); mtx_init(&pi->pi_lock, pi->lockname, 0, MTX_DEF); if (is_10G_port(pi)) { n10g++; pi->tmr_idx = tmr_idx_10g; pi->pktc_idx = pktc_idx_10g; } else { n1g++; pi->tmr_idx = tmr_idx_1g; pi->pktc_idx = pktc_idx_1g; } pi->xact_addr_filt = -1; pi->qsize_rxq = max(qsize_rxq, 128); while (pi->qsize_rxq & 7) pi->qsize_rxq++; pi->qsize_txq = max(qsize_txq, 128); if (pi->qsize_rxq != qsize_rxq) { device_printf(dev, "using %d instead of %d as the rx queue size.\n", pi->qsize_rxq, qsize_rxq); } if (pi->qsize_txq != qsize_txq) { device_printf(dev, "using %d instead of %d as the tx queue size.\n", pi->qsize_txq, qsize_txq); } pi->dev = device_add_child(dev, "cxgbe", -1); if (pi->dev == NULL) { device_printf(dev, "failed to add device for port %d.\n", i); rc = ENXIO; goto done; } device_set_softc(pi->dev, pi); setbit(&sc->registered_device_map, i); } if (sc->registered_device_map == 0) { device_printf(dev, "no usable ports\n"); rc = ENXIO; goto done; } /* * Interrupt type, # of interrupts, # of rx/tx queues, etc. */ rc = cfg_itype_and_nqueues(sc, n10g, n1g, &iaq); if (rc != 0) goto done; /* error message displayed already */ sc->intr_type = iaq.intr_type; sc->intr_count = iaq.nirq; s = &sc->sge; s->nrxq = n10g * iaq.nrxq10g + n1g * iaq.nrxq1g; s->ntxq = n10g * iaq.ntxq10g + n1g * iaq.ntxq1g; s->neq = s->ntxq + s->nrxq; /* the free list in an rxq is an eq */ s->neq += NCHAN; /* control queues, 1 per hw channel */ s->niq = s->nrxq + 1; /* 1 extra for firmware event queue */ if (iaq.intr_fwd) { sc->flags |= INTR_FWD; s->niq += NFIQ(sc); /* forwarded interrupt queues */ s->fiq = malloc(NFIQ(sc) * sizeof(struct sge_iq), M_CXGBE, M_ZERO | M_WAITOK); } s->ctrlq = malloc(NCHAN * sizeof(struct sge_ctrlq), M_CXGBE, M_ZERO | M_WAITOK); s->rxq = malloc(s->nrxq * sizeof(struct sge_rxq), M_CXGBE, M_ZERO | M_WAITOK); s->txq = malloc(s->ntxq * sizeof(struct sge_txq), M_CXGBE, M_ZERO | M_WAITOK); s->iqmap = malloc(s->niq * sizeof(struct sge_iq *), M_CXGBE, M_ZERO | M_WAITOK); s->eqmap = malloc(s->neq * sizeof(struct sge_eq *), M_CXGBE, M_ZERO | M_WAITOK); sc->irq = malloc(sc->intr_count * sizeof(struct irq), M_CXGBE, M_ZERO | M_WAITOK); t4_sysctls(sc); /* * Second pass over the ports. This time we know the number of rx and * tx queues that each port should get. */ rqidx = tqidx = 0; for_each_port(sc, i) { struct port_info *pi = sc->port[i]; if (pi == NULL) continue; pi->first_rxq = rqidx; pi->nrxq = is_10G_port(pi) ? iaq.nrxq10g : iaq.nrxq1g; pi->first_txq = tqidx; pi->ntxq = is_10G_port(pi) ? iaq.ntxq10g : iaq.ntxq1g; rqidx += pi->nrxq; tqidx += pi->ntxq; } rc = bus_generic_attach(dev); if (rc != 0) { device_printf(dev, "failed to attach all child ports: %d\n", rc); goto done; } #ifdef INVARIANTS device_printf(dev, "%p, %d ports (0x%x), %d intr_type, %d intr_count\n", sc, sc->params.nports, sc->params.portvec, sc->intr_type, sc->intr_count); #endif t4_set_desc(sc); done: if (rc != 0) t4_detach(dev); return (rc); } /* * Idempotent */ static int t4_detach(device_t dev) { struct adapter *sc; struct port_info *pi; int i; sc = device_get_softc(dev); if (sc->cdev) destroy_dev(sc->cdev); bus_generic_detach(dev); for (i = 0; i < MAX_NPORTS; i++) { pi = sc->port[i]; if (pi) { t4_free_vi(pi->adapter, sc->mbox, sc->pf, 0, pi->viid); if (pi->dev) device_delete_child(dev, pi->dev); mtx_destroy(&pi->pi_lock); free(pi, M_CXGBE); } } if (sc->flags & FW_OK) t4_fw_bye(sc, sc->mbox); if (sc->intr_type == INTR_MSI || sc->intr_type == INTR_MSIX) pci_release_msi(dev); if (sc->regs_res) bus_release_resource(dev, SYS_RES_MEMORY, sc->regs_rid, sc->regs_res); if (sc->msix_res) bus_release_resource(dev, SYS_RES_MEMORY, sc->msix_rid, sc->msix_res); free(sc->irq, M_CXGBE); free(sc->sge.rxq, M_CXGBE); free(sc->sge.txq, M_CXGBE); free(sc->sge.ctrlq, M_CXGBE); free(sc->sge.fiq, M_CXGBE); free(sc->sge.iqmap, M_CXGBE); free(sc->sge.eqmap, M_CXGBE); + free(sc->tids.ftid_tab, M_CXGBE); t4_destroy_dma_tag(sc); mtx_destroy(&sc->sc_lock); bzero(sc, sizeof(*sc)); return (0); } static int cxgbe_probe(device_t dev) { char buf[128]; struct port_info *pi = device_get_softc(dev); snprintf(buf, sizeof(buf), "Port %d", pi->port_id); device_set_desc_copy(dev, buf); return (BUS_PROBE_DEFAULT); } #define T4_CAP (IFCAP_VLAN_HWTAGGING | IFCAP_VLAN_MTU | IFCAP_HWCSUM | \ IFCAP_VLAN_HWCSUM | IFCAP_TSO | IFCAP_JUMBO_MTU | IFCAP_LRO | \ IFCAP_VLAN_HWTSO) #define T4_CAP_ENABLE (T4_CAP & ~IFCAP_TSO6) static int cxgbe_attach(device_t dev) { struct port_info *pi = device_get_softc(dev); struct ifnet *ifp; /* Allocate an ifnet and set it up */ ifp = if_alloc(IFT_ETHER); if (ifp == NULL) { device_printf(dev, "Cannot allocate ifnet\n"); return (ENOMEM); } pi->ifp = ifp; ifp->if_softc = pi; callout_init(&pi->tick, CALLOUT_MPSAFE); pi->tq = taskqueue_create("cxgbe_taskq", M_NOWAIT, taskqueue_thread_enqueue, &pi->tq); if (pi->tq == NULL) { device_printf(dev, "failed to allocate port task queue\n"); if_free(pi->ifp); return (ENOMEM); } taskqueue_start_threads(&pi->tq, 1, PI_NET, "%s taskq", device_get_nameunit(dev)); if_initname(ifp, device_get_name(dev), device_get_unit(dev)); ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST; ifp->if_init = cxgbe_init; ifp->if_ioctl = cxgbe_ioctl; ifp->if_start = cxgbe_start; ifp->if_transmit = cxgbe_transmit; ifp->if_qflush = cxgbe_qflush; ifp->if_snd.ifq_drv_maxlen = 1024; IFQ_SET_MAXLEN(&ifp->if_snd, ifp->if_snd.ifq_drv_maxlen); IFQ_SET_READY(&ifp->if_snd); ifp->if_capabilities = T4_CAP; ifp->if_capenable = T4_CAP_ENABLE; ifp->if_hwassist = CSUM_TCP | CSUM_UDP | CSUM_IP | CSUM_TSO; /* Initialize ifmedia for this port */ ifmedia_init(&pi->media, IFM_IMASK, cxgbe_media_change, cxgbe_media_status); build_medialist(pi); ether_ifattach(ifp, pi->hw_addr); #ifdef INVARIANTS device_printf(dev, "%p, %d txq, %d rxq\n", pi, pi->ntxq, pi->nrxq); #endif cxgbe_sysctls(pi); return (0); } static int cxgbe_detach(device_t dev) { struct port_info *pi = device_get_softc(dev); struct adapter *sc = pi->adapter; int rc; /* Tell if_ioctl and if_init that the port is going away */ ADAPTER_LOCK(sc); SET_DOOMED(pi); wakeup(&sc->flags); while (IS_BUSY(sc)) mtx_sleep(&sc->flags, &sc->sc_lock, 0, "t4detach", 0); SET_BUSY(sc); ADAPTER_UNLOCK(sc); rc = cxgbe_uninit_synchronized(pi); if (rc != 0) device_printf(dev, "port uninit failed: %d.\n", rc); taskqueue_free(pi->tq); ifmedia_removeall(&pi->media); ether_ifdetach(pi->ifp); if_free(pi->ifp); ADAPTER_LOCK(sc); CLR_BUSY(sc); wakeup_one(&sc->flags); ADAPTER_UNLOCK(sc); return (0); } static void cxgbe_init(void *arg) { struct port_info *pi = arg; struct adapter *sc = pi->adapter; ADAPTER_LOCK(sc); cxgbe_init_locked(pi); /* releases adapter lock */ ADAPTER_LOCK_ASSERT_NOTOWNED(sc); } static int cxgbe_ioctl(struct ifnet *ifp, unsigned long cmd, caddr_t data) { int rc = 0, mtu, flags; struct port_info *pi = ifp->if_softc; struct adapter *sc = pi->adapter; struct ifreq *ifr = (struct ifreq *)data; uint32_t mask; switch (cmd) { case SIOCSIFMTU: ADAPTER_LOCK(sc); rc = IS_DOOMED(pi) ? ENXIO : (IS_BUSY(sc) ? EBUSY : 0); if (rc) { fail: ADAPTER_UNLOCK(sc); return (rc); } mtu = ifr->ifr_mtu; if ((mtu < ETHERMIN) || (mtu > ETHERMTU_JUMBO)) { rc = EINVAL; } else { ifp->if_mtu = mtu; if (ifp->if_drv_flags & IFF_DRV_RUNNING) { t4_update_fl_bufsize(ifp); PORT_LOCK(pi); rc = update_mac_settings(pi, XGMAC_MTU); PORT_UNLOCK(pi); } } ADAPTER_UNLOCK(sc); break; case SIOCSIFFLAGS: ADAPTER_LOCK(sc); if (IS_DOOMED(pi)) { rc = ENXIO; goto fail; } if (ifp->if_flags & IFF_UP) { if (ifp->if_drv_flags & IFF_DRV_RUNNING) { flags = pi->if_flags; if ((ifp->if_flags ^ flags) & (IFF_PROMISC | IFF_ALLMULTI)) { if (IS_BUSY(sc)) { rc = EBUSY; goto fail; } PORT_LOCK(pi); rc = update_mac_settings(pi, XGMAC_PROMISC | XGMAC_ALLMULTI); PORT_UNLOCK(pi); } ADAPTER_UNLOCK(sc); } else rc = cxgbe_init_locked(pi); pi->if_flags = ifp->if_flags; } else if (ifp->if_drv_flags & IFF_DRV_RUNNING) rc = cxgbe_uninit_locked(pi); else ADAPTER_UNLOCK(sc); ADAPTER_LOCK_ASSERT_NOTOWNED(sc); break; case SIOCADDMULTI: case SIOCDELMULTI: /* these two can be called with a mutex held :-( */ ADAPTER_LOCK(sc); rc = IS_DOOMED(pi) ? ENXIO : (IS_BUSY(sc) ? EBUSY : 0); if (rc) goto fail; if (ifp->if_drv_flags & IFF_DRV_RUNNING) { PORT_LOCK(pi); rc = update_mac_settings(pi, XGMAC_MCADDRS); PORT_UNLOCK(pi); } ADAPTER_UNLOCK(sc); break; case SIOCSIFCAP: ADAPTER_LOCK(sc); rc = IS_DOOMED(pi) ? ENXIO : (IS_BUSY(sc) ? EBUSY : 0); if (rc) goto fail; mask = ifr->ifr_reqcap ^ ifp->if_capenable; if (mask & IFCAP_TXCSUM) { ifp->if_capenable ^= IFCAP_TXCSUM; ifp->if_hwassist ^= (CSUM_TCP | CSUM_UDP | CSUM_IP); if (IFCAP_TSO & ifp->if_capenable && !(IFCAP_TXCSUM & ifp->if_capenable)) { ifp->if_capenable &= ~IFCAP_TSO; ifp->if_hwassist &= ~CSUM_TSO; if_printf(ifp, "tso disabled due to -txcsum.\n"); } } if (mask & IFCAP_RXCSUM) ifp->if_capenable ^= IFCAP_RXCSUM; if (mask & IFCAP_TSO4) { ifp->if_capenable ^= IFCAP_TSO4; if (IFCAP_TSO & ifp->if_capenable) { if (IFCAP_TXCSUM & ifp->if_capenable) ifp->if_hwassist |= CSUM_TSO; else { ifp->if_capenable &= ~IFCAP_TSO; ifp->if_hwassist &= ~CSUM_TSO; if_printf(ifp, "enable txcsum first.\n"); rc = EAGAIN; } } else ifp->if_hwassist &= ~CSUM_TSO; } if (mask & IFCAP_LRO) { #ifdef INET int i; struct sge_rxq *rxq; ifp->if_capenable ^= IFCAP_LRO; for_each_rxq(pi, i, rxq) { if (ifp->if_capenable & IFCAP_LRO) rxq->flags |= RXQ_LRO_ENABLED; else rxq->flags &= ~RXQ_LRO_ENABLED; } #endif } #ifndef TCP_OFFLOAD_DISABLE if (mask & IFCAP_TOE4) { rc = EOPNOTSUPP; } #endif if (mask & IFCAP_VLAN_HWTAGGING) { ifp->if_capenable ^= IFCAP_VLAN_HWTAGGING; if (ifp->if_drv_flags & IFF_DRV_RUNNING) { PORT_LOCK(pi); rc = update_mac_settings(pi, XGMAC_VLANEX); PORT_UNLOCK(pi); } } if (mask & IFCAP_VLAN_MTU) { ifp->if_capenable ^= IFCAP_VLAN_MTU; /* Need to find out how to disable auto-mtu-inflation */ } if (mask & IFCAP_VLAN_HWTSO) ifp->if_capenable ^= IFCAP_VLAN_HWTSO; if (mask & IFCAP_VLAN_HWCSUM) ifp->if_capenable ^= IFCAP_VLAN_HWCSUM; #ifdef VLAN_CAPABILITIES VLAN_CAPABILITIES(ifp); #endif ADAPTER_UNLOCK(sc); break; case SIOCSIFMEDIA: case SIOCGIFMEDIA: ifmedia_ioctl(ifp, ifr, &pi->media, cmd); break; default: rc = ether_ioctl(ifp, cmd, data); } return (rc); } static void cxgbe_start(struct ifnet *ifp) { struct port_info *pi = ifp->if_softc; struct sge_txq *txq; int i; for_each_txq(pi, i, txq) { if (TXQ_TRYLOCK(txq)) { txq_start(ifp, txq); TXQ_UNLOCK(txq); } } } static int cxgbe_transmit(struct ifnet *ifp, struct mbuf *m) { struct port_info *pi = ifp->if_softc; struct adapter *sc = pi->adapter; struct sge_txq *txq = &sc->sge.txq[pi->first_txq]; struct buf_ring *br; int rc; M_ASSERTPKTHDR(m); if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { m_freem(m); return (0); } if (m->m_flags & M_FLOWID) txq += (m->m_pkthdr.flowid % pi->ntxq); br = txq->br; if (TXQ_TRYLOCK(txq) == 0) { /* * XXX: make sure that this packet really is sent out. There is * a small race where t4_eth_tx may stop draining the drbr and * goes away, just before we enqueued this mbuf. */ return (drbr_enqueue(ifp, br, m)); } /* * txq->m is the mbuf that is held up due to a temporary shortage of * resources and it should be put on the wire first. Then what's in * drbr and finally the mbuf that was just passed in to us. * * Return code should indicate the fate of the mbuf that was passed in * this time. */ TXQ_LOCK_ASSERT_OWNED(txq); if (drbr_needs_enqueue(ifp, br) || txq->m) { /* Queued for transmission. */ rc = drbr_enqueue(ifp, br, m); m = txq->m ? txq->m : drbr_dequeue(ifp, br); (void) t4_eth_tx(ifp, txq, m); TXQ_UNLOCK(txq); return (rc); } /* Direct transmission. */ rc = t4_eth_tx(ifp, txq, m); if (rc != 0 && txq->m) rc = 0; /* held, will be transmitted soon (hopefully) */ TXQ_UNLOCK(txq); return (rc); } static void cxgbe_qflush(struct ifnet *ifp) { struct port_info *pi = ifp->if_softc; struct sge_txq *txq; int i; struct mbuf *m; /* queues do not exist if !IFF_DRV_RUNNING. */ if (ifp->if_drv_flags & IFF_DRV_RUNNING) { for_each_txq(pi, i, txq) { TXQ_LOCK(txq); m_freem(txq->m); while ((m = buf_ring_dequeue_sc(txq->br)) != NULL) m_freem(m); TXQ_UNLOCK(txq); } } if_qflush(ifp); } static int cxgbe_media_change(struct ifnet *ifp) { struct port_info *pi = ifp->if_softc; device_printf(pi->dev, "%s unimplemented.\n", __func__); return (EOPNOTSUPP); } static void cxgbe_media_status(struct ifnet *ifp, struct ifmediareq *ifmr) { struct port_info *pi = ifp->if_softc; struct ifmedia_entry *cur = pi->media.ifm_cur; int speed = pi->link_cfg.speed; int data = (pi->port_type << 8) | pi->mod_type; if (cur->ifm_data != data) { build_medialist(pi); cur = pi->media.ifm_cur; } ifmr->ifm_status = IFM_AVALID; if (!pi->link_cfg.link_ok) return; ifmr->ifm_status |= IFM_ACTIVE; /* active and current will differ iff current media is autoselect. */ if (IFM_SUBTYPE(cur->ifm_media) != IFM_AUTO) return; ifmr->ifm_active = IFM_ETHER | IFM_FDX; if (speed == SPEED_10000) ifmr->ifm_active |= IFM_10G_T; else if (speed == SPEED_1000) ifmr->ifm_active |= IFM_1000_T; else if (speed == SPEED_100) ifmr->ifm_active |= IFM_100_TX; else if (speed == SPEED_10) ifmr->ifm_active |= IFM_10_T; else KASSERT(0, ("%s: link up but speed unknown (%u)", __func__, speed)); } void t4_fatal_err(struct adapter *sc) { t4_set_reg_field(sc, A_SGE_CONTROL, F_GLOBALENABLE, 0); t4_intr_disable(sc); log(LOG_EMERG, "%s: encountered fatal error, adapter stopped.\n", device_get_nameunit(sc->dev)); } static int map_bars(struct adapter *sc) { sc->regs_rid = PCIR_BAR(0); sc->regs_res = bus_alloc_resource_any(sc->dev, SYS_RES_MEMORY, &sc->regs_rid, RF_ACTIVE); if (sc->regs_res == NULL) { device_printf(sc->dev, "cannot map registers.\n"); return (ENXIO); } sc->bt = rman_get_bustag(sc->regs_res); sc->bh = rman_get_bushandle(sc->regs_res); sc->mmio_len = rman_get_size(sc->regs_res); sc->msix_rid = PCIR_BAR(4); sc->msix_res = bus_alloc_resource_any(sc->dev, SYS_RES_MEMORY, &sc->msix_rid, RF_ACTIVE); if (sc->msix_res == NULL) { device_printf(sc->dev, "cannot map MSI-X BAR.\n"); return (ENXIO); } return (0); } static void setup_memwin(struct adapter *sc) { u_long bar0; bar0 = rman_get_start(sc->regs_res); t4_write_reg(sc, PCIE_MEM_ACCESS_REG(A_PCIE_MEM_ACCESS_BASE_WIN, 0), (bar0 + MEMWIN0_BASE) | V_BIR(0) | V_WINDOW(ilog2(MEMWIN0_APERTURE) - 10)); t4_write_reg(sc, PCIE_MEM_ACCESS_REG(A_PCIE_MEM_ACCESS_BASE_WIN, 1), (bar0 + MEMWIN1_BASE) | V_BIR(0) | V_WINDOW(ilog2(MEMWIN1_APERTURE) - 10)); t4_write_reg(sc, PCIE_MEM_ACCESS_REG(A_PCIE_MEM_ACCESS_BASE_WIN, 2), (bar0 + MEMWIN2_BASE) | V_BIR(0) | V_WINDOW(ilog2(MEMWIN2_APERTURE) - 10)); } static int cfg_itype_and_nqueues(struct adapter *sc, int n10g, int n1g, struct intrs_and_queues *iaq) { int rc, itype, navail, nc, nrxq10g, nrxq1g; bzero(iaq, sizeof(*iaq)); nc = mp_ncpus; /* our snapshot of the number of CPUs */ for (itype = INTR_MSIX; itype; itype >>= 1) { if ((itype & intr_types) == 0) continue; /* not allowed */ if (itype == INTR_MSIX) navail = pci_msix_count(sc->dev); else if (itype == INTR_MSI) navail = pci_msi_count(sc->dev); else navail = 1; if (navail == 0) continue; iaq->intr_type = itype; iaq->ntxq10g = min(nc, max_ntxq_10g); iaq->ntxq1g = min(nc, max_ntxq_1g); nrxq10g = min(nc, max_nrxq_10g); nrxq1g = min(nc, max_nrxq_1g); /* Extra 2 is for a) error interrupt b) firmware event */ iaq->nirq = n10g * nrxq10g + n1g * nrxq1g + 2; if (iaq->nirq <= navail && intr_fwd == 0) { if (itype == INTR_MSI && !powerof2(iaq->nirq)) goto fwd; /* One for err, one for fwq, and one for each rxq */ iaq->intr_fwd = 0; iaq->nrxq10g = nrxq10g; iaq->nrxq1g = nrxq1g; } else { fwd: iaq->intr_fwd = 1; if (navail > nc) { if (itype == INTR_MSIX) navail = nc + 1; /* navail is and must remain a pow2 for MSI */ if (itype == INTR_MSI) { KASSERT(powerof2(navail), ("%d not power of 2", navail)); while (navail / 2 > nc) navail /= 2; } } iaq->nirq = navail; /* total # of interrupts */ /* * If we have multiple vectors available reserve one * exclusively for errors. The rest will be shared by * the fwq and data. */ if (navail > 1) navail--; iaq->nrxq10g = min(nrxq10g, navail); iaq->nrxq1g = min(nrxq1g, navail); } navail = iaq->nirq; rc = 0; if (itype == INTR_MSIX) rc = pci_alloc_msix(sc->dev, &navail); else if (itype == INTR_MSI) rc = pci_alloc_msi(sc->dev, &navail); if (rc == 0) { if (navail == iaq->nirq) return (0); /* * Didn't get the number requested. Use whatever number * the kernel is willing to allocate (it's in navail). */ pci_release_msi(sc->dev); goto fwd; } device_printf(sc->dev, "failed to allocate vectors:%d, type=%d, req=%d, rcvd=%d\n", itype, rc, iaq->nirq, navail); } device_printf(sc->dev, "failed to find a usable interrupt type. " "allowed=%d, msi-x=%d, msi=%d, intx=1", intr_types, pci_msix_count(sc->dev), pci_msi_count(sc->dev)); return (ENXIO); } /* * Install a compatible firmware (if required), establish contact with it, * become the master, and reset the device. */ static int prep_firmware(struct adapter *sc) { const struct firmware *fw; int rc; enum dev_state state; /* Check firmware version and install a different one if necessary */ rc = t4_check_fw_version(sc); if (rc != 0 || force_firmware_install) { uint32_t v = 0; fw = firmware_get(T4_FWNAME); if (fw != NULL) { const struct fw_hdr *hdr = (const void *)fw->data; v = ntohl(hdr->fw_ver); /* * The firmware module will not be used if it isn't the * same major version as what the driver was compiled * with. This check trumps force_firmware_install. */ if (G_FW_HDR_FW_VER_MAJOR(v) != FW_VERSION_MAJOR) { device_printf(sc->dev, "Found firmware image but version %d " "can not be used with this driver (%d)\n", G_FW_HDR_FW_VER_MAJOR(v), FW_VERSION_MAJOR); firmware_put(fw, FIRMWARE_UNLOAD); fw = NULL; } } if (fw == NULL && (rc < 0 || force_firmware_install)) { device_printf(sc->dev, "No usable firmware. " "card has %d.%d.%d, driver compiled with %d.%d.%d, " "force_firmware_install%s set", G_FW_HDR_FW_VER_MAJOR(sc->params.fw_vers), G_FW_HDR_FW_VER_MINOR(sc->params.fw_vers), G_FW_HDR_FW_VER_MICRO(sc->params.fw_vers), FW_VERSION_MAJOR, FW_VERSION_MINOR, FW_VERSION_MICRO, force_firmware_install ? "" : " not"); return (EAGAIN); } /* * Always upgrade, even for minor/micro/build mismatches. * Downgrade only for a major version mismatch or if * force_firmware_install was specified. */ if (fw != NULL && (rc < 0 || force_firmware_install || v > sc->params.fw_vers)) { device_printf(sc->dev, "installing firmware %d.%d.%d.%d on card.\n", G_FW_HDR_FW_VER_MAJOR(v), G_FW_HDR_FW_VER_MINOR(v), G_FW_HDR_FW_VER_MICRO(v), G_FW_HDR_FW_VER_BUILD(v)); rc = -t4_load_fw(sc, fw->data, fw->datasize); if (rc != 0) { device_printf(sc->dev, "failed to install firmware: %d\n", rc); firmware_put(fw, FIRMWARE_UNLOAD); return (rc); } else { /* refresh */ (void) t4_check_fw_version(sc); } } if (fw != NULL) firmware_put(fw, FIRMWARE_UNLOAD); } /* Contact firmware, request master */ rc = t4_fw_hello(sc, sc->mbox, sc->mbox, MASTER_MUST, &state); if (rc < 0) { rc = -rc; device_printf(sc->dev, "failed to connect to the firmware: %d.\n", rc); return (rc); } /* Reset device */ rc = -t4_fw_reset(sc, sc->mbox, F_PIORSTMODE | F_PIORST); if (rc != 0) { device_printf(sc->dev, "firmware reset failed: %d.\n", rc); if (rc != ETIMEDOUT && rc != EIO) t4_fw_bye(sc, sc->mbox); return (rc); } snprintf(sc->fw_version, sizeof(sc->fw_version), "%u.%u.%u.%u", G_FW_HDR_FW_VER_MAJOR(sc->params.fw_vers), G_FW_HDR_FW_VER_MINOR(sc->params.fw_vers), G_FW_HDR_FW_VER_MICRO(sc->params.fw_vers), G_FW_HDR_FW_VER_BUILD(sc->params.fw_vers)); sc->flags |= FW_OK; return (0); } static int get_capabilities(struct adapter *sc, struct fw_caps_config_cmd *caps) { int rc; bzero(caps, sizeof(*caps)); caps->op_to_write = htobe32(V_FW_CMD_OP(FW_CAPS_CONFIG_CMD) | F_FW_CMD_REQUEST | F_FW_CMD_READ); caps->retval_len16 = htobe32(FW_LEN16(*caps)); rc = -t4_wr_mbox(sc, sc->mbox, caps, sizeof(*caps), caps); if (rc != 0) return (rc); if (caps->niccaps & htobe16(FW_CAPS_CONFIG_NIC_VM)) caps->niccaps ^= htobe16(FW_CAPS_CONFIG_NIC_VM); caps->op_to_write = htobe32(V_FW_CMD_OP(FW_CAPS_CONFIG_CMD) | F_FW_CMD_REQUEST | F_FW_CMD_WRITE); rc = -t4_wr_mbox(sc, sc->mbox, caps, sizeof(*caps), NULL); return (rc); } static int get_params(struct adapter *sc, struct fw_caps_config_cmd *caps) { int rc; uint32_t params[7], val[7]; #define FW_PARAM_DEV(param) \ (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_DEV) | \ V_FW_PARAMS_PARAM_X(FW_PARAMS_PARAM_DEV_##param)) #define FW_PARAM_PFVF(param) \ (V_FW_PARAMS_MNEM(FW_PARAMS_MNEM_PFVF) | \ V_FW_PARAMS_PARAM_X(FW_PARAMS_PARAM_PFVF_##param)) params[0] = FW_PARAM_DEV(PORTVEC); params[1] = FW_PARAM_PFVF(IQFLINT_START); params[2] = FW_PARAM_PFVF(EQ_START); params[3] = FW_PARAM_PFVF(FILTER_START); params[4] = FW_PARAM_PFVF(FILTER_END); rc = -t4_query_params(sc, sc->mbox, sc->pf, 0, 5, params, val); if (rc != 0) { device_printf(sc->dev, "failed to query parameters: %d.\n", rc); goto done; } sc->params.portvec = val[0]; sc->params.nports = 0; while (val[0]) { sc->params.nports++; val[0] &= val[0] - 1; } sc->sge.iq_start = val[1]; sc->sge.eq_start = val[2]; sc->tids.ftid_base = val[3]; sc->tids.nftids = val[4] - val[3] + 1; if (caps->toecaps) { /* query offload-related parameters */ params[0] = FW_PARAM_DEV(NTID); params[1] = FW_PARAM_PFVF(SERVER_START); params[2] = FW_PARAM_PFVF(SERVER_END); params[3] = FW_PARAM_PFVF(TDDP_START); params[4] = FW_PARAM_PFVF(TDDP_END); params[5] = FW_PARAM_DEV(FLOWC_BUFFIFO_SZ); rc = -t4_query_params(sc, sc->mbox, sc->pf, 0, 6, params, val); if (rc != 0) { device_printf(sc->dev, "failed to query TOE parameters: %d.\n", rc); goto done; } sc->tids.ntids = val[0]; sc->tids.natids = min(sc->tids.ntids / 2, MAX_ATIDS); sc->tids.stid_base = val[1]; sc->tids.nstids = val[2] - val[1] + 1; sc->vres.ddp.start = val[3]; sc->vres.ddp.size = val[4] - val[3] + 1; sc->params.ofldq_wr_cred = val[5]; sc->params.offload = 1; } if (caps->rdmacaps) { params[0] = FW_PARAM_PFVF(STAG_START); params[1] = FW_PARAM_PFVF(STAG_END); params[2] = FW_PARAM_PFVF(RQ_START); params[3] = FW_PARAM_PFVF(RQ_END); params[4] = FW_PARAM_PFVF(PBL_START); params[5] = FW_PARAM_PFVF(PBL_END); rc = -t4_query_params(sc, sc->mbox, sc->pf, 0, 6, params, val); if (rc != 0) { device_printf(sc->dev, "failed to query RDMA parameters: %d.\n", rc); goto done; } sc->vres.stag.start = val[0]; sc->vres.stag.size = val[1] - val[0] + 1; sc->vres.rq.start = val[2]; sc->vres.rq.size = val[3] - val[2] + 1; sc->vres.pbl.start = val[4]; sc->vres.pbl.size = val[5] - val[4] + 1; } if (caps->iscsicaps) { params[0] = FW_PARAM_PFVF(ISCSI_START); params[1] = FW_PARAM_PFVF(ISCSI_END); rc = -t4_query_params(sc, sc->mbox, sc->pf, 0, 2, params, val); if (rc != 0) { device_printf(sc->dev, "failed to query iSCSI parameters: %d.\n", rc); goto done; } sc->vres.iscsi.start = val[0]; sc->vres.iscsi.size = val[1] - val[0] + 1; } #undef FW_PARAM_PFVF #undef FW_PARAM_DEV done: return (rc); } static void t4_set_desc(struct adapter *sc) { char buf[128]; struct adapter_params *p = &sc->params; snprintf(buf, sizeof(buf), "Chelsio %s (rev %d) %d port %sNIC PCIe-x%d %d %s, S/N:%s, E/C:%s", p->vpd.id, p->rev, p->nports, is_offload(sc) ? "R" : "", p->pci.width, sc->intr_count, sc->intr_type == INTR_MSIX ? "MSI-X" : (sc->intr_type == INTR_MSI ? "MSI" : "INTx"), p->vpd.sn, p->vpd.ec); device_set_desc_copy(sc->dev, buf); } static void build_medialist(struct port_info *pi) { struct ifmedia *media = &pi->media; int data, m; PORT_LOCK(pi); ifmedia_removeall(media); m = IFM_ETHER | IFM_FDX; data = (pi->port_type << 8) | pi->mod_type; switch(pi->port_type) { case FW_PORT_TYPE_BT_XFI: ifmedia_add(media, m | IFM_10G_T, data, NULL); break; case FW_PORT_TYPE_BT_XAUI: ifmedia_add(media, m | IFM_10G_T, data, NULL); /* fall through */ case FW_PORT_TYPE_BT_SGMII: ifmedia_add(media, m | IFM_1000_T, data, NULL); ifmedia_add(media, m | IFM_100_TX, data, NULL); ifmedia_add(media, IFM_ETHER | IFM_AUTO, data, NULL); ifmedia_set(media, IFM_ETHER | IFM_AUTO); break; case FW_PORT_TYPE_CX4: ifmedia_add(media, m | IFM_10G_CX4, data, NULL); ifmedia_set(media, m | IFM_10G_CX4); break; case FW_PORT_TYPE_SFP: case FW_PORT_TYPE_FIBER_XFI: case FW_PORT_TYPE_FIBER_XAUI: switch (pi->mod_type) { case FW_PORT_MOD_TYPE_LR: ifmedia_add(media, m | IFM_10G_LR, data, NULL); ifmedia_set(media, m | IFM_10G_LR); break; case FW_PORT_MOD_TYPE_SR: ifmedia_add(media, m | IFM_10G_SR, data, NULL); ifmedia_set(media, m | IFM_10G_SR); break; case FW_PORT_MOD_TYPE_LRM: ifmedia_add(media, m | IFM_10G_LRM, data, NULL); ifmedia_set(media, m | IFM_10G_LRM); break; case FW_PORT_MOD_TYPE_TWINAX_PASSIVE: case FW_PORT_MOD_TYPE_TWINAX_ACTIVE: ifmedia_add(media, m | IFM_10G_TWINAX, data, NULL); ifmedia_set(media, m | IFM_10G_TWINAX); break; case FW_PORT_MOD_TYPE_NONE: m &= ~IFM_FDX; ifmedia_add(media, m | IFM_NONE, data, NULL); ifmedia_set(media, m | IFM_NONE); break; case FW_PORT_MOD_TYPE_NA: case FW_PORT_MOD_TYPE_ER: default: ifmedia_add(media, m | IFM_UNKNOWN, data, NULL); ifmedia_set(media, m | IFM_UNKNOWN); break; } break; case FW_PORT_TYPE_KX4: case FW_PORT_TYPE_KX: case FW_PORT_TYPE_KR: default: ifmedia_add(media, m | IFM_UNKNOWN, data, NULL); ifmedia_set(media, m | IFM_UNKNOWN); break; } PORT_UNLOCK(pi); } /* * Program the port's XGMAC based on parameters in ifnet. The caller also * indicates which parameters should be programmed (the rest are left alone). */ static int update_mac_settings(struct port_info *pi, int flags) { int rc; struct ifnet *ifp = pi->ifp; struct adapter *sc = pi->adapter; int mtu = -1, promisc = -1, allmulti = -1, vlanex = -1; PORT_LOCK_ASSERT_OWNED(pi); KASSERT(flags, ("%s: not told what to update.", __func__)); if (flags & XGMAC_MTU) mtu = ifp->if_mtu; if (flags & XGMAC_PROMISC) promisc = ifp->if_flags & IFF_PROMISC ? 1 : 0; if (flags & XGMAC_ALLMULTI) allmulti = ifp->if_flags & IFF_ALLMULTI ? 1 : 0; if (flags & XGMAC_VLANEX) vlanex = ifp->if_capenable & IFCAP_VLAN_HWTAGGING ? 1 : 0; rc = -t4_set_rxmode(sc, sc->mbox, pi->viid, mtu, promisc, allmulti, 1, vlanex, false); if (rc) { if_printf(ifp, "set_rxmode (%x) failed: %d\n", flags, rc); return (rc); } if (flags & XGMAC_UCADDR) { uint8_t ucaddr[ETHER_ADDR_LEN]; bcopy(IF_LLADDR(ifp), ucaddr, sizeof(ucaddr)); rc = t4_change_mac(sc, sc->mbox, pi->viid, pi->xact_addr_filt, ucaddr, true, true); if (rc < 0) { rc = -rc; if_printf(ifp, "change_mac failed: %d\n", rc); return (rc); } else { pi->xact_addr_filt = rc; rc = 0; } } if (flags & XGMAC_MCADDRS) { const uint8_t *mcaddr; int del = 1; uint64_t hash = 0; struct ifmultiaddr *ifma; if_maddr_rlock(ifp); TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) { if (ifma->ifma_addr->sa_family != AF_LINK) continue; mcaddr = LLADDR((struct sockaddr_dl *)ifma->ifma_addr); rc = t4_alloc_mac_filt(sc, sc->mbox, pi->viid, del, 1, &mcaddr, NULL, &hash, 0); if (rc < 0) { rc = -rc; if_printf(ifp, "failed to add mc address" " %02x:%02x:%02x:%02x:%02x:%02x rc=%d\n", mcaddr[0], mcaddr[1], mcaddr[2], mcaddr[3], mcaddr[4], mcaddr[5], rc); goto mcfail; } del = 0; } rc = -t4_set_addr_hash(sc, sc->mbox, pi->viid, 0, hash, 0); if (rc != 0) if_printf(ifp, "failed to set mc address hash: %d", rc); mcfail: if_maddr_runlock(ifp); } return (rc); } static int cxgbe_init_locked(struct port_info *pi) { struct adapter *sc = pi->adapter; int rc = 0; ADAPTER_LOCK_ASSERT_OWNED(sc); while (!IS_DOOMED(pi) && IS_BUSY(sc)) { if (mtx_sleep(&sc->flags, &sc->sc_lock, PCATCH, "t4init", 0)) { rc = EINTR; goto done; } } if (IS_DOOMED(pi)) { rc = ENXIO; goto done; } KASSERT(!IS_BUSY(sc), ("%s: controller busy.", __func__)); /* Give up the adapter lock, port init code can sleep. */ SET_BUSY(sc); ADAPTER_UNLOCK(sc); rc = cxgbe_init_synchronized(pi); done: ADAPTER_LOCK(sc); KASSERT(IS_BUSY(sc), ("%s: controller not busy.", __func__)); CLR_BUSY(sc); wakeup_one(&sc->flags); ADAPTER_UNLOCK(sc); return (rc); } static int cxgbe_init_synchronized(struct port_info *pi) { struct adapter *sc = pi->adapter; struct ifnet *ifp = pi->ifp; int rc = 0, i; uint16_t *rss; struct sge_rxq *rxq; ADAPTER_LOCK_ASSERT_NOTOWNED(sc); if (isset(&sc->open_device_map, pi->port_id)) { KASSERT(ifp->if_drv_flags & IFF_DRV_RUNNING, ("mismatch between open_device_map and if_drv_flags")); return (0); /* already running */ } if (sc->open_device_map == 0 && ((rc = first_port_up(sc)) != 0)) return (rc); /* error message displayed already */ /* * Allocate tx/rx/fl queues for this port. */ rc = t4_setup_eth_queues(pi); if (rc != 0) goto done; /* error message displayed already */ /* * Setup RSS for this port. */ rss = malloc(pi->nrxq * sizeof (*rss), M_CXGBE, M_ZERO | M_WAITOK); for_each_rxq(pi, i, rxq) { rss[i] = rxq->iq.abs_id; } rc = -t4_config_rss_range(sc, sc->mbox, pi->viid, 0, pi->rss_size, rss, pi->nrxq); free(rss, M_CXGBE); if (rc != 0) { if_printf(ifp, "rss_config failed: %d\n", rc); goto done; } PORT_LOCK(pi); rc = update_mac_settings(pi, XGMAC_ALL); PORT_UNLOCK(pi); if (rc) goto done; /* error message displayed already */ rc = -t4_link_start(sc, sc->mbox, pi->tx_chan, &pi->link_cfg); if (rc != 0) { if_printf(ifp, "start_link failed: %d\n", rc); goto done; } rc = -t4_enable_vi(sc, sc->mbox, pi->viid, true, true); if (rc != 0) { if_printf(ifp, "enable_vi failed: %d\n", rc); goto done; } pi->flags |= VI_ENABLED; /* all ok */ setbit(&sc->open_device_map, pi->port_id); ifp->if_drv_flags |= IFF_DRV_RUNNING; ifp->if_drv_flags &= ~IFF_DRV_OACTIVE; callout_reset(&pi->tick, hz, cxgbe_tick, pi); done: if (rc != 0) cxgbe_uninit_synchronized(pi); return (rc); } static int cxgbe_uninit_locked(struct port_info *pi) { struct adapter *sc = pi->adapter; int rc; ADAPTER_LOCK_ASSERT_OWNED(sc); while (!IS_DOOMED(pi) && IS_BUSY(sc)) { if (mtx_sleep(&sc->flags, &sc->sc_lock, PCATCH, "t4uninit", 0)) { rc = EINTR; goto done; } } if (IS_DOOMED(pi)) { rc = ENXIO; goto done; } KASSERT(!IS_BUSY(sc), ("%s: controller busy.", __func__)); SET_BUSY(sc); ADAPTER_UNLOCK(sc); rc = cxgbe_uninit_synchronized(pi); ADAPTER_LOCK(sc); KASSERT(IS_BUSY(sc), ("%s: controller not busy.", __func__)); CLR_BUSY(sc); wakeup_one(&sc->flags); done: ADAPTER_UNLOCK(sc); return (rc); } /* * Idempotent. */ static int cxgbe_uninit_synchronized(struct port_info *pi) { struct adapter *sc = pi->adapter; struct ifnet *ifp = pi->ifp; int rc; /* * taskqueue_drain may cause a deadlock if the adapter lock is held. */ ADAPTER_LOCK_ASSERT_NOTOWNED(sc); /* * Clear this port's bit from the open device map, and then drain * tasks and callouts. */ clrbit(&sc->open_device_map, pi->port_id); PORT_LOCK(pi); ifp->if_drv_flags &= ~(IFF_DRV_RUNNING | IFF_DRV_OACTIVE); callout_stop(&pi->tick); PORT_UNLOCK(pi); callout_drain(&pi->tick); /* * Stop and then free the queues' resources, including the queues * themselves. * * XXX: we could just stop the queues here (on ifconfig down) and free * them later (on port detach), but having up/down go through the entire * allocate/activate/deactivate/free sequence is a good way to find * leaks and bugs. */ rc = t4_teardown_eth_queues(pi); if (rc != 0) if_printf(ifp, "teardown failed: %d\n", rc); if (pi->flags & VI_ENABLED) { rc = -t4_enable_vi(sc, sc->mbox, pi->viid, false, false); if (rc) if_printf(ifp, "disable_vi failed: %d\n", rc); else pi->flags &= ~VI_ENABLED; } pi->link_cfg.link_ok = 0; pi->link_cfg.speed = 0; t4_os_link_changed(sc, pi->port_id, 0); if (sc->open_device_map == 0) last_port_down(sc); return (0); } #define T4_ALLOC_IRQ(sc, irqid, rid, handler, arg, name) do { \ rc = t4_alloc_irq(sc, &sc->irq[irqid], rid, handler, arg, name); \ if (rc != 0) \ goto done; \ } while (0) static int first_port_up(struct adapter *sc) { int rc, i; char name[8]; ADAPTER_LOCK_ASSERT_NOTOWNED(sc); /* * queues that belong to the adapter (not any particular port). */ rc = t4_setup_adapter_queues(sc); if (rc != 0) goto done; /* * Setup interrupts. */ if (sc->intr_count == 1) { KASSERT(sc->flags & INTR_FWD, ("%s: single interrupt but not forwarded?", __func__)); T4_ALLOC_IRQ(sc, 0, 0, t4_intr_all, sc, "all"); } else { /* Multiple interrupts. The first one is always error intr */ T4_ALLOC_IRQ(sc, 0, 1, t4_intr_err, sc, "err"); if (sc->flags & INTR_FWD) { /* The rest are shared by the fwq and all data intr */ for (i = 1; i < sc->intr_count; i++) { snprintf(name, sizeof(name), "mux%d", i - 1); T4_ALLOC_IRQ(sc, i, i + 1, t4_intr_fwd, &sc->sge.fiq[i - 1], name); } } else { struct port_info *pi; int p, q; T4_ALLOC_IRQ(sc, 1, 2, t4_intr_evt, &sc->sge.fwq, "evt"); p = q = 0; pi = sc->port[p]; for (i = 2; i < sc->intr_count; i++) { snprintf(name, sizeof(name), "p%dq%d", p, q); if (++q >= pi->nrxq) { p++; q = 0; pi = sc->port[p]; } T4_ALLOC_IRQ(sc, i, i + 1, t4_intr_data, &sc->sge.rxq[i - 2], name); } } } t4_intr_enable(sc); sc->flags |= FULL_INIT_DONE; done: if (rc != 0) last_port_down(sc); return (rc); } #undef T4_ALLOC_IRQ /* * Idempotent. */ static int last_port_down(struct adapter *sc) { int i; ADAPTER_LOCK_ASSERT_NOTOWNED(sc); t4_intr_disable(sc); t4_teardown_adapter_queues(sc); for (i = 0; i < sc->intr_count; i++) t4_free_irq(sc, &sc->irq[i]); sc->flags &= ~FULL_INIT_DONE; return (0); } static int t4_alloc_irq(struct adapter *sc, struct irq *irq, int rid, iq_intr_handler_t *handler, void *arg, char *name) { int rc; irq->rid = rid; irq->res = bus_alloc_resource_any(sc->dev, SYS_RES_IRQ, &irq->rid, RF_SHAREABLE | RF_ACTIVE); if (irq->res == NULL) { device_printf(sc->dev, "failed to allocate IRQ for rid %d, name %s.\n", rid, name); return (ENOMEM); } rc = bus_setup_intr(sc->dev, irq->res, INTR_MPSAFE | INTR_TYPE_NET, NULL, handler, arg, &irq->tag); if (rc != 0) { device_printf(sc->dev, "failed to setup interrupt for rid %d, name %s: %d\n", rid, name, rc); } else if (name) bus_describe_intr(sc->dev, irq->res, irq->tag, name); return (rc); } static int t4_free_irq(struct adapter *sc, struct irq *irq) { if (irq->tag) bus_teardown_intr(sc->dev, irq->res, irq->tag); if (irq->res) bus_release_resource(sc->dev, SYS_RES_IRQ, irq->rid, irq->res); bzero(irq, sizeof(*irq)); return (0); } static void reg_block_dump(struct adapter *sc, uint8_t *buf, unsigned int start, unsigned int end) { uint32_t *p = (uint32_t *)(buf + start); for ( ; start <= end; start += sizeof(uint32_t)) *p++ = t4_read_reg(sc, start); } static void t4_get_regs(struct adapter *sc, struct t4_regdump *regs, uint8_t *buf) { int i; static const unsigned int reg_ranges[] = { 0x1008, 0x1108, 0x1180, 0x11b4, 0x11fc, 0x123c, 0x1300, 0x173c, 0x1800, 0x18fc, 0x3000, 0x30d8, 0x30e0, 0x5924, 0x5960, 0x59d4, 0x5a00, 0x5af8, 0x6000, 0x6098, 0x6100, 0x6150, 0x6200, 0x6208, 0x6240, 0x6248, 0x6280, 0x6338, 0x6370, 0x638c, 0x6400, 0x643c, 0x6500, 0x6524, 0x6a00, 0x6a38, 0x6a60, 0x6a78, 0x6b00, 0x6b84, 0x6bf0, 0x6c84, 0x6cf0, 0x6d84, 0x6df0, 0x6e84, 0x6ef0, 0x6f84, 0x6ff0, 0x7084, 0x70f0, 0x7184, 0x71f0, 0x7284, 0x72f0, 0x7384, 0x73f0, 0x7450, 0x7500, 0x7530, 0x7600, 0x761c, 0x7680, 0x76cc, 0x7700, 0x7798, 0x77c0, 0x77fc, 0x7900, 0x79fc, 0x7b00, 0x7c38, 0x7d00, 0x7efc, 0x8dc0, 0x8e1c, 0x8e30, 0x8e78, 0x8ea0, 0x8f6c, 0x8fc0, 0x9074, 0x90fc, 0x90fc, 0x9400, 0x9458, 0x9600, 0x96bc, 0x9800, 0x9808, 0x9820, 0x983c, 0x9850, 0x9864, 0x9c00, 0x9c6c, 0x9c80, 0x9cec, 0x9d00, 0x9d6c, 0x9d80, 0x9dec, 0x9e00, 0x9e6c, 0x9e80, 0x9eec, 0x9f00, 0x9f6c, 0x9f80, 0x9fec, 0xd004, 0xd03c, 0xdfc0, 0xdfe0, 0xe000, 0xea7c, 0xf000, 0x11190, 0x19040, 0x19124, 0x19150, 0x191b0, 0x191d0, 0x191e8, 0x19238, 0x1924c, 0x193f8, 0x19474, 0x19490, 0x194f8, 0x19800, 0x19f30, 0x1a000, 0x1a06c, 0x1a0b0, 0x1a120, 0x1a128, 0x1a138, 0x1a190, 0x1a1c4, 0x1a1fc, 0x1a1fc, 0x1e040, 0x1e04c, 0x1e240, 0x1e28c, 0x1e2c0, 0x1e2c0, 0x1e2e0, 0x1e2e0, 0x1e300, 0x1e384, 0x1e3c0, 0x1e3c8, 0x1e440, 0x1e44c, 0x1e640, 0x1e68c, 0x1e6c0, 0x1e6c0, 0x1e6e0, 0x1e6e0, 0x1e700, 0x1e784, 0x1e7c0, 0x1e7c8, 0x1e840, 0x1e84c, 0x1ea40, 0x1ea8c, 0x1eac0, 0x1eac0, 0x1eae0, 0x1eae0, 0x1eb00, 0x1eb84, 0x1ebc0, 0x1ebc8, 0x1ec40, 0x1ec4c, 0x1ee40, 0x1ee8c, 0x1eec0, 0x1eec0, 0x1eee0, 0x1eee0, 0x1ef00, 0x1ef84, 0x1efc0, 0x1efc8, 0x1f040, 0x1f04c, 0x1f240, 0x1f28c, 0x1f2c0, 0x1f2c0, 0x1f2e0, 0x1f2e0, 0x1f300, 0x1f384, 0x1f3c0, 0x1f3c8, 0x1f440, 0x1f44c, 0x1f640, 0x1f68c, 0x1f6c0, 0x1f6c0, 0x1f6e0, 0x1f6e0, 0x1f700, 0x1f784, 0x1f7c0, 0x1f7c8, 0x1f840, 0x1f84c, 0x1fa40, 0x1fa8c, 0x1fac0, 0x1fac0, 0x1fae0, 0x1fae0, 0x1fb00, 0x1fb84, 0x1fbc0, 0x1fbc8, 0x1fc40, 0x1fc4c, 0x1fe40, 0x1fe8c, 0x1fec0, 0x1fec0, 0x1fee0, 0x1fee0, 0x1ff00, 0x1ff84, 0x1ffc0, 0x1ffc8, 0x20000, 0x2002c, 0x20100, 0x2013c, 0x20190, 0x201c8, 0x20200, 0x20318, 0x20400, 0x20528, 0x20540, 0x20614, 0x21000, 0x21040, 0x2104c, 0x21060, 0x210c0, 0x210ec, 0x21200, 0x21268, 0x21270, 0x21284, 0x212fc, 0x21388, 0x21400, 0x21404, 0x21500, 0x21518, 0x2152c, 0x2153c, 0x21550, 0x21554, 0x21600, 0x21600, 0x21608, 0x21628, 0x21630, 0x2163c, 0x21700, 0x2171c, 0x21780, 0x2178c, 0x21800, 0x21c38, 0x21c80, 0x21d7c, 0x21e00, 0x21e04, 0x22000, 0x2202c, 0x22100, 0x2213c, 0x22190, 0x221c8, 0x22200, 0x22318, 0x22400, 0x22528, 0x22540, 0x22614, 0x23000, 0x23040, 0x2304c, 0x23060, 0x230c0, 0x230ec, 0x23200, 0x23268, 0x23270, 0x23284, 0x232fc, 0x23388, 0x23400, 0x23404, 0x23500, 0x23518, 0x2352c, 0x2353c, 0x23550, 0x23554, 0x23600, 0x23600, 0x23608, 0x23628, 0x23630, 0x2363c, 0x23700, 0x2371c, 0x23780, 0x2378c, 0x23800, 0x23c38, 0x23c80, 0x23d7c, 0x23e00, 0x23e04, 0x24000, 0x2402c, 0x24100, 0x2413c, 0x24190, 0x241c8, 0x24200, 0x24318, 0x24400, 0x24528, 0x24540, 0x24614, 0x25000, 0x25040, 0x2504c, 0x25060, 0x250c0, 0x250ec, 0x25200, 0x25268, 0x25270, 0x25284, 0x252fc, 0x25388, 0x25400, 0x25404, 0x25500, 0x25518, 0x2552c, 0x2553c, 0x25550, 0x25554, 0x25600, 0x25600, 0x25608, 0x25628, 0x25630, 0x2563c, 0x25700, 0x2571c, 0x25780, 0x2578c, 0x25800, 0x25c38, 0x25c80, 0x25d7c, 0x25e00, 0x25e04, 0x26000, 0x2602c, 0x26100, 0x2613c, 0x26190, 0x261c8, 0x26200, 0x26318, 0x26400, 0x26528, 0x26540, 0x26614, 0x27000, 0x27040, 0x2704c, 0x27060, 0x270c0, 0x270ec, 0x27200, 0x27268, 0x27270, 0x27284, 0x272fc, 0x27388, 0x27400, 0x27404, 0x27500, 0x27518, 0x2752c, 0x2753c, 0x27550, 0x27554, 0x27600, 0x27600, 0x27608, 0x27628, 0x27630, 0x2763c, 0x27700, 0x2771c, 0x27780, 0x2778c, 0x27800, 0x27c38, 0x27c80, 0x27d7c, 0x27e00, 0x27e04 }; regs->version = 4 | (sc->params.rev << 10); for (i = 0; i < ARRAY_SIZE(reg_ranges); i += 2) reg_block_dump(sc, buf, reg_ranges[i], reg_ranges[i + 1]); } static void cxgbe_tick(void *arg) { struct port_info *pi = arg; struct ifnet *ifp = pi->ifp; struct sge_txq *txq; int i, drops; struct port_stats *s = &pi->stats; PORT_LOCK(pi); if (!(ifp->if_drv_flags & IFF_DRV_RUNNING)) { PORT_UNLOCK(pi); return; /* without scheduling another callout */ } t4_get_port_stats(pi->adapter, pi->tx_chan, s); ifp->if_opackets = s->tx_frames; ifp->if_ipackets = s->rx_frames; ifp->if_obytes = s->tx_octets; ifp->if_ibytes = s->rx_octets; ifp->if_omcasts = s->tx_mcast_frames; ifp->if_imcasts = s->rx_mcast_frames; ifp->if_iqdrops = s->rx_ovflow0 + s->rx_ovflow1 + s->rx_ovflow2 + s->rx_ovflow3; drops = s->tx_drop; for_each_txq(pi, i, txq) drops += txq->br->br_drops; ifp->if_snd.ifq_drops = drops; ifp->if_oerrors = s->tx_error_frames; ifp->if_ierrors = s->rx_jabber + s->rx_runt + s->rx_too_long + s->rx_fcs_err + s->rx_len_err; callout_schedule(&pi->tick, hz); PORT_UNLOCK(pi); } static int t4_sysctls(struct adapter *sc) { struct sysctl_ctx_list *ctx; struct sysctl_oid *oid; struct sysctl_oid_list *children; ctx = device_get_sysctl_ctx(sc->dev); oid = device_get_sysctl_tree(sc->dev); children = SYSCTL_CHILDREN(oid); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "nports", CTLFLAG_RD, &sc->params.nports, 0, "# of ports"); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "hw_revision", CTLFLAG_RD, &sc->params.rev, 0, "chip hardware revision"); SYSCTL_ADD_STRING(ctx, children, OID_AUTO, "firmware_version", CTLFLAG_RD, &sc->fw_version, 0, "firmware version"); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "TOE", CTLFLAG_RD, &sc->params.offload, 0, "hardware is capable of TCP offload"); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "core_clock", CTLFLAG_RD, &sc->params.vpd.cclk, 0, "core clock frequency (in KHz)"); SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "holdoff_timers", CTLTYPE_STRING | CTLFLAG_RD, &intr_timer, sizeof(intr_timer), sysctl_int_array, "A", "interrupt holdoff timer values (us)"); SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "holdoff_pkt_counts", CTLTYPE_STRING | CTLFLAG_RD, &intr_pktcount, sizeof(intr_pktcount), sysctl_int_array, "A", "interrupt holdoff packet counter values"); return (0); } static int cxgbe_sysctls(struct port_info *pi) { struct sysctl_ctx_list *ctx; struct sysctl_oid *oid; struct sysctl_oid_list *children; ctx = device_get_sysctl_ctx(pi->dev); /* * dev.cxgbe.X. */ oid = device_get_sysctl_tree(pi->dev); children = SYSCTL_CHILDREN(oid); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "nrxq", CTLFLAG_RD, &pi->nrxq, 0, "# of rx queues"); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "ntxq", CTLFLAG_RD, &pi->ntxq, 0, "# of tx queues"); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "first_rxq", CTLFLAG_RD, &pi->first_rxq, 0, "index of first rx queue"); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "first_txq", CTLFLAG_RD, &pi->first_txq, 0, "index of first tx queue"); SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "holdoff_tmr_idx", CTLTYPE_INT | CTLFLAG_RW, pi, 0, sysctl_holdoff_tmr_idx, "I", "holdoff timer index"); SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "holdoff_pktc_idx", CTLTYPE_INT | CTLFLAG_RW, pi, 0, sysctl_holdoff_pktc_idx, "I", "holdoff packet counter index"); SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "qsize_rxq", CTLTYPE_INT | CTLFLAG_RW, pi, 0, sysctl_qsize_rxq, "I", "rx queue size"); SYSCTL_ADD_PROC(ctx, children, OID_AUTO, "qsize_txq", CTLTYPE_INT | CTLFLAG_RW, pi, 0, sysctl_qsize_txq, "I", "tx queue size"); /* * dev.cxgbe.X.stats. */ oid = SYSCTL_ADD_NODE(ctx, children, OID_AUTO, "stats", CTLFLAG_RD, NULL, "port statistics"); children = SYSCTL_CHILDREN(oid); #define SYSCTL_ADD_T4_REG64(pi, name, desc, reg) \ SYSCTL_ADD_OID(ctx, children, OID_AUTO, name, \ CTLTYPE_U64 | CTLFLAG_RD, pi->adapter, reg, \ sysctl_handle_t4_reg64, "QU", desc) SYSCTL_ADD_T4_REG64(pi, "tx_octets", "# of octets in good frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_BYTES_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames", "total # of good frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_FRAMES_L)); SYSCTL_ADD_T4_REG64(pi, "tx_bcast_frames", "# of broadcast frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_BCAST_L)); SYSCTL_ADD_T4_REG64(pi, "tx_mcast_frames", "# of multicast frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_MCAST_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ucast_frames", "# of unicast frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_UCAST_L)); SYSCTL_ADD_T4_REG64(pi, "tx_error_frames", "# of error frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_ERROR_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames_64", "# of tx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_64B_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames_65_127", "# of tx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_65B_127B_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames_128_255", "# of tx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_128B_255B_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames_256_511", "# of tx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_256B_511B_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames_512_1023", "# of tx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_512B_1023B_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames_1024_1518", "# of tx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_1024B_1518B_L)); SYSCTL_ADD_T4_REG64(pi, "tx_frames_1519_max", "# of tx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_1519B_MAX_L)); SYSCTL_ADD_T4_REG64(pi, "tx_drop", "# of dropped tx frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_DROP_L)); SYSCTL_ADD_T4_REG64(pi, "tx_pause", "# of pause frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PAUSE_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp0", "# of PPP prio 0 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP0_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp1", "# of PPP prio 1 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP1_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp2", "# of PPP prio 2 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP2_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp3", "# of PPP prio 3 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP3_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp4", "# of PPP prio 4 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP4_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp5", "# of PPP prio 5 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP5_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp6", "# of PPP prio 6 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP6_L)); SYSCTL_ADD_T4_REG64(pi, "tx_ppp7", "# of PPP prio 7 frames transmitted", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_TX_PORT_PPP7_L)); SYSCTL_ADD_T4_REG64(pi, "rx_octets", "# of octets in good frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_BYTES_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames", "total # of good frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_FRAMES_L)); SYSCTL_ADD_T4_REG64(pi, "rx_bcast_frames", "# of broadcast frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_BCAST_L)); SYSCTL_ADD_T4_REG64(pi, "rx_mcast_frames", "# of multicast frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_MCAST_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ucast_frames", "# of unicast frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_UCAST_L)); SYSCTL_ADD_T4_REG64(pi, "rx_too_long", "# of frames exceeding MTU", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_MTU_ERROR_L)); SYSCTL_ADD_T4_REG64(pi, "rx_jabber", "# of jabber frames", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_MTU_CRC_ERROR_L)); SYSCTL_ADD_T4_REG64(pi, "rx_fcs_err", "# of frames received with bad FCS", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_CRC_ERROR_L)); SYSCTL_ADD_T4_REG64(pi, "rx_len_err", "# of frames received with length error", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_LEN_ERROR_L)); SYSCTL_ADD_T4_REG64(pi, "rx_symbol_err", "symbol errors", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_SYM_ERROR_L)); SYSCTL_ADD_T4_REG64(pi, "rx_runt", "# of short frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_LESS_64B_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames_64", "# of rx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_64B_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames_65_127", "# of rx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_65B_127B_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames_128_255", "# of rx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_128B_255B_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames_256_511", "# of rx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_256B_511B_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames_512_1023", "# of rx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_512B_1023B_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames_1024_1518", "# of rx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_1024B_1518B_L)); SYSCTL_ADD_T4_REG64(pi, "rx_frames_1519_max", "# of rx frames in this range", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_1519B_MAX_L)); SYSCTL_ADD_T4_REG64(pi, "rx_pause", "# of pause frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PAUSE_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp0", "# of PPP prio 0 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP0_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp1", "# of PPP prio 1 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP1_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp2", "# of PPP prio 2 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP2_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp3", "# of PPP prio 3 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP3_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp4", "# of PPP prio 4 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP4_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp5", "# of PPP prio 5 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP5_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp6", "# of PPP prio 6 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP6_L)); SYSCTL_ADD_T4_REG64(pi, "rx_ppp7", "# of PPP prio 7 frames received", PORT_REG(pi->tx_chan, A_MPS_PORT_STAT_RX_PORT_PPP7_L)); #undef SYSCTL_ADD_T4_REG64 #define SYSCTL_ADD_T4_PORTSTAT(name, desc) \ SYSCTL_ADD_UQUAD(ctx, children, OID_AUTO, #name, CTLFLAG_RD, \ &pi->stats.name, desc) /* We get these from port_stats and they may be stale by upto 1s */ SYSCTL_ADD_T4_PORTSTAT(rx_ovflow0, "# drops due to buffer-group 0 overflows"); SYSCTL_ADD_T4_PORTSTAT(rx_ovflow1, "# drops due to buffer-group 1 overflows"); SYSCTL_ADD_T4_PORTSTAT(rx_ovflow2, "# drops due to buffer-group 2 overflows"); SYSCTL_ADD_T4_PORTSTAT(rx_ovflow3, "# drops due to buffer-group 3 overflows"); SYSCTL_ADD_T4_PORTSTAT(rx_trunc0, "# of buffer-group 0 truncated packets"); SYSCTL_ADD_T4_PORTSTAT(rx_trunc1, "# of buffer-group 1 truncated packets"); SYSCTL_ADD_T4_PORTSTAT(rx_trunc2, "# of buffer-group 2 truncated packets"); SYSCTL_ADD_T4_PORTSTAT(rx_trunc3, "# of buffer-group 3 truncated packets"); #undef SYSCTL_ADD_T4_PORTSTAT return (0); } static int sysctl_int_array(SYSCTL_HANDLER_ARGS) { int rc, *i; struct sbuf sb; sbuf_new(&sb, NULL, 32, SBUF_AUTOEXTEND); for (i = arg1; arg2; arg2 -= sizeof(int), i++) sbuf_printf(&sb, "%d ", *i); sbuf_trim(&sb); sbuf_finish(&sb); rc = sysctl_handle_string(oidp, sbuf_data(&sb), sbuf_len(&sb), req); sbuf_delete(&sb); return (rc); } static int sysctl_holdoff_tmr_idx(SYSCTL_HANDLER_ARGS) { struct port_info *pi = arg1; struct adapter *sc = pi->adapter; struct sge_rxq *rxq; int idx, rc, i; idx = pi->tmr_idx; rc = sysctl_handle_int(oidp, &idx, 0, req); if (rc != 0 || req->newptr == NULL) return (rc); if (idx < 0 || idx >= SGE_NTIMERS) return (EINVAL); ADAPTER_LOCK(sc); rc = IS_DOOMED(pi) ? ENXIO : (IS_BUSY(sc) ? EBUSY : 0); if (rc == 0) { for_each_rxq(pi, i, rxq) { rxq->iq.intr_params = V_QINTR_TIMER_IDX(idx) | V_QINTR_CNT_EN(pi->pktc_idx != -1); } pi->tmr_idx = idx; } ADAPTER_UNLOCK(sc); return (rc); } static int sysctl_holdoff_pktc_idx(SYSCTL_HANDLER_ARGS) { struct port_info *pi = arg1; struct adapter *sc = pi->adapter; int idx, rc; idx = pi->pktc_idx; rc = sysctl_handle_int(oidp, &idx, 0, req); if (rc != 0 || req->newptr == NULL) return (rc); if (idx < -1 || idx >= SGE_NCOUNTERS) return (EINVAL); ADAPTER_LOCK(sc); rc = IS_DOOMED(pi) ? ENXIO : (IS_BUSY(sc) ? EBUSY : 0); if (rc == 0 && pi->ifp->if_drv_flags & IFF_DRV_RUNNING) rc = EBUSY; /* can be changed only when port is down */ if (rc == 0) pi->pktc_idx = idx; ADAPTER_UNLOCK(sc); return (rc); } static int sysctl_qsize_rxq(SYSCTL_HANDLER_ARGS) { struct port_info *pi = arg1; struct adapter *sc = pi->adapter; int qsize, rc; qsize = pi->qsize_rxq; rc = sysctl_handle_int(oidp, &qsize, 0, req); if (rc != 0 || req->newptr == NULL) return (rc); if (qsize < 128 || (qsize & 7)) return (EINVAL); ADAPTER_LOCK(sc); rc = IS_DOOMED(pi) ? ENXIO : (IS_BUSY(sc) ? EBUSY : 0); if (rc == 0 && pi->ifp->if_drv_flags & IFF_DRV_RUNNING) rc = EBUSY; /* can be changed only when port is down */ if (rc == 0) pi->qsize_rxq = qsize; ADAPTER_UNLOCK(sc); return (rc); } static int sysctl_qsize_txq(SYSCTL_HANDLER_ARGS) { struct port_info *pi = arg1; struct adapter *sc = pi->adapter; int qsize, rc; qsize = pi->qsize_txq; rc = sysctl_handle_int(oidp, &qsize, 0, req); if (rc != 0 || req->newptr == NULL) return (rc); if (qsize < 128) return (EINVAL); ADAPTER_LOCK(sc); rc = IS_DOOMED(pi) ? ENXIO : (IS_BUSY(sc) ? EBUSY : 0); if (rc == 0 && pi->ifp->if_drv_flags & IFF_DRV_RUNNING) rc = EBUSY; /* can be changed only when port is down */ if (rc == 0) pi->qsize_txq = qsize; ADAPTER_UNLOCK(sc); return (rc); } static int sysctl_handle_t4_reg64(SYSCTL_HANDLER_ARGS) { struct adapter *sc = arg1; int reg = arg2; uint64_t val; val = t4_read_reg64(sc, reg); return (sysctl_handle_64(oidp, &val, 0, req)); } static inline void txq_start(struct ifnet *ifp, struct sge_txq *txq) { struct buf_ring *br; struct mbuf *m; TXQ_LOCK_ASSERT_OWNED(txq); br = txq->br; m = txq->m ? txq->m : drbr_dequeue(ifp, br); if (m) t4_eth_tx(ifp, txq, m); } void cxgbe_txq_start(void *arg, int count) { struct sge_txq *txq = arg; TXQ_LOCK(txq); if (txq->eq.flags & EQ_CRFLUSHED) { txq->eq.flags &= ~EQ_CRFLUSHED; txq_start(txq->ifp, txq); } else wakeup_one(txq); /* txq is going away, wakeup free_txq */ TXQ_UNLOCK(txq); } +static uint32_t +fconf_to_mode(uint32_t fconf) +{ + uint32_t mode; + + mode = T4_FILTER_IPv4 | T4_FILTER_IPv6 | T4_FILTER_IP_SADDR | + T4_FILTER_IP_DADDR | T4_FILTER_IP_SPORT | T4_FILTER_IP_DPORT; + + if (fconf & F_FRAGMENTATION) + mode |= T4_FILTER_IP_FRAGMENT; + + if (fconf & F_MPSHITTYPE) + mode |= T4_FILTER_MPS_HIT_TYPE; + + if (fconf & F_MACMATCH) + mode |= T4_FILTER_MAC_IDX; + + if (fconf & F_ETHERTYPE) + mode |= T4_FILTER_ETH_TYPE; + + if (fconf & F_PROTOCOL) + mode |= T4_FILTER_IP_PROTO; + + if (fconf & F_TOS) + mode |= T4_FILTER_IP_TOS; + + if (fconf & F_VLAN) + mode |= T4_FILTER_IVLAN; + + if (fconf & F_VNIC_ID) + mode |= T4_FILTER_OVLAN; + + if (fconf & F_PORT) + mode |= T4_FILTER_PORT; + + if (fconf & F_FCOE) + mode |= T4_FILTER_FCoE; + + return (mode); +} + +static uint32_t +mode_to_fconf(uint32_t mode) +{ + uint32_t fconf = 0; + + if (mode & T4_FILTER_IP_FRAGMENT) + fconf |= F_FRAGMENTATION; + + if (mode & T4_FILTER_MPS_HIT_TYPE) + fconf |= F_MPSHITTYPE; + + if (mode & T4_FILTER_MAC_IDX) + fconf |= F_MACMATCH; + + if (mode & T4_FILTER_ETH_TYPE) + fconf |= F_ETHERTYPE; + + if (mode & T4_FILTER_IP_PROTO) + fconf |= F_PROTOCOL; + + if (mode & T4_FILTER_IP_TOS) + fconf |= F_TOS; + + if (mode & T4_FILTER_IVLAN) + fconf |= F_VLAN; + + if (mode & T4_FILTER_OVLAN) + fconf |= F_VNIC_ID; + + if (mode & T4_FILTER_PORT) + fconf |= F_PORT; + + if (mode & T4_FILTER_FCoE) + fconf |= F_FCOE; + + return (fconf); +} + +static uint32_t +fspec_to_fconf(struct t4_filter_specification *fs) +{ + uint32_t fconf = 0; + + if (fs->val.frag || fs->mask.frag) + fconf |= F_FRAGMENTATION; + + if (fs->val.matchtype || fs->mask.matchtype) + fconf |= F_MPSHITTYPE; + + if (fs->val.macidx || fs->mask.macidx) + fconf |= F_MACMATCH; + + if (fs->val.ethtype || fs->mask.ethtype) + fconf |= F_ETHERTYPE; + + if (fs->val.proto || fs->mask.proto) + fconf |= F_PROTOCOL; + + if (fs->val.tos || fs->mask.tos) + fconf |= F_TOS; + + if (fs->val.ivlan_vld || fs->mask.ivlan_vld) + fconf |= F_VLAN; + + if (fs->val.ovlan_vld || fs->mask.ovlan_vld) + fconf |= F_VNIC_ID; + + if (fs->val.iport || fs->mask.iport) + fconf |= F_PORT; + + if (fs->val.fcoe || fs->mask.fcoe) + fconf |= F_FCOE; + + return (fconf); +} + +static int +get_filter_mode(struct adapter *sc, uint32_t *mode) +{ + uint32_t fconf; + + t4_read_indirect(sc, A_TP_PIO_ADDR, A_TP_PIO_DATA, &fconf, 1, + A_TP_VLAN_PRI_MAP); + + *mode = fconf_to_mode(fconf); + + return (0); +} + +static int +set_filter_mode(struct adapter *sc, uint32_t mode) +{ + uint32_t fconf; + int rc; + + fconf = mode_to_fconf(mode); + + ADAPTER_LOCK(sc); + if (IS_BUSY(sc)) { + rc = EAGAIN; + goto done; + } + + if (sc->tids.ftids_in_use > 0) { + rc = EBUSY; + goto done; + } + + rc = -t4_set_filter_mode(sc, fconf); +done: + ADAPTER_UNLOCK(sc); + return (rc); +} + +static int +get_filter(struct adapter *sc, struct t4_filter *t) +{ + int i, nfilters = sc->tids.nftids; + struct filter_entry *f; + + ADAPTER_LOCK_ASSERT_OWNED(sc); + + if (IS_BUSY(sc)) + return (EAGAIN); + + if (sc->tids.ftids_in_use == 0 || sc->tids.ftid_tab == NULL || + t->idx >= nfilters) { + t->idx = 0xffffffff; + return (0); + } + + f = &sc->tids.ftid_tab[t->idx]; + for (i = t->idx; i < nfilters; i++, f++) { + if (f->valid) { + t->idx = i; + t->fs = f->fs; + t->hits = 0; /* XXX implement */ + + return (0); + } + } + + t->idx = 0xffffffff; + return (0); +} + +static int +set_filter(struct adapter *sc, struct t4_filter *t) +{ + uint32_t fconf; + unsigned int nfilters, nports; + struct filter_entry *f; + int i; + + ADAPTER_LOCK_ASSERT_OWNED(sc); + + nfilters = sc->tids.nftids; + nports = sc->params.nports; + + if (nfilters == 0) + return (ENOTSUP); + + if (!(sc->flags & FULL_INIT_DONE)) + return (EAGAIN); + + if (t->idx >= nfilters) + return (EINVAL); + + /* Validate against the global filter mode */ + t4_read_indirect(sc, A_TP_PIO_ADDR, A_TP_PIO_DATA, &fconf, 1, + A_TP_VLAN_PRI_MAP); + if ((fconf | fspec_to_fconf(&t->fs)) != fconf) + return (E2BIG); + + if (t->fs.action == FILTER_SWITCH && t->fs.eport >= nports) + return (EINVAL); + + if (t->fs.val.iport >= nports) + return (EINVAL); + + /* Can't specify an iq if not steering to it */ + if (!t->fs.dirsteer && t->fs.iq) + return (EINVAL); + + /* IPv6 filter idx must be 4 aligned */ + if (t->fs.type == 1 && + ((t->idx & 0x3) || t->idx + 4 >= nfilters)) + return (EINVAL); + + if (sc->tids.ftid_tab == NULL) { + KASSERT(sc->tids.ftids_in_use == 0, + ("%s: no memory allocated but filters_in_use > 0", + __func__)); + + sc->tids.ftid_tab = malloc(sizeof (struct filter_entry) * + nfilters, M_CXGBE, M_NOWAIT | M_ZERO); + if (sc->tids.ftid_tab == NULL) + return (ENOMEM); + } + + for (i = 0; i < 4; i++) { + f = &sc->tids.ftid_tab[t->idx + i]; + + if (f->pending || f->valid) + return (EBUSY); + if (f->locked) + return (EPERM); + + if (t->fs.type == 0) + break; + } + + f = &sc->tids.ftid_tab[t->idx]; + f->fs = t->fs; + + return set_filter_wr(sc, t->idx); +} + +static int +del_filter(struct adapter *sc, struct t4_filter *t) +{ + unsigned int nfilters; + struct filter_entry *f; + + ADAPTER_LOCK_ASSERT_OWNED(sc); + + if (IS_BUSY(sc)) + return (EAGAIN); + + nfilters = sc->tids.nftids; + + if (nfilters == 0) + return (ENOTSUP); + + if (sc->tids.ftid_tab == NULL || sc->tids.ftids_in_use == 0 || + t->idx >= nfilters) + return (EINVAL); + + if (!(sc->flags & FULL_INIT_DONE)) + return (EAGAIN); + + f = &sc->tids.ftid_tab[t->idx]; + + if (f->pending) + return (EBUSY); + if (f->locked) + return (EPERM); + + if (f->valid) { + t->fs = f->fs; /* extra info for the caller */ + return del_filter_wr(sc, t->idx); + } + + return (0); +} + +/* XXX: L2T */ +static void +clear_filter(struct adapter *sc, struct filter_entry *f) +{ + (void) sc; + bzero(f, sizeof (*f)); +} + +static int +set_filter_wr(struct adapter *sc, int fidx) +{ + int rc; + struct filter_entry *f = &sc->tids.ftid_tab[fidx]; + struct mbuf *m; + struct fw_filter_wr *fwr; + unsigned int ftid; + + ADAPTER_LOCK_ASSERT_OWNED(sc); + + if (f->fs.newdmac || f->fs.newvlan) + return (ENOTSUP); /* XXX: fix after L2T code */ + + ftid = sc->tids.ftid_base + fidx; + + m = m_gethdr(M_NOWAIT, MT_DATA); + if (m == NULL) + return (ENOMEM); + + fwr = mtod(m, struct fw_filter_wr *); + m->m_len = m->m_pkthdr.len = sizeof(*fwr); + bzero(fwr, sizeof (*fwr)); + + fwr->op_pkd = htobe32(V_FW_WR_OP(FW_FILTER_WR)); + fwr->len16_pkd = htobe32(FW_LEN16(*fwr)); + fwr->tid_to_iq = + htobe32(V_FW_FILTER_WR_TID(ftid) | + V_FW_FILTER_WR_RQTYPE(f->fs.type) | + V_FW_FILTER_WR_NOREPLY(0) | + V_FW_FILTER_WR_IQ(f->fs.iq)); + fwr->del_filter_to_l2tix = + htobe32(V_FW_FILTER_WR_RPTTID(f->fs.rpttid) | + V_FW_FILTER_WR_DROP(f->fs.action == FILTER_DROP) | + V_FW_FILTER_WR_DIRSTEER(f->fs.dirsteer) | + V_FW_FILTER_WR_MASKHASH(f->fs.maskhash) | + V_FW_FILTER_WR_DIRSTEERHASH(f->fs.dirsteerhash) | + V_FW_FILTER_WR_LPBK(f->fs.action == FILTER_SWITCH) | + V_FW_FILTER_WR_DMAC(f->fs.newdmac) | + V_FW_FILTER_WR_SMAC(f->fs.newsmac) | + V_FW_FILTER_WR_INSVLAN(f->fs.newvlan == VLAN_INSERT || + f->fs.newvlan == VLAN_REWRITE) | + V_FW_FILTER_WR_RMVLAN(f->fs.newvlan == VLAN_REMOVE || + f->fs.newvlan == VLAN_REWRITE) | + V_FW_FILTER_WR_HITCNTS(f->fs.hitcnts) | + V_FW_FILTER_WR_TXCHAN(f->fs.eport) | + V_FW_FILTER_WR_PRIO(f->fs.prio) | + V_FW_FILTER_WR_L2TIX(0)); /* XXX: L2T */ + fwr->ethtype = htobe16(f->fs.val.ethtype); + fwr->ethtypem = htobe16(f->fs.mask.ethtype); + fwr->frag_to_ovlan_vldm = + (V_FW_FILTER_WR_FRAG(f->fs.val.frag) | + V_FW_FILTER_WR_FRAGM(f->fs.mask.frag) | + V_FW_FILTER_WR_IVLAN_VLD(f->fs.val.ivlan_vld) | + V_FW_FILTER_WR_OVLAN_VLD(f->fs.val.ovlan_vld) | + V_FW_FILTER_WR_IVLAN_VLDM(f->fs.mask.ivlan_vld) | + V_FW_FILTER_WR_OVLAN_VLDM(f->fs.mask.ovlan_vld)); + fwr->smac_sel = 0; + fwr->rx_chan_rx_rpl_iq = htobe16(V_FW_FILTER_WR_RX_CHAN(0) | + V_FW_FILTER_WR_RX_RPL_IQ(sc->sge.fwq.abs_id)); + fwr->maci_to_matchtypem = + htobe32(V_FW_FILTER_WR_MACI(f->fs.val.macidx) | + V_FW_FILTER_WR_MACIM(f->fs.mask.macidx) | + V_FW_FILTER_WR_FCOE(f->fs.val.fcoe) | + V_FW_FILTER_WR_FCOEM(f->fs.mask.fcoe) | + V_FW_FILTER_WR_PORT(f->fs.val.iport) | + V_FW_FILTER_WR_PORTM(f->fs.mask.iport) | + V_FW_FILTER_WR_MATCHTYPE(f->fs.val.matchtype) | + V_FW_FILTER_WR_MATCHTYPEM(f->fs.mask.matchtype)); + fwr->ptcl = f->fs.val.proto; + fwr->ptclm = f->fs.mask.proto; + fwr->ttyp = f->fs.val.tos; + fwr->ttypm = f->fs.mask.tos; + fwr->ivlan = htobe16(f->fs.val.ivlan); + fwr->ivlanm = htobe16(f->fs.mask.ivlan); + fwr->ovlan = htobe16(f->fs.val.ovlan); + fwr->ovlanm = htobe16(f->fs.mask.ovlan); + bcopy(f->fs.val.dip, fwr->lip, sizeof (fwr->lip)); + bcopy(f->fs.mask.dip, fwr->lipm, sizeof (fwr->lipm)); + bcopy(f->fs.val.sip, fwr->fip, sizeof (fwr->fip)); + bcopy(f->fs.mask.sip, fwr->fipm, sizeof (fwr->fipm)); + fwr->lp = htobe16(f->fs.val.dport); + fwr->lpm = htobe16(f->fs.mask.dport); + fwr->fp = htobe16(f->fs.val.sport); + fwr->fpm = htobe16(f->fs.mask.sport); + if (f->fs.newsmac) + bcopy(f->fs.smac, fwr->sma, sizeof (fwr->sma)); + + f->pending = 1; + sc->tids.ftids_in_use++; + rc = t4_mgmt_tx(sc, m); + if (rc != 0) { + sc->tids.ftids_in_use--; + m_freem(m); + clear_filter(sc, f); + } + return (rc); +} + +static int +del_filter_wr(struct adapter *sc, int fidx) +{ + struct filter_entry *f = &sc->tids.ftid_tab[fidx]; + struct mbuf *m; + struct fw_filter_wr *fwr; + unsigned int rc, ftid; + + ADAPTER_LOCK_ASSERT_OWNED(sc); + + ftid = sc->tids.ftid_base + fidx; + + m = m_gethdr(M_NOWAIT, MT_DATA); + if (m == NULL) + return (ENOMEM); + + fwr = mtod(m, struct fw_filter_wr *); + m->m_len = m->m_pkthdr.len = sizeof(*fwr); + bzero(fwr, sizeof (*fwr)); + + t4_mk_filtdelwr(ftid, fwr, sc->sge.fwq.abs_id); + + f->pending = 1; + rc = t4_mgmt_tx(sc, m); + if (rc != 0) { + f->pending = 0; + m_freem(m); + } + return (rc); +} + +/* XXX move intr handlers to main.c and make this static */ +void +filter_rpl(struct adapter *sc, const struct cpl_set_tcb_rpl *rpl) +{ + unsigned int idx = GET_TID(rpl); + + if (idx >= sc->tids.ftid_base && + (idx -= sc->tids.ftid_base) < sc->tids.nftids) { + unsigned int rc = G_COOKIE(rpl->cookie); + struct filter_entry *f = &sc->tids.ftid_tab[idx]; + + if (rc == FW_FILTER_WR_FLT_DELETED) { + /* + * Clear the filter when we get confirmation from the + * hardware that the filter has been deleted. + */ + clear_filter(sc, f); + sc->tids.ftids_in_use--; + } else if (rc == FW_FILTER_WR_SMT_TBL_FULL) { + device_printf(sc->dev, + "filter %u setup failed due to full SMT\n", idx); + clear_filter(sc, f); + sc->tids.ftids_in_use--; + } else if (rc == FW_FILTER_WR_FLT_ADDED) { + f->smtidx = (be64toh(rpl->oldval) >> 24) & 0xff; + f->pending = 0; /* asynchronous setup completed */ + f->valid = 1; + } else { + /* + * Something went wrong. Issue a warning about the + * problem and clear everything out. + */ + device_printf(sc->dev, + "filter %u setup failed with error %u\n", idx, rc); + clear_filter(sc, f); + sc->tids.ftids_in_use--; + } + } +} + int t4_os_find_pci_capability(struct adapter *sc, int cap) { device_t dev; struct pci_devinfo *dinfo; pcicfgregs *cfg; uint32_t status; uint8_t ptr; dev = sc->dev; dinfo = device_get_ivars(dev); cfg = &dinfo->cfg; status = pci_read_config(dev, PCIR_STATUS, 2); if (!(status & PCIM_STATUS_CAPPRESENT)) return (0); switch (cfg->hdrtype & PCIM_HDRTYPE) { case 0: case 1: ptr = PCIR_CAP_PTR; break; case 2: ptr = PCIR_CAP_PTR_2; break; default: return (0); break; } ptr = pci_read_config(dev, ptr, 1); while (ptr != 0) { if (pci_read_config(dev, ptr + PCICAP_ID, 1) == cap) return (ptr); ptr = pci_read_config(dev, ptr + PCICAP_NEXTPTR, 1); } return (0); } int t4_os_pci_save_state(struct adapter *sc) { device_t dev; struct pci_devinfo *dinfo; dev = sc->dev; dinfo = device_get_ivars(dev); pci_cfg_save(dev, dinfo, 0); return (0); } int t4_os_pci_restore_state(struct adapter *sc) { device_t dev; struct pci_devinfo *dinfo; dev = sc->dev; dinfo = device_get_ivars(dev); pci_cfg_restore(dev, dinfo); return (0); } void t4_os_portmod_changed(const struct adapter *sc, int idx) { struct port_info *pi = sc->port[idx]; static const char *mod_str[] = { NULL, "LR", "SR", "ER", "TWINAX", "active TWINAX", "LRM" }; if (pi->mod_type == FW_PORT_MOD_TYPE_NONE) if_printf(pi->ifp, "transceiver unplugged.\n"); else if (pi->mod_type == FW_PORT_MOD_TYPE_UNKNOWN) if_printf(pi->ifp, "unknown transceiver inserted.\n"); else if (pi->mod_type == FW_PORT_MOD_TYPE_NOTSUPPORTED) if_printf(pi->ifp, "unsupported transceiver inserted.\n"); else if (pi->mod_type > 0 && pi->mod_type < ARRAY_SIZE(mod_str)) { if_printf(pi->ifp, "%s transceiver inserted.\n", mod_str[pi->mod_type]); } else { if_printf(pi->ifp, "transceiver (type %d) inserted.\n", pi->mod_type); } } void t4_os_link_changed(struct adapter *sc, int idx, int link_stat) { struct port_info *pi = sc->port[idx]; struct ifnet *ifp = pi->ifp; if (link_stat) { ifp->if_baudrate = IF_Mbps(pi->link_cfg.speed); if_link_state_change(ifp, LINK_STATE_UP); } else if_link_state_change(ifp, LINK_STATE_DOWN); } static int t4_open(struct cdev *dev, int flags, int type, struct thread *td) { return (0); } static int t4_close(struct cdev *dev, int flags, int type, struct thread *td) { return (0); } static int t4_ioctl(struct cdev *dev, unsigned long cmd, caddr_t data, int fflag, struct thread *td) { int rc; struct adapter *sc = dev->si_drv1; rc = priv_check(td, PRIV_DRIVER); if (rc != 0) return (rc); switch (cmd) { case CHELSIO_T4_GETREG: { struct t4_reg *edata = (struct t4_reg *)data; if ((edata->addr & 0x3) != 0 || edata->addr >= sc->mmio_len) return (EFAULT); if (edata->size == 4) edata->val = t4_read_reg(sc, edata->addr); else if (edata->size == 8) edata->val = t4_read_reg64(sc, edata->addr); else return (EINVAL); break; } case CHELSIO_T4_SETREG: { struct t4_reg *edata = (struct t4_reg *)data; if ((edata->addr & 0x3) != 0 || edata->addr >= sc->mmio_len) return (EFAULT); if (edata->size == 4) { if (edata->val & 0xffffffff00000000) return (EINVAL); t4_write_reg(sc, edata->addr, (uint32_t) edata->val); } else if (edata->size == 8) t4_write_reg64(sc, edata->addr, edata->val); else return (EINVAL); break; } case CHELSIO_T4_REGDUMP: { struct t4_regdump *regs = (struct t4_regdump *)data; int reglen = T4_REGDUMP_SIZE; uint8_t *buf; if (regs->len < reglen) { regs->len = reglen; /* hint to the caller */ return (ENOBUFS); } regs->len = reglen; buf = malloc(reglen, M_CXGBE, M_WAITOK | M_ZERO); t4_get_regs(sc, regs, buf); rc = copyout(buf, regs->data, reglen); free(buf, M_CXGBE); break; } + case CHELSIO_T4_GET_FILTER_MODE: + rc = get_filter_mode(sc, (uint32_t *)data); + break; + case CHELSIO_T4_SET_FILTER_MODE: + rc = set_filter_mode(sc, *(uint32_t *)data); + break; + case CHELSIO_T4_GET_FILTER: + ADAPTER_LOCK(sc); + rc = get_filter(sc, (struct t4_filter *)data); + ADAPTER_UNLOCK(sc); + break; + case CHELSIO_T4_SET_FILTER: + ADAPTER_LOCK(sc); + rc = set_filter(sc, (struct t4_filter *)data); + ADAPTER_UNLOCK(sc); + break; + case CHELSIO_T4_DEL_FILTER: + ADAPTER_LOCK(sc); + rc = del_filter(sc, (struct t4_filter *)data); + ADAPTER_UNLOCK(sc); + break; default: rc = EINVAL; } return (rc); } static int t4_mod_event(module_t mod, int cmd, void *arg) { if (cmd == MOD_LOAD) t4_sge_modload(); return (0); } static devclass_t t4_devclass; static devclass_t cxgbe_devclass; DRIVER_MODULE(t4nex, pci, t4_driver, t4_devclass, t4_mod_event, 0); MODULE_VERSION(t4nex, 1); DRIVER_MODULE(cxgbe, t4nex, cxgbe_driver, cxgbe_devclass, 0, 0); MODULE_VERSION(cxgbe, 1); Index: projects/largeSMP/sys/dev/cxgbe/t4_sge.c =================================================================== --- projects/largeSMP/sys/dev/cxgbe/t4_sge.c (revision 221494) +++ projects/largeSMP/sys/dev/cxgbe/t4_sge.c (revision 221495) @@ -1,2716 +1,2735 @@ /*- * Copyright (c) 2011 Chelsio Communications, Inc. * All rights reserved. * Written by: Navdeep Parhar * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "common/common.h" #include "common/t4_regs.h" #include "common/t4_regs_values.h" #include "common/t4_msg.h" #include "common/t4fw_interface.h" struct fl_buf_info { int size; int type; uma_zone_t zone; }; /* Filled up by t4_sge_modload */ static struct fl_buf_info fl_buf_info[FL_BUF_SIZES]; #define FL_BUF_SIZE(x) (fl_buf_info[x].size) #define FL_BUF_TYPE(x) (fl_buf_info[x].type) #define FL_BUF_ZONE(x) (fl_buf_info[x].zone) enum { FL_PKTSHIFT = 2 }; #define FL_ALIGN min(CACHE_LINE_SIZE, 32) #if CACHE_LINE_SIZE > 64 #define SPG_LEN 128 #else #define SPG_LEN 64 #endif /* Used to track coalesced tx work request */ struct txpkts { uint64_t *flitp; /* ptr to flit where next pkt should start */ uint8_t npkt; /* # of packets in this work request */ uint8_t nflits; /* # of flits used by this work request */ uint16_t plen; /* total payload (sum of all packets) */ }; /* A packet's SGL. This + m_pkthdr has all info needed for tx */ struct sgl { int nsegs; /* # of segments in the SGL, 0 means imm. tx */ int nflits; /* # of flits needed for the SGL */ bus_dma_segment_t seg[TX_SGL_SEGS]; }; static inline void init_iq(struct sge_iq *, struct adapter *, int, int, int, int, iq_intr_handler_t *, char *); static inline void init_fl(struct sge_fl *, int, char *); static inline void init_eq(struct sge_eq *, int, char *); static int alloc_ring(struct adapter *, size_t, bus_dma_tag_t *, bus_dmamap_t *, bus_addr_t *, void **); static int free_ring(struct adapter *, bus_dma_tag_t, bus_dmamap_t, bus_addr_t, void *); static int alloc_iq_fl(struct port_info *, struct sge_iq *, struct sge_fl *, int); static int free_iq_fl(struct port_info *, struct sge_iq *, struct sge_fl *); static int alloc_iq(struct sge_iq *, int); static int free_iq(struct sge_iq *); static int alloc_rxq(struct port_info *, struct sge_rxq *, int, int); static int free_rxq(struct port_info *, struct sge_rxq *); static int alloc_ctrlq(struct adapter *, struct sge_ctrlq *, int); static int free_ctrlq(struct adapter *, struct sge_ctrlq *); static int alloc_txq(struct port_info *, struct sge_txq *, int); static int free_txq(struct port_info *, struct sge_txq *); static void oneseg_dma_callback(void *, bus_dma_segment_t *, int, int); static inline bool is_new_response(const struct sge_iq *, struct rsp_ctrl **); static inline void iq_next(struct sge_iq *); static inline void ring_fl_db(struct adapter *, struct sge_fl *); static void refill_fl(struct adapter *, struct sge_fl *, int, int); static int alloc_fl_sdesc(struct sge_fl *); static void free_fl_sdesc(struct sge_fl *); static int alloc_tx_maps(struct sge_txq *); static void free_tx_maps(struct sge_txq *); static void set_fl_tag_idx(struct sge_fl *, int); static int get_pkt_sgl(struct sge_txq *, struct mbuf **, struct sgl *, int); static int free_pkt_sgl(struct sge_txq *, struct sgl *); static int write_txpkt_wr(struct port_info *, struct sge_txq *, struct mbuf *, struct sgl *); static int add_to_txpkts(struct port_info *, struct sge_txq *, struct txpkts *, struct mbuf *, struct sgl *); static void write_txpkts_wr(struct sge_txq *, struct txpkts *); static inline void write_ulp_cpl_sgl(struct port_info *, struct sge_txq *, struct txpkts *, struct mbuf *, struct sgl *); static int write_sgl_to_txd(struct sge_eq *, struct sgl *, caddr_t *); static inline void copy_to_txd(struct sge_eq *, caddr_t, caddr_t *, int); static inline void ring_eq_db(struct adapter *, struct sge_eq *); static inline int reclaimable(struct sge_eq *); static int reclaim_tx_descs(struct sge_txq *, int, int); static void write_eqflush_wr(struct sge_eq *); static __be64 get_flit(bus_dma_segment_t *, int, int); static int handle_sge_egr_update(struct adapter *, const struct cpl_sge_egr_update *); static int ctrl_tx(struct adapter *, struct sge_ctrlq *, struct mbuf *); +extern void filter_rpl(struct adapter *, const struct cpl_set_tcb_rpl *); + /* * Called on MOD_LOAD and fills up fl_buf_info[]. */ void t4_sge_modload(void) { int i; int bufsize[FL_BUF_SIZES] = { MCLBYTES, #if MJUMPAGESIZE != MCLBYTES MJUMPAGESIZE, #endif MJUM9BYTES, MJUM16BYTES }; for (i = 0; i < FL_BUF_SIZES; i++) { FL_BUF_SIZE(i) = bufsize[i]; FL_BUF_TYPE(i) = m_gettype(bufsize[i]); FL_BUF_ZONE(i) = m_getzone(bufsize[i]); } } /** * t4_sge_init - initialize SGE * @sc: the adapter * * Performs SGE initialization needed every time after a chip reset. * We do not initialize any of the queues here, instead the driver * top-level must request them individually. */ void t4_sge_init(struct adapter *sc) { struct sge *s = &sc->sge; int i; t4_set_reg_field(sc, A_SGE_CONTROL, V_PKTSHIFT(M_PKTSHIFT) | V_INGPADBOUNDARY(M_INGPADBOUNDARY) | F_EGRSTATUSPAGESIZE, V_INGPADBOUNDARY(ilog2(FL_ALIGN) - 5) | V_PKTSHIFT(FL_PKTSHIFT) | F_RXPKTCPLMODE | V_EGRSTATUSPAGESIZE(SPG_LEN == 128)); t4_set_reg_field(sc, A_SGE_HOST_PAGE_SIZE, V_HOSTPAGESIZEPF0(M_HOSTPAGESIZEPF0), V_HOSTPAGESIZEPF0(PAGE_SHIFT - 10)); for (i = 0; i < FL_BUF_SIZES; i++) { t4_write_reg(sc, A_SGE_FL_BUFFER_SIZE0 + (4 * i), FL_BUF_SIZE(i)); } t4_write_reg(sc, A_SGE_INGRESS_RX_THRESHOLD, V_THRESHOLD_0(s->counter_val[0]) | V_THRESHOLD_1(s->counter_val[1]) | V_THRESHOLD_2(s->counter_val[2]) | V_THRESHOLD_3(s->counter_val[3])); t4_write_reg(sc, A_SGE_TIMER_VALUE_0_AND_1, V_TIMERVALUE0(us_to_core_ticks(sc, s->timer_val[0])) | V_TIMERVALUE1(us_to_core_ticks(sc, s->timer_val[1]))); t4_write_reg(sc, A_SGE_TIMER_VALUE_2_AND_3, V_TIMERVALUE2(us_to_core_ticks(sc, s->timer_val[2])) | V_TIMERVALUE3(us_to_core_ticks(sc, s->timer_val[3]))); t4_write_reg(sc, A_SGE_TIMER_VALUE_4_AND_5, V_TIMERVALUE4(us_to_core_ticks(sc, s->timer_val[4])) | V_TIMERVALUE5(us_to_core_ticks(sc, s->timer_val[5]))); } int t4_create_dma_tag(struct adapter *sc) { int rc; rc = bus_dma_tag_create(bus_get_dma_tag(sc->dev), 1, 0, BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL, BUS_SPACE_MAXSIZE, BUS_SPACE_UNRESTRICTED, BUS_SPACE_MAXSIZE, BUS_DMA_ALLOCNOW, NULL, NULL, &sc->dmat); if (rc != 0) { device_printf(sc->dev, "failed to create main DMA tag: %d\n", rc); } return (rc); } int t4_destroy_dma_tag(struct adapter *sc) { if (sc->dmat) bus_dma_tag_destroy(sc->dmat); return (0); } /* * Allocate and initialize the firmware event queue, control queues, and the * forwarded interrupt queues (if any). The adapter owns all these queues as * they are not associated with any particular port. * * Returns errno on failure. Resources allocated up to that point may still be * allocated. Caller is responsible for cleanup in case this function fails. */ int t4_setup_adapter_queues(struct adapter *sc) { int i, rc; struct sge_iq *iq, *fwq; struct sge_ctrlq *ctrlq; iq_intr_handler_t *handler; char name[16]; ADAPTER_LOCK_ASSERT_NOTOWNED(sc); if (sysctl_ctx_init(&sc->ctx) == 0) { struct sysctl_oid *oid = device_get_sysctl_tree(sc->dev); struct sysctl_oid_list *children = SYSCTL_CHILDREN(oid); sc->oid_ctrlq = SYSCTL_ADD_NODE(&sc->ctx, children, OID_AUTO, "ctrlq", CTLFLAG_RD, NULL, "ctrl queues"); } fwq = &sc->sge.fwq; if (sc->flags & INTR_FWD) { iq = &sc->sge.fiq[0]; /* * Forwarded interrupt queues - allocate 1 if there's only 1 * vector available, one less than the number of vectors * otherwise (the first vector is reserved for the error * interrupt in that case). */ i = sc->intr_count > 1 ? 1 : 0; for (; i < sc->intr_count; i++, iq++) { snprintf(name, sizeof(name), "%s fiq%d", device_get_nameunit(sc->dev), i); init_iq(iq, sc, 0, 0, (sc->sge.nrxq + 1) * 2, 16, NULL, name); rc = alloc_iq(iq, i); if (rc != 0) { device_printf(sc->dev, "failed to create fwd intr queue %d: %d\n", i, rc); return (rc); } } handler = t4_evt_rx; i = 0; /* forward fwq's interrupt to the first fiq */ } else { handler = NULL; i = 1; /* fwq should use vector 1 (0 is used by error) */ } snprintf(name, sizeof(name), "%s fwq", device_get_nameunit(sc->dev)); init_iq(fwq, sc, 0, 0, FW_IQ_QSIZE, FW_IQ_ESIZE, handler, name); rc = alloc_iq(fwq, i); if (rc != 0) { device_printf(sc->dev, "failed to create firmware event queue: %d\n", rc); return (rc); } /* * Control queues - one per hardware channel. */ ctrlq = &sc->sge.ctrlq[0]; for (i = 0; i < NCHAN; i++, ctrlq++) { snprintf(name, sizeof(name), "%s ctrlq%d", device_get_nameunit(sc->dev), i); init_eq(&ctrlq->eq, CTRL_EQ_QSIZE, name); rc = alloc_ctrlq(sc, ctrlq, i); if (rc != 0) { device_printf(sc->dev, "failed to create control queue %d: %d\n", i, rc); return (rc); } } return (rc); } /* * Idempotent */ int t4_teardown_adapter_queues(struct adapter *sc) { int i; struct sge_iq *iq; ADAPTER_LOCK_ASSERT_NOTOWNED(sc); /* Do this before freeing the queues */ if (sc->oid_ctrlq) { sysctl_ctx_free(&sc->ctx); sc->oid_ctrlq = NULL; } for (i = 0; i < NCHAN; i++) free_ctrlq(sc, &sc->sge.ctrlq[i]); iq = &sc->sge.fwq; free_iq(iq); if (sc->flags & INTR_FWD) { for (i = 0; i < NFIQ(sc); i++) { iq = &sc->sge.fiq[i]; free_iq(iq); } } return (0); } int t4_setup_eth_queues(struct port_info *pi) { int rc = 0, i, intr_idx; struct sge_rxq *rxq; struct sge_txq *txq; char name[16]; struct adapter *sc = pi->adapter; if (sysctl_ctx_init(&pi->ctx) == 0) { struct sysctl_oid *oid = device_get_sysctl_tree(pi->dev); struct sysctl_oid_list *children = SYSCTL_CHILDREN(oid); pi->oid_rxq = SYSCTL_ADD_NODE(&pi->ctx, children, OID_AUTO, "rxq", CTLFLAG_RD, NULL, "rx queues"); pi->oid_txq = SYSCTL_ADD_NODE(&pi->ctx, children, OID_AUTO, "txq", CTLFLAG_RD, NULL, "tx queues"); } for_each_rxq(pi, i, rxq) { snprintf(name, sizeof(name), "%s rxq%d-iq", device_get_nameunit(pi->dev), i); init_iq(&rxq->iq, sc, pi->tmr_idx, pi->pktc_idx, pi->qsize_rxq, RX_IQ_ESIZE, sc->flags & INTR_FWD ? t4_eth_rx : NULL, name); snprintf(name, sizeof(name), "%s rxq%d-fl", device_get_nameunit(pi->dev), i); init_fl(&rxq->fl, pi->qsize_rxq / 8, name); if (sc->flags & INTR_FWD) intr_idx = (pi->first_rxq + i) % NFIQ(sc); else intr_idx = pi->first_rxq + i + 2; rc = alloc_rxq(pi, rxq, intr_idx, i); if (rc != 0) goto done; intr_idx++; } for_each_txq(pi, i, txq) { snprintf(name, sizeof(name), "%s txq%d", device_get_nameunit(pi->dev), i); init_eq(&txq->eq, pi->qsize_txq, name); rc = alloc_txq(pi, txq, i); if (rc != 0) goto done; } done: if (rc) t4_teardown_eth_queues(pi); return (rc); } /* * Idempotent */ int t4_teardown_eth_queues(struct port_info *pi) { int i; struct sge_rxq *rxq; struct sge_txq *txq; /* Do this before freeing the queues */ if (pi->oid_txq || pi->oid_rxq) { sysctl_ctx_free(&pi->ctx); pi->oid_txq = pi->oid_rxq = NULL; } for_each_txq(pi, i, txq) { free_txq(pi, txq); } for_each_rxq(pi, i, rxq) { free_rxq(pi, rxq); } return (0); } /* Deals with errors and forwarded interrupts */ void t4_intr_all(void *arg) { struct adapter *sc = arg; t4_intr_err(arg); t4_intr_fwd(&sc->sge.fiq[0]); } /* Deals with forwarded interrupts on the given ingress queue */ void t4_intr_fwd(void *arg) { struct sge_iq *iq = arg, *q; struct adapter *sc = iq->adapter; struct rsp_ctrl *ctrl; int ndesc_pending = 0, ndesc_total = 0; int qid; if (!atomic_cmpset_32(&iq->state, IQS_IDLE, IQS_BUSY)) return; while (is_new_response(iq, &ctrl)) { rmb(); /* Only interrupt muxing expected on this queue */ KASSERT(G_RSPD_TYPE(ctrl->u.type_gen) == X_RSPD_TYPE_INTR, ("unexpected event on forwarded interrupt queue: %x", G_RSPD_TYPE(ctrl->u.type_gen))); qid = ntohl(ctrl->pldbuflen_qid) - sc->sge.iq_start; q = sc->sge.iqmap[qid]; q->handler(q); ndesc_total++; if (++ndesc_pending >= iq->qsize / 4) { t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_CIDXINC(ndesc_pending) | V_INGRESSQID(iq->cntxt_id) | V_SEINTARM( V_QINTR_TIMER_IDX(X_TIMERREG_UPDATE_CIDX))); ndesc_pending = 0; } iq_next(iq); } - if (ndesc_total > 0) { - t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), - V_CIDXINC(ndesc_pending) | V_INGRESSQID((u32)iq->cntxt_id) | - V_SEINTARM(iq->intr_params)); - } + t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_CIDXINC(ndesc_pending) | + V_INGRESSQID((u32)iq->cntxt_id) | V_SEINTARM(iq->intr_params)); atomic_cmpset_32(&iq->state, IQS_BUSY, IQS_IDLE); } /* Deals with error interrupts */ void t4_intr_err(void *arg) { struct adapter *sc = arg; if (sc->intr_type == INTR_INTX) t4_write_reg(sc, MYPF_REG(A_PCIE_PF_CLI), 0); t4_slow_intr_handler(sc); } /* Deals with the firmware event queue */ void t4_intr_evt(void *arg) { struct sge_iq *iq = arg; if (!atomic_cmpset_32(&iq->state, IQS_IDLE, IQS_BUSY)) return; t4_evt_rx(arg); atomic_cmpset_32(&iq->state, IQS_BUSY, IQS_IDLE); } void t4_intr_data(void *arg) { struct sge_iq *iq = arg; if (!atomic_cmpset_32(&iq->state, IQS_IDLE, IQS_BUSY)) return; t4_eth_rx(arg); atomic_cmpset_32(&iq->state, IQS_BUSY, IQS_IDLE); } void t4_evt_rx(void *arg) { struct sge_iq *iq = arg; struct adapter *sc = iq->adapter; struct rsp_ctrl *ctrl; const struct rss_header *rss; int ndesc_pending = 0, ndesc_total = 0; KASSERT(iq == &sc->sge.fwq, ("%s: unexpected ingress queue", __func__)); while (is_new_response(iq, &ctrl)) { rmb(); rss = (const void *)iq->cdesc; /* Should only get CPL on this queue */ KASSERT(G_RSPD_TYPE(ctrl->u.type_gen) == X_RSPD_TYPE_CPL, ("%s: unexpected type %d", __func__, G_RSPD_TYPE(ctrl->u.type_gen))); switch (rss->opcode) { case CPL_FW4_MSG: case CPL_FW6_MSG: { const struct cpl_fw6_msg *cpl; cpl = (const void *)(rss + 1); if (cpl->type == FW6_TYPE_CMD_RPL) t4_handle_fw_rpl(sc, cpl->data); break; } case CPL_SGE_EGR_UPDATE: handle_sge_egr_update(sc, (const void *)(rss + 1)); break; - + case CPL_SET_TCB_RPL: + filter_rpl(sc, (const void *) (rss + 1)); + break; default: device_printf(sc->dev, "can't handle CPL opcode %d.", rss->opcode); } ndesc_total++; if (++ndesc_pending >= iq->qsize / 4) { t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_CIDXINC(ndesc_pending) | V_INGRESSQID(iq->cntxt_id) | V_SEINTARM( V_QINTR_TIMER_IDX(X_TIMERREG_UPDATE_CIDX))); ndesc_pending = 0; } iq_next(iq); } - if (ndesc_total > 0) { - t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), - V_CIDXINC(ndesc_pending) | V_INGRESSQID(iq->cntxt_id) | - V_SEINTARM(iq->intr_params)); - } + t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_CIDXINC(ndesc_pending) | + V_INGRESSQID(iq->cntxt_id) | V_SEINTARM(iq->intr_params)); } +#ifdef T4_PKT_TIMESTAMP +#define RX_COPY_THRESHOLD (MINCLSIZE - 8) +#else +#define RX_COPY_THRESHOLD MINCLSIZE +#endif + void t4_eth_rx(void *arg) { struct sge_rxq *rxq = arg; struct sge_iq *iq = arg; struct adapter *sc = iq->adapter; struct rsp_ctrl *ctrl; struct ifnet *ifp = rxq->ifp; struct sge_fl *fl = &rxq->fl; struct fl_sdesc *sd = &fl->sdesc[fl->cidx], *sd_next; const struct rss_header *rss; const struct cpl_rx_pkt *cpl; uint32_t len; int ndescs = 0, i; struct mbuf *m0, *m; #ifdef INET struct lro_ctrl *lro = &rxq->lro; struct lro_entry *l; #endif prefetch(sd->m); prefetch(sd->cl); iq->intr_next = iq->intr_params; while (is_new_response(iq, &ctrl)) { rmb(); rss = (const void *)iq->cdesc; i = G_RSPD_TYPE(ctrl->u.type_gen); if (__predict_false(i == X_RSPD_TYPE_CPL)) { /* Can't be anything except an egress update */ KASSERT(rss->opcode == CPL_SGE_EGR_UPDATE, ("%s: unexpected CPL %x", __func__, rss->opcode)); handle_sge_egr_update(sc, (const void *)(rss + 1)); goto nextdesc; } KASSERT(i == X_RSPD_TYPE_FLBUF && rss->opcode == CPL_RX_PKT, ("%s: unexpected CPL %x rsp %d", __func__, rss->opcode, i)); sd_next = sd + 1; if (__predict_false(fl->cidx + 1 == fl->cap)) sd_next = fl->sdesc; prefetch(sd_next->m); prefetch(sd_next->cl); cpl = (const void *)(rss + 1); m0 = sd->m; sd->m = NULL; /* consumed */ len = be32toh(ctrl->pldbuflen_qid); if (__predict_false((len & F_RSPD_NEWBUF) == 0)) panic("%s: cannot handle packed frames", __func__); len = G_RSPD_LEN(len); bus_dmamap_sync(fl->tag[sd->tag_idx], sd->map, BUS_DMASYNC_POSTREAD); m_init(m0, NULL, 0, M_NOWAIT, MT_DATA, M_PKTHDR); - if (len < MINCLSIZE) { + +#ifdef T4_PKT_TIMESTAMP + *mtod(m0, uint64_t *) = + be64toh(ctrl->u.last_flit & 0xfffffffffffffff); + m0->m_data += 8; + + /* + * 60 bit timestamp value is *(uint64_t *)m0->m_pktdat. Note + * that it is in the leading free-space (see M_LEADINGSPACE) in + * the mbuf. The kernel can clobber it during a pullup, + * m_copymdata, etc. You need to make sure that the mbuf + * reaches you unmolested if you care about the timestamp. + */ +#endif + + if (len < RX_COPY_THRESHOLD) { /* copy data to mbuf, buffer will be recycled */ bcopy(sd->cl, mtod(m0, caddr_t), len); m0->m_len = len; } else { bus_dmamap_unload(fl->tag[sd->tag_idx], sd->map); m_cljset(m0, sd->cl, FL_BUF_TYPE(sd->tag_idx)); sd->cl = NULL; /* consumed */ m0->m_len = min(len, FL_BUF_SIZE(sd->tag_idx)); } len -= FL_PKTSHIFT; m0->m_len -= FL_PKTSHIFT; m0->m_data += FL_PKTSHIFT; m0->m_pkthdr.len = len; m0->m_pkthdr.rcvif = ifp; m0->m_flags |= M_FLOWID; m0->m_pkthdr.flowid = rss->hash_val; if (cpl->csum_calc && !cpl->err_vec && ifp->if_capenable & IFCAP_RXCSUM) { m0->m_pkthdr.csum_flags |= (CSUM_IP_CHECKED | CSUM_IP_VALID | CSUM_DATA_VALID | CSUM_PSEUDO_HDR); if (cpl->ip_frag) m0->m_pkthdr.csum_data = be16toh(cpl->csum); else m0->m_pkthdr.csum_data = 0xffff; rxq->rxcsum++; } if (cpl->vlan_ex) { m0->m_pkthdr.ether_vtag = be16toh(cpl->vlan); m0->m_flags |= M_VLANTAG; rxq->vlan_extraction++; } i = 1; /* # of fl sdesc used */ sd = sd_next; if (__predict_false(++fl->cidx == fl->cap)) fl->cidx = 0; len -= m0->m_len; m = m0; while (len) { i++; sd_next = sd + 1; if (__predict_false(fl->cidx + 1 == fl->cap)) sd_next = fl->sdesc; prefetch(sd_next->m); prefetch(sd_next->cl); m->m_next = sd->m; sd->m = NULL; /* consumed */ m = m->m_next; bus_dmamap_sync(fl->tag[sd->tag_idx], sd->map, BUS_DMASYNC_POSTREAD); m_init(m, NULL, 0, M_NOWAIT, MT_DATA, 0); if (len <= MLEN) { bcopy(sd->cl, mtod(m, caddr_t), len); m->m_len = len; } else { bus_dmamap_unload(fl->tag[sd->tag_idx], sd->map); m_cljset(m, sd->cl, FL_BUF_TYPE(sd->tag_idx)); sd->cl = NULL; /* consumed */ m->m_len = min(len, FL_BUF_SIZE(sd->tag_idx)); } i++; sd = sd_next; if (__predict_false(++fl->cidx == fl->cap)) fl->cidx = 0; len -= m->m_len; } #ifdef INET if (cpl->l2info & htobe32(F_RXF_LRO) && rxq->flags & RXQ_LRO_ENABLED && tcp_lro_rx(lro, m0, 0) == 0) { /* queued for LRO */ } else #endif ifp->if_input(ifp, m0); FL_LOCK(fl); fl->needed += i; if (fl->needed >= 32) refill_fl(sc, fl, 64, 32); FL_UNLOCK(fl); nextdesc: ndescs++; iq_next(iq); if (ndescs > 32) { t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_CIDXINC(ndescs) | V_INGRESSQID((u32)iq->cntxt_id) | V_SEINTARM(V_QINTR_TIMER_IDX(X_TIMERREG_UPDATE_CIDX))); ndescs = 0; } } #ifdef INET while (!SLIST_EMPTY(&lro->lro_active)) { l = SLIST_FIRST(&lro->lro_active); SLIST_REMOVE_HEAD(&lro->lro_active, next); tcp_lro_flush(lro, l); } #endif t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_CIDXINC(ndescs) | V_INGRESSQID((u32)iq->cntxt_id) | V_SEINTARM(iq->intr_next)); FL_LOCK(fl); if (fl->needed >= 32) refill_fl(sc, fl, 128, 8); FL_UNLOCK(fl); } int t4_mgmt_tx(struct adapter *sc, struct mbuf *m) { return ctrl_tx(sc, &sc->sge.ctrlq[0], m); } /* Per-packet header in a coalesced tx WR, before the SGL starts (in flits) */ #define TXPKTS_PKT_HDR ((\ sizeof(struct ulp_txpkt) + \ sizeof(struct ulptx_idata) + \ sizeof(struct cpl_tx_pkt_core) \ ) / 8) /* Header of a coalesced tx WR, before SGL of first packet (in flits) */ #define TXPKTS_WR_HDR (\ sizeof(struct fw_eth_tx_pkts_wr) / 8 + \ TXPKTS_PKT_HDR) /* Header of a tx WR, before SGL of first packet (in flits) */ #define TXPKT_WR_HDR ((\ sizeof(struct fw_eth_tx_pkt_wr) + \ sizeof(struct cpl_tx_pkt_core) \ ) / 8 ) /* Header of a tx LSO WR, before SGL of first packet (in flits) */ #define TXPKT_LSO_WR_HDR ((\ sizeof(struct fw_eth_tx_pkt_wr) + \ sizeof(struct cpl_tx_pkt_lso) + \ sizeof(struct cpl_tx_pkt_core) \ ) / 8 ) int t4_eth_tx(struct ifnet *ifp, struct sge_txq *txq, struct mbuf *m) { struct port_info *pi = (void *)ifp->if_softc; struct adapter *sc = pi->adapter; struct sge_eq *eq = &txq->eq; struct buf_ring *br = txq->br; struct mbuf *next; int rc, coalescing, can_reclaim; struct txpkts txpkts; struct sgl sgl; TXQ_LOCK_ASSERT_OWNED(txq); KASSERT(m, ("%s: called with nothing to do.", __func__)); prefetch(&eq->desc[eq->pidx]); prefetch(&txq->sdesc[eq->pidx]); txpkts.npkt = 0;/* indicates there's nothing in txpkts */ coalescing = 0; if (eq->avail < 8) reclaim_tx_descs(txq, 0, 8); for (; m; m = next ? next : drbr_dequeue(ifp, br)) { if (eq->avail < 8) break; next = m->m_nextpkt; m->m_nextpkt = NULL; if (next || buf_ring_peek(br)) coalescing = 1; rc = get_pkt_sgl(txq, &m, &sgl, coalescing); if (rc != 0) { if (rc == ENOMEM) { /* Short of resources, suspend tx */ m->m_nextpkt = next; break; } /* * Unrecoverable error for this packet, throw it away * and move on to the next. get_pkt_sgl may already * have freed m (it will be NULL in that case and the * m_freem here is still safe). */ m_freem(m); continue; } if (coalescing && add_to_txpkts(pi, txq, &txpkts, m, &sgl) == 0) { /* Successfully absorbed into txpkts */ write_ulp_cpl_sgl(pi, txq, &txpkts, m, &sgl); goto doorbell; } /* * We weren't coalescing to begin with, or current frame could * not be coalesced (add_to_txpkts flushes txpkts if a frame * given to it can't be coalesced). Either way there should be * nothing in txpkts. */ KASSERT(txpkts.npkt == 0, ("%s: txpkts not empty: %d", __func__, txpkts.npkt)); /* We're sending out individual packets now */ coalescing = 0; if (eq->avail < 8) reclaim_tx_descs(txq, 0, 8); rc = write_txpkt_wr(pi, txq, m, &sgl); if (rc != 0) { /* Short of hardware descriptors, suspend tx */ /* * This is an unlikely but expensive failure. We've * done all the hard work (DMA mappings etc.) and now we * can't send out the packet. What's worse, we have to * spend even more time freeing up everything in sgl. */ txq->no_desc++; free_pkt_sgl(txq, &sgl); m->m_nextpkt = next; break; } ETHER_BPF_MTAP(ifp, m); if (sgl.nsegs == 0) m_freem(m); doorbell: /* Fewer and fewer doorbells as the queue fills up */ if (eq->pending >= (1 << (fls(eq->qsize - eq->avail) / 2))) ring_eq_db(sc, eq); can_reclaim = reclaimable(eq); if (can_reclaim >= 32) reclaim_tx_descs(txq, can_reclaim, 32); } if (txpkts.npkt > 0) write_txpkts_wr(txq, &txpkts); /* * m not NULL means there was an error but we haven't thrown it away. * This can happen when we're short of tx descriptors (no_desc) or maybe * even DMA maps (no_dmamap). Either way, a credit flush and reclaim * will get things going again. * * If eq->avail is already 0 we know a credit flush was requested in the * WR that reduced it to 0 so we don't need another flush (we don't have * any descriptor for a flush WR anyway, duh). */ if (m && eq->avail > 0 && !(eq->flags & EQ_CRFLUSHED)) { struct tx_sdesc *txsd = &txq->sdesc[eq->pidx]; txsd->desc_used = 1; txsd->credits = 0; write_eqflush_wr(eq); } txq->m = m; if (eq->pending) ring_eq_db(sc, eq); can_reclaim = reclaimable(eq); if (can_reclaim >= 32) reclaim_tx_descs(txq, can_reclaim, 128); return (0); } void t4_update_fl_bufsize(struct ifnet *ifp) { struct port_info *pi = ifp->if_softc; struct sge_rxq *rxq; struct sge_fl *fl; int i; for_each_rxq(pi, i, rxq) { fl = &rxq->fl; FL_LOCK(fl); set_fl_tag_idx(fl, ifp->if_mtu); FL_UNLOCK(fl); } } /* * A non-NULL handler indicates this iq will not receive direct interrupts, the * handler will be invoked by a forwarded interrupt queue. */ static inline void init_iq(struct sge_iq *iq, struct adapter *sc, int tmr_idx, int pktc_idx, int qsize, int esize, iq_intr_handler_t *handler, char *name) { KASSERT(tmr_idx >= 0 && tmr_idx < SGE_NTIMERS, ("%s: bad tmr_idx %d", __func__, tmr_idx)); KASSERT(pktc_idx < SGE_NCOUNTERS, /* -ve is ok, means don't use */ ("%s: bad pktc_idx %d", __func__, pktc_idx)); iq->flags = 0; iq->adapter = sc; iq->intr_params = V_QINTR_TIMER_IDX(tmr_idx) | V_QINTR_CNT_EN(pktc_idx >= 0); iq->intr_pktc_idx = pktc_idx; iq->qsize = roundup(qsize, 16); /* See FW_IQ_CMD/iqsize */ iq->esize = max(esize, 16); /* See FW_IQ_CMD/iqesize */ iq->handler = handler; strlcpy(iq->lockname, name, sizeof(iq->lockname)); } static inline void init_fl(struct sge_fl *fl, int qsize, char *name) { fl->qsize = qsize; strlcpy(fl->lockname, name, sizeof(fl->lockname)); } static inline void init_eq(struct sge_eq *eq, int qsize, char *name) { eq->qsize = qsize; strlcpy(eq->lockname, name, sizeof(eq->lockname)); } static int alloc_ring(struct adapter *sc, size_t len, bus_dma_tag_t *tag, bus_dmamap_t *map, bus_addr_t *pa, void **va) { int rc; rc = bus_dma_tag_create(sc->dmat, 512, 0, BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL, len, 1, len, 0, NULL, NULL, tag); if (rc != 0) { device_printf(sc->dev, "cannot allocate DMA tag: %d\n", rc); goto done; } rc = bus_dmamem_alloc(*tag, va, BUS_DMA_WAITOK | BUS_DMA_COHERENT | BUS_DMA_ZERO, map); if (rc != 0) { device_printf(sc->dev, "cannot allocate DMA memory: %d\n", rc); goto done; } rc = bus_dmamap_load(*tag, *map, *va, len, oneseg_dma_callback, pa, 0); if (rc != 0) { device_printf(sc->dev, "cannot load DMA map: %d\n", rc); goto done; } done: if (rc) free_ring(sc, *tag, *map, *pa, *va); return (rc); } static int free_ring(struct adapter *sc, bus_dma_tag_t tag, bus_dmamap_t map, bus_addr_t pa, void *va) { if (pa) bus_dmamap_unload(tag, map); if (va) bus_dmamem_free(tag, va, map); if (tag) bus_dma_tag_destroy(tag); return (0); } /* * Allocates the ring for an ingress queue and an optional freelist. If the * freelist is specified it will be allocated and then associated with the * ingress queue. * * Returns errno on failure. Resources allocated up to that point may still be * allocated. Caller is responsible for cleanup in case this function fails. * * If the ingress queue will take interrupts directly (iq->handler == NULL) then * the intr_idx specifies the vector, starting from 0. Otherwise it specifies * the index of the queue to which its interrupts will be forwarded. */ static int alloc_iq_fl(struct port_info *pi, struct sge_iq *iq, struct sge_fl *fl, int intr_idx) { int rc, i, cntxt_id; size_t len; struct fw_iq_cmd c; struct adapter *sc = iq->adapter; __be32 v = 0; /* The adapter queues are nominally allocated in port[0]'s name */ if (pi == NULL) pi = sc->port[0]; len = iq->qsize * iq->esize; rc = alloc_ring(sc, len, &iq->desc_tag, &iq->desc_map, &iq->ba, (void **)&iq->desc); if (rc != 0) return (rc); bzero(&c, sizeof(c)); c.op_to_vfn = htobe32(V_FW_CMD_OP(FW_IQ_CMD) | F_FW_CMD_REQUEST | F_FW_CMD_WRITE | F_FW_CMD_EXEC | V_FW_IQ_CMD_PFN(sc->pf) | V_FW_IQ_CMD_VFN(0)); c.alloc_to_len16 = htobe32(F_FW_IQ_CMD_ALLOC | F_FW_IQ_CMD_IQSTART | FW_LEN16(c)); /* Special handling for firmware event queue */ if (iq == &sc->sge.fwq) v |= F_FW_IQ_CMD_IQASYNCH; if (iq->handler) { KASSERT(intr_idx < NFIQ(sc), ("%s: invalid indirect intr_idx %d", __func__, intr_idx)); v |= F_FW_IQ_CMD_IQANDST; v |= V_FW_IQ_CMD_IQANDSTINDEX(sc->sge.fiq[intr_idx].abs_id); } else { KASSERT(intr_idx < sc->intr_count, ("%s: invalid direct intr_idx %d", __func__, intr_idx)); v |= V_FW_IQ_CMD_IQANDSTINDEX(intr_idx); } c.type_to_iqandstindex = htobe32(v | V_FW_IQ_CMD_TYPE(FW_IQ_TYPE_FL_INT_CAP) | V_FW_IQ_CMD_VIID(pi->viid) | V_FW_IQ_CMD_IQANUD(X_UPDATEDELIVERY_INTERRUPT)); c.iqdroprss_to_iqesize = htobe16(V_FW_IQ_CMD_IQPCIECH(pi->tx_chan) | F_FW_IQ_CMD_IQGTSMODE | V_FW_IQ_CMD_IQINTCNTTHRESH(iq->intr_pktc_idx) | V_FW_IQ_CMD_IQESIZE(ilog2(iq->esize) - 4)); c.iqsize = htobe16(iq->qsize); c.iqaddr = htobe64(iq->ba); if (fl) { mtx_init(&fl->fl_lock, fl->lockname, NULL, MTX_DEF); for (i = 0; i < FL_BUF_SIZES; i++) { /* * A freelist buffer must be 16 byte aligned as the SGE * uses the low 4 bits of the bus addr to figure out the * buffer size. */ rc = bus_dma_tag_create(sc->dmat, 16, 0, BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL, FL_BUF_SIZE(i), 1, FL_BUF_SIZE(i), BUS_DMA_ALLOCNOW, NULL, NULL, &fl->tag[i]); if (rc != 0) { device_printf(sc->dev, "failed to create fl DMA tag[%d]: %d\n", i, rc); return (rc); } } len = fl->qsize * RX_FL_ESIZE; rc = alloc_ring(sc, len, &fl->desc_tag, &fl->desc_map, &fl->ba, (void **)&fl->desc); if (rc) return (rc); /* Allocate space for one software descriptor per buffer. */ fl->cap = (fl->qsize - SPG_LEN / RX_FL_ESIZE) * 8; FL_LOCK(fl); set_fl_tag_idx(fl, pi->ifp->if_mtu); rc = alloc_fl_sdesc(fl); FL_UNLOCK(fl); if (rc != 0) { device_printf(sc->dev, "failed to setup fl software descriptors: %d\n", rc); return (rc); } fl->needed = fl->cap; c.iqns_to_fl0congen = htobe32(V_FW_IQ_CMD_FL0HOSTFCMODE(X_HOSTFCMODE_NONE)); c.fl0dcaen_to_fl0cidxfthresh = htobe16(V_FW_IQ_CMD_FL0FBMIN(X_FETCHBURSTMIN_64B) | V_FW_IQ_CMD_FL0FBMAX(X_FETCHBURSTMAX_512B)); c.fl0size = htobe16(fl->qsize); c.fl0addr = htobe64(fl->ba); } rc = -t4_wr_mbox(sc, sc->mbox, &c, sizeof(c), &c); if (rc != 0) { device_printf(sc->dev, "failed to create ingress queue: %d\n", rc); return (rc); } iq->cdesc = iq->desc; iq->cidx = 0; iq->gen = 1; iq->intr_next = iq->intr_params; iq->cntxt_id = be16toh(c.iqid); iq->abs_id = be16toh(c.physiqid); iq->flags |= (IQ_ALLOCATED | IQ_STARTED); cntxt_id = iq->cntxt_id - sc->sge.iq_start; KASSERT(cntxt_id < sc->sge.niq, ("%s: iq->cntxt_id (%d) more than the max (%d)", __func__, cntxt_id, sc->sge.niq - 1)); sc->sge.iqmap[cntxt_id] = iq; if (fl) { fl->cntxt_id = be16toh(c.fl0id); fl->pidx = fl->cidx = 0; cntxt_id = fl->cntxt_id - sc->sge.eq_start; KASSERT(cntxt_id < sc->sge.neq, ("%s: fl->cntxt_id (%d) more than the max (%d)", __func__, cntxt_id, sc->sge.neq - 1)); sc->sge.eqmap[cntxt_id] = (void *)fl; FL_LOCK(fl); refill_fl(sc, fl, -1, 8); FL_UNLOCK(fl); } /* Enable IQ interrupts */ atomic_store_rel_32(&iq->state, IQS_IDLE); t4_write_reg(sc, MYPF_REG(A_SGE_PF_GTS), V_SEINTARM(iq->intr_params) | V_INGRESSQID(iq->cntxt_id)); return (0); } /* * This can be called with the iq/fl in any state - fully allocated and * functional, partially allocated, even all-zeroed out. */ static int free_iq_fl(struct port_info *pi, struct sge_iq *iq, struct sge_fl *fl) { int i, rc; struct adapter *sc = iq->adapter; device_t dev; if (sc == NULL) return (0); /* nothing to do */ dev = pi ? pi->dev : sc->dev; if (iq->flags & IQ_STARTED) { rc = -t4_iq_start_stop(sc, sc->mbox, 0, sc->pf, 0, iq->cntxt_id, fl ? fl->cntxt_id : 0xffff, 0xffff); if (rc != 0) { device_printf(dev, "failed to stop queue %p: %d\n", iq, rc); return (rc); } iq->flags &= ~IQ_STARTED; /* Synchronize with the interrupt handler */ while (!atomic_cmpset_32(&iq->state, IQS_IDLE, IQS_DISABLED)) pause("iqfree", hz / 1000); } if (iq->flags & IQ_ALLOCATED) { rc = -t4_iq_free(sc, sc->mbox, sc->pf, 0, FW_IQ_TYPE_FL_INT_CAP, iq->cntxt_id, fl ? fl->cntxt_id : 0xffff, 0xffff); if (rc != 0) { device_printf(dev, "failed to free queue %p: %d\n", iq, rc); return (rc); } iq->flags &= ~IQ_ALLOCATED; } free_ring(sc, iq->desc_tag, iq->desc_map, iq->ba, iq->desc); bzero(iq, sizeof(*iq)); if (fl) { free_ring(sc, fl->desc_tag, fl->desc_map, fl->ba, fl->desc); if (fl->sdesc) { FL_LOCK(fl); free_fl_sdesc(fl); FL_UNLOCK(fl); } if (mtx_initialized(&fl->fl_lock)) mtx_destroy(&fl->fl_lock); for (i = 0; i < FL_BUF_SIZES; i++) { if (fl->tag[i]) bus_dma_tag_destroy(fl->tag[i]); } bzero(fl, sizeof(*fl)); } return (0); } static int alloc_iq(struct sge_iq *iq, int intr_idx) { return alloc_iq_fl(NULL, iq, NULL, intr_idx); } static int free_iq(struct sge_iq *iq) { return free_iq_fl(NULL, iq, NULL); } static int alloc_rxq(struct port_info *pi, struct sge_rxq *rxq, int intr_idx, int idx) { int rc; struct sysctl_oid *oid; struct sysctl_oid_list *children; char name[16]; rc = alloc_iq_fl(pi, &rxq->iq, &rxq->fl, intr_idx); if (rc != 0) return (rc); #ifdef INET rc = tcp_lro_init(&rxq->lro); if (rc != 0) return (rc); rxq->lro.ifp = pi->ifp; /* also indicates LRO init'ed */ if (pi->ifp->if_capenable & IFCAP_LRO) rxq->flags |= RXQ_LRO_ENABLED; #endif rxq->ifp = pi->ifp; children = SYSCTL_CHILDREN(pi->oid_rxq); snprintf(name, sizeof(name), "%d", idx); oid = SYSCTL_ADD_NODE(&pi->ctx, children, OID_AUTO, name, CTLFLAG_RD, NULL, "rx queue"); children = SYSCTL_CHILDREN(oid); #ifdef INET SYSCTL_ADD_INT(&pi->ctx, children, OID_AUTO, "lro_queued", CTLFLAG_RD, &rxq->lro.lro_queued, 0, NULL); SYSCTL_ADD_INT(&pi->ctx, children, OID_AUTO, "lro_flushed", CTLFLAG_RD, &rxq->lro.lro_flushed, 0, NULL); #endif SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "rxcsum", CTLFLAG_RD, &rxq->rxcsum, "# of times hardware assisted with checksum"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "vlan_extraction", CTLFLAG_RD, &rxq->vlan_extraction, "# of times hardware extracted 802.1Q tag"); return (rc); } static int free_rxq(struct port_info *pi, struct sge_rxq *rxq) { int rc; #ifdef INET if (rxq->lro.ifp) { tcp_lro_free(&rxq->lro); rxq->lro.ifp = NULL; } #endif rc = free_iq_fl(pi, &rxq->iq, &rxq->fl); if (rc == 0) bzero(rxq, sizeof(*rxq)); return (rc); } static int alloc_ctrlq(struct adapter *sc, struct sge_ctrlq *ctrlq, int idx) { int rc, cntxt_id; size_t len; struct fw_eq_ctrl_cmd c; struct sge_eq *eq = &ctrlq->eq; char name[16]; struct sysctl_oid *oid; struct sysctl_oid_list *children; mtx_init(&eq->eq_lock, eq->lockname, NULL, MTX_DEF); len = eq->qsize * CTRL_EQ_ESIZE; rc = alloc_ring(sc, len, &eq->desc_tag, &eq->desc_map, &eq->ba, (void **)&eq->desc); if (rc) return (rc); eq->cap = eq->qsize - SPG_LEN / CTRL_EQ_ESIZE; eq->spg = (void *)&eq->desc[eq->cap]; eq->avail = eq->cap - 1; /* one less to avoid cidx = pidx */ eq->iqid = sc->sge.fwq.cntxt_id; bzero(&c, sizeof(c)); c.op_to_vfn = htobe32(V_FW_CMD_OP(FW_EQ_CTRL_CMD) | F_FW_CMD_REQUEST | F_FW_CMD_WRITE | F_FW_CMD_EXEC | V_FW_EQ_CTRL_CMD_PFN(sc->pf) | V_FW_EQ_CTRL_CMD_VFN(0)); c.alloc_to_len16 = htobe32(F_FW_EQ_CTRL_CMD_ALLOC | F_FW_EQ_CTRL_CMD_EQSTART | FW_LEN16(c)); c.cmpliqid_eqid = htonl(V_FW_EQ_CTRL_CMD_CMPLIQID(eq->iqid)); /* XXX */ c.physeqid_pkd = htobe32(0); c.fetchszm_to_iqid = htobe32(V_FW_EQ_CTRL_CMD_HOSTFCMODE(X_HOSTFCMODE_STATUS_PAGE) | V_FW_EQ_CTRL_CMD_PCIECHN(idx) | V_FW_EQ_CTRL_CMD_IQID(eq->iqid)); c.dcaen_to_eqsize = htobe32(V_FW_EQ_CTRL_CMD_FBMIN(X_FETCHBURSTMIN_64B) | V_FW_EQ_CTRL_CMD_FBMAX(X_FETCHBURSTMAX_512B) | V_FW_EQ_CTRL_CMD_CIDXFTHRESH(X_CIDXFLUSHTHRESH_32) | V_FW_EQ_CTRL_CMD_EQSIZE(eq->qsize)); c.eqaddr = htobe64(eq->ba); rc = -t4_wr_mbox(sc, sc->mbox, &c, sizeof(c), &c); if (rc != 0) { device_printf(sc->dev, "failed to create control queue %d: %d\n", idx, rc); return (rc); } eq->pidx = eq->cidx = 0; eq->cntxt_id = G_FW_EQ_CTRL_CMD_EQID(be32toh(c.cmpliqid_eqid)); eq->flags |= (EQ_ALLOCATED | EQ_STARTED); cntxt_id = eq->cntxt_id - sc->sge.eq_start; KASSERT(cntxt_id < sc->sge.neq, ("%s: eq->cntxt_id (%d) more than the max (%d)", __func__, cntxt_id, sc->sge.neq - 1)); sc->sge.eqmap[cntxt_id] = eq; children = SYSCTL_CHILDREN(sc->oid_ctrlq); snprintf(name, sizeof(name), "%d", idx); oid = SYSCTL_ADD_NODE(&sc->ctx, children, OID_AUTO, name, CTLFLAG_RD, NULL, "ctrl queue"); children = SYSCTL_CHILDREN(oid); SYSCTL_ADD_UQUAD(&sc->ctx, children, OID_AUTO, "total_wrs", CTLFLAG_RD, &ctrlq->total_wrs, "total # of work requests"); SYSCTL_ADD_UINT(&sc->ctx, children, OID_AUTO, "no_desc", CTLFLAG_RD, &ctrlq->no_desc, 0, "# of times ctrlq ran out of hardware descriptors"); SYSCTL_ADD_UINT(&sc->ctx, children, OID_AUTO, "too_long", CTLFLAG_RD, &ctrlq->too_long, 0, "# of oversized work requests"); return (rc); } static int free_ctrlq(struct adapter *sc, struct sge_ctrlq *ctrlq) { int rc; struct sge_eq *eq = &ctrlq->eq; if (eq->flags & (EQ_ALLOCATED | EQ_STARTED)) { rc = -t4_ctrl_eq_free(sc, sc->mbox, sc->pf, 0, eq->cntxt_id); if (rc != 0) { device_printf(sc->dev, "failed to free ctrl queue %p: %d\n", eq, rc); return (rc); } eq->flags &= ~(EQ_ALLOCATED | EQ_STARTED); } free_ring(sc, eq->desc_tag, eq->desc_map, eq->ba, eq->desc); if (mtx_initialized(&eq->eq_lock)) mtx_destroy(&eq->eq_lock); bzero(ctrlq, sizeof(*ctrlq)); return (0); } static int alloc_txq(struct port_info *pi, struct sge_txq *txq, int idx) { int rc, cntxt_id; size_t len; struct adapter *sc = pi->adapter; struct fw_eq_eth_cmd c; struct sge_eq *eq = &txq->eq; char name[16]; struct sysctl_oid *oid; struct sysctl_oid_list *children; txq->ifp = pi->ifp; TASK_INIT(&txq->resume_tx, 0, cxgbe_txq_start, txq); mtx_init(&eq->eq_lock, eq->lockname, NULL, MTX_DEF); len = eq->qsize * TX_EQ_ESIZE; rc = alloc_ring(sc, len, &eq->desc_tag, &eq->desc_map, &eq->ba, (void **)&eq->desc); if (rc) return (rc); eq->cap = eq->qsize - SPG_LEN / TX_EQ_ESIZE; eq->spg = (void *)&eq->desc[eq->cap]; eq->avail = eq->cap - 1; /* one less to avoid cidx = pidx */ txq->sdesc = malloc(eq->cap * sizeof(struct tx_sdesc), M_CXGBE, M_ZERO | M_WAITOK); txq->br = buf_ring_alloc(eq->qsize, M_CXGBE, M_WAITOK, &eq->eq_lock); eq->iqid = sc->sge.rxq[pi->first_rxq].iq.cntxt_id; rc = bus_dma_tag_create(sc->dmat, 1, 0, BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL, 64 * 1024, TX_SGL_SEGS, BUS_SPACE_MAXSIZE, BUS_DMA_ALLOCNOW, NULL, NULL, &txq->tx_tag); if (rc != 0) { device_printf(sc->dev, "failed to create tx DMA tag: %d\n", rc); return (rc); } rc = alloc_tx_maps(txq); if (rc != 0) { device_printf(sc->dev, "failed to setup tx DMA maps: %d\n", rc); return (rc); } bzero(&c, sizeof(c)); c.op_to_vfn = htobe32(V_FW_CMD_OP(FW_EQ_ETH_CMD) | F_FW_CMD_REQUEST | F_FW_CMD_WRITE | F_FW_CMD_EXEC | V_FW_EQ_ETH_CMD_PFN(sc->pf) | V_FW_EQ_ETH_CMD_VFN(0)); c.alloc_to_len16 = htobe32(F_FW_EQ_ETH_CMD_ALLOC | F_FW_EQ_ETH_CMD_EQSTART | FW_LEN16(c)); c.viid_pkd = htobe32(V_FW_EQ_ETH_CMD_VIID(pi->viid)); c.fetchszm_to_iqid = htobe32(V_FW_EQ_ETH_CMD_HOSTFCMODE(X_HOSTFCMODE_STATUS_PAGE) | V_FW_EQ_ETH_CMD_PCIECHN(pi->tx_chan) | V_FW_EQ_ETH_CMD_IQID(eq->iqid)); c.dcaen_to_eqsize = htobe32(V_FW_EQ_ETH_CMD_FBMIN(X_FETCHBURSTMIN_64B) | V_FW_EQ_ETH_CMD_FBMAX(X_FETCHBURSTMAX_512B) | V_FW_EQ_ETH_CMD_CIDXFTHRESH(X_CIDXFLUSHTHRESH_32) | V_FW_EQ_ETH_CMD_EQSIZE(eq->qsize)); c.eqaddr = htobe64(eq->ba); rc = -t4_wr_mbox(sc, sc->mbox, &c, sizeof(c), &c); if (rc != 0) { device_printf(pi->dev, "failed to create egress queue: %d\n", rc); return (rc); } eq->pidx = eq->cidx = 0; eq->cntxt_id = G_FW_EQ_ETH_CMD_EQID(be32toh(c.eqid_pkd)); eq->flags |= (EQ_ALLOCATED | EQ_STARTED); cntxt_id = eq->cntxt_id - sc->sge.eq_start; KASSERT(cntxt_id < sc->sge.neq, ("%s: eq->cntxt_id (%d) more than the max (%d)", __func__, cntxt_id, sc->sge.neq - 1)); sc->sge.eqmap[cntxt_id] = eq; children = SYSCTL_CHILDREN(pi->oid_txq); snprintf(name, sizeof(name), "%d", idx); oid = SYSCTL_ADD_NODE(&pi->ctx, children, OID_AUTO, name, CTLFLAG_RD, NULL, "tx queue"); children = SYSCTL_CHILDREN(oid); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "txcsum", CTLFLAG_RD, &txq->txcsum, "# of times hardware assisted with checksum"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "vlan_insertion", CTLFLAG_RD, &txq->vlan_insertion, "# of times hardware inserted 802.1Q tag"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "tso_wrs", CTLFLAG_RD, &txq->tso_wrs, "# of IPv4 TSO work requests"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "imm_wrs", CTLFLAG_RD, &txq->imm_wrs, "# of work requests with immediate data"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "sgl_wrs", CTLFLAG_RD, &txq->sgl_wrs, "# of work requests with direct SGL"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "txpkt_wrs", CTLFLAG_RD, &txq->txpkt_wrs, "# of txpkt work requests (one pkt/WR)"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "txpkts_wrs", CTLFLAG_RD, &txq->txpkts_wrs, "# of txpkts work requests (multiple pkts/WR)"); SYSCTL_ADD_UQUAD(&pi->ctx, children, OID_AUTO, "txpkts_pkts", CTLFLAG_RD, &txq->txpkts_pkts, "# of frames tx'd using txpkts work requests"); SYSCTL_ADD_UINT(&pi->ctx, children, OID_AUTO, "no_dmamap", CTLFLAG_RD, &txq->no_dmamap, 0, "# of times txq ran out of DMA maps"); SYSCTL_ADD_UINT(&pi->ctx, children, OID_AUTO, "no_desc", CTLFLAG_RD, &txq->no_desc, 0, "# of times txq ran out of hardware descriptors"); SYSCTL_ADD_UINT(&pi->ctx, children, OID_AUTO, "egr_update", CTLFLAG_RD, &txq->egr_update, 0, "egress update notifications from the SGE"); return (rc); } static int free_txq(struct port_info *pi, struct sge_txq *txq) { int rc; struct adapter *sc = pi->adapter; struct sge_eq *eq = &txq->eq; if (eq->flags & (EQ_ALLOCATED | EQ_STARTED)) { /* * Wait for the response to a credit flush if there's one * pending. Clearing the flag tells handle_sge_egr_update or * cxgbe_txq_start (depending on how far the response has made * it) that they should ignore the response and wake up free_txq * instead. * * The interface has been marked down by the time we get here * (both IFF_UP and IFF_DRV_RUNNING cleared). qflush has * emptied the tx buf_rings and we know nothing new is being * queued for tx so we don't have to worry about a new credit * flush request. */ TXQ_LOCK(txq); if (eq->flags & EQ_CRFLUSHED) { eq->flags &= ~EQ_CRFLUSHED; msleep(txq, &eq->eq_lock, 0, "crflush", 0); } TXQ_UNLOCK(txq); rc = -t4_eth_eq_free(sc, sc->mbox, sc->pf, 0, eq->cntxt_id); if (rc != 0) { device_printf(pi->dev, "failed to free egress queue %p: %d\n", eq, rc); return (rc); } eq->flags &= ~(EQ_ALLOCATED | EQ_STARTED); } free_ring(sc, eq->desc_tag, eq->desc_map, eq->ba, eq->desc); free(txq->sdesc, M_CXGBE); if (txq->maps) free_tx_maps(txq); buf_ring_free(txq->br, M_CXGBE); if (txq->tx_tag) bus_dma_tag_destroy(txq->tx_tag); if (mtx_initialized(&eq->eq_lock)) mtx_destroy(&eq->eq_lock); bzero(txq, sizeof(*txq)); return (0); } static void oneseg_dma_callback(void *arg, bus_dma_segment_t *segs, int nseg, int error) { bus_addr_t *ba = arg; KASSERT(nseg == 1, ("%s meant for single segment mappings only.", __func__)); *ba = error ? 0 : segs->ds_addr; } static inline bool is_new_response(const struct sge_iq *iq, struct rsp_ctrl **ctrl) { *ctrl = (void *)((uintptr_t)iq->cdesc + (iq->esize - sizeof(struct rsp_ctrl))); return (((*ctrl)->u.type_gen >> S_RSPD_GEN) == iq->gen); } static inline void iq_next(struct sge_iq *iq) { iq->cdesc = (void *) ((uintptr_t)iq->cdesc + iq->esize); if (__predict_false(++iq->cidx == iq->qsize - 1)) { iq->cidx = 0; iq->gen ^= 1; iq->cdesc = iq->desc; } } #define FL_HW_IDX(x) ((x) >> 3) static inline void ring_fl_db(struct adapter *sc, struct sge_fl *fl) { int ndesc = fl->pending / 8; if (FL_HW_IDX(fl->pidx) == FL_HW_IDX(fl->cidx)) ndesc--; /* hold back one credit */ if (ndesc <= 0) return; /* nothing to do */ wmb(); t4_write_reg(sc, MYPF_REG(A_SGE_PF_KDOORBELL), F_DBPRIO | V_QID(fl->cntxt_id) | V_PIDX(ndesc)); fl->pending -= ndesc * 8; } /* * Fill up the freelist by upto nbufs and ring its doorbell if the number of * buffers ready to be handed to the hardware >= dbthresh. */ static void refill_fl(struct adapter *sc, struct sge_fl *fl, int nbufs, int dbthresh) { __be64 *d = &fl->desc[fl->pidx]; struct fl_sdesc *sd = &fl->sdesc[fl->pidx]; bus_dma_tag_t tag; bus_addr_t pa; caddr_t cl; int rc; FL_LOCK_ASSERT_OWNED(fl); if (nbufs < 0 || nbufs > fl->needed) nbufs = fl->needed; while (nbufs--) { if (sd->cl != NULL) { /* * This happens when a frame small enough to fit * entirely in an mbuf was received in cl last time. * We'd held on to cl and can reuse it now. Note that * we reuse a cluster of the old size if fl->tag_idx is * no longer the same as sd->tag_idx. */ KASSERT(*d == sd->ba_tag, ("%s: recyling problem at pidx %d", __func__, fl->pidx)); d++; goto recycled; } if (fl->tag_idx != sd->tag_idx) { bus_dmamap_t map; bus_dma_tag_t newtag = fl->tag[fl->tag_idx]; bus_dma_tag_t oldtag = fl->tag[sd->tag_idx]; /* * An MTU change can get us here. Discard the old map * which was created with the old tag, but only if * we're able to get a new one. */ rc = bus_dmamap_create(newtag, 0, &map); if (rc == 0) { bus_dmamap_destroy(oldtag, sd->map); sd->map = map; sd->tag_idx = fl->tag_idx; } } tag = fl->tag[sd->tag_idx]; cl = m_cljget(NULL, M_NOWAIT, FL_BUF_SIZE(sd->tag_idx)); if (cl == NULL) break; rc = bus_dmamap_load(tag, sd->map, cl, FL_BUF_SIZE(sd->tag_idx), oneseg_dma_callback, &pa, 0); if (rc != 0 || pa == 0) { fl->dmamap_failed++; uma_zfree(FL_BUF_ZONE(sd->tag_idx), cl); break; } sd->cl = cl; *d++ = htobe64(pa | sd->tag_idx); #ifdef INVARIANTS sd->ba_tag = htobe64(pa | sd->tag_idx); #endif recycled: /* sd->m is never recycled, should always be NULL */ KASSERT(sd->m == NULL, ("%s: stray mbuf", __func__)); sd->m = m_gethdr(M_NOWAIT, MT_NOINIT); if (sd->m == NULL) break; fl->pending++; fl->needed--; sd++; if (++fl->pidx == fl->cap) { fl->pidx = 0; sd = fl->sdesc; d = fl->desc; } } if (fl->pending >= dbthresh) ring_fl_db(sc, fl); } static int alloc_fl_sdesc(struct sge_fl *fl) { struct fl_sdesc *sd; bus_dma_tag_t tag; int i, rc; FL_LOCK_ASSERT_OWNED(fl); fl->sdesc = malloc(fl->cap * sizeof(struct fl_sdesc), M_CXGBE, M_ZERO | M_WAITOK); tag = fl->tag[fl->tag_idx]; sd = fl->sdesc; for (i = 0; i < fl->cap; i++, sd++) { sd->tag_idx = fl->tag_idx; rc = bus_dmamap_create(tag, 0, &sd->map); if (rc != 0) goto failed; } return (0); failed: while (--i >= 0) { sd--; bus_dmamap_destroy(tag, sd->map); if (sd->m) { m_init(sd->m, NULL, 0, M_NOWAIT, MT_DATA, 0); m_free(sd->m); sd->m = NULL; } } KASSERT(sd == fl->sdesc, ("%s: EDOOFUS", __func__)); free(fl->sdesc, M_CXGBE); fl->sdesc = NULL; return (rc); } static void free_fl_sdesc(struct sge_fl *fl) { struct fl_sdesc *sd; int i; FL_LOCK_ASSERT_OWNED(fl); sd = fl->sdesc; for (i = 0; i < fl->cap; i++, sd++) { if (sd->m) { m_init(sd->m, NULL, 0, M_NOWAIT, MT_DATA, 0); m_free(sd->m); sd->m = NULL; } if (sd->cl) { bus_dmamap_unload(fl->tag[sd->tag_idx], sd->map); uma_zfree(FL_BUF_ZONE(sd->tag_idx), sd->cl); sd->cl = NULL; } bus_dmamap_destroy(fl->tag[sd->tag_idx], sd->map); } free(fl->sdesc, M_CXGBE); fl->sdesc = NULL; } static int alloc_tx_maps(struct sge_txq *txq) { struct tx_map *txm; int i, rc, count; /* * We can stuff ~10 frames in an 8-descriptor txpkts WR (8 is the SGE * limit for any WR). txq->no_dmamap events shouldn't occur if maps is * sized for the worst case. */ count = txq->eq.qsize * 10 / 8; txq->map_total = txq->map_avail = count; txq->map_cidx = txq->map_pidx = 0; txq->maps = malloc(count * sizeof(struct tx_map), M_CXGBE, M_ZERO | M_WAITOK); txm = txq->maps; for (i = 0; i < count; i++, txm++) { rc = bus_dmamap_create(txq->tx_tag, 0, &txm->map); if (rc != 0) goto failed; } return (0); failed: while (--i >= 0) { txm--; bus_dmamap_destroy(txq->tx_tag, txm->map); } KASSERT(txm == txq->maps, ("%s: EDOOFUS", __func__)); free(txq->maps, M_CXGBE); txq->maps = NULL; return (rc); } static void free_tx_maps(struct sge_txq *txq) { struct tx_map *txm; int i; txm = txq->maps; for (i = 0; i < txq->map_total; i++, txm++) { if (txm->m) { bus_dmamap_unload(txq->tx_tag, txm->map); m_freem(txm->m); txm->m = NULL; } bus_dmamap_destroy(txq->tx_tag, txm->map); } free(txq->maps, M_CXGBE); txq->maps = NULL; } /* * We'll do immediate data tx for non-TSO, but only when not coalescing. We're * willing to use upto 2 hardware descriptors which means a maximum of 96 bytes * of immediate data. */ #define IMM_LEN ( \ 2 * TX_EQ_ESIZE \ - sizeof(struct fw_eth_tx_pkt_wr) \ - sizeof(struct cpl_tx_pkt_core)) /* * Returns non-zero on failure, no need to cleanup anything in that case. * * Note 1: We always try to defrag the mbuf if required and return EFBIG only * if the resulting chain still won't fit in a tx descriptor. * * Note 2: We'll pullup the mbuf chain if TSO is requested and the first mbuf * does not have the TCP header in it. */ static int get_pkt_sgl(struct sge_txq *txq, struct mbuf **fp, struct sgl *sgl, int sgl_only) { struct mbuf *m = *fp; struct tx_map *txm; int rc, defragged = 0, n; TXQ_LOCK_ASSERT_OWNED(txq); if (m->m_pkthdr.tso_segsz) sgl_only = 1; /* Do not allow immediate data with LSO */ start: sgl->nsegs = 0; if (m->m_pkthdr.len <= IMM_LEN && !sgl_only) return (0); /* nsegs = 0 tells caller to use imm. tx */ if (txq->map_avail == 0) { txq->no_dmamap++; return (ENOMEM); } txm = &txq->maps[txq->map_pidx]; if (m->m_pkthdr.tso_segsz && m->m_len < 50) { *fp = m_pullup(m, 50); m = *fp; if (m == NULL) return (ENOBUFS); } rc = bus_dmamap_load_mbuf_sg(txq->tx_tag, txm->map, m, sgl->seg, &sgl->nsegs, BUS_DMA_NOWAIT); if (rc == EFBIG && defragged == 0) { m = m_defrag(m, M_DONTWAIT); if (m == NULL) return (EFBIG); defragged = 1; *fp = m; goto start; } if (rc != 0) return (rc); txm->m = m; txq->map_avail--; if (++txq->map_pidx == txq->map_total) txq->map_pidx = 0; KASSERT(sgl->nsegs > 0 && sgl->nsegs <= TX_SGL_SEGS, ("%s: bad DMA mapping (%d segments)", __func__, sgl->nsegs)); /* * Store the # of flits required to hold this frame's SGL in nflits. An * SGL has a (ULPTX header + len0, addr0) tuple optionally followed by * multiple (len0 + len1, addr0, addr1) tuples. If addr1 is not used * then len1 must be set to 0. */ n = sgl->nsegs - 1; sgl->nflits = (3 * n) / 2 + (n & 1) + 2; return (0); } /* * Releases all the txq resources used up in the specified sgl. */ static int free_pkt_sgl(struct sge_txq *txq, struct sgl *sgl) { struct tx_map *txm; TXQ_LOCK_ASSERT_OWNED(txq); if (sgl->nsegs == 0) return (0); /* didn't use any map */ /* 1 pkt uses exactly 1 map, back it out */ txq->map_avail++; if (txq->map_pidx > 0) txq->map_pidx--; else txq->map_pidx = txq->map_total - 1; txm = &txq->maps[txq->map_pidx]; bus_dmamap_unload(txq->tx_tag, txm->map); txm->m = NULL; return (0); } static int write_txpkt_wr(struct port_info *pi, struct sge_txq *txq, struct mbuf *m, struct sgl *sgl) { struct sge_eq *eq = &txq->eq; struct fw_eth_tx_pkt_wr *wr; struct cpl_tx_pkt_core *cpl; uint32_t ctrl; /* used in many unrelated places */ uint64_t ctrl1; int nflits, ndesc, pktlen; struct tx_sdesc *txsd; caddr_t dst; TXQ_LOCK_ASSERT_OWNED(txq); pktlen = m->m_pkthdr.len; /* * Do we have enough flits to send this frame out? */ ctrl = sizeof(struct cpl_tx_pkt_core); if (m->m_pkthdr.tso_segsz) { nflits = TXPKT_LSO_WR_HDR; ctrl += sizeof(struct cpl_tx_pkt_lso); } else nflits = TXPKT_WR_HDR; if (sgl->nsegs > 0) nflits += sgl->nflits; else { nflits += howmany(pktlen, 8); ctrl += pktlen; } ndesc = howmany(nflits, 8); if (ndesc > eq->avail) return (ENOMEM); /* Firmware work request header */ wr = (void *)&eq->desc[eq->pidx]; wr->op_immdlen = htobe32(V_FW_WR_OP(FW_ETH_TX_PKT_WR) | V_FW_WR_IMMDLEN(ctrl)); ctrl = V_FW_WR_LEN16(howmany(nflits, 2)); if (eq->avail == ndesc && !(eq->flags & EQ_CRFLUSHED)) { ctrl |= F_FW_WR_EQUEQ | F_FW_WR_EQUIQ; eq->flags |= EQ_CRFLUSHED; } wr->equiq_to_len16 = htobe32(ctrl); wr->r3 = 0; if (m->m_pkthdr.tso_segsz) { struct cpl_tx_pkt_lso *lso = (void *)(wr + 1); struct ether_header *eh; struct ip *ip; struct tcphdr *tcp; ctrl = V_LSO_OPCODE(CPL_TX_PKT_LSO) | F_LSO_FIRST_SLICE | F_LSO_LAST_SLICE; eh = mtod(m, struct ether_header *); if (eh->ether_type == htons(ETHERTYPE_VLAN)) { ctrl |= V_LSO_ETHHDR_LEN(1); ip = (void *)((struct ether_vlan_header *)eh + 1); } else ip = (void *)(eh + 1); tcp = (void *)((uintptr_t)ip + ip->ip_hl * 4); ctrl |= V_LSO_IPHDR_LEN(ip->ip_hl) | V_LSO_TCPHDR_LEN(tcp->th_off); lso->lso_ctrl = htobe32(ctrl); lso->ipid_ofst = htobe16(0); lso->mss = htobe16(m->m_pkthdr.tso_segsz); lso->seqno_offset = htobe32(0); lso->len = htobe32(pktlen); cpl = (void *)(lso + 1); txq->tso_wrs++; } else cpl = (void *)(wr + 1); /* Checksum offload */ ctrl1 = 0; if (!(m->m_pkthdr.csum_flags & CSUM_IP)) ctrl1 |= F_TXPKT_IPCSUM_DIS; if (!(m->m_pkthdr.csum_flags & (CSUM_TCP | CSUM_UDP))) ctrl1 |= F_TXPKT_L4CSUM_DIS; if (m->m_pkthdr.csum_flags & (CSUM_IP | CSUM_TCP | CSUM_UDP)) txq->txcsum++; /* some hardware assistance provided */ /* VLAN tag insertion */ if (m->m_flags & M_VLANTAG) { ctrl1 |= F_TXPKT_VLAN_VLD | V_TXPKT_VLAN(m->m_pkthdr.ether_vtag); txq->vlan_insertion++; } /* CPL header */ cpl->ctrl0 = htobe32(V_TXPKT_OPCODE(CPL_TX_PKT) | V_TXPKT_INTF(pi->tx_chan) | V_TXPKT_PF(pi->adapter->pf)); cpl->pack = 0; cpl->len = htobe16(pktlen); cpl->ctrl1 = htobe64(ctrl1); /* Software descriptor */ txsd = &txq->sdesc[eq->pidx]; txsd->desc_used = ndesc; eq->pending += ndesc; eq->avail -= ndesc; eq->pidx += ndesc; if (eq->pidx >= eq->cap) eq->pidx -= eq->cap; /* SGL */ dst = (void *)(cpl + 1); if (sgl->nsegs > 0) { txsd->credits = 1; txq->sgl_wrs++; write_sgl_to_txd(eq, sgl, &dst); } else { txsd->credits = 0; txq->imm_wrs++; for (; m; m = m->m_next) { copy_to_txd(eq, mtod(m, caddr_t), &dst, m->m_len); #ifdef INVARIANTS pktlen -= m->m_len; #endif } #ifdef INVARIANTS KASSERT(pktlen == 0, ("%s: %d bytes left.", __func__, pktlen)); #endif } txq->txpkt_wrs++; return (0); } /* * Returns 0 to indicate that m has been accepted into a coalesced tx work * request. It has either been folded into txpkts or txpkts was flushed and m * has started a new coalesced work request (as the first frame in a fresh * txpkts). * * Returns non-zero to indicate a failure - caller is responsible for * transmitting m, if there was anything in txpkts it has been flushed. */ static int add_to_txpkts(struct port_info *pi, struct sge_txq *txq, struct txpkts *txpkts, struct mbuf *m, struct sgl *sgl) { struct sge_eq *eq = &txq->eq; int can_coalesce; struct tx_sdesc *txsd; int flits; TXQ_LOCK_ASSERT_OWNED(txq); if (txpkts->npkt > 0) { flits = TXPKTS_PKT_HDR + sgl->nflits; can_coalesce = m->m_pkthdr.tso_segsz == 0 && txpkts->nflits + flits <= TX_WR_FLITS && txpkts->nflits + flits <= eq->avail * 8 && txpkts->plen + m->m_pkthdr.len < 65536; if (can_coalesce) { txpkts->npkt++; txpkts->nflits += flits; txpkts->plen += m->m_pkthdr.len; txsd = &txq->sdesc[eq->pidx]; txsd->credits++; return (0); } /* * Couldn't coalesce m into txpkts. The first order of business * is to send txpkts on its way. Then we'll revisit m. */ write_txpkts_wr(txq, txpkts); } /* * Check if we can start a new coalesced tx work request with m as * the first packet in it. */ KASSERT(txpkts->npkt == 0, ("%s: txpkts not empty", __func__)); flits = TXPKTS_WR_HDR + sgl->nflits; can_coalesce = m->m_pkthdr.tso_segsz == 0 && flits <= eq->avail * 8 && flits <= TX_WR_FLITS; if (can_coalesce == 0) return (EINVAL); /* * Start a fresh coalesced tx WR with m as the first frame in it. */ txpkts->npkt = 1; txpkts->nflits = flits; txpkts->flitp = &eq->desc[eq->pidx].flit[2]; txpkts->plen = m->m_pkthdr.len; txsd = &txq->sdesc[eq->pidx]; txsd->credits = 1; return (0); } /* * Note that write_txpkts_wr can never run out of hardware descriptors (but * write_txpkt_wr can). add_to_txpkts ensures that a frame is accepted for * coalescing only if sufficient hardware descriptors are available. */ static void write_txpkts_wr(struct sge_txq *txq, struct txpkts *txpkts) { struct sge_eq *eq = &txq->eq; struct fw_eth_tx_pkts_wr *wr; struct tx_sdesc *txsd; uint32_t ctrl; int ndesc; TXQ_LOCK_ASSERT_OWNED(txq); ndesc = howmany(txpkts->nflits, 8); wr = (void *)&eq->desc[eq->pidx]; wr->op_immdlen = htobe32(V_FW_WR_OP(FW_ETH_TX_PKTS_WR) | V_FW_WR_IMMDLEN(0)); /* immdlen does not matter in this WR */ ctrl = V_FW_WR_LEN16(howmany(txpkts->nflits, 2)); if (eq->avail == ndesc && !(eq->flags & EQ_CRFLUSHED)) { ctrl |= F_FW_WR_EQUEQ | F_FW_WR_EQUIQ; eq->flags |= EQ_CRFLUSHED; } wr->equiq_to_len16 = htobe32(ctrl); wr->plen = htobe16(txpkts->plen); wr->npkt = txpkts->npkt; wr->r3 = wr->r4 = 0; /* Everything else already written */ txsd = &txq->sdesc[eq->pidx]; txsd->desc_used = ndesc; KASSERT(eq->avail >= ndesc, ("%s: out of descriptors", __func__)); eq->pending += ndesc; eq->avail -= ndesc; eq->pidx += ndesc; if (eq->pidx >= eq->cap) eq->pidx -= eq->cap; txq->txpkts_pkts += txpkts->npkt; txq->txpkts_wrs++; txpkts->npkt = 0; /* emptied */ } static inline void write_ulp_cpl_sgl(struct port_info *pi, struct sge_txq *txq, struct txpkts *txpkts, struct mbuf *m, struct sgl *sgl) { struct ulp_txpkt *ulpmc; struct ulptx_idata *ulpsc; struct cpl_tx_pkt_core *cpl; struct sge_eq *eq = &txq->eq; uintptr_t flitp, start, end; uint64_t ctrl; caddr_t dst; KASSERT(txpkts->npkt > 0, ("%s: txpkts is empty", __func__)); start = (uintptr_t)eq->desc; end = (uintptr_t)eq->spg; /* Checksum offload */ ctrl = 0; if (!(m->m_pkthdr.csum_flags & CSUM_IP)) ctrl |= F_TXPKT_IPCSUM_DIS; if (!(m->m_pkthdr.csum_flags & (CSUM_TCP | CSUM_UDP))) ctrl |= F_TXPKT_L4CSUM_DIS; if (m->m_pkthdr.csum_flags & (CSUM_IP | CSUM_TCP | CSUM_UDP)) txq->txcsum++; /* some hardware assistance provided */ /* VLAN tag insertion */ if (m->m_flags & M_VLANTAG) { ctrl |= F_TXPKT_VLAN_VLD | V_TXPKT_VLAN(m->m_pkthdr.ether_vtag); txq->vlan_insertion++; } /* * The previous packet's SGL must have ended at a 16 byte boundary (this * is required by the firmware/hardware). It follows that flitp cannot * wrap around between the ULPTX master command and ULPTX subcommand (8 * bytes each), and that it can not wrap around in the middle of the * cpl_tx_pkt_core either. */ flitp = (uintptr_t)txpkts->flitp; KASSERT((flitp & 0xf) == 0, ("%s: last SGL did not end at 16 byte boundary: %p", __func__, txpkts->flitp)); /* ULP master command */ ulpmc = (void *)flitp; ulpmc->cmd_dest = htonl(V_ULPTX_CMD(ULP_TX_PKT) | V_ULP_TXPKT_DEST(0) | V_ULP_TXPKT_FID(eq->iqid)); ulpmc->len = htonl(howmany(sizeof(*ulpmc) + sizeof(*ulpsc) + sizeof(*cpl) + 8 * sgl->nflits, 16)); /* ULP subcommand */ ulpsc = (void *)(ulpmc + 1); ulpsc->cmd_more = htobe32(V_ULPTX_CMD((u32)ULP_TX_SC_IMM) | F_ULP_TX_SC_MORE); ulpsc->len = htobe32(sizeof(struct cpl_tx_pkt_core)); flitp += sizeof(*ulpmc) + sizeof(*ulpsc); if (flitp == end) flitp = start; /* CPL_TX_PKT */ cpl = (void *)flitp; cpl->ctrl0 = htobe32(V_TXPKT_OPCODE(CPL_TX_PKT) | V_TXPKT_INTF(pi->tx_chan) | V_TXPKT_PF(pi->adapter->pf)); cpl->pack = 0; cpl->len = htobe16(m->m_pkthdr.len); cpl->ctrl1 = htobe64(ctrl); flitp += sizeof(*cpl); if (flitp == end) flitp = start; /* SGL for this frame */ dst = (caddr_t)flitp; txpkts->nflits += write_sgl_to_txd(eq, sgl, &dst); txpkts->flitp = (void *)dst; KASSERT(((uintptr_t)dst & 0xf) == 0, ("%s: SGL ends at %p (not a 16 byte boundary)", __func__, dst)); } /* * If the SGL ends on an address that is not 16 byte aligned, this function will * add a 0 filled flit at the end. It returns 1 in that case. */ static int write_sgl_to_txd(struct sge_eq *eq, struct sgl *sgl, caddr_t *to) { __be64 *flitp, *end; struct ulptx_sgl *usgl; bus_dma_segment_t *seg; int i, padded; KASSERT(sgl->nsegs > 0 && sgl->nflits > 0, ("%s: bad SGL - nsegs=%d, nflits=%d", __func__, sgl->nsegs, sgl->nflits)); KASSERT(((uintptr_t)(*to) & 0xf) == 0, ("%s: SGL must start at a 16 byte boundary: %p", __func__, *to)); flitp = (__be64 *)(*to); end = flitp + sgl->nflits; seg = &sgl->seg[0]; usgl = (void *)flitp; /* * We start at a 16 byte boundary somewhere inside the tx descriptor * ring, so we're at least 16 bytes away from the status page. There is * no chance of a wrap around in the middle of usgl (which is 16 bytes). */ usgl->cmd_nsge = htobe32(V_ULPTX_CMD(ULP_TX_SC_DSGL) | V_ULPTX_NSGE(sgl->nsegs)); usgl->len0 = htobe32(seg->ds_len); usgl->addr0 = htobe64(seg->ds_addr); seg++; if ((uintptr_t)end <= (uintptr_t)eq->spg) { /* Won't wrap around at all */ for (i = 0; i < sgl->nsegs - 1; i++, seg++) { usgl->sge[i / 2].len[i & 1] = htobe32(seg->ds_len); usgl->sge[i / 2].addr[i & 1] = htobe64(seg->ds_addr); } if (i & 1) usgl->sge[i / 2].len[1] = htobe32(0); } else { /* Will wrap somewhere in the rest of the SGL */ /* 2 flits already written, write the rest flit by flit */ flitp = (void *)(usgl + 1); for (i = 0; i < sgl->nflits - 2; i++) { if ((uintptr_t)flitp == (uintptr_t)eq->spg) flitp = (void *)eq->desc; *flitp++ = get_flit(seg, sgl->nsegs - 1, i); } end = flitp; } if ((uintptr_t)end & 0xf) { *(uint64_t *)end = 0; end++; padded = 1; } else padded = 0; if ((uintptr_t)end == (uintptr_t)eq->spg) *to = (void *)eq->desc; else *to = (void *)end; return (padded); } static inline void copy_to_txd(struct sge_eq *eq, caddr_t from, caddr_t *to, int len) { if ((uintptr_t)(*to) + len <= (uintptr_t)eq->spg) { bcopy(from, *to, len); (*to) += len; } else { int portion = (uintptr_t)eq->spg - (uintptr_t)(*to); bcopy(from, *to, portion); from += portion; portion = len - portion; /* remaining */ bcopy(from, (void *)eq->desc, portion); (*to) = (caddr_t)eq->desc + portion; } } static inline void ring_eq_db(struct adapter *sc, struct sge_eq *eq) { wmb(); t4_write_reg(sc, MYPF_REG(A_SGE_PF_KDOORBELL), V_QID(eq->cntxt_id) | V_PIDX(eq->pending)); eq->pending = 0; } static inline int reclaimable(struct sge_eq *eq) { unsigned int cidx; cidx = eq->spg->cidx; /* stable snapshot */ cidx = be16_to_cpu(cidx); if (cidx >= eq->cidx) return (cidx - eq->cidx); else return (cidx + eq->cap - eq->cidx); } /* * There are "can_reclaim" tx descriptors ready to be reclaimed. Reclaim as * many as possible but stop when there are around "n" mbufs to free. * * The actual number reclaimed is provided as the return value. */ static int reclaim_tx_descs(struct sge_txq *txq, int can_reclaim, int n) { struct tx_sdesc *txsd; struct tx_map *txm; unsigned int reclaimed, maps; struct sge_eq *eq = &txq->eq; EQ_LOCK_ASSERT_OWNED(eq); if (can_reclaim == 0) can_reclaim = reclaimable(eq); maps = reclaimed = 0; while (can_reclaim && maps < n) { int ndesc; txsd = &txq->sdesc[eq->cidx]; ndesc = txsd->desc_used; /* Firmware doesn't return "partial" credits. */ KASSERT(can_reclaim >= ndesc, ("%s: unexpected number of credits: %d, %d", __func__, can_reclaim, ndesc)); maps += txsd->credits; reclaimed += ndesc; can_reclaim -= ndesc; eq->cidx += ndesc; if (__predict_false(eq->cidx >= eq->cap)) eq->cidx -= eq->cap; } txm = &txq->maps[txq->map_cidx]; if (maps) prefetch(txm->m); eq->avail += reclaimed; KASSERT(eq->avail < eq->cap, /* avail tops out at (cap - 1) */ ("%s: too many descriptors available", __func__)); txq->map_avail += maps; KASSERT(txq->map_avail <= txq->map_total, ("%s: too many maps available", __func__)); while (maps--) { struct tx_map *next; next = txm + 1; if (__predict_false(txq->map_cidx + 1 == txq->map_total)) next = txq->maps; prefetch(next->m); bus_dmamap_unload(txq->tx_tag, txm->map); m_freem(txm->m); txm->m = NULL; txm = next; if (__predict_false(++txq->map_cidx == txq->map_total)) txq->map_cidx = 0; } return (reclaimed); } static void write_eqflush_wr(struct sge_eq *eq) { struct fw_eq_flush_wr *wr; EQ_LOCK_ASSERT_OWNED(eq); KASSERT(eq->avail > 0, ("%s: no descriptors left.", __func__)); wr = (void *)&eq->desc[eq->pidx]; bzero(wr, sizeof(*wr)); wr->opcode = FW_EQ_FLUSH_WR; wr->equiq_to_len16 = htobe32(V_FW_WR_LEN16(sizeof(*wr) / 16) | F_FW_WR_EQUEQ | F_FW_WR_EQUIQ); eq->flags |= EQ_CRFLUSHED; eq->pending++; eq->avail--; if (++eq->pidx == eq->cap) eq->pidx = 0; } static __be64 get_flit(bus_dma_segment_t *sgl, int nsegs, int idx) { int i = (idx / 3) * 2; switch (idx % 3) { case 0: { __be64 rc; rc = htobe32(sgl[i].ds_len); if (i + 1 < nsegs) rc |= (uint64_t)htobe32(sgl[i + 1].ds_len) << 32; return (rc); } case 1: return htobe64(sgl[i].ds_addr); case 2: return htobe64(sgl[i + 1].ds_addr); } return (0); } static void set_fl_tag_idx(struct sge_fl *fl, int mtu) { int i; FL_LOCK_ASSERT_OWNED(fl); for (i = 0; i < FL_BUF_SIZES - 1; i++) { if (FL_BUF_SIZE(i) >= (mtu + FL_PKTSHIFT)) break; } fl->tag_idx = i; } static int handle_sge_egr_update(struct adapter *sc, const struct cpl_sge_egr_update *cpl) { unsigned int qid = G_EGR_QID(ntohl(cpl->opcode_qid)); struct sge *s = &sc->sge; struct sge_txq *txq; struct port_info *pi; txq = (void *)s->eqmap[qid - s->eq_start]; TXQ_LOCK(txq); if (txq->eq.flags & EQ_CRFLUSHED) { pi = txq->ifp->if_softc; taskqueue_enqueue(pi->tq, &txq->resume_tx); txq->egr_update++; } else wakeup_one(txq); /* txq is going away, wakeup free_txq */ TXQ_UNLOCK(txq); return (0); } /* * m0 is freed on successful transmission. */ static int ctrl_tx(struct adapter *sc, struct sge_ctrlq *ctrlq, struct mbuf *m0) { struct sge_eq *eq = &ctrlq->eq; int rc = 0, ndesc; int can_reclaim; caddr_t dst; struct mbuf *m; M_ASSERTPKTHDR(m0); if (m0->m_pkthdr.len > SGE_MAX_WR_LEN) { ctrlq->too_long++; return (EMSGSIZE); } ndesc = howmany(m0->m_pkthdr.len, CTRL_EQ_ESIZE); EQ_LOCK(eq); can_reclaim = reclaimable(eq); eq->cidx += can_reclaim; eq->avail += can_reclaim; if (__predict_false(eq->cidx >= eq->cap)) eq->cidx -= eq->cap; if (eq->avail < ndesc) { rc = EAGAIN; ctrlq->no_desc++; goto failed; } dst = (void *)&eq->desc[eq->pidx]; for (m = m0; m; m = m->m_next) copy_to_txd(eq, mtod(m, caddr_t), &dst, m->m_len); eq->pidx += ndesc; if (__predict_false(eq->pidx >= eq->cap)) eq->pidx -= eq->cap; eq->pending += ndesc; ctrlq->total_wrs++; ring_eq_db(sc, eq); failed: EQ_UNLOCK(eq); if (rc == 0) m_freem(m0); return (rc); } Index: projects/largeSMP/sys/dev/mii/brgphy.c =================================================================== --- projects/largeSMP/sys/dev/mii/brgphy.c (revision 221494) +++ projects/largeSMP/sys/dev/mii/brgphy.c (revision 221495) @@ -1,1047 +1,1047 @@ /*- * Copyright (c) 2000 * Bill Paul . All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by Bill Paul. * 4. Neither the name of the author nor the names of any co-contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY Bill Paul AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL Bill Paul OR THE VOICES IN HIS HEAD * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF * THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); /* * Driver for the Broadcom BCM54xx/57xx 1000baseTX PHY. */ #include #include #include #include #include #include #include #include #include #include #include #include "miidevs.h" #include #include #include #include #include #include #include #include "miibus_if.h" static int brgphy_probe(device_t); static int brgphy_attach(device_t); struct brgphy_softc { struct mii_softc mii_sc; int serdes_flags; /* Keeps track of the serdes type used */ #define BRGPHY_5706S 0x0001 #define BRGPHY_5708S 0x0002 #define BRGPHY_NOANWAIT 0x0004 #define BRGPHY_5709S 0x0008 int bce_phy_flags; /* PHY flags transferred from the MAC driver */ }; static device_method_t brgphy_methods[] = { /* device interface */ DEVMETHOD(device_probe, brgphy_probe), DEVMETHOD(device_attach, brgphy_attach), DEVMETHOD(device_detach, mii_phy_detach), DEVMETHOD(device_shutdown, bus_generic_shutdown), { 0, 0 } }; static devclass_t brgphy_devclass; static driver_t brgphy_driver = { "brgphy", brgphy_methods, sizeof(struct brgphy_softc) }; DRIVER_MODULE(brgphy, miibus, brgphy_driver, brgphy_devclass, 0, 0); static int brgphy_service(struct mii_softc *, struct mii_data *, int); static void brgphy_setmedia(struct mii_softc *, int); static void brgphy_status(struct mii_softc *); static void brgphy_mii_phy_auto(struct mii_softc *, int); static void brgphy_reset(struct mii_softc *); static void brgphy_enable_loopback(struct mii_softc *); static void bcm5401_load_dspcode(struct mii_softc *); static void bcm5411_load_dspcode(struct mii_softc *); static void bcm54k2_load_dspcode(struct mii_softc *); static void brgphy_fixup_5704_a0_bug(struct mii_softc *); static void brgphy_fixup_adc_bug(struct mii_softc *); static void brgphy_fixup_adjust_trim(struct mii_softc *); static void brgphy_fixup_ber_bug(struct mii_softc *); static void brgphy_fixup_crc_bug(struct mii_softc *); static void brgphy_fixup_jitter_bug(struct mii_softc *); static void brgphy_ethernet_wirespeed(struct mii_softc *); static void brgphy_jumbo_settings(struct mii_softc *, u_long); static const struct mii_phydesc brgphys[] = { MII_PHY_DESC(BROADCOM, BCM5400), MII_PHY_DESC(BROADCOM, BCM5401), MII_PHY_DESC(BROADCOM, BCM5411), MII_PHY_DESC(BROADCOM, BCM54K2), MII_PHY_DESC(BROADCOM, BCM5701), MII_PHY_DESC(BROADCOM, BCM5703), MII_PHY_DESC(BROADCOM, BCM5704), MII_PHY_DESC(BROADCOM, BCM5705), MII_PHY_DESC(BROADCOM, BCM5706), MII_PHY_DESC(BROADCOM, BCM5714), MII_PHY_DESC(BROADCOM, BCM5421), MII_PHY_DESC(BROADCOM, BCM5750), MII_PHY_DESC(BROADCOM, BCM5752), MII_PHY_DESC(BROADCOM, BCM5780), MII_PHY_DESC(BROADCOM, BCM5708C), MII_PHY_DESC(BROADCOM2, BCM5482), MII_PHY_DESC(BROADCOM2, BCM5708S), MII_PHY_DESC(BROADCOM2, BCM5709C), MII_PHY_DESC(BROADCOM2, BCM5709S), MII_PHY_DESC(BROADCOM2, BCM5709CAX), MII_PHY_DESC(BROADCOM2, BCM5722), MII_PHY_DESC(BROADCOM2, BCM5755), MII_PHY_DESC(BROADCOM2, BCM5754), MII_PHY_DESC(BROADCOM2, BCM5761), MII_PHY_DESC(BROADCOM2, BCM5784), MII_PHY_DESC(BROADCOM3, BCM5717C), MII_PHY_DESC(BROADCOM3, BCM57765), MII_PHY_DESC(xxBROADCOM_ALT1, BCM5906), MII_PHY_END }; static const struct mii_phy_funcs brgphy_funcs = { brgphy_service, brgphy_status, brgphy_reset }; #define HS21_PRODUCT_ID "IBM eServer BladeCenter HS21" #define HS21_BCM_CHIPID 0x57081021 static int detect_hs21(struct bce_softc *bce_sc) { char *sysenv; int found; found = 0; if (bce_sc->bce_chipid == HS21_BCM_CHIPID) { sysenv = getenv("smbios.system.product"); if (sysenv != NULL) { if (strncmp(sysenv, HS21_PRODUCT_ID, strlen(HS21_PRODUCT_ID)) == 0) found = 1; freeenv(sysenv); } } return (found); } /* Search for our PHY in the list of known PHYs */ static int brgphy_probe(device_t dev) { return (mii_phy_dev_probe(dev, brgphys, BUS_PROBE_DEFAULT)); } /* Attach the PHY to the MII bus */ static int brgphy_attach(device_t dev) { struct brgphy_softc *bsc; struct bge_softc *bge_sc = NULL; struct bce_softc *bce_sc = NULL; struct mii_softc *sc; struct ifnet *ifp; bsc = device_get_softc(dev); sc = &bsc->mii_sc; mii_phy_dev_attach(dev, MIIF_NOISOLATE | MIIF_NOMANPAUSE, &brgphy_funcs, 0); bsc->serdes_flags = 0; /* Handle any special cases based on the PHY ID */ switch (sc->mii_mpd_oui) { case MII_OUI_BROADCOM: switch (sc->mii_mpd_model) { case MII_MODEL_BROADCOM_BCM5706: case MII_MODEL_BROADCOM_BCM5714: /* * The 5464 PHY used in the 5706 supports both copper * and fiber interfaces over GMII. Need to check the * shadow registers to see which mode is actually * in effect, and therefore whether we have 5706C or * 5706S. */ PHY_WRITE(sc, BRGPHY_MII_SHADOW_1C, BRGPHY_SHADOW_1C_MODE_CTRL); if (PHY_READ(sc, BRGPHY_MII_SHADOW_1C) & BRGPHY_SHADOW_1C_ENA_1000X) { bsc->serdes_flags |= BRGPHY_5706S; sc->mii_flags |= MIIF_HAVEFIBER; } break; } break; case MII_OUI_BROADCOM2: switch (sc->mii_mpd_model) { case MII_MODEL_BROADCOM2_BCM5708S: bsc->serdes_flags |= BRGPHY_5708S; sc->mii_flags |= MIIF_HAVEFIBER; break; case MII_MODEL_BROADCOM2_BCM5709S: bsc->serdes_flags |= BRGPHY_5709S; sc->mii_flags |= MIIF_HAVEFIBER; break; } break; } ifp = sc->mii_pdata->mii_ifp; /* Find the MAC driver associated with this PHY. */ if (strcmp(ifp->if_dname, "bge") == 0) { bge_sc = ifp->if_softc; } else if (strcmp(ifp->if_dname, "bce") == 0) { bce_sc = ifp->if_softc; } PHY_RESET(sc); /* Read the PHY's capabilities. */ sc->mii_capabilities = PHY_READ(sc, MII_BMSR) & sc->mii_capmask; if (sc->mii_capabilities & BMSR_EXTSTAT) sc->mii_extcapabilities = PHY_READ(sc, MII_EXTSR); device_printf(dev, " "); #define ADD(m, c) ifmedia_add(&sc->mii_pdata->mii_media, (m), (c), NULL) /* Add the supported media types */ if ((sc->mii_flags & MIIF_HAVEFIBER) == 0) { mii_phy_add_media(sc); printf("\n"); } else { sc->mii_anegticks = MII_ANEGTICKS_GIGE; ADD(IFM_MAKEWORD(IFM_ETHER, IFM_1000_SX, IFM_FDX, sc->mii_inst), BRGPHY_S1000 | BRGPHY_BMCR_FDX); printf("1000baseSX-FDX, "); /* 2.5G support is a software enabled feature on the 5708S and 5709S. */ if (bce_sc && (bce_sc->bce_phy_flags & BCE_PHY_2_5G_CAPABLE_FLAG)) { ADD(IFM_MAKEWORD(IFM_ETHER, IFM_2500_SX, IFM_FDX, sc->mii_inst), 0); printf("2500baseSX-FDX, "); } else if ((bsc->serdes_flags & BRGPHY_5708S) && bce_sc && (detect_hs21(bce_sc) != 0)) { /* * There appears to be certain silicon revision * in IBM HS21 blades that is having issues with * this driver wating for the auto-negotiation to * complete. This happens with a specific chip id * only and when the 1000baseSX-FDX is the only * mode. Workaround this issue since it's unlikely * to be ever addressed. */ printf("auto-neg workaround, "); bsc->serdes_flags |= BRGPHY_NOANWAIT; } ADD(IFM_MAKEWORD(IFM_ETHER, IFM_AUTO, 0, sc->mii_inst), 0); printf("auto\n"); } #undef ADD MIIBUS_MEDIAINIT(sc->mii_dev); return (0); } static int brgphy_service(struct mii_softc *sc, struct mii_data *mii, int cmd) { struct ifmedia_entry *ife = mii->mii_media.ifm_cur; int val; switch (cmd) { case MII_POLLSTAT: break; case MII_MEDIACHG: /* If the interface is not up, don't do anything. */ if ((mii->mii_ifp->if_flags & IFF_UP) == 0) break; /* Todo: Why is this here? Is it really needed? */ PHY_RESET(sc); /* XXX hardware bug work-around */ switch (IFM_SUBTYPE(ife->ifm_media)) { case IFM_AUTO: brgphy_mii_phy_auto(sc, ife->ifm_media); break; case IFM_2500_SX: case IFM_1000_SX: case IFM_1000_T: case IFM_100_TX: case IFM_10_T: brgphy_setmedia(sc, ife->ifm_media); break; default: return (EINVAL); } break; case MII_TICK: /* Bail if the interface isn't up. */ if ((mii->mii_ifp->if_flags & IFF_UP) == 0) return (0); /* Bail if autoneg isn't in process. */ if (IFM_SUBTYPE(ife->ifm_media) != IFM_AUTO) { sc->mii_ticks = 0; break; } /* * Check to see if we have link. If we do, we don't * need to restart the autonegotiation process. */ val = PHY_READ(sc, MII_BMSR) | PHY_READ(sc, MII_BMSR); if (val & BMSR_LINK) { sc->mii_ticks = 0; /* Reset autoneg timer. */ break; } /* Announce link loss right after it happens. */ if (sc->mii_ticks++ == 0) break; /* Only retry autonegotiation every mii_anegticks seconds. */ if (sc->mii_ticks <= sc->mii_anegticks) break; /* Retry autonegotiation */ sc->mii_ticks = 0; brgphy_mii_phy_auto(sc, ife->ifm_media); break; } /* Update the media status. */ PHY_STATUS(sc); /* * Callback if something changed. Note that we need to poke * the DSP on the Broadcom PHYs if the media changes. */ if (sc->mii_media_active != mii->mii_media_active || sc->mii_media_status != mii->mii_media_status || cmd == MII_MEDIACHG) { switch (sc->mii_mpd_oui) { case MII_OUI_BROADCOM: switch (sc->mii_mpd_model) { case MII_MODEL_BROADCOM_BCM5400: bcm5401_load_dspcode(sc); break; case MII_MODEL_BROADCOM_BCM5401: if (sc->mii_mpd_rev == 1 || sc->mii_mpd_rev == 3) bcm5401_load_dspcode(sc); break; case MII_MODEL_BROADCOM_BCM5411: bcm5411_load_dspcode(sc); break; case MII_MODEL_BROADCOM_BCM54K2: bcm54k2_load_dspcode(sc); break; } break; } } mii_phy_update(sc, cmd); return (0); } /****************************************************************************/ /* Sets the PHY link speed. */ /* */ /* Returns: */ /* None */ /****************************************************************************/ static void brgphy_setmedia(struct mii_softc *sc, int media) { int bmcr = 0, gig; switch (IFM_SUBTYPE(media)) { case IFM_2500_SX: break; case IFM_1000_SX: case IFM_1000_T: bmcr = BRGPHY_S1000; break; case IFM_100_TX: bmcr = BRGPHY_S100; break; case IFM_10_T: default: bmcr = BRGPHY_S10; break; } if ((media & IFM_FDX) != 0) { bmcr |= BRGPHY_BMCR_FDX; gig = BRGPHY_1000CTL_AFD; } else { gig = BRGPHY_1000CTL_AHD; } /* Force loopback to disconnect PHY from Ethernet medium. */ brgphy_enable_loopback(sc); PHY_WRITE(sc, BRGPHY_MII_1000CTL, 0); PHY_WRITE(sc, BRGPHY_MII_ANAR, BRGPHY_SEL_TYPE); if (IFM_SUBTYPE(media) != IFM_1000_T && IFM_SUBTYPE(media) != IFM_1000_SX) { PHY_WRITE(sc, BRGPHY_MII_BMCR, bmcr); return; } if (IFM_SUBTYPE(media) == IFM_1000_T) { gig |= BRGPHY_1000CTL_MSE; if ((media & IFM_ETH_MASTER) != 0) gig |= BRGPHY_1000CTL_MSC; } PHY_WRITE(sc, BRGPHY_MII_1000CTL, gig); PHY_WRITE(sc, BRGPHY_MII_BMCR, bmcr | BRGPHY_BMCR_AUTOEN | BRGPHY_BMCR_STARTNEG); } /****************************************************************************/ /* Set the media status based on the PHY settings. */ /* */ /* Returns: */ /* None */ /****************************************************************************/ static void brgphy_status(struct mii_softc *sc) { struct brgphy_softc *bsc = (struct brgphy_softc *)sc; struct mii_data *mii = sc->mii_pdata; int aux, bmcr, bmsr, val, xstat; u_int flowstat; mii->mii_media_status = IFM_AVALID; mii->mii_media_active = IFM_ETHER; bmsr = PHY_READ(sc, BRGPHY_MII_BMSR) | PHY_READ(sc, BRGPHY_MII_BMSR); bmcr = PHY_READ(sc, BRGPHY_MII_BMCR); if (bmcr & BRGPHY_BMCR_LOOP) { mii->mii_media_active |= IFM_LOOP; } if ((bmcr & BRGPHY_BMCR_AUTOEN) && (bmsr & BRGPHY_BMSR_ACOMP) == 0 && (bsc->serdes_flags & BRGPHY_NOANWAIT) == 0) { /* Erg, still trying, I guess... */ mii->mii_media_active |= IFM_NONE; return; } if ((sc->mii_flags & MIIF_HAVEFIBER) == 0) { /* * NB: reading the ANAR, ANLPAR or 1000STS after the AUXSTS * wedges at least the PHY of BCM5704 (but not others). */ flowstat = mii_phy_flowstatus(sc); xstat = PHY_READ(sc, BRGPHY_MII_1000STS); aux = PHY_READ(sc, BRGPHY_MII_AUXSTS); /* If copper link is up, get the negotiated speed/duplex. */ if (aux & BRGPHY_AUXSTS_LINK) { mii->mii_media_status |= IFM_ACTIVE; switch (aux & BRGPHY_AUXSTS_AN_RES) { case BRGPHY_RES_1000FD: mii->mii_media_active |= IFM_1000_T | IFM_FDX; break; case BRGPHY_RES_1000HD: mii->mii_media_active |= IFM_1000_T | IFM_HDX; break; case BRGPHY_RES_100FD: mii->mii_media_active |= IFM_100_TX | IFM_FDX; break; case BRGPHY_RES_100T4: mii->mii_media_active |= IFM_100_T4; break; case BRGPHY_RES_100HD: mii->mii_media_active |= IFM_100_TX | IFM_HDX; break; case BRGPHY_RES_10FD: mii->mii_media_active |= IFM_10_T | IFM_FDX; break; case BRGPHY_RES_10HD: mii->mii_media_active |= IFM_10_T | IFM_HDX; break; default: mii->mii_media_active |= IFM_NONE; break; } if ((mii->mii_media_active & IFM_FDX) != 0) mii->mii_media_active |= flowstat; if (IFM_SUBTYPE(mii->mii_media_active) == IFM_1000_T && (xstat & BRGPHY_1000STS_MSR) != 0) mii->mii_media_active |= IFM_ETH_MASTER; } } else { /* Todo: Add support for flow control. */ /* If serdes link is up, get the negotiated speed/duplex. */ if (bmsr & BRGPHY_BMSR_LINK) { mii->mii_media_status |= IFM_ACTIVE; } /* Check the link speed/duplex based on the PHY type. */ if (bsc->serdes_flags & BRGPHY_5706S) { mii->mii_media_active |= IFM_1000_SX; /* If autoneg enabled, read negotiated duplex settings */ if (bmcr & BRGPHY_BMCR_AUTOEN) { val = PHY_READ(sc, BRGPHY_SERDES_ANAR) & PHY_READ(sc, BRGPHY_SERDES_ANLPAR); if (val & BRGPHY_SERDES_ANAR_FDX) mii->mii_media_active |= IFM_FDX; else mii->mii_media_active |= IFM_HDX; } } else if (bsc->serdes_flags & BRGPHY_5708S) { PHY_WRITE(sc, BRGPHY_5708S_BLOCK_ADDR, BRGPHY_5708S_DIG_PG0); xstat = PHY_READ(sc, BRGPHY_5708S_PG0_1000X_STAT1); /* Check for MRBE auto-negotiated speed results. */ switch (xstat & BRGPHY_5708S_PG0_1000X_STAT1_SPEED_MASK) { case BRGPHY_5708S_PG0_1000X_STAT1_SPEED_10: mii->mii_media_active |= IFM_10_FL; break; case BRGPHY_5708S_PG0_1000X_STAT1_SPEED_100: mii->mii_media_active |= IFM_100_FX; break; case BRGPHY_5708S_PG0_1000X_STAT1_SPEED_1G: mii->mii_media_active |= IFM_1000_SX; break; case BRGPHY_5708S_PG0_1000X_STAT1_SPEED_25G: mii->mii_media_active |= IFM_2500_SX; break; } /* Check for MRBE auto-negotiated duplex results. */ if (xstat & BRGPHY_5708S_PG0_1000X_STAT1_FDX) mii->mii_media_active |= IFM_FDX; else mii->mii_media_active |= IFM_HDX; } else if (bsc->serdes_flags & BRGPHY_5709S) { /* Select GP Status Block of the AN MMD, get autoneg results. */ PHY_WRITE(sc, BRGPHY_BLOCK_ADDR, BRGPHY_BLOCK_ADDR_GP_STATUS); xstat = PHY_READ(sc, BRGPHY_GP_STATUS_TOP_ANEG_STATUS); /* Restore IEEE0 block (assumed in all brgphy(4) code). */ PHY_WRITE(sc, BRGPHY_BLOCK_ADDR, BRGPHY_BLOCK_ADDR_COMBO_IEEE0); /* Check for MRBE auto-negotiated speed results. */ switch (xstat & BRGPHY_GP_STATUS_TOP_ANEG_SPEED_MASK) { case BRGPHY_GP_STATUS_TOP_ANEG_SPEED_10: mii->mii_media_active |= IFM_10_FL; break; case BRGPHY_GP_STATUS_TOP_ANEG_SPEED_100: mii->mii_media_active |= IFM_100_FX; break; case BRGPHY_GP_STATUS_TOP_ANEG_SPEED_1G: mii->mii_media_active |= IFM_1000_SX; break; case BRGPHY_GP_STATUS_TOP_ANEG_SPEED_25G: mii->mii_media_active |= IFM_2500_SX; break; } /* Check for MRBE auto-negotiated duplex results. */ if (xstat & BRGPHY_GP_STATUS_TOP_ANEG_FDX) mii->mii_media_active |= IFM_FDX; else mii->mii_media_active |= IFM_HDX; } } } static void brgphy_mii_phy_auto(struct mii_softc *sc, int media) { int anar, ktcr = 0; PHY_RESET(sc); if ((sc->mii_flags & MIIF_HAVEFIBER) == 0) { anar = BMSR_MEDIA_TO_ANAR(sc->mii_capabilities) | ANAR_CSMA; if ((media & IFM_FLOW) != 0 || (sc->mii_flags & MIIF_FORCEPAUSE) != 0) anar |= BRGPHY_ANAR_PC | BRGPHY_ANAR_ASP; PHY_WRITE(sc, BRGPHY_MII_ANAR, anar); } else { anar = BRGPHY_SERDES_ANAR_FDX | BRGPHY_SERDES_ANAR_HDX; if ((media & IFM_FLOW) != 0 || (sc->mii_flags & MIIF_FORCEPAUSE) != 0) anar |= BRGPHY_SERDES_ANAR_BOTH_PAUSE; PHY_WRITE(sc, BRGPHY_SERDES_ANAR, anar); } ktcr = BRGPHY_1000CTL_AFD | BRGPHY_1000CTL_AHD; if (sc->mii_mpd_model == MII_MODEL_BROADCOM_BCM5701) ktcr |= BRGPHY_1000CTL_MSE | BRGPHY_1000CTL_MSC; PHY_WRITE(sc, BRGPHY_MII_1000CTL, ktcr); ktcr = PHY_READ(sc, BRGPHY_MII_1000CTL); PHY_WRITE(sc, BRGPHY_MII_BMCR, BRGPHY_BMCR_AUTOEN | BRGPHY_BMCR_STARTNEG); PHY_WRITE(sc, BRGPHY_MII_IMR, 0xFF00); } /* Enable loopback to force the link down. */ static void brgphy_enable_loopback(struct mii_softc *sc) { int i; PHY_WRITE(sc, BRGPHY_MII_BMCR, BRGPHY_BMCR_LOOP); for (i = 0; i < 15000; i++) { if (!(PHY_READ(sc, BRGPHY_MII_BMSR) & BRGPHY_BMSR_LINK)) break; DELAY(10); } } /* Turn off tap power management on 5401. */ static void bcm5401_load_dspcode(struct mii_softc *sc) { static const struct { int reg; uint16_t val; } dspcode[] = { { BRGPHY_MII_AUXCTL, 0x0c20 }, { BRGPHY_MII_DSP_ADDR_REG, 0x0012 }, { BRGPHY_MII_DSP_RW_PORT, 0x1804 }, { BRGPHY_MII_DSP_ADDR_REG, 0x0013 }, { BRGPHY_MII_DSP_RW_PORT, 0x1204 }, { BRGPHY_MII_DSP_ADDR_REG, 0x8006 }, { BRGPHY_MII_DSP_RW_PORT, 0x0132 }, { BRGPHY_MII_DSP_ADDR_REG, 0x8006 }, { BRGPHY_MII_DSP_RW_PORT, 0x0232 }, { BRGPHY_MII_DSP_ADDR_REG, 0x201f }, { BRGPHY_MII_DSP_RW_PORT, 0x0a20 }, { 0, 0 }, }; int i; for (i = 0; dspcode[i].reg != 0; i++) PHY_WRITE(sc, dspcode[i].reg, dspcode[i].val); DELAY(40); } static void bcm5411_load_dspcode(struct mii_softc *sc) { static const struct { int reg; uint16_t val; } dspcode[] = { { 0x1c, 0x8c23 }, { 0x1c, 0x8ca3 }, { 0x1c, 0x8c23 }, { 0, 0 }, }; int i; for (i = 0; dspcode[i].reg != 0; i++) PHY_WRITE(sc, dspcode[i].reg, dspcode[i].val); } void bcm54k2_load_dspcode(struct mii_softc *sc) { static const struct { int reg; uint16_t val; } dspcode[] = { { 4, 0x01e1 }, { 9, 0x0300 }, { 0, 0 }, }; int i; for (i = 0; dspcode[i].reg != 0; i++) PHY_WRITE(sc, dspcode[i].reg, dspcode[i].val); } static void brgphy_fixup_5704_a0_bug(struct mii_softc *sc) { static const struct { int reg; uint16_t val; } dspcode[] = { { 0x1c, 0x8d68 }, { 0x1c, 0x8d68 }, { 0, 0 }, }; int i; for (i = 0; dspcode[i].reg != 0; i++) PHY_WRITE(sc, dspcode[i].reg, dspcode[i].val); } static void brgphy_fixup_adc_bug(struct mii_softc *sc) { static const struct { int reg; uint16_t val; } dspcode[] = { { BRGPHY_MII_AUXCTL, 0x0c00 }, { BRGPHY_MII_DSP_ADDR_REG, 0x201f }, { BRGPHY_MII_DSP_RW_PORT, 0x2aaa }, { 0, 0 }, }; int i; for (i = 0; dspcode[i].reg != 0; i++) PHY_WRITE(sc, dspcode[i].reg, dspcode[i].val); } static void brgphy_fixup_adjust_trim(struct mii_softc *sc) { static const struct { int reg; uint16_t val; } dspcode[] = { { BRGPHY_MII_AUXCTL, 0x0c00 }, { BRGPHY_MII_DSP_ADDR_REG, 0x000a }, { BRGPHY_MII_DSP_RW_PORT, 0x110b }, { BRGPHY_MII_TEST1, 0x0014 }, { BRGPHY_MII_AUXCTL, 0x0400 }, { 0, 0 }, }; int i; for (i = 0; dspcode[i].reg != 0; i++) PHY_WRITE(sc, dspcode[i].reg, dspcode[i].val); } static void brgphy_fixup_ber_bug(struct mii_softc *sc) { static const struct { int reg; uint16_t val; } dspcode[] = { { BRGPHY_MII_AUXCTL, 0x0c00 }, { BRGPHY_MII_DSP_ADDR_REG, 0x000a }, { BRGPHY_MII_DSP_RW_PORT, 0x310b }, { BRGPHY_MII_DSP_ADDR_REG, 0x201f }, { BRGPHY_MII_DSP_RW_PORT, 0x9506 }, { BRGPHY_MII_DSP_ADDR_REG, 0x401f }, { BRGPHY_MII_DSP_RW_PORT, 0x14e2 }, { BRGPHY_MII_AUXCTL, 0x0400 }, { 0, 0 }, }; int i; for (i = 0; dspcode[i].reg != 0; i++) PHY_WRITE(sc, dspcode[i].reg, dspcode[i].val); } static void brgphy_fixup_crc_bug(struct mii_softc *sc) { static const struct { int reg; uint16_t val; } dspcode[] = { { BRGPHY_MII_DSP_RW_PORT, 0x0a75 }, { 0x1c, 0x8c68 }, { 0x1c, 0x8d68 }, { 0x1c, 0x8c68 }, { 0, 0 }, }; int i; for (i = 0; dspcode[i].reg != 0; i++) PHY_WRITE(sc, dspcode[i].reg, dspcode[i].val); } static void brgphy_fixup_jitter_bug(struct mii_softc *sc) { static const struct { int reg; uint16_t val; } dspcode[] = { { BRGPHY_MII_AUXCTL, 0x0c00 }, { BRGPHY_MII_DSP_ADDR_REG, 0x000a }, { BRGPHY_MII_DSP_RW_PORT, 0x010b }, { BRGPHY_MII_AUXCTL, 0x0400 }, { 0, 0 }, }; int i; for (i = 0; dspcode[i].reg != 0; i++) PHY_WRITE(sc, dspcode[i].reg, dspcode[i].val); } static void brgphy_fixup_disable_early_dac(struct mii_softc *sc) { uint32_t val; PHY_WRITE(sc, BRGPHY_MII_DSP_ADDR_REG, 0x0f08); val = PHY_READ(sc, BRGPHY_MII_DSP_RW_PORT); val &= ~(1 << 8); PHY_WRITE(sc, BRGPHY_MII_DSP_RW_PORT, val); } static void brgphy_ethernet_wirespeed(struct mii_softc *sc) { uint32_t val; /* Enable Ethernet@WireSpeed. */ PHY_WRITE(sc, BRGPHY_MII_AUXCTL, 0x7007); val = PHY_READ(sc, BRGPHY_MII_AUXCTL); PHY_WRITE(sc, BRGPHY_MII_AUXCTL, val | (1 << 15) | (1 << 4)); } static void brgphy_jumbo_settings(struct mii_softc *sc, u_long mtu) { uint32_t val; /* Set or clear jumbo frame settings in the PHY. */ if (mtu > ETHER_MAX_LEN) { if (sc->mii_mpd_model == MII_MODEL_BROADCOM_BCM5401) { /* BCM5401 PHY cannot read-modify-write. */ PHY_WRITE(sc, BRGPHY_MII_AUXCTL, 0x4c20); } else { PHY_WRITE(sc, BRGPHY_MII_AUXCTL, 0x7); val = PHY_READ(sc, BRGPHY_MII_AUXCTL); PHY_WRITE(sc, BRGPHY_MII_AUXCTL, val | BRGPHY_AUXCTL_LONG_PKT); } val = PHY_READ(sc, BRGPHY_MII_PHY_EXTCTL); PHY_WRITE(sc, BRGPHY_MII_PHY_EXTCTL, val | BRGPHY_PHY_EXTCTL_HIGH_LA); } else { PHY_WRITE(sc, BRGPHY_MII_AUXCTL, 0x7); val = PHY_READ(sc, BRGPHY_MII_AUXCTL); PHY_WRITE(sc, BRGPHY_MII_AUXCTL, val & ~(BRGPHY_AUXCTL_LONG_PKT | 0x7)); val = PHY_READ(sc, BRGPHY_MII_PHY_EXTCTL); PHY_WRITE(sc, BRGPHY_MII_PHY_EXTCTL, val & ~BRGPHY_PHY_EXTCTL_HIGH_LA); } } static void brgphy_reset(struct mii_softc *sc) { struct bge_softc *bge_sc = NULL; struct bce_softc *bce_sc = NULL; struct ifnet *ifp; int val; /* Perform a standard PHY reset. */ mii_phy_reset(sc); /* Handle any PHY specific procedures following the reset. */ switch (sc->mii_mpd_oui) { case MII_OUI_BROADCOM: switch (sc->mii_mpd_model) { case MII_MODEL_BROADCOM_BCM5400: bcm5401_load_dspcode(sc); break; case MII_MODEL_BROADCOM_BCM5401: if (sc->mii_mpd_rev == 1 || sc->mii_mpd_rev == 3) bcm5401_load_dspcode(sc); break; case MII_MODEL_BROADCOM_BCM5411: bcm5411_load_dspcode(sc); break; case MII_MODEL_BROADCOM_BCM54K2: bcm54k2_load_dspcode(sc); break; } break; } ifp = sc->mii_pdata->mii_ifp; /* Find the driver associated with this PHY. */ if (strcmp(ifp->if_dname, "bge") == 0) { bge_sc = ifp->if_softc; } else if (strcmp(ifp->if_dname, "bce") == 0) { bce_sc = ifp->if_softc; } if (bge_sc) { /* Fix up various bugs */ if (bge_sc->bge_phy_flags & BGE_PHY_5704_A0_BUG) brgphy_fixup_5704_a0_bug(sc); if (bge_sc->bge_phy_flags & BGE_PHY_ADC_BUG) brgphy_fixup_adc_bug(sc); if (bge_sc->bge_phy_flags & BGE_PHY_ADJUST_TRIM) brgphy_fixup_adjust_trim(sc); if (bge_sc->bge_phy_flags & BGE_PHY_BER_BUG) brgphy_fixup_ber_bug(sc); if (bge_sc->bge_phy_flags & BGE_PHY_CRC_BUG) brgphy_fixup_crc_bug(sc); if (bge_sc->bge_phy_flags & BGE_PHY_JITTER_BUG) brgphy_fixup_jitter_bug(sc); brgphy_jumbo_settings(sc, ifp->if_mtu); - if (bge_sc->bge_phy_flags & BGE_PHY_WIRESPEED) + if ((bge_sc->bge_phy_flags & BGE_PHY_NO_WIRESPEED) == 0) brgphy_ethernet_wirespeed(sc); /* Enable Link LED on Dell boxes */ if (bge_sc->bge_phy_flags & BGE_PHY_NO_3LED) { PHY_WRITE(sc, BRGPHY_MII_PHY_EXTCTL, PHY_READ(sc, BRGPHY_MII_PHY_EXTCTL) & ~BRGPHY_PHY_EXTCTL_3_LED); } /* Adjust output voltage (From Linux driver) */ if (bge_sc->bge_asicrev == BGE_ASICREV_BCM5906) PHY_WRITE(sc, BRGPHY_MII_EPHY_PTEST, 0x12); } else if (bce_sc) { if (BCE_CHIP_NUM(bce_sc) == BCE_CHIP_NUM_5708 && (bce_sc->bce_phy_flags & BCE_PHY_SERDES_FLAG)) { /* Store autoneg capabilities/results in digital block (Page 0) */ PHY_WRITE(sc, BRGPHY_5708S_BLOCK_ADDR, BRGPHY_5708S_DIG3_PG2); PHY_WRITE(sc, BRGPHY_5708S_PG2_DIGCTL_3_0, BRGPHY_5708S_PG2_DIGCTL_3_0_USE_IEEE); PHY_WRITE(sc, BRGPHY_5708S_BLOCK_ADDR, BRGPHY_5708S_DIG_PG0); /* Enable fiber mode and autodetection */ PHY_WRITE(sc, BRGPHY_5708S_PG0_1000X_CTL1, PHY_READ(sc, BRGPHY_5708S_PG0_1000X_CTL1) | BRGPHY_5708S_PG0_1000X_CTL1_AUTODET_EN | BRGPHY_5708S_PG0_1000X_CTL1_FIBER_MODE); /* Enable parallel detection */ PHY_WRITE(sc, BRGPHY_5708S_PG0_1000X_CTL2, PHY_READ(sc, BRGPHY_5708S_PG0_1000X_CTL2) | BRGPHY_5708S_PG0_1000X_CTL2_PAR_DET_EN); /* Advertise 2.5G support through next page during autoneg */ if (bce_sc->bce_phy_flags & BCE_PHY_2_5G_CAPABLE_FLAG) PHY_WRITE(sc, BRGPHY_5708S_ANEG_NXT_PG_XMIT1, PHY_READ(sc, BRGPHY_5708S_ANEG_NXT_PG_XMIT1) | BRGPHY_5708S_ANEG_NXT_PG_XMIT1_25G); /* Increase TX signal amplitude */ if ((BCE_CHIP_ID(bce_sc) == BCE_CHIP_ID_5708_A0) || (BCE_CHIP_ID(bce_sc) == BCE_CHIP_ID_5708_B0) || (BCE_CHIP_ID(bce_sc) == BCE_CHIP_ID_5708_B1)) { PHY_WRITE(sc, BRGPHY_5708S_BLOCK_ADDR, BRGPHY_5708S_TX_MISC_PG5); PHY_WRITE(sc, BRGPHY_5708S_PG5_TXACTL1, PHY_READ(sc, BRGPHY_5708S_PG5_TXACTL1) & ~0x30); PHY_WRITE(sc, BRGPHY_5708S_BLOCK_ADDR, BRGPHY_5708S_DIG_PG0); } /* Backplanes use special driver/pre-driver/pre-emphasis values. */ if ((bce_sc->bce_shared_hw_cfg & BCE_SHARED_HW_CFG_PHY_BACKPLANE) && (bce_sc->bce_port_hw_cfg & BCE_PORT_HW_CFG_CFG_TXCTL3_MASK)) { PHY_WRITE(sc, BRGPHY_5708S_BLOCK_ADDR, BRGPHY_5708S_TX_MISC_PG5); PHY_WRITE(sc, BRGPHY_5708S_PG5_TXACTL3, bce_sc->bce_port_hw_cfg & BCE_PORT_HW_CFG_CFG_TXCTL3_MASK); PHY_WRITE(sc, BRGPHY_5708S_BLOCK_ADDR, BRGPHY_5708S_DIG_PG0); } } else if (BCE_CHIP_NUM(bce_sc) == BCE_CHIP_NUM_5709 && (bce_sc->bce_phy_flags & BCE_PHY_SERDES_FLAG)) { /* Select the SerDes Digital block of the AN MMD. */ PHY_WRITE(sc, BRGPHY_BLOCK_ADDR, BRGPHY_BLOCK_ADDR_SERDES_DIG); val = PHY_READ(sc, BRGPHY_SERDES_DIG_1000X_CTL1); val &= ~BRGPHY_SD_DIG_1000X_CTL1_AUTODET; val |= BRGPHY_SD_DIG_1000X_CTL1_FIBER; PHY_WRITE(sc, BRGPHY_SERDES_DIG_1000X_CTL1, val); /* Select the Over 1G block of the AN MMD. */ PHY_WRITE(sc, BRGPHY_BLOCK_ADDR, BRGPHY_BLOCK_ADDR_OVER_1G); /* Enable autoneg "Next Page" to advertise 2.5G support. */ val = PHY_READ(sc, BRGPHY_OVER_1G_UNFORMAT_PG1); if (bce_sc->bce_phy_flags & BCE_PHY_2_5G_CAPABLE_FLAG) val |= BRGPHY_5708S_ANEG_NXT_PG_XMIT1_25G; else val &= ~BRGPHY_5708S_ANEG_NXT_PG_XMIT1_25G; PHY_WRITE(sc, BRGPHY_OVER_1G_UNFORMAT_PG1, val); /* Select the Multi-Rate Backplane Ethernet block of the AN MMD. */ PHY_WRITE(sc, BRGPHY_BLOCK_ADDR, BRGPHY_BLOCK_ADDR_MRBE); /* Enable MRBE speed autoneg. */ val = PHY_READ(sc, BRGPHY_MRBE_MSG_PG5_NP); val |= BRGPHY_MRBE_MSG_PG5_NP_MBRE | BRGPHY_MRBE_MSG_PG5_NP_T2; PHY_WRITE(sc, BRGPHY_MRBE_MSG_PG5_NP, val); /* Select the Clause 73 User B0 block of the AN MMD. */ PHY_WRITE(sc, BRGPHY_BLOCK_ADDR, BRGPHY_BLOCK_ADDR_CL73_USER_B0); /* Enable MRBE speed autoneg. */ PHY_WRITE(sc, BRGPHY_CL73_USER_B0_MBRE_CTL1, BRGPHY_CL73_USER_B0_MBRE_CTL1_NP_AFT_BP | BRGPHY_CL73_USER_B0_MBRE_CTL1_STA_MGR | BRGPHY_CL73_USER_B0_MBRE_CTL1_ANEG); /* Restore IEEE0 block (assumed in all brgphy(4) code). */ PHY_WRITE(sc, BRGPHY_BLOCK_ADDR, BRGPHY_BLOCK_ADDR_COMBO_IEEE0); } else if (BCE_CHIP_NUM(bce_sc) == BCE_CHIP_NUM_5709) { if ((BCE_CHIP_REV(bce_sc) == BCE_CHIP_REV_Ax) || (BCE_CHIP_REV(bce_sc) == BCE_CHIP_REV_Bx)) brgphy_fixup_disable_early_dac(sc); brgphy_jumbo_settings(sc, ifp->if_mtu); brgphy_ethernet_wirespeed(sc); } else { brgphy_fixup_ber_bug(sc); brgphy_jumbo_settings(sc, ifp->if_mtu); brgphy_ethernet_wirespeed(sc); } } } Index: projects/largeSMP/sys/fs/nfs/nfsproto.h =================================================================== --- projects/largeSMP/sys/fs/nfs/nfsproto.h (revision 221494) +++ projects/largeSMP/sys/fs/nfs/nfsproto.h (revision 221495) @@ -1,1033 +1,1041 @@ /*- * Copyright (c) 1989, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Rick Macklem at The University of Guelph. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #ifndef _NFS_NFSPROTO_H_ #define _NFS_NFSPROTO_H_ /* * nfs definitions as per the Version 2, 3 and 4 specs */ /* * Constants as defined in the NFS Version 2, 3 and 4 specs. * "NFS: Network File System Protocol Specification" RFC1094 * and in the "NFS: Network File System Version 3 Protocol * Specification" */ #define NFS_PORT 2049 #define NFS_PROG 100003 #define NFS_CALLBCKPROG 0x40000000 /* V4 only */ #define NFS_VER2 2 #define NFS_VER3 3 #define NFS_VER4 4 #define NFS_V2MAXDATA 8192 #define NFS_MAXDGRAMDATA 16384 #define NFS_MAXDATA NFS_MAXBSIZE #define NFS_MAXPATHLEN 1024 #define NFS_MAXNAMLEN 255 #define NFS_MAXPKTHDR 404 #define NFS_MAXPACKET (NFS_MAXDATA + 2048) #define NFS_MINPACKET 20 #define NFS_FABLKSIZE 512 /* Size in bytes of a block wrt fa_blocks */ #define NFSV4_MINORVERSION 0 /* V4 Minor version */ #define NFSV4_CBVERS 1 /* V4 CB Version */ #define NFSV4_SMALLSTR 50 /* Strings small enough for stack */ /* Stat numbers for rpc returns (version 2, 3 and 4) */ +/* + * These numbers are hard-wired in the RFCs, so they can't be changed. + * The code currently assumes that the ones < 10000 are the same as + * sys/errno.h and that sys/errno.h will never go as high as 10000. + * If the value in sys/errno.h of any entry listed below is changed, + * the NFS code must be modified to do the mapping between them. + * (You can ignore NFSERR_WFLUSH, since it is never actually used.) + */ #define NFSERR_OK 0 #define NFSERR_PERM 1 #define NFSERR_NOENT 2 #define NFSERR_IO 5 #define NFSERR_NXIO 6 #define NFSERR_ACCES 13 #define NFSERR_EXIST 17 #define NFSERR_XDEV 18 /* Version 3, 4 only */ #define NFSERR_NODEV 19 #define NFSERR_NOTDIR 20 #define NFSERR_ISDIR 21 #define NFSERR_INVAL 22 /* Version 3, 4 only */ #define NFSERR_FBIG 27 #define NFSERR_NOSPC 28 #define NFSERR_ROFS 30 #define NFSERR_MLINK 31 /* Version 3, 4 only */ #define NFSERR_NAMETOL 63 #define NFSERR_NOTEMPTY 66 #define NFSERR_DQUOT 69 #define NFSERR_STALE 70 #define NFSERR_REMOTE 71 /* Version 3 only */ #define NFSERR_WFLUSH 99 /* Version 2 only */ #define NFSERR_BADHANDLE 10001 /* These are Version 3, 4 only */ #define NFSERR_NOT_SYNC 10002 /* Version 3 Only */ #define NFSERR_BAD_COOKIE 10003 #define NFSERR_NOTSUPP 10004 #define NFSERR_TOOSMALL 10005 #define NFSERR_SERVERFAULT 10006 #define NFSERR_BADTYPE 10007 #define NFSERR_DELAY 10008 /* Called NFSERR_JUKEBOX for V3 */ #define NFSERR_SAME 10009 /* These are Version 4 only */ #define NFSERR_DENIED 10010 #define NFSERR_EXPIRED 10011 #define NFSERR_LOCKED 10012 #define NFSERR_GRACE 10013 #define NFSERR_FHEXPIRED 10014 #define NFSERR_SHAREDENIED 10015 #define NFSERR_WRONGSEC 10016 #define NFSERR_CLIDINUSE 10017 #define NFSERR_RESOURCE 10018 #define NFSERR_MOVED 10019 #define NFSERR_NOFILEHANDLE 10020 #define NFSERR_MINORVERMISMATCH 10021 #define NFSERR_STALECLIENTID 10022 #define NFSERR_STALESTATEID 10023 #define NFSERR_OLDSTATEID 10024 #define NFSERR_BADSTATEID 10025 #define NFSERR_BADSEQID 10026 #define NFSERR_NOTSAME 10027 #define NFSERR_LOCKRANGE 10028 #define NFSERR_SYMLINK 10029 #define NFSERR_RESTOREFH 10030 #define NFSERR_LEASEMOVED 10031 #define NFSERR_ATTRNOTSUPP 10032 #define NFSERR_NOGRACE 10033 #define NFSERR_RECLAIMBAD 10034 #define NFSERR_RECLAIMCONFLICT 10035 #define NFSERR_BADXDR 10036 #define NFSERR_LOCKSHELD 10037 #define NFSERR_OPENMODE 10038 #define NFSERR_BADOWNER 10039 #define NFSERR_BADCHAR 10040 #define NFSERR_BADNAME 10041 #define NFSERR_BADRANGE 10042 #define NFSERR_LOCKNOTSUPP 10043 #define NFSERR_OPILLEGAL 10044 #define NFSERR_DEADLOCK 10045 #define NFSERR_FILEOPEN 10046 #define NFSERR_ADMINREVOKED 10047 #define NFSERR_CBPATHDOWN 10048 #define NFSERR_STALEWRITEVERF 30001 /* Fake return for nfs_commit() */ #define NFSERR_DONTREPLY 30003 /* Don't process request */ #define NFSERR_RETVOID 30004 /* Return void, not error */ #define NFSERR_REPLYFROMCACHE 30005 /* Reply from recent request cache */ #define NFSERR_STALEDONTRECOVER 30006 /* Don't initiate recovery */ #define NFSERR_RPCERR 0x40000000 /* Mark an RPC layer error */ #define NFSERR_AUTHERR 0x80000000 /* Mark an authentication error */ #define NFSERR_RPCMISMATCH (NFSERR_RPCERR | RPC_MISMATCH) #define NFSERR_PROGUNAVAIL (NFSERR_RPCERR | RPC_PROGUNAVAIL) #define NFSERR_PROGMISMATCH (NFSERR_RPCERR | RPC_PROGMISMATCH) #define NFSERR_PROGNOTV4 (NFSERR_RPCERR | 0xffff) #define NFSERR_PROCUNAVAIL (NFSERR_RPCERR | RPC_PROCUNAVAIL) #define NFSERR_GARBAGE (NFSERR_RPCERR | RPC_GARBAGE) /* Sizes in bytes of various nfs rpc components */ #define NFSX_UNSIGNED 4 #define NFSX_HYPER (2 * NFSX_UNSIGNED) /* specific to NFS Version 2 */ #define NFSX_V2FH 32 #define NFSX_V2FATTR 68 #define NFSX_V2SATTR 32 #define NFSX_V2COOKIE 4 #define NFSX_V2STATFS 20 /* specific to NFS Version 3 */ #define NFSX_V3FHMAX 64 /* max. allowed by protocol */ #define NFSX_V3FATTR 84 #define NFSX_V3SATTR 60 /* max. all fields filled in */ #define NFSX_V3SRVSATTR (sizeof (struct nfsv3_sattr)) #define NFSX_V3POSTOPATTR (NFSX_V3FATTR + NFSX_UNSIGNED) #define NFSX_V3WCCDATA (NFSX_V3POSTOPATTR + 8 * NFSX_UNSIGNED) #define NFSX_V3STATFS 52 #define NFSX_V3FSINFO 48 #define NFSX_V3PATHCONF 24 /* specific to NFS Version 4 */ #define NFSX_V4FHMAX 128 #define NFSX_V4FSID (2 * NFSX_HYPER) #define NFSX_V4SPECDATA (2 * NFSX_UNSIGNED) #define NFSX_V4TIME (NFSX_HYPER + NFSX_UNSIGNED) #define NFSX_V4SETTIME (NFSX_UNSIGNED + NFSX_V4TIME) /* sizes common to multiple NFS versions */ #define NFSX_FHMAX (NFSX_V4FHMAX) #define NFSX_MYFH (sizeof (fhandle_t)) /* size this server uses */ #define NFSX_VERF 8 #define NFSX_STATEIDOTHER 12 #define NFSX_STATEID (NFSX_UNSIGNED + NFSX_STATEIDOTHER) #define NFSX_GSSH 12 /* variants for multiple versions */ #define NFSX_STATFS(v3) ((v3) ? NFSX_V3STATFS : NFSX_V2STATFS) /* nfs rpc procedure numbers (before version mapping) */ #define NFSPROC_NULL 0 #define NFSPROC_GETATTR 1 #define NFSPROC_SETATTR 2 #define NFSPROC_LOOKUP 3 #define NFSPROC_ACCESS 4 #define NFSPROC_READLINK 5 #define NFSPROC_READ 6 #define NFSPROC_WRITE 7 #define NFSPROC_CREATE 8 #define NFSPROC_MKDIR 9 #define NFSPROC_SYMLINK 10 #define NFSPROC_MKNOD 11 #define NFSPROC_REMOVE 12 #define NFSPROC_RMDIR 13 #define NFSPROC_RENAME 14 #define NFSPROC_LINK 15 #define NFSPROC_READDIR 16 #define NFSPROC_READDIRPLUS 17 #define NFSPROC_FSSTAT 18 #define NFSPROC_FSINFO 19 #define NFSPROC_PATHCONF 20 #define NFSPROC_COMMIT 21 /* * NFSPROC_NOOP is a fake op# that can't be the same as any V2/3/4 Procedure * or Operation#. Since the NFS V4 Op #s go higher, use NFSV4OP_NOPS, which * is one greater than the highest Op#. */ #define NFSPROC_NOOP NFSV4OP_NOPS /* Actual Version 2 procedure numbers */ #define NFSV2PROC_NULL 0 #define NFSV2PROC_GETATTR 1 #define NFSV2PROC_SETATTR 2 #define NFSV2PROC_NOOP 3 #define NFSV2PROC_ROOT NFSV2PROC_NOOP /* Obsolete */ #define NFSV2PROC_LOOKUP 4 #define NFSV2PROC_READLINK 5 #define NFSV2PROC_READ 6 #define NFSV2PROC_WRITECACHE NFSV2PROC_NOOP /* Obsolete */ #define NFSV2PROC_WRITE 8 #define NFSV2PROC_CREATE 9 #define NFSV2PROC_REMOVE 10 #define NFSV2PROC_RENAME 11 #define NFSV2PROC_LINK 12 #define NFSV2PROC_SYMLINK 13 #define NFSV2PROC_MKDIR 14 #define NFSV2PROC_RMDIR 15 #define NFSV2PROC_READDIR 16 #define NFSV2PROC_STATFS 17 /* * V4 Procedure numbers */ #define NFSV4PROC_COMPOUND 1 #define NFSV4PROC_CBNULL 0 #define NFSV4PROC_CBCOMPOUND 1 /* * Constants used by the Version 3 and 4 protocols for various RPCs */ #define NFSV3SATTRTIME_DONTCHANGE 0 #define NFSV3SATTRTIME_TOSERVER 1 #define NFSV3SATTRTIME_TOCLIENT 2 #define NFSV4SATTRTIME_TOSERVER 0 #define NFSV4SATTRTIME_TOCLIENT 1 #define NFSV4LOCKT_READ 1 #define NFSV4LOCKT_WRITE 2 #define NFSV4LOCKT_READW 3 #define NFSV4LOCKT_WRITEW 4 #define NFSV4LOCKT_RELEASE 5 #define NFSV4OPEN_NOCREATE 0 #define NFSV4OPEN_CREATE 1 #define NFSV4OPEN_CLAIMNULL 0 #define NFSV4OPEN_CLAIMPREVIOUS 1 #define NFSV4OPEN_CLAIMDELEGATECUR 2 #define NFSV4OPEN_CLAIMDELEGATEPREV 3 #define NFSV4OPEN_DELEGATENONE 0 #define NFSV4OPEN_DELEGATEREAD 1 #define NFSV4OPEN_DELEGATEWRITE 2 #define NFSV4OPEN_LIMITSIZE 1 #define NFSV4OPEN_LIMITBLOCKS 2 /* * Nfs V4 ACE stuff */ #define NFSV4ACE_ALLOWEDTYPE 0x00000000 #define NFSV4ACE_DENIEDTYPE 0x00000001 #define NFSV4ACE_AUDITTYPE 0x00000002 #define NFSV4ACE_ALARMTYPE 0x00000003 #define NFSV4ACE_SUPALLOWED 0x00000001 #define NFSV4ACE_SUPDENIED 0x00000002 #define NFSV4ACE_SUPAUDIT 0x00000004 #define NFSV4ACE_SUPALARM 0x00000008 #define NFSV4ACE_SUPTYPES (NFSV4ACE_SUPALLOWED | NFSV4ACE_SUPDENIED) #define NFSV4ACE_FILEINHERIT 0x00000001 #define NFSV4ACE_DIRECTORYINHERIT 0x00000002 #define NFSV4ACE_NOPROPAGATEINHERIT 0x00000004 #define NFSV4ACE_INHERITONLY 0x00000008 #define NFSV4ACE_SUCCESSFULACCESS 0x00000010 #define NFSV4ACE_FAILEDACCESS 0x00000020 #define NFSV4ACE_IDENTIFIERGROUP 0x00000040 #define NFSV4ACE_READDATA 0x00000001 #define NFSV4ACE_LISTDIRECTORY 0x00000001 #define NFSV4ACE_WRITEDATA 0x00000002 #define NFSV4ACE_ADDFILE 0x00000002 #define NFSV4ACE_APPENDDATA 0x00000004 #define NFSV4ACE_ADDSUBDIRECTORY 0x00000004 #define NFSV4ACE_READNAMEDATTR 0x00000008 #define NFSV4ACE_WRITENAMEDATTR 0x00000010 #define NFSV4ACE_EXECUTE 0x00000020 #define NFSV4ACE_SEARCH 0x00000020 #define NFSV4ACE_DELETECHILD 0x00000040 #define NFSV4ACE_READATTRIBUTES 0x00000080 #define NFSV4ACE_WRITEATTRIBUTES 0x00000100 #define NFSV4ACE_DELETE 0x00010000 #define NFSV4ACE_READACL 0x00020000 #define NFSV4ACE_WRITEACL 0x00040000 #define NFSV4ACE_WRITEOWNER 0x00080000 #define NFSV4ACE_SYNCHRONIZE 0x00100000 /* * Here are the mappings between mode bits and acl mask bits for * directories and other files. * (Named attributes have not been included, since named attributes are * not yet supported.) * The mailing list seems to indicate that NFSV4ACE_EXECUTE refers to * searching a directory, although I can't find a statement of that in * the RFC. */ #define NFSV4ACE_ALLFILESMASK (NFSV4ACE_READATTRIBUTES | NFSV4ACE_READACL) #define NFSV4ACE_OWNERMASK (NFSV4ACE_WRITEATTRIBUTES | NFSV4ACE_WRITEACL) #define NFSV4ACE_DIRREADMASK NFSV4ACE_LISTDIRECTORY #define NFSV4ACE_DIREXECUTEMASK NFSV4ACE_EXECUTE #define NFSV4ACE_DIRWRITEMASK (NFSV4ACE_ADDFILE | \ NFSV4ACE_ADDSUBDIRECTORY | NFSV4ACE_DELETECHILD) #define NFSV4ACE_READMASK NFSV4ACE_READDATA #define NFSV4ACE_WRITEMASK (NFSV4ACE_WRITEDATA | NFSV4ACE_APPENDDATA) #define NFSV4ACE_EXECUTEMASK NFSV4ACE_EXECUTE #define NFSV4ACE_ALLFILEBITS (NFSV4ACE_READMASK | NFSV4ACE_WRITEMASK | \ NFSV4ACE_EXECUTEMASK | NFSV4ACE_SYNCHRONIZE) #define NFSV4ACE_ALLDIRBITS (NFSV4ACE_DIRREADMASK | \ NFSV4ACE_DIRWRITEMASK | NFSV4ACE_DIREXECUTEMASK) #define NFSV4ACE_AUDITMASK 0x0 /* * These GENERIC masks are not used and are no longer believed to be useful. */ #define NFSV4ACE_GENERICREAD 0x00120081 #define NFSV4ACE_GENERICWRITE 0x00160106 #define NFSV4ACE_GENERICEXECUTE 0x001200a0 #define NFSSTATEID_PUTALLZERO 0 #define NFSSTATEID_PUTALLONE 1 #define NFSSTATEID_PUTSTATEID 2 /* * Bits for share access and deny. */ #define NFSV4OPEN_ACCESSREAD 0x00000001 #define NFSV4OPEN_ACCESSWRITE 0x00000002 #define NFSV4OPEN_ACCESSBOTH 0x00000003 #define NFSV4OPEN_DENYNONE 0x00000000 #define NFSV4OPEN_DENYREAD 0x00000001 #define NFSV4OPEN_DENYWRITE 0x00000002 #define NFSV4OPEN_DENYBOTH 0x00000003 /* * Open result flags * (The first two are in the spec. The rest are used internally.) */ #define NFSV4OPEN_RESULTCONFIRM 0x00000002 #define NFSV4OPEN_LOCKTYPEPOSIX 0x00000004 #define NFSV4OPEN_RFLAGS \ (NFSV4OPEN_RESULTCONFIRM | NFSV4OPEN_LOCKTYPEPOSIX) #define NFSV4OPEN_RECALL 0x00010000 #define NFSV4OPEN_READDELEGATE 0x00020000 #define NFSV4OPEN_WRITEDELEGATE 0x00040000 /* * NFS V4 File Handle types */ #define NFSV4FHTYPE_PERSISTENT 0x0 #define NFSV4FHTYPE_NOEXPIREWITHOPEN 0x1 #define NFSV4FHTYPE_VOLATILEANY 0x2 #define NFSV4FHTYPE_VOLATILEMIGRATE 0x4 #define NFSV4FHTYPE_VOLATILERENAME 0x8 /* * Maximum size of V4 opaque strings. */ #define NFSV4_OPAQUELIMIT 1024 /* * These are the same for V3 and V4. */ #define NFSACCESS_READ 0x01 #define NFSACCESS_LOOKUP 0x02 #define NFSACCESS_MODIFY 0x04 #define NFSACCESS_EXTEND 0x08 #define NFSACCESS_DELETE 0x10 #define NFSACCESS_EXECUTE 0x20 #define NFSWRITE_UNSTABLE 0 #define NFSWRITE_DATASYNC 1 #define NFSWRITE_FILESYNC 2 #define NFSCREATE_UNCHECKED 0 #define NFSCREATE_GUARDED 1 #define NFSCREATE_EXCLUSIVE 2 #define NFSV3FSINFO_LINK 0x01 #define NFSV3FSINFO_SYMLINK 0x02 #define NFSV3FSINFO_HOMOGENEOUS 0x08 #define NFSV3FSINFO_CANSETTIME 0x10 /* Conversion macros */ #define vtonfsv2_mode(t,m) \ txdr_unsigned(((t) == VFIFO) ? MAKEIMODE(VCHR, (m)) : \ MAKEIMODE((t), (m))) #define vtonfsv34_mode(m) txdr_unsigned((m) & 07777) #define nfstov_mode(a) (fxdr_unsigned(u_int16_t, (a))&07777) #define vtonfsv2_type(a) (((u_int32_t)(a)) >= 9 ? txdr_unsigned(NFNON) : \ txdr_unsigned(newnfsv2_type[((u_int32_t)(a))])) #define vtonfsv34_type(a) (((u_int32_t)(a)) >= 9 ? txdr_unsigned(NFNON) : \ txdr_unsigned(nfsv34_type[((u_int32_t)(a))])) #define nfsv2tov_type(a) newnv2tov_type[fxdr_unsigned(u_int32_t,(a))&0x7] #define nfsv34tov_type(a) nv34tov_type[fxdr_unsigned(u_int32_t,(a))&0x7] #define vtonfs_dtype(a) (((u_int32_t)(a)) >= 9 ? IFTODT(VTTOIF(VNON)) : \ IFTODT(VTTOIF(a))) /* File types */ typedef enum { NFNON=0, NFREG=1, NFDIR=2, NFBLK=3, NFCHR=4, NFLNK=5, NFSOCK=6, NFFIFO=7, NFATTRDIR=8, NFNAMEDATTR=9 } nfstype; /* Structs for common parts of the rpc's */ struct nfsv2_time { u_int32_t nfsv2_sec; u_int32_t nfsv2_usec; }; typedef struct nfsv2_time nfstime2; struct nfsv3_time { u_int32_t nfsv3_sec; u_int32_t nfsv3_nsec; }; typedef struct nfsv3_time nfstime3; struct nfsv4_time { u_int32_t nfsv4_highsec; u_int32_t nfsv4_sec; u_int32_t nfsv4_nsec; }; typedef struct nfsv4_time nfstime4; /* * Quads are defined as arrays of 2 longs to ensure dense packing for the * protocol and to facilitate xdr conversion. */ struct nfs_uquad { u_int32_t nfsuquad[2]; }; typedef struct nfs_uquad nfsuint64; /* * Used to convert between two u_longs and a u_quad_t. */ union nfs_quadconvert { u_int32_t lval[2]; u_quad_t qval; }; typedef union nfs_quadconvert nfsquad_t; /* * NFS Version 3 special file number. */ struct nfsv3_spec { u_int32_t specdata1; u_int32_t specdata2; }; typedef struct nfsv3_spec nfsv3spec; /* * File attributes and setable attributes. These structures cover both * NFS version 2 and the version 3 protocol. Note that the union is only * used so that one pointer can refer to both variants. These structures * go out on the wire and must be densely packed, so no quad data types * are used. (all fields are longs or u_longs or structures of same) * NB: You can't do sizeof(struct nfs_fattr), you must use the * NFSX_FATTR(v3) macro. */ struct nfs_fattr { u_int32_t fa_type; u_int32_t fa_mode; u_int32_t fa_nlink; u_int32_t fa_uid; u_int32_t fa_gid; union { struct { u_int32_t nfsv2fa_size; u_int32_t nfsv2fa_blocksize; u_int32_t nfsv2fa_rdev; u_int32_t nfsv2fa_blocks; u_int32_t nfsv2fa_fsid; u_int32_t nfsv2fa_fileid; nfstime2 nfsv2fa_atime; nfstime2 nfsv2fa_mtime; nfstime2 nfsv2fa_ctime; } fa_nfsv2; struct { nfsuint64 nfsv3fa_size; nfsuint64 nfsv3fa_used; nfsv3spec nfsv3fa_rdev; nfsuint64 nfsv3fa_fsid; nfsuint64 nfsv3fa_fileid; nfstime3 nfsv3fa_atime; nfstime3 nfsv3fa_mtime; nfstime3 nfsv3fa_ctime; } fa_nfsv3; } fa_un; }; /* and some ugly defines for accessing union components */ #define fa2_size fa_un.fa_nfsv2.nfsv2fa_size #define fa2_blocksize fa_un.fa_nfsv2.nfsv2fa_blocksize #define fa2_rdev fa_un.fa_nfsv2.nfsv2fa_rdev #define fa2_blocks fa_un.fa_nfsv2.nfsv2fa_blocks #define fa2_fsid fa_un.fa_nfsv2.nfsv2fa_fsid #define fa2_fileid fa_un.fa_nfsv2.nfsv2fa_fileid #define fa2_atime fa_un.fa_nfsv2.nfsv2fa_atime #define fa2_mtime fa_un.fa_nfsv2.nfsv2fa_mtime #define fa2_ctime fa_un.fa_nfsv2.nfsv2fa_ctime #define fa3_size fa_un.fa_nfsv3.nfsv3fa_size #define fa3_used fa_un.fa_nfsv3.nfsv3fa_used #define fa3_rdev fa_un.fa_nfsv3.nfsv3fa_rdev #define fa3_fsid fa_un.fa_nfsv3.nfsv3fa_fsid #define fa3_fileid fa_un.fa_nfsv3.nfsv3fa_fileid #define fa3_atime fa_un.fa_nfsv3.nfsv3fa_atime #define fa3_mtime fa_un.fa_nfsv3.nfsv3fa_mtime #define fa3_ctime fa_un.fa_nfsv3.nfsv3fa_ctime struct nfsv2_sattr { u_int32_t sa_mode; u_int32_t sa_uid; u_int32_t sa_gid; u_int32_t sa_size; nfstime2 sa_atime; nfstime2 sa_mtime; }; /* * NFS Version 3 sattr structure for the new node creation case. */ struct nfsv3_sattr { u_int32_t sa_modetrue; u_int32_t sa_mode; u_int32_t sa_uidfalse; u_int32_t sa_gidfalse; u_int32_t sa_sizefalse; u_int32_t sa_atimetype; nfstime3 sa_atime; u_int32_t sa_mtimetype; nfstime3 sa_mtime; }; /* * The attribute bits used for V4. * NFSATTRBIT_xxx defines the attribute# (and its bit position) * NFSATTRBM_xxx is a 32bit mask with the correct bit set within the * appropriate 32bit word. * NFSATTRBIT_MAX is one greater than the largest NFSATTRBIT_xxx */ #define NFSATTRBIT_SUPPORTEDATTRS 0 #define NFSATTRBIT_TYPE 1 #define NFSATTRBIT_FHEXPIRETYPE 2 #define NFSATTRBIT_CHANGE 3 #define NFSATTRBIT_SIZE 4 #define NFSATTRBIT_LINKSUPPORT 5 #define NFSATTRBIT_SYMLINKSUPPORT 6 #define NFSATTRBIT_NAMEDATTR 7 #define NFSATTRBIT_FSID 8 #define NFSATTRBIT_UNIQUEHANDLES 9 #define NFSATTRBIT_LEASETIME 10 #define NFSATTRBIT_RDATTRERROR 11 #define NFSATTRBIT_ACL 12 #define NFSATTRBIT_ACLSUPPORT 13 #define NFSATTRBIT_ARCHIVE 14 #define NFSATTRBIT_CANSETTIME 15 #define NFSATTRBIT_CASEINSENSITIVE 16 #define NFSATTRBIT_CASEPRESERVING 17 #define NFSATTRBIT_CHOWNRESTRICTED 18 #define NFSATTRBIT_FILEHANDLE 19 #define NFSATTRBIT_FILEID 20 #define NFSATTRBIT_FILESAVAIL 21 #define NFSATTRBIT_FILESFREE 22 #define NFSATTRBIT_FILESTOTAL 23 #define NFSATTRBIT_FSLOCATIONS 24 #define NFSATTRBIT_HIDDEN 25 #define NFSATTRBIT_HOMOGENEOUS 26 #define NFSATTRBIT_MAXFILESIZE 27 #define NFSATTRBIT_MAXLINK 28 #define NFSATTRBIT_MAXNAME 29 #define NFSATTRBIT_MAXREAD 30 #define NFSATTRBIT_MAXWRITE 31 #define NFSATTRBIT_MIMETYPE 32 #define NFSATTRBIT_MODE 33 #define NFSATTRBIT_NOTRUNC 34 #define NFSATTRBIT_NUMLINKS 35 #define NFSATTRBIT_OWNER 36 #define NFSATTRBIT_OWNERGROUP 37 #define NFSATTRBIT_QUOTAHARD 38 #define NFSATTRBIT_QUOTASOFT 39 #define NFSATTRBIT_QUOTAUSED 40 #define NFSATTRBIT_RAWDEV 41 #define NFSATTRBIT_SPACEAVAIL 42 #define NFSATTRBIT_SPACEFREE 43 #define NFSATTRBIT_SPACETOTAL 44 #define NFSATTRBIT_SPACEUSED 45 #define NFSATTRBIT_SYSTEM 46 #define NFSATTRBIT_TIMEACCESS 47 #define NFSATTRBIT_TIMEACCESSSET 48 #define NFSATTRBIT_TIMEBACKUP 49 #define NFSATTRBIT_TIMECREATE 50 #define NFSATTRBIT_TIMEDELTA 51 #define NFSATTRBIT_TIMEMETADATA 52 #define NFSATTRBIT_TIMEMODIFY 53 #define NFSATTRBIT_TIMEMODIFYSET 54 #define NFSATTRBIT_MOUNTEDONFILEID 55 #define NFSATTRBM_SUPPORTEDATTRS 0x00000001 #define NFSATTRBM_TYPE 0x00000002 #define NFSATTRBM_FHEXPIRETYPE 0x00000004 #define NFSATTRBM_CHANGE 0x00000008 #define NFSATTRBM_SIZE 0x00000010 #define NFSATTRBM_LINKSUPPORT 0x00000020 #define NFSATTRBM_SYMLINKSUPPORT 0x00000040 #define NFSATTRBM_NAMEDATTR 0x00000080 #define NFSATTRBM_FSID 0x00000100 #define NFSATTRBM_UNIQUEHANDLES 0x00000200 #define NFSATTRBM_LEASETIME 0x00000400 #define NFSATTRBM_RDATTRERROR 0x00000800 #define NFSATTRBM_ACL 0x00001000 #define NFSATTRBM_ACLSUPPORT 0x00002000 #define NFSATTRBM_ARCHIVE 0x00004000 #define NFSATTRBM_CANSETTIME 0x00008000 #define NFSATTRBM_CASEINSENSITIVE 0x00010000 #define NFSATTRBM_CASEPRESERVING 0x00020000 #define NFSATTRBM_CHOWNRESTRICTED 0x00040000 #define NFSATTRBM_FILEHANDLE 0x00080000 #define NFSATTRBM_FILEID 0x00100000 #define NFSATTRBM_FILESAVAIL 0x00200000 #define NFSATTRBM_FILESFREE 0x00400000 #define NFSATTRBM_FILESTOTAL 0x00800000 #define NFSATTRBM_FSLOCATIONS 0x01000000 #define NFSATTRBM_HIDDEN 0x02000000 #define NFSATTRBM_HOMOGENEOUS 0x04000000 #define NFSATTRBM_MAXFILESIZE 0x08000000 #define NFSATTRBM_MAXLINK 0x10000000 #define NFSATTRBM_MAXNAME 0x20000000 #define NFSATTRBM_MAXREAD 0x40000000 #define NFSATTRBM_MAXWRITE 0x80000000 #define NFSATTRBM_MIMETYPE 0x00000001 #define NFSATTRBM_MODE 0x00000002 #define NFSATTRBM_NOTRUNC 0x00000004 #define NFSATTRBM_NUMLINKS 0x00000008 #define NFSATTRBM_OWNER 0x00000010 #define NFSATTRBM_OWNERGROUP 0x00000020 #define NFSATTRBM_QUOTAHARD 0x00000040 #define NFSATTRBM_QUOTASOFT 0x00000080 #define NFSATTRBM_QUOTAUSED 0x00000100 #define NFSATTRBM_RAWDEV 0x00000200 #define NFSATTRBM_SPACEAVAIL 0x00000400 #define NFSATTRBM_SPACEFREE 0x00000800 #define NFSATTRBM_SPACETOTAL 0x00001000 #define NFSATTRBM_SPACEUSED 0x00002000 #define NFSATTRBM_SYSTEM 0x00004000 #define NFSATTRBM_TIMEACCESS 0x00008000 #define NFSATTRBM_TIMEACCESSSET 0x00010000 #define NFSATTRBM_TIMEBACKUP 0x00020000 #define NFSATTRBM_TIMECREATE 0x00040000 #define NFSATTRBM_TIMEDELTA 0x00080000 #define NFSATTRBM_TIMEMETADATA 0x00100000 #define NFSATTRBM_TIMEMODIFY 0x00200000 #define NFSATTRBM_TIMEMODIFYSET 0x00400000 #define NFSATTRBM_MOUNTEDONFILEID 0x00800000 #define NFSATTRBIT_MAX 56 /* * Sets of attributes that are supported, by words in the bitmap. */ /* * NFSATTRBIT_SUPPORTED - SUPP0 - bits 0<->31 * SUPP1 - bits 32<->63 */ #define NFSATTRBIT_SUPP0 \ (NFSATTRBM_SUPPORTEDATTRS | \ NFSATTRBM_TYPE | \ NFSATTRBM_FHEXPIRETYPE | \ NFSATTRBM_CHANGE | \ NFSATTRBM_SIZE | \ NFSATTRBM_LINKSUPPORT | \ NFSATTRBM_SYMLINKSUPPORT | \ NFSATTRBM_NAMEDATTR | \ NFSATTRBM_FSID | \ NFSATTRBM_UNIQUEHANDLES | \ NFSATTRBM_LEASETIME | \ NFSATTRBM_RDATTRERROR | \ NFSATTRBM_ACL | \ NFSATTRBM_ACLSUPPORT | \ NFSATTRBM_CANSETTIME | \ NFSATTRBM_CASEINSENSITIVE | \ NFSATTRBM_CASEPRESERVING | \ NFSATTRBM_CHOWNRESTRICTED | \ NFSATTRBM_FILEHANDLE | \ NFSATTRBM_FILEID | \ NFSATTRBM_FILESAVAIL | \ NFSATTRBM_FILESFREE | \ NFSATTRBM_FILESTOTAL | \ NFSATTRBM_FSLOCATIONS | \ NFSATTRBM_HOMOGENEOUS | \ NFSATTRBM_MAXFILESIZE | \ NFSATTRBM_MAXLINK | \ NFSATTRBM_MAXNAME | \ NFSATTRBM_MAXREAD | \ NFSATTRBM_MAXWRITE) /* * NFSATTRBIT_S1 - subset of SUPP1 - OR of the following bits: */ #define NFSATTRBIT_S1 \ (NFSATTRBM_MODE | \ NFSATTRBM_NOTRUNC | \ NFSATTRBM_NUMLINKS | \ NFSATTRBM_OWNER | \ NFSATTRBM_OWNERGROUP | \ NFSATTRBM_RAWDEV | \ NFSATTRBM_SPACEAVAIL | \ NFSATTRBM_SPACEFREE | \ NFSATTRBM_SPACETOTAL | \ NFSATTRBM_SPACEUSED | \ NFSATTRBM_TIMEACCESS | \ NFSATTRBM_TIMEDELTA | \ NFSATTRBM_TIMEMETADATA | \ NFSATTRBM_TIMEMODIFY | \ NFSATTRBM_MOUNTEDONFILEID) #ifdef QUOTA /* * If QUOTA OR in NFSATTRBIT_QUOTAHARD, NFSATTRBIT_QUOTASOFT and * NFSATTRBIT_QUOTAUSED. */ #define NFSATTRBIT_SUPP1 (NFSATTRBIT_S1 | \ NFSATTRBM_QUOTAHARD | \ NFSATTRBM_QUOTASOFT | \ NFSATTRBM_QUOTAUSED) #else #define NFSATTRBIT_SUPP1 NFSATTRBIT_S1 #endif /* * NFSATTRBIT_SUPPSETONLY is the OR of NFSATTRBIT_TIMEACCESSSET and * NFSATTRBIT_TIMEMODIFYSET. */ #define NFSATTRBIT_SUPPSETONLY (NFSATTRBM_TIMEACCESSSET | \ NFSATTRBM_TIMEMODIFYSET) /* * NFSATTRBIT_SETABLE - SETABLE0 - bits 0<->31 * SETABLE1 - bits 32<->63 */ #define NFSATTRBIT_SETABLE0 \ (NFSATTRBM_SIZE | \ NFSATTRBM_ACL) #define NFSATTRBIT_SETABLE1 \ (NFSATTRBM_MODE | \ NFSATTRBM_OWNER | \ NFSATTRBM_OWNERGROUP | \ NFSATTRBM_TIMEACCESSSET | \ NFSATTRBM_TIMEMODIFYSET) /* * Set of attributes that the getattr vnode op needs. * OR of the following bits. * NFSATTRBIT_GETATTR0 - bits 0<->31 */ #define NFSATTRBIT_GETATTR0 \ (NFSATTRBM_SUPPORTEDATTRS | \ NFSATTRBM_TYPE | \ NFSATTRBM_CHANGE | \ NFSATTRBM_SIZE | \ NFSATTRBM_FSID | \ NFSATTRBM_FILEID | \ NFSATTRBM_MAXREAD) /* * NFSATTRBIT_GETATTR1 - bits 32<->63 */ #define NFSATTRBIT_GETATTR1 \ (NFSATTRBM_MODE | \ NFSATTRBM_NUMLINKS | \ NFSATTRBM_OWNER | \ NFSATTRBM_OWNERGROUP | \ NFSATTRBM_RAWDEV | \ NFSATTRBM_SPACEUSED | \ NFSATTRBM_TIMEACCESS | \ NFSATTRBM_TIMEMETADATA | \ NFSATTRBM_TIMEMODIFY) /* * Subset of the above that the Write RPC gets. * OR of the following bits. * NFSATTRBIT_WRITEGETATTR0 - bits 0<->31 */ #define NFSATTRBIT_WRITEGETATTR0 \ (NFSATTRBM_CHANGE | \ NFSATTRBM_SIZE | \ NFSATTRBM_FSID) /* * NFSATTRBIT_WRITEGETATTR1 - bits 32<->63 */ #define NFSATTRBIT_WRITEGETATTR1 \ (NFSATTRBM_TIMEMETADATA | \ NFSATTRBM_TIMEMODIFY) /* * Set of attributes that the wccattr operation op needs. * OR of the following bits. * NFSATTRBIT_WCCATTR0 - bits 0<->31 */ #define NFSATTRBIT_WCCATTR0 0 /* * NFSATTRBIT_WCCATTR1 - bits 32<->63 */ #define NFSATTRBIT_WCCATTR1 \ (NFSATTRBM_TIMEMODIFY) /* * NFSATTRBIT_CBGETATTR0 - bits 0<->31 */ #define NFSATTRBIT_CBGETATTR0 (NFSATTRBM_CHANGE | NFSATTRBM_SIZE) /* * NFSATTRBIT_CBGETATTR1 - bits 32<->63 */ #define NFSATTRBIT_CBGETATTR1 0x0 /* * Sets of attributes that require a VFS_STATFS() call to get the * values of. * NFSATTRBIT_STATFS0 - bits 0<->31 */ #define NFSATTRBIT_STATFS0 \ (NFSATTRBM_LINKSUPPORT | \ NFSATTRBM_SYMLINKSUPPORT | \ NFSATTRBM_CANSETTIME | \ NFSATTRBM_FILESAVAIL | \ NFSATTRBM_FILESFREE | \ NFSATTRBM_FILESTOTAL | \ NFSATTRBM_HOMOGENEOUS | \ NFSATTRBM_MAXFILESIZE | \ NFSATTRBM_MAXNAME | \ NFSATTRBM_MAXREAD | \ NFSATTRBM_MAXWRITE) /* * NFSATTRBIT_STATFS1 - bits 32<->63 */ #define NFSATTRBIT_STATFS1 \ (NFSATTRBM_QUOTAHARD | \ NFSATTRBM_QUOTASOFT | \ NFSATTRBM_QUOTAUSED | \ NFSATTRBM_SPACEAVAIL | \ NFSATTRBM_SPACEFREE | \ NFSATTRBM_SPACETOTAL | \ NFSATTRBM_SPACEUSED | \ NFSATTRBM_TIMEDELTA) /* * These are the bits that are needed by the nfs_statfs() call. * (The regular getattr bits are or'd in so the vnode gets the correct * type, etc.) * NFSGETATTRBIT_STATFS0 - bits 0<->31 */ #define NFSGETATTRBIT_STATFS0 (NFSATTRBIT_GETATTR0 | \ NFSATTRBM_LINKSUPPORT | \ NFSATTRBM_SYMLINKSUPPORT | \ NFSATTRBM_CANSETTIME | \ NFSATTRBM_FILESFREE | \ NFSATTRBM_FILESTOTAL | \ NFSATTRBM_HOMOGENEOUS | \ NFSATTRBM_MAXFILESIZE | \ NFSATTRBM_MAXNAME | \ NFSATTRBM_MAXREAD | \ NFSATTRBM_MAXWRITE) /* * NFSGETATTRBIT_STATFS1 - bits 32<->63 */ #define NFSGETATTRBIT_STATFS1 (NFSATTRBIT_GETATTR1 | \ NFSATTRBM_SPACEAVAIL | \ NFSATTRBM_SPACEFREE | \ NFSATTRBM_SPACETOTAL | \ NFSATTRBM_TIMEDELTA) /* * Set of attributes for the equivalent of an nfsv3 pathconf rpc. * NFSGETATTRBIT_PATHCONF0 - bits 0<->31 */ #define NFSGETATTRBIT_PATHCONF0 (NFSATTRBIT_GETATTR0 | \ NFSATTRBM_CASEINSENSITIVE | \ NFSATTRBM_CASEPRESERVING | \ NFSATTRBM_CHOWNRESTRICTED | \ NFSATTRBM_MAXLINK | \ NFSATTRBM_MAXNAME) /* * NFSGETATTRBIT_PATHCONF1 - bits 32<->63 */ #define NFSGETATTRBIT_PATHCONF1 (NFSATTRBIT_GETATTR1 | \ NFSATTRBM_NOTRUNC) /* * Sets of attributes required by readdir and readdirplus. * NFSATTRBIT_READDIRPLUS0 (NFSATTRBIT_GETATTR0 | NFSATTRBIT_FILEHANDLE | * NFSATTRBIT_RDATTRERROR) */ #define NFSATTRBIT_READDIRPLUS0 (NFSATTRBIT_GETATTR0 | NFSATTRBM_FILEHANDLE | \ NFSATTRBM_RDATTRERROR) #define NFSATTRBIT_READDIRPLUS1 NFSATTRBIT_GETATTR1 /* * Set of attributes supported by Referral vnodes. */ #define NFSATTRBIT_REFERRAL0 (NFSATTRBM_TYPE | NFSATTRBM_FSID | \ NFSATTRBM_RDATTRERROR | NFSATTRBM_FSLOCATIONS) #define NFSATTRBIT_REFERRAL1 NFSATTRBM_MOUNTEDONFILEID /* * Structure for data handled by the statfs rpc. Since some fields are * u_int64_t, this cannot be used for copying data on/off the wire, due * to alignment concerns. */ struct nfsstatfs { union { struct { u_int32_t nfsv2sf_tsize; u_int32_t nfsv2sf_bsize; u_int32_t nfsv2sf_blocks; u_int32_t nfsv2sf_bfree; u_int32_t nfsv2sf_bavail; } sf_nfsv2; struct { u_int64_t nfsv3sf_tbytes; u_int64_t nfsv3sf_fbytes; u_int64_t nfsv3sf_abytes; u_int64_t nfsv3sf_tfiles; u_int64_t nfsv3sf_ffiles; u_int64_t nfsv3sf_afiles; u_int32_t nfsv3sf_invarsec; } sf_nfsv3; } sf_un; }; #define sf_tsize sf_un.sf_nfsv2.nfsv2sf_tsize #define sf_bsize sf_un.sf_nfsv2.nfsv2sf_bsize #define sf_blocks sf_un.sf_nfsv2.nfsv2sf_blocks #define sf_bfree sf_un.sf_nfsv2.nfsv2sf_bfree #define sf_bavail sf_un.sf_nfsv2.nfsv2sf_bavail #define sf_tbytes sf_un.sf_nfsv3.nfsv3sf_tbytes #define sf_fbytes sf_un.sf_nfsv3.nfsv3sf_fbytes #define sf_abytes sf_un.sf_nfsv3.nfsv3sf_abytes #define sf_tfiles sf_un.sf_nfsv3.nfsv3sf_tfiles #define sf_ffiles sf_un.sf_nfsv3.nfsv3sf_ffiles #define sf_afiles sf_un.sf_nfsv3.nfsv3sf_afiles #define sf_invarsec sf_un.sf_nfsv3.nfsv3sf_invarsec /* * Now defined using u_int64_t for the 64 bit field(s). * (Cannot be used to move data on/off the wire, due to alignment concerns.) */ struct nfsfsinfo { u_int32_t fs_rtmax; u_int32_t fs_rtpref; u_int32_t fs_rtmult; u_int32_t fs_wtmax; u_int32_t fs_wtpref; u_int32_t fs_wtmult; u_int32_t fs_dtpref; u_int64_t fs_maxfilesize; struct timespec fs_timedelta; u_int32_t fs_properties; }; /* * Bits for fs_properties */ #define NFSV3_FSFLINK 0x1 #define NFSV3_FSFSYMLINK 0x2 #define NFSV3_FSFHOMOGENEOUS 0x4 #define NFSV3_FSFCANSETTIME 0x8 /* * Yikes, overload fs_rtmult as fs_maxname for V4. */ #define fs_maxname fs_rtmult struct nfsv3_pathconf { u_int32_t pc_linkmax; u_int32_t pc_namemax; u_int32_t pc_notrunc; u_int32_t pc_chownrestricted; u_int32_t pc_caseinsensitive; u_int32_t pc_casepreserving; }; /* * NFS V4 data structures. */ struct nfsv4stateid { u_int32_t seqid; u_int32_t other[NFSX_STATEIDOTHER / NFSX_UNSIGNED]; }; typedef struct nfsv4stateid nfsv4stateid_t; #endif /* _NFS_NFSPROTO_H_ */ Index: projects/largeSMP/sys/fs/nfsclient/nfs_clport.c =================================================================== --- projects/largeSMP/sys/fs/nfsclient/nfs_clport.c (revision 221494) +++ projects/largeSMP/sys/fs/nfsclient/nfs_clport.c (revision 221495) @@ -1,1284 +1,1296 @@ /*- * Copyright (c) 1989, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Rick Macklem at The University of Guelph. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * */ #include __FBSDID("$FreeBSD$"); /* * generally, I don't like #includes inside .h files, but it seems to * be the easiest way to handle the port. */ #include #include #include #include extern u_int32_t newnfs_true, newnfs_false, newnfs_xdrneg1; extern struct vop_vector newnfs_vnodeops; extern struct vop_vector newnfs_fifoops; extern uma_zone_t newnfsnode_zone; extern struct buf_ops buf_ops_newnfs; extern int ncl_pbuf_freecnt; extern short nfsv4_cbport; extern int nfscl_enablecallb; extern int nfs_numnfscbd; extern int nfscl_inited; struct mtx nfs_clstate_mutex; struct mtx ncl_iod_mutex; NFSDLOCKMUTEX; extern void (*ncl_call_invalcaches)(struct vnode *); /* * Comparison function for vfs_hash functions. */ int newnfs_vncmpf(struct vnode *vp, void *arg) { struct nfsfh *nfhp = (struct nfsfh *)arg; struct nfsnode *np = VTONFS(vp); if (np->n_fhp->nfh_len != nfhp->nfh_len || NFSBCMP(np->n_fhp->nfh_fh, nfhp->nfh_fh, nfhp->nfh_len)) return (1); return (0); } /* * Look up a vnode/nfsnode by file handle. * Callers must check for mount points!! * In all cases, a pointer to a * nfsnode structure is returned. * This variant takes a "struct nfsfh *" as second argument and uses * that structure up, either by hanging off the nfsnode or FREEing it. */ int nfscl_nget(struct mount *mntp, struct vnode *dvp, struct nfsfh *nfhp, struct componentname *cnp, struct thread *td, struct nfsnode **npp, void *stuff, int lkflags) { struct nfsnode *np, *dnp; struct vnode *vp, *nvp; struct nfsv4node *newd, *oldd; int error; u_int hash; struct nfsmount *nmp; nmp = VFSTONFS(mntp); dnp = VTONFS(dvp); *npp = NULL; hash = fnv_32_buf(nfhp->nfh_fh, nfhp->nfh_len, FNV1_32_INIT); error = vfs_hash_get(mntp, hash, lkflags, td, &nvp, newnfs_vncmpf, nfhp); if (error == 0 && nvp != NULL) { /* * I believe there is a slight chance that vgonel() could * get called on this vnode between when vn_lock() drops * the VI_LOCK() and vget() acquires it again, so that it * hasn't yet had v_usecount incremented. If this were to * happen, the VI_DOOMED flag would be set, so check for * that here. Since we now have the v_usecount incremented, * we should be ok until we vrele() it, if the VI_DOOMED * flag isn't set now. */ VI_LOCK(nvp); if ((nvp->v_iflag & VI_DOOMED)) { VI_UNLOCK(nvp); vrele(nvp); error = ENOENT; } else { VI_UNLOCK(nvp); } } if (error) { FREE((caddr_t)nfhp, M_NFSFH); return (error); } if (nvp != NULL) { np = VTONFS(nvp); /* * For NFSv4, check to see if it is the same name and * replace the name, if it is different. */ oldd = newd = NULL; if ((nmp->nm_flag & NFSMNT_NFSV4) && np->n_v4 != NULL && nvp->v_type == VREG && (np->n_v4->n4_namelen != cnp->cn_namelen || NFSBCMP(cnp->cn_nameptr, NFS4NODENAME(np->n_v4), cnp->cn_namelen) || dnp->n_fhp->nfh_len != np->n_v4->n4_fhlen || NFSBCMP(dnp->n_fhp->nfh_fh, np->n_v4->n4_data, dnp->n_fhp->nfh_len))) { MALLOC(newd, struct nfsv4node *, sizeof (struct nfsv4node) + dnp->n_fhp->nfh_len + + cnp->cn_namelen - 1, M_NFSV4NODE, M_WAITOK); NFSLOCKNODE(np); if (newd != NULL && np->n_v4 != NULL && nvp->v_type == VREG && (np->n_v4->n4_namelen != cnp->cn_namelen || NFSBCMP(cnp->cn_nameptr, NFS4NODENAME(np->n_v4), cnp->cn_namelen) || dnp->n_fhp->nfh_len != np->n_v4->n4_fhlen || NFSBCMP(dnp->n_fhp->nfh_fh, np->n_v4->n4_data, dnp->n_fhp->nfh_len))) { oldd = np->n_v4; np->n_v4 = newd; newd = NULL; np->n_v4->n4_fhlen = dnp->n_fhp->nfh_len; np->n_v4->n4_namelen = cnp->cn_namelen; NFSBCOPY(dnp->n_fhp->nfh_fh, np->n_v4->n4_data, dnp->n_fhp->nfh_len); NFSBCOPY(cnp->cn_nameptr, NFS4NODENAME(np->n_v4), cnp->cn_namelen); } NFSUNLOCKNODE(np); } if (newd != NULL) FREE((caddr_t)newd, M_NFSV4NODE); if (oldd != NULL) FREE((caddr_t)oldd, M_NFSV4NODE); *npp = np; FREE((caddr_t)nfhp, M_NFSFH); return (0); } /* * Allocate before getnewvnode since doing so afterward * might cause a bogus v_data pointer to get dereferenced * elsewhere if zalloc should block. */ np = uma_zalloc(newnfsnode_zone, M_WAITOK | M_ZERO); error = getnewvnode("newnfs", mntp, &newnfs_vnodeops, &nvp); if (error) { uma_zfree(newnfsnode_zone, np); FREE((caddr_t)nfhp, M_NFSFH); return (error); } vp = nvp; vp->v_bufobj.bo_ops = &buf_ops_newnfs; vp->v_data = np; np->n_vnode = vp; /* * Initialize the mutex even if the vnode is going to be a loser. * This simplifies the logic in reclaim, which can then unconditionally * destroy the mutex (in the case of the loser, or if hash_insert * happened to return an error no special casing is needed). */ mtx_init(&np->n_mtx, "NEWNFSnode lock", NULL, MTX_DEF | MTX_DUPOK); /* * Are we getting the root? If so, make sure the vnode flags * are correct */ if ((nfhp->nfh_len == nmp->nm_fhsize) && !bcmp(nfhp->nfh_fh, nmp->nm_fh, nfhp->nfh_len)) { if (vp->v_type == VNON) vp->v_type = VDIR; vp->v_vflag |= VV_ROOT; } np->n_fhp = nfhp; /* * For NFSv4, we have to attach the directory file handle and * file name, so that Open Ops can be done later. */ if (nmp->nm_flag & NFSMNT_NFSV4) { MALLOC(np->n_v4, struct nfsv4node *, sizeof (struct nfsv4node) + dnp->n_fhp->nfh_len + cnp->cn_namelen - 1, M_NFSV4NODE, M_WAITOK); np->n_v4->n4_fhlen = dnp->n_fhp->nfh_len; np->n_v4->n4_namelen = cnp->cn_namelen; NFSBCOPY(dnp->n_fhp->nfh_fh, np->n_v4->n4_data, dnp->n_fhp->nfh_len); NFSBCOPY(cnp->cn_nameptr, NFS4NODENAME(np->n_v4), cnp->cn_namelen); } else { np->n_v4 = NULL; } /* * NFS supports recursive and shared locking. */ lockmgr(vp->v_vnlock, LK_EXCLUSIVE | LK_NOWITNESS, NULL); VN_LOCK_AREC(vp); VN_LOCK_ASHARE(vp); error = insmntque(vp, mntp); if (error != 0) { *npp = NULL; mtx_destroy(&np->n_mtx); FREE((caddr_t)nfhp, M_NFSFH); if (np->n_v4 != NULL) FREE((caddr_t)np->n_v4, M_NFSV4NODE); uma_zfree(newnfsnode_zone, np); return (error); } error = vfs_hash_insert(vp, hash, lkflags, td, &nvp, newnfs_vncmpf, nfhp); if (error) return (error); if (nvp != NULL) { *npp = VTONFS(nvp); /* vfs_hash_insert() vput()'s the losing vnode */ return (0); } *npp = np; return (0); } /* * Anothe variant of nfs_nget(). This one is only used by reopen. It * takes almost the same args as nfs_nget(), but only succeeds if an entry * exists in the cache. (Since files should already be "open" with a * vnode ref cnt on the node when reopen calls this, it should always * succeed.) * Also, don't get a vnode lock, since it may already be locked by some * other process that is handling it. This is ok, since all other threads * on the client are blocked by the nfsc_lock being exclusively held by the * caller of this function. */ int nfscl_ngetreopen(struct mount *mntp, u_int8_t *fhp, int fhsize, struct thread *td, struct nfsnode **npp) { struct vnode *nvp; u_int hash; struct nfsfh *nfhp; int error; *npp = NULL; /* For forced dismounts, just return error. */ if ((mntp->mnt_kern_flag & MNTK_UNMOUNTF)) return (EINTR); MALLOC(nfhp, struct nfsfh *, sizeof (struct nfsfh) + fhsize, M_NFSFH, M_WAITOK); bcopy(fhp, &nfhp->nfh_fh[0], fhsize); nfhp->nfh_len = fhsize; hash = fnv_32_buf(fhp, fhsize, FNV1_32_INIT); /* * First, try to get the vnode locked, but don't block for the lock. */ error = vfs_hash_get(mntp, hash, (LK_EXCLUSIVE | LK_NOWAIT), td, &nvp, newnfs_vncmpf, nfhp); if (error == 0 && nvp != NULL) { VOP_UNLOCK(nvp, 0); } else if (error == EBUSY) { /* * The LK_EXCLOTHER lock type tells nfs_lock1() to not try * and lock the vnode, but just get a v_usecount on it. * LK_NOWAIT is set so that when vget() returns ENOENT, * vfs_hash_get() fails instead of looping. * If this succeeds, it is safe so long as a vflush() with * FORCECLOSE has not been done. Since the Renew thread is * stopped and the MNTK_UNMOUNTF flag is set before doing * a vflush() with FORCECLOSE, we should be ok here. */ if ((mntp->mnt_kern_flag & MNTK_UNMOUNTF)) error = EINTR; else error = vfs_hash_get(mntp, hash, (LK_EXCLOTHER | LK_NOWAIT), td, &nvp, newnfs_vncmpf, nfhp); } FREE(nfhp, M_NFSFH); if (error) return (error); if (nvp != NULL) { *npp = VTONFS(nvp); return (0); } return (EINVAL); } /* * Load the attribute cache (that lives in the nfsnode entry) with * the attributes of the second argument and * Iff vaper not NULL * copy the attributes to *vaper * Similar to nfs_loadattrcache(), except the attributes are passed in * instead of being parsed out of the mbuf list. */ int nfscl_loadattrcache(struct vnode **vpp, struct nfsvattr *nap, void *nvaper, void *stuff, int writeattr, int dontshrink) { struct vnode *vp = *vpp; struct vattr *vap, *nvap = &nap->na_vattr, *vaper = nvaper; struct nfsnode *np; struct nfsmount *nmp; struct timespec mtime_save; /* * If v_type == VNON it is a new node, so fill in the v_type, * n_mtime fields. Check to see if it represents a special * device, and if so, check for a possible alias. Once the * correct vnode has been obtained, fill in the rest of the * information. */ np = VTONFS(vp); NFSLOCKNODE(np); if (vp->v_type != nvap->va_type) { vp->v_type = nvap->va_type; if (vp->v_type == VFIFO) vp->v_op = &newnfs_fifoops; np->n_mtime = nvap->va_mtime; } nmp = VFSTONFS(vp->v_mount); vap = &np->n_vattr.na_vattr; mtime_save = vap->va_mtime; if (writeattr) { np->n_vattr.na_filerev = nap->na_filerev; np->n_vattr.na_size = nap->na_size; np->n_vattr.na_mtime = nap->na_mtime; np->n_vattr.na_ctime = nap->na_ctime; np->n_vattr.na_fsid = nap->na_fsid; } else { NFSBCOPY((caddr_t)nap, (caddr_t)&np->n_vattr, sizeof (struct nfsvattr)); } /* * For NFSv4, if the node's fsid is not equal to the mount point's * fsid, return the low order 32bits of the node's fsid. This * allows getcwd(3) to work. There is a chance that the fsid might * be the same as a local fs, but since this is in an NFS mount * point, I don't think that will cause any problems? */ if (NFSHASNFSV4(nmp) && NFSHASHASSETFSID(nmp) && (nmp->nm_fsid[0] != np->n_vattr.na_filesid[0] || nmp->nm_fsid[1] != np->n_vattr.na_filesid[1])) { /* * va_fsid needs to be set to some value derived from * np->n_vattr.na_filesid that is not equal * vp->v_mount->mnt_stat.f_fsid[0], so that it changes * from the value used for the top level server volume * in the mounted subtree. */ if (vp->v_mount->mnt_stat.f_fsid.val[0] != (uint32_t)np->n_vattr.na_filesid[0]) vap->va_fsid = (uint32_t)np->n_vattr.na_filesid[0]; else vap->va_fsid = (uint32_t)hash32_buf( np->n_vattr.na_filesid, 2 * sizeof(uint64_t), 0); } else vap->va_fsid = vp->v_mount->mnt_stat.f_fsid.val[0]; np->n_attrstamp = time_second; if (vap->va_size != np->n_size) { if (vap->va_type == VREG) { if (dontshrink && vap->va_size < np->n_size) { /* * We've been told not to shrink the file; * zero np->n_attrstamp to indicate that * the attributes are stale. */ vap->va_size = np->n_size; np->n_attrstamp = 0; } else if (np->n_flag & NMODIFIED) { /* * We've modified the file: Use the larger * of our size, and the server's size. */ if (vap->va_size < np->n_size) { vap->va_size = np->n_size; } else { np->n_size = vap->va_size; np->n_flag |= NSIZECHANGED; } } else { np->n_size = vap->va_size; np->n_flag |= NSIZECHANGED; } vnode_pager_setsize(vp, np->n_size); } else { np->n_size = vap->va_size; } } /* * The following checks are added to prevent a race between (say) * a READDIR+ and a WRITE. * READDIR+, WRITE requests sent out. * READDIR+ resp, WRITE resp received on client. * However, the WRITE resp was handled before the READDIR+ resp * causing the post op attrs from the write to be loaded first * and the attrs from the READDIR+ to be loaded later. If this * happens, we have stale attrs loaded into the attrcache. * We detect this by for the mtime moving back. We invalidate the * attrcache when this happens. */ if (timespeccmp(&mtime_save, &vap->va_mtime, >)) /* Size changed or mtime went backwards */ np->n_attrstamp = 0; if (vaper != NULL) { NFSBCOPY((caddr_t)vap, (caddr_t)vaper, sizeof(*vap)); if (np->n_flag & NCHG) { if (np->n_flag & NACC) vaper->va_atime = np->n_atim; if (np->n_flag & NUPD) vaper->va_mtime = np->n_mtim; } } NFSUNLOCKNODE(np); return (0); } /* * Fill in the client id name. For these bytes: * 1 - they must be unique * 2 - they should be persistent across client reboots * 1 is more critical than 2 * Use the mount point's unique id plus either the uuid or, if that * isn't set, random junk. */ void nfscl_fillclid(u_int64_t clval, char *uuid, u_int8_t *cp, u_int16_t idlen) { int uuidlen; /* * First, put in the 64bit mount point identifier. */ if (idlen >= sizeof (u_int64_t)) { NFSBCOPY((caddr_t)&clval, cp, sizeof (u_int64_t)); cp += sizeof (u_int64_t); idlen -= sizeof (u_int64_t); } /* * If uuid is non-zero length, use it. */ uuidlen = strlen(uuid); if (uuidlen > 0 && idlen >= uuidlen) { NFSBCOPY(uuid, cp, uuidlen); cp += uuidlen; idlen -= uuidlen; } /* * This only normally happens if the uuid isn't set. */ while (idlen > 0) { *cp++ = (u_int8_t)(arc4random() % 256); idlen--; } } /* * Fill in a lock owner name. For now, pid + the process's creation time. */ void nfscl_filllockowner(struct thread *td, u_int8_t *cp) { union { u_int32_t lval; u_int8_t cval[4]; } tl; struct proc *p; if (td == NULL) { printf("NULL td\n"); bzero(cp, 12); return; } p = td->td_proc; if (p == NULL) { printf("NULL pid\n"); bzero(cp, 12); return; } tl.lval = p->p_pid; *cp++ = tl.cval[0]; *cp++ = tl.cval[1]; *cp++ = tl.cval[2]; *cp++ = tl.cval[3]; if (p->p_stats == NULL) { printf("pstats null\n"); bzero(cp, 8); return; } tl.lval = p->p_stats->p_start.tv_sec; *cp++ = tl.cval[0]; *cp++ = tl.cval[1]; *cp++ = tl.cval[2]; *cp++ = tl.cval[3]; tl.lval = p->p_stats->p_start.tv_usec; *cp++ = tl.cval[0]; *cp++ = tl.cval[1]; *cp++ = tl.cval[2]; *cp = tl.cval[3]; } /* * Find the parent process for the thread passed in as an argument. * If none exists, return NULL, otherwise return a thread for the parent. * (Can be any of the threads, since it is only used for td->td_proc.) */ NFSPROC_T * nfscl_getparent(struct thread *td) { struct proc *p; struct thread *ptd; if (td == NULL) return (NULL); p = td->td_proc; if (p->p_pid == 0) return (NULL); p = p->p_pptr; if (p == NULL) return (NULL); ptd = TAILQ_FIRST(&p->p_threads); return (ptd); } /* * Start up the renew kernel thread. */ static void start_nfscl(void *arg) { struct nfsclclient *clp; struct thread *td; clp = (struct nfsclclient *)arg; td = TAILQ_FIRST(&clp->nfsc_renewthread->p_threads); nfscl_renewthread(clp, td); kproc_exit(0); } void nfscl_start_renewthread(struct nfsclclient *clp) { kproc_create(start_nfscl, (void *)clp, &clp->nfsc_renewthread, 0, 0, "nfscl"); } /* * Handle wcc_data. * For NFSv4, it assumes that nfsv4_wccattr() was used to set up the getattr * as the first Op after PutFH. * (For NFSv4, the postop attributes are after the Op, so they can't be * parsed here. A separate call to nfscl_postop_attr() is required.) */ int nfscl_wcc_data(struct nfsrv_descript *nd, struct vnode *vp, struct nfsvattr *nap, int *flagp, int *wccflagp, void *stuff) { u_int32_t *tl; struct nfsnode *np = VTONFS(vp); struct nfsvattr nfsva; int error = 0; if (wccflagp != NULL) *wccflagp = 0; if (nd->nd_flag & ND_NFSV3) { *flagp = 0; NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (*tl == newnfs_true) { NFSM_DISSECT(tl, u_int32_t *, 6 * NFSX_UNSIGNED); if (wccflagp != NULL) { mtx_lock(&np->n_mtx); *wccflagp = (np->n_mtime.tv_sec == fxdr_unsigned(u_int32_t, *(tl + 2)) && np->n_mtime.tv_nsec == fxdr_unsigned(u_int32_t, *(tl + 3))); mtx_unlock(&np->n_mtx); } } error = nfscl_postop_attr(nd, nap, flagp, stuff); } else if ((nd->nd_flag & (ND_NOMOREDATA | ND_NFSV4 | ND_V4WCCATTR)) == (ND_NFSV4 | ND_V4WCCATTR)) { error = nfsv4_loadattr(nd, NULL, &nfsva, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL); if (error) return (error); /* * Get rid of Op# and status for next op. */ NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); if (*++tl) nd->nd_flag |= ND_NOMOREDATA; if (wccflagp != NULL && nfsva.na_vattr.va_mtime.tv_sec != 0) { mtx_lock(&np->n_mtx); *wccflagp = (np->n_mtime.tv_sec == nfsva.na_vattr.va_mtime.tv_sec && np->n_mtime.tv_nsec == nfsva.na_vattr.va_mtime.tv_sec); mtx_unlock(&np->n_mtx); } } nfsmout: return (error); } /* * Get postop attributes. */ int nfscl_postop_attr(struct nfsrv_descript *nd, struct nfsvattr *nap, int *retp, void *stuff) { u_int32_t *tl; int error = 0; *retp = 0; if (nd->nd_flag & ND_NOMOREDATA) return (error); if (nd->nd_flag & ND_NFSV3) { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); *retp = fxdr_unsigned(int, *tl); } else if (nd->nd_flag & ND_NFSV4) { /* * For NFSv4, the postop attr are at the end, so no point * in looking if nd_repstat != 0. */ if (!nd->nd_repstat) { NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); if (*(tl + 1)) /* should never happen since nd_repstat != 0 */ nd->nd_flag |= ND_NOMOREDATA; else *retp = 1; } } else if (!nd->nd_repstat) { /* For NFSv2, the attributes are here iff nd_repstat == 0 */ *retp = 1; } if (*retp) { error = nfsm_loadattr(nd, nap); if (error) *retp = 0; } nfsmout: return (error); } /* * Fill in the setable attributes. The full argument indicates whether * to fill in them all or just mode and time. */ void nfscl_fillsattr(struct nfsrv_descript *nd, struct vattr *vap, struct vnode *vp, int flags, u_int32_t rdev) { u_int32_t *tl; struct nfsv2_sattr *sp; nfsattrbit_t attrbits; struct timeval curtime; switch (nd->nd_flag & (ND_NFSV2 | ND_NFSV3 | ND_NFSV4)) { case ND_NFSV2: NFSM_BUILD(sp, struct nfsv2_sattr *, NFSX_V2SATTR); if (vap->va_mode == (mode_t)VNOVAL) sp->sa_mode = newnfs_xdrneg1; else sp->sa_mode = vtonfsv2_mode(vap->va_type, vap->va_mode); if (vap->va_uid == (uid_t)VNOVAL) sp->sa_uid = newnfs_xdrneg1; else sp->sa_uid = txdr_unsigned(vap->va_uid); if (vap->va_gid == (gid_t)VNOVAL) sp->sa_gid = newnfs_xdrneg1; else sp->sa_gid = txdr_unsigned(vap->va_gid); if (flags & NFSSATTR_SIZE0) sp->sa_size = 0; else if (flags & NFSSATTR_SIZENEG1) sp->sa_size = newnfs_xdrneg1; else if (flags & NFSSATTR_SIZERDEV) sp->sa_size = txdr_unsigned(rdev); else sp->sa_size = txdr_unsigned(vap->va_size); txdr_nfsv2time(&vap->va_atime, &sp->sa_atime); txdr_nfsv2time(&vap->va_mtime, &sp->sa_mtime); break; case ND_NFSV3: getmicrotime(&curtime); if (vap->va_mode != (mode_t)VNOVAL) { NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = newnfs_true; *tl = txdr_unsigned(vap->va_mode); } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_false; } if ((flags & NFSSATTR_FULL) && vap->va_uid != (uid_t)VNOVAL) { NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = newnfs_true; *tl = txdr_unsigned(vap->va_uid); } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_false; } if ((flags & NFSSATTR_FULL) && vap->va_gid != (gid_t)VNOVAL) { NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = newnfs_true; *tl = txdr_unsigned(vap->va_gid); } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_false; } if ((flags & NFSSATTR_FULL) && vap->va_size != VNOVAL) { NFSM_BUILD(tl, u_int32_t *, 3 * NFSX_UNSIGNED); *tl++ = newnfs_true; txdr_hyper(vap->va_size, tl); } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_false; } if (vap->va_atime.tv_sec != VNOVAL) { if (vap->va_atime.tv_sec != curtime.tv_sec) { NFSM_BUILD(tl, u_int32_t *, 3 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV3SATTRTIME_TOCLIENT); txdr_nfsv3time(&vap->va_atime, tl); } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV3SATTRTIME_TOSERVER); } } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV3SATTRTIME_DONTCHANGE); } if (vap->va_mtime.tv_sec != VNOVAL) { if (vap->va_mtime.tv_sec != curtime.tv_sec) { NFSM_BUILD(tl, u_int32_t *, 3 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV3SATTRTIME_TOCLIENT); txdr_nfsv3time(&vap->va_mtime, tl); } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV3SATTRTIME_TOSERVER); } } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV3SATTRTIME_DONTCHANGE); } break; case ND_NFSV4: NFSZERO_ATTRBIT(&attrbits); if (vap->va_mode != (mode_t)VNOVAL) NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_MODE); if ((flags & NFSSATTR_FULL) && vap->va_uid != (uid_t)VNOVAL) NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_OWNER); if ((flags & NFSSATTR_FULL) && vap->va_gid != (gid_t)VNOVAL) NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_OWNERGROUP); if ((flags & NFSSATTR_FULL) && vap->va_size != VNOVAL) NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_SIZE); if (vap->va_atime.tv_sec != VNOVAL) NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_TIMEACCESSSET); if (vap->va_mtime.tv_sec != VNOVAL) NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_TIMEMODIFYSET); (void) nfsv4_fillattr(nd, vp->v_mount, vp, NULL, vap, NULL, 0, &attrbits, NULL, NULL, 0, 0, 0, 0, (uint64_t)0); break; }; } /* * nfscl_request() - mostly a wrapper for newnfs_request(). */ int nfscl_request(struct nfsrv_descript *nd, struct vnode *vp, NFSPROC_T *p, struct ucred *cred, void *stuff) { int ret, vers; struct nfsmount *nmp; nmp = VFSTONFS(vp->v_mount); if (nd->nd_flag & ND_NFSV4) vers = NFS_VER4; else if (nd->nd_flag & ND_NFSV3) vers = NFS_VER3; else vers = NFS_VER2; ret = newnfs_request(nd, nmp, NULL, &nmp->nm_sockreq, vp, p, cred, NFS_PROG, vers, NULL, 1, NULL); return (ret); } /* * fill in this bsden's variant of statfs using nfsstatfs. */ void nfscl_loadsbinfo(struct nfsmount *nmp, struct nfsstatfs *sfp, void *statfs) { struct statfs *sbp = (struct statfs *)statfs; - nfsquad_t tquad; if (nmp->nm_flag & (NFSMNT_NFSV3 | NFSMNT_NFSV4)) { sbp->f_bsize = NFS_FABLKSIZE; - tquad.qval = sfp->sf_tbytes; - sbp->f_blocks = (long)(tquad.qval / ((u_quad_t)NFS_FABLKSIZE)); - tquad.qval = sfp->sf_fbytes; - sbp->f_bfree = (long)(tquad.qval / ((u_quad_t)NFS_FABLKSIZE)); - tquad.qval = sfp->sf_abytes; - sbp->f_bavail = (long)(tquad.qval / ((u_quad_t)NFS_FABLKSIZE)); - tquad.qval = sfp->sf_tfiles; - sbp->f_files = (tquad.lval[0] & 0x7fffffff); - tquad.qval = sfp->sf_ffiles; - sbp->f_ffree = (tquad.lval[0] & 0x7fffffff); + sbp->f_blocks = sfp->sf_tbytes / NFS_FABLKSIZE; + sbp->f_bfree = sfp->sf_fbytes / NFS_FABLKSIZE; + /* + * Although sf_abytes is uint64_t and f_bavail is int64_t, + * the value after dividing by NFS_FABLKSIZE is small + * enough that it will fit in 63bits, so it is ok to + * assign it to f_bavail without fear that it will become + * negative. + */ + sbp->f_bavail = sfp->sf_abytes / NFS_FABLKSIZE; + sbp->f_files = sfp->sf_tfiles; + /* Since f_ffree is int64_t, clip it to 63bits. */ + if (sfp->sf_ffiles > INT64_MAX) + sbp->f_ffree = INT64_MAX; + else + sbp->f_ffree = sfp->sf_ffiles; } else if ((nmp->nm_flag & NFSMNT_NFSV4) == 0) { + /* + * The type casts to (int32_t) ensure that this code is + * compatible with the old NFS client, in that it will + * propagate bit31 to the high order bits. This may or may + * not be correct for NFSv2, but since it is a legacy + * environment, I'd rather retain backwards compatibility. + */ sbp->f_bsize = (int32_t)sfp->sf_bsize; sbp->f_blocks = (int32_t)sfp->sf_blocks; sbp->f_bfree = (int32_t)sfp->sf_bfree; sbp->f_bavail = (int32_t)sfp->sf_bavail; sbp->f_files = 0; sbp->f_ffree = 0; } } /* * Use the fsinfo stuff to update the mount point. */ void nfscl_loadfsinfo(struct nfsmount *nmp, struct nfsfsinfo *fsp) { if ((nmp->nm_wsize == 0 || fsp->fs_wtpref < nmp->nm_wsize) && fsp->fs_wtpref >= NFS_FABLKSIZE) nmp->nm_wsize = (fsp->fs_wtpref + NFS_FABLKSIZE - 1) & ~(NFS_FABLKSIZE - 1); if (fsp->fs_wtmax < nmp->nm_wsize && fsp->fs_wtmax > 0) { nmp->nm_wsize = fsp->fs_wtmax & ~(NFS_FABLKSIZE - 1); if (nmp->nm_wsize == 0) nmp->nm_wsize = fsp->fs_wtmax; } if (nmp->nm_wsize < NFS_FABLKSIZE) nmp->nm_wsize = NFS_FABLKSIZE; if ((nmp->nm_rsize == 0 || fsp->fs_rtpref < nmp->nm_rsize) && fsp->fs_rtpref >= NFS_FABLKSIZE) nmp->nm_rsize = (fsp->fs_rtpref + NFS_FABLKSIZE - 1) & ~(NFS_FABLKSIZE - 1); if (fsp->fs_rtmax < nmp->nm_rsize && fsp->fs_rtmax > 0) { nmp->nm_rsize = fsp->fs_rtmax & ~(NFS_FABLKSIZE - 1); if (nmp->nm_rsize == 0) nmp->nm_rsize = fsp->fs_rtmax; } if (nmp->nm_rsize < NFS_FABLKSIZE) nmp->nm_rsize = NFS_FABLKSIZE; if ((nmp->nm_readdirsize == 0 || fsp->fs_dtpref < nmp->nm_readdirsize) && fsp->fs_dtpref >= NFS_DIRBLKSIZ) nmp->nm_readdirsize = (fsp->fs_dtpref + NFS_DIRBLKSIZ - 1) & ~(NFS_DIRBLKSIZ - 1); if (fsp->fs_rtmax < nmp->nm_readdirsize && fsp->fs_rtmax > 0) { nmp->nm_readdirsize = fsp->fs_rtmax & ~(NFS_DIRBLKSIZ - 1); if (nmp->nm_readdirsize == 0) nmp->nm_readdirsize = fsp->fs_rtmax; } if (nmp->nm_readdirsize < NFS_DIRBLKSIZ) nmp->nm_readdirsize = NFS_DIRBLKSIZ; if (fsp->fs_maxfilesize > 0 && fsp->fs_maxfilesize < nmp->nm_maxfilesize) nmp->nm_maxfilesize = fsp->fs_maxfilesize; nmp->nm_mountp->mnt_stat.f_iosize = newnfs_iosize(nmp); nmp->nm_state |= NFSSTA_GOTFSINFO; } /* * Get a pointer to my IP addrress and return it. * Return NULL if you can't find one. */ u_int8_t * nfscl_getmyip(struct nfsmount *nmp, int *isinet6p) { struct sockaddr_in sad, *sin; struct rtentry *rt; u_int8_t *retp = NULL; static struct in_addr laddr; *isinet6p = 0; /* * Loop up a route for the destination address. */ if (nmp->nm_nam->sa_family == AF_INET) { bzero(&sad, sizeof (sad)); sin = (struct sockaddr_in *)nmp->nm_nam; sad.sin_family = AF_INET; sad.sin_len = sizeof (struct sockaddr_in); sad.sin_addr.s_addr = sin->sin_addr.s_addr; rt = rtalloc1((struct sockaddr *)&sad, 0, 0UL); if (rt != NULL) { if (rt->rt_ifp != NULL && rt->rt_ifa != NULL && ((rt->rt_ifp->if_flags & IFF_LOOPBACK) == 0) && rt->rt_ifa->ifa_addr->sa_family == AF_INET) { sin = (struct sockaddr_in *) rt->rt_ifa->ifa_addr; laddr.s_addr = sin->sin_addr.s_addr; retp = (u_int8_t *)&laddr; } RTFREE_LOCKED(rt); } #ifdef INET6 } else if (nmp->nm_nam->sa_family == AF_INET6) { struct sockaddr_in6 sad6, *sin6; static struct in6_addr laddr6; bzero(&sad6, sizeof (sad6)); sin6 = (struct sockaddr_in6 *)nmp->nm_nam; sad6.sin6_family = AF_INET6; sad6.sin6_len = sizeof (struct sockaddr_in6); sad6.sin6_addr = sin6->sin6_addr; rt = rtalloc1((struct sockaddr *)&sad6, 0, 0UL); if (rt != NULL) { if (rt->rt_ifp != NULL && rt->rt_ifa != NULL && ((rt->rt_ifp->if_flags & IFF_LOOPBACK) == 0) && rt->rt_ifa->ifa_addr->sa_family == AF_INET6) { sin6 = (struct sockaddr_in6 *) rt->rt_ifa->ifa_addr; laddr6 = sin6->sin6_addr; retp = (u_int8_t *)&laddr6; *isinet6p = 1; } RTFREE_LOCKED(rt); } #endif } return (retp); } /* * Copy NFS uid, gids from the cred structure. */ void newnfs_copyincred(struct ucred *cr, struct nfscred *nfscr) { int i; KASSERT(cr->cr_ngroups >= 0, ("newnfs_copyincred: negative cr_ngroups")); nfscr->nfsc_uid = cr->cr_uid; nfscr->nfsc_ngroups = MIN(cr->cr_ngroups, NFS_MAXGRPS + 1); for (i = 0; i < nfscr->nfsc_ngroups; i++) nfscr->nfsc_groups[i] = cr->cr_groups[i]; } /* * Do any client specific initialization. */ void nfscl_init(void) { static int inited = 0; if (inited) return; inited = 1; nfscl_inited = 1; ncl_pbuf_freecnt = nswbuf / 2 + 1; } /* * Check each of the attributes to be set, to ensure they aren't already * the correct value. Disable setting ones already correct. */ int nfscl_checksattr(struct vattr *vap, struct nfsvattr *nvap) { if (vap->va_mode != (mode_t)VNOVAL) { if (vap->va_mode == nvap->na_mode) vap->va_mode = (mode_t)VNOVAL; } if (vap->va_uid != (uid_t)VNOVAL) { if (vap->va_uid == nvap->na_uid) vap->va_uid = (uid_t)VNOVAL; } if (vap->va_gid != (gid_t)VNOVAL) { if (vap->va_gid == nvap->na_gid) vap->va_gid = (gid_t)VNOVAL; } if (vap->va_size != VNOVAL) { if (vap->va_size == nvap->na_size) vap->va_size = VNOVAL; } /* * We are normally called with only a partially initialized * VAP. Since the NFSv3 spec says that server may use the * file attributes to store the verifier, the spec requires * us to do a SETATTR RPC. FreeBSD servers store the verifier * in atime, but we can't really assume that all servers will * so we ensure that our SETATTR sets both atime and mtime. */ if (vap->va_mtime.tv_sec == VNOVAL) vfs_timestamp(&vap->va_mtime); if (vap->va_atime.tv_sec == VNOVAL) vap->va_atime = vap->va_mtime; return (1); } /* * Map nfsv4 errors to errno.h errors. * The uid and gid arguments are only used for NFSERR_BADOWNER and that * error should only be returned for the Open, Create and Setattr Ops. * As such, most calls can just pass in 0 for those arguments. */ APPLESTATIC int nfscl_maperr(struct thread *td, int error, uid_t uid, gid_t gid) { struct proc *p; if (error < 10000) return (error); if (td != NULL) p = td->td_proc; else p = NULL; switch (error) { case NFSERR_BADOWNER: tprintf(p, LOG_INFO, "No name and/or group mapping for uid,gid:(%d,%d)\n", uid, gid); return (EPERM); case NFSERR_STALECLIENTID: case NFSERR_STALESTATEID: case NFSERR_EXPIRED: case NFSERR_BADSTATEID: printf("nfsv4 recover err returned %d\n", error); return (EIO); case NFSERR_BADHANDLE: case NFSERR_SERVERFAULT: case NFSERR_BADTYPE: case NFSERR_FHEXPIRED: case NFSERR_RESOURCE: case NFSERR_MOVED: case NFSERR_NOFILEHANDLE: case NFSERR_MINORVERMISMATCH: case NFSERR_OLDSTATEID: case NFSERR_BADSEQID: case NFSERR_LEASEMOVED: case NFSERR_RECLAIMBAD: case NFSERR_BADXDR: case NFSERR_BADCHAR: case NFSERR_BADNAME: case NFSERR_OPILLEGAL: printf("nfsv4 client/server protocol prob err=%d\n", error); return (EIO); default: tprintf(p, LOG_INFO, "nfsv4 err=%d\n", error); return (EIO); }; } /* * Locate a process by number; return only "live" processes -- i.e., neither * zombies nor newly born but incompletely initialized processes. By not * returning processes in the PRS_NEW state, we allow callers to avoid * testing for that condition to avoid dereferencing p_ucred, et al. * Identical to pfind() in kern_proc.c, except it assume the list is * already locked. */ static struct proc * pfind_locked(pid_t pid) { struct proc *p; LIST_FOREACH(p, PIDHASH(pid), p_hash) if (p->p_pid == pid) { PROC_LOCK(p); if (p->p_state == PRS_NEW) { PROC_UNLOCK(p); p = NULL; } break; } return (p); } /* * Check to see if the process for this owner exists. Return 1 if it doesn't * and 0 otherwise. */ int nfscl_procdoesntexist(u_int8_t *own) { union { u_int32_t lval; u_int8_t cval[4]; } tl; struct proc *p; pid_t pid; int ret = 0; tl.cval[0] = *own++; tl.cval[1] = *own++; tl.cval[2] = *own++; tl.cval[3] = *own++; pid = tl.lval; p = pfind_locked(pid); if (p == NULL) return (1); if (p->p_stats == NULL) { PROC_UNLOCK(p); return (0); } tl.cval[0] = *own++; tl.cval[1] = *own++; tl.cval[2] = *own++; tl.cval[3] = *own++; if (tl.lval != p->p_stats->p_start.tv_sec) { ret = 1; } else { tl.cval[0] = *own++; tl.cval[1] = *own++; tl.cval[2] = *own++; tl.cval[3] = *own; if (tl.lval != p->p_stats->p_start.tv_usec) ret = 1; } PROC_UNLOCK(p); return (ret); } /* * - nfs pseudo system call for the client */ /* * MPSAFE */ static int nfssvc_nfscl(struct thread *td, struct nfssvc_args *uap) { struct file *fp; struct nfscbd_args nfscbdarg; struct nfsd_nfscbd_args nfscbdarg2; int error; if (uap->flag & NFSSVC_CBADDSOCK) { error = copyin(uap->argp, (caddr_t)&nfscbdarg, sizeof(nfscbdarg)); if (error) return (error); if ((error = fget(td, nfscbdarg.sock, &fp)) != 0) { return (error); } if (fp->f_type != DTYPE_SOCKET) { fdrop(fp, td); return (EPERM); } error = nfscbd_addsock(fp); fdrop(fp, td); if (!error && nfscl_enablecallb == 0) { nfsv4_cbport = nfscbdarg.port; nfscl_enablecallb = 1; } } else if (uap->flag & NFSSVC_NFSCBD) { if (uap->argp == NULL) return (EINVAL); error = copyin(uap->argp, (caddr_t)&nfscbdarg2, sizeof(nfscbdarg2)); if (error) return (error); error = nfscbd_nfsd(td, &nfscbdarg2); } else { error = EINVAL; } return (error); } extern int (*nfsd_call_nfscl)(struct thread *, struct nfssvc_args *); /* * Called once to initialize data structures... */ static int nfscl_modevent(module_t mod, int type, void *data) { int error = 0; static int loaded = 0; switch (type) { case MOD_LOAD: if (loaded) return (0); newnfs_portinit(); mtx_init(&nfs_clstate_mutex, "nfs_clstate_mutex", NULL, MTX_DEF); mtx_init(&ncl_iod_mutex, "ncl_iod_mutex", NULL, MTX_DEF); nfscl_init(); NFSD_LOCK(); nfsrvd_cbinit(0); NFSD_UNLOCK(); ncl_call_invalcaches = ncl_invalcaches; nfsd_call_nfscl = nfssvc_nfscl; loaded = 1; break; case MOD_UNLOAD: if (nfs_numnfscbd != 0) { error = EBUSY; break; } /* * XXX: Unloading of nfscl module is unsupported. */ #if 0 ncl_call_invalcaches = NULL; nfsd_call_nfscl = NULL; /* and get rid of the mutexes */ mtx_destroy(&nfs_clstate_mutex); mtx_destroy(&ncl_iod_mutex); loaded = 0; break; #else /* FALLTHROUGH */ #endif default: error = EOPNOTSUPP; break; } return error; } static moduledata_t nfscl_mod = { "nfscl", nfscl_modevent, NULL, }; DECLARE_MODULE(nfscl, nfscl_mod, SI_SUB_VFS, SI_ORDER_FIRST); /* So that loader and kldload(2) can find us, wherever we are.. */ MODULE_VERSION(nfscl, 1); MODULE_DEPEND(nfscl, nfscommon, 1, 1, 1); MODULE_DEPEND(nfscl, krpc, 1, 1, 1); MODULE_DEPEND(nfscl, nfssvc, 1, 1, 1); MODULE_DEPEND(nfscl, nfslock, 1, 1, 1); Index: projects/largeSMP/sys/geom/cache/g_cache.c =================================================================== --- projects/largeSMP/sys/geom/cache/g_cache.c (revision 221494) +++ projects/largeSMP/sys/geom/cache/g_cache.c (revision 221495) @@ -1,1039 +1,1014 @@ /*- * Copyright (c) 2006 Ruslan Ermilov * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include FEATURE(geom_cache, "GEOM cache module"); static MALLOC_DEFINE(M_GCACHE, "gcache_data", "GEOM_CACHE Data"); SYSCTL_DECL(_kern_geom); SYSCTL_NODE(_kern_geom, OID_AUTO, cache, CTLFLAG_RW, 0, "GEOM_CACHE stuff"); static u_int g_cache_debug = 0; SYSCTL_UINT(_kern_geom_cache, OID_AUTO, debug, CTLFLAG_RW, &g_cache_debug, 0, "Debug level"); static u_int g_cache_enable = 1; SYSCTL_UINT(_kern_geom_cache, OID_AUTO, enable, CTLFLAG_RW, &g_cache_enable, 0, ""); static u_int g_cache_timeout = 10; SYSCTL_UINT(_kern_geom_cache, OID_AUTO, timeout, CTLFLAG_RW, &g_cache_timeout, 0, ""); static u_int g_cache_idletime = 5; SYSCTL_UINT(_kern_geom_cache, OID_AUTO, idletime, CTLFLAG_RW, &g_cache_idletime, 0, ""); static u_int g_cache_used_lo = 5; static u_int g_cache_used_hi = 20; static int sysctl_handle_pct(SYSCTL_HANDLER_ARGS) { u_int val = *(u_int *)arg1; int error; error = sysctl_handle_int(oidp, &val, 0, req); if (error || !req->newptr) return (error); if (val < 0 || val > 100) return (EINVAL); if ((arg1 == &g_cache_used_lo && val > g_cache_used_hi) || (arg1 == &g_cache_used_hi && g_cache_used_lo > val)) return (EINVAL); *(u_int *)arg1 = val; return (0); } SYSCTL_PROC(_kern_geom_cache, OID_AUTO, used_lo, CTLTYPE_UINT|CTLFLAG_RW, &g_cache_used_lo, 0, sysctl_handle_pct, "IU", ""); SYSCTL_PROC(_kern_geom_cache, OID_AUTO, used_hi, CTLTYPE_UINT|CTLFLAG_RW, &g_cache_used_hi, 0, sysctl_handle_pct, "IU", ""); static int g_cache_destroy(struct g_cache_softc *sc, boolean_t force); static g_ctl_destroy_geom_t g_cache_destroy_geom; static g_taste_t g_cache_taste; static g_ctl_req_t g_cache_config; static g_dumpconf_t g_cache_dumpconf; struct g_class g_cache_class = { .name = G_CACHE_CLASS_NAME, .version = G_VERSION, .ctlreq = g_cache_config, .taste = g_cache_taste, .destroy_geom = g_cache_destroy_geom }; #define OFF2BNO(off, sc) ((off) >> (sc)->sc_bshift) #define BNO2OFF(bno, sc) ((bno) << (sc)->sc_bshift) static struct g_cache_desc * g_cache_alloc(struct g_cache_softc *sc) { struct g_cache_desc *dp; mtx_assert(&sc->sc_mtx, MA_OWNED); if (!TAILQ_EMPTY(&sc->sc_usedlist)) { dp = TAILQ_FIRST(&sc->sc_usedlist); TAILQ_REMOVE(&sc->sc_usedlist, dp, d_used); sc->sc_nused--; dp->d_flags = 0; LIST_REMOVE(dp, d_next); return (dp); } if (sc->sc_nent > sc->sc_maxent) { sc->sc_cachefull++; return (NULL); } dp = malloc(sizeof(*dp), M_GCACHE, M_NOWAIT | M_ZERO); if (dp == NULL) return (NULL); dp->d_data = uma_zalloc(sc->sc_zone, M_NOWAIT); if (dp->d_data == NULL) { free(dp, M_GCACHE); return (NULL); } sc->sc_nent++; return (dp); } static void g_cache_free(struct g_cache_softc *sc, struct g_cache_desc *dp) { mtx_assert(&sc->sc_mtx, MA_OWNED); uma_zfree(sc->sc_zone, dp->d_data); free(dp, M_GCACHE); sc->sc_nent--; } static void g_cache_free_used(struct g_cache_softc *sc) { struct g_cache_desc *dp; u_int n; mtx_assert(&sc->sc_mtx, MA_OWNED); n = g_cache_used_lo * sc->sc_maxent / 100; while (sc->sc_nused > n) { KASSERT(!TAILQ_EMPTY(&sc->sc_usedlist), ("used list empty")); dp = TAILQ_FIRST(&sc->sc_usedlist); TAILQ_REMOVE(&sc->sc_usedlist, dp, d_used); sc->sc_nused--; LIST_REMOVE(dp, d_next); g_cache_free(sc, dp); } } static void g_cache_deliver(struct g_cache_softc *sc, struct bio *bp, struct g_cache_desc *dp, int error) { off_t off1, off, len; mtx_assert(&sc->sc_mtx, MA_OWNED); KASSERT(OFF2BNO(bp->bio_offset, sc) <= dp->d_bno, ("wrong entry")); KASSERT(OFF2BNO(bp->bio_offset + bp->bio_length - 1, sc) >= dp->d_bno, ("wrong entry")); off1 = BNO2OFF(dp->d_bno, sc); off = MAX(bp->bio_offset, off1); len = MIN(bp->bio_offset + bp->bio_length, off1 + sc->sc_bsize) - off; if (bp->bio_error == 0) bp->bio_error = error; if (bp->bio_error == 0) { bcopy(dp->d_data + (off - off1), bp->bio_data + (off - bp->bio_offset), len); } bp->bio_completed += len; KASSERT(bp->bio_completed <= bp->bio_length, ("extra data")); if (bp->bio_completed == bp->bio_length) { if (bp->bio_error != 0) bp->bio_completed = 0; g_io_deliver(bp, bp->bio_error); } if (dp->d_flags & D_FLAG_USED) { TAILQ_REMOVE(&sc->sc_usedlist, dp, d_used); TAILQ_INSERT_TAIL(&sc->sc_usedlist, dp, d_used); } else if (OFF2BNO(off + len, sc) > dp->d_bno) { TAILQ_INSERT_TAIL(&sc->sc_usedlist, dp, d_used); sc->sc_nused++; dp->d_flags |= D_FLAG_USED; } dp->d_atime = time_uptime; } static void g_cache_done(struct bio *bp) { struct g_cache_softc *sc; struct g_cache_desc *dp; struct bio *bp2, *tmpbp; sc = bp->bio_from->geom->softc; KASSERT(G_CACHE_DESC1(bp) == sc, ("corrupt bio_caller in g_cache_done()")); dp = G_CACHE_DESC2(bp); mtx_lock(&sc->sc_mtx); bp2 = dp->d_biolist; while (bp2 != NULL) { KASSERT(G_CACHE_NEXT_BIO1(bp2) == sc, ("corrupt bio_driver in g_cache_done()")); tmpbp = G_CACHE_NEXT_BIO2(bp2); g_cache_deliver(sc, bp2, dp, bp->bio_error); bp2 = tmpbp; } dp->d_biolist = NULL; if (dp->d_flags & D_FLAG_INVALID) { sc->sc_invalid--; g_cache_free(sc, dp); } else if (bp->bio_error) { LIST_REMOVE(dp, d_next); if (dp->d_flags & D_FLAG_USED) { TAILQ_REMOVE(&sc->sc_usedlist, dp, d_used); sc->sc_nused--; } g_cache_free(sc, dp); } mtx_unlock(&sc->sc_mtx); g_destroy_bio(bp); } static struct g_cache_desc * g_cache_lookup(struct g_cache_softc *sc, off_t bno) { struct g_cache_desc *dp; mtx_assert(&sc->sc_mtx, MA_OWNED); LIST_FOREACH(dp, &sc->sc_desclist[G_CACHE_BUCKET(bno)], d_next) if (dp->d_bno == bno) return (dp); return (NULL); } static int g_cache_read(struct g_cache_softc *sc, struct bio *bp) { struct bio *cbp; struct g_cache_desc *dp; mtx_lock(&sc->sc_mtx); dp = g_cache_lookup(sc, OFF2BNO(bp->bio_offset + bp->bio_completed, sc)); if (dp != NULL) { /* Add to waiters list or deliver. */ sc->sc_cachehits++; if (dp->d_biolist != NULL) { G_CACHE_NEXT_BIO1(bp) = sc; G_CACHE_NEXT_BIO2(bp) = dp->d_biolist; dp->d_biolist = bp; } else g_cache_deliver(sc, bp, dp, 0); mtx_unlock(&sc->sc_mtx); return (0); } /* Cache miss. Allocate entry and schedule bio. */ sc->sc_cachemisses++; dp = g_cache_alloc(sc); if (dp == NULL) { mtx_unlock(&sc->sc_mtx); return (ENOMEM); } cbp = g_clone_bio(bp); if (cbp == NULL) { g_cache_free(sc, dp); mtx_unlock(&sc->sc_mtx); return (ENOMEM); } dp->d_bno = OFF2BNO(bp->bio_offset + bp->bio_completed, sc); G_CACHE_NEXT_BIO1(bp) = sc; G_CACHE_NEXT_BIO2(bp) = NULL; dp->d_biolist = bp; LIST_INSERT_HEAD(&sc->sc_desclist[G_CACHE_BUCKET(dp->d_bno)], dp, d_next); mtx_unlock(&sc->sc_mtx); G_CACHE_DESC1(cbp) = sc; G_CACHE_DESC2(cbp) = dp; cbp->bio_done = g_cache_done; cbp->bio_offset = BNO2OFF(dp->d_bno, sc); cbp->bio_data = dp->d_data; cbp->bio_length = sc->sc_bsize; g_io_request(cbp, LIST_FIRST(&bp->bio_to->geom->consumer)); return (0); } static void g_cache_invalidate(struct g_cache_softc *sc, struct bio *bp) { struct g_cache_desc *dp; off_t bno, lim; mtx_lock(&sc->sc_mtx); bno = OFF2BNO(bp->bio_offset, sc); lim = OFF2BNO(bp->bio_offset + bp->bio_length - 1, sc); do { if ((dp = g_cache_lookup(sc, bno)) != NULL) { LIST_REMOVE(dp, d_next); if (dp->d_flags & D_FLAG_USED) { TAILQ_REMOVE(&sc->sc_usedlist, dp, d_used); sc->sc_nused--; } if (dp->d_biolist == NULL) g_cache_free(sc, dp); else { dp->d_flags = D_FLAG_INVALID; sc->sc_invalid++; } } bno++; } while (bno <= lim); mtx_unlock(&sc->sc_mtx); } static void g_cache_start(struct bio *bp) { struct g_cache_softc *sc; struct g_geom *gp; struct g_cache_desc *dp; struct bio *cbp; gp = bp->bio_to->geom; sc = gp->softc; G_CACHE_LOGREQ(bp, "Request received."); switch (bp->bio_cmd) { case BIO_READ: sc->sc_reads++; sc->sc_readbytes += bp->bio_length; if (!g_cache_enable) break; if (bp->bio_offset + bp->bio_length > sc->sc_tail) break; if (OFF2BNO(bp->bio_offset, sc) == OFF2BNO(bp->bio_offset + bp->bio_length - 1, sc)) { sc->sc_cachereads++; sc->sc_cachereadbytes += bp->bio_length; if (g_cache_read(sc, bp) == 0) return; sc->sc_cachereads--; sc->sc_cachereadbytes -= bp->bio_length; break; } else if (OFF2BNO(bp->bio_offset, sc) + 1 == OFF2BNO(bp->bio_offset + bp->bio_length - 1, sc)) { mtx_lock(&sc->sc_mtx); dp = g_cache_lookup(sc, OFF2BNO(bp->bio_offset, sc)); if (dp == NULL || dp->d_biolist != NULL) { mtx_unlock(&sc->sc_mtx); break; } sc->sc_cachereads++; sc->sc_cachereadbytes += bp->bio_length; g_cache_deliver(sc, bp, dp, 0); mtx_unlock(&sc->sc_mtx); if (g_cache_read(sc, bp) == 0) return; sc->sc_cachereads--; sc->sc_cachereadbytes -= bp->bio_length; break; } break; case BIO_WRITE: sc->sc_writes++; sc->sc_wrotebytes += bp->bio_length; g_cache_invalidate(sc, bp); break; } cbp = g_clone_bio(bp); if (cbp == NULL) { g_io_deliver(bp, ENOMEM); return; } cbp->bio_done = g_std_done; G_CACHE_LOGREQ(cbp, "Sending request."); g_io_request(cbp, LIST_FIRST(&gp->consumer)); } static void g_cache_go(void *arg) { struct g_cache_softc *sc = arg; struct g_cache_desc *dp; int i; mtx_assert(&sc->sc_mtx, MA_OWNED); /* Forcibly mark idle ready entries as used. */ for (i = 0; i < G_CACHE_BUCKETS; i++) { LIST_FOREACH(dp, &sc->sc_desclist[i], d_next) { if (dp->d_flags & D_FLAG_USED || dp->d_biolist != NULL || time_uptime - dp->d_atime < g_cache_idletime) continue; TAILQ_INSERT_TAIL(&sc->sc_usedlist, dp, d_used); sc->sc_nused++; dp->d_flags |= D_FLAG_USED; } } /* Keep the number of used entries low. */ if (sc->sc_nused > g_cache_used_hi * sc->sc_maxent / 100) g_cache_free_used(sc); callout_reset(&sc->sc_callout, g_cache_timeout * hz, g_cache_go, sc); } static int g_cache_access(struct g_provider *pp, int dr, int dw, int de) { struct g_geom *gp; struct g_consumer *cp; int error; gp = pp->geom; cp = LIST_FIRST(&gp->consumer); error = g_access(cp, dr, dw, de); return (error); } static void g_cache_orphan(struct g_consumer *cp) { g_topology_assert(); g_cache_destroy(cp->geom->softc, 1); } static struct g_cache_softc * g_cache_find_device(struct g_class *mp, const char *name) { struct g_geom *gp; LIST_FOREACH(gp, &mp->geom, geom) { if (strcmp(gp->name, name) == 0) return (gp->softc); } return (NULL); } static struct g_geom * g_cache_create(struct g_class *mp, struct g_provider *pp, const struct g_cache_metadata *md, u_int type) { struct g_cache_softc *sc; struct g_geom *gp; struct g_provider *newpp; struct g_consumer *cp; u_int bshift; int i; g_topology_assert(); gp = NULL; newpp = NULL; cp = NULL; G_CACHE_DEBUG(1, "Creating device %s.", md->md_name); /* Cache size is minimum 100. */ if (md->md_size < 100) { G_CACHE_DEBUG(0, "Invalid size for device %s.", md->md_name); return (NULL); } /* Block size restrictions. */ bshift = ffs(md->md_bsize) - 1; if (md->md_bsize == 0 || md->md_bsize > MAXPHYS || md->md_bsize != 1 << bshift || (md->md_bsize % pp->sectorsize) != 0) { G_CACHE_DEBUG(0, "Invalid blocksize for provider %s.", pp->name); return (NULL); } /* Check for duplicate unit. */ if (g_cache_find_device(mp, (const char *)&md->md_name) != NULL) { G_CACHE_DEBUG(0, "Provider %s already exists.", md->md_name); return (NULL); } gp = g_new_geomf(mp, md->md_name); - if (gp == NULL) { - G_CACHE_DEBUG(0, "Cannot create geom %s.", md->md_name); - return (NULL); - } - gp->softc = NULL; /* for a moment */ - sc = g_malloc(sizeof(*sc), M_WAITOK | M_ZERO); sc->sc_type = type; sc->sc_bshift = bshift; sc->sc_bsize = 1 << bshift; sc->sc_zone = uma_zcreate("gcache", sc->sc_bsize, NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0); mtx_init(&sc->sc_mtx, "GEOM CACHE mutex", NULL, MTX_DEF); for (i = 0; i < G_CACHE_BUCKETS; i++) LIST_INIT(&sc->sc_desclist[i]); TAILQ_INIT(&sc->sc_usedlist); sc->sc_maxent = md->md_size; callout_init_mtx(&sc->sc_callout, &sc->sc_mtx, 0); gp->softc = sc; sc->sc_geom = gp; gp->start = g_cache_start; gp->orphan = g_cache_orphan; gp->access = g_cache_access; gp->dumpconf = g_cache_dumpconf; newpp = g_new_providerf(gp, "cache/%s", gp->name); - if (newpp == NULL) { - G_CACHE_DEBUG(0, "Cannot create provider cache/%s.", gp->name); - goto fail; - } newpp->sectorsize = pp->sectorsize; newpp->mediasize = pp->mediasize; if (type == G_CACHE_TYPE_AUTOMATIC) newpp->mediasize -= pp->sectorsize; sc->sc_tail = BNO2OFF(OFF2BNO(newpp->mediasize, sc), sc); cp = g_new_consumer(gp); - if (cp == NULL) { - G_CACHE_DEBUG(0, "Cannot create consumer for %s.", gp->name); - goto fail; - } if (g_attach(cp, pp) != 0) { G_CACHE_DEBUG(0, "Cannot attach to provider %s.", pp->name); - goto fail; + g_destroy_consumer(cp); + g_destroy_provider(newpp); + mtx_destroy(&sc->sc_mtx); + g_free(sc); + g_destroy_geom(gp); + return (NULL); } g_error_provider(newpp, 0); G_CACHE_DEBUG(0, "Device %s created.", gp->name); callout_reset(&sc->sc_callout, g_cache_timeout * hz, g_cache_go, sc); return (gp); -fail: - if (cp != NULL) { - if (cp->provider != NULL) - g_detach(cp); - g_destroy_consumer(cp); - } - if (newpp != NULL) - g_destroy_provider(newpp); - if (gp != NULL) { - if (gp->softc != NULL) { - mtx_destroy(&sc->sc_mtx); - g_free(gp->softc); - } - g_destroy_geom(gp); - } - return (NULL); } static int g_cache_destroy(struct g_cache_softc *sc, boolean_t force) { struct g_geom *gp; struct g_provider *pp; struct g_cache_desc *dp, *dp2; int i; g_topology_assert(); if (sc == NULL) return (ENXIO); gp = sc->sc_geom; pp = LIST_FIRST(&gp->provider); if (pp != NULL && (pp->acr != 0 || pp->acw != 0 || pp->ace != 0)) { if (force) { G_CACHE_DEBUG(0, "Device %s is still open, so it " "can't be definitely removed.", pp->name); } else { G_CACHE_DEBUG(1, "Device %s is still open (r%dw%de%d).", pp->name, pp->acr, pp->acw, pp->ace); return (EBUSY); } } else { G_CACHE_DEBUG(0, "Device %s removed.", gp->name); } callout_drain(&sc->sc_callout); mtx_lock(&sc->sc_mtx); for (i = 0; i < G_CACHE_BUCKETS; i++) { dp = LIST_FIRST(&sc->sc_desclist[i]); while (dp != NULL) { dp2 = LIST_NEXT(dp, d_next); g_cache_free(sc, dp); dp = dp2; } } mtx_unlock(&sc->sc_mtx); mtx_destroy(&sc->sc_mtx); uma_zdestroy(sc->sc_zone); g_free(sc); gp->softc = NULL; g_wither_geom(gp, ENXIO); return (0); } static int g_cache_destroy_geom(struct gctl_req *req, struct g_class *mp, struct g_geom *gp) { return (g_cache_destroy(gp->softc, 0)); } static int g_cache_read_metadata(struct g_consumer *cp, struct g_cache_metadata *md) { struct g_provider *pp; u_char *buf; int error; g_topology_assert(); error = g_access(cp, 1, 0, 0); if (error != 0) return (error); pp = cp->provider; g_topology_unlock(); buf = g_read_data(cp, pp->mediasize - pp->sectorsize, pp->sectorsize, &error); g_topology_lock(); g_access(cp, -1, 0, 0); if (buf == NULL) return (error); /* Decode metadata. */ cache_metadata_decode(buf, md); g_free(buf); return (0); } static int g_cache_write_metadata(struct g_consumer *cp, struct g_cache_metadata *md) { struct g_provider *pp; u_char *buf; int error; g_topology_assert(); error = g_access(cp, 0, 1, 0); if (error != 0) return (error); pp = cp->provider; buf = malloc((size_t)pp->sectorsize, M_GCACHE, M_WAITOK | M_ZERO); cache_metadata_encode(md, buf); g_topology_unlock(); error = g_write_data(cp, pp->mediasize - pp->sectorsize, buf, pp->sectorsize); g_topology_lock(); g_access(cp, 0, -1, 0); free(buf, M_GCACHE); return (error); } static struct g_geom * g_cache_taste(struct g_class *mp, struct g_provider *pp, int flags __unused) { struct g_cache_metadata md; struct g_consumer *cp; struct g_geom *gp; int error; g_trace(G_T_TOPOLOGY, "%s(%s, %s)", __func__, mp->name, pp->name); g_topology_assert(); G_CACHE_DEBUG(3, "Tasting %s.", pp->name); gp = g_new_geomf(mp, "cache:taste"); gp->start = g_cache_start; gp->orphan = g_cache_orphan; gp->access = g_cache_access; cp = g_new_consumer(gp); g_attach(cp, pp); error = g_cache_read_metadata(cp, &md); g_detach(cp); g_destroy_consumer(cp); g_destroy_geom(gp); if (error != 0) return (NULL); if (strcmp(md.md_magic, G_CACHE_MAGIC) != 0) return (NULL); if (md.md_version > G_CACHE_VERSION) { printf("geom_cache.ko module is too old to handle %s.\n", pp->name); return (NULL); } if (md.md_provsize != pp->mediasize) return (NULL); gp = g_cache_create(mp, pp, &md, G_CACHE_TYPE_AUTOMATIC); if (gp == NULL) { G_CACHE_DEBUG(0, "Can't create %s.", md.md_name); return (NULL); } return (gp); } static void g_cache_ctl_create(struct gctl_req *req, struct g_class *mp) { struct g_cache_metadata md; struct g_provider *pp; struct g_geom *gp; intmax_t *bsize, *size; const char *name; int *nargs; g_topology_assert(); nargs = gctl_get_paraml(req, "nargs", sizeof(*nargs)); if (nargs == NULL) { gctl_error(req, "No '%s' argument", "nargs"); return; } if (*nargs != 2) { gctl_error(req, "Invalid number of arguments."); return; } strlcpy(md.md_magic, G_CACHE_MAGIC, sizeof(md.md_magic)); md.md_version = G_CACHE_VERSION; name = gctl_get_asciiparam(req, "arg0"); if (name == NULL) { gctl_error(req, "No 'arg0' argument"); return; } strlcpy(md.md_name, name, sizeof(md.md_name)); size = gctl_get_paraml(req, "size", sizeof(*size)); if (size == NULL) { gctl_error(req, "No '%s' argument", "size"); return; } if ((u_int)*size < 100) { gctl_error(req, "Invalid '%s' argument", "size"); return; } md.md_size = (u_int)*size; bsize = gctl_get_paraml(req, "blocksize", sizeof(*bsize)); if (bsize == NULL) { gctl_error(req, "No '%s' argument", "blocksize"); return; } if (*bsize < 0) { gctl_error(req, "Invalid '%s' argument", "blocksize"); return; } md.md_bsize = (u_int)*bsize; /* This field is not important here. */ md.md_provsize = 0; name = gctl_get_asciiparam(req, "arg1"); if (name == NULL) { gctl_error(req, "No 'arg1' argument"); return; } if (strncmp(name, "/dev/", strlen("/dev/")) == 0) name += strlen("/dev/"); pp = g_provider_by_name(name); if (pp == NULL) { G_CACHE_DEBUG(1, "Provider %s is invalid.", name); gctl_error(req, "Provider %s is invalid.", name); return; } gp = g_cache_create(mp, pp, &md, G_CACHE_TYPE_MANUAL); if (gp == NULL) { gctl_error(req, "Can't create %s.", md.md_name); return; } } static void g_cache_ctl_configure(struct gctl_req *req, struct g_class *mp) { struct g_cache_metadata md; struct g_cache_softc *sc; struct g_consumer *cp; intmax_t *bsize, *size; const char *name; int error, *nargs; g_topology_assert(); nargs = gctl_get_paraml(req, "nargs", sizeof(*nargs)); if (nargs == NULL) { gctl_error(req, "No '%s' argument", "nargs"); return; } if (*nargs != 1) { gctl_error(req, "Missing device."); return; } name = gctl_get_asciiparam(req, "arg0"); if (name == NULL) { gctl_error(req, "No 'arg0' argument"); return; } sc = g_cache_find_device(mp, name); if (sc == NULL) { G_CACHE_DEBUG(1, "Device %s is invalid.", name); gctl_error(req, "Device %s is invalid.", name); return; } size = gctl_get_paraml(req, "size", sizeof(*size)); if (size == NULL) { gctl_error(req, "No '%s' argument", "size"); return; } if ((u_int)*size != 0 && (u_int)*size < 100) { gctl_error(req, "Invalid '%s' argument", "size"); return; } if ((u_int)*size != 0) sc->sc_maxent = (u_int)*size; bsize = gctl_get_paraml(req, "blocksize", sizeof(*bsize)); if (bsize == NULL) { gctl_error(req, "No '%s' argument", "blocksize"); return; } if (*bsize < 0) { gctl_error(req, "Invalid '%s' argument", "blocksize"); return; } if (sc->sc_type != G_CACHE_TYPE_AUTOMATIC) return; strlcpy(md.md_name, name, sizeof(md.md_name)); strlcpy(md.md_magic, G_CACHE_MAGIC, sizeof(md.md_magic)); md.md_version = G_CACHE_VERSION; if ((u_int)*size != 0) md.md_size = (u_int)*size; else md.md_size = sc->sc_maxent; if ((u_int)*bsize != 0) md.md_bsize = (u_int)*bsize; else md.md_bsize = sc->sc_bsize; cp = LIST_FIRST(&sc->sc_geom->consumer); md.md_provsize = cp->provider->mediasize; error = g_cache_write_metadata(cp, &md); if (error == 0) G_CACHE_DEBUG(2, "Metadata on %s updated.", cp->provider->name); else G_CACHE_DEBUG(0, "Cannot update metadata on %s (error=%d).", cp->provider->name, error); } static void g_cache_ctl_destroy(struct gctl_req *req, struct g_class *mp) { int *nargs, *force, error, i; struct g_cache_softc *sc; const char *name; char param[16]; g_topology_assert(); nargs = gctl_get_paraml(req, "nargs", sizeof(*nargs)); if (nargs == NULL) { gctl_error(req, "No '%s' argument", "nargs"); return; } if (*nargs <= 0) { gctl_error(req, "Missing device(s)."); return; } force = gctl_get_paraml(req, "force", sizeof(*force)); if (force == NULL) { gctl_error(req, "No 'force' argument"); return; } for (i = 0; i < *nargs; i++) { snprintf(param, sizeof(param), "arg%d", i); name = gctl_get_asciiparam(req, param); if (name == NULL) { gctl_error(req, "No 'arg%d' argument", i); return; } sc = g_cache_find_device(mp, name); if (sc == NULL) { G_CACHE_DEBUG(1, "Device %s is invalid.", name); gctl_error(req, "Device %s is invalid.", name); return; } error = g_cache_destroy(sc, *force); if (error != 0) { gctl_error(req, "Cannot destroy device %s (error=%d).", sc->sc_name, error); return; } } } static void g_cache_ctl_reset(struct gctl_req *req, struct g_class *mp) { struct g_cache_softc *sc; const char *name; char param[16]; int i, *nargs; g_topology_assert(); nargs = gctl_get_paraml(req, "nargs", sizeof(*nargs)); if (nargs == NULL) { gctl_error(req, "No '%s' argument", "nargs"); return; } if (*nargs <= 0) { gctl_error(req, "Missing device(s)."); return; } for (i = 0; i < *nargs; i++) { snprintf(param, sizeof(param), "arg%d", i); name = gctl_get_asciiparam(req, param); if (name == NULL) { gctl_error(req, "No 'arg%d' argument", i); return; } sc = g_cache_find_device(mp, name); if (sc == NULL) { G_CACHE_DEBUG(1, "Device %s is invalid.", name); gctl_error(req, "Device %s is invalid.", name); return; } sc->sc_reads = 0; sc->sc_readbytes = 0; sc->sc_cachereads = 0; sc->sc_cachereadbytes = 0; sc->sc_cachehits = 0; sc->sc_cachemisses = 0; sc->sc_cachefull = 0; sc->sc_writes = 0; sc->sc_wrotebytes = 0; } } static void g_cache_config(struct gctl_req *req, struct g_class *mp, const char *verb) { uint32_t *version; g_topology_assert(); version = gctl_get_paraml(req, "version", sizeof(*version)); if (version == NULL) { gctl_error(req, "No '%s' argument.", "version"); return; } if (*version != G_CACHE_VERSION) { gctl_error(req, "Userland and kernel parts are out of sync."); return; } if (strcmp(verb, "create") == 0) { g_cache_ctl_create(req, mp); return; } else if (strcmp(verb, "configure") == 0) { g_cache_ctl_configure(req, mp); return; } else if (strcmp(verb, "destroy") == 0 || strcmp(verb, "stop") == 0) { g_cache_ctl_destroy(req, mp); return; } else if (strcmp(verb, "reset") == 0) { g_cache_ctl_reset(req, mp); return; } gctl_error(req, "Unknown verb."); } static void g_cache_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp) { struct g_cache_softc *sc; if (pp != NULL || cp != NULL) return; sc = gp->softc; sbuf_printf(sb, "%s%u\n", indent, sc->sc_maxent); sbuf_printf(sb, "%s%u\n", indent, sc->sc_bsize); sbuf_printf(sb, "%s%ju\n", indent, (uintmax_t)sc->sc_tail); sbuf_printf(sb, "%s%u\n", indent, sc->sc_nent); sbuf_printf(sb, "%s%u\n", indent, sc->sc_nused); sbuf_printf(sb, "%s%u\n", indent, sc->sc_invalid); sbuf_printf(sb, "%s%ju\n", indent, sc->sc_reads); sbuf_printf(sb, "%s%ju\n", indent, sc->sc_readbytes); sbuf_printf(sb, "%s%ju\n", indent, sc->sc_cachereads); sbuf_printf(sb, "%s%ju\n", indent, sc->sc_cachereadbytes); sbuf_printf(sb, "%s%ju\n", indent, sc->sc_cachehits); sbuf_printf(sb, "%s%ju\n", indent, sc->sc_cachemisses); sbuf_printf(sb, "%s%ju\n", indent, sc->sc_cachefull); sbuf_printf(sb, "%s%ju\n", indent, sc->sc_writes); sbuf_printf(sb, "%s%ju\n", indent, sc->sc_wrotebytes); } DECLARE_GEOM_CLASS(g_cache_class, g_cache); Index: projects/largeSMP/sys/geom/concat/g_concat.c =================================================================== --- projects/largeSMP/sys/geom/concat/g_concat.c (revision 221494) +++ projects/largeSMP/sys/geom/concat/g_concat.c (revision 221495) @@ -1,966 +1,964 @@ /*- * Copyright (c) 2004-2005 Pawel Jakub Dawidek * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include FEATURE(geom_concat, "GEOM concatenation support"); static MALLOC_DEFINE(M_CONCAT, "concat_data", "GEOM_CONCAT Data"); SYSCTL_DECL(_kern_geom); SYSCTL_NODE(_kern_geom, OID_AUTO, concat, CTLFLAG_RW, 0, "GEOM_CONCAT stuff"); static u_int g_concat_debug = 0; TUNABLE_INT("kern.geom.concat.debug", &g_concat_debug); SYSCTL_UINT(_kern_geom_concat, OID_AUTO, debug, CTLFLAG_RW, &g_concat_debug, 0, "Debug level"); static int g_concat_destroy(struct g_concat_softc *sc, boolean_t force); static int g_concat_destroy_geom(struct gctl_req *req, struct g_class *mp, struct g_geom *gp); static g_taste_t g_concat_taste; static g_ctl_req_t g_concat_config; static g_dumpconf_t g_concat_dumpconf; struct g_class g_concat_class = { .name = G_CONCAT_CLASS_NAME, .version = G_VERSION, .ctlreq = g_concat_config, .taste = g_concat_taste, .destroy_geom = g_concat_destroy_geom }; /* * Greatest Common Divisor. */ static u_int gcd(u_int a, u_int b) { u_int c; while (b != 0) { c = a; a = b; b = (c % b); } return (a); } /* * Least Common Multiple. */ static u_int lcm(u_int a, u_int b) { return ((a * b) / gcd(a, b)); } /* * Return the number of valid disks. */ static u_int g_concat_nvalid(struct g_concat_softc *sc) { u_int i, no; no = 0; for (i = 0; i < sc->sc_ndisks; i++) { if (sc->sc_disks[i].d_consumer != NULL) no++; } return (no); } static void g_concat_remove_disk(struct g_concat_disk *disk) { struct g_consumer *cp; struct g_concat_softc *sc; KASSERT(disk->d_consumer != NULL, ("Non-valid disk in %s.", __func__)); sc = disk->d_softc; cp = disk->d_consumer; G_CONCAT_DEBUG(0, "Disk %s removed from %s.", cp->provider->name, sc->sc_name); disk->d_consumer = NULL; if (sc->sc_provider != NULL) { sc->sc_provider->flags |= G_PF_WITHER; g_orphan_provider(sc->sc_provider, ENXIO); sc->sc_provider = NULL; G_CONCAT_DEBUG(0, "Device %s removed.", sc->sc_name); } if (cp->acr > 0 || cp->acw > 0 || cp->ace > 0) g_access(cp, -cp->acr, -cp->acw, -cp->ace); g_detach(cp); g_destroy_consumer(cp); } static void g_concat_orphan(struct g_consumer *cp) { struct g_concat_softc *sc; struct g_concat_disk *disk; struct g_geom *gp; g_topology_assert(); gp = cp->geom; sc = gp->softc; if (sc == NULL) return; disk = cp->private; if (disk == NULL) /* Possible? */ return; g_concat_remove_disk(disk); /* If there are no valid disks anymore, remove device. */ if (g_concat_nvalid(sc) == 0) g_concat_destroy(sc, 1); } static int g_concat_access(struct g_provider *pp, int dr, int dw, int de) { struct g_consumer *cp1, *cp2; struct g_concat_softc *sc; struct g_geom *gp; int error; gp = pp->geom; sc = gp->softc; if (sc == NULL) { /* * It looks like geom is being withered. * In that case we allow only negative requests. */ KASSERT(dr <= 0 && dw <= 0 && de <= 0, ("Positive access request (device=%s).", pp->name)); if ((pp->acr + dr) == 0 && (pp->acw + dw) == 0 && (pp->ace + de) == 0) { G_CONCAT_DEBUG(0, "Device %s definitely destroyed.", gp->name); } return (0); } /* On first open, grab an extra "exclusive" bit */ if (pp->acr == 0 && pp->acw == 0 && pp->ace == 0) de++; /* ... and let go of it on last close */ if ((pp->acr + dr) == 0 && (pp->acw + dw) == 0 && (pp->ace + de) == 0) de--; error = ENXIO; LIST_FOREACH(cp1, &gp->consumer, consumer) { error = g_access(cp1, dr, dw, de); if (error == 0) continue; /* * If we fail here, backout all previous changes. */ LIST_FOREACH(cp2, &gp->consumer, consumer) { if (cp1 == cp2) return (error); g_access(cp2, -dr, -dw, -de); } /* NOTREACHED */ } return (error); } static void g_concat_kernel_dump(struct bio *bp) { struct g_concat_softc *sc; struct g_concat_disk *disk; struct bio *cbp; struct g_kerneldump *gkd; u_int i; sc = bp->bio_to->geom->softc; gkd = (struct g_kerneldump *)bp->bio_data; for (i = 0; i < sc->sc_ndisks; i++) { if (sc->sc_disks[i].d_start <= gkd->offset && sc->sc_disks[i].d_end > gkd->offset) break; } if (i == sc->sc_ndisks) g_io_deliver(bp, EOPNOTSUPP); disk = &sc->sc_disks[i]; gkd->offset -= disk->d_start; if (gkd->length > disk->d_end - disk->d_start - gkd->offset) gkd->length = disk->d_end - disk->d_start - gkd->offset; cbp = g_clone_bio(bp); if (cbp == NULL) { g_io_deliver(bp, ENOMEM); return; } cbp->bio_done = g_std_done; g_io_request(cbp, disk->d_consumer); G_CONCAT_DEBUG(1, "Kernel dump will go to %s.", disk->d_consumer->provider->name); } static void g_concat_flush(struct g_concat_softc *sc, struct bio *bp) { struct bio_queue_head queue; struct g_consumer *cp; struct bio *cbp; u_int no; bioq_init(&queue); for (no = 0; no < sc->sc_ndisks; no++) { cbp = g_clone_bio(bp); if (cbp == NULL) { for (cbp = bioq_first(&queue); cbp != NULL; cbp = bioq_first(&queue)) { bioq_remove(&queue, cbp); g_destroy_bio(cbp); } if (bp->bio_error == 0) bp->bio_error = ENOMEM; g_io_deliver(bp, bp->bio_error); return; } bioq_insert_tail(&queue, cbp); cbp->bio_done = g_std_done; cbp->bio_caller1 = sc->sc_disks[no].d_consumer; cbp->bio_to = sc->sc_disks[no].d_consumer->provider; } for (cbp = bioq_first(&queue); cbp != NULL; cbp = bioq_first(&queue)) { bioq_remove(&queue, cbp); G_CONCAT_LOGREQ(cbp, "Sending request."); cp = cbp->bio_caller1; cbp->bio_caller1 = NULL; g_io_request(cbp, cp); } } static void g_concat_start(struct bio *bp) { struct bio_queue_head queue; struct g_concat_softc *sc; struct g_concat_disk *disk; struct g_provider *pp; off_t offset, end, length, off, len; struct bio *cbp; char *addr; u_int no; pp = bp->bio_to; sc = pp->geom->softc; /* * If sc == NULL, provider's error should be set and g_concat_start() * should not be called at all. */ KASSERT(sc != NULL, ("Provider's error should be set (error=%d)(device=%s).", bp->bio_to->error, bp->bio_to->name)); G_CONCAT_LOGREQ(bp, "Request received."); switch (bp->bio_cmd) { case BIO_READ: case BIO_WRITE: case BIO_DELETE: break; case BIO_FLUSH: g_concat_flush(sc, bp); return; case BIO_GETATTR: if (strcmp("GEOM::kerneldump", bp->bio_attribute) == 0) { g_concat_kernel_dump(bp); return; } /* To which provider it should be delivered? */ /* FALLTHROUGH */ default: g_io_deliver(bp, EOPNOTSUPP); return; } offset = bp->bio_offset; length = bp->bio_length; addr = bp->bio_data; end = offset + length; bioq_init(&queue); for (no = 0; no < sc->sc_ndisks; no++) { disk = &sc->sc_disks[no]; if (disk->d_end <= offset) continue; if (disk->d_start >= end) break; off = offset - disk->d_start; len = MIN(length, disk->d_end - offset); length -= len; offset += len; cbp = g_clone_bio(bp); if (cbp == NULL) { for (cbp = bioq_first(&queue); cbp != NULL; cbp = bioq_first(&queue)) { bioq_remove(&queue, cbp); g_destroy_bio(cbp); } if (bp->bio_error == 0) bp->bio_error = ENOMEM; g_io_deliver(bp, bp->bio_error); return; } bioq_insert_tail(&queue, cbp); /* * Fill in the component buf structure. */ cbp->bio_done = g_std_done; cbp->bio_offset = off; cbp->bio_data = addr; addr += len; cbp->bio_length = len; cbp->bio_to = disk->d_consumer->provider; cbp->bio_caller1 = disk; if (length == 0) break; } KASSERT(length == 0, ("Length is still greater than 0 (class=%s, name=%s).", bp->bio_to->geom->class->name, bp->bio_to->geom->name)); for (cbp = bioq_first(&queue); cbp != NULL; cbp = bioq_first(&queue)) { bioq_remove(&queue, cbp); G_CONCAT_LOGREQ(cbp, "Sending request."); disk = cbp->bio_caller1; cbp->bio_caller1 = NULL; g_io_request(cbp, disk->d_consumer); } } static void g_concat_check_and_run(struct g_concat_softc *sc) { struct g_concat_disk *disk; struct g_provider *pp; u_int no, sectorsize = 0; off_t start; if (g_concat_nvalid(sc) != sc->sc_ndisks) return; pp = g_new_providerf(sc->sc_geom, "concat/%s", sc->sc_name); start = 0; for (no = 0; no < sc->sc_ndisks; no++) { disk = &sc->sc_disks[no]; disk->d_start = start; disk->d_end = disk->d_start + disk->d_consumer->provider->mediasize; if (sc->sc_type == G_CONCAT_TYPE_AUTOMATIC) disk->d_end -= disk->d_consumer->provider->sectorsize; start = disk->d_end; if (no == 0) sectorsize = disk->d_consumer->provider->sectorsize; else { sectorsize = lcm(sectorsize, disk->d_consumer->provider->sectorsize); } } pp->sectorsize = sectorsize; /* We have sc->sc_disks[sc->sc_ndisks - 1].d_end in 'start'. */ pp->mediasize = start; pp->stripesize = sc->sc_disks[0].d_consumer->provider->stripesize; pp->stripeoffset = sc->sc_disks[0].d_consumer->provider->stripeoffset; sc->sc_provider = pp; g_error_provider(pp, 0); G_CONCAT_DEBUG(0, "Device %s activated.", sc->sc_name); } static int g_concat_read_metadata(struct g_consumer *cp, struct g_concat_metadata *md) { struct g_provider *pp; u_char *buf; int error; g_topology_assert(); error = g_access(cp, 1, 0, 0); if (error != 0) return (error); pp = cp->provider; g_topology_unlock(); buf = g_read_data(cp, pp->mediasize - pp->sectorsize, pp->sectorsize, &error); g_topology_lock(); g_access(cp, -1, 0, 0); if (buf == NULL) return (error); /* Decode metadata. */ concat_metadata_decode(buf, md); g_free(buf); return (0); } /* * Add disk to given device. */ static int g_concat_add_disk(struct g_concat_softc *sc, struct g_provider *pp, u_int no) { struct g_concat_disk *disk; struct g_consumer *cp, *fcp; struct g_geom *gp; int error; /* Metadata corrupted? */ if (no >= sc->sc_ndisks) return (EINVAL); disk = &sc->sc_disks[no]; /* Check if disk is not already attached. */ if (disk->d_consumer != NULL) return (EEXIST); gp = sc->sc_geom; fcp = LIST_FIRST(&gp->consumer); cp = g_new_consumer(gp); error = g_attach(cp, pp); if (error != 0) { g_destroy_consumer(cp); return (error); } if (fcp != NULL && (fcp->acr > 0 || fcp->acw > 0 || fcp->ace > 0)) { error = g_access(cp, fcp->acr, fcp->acw, fcp->ace); if (error != 0) { g_detach(cp); g_destroy_consumer(cp); return (error); } } if (sc->sc_type == G_CONCAT_TYPE_AUTOMATIC) { struct g_concat_metadata md; /* Re-read metadata. */ error = g_concat_read_metadata(cp, &md); if (error != 0) goto fail; if (strcmp(md.md_magic, G_CONCAT_MAGIC) != 0 || strcmp(md.md_name, sc->sc_name) != 0 || md.md_id != sc->sc_id) { G_CONCAT_DEBUG(0, "Metadata on %s changed.", pp->name); goto fail; } } cp->private = disk; disk->d_consumer = cp; disk->d_softc = sc; disk->d_start = 0; /* not yet */ disk->d_end = 0; /* not yet */ G_CONCAT_DEBUG(0, "Disk %s attached to %s.", pp->name, sc->sc_name); g_concat_check_and_run(sc); return (0); fail: if (fcp != NULL && (fcp->acr > 0 || fcp->acw > 0 || fcp->ace > 0)) g_access(cp, -fcp->acr, -fcp->acw, -fcp->ace); g_detach(cp); g_destroy_consumer(cp); return (error); } static struct g_geom * g_concat_create(struct g_class *mp, const struct g_concat_metadata *md, u_int type) { struct g_concat_softc *sc; struct g_geom *gp; u_int no; G_CONCAT_DEBUG(1, "Creating device %s (id=%u).", md->md_name, md->md_id); /* One disks is minimum. */ if (md->md_all < 1) return (NULL); /* Check for duplicate unit */ LIST_FOREACH(gp, &mp->geom, geom) { sc = gp->softc; if (sc != NULL && strcmp(sc->sc_name, md->md_name) == 0) { G_CONCAT_DEBUG(0, "Device %s already configured.", gp->name); return (NULL); } } gp = g_new_geomf(mp, "%s", md->md_name); - gp->softc = NULL; /* for a moment */ - sc = malloc(sizeof(*sc), M_CONCAT, M_WAITOK | M_ZERO); gp->start = g_concat_start; gp->spoiled = g_concat_orphan; gp->orphan = g_concat_orphan; gp->access = g_concat_access; gp->dumpconf = g_concat_dumpconf; sc->sc_id = md->md_id; sc->sc_ndisks = md->md_all; sc->sc_disks = malloc(sizeof(struct g_concat_disk) * sc->sc_ndisks, M_CONCAT, M_WAITOK | M_ZERO); for (no = 0; no < sc->sc_ndisks; no++) sc->sc_disks[no].d_consumer = NULL; sc->sc_type = type; gp->softc = sc; sc->sc_geom = gp; sc->sc_provider = NULL; G_CONCAT_DEBUG(0, "Device %s created (id=%u).", sc->sc_name, sc->sc_id); return (gp); } static int g_concat_destroy(struct g_concat_softc *sc, boolean_t force) { struct g_provider *pp; struct g_geom *gp; u_int no; g_topology_assert(); if (sc == NULL) return (ENXIO); pp = sc->sc_provider; if (pp != NULL && (pp->acr != 0 || pp->acw != 0 || pp->ace != 0)) { if (force) { G_CONCAT_DEBUG(0, "Device %s is still open, so it " "can't be definitely removed.", pp->name); } else { G_CONCAT_DEBUG(1, "Device %s is still open (r%dw%de%d).", pp->name, pp->acr, pp->acw, pp->ace); return (EBUSY); } } for (no = 0; no < sc->sc_ndisks; no++) { if (sc->sc_disks[no].d_consumer != NULL) g_concat_remove_disk(&sc->sc_disks[no]); } gp = sc->sc_geom; gp->softc = NULL; KASSERT(sc->sc_provider == NULL, ("Provider still exists? (device=%s)", gp->name)); free(sc->sc_disks, M_CONCAT); free(sc, M_CONCAT); pp = LIST_FIRST(&gp->provider); if (pp == NULL || (pp->acr == 0 && pp->acw == 0 && pp->ace == 0)) G_CONCAT_DEBUG(0, "Device %s destroyed.", gp->name); g_wither_geom(gp, ENXIO); return (0); } static int g_concat_destroy_geom(struct gctl_req *req __unused, struct g_class *mp __unused, struct g_geom *gp) { struct g_concat_softc *sc; sc = gp->softc; return (g_concat_destroy(sc, 0)); } static struct g_geom * g_concat_taste(struct g_class *mp, struct g_provider *pp, int flags __unused) { struct g_concat_metadata md; struct g_concat_softc *sc; struct g_consumer *cp; struct g_geom *gp; int error; g_trace(G_T_TOPOLOGY, "%s(%s, %s)", __func__, mp->name, pp->name); g_topology_assert(); /* Skip providers that are already open for writing. */ if (pp->acw > 0) return (NULL); G_CONCAT_DEBUG(3, "Tasting %s.", pp->name); gp = g_new_geomf(mp, "concat:taste"); gp->start = g_concat_start; gp->access = g_concat_access; gp->orphan = g_concat_orphan; cp = g_new_consumer(gp); g_attach(cp, pp); error = g_concat_read_metadata(cp, &md); g_detach(cp); g_destroy_consumer(cp); g_destroy_geom(gp); if (error != 0) return (NULL); gp = NULL; if (strcmp(md.md_magic, G_CONCAT_MAGIC) != 0) return (NULL); if (md.md_version > G_CONCAT_VERSION) { printf("geom_concat.ko module is too old to handle %s.\n", pp->name); return (NULL); } /* * Backward compatibility: */ /* There was no md_provider field in earlier versions of metadata. */ if (md.md_version < 3) bzero(md.md_provider, sizeof(md.md_provider)); /* There was no md_provsize field in earlier versions of metadata. */ if (md.md_version < 4) md.md_provsize = pp->mediasize; if (md.md_provider[0] != '\0' && !g_compare_names(md.md_provider, pp->name)) return (NULL); if (md.md_provsize != pp->mediasize) return (NULL); /* * Let's check if device already exists. */ sc = NULL; LIST_FOREACH(gp, &mp->geom, geom) { sc = gp->softc; if (sc == NULL) continue; if (sc->sc_type != G_CONCAT_TYPE_AUTOMATIC) continue; if (strcmp(md.md_name, sc->sc_name) != 0) continue; if (md.md_id != sc->sc_id) continue; break; } if (gp != NULL) { G_CONCAT_DEBUG(1, "Adding disk %s to %s.", pp->name, gp->name); error = g_concat_add_disk(sc, pp, md.md_no); if (error != 0) { G_CONCAT_DEBUG(0, "Cannot add disk %s to %s (error=%d).", pp->name, gp->name, error); return (NULL); } } else { gp = g_concat_create(mp, &md, G_CONCAT_TYPE_AUTOMATIC); if (gp == NULL) { G_CONCAT_DEBUG(0, "Cannot create device %s.", md.md_name); return (NULL); } sc = gp->softc; G_CONCAT_DEBUG(1, "Adding disk %s to %s.", pp->name, gp->name); error = g_concat_add_disk(sc, pp, md.md_no); if (error != 0) { G_CONCAT_DEBUG(0, "Cannot add disk %s to %s (error=%d).", pp->name, gp->name, error); g_concat_destroy(sc, 1); return (NULL); } } return (gp); } static void g_concat_ctl_create(struct gctl_req *req, struct g_class *mp) { u_int attached, no; struct g_concat_metadata md; struct g_provider *pp; struct g_concat_softc *sc; struct g_geom *gp; struct sbuf *sb; const char *name; char param[16]; int *nargs; g_topology_assert(); nargs = gctl_get_paraml(req, "nargs", sizeof(*nargs)); if (nargs == NULL) { gctl_error(req, "No '%s' argument.", "nargs"); return; } if (*nargs < 2) { gctl_error(req, "Too few arguments."); return; } strlcpy(md.md_magic, G_CONCAT_MAGIC, sizeof(md.md_magic)); md.md_version = G_CONCAT_VERSION; name = gctl_get_asciiparam(req, "arg0"); if (name == NULL) { gctl_error(req, "No 'arg%u' argument.", 0); return; } strlcpy(md.md_name, name, sizeof(md.md_name)); md.md_id = arc4random(); md.md_no = 0; md.md_all = *nargs - 1; bzero(md.md_provider, sizeof(md.md_provider)); /* This field is not important here. */ md.md_provsize = 0; /* Check all providers are valid */ for (no = 1; no < *nargs; no++) { snprintf(param, sizeof(param), "arg%u", no); name = gctl_get_asciiparam(req, param); if (name == NULL) { gctl_error(req, "No 'arg%u' argument.", no); return; } if (strncmp(name, "/dev/", strlen("/dev/")) == 0) name += strlen("/dev/"); pp = g_provider_by_name(name); if (pp == NULL) { G_CONCAT_DEBUG(1, "Disk %s is invalid.", name); gctl_error(req, "Disk %s is invalid.", name); return; } } gp = g_concat_create(mp, &md, G_CONCAT_TYPE_MANUAL); if (gp == NULL) { gctl_error(req, "Can't configure %s.", md.md_name); return; } sc = gp->softc; sb = sbuf_new_auto(); sbuf_printf(sb, "Can't attach disk(s) to %s:", gp->name); for (attached = 0, no = 1; no < *nargs; no++) { snprintf(param, sizeof(param), "arg%u", no); name = gctl_get_asciiparam(req, param); if (name == NULL) { gctl_error(req, "No 'arg%d' argument.", no); return; } if (strncmp(name, "/dev/", strlen("/dev/")) == 0) name += strlen("/dev/"); pp = g_provider_by_name(name); KASSERT(pp != NULL, ("Provider %s disappear?!", name)); if (g_concat_add_disk(sc, pp, no - 1) != 0) { G_CONCAT_DEBUG(1, "Disk %u (%s) not attached to %s.", no, pp->name, gp->name); sbuf_printf(sb, " %s", pp->name); continue; } attached++; } sbuf_finish(sb); if (md.md_all != attached) { g_concat_destroy(gp->softc, 1); gctl_error(req, "%s", sbuf_data(sb)); } sbuf_delete(sb); } static struct g_concat_softc * g_concat_find_device(struct g_class *mp, const char *name) { struct g_concat_softc *sc; struct g_geom *gp; LIST_FOREACH(gp, &mp->geom, geom) { sc = gp->softc; if (sc == NULL) continue; if (strcmp(sc->sc_name, name) == 0) return (sc); } return (NULL); } static void g_concat_ctl_destroy(struct gctl_req *req, struct g_class *mp) { struct g_concat_softc *sc; int *force, *nargs, error; const char *name; char param[16]; u_int i; g_topology_assert(); nargs = gctl_get_paraml(req, "nargs", sizeof(*nargs)); if (nargs == NULL) { gctl_error(req, "No '%s' argument.", "nargs"); return; } if (*nargs <= 0) { gctl_error(req, "Missing device(s)."); return; } force = gctl_get_paraml(req, "force", sizeof(*force)); if (force == NULL) { gctl_error(req, "No '%s' argument.", "force"); return; } for (i = 0; i < (u_int)*nargs; i++) { snprintf(param, sizeof(param), "arg%u", i); name = gctl_get_asciiparam(req, param); if (name == NULL) { gctl_error(req, "No 'arg%u' argument.", i); return; } sc = g_concat_find_device(mp, name); if (sc == NULL) { gctl_error(req, "No such device: %s.", name); return; } error = g_concat_destroy(sc, *force); if (error != 0) { gctl_error(req, "Cannot destroy device %s (error=%d).", sc->sc_name, error); return; } } } static void g_concat_config(struct gctl_req *req, struct g_class *mp, const char *verb) { uint32_t *version; g_topology_assert(); version = gctl_get_paraml(req, "version", sizeof(*version)); if (version == NULL) { gctl_error(req, "No '%s' argument.", "version"); return; } if (*version != G_CONCAT_VERSION) { gctl_error(req, "Userland and kernel parts are out of sync."); return; } if (strcmp(verb, "create") == 0) { g_concat_ctl_create(req, mp); return; } else if (strcmp(verb, "destroy") == 0 || strcmp(verb, "stop") == 0) { g_concat_ctl_destroy(req, mp); return; } gctl_error(req, "Unknown verb."); } static void g_concat_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp) { struct g_concat_softc *sc; g_topology_assert(); sc = gp->softc; if (sc == NULL) return; if (pp != NULL) { /* Nothing here. */ } else if (cp != NULL) { struct g_concat_disk *disk; disk = cp->private; if (disk == NULL) return; sbuf_printf(sb, "%s%jd\n", indent, (intmax_t)disk->d_end); sbuf_printf(sb, "%s%jd\n", indent, (intmax_t)disk->d_start); } else { sbuf_printf(sb, "%s%u\n", indent, (u_int)sc->sc_id); sbuf_printf(sb, "%s", indent); switch (sc->sc_type) { case G_CONCAT_TYPE_AUTOMATIC: sbuf_printf(sb, "AUTOMATIC"); break; case G_CONCAT_TYPE_MANUAL: sbuf_printf(sb, "MANUAL"); break; default: sbuf_printf(sb, "UNKNOWN"); break; } sbuf_printf(sb, "\n"); sbuf_printf(sb, "%sTotal=%u, Online=%u\n", indent, sc->sc_ndisks, g_concat_nvalid(sc)); sbuf_printf(sb, "%s", indent); if (sc->sc_provider != NULL && sc->sc_provider->error == 0) sbuf_printf(sb, "UP"); else sbuf_printf(sb, "DOWN"); sbuf_printf(sb, "\n"); } } DECLARE_GEOM_CLASS(g_concat_class, g_concat); Index: projects/largeSMP/sys/geom/eli/g_eli.c =================================================================== --- projects/largeSMP/sys/geom/eli/g_eli.c (revision 221494) +++ projects/largeSMP/sys/geom/eli/g_eli.c (revision 221495) @@ -1,1271 +1,1269 @@ /*- * Copyright (c) 2005-2011 Pawel Jakub Dawidek * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include FEATURE(geom_eli, "GEOM crypto module"); MALLOC_DEFINE(M_ELI, "eli data", "GEOM_ELI Data"); SYSCTL_DECL(_kern_geom); SYSCTL_NODE(_kern_geom, OID_AUTO, eli, CTLFLAG_RW, 0, "GEOM_ELI stuff"); int g_eli_debug = 0; TUNABLE_INT("kern.geom.eli.debug", &g_eli_debug); SYSCTL_INT(_kern_geom_eli, OID_AUTO, debug, CTLFLAG_RW, &g_eli_debug, 0, "Debug level"); static u_int g_eli_tries = 3; TUNABLE_INT("kern.geom.eli.tries", &g_eli_tries); SYSCTL_UINT(_kern_geom_eli, OID_AUTO, tries, CTLFLAG_RW, &g_eli_tries, 0, "Number of tries for entering the passphrase"); static u_int g_eli_visible_passphrase = GETS_NOECHO; TUNABLE_INT("kern.geom.eli.visible_passphrase", &g_eli_visible_passphrase); SYSCTL_UINT(_kern_geom_eli, OID_AUTO, visible_passphrase, CTLFLAG_RW, &g_eli_visible_passphrase, 0, "Visibility of passphrase prompt (0 = invisible, 1 = visible, 2 = asterisk)"); u_int g_eli_overwrites = G_ELI_OVERWRITES; TUNABLE_INT("kern.geom.eli.overwrites", &g_eli_overwrites); SYSCTL_UINT(_kern_geom_eli, OID_AUTO, overwrites, CTLFLAG_RW, &g_eli_overwrites, 0, "Number of times on-disk keys should be overwritten when destroying them"); static u_int g_eli_threads = 0; TUNABLE_INT("kern.geom.eli.threads", &g_eli_threads); SYSCTL_UINT(_kern_geom_eli, OID_AUTO, threads, CTLFLAG_RW, &g_eli_threads, 0, "Number of threads doing crypto work"); u_int g_eli_batch = 0; TUNABLE_INT("kern.geom.eli.batch", &g_eli_batch); SYSCTL_UINT(_kern_geom_eli, OID_AUTO, batch, CTLFLAG_RW, &g_eli_batch, 0, "Use crypto operations batching"); static eventhandler_tag g_eli_pre_sync = NULL; static int g_eli_destroy_geom(struct gctl_req *req, struct g_class *mp, struct g_geom *gp); static void g_eli_init(struct g_class *mp); static void g_eli_fini(struct g_class *mp); static g_taste_t g_eli_taste; static g_dumpconf_t g_eli_dumpconf; struct g_class g_eli_class = { .name = G_ELI_CLASS_NAME, .version = G_VERSION, .ctlreq = g_eli_config, .taste = g_eli_taste, .destroy_geom = g_eli_destroy_geom, .init = g_eli_init, .fini = g_eli_fini }; /* * Code paths: * BIO_READ: * g_eli_start -> g_eli_crypto_read -> g_io_request -> g_eli_read_done -> g_eli_crypto_run -> g_eli_crypto_read_done -> g_io_deliver * BIO_WRITE: * g_eli_start -> g_eli_crypto_run -> g_eli_crypto_write_done -> g_io_request -> g_eli_write_done -> g_io_deliver */ /* * EAGAIN from crypto(9) means, that we were probably balanced to another crypto * accelerator or something like this. * The function updates the SID and rerun the operation. */ int g_eli_crypto_rerun(struct cryptop *crp) { struct g_eli_softc *sc; struct g_eli_worker *wr; struct bio *bp; int error; bp = (struct bio *)crp->crp_opaque; sc = bp->bio_to->geom->softc; LIST_FOREACH(wr, &sc->sc_workers, w_next) { if (wr->w_number == bp->bio_pflags) break; } KASSERT(wr != NULL, ("Invalid worker (%u).", bp->bio_pflags)); G_ELI_DEBUG(1, "Rerunning crypto %s request (sid: %ju -> %ju).", bp->bio_cmd == BIO_READ ? "READ" : "WRITE", (uintmax_t)wr->w_sid, (uintmax_t)crp->crp_sid); wr->w_sid = crp->crp_sid; crp->crp_etype = 0; error = crypto_dispatch(crp); if (error == 0) return (0); G_ELI_DEBUG(1, "%s: crypto_dispatch() returned %d.", __func__, error); crp->crp_etype = error; return (error); } /* * The function is called afer reading encrypted data from the provider. * * g_eli_start -> g_eli_crypto_read -> g_io_request -> G_ELI_READ_DONE -> g_eli_crypto_run -> g_eli_crypto_read_done -> g_io_deliver */ void g_eli_read_done(struct bio *bp) { struct g_eli_softc *sc; struct bio *pbp; G_ELI_LOGREQ(2, bp, "Request done."); pbp = bp->bio_parent; if (pbp->bio_error == 0) pbp->bio_error = bp->bio_error; g_destroy_bio(bp); /* * Do we have all sectors already? */ pbp->bio_inbed++; if (pbp->bio_inbed < pbp->bio_children) return; sc = pbp->bio_to->geom->softc; if (pbp->bio_error != 0) { G_ELI_LOGREQ(0, pbp, "%s() failed", __func__); pbp->bio_completed = 0; if (pbp->bio_driver2 != NULL) { free(pbp->bio_driver2, M_ELI); pbp->bio_driver2 = NULL; } g_io_deliver(pbp, pbp->bio_error); atomic_subtract_int(&sc->sc_inflight, 1); return; } mtx_lock(&sc->sc_queue_mtx); bioq_insert_tail(&sc->sc_queue, pbp); mtx_unlock(&sc->sc_queue_mtx); wakeup(sc); } /* * The function is called after we encrypt and write data. * * g_eli_start -> g_eli_crypto_run -> g_eli_crypto_write_done -> g_io_request -> G_ELI_WRITE_DONE -> g_io_deliver */ void g_eli_write_done(struct bio *bp) { struct g_eli_softc *sc; struct bio *pbp; G_ELI_LOGREQ(2, bp, "Request done."); pbp = bp->bio_parent; if (pbp->bio_error == 0) { if (bp->bio_error != 0) pbp->bio_error = bp->bio_error; } g_destroy_bio(bp); /* * Do we have all sectors already? */ pbp->bio_inbed++; if (pbp->bio_inbed < pbp->bio_children) return; free(pbp->bio_driver2, M_ELI); pbp->bio_driver2 = NULL; if (pbp->bio_error != 0) { G_ELI_LOGREQ(0, pbp, "Crypto WRITE request failed (error=%d).", pbp->bio_error); pbp->bio_completed = 0; } /* * Write is finished, send it up. */ pbp->bio_completed = pbp->bio_length; sc = pbp->bio_to->geom->softc; g_io_deliver(pbp, pbp->bio_error); atomic_subtract_int(&sc->sc_inflight, 1); } /* * This function should never be called, but GEOM made as it set ->orphan() * method for every geom. */ static void g_eli_orphan_spoil_assert(struct g_consumer *cp) { panic("Function %s() called for %s.", __func__, cp->geom->name); } static void g_eli_orphan(struct g_consumer *cp) { struct g_eli_softc *sc; g_topology_assert(); sc = cp->geom->softc; if (sc == NULL) return; g_eli_destroy(sc, TRUE); } /* * BIO_READ: * G_ELI_START -> g_eli_crypto_read -> g_io_request -> g_eli_read_done -> g_eli_crypto_run -> g_eli_crypto_read_done -> g_io_deliver * BIO_WRITE: * G_ELI_START -> g_eli_crypto_run -> g_eli_crypto_write_done -> g_io_request -> g_eli_write_done -> g_io_deliver */ static void g_eli_start(struct bio *bp) { struct g_eli_softc *sc; struct g_consumer *cp; struct bio *cbp; sc = bp->bio_to->geom->softc; KASSERT(sc != NULL, ("Provider's error should be set (error=%d)(device=%s).", bp->bio_to->error, bp->bio_to->name)); G_ELI_LOGREQ(2, bp, "Request received."); switch (bp->bio_cmd) { case BIO_READ: case BIO_WRITE: case BIO_GETATTR: case BIO_FLUSH: break; case BIO_DELETE: /* * We could eventually support BIO_DELETE request. * It could be done by overwritting requested sector with * random data g_eli_overwrites number of times. */ default: g_io_deliver(bp, EOPNOTSUPP); return; } cbp = g_clone_bio(bp); if (cbp == NULL) { g_io_deliver(bp, ENOMEM); return; } bp->bio_driver1 = cbp; bp->bio_pflags = G_ELI_NEW_BIO; switch (bp->bio_cmd) { case BIO_READ: if (!(sc->sc_flags & G_ELI_FLAG_AUTH)) { g_eli_crypto_read(sc, bp, 0); break; } /* FALLTHROUGH */ case BIO_WRITE: mtx_lock(&sc->sc_queue_mtx); bioq_insert_tail(&sc->sc_queue, bp); mtx_unlock(&sc->sc_queue_mtx); wakeup(sc); break; case BIO_GETATTR: case BIO_FLUSH: cbp->bio_done = g_std_done; cp = LIST_FIRST(&sc->sc_geom->consumer); cbp->bio_to = cp->provider; G_ELI_LOGREQ(2, cbp, "Sending request."); g_io_request(cbp, cp); break; } } static int g_eli_newsession(struct g_eli_worker *wr) { struct g_eli_softc *sc; struct cryptoini crie, cria; int error; sc = wr->w_softc; bzero(&crie, sizeof(crie)); crie.cri_alg = sc->sc_ealgo; crie.cri_klen = sc->sc_ekeylen; if (sc->sc_ealgo == CRYPTO_AES_XTS) crie.cri_klen <<= 1; crie.cri_key = sc->sc_ekey; if (sc->sc_flags & G_ELI_FLAG_AUTH) { bzero(&cria, sizeof(cria)); cria.cri_alg = sc->sc_aalgo; cria.cri_klen = sc->sc_akeylen; cria.cri_key = sc->sc_akey; crie.cri_next = &cria; } switch (sc->sc_crypto) { case G_ELI_CRYPTO_SW: error = crypto_newsession(&wr->w_sid, &crie, CRYPTOCAP_F_SOFTWARE); break; case G_ELI_CRYPTO_HW: error = crypto_newsession(&wr->w_sid, &crie, CRYPTOCAP_F_HARDWARE); break; case G_ELI_CRYPTO_UNKNOWN: error = crypto_newsession(&wr->w_sid, &crie, CRYPTOCAP_F_HARDWARE); if (error == 0) { mtx_lock(&sc->sc_queue_mtx); if (sc->sc_crypto == G_ELI_CRYPTO_UNKNOWN) sc->sc_crypto = G_ELI_CRYPTO_HW; mtx_unlock(&sc->sc_queue_mtx); } else { error = crypto_newsession(&wr->w_sid, &crie, CRYPTOCAP_F_SOFTWARE); mtx_lock(&sc->sc_queue_mtx); if (sc->sc_crypto == G_ELI_CRYPTO_UNKNOWN) sc->sc_crypto = G_ELI_CRYPTO_SW; mtx_unlock(&sc->sc_queue_mtx); } break; default: panic("%s: invalid condition", __func__); } return (error); } static void g_eli_freesession(struct g_eli_worker *wr) { crypto_freesession(wr->w_sid); } static void g_eli_cancel(struct g_eli_softc *sc) { struct bio *bp; mtx_assert(&sc->sc_queue_mtx, MA_OWNED); while ((bp = bioq_takefirst(&sc->sc_queue)) != NULL) { KASSERT(bp->bio_pflags == G_ELI_NEW_BIO, ("Not new bio when canceling (bp=%p).", bp)); g_io_deliver(bp, ENXIO); } } static struct bio * g_eli_takefirst(struct g_eli_softc *sc) { struct bio *bp; mtx_assert(&sc->sc_queue_mtx, MA_OWNED); if (!(sc->sc_flags & G_ELI_FLAG_SUSPEND)) return (bioq_takefirst(&sc->sc_queue)); /* * Device suspended, so we skip new I/O requests. */ TAILQ_FOREACH(bp, &sc->sc_queue.queue, bio_queue) { if (bp->bio_pflags != G_ELI_NEW_BIO) break; } if (bp != NULL) bioq_remove(&sc->sc_queue, bp); return (bp); } /* * This is the main function for kernel worker thread when we don't have * hardware acceleration and we have to do cryptography in software. * Dedicated thread is needed, so we don't slow down g_up/g_down GEOM * threads with crypto work. */ static void g_eli_worker(void *arg) { struct g_eli_softc *sc; struct g_eli_worker *wr; struct bio *bp; int error; wr = arg; sc = wr->w_softc; #ifdef SMP /* Before sched_bind() to a CPU, wait for all CPUs to go on-line. */ if (mp_ncpus > 1 && sc->sc_crypto == G_ELI_CRYPTO_SW && g_eli_threads == 0) { while (!smp_started) tsleep(wr, 0, "geli:smp", hz / 4); } #endif thread_lock(curthread); sched_prio(curthread, PUSER); if (sc->sc_crypto == G_ELI_CRYPTO_SW && g_eli_threads == 0) sched_bind(curthread, wr->w_number); thread_unlock(curthread); G_ELI_DEBUG(1, "Thread %s started.", curthread->td_proc->p_comm); for (;;) { mtx_lock(&sc->sc_queue_mtx); again: bp = g_eli_takefirst(sc); if (bp == NULL) { if (sc->sc_flags & G_ELI_FLAG_DESTROY) { g_eli_cancel(sc); LIST_REMOVE(wr, w_next); g_eli_freesession(wr); free(wr, M_ELI); G_ELI_DEBUG(1, "Thread %s exiting.", curthread->td_proc->p_comm); wakeup(&sc->sc_workers); mtx_unlock(&sc->sc_queue_mtx); kproc_exit(0); } while (sc->sc_flags & G_ELI_FLAG_SUSPEND) { if (sc->sc_inflight > 0) { G_ELI_DEBUG(0, "inflight=%d", sc->sc_inflight); /* * We still have inflight BIOs, so * sleep and retry. */ msleep(sc, &sc->sc_queue_mtx, PRIBIO, "geli:inf", hz / 5); goto again; } /* * Suspend requested, mark the worker as * suspended and go to sleep. */ if (wr->w_active) { g_eli_freesession(wr); wr->w_active = FALSE; } wakeup(&sc->sc_workers); msleep(sc, &sc->sc_queue_mtx, PRIBIO, "geli:suspend", 0); if (!wr->w_active && !(sc->sc_flags & G_ELI_FLAG_SUSPEND)) { error = g_eli_newsession(wr); KASSERT(error == 0, ("g_eli_newsession() failed on resume (error=%d)", error)); wr->w_active = TRUE; } goto again; } msleep(sc, &sc->sc_queue_mtx, PDROP, "geli:w", 0); continue; } if (bp->bio_pflags == G_ELI_NEW_BIO) atomic_add_int(&sc->sc_inflight, 1); mtx_unlock(&sc->sc_queue_mtx); if (bp->bio_pflags == G_ELI_NEW_BIO) { bp->bio_pflags = 0; if (sc->sc_flags & G_ELI_FLAG_AUTH) { if (bp->bio_cmd == BIO_READ) g_eli_auth_read(sc, bp); else g_eli_auth_run(wr, bp); } else { if (bp->bio_cmd == BIO_READ) g_eli_crypto_read(sc, bp, 1); else g_eli_crypto_run(wr, bp); } } else { if (sc->sc_flags & G_ELI_FLAG_AUTH) g_eli_auth_run(wr, bp); else g_eli_crypto_run(wr, bp); } } } /* * Here we generate IV. It is unique for every sector. */ void g_eli_crypto_ivgen(struct g_eli_softc *sc, off_t offset, u_char *iv, size_t size) { uint8_t off[8]; if ((sc->sc_flags & G_ELI_FLAG_NATIVE_BYTE_ORDER) != 0) bcopy(&offset, off, sizeof(off)); else le64enc(off, (uint64_t)offset); switch (sc->sc_ealgo) { case CRYPTO_AES_XTS: bcopy(off, iv, sizeof(off)); bzero(iv + sizeof(off), size - sizeof(off)); break; default: { u_char hash[SHA256_DIGEST_LENGTH]; SHA256_CTX ctx; /* Copy precalculated SHA256 context for IV-Key. */ bcopy(&sc->sc_ivctx, &ctx, sizeof(ctx)); SHA256_Update(&ctx, off, sizeof(off)); SHA256_Final(hash, &ctx); bcopy(hash, iv, MIN(sizeof(hash), size)); break; } } } int g_eli_read_metadata(struct g_class *mp, struct g_provider *pp, struct g_eli_metadata *md) { struct g_geom *gp; struct g_consumer *cp; u_char *buf = NULL; int error; g_topology_assert(); gp = g_new_geomf(mp, "eli:taste"); gp->start = g_eli_start; gp->access = g_std_access; /* * g_eli_read_metadata() is always called from the event thread. * Our geom is created and destroyed in the same event, so there * could be no orphan nor spoil event in the meantime. */ gp->orphan = g_eli_orphan_spoil_assert; gp->spoiled = g_eli_orphan_spoil_assert; cp = g_new_consumer(gp); error = g_attach(cp, pp); if (error != 0) goto end; error = g_access(cp, 1, 0, 0); if (error != 0) goto end; g_topology_unlock(); buf = g_read_data(cp, pp->mediasize - pp->sectorsize, pp->sectorsize, &error); g_topology_lock(); if (buf == NULL) goto end; eli_metadata_decode(buf, md); end: if (buf != NULL) g_free(buf); if (cp->provider != NULL) { if (cp->acr == 1) g_access(cp, -1, 0, 0); g_detach(cp); } g_destroy_consumer(cp); g_destroy_geom(gp); return (error); } /* * The function is called when we had last close on provider and user requested * to close it when this situation occur. */ static void g_eli_last_close(struct g_eli_softc *sc) { struct g_geom *gp; struct g_provider *pp; char ppname[64]; int error; g_topology_assert(); gp = sc->sc_geom; pp = LIST_FIRST(&gp->provider); strlcpy(ppname, pp->name, sizeof(ppname)); error = g_eli_destroy(sc, TRUE); KASSERT(error == 0, ("Cannot detach %s on last close (error=%d).", ppname, error)); G_ELI_DEBUG(0, "Detached %s on last close.", ppname); } int g_eli_access(struct g_provider *pp, int dr, int dw, int de) { struct g_eli_softc *sc; struct g_geom *gp; gp = pp->geom; sc = gp->softc; if (dw > 0) { if (sc->sc_flags & G_ELI_FLAG_RO) { /* Deny write attempts. */ return (EROFS); } /* Someone is opening us for write, we need to remember that. */ sc->sc_flags |= G_ELI_FLAG_WOPEN; return (0); } /* Is this the last close? */ if (pp->acr + dr > 0 || pp->acw + dw > 0 || pp->ace + de > 0) return (0); /* * Automatically detach on last close if requested. */ if ((sc->sc_flags & G_ELI_FLAG_RW_DETACH) || (sc->sc_flags & G_ELI_FLAG_WOPEN)) { g_eli_last_close(sc); } return (0); } static int g_eli_cpu_is_disabled(int cpu) { #ifdef SMP return ((hlt_cpus_mask & (1 << cpu)) != 0); #else return (0); #endif } struct g_geom * g_eli_create(struct gctl_req *req, struct g_class *mp, struct g_provider *bpp, const struct g_eli_metadata *md, const u_char *mkey, int nkey) { struct g_eli_softc *sc; struct g_eli_worker *wr; struct g_geom *gp; struct g_provider *pp; struct g_consumer *cp; u_int i, threads; int error; G_ELI_DEBUG(1, "Creating device %s%s.", bpp->name, G_ELI_SUFFIX); gp = g_new_geomf(mp, "%s%s", bpp->name, G_ELI_SUFFIX); - gp->softc = NULL; /* for a moment */ - sc = malloc(sizeof(*sc), M_ELI, M_WAITOK | M_ZERO); gp->start = g_eli_start; /* * Spoiling cannot happen actually, because we keep provider open for * writing all the time or provider is read-only. */ gp->spoiled = g_eli_orphan_spoil_assert; gp->orphan = g_eli_orphan; gp->dumpconf = g_eli_dumpconf; /* * If detach-on-last-close feature is not enabled and we don't operate * on read-only provider, we can simply use g_std_access(). */ if (md->md_flags & (G_ELI_FLAG_WO_DETACH | G_ELI_FLAG_RO)) gp->access = g_eli_access; else gp->access = g_std_access; sc->sc_inflight = 0; sc->sc_crypto = G_ELI_CRYPTO_UNKNOWN; sc->sc_flags = md->md_flags; /* Backward compatibility. */ if (md->md_version < 4) sc->sc_flags |= G_ELI_FLAG_NATIVE_BYTE_ORDER; if (md->md_version < 5) sc->sc_flags |= G_ELI_FLAG_SINGLE_KEY; sc->sc_ealgo = md->md_ealgo; sc->sc_nkey = nkey; if (sc->sc_flags & G_ELI_FLAG_AUTH) { sc->sc_akeylen = sizeof(sc->sc_akey) * 8; sc->sc_aalgo = md->md_aalgo; sc->sc_alen = g_eli_hashlen(sc->sc_aalgo); sc->sc_data_per_sector = bpp->sectorsize - sc->sc_alen; /* * Some hash functions (like SHA1 and RIPEMD160) generates hash * which length is not multiple of 128 bits, but we want data * length to be multiple of 128, so we can encrypt without * padding. The line below rounds down data length to multiple * of 128 bits. */ sc->sc_data_per_sector -= sc->sc_data_per_sector % 16; sc->sc_bytes_per_sector = (md->md_sectorsize - 1) / sc->sc_data_per_sector + 1; sc->sc_bytes_per_sector *= bpp->sectorsize; } gp->softc = sc; sc->sc_geom = gp; bioq_init(&sc->sc_queue); mtx_init(&sc->sc_queue_mtx, "geli:queue", NULL, MTX_DEF); mtx_init(&sc->sc_ekeys_lock, "geli:ekeys", NULL, MTX_DEF); pp = NULL; cp = g_new_consumer(gp); error = g_attach(cp, bpp); if (error != 0) { if (req != NULL) { gctl_error(req, "Cannot attach to %s (error=%d).", bpp->name, error); } else { G_ELI_DEBUG(1, "Cannot attach to %s (error=%d).", bpp->name, error); } goto failed; } /* * Keep provider open all the time, so we can run critical tasks, * like Master Keys deletion, without wondering if we can open * provider or not. * We don't open provider for writing only when user requested read-only * access. */ if (sc->sc_flags & G_ELI_FLAG_RO) error = g_access(cp, 1, 0, 1); else error = g_access(cp, 1, 1, 1); if (error != 0) { if (req != NULL) { gctl_error(req, "Cannot access %s (error=%d).", bpp->name, error); } else { G_ELI_DEBUG(1, "Cannot access %s (error=%d).", bpp->name, error); } goto failed; } sc->sc_sectorsize = md->md_sectorsize; sc->sc_mediasize = bpp->mediasize; if (!(sc->sc_flags & G_ELI_FLAG_ONETIME)) sc->sc_mediasize -= bpp->sectorsize; if (!(sc->sc_flags & G_ELI_FLAG_AUTH)) sc->sc_mediasize -= (sc->sc_mediasize % sc->sc_sectorsize); else { sc->sc_mediasize /= sc->sc_bytes_per_sector; sc->sc_mediasize *= sc->sc_sectorsize; } /* * Remember the keys in our softc structure. */ g_eli_mkey_propagate(sc, mkey); sc->sc_ekeylen = md->md_keylen; LIST_INIT(&sc->sc_workers); threads = g_eli_threads; if (threads == 0) threads = mp_ncpus; else if (threads > mp_ncpus) { /* There is really no need for too many worker threads. */ threads = mp_ncpus; G_ELI_DEBUG(0, "Reducing number of threads to %u.", threads); } for (i = 0; i < threads; i++) { if (g_eli_cpu_is_disabled(i)) { G_ELI_DEBUG(1, "%s: CPU %u disabled, skipping.", bpp->name, i); continue; } wr = malloc(sizeof(*wr), M_ELI, M_WAITOK | M_ZERO); wr->w_softc = sc; wr->w_number = i; wr->w_active = TRUE; error = g_eli_newsession(wr); if (error != 0) { free(wr, M_ELI); if (req != NULL) { gctl_error(req, "Cannot set up crypto session " "for %s (error=%d).", bpp->name, error); } else { G_ELI_DEBUG(1, "Cannot set up crypto session " "for %s (error=%d).", bpp->name, error); } goto failed; } error = kproc_create(g_eli_worker, wr, &wr->w_proc, 0, 0, "g_eli[%u] %s", i, bpp->name); if (error != 0) { g_eli_freesession(wr); free(wr, M_ELI); if (req != NULL) { gctl_error(req, "Cannot create kernel thread " "for %s (error=%d).", bpp->name, error); } else { G_ELI_DEBUG(1, "Cannot create kernel thread " "for %s (error=%d).", bpp->name, error); } goto failed; } LIST_INSERT_HEAD(&sc->sc_workers, wr, w_next); /* If we have hardware support, one thread is enough. */ if (sc->sc_crypto == G_ELI_CRYPTO_HW) break; } /* * Create decrypted provider. */ pp = g_new_providerf(gp, "%s%s", bpp->name, G_ELI_SUFFIX); pp->mediasize = sc->sc_mediasize; pp->sectorsize = sc->sc_sectorsize; g_error_provider(pp, 0); G_ELI_DEBUG(0, "Device %s created.", pp->name); G_ELI_DEBUG(0, "Encryption: %s %u", g_eli_algo2str(sc->sc_ealgo), sc->sc_ekeylen); if (sc->sc_flags & G_ELI_FLAG_AUTH) G_ELI_DEBUG(0, " Integrity: %s", g_eli_algo2str(sc->sc_aalgo)); G_ELI_DEBUG(0, " Crypto: %s", sc->sc_crypto == G_ELI_CRYPTO_SW ? "software" : "hardware"); return (gp); failed: mtx_lock(&sc->sc_queue_mtx); sc->sc_flags |= G_ELI_FLAG_DESTROY; wakeup(sc); /* * Wait for kernel threads self destruction. */ while (!LIST_EMPTY(&sc->sc_workers)) { msleep(&sc->sc_workers, &sc->sc_queue_mtx, PRIBIO, "geli:destroy", 0); } mtx_destroy(&sc->sc_queue_mtx); if (cp->provider != NULL) { if (cp->acr == 1) g_access(cp, -1, -1, -1); g_detach(cp); } g_destroy_consumer(cp); g_destroy_geom(gp); g_eli_key_destroy(sc); bzero(sc, sizeof(*sc)); free(sc, M_ELI); return (NULL); } int g_eli_destroy(struct g_eli_softc *sc, boolean_t force) { struct g_geom *gp; struct g_provider *pp; g_topology_assert(); if (sc == NULL) return (ENXIO); gp = sc->sc_geom; pp = LIST_FIRST(&gp->provider); if (pp != NULL && (pp->acr != 0 || pp->acw != 0 || pp->ace != 0)) { if (force) { G_ELI_DEBUG(1, "Device %s is still open, so it " "cannot be definitely removed.", pp->name); } else { G_ELI_DEBUG(1, "Device %s is still open (r%dw%de%d).", pp->name, pp->acr, pp->acw, pp->ace); return (EBUSY); } } mtx_lock(&sc->sc_queue_mtx); sc->sc_flags |= G_ELI_FLAG_DESTROY; wakeup(sc); while (!LIST_EMPTY(&sc->sc_workers)) { msleep(&sc->sc_workers, &sc->sc_queue_mtx, PRIBIO, "geli:destroy", 0); } mtx_destroy(&sc->sc_queue_mtx); gp->softc = NULL; g_eli_key_destroy(sc); bzero(sc, sizeof(*sc)); free(sc, M_ELI); if (pp == NULL || (pp->acr == 0 && pp->acw == 0 && pp->ace == 0)) G_ELI_DEBUG(0, "Device %s destroyed.", gp->name); g_wither_geom_close(gp, ENXIO); return (0); } static int g_eli_destroy_geom(struct gctl_req *req __unused, struct g_class *mp __unused, struct g_geom *gp) { struct g_eli_softc *sc; sc = gp->softc; return (g_eli_destroy(sc, FALSE)); } static int g_eli_keyfiles_load(struct hmac_ctx *ctx, const char *provider) { u_char *keyfile, *data; char *file, name[64]; size_t size; int i; for (i = 0; ; i++) { snprintf(name, sizeof(name), "%s:geli_keyfile%d", provider, i); keyfile = preload_search_by_type(name); if (keyfile == NULL) return (i); /* Return number of loaded keyfiles. */ data = preload_fetch_addr(keyfile); if (data == NULL) { G_ELI_DEBUG(0, "Cannot find key file data for %s.", name); return (0); } size = preload_fetch_size(keyfile); if (size == 0) { G_ELI_DEBUG(0, "Cannot find key file size for %s.", name); return (0); } file = preload_search_info(keyfile, MODINFO_NAME); if (file == NULL) { G_ELI_DEBUG(0, "Cannot find key file name for %s.", name); return (0); } G_ELI_DEBUG(1, "Loaded keyfile %s for %s (type: %s).", file, provider, name); g_eli_crypto_hmac_update(ctx, data, size); } } static void g_eli_keyfiles_clear(const char *provider) { u_char *keyfile, *data; char name[64]; size_t size; int i; for (i = 0; ; i++) { snprintf(name, sizeof(name), "%s:geli_keyfile%d", provider, i); keyfile = preload_search_by_type(name); if (keyfile == NULL) return; data = preload_fetch_addr(keyfile); size = preload_fetch_size(keyfile); if (data != NULL && size != 0) bzero(data, size); } } /* * Tasting is only made on boot. * We detect providers which should be attached before root is mounted. */ static struct g_geom * g_eli_taste(struct g_class *mp, struct g_provider *pp, int flags __unused) { struct g_eli_metadata md; struct g_geom *gp; struct hmac_ctx ctx; char passphrase[256]; u_char key[G_ELI_USERKEYLEN], mkey[G_ELI_DATAIVKEYLEN]; u_int i, nkey, nkeyfiles, tries; int error; g_trace(G_T_TOPOLOGY, "%s(%s, %s)", __func__, mp->name, pp->name); g_topology_assert(); if (root_mounted() || g_eli_tries == 0) return (NULL); G_ELI_DEBUG(3, "Tasting %s.", pp->name); error = g_eli_read_metadata(mp, pp, &md); if (error != 0) return (NULL); gp = NULL; if (strcmp(md.md_magic, G_ELI_MAGIC) != 0) return (NULL); if (md.md_version > G_ELI_VERSION) { printf("geom_eli.ko module is too old to handle %s.\n", pp->name); return (NULL); } if (md.md_provsize != pp->mediasize) return (NULL); /* Should we attach it on boot? */ if (!(md.md_flags & G_ELI_FLAG_BOOT)) return (NULL); if (md.md_keys == 0x00) { G_ELI_DEBUG(0, "No valid keys on %s.", pp->name); return (NULL); } if (md.md_iterations == -1) { /* If there is no passphrase, we try only once. */ tries = 1; } else { /* Ask for the passphrase no more than g_eli_tries times. */ tries = g_eli_tries; } for (i = 0; i < tries; i++) { g_eli_crypto_hmac_init(&ctx, NULL, 0); /* * Load all key files. */ nkeyfiles = g_eli_keyfiles_load(&ctx, pp->name); if (nkeyfiles == 0 && md.md_iterations == -1) { /* * No key files and no passphrase, something is * definitely wrong here. * geli(8) doesn't allow for such situation, so assume * that there was really no passphrase and in that case * key files are no properly defined in loader.conf. */ G_ELI_DEBUG(0, "Found no key files in loader.conf for %s.", pp->name); return (NULL); } /* Ask for the passphrase if defined. */ if (md.md_iterations >= 0) { printf("Enter passphrase for %s: ", pp->name); gets(passphrase, sizeof(passphrase), g_eli_visible_passphrase); } /* * Prepare Derived-Key from the user passphrase. */ if (md.md_iterations == 0) { g_eli_crypto_hmac_update(&ctx, md.md_salt, sizeof(md.md_salt)); g_eli_crypto_hmac_update(&ctx, passphrase, strlen(passphrase)); bzero(passphrase, sizeof(passphrase)); } else if (md.md_iterations > 0) { u_char dkey[G_ELI_USERKEYLEN]; pkcs5v2_genkey(dkey, sizeof(dkey), md.md_salt, sizeof(md.md_salt), passphrase, md.md_iterations); bzero(passphrase, sizeof(passphrase)); g_eli_crypto_hmac_update(&ctx, dkey, sizeof(dkey)); bzero(dkey, sizeof(dkey)); } g_eli_crypto_hmac_final(&ctx, key, 0); /* * Decrypt Master-Key. */ error = g_eli_mkey_decrypt(&md, key, mkey, &nkey); bzero(key, sizeof(key)); if (error == -1) { if (i == tries - 1) { G_ELI_DEBUG(0, "Wrong key for %s. No tries left.", pp->name); g_eli_keyfiles_clear(pp->name); return (NULL); } G_ELI_DEBUG(0, "Wrong key for %s. Tries left: %u.", pp->name, tries - i - 1); /* Try again. */ continue; } else if (error > 0) { G_ELI_DEBUG(0, "Cannot decrypt Master Key for %s (error=%d).", pp->name, error); g_eli_keyfiles_clear(pp->name); return (NULL); } G_ELI_DEBUG(1, "Using Master Key %u for %s.", nkey, pp->name); break; } /* * We have correct key, let's attach provider. */ gp = g_eli_create(NULL, mp, pp, &md, mkey, nkey); bzero(mkey, sizeof(mkey)); bzero(&md, sizeof(md)); if (gp == NULL) { G_ELI_DEBUG(0, "Cannot create device %s%s.", pp->name, G_ELI_SUFFIX); return (NULL); } return (gp); } static void g_eli_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp) { struct g_eli_softc *sc; g_topology_assert(); sc = gp->softc; if (sc == NULL) return; if (pp != NULL || cp != NULL) return; /* Nothing here. */ sbuf_printf(sb, "%s%ju", indent, (uintmax_t)sc->sc_ekeys_total); sbuf_printf(sb, "%s%ju", indent, (uintmax_t)sc->sc_ekeys_allocated); sbuf_printf(sb, "%s", indent); if (sc->sc_flags == 0) sbuf_printf(sb, "NONE"); else { int first = 1; #define ADD_FLAG(flag, name) do { \ if (sc->sc_flags & (flag)) { \ if (!first) \ sbuf_printf(sb, ", "); \ else \ first = 0; \ sbuf_printf(sb, name); \ } \ } while (0) ADD_FLAG(G_ELI_FLAG_SUSPEND, "SUSPEND"); ADD_FLAG(G_ELI_FLAG_SINGLE_KEY, "SINGLE-KEY"); ADD_FLAG(G_ELI_FLAG_NATIVE_BYTE_ORDER, "NATIVE-BYTE-ORDER"); ADD_FLAG(G_ELI_FLAG_ONETIME, "ONETIME"); ADD_FLAG(G_ELI_FLAG_BOOT, "BOOT"); ADD_FLAG(G_ELI_FLAG_WO_DETACH, "W-DETACH"); ADD_FLAG(G_ELI_FLAG_RW_DETACH, "RW-DETACH"); ADD_FLAG(G_ELI_FLAG_AUTH, "AUTH"); ADD_FLAG(G_ELI_FLAG_WOPEN, "W-OPEN"); ADD_FLAG(G_ELI_FLAG_DESTROY, "DESTROY"); ADD_FLAG(G_ELI_FLAG_RO, "READ-ONLY"); #undef ADD_FLAG } sbuf_printf(sb, "\n"); if (!(sc->sc_flags & G_ELI_FLAG_ONETIME)) { sbuf_printf(sb, "%s%u\n", indent, sc->sc_nkey); } sbuf_printf(sb, "%s", indent); switch (sc->sc_crypto) { case G_ELI_CRYPTO_HW: sbuf_printf(sb, "hardware"); break; case G_ELI_CRYPTO_SW: sbuf_printf(sb, "software"); break; default: sbuf_printf(sb, "UNKNOWN"); break; } sbuf_printf(sb, "\n"); if (sc->sc_flags & G_ELI_FLAG_AUTH) { sbuf_printf(sb, "%s%s\n", indent, g_eli_algo2str(sc->sc_aalgo)); } sbuf_printf(sb, "%s%u\n", indent, sc->sc_ekeylen); sbuf_printf(sb, "%s%s\n", indent, g_eli_algo2str(sc->sc_ealgo)); sbuf_printf(sb, "%s%s\n", indent, (sc->sc_flags & G_ELI_FLAG_SUSPEND) ? "SUSPENDED" : "ACTIVE"); } static void g_eli_shutdown_pre_sync(void *arg, int howto) { struct g_class *mp; struct g_geom *gp, *gp2; struct g_provider *pp; struct g_eli_softc *sc; int error; mp = arg; DROP_GIANT(); g_topology_lock(); LIST_FOREACH_SAFE(gp, &mp->geom, geom, gp2) { sc = gp->softc; if (sc == NULL) continue; pp = LIST_FIRST(&gp->provider); KASSERT(pp != NULL, ("No provider? gp=%p (%s)", gp, gp->name)); if (pp->acr + pp->acw + pp->ace == 0) error = g_eli_destroy(sc, TRUE); else { sc->sc_flags |= G_ELI_FLAG_RW_DETACH; gp->access = g_eli_access; } } g_topology_unlock(); PICKUP_GIANT(); } static void g_eli_init(struct g_class *mp) { g_eli_pre_sync = EVENTHANDLER_REGISTER(shutdown_pre_sync, g_eli_shutdown_pre_sync, mp, SHUTDOWN_PRI_FIRST); if (g_eli_pre_sync == NULL) G_ELI_DEBUG(0, "Warning! Cannot register shutdown event."); } static void g_eli_fini(struct g_class *mp) { if (g_eli_pre_sync != NULL) EVENTHANDLER_DEREGISTER(shutdown_pre_sync, g_eli_pre_sync); } DECLARE_GEOM_CLASS(g_eli_class, g_eli); MODULE_DEPEND(g_eli, crypto, 1, 1, 1); Index: projects/largeSMP/sys/geom/journal/g_journal.c =================================================================== --- projects/largeSMP/sys/geom/journal/g_journal.c (revision 221494) +++ projects/largeSMP/sys/geom/journal/g_journal.c (revision 221495) @@ -1,3062 +1,3061 @@ /*- * Copyright (c) 2005-2006 Pawel Jakub Dawidek * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef GJ_MEMDEBUG #include #include #endif #include #include #include #include FEATURE(geom_journal, "GEOM journaling support"); /* * On-disk journal format: * * JH - Journal header * RH - Record header * * %%%%%% ****** +------+ +------+ ****** +------+ %%%%%% * % JH % * RH * | Data | | Data | ... * RH * | Data | ... % JH % ... * %%%%%% ****** +------+ +------+ ****** +------+ %%%%%% * */ CTASSERT(sizeof(struct g_journal_header) <= 512); CTASSERT(sizeof(struct g_journal_record_header) <= 512); static MALLOC_DEFINE(M_JOURNAL, "journal_data", "GEOM_JOURNAL Data"); static struct mtx g_journal_cache_mtx; MTX_SYSINIT(g_journal_cache, &g_journal_cache_mtx, "cache usage", MTX_DEF); const struct g_journal_desc *g_journal_filesystems[] = { &g_journal_ufs, NULL }; SYSCTL_DECL(_kern_geom); int g_journal_debug = 0; TUNABLE_INT("kern.geom.journal.debug", &g_journal_debug); static u_int g_journal_switch_time = 10; static u_int g_journal_force_switch = 70; static u_int g_journal_parallel_flushes = 16; static u_int g_journal_parallel_copies = 16; static u_int g_journal_accept_immediately = 64; static u_int g_journal_record_entries = GJ_RECORD_HEADER_NENTRIES; static u_int g_journal_do_optimize = 1; SYSCTL_NODE(_kern_geom, OID_AUTO, journal, CTLFLAG_RW, 0, "GEOM_JOURNAL stuff"); SYSCTL_INT(_kern_geom_journal, OID_AUTO, debug, CTLFLAG_RW, &g_journal_debug, 0, "Debug level"); SYSCTL_UINT(_kern_geom_journal, OID_AUTO, switch_time, CTLFLAG_RW, &g_journal_switch_time, 0, "Switch journals every N seconds"); SYSCTL_UINT(_kern_geom_journal, OID_AUTO, force_switch, CTLFLAG_RW, &g_journal_force_switch, 0, "Force switch when journal is N% full"); SYSCTL_UINT(_kern_geom_journal, OID_AUTO, parallel_flushes, CTLFLAG_RW, &g_journal_parallel_flushes, 0, "Number of flush I/O requests to send in parallel"); SYSCTL_UINT(_kern_geom_journal, OID_AUTO, accept_immediately, CTLFLAG_RW, &g_journal_accept_immediately, 0, "Number of I/O requests accepted immediately"); SYSCTL_UINT(_kern_geom_journal, OID_AUTO, parallel_copies, CTLFLAG_RW, &g_journal_parallel_copies, 0, "Number of copy I/O requests to send in parallel"); static int g_journal_record_entries_sysctl(SYSCTL_HANDLER_ARGS) { u_int entries; int error; entries = g_journal_record_entries; error = sysctl_handle_int(oidp, &entries, 0, req); if (error != 0 || req->newptr == NULL) return (error); if (entries < 1 || entries > GJ_RECORD_HEADER_NENTRIES) return (EINVAL); g_journal_record_entries = entries; return (0); } SYSCTL_PROC(_kern_geom_journal, OID_AUTO, record_entries, CTLTYPE_UINT | CTLFLAG_RW, NULL, 0, g_journal_record_entries_sysctl, "I", "Maximum number of entires in one journal record"); SYSCTL_UINT(_kern_geom_journal, OID_AUTO, optimize, CTLFLAG_RW, &g_journal_do_optimize, 0, "Try to combine bios on flush and copy"); static u_int g_journal_cache_used = 0; static u_int g_journal_cache_limit = 64 * 1024 * 1024; TUNABLE_INT("kern.geom.journal.cache.limit", &g_journal_cache_limit); static u_int g_journal_cache_divisor = 2; TUNABLE_INT("kern.geom.journal.cache.divisor", &g_journal_cache_divisor); static u_int g_journal_cache_switch = 90; static u_int g_journal_cache_misses = 0; static u_int g_journal_cache_alloc_failures = 0; static u_int g_journal_cache_low = 0; SYSCTL_NODE(_kern_geom_journal, OID_AUTO, cache, CTLFLAG_RW, 0, "GEOM_JOURNAL cache"); SYSCTL_UINT(_kern_geom_journal_cache, OID_AUTO, used, CTLFLAG_RD, &g_journal_cache_used, 0, "Number of allocated bytes"); static int g_journal_cache_limit_sysctl(SYSCTL_HANDLER_ARGS) { u_int limit; int error; limit = g_journal_cache_limit; error = sysctl_handle_int(oidp, &limit, 0, req); if (error != 0 || req->newptr == NULL) return (error); g_journal_cache_limit = limit; g_journal_cache_low = (limit / 100) * g_journal_cache_switch; return (0); } SYSCTL_PROC(_kern_geom_journal_cache, OID_AUTO, limit, CTLTYPE_UINT | CTLFLAG_RW, NULL, 0, g_journal_cache_limit_sysctl, "I", "Maximum number of allocated bytes"); SYSCTL_UINT(_kern_geom_journal_cache, OID_AUTO, divisor, CTLFLAG_RDTUN, &g_journal_cache_divisor, 0, "(kmem_size / kern.geom.journal.cache.divisor) == cache size"); static int g_journal_cache_switch_sysctl(SYSCTL_HANDLER_ARGS) { u_int cswitch; int error; cswitch = g_journal_cache_switch; error = sysctl_handle_int(oidp, &cswitch, 0, req); if (error != 0 || req->newptr == NULL) return (error); if (cswitch < 0 || cswitch > 100) return (EINVAL); g_journal_cache_switch = cswitch; g_journal_cache_low = (g_journal_cache_limit / 100) * cswitch; return (0); } SYSCTL_PROC(_kern_geom_journal_cache, OID_AUTO, switch, CTLTYPE_UINT | CTLFLAG_RW, NULL, 0, g_journal_cache_switch_sysctl, "I", "Force switch when we hit this percent of cache use"); SYSCTL_UINT(_kern_geom_journal_cache, OID_AUTO, misses, CTLFLAG_RW, &g_journal_cache_misses, 0, "Number of cache misses"); SYSCTL_UINT(_kern_geom_journal_cache, OID_AUTO, alloc_failures, CTLFLAG_RW, &g_journal_cache_alloc_failures, 0, "Memory allocation failures"); static u_long g_journal_stats_bytes_skipped = 0; static u_long g_journal_stats_combined_ios = 0; static u_long g_journal_stats_switches = 0; static u_long g_journal_stats_wait_for_copy = 0; static u_long g_journal_stats_journal_full = 0; static u_long g_journal_stats_low_mem = 0; SYSCTL_NODE(_kern_geom_journal, OID_AUTO, stats, CTLFLAG_RW, 0, "GEOM_JOURNAL statistics"); SYSCTL_ULONG(_kern_geom_journal_stats, OID_AUTO, skipped_bytes, CTLFLAG_RW, &g_journal_stats_bytes_skipped, 0, "Number of skipped bytes"); SYSCTL_ULONG(_kern_geom_journal_stats, OID_AUTO, combined_ios, CTLFLAG_RW, &g_journal_stats_combined_ios, 0, "Number of combined I/O requests"); SYSCTL_ULONG(_kern_geom_journal_stats, OID_AUTO, switches, CTLFLAG_RW, &g_journal_stats_switches, 0, "Number of journal switches"); SYSCTL_ULONG(_kern_geom_journal_stats, OID_AUTO, wait_for_copy, CTLFLAG_RW, &g_journal_stats_wait_for_copy, 0, "Wait for journal copy on switch"); SYSCTL_ULONG(_kern_geom_journal_stats, OID_AUTO, journal_full, CTLFLAG_RW, &g_journal_stats_journal_full, 0, "Number of times journal was almost full."); SYSCTL_ULONG(_kern_geom_journal_stats, OID_AUTO, low_mem, CTLFLAG_RW, &g_journal_stats_low_mem, 0, "Number of times low_mem hook was called."); static g_taste_t g_journal_taste; static g_ctl_req_t g_journal_config; static g_dumpconf_t g_journal_dumpconf; static g_init_t g_journal_init; static g_fini_t g_journal_fini; struct g_class g_journal_class = { .name = G_JOURNAL_CLASS_NAME, .version = G_VERSION, .taste = g_journal_taste, .ctlreq = g_journal_config, .dumpconf = g_journal_dumpconf, .init = g_journal_init, .fini = g_journal_fini }; static int g_journal_destroy(struct g_journal_softc *sc); static void g_journal_metadata_update(struct g_journal_softc *sc); static void g_journal_switch_wait(struct g_journal_softc *sc); #define GJ_SWITCHER_WORKING 0 #define GJ_SWITCHER_DIE 1 #define GJ_SWITCHER_DIED 2 static int g_journal_switcher_state = GJ_SWITCHER_WORKING; static int g_journal_switcher_wokenup = 0; static int g_journal_sync_requested = 0; #ifdef GJ_MEMDEBUG struct meminfo { size_t mi_size; struct stack mi_stack; }; #endif /* * We use our own malloc/realloc/free funtions, so we can collect statistics * and force journal switch when we're running out of cache. */ static void * gj_malloc(size_t size, int flags) { void *p; #ifdef GJ_MEMDEBUG struct meminfo *mi; #endif mtx_lock(&g_journal_cache_mtx); if (g_journal_cache_limit > 0 && !g_journal_switcher_wokenup && g_journal_cache_used + size > g_journal_cache_low) { GJ_DEBUG(1, "No cache, waking up the switcher."); g_journal_switcher_wokenup = 1; wakeup(&g_journal_switcher_state); } if ((flags & M_NOWAIT) && g_journal_cache_limit > 0 && g_journal_cache_used + size > g_journal_cache_limit) { mtx_unlock(&g_journal_cache_mtx); g_journal_cache_alloc_failures++; return (NULL); } g_journal_cache_used += size; mtx_unlock(&g_journal_cache_mtx); flags &= ~M_NOWAIT; #ifndef GJ_MEMDEBUG p = malloc(size, M_JOURNAL, flags | M_WAITOK); #else mi = malloc(sizeof(*mi) + size, M_JOURNAL, flags | M_WAITOK); p = (u_char *)mi + sizeof(*mi); mi->mi_size = size; stack_save(&mi->mi_stack); #endif return (p); } static void gj_free(void *p, size_t size) { #ifdef GJ_MEMDEBUG struct meminfo *mi; #endif KASSERT(p != NULL, ("p=NULL")); KASSERT(size > 0, ("size=0")); mtx_lock(&g_journal_cache_mtx); KASSERT(g_journal_cache_used >= size, ("Freeing too much?")); g_journal_cache_used -= size; mtx_unlock(&g_journal_cache_mtx); #ifdef GJ_MEMDEBUG mi = p = (void *)((u_char *)p - sizeof(*mi)); if (mi->mi_size != size) { printf("GJOURNAL: Size mismatch! %zu != %zu\n", size, mi->mi_size); printf("GJOURNAL: Alloc backtrace:\n"); stack_print(&mi->mi_stack); printf("GJOURNAL: Free backtrace:\n"); kdb_backtrace(); } #endif free(p, M_JOURNAL); } static void * gj_realloc(void *p, size_t size, size_t oldsize) { void *np; #ifndef GJ_MEMDEBUG mtx_lock(&g_journal_cache_mtx); g_journal_cache_used -= oldsize; g_journal_cache_used += size; mtx_unlock(&g_journal_cache_mtx); np = realloc(p, size, M_JOURNAL, M_WAITOK); #else np = gj_malloc(size, M_WAITOK); bcopy(p, np, MIN(oldsize, size)); gj_free(p, oldsize); #endif return (np); } static void g_journal_check_overflow(struct g_journal_softc *sc) { off_t length, used; if ((sc->sc_active.jj_offset < sc->sc_inactive.jj_offset && sc->sc_journal_offset >= sc->sc_inactive.jj_offset) || (sc->sc_active.jj_offset > sc->sc_inactive.jj_offset && sc->sc_journal_offset >= sc->sc_inactive.jj_offset && sc->sc_journal_offset < sc->sc_active.jj_offset)) { panic("Journal overflow (joffset=%jd active=%jd inactive=%jd)", (intmax_t)sc->sc_journal_offset, (intmax_t)sc->sc_active.jj_offset, (intmax_t)sc->sc_inactive.jj_offset); } if (sc->sc_active.jj_offset < sc->sc_inactive.jj_offset) { length = sc->sc_inactive.jj_offset - sc->sc_active.jj_offset; used = sc->sc_journal_offset - sc->sc_active.jj_offset; } else { length = sc->sc_jend - sc->sc_active.jj_offset; length += sc->sc_inactive.jj_offset - sc->sc_jstart; if (sc->sc_journal_offset >= sc->sc_active.jj_offset) used = sc->sc_journal_offset - sc->sc_active.jj_offset; else { used = sc->sc_jend - sc->sc_active.jj_offset; used += sc->sc_journal_offset - sc->sc_jstart; } } /* Already woken up? */ if (g_journal_switcher_wokenup) return; /* * If the active journal takes more than g_journal_force_switch precent * of free journal space, we force journal switch. */ KASSERT(length > 0, ("length=%jd used=%jd active=%jd inactive=%jd joffset=%jd", (intmax_t)length, (intmax_t)used, (intmax_t)sc->sc_active.jj_offset, (intmax_t)sc->sc_inactive.jj_offset, (intmax_t)sc->sc_journal_offset)); if ((used * 100) / length > g_journal_force_switch) { g_journal_stats_journal_full++; GJ_DEBUG(1, "Journal %s %jd%% full, forcing journal switch.", sc->sc_name, (used * 100) / length); mtx_lock(&g_journal_cache_mtx); g_journal_switcher_wokenup = 1; wakeup(&g_journal_switcher_state); mtx_unlock(&g_journal_cache_mtx); } } static void g_journal_orphan(struct g_consumer *cp) { struct g_journal_softc *sc; char name[256]; int error; g_topology_assert(); sc = cp->geom->softc; strlcpy(name, cp->provider->name, sizeof(name)); GJ_DEBUG(0, "Lost provider %s.", name); if (sc == NULL) return; error = g_journal_destroy(sc); if (error == 0) GJ_DEBUG(0, "Journal %s destroyed.", name); else { GJ_DEBUG(0, "Cannot destroy journal %s (error=%d). " "Destroy it manually after last close.", sc->sc_name, error); } } static int g_journal_access(struct g_provider *pp, int acr, int acw, int ace) { struct g_journal_softc *sc; int dcr, dcw, dce; g_topology_assert(); GJ_DEBUG(2, "Access request for %s: r%dw%de%d.", pp->name, acr, acw, ace); dcr = pp->acr + acr; dcw = pp->acw + acw; dce = pp->ace + ace; sc = pp->geom->softc; if (sc == NULL || (sc->sc_flags & GJF_DEVICE_DESTROY)) { if (acr <= 0 && acw <= 0 && ace <= 0) return (0); else return (ENXIO); } if (pp->acw == 0 && dcw > 0) { GJ_DEBUG(1, "Marking %s as dirty.", sc->sc_name); sc->sc_flags &= ~GJF_DEVICE_CLEAN; g_topology_unlock(); g_journal_metadata_update(sc); g_topology_lock(); } /* else if (pp->acw == 0 && dcw > 0 && JEMPTY(sc)) { GJ_DEBUG(1, "Marking %s as clean.", sc->sc_name); sc->sc_flags |= GJF_DEVICE_CLEAN; g_topology_unlock(); g_journal_metadata_update(sc); g_topology_lock(); } */ return (0); } static void g_journal_header_encode(struct g_journal_header *hdr, u_char *data) { bcopy(GJ_HEADER_MAGIC, data, sizeof(GJ_HEADER_MAGIC)); data += sizeof(GJ_HEADER_MAGIC); le32enc(data, hdr->jh_journal_id); data += 4; le32enc(data, hdr->jh_journal_next_id); } static int g_journal_header_decode(const u_char *data, struct g_journal_header *hdr) { bcopy(data, hdr->jh_magic, sizeof(hdr->jh_magic)); data += sizeof(hdr->jh_magic); if (bcmp(hdr->jh_magic, GJ_HEADER_MAGIC, sizeof(GJ_HEADER_MAGIC)) != 0) return (EINVAL); hdr->jh_journal_id = le32dec(data); data += 4; hdr->jh_journal_next_id = le32dec(data); return (0); } static void g_journal_flush_cache(struct g_journal_softc *sc) { struct bintime bt; int error; if (sc->sc_bio_flush == 0) return; GJ_TIMER_START(1, &bt); if (sc->sc_bio_flush & GJ_FLUSH_JOURNAL) { error = g_io_flush(sc->sc_jconsumer); GJ_DEBUG(error == 0 ? 2 : 0, "Flush cache of %s: error=%d.", sc->sc_jconsumer->provider->name, error); } if (sc->sc_bio_flush & GJ_FLUSH_DATA) { /* * TODO: This could be called in parallel with the * previous call. */ error = g_io_flush(sc->sc_dconsumer); GJ_DEBUG(error == 0 ? 2 : 0, "Flush cache of %s: error=%d.", sc->sc_dconsumer->provider->name, error); } GJ_TIMER_STOP(1, &bt, "Cache flush time"); } static int g_journal_write_header(struct g_journal_softc *sc) { struct g_journal_header hdr; struct g_consumer *cp; u_char *buf; int error; cp = sc->sc_jconsumer; buf = gj_malloc(cp->provider->sectorsize, M_WAITOK); strlcpy(hdr.jh_magic, GJ_HEADER_MAGIC, sizeof(hdr.jh_magic)); hdr.jh_journal_id = sc->sc_journal_id; hdr.jh_journal_next_id = sc->sc_journal_next_id; g_journal_header_encode(&hdr, buf); error = g_write_data(cp, sc->sc_journal_offset, buf, cp->provider->sectorsize); /* if (error == 0) */ sc->sc_journal_offset += cp->provider->sectorsize; gj_free(buf, cp->provider->sectorsize); return (error); } /* * Every journal record has a header and data following it. * Functions below are used to decode the header before storing it to * little endian and to encode it after reading to system endianess. */ static void g_journal_record_header_encode(struct g_journal_record_header *hdr, u_char *data) { struct g_journal_entry *ent; u_int i; bcopy(GJ_RECORD_HEADER_MAGIC, data, sizeof(GJ_RECORD_HEADER_MAGIC)); data += sizeof(GJ_RECORD_HEADER_MAGIC); le32enc(data, hdr->jrh_journal_id); data += 8; le16enc(data, hdr->jrh_nentries); data += 2; bcopy(hdr->jrh_sum, data, sizeof(hdr->jrh_sum)); data += 8; for (i = 0; i < hdr->jrh_nentries; i++) { ent = &hdr->jrh_entries[i]; le64enc(data, ent->je_joffset); data += 8; le64enc(data, ent->je_offset); data += 8; le64enc(data, ent->je_length); data += 8; } } static int g_journal_record_header_decode(const u_char *data, struct g_journal_record_header *hdr) { struct g_journal_entry *ent; u_int i; bcopy(data, hdr->jrh_magic, sizeof(hdr->jrh_magic)); data += sizeof(hdr->jrh_magic); if (strcmp(hdr->jrh_magic, GJ_RECORD_HEADER_MAGIC) != 0) return (EINVAL); hdr->jrh_journal_id = le32dec(data); data += 8; hdr->jrh_nentries = le16dec(data); data += 2; if (hdr->jrh_nentries > GJ_RECORD_HEADER_NENTRIES) return (EINVAL); bcopy(data, hdr->jrh_sum, sizeof(hdr->jrh_sum)); data += 8; for (i = 0; i < hdr->jrh_nentries; i++) { ent = &hdr->jrh_entries[i]; ent->je_joffset = le64dec(data); data += 8; ent->je_offset = le64dec(data); data += 8; ent->je_length = le64dec(data); data += 8; } return (0); } /* * Function reads metadata from a provider (via the given consumer), decodes * it to system endianess and verifies its correctness. */ static int g_journal_metadata_read(struct g_consumer *cp, struct g_journal_metadata *md) { struct g_provider *pp; u_char *buf; int error; g_topology_assert(); error = g_access(cp, 1, 0, 0); if (error != 0) return (error); pp = cp->provider; g_topology_unlock(); /* Metadata is stored in last sector. */ buf = g_read_data(cp, pp->mediasize - pp->sectorsize, pp->sectorsize, &error); g_topology_lock(); g_access(cp, -1, 0, 0); if (buf == NULL) { GJ_DEBUG(1, "Cannot read metadata from %s (error=%d).", cp->provider->name, error); return (error); } /* Decode metadata. */ error = journal_metadata_decode(buf, md); g_free(buf); /* Is this is gjournal provider at all? */ if (strcmp(md->md_magic, G_JOURNAL_MAGIC) != 0) return (EINVAL); /* * Are we able to handle this version of metadata? * We only maintain backward compatibility. */ if (md->md_version > G_JOURNAL_VERSION) { GJ_DEBUG(0, "Kernel module is too old to handle metadata from %s.", cp->provider->name); return (EINVAL); } /* Is checksum correct? */ if (error != 0) { GJ_DEBUG(0, "MD5 metadata hash mismatch for provider %s.", cp->provider->name); return (error); } return (0); } /* * Two functions below are responsible for updating metadata. * Only metadata on the data provider is updated (we need to update * information about active journal in there). */ static void g_journal_metadata_done(struct bio *bp) { /* * There is not much we can do on error except informing about it. */ if (bp->bio_error != 0) { GJ_LOGREQ(0, bp, "Cannot update metadata (error=%d).", bp->bio_error); } else { GJ_LOGREQ(2, bp, "Metadata updated."); } gj_free(bp->bio_data, bp->bio_length); g_destroy_bio(bp); } static void g_journal_metadata_update(struct g_journal_softc *sc) { struct g_journal_metadata md; struct g_consumer *cp; struct bio *bp; u_char *sector; cp = sc->sc_dconsumer; sector = gj_malloc(cp->provider->sectorsize, M_WAITOK); strlcpy(md.md_magic, G_JOURNAL_MAGIC, sizeof(md.md_magic)); md.md_version = G_JOURNAL_VERSION; md.md_id = sc->sc_id; md.md_type = sc->sc_orig_type; md.md_jstart = sc->sc_jstart; md.md_jend = sc->sc_jend; md.md_joffset = sc->sc_inactive.jj_offset; md.md_jid = sc->sc_journal_previous_id; md.md_flags = 0; if (sc->sc_flags & GJF_DEVICE_CLEAN) md.md_flags |= GJ_FLAG_CLEAN; if (sc->sc_flags & GJF_DEVICE_HARDCODED) strlcpy(md.md_provider, sc->sc_name, sizeof(md.md_provider)); else bzero(md.md_provider, sizeof(md.md_provider)); md.md_provsize = cp->provider->mediasize; journal_metadata_encode(&md, sector); /* * Flush the cache, so we know all data are on disk. * We write here informations like "journal is consistent", so we need * to be sure it is. Without BIO_FLUSH here, we can end up in situation * where metadata is stored on disk, but not all data. */ g_journal_flush_cache(sc); bp = g_alloc_bio(); bp->bio_offset = cp->provider->mediasize - cp->provider->sectorsize; bp->bio_length = cp->provider->sectorsize; bp->bio_data = sector; bp->bio_cmd = BIO_WRITE; if (!(sc->sc_flags & GJF_DEVICE_DESTROY)) { bp->bio_done = g_journal_metadata_done; g_io_request(bp, cp); } else { bp->bio_done = NULL; g_io_request(bp, cp); biowait(bp, "gjmdu"); g_journal_metadata_done(bp); } /* * Be sure metadata reached the disk. */ g_journal_flush_cache(sc); } /* * This is where the I/O request comes from the GEOM. */ static void g_journal_start(struct bio *bp) { struct g_journal_softc *sc; sc = bp->bio_to->geom->softc; GJ_LOGREQ(3, bp, "Request received."); switch (bp->bio_cmd) { case BIO_READ: case BIO_WRITE: mtx_lock(&sc->sc_mtx); bioq_insert_tail(&sc->sc_regular_queue, bp); wakeup(sc); mtx_unlock(&sc->sc_mtx); return; case BIO_GETATTR: if (strcmp(bp->bio_attribute, "GJOURNAL::provider") == 0) { strlcpy(bp->bio_data, bp->bio_to->name, bp->bio_length); bp->bio_completed = strlen(bp->bio_to->name) + 1; g_io_deliver(bp, 0); return; } /* FALLTHROUGH */ case BIO_DELETE: default: g_io_deliver(bp, EOPNOTSUPP); return; } } static void g_journal_std_done(struct bio *bp) { struct g_journal_softc *sc; sc = bp->bio_from->geom->softc; mtx_lock(&sc->sc_mtx); bioq_insert_tail(&sc->sc_back_queue, bp); wakeup(sc); mtx_unlock(&sc->sc_mtx); } static struct bio * g_journal_new_bio(off_t start, off_t end, off_t joffset, u_char *data, int flags) { struct bio *bp; bp = g_alloc_bio(); bp->bio_offset = start; bp->bio_joffset = joffset; bp->bio_length = end - start; bp->bio_cmd = BIO_WRITE; bp->bio_done = g_journal_std_done; if (data == NULL) bp->bio_data = NULL; else { bp->bio_data = gj_malloc(bp->bio_length, flags); if (bp->bio_data != NULL) bcopy(data, bp->bio_data, bp->bio_length); } return (bp); } #define g_journal_insert_bio(head, bp, flags) \ g_journal_insert((head), (bp)->bio_offset, \ (bp)->bio_offset + (bp)->bio_length, (bp)->bio_joffset, \ (bp)->bio_data, flags) /* * The function below does a lot more than just inserting bio to the queue. * It keeps the queue sorted by offset and ensures that there are no doubled * data (it combines bios where ranges overlap). * * The function returns the number of bios inserted (as bio can be splitted). */ static int g_journal_insert(struct bio **head, off_t nstart, off_t nend, off_t joffset, u_char *data, int flags) { struct bio *nbp, *cbp, *pbp; off_t cstart, cend; u_char *tmpdata; int n; GJ_DEBUG(3, "INSERT(%p): (%jd, %jd, %jd)", *head, nstart, nend, joffset); n = 0; pbp = NULL; GJQ_FOREACH(*head, cbp) { cstart = cbp->bio_offset; cend = cbp->bio_offset + cbp->bio_length; if (nstart >= cend) { /* * +-------------+ * | | * | current | +-------------+ * | bio | | | * | | | new | * +-------------+ | bio | * | | * +-------------+ */ GJ_DEBUG(3, "INSERT(%p): 1", *head); } else if (nend <= cstart) { /* * +-------------+ * | | * +-------------+ | current | * | | | bio | * | new | | | * | bio | +-------------+ * | | * +-------------+ */ nbp = g_journal_new_bio(nstart, nend, joffset, data, flags); if (pbp == NULL) *head = nbp; else pbp->bio_next = nbp; nbp->bio_next = cbp; n++; GJ_DEBUG(3, "INSERT(%p): 2 (nbp=%p pbp=%p)", *head, nbp, pbp); goto end; } else if (nstart <= cstart && nend >= cend) { /* * +-------------+ +-------------+ * | current bio | | current bio | * +---+-------------+---+ +-------------+---+ * | | | | | | | * | | | | | | | * | +-------------+ | +-------------+ | * | new bio | | new bio | * +---------------------+ +-----------------+ * * +-------------+ +-------------+ * | current bio | | current bio | * +---+-------------+ +-------------+ * | | | | | * | | | | | * | +-------------+ +-------------+ * | new bio | | new bio | * +-----------------+ +-------------+ */ g_journal_stats_bytes_skipped += cbp->bio_length; cbp->bio_offset = nstart; cbp->bio_joffset = joffset; cbp->bio_length = cend - nstart; if (cbp->bio_data != NULL) { gj_free(cbp->bio_data, cend - cstart); cbp->bio_data = NULL; } if (data != NULL) { cbp->bio_data = gj_malloc(cbp->bio_length, flags); if (cbp->bio_data != NULL) { bcopy(data, cbp->bio_data, cbp->bio_length); } data += cend - nstart; } joffset += cend - nstart; nstart = cend; GJ_DEBUG(3, "INSERT(%p): 3 (cbp=%p)", *head, cbp); } else if (nstart > cstart && nend >= cend) { /* * +-----------------+ +-------------+ * | current bio | | current bio | * | +-------------+ | +---------+---+ * | | | | | | | * | | | | | | | * +---+-------------+ +---+---------+ | * | new bio | | new bio | * +-------------+ +-------------+ */ g_journal_stats_bytes_skipped += cend - nstart; nbp = g_journal_new_bio(nstart, cend, joffset, data, flags); nbp->bio_next = cbp->bio_next; cbp->bio_next = nbp; cbp->bio_length = nstart - cstart; if (cbp->bio_data != NULL) { cbp->bio_data = gj_realloc(cbp->bio_data, cbp->bio_length, cend - cstart); } if (data != NULL) data += cend - nstart; joffset += cend - nstart; nstart = cend; n++; GJ_DEBUG(3, "INSERT(%p): 4 (cbp=%p)", *head, cbp); } else if (nstart > cstart && nend < cend) { /* * +---------------------+ * | current bio | * | +-------------+ | * | | | | * | | | | * +---+-------------+---+ * | new bio | * +-------------+ */ g_journal_stats_bytes_skipped += nend - nstart; nbp = g_journal_new_bio(nstart, nend, joffset, data, flags); nbp->bio_next = cbp->bio_next; cbp->bio_next = nbp; if (cbp->bio_data == NULL) tmpdata = NULL; else tmpdata = cbp->bio_data + nend - cstart; nbp = g_journal_new_bio(nend, cend, cbp->bio_joffset + nend - cstart, tmpdata, flags); nbp->bio_next = ((struct bio *)cbp->bio_next)->bio_next; ((struct bio *)cbp->bio_next)->bio_next = nbp; cbp->bio_length = nstart - cstart; if (cbp->bio_data != NULL) { cbp->bio_data = gj_realloc(cbp->bio_data, cbp->bio_length, cend - cstart); } n += 2; GJ_DEBUG(3, "INSERT(%p): 5 (cbp=%p)", *head, cbp); goto end; } else if (nstart <= cstart && nend < cend) { /* * +-----------------+ +-------------+ * | current bio | | current bio | * +-------------+ | +---+---------+ | * | | | | | | | * | | | | | | | * +-------------+---+ | +---------+---+ * | new bio | | new bio | * +-------------+ +-------------+ */ g_journal_stats_bytes_skipped += nend - nstart; nbp = g_journal_new_bio(nstart, nend, joffset, data, flags); if (pbp == NULL) *head = nbp; else pbp->bio_next = nbp; nbp->bio_next = cbp; cbp->bio_offset = nend; cbp->bio_length = cend - nend; cbp->bio_joffset += nend - cstart; tmpdata = cbp->bio_data; if (tmpdata != NULL) { cbp->bio_data = gj_malloc(cbp->bio_length, flags); if (cbp->bio_data != NULL) { bcopy(tmpdata + nend - cstart, cbp->bio_data, cbp->bio_length); } gj_free(tmpdata, cend - cstart); } n++; GJ_DEBUG(3, "INSERT(%p): 6 (cbp=%p)", *head, cbp); goto end; } if (nstart == nend) goto end; pbp = cbp; } nbp = g_journal_new_bio(nstart, nend, joffset, data, flags); if (pbp == NULL) *head = nbp; else pbp->bio_next = nbp; nbp->bio_next = NULL; n++; GJ_DEBUG(3, "INSERT(%p): 8 (nbp=%p pbp=%p)", *head, nbp, pbp); end: if (g_journal_debug >= 3) { GJQ_FOREACH(*head, cbp) { GJ_DEBUG(3, "ELEMENT: %p (%jd, %jd, %jd, %p)", cbp, (intmax_t)cbp->bio_offset, (intmax_t)cbp->bio_length, (intmax_t)cbp->bio_joffset, cbp->bio_data); } GJ_DEBUG(3, "INSERT(%p): DONE %d", *head, n); } return (n); } /* * The function combines neighbour bios trying to squeeze as much data as * possible into one bio. * * The function returns the number of bios combined (negative value). */ static int g_journal_optimize(struct bio *head) { struct bio *cbp, *pbp; int n; n = 0; pbp = NULL; GJQ_FOREACH(head, cbp) { /* Skip bios which has to be read first. */ if (cbp->bio_data == NULL) { pbp = NULL; continue; } /* There is no previous bio yet. */ if (pbp == NULL) { pbp = cbp; continue; } /* Is this a neighbour bio? */ if (pbp->bio_offset + pbp->bio_length != cbp->bio_offset) { /* Be sure that bios queue is sorted. */ KASSERT(pbp->bio_offset + pbp->bio_length < cbp->bio_offset, ("poffset=%jd plength=%jd coffset=%jd", (intmax_t)pbp->bio_offset, (intmax_t)pbp->bio_length, (intmax_t)cbp->bio_offset)); pbp = cbp; continue; } /* Be sure we don't end up with too big bio. */ if (pbp->bio_length + cbp->bio_length > MAXPHYS) { pbp = cbp; continue; } /* Ok, we can join bios. */ GJ_LOGREQ(4, pbp, "Join: "); GJ_LOGREQ(4, cbp, "and: "); pbp->bio_data = gj_realloc(pbp->bio_data, pbp->bio_length + cbp->bio_length, pbp->bio_length); bcopy(cbp->bio_data, pbp->bio_data + pbp->bio_length, cbp->bio_length); gj_free(cbp->bio_data, cbp->bio_length); pbp->bio_length += cbp->bio_length; pbp->bio_next = cbp->bio_next; g_destroy_bio(cbp); cbp = pbp; g_journal_stats_combined_ios++; n--; GJ_LOGREQ(4, pbp, "Got: "); } return (n); } /* * TODO: Update comment. * These are functions responsible for copying one portion of data from journal * to the destination provider. * The order goes like this: * 1. Read the header, which contains informations about data blocks * following it. * 2. Read the data blocks from the journal. * 3. Write the data blocks on the data provider. * * g_journal_copy_start() * g_journal_copy_done() - got finished write request, logs potential errors. */ /* * When there is no data in cache, this function is used to read it. */ static void g_journal_read_first(struct g_journal_softc *sc, struct bio *bp) { struct bio *cbp; /* * We were short in memory, so data was freed. * In that case we need to read it back from journal. */ cbp = g_alloc_bio(); cbp->bio_cflags = bp->bio_cflags; cbp->bio_parent = bp; cbp->bio_offset = bp->bio_joffset; cbp->bio_length = bp->bio_length; cbp->bio_data = gj_malloc(bp->bio_length, M_WAITOK); cbp->bio_cmd = BIO_READ; cbp->bio_done = g_journal_std_done; GJ_LOGREQ(4, cbp, "READ FIRST"); g_io_request(cbp, sc->sc_jconsumer); g_journal_cache_misses++; } static void g_journal_copy_send(struct g_journal_softc *sc) { struct bio *bioq, *bp, *lbp; bioq = lbp = NULL; mtx_lock(&sc->sc_mtx); for (; sc->sc_copy_in_progress < g_journal_parallel_copies;) { bp = GJQ_FIRST(sc->sc_inactive.jj_queue); if (bp == NULL) break; GJQ_REMOVE(sc->sc_inactive.jj_queue, bp); sc->sc_copy_in_progress++; GJQ_INSERT_AFTER(bioq, bp, lbp); lbp = bp; } mtx_unlock(&sc->sc_mtx); if (g_journal_do_optimize) sc->sc_copy_in_progress += g_journal_optimize(bioq); while ((bp = GJQ_FIRST(bioq)) != NULL) { GJQ_REMOVE(bioq, bp); GJQ_INSERT_HEAD(sc->sc_copy_queue, bp); bp->bio_cflags = GJ_BIO_COPY; if (bp->bio_data == NULL) g_journal_read_first(sc, bp); else { bp->bio_joffset = 0; GJ_LOGREQ(4, bp, "SEND"); g_io_request(bp, sc->sc_dconsumer); } } } static void g_journal_copy_start(struct g_journal_softc *sc) { /* * Remember in metadata that we're starting to copy journaled data * to the data provider. * In case of power failure, we will copy these data once again on boot. */ if (!sc->sc_journal_copying) { sc->sc_journal_copying = 1; GJ_DEBUG(1, "Starting copy of journal."); g_journal_metadata_update(sc); } g_journal_copy_send(sc); } /* * Data block has been read from the journal provider. */ static int g_journal_copy_read_done(struct bio *bp) { struct g_journal_softc *sc; struct g_consumer *cp; struct bio *pbp; KASSERT(bp->bio_cflags == GJ_BIO_COPY, ("Invalid bio (%d != %d).", bp->bio_cflags, GJ_BIO_COPY)); sc = bp->bio_from->geom->softc; pbp = bp->bio_parent; if (bp->bio_error != 0) { GJ_DEBUG(0, "Error while reading data from %s (error=%d).", bp->bio_to->name, bp->bio_error); /* * We will not be able to deliver WRITE request as well. */ gj_free(bp->bio_data, bp->bio_length); g_destroy_bio(pbp); g_destroy_bio(bp); sc->sc_copy_in_progress--; return (1); } pbp->bio_data = bp->bio_data; cp = sc->sc_dconsumer; g_io_request(pbp, cp); GJ_LOGREQ(4, bp, "READ DONE"); g_destroy_bio(bp); return (0); } /* * Data block has been written to the data provider. */ static void g_journal_copy_write_done(struct bio *bp) { struct g_journal_softc *sc; KASSERT(bp->bio_cflags == GJ_BIO_COPY, ("Invalid bio (%d != %d).", bp->bio_cflags, GJ_BIO_COPY)); sc = bp->bio_from->geom->softc; sc->sc_copy_in_progress--; if (bp->bio_error != 0) { GJ_LOGREQ(0, bp, "[copy] Error while writing data (error=%d)", bp->bio_error); } GJQ_REMOVE(sc->sc_copy_queue, bp); gj_free(bp->bio_data, bp->bio_length); GJ_LOGREQ(4, bp, "DONE"); g_destroy_bio(bp); if (sc->sc_copy_in_progress == 0) { /* * This was the last write request for this journal. */ GJ_DEBUG(1, "Data has been copied."); sc->sc_journal_copying = 0; } } static void g_journal_flush_done(struct bio *bp); /* * Flush one record onto active journal provider. */ static void g_journal_flush(struct g_journal_softc *sc) { struct g_journal_record_header hdr; struct g_journal_entry *ent; struct g_provider *pp; struct bio **bioq; struct bio *bp, *fbp, *pbp; off_t joffset, size; u_char *data, hash[16]; MD5_CTX ctx; u_int i; if (sc->sc_current_count == 0) return; size = 0; pp = sc->sc_jprovider; GJ_VALIDATE_OFFSET(sc->sc_journal_offset, sc); joffset = sc->sc_journal_offset; GJ_DEBUG(2, "Storing %d journal entries on %s at %jd.", sc->sc_current_count, pp->name, (intmax_t)joffset); /* * Store 'journal id', so we know to which journal this record belongs. */ hdr.jrh_journal_id = sc->sc_journal_id; /* Could be less than g_journal_record_entries if called due timeout. */ hdr.jrh_nentries = MIN(sc->sc_current_count, g_journal_record_entries); strlcpy(hdr.jrh_magic, GJ_RECORD_HEADER_MAGIC, sizeof(hdr.jrh_magic)); bioq = &sc->sc_active.jj_queue; pbp = sc->sc_flush_queue; fbp = g_alloc_bio(); fbp->bio_parent = NULL; fbp->bio_cflags = GJ_BIO_JOURNAL; fbp->bio_offset = -1; fbp->bio_joffset = joffset; fbp->bio_length = pp->sectorsize; fbp->bio_cmd = BIO_WRITE; fbp->bio_done = g_journal_std_done; GJQ_INSERT_AFTER(sc->sc_flush_queue, fbp, pbp); pbp = fbp; fbp->bio_to = pp; GJ_LOGREQ(4, fbp, "FLUSH_OUT"); joffset += pp->sectorsize; sc->sc_flush_count++; if (sc->sc_flags & GJF_DEVICE_CHECKSUM) MD5Init(&ctx); for (i = 0; i < hdr.jrh_nentries; i++) { bp = sc->sc_current_queue; KASSERT(bp != NULL, ("NULL bp")); bp->bio_to = pp; GJ_LOGREQ(4, bp, "FLUSHED"); sc->sc_current_queue = bp->bio_next; bp->bio_next = NULL; sc->sc_current_count--; /* Add to the header. */ ent = &hdr.jrh_entries[i]; ent->je_offset = bp->bio_offset; ent->je_joffset = joffset; ent->je_length = bp->bio_length; size += ent->je_length; data = bp->bio_data; if (sc->sc_flags & GJF_DEVICE_CHECKSUM) MD5Update(&ctx, data, ent->je_length); bzero(bp, sizeof(*bp)); bp->bio_cflags = GJ_BIO_JOURNAL; bp->bio_offset = ent->je_offset; bp->bio_joffset = ent->je_joffset; bp->bio_length = ent->je_length; bp->bio_data = data; bp->bio_cmd = BIO_WRITE; bp->bio_done = g_journal_std_done; GJQ_INSERT_AFTER(sc->sc_flush_queue, bp, pbp); pbp = bp; bp->bio_to = pp; GJ_LOGREQ(4, bp, "FLUSH_OUT"); joffset += bp->bio_length; sc->sc_flush_count++; /* * Add request to the active sc_journal_queue queue. * This is our cache. After journal switch we don't have to * read the data from the inactive journal, because we keep * it in memory. */ g_journal_insert(bioq, ent->je_offset, ent->je_offset + ent->je_length, ent->je_joffset, data, M_NOWAIT); } /* * After all requests, store valid header. */ data = gj_malloc(pp->sectorsize, M_WAITOK); if (sc->sc_flags & GJF_DEVICE_CHECKSUM) { MD5Final(hash, &ctx); bcopy(hash, hdr.jrh_sum, sizeof(hdr.jrh_sum)); } g_journal_record_header_encode(&hdr, data); fbp->bio_data = data; sc->sc_journal_offset = joffset; g_journal_check_overflow(sc); } /* * Flush request finished. */ static void g_journal_flush_done(struct bio *bp) { struct g_journal_softc *sc; struct g_consumer *cp; KASSERT((bp->bio_cflags & GJ_BIO_MASK) == GJ_BIO_JOURNAL, ("Invalid bio (%d != %d).", bp->bio_cflags, GJ_BIO_JOURNAL)); cp = bp->bio_from; sc = cp->geom->softc; sc->sc_flush_in_progress--; if (bp->bio_error != 0) { GJ_LOGREQ(0, bp, "[flush] Error while writing data (error=%d)", bp->bio_error); } gj_free(bp->bio_data, bp->bio_length); GJ_LOGREQ(4, bp, "DONE"); g_destroy_bio(bp); } static void g_journal_release_delayed(struct g_journal_softc *sc); static void g_journal_flush_send(struct g_journal_softc *sc) { struct g_consumer *cp; struct bio *bioq, *bp, *lbp; cp = sc->sc_jconsumer; bioq = lbp = NULL; while (sc->sc_flush_in_progress < g_journal_parallel_flushes) { /* Send one flush requests to the active journal. */ bp = GJQ_FIRST(sc->sc_flush_queue); if (bp != NULL) { GJQ_REMOVE(sc->sc_flush_queue, bp); sc->sc_flush_count--; bp->bio_offset = bp->bio_joffset; bp->bio_joffset = 0; sc->sc_flush_in_progress++; GJQ_INSERT_AFTER(bioq, bp, lbp); lbp = bp; } /* Try to release delayed requests. */ g_journal_release_delayed(sc); /* If there are no requests to flush, leave. */ if (GJQ_FIRST(sc->sc_flush_queue) == NULL) break; } if (g_journal_do_optimize) sc->sc_flush_in_progress += g_journal_optimize(bioq); while ((bp = GJQ_FIRST(bioq)) != NULL) { GJQ_REMOVE(bioq, bp); GJ_LOGREQ(3, bp, "Flush request send"); g_io_request(bp, cp); } } static void g_journal_add_current(struct g_journal_softc *sc, struct bio *bp) { int n; GJ_LOGREQ(4, bp, "CURRENT %d", sc->sc_current_count); n = g_journal_insert_bio(&sc->sc_current_queue, bp, M_WAITOK); sc->sc_current_count += n; n = g_journal_optimize(sc->sc_current_queue); sc->sc_current_count += n; /* * For requests which are added to the current queue we deliver * response immediately. */ bp->bio_completed = bp->bio_length; g_io_deliver(bp, 0); if (sc->sc_current_count >= g_journal_record_entries) { /* * Let's flush one record onto active journal provider. */ g_journal_flush(sc); } } static void g_journal_release_delayed(struct g_journal_softc *sc) { struct bio *bp; for (;;) { /* The flush queue is full, exit. */ if (sc->sc_flush_count >= g_journal_accept_immediately) return; bp = bioq_takefirst(&sc->sc_delayed_queue); if (bp == NULL) return; sc->sc_delayed_count--; g_journal_add_current(sc, bp); } } /* * Add I/O request to the current queue. If we have enough requests for one * journal record we flush them onto active journal provider. */ static void g_journal_add_request(struct g_journal_softc *sc, struct bio *bp) { /* * The flush queue is full, we need to delay the request. */ if (sc->sc_delayed_count > 0 || sc->sc_flush_count >= g_journal_accept_immediately) { GJ_LOGREQ(4, bp, "DELAYED"); bioq_insert_tail(&sc->sc_delayed_queue, bp); sc->sc_delayed_count++; return; } KASSERT(TAILQ_EMPTY(&sc->sc_delayed_queue.queue), ("DELAYED queue not empty.")); g_journal_add_current(sc, bp); } static void g_journal_read_done(struct bio *bp); /* * Try to find requested data in cache. */ static struct bio * g_journal_read_find(struct bio *head, int sorted, struct bio *pbp, off_t ostart, off_t oend) { off_t cstart, cend; struct bio *bp; GJQ_FOREACH(head, bp) { if (bp->bio_offset == -1) continue; cstart = MAX(ostart, bp->bio_offset); cend = MIN(oend, bp->bio_offset + bp->bio_length); if (cend <= ostart) continue; else if (cstart >= oend) { if (!sorted) continue; else { bp = NULL; break; } } if (bp->bio_data == NULL) break; GJ_DEBUG(3, "READ(%p): (%jd, %jd) (bp=%p)", head, cstart, cend, bp); bcopy(bp->bio_data + cstart - bp->bio_offset, pbp->bio_data + cstart - pbp->bio_offset, cend - cstart); pbp->bio_completed += cend - cstart; if (pbp->bio_completed == pbp->bio_length) { /* * Cool, the whole request was in cache, deliver happy * message. */ g_io_deliver(pbp, 0); return (pbp); } break; } return (bp); } /* * Try to find requested data in cache. */ static struct bio * g_journal_read_queue_find(struct bio_queue *head, struct bio *pbp, off_t ostart, off_t oend) { off_t cstart, cend; struct bio *bp; TAILQ_FOREACH(bp, head, bio_queue) { cstart = MAX(ostart, bp->bio_offset); cend = MIN(oend, bp->bio_offset + bp->bio_length); if (cend <= ostart) continue; else if (cstart >= oend) continue; KASSERT(bp->bio_data != NULL, ("%s: bio_data == NULL", __func__)); GJ_DEBUG(3, "READ(%p): (%jd, %jd) (bp=%p)", head, cstart, cend, bp); bcopy(bp->bio_data + cstart - bp->bio_offset, pbp->bio_data + cstart - pbp->bio_offset, cend - cstart); pbp->bio_completed += cend - cstart; if (pbp->bio_completed == pbp->bio_length) { /* * Cool, the whole request was in cache, deliver happy * message. */ g_io_deliver(pbp, 0); return (pbp); } break; } return (bp); } /* * This function is used for colecting data on read. * The complexity is because parts of the data can be stored in four different * places: * - in delayed requests * - in memory - the data not yet send to the active journal provider * - in requests which are going to be sent to the active journal * - in the active journal * - in the inactive journal * - in the data provider */ static void g_journal_read(struct g_journal_softc *sc, struct bio *pbp, off_t ostart, off_t oend) { struct bio *bp, *nbp, *head; off_t cstart, cend; u_int i, sorted = 0; GJ_DEBUG(3, "READ: (%jd, %jd)", ostart, oend); cstart = cend = -1; bp = NULL; head = NULL; for (i = 0; i <= 5; i++) { switch (i) { case 0: /* Delayed requests. */ head = NULL; sorted = 0; break; case 1: /* Not-yet-send data. */ head = sc->sc_current_queue; sorted = 1; break; case 2: /* In-flight to the active journal. */ head = sc->sc_flush_queue; sorted = 0; break; case 3: /* Active journal. */ head = sc->sc_active.jj_queue; sorted = 1; break; case 4: /* Inactive journal. */ /* * XXX: Here could be a race with g_journal_lowmem(). */ head = sc->sc_inactive.jj_queue; sorted = 1; break; case 5: /* In-flight to the data provider. */ head = sc->sc_copy_queue; sorted = 0; break; default: panic("gjournal %s: i=%d", __func__, i); } if (i == 0) bp = g_journal_read_queue_find(&sc->sc_delayed_queue.queue, pbp, ostart, oend); else bp = g_journal_read_find(head, sorted, pbp, ostart, oend); if (bp == pbp) { /* Got the whole request. */ GJ_DEBUG(2, "Got the whole request from %u.", i); return; } else if (bp != NULL) { cstart = MAX(ostart, bp->bio_offset); cend = MIN(oend, bp->bio_offset + bp->bio_length); GJ_DEBUG(2, "Got part of the request from %u (%jd-%jd).", i, (intmax_t)cstart, (intmax_t)cend); break; } } if (bp != NULL) { if (bp->bio_data == NULL) { nbp = g_duplicate_bio(pbp); nbp->bio_cflags = GJ_BIO_READ; nbp->bio_data = pbp->bio_data + cstart - pbp->bio_offset; nbp->bio_offset = bp->bio_joffset + cstart - bp->bio_offset; nbp->bio_length = cend - cstart; nbp->bio_done = g_journal_read_done; g_io_request(nbp, sc->sc_jconsumer); } /* * If we don't have the whole request yet, call g_journal_read() * recursively. */ if (ostart < cstart) g_journal_read(sc, pbp, ostart, cstart); if (oend > cend) g_journal_read(sc, pbp, cend, oend); } else { /* * No data in memory, no data in journal. * Its time for asking data provider. */ GJ_DEBUG(3, "READ(data): (%jd, %jd)", ostart, oend); nbp = g_duplicate_bio(pbp); nbp->bio_cflags = GJ_BIO_READ; nbp->bio_data = pbp->bio_data + ostart - pbp->bio_offset; nbp->bio_offset = ostart; nbp->bio_length = oend - ostart; nbp->bio_done = g_journal_read_done; g_io_request(nbp, sc->sc_dconsumer); /* We have the whole request, return here. */ return; } } /* * Function responsible for handling finished READ requests. * Actually, g_std_done() could be used here, the only difference is that we * log error. */ static void g_journal_read_done(struct bio *bp) { struct bio *pbp; KASSERT(bp->bio_cflags == GJ_BIO_READ, ("Invalid bio (%d != %d).", bp->bio_cflags, GJ_BIO_READ)); pbp = bp->bio_parent; pbp->bio_inbed++; pbp->bio_completed += bp->bio_length; if (bp->bio_error != 0) { if (pbp->bio_error == 0) pbp->bio_error = bp->bio_error; GJ_DEBUG(0, "Error while reading data from %s (error=%d).", bp->bio_to->name, bp->bio_error); } g_destroy_bio(bp); if (pbp->bio_children == pbp->bio_inbed && pbp->bio_completed == pbp->bio_length) { /* We're done. */ g_io_deliver(pbp, 0); } } /* * Deactive current journal and active next one. */ static void g_journal_switch(struct g_journal_softc *sc) { struct g_provider *pp; if (JEMPTY(sc)) { GJ_DEBUG(3, "No need for %s switch.", sc->sc_name); pp = LIST_FIRST(&sc->sc_geom->provider); if (!(sc->sc_flags & GJF_DEVICE_CLEAN) && pp->acw == 0) { sc->sc_flags |= GJF_DEVICE_CLEAN; GJ_DEBUG(1, "Marking %s as clean.", sc->sc_name); g_journal_metadata_update(sc); } } else { GJ_DEBUG(3, "Switching journal %s.", sc->sc_geom->name); pp = sc->sc_jprovider; sc->sc_journal_previous_id = sc->sc_journal_id; sc->sc_journal_id = sc->sc_journal_next_id; sc->sc_journal_next_id = arc4random(); GJ_VALIDATE_OFFSET(sc->sc_journal_offset, sc); g_journal_write_header(sc); sc->sc_inactive.jj_offset = sc->sc_active.jj_offset; sc->sc_inactive.jj_queue = sc->sc_active.jj_queue; sc->sc_active.jj_offset = sc->sc_journal_offset - pp->sectorsize; sc->sc_active.jj_queue = NULL; /* * Switch is done, start copying data from the (now) inactive * journal to the data provider. */ g_journal_copy_start(sc); } mtx_lock(&sc->sc_mtx); sc->sc_flags &= ~GJF_DEVICE_SWITCH; mtx_unlock(&sc->sc_mtx); } static void g_journal_initialize(struct g_journal_softc *sc) { sc->sc_journal_id = arc4random(); sc->sc_journal_next_id = arc4random(); sc->sc_journal_previous_id = sc->sc_journal_id; sc->sc_journal_offset = sc->sc_jstart; sc->sc_inactive.jj_offset = sc->sc_jstart; g_journal_write_header(sc); sc->sc_active.jj_offset = sc->sc_jstart; } static void g_journal_mark_as_dirty(struct g_journal_softc *sc) { const struct g_journal_desc *desc; int i; GJ_DEBUG(1, "Marking file system %s as dirty.", sc->sc_name); for (i = 0; (desc = g_journal_filesystems[i]) != NULL; i++) desc->jd_dirty(sc->sc_dconsumer); } /* * Function read record header from the given journal. * It is very simlar to g_read_data(9), but it doesn't allocate memory for bio * and data on every call. */ static int g_journal_sync_read(struct g_consumer *cp, struct bio *bp, off_t offset, void *data) { int error; bzero(bp, sizeof(*bp)); bp->bio_cmd = BIO_READ; bp->bio_done = NULL; bp->bio_offset = offset; bp->bio_length = cp->provider->sectorsize; bp->bio_data = data; g_io_request(bp, cp); error = biowait(bp, "gjs_read"); return (error); } #if 0 /* * Function is called when we start the journal device and we detect that * one of the journals was not fully copied. * The purpose of this function is to read all records headers from journal * and placed them in the inactive queue, so we can start journal * synchronization process and the journal provider itself. * Design decision was taken to not synchronize the whole journal here as it * can take too much time. Reading headers only and delaying synchronization * process until after journal provider is started should be the best choice. */ #endif static void g_journal_sync(struct g_journal_softc *sc) { struct g_journal_record_header rhdr; struct g_journal_entry *ent; struct g_journal_header jhdr; struct g_consumer *cp; struct bio *bp, *fbp, *tbp; off_t joffset, offset; u_char *buf, sum[16]; uint64_t id; MD5_CTX ctx; int error, found, i; found = 0; fbp = NULL; cp = sc->sc_jconsumer; bp = g_alloc_bio(); buf = gj_malloc(cp->provider->sectorsize, M_WAITOK); offset = joffset = sc->sc_inactive.jj_offset = sc->sc_journal_offset; GJ_DEBUG(2, "Looking for termination at %jd.", (intmax_t)joffset); /* * Read and decode first journal header. */ error = g_journal_sync_read(cp, bp, offset, buf); if (error != 0) { GJ_DEBUG(0, "Error while reading journal header from %s.", cp->provider->name); goto end; } error = g_journal_header_decode(buf, &jhdr); if (error != 0) { GJ_DEBUG(0, "Cannot decode journal header from %s.", cp->provider->name); goto end; } id = sc->sc_journal_id; if (jhdr.jh_journal_id != sc->sc_journal_id) { GJ_DEBUG(1, "Journal ID mismatch at %jd (0x%08x != 0x%08x).", (intmax_t)offset, (u_int)jhdr.jh_journal_id, (u_int)id); goto end; } offset += cp->provider->sectorsize; id = sc->sc_journal_next_id = jhdr.jh_journal_next_id; for (;;) { /* * If the biggest record won't fit, look for a record header or * journal header from the begining. */ GJ_VALIDATE_OFFSET(offset, sc); error = g_journal_sync_read(cp, bp, offset, buf); if (error != 0) { /* * Not good. Having an error while reading header * means, that we cannot read next headers and in * consequence we cannot find termination. */ GJ_DEBUG(0, "Error while reading record header from %s.", cp->provider->name); break; } error = g_journal_record_header_decode(buf, &rhdr); if (error != 0) { GJ_DEBUG(2, "Not a record header at %jd (error=%d).", (intmax_t)offset, error); /* * This is not a record header. * If we are lucky, this is next journal header. */ error = g_journal_header_decode(buf, &jhdr); if (error != 0) { GJ_DEBUG(1, "Not a journal header at %jd (error=%d).", (intmax_t)offset, error); /* * Nope, this is not journal header, which * bascially means that journal is not * terminated properly. */ error = ENOENT; break; } /* * Ok. This is header of _some_ journal. Now we need to * verify if this is header of the _next_ journal. */ if (jhdr.jh_journal_id != id) { GJ_DEBUG(1, "Journal ID mismatch at %jd " "(0x%08x != 0x%08x).", (intmax_t)offset, (u_int)jhdr.jh_journal_id, (u_int)id); error = ENOENT; break; } /* Found termination. */ found++; GJ_DEBUG(1, "Found termination at %jd (id=0x%08x).", (intmax_t)offset, (u_int)id); sc->sc_active.jj_offset = offset; sc->sc_journal_offset = offset + cp->provider->sectorsize; sc->sc_journal_id = id; id = sc->sc_journal_next_id = jhdr.jh_journal_next_id; while ((tbp = fbp) != NULL) { fbp = tbp->bio_next; GJ_LOGREQ(3, tbp, "Adding request."); g_journal_insert_bio(&sc->sc_inactive.jj_queue, tbp, M_WAITOK); } /* Skip journal's header. */ offset += cp->provider->sectorsize; continue; } /* Skip record's header. */ offset += cp->provider->sectorsize; /* * Add information about every record entry to the inactive * queue. */ if (sc->sc_flags & GJF_DEVICE_CHECKSUM) MD5Init(&ctx); for (i = 0; i < rhdr.jrh_nentries; i++) { ent = &rhdr.jrh_entries[i]; GJ_DEBUG(3, "Insert entry: %jd %jd.", (intmax_t)ent->je_offset, (intmax_t)ent->je_length); g_journal_insert(&fbp, ent->je_offset, ent->je_offset + ent->je_length, ent->je_joffset, NULL, M_WAITOK); if (sc->sc_flags & GJF_DEVICE_CHECKSUM) { u_char *buf2; /* * TODO: Should use faster function (like * g_journal_sync_read()). */ buf2 = g_read_data(cp, offset, ent->je_length, NULL); if (buf2 == NULL) GJ_DEBUG(0, "Cannot read data at %jd.", (intmax_t)offset); else { MD5Update(&ctx, buf2, ent->je_length); g_free(buf2); } } /* Skip entry's data. */ offset += ent->je_length; } if (sc->sc_flags & GJF_DEVICE_CHECKSUM) { MD5Final(sum, &ctx); if (bcmp(sum, rhdr.jrh_sum, sizeof(rhdr.jrh_sum)) != 0) { GJ_DEBUG(0, "MD5 hash mismatch at %jd!", (intmax_t)offset); } } } end: gj_free(bp->bio_data, cp->provider->sectorsize); g_destroy_bio(bp); /* Remove bios from unterminated journal. */ while ((tbp = fbp) != NULL) { fbp = tbp->bio_next; g_destroy_bio(tbp); } if (found < 1 && joffset > 0) { GJ_DEBUG(0, "Journal on %s is broken/corrupted. Initializing.", sc->sc_name); while ((tbp = sc->sc_inactive.jj_queue) != NULL) { sc->sc_inactive.jj_queue = tbp->bio_next; g_destroy_bio(tbp); } g_journal_initialize(sc); g_journal_mark_as_dirty(sc); } else { GJ_DEBUG(0, "Journal %s consistent.", sc->sc_name); g_journal_copy_start(sc); } } /* * Wait for requests. * If we have requests in the current queue, flush them after 3 seconds from the * last flush. In this way we don't wait forever (or for journal switch) with * storing not full records on journal. */ static void g_journal_wait(struct g_journal_softc *sc, time_t last_write) { int error, timeout; GJ_DEBUG(3, "%s: enter", __func__); if (sc->sc_current_count == 0) { if (g_journal_debug < 2) msleep(sc, &sc->sc_mtx, PRIBIO | PDROP, "gj:work", 0); else { /* * If we have debug turned on, show number of elements * in various queues. */ for (;;) { error = msleep(sc, &sc->sc_mtx, PRIBIO, "gj:work", hz * 3); if (error == 0) { mtx_unlock(&sc->sc_mtx); break; } GJ_DEBUG(3, "Report: current count=%d", sc->sc_current_count); GJ_DEBUG(3, "Report: flush count=%d", sc->sc_flush_count); GJ_DEBUG(3, "Report: flush in progress=%d", sc->sc_flush_in_progress); GJ_DEBUG(3, "Report: copy in progress=%d", sc->sc_copy_in_progress); GJ_DEBUG(3, "Report: delayed=%d", sc->sc_delayed_count); } } GJ_DEBUG(3, "%s: exit 1", __func__); return; } /* * Flush even not full records every 3 seconds. */ timeout = (last_write + 3 - time_second) * hz; if (timeout <= 0) { mtx_unlock(&sc->sc_mtx); g_journal_flush(sc); g_journal_flush_send(sc); GJ_DEBUG(3, "%s: exit 2", __func__); return; } error = msleep(sc, &sc->sc_mtx, PRIBIO | PDROP, "gj:work", timeout); if (error == EWOULDBLOCK) g_journal_flush_send(sc); GJ_DEBUG(3, "%s: exit 3", __func__); } /* * Worker thread. */ static void g_journal_worker(void *arg) { struct g_journal_softc *sc; struct g_geom *gp; struct g_provider *pp; struct bio *bp; time_t last_write; int type; thread_lock(curthread); sched_prio(curthread, PRIBIO); thread_unlock(curthread); sc = arg; type = 0; /* gcc */ if (sc->sc_flags & GJF_DEVICE_CLEAN) { GJ_DEBUG(0, "Journal %s clean.", sc->sc_name); g_journal_initialize(sc); } else { g_journal_sync(sc); } /* * Check if we can use BIO_FLUSH. */ sc->sc_bio_flush = 0; if (g_io_flush(sc->sc_jconsumer) == 0) { sc->sc_bio_flush |= GJ_FLUSH_JOURNAL; GJ_DEBUG(1, "BIO_FLUSH supported by %s.", sc->sc_jconsumer->provider->name); } else { GJ_DEBUG(0, "BIO_FLUSH not supported by %s.", sc->sc_jconsumer->provider->name); } if (sc->sc_jconsumer != sc->sc_dconsumer) { if (g_io_flush(sc->sc_dconsumer) == 0) { sc->sc_bio_flush |= GJ_FLUSH_DATA; GJ_DEBUG(1, "BIO_FLUSH supported by %s.", sc->sc_dconsumer->provider->name); } else { GJ_DEBUG(0, "BIO_FLUSH not supported by %s.", sc->sc_dconsumer->provider->name); } } gp = sc->sc_geom; g_topology_lock(); pp = g_new_providerf(gp, "%s.journal", sc->sc_name); - KASSERT(pp != NULL, ("Cannot create %s.journal.", sc->sc_name)); pp->mediasize = sc->sc_mediasize; /* * There could be a problem when data provider and journal providers * have different sectorsize, but such scenario is prevented on journal * creation. */ pp->sectorsize = sc->sc_sectorsize; g_error_provider(pp, 0); g_topology_unlock(); last_write = time_second; if (sc->sc_rootmount != NULL) { GJ_DEBUG(1, "root_mount_rel %p", sc->sc_rootmount); root_mount_rel(sc->sc_rootmount); sc->sc_rootmount = NULL; } for (;;) { /* Get first request from the queue. */ mtx_lock(&sc->sc_mtx); bp = bioq_first(&sc->sc_back_queue); if (bp != NULL) type = (bp->bio_cflags & GJ_BIO_MASK); if (bp == NULL) { bp = bioq_first(&sc->sc_regular_queue); if (bp != NULL) type = GJ_BIO_REGULAR; } if (bp == NULL) { try_switch: if ((sc->sc_flags & GJF_DEVICE_SWITCH) || (sc->sc_flags & GJF_DEVICE_DESTROY)) { if (sc->sc_current_count > 0) { mtx_unlock(&sc->sc_mtx); g_journal_flush(sc); g_journal_flush_send(sc); continue; } if (sc->sc_flush_in_progress > 0) goto sleep; if (sc->sc_copy_in_progress > 0) goto sleep; } if (sc->sc_flags & GJF_DEVICE_SWITCH) { mtx_unlock(&sc->sc_mtx); g_journal_switch(sc); wakeup(&sc->sc_journal_copying); continue; } if (sc->sc_flags & GJF_DEVICE_DESTROY) { GJ_DEBUG(1, "Shutting down worker " "thread for %s.", gp->name); sc->sc_worker = NULL; wakeup(&sc->sc_worker); mtx_unlock(&sc->sc_mtx); kproc_exit(0); } sleep: g_journal_wait(sc, last_write); continue; } /* * If we're in switch process, we need to delay all new * write requests until its done. */ if ((sc->sc_flags & GJF_DEVICE_SWITCH) && type == GJ_BIO_REGULAR && bp->bio_cmd == BIO_WRITE) { GJ_LOGREQ(2, bp, "WRITE on SWITCH"); goto try_switch; } if (type == GJ_BIO_REGULAR) bioq_remove(&sc->sc_regular_queue, bp); else bioq_remove(&sc->sc_back_queue, bp); mtx_unlock(&sc->sc_mtx); switch (type) { case GJ_BIO_REGULAR: /* Regular request. */ switch (bp->bio_cmd) { case BIO_READ: g_journal_read(sc, bp, bp->bio_offset, bp->bio_offset + bp->bio_length); break; case BIO_WRITE: last_write = time_second; g_journal_add_request(sc, bp); g_journal_flush_send(sc); break; default: panic("Invalid bio_cmd (%d).", bp->bio_cmd); } break; case GJ_BIO_COPY: switch (bp->bio_cmd) { case BIO_READ: if (g_journal_copy_read_done(bp)) g_journal_copy_send(sc); break; case BIO_WRITE: g_journal_copy_write_done(bp); g_journal_copy_send(sc); break; default: panic("Invalid bio_cmd (%d).", bp->bio_cmd); } break; case GJ_BIO_JOURNAL: g_journal_flush_done(bp); g_journal_flush_send(sc); break; case GJ_BIO_READ: default: panic("Invalid bio (%d).", type); } } } static void g_journal_destroy_event(void *arg, int flags __unused) { struct g_journal_softc *sc; g_topology_assert(); sc = arg; g_journal_destroy(sc); } static void g_journal_timeout(void *arg) { struct g_journal_softc *sc; sc = arg; GJ_DEBUG(0, "Timeout. Journal %s cannot be completed.", sc->sc_geom->name); g_post_event(g_journal_destroy_event, sc, M_NOWAIT, NULL); } static struct g_geom * g_journal_create(struct g_class *mp, struct g_provider *pp, const struct g_journal_metadata *md) { struct g_journal_softc *sc; struct g_geom *gp; struct g_consumer *cp; int error; sc = NULL; /* gcc */ g_topology_assert(); /* * There are two possibilities: * 1. Data and both journals are on the same provider. * 2. Data and journals are all on separated providers. */ /* Look for journal device with the same ID. */ LIST_FOREACH(gp, &mp->geom, geom) { sc = gp->softc; if (sc == NULL) continue; if (sc->sc_id == md->md_id) break; } if (gp == NULL) sc = NULL; else if (sc != NULL && (sc->sc_type & md->md_type) != 0) { GJ_DEBUG(1, "Journal device %u already configured.", sc->sc_id); return (NULL); } if (md->md_type == 0 || (md->md_type & ~GJ_TYPE_COMPLETE) != 0) { GJ_DEBUG(0, "Invalid type on %s.", pp->name); return (NULL); } if (md->md_type & GJ_TYPE_DATA) { GJ_DEBUG(0, "Journal %u: %s contains data.", md->md_id, pp->name); } if (md->md_type & GJ_TYPE_JOURNAL) { GJ_DEBUG(0, "Journal %u: %s contains journal.", md->md_id, pp->name); } if (sc == NULL) { /* Action geom. */ sc = malloc(sizeof(*sc), M_JOURNAL, M_WAITOK | M_ZERO); sc->sc_id = md->md_id; sc->sc_type = 0; sc->sc_flags = 0; sc->sc_worker = NULL; gp = g_new_geomf(mp, "gjournal %u", sc->sc_id); gp->start = g_journal_start; gp->orphan = g_journal_orphan; gp->access = g_journal_access; gp->softc = sc; gp->flags |= G_GEOM_VOLATILE_BIO; sc->sc_geom = gp; mtx_init(&sc->sc_mtx, "gjournal", NULL, MTX_DEF); bioq_init(&sc->sc_back_queue); bioq_init(&sc->sc_regular_queue); bioq_init(&sc->sc_delayed_queue); sc->sc_delayed_count = 0; sc->sc_current_queue = NULL; sc->sc_current_count = 0; sc->sc_flush_queue = NULL; sc->sc_flush_count = 0; sc->sc_flush_in_progress = 0; sc->sc_copy_queue = NULL; sc->sc_copy_in_progress = 0; sc->sc_inactive.jj_queue = NULL; sc->sc_active.jj_queue = NULL; sc->sc_rootmount = root_mount_hold("GJOURNAL"); GJ_DEBUG(1, "root_mount_hold %p", sc->sc_rootmount); callout_init(&sc->sc_callout, CALLOUT_MPSAFE); if (md->md_type != GJ_TYPE_COMPLETE) { /* * Journal and data are on separate providers. * At this point we have only one of them. * We setup a timeout in case the other part will not * appear, so we won't wait forever. */ callout_reset(&sc->sc_callout, 5 * hz, g_journal_timeout, sc); } } /* Remember type of the data provider. */ if (md->md_type & GJ_TYPE_DATA) sc->sc_orig_type = md->md_type; sc->sc_type |= md->md_type; cp = NULL; if (md->md_type & GJ_TYPE_DATA) { if (md->md_flags & GJ_FLAG_CLEAN) sc->sc_flags |= GJF_DEVICE_CLEAN; if (md->md_flags & GJ_FLAG_CHECKSUM) sc->sc_flags |= GJF_DEVICE_CHECKSUM; cp = g_new_consumer(gp); error = g_attach(cp, pp); KASSERT(error == 0, ("Cannot attach to %s (error=%d).", pp->name, error)); error = g_access(cp, 1, 1, 1); if (error != 0) { GJ_DEBUG(0, "Cannot access %s (error=%d).", pp->name, error); g_journal_destroy(sc); return (NULL); } sc->sc_dconsumer = cp; sc->sc_mediasize = pp->mediasize - pp->sectorsize; sc->sc_sectorsize = pp->sectorsize; sc->sc_jstart = md->md_jstart; sc->sc_jend = md->md_jend; if (md->md_provider[0] != '\0') sc->sc_flags |= GJF_DEVICE_HARDCODED; sc->sc_journal_offset = md->md_joffset; sc->sc_journal_id = md->md_jid; sc->sc_journal_previous_id = md->md_jid; } if (md->md_type & GJ_TYPE_JOURNAL) { if (cp == NULL) { cp = g_new_consumer(gp); error = g_attach(cp, pp); KASSERT(error == 0, ("Cannot attach to %s (error=%d).", pp->name, error)); error = g_access(cp, 1, 1, 1); if (error != 0) { GJ_DEBUG(0, "Cannot access %s (error=%d).", pp->name, error); g_journal_destroy(sc); return (NULL); } } else { /* * Journal is on the same provider as data, which means * that data provider ends where journal starts. */ sc->sc_mediasize = md->md_jstart; } sc->sc_jconsumer = cp; } if ((sc->sc_type & GJ_TYPE_COMPLETE) != GJ_TYPE_COMPLETE) { /* Journal is not complete yet. */ return (gp); } else { /* Journal complete, cancel timeout. */ callout_drain(&sc->sc_callout); } error = kproc_create(g_journal_worker, sc, &sc->sc_worker, 0, 0, "g_journal %s", sc->sc_name); if (error != 0) { GJ_DEBUG(0, "Cannot create worker thread for %s.journal.", sc->sc_name); g_journal_destroy(sc); return (NULL); } return (gp); } static void g_journal_destroy_consumer(void *arg, int flags __unused) { struct g_consumer *cp; g_topology_assert(); cp = arg; g_detach(cp); g_destroy_consumer(cp); } static int g_journal_destroy(struct g_journal_softc *sc) { struct g_geom *gp; struct g_provider *pp; struct g_consumer *cp; g_topology_assert(); if (sc == NULL) return (ENXIO); gp = sc->sc_geom; pp = LIST_FIRST(&gp->provider); if (pp != NULL) { if (pp->acr != 0 || pp->acw != 0 || pp->ace != 0) { GJ_DEBUG(1, "Device %s is still open (r%dw%de%d).", pp->name, pp->acr, pp->acw, pp->ace); return (EBUSY); } g_error_provider(pp, ENXIO); g_journal_flush(sc); g_journal_flush_send(sc); g_journal_switch(sc); } sc->sc_flags |= (GJF_DEVICE_DESTROY | GJF_DEVICE_CLEAN); g_topology_unlock(); if (sc->sc_rootmount != NULL) { GJ_DEBUG(1, "root_mount_rel %p", sc->sc_rootmount); root_mount_rel(sc->sc_rootmount); sc->sc_rootmount = NULL; } callout_drain(&sc->sc_callout); mtx_lock(&sc->sc_mtx); wakeup(sc); while (sc->sc_worker != NULL) msleep(&sc->sc_worker, &sc->sc_mtx, PRIBIO, "gj:destroy", 0); mtx_unlock(&sc->sc_mtx); if (pp != NULL) { GJ_DEBUG(1, "Marking %s as clean.", sc->sc_name); g_journal_metadata_update(sc); g_topology_lock(); pp->flags |= G_PF_WITHER; g_orphan_provider(pp, ENXIO); } else { g_topology_lock(); } mtx_destroy(&sc->sc_mtx); if (sc->sc_current_count != 0) { GJ_DEBUG(0, "Warning! Number of current requests %d.", sc->sc_current_count); } LIST_FOREACH(cp, &gp->consumer, consumer) { if (cp->acr + cp->acw + cp->ace > 0) g_access(cp, -1, -1, -1); /* * We keep all consumers open for writting, so if I'll detach * and destroy consumer here, I'll get providers for taste, so * journal will be started again. * Sending an event here, prevents this from happening. */ g_post_event(g_journal_destroy_consumer, cp, M_WAITOK, NULL); } gp->softc = NULL; g_wither_geom(gp, ENXIO); free(sc, M_JOURNAL); return (0); } static void g_journal_taste_orphan(struct g_consumer *cp) { KASSERT(1 == 0, ("%s called while tasting %s.", __func__, cp->provider->name)); } static struct g_geom * g_journal_taste(struct g_class *mp, struct g_provider *pp, int flags __unused) { struct g_journal_metadata md; struct g_consumer *cp; struct g_geom *gp; int error; g_topology_assert(); g_trace(G_T_TOPOLOGY, "%s(%s, %s)", __func__, mp->name, pp->name); GJ_DEBUG(2, "Tasting %s.", pp->name); if (pp->geom->class == mp) return (NULL); gp = g_new_geomf(mp, "journal:taste"); /* This orphan function should be never called. */ gp->orphan = g_journal_taste_orphan; cp = g_new_consumer(gp); g_attach(cp, pp); error = g_journal_metadata_read(cp, &md); g_detach(cp); g_destroy_consumer(cp); g_destroy_geom(gp); if (error != 0) return (NULL); gp = NULL; if (md.md_provider[0] != '\0' && !g_compare_names(md.md_provider, pp->name)) return (NULL); if (md.md_provsize != 0 && md.md_provsize != pp->mediasize) return (NULL); if (g_journal_debug >= 2) journal_metadata_dump(&md); gp = g_journal_create(mp, pp, &md); return (gp); } static struct g_journal_softc * g_journal_find_device(struct g_class *mp, const char *name) { struct g_journal_softc *sc; struct g_geom *gp; struct g_provider *pp; if (strncmp(name, "/dev/", 5) == 0) name += 5; LIST_FOREACH(gp, &mp->geom, geom) { sc = gp->softc; if (sc == NULL) continue; if (sc->sc_flags & GJF_DEVICE_DESTROY) continue; if ((sc->sc_type & GJ_TYPE_COMPLETE) != GJ_TYPE_COMPLETE) continue; pp = LIST_FIRST(&gp->provider); if (strcmp(sc->sc_name, name) == 0) return (sc); if (pp != NULL && strcmp(pp->name, name) == 0) return (sc); } return (NULL); } static void g_journal_ctl_destroy(struct gctl_req *req, struct g_class *mp) { struct g_journal_softc *sc; const char *name; char param[16]; int *nargs; int error, i; g_topology_assert(); nargs = gctl_get_paraml(req, "nargs", sizeof(*nargs)); if (nargs == NULL) { gctl_error(req, "No '%s' argument.", "nargs"); return; } if (*nargs <= 0) { gctl_error(req, "Missing device(s)."); return; } for (i = 0; i < *nargs; i++) { snprintf(param, sizeof(param), "arg%d", i); name = gctl_get_asciiparam(req, param); if (name == NULL) { gctl_error(req, "No 'arg%d' argument.", i); return; } sc = g_journal_find_device(mp, name); if (sc == NULL) { gctl_error(req, "No such device: %s.", name); return; } error = g_journal_destroy(sc); if (error != 0) { gctl_error(req, "Cannot destroy device %s (error=%d).", LIST_FIRST(&sc->sc_geom->provider)->name, error); return; } } } static void g_journal_ctl_sync(struct gctl_req *req __unused, struct g_class *mp __unused) { g_topology_assert(); g_topology_unlock(); g_journal_sync_requested++; wakeup(&g_journal_switcher_state); while (g_journal_sync_requested > 0) tsleep(&g_journal_sync_requested, PRIBIO, "j:sreq", hz / 2); g_topology_lock(); } static void g_journal_config(struct gctl_req *req, struct g_class *mp, const char *verb) { uint32_t *version; g_topology_assert(); version = gctl_get_paraml(req, "version", sizeof(*version)); if (version == NULL) { gctl_error(req, "No '%s' argument.", "version"); return; } if (*version != G_JOURNAL_VERSION) { gctl_error(req, "Userland and kernel parts are out of sync."); return; } if (strcmp(verb, "destroy") == 0 || strcmp(verb, "stop") == 0) { g_journal_ctl_destroy(req, mp); return; } else if (strcmp(verb, "sync") == 0) { g_journal_ctl_sync(req, mp); return; } gctl_error(req, "Unknown verb."); } static void g_journal_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp) { struct g_journal_softc *sc; g_topology_assert(); sc = gp->softc; if (sc == NULL) return; if (pp != NULL) { /* Nothing here. */ } else if (cp != NULL) { int first = 1; sbuf_printf(sb, "%s", indent); if (cp == sc->sc_dconsumer) { sbuf_printf(sb, "Data"); first = 0; } if (cp == sc->sc_jconsumer) { if (!first) sbuf_printf(sb, ","); sbuf_printf(sb, "Journal"); } sbuf_printf(sb, "\n"); if (cp == sc->sc_jconsumer) { sbuf_printf(sb, "%jd\n", (intmax_t)sc->sc_jstart); sbuf_printf(sb, "%jd\n", (intmax_t)sc->sc_jend); } } else { sbuf_printf(sb, "%s%u\n", indent, (u_int)sc->sc_id); } } static eventhandler_tag g_journal_event_shutdown = NULL; static eventhandler_tag g_journal_event_lowmem = NULL; static void g_journal_shutdown(void *arg, int howto __unused) { struct g_class *mp; struct g_geom *gp, *gp2; if (panicstr != NULL) return; mp = arg; DROP_GIANT(); g_topology_lock(); LIST_FOREACH_SAFE(gp, &mp->geom, geom, gp2) { if (gp->softc == NULL) continue; GJ_DEBUG(0, "Shutting down geom %s.", gp->name); g_journal_destroy(gp->softc); } g_topology_unlock(); PICKUP_GIANT(); } /* * Free cached requests from inactive queue in case of low memory. * We free GJ_FREE_AT_ONCE elements at once. */ #define GJ_FREE_AT_ONCE 4 static void g_journal_lowmem(void *arg, int howto __unused) { struct g_journal_softc *sc; struct g_class *mp; struct g_geom *gp; struct bio *bp; u_int nfree = GJ_FREE_AT_ONCE; g_journal_stats_low_mem++; mp = arg; DROP_GIANT(); g_topology_lock(); LIST_FOREACH(gp, &mp->geom, geom) { sc = gp->softc; if (sc == NULL || (sc->sc_flags & GJF_DEVICE_DESTROY)) continue; mtx_lock(&sc->sc_mtx); for (bp = sc->sc_inactive.jj_queue; nfree > 0 && bp != NULL; nfree--, bp = bp->bio_next) { /* * This is safe to free the bio_data, because: * 1. If bio_data is NULL it will be read from the * inactive journal. * 2. If bp is sent down, it is first removed from the * inactive queue, so it's impossible to free the * data from under in-flight bio. * On the other hand, freeing elements from the active * queue, is not safe. */ if (bp->bio_data != NULL) { GJ_DEBUG(2, "Freeing data from %s.", sc->sc_name); gj_free(bp->bio_data, bp->bio_length); bp->bio_data = NULL; } } mtx_unlock(&sc->sc_mtx); if (nfree == 0) break; } g_topology_unlock(); PICKUP_GIANT(); } static void g_journal_switcher(void *arg); static void g_journal_init(struct g_class *mp) { int error; /* Pick a conservative value if provided value sucks. */ if (g_journal_cache_divisor <= 0 || (vm_kmem_size / g_journal_cache_divisor == 0)) { g_journal_cache_divisor = 5; } if (g_journal_cache_limit > 0) { g_journal_cache_limit = vm_kmem_size / g_journal_cache_divisor; g_journal_cache_low = (g_journal_cache_limit / 100) * g_journal_cache_switch; } g_journal_event_shutdown = EVENTHANDLER_REGISTER(shutdown_post_sync, g_journal_shutdown, mp, EVENTHANDLER_PRI_FIRST); if (g_journal_event_shutdown == NULL) GJ_DEBUG(0, "Warning! Cannot register shutdown event."); g_journal_event_lowmem = EVENTHANDLER_REGISTER(vm_lowmem, g_journal_lowmem, mp, EVENTHANDLER_PRI_FIRST); if (g_journal_event_lowmem == NULL) GJ_DEBUG(0, "Warning! Cannot register lowmem event."); error = kproc_create(g_journal_switcher, mp, NULL, 0, 0, "g_journal switcher"); KASSERT(error == 0, ("Cannot create switcher thread.")); } static void g_journal_fini(struct g_class *mp) { if (g_journal_event_shutdown != NULL) { EVENTHANDLER_DEREGISTER(shutdown_post_sync, g_journal_event_shutdown); } if (g_journal_event_lowmem != NULL) EVENTHANDLER_DEREGISTER(vm_lowmem, g_journal_event_lowmem); g_journal_switcher_state = GJ_SWITCHER_DIE; wakeup(&g_journal_switcher_state); while (g_journal_switcher_state != GJ_SWITCHER_DIED) tsleep(&g_journal_switcher_state, PRIBIO, "jfini:wait", hz / 5); GJ_DEBUG(1, "Switcher died."); } DECLARE_GEOM_CLASS(g_journal_class, g_journal); static const struct g_journal_desc * g_journal_find_desc(const char *fstype) { const struct g_journal_desc *desc; int i; for (desc = g_journal_filesystems[i = 0]; desc != NULL; desc = g_journal_filesystems[++i]) { if (strcmp(desc->jd_fstype, fstype) == 0) break; } return (desc); } static void g_journal_switch_wait(struct g_journal_softc *sc) { struct bintime bt; mtx_assert(&sc->sc_mtx, MA_OWNED); if (g_journal_debug >= 2) { if (sc->sc_flush_in_progress > 0) { GJ_DEBUG(2, "%d requests flushing.", sc->sc_flush_in_progress); } if (sc->sc_copy_in_progress > 0) { GJ_DEBUG(2, "%d requests copying.", sc->sc_copy_in_progress); } if (sc->sc_flush_count > 0) { GJ_DEBUG(2, "%d requests to flush.", sc->sc_flush_count); } if (sc->sc_delayed_count > 0) { GJ_DEBUG(2, "%d requests delayed.", sc->sc_delayed_count); } } g_journal_stats_switches++; if (sc->sc_copy_in_progress > 0) g_journal_stats_wait_for_copy++; GJ_TIMER_START(1, &bt); sc->sc_flags &= ~GJF_DEVICE_BEFORE_SWITCH; sc->sc_flags |= GJF_DEVICE_SWITCH; wakeup(sc); while (sc->sc_flags & GJF_DEVICE_SWITCH) { msleep(&sc->sc_journal_copying, &sc->sc_mtx, PRIBIO, "gj:switch", 0); } GJ_TIMER_STOP(1, &bt, "Switch time of %s", sc->sc_name); } static void g_journal_do_switch(struct g_class *classp) { struct g_journal_softc *sc; const struct g_journal_desc *desc; struct g_geom *gp; struct mount *mp; struct bintime bt; char *mountpoint; int error, vfslocked; DROP_GIANT(); g_topology_lock(); LIST_FOREACH(gp, &classp->geom, geom) { sc = gp->softc; if (sc == NULL) continue; if (sc->sc_flags & GJF_DEVICE_DESTROY) continue; if ((sc->sc_type & GJ_TYPE_COMPLETE) != GJ_TYPE_COMPLETE) continue; mtx_lock(&sc->sc_mtx); sc->sc_flags |= GJF_DEVICE_BEFORE_SWITCH; mtx_unlock(&sc->sc_mtx); } g_topology_unlock(); PICKUP_GIANT(); mtx_lock(&mountlist_mtx); TAILQ_FOREACH(mp, &mountlist, mnt_list) { if (mp->mnt_gjprovider == NULL) continue; if (mp->mnt_flag & MNT_RDONLY) continue; desc = g_journal_find_desc(mp->mnt_stat.f_fstypename); if (desc == NULL) continue; if (vfs_busy(mp, MBF_NOWAIT | MBF_MNTLSTLOCK)) continue; /* mtx_unlock(&mountlist_mtx) was done inside vfs_busy() */ DROP_GIANT(); g_topology_lock(); sc = g_journal_find_device(classp, mp->mnt_gjprovider); g_topology_unlock(); PICKUP_GIANT(); if (sc == NULL) { GJ_DEBUG(0, "Cannot find journal geom for %s.", mp->mnt_gjprovider); goto next; } else if (JEMPTY(sc)) { mtx_lock(&sc->sc_mtx); sc->sc_flags &= ~GJF_DEVICE_BEFORE_SWITCH; mtx_unlock(&sc->sc_mtx); GJ_DEBUG(3, "No need for %s switch.", sc->sc_name); goto next; } mountpoint = mp->mnt_stat.f_mntonname; vfslocked = VFS_LOCK_GIANT(mp); error = vn_start_write(NULL, &mp, V_WAIT); if (error != 0) { VFS_UNLOCK_GIANT(vfslocked); GJ_DEBUG(0, "vn_start_write(%s) failed (error=%d).", mountpoint, error); goto next; } MNT_ILOCK(mp); mp->mnt_noasync++; mp->mnt_kern_flag &= ~MNTK_ASYNC; MNT_IUNLOCK(mp); GJ_TIMER_START(1, &bt); vfs_msync(mp, MNT_NOWAIT); GJ_TIMER_STOP(1, &bt, "Msync time of %s", mountpoint); GJ_TIMER_START(1, &bt); error = VFS_SYNC(mp, MNT_NOWAIT); if (error == 0) GJ_TIMER_STOP(1, &bt, "Sync time of %s", mountpoint); else { GJ_DEBUG(0, "Cannot sync file system %s (error=%d).", mountpoint, error); } MNT_ILOCK(mp); mp->mnt_noasync--; if ((mp->mnt_flag & MNT_ASYNC) != 0 && mp->mnt_noasync == 0) mp->mnt_kern_flag |= MNTK_ASYNC; MNT_IUNLOCK(mp); vn_finished_write(mp); if (error != 0) { VFS_UNLOCK_GIANT(vfslocked); goto next; } /* * Send BIO_FLUSH before freezing the file system, so it can be * faster after the freeze. */ GJ_TIMER_START(1, &bt); g_journal_flush_cache(sc); GJ_TIMER_STOP(1, &bt, "BIO_FLUSH time of %s", sc->sc_name); GJ_TIMER_START(1, &bt); error = vfs_write_suspend(mp); VFS_UNLOCK_GIANT(vfslocked); GJ_TIMER_STOP(1, &bt, "Suspend time of %s", mountpoint); if (error != 0) { GJ_DEBUG(0, "Cannot suspend file system %s (error=%d).", mountpoint, error); goto next; } error = desc->jd_clean(mp); if (error != 0) goto next; mtx_lock(&sc->sc_mtx); g_journal_switch_wait(sc); mtx_unlock(&sc->sc_mtx); vfs_write_resume(mp); next: mtx_lock(&mountlist_mtx); vfs_unbusy(mp); } mtx_unlock(&mountlist_mtx); sc = NULL; for (;;) { DROP_GIANT(); g_topology_lock(); LIST_FOREACH(gp, &g_journal_class.geom, geom) { sc = gp->softc; if (sc == NULL) continue; mtx_lock(&sc->sc_mtx); if ((sc->sc_type & GJ_TYPE_COMPLETE) == GJ_TYPE_COMPLETE && !(sc->sc_flags & GJF_DEVICE_DESTROY) && (sc->sc_flags & GJF_DEVICE_BEFORE_SWITCH)) { break; } mtx_unlock(&sc->sc_mtx); sc = NULL; } g_topology_unlock(); PICKUP_GIANT(); if (sc == NULL) break; mtx_assert(&sc->sc_mtx, MA_OWNED); g_journal_switch_wait(sc); mtx_unlock(&sc->sc_mtx); } } /* * TODO: Switcher thread should be started on first geom creation and killed on * last geom destruction. */ static void g_journal_switcher(void *arg) { struct g_class *mp; struct bintime bt; int error; mp = arg; curthread->td_pflags |= TDP_NORUNNINGBUF; for (;;) { g_journal_switcher_wokenup = 0; error = tsleep(&g_journal_switcher_state, PRIBIO, "jsw:wait", g_journal_switch_time * hz); if (g_journal_switcher_state == GJ_SWITCHER_DIE) { g_journal_switcher_state = GJ_SWITCHER_DIED; GJ_DEBUG(1, "Switcher exiting."); wakeup(&g_journal_switcher_state); kproc_exit(0); } if (error == 0 && g_journal_sync_requested == 0) { GJ_DEBUG(1, "Out of cache, force switch (used=%u " "limit=%u).", g_journal_cache_used, g_journal_cache_limit); } GJ_TIMER_START(1, &bt); g_journal_do_switch(mp); GJ_TIMER_STOP(1, &bt, "Entire switch time"); if (g_journal_sync_requested > 0) { g_journal_sync_requested = 0; wakeup(&g_journal_sync_requested); } } } Index: projects/largeSMP/sys/geom/mountver/g_mountver.c =================================================================== --- projects/largeSMP/sys/geom/mountver/g_mountver.c (revision 221494) +++ projects/largeSMP/sys/geom/mountver/g_mountver.c (revision 221495) @@ -1,645 +1,624 @@ /*- * Copyright (c) 2010 Edward Tomasz Napierala * Copyright (c) 2004-2006 Pawel Jakub Dawidek * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include SYSCTL_DECL(_kern_geom); SYSCTL_NODE(_kern_geom, OID_AUTO, mountver, CTLFLAG_RW, 0, "GEOM_MOUNTVER stuff"); static u_int g_mountver_debug = 0; static u_int g_mountver_check_ident = 1; SYSCTL_UINT(_kern_geom_mountver, OID_AUTO, debug, CTLFLAG_RW, &g_mountver_debug, 0, "Debug level"); SYSCTL_UINT(_kern_geom_mountver, OID_AUTO, check_ident, CTLFLAG_RW, &g_mountver_check_ident, 0, "Check disk ident when reattaching"); static eventhandler_tag g_mountver_pre_sync = NULL; static void g_mountver_queue(struct bio *bp); static void g_mountver_orphan(struct g_consumer *cp); static int g_mountver_destroy(struct g_geom *gp, boolean_t force); static g_taste_t g_mountver_taste; static int g_mountver_destroy_geom(struct gctl_req *req, struct g_class *mp, struct g_geom *gp); static void g_mountver_config(struct gctl_req *req, struct g_class *mp, const char *verb); static void g_mountver_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp); static void g_mountver_init(struct g_class *mp); static void g_mountver_fini(struct g_class *mp); struct g_class g_mountver_class = { .name = G_MOUNTVER_CLASS_NAME, .version = G_VERSION, .ctlreq = g_mountver_config, .taste = g_mountver_taste, .destroy_geom = g_mountver_destroy_geom, .init = g_mountver_init, .fini = g_mountver_fini }; static void g_mountver_done(struct bio *bp) { struct g_geom *gp; struct bio *pbp; if (bp->bio_error != ENXIO) { g_std_done(bp); return; } /* * When the device goes away, it's possible that few requests * will be completed with ENXIO before g_mountver_orphan() * gets called. To work around that, we have to queue requests * that failed with ENXIO, in order to send them later. */ gp = bp->bio_from->geom; pbp = bp->bio_parent; KASSERT(pbp->bio_to == LIST_FIRST(&gp->provider), ("parent request was for someone else")); g_destroy_bio(bp); pbp->bio_inbed++; g_mountver_queue(pbp); } static void g_mountver_send(struct bio *bp) { struct g_geom *gp; struct bio *cbp; gp = bp->bio_to->geom; cbp = g_clone_bio(bp); if (cbp == NULL) { g_io_deliver(bp, ENOMEM); return; } cbp->bio_done = g_mountver_done; g_io_request(cbp, LIST_FIRST(&gp->consumer)); } static void g_mountver_queue(struct bio *bp) { struct g_mountver_softc *sc; struct g_geom *gp; gp = bp->bio_to->geom; sc = gp->softc; mtx_lock(&sc->sc_mtx); TAILQ_INSERT_TAIL(&sc->sc_queue, bp, bio_queue); mtx_unlock(&sc->sc_mtx); } static void g_mountver_send_queued(struct g_geom *gp) { struct g_mountver_softc *sc; struct bio *bp; sc = gp->softc; mtx_lock(&sc->sc_mtx); while ((bp = TAILQ_FIRST(&sc->sc_queue)) != NULL) { TAILQ_REMOVE(&sc->sc_queue, bp, bio_queue); G_MOUNTVER_LOGREQ(bp, "Sending queued request."); g_mountver_send(bp); } mtx_unlock(&sc->sc_mtx); } static void g_mountver_discard_queued(struct g_geom *gp) { struct g_mountver_softc *sc; struct bio *bp; sc = gp->softc; mtx_lock(&sc->sc_mtx); while ((bp = TAILQ_FIRST(&sc->sc_queue)) != NULL) { TAILQ_REMOVE(&sc->sc_queue, bp, bio_queue); G_MOUNTVER_LOGREQ(bp, "Discarding queued request."); g_io_deliver(bp, ENXIO); } mtx_unlock(&sc->sc_mtx); } static void g_mountver_start(struct bio *bp) { struct g_mountver_softc *sc; struct g_geom *gp; gp = bp->bio_to->geom; sc = gp->softc; G_MOUNTVER_LOGREQ(bp, "Request received."); /* * It is possible that some bios were returned with ENXIO, even though * orphaning didn't happen yet. In that case, queue all subsequent * requests in order to maintain ordering. */ if (sc->sc_orphaned || !TAILQ_EMPTY(&sc->sc_queue)) { G_MOUNTVER_LOGREQ(bp, "Queueing request."); g_mountver_queue(bp); if (!sc->sc_orphaned) g_mountver_send_queued(gp); } else { G_MOUNTVER_LOGREQ(bp, "Sending request."); g_mountver_send(bp); } } static int g_mountver_access(struct g_provider *pp, int dr, int dw, int de) { struct g_mountver_softc *sc; struct g_geom *gp; struct g_consumer *cp; g_topology_assert(); gp = pp->geom; cp = LIST_FIRST(&gp->consumer); sc = gp->softc; if (sc == NULL && dr <= 0 && dw <= 0 && de <= 0) return (0); KASSERT(sc != NULL, ("Trying to access withered provider \"%s\".", pp->name)); sc->sc_access_r += dr; sc->sc_access_w += dw; sc->sc_access_e += de; if (sc->sc_orphaned) return (0); return (g_access(cp, dr, dw, de)); } static int g_mountver_create(struct gctl_req *req, struct g_class *mp, struct g_provider *pp) { struct g_mountver_softc *sc; struct g_geom *gp; struct g_provider *newpp; struct g_consumer *cp; char name[64]; int error; int identsize = DISK_IDENT_SIZE; g_topology_assert(); gp = NULL; newpp = NULL; cp = NULL; snprintf(name, sizeof(name), "%s%s", pp->name, G_MOUNTVER_SUFFIX); LIST_FOREACH(gp, &mp->geom, geom) { if (strcmp(gp->name, name) == 0) { gctl_error(req, "Provider %s already exists.", name); return (EEXIST); } } gp = g_new_geomf(mp, name); - if (gp == NULL) { - gctl_error(req, "Cannot create geom %s.", name); - return (ENOMEM); - } sc = g_malloc(sizeof(*sc), M_WAITOK | M_ZERO); mtx_init(&sc->sc_mtx, "gmountver", NULL, MTX_DEF); TAILQ_INIT(&sc->sc_queue); sc->sc_provider_name = strdup(pp->name, M_GEOM); gp->softc = sc; gp->start = g_mountver_start; gp->orphan = g_mountver_orphan; gp->access = g_mountver_access; gp->dumpconf = g_mountver_dumpconf; newpp = g_new_providerf(gp, gp->name); - if (newpp == NULL) { - gctl_error(req, "Cannot create provider %s.", name); - error = ENOMEM; - goto fail; - } newpp->mediasize = pp->mediasize; newpp->sectorsize = pp->sectorsize; cp = g_new_consumer(gp); - if (cp == NULL) { - gctl_error(req, "Cannot create consumer for %s.", gp->name); - error = ENOMEM; - goto fail; - } error = g_attach(cp, pp); if (error != 0) { gctl_error(req, "Cannot attach to provider %s.", pp->name); goto fail; } error = g_access(cp, 1, 0, 0); if (error != 0) { gctl_error(req, "Cannot access provider %s.", pp->name); goto fail; } error = g_io_getattr("GEOM::ident", cp, &identsize, sc->sc_ident); g_access(cp, -1, 0, 0); if (error != 0) { if (g_mountver_check_ident) { gctl_error(req, "Cannot get disk ident from %s; error = %d.", pp->name, error); goto fail; } G_MOUNTVER_DEBUG(0, "Cannot get disk ident from %s; error = %d.", pp->name, error); sc->sc_ident[0] = '\0'; } g_error_provider(newpp, 0); G_MOUNTVER_DEBUG(0, "Device %s created.", gp->name); return (0); fail: - if (sc->sc_provider_name != NULL) - g_free(sc->sc_provider_name); - if (cp != NULL) { - if (cp->provider != NULL) - g_detach(cp); - g_destroy_consumer(cp); - } - if (newpp != NULL) - g_destroy_provider(newpp); - if (gp != NULL) { - if (gp->softc != NULL) - g_free(gp->softc); - g_destroy_geom(gp); - } + g_free(sc->sc_provider_name); + if (cp->provider != NULL) + g_detach(cp); + g_destroy_consumer(cp); + g_destroy_provider(newpp); + g_free(gp->softc); + g_destroy_geom(gp); return (error); } static int g_mountver_destroy(struct g_geom *gp, boolean_t force) { struct g_mountver_softc *sc; struct g_provider *pp; g_topology_assert(); if (gp->softc == NULL) return (ENXIO); sc = gp->softc; pp = LIST_FIRST(&gp->provider); if (pp != NULL && (pp->acr != 0 || pp->acw != 0 || pp->ace != 0)) { if (force) { G_MOUNTVER_DEBUG(0, "Device %s is still open, so it " "can't be definitely removed.", pp->name); } else { G_MOUNTVER_DEBUG(1, "Device %s is still open (r%dw%de%d).", pp->name, pp->acr, pp->acw, pp->ace); return (EBUSY); } } else { G_MOUNTVER_DEBUG(0, "Device %s removed.", gp->name); } if (pp != NULL) g_orphan_provider(pp, ENXIO); g_mountver_discard_queued(gp); g_free(sc->sc_provider_name); g_free(gp->softc); gp->softc = NULL; g_wither_geom(gp, ENXIO); return (0); } static int g_mountver_destroy_geom(struct gctl_req *req, struct g_class *mp, struct g_geom *gp) { return (g_mountver_destroy(gp, 0)); } static void g_mountver_ctl_create(struct gctl_req *req, struct g_class *mp) { struct g_provider *pp; const char *name; char param[16]; int i, *nargs; g_topology_assert(); nargs = gctl_get_paraml(req, "nargs", sizeof(*nargs)); if (nargs == NULL) { gctl_error(req, "No '%s' argument", "nargs"); return; } if (*nargs <= 0) { gctl_error(req, "Missing device(s)."); return; } for (i = 0; i < *nargs; i++) { snprintf(param, sizeof(param), "arg%d", i); name = gctl_get_asciiparam(req, param); if (name == NULL) { gctl_error(req, "No 'arg%d' argument", i); return; } if (strncmp(name, "/dev/", strlen("/dev/")) == 0) name += strlen("/dev/"); pp = g_provider_by_name(name); if (pp == NULL) { G_MOUNTVER_DEBUG(1, "Provider %s is invalid.", name); gctl_error(req, "Provider %s is invalid.", name); return; } if (g_mountver_create(req, mp, pp) != 0) return; } } static struct g_geom * g_mountver_find_geom(struct g_class *mp, const char *name) { struct g_geom *gp; LIST_FOREACH(gp, &mp->geom, geom) { if (strcmp(gp->name, name) == 0) return (gp); } return (NULL); } static void g_mountver_ctl_destroy(struct gctl_req *req, struct g_class *mp) { int *nargs, *force, error, i; struct g_geom *gp; const char *name; char param[16]; g_topology_assert(); nargs = gctl_get_paraml(req, "nargs", sizeof(*nargs)); if (nargs == NULL) { gctl_error(req, "No '%s' argument", "nargs"); return; } if (*nargs <= 0) { gctl_error(req, "Missing device(s)."); return; } force = gctl_get_paraml(req, "force", sizeof(*force)); if (force == NULL) { gctl_error(req, "No 'force' argument"); return; } for (i = 0; i < *nargs; i++) { snprintf(param, sizeof(param), "arg%d", i); name = gctl_get_asciiparam(req, param); if (name == NULL) { gctl_error(req, "No 'arg%d' argument", i); return; } if (strncmp(name, "/dev/", strlen("/dev/")) == 0) name += strlen("/dev/"); gp = g_mountver_find_geom(mp, name); if (gp == NULL) { G_MOUNTVER_DEBUG(1, "Device %s is invalid.", name); gctl_error(req, "Device %s is invalid.", name); return; } error = g_mountver_destroy(gp, *force); if (error != 0) { gctl_error(req, "Cannot destroy device %s (error=%d).", gp->name, error); return; } } } static void g_mountver_orphan(struct g_consumer *cp) { struct g_mountver_softc *sc; g_topology_assert(); sc = cp->geom->softc; sc->sc_orphaned = 1; if (cp->acr > 0 || cp->acw > 0 || cp->ace > 0) g_access(cp, -cp->acr, -cp->acw, -cp->ace); g_detach(cp); G_MOUNTVER_DEBUG(0, "%s is offline. Mount verification in progress.", sc->sc_provider_name); } static int g_mountver_ident_matches(struct g_geom *gp) { struct g_consumer *cp; struct g_mountver_softc *sc; char ident[DISK_IDENT_SIZE]; int error, identsize = DISK_IDENT_SIZE; sc = gp->softc; cp = LIST_FIRST(&gp->consumer); if (g_mountver_check_ident == 0) return (0); error = g_access(cp, 1, 0, 0); if (error != 0) { G_MOUNTVER_DEBUG(0, "Cannot access %s; " "not attaching; error = %d.", gp->name, error); return (1); } error = g_io_getattr("GEOM::ident", cp, &identsize, ident); g_access(cp, -1, 0, 0); if (error != 0) { G_MOUNTVER_DEBUG(0, "Cannot get disk ident for %s; " "not attaching; error = %d.", gp->name, error); return (1); } if (strcmp(ident, sc->sc_ident) != 0) { G_MOUNTVER_DEBUG(1, "Disk ident for %s (\"%s\") is different " "from expected \"%s\", not attaching.", gp->name, ident, sc->sc_ident); return (1); } return (0); } static struct g_geom * g_mountver_taste(struct g_class *mp, struct g_provider *pp, int flags __unused) { struct g_mountver_softc *sc; struct g_consumer *cp; struct g_geom *gp; int error; g_topology_assert(); g_trace(G_T_TOPOLOGY, "%s(%s, %s)", __func__, mp->name, pp->name); G_MOUNTVER_DEBUG(2, "Tasting %s.", pp->name); /* * Let's check if device already exists. */ LIST_FOREACH(gp, &mp->geom, geom) { sc = gp->softc; if (sc == NULL) continue; /* Already attached? */ if (pp == LIST_FIRST(&gp->provider)) return (NULL); if (sc->sc_orphaned && strcmp(pp->name, sc->sc_provider_name) == 0) break; } if (gp == NULL) return (NULL); cp = LIST_FIRST(&gp->consumer); g_attach(cp, pp); error = g_mountver_ident_matches(gp); if (error != 0) { g_detach(cp); return (NULL); } if (sc->sc_access_r > 0 || sc->sc_access_w > 0 || sc->sc_access_e > 0) { error = g_access(cp, sc->sc_access_r, sc->sc_access_w, sc->sc_access_e); if (error != 0) { G_MOUNTVER_DEBUG(0, "Cannot access %s; error = %d.", pp->name, error); g_detach(cp); return (NULL); } } g_mountver_send_queued(gp); sc->sc_orphaned = 0; G_MOUNTVER_DEBUG(0, "%s has completed mount verification.", sc->sc_provider_name); return (gp); } static void g_mountver_config(struct gctl_req *req, struct g_class *mp, const char *verb) { uint32_t *version; g_topology_assert(); version = gctl_get_paraml(req, "version", sizeof(*version)); if (version == NULL) { gctl_error(req, "No '%s' argument.", "version"); return; } if (*version != G_MOUNTVER_VERSION) { gctl_error(req, "Userland and kernel parts are out of sync."); return; } if (strcmp(verb, "create") == 0) { g_mountver_ctl_create(req, mp); return; } else if (strcmp(verb, "destroy") == 0) { g_mountver_ctl_destroy(req, mp); return; } gctl_error(req, "Unknown verb."); } static void g_mountver_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp) { struct g_mountver_softc *sc; if (pp != NULL || cp != NULL) return; sc = gp->softc; sbuf_printf(sb, "%s%s\n", indent, sc->sc_orphaned ? "OFFLINE" : "ONLINE"); sbuf_printf(sb, "%s%s\n", indent, sc->sc_provider_name); sbuf_printf(sb, "%s%s\n", indent, sc->sc_ident); } static void g_mountver_shutdown_pre_sync(void *arg, int howto) { struct g_class *mp; struct g_geom *gp, *gp2; mp = arg; DROP_GIANT(); g_topology_lock(); LIST_FOREACH_SAFE(gp, &mp->geom, geom, gp2) g_mountver_destroy(gp, 1); g_topology_unlock(); PICKUP_GIANT(); } static void g_mountver_init(struct g_class *mp) { g_mountver_pre_sync = EVENTHANDLER_REGISTER(shutdown_pre_sync, g_mountver_shutdown_pre_sync, mp, SHUTDOWN_PRI_FIRST); if (g_mountver_pre_sync == NULL) G_MOUNTVER_DEBUG(0, "Warning! Cannot register shutdown event."); } static void g_mountver_fini(struct g_class *mp) { if (g_mountver_pre_sync != NULL) EVENTHANDLER_DEREGISTER(shutdown_pre_sync, g_mountver_pre_sync); } DECLARE_GEOM_CLASS(g_mountver_class, g_mountver); Index: projects/largeSMP/sys/geom/multipath/g_multipath.c =================================================================== --- projects/largeSMP/sys/geom/multipath/g_multipath.c (revision 221494) +++ projects/largeSMP/sys/geom/multipath/g_multipath.c (revision 221495) @@ -1,793 +1,779 @@ /*- * Copyright (c) 2006-2007 Matthew Jacob * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ /* * Based upon work by Pawel Jakub Dawidek for all of the * fine geom examples, and by Poul Henning Kamp for GEOM * itself, all of which is most gratefully acknowledged. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include FEATURE(geom_multipath, "GEOM multipath support"); SYSCTL_DECL(_kern_geom); SYSCTL_NODE(_kern_geom, OID_AUTO, multipath, CTLFLAG_RW, 0, "GEOM_MULTIPATH tunables"); static u_int g_multipath_debug = 0; SYSCTL_UINT(_kern_geom_multipath, OID_AUTO, debug, CTLFLAG_RW, &g_multipath_debug, 0, "Debug level"); static enum { GKT_NIL, GKT_RUN, GKT_DIE } g_multipath_kt_state; static struct bio_queue_head gmtbq; static struct mtx gmtbq_mtx; static void g_multipath_orphan(struct g_consumer *); static void g_multipath_start(struct bio *); static void g_multipath_done(struct bio *); static void g_multipath_done_error(struct bio *); static void g_multipath_kt(void *); static int g_multipath_destroy(struct g_geom *); static int g_multipath_destroy_geom(struct gctl_req *, struct g_class *, struct g_geom *); static struct g_geom *g_multipath_find_geom(struct g_class *, const char *); static int g_multipath_rotate(struct g_geom *); static g_taste_t g_multipath_taste; static g_ctl_req_t g_multipath_config; static g_init_t g_multipath_init; static g_fini_t g_multipath_fini; struct g_class g_multipath_class = { .name = G_MULTIPATH_CLASS_NAME, .version = G_VERSION, .ctlreq = g_multipath_config, .taste = g_multipath_taste, .destroy_geom = g_multipath_destroy_geom, .init = g_multipath_init, .fini = g_multipath_fini }; #define MP_BAD 0x1 #define MP_POSTED 0x2 static void g_mpd(void *arg, int flags __unused) { struct g_consumer *cp; g_topology_assert(); cp = arg; if (cp->acr > 0 || cp->acw > 0 || cp->ace > 0) g_access(cp, -cp->acr, -cp->acw, -cp->ace); if (cp->provider) { printf("GEOM_MULTIPATH: %s removed from %s\n", cp->provider->name, cp->geom->name); g_detach(cp); } g_destroy_consumer(cp); } static void g_multipath_orphan(struct g_consumer *cp) { if ((cp->index & MP_POSTED) == 0) { cp->index |= MP_POSTED; printf("GEOM_MULTIPATH: %s orphaned in %s\n", cp->provider->name, cp->geom->name); g_mpd(cp, 0); } } static void g_multipath_start(struct bio *bp) { struct g_multipath_softc *sc; struct g_geom *gp; struct g_consumer *cp; struct bio *cbp; gp = bp->bio_to->geom; sc = gp->softc; KASSERT(sc != NULL, ("NULL sc")); cp = sc->cp_active; if (cp == NULL) { g_io_deliver(bp, ENXIO); return; } cbp = g_clone_bio(bp); if (cbp == NULL) { g_io_deliver(bp, ENOMEM); return; } cbp->bio_done = g_multipath_done; g_io_request(cbp, cp); } static void g_multipath_done(struct bio *bp) { if (bp->bio_error == ENXIO || bp->bio_error == EIO) { mtx_lock(&gmtbq_mtx); bioq_insert_tail(&gmtbq, bp); wakeup(&g_multipath_kt_state); mtx_unlock(&gmtbq_mtx); } else { g_std_done(bp); } } static void g_multipath_done_error(struct bio *bp) { struct bio *pbp; struct g_geom *gp; struct g_multipath_softc *sc; struct g_consumer *cp; struct g_provider *pp; /* * If we had a failure, we have to check first to see * whether the consumer it failed on was the currently * active consumer (i.e., this is the first in perhaps * a number of failures). If so, we then switch consumers * to the next available consumer. */ g_topology_lock(); pbp = bp->bio_parent; gp = pbp->bio_to->geom; sc = gp->softc; cp = bp->bio_from; pp = cp->provider; cp->index |= MP_BAD; if (cp->nend == cp->nstart && pp->nend == pp->nstart) { cp->index |= MP_POSTED; g_post_event(g_mpd, cp, M_NOWAIT, NULL); } if (cp == sc->cp_active) { struct g_consumer *lcp; printf("GEOM_MULTIPATH: %s failed in %s\n", pp->name, sc->sc_name); sc->cp_active = NULL; LIST_FOREACH(lcp, &gp->consumer, consumer) { if ((lcp->index & MP_BAD) == 0) { sc->cp_active = lcp; break; } } if (sc->cp_active == NULL || sc->cp_active->provider == NULL) { printf("GEOM_MULTIPATH: out of providers for %s\n", sc->sc_name); g_topology_unlock(); return; } else { printf("GEOM_MULTIPATH: %s now active path in %s\n", sc->cp_active->provider->name, sc->sc_name); } } g_topology_unlock(); /* * If we can fruitfully restart the I/O, do so. */ if (sc->cp_active) { g_destroy_bio(bp); pbp->bio_children--; g_multipath_start(pbp); } else { g_std_done(bp); } } static void g_multipath_kt(void *arg) { g_multipath_kt_state = GKT_RUN; mtx_lock(&gmtbq_mtx); while (g_multipath_kt_state == GKT_RUN) { for (;;) { struct bio *bp; bp = bioq_takefirst(&gmtbq); if (bp == NULL) break; mtx_unlock(&gmtbq_mtx); g_multipath_done_error(bp); mtx_lock(&gmtbq_mtx); } msleep(&g_multipath_kt_state, &gmtbq_mtx, PRIBIO, "gkt:wait", hz / 10); } mtx_unlock(&gmtbq_mtx); wakeup(&g_multipath_kt_state); kproc_exit(0); } static int g_multipath_access(struct g_provider *pp, int dr, int dw, int de) { struct g_geom *gp; struct g_consumer *cp, *badcp = NULL; int error; gp = pp->geom; LIST_FOREACH(cp, &gp->consumer, consumer) { error = g_access(cp, dr, dw, de); if (error) { badcp = cp; goto fail; } } return (0); fail: LIST_FOREACH(cp, &gp->consumer, consumer) { if (cp == badcp) break; (void) g_access(cp, -dr, -dw, -de); } return (error); } static struct g_geom * g_multipath_create(struct g_class *mp, struct g_multipath_metadata *md) { struct g_multipath_softc *sc; struct g_geom *gp; struct g_provider *pp; g_topology_assert(); LIST_FOREACH(gp, &mp->geom, geom) { if (strcmp(gp->name, md->md_name) == 0) { printf("GEOM_MULTIPATH: name %s already exists\n", md->md_name); return (NULL); } } gp = g_new_geomf(mp, md->md_name); - if (gp == NULL) - goto fail; - sc = g_malloc(sizeof(*sc), M_WAITOK | M_ZERO); gp->softc = sc; gp->start = g_multipath_start; gp->orphan = g_multipath_orphan; gp->access = g_multipath_access; memcpy(sc->sc_uuid, md->md_uuid, sizeof (sc->sc_uuid)); memcpy(sc->sc_name, md->md_name, sizeof (sc->sc_name)); pp = g_new_providerf(gp, "multipath/%s", md->md_name); - if (pp == NULL) - goto fail; /* limit the provider to not have it stomp on metadata */ pp->mediasize = md->md_size - md->md_sectorsize; pp->sectorsize = md->md_sectorsize; sc->pp = pp; g_error_provider(pp, 0); return (gp); -fail: - if (gp != NULL) { - if (gp->softc != NULL) - g_free(gp->softc); - g_destroy_geom(gp); - } - return (NULL); } static int g_multipath_add_disk(struct g_geom *gp, struct g_provider *pp) { struct g_multipath_softc *sc; struct g_consumer *cp, *nxtcp; int error; g_topology_assert(); sc = gp->softc; KASSERT(sc, ("no softc")); /* * Make sure that the passed provider isn't already attached */ LIST_FOREACH(cp, &gp->consumer, consumer) { if (cp->provider == pp) break; } if (cp) { printf("GEOM_MULTIPATH: provider %s already attached to %s\n", pp->name, gp->name); return (EEXIST); } nxtcp = LIST_FIRST(&gp->consumer); cp = g_new_consumer(gp); - if (cp == NULL) - return (ENOMEM); error = g_attach(cp, pp); if (error != 0) { printf("GEOM_MULTIPATH: cannot attach %s to %s", pp->name, sc->sc_name); g_destroy_consumer(cp); return (error); } cp->private = sc; cp->index = 0; /* * Set access permissions on new consumer to match other consumers */ if (nxtcp && (nxtcp->acr + nxtcp->acw + nxtcp->ace)) { error = g_access(cp, nxtcp->acr, nxtcp->acw, nxtcp->ace); if (error) { printf("GEOM_MULTIPATH: cannot set access in " "attaching %s to %s/%s (%d)\n", pp->name, sc->sc_name, sc->sc_uuid, error); g_detach(cp); g_destroy_consumer(cp); return (error); } } printf("GEOM_MULTIPATH: adding %s to %s/%s\n", pp->name, sc->sc_name, sc->sc_uuid); if (sc->cp_active == NULL) { sc->cp_active = cp; printf("GEOM_MULTIPATH: %s now active path in %s\n", pp->name, sc->sc_name); } return (0); } static int g_multipath_destroy(struct g_geom *gp) { struct g_provider *pp; g_topology_assert(); if (gp->softc == NULL) return (ENXIO); pp = LIST_FIRST(&gp->provider); if (pp != NULL && (pp->acr != 0 || pp->acw != 0 || pp->ace != 0)) return (EBUSY); printf("GEOM_MULTIPATH: destroying %s\n", gp->name); g_free(gp->softc); gp->softc = NULL; g_wither_geom(gp, ENXIO); return (0); } static int g_multipath_destroy_geom(struct gctl_req *req, struct g_class *mp, struct g_geom *gp) { return (g_multipath_destroy(gp)); } static int g_multipath_rotate(struct g_geom *gp) { struct g_consumer *lcp; struct g_multipath_softc *sc = gp->softc; g_topology_assert(); if (sc == NULL) return (ENXIO); LIST_FOREACH(lcp, &gp->consumer, consumer) { if ((lcp->index & MP_BAD) == 0) { if (sc->cp_active != lcp) { break; } } } if (lcp) { sc->cp_active = lcp; printf("GEOM_MULTIPATH: %s now active path in %s\n", lcp->provider->name, sc->sc_name); } return (0); } static void g_multipath_init(struct g_class *mp) { bioq_init(&gmtbq); mtx_init(&gmtbq_mtx, "gmtbq", NULL, MTX_DEF); if (kproc_create(g_multipath_kt, mp, NULL, 0, 0, "g_mp_kt") == 0) g_multipath_kt_state = GKT_RUN; } static void g_multipath_fini(struct g_class *mp) { if (g_multipath_kt_state == GKT_RUN) { mtx_lock(&gmtbq_mtx); g_multipath_kt_state = GKT_DIE; wakeup(&g_multipath_kt_state); msleep(&g_multipath_kt_state, &gmtbq_mtx, PRIBIO, "gmp:fini", 0); mtx_unlock(&gmtbq_mtx); } } static int g_multipath_read_metadata(struct g_consumer *cp, struct g_multipath_metadata *md) { struct g_provider *pp; u_char *buf; int error; g_topology_assert(); error = g_access(cp, 1, 0, 0); if (error != 0) return (error); pp = cp->provider; g_topology_unlock(); buf = g_read_data(cp, pp->mediasize - pp->sectorsize, pp->sectorsize, &error); g_topology_lock(); g_access(cp, -1, 0, 0); if (buf == NULL) return (error); multipath_metadata_decode(buf, md); g_free(buf); return (0); } static struct g_geom * g_multipath_taste(struct g_class *mp, struct g_provider *pp, int flags __unused) { struct g_multipath_metadata md; struct g_multipath_softc *sc; struct g_consumer *cp; struct g_geom *gp, *gp1; int error, isnew; g_topology_assert(); gp = g_new_geomf(mp, "multipath:taste"); gp->start = g_multipath_start; gp->access = g_multipath_access; gp->orphan = g_multipath_orphan; cp = g_new_consumer(gp); g_attach(cp, pp); error = g_multipath_read_metadata(cp, &md); g_detach(cp); g_destroy_consumer(cp); g_destroy_geom(gp); if (error != 0) return (NULL); gp = NULL; if (strcmp(md.md_magic, G_MULTIPATH_MAGIC) != 0) { if (g_multipath_debug) printf("%s is not MULTIPATH\n", pp->name); return (NULL); } if (md.md_version != G_MULTIPATH_VERSION) { printf("%s has version %d multipath id- this module is version " " %d: rejecting\n", pp->name, md.md_version, G_MULTIPATH_VERSION); return (NULL); } if (g_multipath_debug) printf("MULTIPATH: %s/%s\n", md.md_name, md.md_uuid); /* * Let's check if such a device already is present. We check against * uuid alone first because that's the true distinguishor. If that * passes, then we check for name conflicts. If there are conflicts, * modify the name. * * The whole purpose of this is to solve the problem that people don't * pick good unique names, but good unique names (like uuids) are a * pain to use. So, we allow people to build GEOMs with friendly names * and uuids, and modify the names in case there's a collision. */ sc = NULL; LIST_FOREACH(gp, &mp->geom, geom) { sc = gp->softc; if (sc == NULL) continue; if (strncmp(md.md_uuid, sc->sc_uuid, sizeof(md.md_uuid)) == 0) break; } LIST_FOREACH(gp1, &mp->geom, geom) { if (gp1 == gp) continue; sc = gp1->softc; if (sc == NULL) continue; if (strncmp(md.md_name, sc->sc_name, sizeof(md.md_name)) == 0) break; } /* * If gp is NULL, we had no extant MULTIPATH geom with this uuid. * * If gp1 is *not* NULL, that means we have a MULTIPATH geom extant * with the same name (but a different UUID). * * If gp is NULL, then modify the name with a random number and * complain, but allow the creation of the geom to continue. * * If gp is *not* NULL, just use the geom's name as we're attaching * this disk to the (previously generated) name. */ if (gp1) { sc = gp1->softc; if (gp == NULL) { char buf[16]; u_long rand = random(); snprintf(buf, sizeof (buf), "%s-%lu", md.md_name, rand); printf("GEOM_MULTIPATH: geom %s/%s exists already\n", sc->sc_name, sc->sc_uuid); printf("GEOM_MULTIPATH: %s will be (temporarily) %s\n", md.md_uuid, buf); strlcpy(md.md_name, buf, sizeof(md.md_name)); } else { strlcpy(md.md_name, sc->sc_name, sizeof(md.md_name)); } } if (gp == NULL) { gp = g_multipath_create(mp, &md); if (gp == NULL) { printf("GEOM_MULTIPATH: cannot create geom %s/%s\n", md.md_name, md.md_uuid); return (NULL); } isnew = 1; } else { isnew = 0; } sc = gp->softc; KASSERT(sc != NULL, ("sc is NULL")); error = g_multipath_add_disk(gp, pp); if (error != 0) { if (isnew) g_multipath_destroy(gp); return (NULL); } return (gp); } static void g_multipath_ctl_add(struct gctl_req *req, struct g_class *mp) { struct g_geom *gp; struct g_consumer *cp; struct g_provider *pp, *pp0; const char *name, *mpname; static const char devpf[6] = "/dev/"; g_topology_assert(); mpname = gctl_get_asciiparam(req, "arg0"); if (mpname == NULL) { gctl_error(req, "No 'arg0' argument"); return; } gp = g_multipath_find_geom(mp, mpname); if (gp == NULL) { gctl_error(req, "Device %s is invalid", mpname); return; } name = gctl_get_asciiparam(req, "arg1"); if (name == NULL) { gctl_error(req, "No 'arg1' argument"); return; } if (strncmp(name, devpf, 5) == 0) name += 5; pp = g_provider_by_name(name); if (pp == NULL) { gctl_error(req, "Provider %s is invalid", name); return; } /* * Check to make sure parameters match, if we already have one. */ cp = LIST_FIRST(&gp->consumer); if (cp) { pp0 = cp->provider; } else { pp0 = NULL; } if (pp0) { if (pp0 == pp) { gctl_error(req, "providers %s and %s are the same", pp0->name, pp->name); return; } if (pp0->mediasize != pp->mediasize) { gctl_error(req, "Provider %s is %jd; Provider %s is %jd", pp0->name, (intmax_t) pp0->mediasize, pp->name, (intmax_t) pp->mediasize); return; } if (pp0->sectorsize != pp->sectorsize) { gctl_error(req, "Provider %s has sectorsize %u; Provider %s " "has sectorsize %u", pp0->name, pp0->sectorsize, pp->name, pp->sectorsize); return; } } /* * Now add.... */ (void) g_multipath_add_disk(gp, pp); } static struct g_geom * g_multipath_find_geom(struct g_class *mp, const char *name) { struct g_geom *gp; LIST_FOREACH(gp, &mp->geom, geom) { if (strcmp(gp->name, name) == 0) { return (gp); } } return (NULL); } static void g_multipath_ctl_destroy(struct gctl_req *req, struct g_class *mp) { struct g_geom *gp; const char *name; int error; g_topology_assert(); name = gctl_get_asciiparam(req, "arg0"); if (name == NULL) { gctl_error(req, "No 'arg0' argument"); return; } gp = g_multipath_find_geom(mp, name); if (gp == NULL) { gctl_error(req, "Device %s is invalid", name); return; } error = g_multipath_destroy(gp); if (error != 0) { gctl_error(req, "failed to destroy %s (err=%d)", name, error); } } static void g_multipath_ctl_rotate(struct gctl_req *req, struct g_class *mp) { struct g_geom *gp; const char *name; int error; g_topology_assert(); name = gctl_get_asciiparam(req, "arg0"); if (name == NULL) { gctl_error(req, "No 'arg0' argument"); return; } gp = g_multipath_find_geom(mp, name); if (gp == NULL) { gctl_error(req, "Device %s is invalid", name); return; } error = g_multipath_rotate(gp); if (error != 0) { gctl_error(req, "failed to rotate %s (err=%d)", name, error); } } static void g_multipath_ctl_getactive(struct gctl_req *req, struct g_class *mp) { struct sbuf *sb; struct g_geom *gp; struct g_multipath_softc *sc; const char *name; sb = sbuf_new_auto(); g_topology_assert(); name = gctl_get_asciiparam(req, "arg0"); if (name == NULL) { gctl_error(req, "No 'arg0' argument"); return; } gp = g_multipath_find_geom(mp, name); if (gp == NULL) { gctl_error(req, "Device %s is invalid", name); return; } sc = gp->softc; if (sc->cp_active && sc->cp_active->provider) { sbuf_printf(sb, "%s\n", sc->cp_active->provider->name); } else { sbuf_printf(sb, "none\n"); } sbuf_finish(sb); gctl_set_param_err(req, "output", sbuf_data(sb), sbuf_len(sb) + 1); sbuf_delete(sb); } static void g_multipath_config(struct gctl_req *req, struct g_class *mp, const char *verb) { uint32_t *version; g_topology_assert(); version = gctl_get_paraml(req, "version", sizeof(*version)); if (version == NULL) { gctl_error(req, "No 'version' argument"); } else if (*version != G_MULTIPATH_VERSION) { gctl_error(req, "Userland and kernel parts are out of sync"); } else if (strcmp(verb, "add") == 0) { g_multipath_ctl_add(req, mp); } else if (strcmp(verb, "destroy") == 0) { g_multipath_ctl_destroy(req, mp); } else if (strcmp(verb, "rotate") == 0) { g_multipath_ctl_rotate(req, mp); } else if (strcmp(verb, "getactive") == 0) { g_multipath_ctl_getactive(req, mp); } else { gctl_error(req, "Unknown verb %s", verb); } } DECLARE_GEOM_CLASS(g_multipath_class, g_multipath); Index: projects/largeSMP/sys/geom/sched/g_sched.c =================================================================== --- projects/largeSMP/sys/geom/sched/g_sched.c (revision 221494) +++ projects/largeSMP/sys/geom/sched/g_sched.c (revision 221495) @@ -1,1914 +1,1888 @@ /*- * Copyright (c) 2009-2010 Fabio Checconi * Copyright (c) 2009-2010 Luigi Rizzo, Universita` di Pisa * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ /* * $Id$ * $FreeBSD$ * * Main control module for geom-based disk schedulers ('sched'). * * USER VIEW * A 'sched' node is typically inserted transparently between * an existing provider pp and its original geom gp * * [pp --> gp ..] * * using the command "geom sched insert " and * resulting in the following topology * * [pp --> sched_gp --> cp] [new_pp --> gp ... ] * * Deletion "geom sched destroy .sched." restores the * original chain. The normal "geom sched create " * is also supported. * * INTERNALS * Internally, the 'sched' uses the following data structures * * geom{} g_sched_softc{} g_gsched{} * +----------+ +---------------+ +-------------+ * | softc *-|--->| sc_gsched *-|-->| gs_init | * | ... | | | | gs_fini | * | | | [ hash table] | | gs_start | * +----------+ | | | ... | * | | +-------------+ * | | * | | g_*_softc{} * | | +-------------+ * | sc_data *-|-->| | * +---------------+ | algorithm- | * | specific | * +-------------+ * * A g_sched_softc{} is created with a "geom sched insert" call. * In turn this instantiates a specific scheduling algorithm, * which sets sc_gsched to point to the algorithm callbacks, * and calls gs_init() to create the g_*_softc{} . * The other callbacks (gs_start, gs_next, ...) are invoked * as needed * * g_sched_softc{} is defined in g_sched.h and mostly used here; * g_gsched{}, and the gs_callbacks, are documented in gs_scheduler.h; * g_*_softc{} is defined/implemented by each algorithm (gs_*.c) * * DATA MOVING * When a bio is received on the provider, it goes to the * g_sched_start() which calls gs_start() to initially queue it; * then we call g_sched_dispatch() that loops around gs_next() * to select zero or more bio's to be sent downstream. * * g_sched_dispatch() can also be called as a result of a timeout, * e.g. when doing anticipation or pacing requests. * * When a bio comes back, it goes to g_sched_done() which in turn * calls gs_done(). The latter does any necessary housekeeping in * the scheduling algorithm, and may decide to call g_sched_dispatch() * to send more bio's downstream. * * If an algorithm needs per-flow queues, these are created * calling gs_init_class() and destroyed with gs_fini_class(), * and they are also inserted in the hash table implemented in * the g_sched_softc{} * * If an algorithm is replaced, or a transparently-inserted node is * removed with "geom sched destroy", we need to remove all references * to the g_*_softc{} and g_sched_softc from the bio's still in * the scheduler. g_sched_forced_dispatch() helps doing this. * XXX need to explain better. */ #include #include #include #include #include #include #include #include #include #include #include #include #include /* we access curthread */ #include #include "gs_scheduler.h" #include "g_sched.h" /* geom hooks */ /* * Size of the per-geom hash table storing traffic classes. * We may decide to change it at a later time, it has no ABI * implications as it is only used for run-time allocations. */ #define G_SCHED_HASH_SIZE 32 static int g_sched_destroy(struct g_geom *gp, boolean_t force); static int g_sched_destroy_geom(struct gctl_req *req, struct g_class *mp, struct g_geom *gp); static void g_sched_config(struct gctl_req *req, struct g_class *mp, const char *verb); static struct g_geom *g_sched_taste(struct g_class *mp, struct g_provider *pp, int flags __unused); static void g_sched_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp); static void g_sched_init(struct g_class *mp); static void g_sched_fini(struct g_class *mp); static int g_sched_ioctl(struct g_provider *pp, u_long cmd, void *data, int fflag, struct thread *td); struct g_class g_sched_class = { .name = G_SCHED_CLASS_NAME, .version = G_VERSION, .ctlreq = g_sched_config, .taste = g_sched_taste, .destroy_geom = g_sched_destroy_geom, .init = g_sched_init, .ioctl = g_sched_ioctl, .fini = g_sched_fini }; MALLOC_DEFINE(M_GEOM_SCHED, "GEOM_SCHED", "Geom schedulers data structures"); /* * Global variables describing the state of the geom_sched module. * There is only one static instance of this structure. */ LIST_HEAD(gs_list, g_gsched); /* type, link field */ struct geom_sched_vars { struct mtx gs_mtx; struct gs_list gs_scheds; /* list of algorithms */ u_int gs_debug; u_int gs_sched_count; /* how many algorithms ? */ u_int gs_patched; /* g_io_request was patched */ u_int gs_initialized; u_int gs_expire_secs; /* expiration of hash entries */ struct bio_queue_head gs_pending; u_int gs_npending; /* The following are for stats, usually protected by gs_mtx. */ u_long gs_requests; /* total requests */ u_long gs_done; /* total done */ u_int gs_in_flight; /* requests in flight */ u_int gs_writes_in_flight; u_int gs_bytes_in_flight; u_int gs_write_bytes_in_flight; char gs_names[256]; /* names of schedulers */ }; static struct geom_sched_vars me = { .gs_expire_secs = 10, }; SYSCTL_DECL(_kern_geom); SYSCTL_NODE(_kern_geom, OID_AUTO, sched, CTLFLAG_RW, 0, "GEOM_SCHED stuff"); SYSCTL_UINT(_kern_geom_sched, OID_AUTO, in_flight_wb, CTLFLAG_RD, &me.gs_write_bytes_in_flight, 0, "Write bytes in flight"); SYSCTL_UINT(_kern_geom_sched, OID_AUTO, in_flight_b, CTLFLAG_RD, &me.gs_bytes_in_flight, 0, "Bytes in flight"); SYSCTL_UINT(_kern_geom_sched, OID_AUTO, in_flight_w, CTLFLAG_RD, &me.gs_writes_in_flight, 0, "Write Requests in flight"); SYSCTL_UINT(_kern_geom_sched, OID_AUTO, in_flight, CTLFLAG_RD, &me.gs_in_flight, 0, "Requests in flight"); SYSCTL_ULONG(_kern_geom_sched, OID_AUTO, done, CTLFLAG_RD, &me.gs_done, 0, "Total done"); SYSCTL_ULONG(_kern_geom_sched, OID_AUTO, requests, CTLFLAG_RD, &me.gs_requests, 0, "Total requests"); SYSCTL_STRING(_kern_geom_sched, OID_AUTO, algorithms, CTLFLAG_RD, &me.gs_names, 0, "Algorithm names"); SYSCTL_UINT(_kern_geom_sched, OID_AUTO, alg_count, CTLFLAG_RD, &me.gs_sched_count, 0, "Number of algorithms"); SYSCTL_UINT(_kern_geom_sched, OID_AUTO, debug, CTLFLAG_RW, &me.gs_debug, 0, "Debug level"); SYSCTL_UINT(_kern_geom_sched, OID_AUTO, expire_secs, CTLFLAG_RW, &me.gs_expire_secs, 0, "Expire time in seconds"); /* * g_sched calls the scheduler algorithms with this lock held. * The locking functions are exposed so the scheduler algorithms can also * protect themselves e.g. when running a callout handler. */ void g_sched_lock(struct g_geom *gp) { struct g_sched_softc *sc = gp->softc; mtx_lock(&sc->sc_mtx); } void g_sched_unlock(struct g_geom *gp) { struct g_sched_softc *sc = gp->softc; mtx_unlock(&sc->sc_mtx); } /* * Support functions to handle references to the module, * which are coming from devices using this scheduler. */ static inline void g_gsched_ref(struct g_gsched *gsp) { atomic_add_int(&gsp->gs_refs, 1); } static inline void g_gsched_unref(struct g_gsched *gsp) { atomic_add_int(&gsp->gs_refs, -1); } /* * Update the stats when this request is done. */ static void g_sched_update_stats(struct bio *bio) { me.gs_done++; me.gs_in_flight--; me.gs_bytes_in_flight -= bio->bio_length; if (bio->bio_cmd & BIO_WRITE) { me.gs_writes_in_flight--; me.gs_write_bytes_in_flight -= bio->bio_length; } } /* * Dispatch any pending request. */ static void g_sched_forced_dispatch(struct g_geom *gp) { struct g_sched_softc *sc = gp->softc; struct g_gsched *gsp = sc->sc_gsched; struct bio *bp; KASSERT(mtx_owned(&sc->sc_mtx), ("sc_mtx not owned during forced dispatch")); while ((bp = gsp->gs_next(sc->sc_data, 1)) != NULL) g_io_request(bp, LIST_FIRST(&gp->consumer)); } /* * The main dispatch loop, called either here after the start * routine, or by scheduling algorithms when they receive a timeout * or a 'done' notification. Does not share code with the forced * dispatch path, since the gs_done() callback can call us. */ void g_sched_dispatch(struct g_geom *gp) { struct g_sched_softc *sc = gp->softc; struct g_gsched *gsp = sc->sc_gsched; struct bio *bp; KASSERT(mtx_owned(&sc->sc_mtx), ("sc_mtx not owned during dispatch")); if ((sc->sc_flags & G_SCHED_FLUSHING)) return; while ((bp = gsp->gs_next(sc->sc_data, 0)) != NULL) g_io_request(bp, LIST_FIRST(&gp->consumer)); } /* * Recent (8.0 and above) versions of FreeBSD have support to * register classifiers of disk requests. The classifier is * invoked by g_io_request(), and stores the information into * bp->bio_classifier1. * * Support for older versions, which is left here only for * documentation purposes, relies on two hacks: * 1. classification info is written into the bio_caller1 * field of the topmost node in the bio chain. This field * is rarely used, but this module is incompatible with * those that use bio_caller1 for other purposes, * such as ZFS and gjournal; * 2. g_io_request() is patched in-memory when the module is * loaded, so that the function calls a classifier as its * first thing. g_io_request() is restored when the module * is unloaded. This functionality is only supported for * x86 and amd64, other architectures need source code changes. */ /* * Lookup the identity of the issuer of the original request. * In the current implementation we use the curthread of the * issuer, but different mechanisms may be implemented later * so we do not make assumptions on the return value which for * us is just an opaque identifier. */ static inline u_long g_sched_classify(struct bio *bp) { #if __FreeBSD_version > 800098 /* we have classifier fields in the struct bio */ #define HAVE_BIO_CLASSIFIER return ((u_long)bp->bio_classifier1); #else #warning old version!!! while (bp->bio_parent != NULL) bp = bp->bio_parent; return ((u_long)bp->bio_caller1); #endif } /* Return the hash chain for the given key. */ static inline struct g_hash * g_sched_hash(struct g_sched_softc *sc, u_long key) { return (&sc->sc_hash[key & sc->sc_mask]); } /* * Helper function for the children classes, which takes * a geom and a bio and returns the private descriptor * associated to the request. This involves fetching * the classification field and [al]locating the * corresponding entry in the hash table. */ void * g_sched_get_class(struct g_geom *gp, struct bio *bp) { struct g_sched_softc *sc; struct g_sched_class *gsc; struct g_gsched *gsp; struct g_hash *bucket; u_long key; sc = gp->softc; key = g_sched_classify(bp); bucket = g_sched_hash(sc, key); LIST_FOREACH(gsc, bucket, gsc_clist) { if (key == gsc->gsc_key) { gsc->gsc_refs++; return (gsc->gsc_priv); } } gsp = sc->sc_gsched; gsc = malloc(sizeof(*gsc) + gsp->gs_priv_size, M_GEOM_SCHED, M_NOWAIT | M_ZERO); if (!gsc) return (NULL); if (gsp->gs_init_class(sc->sc_data, gsc->gsc_priv)) { free(gsc, M_GEOM_SCHED); return (NULL); } gsc->gsc_refs = 2; /* 1 for the hash table, 1 for the caller. */ gsc->gsc_key = key; LIST_INSERT_HEAD(bucket, gsc, gsc_clist); gsc->gsc_expire = ticks + me.gs_expire_secs * hz; return (gsc->gsc_priv); } /* * Release a reference to the per-client descriptor, */ void g_sched_put_class(struct g_geom *gp, void *priv) { struct g_sched_class *gsc; struct g_sched_softc *sc; gsc = g_sched_priv2class(priv); gsc->gsc_expire = ticks + me.gs_expire_secs * hz; if (--gsc->gsc_refs > 0) return; sc = gp->softc; sc->sc_gsched->gs_fini_class(sc->sc_data, priv); LIST_REMOVE(gsc, gsc_clist); free(gsc, M_GEOM_SCHED); } static void g_sched_hash_fini(struct g_geom *gp, struct g_hash *hp, u_long mask, struct g_gsched *gsp, void *data) { struct g_sched_class *cp, *cp2; int i; if (!hp) return; if (data && gsp->gs_hash_unref) gsp->gs_hash_unref(data); for (i = 0; i < G_SCHED_HASH_SIZE; i++) { LIST_FOREACH_SAFE(cp, &hp[i], gsc_clist, cp2) g_sched_put_class(gp, cp->gsc_priv); } hashdestroy(hp, M_GEOM_SCHED, mask); } static struct g_hash * g_sched_hash_init(struct g_gsched *gsp, u_long *mask, int flags) { struct g_hash *hash; if (gsp->gs_priv_size == 0) return (NULL); hash = hashinit_flags(G_SCHED_HASH_SIZE, M_GEOM_SCHED, mask, flags); return (hash); } static void g_sched_flush_classes(struct g_geom *gp) { struct g_sched_softc *sc; struct g_sched_class *cp, *cp2; int i; sc = gp->softc; if (!sc->sc_hash || ticks - sc->sc_flush_ticks <= 0) return; for (i = 0; i < G_SCHED_HASH_SIZE; i++) { LIST_FOREACH_SAFE(cp, &sc->sc_hash[i], gsc_clist, cp2) { if (cp->gsc_refs == 1 && ticks - cp->gsc_expire > 0) g_sched_put_class(gp, cp->gsc_priv); } } sc->sc_flush_ticks = ticks + me.gs_expire_secs * hz; } /* * Wait for the completion of any outstanding request. To ensure * that this does not take forever the caller has to make sure that * no new request enter the scehduler before calling us. * * Must be called with the gp mutex held and topology locked. */ static int g_sched_wait_pending(struct g_geom *gp) { struct g_sched_softc *sc = gp->softc; int endticks = ticks + hz; g_topology_assert(); while (sc->sc_pending && endticks - ticks >= 0) msleep(gp, &sc->sc_mtx, 0, "sched_wait_pending", hz / 4); return (sc->sc_pending ? ETIMEDOUT : 0); } static int g_sched_remove_locked(struct g_geom *gp, struct g_gsched *gsp) { struct g_sched_softc *sc = gp->softc; int error; /* Set the flushing flag: new bios will not enter the scheduler. */ sc->sc_flags |= G_SCHED_FLUSHING; g_sched_forced_dispatch(gp); error = g_sched_wait_pending(gp); if (error) goto failed; /* No more requests pending or in flight from the old gsp. */ g_sched_hash_fini(gp, sc->sc_hash, sc->sc_mask, gsp, sc->sc_data); sc->sc_hash = NULL; /* * Avoid deadlock here by releasing the gp mutex and reacquiring * it once done. It should be safe, since no reconfiguration or * destruction can take place due to the geom topology lock; no * new request can use the current sc_data since we flagged the * geom as being flushed. */ g_sched_unlock(gp); gsp->gs_fini(sc->sc_data); g_sched_lock(gp); sc->sc_gsched = NULL; sc->sc_data = NULL; g_gsched_unref(gsp); failed: sc->sc_flags &= ~G_SCHED_FLUSHING; return (error); } static int g_sched_remove(struct g_geom *gp, struct g_gsched *gsp) { int error; g_sched_lock(gp); error = g_sched_remove_locked(gp, gsp); /* gsp is surely non-null */ g_sched_unlock(gp); return (error); } /* * Support function for create/taste -- locate the desired * algorithm and grab a reference to it. */ static struct g_gsched * g_gsched_find(const char *name) { struct g_gsched *gsp = NULL; mtx_lock(&me.gs_mtx); LIST_FOREACH(gsp, &me.gs_scheds, glist) { if (strcmp(name, gsp->gs_name) == 0) { g_gsched_ref(gsp); break; } } mtx_unlock(&me.gs_mtx); return (gsp); } /* * Rebuild the list of scheduler names. * To be called with me.gs_mtx lock held. */ static void g_gsched_build_names(struct g_gsched *gsp) { int pos, l; struct g_gsched *cur; pos = 0; LIST_FOREACH(cur, &me.gs_scheds, glist) { l = strlen(cur->gs_name); if (l + pos + 1 + 1 < sizeof(me.gs_names)) { if (pos != 0) me.gs_names[pos++] = ' '; strcpy(me.gs_names + pos, cur->gs_name); pos += l; } } me.gs_names[pos] = '\0'; } /* * Register or unregister individual scheduling algorithms. */ static int g_gsched_register(struct g_gsched *gsp) { struct g_gsched *cur; int error = 0; mtx_lock(&me.gs_mtx); LIST_FOREACH(cur, &me.gs_scheds, glist) { if (strcmp(gsp->gs_name, cur->gs_name) == 0) break; } if (cur != NULL) { G_SCHED_DEBUG(0, "A scheduler named %s already" "exists.", gsp->gs_name); error = EEXIST; } else { LIST_INSERT_HEAD(&me.gs_scheds, gsp, glist); gsp->gs_refs = 1; me.gs_sched_count++; g_gsched_build_names(gsp); } mtx_unlock(&me.gs_mtx); return (error); } struct g_gsched_unregparm { struct g_gsched *gup_gsp; int gup_error; }; static void g_gsched_unregister(void *arg, int flag) { struct g_gsched_unregparm *parm = arg; struct g_gsched *gsp = parm->gup_gsp, *cur, *tmp; struct g_sched_softc *sc; struct g_geom *gp, *gp_tmp; int error; parm->gup_error = 0; g_topology_assert(); if (flag == EV_CANCEL) return; mtx_lock(&me.gs_mtx); LIST_FOREACH_SAFE(gp, &g_sched_class.geom, geom, gp_tmp) { if (gp->class != &g_sched_class) continue; /* Should not happen. */ sc = gp->softc; if (sc->sc_gsched == gsp) { error = g_sched_remove(gp, gsp); if (error) goto failed; } } LIST_FOREACH_SAFE(cur, &me.gs_scheds, glist, tmp) { if (cur != gsp) continue; if (gsp->gs_refs != 1) { G_SCHED_DEBUG(0, "%s still in use.", gsp->gs_name); parm->gup_error = EBUSY; } else { LIST_REMOVE(gsp, glist); me.gs_sched_count--; g_gsched_build_names(gsp); } break; } if (cur == NULL) { G_SCHED_DEBUG(0, "%s not registered.", gsp->gs_name); parm->gup_error = ENOENT; } failed: mtx_unlock(&me.gs_mtx); } static inline void g_gsched_global_init(void) { if (!me.gs_initialized) { G_SCHED_DEBUG(0, "Initializing global data."); mtx_init(&me.gs_mtx, "gsched", NULL, MTX_DEF); LIST_INIT(&me.gs_scheds); gs_bioq_init(&me.gs_pending); me.gs_initialized = 1; } } /* * Module event called when a scheduling algorithm module is loaded or * unloaded. */ int g_gsched_modevent(module_t mod, int cmd, void *arg) { struct g_gsched *gsp = arg; struct g_gsched_unregparm parm; int error; G_SCHED_DEBUG(0, "Modevent %d.", cmd); /* * If the module is loaded at boot, the geom thread that calls * g_sched_init() might actually run after g_gsched_modevent(), * so make sure that the module is properly initialized. */ g_gsched_global_init(); error = EOPNOTSUPP; switch (cmd) { case MOD_LOAD: error = g_gsched_register(gsp); G_SCHED_DEBUG(0, "Loaded module %s error %d.", gsp->gs_name, error); if (error == 0) g_retaste(&g_sched_class); break; case MOD_UNLOAD: parm.gup_gsp = gsp; parm.gup_error = 0; error = g_waitfor_event(g_gsched_unregister, &parm, M_WAITOK, NULL); if (error == 0) error = parm.gup_error; G_SCHED_DEBUG(0, "Unloaded module %s error %d.", gsp->gs_name, error); break; }; return (error); } #ifdef KTR #define TRC_BIO_EVENT(e, bp) g_sched_trace_bio_ ## e (bp) static inline char g_sched_type(struct bio *bp) { if (0 != (bp->bio_cmd & BIO_READ)) return ('R'); else if (0 != (bp->bio_cmd & BIO_WRITE)) return ('W'); return ('U'); } static inline void g_sched_trace_bio_START(struct bio *bp) { CTR5(KTR_GSCHED, "S %lu %c %lu/%lu %lu", g_sched_classify(bp), g_sched_type(bp), bp->bio_offset / ULONG_MAX, bp->bio_offset, bp->bio_length); } static inline void g_sched_trace_bio_DONE(struct bio *bp) { CTR5(KTR_GSCHED, "D %lu %c %lu/%lu %lu", g_sched_classify(bp), g_sched_type(bp), bp->bio_offset / ULONG_MAX, bp->bio_offset, bp->bio_length); } #else /* !KTR */ #define TRC_BIO_EVENT(e, bp) #endif /* !KTR */ /* * g_sched_done() and g_sched_start() dispatch the geom requests to * the scheduling algorithm in use. */ static void g_sched_done(struct bio *bio) { struct g_geom *gp = bio->bio_caller2; struct g_sched_softc *sc = gp->softc; TRC_BIO_EVENT(DONE, bio); KASSERT(bio->bio_caller1, ("null bio_caller1 in g_sched_done")); g_sched_lock(gp); g_sched_update_stats(bio); sc->sc_gsched->gs_done(sc->sc_data, bio); if (!--sc->sc_pending) wakeup(gp); g_sched_flush_classes(gp); g_sched_unlock(gp); g_std_done(bio); } static void g_sched_start(struct bio *bp) { struct g_geom *gp = bp->bio_to->geom; struct g_sched_softc *sc = gp->softc; struct bio *cbp; TRC_BIO_EVENT(START, bp); G_SCHED_LOGREQ(bp, "Request received."); cbp = g_clone_bio(bp); if (cbp == NULL) { g_io_deliver(bp, ENOMEM); return; } cbp->bio_done = g_sched_done; cbp->bio_to = LIST_FIRST(&gp->provider); KASSERT(cbp->bio_to != NULL, ("NULL provider")); /* We only schedule reads and writes. */ if (0 == (bp->bio_cmd & (BIO_READ | BIO_WRITE))) goto bypass; G_SCHED_LOGREQ(cbp, "Sending request."); g_sched_lock(gp); /* * Call the algorithm's gs_start to queue the request in the * scheduler. If gs_start fails then pass the request down, * otherwise call g_sched_dispatch() which tries to push * one or more requests down. */ if (!sc->sc_gsched || (sc->sc_flags & G_SCHED_FLUSHING) || sc->sc_gsched->gs_start(sc->sc_data, cbp)) { g_sched_unlock(gp); goto bypass; } /* * We use bio_caller1 to mark requests that are scheduled * so make sure it is not NULL. */ if (cbp->bio_caller1 == NULL) cbp->bio_caller1 = &me; /* anything not NULL */ cbp->bio_caller2 = gp; sc->sc_pending++; /* Update general stats. */ me.gs_in_flight++; me.gs_requests++; me.gs_bytes_in_flight += bp->bio_length; if (bp->bio_cmd & BIO_WRITE) { me.gs_writes_in_flight++; me.gs_write_bytes_in_flight += bp->bio_length; } g_sched_dispatch(gp); g_sched_unlock(gp); return; bypass: cbp->bio_done = g_std_done; cbp->bio_caller1 = NULL; /* not scheduled */ g_io_request(cbp, LIST_FIRST(&gp->consumer)); } /* * The next few functions are the geom glue. */ static void g_sched_orphan(struct g_consumer *cp) { g_topology_assert(); g_sched_destroy(cp->geom, 1); } static int g_sched_access(struct g_provider *pp, int dr, int dw, int de) { struct g_geom *gp; struct g_consumer *cp; int error; gp = pp->geom; cp = LIST_FIRST(&gp->consumer); error = g_access(cp, dr, dw, de); return (error); } static void g_sched_temporary_start(struct bio *bio) { mtx_lock(&me.gs_mtx); me.gs_npending++; gs_bioq_disksort(&me.gs_pending, bio); mtx_unlock(&me.gs_mtx); } static void g_sched_flush_pending(g_start_t *start) { struct bio *bp; while ((bp = gs_bioq_takefirst(&me.gs_pending))) start(bp); } static int g_insert_proxy(struct g_geom *gp, struct g_provider *newpp, struct g_geom *dstgp, struct g_provider *pp, struct g_consumer *cp) { struct g_sched_softc *sc = gp->softc; g_start_t *saved_start, *flush = g_sched_start; int error = 0, endticks = ticks + hz; g_cancel_event(newpp); /* prevent taste() */ /* copy private fields */ newpp->private = pp->private; newpp->index = pp->index; /* Queue all the early requests coming for us. */ me.gs_npending = 0; saved_start = pp->geom->start; dstgp->start = g_sched_temporary_start; while (pp->nstart - pp->nend != me.gs_npending && endticks - ticks >= 0) tsleep(pp, PRIBIO, "-", hz/10); if (pp->nstart - pp->nend != me.gs_npending) { flush = saved_start; error = ETIMEDOUT; goto fail; } /* link pp to this geom */ LIST_REMOVE(pp, provider); pp->geom = gp; LIST_INSERT_HEAD(&gp->provider, pp, provider); /* * replicate the counts from the parent in the * new provider and consumer nodes */ cp->acr = newpp->acr = pp->acr; cp->acw = newpp->acw = pp->acw; cp->ace = newpp->ace = pp->ace; sc->sc_flags |= G_SCHED_PROXYING; fail: dstgp->start = saved_start; g_sched_flush_pending(flush); return (error); } /* * Create a geom node for the device passed as *pp. * If successful, add a reference to this gsp. */ static int g_sched_create(struct gctl_req *req, struct g_class *mp, struct g_provider *pp, struct g_gsched *gsp, int proxy) { struct g_sched_softc *sc = NULL; struct g_geom *gp, *dstgp; struct g_provider *newpp = NULL; struct g_consumer *cp = NULL; char name[64]; int error; g_topology_assert(); snprintf(name, sizeof(name), "%s%s", pp->name, G_SCHED_SUFFIX); LIST_FOREACH(gp, &mp->geom, geom) { if (strcmp(gp->name, name) == 0) { gctl_error(req, "Geom %s already exists.", name); return (EEXIST); } } gp = g_new_geomf(mp, name); dstgp = proxy ? pp->geom : gp; /* where do we link the provider */ - if (gp == NULL) { - gctl_error(req, "Cannot create geom %s.", name); - error = ENOMEM; - goto fail; - } sc = g_malloc(sizeof(*sc), M_WAITOK | M_ZERO); sc->sc_gsched = gsp; sc->sc_data = gsp->gs_init(gp); if (sc->sc_data == NULL) { error = ENOMEM; goto fail; } sc->sc_hash = g_sched_hash_init(gsp, &sc->sc_mask, HASH_WAITOK); /* * Do not initialize the flush mechanism, will be initialized * on the first insertion on the hash table. */ mtx_init(&sc->sc_mtx, "g_sched_mtx", NULL, MTX_DEF); gp->softc = sc; gp->start = g_sched_start; gp->orphan = g_sched_orphan; gp->access = g_sched_access; gp->dumpconf = g_sched_dumpconf; newpp = g_new_providerf(dstgp, gp->name); - if (newpp == NULL) { - gctl_error(req, "Cannot create provider %s.", name); - error = ENOMEM; - goto fail; - } - newpp->mediasize = pp->mediasize; newpp->sectorsize = pp->sectorsize; cp = g_new_consumer(gp); - if (cp == NULL) { - gctl_error(req, "Cannot create consumer for %s.", - gp->name); - error = ENOMEM; - goto fail; - } - error = g_attach(cp, proxy ? newpp : pp); if (error != 0) { gctl_error(req, "Cannot attach to provider %s.", pp->name); goto fail; } g_error_provider(newpp, 0); if (proxy) { error = g_insert_proxy(gp, newpp, dstgp, pp, cp); if (error) goto fail; } G_SCHED_DEBUG(0, "Device %s created.", gp->name); g_gsched_ref(gsp); return (0); fail: if (cp != NULL) { if (cp->provider != NULL) g_detach(cp); g_destroy_consumer(cp); } - if (newpp != NULL) g_destroy_provider(newpp); - - if (sc && sc->sc_hash) { + if (sc->sc_hash) g_sched_hash_fini(gp, sc->sc_hash, sc->sc_mask, gsp, sc->sc_data); - } - - if (sc && sc->sc_data) + if (sc->sc_data) gsp->gs_fini(sc->sc_data); - - if (gp != NULL) { - if (gp->softc != NULL) - g_free(gp->softc); - g_destroy_geom(gp); - } + g_free(gp->softc); + g_destroy_geom(gp); return (error); } /* * Support for dynamic switching of scheduling algorithms. * First initialize the data structures for the new algorithm, * then call g_sched_remove_locked() to flush all references * to the old one, finally link the new algorithm. */ static int g_sched_change_algo(struct gctl_req *req, struct g_class *mp, struct g_provider *pp, struct g_gsched *gsp) { struct g_sched_softc *sc; struct g_geom *gp; struct g_hash *newh; void *data; u_long mask; int error = 0; gp = pp->geom; sc = gp->softc; data = gsp->gs_init(gp); if (data == NULL) return (ENOMEM); newh = g_sched_hash_init(gsp, &mask, HASH_WAITOK); if (gsp->gs_priv_size && !newh) { error = ENOMEM; goto fail; } g_sched_lock(gp); if (sc->sc_gsched) { /* can be NULL in some cases */ error = g_sched_remove_locked(gp, sc->sc_gsched); if (error) goto fail; } g_gsched_ref(gsp); sc->sc_gsched = gsp; sc->sc_data = data; sc->sc_hash = newh; sc->sc_mask = mask; g_sched_unlock(gp); return (0); fail: if (newh) g_sched_hash_fini(gp, newh, mask, gsp, data); if (data) gsp->gs_fini(data); g_sched_unlock(gp); return (error); } /* * Stop the request flow directed to the proxy, redirecting the new * requests to the me.gs_pending queue. */ static struct g_provider * g_detach_proxy(struct g_geom *gp) { struct g_consumer *cp; struct g_provider *pp, *newpp; do { pp = LIST_FIRST(&gp->provider); if (pp == NULL) break; cp = LIST_FIRST(&gp->consumer); if (cp == NULL) break; newpp = cp->provider; if (newpp == NULL) break; me.gs_npending = 0; pp->geom->start = g_sched_temporary_start; return (pp); } while (0); printf("%s error detaching proxy %s\n", __FUNCTION__, gp->name); return (NULL); } static void g_sched_blackhole(struct bio *bp) { g_io_deliver(bp, ENXIO); } static inline void g_reparent_provider(struct g_provider *pp, struct g_geom *gp, struct g_provider *newpp) { LIST_REMOVE(pp, provider); if (newpp) { pp->private = newpp->private; pp->index = newpp->index; } pp->geom = gp; LIST_INSERT_HEAD(&gp->provider, pp, provider); } static inline void g_unproxy_provider(struct g_provider *oldpp, struct g_provider *newpp) { struct g_geom *gp = oldpp->geom; g_reparent_provider(oldpp, newpp->geom, newpp); /* * Hackish: let the system destroy the old provider for us, just * in case someone attached a consumer to it, in which case a * direct call to g_destroy_provider() would not work. */ g_reparent_provider(newpp, gp, NULL); } /* * Complete the proxy destruction, linking the old provider to its * original geom, and destroying the proxy provider. Also take care * of issuing the pending requests collected in me.gs_pending (if any). */ static int g_destroy_proxy(struct g_geom *gp, struct g_provider *oldpp) { struct g_consumer *cp; struct g_provider *newpp; do { cp = LIST_FIRST(&gp->consumer); if (cp == NULL) break; newpp = cp->provider; if (newpp == NULL) break; /* Relink the provider to its original geom. */ g_unproxy_provider(oldpp, newpp); /* Detach consumer from provider, and destroy provider. */ cp->acr = newpp->acr = 0; cp->acw = newpp->acw = 0; cp->ace = newpp->ace = 0; g_detach(cp); /* Send the pending bios through the right start function. */ g_sched_flush_pending(oldpp->geom->start); return (0); } while (0); printf("%s error destroying proxy %s\n", __FUNCTION__, gp->name); /* We cannot send the pending bios anywhere... */ g_sched_flush_pending(g_sched_blackhole); return (EINVAL); } static int g_sched_destroy(struct g_geom *gp, boolean_t force) { struct g_provider *pp, *oldpp = NULL; struct g_sched_softc *sc; struct g_gsched *gsp; int error; g_topology_assert(); sc = gp->softc; if (sc == NULL) return (ENXIO); if (!(sc->sc_flags & G_SCHED_PROXYING)) { pp = LIST_FIRST(&gp->provider); if (pp && (pp->acr != 0 || pp->acw != 0 || pp->ace != 0)) { const char *msg = force ? "but we force removal" : "cannot remove"; G_SCHED_DEBUG(!force, "Device %s is still open (r%dw%de%d), %s.", pp->name, pp->acr, pp->acw, pp->ace, msg); if (!force) return (EBUSY); } else { G_SCHED_DEBUG(0, "Device %s removed.", gp->name); } } else oldpp = g_detach_proxy(gp); gsp = sc->sc_gsched; if (gsp) { /* * XXX bad hack here: force a dispatch to release * any reference to the hash table still held by * the scheduler. */ g_sched_lock(gp); /* * We are dying here, no new requests should enter * the scheduler. This is granted by the topolgy, * either in case we were proxying (new bios are * being redirected) or not (see the access check * above). */ g_sched_forced_dispatch(gp); error = g_sched_wait_pending(gp); if (error) { /* * Not all the requests came home: this might happen * under heavy load, or if we were waiting for any * bio which is served in the event path (see * geom_slice.c for an example of how this can * happen). Try to restore a working configuration * if we can fail. */ if ((sc->sc_flags & G_SCHED_PROXYING) && oldpp) { g_sched_flush_pending(force ? g_sched_blackhole : g_sched_start); } /* * In the forced destroy case there is not so much * we can do, we have pending bios that will call * g_sched_done() somehow, and we don't want them * to crash the system using freed memory. We tell * the user that something went wrong, and leak some * memory here. * Note: the callers using force = 1 ignore the * return value. */ if (force) { G_SCHED_DEBUG(0, "Pending requests while " " destroying geom, some memory leaked."); } return (error); } g_sched_unlock(gp); g_sched_hash_fini(gp, sc->sc_hash, sc->sc_mask, gsp, sc->sc_data); sc->sc_hash = NULL; gsp->gs_fini(sc->sc_data); g_gsched_unref(gsp); sc->sc_gsched = NULL; } if ((sc->sc_flags & G_SCHED_PROXYING) && oldpp) { error = g_destroy_proxy(gp, oldpp); if (error) { if (force) { G_SCHED_DEBUG(0, "Unrecoverable error while " "destroying a proxy geom, leaking some " " memory."); } return (error); } } mtx_destroy(&sc->sc_mtx); g_free(gp->softc); gp->softc = NULL; g_wither_geom(gp, ENXIO); return (error); } static int g_sched_destroy_geom(struct gctl_req *req, struct g_class *mp, struct g_geom *gp) { return (g_sched_destroy(gp, 0)); } /* * Functions related to the classification of requests. * * On recent FreeBSD versions (8.0 and above), we store a reference * to the issuer of a request in bp->bio_classifier1 as soon * as the bio is posted to the geom queue (and not later, because * requests are managed by the g_down thread afterwards). * * On older versions of the system (but this code is not used * in any existing release), we [ab]use the caller1 field in the * root element of the bio tree to store the classification info. * The marking is done at the beginning of g_io_request() * and only if we find that the field is NULL. * * To avoid rebuilding the kernel, this module will patch the * initial part of g_io_request() so it jumps to some hand-coded * assembly that does the marking and then executes the original * body of g_io_request(). * * fake_ioreq[] is architecture-specific machine code * that implements the above. CODE_SIZE, STORE_SIZE etc. * are constants used in the patching routine. Look at the * code in g_ioreq_patch() for the details. */ #ifndef HAVE_BIO_CLASSIFIER /* * Support for old FreeBSD versions */ #if defined(__i386__) #define CODE_SIZE 29 #define STORE_SIZE 5 #define EPILOGUE 5 #define SIZE (CODE_SIZE + STORE_SIZE + EPILOGUE) static u_char fake_ioreq[SIZE] = { 0x8b, 0x44, 0x24, 0x04, /* mov bp, %eax */ /* 1: */ 0x89, 0xc2, /* mov %eax, %edx # edx = bp */ 0x8b, 0x40, 0x64, /* mov bp->bio_parent, %eax */ 0x85, 0xc0, /* test %eax, %eax */ 0x75, 0xf7, /* jne 1b */ 0x8b, 0x42, 0x30, /* mov bp->bp_caller1, %eax */ 0x85, 0xc0, /* test %eax, %eax */ 0x75, 0x09, /* jne 2f */ 0x64, 0xa1, 0x00, 0x00, /* mov %fs:0, %eax */ 0x00, 0x00, 0x89, 0x42, 0x30, /* mov %eax, bp->bio_caller1 */ /* 2: */ 0x55, 0x89, 0xe5, 0x57, 0x56, 0xe9, 0x00, 0x00, 0x00, 0x00, /* jmp back... */ }; #elif defined(__amd64) #define CODE_SIZE 38 #define STORE_SIZE 6 #define EPILOGUE 5 #define SIZE (CODE_SIZE + STORE_SIZE + EPILOGUE) static u_char fake_ioreq[SIZE] = { 0x48, 0x89, 0xf8, /* mov bp, %rax */ /* 1: */ 0x48, 0x89, 0xc2, /* mov %rax, %rdx # rdx = bp */ 0x48, 0x8b, 0x82, 0xa8, /* mov bp->bio_parent, %rax */ 0x00, 0x00, 0x00, 0x48, 0x85, 0xc0, /* test %rax, %rax */ 0x75, 0xf1, /* jne 1b */ 0x48, 0x83, 0x7a, 0x58, /* cmp $0, bp->bp_caller1 */ 0x00, 0x75, 0x0d, /* jne 2f */ 0x65, 0x48, 0x8b, 0x04, /* mov %gs:0, %rax */ 0x25, 0x00, 0x00, 0x00, 0x00, 0x48, 0x89, 0x42, 0x58, /* mov %rax, bp->bio_caller1 */ /* 2: */ 0x55, 0x48, 0x89, 0xe5, 0x41, 0x56, 0xe9, 0x00, 0x00, 0x00, 0x00, /* jmp back... */ }; #else /* neither x86 nor amd64 */ static void g_new_io_request(struct bio *bp, struct g_consumer *cp) { struct bio *top = bp; /* * bio classification: if bio_caller1 is available in the * root of the 'struct bio' tree, store there the thread id * of the thread that originated the request. * More sophisticated classification schemes can be used. */ while (top->bio_parent) top = top->bio_parent; if (top->bio_caller1 == NULL) top->bio_caller1 = curthread; } #error please add the code above in g_new_io_request() to the beginning of \ /sys/geom/geom_io.c::g_io_request(), and remove this line. #endif /* end of arch-specific code */ static int g_ioreq_patch(void) { u_char *original; u_long ofs; int found; if (me.gs_patched) return (-1); original = (u_char *)g_io_request; found = !bcmp(original, fake_ioreq + CODE_SIZE, STORE_SIZE); if (!found) return (-1); /* Jump back to the original + STORE_SIZE. */ ofs = (original + STORE_SIZE) - (fake_ioreq + SIZE); bcopy(&ofs, fake_ioreq + CODE_SIZE + STORE_SIZE + 1, 4); /* Patch the original address with a jump to the trampoline. */ *original = 0xe9; /* jump opcode */ ofs = fake_ioreq - (original + 5); bcopy(&ofs, original + 1, 4); me.gs_patched = 1; return (0); } /* * Restore the original code, this is easy. */ static void g_ioreq_restore(void) { u_char *original; if (me.gs_patched) { original = (u_char *)g_io_request; bcopy(fake_ioreq + CODE_SIZE, original, STORE_SIZE); me.gs_patched = 0; } } static inline void g_classifier_ini(void) { g_ioreq_patch(); } static inline void g_classifier_fini(void) { g_ioreq_restore(); } /*--- end of support code for older FreeBSD versions */ #else /* HAVE_BIO_CLASSIFIER */ /* * Classifier support for recent FreeBSD versions: we use * a very simple classifier, only use curthread to tag a request. * The classifier is registered at module load, and unregistered * at module unload. */ static int g_sched_tag(void *arg, struct bio *bp) { bp->bio_classifier1 = curthread; return (1); } static struct g_classifier_hook g_sched_classifier = { .func = g_sched_tag, }; static inline void g_classifier_ini(void) { g_register_classifier(&g_sched_classifier); } static inline void g_classifier_fini(void) { g_unregister_classifier(&g_sched_classifier); } #endif /* HAVE_BIO_CLASSIFIER */ static void g_sched_init(struct g_class *mp) { g_gsched_global_init(); G_SCHED_DEBUG(0, "Loading: mp = %p, g_sched_class = %p.", mp, &g_sched_class); /* Patch g_io_request to store classification info in the bio. */ g_classifier_ini(); } static void g_sched_fini(struct g_class *mp) { g_classifier_fini(); G_SCHED_DEBUG(0, "Unloading..."); KASSERT(LIST_EMPTY(&me.gs_scheds), ("still registered schedulers")); mtx_destroy(&me.gs_mtx); } static int g_sched_ioctl(struct g_provider *pp, u_long cmd, void *data, int fflag, struct thread *td) { struct g_consumer *cp; struct g_geom *gp; cp = LIST_FIRST(&pp->geom->consumer); if (cp == NULL) return (ENOIOCTL); gp = cp->provider->geom; if (gp->ioctl == NULL) return (ENOIOCTL); return (gp->ioctl(cp->provider, cmd, data, fflag, td)); } /* * Read the i-th argument for a request, skipping the /dev/ * prefix if present. */ static const char * g_sched_argi(struct gctl_req *req, int i) { static const char *dev_prefix = "/dev/"; const char *name; char param[16]; int l = strlen(dev_prefix); snprintf(param, sizeof(param), "arg%d", i); name = gctl_get_asciiparam(req, param); if (name == NULL) gctl_error(req, "No 'arg%d' argument", i); else if (strncmp(name, dev_prefix, l) == 0) name += l; return (name); } /* * Fetch nargs and do appropriate checks. */ static int g_sched_get_nargs(struct gctl_req *req) { int *nargs; nargs = gctl_get_paraml(req, "nargs", sizeof(*nargs)); if (nargs == NULL) { gctl_error(req, "No 'nargs' argument"); return (0); } if (*nargs <= 0) gctl_error(req, "Missing device(s)."); return (*nargs); } /* * Check whether we should add the class on certain volumes when * this geom is created. Right now this is under control of a kenv * variable containing the names of all devices that we care about. * Probably we should only support transparent insertion as the * preferred mode of operation. */ static struct g_geom * g_sched_taste(struct g_class *mp, struct g_provider *pp, int flags __unused) { struct g_gsched *gsp = NULL; /* the . algorithm we want */ const char *s; /* generic string pointer */ const char *taste_names; /* devices we like */ int l; g_trace(G_T_TOPOLOGY, "%s(%s, %s)", __func__, mp->name, pp->name); g_topology_assert(); G_SCHED_DEBUG(2, "Tasting %s.", pp->name); do { /* do not taste on ourselves */ if (pp->geom->class == mp) break; taste_names = getenv("geom.sched.taste"); if (taste_names == NULL) break; l = strlen(pp->name); for (s = taste_names; *s && (s = strstr(s, pp->name)); s++) { /* further checks for an exact match */ if ( (s == taste_names || s[-1] == ' ') && (s[l] == '\0' || s[l] == ' ') ) break; } if (s == NULL) break; G_SCHED_DEBUG(0, "Attach device %s match [%s]\n", pp->name, s); /* look up the provider name in the list */ s = getenv("geom.sched.algo"); if (s == NULL) s = "rr"; gsp = g_gsched_find(s); /* also get a reference */ if (gsp == NULL) { G_SCHED_DEBUG(0, "Bad '%s' algorithm.", s); break; } /* XXX create with 1 as last argument ? */ g_sched_create(NULL, mp, pp, gsp, 0); g_gsched_unref(gsp); } while (0); return NULL; } static void g_sched_ctl_create(struct gctl_req *req, struct g_class *mp, int proxy) { struct g_provider *pp; struct g_gsched *gsp; const char *name; int i, nargs; g_topology_assert(); name = gctl_get_asciiparam(req, "algo"); if (name == NULL) { gctl_error(req, "No '%s' argument", "algo"); return; } gsp = g_gsched_find(name); /* also get a reference */ if (gsp == NULL) { gctl_error(req, "Bad algorithm '%s'", name); return; } nargs = g_sched_get_nargs(req); /* * Run on the arguments, and break on any error. * We look for a device name, but skip the /dev/ prefix if any. */ for (i = 0; i < nargs; i++) { name = g_sched_argi(req, i); if (name == NULL) break; pp = g_provider_by_name(name); if (pp == NULL) { G_SCHED_DEBUG(1, "Provider %s is invalid.", name); gctl_error(req, "Provider %s is invalid.", name); break; } if (g_sched_create(req, mp, pp, gsp, proxy) != 0) break; } g_gsched_unref(gsp); } static void g_sched_ctl_configure(struct gctl_req *req, struct g_class *mp) { struct g_provider *pp; struct g_gsched *gsp; const char *name; int i, nargs; g_topology_assert(); name = gctl_get_asciiparam(req, "algo"); if (name == NULL) { gctl_error(req, "No '%s' argument", "algo"); return; } gsp = g_gsched_find(name); /* also get a reference */ if (gsp == NULL) { gctl_error(req, "Bad algorithm '%s'", name); return; } nargs = g_sched_get_nargs(req); /* * Run on the arguments, and break on any error. * We look for a device name, but skip the /dev/ prefix if any. */ for (i = 0; i < nargs; i++) { name = g_sched_argi(req, i); if (name == NULL) break; pp = g_provider_by_name(name); if (pp == NULL || pp->geom->class != mp) { G_SCHED_DEBUG(1, "Provider %s is invalid.", name); gctl_error(req, "Provider %s is invalid.", name); break; } if (g_sched_change_algo(req, mp, pp, gsp) != 0) break; } g_gsched_unref(gsp); } static struct g_geom * g_sched_find_geom(struct g_class *mp, const char *name) { struct g_geom *gp; LIST_FOREACH(gp, &mp->geom, geom) { if (strcmp(gp->name, name) == 0) return (gp); } return (NULL); } static void g_sched_ctl_destroy(struct gctl_req *req, struct g_class *mp) { int nargs, *force, error, i; struct g_geom *gp; const char *name; g_topology_assert(); nargs = g_sched_get_nargs(req); force = gctl_get_paraml(req, "force", sizeof(*force)); if (force == NULL) { gctl_error(req, "No 'force' argument"); return; } for (i = 0; i < nargs; i++) { name = g_sched_argi(req, i); if (name == NULL) break; gp = g_sched_find_geom(mp, name); if (gp == NULL) { G_SCHED_DEBUG(1, "Device %s is invalid.", name); gctl_error(req, "Device %s is invalid.", name); break; } error = g_sched_destroy(gp, *force); if (error != 0) { gctl_error(req, "Cannot destroy device %s (error=%d).", gp->name, error); break; } } } static void g_sched_config(struct gctl_req *req, struct g_class *mp, const char *verb) { uint32_t *version; g_topology_assert(); version = gctl_get_paraml(req, "version", sizeof(*version)); if (version == NULL) { gctl_error(req, "No '%s' argument.", "version"); return; } if (*version != G_SCHED_VERSION) { gctl_error(req, "Userland and kernel parts are " "out of sync."); return; } if (strcmp(verb, "create") == 0) { g_sched_ctl_create(req, mp, 0); return; } else if (strcmp(verb, "insert") == 0) { g_sched_ctl_create(req, mp, 1); return; } else if (strcmp(verb, "configure") == 0) { g_sched_ctl_configure(req, mp); return; } else if (strcmp(verb, "destroy") == 0) { g_sched_ctl_destroy(req, mp); return; } gctl_error(req, "Unknown verb."); } static void g_sched_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp) { struct g_sched_softc *sc = gp->softc; struct g_gsched *gsp = sc->sc_gsched; if (indent == NULL) { /* plaintext */ sbuf_printf(sb, " algo %s", gsp ? gsp->gs_name : "--"); } if (gsp != NULL && gsp->gs_dumpconf) gsp->gs_dumpconf(sb, indent, gp, cp, pp); } DECLARE_GEOM_CLASS(g_sched_class, g_sched); MODULE_VERSION(geom_sched, 0); Index: projects/largeSMP/sys/geom/shsec/g_shsec.c =================================================================== --- projects/largeSMP/sys/geom/shsec/g_shsec.c (revision 221494) +++ projects/largeSMP/sys/geom/shsec/g_shsec.c (revision 221495) @@ -1,838 +1,836 @@ /*- * Copyright (c) 2005 Pawel Jakub Dawidek * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include FEATURE(geom_shsec, "GEOM shared secret device support"); static MALLOC_DEFINE(M_SHSEC, "shsec_data", "GEOM_SHSEC Data"); static uma_zone_t g_shsec_zone; static int g_shsec_destroy(struct g_shsec_softc *sc, boolean_t force); static int g_shsec_destroy_geom(struct gctl_req *req, struct g_class *mp, struct g_geom *gp); static g_taste_t g_shsec_taste; static g_ctl_req_t g_shsec_config; static g_dumpconf_t g_shsec_dumpconf; static g_init_t g_shsec_init; static g_fini_t g_shsec_fini; struct g_class g_shsec_class = { .name = G_SHSEC_CLASS_NAME, .version = G_VERSION, .ctlreq = g_shsec_config, .taste = g_shsec_taste, .destroy_geom = g_shsec_destroy_geom, .init = g_shsec_init, .fini = g_shsec_fini }; SYSCTL_DECL(_kern_geom); SYSCTL_NODE(_kern_geom, OID_AUTO, shsec, CTLFLAG_RW, 0, "GEOM_SHSEC stuff"); static u_int g_shsec_debug = 0; TUNABLE_INT("kern.geom.shsec.debug", &g_shsec_debug); SYSCTL_UINT(_kern_geom_shsec, OID_AUTO, debug, CTLFLAG_RW, &g_shsec_debug, 0, "Debug level"); static u_int g_shsec_maxmem = MAXPHYS * 100; TUNABLE_INT("kern.geom.shsec.maxmem", &g_shsec_maxmem); SYSCTL_UINT(_kern_geom_shsec, OID_AUTO, maxmem, CTLFLAG_RD, &g_shsec_maxmem, 0, "Maximum memory that can be allocated for I/O (in bytes)"); static u_int g_shsec_alloc_failed = 0; SYSCTL_UINT(_kern_geom_shsec, OID_AUTO, alloc_failed, CTLFLAG_RD, &g_shsec_alloc_failed, 0, "How many times I/O allocation failed"); /* * Greatest Common Divisor. */ static u_int gcd(u_int a, u_int b) { u_int c; while (b != 0) { c = a; a = b; b = (c % b); } return (a); } /* * Least Common Multiple. */ static u_int lcm(u_int a, u_int b) { return ((a * b) / gcd(a, b)); } static void g_shsec_init(struct g_class *mp __unused) { g_shsec_zone = uma_zcreate("g_shsec_zone", MAXPHYS, NULL, NULL, NULL, NULL, 0, 0); g_shsec_maxmem -= g_shsec_maxmem % MAXPHYS; uma_zone_set_max(g_shsec_zone, g_shsec_maxmem / MAXPHYS); } static void g_shsec_fini(struct g_class *mp __unused) { uma_zdestroy(g_shsec_zone); } /* * Return the number of valid disks. */ static u_int g_shsec_nvalid(struct g_shsec_softc *sc) { u_int i, no; no = 0; for (i = 0; i < sc->sc_ndisks; i++) { if (sc->sc_disks[i] != NULL) no++; } return (no); } static void g_shsec_remove_disk(struct g_consumer *cp) { struct g_shsec_softc *sc; u_int no; KASSERT(cp != NULL, ("Non-valid disk in %s.", __func__)); sc = (struct g_shsec_softc *)cp->private; KASSERT(sc != NULL, ("NULL sc in %s.", __func__)); no = cp->index; G_SHSEC_DEBUG(0, "Disk %s removed from %s.", cp->provider->name, sc->sc_name); sc->sc_disks[no] = NULL; if (sc->sc_provider != NULL) { g_orphan_provider(sc->sc_provider, ENXIO); sc->sc_provider = NULL; G_SHSEC_DEBUG(0, "Device %s removed.", sc->sc_name); } if (cp->acr > 0 || cp->acw > 0 || cp->ace > 0) g_access(cp, -cp->acr, -cp->acw, -cp->ace); g_detach(cp); g_destroy_consumer(cp); } static void g_shsec_orphan(struct g_consumer *cp) { struct g_shsec_softc *sc; struct g_geom *gp; g_topology_assert(); gp = cp->geom; sc = gp->softc; if (sc == NULL) return; g_shsec_remove_disk(cp); /* If there are no valid disks anymore, remove device. */ if (g_shsec_nvalid(sc) == 0) g_shsec_destroy(sc, 1); } static int g_shsec_access(struct g_provider *pp, int dr, int dw, int de) { struct g_consumer *cp1, *cp2; struct g_shsec_softc *sc; struct g_geom *gp; int error; gp = pp->geom; sc = gp->softc; if (sc == NULL) { /* * It looks like geom is being withered. * In that case we allow only negative requests. */ KASSERT(dr <= 0 && dw <= 0 && de <= 0, ("Positive access request (device=%s).", pp->name)); if ((pp->acr + dr) == 0 && (pp->acw + dw) == 0 && (pp->ace + de) == 0) { G_SHSEC_DEBUG(0, "Device %s definitely destroyed.", gp->name); } return (0); } /* On first open, grab an extra "exclusive" bit */ if (pp->acr == 0 && pp->acw == 0 && pp->ace == 0) de++; /* ... and let go of it on last close */ if ((pp->acr + dr) == 0 && (pp->acw + dw) == 0 && (pp->ace + de) == 0) de--; error = ENXIO; LIST_FOREACH(cp1, &gp->consumer, consumer) { error = g_access(cp1, dr, dw, de); if (error == 0) continue; /* * If we fail here, backout all previous changes. */ LIST_FOREACH(cp2, &gp->consumer, consumer) { if (cp1 == cp2) return (error); g_access(cp2, -dr, -dw, -de); } /* NOTREACHED */ } return (error); } static void g_shsec_xor1(uint32_t *src, uint32_t *dst, ssize_t len) { for (; len > 0; len -= sizeof(uint32_t), dst++) *dst = *dst ^ *src++; KASSERT(len == 0, ("len != 0 (len=%zd)", len)); } static void g_shsec_done(struct bio *bp) { struct g_shsec_softc *sc; struct bio *pbp; pbp = bp->bio_parent; sc = pbp->bio_to->geom->softc; if (bp->bio_error == 0) G_SHSEC_LOGREQ(2, bp, "Request done."); else { G_SHSEC_LOGREQ(0, bp, "Request failed (error=%d).", bp->bio_error); if (pbp->bio_error == 0) pbp->bio_error = bp->bio_error; } if (pbp->bio_cmd == BIO_READ) { if ((pbp->bio_pflags & G_SHSEC_BFLAG_FIRST) != 0) { bcopy(bp->bio_data, pbp->bio_data, pbp->bio_length); pbp->bio_pflags = 0; } else { g_shsec_xor1((uint32_t *)bp->bio_data, (uint32_t *)pbp->bio_data, (ssize_t)pbp->bio_length); } } bzero(bp->bio_data, bp->bio_length); uma_zfree(g_shsec_zone, bp->bio_data); g_destroy_bio(bp); pbp->bio_inbed++; if (pbp->bio_children == pbp->bio_inbed) { pbp->bio_completed = pbp->bio_length; g_io_deliver(pbp, pbp->bio_error); } } static void g_shsec_xor2(uint32_t *rand, uint32_t *dst, ssize_t len) { for (; len > 0; len -= sizeof(uint32_t), dst++) { *rand = arc4random(); *dst = *dst ^ *rand++; } KASSERT(len == 0, ("len != 0 (len=%zd)", len)); } static void g_shsec_start(struct bio *bp) { TAILQ_HEAD(, bio) queue = TAILQ_HEAD_INITIALIZER(queue); struct g_shsec_softc *sc; struct bio *cbp; uint32_t *dst; ssize_t len; u_int no; int error; sc = bp->bio_to->geom->softc; /* * If sc == NULL, provider's error should be set and g_shsec_start() * should not be called at all. */ KASSERT(sc != NULL, ("Provider's error should be set (error=%d)(device=%s).", bp->bio_to->error, bp->bio_to->name)); G_SHSEC_LOGREQ(2, bp, "Request received."); switch (bp->bio_cmd) { case BIO_READ: case BIO_WRITE: case BIO_FLUSH: /* * Only those requests are supported. */ break; case BIO_DELETE: case BIO_GETATTR: /* To which provider it should be delivered? */ default: g_io_deliver(bp, EOPNOTSUPP); return; } /* * Allocate all bios first and calculate XOR. */ dst = NULL; len = bp->bio_length; if (bp->bio_cmd == BIO_READ) bp->bio_pflags = G_SHSEC_BFLAG_FIRST; for (no = 0; no < sc->sc_ndisks; no++) { cbp = g_clone_bio(bp); if (cbp == NULL) { error = ENOMEM; goto failure; } TAILQ_INSERT_TAIL(&queue, cbp, bio_queue); /* * Fill in the component buf structure. */ cbp->bio_done = g_shsec_done; cbp->bio_data = uma_zalloc(g_shsec_zone, M_NOWAIT); if (cbp->bio_data == NULL) { g_shsec_alloc_failed++; error = ENOMEM; goto failure; } cbp->bio_caller2 = sc->sc_disks[no]; if (bp->bio_cmd == BIO_WRITE) { if (no == 0) { dst = (uint32_t *)cbp->bio_data; bcopy(bp->bio_data, dst, len); } else { g_shsec_xor2((uint32_t *)cbp->bio_data, dst, len); } } } /* * Fire off all allocated requests! */ while ((cbp = TAILQ_FIRST(&queue)) != NULL) { struct g_consumer *cp; TAILQ_REMOVE(&queue, cbp, bio_queue); cp = cbp->bio_caller2; cbp->bio_caller2 = NULL; cbp->bio_to = cp->provider; G_SHSEC_LOGREQ(2, cbp, "Sending request."); g_io_request(cbp, cp); } return; failure: while ((cbp = TAILQ_FIRST(&queue)) != NULL) { TAILQ_REMOVE(&queue, cbp, bio_queue); bp->bio_children--; if (cbp->bio_data != NULL) { bzero(cbp->bio_data, cbp->bio_length); uma_zfree(g_shsec_zone, cbp->bio_data); } g_destroy_bio(cbp); } if (bp->bio_error == 0) bp->bio_error = error; g_io_deliver(bp, bp->bio_error); } static void g_shsec_check_and_run(struct g_shsec_softc *sc) { off_t mediasize, ms; u_int no, sectorsize = 0; if (g_shsec_nvalid(sc) != sc->sc_ndisks) return; sc->sc_provider = g_new_providerf(sc->sc_geom, "shsec/%s", sc->sc_name); /* * Find the smallest disk. */ mediasize = sc->sc_disks[0]->provider->mediasize; mediasize -= sc->sc_disks[0]->provider->sectorsize; sectorsize = sc->sc_disks[0]->provider->sectorsize; for (no = 1; no < sc->sc_ndisks; no++) { ms = sc->sc_disks[no]->provider->mediasize; ms -= sc->sc_disks[no]->provider->sectorsize; if (ms < mediasize) mediasize = ms; sectorsize = lcm(sectorsize, sc->sc_disks[no]->provider->sectorsize); } sc->sc_provider->sectorsize = sectorsize; sc->sc_provider->mediasize = mediasize; g_error_provider(sc->sc_provider, 0); G_SHSEC_DEBUG(0, "Device %s activated.", sc->sc_name); } static int g_shsec_read_metadata(struct g_consumer *cp, struct g_shsec_metadata *md) { struct g_provider *pp; u_char *buf; int error; g_topology_assert(); error = g_access(cp, 1, 0, 0); if (error != 0) return (error); pp = cp->provider; g_topology_unlock(); buf = g_read_data(cp, pp->mediasize - pp->sectorsize, pp->sectorsize, &error); g_topology_lock(); g_access(cp, -1, 0, 0); if (buf == NULL) return (error); /* Decode metadata. */ shsec_metadata_decode(buf, md); g_free(buf); return (0); } /* * Add disk to given device. */ static int g_shsec_add_disk(struct g_shsec_softc *sc, struct g_provider *pp, u_int no) { struct g_consumer *cp, *fcp; struct g_geom *gp; struct g_shsec_metadata md; int error; /* Metadata corrupted? */ if (no >= sc->sc_ndisks) return (EINVAL); /* Check if disk is not already attached. */ if (sc->sc_disks[no] != NULL) return (EEXIST); gp = sc->sc_geom; fcp = LIST_FIRST(&gp->consumer); cp = g_new_consumer(gp); error = g_attach(cp, pp); if (error != 0) { g_destroy_consumer(cp); return (error); } if (fcp != NULL && (fcp->acr > 0 || fcp->acw > 0 || fcp->ace > 0)) { error = g_access(cp, fcp->acr, fcp->acw, fcp->ace); if (error != 0) { g_detach(cp); g_destroy_consumer(cp); return (error); } } /* Reread metadata. */ error = g_shsec_read_metadata(cp, &md); if (error != 0) goto fail; if (strcmp(md.md_magic, G_SHSEC_MAGIC) != 0 || strcmp(md.md_name, sc->sc_name) != 0 || md.md_id != sc->sc_id) { G_SHSEC_DEBUG(0, "Metadata on %s changed.", pp->name); goto fail; } cp->private = sc; cp->index = no; sc->sc_disks[no] = cp; G_SHSEC_DEBUG(0, "Disk %s attached to %s.", pp->name, sc->sc_name); g_shsec_check_and_run(sc); return (0); fail: if (fcp != NULL && (fcp->acr > 0 || fcp->acw > 0 || fcp->ace > 0)) g_access(cp, -fcp->acr, -fcp->acw, -fcp->ace); g_detach(cp); g_destroy_consumer(cp); return (error); } static struct g_geom * g_shsec_create(struct g_class *mp, const struct g_shsec_metadata *md) { struct g_shsec_softc *sc; struct g_geom *gp; u_int no; G_SHSEC_DEBUG(1, "Creating device %s (id=%u).", md->md_name, md->md_id); /* Two disks is minimum. */ if (md->md_all < 2) { G_SHSEC_DEBUG(0, "Too few disks defined for %s.", md->md_name); return (NULL); } /* Check for duplicate unit */ LIST_FOREACH(gp, &mp->geom, geom) { sc = gp->softc; if (sc != NULL && strcmp(sc->sc_name, md->md_name) == 0) { G_SHSEC_DEBUG(0, "Device %s already configured.", sc->sc_name); return (NULL); } } gp = g_new_geomf(mp, "%s", md->md_name); - gp->softc = NULL; /* for a moment */ - sc = malloc(sizeof(*sc), M_SHSEC, M_WAITOK | M_ZERO); gp->start = g_shsec_start; gp->spoiled = g_shsec_orphan; gp->orphan = g_shsec_orphan; gp->access = g_shsec_access; gp->dumpconf = g_shsec_dumpconf; sc->sc_id = md->md_id; sc->sc_ndisks = md->md_all; sc->sc_disks = malloc(sizeof(struct g_consumer *) * sc->sc_ndisks, M_SHSEC, M_WAITOK | M_ZERO); for (no = 0; no < sc->sc_ndisks; no++) sc->sc_disks[no] = NULL; gp->softc = sc; sc->sc_geom = gp; sc->sc_provider = NULL; G_SHSEC_DEBUG(0, "Device %s created (id=%u).", sc->sc_name, sc->sc_id); return (gp); } static int g_shsec_destroy(struct g_shsec_softc *sc, boolean_t force) { struct g_provider *pp; struct g_geom *gp; u_int no; g_topology_assert(); if (sc == NULL) return (ENXIO); pp = sc->sc_provider; if (pp != NULL && (pp->acr != 0 || pp->acw != 0 || pp->ace != 0)) { if (force) { G_SHSEC_DEBUG(0, "Device %s is still open, so it " "can't be definitely removed.", pp->name); } else { G_SHSEC_DEBUG(1, "Device %s is still open (r%dw%de%d).", pp->name, pp->acr, pp->acw, pp->ace); return (EBUSY); } } for (no = 0; no < sc->sc_ndisks; no++) { if (sc->sc_disks[no] != NULL) g_shsec_remove_disk(sc->sc_disks[no]); } gp = sc->sc_geom; gp->softc = NULL; KASSERT(sc->sc_provider == NULL, ("Provider still exists? (device=%s)", gp->name)); free(sc->sc_disks, M_SHSEC); free(sc, M_SHSEC); pp = LIST_FIRST(&gp->provider); if (pp == NULL || (pp->acr == 0 && pp->acw == 0 && pp->ace == 0)) G_SHSEC_DEBUG(0, "Device %s destroyed.", gp->name); g_wither_geom(gp, ENXIO); return (0); } static int g_shsec_destroy_geom(struct gctl_req *req __unused, struct g_class *mp __unused, struct g_geom *gp) { struct g_shsec_softc *sc; sc = gp->softc; return (g_shsec_destroy(sc, 0)); } static struct g_geom * g_shsec_taste(struct g_class *mp, struct g_provider *pp, int flags __unused) { struct g_shsec_metadata md; struct g_shsec_softc *sc; struct g_consumer *cp; struct g_geom *gp; int error; g_trace(G_T_TOPOLOGY, "%s(%s, %s)", __func__, mp->name, pp->name); g_topology_assert(); /* Skip providers that are already open for writing. */ if (pp->acw > 0) return (NULL); G_SHSEC_DEBUG(3, "Tasting %s.", pp->name); gp = g_new_geomf(mp, "shsec:taste"); gp->start = g_shsec_start; gp->access = g_shsec_access; gp->orphan = g_shsec_orphan; cp = g_new_consumer(gp); g_attach(cp, pp); error = g_shsec_read_metadata(cp, &md); g_detach(cp); g_destroy_consumer(cp); g_destroy_geom(gp); if (error != 0) return (NULL); gp = NULL; if (strcmp(md.md_magic, G_SHSEC_MAGIC) != 0) return (NULL); if (md.md_version > G_SHSEC_VERSION) { G_SHSEC_DEBUG(0, "Kernel module is too old to handle %s.\n", pp->name); return (NULL); } /* * Backward compatibility: */ /* There was no md_provsize field in earlier versions of metadata. */ if (md.md_version < 1) md.md_provsize = pp->mediasize; if (md.md_provider[0] != '\0' && !g_compare_names(md.md_provider, pp->name)) return (NULL); if (md.md_provsize != pp->mediasize) return (NULL); /* * Let's check if device already exists. */ sc = NULL; LIST_FOREACH(gp, &mp->geom, geom) { sc = gp->softc; if (sc == NULL) continue; if (strcmp(md.md_name, sc->sc_name) != 0) continue; if (md.md_id != sc->sc_id) continue; break; } if (gp != NULL) { G_SHSEC_DEBUG(1, "Adding disk %s to %s.", pp->name, gp->name); error = g_shsec_add_disk(sc, pp, md.md_no); if (error != 0) { G_SHSEC_DEBUG(0, "Cannot add disk %s to %s (error=%d).", pp->name, gp->name, error); return (NULL); } } else { gp = g_shsec_create(mp, &md); if (gp == NULL) { G_SHSEC_DEBUG(0, "Cannot create device %s.", md.md_name); return (NULL); } sc = gp->softc; G_SHSEC_DEBUG(1, "Adding disk %s to %s.", pp->name, gp->name); error = g_shsec_add_disk(sc, pp, md.md_no); if (error != 0) { G_SHSEC_DEBUG(0, "Cannot add disk %s to %s (error=%d).", pp->name, gp->name, error); g_shsec_destroy(sc, 1); return (NULL); } } return (gp); } static struct g_shsec_softc * g_shsec_find_device(struct g_class *mp, const char *name) { struct g_shsec_softc *sc; struct g_geom *gp; LIST_FOREACH(gp, &mp->geom, geom) { sc = gp->softc; if (sc == NULL) continue; if (strcmp(sc->sc_name, name) == 0) return (sc); } return (NULL); } static void g_shsec_ctl_destroy(struct gctl_req *req, struct g_class *mp) { struct g_shsec_softc *sc; int *force, *nargs, error; const char *name; char param[16]; u_int i; g_topology_assert(); nargs = gctl_get_paraml(req, "nargs", sizeof(*nargs)); if (nargs == NULL) { gctl_error(req, "No '%s' argument.", "nargs"); return; } if (*nargs <= 0) { gctl_error(req, "Missing device(s)."); return; } force = gctl_get_paraml(req, "force", sizeof(*force)); if (force == NULL) { gctl_error(req, "No '%s' argument.", "force"); return; } for (i = 0; i < (u_int)*nargs; i++) { snprintf(param, sizeof(param), "arg%u", i); name = gctl_get_asciiparam(req, param); if (name == NULL) { gctl_error(req, "No 'arg%u' argument.", i); return; } sc = g_shsec_find_device(mp, name); if (sc == NULL) { gctl_error(req, "No such device: %s.", name); return; } error = g_shsec_destroy(sc, *force); if (error != 0) { gctl_error(req, "Cannot destroy device %s (error=%d).", sc->sc_name, error); return; } } } static void g_shsec_config(struct gctl_req *req, struct g_class *mp, const char *verb) { uint32_t *version; g_topology_assert(); version = gctl_get_paraml(req, "version", sizeof(*version)); if (version == NULL) { gctl_error(req, "No '%s' argument.", "version"); return; } if (*version != G_SHSEC_VERSION) { gctl_error(req, "Userland and kernel parts are out of sync."); return; } if (strcmp(verb, "stop") == 0) { g_shsec_ctl_destroy(req, mp); return; } gctl_error(req, "Unknown verb."); } static void g_shsec_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp) { struct g_shsec_softc *sc; sc = gp->softc; if (sc == NULL) return; if (pp != NULL) { /* Nothing here. */ } else if (cp != NULL) { sbuf_printf(sb, "%s%u\n", indent, (u_int)cp->index); } else { sbuf_printf(sb, "%s%u\n", indent, (u_int)sc->sc_id); sbuf_printf(sb, "%sTotal=%u, Online=%u\n", indent, sc->sc_ndisks, g_shsec_nvalid(sc)); sbuf_printf(sb, "%s", indent); if (sc->sc_provider != NULL && sc->sc_provider->error == 0) sbuf_printf(sb, "UP"); else sbuf_printf(sb, "DOWN"); sbuf_printf(sb, "\n"); } } DECLARE_GEOM_CLASS(g_shsec_class, g_shsec); Index: projects/largeSMP/sys/geom/stripe/g_stripe.c =================================================================== --- projects/largeSMP/sys/geom/stripe/g_stripe.c (revision 221494) +++ projects/largeSMP/sys/geom/stripe/g_stripe.c (revision 221495) @@ -1,1242 +1,1240 @@ /*- * Copyright (c) 2004-2005 Pawel Jakub Dawidek * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include FEATURE(geom_stripe, "GEOM striping support"); static MALLOC_DEFINE(M_STRIPE, "stripe_data", "GEOM_STRIPE Data"); static uma_zone_t g_stripe_zone; static int g_stripe_destroy(struct g_stripe_softc *sc, boolean_t force); static int g_stripe_destroy_geom(struct gctl_req *req, struct g_class *mp, struct g_geom *gp); static g_taste_t g_stripe_taste; static g_ctl_req_t g_stripe_config; static g_dumpconf_t g_stripe_dumpconf; static g_init_t g_stripe_init; static g_fini_t g_stripe_fini; struct g_class g_stripe_class = { .name = G_STRIPE_CLASS_NAME, .version = G_VERSION, .ctlreq = g_stripe_config, .taste = g_stripe_taste, .destroy_geom = g_stripe_destroy_geom, .init = g_stripe_init, .fini = g_stripe_fini }; SYSCTL_DECL(_kern_geom); SYSCTL_NODE(_kern_geom, OID_AUTO, stripe, CTLFLAG_RW, 0, "GEOM_STRIPE stuff"); static u_int g_stripe_debug = 0; TUNABLE_INT("kern.geom.stripe.debug", &g_stripe_debug); SYSCTL_UINT(_kern_geom_stripe, OID_AUTO, debug, CTLFLAG_RW, &g_stripe_debug, 0, "Debug level"); static int g_stripe_fast = 0; TUNABLE_INT("kern.geom.stripe.fast", &g_stripe_fast); static int g_sysctl_stripe_fast(SYSCTL_HANDLER_ARGS) { int error, fast; fast = g_stripe_fast; error = sysctl_handle_int(oidp, &fast, 0, req); if (error == 0 && req->newptr != NULL) g_stripe_fast = fast; return (error); } SYSCTL_PROC(_kern_geom_stripe, OID_AUTO, fast, CTLTYPE_INT | CTLFLAG_RW, NULL, 0, g_sysctl_stripe_fast, "I", "Fast, but memory-consuming, mode"); static u_int g_stripe_maxmem = MAXPHYS * 100; TUNABLE_INT("kern.geom.stripe.maxmem", &g_stripe_maxmem); SYSCTL_UINT(_kern_geom_stripe, OID_AUTO, maxmem, CTLFLAG_RD, &g_stripe_maxmem, 0, "Maximum memory that can be allocated in \"fast\" mode (in bytes)"); static u_int g_stripe_fast_failed = 0; SYSCTL_UINT(_kern_geom_stripe, OID_AUTO, fast_failed, CTLFLAG_RD, &g_stripe_fast_failed, 0, "How many times \"fast\" mode failed"); /* * Greatest Common Divisor. */ static u_int gcd(u_int a, u_int b) { u_int c; while (b != 0) { c = a; a = b; b = (c % b); } return (a); } /* * Least Common Multiple. */ static u_int lcm(u_int a, u_int b) { return ((a * b) / gcd(a, b)); } static void g_stripe_init(struct g_class *mp __unused) { g_stripe_zone = uma_zcreate("g_stripe_zone", MAXPHYS, NULL, NULL, NULL, NULL, 0, 0); g_stripe_maxmem -= g_stripe_maxmem % MAXPHYS; uma_zone_set_max(g_stripe_zone, g_stripe_maxmem / MAXPHYS); } static void g_stripe_fini(struct g_class *mp __unused) { uma_zdestroy(g_stripe_zone); } /* * Return the number of valid disks. */ static u_int g_stripe_nvalid(struct g_stripe_softc *sc) { u_int i, no; no = 0; for (i = 0; i < sc->sc_ndisks; i++) { if (sc->sc_disks[i] != NULL) no++; } return (no); } static void g_stripe_remove_disk(struct g_consumer *cp) { struct g_stripe_softc *sc; u_int no; KASSERT(cp != NULL, ("Non-valid disk in %s.", __func__)); sc = (struct g_stripe_softc *)cp->private; KASSERT(sc != NULL, ("NULL sc in %s.", __func__)); no = cp->index; G_STRIPE_DEBUG(0, "Disk %s removed from %s.", cp->provider->name, sc->sc_name); sc->sc_disks[no] = NULL; if (sc->sc_provider != NULL) { sc->sc_provider->flags |= G_PF_WITHER; g_orphan_provider(sc->sc_provider, ENXIO); sc->sc_provider = NULL; G_STRIPE_DEBUG(0, "Device %s removed.", sc->sc_name); } if (cp->acr > 0 || cp->acw > 0 || cp->ace > 0) g_access(cp, -cp->acr, -cp->acw, -cp->ace); g_detach(cp); g_destroy_consumer(cp); } static void g_stripe_orphan(struct g_consumer *cp) { struct g_stripe_softc *sc; struct g_geom *gp; g_topology_assert(); gp = cp->geom; sc = gp->softc; if (sc == NULL) return; g_stripe_remove_disk(cp); /* If there are no valid disks anymore, remove device. */ if (g_stripe_nvalid(sc) == 0) g_stripe_destroy(sc, 1); } static int g_stripe_access(struct g_provider *pp, int dr, int dw, int de) { struct g_consumer *cp1, *cp2; struct g_stripe_softc *sc; struct g_geom *gp; int error; gp = pp->geom; sc = gp->softc; if (sc == NULL) { /* * It looks like geom is being withered. * In that case we allow only negative requests. */ KASSERT(dr <= 0 && dw <= 0 && de <= 0, ("Positive access request (device=%s).", pp->name)); if ((pp->acr + dr) == 0 && (pp->acw + dw) == 0 && (pp->ace + de) == 0) { G_STRIPE_DEBUG(0, "Device %s definitely destroyed.", gp->name); } return (0); } /* On first open, grab an extra "exclusive" bit */ if (pp->acr == 0 && pp->acw == 0 && pp->ace == 0) de++; /* ... and let go of it on last close */ if ((pp->acr + dr) == 0 && (pp->acw + dw) == 0 && (pp->ace + de) == 0) de--; error = ENXIO; LIST_FOREACH(cp1, &gp->consumer, consumer) { error = g_access(cp1, dr, dw, de); if (error == 0) continue; /* * If we fail here, backout all previous changes. */ LIST_FOREACH(cp2, &gp->consumer, consumer) { if (cp1 == cp2) return (error); g_access(cp2, -dr, -dw, -de); } /* NOTREACHED */ } return (error); } static void g_stripe_copy(struct g_stripe_softc *sc, char *src, char *dst, off_t offset, off_t length, int mode) { u_int stripesize; size_t len; stripesize = sc->sc_stripesize; len = (size_t)(stripesize - (offset & (stripesize - 1))); do { bcopy(src, dst, len); if (mode) { dst += len + stripesize * (sc->sc_ndisks - 1); src += len; } else { dst += len; src += len + stripesize * (sc->sc_ndisks - 1); } length -= len; KASSERT(length >= 0, ("Length < 0 (stripesize=%zu, offset=%jd, length=%jd).", (size_t)stripesize, (intmax_t)offset, (intmax_t)length)); if (length > stripesize) len = stripesize; else len = length; } while (length > 0); } static void g_stripe_done(struct bio *bp) { struct g_stripe_softc *sc; struct bio *pbp; pbp = bp->bio_parent; sc = pbp->bio_to->geom->softc; if (pbp->bio_error == 0) pbp->bio_error = bp->bio_error; pbp->bio_completed += bp->bio_completed; if (bp->bio_cmd == BIO_READ && bp->bio_caller1 != NULL) { g_stripe_copy(sc, bp->bio_data, bp->bio_caller1, bp->bio_offset, bp->bio_length, 1); bp->bio_data = bp->bio_caller1; bp->bio_caller1 = NULL; } g_destroy_bio(bp); pbp->bio_inbed++; if (pbp->bio_children == pbp->bio_inbed) { if (pbp->bio_driver1 != NULL) uma_zfree(g_stripe_zone, pbp->bio_driver1); g_io_deliver(pbp, pbp->bio_error); } } static int g_stripe_start_fast(struct bio *bp, u_int no, off_t offset, off_t length) { TAILQ_HEAD(, bio) queue = TAILQ_HEAD_INITIALIZER(queue); u_int nparts = 0, stripesize; struct g_stripe_softc *sc; char *addr, *data = NULL; struct bio *cbp; int error; sc = bp->bio_to->geom->softc; addr = bp->bio_data; stripesize = sc->sc_stripesize; cbp = g_clone_bio(bp); if (cbp == NULL) { error = ENOMEM; goto failure; } TAILQ_INSERT_TAIL(&queue, cbp, bio_queue); nparts++; /* * Fill in the component buf structure. */ cbp->bio_done = g_stripe_done; cbp->bio_offset = offset; cbp->bio_data = addr; cbp->bio_caller1 = NULL; cbp->bio_length = length; cbp->bio_caller2 = sc->sc_disks[no]; /* offset -= offset % stripesize; */ offset -= offset & (stripesize - 1); addr += length; length = bp->bio_length - length; for (no++; length > 0; no++, length -= stripesize, addr += stripesize) { if (no > sc->sc_ndisks - 1) { no = 0; offset += stripesize; } if (nparts >= sc->sc_ndisks) { cbp = TAILQ_NEXT(cbp, bio_queue); if (cbp == NULL) cbp = TAILQ_FIRST(&queue); nparts++; /* * Update bio structure. */ /* * MIN() is in case when * (bp->bio_length % sc->sc_stripesize) != 0. */ cbp->bio_length += MIN(stripesize, length); if (cbp->bio_caller1 == NULL) { cbp->bio_caller1 = cbp->bio_data; cbp->bio_data = NULL; if (data == NULL) { data = uma_zalloc(g_stripe_zone, M_NOWAIT); if (data == NULL) { error = ENOMEM; goto failure; } } } } else { cbp = g_clone_bio(bp); if (cbp == NULL) { error = ENOMEM; goto failure; } TAILQ_INSERT_TAIL(&queue, cbp, bio_queue); nparts++; /* * Fill in the component buf structure. */ cbp->bio_done = g_stripe_done; cbp->bio_offset = offset; cbp->bio_data = addr; cbp->bio_caller1 = NULL; /* * MIN() is in case when * (bp->bio_length % sc->sc_stripesize) != 0. */ cbp->bio_length = MIN(stripesize, length); cbp->bio_caller2 = sc->sc_disks[no]; } } if (data != NULL) bp->bio_driver1 = data; /* * Fire off all allocated requests! */ while ((cbp = TAILQ_FIRST(&queue)) != NULL) { struct g_consumer *cp; TAILQ_REMOVE(&queue, cbp, bio_queue); cp = cbp->bio_caller2; cbp->bio_caller2 = NULL; cbp->bio_to = cp->provider; if (cbp->bio_caller1 != NULL) { cbp->bio_data = data; if (bp->bio_cmd == BIO_WRITE) { g_stripe_copy(sc, cbp->bio_caller1, data, cbp->bio_offset, cbp->bio_length, 0); } data += cbp->bio_length; } G_STRIPE_LOGREQ(cbp, "Sending request."); g_io_request(cbp, cp); } return (0); failure: if (data != NULL) uma_zfree(g_stripe_zone, data); while ((cbp = TAILQ_FIRST(&queue)) != NULL) { TAILQ_REMOVE(&queue, cbp, bio_queue); if (cbp->bio_caller1 != NULL) { cbp->bio_data = cbp->bio_caller1; cbp->bio_caller1 = NULL; } bp->bio_children--; g_destroy_bio(cbp); } return (error); } static int g_stripe_start_economic(struct bio *bp, u_int no, off_t offset, off_t length) { TAILQ_HEAD(, bio) queue = TAILQ_HEAD_INITIALIZER(queue); struct g_stripe_softc *sc; uint32_t stripesize; struct bio *cbp; char *addr; int error; sc = bp->bio_to->geom->softc; addr = bp->bio_data; stripesize = sc->sc_stripesize; cbp = g_clone_bio(bp); if (cbp == NULL) { error = ENOMEM; goto failure; } TAILQ_INSERT_TAIL(&queue, cbp, bio_queue); /* * Fill in the component buf structure. */ cbp->bio_done = g_std_done; cbp->bio_offset = offset; cbp->bio_data = addr; cbp->bio_length = length; cbp->bio_caller2 = sc->sc_disks[no]; /* offset -= offset % stripesize; */ offset -= offset & (stripesize - 1); addr += length; length = bp->bio_length - length; for (no++; length > 0; no++, length -= stripesize, addr += stripesize) { if (no > sc->sc_ndisks - 1) { no = 0; offset += stripesize; } cbp = g_clone_bio(bp); if (cbp == NULL) { error = ENOMEM; goto failure; } TAILQ_INSERT_TAIL(&queue, cbp, bio_queue); /* * Fill in the component buf structure. */ cbp->bio_done = g_std_done; cbp->bio_offset = offset; cbp->bio_data = addr; /* * MIN() is in case when * (bp->bio_length % sc->sc_stripesize) != 0. */ cbp->bio_length = MIN(stripesize, length); cbp->bio_caller2 = sc->sc_disks[no]; } /* * Fire off all allocated requests! */ while ((cbp = TAILQ_FIRST(&queue)) != NULL) { struct g_consumer *cp; TAILQ_REMOVE(&queue, cbp, bio_queue); cp = cbp->bio_caller2; cbp->bio_caller2 = NULL; cbp->bio_to = cp->provider; G_STRIPE_LOGREQ(cbp, "Sending request."); g_io_request(cbp, cp); } return (0); failure: while ((cbp = TAILQ_FIRST(&queue)) != NULL) { TAILQ_REMOVE(&queue, cbp, bio_queue); bp->bio_children--; g_destroy_bio(cbp); } return (error); } static void g_stripe_flush(struct g_stripe_softc *sc, struct bio *bp) { struct bio_queue_head queue; struct g_consumer *cp; struct bio *cbp; u_int no; bioq_init(&queue); for (no = 0; no < sc->sc_ndisks; no++) { cbp = g_clone_bio(bp); if (cbp == NULL) { for (cbp = bioq_first(&queue); cbp != NULL; cbp = bioq_first(&queue)) { bioq_remove(&queue, cbp); g_destroy_bio(cbp); } if (bp->bio_error == 0) bp->bio_error = ENOMEM; g_io_deliver(bp, bp->bio_error); return; } bioq_insert_tail(&queue, cbp); cbp->bio_done = g_std_done; cbp->bio_caller1 = sc->sc_disks[no]; cbp->bio_to = sc->sc_disks[no]->provider; } for (cbp = bioq_first(&queue); cbp != NULL; cbp = bioq_first(&queue)) { bioq_remove(&queue, cbp); G_STRIPE_LOGREQ(cbp, "Sending request."); cp = cbp->bio_caller1; cbp->bio_caller1 = NULL; g_io_request(cbp, cp); } } static void g_stripe_start(struct bio *bp) { off_t offset, start, length, nstripe; struct g_stripe_softc *sc; u_int no, stripesize; int error, fast = 0; sc = bp->bio_to->geom->softc; /* * If sc == NULL, provider's error should be set and g_stripe_start() * should not be called at all. */ KASSERT(sc != NULL, ("Provider's error should be set (error=%d)(device=%s).", bp->bio_to->error, bp->bio_to->name)); G_STRIPE_LOGREQ(bp, "Request received."); switch (bp->bio_cmd) { case BIO_READ: case BIO_WRITE: case BIO_DELETE: break; case BIO_FLUSH: g_stripe_flush(sc, bp); return; case BIO_GETATTR: /* To which provider it should be delivered? */ default: g_io_deliver(bp, EOPNOTSUPP); return; } stripesize = sc->sc_stripesize; /* * Calculations are quite messy, but fast I hope. */ /* Stripe number. */ /* nstripe = bp->bio_offset / stripesize; */ nstripe = bp->bio_offset >> (off_t)sc->sc_stripebits; /* Disk number. */ no = nstripe % sc->sc_ndisks; /* Start position in stripe. */ /* start = bp->bio_offset % stripesize; */ start = bp->bio_offset & (stripesize - 1); /* Start position in disk. */ /* offset = (nstripe / sc->sc_ndisks) * stripesize + start; */ offset = ((nstripe / sc->sc_ndisks) << sc->sc_stripebits) + start; /* Length of data to operate. */ length = MIN(bp->bio_length, stripesize - start); /* * Do use "fast" mode when: * 1. "Fast" mode is ON. * and * 2. Request size is less than or equal to MAXPHYS, * which should always be true. * and * 3. Request size is bigger than stripesize * ndisks. If it isn't, * there will be no need to send more than one I/O request to * a provider, so there is nothing to optmize. */ if (g_stripe_fast && bp->bio_length <= MAXPHYS && bp->bio_length >= stripesize * sc->sc_ndisks) { fast = 1; } error = 0; if (fast) { error = g_stripe_start_fast(bp, no, offset, length); if (error != 0) g_stripe_fast_failed++; } /* * Do use "economic" when: * 1. "Economic" mode is ON. * or * 2. "Fast" mode failed. It can only fail if there is no memory. */ if (!fast || error != 0) error = g_stripe_start_economic(bp, no, offset, length); if (error != 0) { if (bp->bio_error == 0) bp->bio_error = error; g_io_deliver(bp, bp->bio_error); } } static void g_stripe_check_and_run(struct g_stripe_softc *sc) { off_t mediasize, ms; u_int no, sectorsize = 0; if (g_stripe_nvalid(sc) != sc->sc_ndisks) return; sc->sc_provider = g_new_providerf(sc->sc_geom, "stripe/%s", sc->sc_name); /* * Find the smallest disk. */ mediasize = sc->sc_disks[0]->provider->mediasize; if (sc->sc_type == G_STRIPE_TYPE_AUTOMATIC) mediasize -= sc->sc_disks[0]->provider->sectorsize; mediasize -= mediasize % sc->sc_stripesize; sectorsize = sc->sc_disks[0]->provider->sectorsize; for (no = 1; no < sc->sc_ndisks; no++) { ms = sc->sc_disks[no]->provider->mediasize; if (sc->sc_type == G_STRIPE_TYPE_AUTOMATIC) ms -= sc->sc_disks[no]->provider->sectorsize; ms -= ms % sc->sc_stripesize; if (ms < mediasize) mediasize = ms; sectorsize = lcm(sectorsize, sc->sc_disks[no]->provider->sectorsize); } sc->sc_provider->sectorsize = sectorsize; sc->sc_provider->mediasize = mediasize * sc->sc_ndisks; sc->sc_provider->stripesize = sc->sc_stripesize; sc->sc_provider->stripeoffset = 0; g_error_provider(sc->sc_provider, 0); G_STRIPE_DEBUG(0, "Device %s activated.", sc->sc_name); } static int g_stripe_read_metadata(struct g_consumer *cp, struct g_stripe_metadata *md) { struct g_provider *pp; u_char *buf; int error; g_topology_assert(); error = g_access(cp, 1, 0, 0); if (error != 0) return (error); pp = cp->provider; g_topology_unlock(); buf = g_read_data(cp, pp->mediasize - pp->sectorsize, pp->sectorsize, &error); g_topology_lock(); g_access(cp, -1, 0, 0); if (buf == NULL) return (error); /* Decode metadata. */ stripe_metadata_decode(buf, md); g_free(buf); return (0); } /* * Add disk to given device. */ static int g_stripe_add_disk(struct g_stripe_softc *sc, struct g_provider *pp, u_int no) { struct g_consumer *cp, *fcp; struct g_geom *gp; int error; /* Metadata corrupted? */ if (no >= sc->sc_ndisks) return (EINVAL); /* Check if disk is not already attached. */ if (sc->sc_disks[no] != NULL) return (EEXIST); gp = sc->sc_geom; fcp = LIST_FIRST(&gp->consumer); cp = g_new_consumer(gp); error = g_attach(cp, pp); if (error != 0) { g_destroy_consumer(cp); return (error); } if (fcp != NULL && (fcp->acr > 0 || fcp->acw > 0 || fcp->ace > 0)) { error = g_access(cp, fcp->acr, fcp->acw, fcp->ace); if (error != 0) { g_detach(cp); g_destroy_consumer(cp); return (error); } } if (sc->sc_type == G_STRIPE_TYPE_AUTOMATIC) { struct g_stripe_metadata md; /* Reread metadata. */ error = g_stripe_read_metadata(cp, &md); if (error != 0) goto fail; if (strcmp(md.md_magic, G_STRIPE_MAGIC) != 0 || strcmp(md.md_name, sc->sc_name) != 0 || md.md_id != sc->sc_id) { G_STRIPE_DEBUG(0, "Metadata on %s changed.", pp->name); goto fail; } } cp->private = sc; cp->index = no; sc->sc_disks[no] = cp; G_STRIPE_DEBUG(0, "Disk %s attached to %s.", pp->name, sc->sc_name); g_stripe_check_and_run(sc); return (0); fail: if (fcp != NULL && (fcp->acr > 0 || fcp->acw > 0 || fcp->ace > 0)) g_access(cp, -fcp->acr, -fcp->acw, -fcp->ace); g_detach(cp); g_destroy_consumer(cp); return (error); } static struct g_geom * g_stripe_create(struct g_class *mp, const struct g_stripe_metadata *md, u_int type) { struct g_stripe_softc *sc; struct g_geom *gp; u_int no; G_STRIPE_DEBUG(1, "Creating device %s (id=%u).", md->md_name, md->md_id); /* Two disks is minimum. */ if (md->md_all < 2) { G_STRIPE_DEBUG(0, "Too few disks defined for %s.", md->md_name); return (NULL); } #if 0 /* Stripe size have to be grater than or equal to sector size. */ if (md->md_stripesize < sectorsize) { G_STRIPE_DEBUG(0, "Invalid stripe size for %s.", md->md_name); return (NULL); } #endif /* Stripe size have to be power of 2. */ if (!powerof2(md->md_stripesize)) { G_STRIPE_DEBUG(0, "Invalid stripe size for %s.", md->md_name); return (NULL); } /* Check for duplicate unit */ LIST_FOREACH(gp, &mp->geom, geom) { sc = gp->softc; if (sc != NULL && strcmp(sc->sc_name, md->md_name) == 0) { G_STRIPE_DEBUG(0, "Device %s already configured.", sc->sc_name); return (NULL); } } gp = g_new_geomf(mp, "%s", md->md_name); - gp->softc = NULL; /* for a moment */ - sc = malloc(sizeof(*sc), M_STRIPE, M_WAITOK | M_ZERO); gp->start = g_stripe_start; gp->spoiled = g_stripe_orphan; gp->orphan = g_stripe_orphan; gp->access = g_stripe_access; gp->dumpconf = g_stripe_dumpconf; sc->sc_id = md->md_id; sc->sc_stripesize = md->md_stripesize; sc->sc_stripebits = bitcount32(sc->sc_stripesize - 1); sc->sc_ndisks = md->md_all; sc->sc_disks = malloc(sizeof(struct g_consumer *) * sc->sc_ndisks, M_STRIPE, M_WAITOK | M_ZERO); for (no = 0; no < sc->sc_ndisks; no++) sc->sc_disks[no] = NULL; sc->sc_type = type; gp->softc = sc; sc->sc_geom = gp; sc->sc_provider = NULL; G_STRIPE_DEBUG(0, "Device %s created (id=%u).", sc->sc_name, sc->sc_id); return (gp); } static int g_stripe_destroy(struct g_stripe_softc *sc, boolean_t force) { struct g_provider *pp; struct g_geom *gp; u_int no; g_topology_assert(); if (sc == NULL) return (ENXIO); pp = sc->sc_provider; if (pp != NULL && (pp->acr != 0 || pp->acw != 0 || pp->ace != 0)) { if (force) { G_STRIPE_DEBUG(0, "Device %s is still open, so it " "can't be definitely removed.", pp->name); } else { G_STRIPE_DEBUG(1, "Device %s is still open (r%dw%de%d).", pp->name, pp->acr, pp->acw, pp->ace); return (EBUSY); } } for (no = 0; no < sc->sc_ndisks; no++) { if (sc->sc_disks[no] != NULL) g_stripe_remove_disk(sc->sc_disks[no]); } gp = sc->sc_geom; gp->softc = NULL; KASSERT(sc->sc_provider == NULL, ("Provider still exists? (device=%s)", gp->name)); free(sc->sc_disks, M_STRIPE); free(sc, M_STRIPE); pp = LIST_FIRST(&gp->provider); if (pp == NULL || (pp->acr == 0 && pp->acw == 0 && pp->ace == 0)) G_STRIPE_DEBUG(0, "Device %s destroyed.", gp->name); g_wither_geom(gp, ENXIO); return (0); } static int g_stripe_destroy_geom(struct gctl_req *req __unused, struct g_class *mp __unused, struct g_geom *gp) { struct g_stripe_softc *sc; sc = gp->softc; return (g_stripe_destroy(sc, 0)); } static struct g_geom * g_stripe_taste(struct g_class *mp, struct g_provider *pp, int flags __unused) { struct g_stripe_metadata md; struct g_stripe_softc *sc; struct g_consumer *cp; struct g_geom *gp; int error; g_trace(G_T_TOPOLOGY, "%s(%s, %s)", __func__, mp->name, pp->name); g_topology_assert(); /* Skip providers that are already open for writing. */ if (pp->acw > 0) return (NULL); G_STRIPE_DEBUG(3, "Tasting %s.", pp->name); gp = g_new_geomf(mp, "stripe:taste"); gp->start = g_stripe_start; gp->access = g_stripe_access; gp->orphan = g_stripe_orphan; cp = g_new_consumer(gp); g_attach(cp, pp); error = g_stripe_read_metadata(cp, &md); g_detach(cp); g_destroy_consumer(cp); g_destroy_geom(gp); if (error != 0) return (NULL); gp = NULL; if (strcmp(md.md_magic, G_STRIPE_MAGIC) != 0) return (NULL); if (md.md_version > G_STRIPE_VERSION) { printf("geom_stripe.ko module is too old to handle %s.\n", pp->name); return (NULL); } /* * Backward compatibility: */ /* There was no md_provider field in earlier versions of metadata. */ if (md.md_version < 2) bzero(md.md_provider, sizeof(md.md_provider)); /* There was no md_provsize field in earlier versions of metadata. */ if (md.md_version < 3) md.md_provsize = pp->mediasize; if (md.md_provider[0] != '\0' && !g_compare_names(md.md_provider, pp->name)) return (NULL); if (md.md_provsize != pp->mediasize) return (NULL); /* * Let's check if device already exists. */ sc = NULL; LIST_FOREACH(gp, &mp->geom, geom) { sc = gp->softc; if (sc == NULL) continue; if (sc->sc_type != G_STRIPE_TYPE_AUTOMATIC) continue; if (strcmp(md.md_name, sc->sc_name) != 0) continue; if (md.md_id != sc->sc_id) continue; break; } if (gp != NULL) { G_STRIPE_DEBUG(1, "Adding disk %s to %s.", pp->name, gp->name); error = g_stripe_add_disk(sc, pp, md.md_no); if (error != 0) { G_STRIPE_DEBUG(0, "Cannot add disk %s to %s (error=%d).", pp->name, gp->name, error); return (NULL); } } else { gp = g_stripe_create(mp, &md, G_STRIPE_TYPE_AUTOMATIC); if (gp == NULL) { G_STRIPE_DEBUG(0, "Cannot create device %s.", md.md_name); return (NULL); } sc = gp->softc; G_STRIPE_DEBUG(1, "Adding disk %s to %s.", pp->name, gp->name); error = g_stripe_add_disk(sc, pp, md.md_no); if (error != 0) { G_STRIPE_DEBUG(0, "Cannot add disk %s to %s (error=%d).", pp->name, gp->name, error); g_stripe_destroy(sc, 1); return (NULL); } } return (gp); } static void g_stripe_ctl_create(struct gctl_req *req, struct g_class *mp) { u_int attached, no; struct g_stripe_metadata md; struct g_provider *pp; struct g_stripe_softc *sc; struct g_geom *gp; struct sbuf *sb; intmax_t *stripesize; const char *name; char param[16]; int *nargs; g_topology_assert(); nargs = gctl_get_paraml(req, "nargs", sizeof(*nargs)); if (nargs == NULL) { gctl_error(req, "No '%s' argument.", "nargs"); return; } if (*nargs <= 2) { gctl_error(req, "Too few arguments."); return; } strlcpy(md.md_magic, G_STRIPE_MAGIC, sizeof(md.md_magic)); md.md_version = G_STRIPE_VERSION; name = gctl_get_asciiparam(req, "arg0"); if (name == NULL) { gctl_error(req, "No 'arg%u' argument.", 0); return; } strlcpy(md.md_name, name, sizeof(md.md_name)); md.md_id = arc4random(); md.md_no = 0; md.md_all = *nargs - 1; stripesize = gctl_get_paraml(req, "stripesize", sizeof(*stripesize)); if (stripesize == NULL) { gctl_error(req, "No '%s' argument.", "stripesize"); return; } md.md_stripesize = *stripesize; bzero(md.md_provider, sizeof(md.md_provider)); /* This field is not important here. */ md.md_provsize = 0; /* Check all providers are valid */ for (no = 1; no < *nargs; no++) { snprintf(param, sizeof(param), "arg%u", no); name = gctl_get_asciiparam(req, param); if (name == NULL) { gctl_error(req, "No 'arg%u' argument.", no); return; } if (strncmp(name, "/dev/", strlen("/dev/")) == 0) name += strlen("/dev/"); pp = g_provider_by_name(name); if (pp == NULL) { G_STRIPE_DEBUG(1, "Disk %s is invalid.", name); gctl_error(req, "Disk %s is invalid.", name); return; } } gp = g_stripe_create(mp, &md, G_STRIPE_TYPE_MANUAL); if (gp == NULL) { gctl_error(req, "Can't configure %s.", md.md_name); return; } sc = gp->softc; sb = sbuf_new_auto(); sbuf_printf(sb, "Can't attach disk(s) to %s:", gp->name); for (attached = 0, no = 1; no < *nargs; no++) { snprintf(param, sizeof(param), "arg%u", no); name = gctl_get_asciiparam(req, param); if (name == NULL) { gctl_error(req, "No 'arg%u' argument.", no); continue; } if (strncmp(name, "/dev/", strlen("/dev/")) == 0) name += strlen("/dev/"); pp = g_provider_by_name(name); KASSERT(pp != NULL, ("Provider %s disappear?!", name)); if (g_stripe_add_disk(sc, pp, no - 1) != 0) { G_STRIPE_DEBUG(1, "Disk %u (%s) not attached to %s.", no, pp->name, gp->name); sbuf_printf(sb, " %s", pp->name); continue; } attached++; } sbuf_finish(sb); if (md.md_all != attached) { g_stripe_destroy(gp->softc, 1); gctl_error(req, "%s", sbuf_data(sb)); } sbuf_delete(sb); } static struct g_stripe_softc * g_stripe_find_device(struct g_class *mp, const char *name) { struct g_stripe_softc *sc; struct g_geom *gp; LIST_FOREACH(gp, &mp->geom, geom) { sc = gp->softc; if (sc == NULL) continue; if (strcmp(sc->sc_name, name) == 0) return (sc); } return (NULL); } static void g_stripe_ctl_destroy(struct gctl_req *req, struct g_class *mp) { struct g_stripe_softc *sc; int *force, *nargs, error; const char *name; char param[16]; u_int i; g_topology_assert(); nargs = gctl_get_paraml(req, "nargs", sizeof(*nargs)); if (nargs == NULL) { gctl_error(req, "No '%s' argument.", "nargs"); return; } if (*nargs <= 0) { gctl_error(req, "Missing device(s)."); return; } force = gctl_get_paraml(req, "force", sizeof(*force)); if (force == NULL) { gctl_error(req, "No '%s' argument.", "force"); return; } for (i = 0; i < (u_int)*nargs; i++) { snprintf(param, sizeof(param), "arg%u", i); name = gctl_get_asciiparam(req, param); if (name == NULL) { gctl_error(req, "No 'arg%u' argument.", i); return; } sc = g_stripe_find_device(mp, name); if (sc == NULL) { gctl_error(req, "No such device: %s.", name); return; } error = g_stripe_destroy(sc, *force); if (error != 0) { gctl_error(req, "Cannot destroy device %s (error=%d).", sc->sc_name, error); return; } } } static void g_stripe_config(struct gctl_req *req, struct g_class *mp, const char *verb) { uint32_t *version; g_topology_assert(); version = gctl_get_paraml(req, "version", sizeof(*version)); if (version == NULL) { gctl_error(req, "No '%s' argument.", "version"); return; } if (*version != G_STRIPE_VERSION) { gctl_error(req, "Userland and kernel parts are out of sync."); return; } if (strcmp(verb, "create") == 0) { g_stripe_ctl_create(req, mp); return; } else if (strcmp(verb, "destroy") == 0 || strcmp(verb, "stop") == 0) { g_stripe_ctl_destroy(req, mp); return; } gctl_error(req, "Unknown verb."); } static void g_stripe_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp) { struct g_stripe_softc *sc; sc = gp->softc; if (sc == NULL) return; if (pp != NULL) { /* Nothing here. */ } else if (cp != NULL) { sbuf_printf(sb, "%s%u\n", indent, (u_int)cp->index); } else { sbuf_printf(sb, "%s%u\n", indent, (u_int)sc->sc_id); sbuf_printf(sb, "%s%u\n", indent, (u_int)sc->sc_stripesize); sbuf_printf(sb, "%s", indent); switch (sc->sc_type) { case G_STRIPE_TYPE_AUTOMATIC: sbuf_printf(sb, "AUTOMATIC"); break; case G_STRIPE_TYPE_MANUAL: sbuf_printf(sb, "MANUAL"); break; default: sbuf_printf(sb, "UNKNOWN"); break; } sbuf_printf(sb, "\n"); sbuf_printf(sb, "%sTotal=%u, Online=%u\n", indent, sc->sc_ndisks, g_stripe_nvalid(sc)); sbuf_printf(sb, "%s", indent); if (sc->sc_provider != NULL && sc->sc_provider->error == 0) sbuf_printf(sb, "UP"); else sbuf_printf(sb, "DOWN"); sbuf_printf(sb, "\n"); } } DECLARE_GEOM_CLASS(g_stripe_class, g_stripe); Index: projects/largeSMP/sys/mips/conf/PB92 =================================================================== --- projects/largeSMP/sys/mips/conf/PB92 (revision 221494) +++ projects/largeSMP/sys/mips/conf/PB92 (revision 221495) @@ -1,100 +1,124 @@ # # PB92 -- Kernel configuration file for FreeBSD/mips for Atheros PB92 reference # board (AR7242) # # $FreeBSD$ # ident PB92 +# XXX The default load address in the Uboot environment is 0x80010000 makeoptions KERNLOADADDR=0x80050000 options HZ=1000 +# The PB92 has 32mb of RAM; hard-code that +options AR71XX_REALMEM=32*1024*1024 + hints "PB92.hints" include "../atheros/std.ar71xx" makeoptions DEBUG=-g #Build kernel with gdb(1) debug symbols makeoptions MODULES_OVERRIDE="" options DDB options KDB options SCHED_4BSD #4BSD scheduler options INET #InterNETworking options INET6 -options NFSCLIENT #Network Filesystem Client +# options NFSCLIENT #Network Filesystem Client options PSEUDOFS #Pseudo-filesystem framework options _KPOSIX_PRIORITY_SCHEDULING #Posix P1003_1B real-time extensions options ALQ -# options NFS_LEGACYRPC -#options NFS_DEBUG - # Debugging for use in -current -options DEADLKRES -options INVARIANTS -options INVARIANT_SUPPORT -options WITNESS -options WITNESS_SKIPSPIN +#options DEADLKRES +#options INVARIANTS +#options INVARIANT_SUPPORT +#options WITNESS +#options WITNESS_SKIPSPIN options FFS #Berkeley Fast Filesystem #options SOFTUPDATES #Enable FFS soft updates support #options UFS_ACL #Support for access control lists #options UFS_DIRHASH #Improve performance on big directories options MD_ROOT options MD_ROOT_SIZE=5120 device geom_uzip options GEOM_UZIP options ROOTDEVNAME=\"ufs:/dev/md0.uzip\" +# PCI bus device pci device ar724x_pci +# NVRAM U-Boot Environment -> Kernel environment +device nvram2env + # Wireless NIC cards options IEEE80211_DEBUG options IEEE80211_SUPPORT_MESH options IEEE80211_SUPPORT_TDMA device wlan # 802.11 support device wlan_wep # 802.11 WEP support device wlan_ccmp # 802.11 CCMP support device wlan_tkip # 802.11 TKIP support device wlan_xauth # 802.11 hostap support device ath # Atheros pci/cardbus NIC's device ath_pci # PCI/PCIe bus glue options ATH_DEBUG options ATH_ENABLE_11N options ATH_DIAGAPI device ath_hal options AH_SUPPORT_AR5416 options AH_DEBUG options AH_DEBUG_ALQ device ath_rate_sample device mii device arge +# USB devices - PB92 has EHCI only + device usb options USB_EHCI_BIG_ENDIAN_DESC # handle big-endian byte order options USB_DEBUG options USB_HOST_ALIGN=32 device ehci +# Mass storage +device scbus +device umass +device da + +# Read MSDOS formatted disks +options GEOM_PART_BSD +options GEOM_PART_MBR +options MSDOSFS + +# GPIO Bus device gpio device gpioled +# SPI and flash device spibus device ar71xx_spi device mx25l +# The flash is statically partitioned; add in that +device geom_map + device ar71xx_wdog +# Serial device uart +# Network twiddling device loop device ether device md device bpf device random device if_bridge Index: projects/largeSMP/sys/mips/conf/PB92.hints =================================================================== --- projects/largeSMP/sys/mips/conf/PB92.hints (revision 221494) +++ projects/largeSMP/sys/mips/conf/PB92.hints (revision 221495) @@ -1,69 +1,113 @@ # $FreeBSD$ hint.apb.0.at="nexus0" hint.apb.0.irq=4 # uart0 hint.uart.0.at="apb0" # see atheros/uart_cpu_ar71xx.c why +3 hint.uart.0.maddr=0x18020003 hint.uart.0.msize=0x18 hint.uart.0.irq=3 -# ohci -#hint.ohci.0.at="apb0" -#hint.ohci.0.maddr=0x1b000000 -#hint.ohci.0.msize=0x01000000 -#hint.ohci.0.irq=1 - #ehci - note the 0x100 offset for the AR913x/AR724x hint.ehci.0.at="nexus0" hint.ehci.0.maddr=0x1b000100 hint.ehci.0.msize=0x00ffff00 hint.ehci.0.irq=1 # pci hint.pcib.0.at="nexus0" hint.pcib.0.irq=0 # arge0 hint.arge.0.at="nexus0" hint.arge.0.maddr=0x19000000 hint.arge.0.msize=0x1000 hint.arge.0.irq=2 # AR8316 workaround for now hint.arge.0.media=1000 hint.arge.0.fduplex=1 hint.arge.0.phymask=0x3 # GPIO hint.gpio.0.at="apb0" hint.gpio.0.maddr=0x18040000 hint.gpio.0.msize=0x1000 hint.gpio.0.irq=2 # Signal leds hint.gpioled.0.at="gpiobus0" hint.gpioled.0.name="sig1" hint.gpioled.0.pins=0x0001 # pin 0 hint.gpioled.1.at="gpiobus0" hint.gpioled.1.name="sig2" hint.gpioled.1.pins=0x0002 # pin 1 hint.gpioled.2.at="gpiobus0" hint.gpioled.2.name="sig3" hint.gpioled.2.pins=0x0800 # pin 11 hint.gpioled.3.at="gpiobus0" hint.gpioled.3.name="sig4" hint.gpioled.3.pins=0x0080 # pin 7 # SPI controller/bus hint.spi.0.at="nexus0" hint.spi.0.maddr=0x1f000000 hint.spi.0.msize=0x10 # SPI flash hint.mx25l.0.at="spibus0" hint.mx25l.0.cs=0 # Watchdog hint.ar71xx_wdog.0.at="nexus0" + +# nvram mapping - XXX ? +hint.nvram.0.base=0x1f030000 +hint.nvram.0.maxsize=0x2000 +hint.nvram.0.flags=3 # 1 = No check, 2 = Format Generic +hint.nvram.1.base=0x1f032000 +hint.nvram.1.maxsize=0x4000 +hint.nvram.1.flags=3 # 1 = No check, 2 = Format Generic + +# GEOM_MAP +# +# From my PB92 environment: +# +# mtdparts=ar7240-nor0:256k(u-boot),64k(u-boot-env),2752k(rootfs),896k(uImage),64k(NVRAM),64k(ART) + +hint.map.0.at="flash/spi0" +hint.map.0.start=0x00000000 +hint.map.0.end=0x00040000 # 256k u-boot +hint.map.0.name="u-boot" +hint.map.0.readonly=1 + +hint.map.1.at="flash/spi0" +hint.map.1.start=0x00040000 +hint.map.1.end=0x00050000 # 64k u-boot-env +hint.map.1.name="u-boot-env" +hint.map.1.readonly=0 + +hint.map.2.at="flash/spi0" +hint.map.2.start=0x00050000 +hint.map.2.end=0x00300000 # 2752k rootfs +hint.map.2.name="rootfs" +hint.map.2.readonly=1 + +hint.map.3.at="flash/spi0" +hint.map.3.start=0x00300000 +hint.map.3.end=0x003e0000 # 896k uImage +hint.map.3.name="uImage" +hint.map.3.readonly=0 + +hint.map.4.at="flash/spi0" +hint.map.4.start=0x003e0000 +hint.map.4.end=0x003f0000 # 64k NVRAM +hint.map.4.name="NVRAM" +hint.map.4.readonly=0 + +hint.map.5.at="flash/spi0" +hint.map.5.start=0x003f0000 +hint.map.5.end=0x00400000 # 64k ART +hint.map.5.name="ART" +hint.map.5.readonly=1 Index: projects/largeSMP/sys/modules/cxgbe/if_cxgbe/Makefile =================================================================== --- projects/largeSMP/sys/modules/cxgbe/if_cxgbe/Makefile (revision 221494) +++ projects/largeSMP/sys/modules/cxgbe/if_cxgbe/Makefile (revision 221495) @@ -1,16 +1,19 @@ # # $FreeBSD$ # CXGBE = ${.CURDIR}/../../../dev/cxgbe .PATH: ${CXGBE} ${CXGBE}/common KMOD = if_cxgbe SRCS = t4_main.c t4_sge.c SRCS+= t4_hw.c SRCS+= device_if.h bus_if.h pci_if.h SRCS+= opt_inet.h CFLAGS+= -I${CXGBE} +# Provide the timestamp of a packet in its header mbuf. +#CFLAGS+= -DT4_PKT_TIMESTAMP + .include Index: projects/largeSMP/sys/netinet/sctp.h =================================================================== --- projects/largeSMP/sys/netinet/sctp.h (revision 221494) +++ projects/largeSMP/sys/netinet/sctp.h (revision 221495) @@ -1,581 +1,588 @@ /*- * Copyright (c) 2001-2008, by Cisco Systems, Inc. All rights reserved. * Copyright (c) 2008-2011, by Randall Stewart. All rights reserved. * Copyright (c) 2008-2011, by Michael Tuexen. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * a) Redistributions of source code must retain the above copyright notice, * this list of conditions and the following disclaimer. * * b) Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in * the documentation and/or other materials provided with the distribution. * * c) Neither the name of Cisco Systems, Inc. nor the names of its * contributors may be used to endorse or promote products derived * from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF * THE POSSIBILITY OF SUCH DAMAGE. */ /* $KAME: sctp.h,v 1.18 2005/03/06 16:04:16 itojun Exp $ */ #include __FBSDID("$FreeBSD$"); #ifndef _NETINET_SCTP_H_ #define _NETINET_SCTP_H_ #include #define SCTP_PACKED __attribute__((packed)) /* * SCTP protocol - RFC2960. */ struct sctphdr { uint16_t src_port; /* source port */ uint16_t dest_port; /* destination port */ uint32_t v_tag; /* verification tag of packet */ uint32_t checksum; /* Adler32 C-Sum */ /* chunks follow... */ } SCTP_PACKED; /* * SCTP Chunks */ struct sctp_chunkhdr { uint8_t chunk_type; /* chunk type */ uint8_t chunk_flags; /* chunk flags */ uint16_t chunk_length; /* chunk length */ /* optional params follow */ } SCTP_PACKED; /* * SCTP chunk parameters */ struct sctp_paramhdr { uint16_t param_type; /* parameter type */ uint16_t param_length; /* parameter length */ } SCTP_PACKED; /* * user socket options: socket API defined */ /* * read-write options */ #define SCTP_RTOINFO 0x00000001 #define SCTP_ASSOCINFO 0x00000002 #define SCTP_INITMSG 0x00000003 #define SCTP_NODELAY 0x00000004 #define SCTP_AUTOCLOSE 0x00000005 #define SCTP_SET_PEER_PRIMARY_ADDR 0x00000006 #define SCTP_PRIMARY_ADDR 0x00000007 #define SCTP_ADAPTATION_LAYER 0x00000008 /* same as above */ #define SCTP_ADAPTION_LAYER 0x00000008 #define SCTP_DISABLE_FRAGMENTS 0x00000009 #define SCTP_PEER_ADDR_PARAMS 0x0000000a #define SCTP_DEFAULT_SEND_PARAM 0x0000000b /* ancillary data/notification interest options */ #define SCTP_EVENTS 0x0000000c /* Without this applied we will give V4 and V6 addresses on a V6 socket */ #define SCTP_I_WANT_MAPPED_V4_ADDR 0x0000000d #define SCTP_MAXSEG 0x0000000e #define SCTP_DELAYED_SACK 0x0000000f #define SCTP_FRAGMENT_INTERLEAVE 0x00000010 #define SCTP_PARTIAL_DELIVERY_POINT 0x00000011 /* authentication support */ #define SCTP_AUTH_CHUNK 0x00000012 #define SCTP_AUTH_KEY 0x00000013 #define SCTP_HMAC_IDENT 0x00000014 #define SCTP_AUTH_ACTIVE_KEY 0x00000015 #define SCTP_AUTH_DELETE_KEY 0x00000016 #define SCTP_USE_EXT_RCVINFO 0x00000017 #define SCTP_AUTO_ASCONF 0x00000018 /* rw */ #define SCTP_MAXBURST 0x00000019 /* rw */ #define SCTP_MAX_BURST 0x00000019 /* rw */ /* assoc level context */ #define SCTP_CONTEXT 0x0000001a /* rw */ /* explicit EOR signalling */ #define SCTP_EXPLICIT_EOR 0x0000001b #define SCTP_REUSE_PORT 0x0000001c /* rw */ #define SCTP_AUTH_DEACTIVATE_KEY 0x0000001d /* * read-only options */ #define SCTP_STATUS 0x00000100 #define SCTP_GET_PEER_ADDR_INFO 0x00000101 /* authentication support */ #define SCTP_PEER_AUTH_CHUNKS 0x00000102 #define SCTP_LOCAL_AUTH_CHUNKS 0x00000103 #define SCTP_GET_ASSOC_NUMBER 0x00000104 /* ro */ #define SCTP_GET_ASSOC_ID_LIST 0x00000105 /* ro */ #define SCTP_TIMEOUTS 0x00000106 /* * user socket options: BSD implementation specific */ /* * Blocking I/O is enabled on any TCP type socket by default. For the UDP * model if this is turned on then the socket buffer is shared for send * resources amongst all associations. The default for the UDP model is that * is SS_NBIO is set. Which means all associations have a separate send * limit BUT they will NOT ever BLOCK instead you will get an error back * EAGAIN if you try to send too much. If you want the blocking semantics you * set this option at the cost of sharing one socket send buffer size amongst * all associations. Peeled off sockets turn this option off and block. But * since both TCP and peeled off sockets have only one assoc per socket this * is fine. It probably does NOT make sense to set this on SS_NBIO on a TCP * model OR peeled off UDP model, but we do allow you to do so. You just use * the normal syscall to toggle SS_NBIO the way you want. * * Blocking I/O is controlled by the SS_NBIO flag on the socket state so_state * field. */ /* these should probably go into sockets API */ #define SCTP_RESET_STREAMS 0x00001004 /* wo */ /* here on down are more implementation specific */ #define SCTP_SET_DEBUG_LEVEL 0x00001005 #define SCTP_CLR_STAT_LOG 0x00001007 /* CMT ON/OFF socket option */ #define SCTP_CMT_ON_OFF 0x00001200 #define SCTP_CMT_USE_DAC 0x00001201 /* JRS - Pluggable Congestion Control Socket option */ #define SCTP_PLUGGABLE_CC 0x00001202 /* RS - Pluggable Stream Scheduling Socket option */ #define SCTP_PLUGGABLE_SS 0x00001203 #define SCTP_SS_VALUE 0x00001204 #define SCTP_CC_OPTION 0x00001205 /* Options for CC * modules */ /* read only */ #define SCTP_GET_SNDBUF_USE 0x00001101 #define SCTP_GET_STAT_LOG 0x00001103 #define SCTP_PCB_STATUS 0x00001104 #define SCTP_GET_NONCE_VALUES 0x00001105 /* Special hook for dynamically setting primary for all assoc's, * this is a write only option that requires root privilege. */ #define SCTP_SET_DYNAMIC_PRIMARY 0x00002001 /* VRF (virtual router feature) and multi-VRF support * options. VRF's provide splits within a router * that give the views of multiple routers. A * standard host, without VRF support, is just * a single VRF. If VRF's are supported then * the transport must be VRF aware. This means * that every socket call coming in must be directed * within the endpoint to one of the VRF's it belongs * to. The endpoint, before binding, may select * the "default" VRF it is in by using a set socket * option with SCTP_VRF_ID. This will also * get propagated to the default VRF. Once the * endpoint binds an address then it CANNOT add * additional VRF's to become a Multi-VRF endpoint. * * Before BINDING additional VRF's can be added with * the SCTP_ADD_VRF_ID call or deleted with * SCTP_DEL_VRF_ID. * * Associations are ALWAYS contained inside a single * VRF. They cannot reside in two (or more) VRF's. Incoming * packets, assuming the router is VRF aware, can always * tell us what VRF they arrived on. A host not supporting * any VRF's will find that the packets always arrived on the * single VRF that the host has. * */ #define SCTP_VRF_ID 0x00003001 #define SCTP_ADD_VRF_ID 0x00003002 #define SCTP_GET_VRF_IDS 0x00003003 #define SCTP_GET_ASOC_VRF 0x00003004 #define SCTP_DEL_VRF_ID 0x00003005 /* * If you enable packet logging you can get * a poor mans ethereal output in binary * form. Note this is a compile option to * the kernel, SCTP_PACKET_LOGGING, and * without it in your kernel you * will get a EOPNOTSUPP */ #define SCTP_GET_PACKET_LOG 0x00004001 /* * hidden implementation specific options these are NOT user visible (should * move out of sctp.h) */ /* sctp_bindx() flags as hidden socket options */ #define SCTP_BINDX_ADD_ADDR 0x00008001 #define SCTP_BINDX_REM_ADDR 0x00008002 /* Hidden socket option that gets the addresses */ #define SCTP_GET_PEER_ADDRESSES 0x00008003 #define SCTP_GET_LOCAL_ADDRESSES 0x00008004 /* return the total count in bytes needed to hold all local addresses bound */ #define SCTP_GET_LOCAL_ADDR_SIZE 0x00008005 /* Return the total count in bytes needed to hold the remote address */ #define SCTP_GET_REMOTE_ADDR_SIZE 0x00008006 /* hidden option for connectx */ #define SCTP_CONNECT_X 0x00008007 /* hidden option for connectx_delayed, part of sendx */ #define SCTP_CONNECT_X_DELAYED 0x00008008 #define SCTP_CONNECT_X_COMPLETE 0x00008009 /* hidden socket option based sctp_peeloff */ #define SCTP_PEELOFF 0x0000800a /* the real worker for sctp_getaddrlen() */ #define SCTP_GET_ADDR_LEN 0x0000800b /* temporary workaround for Apple listen() issue, no args used */ #define SCTP_LISTEN_FIX 0x0000800c /* Debug things that need to be purged */ #define SCTP_SET_INITIAL_DBG_SEQ 0x00009f00 /* JRS - Supported congestion control modules for pluggable * congestion control */ /* Standard TCP Congestion Control */ #define SCTP_CC_RFC2581 0x00000000 /* High Speed TCP Congestion Control (Floyd) */ #define SCTP_CC_HSTCP 0x00000001 /* HTCP Congestion Control */ #define SCTP_CC_HTCP 0x00000002 /* RTCC Congestion Control - RFC2581 plus */ #define SCTP_CC_RTCC 0x00000003 #define SCTP_CC_OPT_RTCC_SETMODE 0x00002000 #define SCTP_CC_OPT_USE_DCCC_ECN 0x00002001 #define SCTP_CC_OPT_STEADY_STEP 0x00002002 +#define SCTP_CMT_OFF 0 +#define SCTP_CMT_BASE 1 +#define SCTP_CMT_RPV1 2 +#define SCTP_CMT_RPV2 3 +#define SCTP_CMT_MPTCP 4 +#define SCTP_CMT_MAX SCTP_CMT_MPTCP + /* RS - Supported stream scheduling modules for pluggable * stream scheduling */ /* Default simple round-robin */ #define SCTP_SS_DEFAULT 0x00000000 /* Real round-robin */ #define SCTP_SS_ROUND_ROBIN 0x00000001 /* Real round-robin per packet */ #define SCTP_SS_ROUND_ROBIN_PACKET 0x00000002 /* Priority */ #define SCTP_SS_PRIORITY 0x00000003 /* Fair Bandwidth */ #define SCTP_SS_FAIR_BANDWITH 0x00000004 /* First-come, first-serve */ #define SCTP_SS_FIRST_COME 0x00000005 /* fragment interleave constants * setting must be one of these or * EINVAL returned. */ #define SCTP_FRAG_LEVEL_0 0x00000000 #define SCTP_FRAG_LEVEL_1 0x00000001 #define SCTP_FRAG_LEVEL_2 0x00000002 /* * user state values */ #define SCTP_CLOSED 0x0000 #define SCTP_BOUND 0x1000 #define SCTP_LISTEN 0x2000 #define SCTP_COOKIE_WAIT 0x0002 #define SCTP_COOKIE_ECHOED 0x0004 #define SCTP_ESTABLISHED 0x0008 #define SCTP_SHUTDOWN_SENT 0x0010 #define SCTP_SHUTDOWN_RECEIVED 0x0020 #define SCTP_SHUTDOWN_ACK_SENT 0x0040 #define SCTP_SHUTDOWN_PENDING 0x0080 /* * SCTP operational error codes (user visible) */ #define SCTP_CAUSE_NO_ERROR 0x0000 #define SCTP_CAUSE_INVALID_STREAM 0x0001 #define SCTP_CAUSE_MISSING_PARAM 0x0002 #define SCTP_CAUSE_STALE_COOKIE 0x0003 #define SCTP_CAUSE_OUT_OF_RESC 0x0004 #define SCTP_CAUSE_UNRESOLVABLE_ADDR 0x0005 #define SCTP_CAUSE_UNRECOG_CHUNK 0x0006 #define SCTP_CAUSE_INVALID_PARAM 0x0007 #define SCTP_CAUSE_UNRECOG_PARAM 0x0008 #define SCTP_CAUSE_NO_USER_DATA 0x0009 #define SCTP_CAUSE_COOKIE_IN_SHUTDOWN 0x000a #define SCTP_CAUSE_RESTART_W_NEWADDR 0x000b #define SCTP_CAUSE_USER_INITIATED_ABT 0x000c #define SCTP_CAUSE_PROTOCOL_VIOLATION 0x000d /* Error causes from RFC5061 */ #define SCTP_CAUSE_DELETING_LAST_ADDR 0x00a0 #define SCTP_CAUSE_RESOURCE_SHORTAGE 0x00a1 #define SCTP_CAUSE_DELETING_SRC_ADDR 0x00a2 #define SCTP_CAUSE_ILLEGAL_ASCONF_ACK 0x00a3 #define SCTP_CAUSE_REQUEST_REFUSED 0x00a4 /* Error causes from nat-draft */ #define SCTP_CAUSE_NAT_COLLIDING_STATE 0x00b0 #define SCTP_CAUSE_NAT_MISSING_STATE 0x00b1 /* Error causes from RFC4895 */ #define SCTP_CAUSE_UNSUPPORTED_HMACID 0x0105 /* * error cause parameters (user visible) */ struct sctp_error_cause { uint16_t code; uint16_t length; /* optional cause-specific info may follow */ } SCTP_PACKED; struct sctp_error_invalid_stream { struct sctp_error_cause cause; /* code=SCTP_ERROR_INVALID_STREAM */ uint16_t stream_id; /* stream id of the DATA in error */ uint16_t reserved; } SCTP_PACKED; struct sctp_error_missing_param { struct sctp_error_cause cause; /* code=SCTP_ERROR_MISSING_PARAM */ uint32_t num_missing_params; /* number of missing parameters */ /* uint16_t param_type's follow */ } SCTP_PACKED; struct sctp_error_stale_cookie { struct sctp_error_cause cause; /* code=SCTP_ERROR_STALE_COOKIE */ uint32_t stale_time; /* time in usec of staleness */ } SCTP_PACKED; struct sctp_error_out_of_resource { struct sctp_error_cause cause; /* code=SCTP_ERROR_OUT_OF_RESOURCES */ } SCTP_PACKED; struct sctp_error_unresolv_addr { struct sctp_error_cause cause; /* code=SCTP_ERROR_UNRESOLVABLE_ADDR */ } SCTP_PACKED; struct sctp_error_unrecognized_chunk { struct sctp_error_cause cause; /* code=SCTP_ERROR_UNRECOG_CHUNK */ struct sctp_chunkhdr ch;/* header from chunk in error */ } SCTP_PACKED; /* * Main SCTP chunk types we place these here so natd and f/w's in user land * can find them. */ /************0x00 series ***********/ #define SCTP_DATA 0x00 #define SCTP_INITIATION 0x01 #define SCTP_INITIATION_ACK 0x02 #define SCTP_SELECTIVE_ACK 0x03 #define SCTP_HEARTBEAT_REQUEST 0x04 #define SCTP_HEARTBEAT_ACK 0x05 #define SCTP_ABORT_ASSOCIATION 0x06 #define SCTP_SHUTDOWN 0x07 #define SCTP_SHUTDOWN_ACK 0x08 #define SCTP_OPERATION_ERROR 0x09 #define SCTP_COOKIE_ECHO 0x0a #define SCTP_COOKIE_ACK 0x0b #define SCTP_ECN_ECHO 0x0c #define SCTP_ECN_CWR 0x0d #define SCTP_SHUTDOWN_COMPLETE 0x0e /* RFC4895 */ #define SCTP_AUTHENTICATION 0x0f /* EY nr_sack chunk id*/ #define SCTP_NR_SELECTIVE_ACK 0x10 /************0x40 series ***********/ /************0x80 series ***********/ /* RFC5061 */ #define SCTP_ASCONF_ACK 0x80 /* draft-ietf-stewart-pktdrpsctp */ #define SCTP_PACKET_DROPPED 0x81 /* draft-ietf-stewart-strreset-xxx */ #define SCTP_STREAM_RESET 0x82 /* RFC4820 */ #define SCTP_PAD_CHUNK 0x84 /************0xc0 series ***********/ /* RFC3758 */ #define SCTP_FORWARD_CUM_TSN 0xc0 /* RFC5061 */ #define SCTP_ASCONF 0xc1 /* ABORT and SHUTDOWN COMPLETE FLAG */ #define SCTP_HAD_NO_TCB 0x01 /* Packet dropped flags */ #define SCTP_FROM_MIDDLE_BOX SCTP_HAD_NO_TCB #define SCTP_BADCRC 0x02 #define SCTP_PACKET_TRUNCATED 0x04 /* Flag for ECN -CWR */ #define SCTP_CWR_REDUCE_OVERRIDE 0x01 #define SCTP_CWR_IN_SAME_WINDOW 0x02 #define SCTP_SAT_NETWORK_MIN 400 /* min ms for RTT to set satellite * time */ #define SCTP_SAT_NETWORK_BURST_INCR 2 /* how many times to multiply maxburst * in sat */ /* Data Chuck Specific Flags */ #define SCTP_DATA_FRAG_MASK 0x03 #define SCTP_DATA_MIDDLE_FRAG 0x00 #define SCTP_DATA_LAST_FRAG 0x01 #define SCTP_DATA_FIRST_FRAG 0x02 #define SCTP_DATA_NOT_FRAG 0x03 #define SCTP_DATA_UNORDERED 0x04 #define SCTP_DATA_SACK_IMMEDIATELY 0x08 /* ECN Nonce: SACK Chunk Specific Flags */ #define SCTP_SACK_NONCE_SUM 0x01 /* CMT DAC algorithm SACK flag */ #define SCTP_SACK_CMT_DAC 0x80 /* * PCB flags (in sctp_flags bitmask). * Note the features and flags are meant * for use by netstat. */ #define SCTP_PCB_FLAGS_UDPTYPE 0x00000001 #define SCTP_PCB_FLAGS_TCPTYPE 0x00000002 #define SCTP_PCB_FLAGS_BOUNDALL 0x00000004 #define SCTP_PCB_FLAGS_ACCEPTING 0x00000008 #define SCTP_PCB_FLAGS_UNBOUND 0x00000010 #define SCTP_PCB_FLAGS_CLOSE_IP 0x00040000 #define SCTP_PCB_FLAGS_WAS_CONNECTED 0x00080000 #define SCTP_PCB_FLAGS_WAS_ABORTED 0x00100000 /* TCP model support */ #define SCTP_PCB_FLAGS_CONNECTED 0x00200000 #define SCTP_PCB_FLAGS_IN_TCPPOOL 0x00400000 #define SCTP_PCB_FLAGS_DONT_WAKE 0x00800000 #define SCTP_PCB_FLAGS_WAKEOUTPUT 0x01000000 #define SCTP_PCB_FLAGS_WAKEINPUT 0x02000000 #define SCTP_PCB_FLAGS_BOUND_V6 0x04000000 #define SCTP_PCB_FLAGS_BLOCKING_IO 0x08000000 #define SCTP_PCB_FLAGS_SOCKET_GONE 0x10000000 #define SCTP_PCB_FLAGS_SOCKET_ALLGONE 0x20000000 #define SCTP_PCB_FLAGS_SOCKET_CANT_READ 0x40000000 /* flags to copy to new PCB */ #define SCTP_PCB_COPY_FLAGS (SCTP_PCB_FLAGS_BOUNDALL|\ SCTP_PCB_FLAGS_WAKEINPUT|\ SCTP_PCB_FLAGS_BOUND_V6) /* * PCB Features (in sctp_features bitmask) */ #define SCTP_PCB_FLAGS_EXT_RCVINFO 0x00000002 #define SCTP_PCB_FLAGS_DONOT_HEARTBEAT 0x00000004 #define SCTP_PCB_FLAGS_FRAG_INTERLEAVE 0x00000008 #define SCTP_PCB_FLAGS_INTERLEAVE_STRMS 0x00000010 #define SCTP_PCB_FLAGS_DO_ASCONF 0x00000020 #define SCTP_PCB_FLAGS_AUTO_ASCONF 0x00000040 #define SCTP_PCB_FLAGS_ZERO_COPY_ACTIVE 0x00000080 /* socket options */ #define SCTP_PCB_FLAGS_NODELAY 0x00000100 #define SCTP_PCB_FLAGS_AUTOCLOSE 0x00000200 #define SCTP_PCB_FLAGS_RECVDATAIOEVNT 0x00000400 #define SCTP_PCB_FLAGS_RECVASSOCEVNT 0x00000800 #define SCTP_PCB_FLAGS_RECVPADDREVNT 0x00001000 #define SCTP_PCB_FLAGS_RECVPEERERR 0x00002000 #define SCTP_PCB_FLAGS_RECVSENDFAILEVNT 0x00004000 #define SCTP_PCB_FLAGS_RECVSHUTDOWNEVNT 0x00008000 #define SCTP_PCB_FLAGS_ADAPTATIONEVNT 0x00010000 #define SCTP_PCB_FLAGS_PDAPIEVNT 0x00020000 #define SCTP_PCB_FLAGS_AUTHEVNT 0x00040000 #define SCTP_PCB_FLAGS_STREAM_RESETEVNT 0x00080000 #define SCTP_PCB_FLAGS_NO_FRAGMENT 0x00100000 #define SCTP_PCB_FLAGS_EXPLICIT_EOR 0x00400000 #define SCTP_PCB_FLAGS_NEEDS_MAPPED_V4 0x00800000 #define SCTP_PCB_FLAGS_MULTIPLE_ASCONFS 0x01000000 #define SCTP_PCB_FLAGS_PORTREUSE 0x02000000 #define SCTP_PCB_FLAGS_DRYEVNT 0x04000000 /*- * mobility_features parameters (by micchie).Note * these features are applied against the * sctp_mobility_features flags.. not the sctp_features * flags. */ #define SCTP_MOBILITY_BASE 0x00000001 #define SCTP_MOBILITY_FASTHANDOFF 0x00000002 #define SCTP_MOBILITY_PRIM_DELETED 0x00000004 #define SCTP_SMALLEST_PMTU 512 /* smallest pmtu allowed when disabling PMTU * discovery */ #include /* This dictates the size of the packet * collection buffer. This only applies * if SCTP_PACKET_LOGGING is enabled in * your config. */ #define SCTP_PACKET_LOG_SIZE 65536 /* Maximum delays and such a user can set for options that * take ms. */ #define SCTP_MAX_SACK_DELAY 500 /* per RFC4960 */ #define SCTP_MAX_HB_INTERVAL 14400000 /* 4 hours in ms */ #define SCTP_MAX_COOKIE_LIFE 3600000 /* 1 hour in ms */ /* Types of logging/KTR tracing that can be enabled via the * sysctl net.inet.sctp.sctp_logging. You must also enable * SUBSYS tracing. * Note that you must have the SCTP option in the kernel * to enable these as well. */ #define SCTP_BLK_LOGGING_ENABLE 0x00000001 #define SCTP_CWND_MONITOR_ENABLE 0x00000002 #define SCTP_CWND_LOGGING_ENABLE 0x00000004 #define SCTP_EARLYFR_LOGGING_ENABLE 0x00000010 #define SCTP_FLIGHT_LOGGING_ENABLE 0x00000020 #define SCTP_FR_LOGGING_ENABLE 0x00000040 #define SCTP_LOCK_LOGGING_ENABLE 0x00000080 #define SCTP_MAP_LOGGING_ENABLE 0x00000100 #define SCTP_MBCNT_LOGGING_ENABLE 0x00000200 #define SCTP_MBUF_LOGGING_ENABLE 0x00000400 #define SCTP_NAGLE_LOGGING_ENABLE 0x00000800 #define SCTP_RECV_RWND_LOGGING_ENABLE 0x00001000 #define SCTP_RTTVAR_LOGGING_ENABLE 0x00002000 #define SCTP_SACK_LOGGING_ENABLE 0x00004000 #define SCTP_SACK_RWND_LOGGING_ENABLE 0x00008000 #define SCTP_SB_LOGGING_ENABLE 0x00010000 #define SCTP_STR_LOGGING_ENABLE 0x00020000 #define SCTP_WAKE_LOGGING_ENABLE 0x00040000 #define SCTP_LOG_MAXBURST_ENABLE 0x00080000 #define SCTP_LOG_RWND_ENABLE 0x00100000 #define SCTP_LOG_SACK_ARRIVALS_ENABLE 0x00200000 #define SCTP_LTRACE_CHUNK_ENABLE 0x00400000 #define SCTP_LTRACE_ERROR_ENABLE 0x00800000 #define SCTP_LAST_PACKET_TRACING 0x01000000 #define SCTP_THRESHOLD_LOGGING 0x02000000 #define SCTP_LOG_AT_SEND_2_SCTP 0x04000000 #define SCTP_LOG_AT_SEND_2_OUTQ 0x08000000 #define SCTP_LOG_TRY_ADVANCE 0x10000000 #undef SCTP_PACKED #endif /* !_NETINET_SCTP_H_ */ Index: projects/largeSMP/sys/netinet/sctp_cc_functions.c =================================================================== --- projects/largeSMP/sys/netinet/sctp_cc_functions.c (revision 221494) +++ projects/largeSMP/sys/netinet/sctp_cc_functions.c (revision 221495) @@ -1,2488 +1,2664 @@ /*- * Copyright (c) 2001-2007, by Cisco Systems, Inc. All rights reserved. * Copyright (c) 2008-2011, by Randall Stewart. All rights reserved. * Copyright (c) 2008-2011, by Michael Tuexen. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * a) Redistributions of source code must retain the above copyright notice, * this list of conditions and the following disclaimer. * * b) Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in * the documentation and/or other materials provided with the distribution. * * c) Neither the name of Cisco Systems, Inc. nor the names of its * contributors may be used to endorse or promote products derived * from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF * THE POSSIBILITY OF SUCH DAMAGE. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include __FBSDID("$FreeBSD$"); +#define SHIFT_MPTCP_MULTI_N 40 +#define SHIFT_MPTCP_MULTI_Z 16 +#define SHIFT_MPTCP_MULTI 8 + static void sctp_set_initial_cc_param(struct sctp_tcb *stcb, struct sctp_nets *net) { struct sctp_association *assoc; uint32_t cwnd_in_mtu; assoc = &stcb->asoc; cwnd_in_mtu = SCTP_BASE_SYSCTL(sctp_initial_cwnd); if (cwnd_in_mtu == 0) { /* Using 0 means that the value of RFC 4960 is used. */ net->cwnd = min((net->mtu * 4), max((2 * net->mtu), SCTP_INITIAL_CWND)); } else { /* * We take the minimum of the burst limit and the initial * congestion window. */ if ((assoc->max_burst > 0) && (cwnd_in_mtu > assoc->max_burst)) cwnd_in_mtu = assoc->max_burst; net->cwnd = (net->mtu - sizeof(struct sctphdr)) * cwnd_in_mtu; } - if (stcb->asoc.sctp_cmt_on_off == 2) { + if ((stcb->asoc.sctp_cmt_on_off == SCTP_CMT_RPV1) || + (stcb->asoc.sctp_cmt_on_off == SCTP_CMT_RPV2)) { /* In case of resource pooling initialize appropriately */ net->cwnd /= assoc->numnets; if (net->cwnd < (net->mtu - sizeof(struct sctphdr))) { net->cwnd = net->mtu - sizeof(struct sctphdr); } } net->ssthresh = assoc->peers_rwnd; SDT_PROBE(sctp, cwnd, net, init, stcb->asoc.my_vtag, ((stcb->sctp_ep->sctp_lport << 16) | (stcb->rport)), net, 0, net->cwnd); if (SCTP_BASE_SYSCTL(sctp_logging_level) & (SCTP_CWND_MONITOR_ENABLE | SCTP_CWND_LOGGING_ENABLE)) { sctp_log_cwnd(stcb, net, 0, SCTP_CWND_INITIALIZATION); } } static void sctp_cwnd_update_after_fr(struct sctp_tcb *stcb, struct sctp_association *asoc) { struct sctp_nets *net; uint32_t t_ssthresh, t_cwnd; + uint64_t t_ucwnd_sbw; /* MT FIXME: Don't compute this over and over again */ t_ssthresh = 0; t_cwnd = 0; - if (asoc->sctp_cmt_on_off == 2) { + t_ucwnd_sbw = 0; + if ((asoc->sctp_cmt_on_off == SCTP_CMT_RPV1) || + (asoc->sctp_cmt_on_off == SCTP_CMT_RPV2)) { TAILQ_FOREACH(net, &asoc->nets, sctp_next) { t_ssthresh += net->ssthresh; t_cwnd += net->cwnd; + if (net->lastsa > 0) { + t_ucwnd_sbw += (uint64_t) net->cwnd / (uint64_t) net->lastsa; + } } + if (t_ucwnd_sbw == 0) { + t_ucwnd_sbw = 1; + } } /*- * CMT fast recovery code. Need to debug. ((sctp_cmt_on_off > 0) && * (net->fast_retran_loss_recovery == 0))) */ TAILQ_FOREACH(net, &asoc->nets, sctp_next) { if ((asoc->fast_retran_loss_recovery == 0) || (asoc->sctp_cmt_on_off > 0)) { /* out of a RFC2582 Fast recovery window? */ if (net->net_ack > 0) { /* * per section 7.2.3, are there any * destinations that had a fast retransmit * to them. If so what we need to do is * adjust ssthresh and cwnd. */ struct sctp_tmit_chunk *lchk; int old_cwnd = net->cwnd; - if (asoc->sctp_cmt_on_off == 2) { - net->ssthresh = (uint32_t) (((uint64_t) 4 * - (uint64_t) net->mtu * - (uint64_t) net->ssthresh) / - (uint64_t) t_ssthresh); + if ((asoc->sctp_cmt_on_off == SCTP_CMT_RPV1) || + (asoc->sctp_cmt_on_off == SCTP_CMT_RPV2)) { + if (asoc->sctp_cmt_on_off == SCTP_CMT_RPV1) { + net->ssthresh = (uint32_t) (((uint64_t) 4 * + (uint64_t) net->mtu * + (uint64_t) net->ssthresh) / + (uint64_t) t_ssthresh); + + } + if (asoc->sctp_cmt_on_off == SCTP_CMT_RPV2) { + uint32_t srtt; + + srtt = net->lastsa; + /* + * lastsa>>3; we don't need + * to devide ... + */ + if (srtt == 0) { + srtt = 1; + } + /* + * Short Version => Equal to + * Contel Version MBe + */ + net->ssthresh = (uint32_t) (((uint64_t) 4 * + (uint64_t) net->mtu * + (uint64_t) net->cwnd) / + ((uint64_t) srtt * + t_ucwnd_sbw)); + /* INCREASE FACTOR */ ; + } if ((net->cwnd > t_cwnd / 2) && (net->ssthresh < net->cwnd - t_cwnd / 2)) { net->ssthresh = net->cwnd - t_cwnd / 2; } if (net->ssthresh < net->mtu) { net->ssthresh = net->mtu; } } else { net->ssthresh = net->cwnd / 2; if (net->ssthresh < (net->mtu * 2)) { net->ssthresh = 2 * net->mtu; } } net->cwnd = net->ssthresh; SDT_PROBE(sctp, cwnd, net, fr, stcb->asoc.my_vtag, ((stcb->sctp_ep->sctp_lport << 16) | (stcb->rport)), net, old_cwnd, net->cwnd); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_FR); } lchk = TAILQ_FIRST(&asoc->send_queue); net->partial_bytes_acked = 0; /* Turn on fast recovery window */ asoc->fast_retran_loss_recovery = 1; if (lchk == NULL) { /* Mark end of the window */ asoc->fast_recovery_tsn = asoc->sending_seq - 1; } else { asoc->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1; } /* * CMT fast recovery -- per destination * recovery variable. */ net->fast_retran_loss_recovery = 1; if (lchk == NULL) { /* Mark end of the window */ net->fast_recovery_tsn = asoc->sending_seq - 1; } else { net->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1; } sctp_timer_stop(SCTP_TIMER_TYPE_SEND, stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_INDATA + SCTP_LOC_32); sctp_timer_start(SCTP_TIMER_TYPE_SEND, stcb->sctp_ep, stcb, net); } } else if (net->net_ack > 0) { /* * Mark a peg that we WOULD have done a cwnd * reduction but RFC2582 prevented this action. */ SCTP_STAT_INCR(sctps_fastretransinrtt); } } } /* Defines for instantaneous bw decisions */ #define SCTP_INST_LOOSING 1 /* Loosing to other flows */ #define SCTP_INST_NEUTRAL 2 /* Neutral, no indication */ #define SCTP_INST_GAINING 3 /* Gaining, step down possible */ static int cc_bw_same(struct sctp_tcb *stcb, struct sctp_nets *net, uint64_t nbw, uint64_t rtt_offset, uint64_t vtag, uint8_t inst_ind) { uint64_t oth, probepoint; probepoint = (((uint64_t) net->cwnd) << 32); if (net->rtt > net->cc_mod.rtcc.lbw_rtt + rtt_offset) { /* * rtt increased we don't update bw.. so we don't update the * rtt either. */ /* Probe point 5 */ probepoint |= ((5 << 16) | 1); SDT_PROBE(sctp, cwnd, net, rttvar, vtag, ((net->cc_mod.rtcc.lbw << 32) | nbw), ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), net->flight_size, probepoint); if ((net->cc_mod.rtcc.steady_step) && (inst_ind != SCTP_INST_LOOSING)) { if (net->cc_mod.rtcc.last_step_state == 5) net->cc_mod.rtcc.step_cnt++; else net->cc_mod.rtcc.step_cnt = 1; net->cc_mod.rtcc.last_step_state = 5; if ((net->cc_mod.rtcc.step_cnt == net->cc_mod.rtcc.steady_step) || ((net->cc_mod.rtcc.step_cnt > net->cc_mod.rtcc.steady_step) && ((net->cc_mod.rtcc.step_cnt % net->cc_mod.rtcc.steady_step) == 0))) { /* Try a step down */ oth = net->cc_mod.rtcc.vol_reduce; oth <<= 16; oth |= net->cc_mod.rtcc.step_cnt; oth <<= 16; oth |= net->cc_mod.rtcc.last_step_state; SDT_PROBE(sctp, cwnd, net, rttstep, vtag, ((net->cc_mod.rtcc.lbw << 32) | nbw), ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), oth, probepoint); if (net->cwnd > (4 * net->mtu)) { net->cwnd -= net->mtu; net->cc_mod.rtcc.vol_reduce++; } else { net->cc_mod.rtcc.step_cnt = 0; } } } return (1); } if (net->rtt < net->cc_mod.rtcc.lbw_rtt - rtt_offset) { /* * rtt decreased, there could be more room. we update both * the bw and the rtt here to lock this in as a good step * down. */ /* Probe point 6 */ probepoint |= ((6 << 16) | 0); SDT_PROBE(sctp, cwnd, net, rttvar, vtag, ((net->cc_mod.rtcc.lbw << 32) | nbw), ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), net->flight_size, probepoint); if (net->cc_mod.rtcc.steady_step) { oth = net->cc_mod.rtcc.vol_reduce; oth <<= 16; oth |= net->cc_mod.rtcc.step_cnt; oth <<= 16; oth |= net->cc_mod.rtcc.last_step_state; SDT_PROBE(sctp, cwnd, net, rttstep, vtag, ((net->cc_mod.rtcc.lbw << 32) | nbw), ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), oth, probepoint); if ((net->cc_mod.rtcc.last_step_state == 5) && (net->cc_mod.rtcc.step_cnt > net->cc_mod.rtcc.steady_step)) { /* Step down worked */ net->cc_mod.rtcc.step_cnt = 0; return (1); } else { net->cc_mod.rtcc.last_step_state = 6; net->cc_mod.rtcc.step_cnt = 0; } } net->cc_mod.rtcc.lbw = nbw; net->cc_mod.rtcc.lbw_rtt = net->rtt; net->cc_mod.rtcc.cwnd_at_bw_set = net->cwnd; if (inst_ind == SCTP_INST_GAINING) return (1); else if (inst_ind == SCTP_INST_NEUTRAL) return (1); else return (0); } /* * Ok bw and rtt remained the same .. no update to any */ /* Probe point 7 */ probepoint |= ((7 << 16) | net->cc_mod.rtcc.ret_from_eq); SDT_PROBE(sctp, cwnd, net, rttvar, vtag, ((net->cc_mod.rtcc.lbw << 32) | nbw), ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), net->flight_size, probepoint); if ((net->cc_mod.rtcc.steady_step) && (inst_ind != SCTP_INST_LOOSING)) { if (net->cc_mod.rtcc.last_step_state == 5) net->cc_mod.rtcc.step_cnt++; else net->cc_mod.rtcc.step_cnt = 1; net->cc_mod.rtcc.last_step_state = 5; if ((net->cc_mod.rtcc.step_cnt == net->cc_mod.rtcc.steady_step) || ((net->cc_mod.rtcc.step_cnt > net->cc_mod.rtcc.steady_step) && ((net->cc_mod.rtcc.step_cnt % net->cc_mod.rtcc.steady_step) == 0))) { /* Try a step down */ if (net->cwnd > (4 * net->mtu)) { net->cwnd -= net->mtu; net->cc_mod.rtcc.vol_reduce++; return (1); } else { net->cc_mod.rtcc.step_cnt = 0; } } } if (inst_ind == SCTP_INST_GAINING) return (1); else if (inst_ind == SCTP_INST_NEUTRAL) return (1); else return ((int)net->cc_mod.rtcc.ret_from_eq); } static int cc_bw_decrease(struct sctp_tcb *stcb, struct sctp_nets *net, uint64_t nbw, uint64_t rtt_offset, uint64_t vtag, uint8_t inst_ind) { uint64_t oth, probepoint; /* Bandwidth decreased. */ probepoint = (((uint64_t) net->cwnd) << 32); if (net->rtt > net->cc_mod.rtcc.lbw_rtt + rtt_offset) { /* rtt increased */ /* Did we add more */ if ((net->cwnd > net->cc_mod.rtcc.cwnd_at_bw_set) && (inst_ind != SCTP_INST_LOOSING)) { /* We caused it maybe.. back off? */ /* PROBE POINT 1 */ probepoint |= ((1 << 16) | 1); SDT_PROBE(sctp, cwnd, net, rttvar, vtag, ((net->cc_mod.rtcc.lbw << 32) | nbw), ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), net->flight_size, probepoint); if (net->cc_mod.rtcc.ret_from_eq) { /* * Switch over to CA if we are less * aggressive */ net->ssthresh = net->cwnd - 1; net->partial_bytes_acked = 0; } return (1); } /* Probe point 2 */ probepoint |= ((2 << 16) | 0); SDT_PROBE(sctp, cwnd, net, rttvar, vtag, ((net->cc_mod.rtcc.lbw << 32) | nbw), ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), net->flight_size, probepoint); /* Someone else - fight for more? */ if (net->cc_mod.rtcc.steady_step) { oth = net->cc_mod.rtcc.vol_reduce; oth <<= 16; oth |= net->cc_mod.rtcc.step_cnt; oth <<= 16; oth |= net->cc_mod.rtcc.last_step_state; SDT_PROBE(sctp, cwnd, net, rttstep, vtag, ((net->cc_mod.rtcc.lbw << 32) | nbw), ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), oth, probepoint); /* * Did we voluntarily give up some? if so take one * back please */ if ((net->cc_mod.rtcc.vol_reduce) && (inst_ind != SCTP_INST_GAINING)) { net->cwnd += net->mtu; net->cc_mod.rtcc.vol_reduce--; } net->cc_mod.rtcc.last_step_state = 2; net->cc_mod.rtcc.step_cnt = 0; } goto out_decision; } else if (net->rtt < net->cc_mod.rtcc.lbw_rtt - rtt_offset) { /* bw & rtt decreased */ /* Probe point 3 */ probepoint |= ((3 << 16) | 0); SDT_PROBE(sctp, cwnd, net, rttvar, vtag, ((net->cc_mod.rtcc.lbw << 32) | nbw), ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), net->flight_size, probepoint); if (net->cc_mod.rtcc.steady_step) { oth = net->cc_mod.rtcc.vol_reduce; oth <<= 16; oth |= net->cc_mod.rtcc.step_cnt; oth <<= 16; oth |= net->cc_mod.rtcc.last_step_state; SDT_PROBE(sctp, cwnd, net, rttstep, vtag, ((net->cc_mod.rtcc.lbw << 32) | nbw), ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), oth, probepoint); if ((net->cc_mod.rtcc.vol_reduce) && (inst_ind != SCTP_INST_GAINING)) { net->cwnd += net->mtu; net->cc_mod.rtcc.vol_reduce--; } net->cc_mod.rtcc.last_step_state = 3; net->cc_mod.rtcc.step_cnt = 0; } goto out_decision; } /* The bw decreased but rtt stayed the same */ /* Probe point 4 */ probepoint |= ((4 << 16) | 0); SDT_PROBE(sctp, cwnd, net, rttvar, vtag, ((net->cc_mod.rtcc.lbw << 32) | nbw), ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), net->flight_size, probepoint); if (net->cc_mod.rtcc.steady_step) { oth = net->cc_mod.rtcc.vol_reduce; oth <<= 16; oth |= net->cc_mod.rtcc.step_cnt; oth <<= 16; oth |= net->cc_mod.rtcc.last_step_state; SDT_PROBE(sctp, cwnd, net, rttstep, vtag, ((net->cc_mod.rtcc.lbw << 32) | nbw), ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), oth, probepoint); if ((net->cc_mod.rtcc.vol_reduce) && (inst_ind != SCTP_INST_GAINING)) { net->cwnd += net->mtu; net->cc_mod.rtcc.vol_reduce--; } net->cc_mod.rtcc.last_step_state = 4; net->cc_mod.rtcc.step_cnt = 0; } out_decision: net->cc_mod.rtcc.lbw = nbw; net->cc_mod.rtcc.lbw_rtt = net->rtt; net->cc_mod.rtcc.cwnd_at_bw_set = net->cwnd; if (inst_ind == SCTP_INST_GAINING) { return (1); } else { return (0); } } static int cc_bw_increase(struct sctp_tcb *stcb, struct sctp_nets *net, uint64_t nbw, uint64_t vtag, uint8_t inst_ind) { uint64_t oth, probepoint; /* * BW increased, so update and return 0, since all actions in our * table say to do the normal CC update. Note that we pay no * attention to the inst_ind since our overall sum is increasing. */ /* PROBE POINT 0 */ probepoint = (((uint64_t) net->cwnd) << 32); SDT_PROBE(sctp, cwnd, net, rttvar, vtag, ((net->cc_mod.rtcc.lbw << 32) | nbw), ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), net->flight_size, probepoint); if (net->cc_mod.rtcc.steady_step) { oth = net->cc_mod.rtcc.vol_reduce; oth <<= 16; oth |= net->cc_mod.rtcc.step_cnt; oth <<= 16; oth |= net->cc_mod.rtcc.last_step_state; SDT_PROBE(sctp, cwnd, net, rttstep, vtag, ((net->cc_mod.rtcc.lbw << 32) | nbw), ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), oth, probepoint); net->cc_mod.rtcc.last_step_state = 0; net->cc_mod.rtcc.step_cnt = 0; net->cc_mod.rtcc.vol_reduce = 0; } net->cc_mod.rtcc.lbw = nbw; net->cc_mod.rtcc.lbw_rtt = net->rtt; net->cc_mod.rtcc.cwnd_at_bw_set = net->cwnd; return (0); } /* RTCC Algoritm to limit growth of cwnd, return * true if you want to NOT allow cwnd growth */ static int cc_bw_limit(struct sctp_tcb *stcb, struct sctp_nets *net, uint64_t nbw) { uint64_t bw_offset, rtt_offset, rtt, vtag, probepoint; uint64_t bytes_for_this_rtt, inst_bw; uint64_t div, inst_off; int bw_shift; uint8_t inst_ind; int ret; /*- * Here we need to see if we want * to limit cwnd growth due to increase * in overall rtt but no increase in bw. * We use the following table to figure * out what we should do. When we return * 0, cc update goes on as planned. If we * return 1, then no cc update happens and cwnd * stays where it is at. * ---------------------------------- * BW | RTT | Action * ********************************* * INC | INC | return 0 * ---------------------------------- * INC | SAME | return 0 * ---------------------------------- * INC | DECR | return 0 * ---------------------------------- * SAME | INC | return 1 * ---------------------------------- * SAME | SAME | return 1 * ---------------------------------- * SAME | DECR | return 0 * ---------------------------------- * DECR | INC | return 0 or 1 based on if we caused. * ---------------------------------- * DECR | SAME | return 0 * ---------------------------------- * DECR | DECR | return 0 * ---------------------------------- * * We are a bit fuzz on what an increase or * decrease is. For BW it is the same if * it did not change within 1/64th. For * RTT it stayed the same if it did not * change within 1/32nd */ bw_shift = SCTP_BASE_SYSCTL(sctp_rttvar_bw); rtt = stcb->asoc.my_vtag; vtag = (rtt << 32) | (((uint32_t) (stcb->sctp_ep->sctp_lport)) << 16) | (stcb->rport); probepoint = (((uint64_t) net->cwnd) << 32); rtt = net->rtt; if (net->cc_mod.rtcc.rtt_set_this_sack) { net->cc_mod.rtcc.rtt_set_this_sack = 0; bytes_for_this_rtt = net->cc_mod.rtcc.bw_bytes - net->cc_mod.rtcc.bw_bytes_at_last_rttc; net->cc_mod.rtcc.bw_bytes_at_last_rttc = net->cc_mod.rtcc.bw_bytes; if (net->rtt) { div = net->rtt / 1000; if (div) { inst_bw = bytes_for_this_rtt / div; inst_off = inst_bw >> bw_shift; if (inst_bw > nbw) inst_ind = SCTP_INST_GAINING; else if ((inst_bw + inst_off) < nbw) inst_ind = SCTP_INST_LOOSING; else inst_ind = SCTP_INST_NEUTRAL; probepoint |= ((0xb << 16) | inst_ind); } else { inst_bw = bytes_for_this_rtt / (uint64_t) (net->rtt); /* Can't determine do not change */ inst_ind = net->cc_mod.rtcc.last_inst_ind; probepoint |= ((0xc << 16) | inst_ind); } } else { inst_bw = bytes_for_this_rtt; /* Can't determine do not change */ inst_ind = net->cc_mod.rtcc.last_inst_ind; probepoint |= ((0xd << 16) | inst_ind); } SDT_PROBE(sctp, cwnd, net, rttvar, vtag, ((nbw << 32) | inst_bw), ((net->cc_mod.rtcc.lbw_rtt << 32) | rtt), net->flight_size, probepoint); } else { /* No rtt measurement, use last one */ inst_ind = net->cc_mod.rtcc.last_inst_ind; } bw_offset = net->cc_mod.rtcc.lbw >> bw_shift; if (nbw > net->cc_mod.rtcc.lbw + bw_offset) { ret = cc_bw_increase(stcb, net, nbw, vtag, inst_ind); goto out; } rtt_offset = net->cc_mod.rtcc.lbw_rtt >> SCTP_BASE_SYSCTL(sctp_rttvar_rtt); if (nbw < net->cc_mod.rtcc.lbw - bw_offset) { ret = cc_bw_decrease(stcb, net, nbw, rtt_offset, vtag, inst_ind); goto out; } /* * If we reach here then we are in a situation where the bw stayed * the same. */ ret = cc_bw_same(stcb, net, nbw, rtt_offset, vtag, inst_ind); out: net->cc_mod.rtcc.last_inst_ind = inst_ind; return (ret); } static void sctp_cwnd_update_after_sack_common(struct sctp_tcb *stcb, struct sctp_association *asoc, int accum_moved, int reneged_all, int will_exit, int use_rtcc) { struct sctp_nets *net; int old_cwnd; uint32_t t_ssthresh, t_cwnd, incr; + uint64_t t_ucwnd_sbw; + uint64_t t_path_mptcp; + uint64_t mptcp_like_alpha; + uint32_t srtt; + uint64_t max_path; /* MT FIXME: Don't compute this over and over again */ t_ssthresh = 0; t_cwnd = 0; - if (stcb->asoc.sctp_cmt_on_off == 2) { + t_ucwnd_sbw = 0; + t_path_mptcp = 0; + mptcp_like_alpha = 1; + if ((stcb->asoc.sctp_cmt_on_off == SCTP_CMT_RPV1) || + (stcb->asoc.sctp_cmt_on_off == SCTP_CMT_RPV2) || + (stcb->asoc.sctp_cmt_on_off == SCTP_CMT_MPTCP)) { + max_path = 0; TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { t_ssthresh += net->ssthresh; t_cwnd += net->cwnd; + /* lastsa>>3; we don't need to devide ... */ + srtt = net->lastsa; + if (srtt > 0) { + uint64_t tmp; + + t_ucwnd_sbw += (uint64_t) net->cwnd / (uint64_t) srtt; + t_path_mptcp += (((uint64_t) net->cwnd) << SHIFT_MPTCP_MULTI_Z) / + (((uint64_t) net->mtu) * (uint64_t) srtt); + tmp = (((uint64_t) net->cwnd) << SHIFT_MPTCP_MULTI_N) / + ((uint64_t) net->mtu * (uint64_t) (srtt * srtt)); + if (tmp > max_path) { + max_path = tmp; + } + } } + if (t_ucwnd_sbw == 0) { + t_ucwnd_sbw = 1; + } + if (t_path_mptcp > 0) { + mptcp_like_alpha = max_path / (t_path_mptcp * t_path_mptcp); + } else { + mptcp_like_alpha = 1; + } } /******************************/ /* update cwnd and Early FR */ /******************************/ TAILQ_FOREACH(net, &asoc->nets, sctp_next) { #ifdef JANA_CMT_FAST_RECOVERY /* * CMT fast recovery code. Need to debug. */ if (net->fast_retran_loss_recovery && net->new_pseudo_cumack) { if (SCTP_TSN_GE(asoc->last_acked_seq, net->fast_recovery_tsn) || SCTP_TSN_GE(net->pseudo_cumack, net->fast_recovery_tsn)) { net->will_exit_fast_recovery = 1; } } #endif if (SCTP_BASE_SYSCTL(sctp_early_fr)) { /* * So, first of all do we need to have a Early FR * timer running? */ if ((!TAILQ_EMPTY(&asoc->sent_queue) && (net->ref_count > 1) && (net->flight_size < net->cwnd)) || (reneged_all)) { /* * yes, so in this case stop it if its * running, and then restart it. Reneging * all is a special case where we want to * run the Early FR timer and then force the * last few unacked to be sent, causing us * to illicit a sack with gaps to force out * the others. */ if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) { SCTP_STAT_INCR(sctps_earlyfrstpidsck2); sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_INDATA + SCTP_LOC_20); } SCTP_STAT_INCR(sctps_earlyfrstrid); sctp_timer_start(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net); } else { /* No, stop it if its running */ if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) { SCTP_STAT_INCR(sctps_earlyfrstpidsck3); sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_INDATA + SCTP_LOC_21); } } } /* if nothing was acked on this destination skip it */ if (net->net_ack == 0) { if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_LOGGING_ENABLE) { sctp_log_cwnd(stcb, net, 0, SCTP_CWND_LOG_FROM_SACK); } continue; } if (net->net_ack2 > 0) { /* * Karn's rule applies to clearing error count, this * is optional. */ net->error_count = 0; if ((net->dest_state & SCTP_ADDR_NOT_REACHABLE) == SCTP_ADDR_NOT_REACHABLE) { /* addr came good */ net->dest_state &= ~SCTP_ADDR_NOT_REACHABLE; net->dest_state |= SCTP_ADDR_REACHABLE; sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_UP, stcb, SCTP_RECEIVED_SACK, (void *)net, SCTP_SO_NOT_LOCKED); /* now was it the primary? if so restore */ if (net->dest_state & SCTP_ADDR_WAS_PRIMARY) { (void)sctp_set_primary_addr(stcb, (struct sockaddr *)NULL, net); } } /* * JRS 5/14/07 - If CMT PF is on and the destination * is in PF state, set the destination to active * state and set the cwnd to one or two MTU's based * on whether PF1 or PF2 is being used. * * Should we stop any running T3 timer here? */ if ((asoc->sctp_cmt_on_off > 0) && (asoc->sctp_cmt_pf > 0) && ((net->dest_state & SCTP_ADDR_PF) == SCTP_ADDR_PF)) { net->dest_state &= ~SCTP_ADDR_PF; old_cwnd = net->cwnd; net->cwnd = net->mtu * asoc->sctp_cmt_pf; SDT_PROBE(sctp, cwnd, net, ack, stcb->asoc.my_vtag, ((stcb->sctp_ep->sctp_lport << 16) | (stcb->rport)), net, old_cwnd, net->cwnd); SCTPDBG(SCTP_DEBUG_INDATA1, "Destination %p moved from PF to reachable with cwnd %d.\n", net, net->cwnd); /* * Since the cwnd value is explicitly set, * skip the code that updates the cwnd * value. */ goto skip_cwnd_update; } } #ifdef JANA_CMT_FAST_RECOVERY /* * CMT fast recovery code */ /* * if (sctp_cmt_on_off > 0 && net->fast_retran_loss_recovery * && net->will_exit_fast_recovery == 0) { @@@ Do something * } else if (sctp_cmt_on_off == 0 && * asoc->fast_retran_loss_recovery && will_exit == 0) { */ #endif if (asoc->fast_retran_loss_recovery && (will_exit == 0) && (asoc->sctp_cmt_on_off == 0)) { /* * If we are in loss recovery we skip any cwnd * update */ goto skip_cwnd_update; } /* * Did any measurements go on for this network? */ if (use_rtcc && (net->cc_mod.rtcc.tls_needs_set > 0)) { uint64_t nbw; /* * At this point our bw_bytes has been updated by * incoming sack information. * * But our bw may not yet be set. * */ if ((net->cc_mod.rtcc.new_tot_time / 1000) > 0) { nbw = net->cc_mod.rtcc.bw_bytes / (net->cc_mod.rtcc.new_tot_time / 1000); } else { nbw = net->cc_mod.rtcc.bw_bytes; } if (net->cc_mod.rtcc.lbw) { if (cc_bw_limit(stcb, net, nbw)) { /* Hold here, no update */ goto skip_cwnd_update; } } else { uint64_t vtag, probepoint; probepoint = (((uint64_t) net->cwnd) << 32); probepoint |= ((0xa << 16) | 0); vtag = (net->rtt << 32) | (((uint32_t) (stcb->sctp_ep->sctp_lport)) << 16) | (stcb->rport); SDT_PROBE(sctp, cwnd, net, rttvar, vtag, nbw, ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), net->flight_size, probepoint); net->cc_mod.rtcc.lbw = nbw; net->cc_mod.rtcc.lbw_rtt = net->rtt; if (net->cc_mod.rtcc.rtt_set_this_sack) { net->cc_mod.rtcc.rtt_set_this_sack = 0; net->cc_mod.rtcc.bw_bytes_at_last_rttc = net->cc_mod.rtcc.bw_bytes; } } } /* * CMT: CUC algorithm. Update cwnd if pseudo-cumack has * moved. */ if (accum_moved || ((asoc->sctp_cmt_on_off > 0) && net->new_pseudo_cumack)) { /* If the cumulative ack moved we can proceed */ if (net->cwnd <= net->ssthresh) { /* We are in slow start */ if (net->flight_size + net->net_ack >= net->cwnd) { - old_cwnd = net->cwnd; - if (stcb->asoc.sctp_cmt_on_off == 2) { - uint32_t limit; + uint32_t limit; + old_cwnd = net->cwnd; + switch (asoc->sctp_cmt_on_off) { + case SCTP_CMT_RPV1: limit = (uint32_t) (((uint64_t) net->mtu * (uint64_t) SCTP_BASE_SYSCTL(sctp_L2_abc_variable) * (uint64_t) net->ssthresh) / (uint64_t) t_ssthresh); incr = (uint32_t) (((uint64_t) net->net_ack * (uint64_t) net->ssthresh) / (uint64_t) t_ssthresh); if (incr > limit) { incr = limit; } if (incr == 0) { incr = 1; } - } else { + break; + case SCTP_CMT_RPV2: + /* + * lastsa>>3; we don't need + * to divide ... + */ + srtt = net->lastsa; + if (srtt == 0) { + srtt = 1; + } + limit = (uint32_t) (((uint64_t) net->mtu * + (uint64_t) SCTP_BASE_SYSCTL(sctp_L2_abc_variable) * + (uint64_t) net->cwnd) / + ((uint64_t) srtt * t_ucwnd_sbw)); + /* INCREASE FACTOR */ + incr = (uint32_t) (((uint64_t) net->net_ack * + (uint64_t) net->cwnd) / + ((uint64_t) srtt * t_ucwnd_sbw)); + /* INCREASE FACTOR */ + if (incr > limit) { + incr = limit; + } + if (incr == 0) { + incr = 1; + } + break; + case SCTP_CMT_MPTCP: + limit = (uint32_t) (((uint64_t) net->mtu * + mptcp_like_alpha * + (uint64_t) SCTP_BASE_SYSCTL(sctp_L2_abc_variable)) >> + SHIFT_MPTCP_MULTI); + incr = (uint32_t) (((uint64_t) net->net_ack * + mptcp_like_alpha) >> + SHIFT_MPTCP_MULTI); + if (incr > limit) { + incr = limit; + } + if (incr > net->net_ack) { + incr = net->net_ack; + } + if (incr > net->mtu) { + incr = net->mtu; + } + break; + default: incr = net->net_ack; if (incr > net->mtu * SCTP_BASE_SYSCTL(sctp_L2_abc_variable)) { incr = net->mtu * SCTP_BASE_SYSCTL(sctp_L2_abc_variable); } + break; } net->cwnd += incr; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, incr, SCTP_CWND_LOG_FROM_SS); } SDT_PROBE(sctp, cwnd, net, ack, stcb->asoc.my_vtag, ((stcb->sctp_ep->sctp_lport << 16) | (stcb->rport)), net, old_cwnd, net->cwnd); } else { if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_LOGGING_ENABLE) { sctp_log_cwnd(stcb, net, net->net_ack, SCTP_CWND_LOG_NOADV_SS); } } } else { /* We are in congestion avoidance */ /* * Add to pba */ net->partial_bytes_acked += net->net_ack; if ((net->flight_size + net->net_ack >= net->cwnd) && (net->partial_bytes_acked >= net->cwnd)) { net->partial_bytes_acked -= net->cwnd; old_cwnd = net->cwnd; - if (asoc->sctp_cmt_on_off == 2) { + switch (asoc->sctp_cmt_on_off) { + case SCTP_CMT_RPV1: incr = (uint32_t) (((uint64_t) net->mtu * (uint64_t) net->ssthresh) / (uint64_t) t_ssthresh); if (incr == 0) { incr = 1; } - } else { + break; + case SCTP_CMT_RPV2: + /* + * lastsa>>3; we don't need + * to divide ... + */ + srtt = net->lastsa; + if (srtt == 0) { + srtt = 1; + } + incr = (uint32_t) ((uint64_t) net->mtu * + (uint64_t) net->cwnd / + ((uint64_t) srtt * + t_ucwnd_sbw)); + /* INCREASE FACTOR */ + if (incr == 0) { + incr = 1; + } + break; + case SCTP_CMT_MPTCP: + incr = (uint32_t) ((mptcp_like_alpha * + (uint64_t) net->cwnd) >> + SHIFT_MPTCP_MULTI); + if (incr > net->mtu) { + incr = net->mtu; + } + break; + default: incr = net->mtu; + break; } net->cwnd += incr; SDT_PROBE(sctp, cwnd, net, ack, stcb->asoc.my_vtag, ((stcb->sctp_ep->sctp_lport << 16) | (stcb->rport)), net, old_cwnd, net->cwnd); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, net->mtu, SCTP_CWND_LOG_FROM_CA); } } else { if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_LOGGING_ENABLE) { sctp_log_cwnd(stcb, net, net->net_ack, SCTP_CWND_LOG_NOADV_CA); } } } } else { if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_LOGGING_ENABLE) { sctp_log_cwnd(stcb, net, net->mtu, SCTP_CWND_LOG_NO_CUMACK); } } skip_cwnd_update: /* * NOW, according to Karn's rule do we need to restore the * RTO timer back? Check our net_ack2. If not set then we * have a ambiguity.. i.e. all data ack'd was sent to more * than one place. */ if (net->net_ack2) { /* restore any doubled timers */ net->RTO = (net->lastsa >> SCTP_RTT_SHIFT) + net->lastsv; if (net->RTO < stcb->asoc.minrto) { net->RTO = stcb->asoc.minrto; } if (net->RTO > stcb->asoc.maxrto) { net->RTO = stcb->asoc.maxrto; } } } } static void sctp_cwnd_update_after_timeout(struct sctp_tcb *stcb, struct sctp_nets *net) { int old_cwnd = net->cwnd; uint32_t t_ssthresh, t_cwnd; + uint64_t t_ucwnd_sbw; /* MT FIXME: Don't compute this over and over again */ t_ssthresh = 0; t_cwnd = 0; - if (stcb->asoc.sctp_cmt_on_off == 2) { + if ((stcb->asoc.sctp_cmt_on_off == SCTP_CMT_RPV1) || + (stcb->asoc.sctp_cmt_on_off == SCTP_CMT_RPV2)) { struct sctp_nets *lnet; + uint32_t srtt; + t_ucwnd_sbw = 0; TAILQ_FOREACH(lnet, &stcb->asoc.nets, sctp_next) { t_ssthresh += lnet->ssthresh; t_cwnd += lnet->cwnd; + srtt = lnet->lastsa; + /* lastsa>>3; we don't need to divide ... */ + if (srtt > 0) { + t_ucwnd_sbw += (uint64_t) lnet->cwnd / (uint64_t) srtt; + } } - net->ssthresh = (uint32_t) (((uint64_t) 4 * - (uint64_t) net->mtu * - (uint64_t) net->ssthresh) / - (uint64_t) t_ssthresh); + if (t_ucwnd_sbw < 1) { + t_ucwnd_sbw = 1; + } + if (stcb->asoc.sctp_cmt_on_off == SCTP_CMT_RPV1) { + net->ssthresh = (uint32_t) (((uint64_t) 4 * + (uint64_t) net->mtu * + (uint64_t) net->ssthresh) / + (uint64_t) t_ssthresh); + } else { + uint64_t cc_delta; + + srtt = net->lastsa; + /* lastsa>>3; we don't need to divide ... */ + if (srtt == 0) { + srtt = 1; + } + cc_delta = t_ucwnd_sbw * (uint64_t) srtt / 2; + if (cc_delta < t_cwnd) { + net->ssthresh = (uint32_t) ((uint64_t) t_cwnd - cc_delta); + } else { + net->ssthresh = net->mtu; + } + } if ((net->cwnd > t_cwnd / 2) && (net->ssthresh < net->cwnd - t_cwnd / 2)) { net->ssthresh = net->cwnd - t_cwnd / 2; } if (net->ssthresh < net->mtu) { net->ssthresh = net->mtu; } } else { net->ssthresh = max(net->cwnd / 2, 4 * net->mtu); } net->cwnd = net->mtu; net->partial_bytes_acked = 0; SDT_PROBE(sctp, cwnd, net, to, stcb->asoc.my_vtag, ((stcb->sctp_ep->sctp_lport << 16) | (stcb->rport)), net, old_cwnd, net->cwnd); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, net->cwnd - old_cwnd, SCTP_CWND_LOG_FROM_RTX); } } static void sctp_cwnd_update_after_ecn_echo_common(struct sctp_tcb *stcb, struct sctp_nets *net, int in_window, int num_pkt_lost, int use_rtcc) { int old_cwnd = net->cwnd; if ((use_rtcc) && (net->lan_type == SCTP_LAN_LOCAL) && (net->cc_mod.rtcc.use_dccc_ecn)) { /* Data center Congestion Control */ if (in_window == 0) { /* * Go to CA with the cwnd at the point we sent the * TSN that was marked with a CE. */ if (net->ecn_prev_cwnd < net->cwnd) { /* Restore to prev cwnd */ net->cwnd = net->ecn_prev_cwnd - (net->mtu * num_pkt_lost); } else { /* Just cut in 1/2 */ net->cwnd /= 2; } /* Drop to CA */ net->ssthresh = net->cwnd - (num_pkt_lost * net->mtu); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_SAT); } } else { /* * Further tuning down required over the drastic * orginal cut */ net->ssthresh -= (net->mtu * num_pkt_lost); net->cwnd -= (net->mtu * num_pkt_lost); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_SAT); } } SCTP_STAT_INCR(sctps_ecnereducedcwnd); } else { if (in_window == 0) { SCTP_STAT_INCR(sctps_ecnereducedcwnd); net->ssthresh = net->cwnd / 2; if (net->ssthresh < net->mtu) { net->ssthresh = net->mtu; /* * here back off the timer as well, to slow * us down */ net->RTO <<= 1; } net->cwnd = net->ssthresh; SDT_PROBE(sctp, cwnd, net, ecn, stcb->asoc.my_vtag, ((stcb->sctp_ep->sctp_lport << 16) | (stcb->rport)), net, old_cwnd, net->cwnd); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_SAT); } } } } static void sctp_cwnd_update_after_packet_dropped(struct sctp_tcb *stcb, struct sctp_nets *net, struct sctp_pktdrop_chunk *cp, uint32_t * bottle_bw, uint32_t * on_queue) { uint32_t bw_avail; int rtt; unsigned int incr; int old_cwnd = net->cwnd; /* need real RTT in msd for this calc */ rtt = net->rtt / 1000; /* get bottle neck bw */ *bottle_bw = ntohl(cp->bottle_bw); /* and whats on queue */ *on_queue = ntohl(cp->current_onq); /* * adjust the on-queue if our flight is more it could be that the * router has not yet gotten data "in-flight" to it */ if (*on_queue < net->flight_size) *on_queue = net->flight_size; /* calculate the available space */ bw_avail = (*bottle_bw * rtt) / 1000; if (bw_avail > *bottle_bw) { /* * Cap the growth to no more than the bottle neck. This can * happen as RTT slides up due to queues. It also means if * you have more than a 1 second RTT with a empty queue you * will be limited to the bottle_bw per second no matter if * other points have 1/2 the RTT and you could get more * out... */ bw_avail = *bottle_bw; } if (*on_queue > bw_avail) { /* * No room for anything else don't allow anything else to be * "added to the fire". */ int seg_inflight, seg_onqueue, my_portion; net->partial_bytes_acked = 0; /* how much are we over queue size? */ incr = *on_queue - bw_avail; if (stcb->asoc.seen_a_sack_this_pkt) { /* * undo any cwnd adjustment that the sack might have * made */ net->cwnd = net->prev_cwnd; } /* Now how much of that is mine? */ seg_inflight = net->flight_size / net->mtu; seg_onqueue = *on_queue / net->mtu; my_portion = (incr * seg_inflight) / seg_onqueue; /* Have I made an adjustment already */ if (net->cwnd > net->flight_size) { /* * for this flight I made an adjustment we need to * decrease the portion by a share our previous * adjustment. */ int diff_adj; diff_adj = net->cwnd - net->flight_size; if (diff_adj > my_portion) my_portion = 0; else my_portion -= diff_adj; } /* * back down to the previous cwnd (assume we have had a sack * before this packet). minus what ever portion of the * overage is my fault. */ net->cwnd -= my_portion; /* we will NOT back down more than 1 MTU */ if (net->cwnd <= net->mtu) { net->cwnd = net->mtu; } /* force into CA */ net->ssthresh = net->cwnd - 1; } else { /* * Take 1/4 of the space left or max burst up .. whichever * is less. */ incr = (bw_avail - *on_queue) >> 2; if ((stcb->asoc.max_burst > 0) && (stcb->asoc.max_burst * net->mtu < incr)) { incr = stcb->asoc.max_burst * net->mtu; } net->cwnd += incr; } if (net->cwnd > bw_avail) { /* We can't exceed the pipe size */ net->cwnd = bw_avail; } if (net->cwnd < net->mtu) { /* We always have 1 MTU */ net->cwnd = net->mtu; } if (net->cwnd - old_cwnd != 0) { /* log only changes */ SDT_PROBE(sctp, cwnd, net, pd, stcb->asoc.my_vtag, ((stcb->sctp_ep->sctp_lport << 16) | (stcb->rport)), net, old_cwnd, net->cwnd); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_SAT); } } } static void sctp_cwnd_update_after_output(struct sctp_tcb *stcb, struct sctp_nets *net, int burst_limit) { int old_cwnd = net->cwnd; if (net->ssthresh < net->cwnd) net->ssthresh = net->cwnd; if (burst_limit) { net->cwnd = (net->flight_size + (burst_limit * net->mtu)); SDT_PROBE(sctp, cwnd, net, bl, stcb->asoc.my_vtag, ((stcb->sctp_ep->sctp_lport << 16) | (stcb->rport)), net, old_cwnd, net->cwnd); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_BRST); } } } static void sctp_cwnd_update_after_fr_timer(struct sctp_inpcb *inp, struct sctp_tcb *stcb, struct sctp_nets *net) { int old_cwnd = net->cwnd; sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_EARLY_FR_TMR, SCTP_SO_NOT_LOCKED); /* * make a small adjustment to cwnd and force to CA. */ if (net->cwnd > net->mtu) /* drop down one MTU after sending */ net->cwnd -= net->mtu; if (net->cwnd < net->ssthresh) /* still in SS move to CA */ net->ssthresh = net->cwnd - 1; SDT_PROBE(sctp, cwnd, net, fr, stcb->asoc.my_vtag, ((stcb->sctp_ep->sctp_lport << 16) | (stcb->rport)), net, old_cwnd, net->cwnd); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, (old_cwnd - net->cwnd), SCTP_CWND_LOG_FROM_FR); } } static void sctp_cwnd_update_after_sack(struct sctp_tcb *stcb, struct sctp_association *asoc, int accum_moved, int reneged_all, int will_exit) { /* Passing a zero argument in last disables the rtcc algoritm */ sctp_cwnd_update_after_sack_common(stcb, asoc, accum_moved, reneged_all, will_exit, 0); } static void sctp_cwnd_update_after_ecn_echo(struct sctp_tcb *stcb, struct sctp_nets *net, int in_window, int num_pkt_lost) { /* Passing a zero argument in last disables the rtcc algoritm */ sctp_cwnd_update_after_ecn_echo_common(stcb, net, in_window, num_pkt_lost, 0); } /* Here starts the RTCCVAR type CC invented by RRS which * is a slight mod to RFC2581. We reuse a common routine or * two since these algoritms are so close and need to * remain the same. */ static void sctp_cwnd_update_rtcc_after_ecn_echo(struct sctp_tcb *stcb, struct sctp_nets *net, int in_window, int num_pkt_lost) { sctp_cwnd_update_after_ecn_echo_common(stcb, net, in_window, num_pkt_lost, 1); } static void sctp_cwnd_update_rtcc_tsn_acknowledged(struct sctp_nets *net, struct sctp_tmit_chunk *tp1) { net->cc_mod.rtcc.bw_bytes += tp1->send_size; } static void sctp_cwnd_prepare_rtcc_net_for_sack(struct sctp_tcb *stcb, struct sctp_nets *net) { if (net->cc_mod.rtcc.tls_needs_set > 0) { /* We had a bw measurment going on */ struct timeval ltls; SCTP_GETPTIME_TIMEVAL(<ls); timevalsub(<ls, &net->cc_mod.rtcc.tls); net->cc_mod.rtcc.new_tot_time = (ltls.tv_sec * 1000000) + ltls.tv_usec; } } static void sctp_cwnd_new_rtcc_transmission_begins(struct sctp_tcb *stcb, struct sctp_nets *net) { uint64_t vtag, probepoint; if (net->cc_mod.rtcc.lbw) { /* Clear the old bw.. we went to 0 in-flight */ vtag = (net->rtt << 32) | (((uint32_t) (stcb->sctp_ep->sctp_lport)) << 16) | (stcb->rport); probepoint = (((uint64_t) net->cwnd) << 32); /* Probe point 8 */ probepoint |= ((8 << 16) | 0); SDT_PROBE(sctp, cwnd, net, rttvar, vtag, ((net->cc_mod.rtcc.lbw << 32) | 0), ((net->cc_mod.rtcc.lbw_rtt << 32) | net->rtt), net->flight_size, probepoint); net->cc_mod.rtcc.lbw_rtt = 0; net->cc_mod.rtcc.cwnd_at_bw_set = 0; net->cc_mod.rtcc.lbw = 0; net->cc_mod.rtcc.bw_bytes_at_last_rttc = 0; net->cc_mod.rtcc.vol_reduce = 0; net->cc_mod.rtcc.bw_tot_time = 0; net->cc_mod.rtcc.bw_bytes = 0; net->cc_mod.rtcc.tls_needs_set = 0; if (net->cc_mod.rtcc.steady_step) { net->cc_mod.rtcc.vol_reduce = 0; net->cc_mod.rtcc.step_cnt = 0; net->cc_mod.rtcc.last_step_state = 0; } if (net->cc_mod.rtcc.ret_from_eq) { /* less aggressive one - reset cwnd too */ uint32_t cwnd_in_mtu, cwnd; cwnd_in_mtu = SCTP_BASE_SYSCTL(sctp_initial_cwnd); if (cwnd_in_mtu == 0) { /* * Using 0 means that the value of RFC 4960 * is used. */ cwnd = min((net->mtu * 4), max((2 * net->mtu), SCTP_INITIAL_CWND)); } else { /* * We take the minimum of the burst limit * and the initial congestion window. */ if ((stcb->asoc.max_burst > 0) && (cwnd_in_mtu > stcb->asoc.max_burst)) cwnd_in_mtu = stcb->asoc.max_burst; cwnd = (net->mtu - sizeof(struct sctphdr)) * cwnd_in_mtu; } if (net->cwnd > cwnd) { /* * Only set if we are not a timeout (i.e. * down to 1 mtu) */ net->cwnd = cwnd; } } } } static void sctp_set_rtcc_initial_cc_param(struct sctp_tcb *stcb, struct sctp_nets *net) { uint64_t vtag, probepoint; sctp_set_initial_cc_param(stcb, net); stcb->asoc.use_precise_time = 1; probepoint = (((uint64_t) net->cwnd) << 32); probepoint |= ((9 << 16) | 0); vtag = (net->rtt << 32) | (((uint32_t) (stcb->sctp_ep->sctp_lport)) << 16) | (stcb->rport); SDT_PROBE(sctp, cwnd, net, rttvar, vtag, 0, 0, 0, probepoint); net->cc_mod.rtcc.lbw_rtt = 0; net->cc_mod.rtcc.cwnd_at_bw_set = 0; net->cc_mod.rtcc.vol_reduce = 0; net->cc_mod.rtcc.lbw = 0; net->cc_mod.rtcc.vol_reduce = 0; net->cc_mod.rtcc.bw_bytes_at_last_rttc = 0; net->cc_mod.rtcc.bw_tot_time = 0; net->cc_mod.rtcc.bw_bytes = 0; net->cc_mod.rtcc.tls_needs_set = 0; net->cc_mod.rtcc.ret_from_eq = SCTP_BASE_SYSCTL(sctp_rttvar_eqret); net->cc_mod.rtcc.steady_step = SCTP_BASE_SYSCTL(sctp_steady_step); net->cc_mod.rtcc.use_dccc_ecn = SCTP_BASE_SYSCTL(sctp_use_dccc_ecn); net->cc_mod.rtcc.step_cnt = 0; net->cc_mod.rtcc.last_step_state = 0; } static int sctp_cwnd_rtcc_socket_option(struct sctp_tcb *stcb, int setorget, struct sctp_cc_option *cc_opt) { struct sctp_nets *net; if (setorget == 1) { /* a set */ if (cc_opt->option == SCTP_CC_OPT_RTCC_SETMODE) { if ((cc_opt->aid_value.assoc_value != 0) && (cc_opt->aid_value.assoc_value != 1)) { return (EINVAL); } TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { net->cc_mod.rtcc.ret_from_eq = cc_opt->aid_value.assoc_value; } } else if (cc_opt->option == SCTP_CC_OPT_USE_DCCC_ECN) { if ((cc_opt->aid_value.assoc_value != 0) && (cc_opt->aid_value.assoc_value != 1)) { return (EINVAL); } TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { net->cc_mod.rtcc.use_dccc_ecn = cc_opt->aid_value.assoc_value; } } else if (cc_opt->option == SCTP_CC_OPT_STEADY_STEP) { TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { net->cc_mod.rtcc.steady_step = cc_opt->aid_value.assoc_value; } } else { return (EINVAL); } } else { /* a get */ if (cc_opt->option == SCTP_CC_OPT_RTCC_SETMODE) { net = TAILQ_FIRST(&stcb->asoc.nets); if (net == NULL) { return (EFAULT); } cc_opt->aid_value.assoc_value = net->cc_mod.rtcc.ret_from_eq; } else if (cc_opt->option == SCTP_CC_OPT_USE_DCCC_ECN) { net = TAILQ_FIRST(&stcb->asoc.nets); if (net == NULL) { return (EFAULT); } cc_opt->aid_value.assoc_value = net->cc_mod.rtcc.use_dccc_ecn; } else if (cc_opt->option == SCTP_CC_OPT_STEADY_STEP) { net = TAILQ_FIRST(&stcb->asoc.nets); if (net == NULL) { return (EFAULT); } cc_opt->aid_value.assoc_value = net->cc_mod.rtcc.steady_step; } else { return (EINVAL); } } return (0); } static void sctp_cwnd_update_rtcc_packet_transmitted(struct sctp_tcb *stcb, struct sctp_nets *net) { if (net->cc_mod.rtcc.tls_needs_set == 0) { SCTP_GETPTIME_TIMEVAL(&net->cc_mod.rtcc.tls); net->cc_mod.rtcc.tls_needs_set = 2; } } static void sctp_cwnd_update_rtcc_after_sack(struct sctp_tcb *stcb, struct sctp_association *asoc, int accum_moved, int reneged_all, int will_exit) { /* Passing a one argument at the last enables the rtcc algoritm */ sctp_cwnd_update_after_sack_common(stcb, asoc, accum_moved, reneged_all, will_exit, 1); } static void sctp_rtt_rtcc_calculated(struct sctp_tcb *stcb, struct sctp_nets *net, struct timeval *now) { net->cc_mod.rtcc.rtt_set_this_sack = 1; } /* Here starts Sally Floyds HS-TCP */ struct sctp_hs_raise_drop { int32_t cwnd; int32_t increase; int32_t drop_percent; }; #define SCTP_HS_TABLE_SIZE 73 struct sctp_hs_raise_drop sctp_cwnd_adjust[SCTP_HS_TABLE_SIZE] = { {38, 1, 50}, /* 0 */ {118, 2, 44}, /* 1 */ {221, 3, 41}, /* 2 */ {347, 4, 38}, /* 3 */ {495, 5, 37}, /* 4 */ {663, 6, 35}, /* 5 */ {851, 7, 34}, /* 6 */ {1058, 8, 33}, /* 7 */ {1284, 9, 32}, /* 8 */ {1529, 10, 31}, /* 9 */ {1793, 11, 30}, /* 10 */ {2076, 12, 29}, /* 11 */ {2378, 13, 28}, /* 12 */ {2699, 14, 28}, /* 13 */ {3039, 15, 27}, /* 14 */ {3399, 16, 27}, /* 15 */ {3778, 17, 26}, /* 16 */ {4177, 18, 26}, /* 17 */ {4596, 19, 25}, /* 18 */ {5036, 20, 25}, /* 19 */ {5497, 21, 24}, /* 20 */ {5979, 22, 24}, /* 21 */ {6483, 23, 23}, /* 22 */ {7009, 24, 23}, /* 23 */ {7558, 25, 22}, /* 24 */ {8130, 26, 22}, /* 25 */ {8726, 27, 22}, /* 26 */ {9346, 28, 21}, /* 27 */ {9991, 29, 21}, /* 28 */ {10661, 30, 21}, /* 29 */ {11358, 31, 20}, /* 30 */ {12082, 32, 20}, /* 31 */ {12834, 33, 20}, /* 32 */ {13614, 34, 19}, /* 33 */ {14424, 35, 19}, /* 34 */ {15265, 36, 19}, /* 35 */ {16137, 37, 19}, /* 36 */ {17042, 38, 18}, /* 37 */ {17981, 39, 18}, /* 38 */ {18955, 40, 18}, /* 39 */ {19965, 41, 17}, /* 40 */ {21013, 42, 17}, /* 41 */ {22101, 43, 17}, /* 42 */ {23230, 44, 17}, /* 43 */ {24402, 45, 16}, /* 44 */ {25618, 46, 16}, /* 45 */ {26881, 47, 16}, /* 46 */ {28193, 48, 16}, /* 47 */ {29557, 49, 15}, /* 48 */ {30975, 50, 15}, /* 49 */ {32450, 51, 15}, /* 50 */ {33986, 52, 15}, /* 51 */ {35586, 53, 14}, /* 52 */ {37253, 54, 14}, /* 53 */ {38992, 55, 14}, /* 54 */ {40808, 56, 14}, /* 55 */ {42707, 57, 13}, /* 56 */ {44694, 58, 13}, /* 57 */ {46776, 59, 13}, /* 58 */ {48961, 60, 13}, /* 59 */ {51258, 61, 13}, /* 60 */ {53677, 62, 12}, /* 61 */ {56230, 63, 12}, /* 62 */ {58932, 64, 12}, /* 63 */ {61799, 65, 12}, /* 64 */ {64851, 66, 11}, /* 65 */ {68113, 67, 11}, /* 66 */ {71617, 68, 11}, /* 67 */ {75401, 69, 10}, /* 68 */ {79517, 70, 10}, /* 69 */ {84035, 71, 10}, /* 70 */ {89053, 72, 10}, /* 71 */ {94717, 73, 9} /* 72 */ }; static void sctp_hs_cwnd_increase(struct sctp_tcb *stcb, struct sctp_nets *net) { int cur_val, i, indx, incr; cur_val = net->cwnd >> 10; indx = SCTP_HS_TABLE_SIZE - 1; #ifdef SCTP_DEBUG printf("HS CC CAlled.\n"); #endif if (cur_val < sctp_cwnd_adjust[0].cwnd) { /* normal mode */ if (net->net_ack > net->mtu) { net->cwnd += net->mtu; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, net->mtu, SCTP_CWND_LOG_FROM_SS); } } else { net->cwnd += net->net_ack; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, net->net_ack, SCTP_CWND_LOG_FROM_SS); } } } else { for (i = net->last_hs_used; i < SCTP_HS_TABLE_SIZE; i++) { if (cur_val < sctp_cwnd_adjust[i].cwnd) { indx = i; break; } } net->last_hs_used = indx; incr = ((sctp_cwnd_adjust[indx].increase) << 10); net->cwnd += incr; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, incr, SCTP_CWND_LOG_FROM_SS); } } } static void sctp_hs_cwnd_decrease(struct sctp_tcb *stcb, struct sctp_nets *net) { int cur_val, i, indx; int old_cwnd = net->cwnd; cur_val = net->cwnd >> 10; if (cur_val < sctp_cwnd_adjust[0].cwnd) { /* normal mode */ net->ssthresh = net->cwnd / 2; if (net->ssthresh < (net->mtu * 2)) { net->ssthresh = 2 * net->mtu; } net->cwnd = net->ssthresh; } else { /* drop by the proper amount */ net->ssthresh = net->cwnd - (int)((net->cwnd / 100) * sctp_cwnd_adjust[net->last_hs_used].drop_percent); net->cwnd = net->ssthresh; /* now where are we */ indx = net->last_hs_used; cur_val = net->cwnd >> 10; /* reset where we are in the table */ if (cur_val < sctp_cwnd_adjust[0].cwnd) { /* feel out of hs */ net->last_hs_used = 0; } else { for (i = indx; i >= 1; i--) { if (cur_val > sctp_cwnd_adjust[i - 1].cwnd) { break; } } net->last_hs_used = indx; } } if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_FR); } } static void sctp_hs_cwnd_update_after_fr(struct sctp_tcb *stcb, struct sctp_association *asoc) { struct sctp_nets *net; /* * CMT fast recovery code. Need to debug. ((sctp_cmt_on_off > 0) && * (net->fast_retran_loss_recovery == 0))) */ TAILQ_FOREACH(net, &asoc->nets, sctp_next) { if ((asoc->fast_retran_loss_recovery == 0) || (asoc->sctp_cmt_on_off > 0)) { /* out of a RFC2582 Fast recovery window? */ if (net->net_ack > 0) { /* * per section 7.2.3, are there any * destinations that had a fast retransmit * to them. If so what we need to do is * adjust ssthresh and cwnd. */ struct sctp_tmit_chunk *lchk; sctp_hs_cwnd_decrease(stcb, net); lchk = TAILQ_FIRST(&asoc->send_queue); net->partial_bytes_acked = 0; /* Turn on fast recovery window */ asoc->fast_retran_loss_recovery = 1; if (lchk == NULL) { /* Mark end of the window */ asoc->fast_recovery_tsn = asoc->sending_seq - 1; } else { asoc->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1; } /* * CMT fast recovery -- per destination * recovery variable. */ net->fast_retran_loss_recovery = 1; if (lchk == NULL) { /* Mark end of the window */ net->fast_recovery_tsn = asoc->sending_seq - 1; } else { net->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1; } sctp_timer_stop(SCTP_TIMER_TYPE_SEND, stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_INDATA + SCTP_LOC_32); sctp_timer_start(SCTP_TIMER_TYPE_SEND, stcb->sctp_ep, stcb, net); } } else if (net->net_ack > 0) { /* * Mark a peg that we WOULD have done a cwnd * reduction but RFC2582 prevented this action. */ SCTP_STAT_INCR(sctps_fastretransinrtt); } } } static void sctp_hs_cwnd_update_after_sack(struct sctp_tcb *stcb, struct sctp_association *asoc, int accum_moved, int reneged_all, int will_exit) { struct sctp_nets *net; /******************************/ /* update cwnd and Early FR */ /******************************/ TAILQ_FOREACH(net, &asoc->nets, sctp_next) { #ifdef JANA_CMT_FAST_RECOVERY /* * CMT fast recovery code. Need to debug. */ if (net->fast_retran_loss_recovery && net->new_pseudo_cumack) { if (SCTP_TSN_GE(asoc->last_acked_seq, net->fast_recovery_tsn) || SCTP_TSN_GE(net->pseudo_cumack, net->fast_recovery_tsn)) { net->will_exit_fast_recovery = 1; } } #endif if (SCTP_BASE_SYSCTL(sctp_early_fr)) { /* * So, first of all do we need to have a Early FR * timer running? */ if ((!TAILQ_EMPTY(&asoc->sent_queue) && (net->ref_count > 1) && (net->flight_size < net->cwnd)) || (reneged_all)) { /* * yes, so in this case stop it if its * running, and then restart it. Reneging * all is a special case where we want to * run the Early FR timer and then force the * last few unacked to be sent, causing us * to illicit a sack with gaps to force out * the others. */ if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) { SCTP_STAT_INCR(sctps_earlyfrstpidsck2); sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_INDATA + SCTP_LOC_20); } SCTP_STAT_INCR(sctps_earlyfrstrid); sctp_timer_start(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net); } else { /* No, stop it if its running */ if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) { SCTP_STAT_INCR(sctps_earlyfrstpidsck3); sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_INDATA + SCTP_LOC_21); } } } /* if nothing was acked on this destination skip it */ if (net->net_ack == 0) { if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_LOGGING_ENABLE) { sctp_log_cwnd(stcb, net, 0, SCTP_CWND_LOG_FROM_SACK); } continue; } if (net->net_ack2 > 0) { /* * Karn's rule applies to clearing error count, this * is optional. */ net->error_count = 0; if ((net->dest_state & SCTP_ADDR_NOT_REACHABLE) == SCTP_ADDR_NOT_REACHABLE) { /* addr came good */ net->dest_state &= ~SCTP_ADDR_NOT_REACHABLE; net->dest_state |= SCTP_ADDR_REACHABLE; sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_UP, stcb, SCTP_RECEIVED_SACK, (void *)net, SCTP_SO_NOT_LOCKED); /* now was it the primary? if so restore */ if (net->dest_state & SCTP_ADDR_WAS_PRIMARY) { (void)sctp_set_primary_addr(stcb, (struct sockaddr *)NULL, net); } } /* * JRS 5/14/07 - If CMT PF is on and the destination * is in PF state, set the destination to active * state and set the cwnd to one or two MTU's based * on whether PF1 or PF2 is being used. * * Should we stop any running T3 timer here? */ if ((asoc->sctp_cmt_on_off > 0) && (asoc->sctp_cmt_pf > 0) && ((net->dest_state & SCTP_ADDR_PF) == SCTP_ADDR_PF)) { net->dest_state &= ~SCTP_ADDR_PF; net->cwnd = net->mtu * asoc->sctp_cmt_pf; SCTPDBG(SCTP_DEBUG_INDATA1, "Destination %p moved from PF to reachable with cwnd %d.\n", net, net->cwnd); /* * Since the cwnd value is explicitly set, * skip the code that updates the cwnd * value. */ goto skip_cwnd_update; } } #ifdef JANA_CMT_FAST_RECOVERY /* * CMT fast recovery code */ /* * if (sctp_cmt_on_off > 0 && net->fast_retran_loss_recovery * && net->will_exit_fast_recovery == 0) { @@@ Do something * } else if (sctp_cmt_on_off == 0 && * asoc->fast_retran_loss_recovery && will_exit == 0) { */ #endif if (asoc->fast_retran_loss_recovery && (will_exit == 0) && (asoc->sctp_cmt_on_off == 0)) { /* * If we are in loss recovery we skip any cwnd * update */ goto skip_cwnd_update; } /* * CMT: CUC algorithm. Update cwnd if pseudo-cumack has * moved. */ if (accum_moved || ((asoc->sctp_cmt_on_off > 0) && net->new_pseudo_cumack)) { /* If the cumulative ack moved we can proceed */ if (net->cwnd <= net->ssthresh) { /* We are in slow start */ if (net->flight_size + net->net_ack >= net->cwnd) { sctp_hs_cwnd_increase(stcb, net); } else { if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_LOGGING_ENABLE) { sctp_log_cwnd(stcb, net, net->net_ack, SCTP_CWND_LOG_NOADV_SS); } } } else { /* We are in congestion avoidance */ net->partial_bytes_acked += net->net_ack; if ((net->flight_size + net->net_ack >= net->cwnd) && (net->partial_bytes_acked >= net->cwnd)) { net->partial_bytes_acked -= net->cwnd; net->cwnd += net->mtu; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, net->mtu, SCTP_CWND_LOG_FROM_CA); } } else { if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_LOGGING_ENABLE) { sctp_log_cwnd(stcb, net, net->net_ack, SCTP_CWND_LOG_NOADV_CA); } } } } else { if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_LOGGING_ENABLE) { sctp_log_cwnd(stcb, net, net->mtu, SCTP_CWND_LOG_NO_CUMACK); } } skip_cwnd_update: /* * NOW, according to Karn's rule do we need to restore the * RTO timer back? Check our net_ack2. If not set then we * have a ambiguity.. i.e. all data ack'd was sent to more * than one place. */ if (net->net_ack2) { /* restore any doubled timers */ net->RTO = (net->lastsa >> SCTP_RTT_SHIFT) + net->lastsv; if (net->RTO < stcb->asoc.minrto) { net->RTO = stcb->asoc.minrto; } if (net->RTO > stcb->asoc.maxrto) { net->RTO = stcb->asoc.maxrto; } } } } /* * H-TCP congestion control. The algorithm is detailed in: * R.N.Shorten, D.J.Leith: * "H-TCP: TCP for high-speed and long-distance networks" * Proc. PFLDnet, Argonne, 2004. * http://www.hamilton.ie/net/htcp3.pdf */ static int use_rtt_scaling = 1; static int use_bandwidth_switch = 1; static inline int between(uint32_t seq1, uint32_t seq2, uint32_t seq3) { return seq3 - seq2 >= seq1 - seq2; } static inline uint32_t htcp_cong_time(struct htcp *ca) { return sctp_get_tick_count() - ca->last_cong; } static inline uint32_t htcp_ccount(struct htcp *ca) { return htcp_cong_time(ca) / ca->minRTT; } static inline void htcp_reset(struct htcp *ca) { ca->undo_last_cong = ca->last_cong; ca->undo_maxRTT = ca->maxRTT; ca->undo_old_maxB = ca->old_maxB; ca->last_cong = sctp_get_tick_count(); } #ifdef SCTP_NOT_USED static uint32_t htcp_cwnd_undo(struct sctp_tcb *stcb, struct sctp_nets *net) { net->cc_mod.htcp_ca.last_cong = net->cc_mod.htcp_ca.undo_last_cong; net->cc_mod.htcp_ca.maxRTT = net->cc_mod.htcp_ca.undo_maxRTT; net->cc_mod.htcp_ca.old_maxB = net->cc_mod.htcp_ca.undo_old_maxB; return max(net->cwnd, ((net->ssthresh / net->mtu << 7) / net->cc_mod.htcp_ca.beta) * net->mtu); } #endif static inline void measure_rtt(struct sctp_tcb *stcb, struct sctp_nets *net) { uint32_t srtt = net->lastsa >> SCTP_RTT_SHIFT; /* keep track of minimum RTT seen so far, minRTT is zero at first */ if (net->cc_mod.htcp_ca.minRTT > srtt || !net->cc_mod.htcp_ca.minRTT) net->cc_mod.htcp_ca.minRTT = srtt; /* max RTT */ if (net->fast_retran_ip == 0 && net->ssthresh < 0xFFFF && htcp_ccount(&net->cc_mod.htcp_ca) > 3) { if (net->cc_mod.htcp_ca.maxRTT < net->cc_mod.htcp_ca.minRTT) net->cc_mod.htcp_ca.maxRTT = net->cc_mod.htcp_ca.minRTT; if (net->cc_mod.htcp_ca.maxRTT < srtt && srtt <= net->cc_mod.htcp_ca.maxRTT + MSEC_TO_TICKS(20)) net->cc_mod.htcp_ca.maxRTT = srtt; } } static void measure_achieved_throughput(struct sctp_tcb *stcb, struct sctp_nets *net) { uint32_t now = sctp_get_tick_count(); if (net->fast_retran_ip == 0) net->cc_mod.htcp_ca.bytes_acked = net->net_ack; if (!use_bandwidth_switch) return; /* achieved throughput calculations */ /* JRS - not 100% sure of this statement */ if (net->fast_retran_ip == 1) { net->cc_mod.htcp_ca.bytecount = 0; net->cc_mod.htcp_ca.lasttime = now; return; } net->cc_mod.htcp_ca.bytecount += net->net_ack; if (net->cc_mod.htcp_ca.bytecount >= net->cwnd - ((net->cc_mod.htcp_ca.alpha >> 7 ? : 1) * net->mtu) && now - net->cc_mod.htcp_ca.lasttime >= net->cc_mod.htcp_ca.minRTT && net->cc_mod.htcp_ca.minRTT > 0) { uint32_t cur_Bi = net->cc_mod.htcp_ca.bytecount / net->mtu * hz / (now - net->cc_mod.htcp_ca.lasttime); if (htcp_ccount(&net->cc_mod.htcp_ca) <= 3) { /* just after backoff */ net->cc_mod.htcp_ca.minB = net->cc_mod.htcp_ca.maxB = net->cc_mod.htcp_ca.Bi = cur_Bi; } else { net->cc_mod.htcp_ca.Bi = (3 * net->cc_mod.htcp_ca.Bi + cur_Bi) / 4; if (net->cc_mod.htcp_ca.Bi > net->cc_mod.htcp_ca.maxB) net->cc_mod.htcp_ca.maxB = net->cc_mod.htcp_ca.Bi; if (net->cc_mod.htcp_ca.minB > net->cc_mod.htcp_ca.maxB) net->cc_mod.htcp_ca.minB = net->cc_mod.htcp_ca.maxB; } net->cc_mod.htcp_ca.bytecount = 0; net->cc_mod.htcp_ca.lasttime = now; } } static inline void htcp_beta_update(struct htcp *ca, uint32_t minRTT, uint32_t maxRTT) { if (use_bandwidth_switch) { uint32_t maxB = ca->maxB; uint32_t old_maxB = ca->old_maxB; ca->old_maxB = ca->maxB; if (!between(5 * maxB, 4 * old_maxB, 6 * old_maxB)) { ca->beta = BETA_MIN; ca->modeswitch = 0; return; } } if (ca->modeswitch && minRTT > (uint32_t) MSEC_TO_TICKS(10) && maxRTT) { ca->beta = (minRTT << 7) / maxRTT; if (ca->beta < BETA_MIN) ca->beta = BETA_MIN; else if (ca->beta > BETA_MAX) ca->beta = BETA_MAX; } else { ca->beta = BETA_MIN; ca->modeswitch = 1; } } static inline void htcp_alpha_update(struct htcp *ca) { uint32_t minRTT = ca->minRTT; uint32_t factor = 1; uint32_t diff = htcp_cong_time(ca); if (diff > (uint32_t) hz) { diff -= hz; factor = 1 + (10 * diff + ((diff / 2) * (diff / 2) / hz)) / hz; } if (use_rtt_scaling && minRTT) { uint32_t scale = (hz << 3) / (10 * minRTT); scale = min(max(scale, 1U << 2), 10U << 3); /* clamping ratio to * interval [0.5,10]<<3 */ factor = (factor << 3) / scale; if (!factor) factor = 1; } ca->alpha = 2 * factor * ((1 << 7) - ca->beta); if (!ca->alpha) ca->alpha = ALPHA_BASE; } /* After we have the rtt data to calculate beta, we'd still prefer to wait one * rtt before we adjust our beta to ensure we are working from a consistent * data. * * This function should be called when we hit a congestion event since only at * that point do we really have a real sense of maxRTT (the queues en route * were getting just too full now). */ static void htcp_param_update(struct sctp_tcb *stcb, struct sctp_nets *net) { uint32_t minRTT = net->cc_mod.htcp_ca.minRTT; uint32_t maxRTT = net->cc_mod.htcp_ca.maxRTT; htcp_beta_update(&net->cc_mod.htcp_ca, minRTT, maxRTT); htcp_alpha_update(&net->cc_mod.htcp_ca); /* * add slowly fading memory for maxRTT to accommodate routing * changes etc */ if (minRTT > 0 && maxRTT > minRTT) net->cc_mod.htcp_ca.maxRTT = minRTT + ((maxRTT - minRTT) * 95) / 100; } static uint32_t htcp_recalc_ssthresh(struct sctp_tcb *stcb, struct sctp_nets *net) { htcp_param_update(stcb, net); return max(((net->cwnd / net->mtu * net->cc_mod.htcp_ca.beta) >> 7) * net->mtu, 2U * net->mtu); } static void htcp_cong_avoid(struct sctp_tcb *stcb, struct sctp_nets *net) { /*- * How to handle these functions? * if (!tcp_is_cwnd_limited(sk, in_flight)) RRS - good question. * return; */ if (net->cwnd <= net->ssthresh) { /* We are in slow start */ if (net->flight_size + net->net_ack >= net->cwnd) { if (net->net_ack > (net->mtu * SCTP_BASE_SYSCTL(sctp_L2_abc_variable))) { net->cwnd += (net->mtu * SCTP_BASE_SYSCTL(sctp_L2_abc_variable)); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, net->mtu, SCTP_CWND_LOG_FROM_SS); } } else { net->cwnd += net->net_ack; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, net->net_ack, SCTP_CWND_LOG_FROM_SS); } } } else { if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_LOGGING_ENABLE) { sctp_log_cwnd(stcb, net, net->net_ack, SCTP_CWND_LOG_NOADV_SS); } } } else { measure_rtt(stcb, net); /* * In dangerous area, increase slowly. In theory this is * net->cwnd += alpha / net->cwnd */ /* What is snd_cwnd_cnt?? */ if (((net->partial_bytes_acked / net->mtu * net->cc_mod.htcp_ca.alpha) >> 7) * net->mtu >= net->cwnd) { /*- * Does SCTP have a cwnd clamp? * if (net->snd_cwnd < net->snd_cwnd_clamp) - Nope (RRS). */ net->cwnd += net->mtu; net->partial_bytes_acked = 0; htcp_alpha_update(&net->cc_mod.htcp_ca); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, net->mtu, SCTP_CWND_LOG_FROM_CA); } } else { net->partial_bytes_acked += net->net_ack; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_LOGGING_ENABLE) { sctp_log_cwnd(stcb, net, net->net_ack, SCTP_CWND_LOG_NOADV_CA); } } net->cc_mod.htcp_ca.bytes_acked = net->mtu; } } #ifdef SCTP_NOT_USED /* Lower bound on congestion window. */ static uint32_t htcp_min_cwnd(struct sctp_tcb *stcb, struct sctp_nets *net) { return net->ssthresh; } #endif static void htcp_init(struct sctp_tcb *stcb, struct sctp_nets *net) { memset(&net->cc_mod.htcp_ca, 0, sizeof(struct htcp)); net->cc_mod.htcp_ca.alpha = ALPHA_BASE; net->cc_mod.htcp_ca.beta = BETA_MIN; net->cc_mod.htcp_ca.bytes_acked = net->mtu; net->cc_mod.htcp_ca.last_cong = sctp_get_tick_count(); } static void sctp_htcp_set_initial_cc_param(struct sctp_tcb *stcb, struct sctp_nets *net) { /* * We take the max of the burst limit times a MTU or the * INITIAL_CWND. We then limit this to 4 MTU's of sending. */ net->cwnd = min((net->mtu * 4), max((2 * net->mtu), SCTP_INITIAL_CWND)); net->ssthresh = stcb->asoc.peers_rwnd; htcp_init(stcb, net); if (SCTP_BASE_SYSCTL(sctp_logging_level) & (SCTP_CWND_MONITOR_ENABLE | SCTP_CWND_LOGGING_ENABLE)) { sctp_log_cwnd(stcb, net, 0, SCTP_CWND_INITIALIZATION); } } static void sctp_htcp_cwnd_update_after_sack(struct sctp_tcb *stcb, struct sctp_association *asoc, int accum_moved, int reneged_all, int will_exit) { struct sctp_nets *net; /******************************/ /* update cwnd and Early FR */ /******************************/ TAILQ_FOREACH(net, &asoc->nets, sctp_next) { #ifdef JANA_CMT_FAST_RECOVERY /* * CMT fast recovery code. Need to debug. */ if (net->fast_retran_loss_recovery && net->new_pseudo_cumack) { if (SCTP_TSN_GE(asoc->last_acked_seq, net->fast_recovery_tsn) || SCTP_TSN_GE(net->pseudo_cumack, net->fast_recovery_tsn)) { net->will_exit_fast_recovery = 1; } } #endif if (SCTP_BASE_SYSCTL(sctp_early_fr)) { /* * So, first of all do we need to have a Early FR * timer running? */ if ((!TAILQ_EMPTY(&asoc->sent_queue) && (net->ref_count > 1) && (net->flight_size < net->cwnd)) || (reneged_all)) { /* * yes, so in this case stop it if its * running, and then restart it. Reneging * all is a special case where we want to * run the Early FR timer and then force the * last few unacked to be sent, causing us * to illicit a sack with gaps to force out * the others. */ if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) { SCTP_STAT_INCR(sctps_earlyfrstpidsck2); sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_INDATA + SCTP_LOC_20); } SCTP_STAT_INCR(sctps_earlyfrstrid); sctp_timer_start(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net); } else { /* No, stop it if its running */ if (SCTP_OS_TIMER_PENDING(&net->fr_timer.timer)) { SCTP_STAT_INCR(sctps_earlyfrstpidsck3); sctp_timer_stop(SCTP_TIMER_TYPE_EARLYFR, stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_INDATA + SCTP_LOC_21); } } } /* if nothing was acked on this destination skip it */ if (net->net_ack == 0) { if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_LOGGING_ENABLE) { sctp_log_cwnd(stcb, net, 0, SCTP_CWND_LOG_FROM_SACK); } continue; } if (net->net_ack2 > 0) { /* * Karn's rule applies to clearing error count, this * is optional. */ net->error_count = 0; if ((net->dest_state & SCTP_ADDR_NOT_REACHABLE) == SCTP_ADDR_NOT_REACHABLE) { /* addr came good */ net->dest_state &= ~SCTP_ADDR_NOT_REACHABLE; net->dest_state |= SCTP_ADDR_REACHABLE; sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_UP, stcb, SCTP_RECEIVED_SACK, (void *)net, SCTP_SO_NOT_LOCKED); /* now was it the primary? if so restore */ if (net->dest_state & SCTP_ADDR_WAS_PRIMARY) { (void)sctp_set_primary_addr(stcb, (struct sockaddr *)NULL, net); } } /* * JRS 5/14/07 - If CMT PF is on and the destination * is in PF state, set the destination to active * state and set the cwnd to one or two MTU's based * on whether PF1 or PF2 is being used. * * Should we stop any running T3 timer here? */ if ((asoc->sctp_cmt_on_off > 0) && (asoc->sctp_cmt_pf > 0) && ((net->dest_state & SCTP_ADDR_PF) == SCTP_ADDR_PF)) { net->dest_state &= ~SCTP_ADDR_PF; net->cwnd = net->mtu * asoc->sctp_cmt_pf; SCTPDBG(SCTP_DEBUG_INDATA1, "Destination %p moved from PF to reachable with cwnd %d.\n", net, net->cwnd); /* * Since the cwnd value is explicitly set, * skip the code that updates the cwnd * value. */ goto skip_cwnd_update; } } #ifdef JANA_CMT_FAST_RECOVERY /* * CMT fast recovery code */ /* * if (sctp_cmt_on_off > 0 && net->fast_retran_loss_recovery * && net->will_exit_fast_recovery == 0) { @@@ Do something * } else if (sctp_cmt_on_off == 0 && * asoc->fast_retran_loss_recovery && will_exit == 0) { */ #endif if (asoc->fast_retran_loss_recovery && will_exit == 0 && (asoc->sctp_cmt_on_off == 0)) { /* * If we are in loss recovery we skip any cwnd * update */ goto skip_cwnd_update; } /* * CMT: CUC algorithm. Update cwnd if pseudo-cumack has * moved. */ if (accum_moved || ((asoc->sctp_cmt_on_off > 0) && net->new_pseudo_cumack)) { htcp_cong_avoid(stcb, net); measure_achieved_throughput(stcb, net); } else { if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_LOGGING_ENABLE) { sctp_log_cwnd(stcb, net, net->mtu, SCTP_CWND_LOG_NO_CUMACK); } } skip_cwnd_update: /* * NOW, according to Karn's rule do we need to restore the * RTO timer back? Check our net_ack2. If not set then we * have a ambiguity.. i.e. all data ack'd was sent to more * than one place. */ if (net->net_ack2) { /* restore any doubled timers */ net->RTO = (net->lastsa >> SCTP_RTT_SHIFT) + net->lastsv; if (net->RTO < stcb->asoc.minrto) { net->RTO = stcb->asoc.minrto; } if (net->RTO > stcb->asoc.maxrto) { net->RTO = stcb->asoc.maxrto; } } } } static void sctp_htcp_cwnd_update_after_fr(struct sctp_tcb *stcb, struct sctp_association *asoc) { struct sctp_nets *net; /* * CMT fast recovery code. Need to debug. ((sctp_cmt_on_off > 0) && * (net->fast_retran_loss_recovery == 0))) */ TAILQ_FOREACH(net, &asoc->nets, sctp_next) { if ((asoc->fast_retran_loss_recovery == 0) || (asoc->sctp_cmt_on_off > 0)) { /* out of a RFC2582 Fast recovery window? */ if (net->net_ack > 0) { /* * per section 7.2.3, are there any * destinations that had a fast retransmit * to them. If so what we need to do is * adjust ssthresh and cwnd. */ struct sctp_tmit_chunk *lchk; int old_cwnd = net->cwnd; /* JRS - reset as if state were changed */ htcp_reset(&net->cc_mod.htcp_ca); net->ssthresh = htcp_recalc_ssthresh(stcb, net); net->cwnd = net->ssthresh; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_FR); } lchk = TAILQ_FIRST(&asoc->send_queue); net->partial_bytes_acked = 0; /* Turn on fast recovery window */ asoc->fast_retran_loss_recovery = 1; if (lchk == NULL) { /* Mark end of the window */ asoc->fast_recovery_tsn = asoc->sending_seq - 1; } else { asoc->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1; } /* * CMT fast recovery -- per destination * recovery variable. */ net->fast_retran_loss_recovery = 1; if (lchk == NULL) { /* Mark end of the window */ net->fast_recovery_tsn = asoc->sending_seq - 1; } else { net->fast_recovery_tsn = lchk->rec.data.TSN_seq - 1; } sctp_timer_stop(SCTP_TIMER_TYPE_SEND, stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_INDATA + SCTP_LOC_32); sctp_timer_start(SCTP_TIMER_TYPE_SEND, stcb->sctp_ep, stcb, net); } } else if (net->net_ack > 0) { /* * Mark a peg that we WOULD have done a cwnd * reduction but RFC2582 prevented this action. */ SCTP_STAT_INCR(sctps_fastretransinrtt); } } } static void sctp_htcp_cwnd_update_after_timeout(struct sctp_tcb *stcb, struct sctp_nets *net) { int old_cwnd = net->cwnd; /* JRS - reset as if the state were being changed to timeout */ htcp_reset(&net->cc_mod.htcp_ca); net->ssthresh = htcp_recalc_ssthresh(stcb, net); net->cwnd = net->mtu; net->partial_bytes_acked = 0; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, net->cwnd - old_cwnd, SCTP_CWND_LOG_FROM_RTX); } } static void sctp_htcp_cwnd_update_after_fr_timer(struct sctp_inpcb *inp, struct sctp_tcb *stcb, struct sctp_nets *net) { int old_cwnd; old_cwnd = net->cwnd; sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_EARLY_FR_TMR, SCTP_SO_NOT_LOCKED); net->cc_mod.htcp_ca.last_cong = sctp_get_tick_count(); /* * make a small adjustment to cwnd and force to CA. */ if (net->cwnd > net->mtu) /* drop down one MTU after sending */ net->cwnd -= net->mtu; if (net->cwnd < net->ssthresh) /* still in SS move to CA */ net->ssthresh = net->cwnd - 1; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, (old_cwnd - net->cwnd), SCTP_CWND_LOG_FROM_FR); } } static void sctp_htcp_cwnd_update_after_ecn_echo(struct sctp_tcb *stcb, struct sctp_nets *net, int in_window, int num_pkt_lost) { int old_cwnd; old_cwnd = net->cwnd; /* JRS - reset hctp as if state changed */ if (in_window == 0) { htcp_reset(&net->cc_mod.htcp_ca); SCTP_STAT_INCR(sctps_ecnereducedcwnd); net->ssthresh = htcp_recalc_ssthresh(stcb, net); if (net->ssthresh < net->mtu) { net->ssthresh = net->mtu; /* here back off the timer as well, to slow us down */ net->RTO <<= 1; } net->cwnd = net->ssthresh; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_CWND_MONITOR_ENABLE) { sctp_log_cwnd(stcb, net, (net->cwnd - old_cwnd), SCTP_CWND_LOG_FROM_SAT); } } } struct sctp_cc_functions sctp_cc_functions[] = { { .sctp_set_initial_cc_param = sctp_set_initial_cc_param, .sctp_cwnd_update_after_sack = sctp_cwnd_update_after_sack, .sctp_cwnd_update_after_fr = sctp_cwnd_update_after_fr, .sctp_cwnd_update_after_timeout = sctp_cwnd_update_after_timeout, .sctp_cwnd_update_after_ecn_echo = sctp_cwnd_update_after_ecn_echo, .sctp_cwnd_update_after_packet_dropped = sctp_cwnd_update_after_packet_dropped, .sctp_cwnd_update_after_output = sctp_cwnd_update_after_output, .sctp_cwnd_update_after_fr_timer = sctp_cwnd_update_after_fr_timer }, { .sctp_set_initial_cc_param = sctp_set_initial_cc_param, .sctp_cwnd_update_after_sack = sctp_hs_cwnd_update_after_sack, .sctp_cwnd_update_after_fr = sctp_hs_cwnd_update_after_fr, .sctp_cwnd_update_after_timeout = sctp_cwnd_update_after_timeout, .sctp_cwnd_update_after_ecn_echo = sctp_cwnd_update_after_ecn_echo, .sctp_cwnd_update_after_packet_dropped = sctp_cwnd_update_after_packet_dropped, .sctp_cwnd_update_after_output = sctp_cwnd_update_after_output, .sctp_cwnd_update_after_fr_timer = sctp_cwnd_update_after_fr_timer }, { .sctp_set_initial_cc_param = sctp_htcp_set_initial_cc_param, .sctp_cwnd_update_after_sack = sctp_htcp_cwnd_update_after_sack, .sctp_cwnd_update_after_fr = sctp_htcp_cwnd_update_after_fr, .sctp_cwnd_update_after_timeout = sctp_htcp_cwnd_update_after_timeout, .sctp_cwnd_update_after_ecn_echo = sctp_htcp_cwnd_update_after_ecn_echo, .sctp_cwnd_update_after_packet_dropped = sctp_cwnd_update_after_packet_dropped, .sctp_cwnd_update_after_output = sctp_cwnd_update_after_output, .sctp_cwnd_update_after_fr_timer = sctp_htcp_cwnd_update_after_fr_timer }, { .sctp_set_initial_cc_param = sctp_set_rtcc_initial_cc_param, .sctp_cwnd_update_after_sack = sctp_cwnd_update_rtcc_after_sack, .sctp_cwnd_update_after_fr = sctp_cwnd_update_after_fr, .sctp_cwnd_update_after_timeout = sctp_cwnd_update_after_timeout, .sctp_cwnd_update_after_ecn_echo = sctp_cwnd_update_rtcc_after_ecn_echo, .sctp_cwnd_update_after_packet_dropped = sctp_cwnd_update_after_packet_dropped, .sctp_cwnd_update_after_output = sctp_cwnd_update_after_output, .sctp_cwnd_update_after_fr_timer = sctp_cwnd_update_after_fr_timer, .sctp_cwnd_update_packet_transmitted = sctp_cwnd_update_rtcc_packet_transmitted, .sctp_cwnd_update_tsn_acknowledged = sctp_cwnd_update_rtcc_tsn_acknowledged, .sctp_cwnd_new_transmission_begins = sctp_cwnd_new_rtcc_transmission_begins, .sctp_cwnd_prepare_net_for_sack = sctp_cwnd_prepare_rtcc_net_for_sack, .sctp_cwnd_socket_option = sctp_cwnd_rtcc_socket_option, .sctp_rtt_calculated = sctp_rtt_rtcc_calculated } }; Index: projects/largeSMP/sys/netinet/sctp_sysctl.h =================================================================== --- projects/largeSMP/sys/netinet/sctp_sysctl.h (revision 221494) +++ projects/largeSMP/sys/netinet/sctp_sysctl.h (revision 221495) @@ -1,575 +1,575 @@ /*- * Copyright (c) 2007, by Cisco Systems, Inc. All rights reserved. * Copyright (c) 2008-2011, by Randall Stewart. All rights reserved. * Copyright (c) 2008-2011, by Michael Tuexen. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * a) Redistributions of source code must retain the above copyright notice, * this list of conditions and the following disclaimer. * * b) Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in * the documentation and/or other materials provided with the distribution. * * c) Neither the name of Cisco Systems, Inc. nor the names of its * contributors may be used to endorse or promote products derived * from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF * THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #ifndef __sctp_sysctl_h__ #define __sctp_sysctl_h__ #include #include struct sctp_sysctl { uint32_t sctp_sendspace; uint32_t sctp_recvspace; uint32_t sctp_auto_asconf; uint32_t sctp_multiple_asconfs; uint32_t sctp_ecn_enable; uint32_t sctp_fr_max_burst_default; uint32_t sctp_strict_sacks; #if !defined(SCTP_WITH_NO_CSUM) uint32_t sctp_no_csum_on_loopback; #endif uint32_t sctp_strict_init; uint32_t sctp_peer_chunk_oh; uint32_t sctp_max_burst_default; uint32_t sctp_max_chunks_on_queue; uint32_t sctp_hashtblsize; uint32_t sctp_pcbtblsize; uint32_t sctp_min_split_point; uint32_t sctp_chunkscale; uint32_t sctp_delayed_sack_time_default; uint32_t sctp_sack_freq_default; uint32_t sctp_system_free_resc_limit; uint32_t sctp_asoc_free_resc_limit; uint32_t sctp_heartbeat_interval_default; uint32_t sctp_pmtu_raise_time_default; uint32_t sctp_shutdown_guard_time_default; uint32_t sctp_secret_lifetime_default; uint32_t sctp_rto_max_default; uint32_t sctp_rto_min_default; uint32_t sctp_rto_initial_default; uint32_t sctp_init_rto_max_default; uint32_t sctp_valid_cookie_life_default; uint32_t sctp_init_rtx_max_default; uint32_t sctp_assoc_rtx_max_default; uint32_t sctp_path_rtx_max_default; uint32_t sctp_add_more_threshold; uint32_t sctp_nr_outgoing_streams_default; uint32_t sctp_cmt_on_off; uint32_t sctp_cmt_use_dac; /* EY 5/5/08 - nr_sack flag variable */ uint32_t sctp_nr_sack_on_off; uint32_t sctp_cmt_pf; uint32_t sctp_use_cwnd_based_maxburst; uint32_t sctp_early_fr; uint32_t sctp_early_fr_msec; uint32_t sctp_asconf_auth_nochk; uint32_t sctp_auth_disable; uint32_t sctp_nat_friendly; uint32_t sctp_L2_abc_variable; uint32_t sctp_mbuf_threshold_count; uint32_t sctp_do_drain; uint32_t sctp_hb_maxburst; uint32_t sctp_abort_if_one_2_one_hits_limit; uint32_t sctp_strict_data_order; uint32_t sctp_min_residual; uint32_t sctp_max_retran_chunk; uint32_t sctp_logging_level; /* JRS - Variable for default congestion control module */ uint32_t sctp_default_cc_module; /* RS - Variable for default stream scheduling module */ uint32_t sctp_default_ss_module; uint32_t sctp_default_frag_interleave; uint32_t sctp_mobility_base; uint32_t sctp_mobility_fasthandoff; uint32_t sctp_inits_include_nat_friendly; uint32_t sctp_rttvar_bw; uint32_t sctp_rttvar_rtt; uint32_t sctp_rttvar_eqret; uint32_t sctp_steady_step; uint32_t sctp_use_dccc_ecn; #if defined(SCTP_LOCAL_TRACE_BUF) struct sctp_log sctp_log; #endif uint32_t sctp_udp_tunneling_for_client_enable; uint32_t sctp_udp_tunneling_port; uint32_t sctp_enable_sack_immediately; uint32_t sctp_vtag_time_wait; uint32_t sctp_buffer_splitting; uint32_t sctp_initial_cwnd; #if defined(SCTP_DEBUG) uint32_t sctp_debug_on; #endif #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) uint32_t sctp_output_unlocked; #endif }; /* * limits for the sysctl variables */ /* maxdgram: Maximum outgoing SCTP buffer size */ #define SCTPCTL_MAXDGRAM_DESC "Maximum outgoing SCTP buffer size" #define SCTPCTL_MAXDGRAM_MIN 0 #define SCTPCTL_MAXDGRAM_MAX 0xFFFFFFFF #define SCTPCTL_MAXDGRAM_DEFAULT 262144 /* 256k */ /* recvspace: Maximum incoming SCTP buffer size */ #define SCTPCTL_RECVSPACE_DESC "Maximum incoming SCTP buffer size" #define SCTPCTL_RECVSPACE_MIN 0 #define SCTPCTL_RECVSPACE_MAX 0xFFFFFFFF #define SCTPCTL_RECVSPACE_DEFAULT 262144 /* 256k */ /* autoasconf: Enable SCTP Auto-ASCONF */ #define SCTPCTL_AUTOASCONF_DESC "Enable SCTP Auto-ASCONF" #define SCTPCTL_AUTOASCONF_MIN 0 #define SCTPCTL_AUTOASCONF_MAX 1 #define SCTPCTL_AUTOASCONF_DEFAULT SCTP_DEFAULT_AUTO_ASCONF /* autoasconf: Enable SCTP Auto-ASCONF */ #define SCTPCTL_MULTIPLEASCONFS_DESC "Enable SCTP Muliple-ASCONFs" #define SCTPCTL_MULTIPLEASCONFS_MIN 0 #define SCTPCTL_MULTIPLEASCONFS_MAX 1 #define SCTPCTL_MULTIPLEASCONFS_DEFAULT SCTP_DEFAULT_MULTIPLE_ASCONFS /* ecn_enable: Enable SCTP ECN */ #define SCTPCTL_ECN_ENABLE_DESC "Enable SCTP ECN" #define SCTPCTL_ECN_ENABLE_MIN 0 #define SCTPCTL_ECN_ENABLE_MAX 1 #define SCTPCTL_ECN_ENABLE_DEFAULT 1 /* strict_sacks: Enable SCTP Strict SACK checking */ #define SCTPCTL_STRICT_SACKS_DESC "Enable SCTP Strict SACK checking" #define SCTPCTL_STRICT_SACKS_MIN 0 #define SCTPCTL_STRICT_SACKS_MAX 1 #define SCTPCTL_STRICT_SACKS_DEFAULT 1 /* loopback_nocsum: Enable NO Csum on packets sent on loopback */ #define SCTPCTL_LOOPBACK_NOCSUM_DESC "Enable NO Csum on packets sent on loopback" #define SCTPCTL_LOOPBACK_NOCSUM_MIN 0 #define SCTPCTL_LOOPBACK_NOCSUM_MAX 1 #define SCTPCTL_LOOPBACK_NOCSUM_DEFAULT 1 /* strict_init: Enable strict INIT/INIT-ACK singleton enforcement */ #define SCTPCTL_STRICT_INIT_DESC "Enable strict INIT/INIT-ACK singleton enforcement" #define SCTPCTL_STRICT_INIT_MIN 0 #define SCTPCTL_STRICT_INIT_MAX 1 #define SCTPCTL_STRICT_INIT_DEFAULT 1 /* peer_chkoh: Amount to debit peers rwnd per chunk sent */ #define SCTPCTL_PEER_CHKOH_DESC "Amount to debit peers rwnd per chunk sent" #define SCTPCTL_PEER_CHKOH_MIN 0 #define SCTPCTL_PEER_CHKOH_MAX 0xFFFFFFFF #define SCTPCTL_PEER_CHKOH_DEFAULT 256 /* maxburst: Default max burst for sctp endpoints */ #define SCTPCTL_MAXBURST_DESC "Default max burst for sctp endpoints" #define SCTPCTL_MAXBURST_MIN 0 #define SCTPCTL_MAXBURST_MAX 0xFFFFFFFF #define SCTPCTL_MAXBURST_DEFAULT SCTP_DEF_MAX_BURST /* fr_maxburst: Default max burst for sctp endpoints when fast retransmitting */ #define SCTPCTL_FRMAXBURST_DESC "Default fr max burst for sctp endpoints" #define SCTPCTL_FRMAXBURST_MIN 0 #define SCTPCTL_FRMAXBURST_MAX 0xFFFFFFFF #define SCTPCTL_FRMAXBURST_DEFAULT SCTP_DEF_FRMAX_BURST /* maxchunks: Default max chunks on queue per asoc */ #define SCTPCTL_MAXCHUNKS_DESC "Default max chunks on queue per asoc" #define SCTPCTL_MAXCHUNKS_MIN 0 #define SCTPCTL_MAXCHUNKS_MAX 0xFFFFFFFF #define SCTPCTL_MAXCHUNKS_DEFAULT SCTP_ASOC_MAX_CHUNKS_ON_QUEUE /* tcbhashsize: Tunable for Hash table sizes */ #define SCTPCTL_TCBHASHSIZE_DESC "Tunable for TCB hash table sizes" #define SCTPCTL_TCBHASHSIZE_MIN 1 #define SCTPCTL_TCBHASHSIZE_MAX 0xFFFFFFFF #define SCTPCTL_TCBHASHSIZE_DEFAULT SCTP_TCBHASHSIZE /* pcbhashsize: Tunable for PCB Hash table sizes */ #define SCTPCTL_PCBHASHSIZE_DESC "Tunable for PCB hash table sizes" #define SCTPCTL_PCBHASHSIZE_MIN 1 #define SCTPCTL_PCBHASHSIZE_MAX 0xFFFFFFFF #define SCTPCTL_PCBHASHSIZE_DEFAULT SCTP_PCBHASHSIZE /* min_split_point: Minimum size when splitting a chunk */ #define SCTPCTL_MIN_SPLIT_POINT_DESC "Minimum size when splitting a chunk" #define SCTPCTL_MIN_SPLIT_POINT_MIN 0 #define SCTPCTL_MIN_SPLIT_POINT_MAX 0xFFFFFFFF #define SCTPCTL_MIN_SPLIT_POINT_DEFAULT SCTP_DEFAULT_SPLIT_POINT_MIN /* chunkscale: Tunable for Scaling of number of chunks and messages */ #define SCTPCTL_CHUNKSCALE_DESC "Tunable for Scaling of number of chunks and messages" #define SCTPCTL_CHUNKSCALE_MIN 1 #define SCTPCTL_CHUNKSCALE_MAX 0xFFFFFFFF #define SCTPCTL_CHUNKSCALE_DEFAULT SCTP_CHUNKQUEUE_SCALE /* delayed_sack_time: Default delayed SACK timer in ms */ #define SCTPCTL_DELAYED_SACK_TIME_DESC "Default delayed SACK timer in ms" #define SCTPCTL_DELAYED_SACK_TIME_MIN 0 #define SCTPCTL_DELAYED_SACK_TIME_MAX 0xFFFFFFFF #define SCTPCTL_DELAYED_SACK_TIME_DEFAULT SCTP_RECV_MSEC /* sack_freq: Default SACK frequency */ #define SCTPCTL_SACK_FREQ_DESC "Default SACK frequency" #define SCTPCTL_SACK_FREQ_MIN 0 #define SCTPCTL_SACK_FREQ_MAX 0xFFFFFFFF #define SCTPCTL_SACK_FREQ_DEFAULT SCTP_DEFAULT_SACK_FREQ /* sys_resource: Max number of cached resources in the system */ #define SCTPCTL_SYS_RESOURCE_DESC "Max number of cached resources in the system" #define SCTPCTL_SYS_RESOURCE_MIN 0 #define SCTPCTL_SYS_RESOURCE_MAX 0xFFFFFFFF #define SCTPCTL_SYS_RESOURCE_DEFAULT SCTP_DEF_SYSTEM_RESC_LIMIT /* asoc_resource: Max number of cached resources in an asoc */ #define SCTPCTL_ASOC_RESOURCE_DESC "Max number of cached resources in an asoc" #define SCTPCTL_ASOC_RESOURCE_MIN 0 #define SCTPCTL_ASOC_RESOURCE_MAX 0xFFFFFFFF #define SCTPCTL_ASOC_RESOURCE_DEFAULT SCTP_DEF_ASOC_RESC_LIMIT /* heartbeat_interval: Default heartbeat interval in ms */ #define SCTPCTL_HEARTBEAT_INTERVAL_DESC "Default heartbeat interval in ms" #define SCTPCTL_HEARTBEAT_INTERVAL_MIN 0 #define SCTPCTL_HEARTBEAT_INTERVAL_MAX 0xFFFFFFFF #define SCTPCTL_HEARTBEAT_INTERVAL_DEFAULT SCTP_HB_DEFAULT_MSEC /* pmtu_raise_time: Default PMTU raise timer in seconds */ #define SCTPCTL_PMTU_RAISE_TIME_DESC "Default PMTU raise timer in seconds" #define SCTPCTL_PMTU_RAISE_TIME_MIN 0 #define SCTPCTL_PMTU_RAISE_TIME_MAX 0xFFFFFFFF #define SCTPCTL_PMTU_RAISE_TIME_DEFAULT SCTP_DEF_PMTU_RAISE_SEC /* shutdown_guard_time: Default shutdown guard timer in seconds */ #define SCTPCTL_SHUTDOWN_GUARD_TIME_DESC "Default shutdown guard timer in seconds" #define SCTPCTL_SHUTDOWN_GUARD_TIME_MIN 0 #define SCTPCTL_SHUTDOWN_GUARD_TIME_MAX 0xFFFFFFFF #define SCTPCTL_SHUTDOWN_GUARD_TIME_DEFAULT SCTP_DEF_MAX_SHUTDOWN_SEC /* secret_lifetime: Default secret lifetime in seconds */ #define SCTPCTL_SECRET_LIFETIME_DESC "Default secret lifetime in seconds" #define SCTPCTL_SECRET_LIFETIME_MIN 0 #define SCTPCTL_SECRET_LIFETIME_MAX 0xFFFFFFFF #define SCTPCTL_SECRET_LIFETIME_DEFAULT SCTP_DEFAULT_SECRET_LIFE_SEC /* rto_max: Default maximum retransmission timeout in ms */ #define SCTPCTL_RTO_MAX_DESC "Default maximum retransmission timeout in ms" #define SCTPCTL_RTO_MAX_MIN 0 #define SCTPCTL_RTO_MAX_MAX 0xFFFFFFFF #define SCTPCTL_RTO_MAX_DEFAULT SCTP_RTO_UPPER_BOUND /* rto_min: Default minimum retransmission timeout in ms */ #define SCTPCTL_RTO_MIN_DESC "Default minimum retransmission timeout in ms" #define SCTPCTL_RTO_MIN_MIN 0 #define SCTPCTL_RTO_MIN_MAX 0xFFFFFFFF #define SCTPCTL_RTO_MIN_DEFAULT SCTP_RTO_LOWER_BOUND /* rto_initial: Default initial retransmission timeout in ms */ #define SCTPCTL_RTO_INITIAL_DESC "Default initial retransmission timeout in ms" #define SCTPCTL_RTO_INITIAL_MIN 0 #define SCTPCTL_RTO_INITIAL_MAX 0xFFFFFFFF #define SCTPCTL_RTO_INITIAL_DEFAULT SCTP_RTO_INITIAL /* init_rto_max: Default maximum retransmission timeout during association setup in ms */ #define SCTPCTL_INIT_RTO_MAX_DESC "Default maximum retransmission timeout during association setup in ms" #define SCTPCTL_INIT_RTO_MAX_MIN 0 #define SCTPCTL_INIT_RTO_MAX_MAX 0xFFFFFFFF #define SCTPCTL_INIT_RTO_MAX_DEFAULT SCTP_RTO_UPPER_BOUND /* valid_cookie_life: Default cookie lifetime in sec */ #define SCTPCTL_VALID_COOKIE_LIFE_DESC "Default cookie lifetime in seconds" #define SCTPCTL_VALID_COOKIE_LIFE_MIN 0 #define SCTPCTL_VALID_COOKIE_LIFE_MAX 0xFFFFFFFF #define SCTPCTL_VALID_COOKIE_LIFE_DEFAULT SCTP_DEFAULT_COOKIE_LIFE /* init_rtx_max: Default maximum number of retransmission for INIT chunks */ #define SCTPCTL_INIT_RTX_MAX_DESC "Default maximum number of retransmission for INIT chunks" #define SCTPCTL_INIT_RTX_MAX_MIN 0 #define SCTPCTL_INIT_RTX_MAX_MAX 0xFFFFFFFF #define SCTPCTL_INIT_RTX_MAX_DEFAULT SCTP_DEF_MAX_INIT /* assoc_rtx_max: Default maximum number of retransmissions per association */ #define SCTPCTL_ASSOC_RTX_MAX_DESC "Default maximum number of retransmissions per association" #define SCTPCTL_ASSOC_RTX_MAX_MIN 0 #define SCTPCTL_ASSOC_RTX_MAX_MAX 0xFFFFFFFF #define SCTPCTL_ASSOC_RTX_MAX_DEFAULT SCTP_DEF_MAX_SEND /* path_rtx_max: Default maximum of retransmissions per path */ #define SCTPCTL_PATH_RTX_MAX_DESC "Default maximum of retransmissions per path" #define SCTPCTL_PATH_RTX_MAX_MIN 0 #define SCTPCTL_PATH_RTX_MAX_MAX 0xFFFFFFFF #define SCTPCTL_PATH_RTX_MAX_DEFAULT SCTP_DEF_MAX_PATH_RTX /* add_more_on_output: When space-wise is it worthwhile to try to add more to a socket send buffer */ #define SCTPCTL_ADD_MORE_ON_OUTPUT_DESC "When space-wise is it worthwhile to try to add more to a socket send buffer" #define SCTPCTL_ADD_MORE_ON_OUTPUT_MIN 0 #define SCTPCTL_ADD_MORE_ON_OUTPUT_MAX 0xFFFFFFFF #define SCTPCTL_ADD_MORE_ON_OUTPUT_DEFAULT SCTP_DEFAULT_ADD_MORE /* outgoing_streams: Default number of outgoing streams */ #define SCTPCTL_OUTGOING_STREAMS_DESC "Default number of outgoing streams" #define SCTPCTL_OUTGOING_STREAMS_MIN 1 #define SCTPCTL_OUTGOING_STREAMS_MAX 65535 #define SCTPCTL_OUTGOING_STREAMS_DEFAULT SCTP_OSTREAM_INITIAL /* cmt_on_off: CMT on/off flag */ #define SCTPCTL_CMT_ON_OFF_DESC "CMT settings" -#define SCTPCTL_CMT_ON_OFF_MIN 0 -#define SCTPCTL_CMT_ON_OFF_MAX 2 -#define SCTPCTL_CMT_ON_OFF_DEFAULT 0 +#define SCTPCTL_CMT_ON_OFF_MIN SCTP_CMT_OFF +#define SCTPCTL_CMT_ON_OFF_MAX SCTP_CMT_MAX +#define SCTPCTL_CMT_ON_OFF_DEFAULT SCTP_CMT_OFF /* EY - nr_sack_on_off: NR_SACK on/off flag */ #define SCTPCTL_NR_SACK_ON_OFF_DESC "NR_SACK on/off flag" #define SCTPCTL_NR_SACK_ON_OFF_MIN 0 #define SCTPCTL_NR_SACK_ON_OFF_MAX 1 #define SCTPCTL_NR_SACK_ON_OFF_DEFAULT 0 /* cmt_use_dac: CMT DAC on/off flag */ #define SCTPCTL_CMT_USE_DAC_DESC "CMT DAC on/off flag" #define SCTPCTL_CMT_USE_DAC_MIN 0 #define SCTPCTL_CMT_USE_DAC_MAX 1 #define SCTPCTL_CMT_USE_DAC_DEFAULT 0 /* JRS 5/2107 - CMT PF type flag */ #define SCTPCTL_CMT_PF_DESC "CMT PF type flag" #define SCTPCTL_CMT_PF_MIN 0 #define SCTPCTL_CMT_PF_MAX 2 #define SCTPCTL_CMT_PF_DEFAULT 0 /* cwnd_maxburst: Use a CWND adjusting maxburst */ #define SCTPCTL_CWND_MAXBURST_DESC "Use a CWND adjusting maxburst" #define SCTPCTL_CWND_MAXBURST_MIN 0 #define SCTPCTL_CWND_MAXBURST_MAX 1 #define SCTPCTL_CWND_MAXBURST_DEFAULT 1 /* early_fast_retran: Early Fast Retransmit with timer */ #define SCTPCTL_EARLY_FAST_RETRAN_DESC "Early Fast Retransmit with timer" #define SCTPCTL_EARLY_FAST_RETRAN_MIN 0 #define SCTPCTL_EARLY_FAST_RETRAN_MAX 0xFFFFFFFF #define SCTPCTL_EARLY_FAST_RETRAN_DEFAULT 0 /* early_fast_retran_msec: Early Fast Retransmit minimum timer value */ #define SCTPCTL_EARLY_FAST_RETRAN_MSEC_DESC "Early Fast Retransmit minimum timer value" #define SCTPCTL_EARLY_FAST_RETRAN_MSEC_MIN 0 #define SCTPCTL_EARLY_FAST_RETRAN_MSEC_MAX 0xFFFFFFFF #define SCTPCTL_EARLY_FAST_RETRAN_MSEC_DEFAULT SCTP_MINFR_MSEC_TIMER /* asconf_auth_nochk: Disable SCTP ASCONF AUTH requirement */ #define SCTPCTL_ASCONF_AUTH_NOCHK_DESC "Disable SCTP ASCONF AUTH requirement" #define SCTPCTL_ASCONF_AUTH_NOCHK_MIN 0 #define SCTPCTL_ASCONF_AUTH_NOCHK_MAX 1 #define SCTPCTL_ASCONF_AUTH_NOCHK_DEFAULT 0 /* auth_disable: Disable SCTP AUTH function */ #define SCTPCTL_AUTH_DISABLE_DESC "Disable SCTP AUTH function" #define SCTPCTL_AUTH_DISABLE_MIN 0 #define SCTPCTL_AUTH_DISABLE_MAX 1 #define SCTPCTL_AUTH_DISABLE_DEFAULT 0 /* nat_friendly: SCTP NAT friendly operation */ #define SCTPCTL_NAT_FRIENDLY_DESC "SCTP NAT friendly operation" #define SCTPCTL_NAT_FRIENDLY_MIN 0 #define SCTPCTL_NAT_FRIENDLY_MAX 1 #define SCTPCTL_NAT_FRIENDLY_DEFAULT 1 /* abc_l_var: SCTP ABC max increase per SACK (L) */ #define SCTPCTL_ABC_L_VAR_DESC "SCTP ABC max increase per SACK (L)" #define SCTPCTL_ABC_L_VAR_MIN 0 #define SCTPCTL_ABC_L_VAR_MAX 0xFFFFFFFF #define SCTPCTL_ABC_L_VAR_DEFAULT 1 /* max_chained_mbufs: Default max number of small mbufs on a chain */ #define SCTPCTL_MAX_CHAINED_MBUFS_DESC "Default max number of small mbufs on a chain" #define SCTPCTL_MAX_CHAINED_MBUFS_MIN 0 #define SCTPCTL_MAX_CHAINED_MBUFS_MAX 0xFFFFFFFF #define SCTPCTL_MAX_CHAINED_MBUFS_DEFAULT SCTP_DEFAULT_MBUFS_IN_CHAIN /* do_sctp_drain: Should SCTP respond to the drain calls */ #define SCTPCTL_DO_SCTP_DRAIN_DESC "Should SCTP respond to the drain calls" #define SCTPCTL_DO_SCTP_DRAIN_MIN 0 #define SCTPCTL_DO_SCTP_DRAIN_MAX 1 #define SCTPCTL_DO_SCTP_DRAIN_DEFAULT 1 /* hb_max_burst: Confirmation Heartbeat max burst? */ #define SCTPCTL_HB_MAX_BURST_DESC "Confirmation Heartbeat max burst" #define SCTPCTL_HB_MAX_BURST_MIN 1 #define SCTPCTL_HB_MAX_BURST_MAX 0xFFFFFFFF #define SCTPCTL_HB_MAX_BURST_DEFAULT SCTP_DEF_HBMAX_BURST /* abort_at_limit: When one-2-one hits qlimit abort */ #define SCTPCTL_ABORT_AT_LIMIT_DESC "When one-2-one hits qlimit abort" #define SCTPCTL_ABORT_AT_LIMIT_MIN 0 #define SCTPCTL_ABORT_AT_LIMIT_MAX 1 #define SCTPCTL_ABORT_AT_LIMIT_DEFAULT 0 /* strict_data_order: Enforce strict data ordering, abort if control inside data */ #define SCTPCTL_STRICT_DATA_ORDER_DESC "Enforce strict data ordering, abort if control inside data" #define SCTPCTL_STRICT_DATA_ORDER_MIN 0 #define SCTPCTL_STRICT_DATA_ORDER_MAX 1 #define SCTPCTL_STRICT_DATA_ORDER_DEFAULT 0 /* min_residual: min residual in a data fragment leftover */ #define SCTPCTL_MIN_RESIDUAL_DESC "Minimum residual data chunk in second part of split" #define SCTPCTL_MIN_RESIDUAL_MIN 20 #define SCTPCTL_MIN_RESIDUAL_MAX 65535 #define SCTPCTL_MIN_RESIDUAL_DEFAULT 1452 /* max_retran_chunk: max chunk retransmissions */ #define SCTPCTL_MAX_RETRAN_CHUNK_DESC "Maximum times an unlucky chunk can be retran'd before assoc abort" #define SCTPCTL_MAX_RETRAN_CHUNK_MIN 0 #define SCTPCTL_MAX_RETRAN_CHUNK_MAX 65535 #define SCTPCTL_MAX_RETRAN_CHUNK_DEFAULT 30 /* sctp_logging: This gives us logging when the options are enabled */ #define SCTPCTL_LOGGING_LEVEL_DESC "Ltrace/KTR trace logging level" #define SCTPCTL_LOGGING_LEVEL_MIN 0 #define SCTPCTL_LOGGING_LEVEL_MAX 0xffffffff #define SCTPCTL_LOGGING_LEVEL_DEFAULT 0 /* JRS - default congestion control module sysctl */ #define SCTPCTL_DEFAULT_CC_MODULE_DESC "Default congestion control module" #define SCTPCTL_DEFAULT_CC_MODULE_MIN 0 #define SCTPCTL_DEFAULT_CC_MODULE_MAX 2 #define SCTPCTL_DEFAULT_CC_MODULE_DEFAULT 0 /* RS - default stream scheduling module sysctl */ #define SCTPCTL_DEFAULT_SS_MODULE_DESC "Default stream scheduling module" #define SCTPCTL_DEFAULT_SS_MODULE_MIN 0 #define SCTPCTL_DEFAULT_SS_MODULE_MAX 5 #define SCTPCTL_DEFAULT_SS_MODULE_DEFAULT 0 /* RRS - default fragment interleave */ #define SCTPCTL_DEFAULT_FRAG_INTERLEAVE_DESC "Default fragment interleave level" #define SCTPCTL_DEFAULT_FRAG_INTERLEAVE_MIN 0 #define SCTPCTL_DEFAULT_FRAG_INTERLEAVE_MAX 2 #define SCTPCTL_DEFAULT_FRAG_INTERLEAVE_DEFAULT 1 /* mobility_base: Enable SCTP mobility support */ #define SCTPCTL_MOBILITY_BASE_DESC "Enable SCTP base mobility" #define SCTPCTL_MOBILITY_BASE_MIN 0 #define SCTPCTL_MOBILITY_BASE_MAX 1 #define SCTPCTL_MOBILITY_BASE_DEFAULT SCTP_DEFAULT_MOBILITY_BASE /* mobility_fasthandoff: Enable SCTP fast handoff support */ #define SCTPCTL_MOBILITY_FASTHANDOFF_DESC "Enable SCTP fast handoff" #define SCTPCTL_MOBILITY_FASTHANDOFF_MIN 0 #define SCTPCTL_MOBILITY_FASTHANDOFF_MAX 1 #define SCTPCTL_MOBILITY_FASTHANDOFF_DEFAULT SCTP_DEFAULT_MOBILITY_FASTHANDOFF /* Enable SCTP/UDP tunneling for clients*/ #define SCTPCTL_UDP_TUNNELING_FOR_CLIENT_ENABLE_DESC "Enable SCTP/UDP tunneling for client" #define SCTPCTL_UDP_TUNNELING_FOR_CLIENT_ENABLE_MIN 0 #define SCTPCTL_UDP_TUNNELING_FOR_CLIENT_ENABLE_MAX 1 #define SCTPCTL_UDP_TUNNELING_FOR_CLIENT_ENABLE_DEFAULT SCTPCTL_UDP_TUNNELING_FOR_CLIENT_ENABLE_MIN /* Enable SCTP/UDP tunneling port */ #define SCTPCTL_UDP_TUNNELING_PORT_DESC "Set the SCTP/UDP tunneling port" #define SCTPCTL_UDP_TUNNELING_PORT_MIN 0 #define SCTPCTL_UDP_TUNNELING_PORT_MAX 65535 #define SCTPCTL_UDP_TUNNELING_PORT_DEFAULT SCTP_OVER_UDP_TUNNELING_PORT /* Enable sending of the SACK-IMMEDIATELY bit */ #define SCTPCTL_SACK_IMMEDIATELY_ENABLE_DESC "Enable sending of the SACK-IMMEDIATELY-bit." #define SCTPCTL_SACK_IMMEDIATELY_ENABLE_MIN 0 #define SCTPCTL_SACK_IMMEDIATELY_ENABLE_MAX 1 #define SCTPCTL_SACK_IMMEDIATELY_ENABLE_DEFAULT SCTPCTL_SACK_IMMEDIATELY_ENABLE_MIN /* Enable sending of the NAT-FRIENDLY message */ #define SCTPCTL_NAT_FRIENDLY_INITS_DESC "Enable sending of the nat-friendly SCTP option on INITs." #define SCTPCTL_NAT_FRIENDLY_INITS_MIN 0 #define SCTPCTL_NAT_FRIENDLY_INITS_MAX 1 #define SCTPCTL_NAT_FRIENDLY_INITS_DEFAULT SCTPCTL_NAT_FRIENDLY_INITS_MIN /* Vtag time wait in seconds */ #define SCTPCTL_TIME_WAIT_DESC "Vtag time wait time in seconds, 0 disables it." #define SCTPCTL_TIME_WAIT_MIN 0 #define SCTPCTL_TIME_WAIT_MAX 0xffffffff #define SCTPCTL_TIME_WAIT_DEFAULT SCTP_TIME_WAIT /* Enable Send/Receive buffer splitting */ #define SCTPCTL_BUFFER_SPLITTING_DESC "Enable send/receive buffer splitting." #define SCTPCTL_BUFFER_SPLITTING_MIN 0 #define SCTPCTL_BUFFER_SPLITTING_MAX 0x3 #define SCTPCTL_BUFFER_SPLITTING_DEFAULT SCTPCTL_BUFFER_SPLITTING_MIN /* Initial congestion window in MTU */ #define SCTPCTL_INITIAL_CWND_DESC "Initial congestion window in MTUs" #define SCTPCTL_INITIAL_CWND_MIN 0 #define SCTPCTL_INITIAL_CWND_MAX 0xffffffff #define SCTPCTL_INITIAL_CWND_DEFAULT 3 /* rttvar smooth avg for bw calc */ #define SCTPCTL_RTTVAR_BW_DESC "Shift amount for bw smoothing on rtt calc" #define SCTPCTL_RTTVAR_BW_MIN 0 #define SCTPCTL_RTTVAR_BW_MAX 32 #define SCTPCTL_RTTVAR_BW_DEFAULT 4 /* rttvar smooth avg for bw calc */ #define SCTPCTL_RTTVAR_RTT_DESC "Shift amount for rtt smoothing on rtt calc" #define SCTPCTL_RTTVAR_RTT_MIN 0 #define SCTPCTL_RTTVAR_RTT_MAX 32 #define SCTPCTL_RTTVAR_RTT_DEFAULT 5 #define SCTPCTL_RTTVAR_EQRET_DESC "What to return when rtt and bw are unchanged" #define SCTPCTL_RTTVAR_EQRET_MIN 0 #define SCTPCTL_RTTVAR_EQRET_MAX 1 #define SCTPCTL_RTTVAR_EQRET_DEFAULT 0 #define SCTPCTL_RTTVAR_STEADYS_DESC "How many the sames it takes to try step down of cwnd" #define SCTPCTL_RTTVAR_STEADYS_MIN 0 #define SCTPCTL_RTTVAR_STEADYS_MAX 0xFFFF #define SCTPCTL_RTTVAR_STEADYS_DEFAULT 20 /* 0 means disable feature */ #define SCTPCTL_RTTVAR_DCCCECN_DESC "Enable for RTCC CC datacenter ECN" #define SCTPCTL_RTTVAR_DCCCECN_MIN 0 #define SCTPCTL_RTTVAR_DCCCECN_MAX 1 #define SCTPCTL_RTTVAR_DCCCECN_DEFAULT 1 /* 0 means disable feature */ #if defined(SCTP_DEBUG) /* debug: Configure debug output */ #define SCTPCTL_DEBUG_DESC "Configure debug output" #define SCTPCTL_DEBUG_MIN 0 #define SCTPCTL_DEBUG_MAX 0xFFFFFFFF #define SCTPCTL_DEBUG_DEFAULT 0 #endif #if defined (__APPLE__) || defined(SCTP_SO_LOCK_TESTING) #define SCTPCTL_OUTPUT_UNLOCKED_DESC "Unlock socket when sending packets down to IP." #define SCTPCTL_OUTPUT_UNLOCKED_MIN 0 #define SCTPCTL_OUTPUT_UNLOCKED_MAX 1 #define SCTPCTL_OUTPUT_UNLOCKED_DEFAULT SCTPCTL_OUTPUT_UNLOCKED_MIN #endif #if defined(_KERNEL) || defined(__Userspace__) #if defined(SYSCTL_DECL) SYSCTL_DECL(_net_inet_sctp); #endif void sctp_init_sysctls(void); #endif /* _KERNEL */ #endif /* __sctp_sysctl_h__ */ Index: projects/largeSMP/sys/netinet/sctp_usrreq.c =================================================================== --- projects/largeSMP/sys/netinet/sctp_usrreq.c (revision 221494) +++ projects/largeSMP/sys/netinet/sctp_usrreq.c (revision 221495) @@ -1,5166 +1,5170 @@ /*- * Copyright (c) 2001-2008, by Cisco Systems, Inc. All rights reserved. * Copyright (c) 2008-2011, by Randall Stewart. All rights reserved. * Copyright (c) 2008-2011, by Michael Tuexen. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * a) Redistributions of source code must retain the above copyright notice, * this list of conditions and the following disclaimer. * * b) Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in * the documentation and/or other materials provided with the distribution. * * c) Neither the name of Cisco Systems, Inc. nor the names of its * contributors may be used to endorse or promote products derived * from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF * THE POSSIBILITY OF SUCH DAMAGE. */ /* $KAME: sctp_usrreq.c,v 1.48 2005/03/07 23:26:08 itojun Exp $ */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #if defined(INET6) #endif #include #include #include #include #include #include #include #include #include #include extern struct sctp_cc_functions sctp_cc_functions[]; extern struct sctp_ss_functions sctp_ss_functions[]; void sctp_init(void) { u_long sb_max_adj; /* Initialize and modify the sysctled variables */ sctp_init_sysctls(); if ((nmbclusters / 8) > SCTP_ASOC_MAX_CHUNKS_ON_QUEUE) SCTP_BASE_SYSCTL(sctp_max_chunks_on_queue) = (nmbclusters / 8); /* * Allow a user to take no more than 1/2 the number of clusters or * the SB_MAX whichever is smaller for the send window. */ sb_max_adj = (u_long)((u_quad_t) (SB_MAX) * MCLBYTES / (MSIZE + MCLBYTES)); SCTP_BASE_SYSCTL(sctp_sendspace) = min(sb_max_adj, (((uint32_t) nmbclusters / 2) * SCTP_DEFAULT_MAXSEGMENT)); /* * Now for the recv window, should we take the same amount? or * should I do 1/2 the SB_MAX instead in the SB_MAX min above. For * now I will just copy. */ SCTP_BASE_SYSCTL(sctp_recvspace) = SCTP_BASE_SYSCTL(sctp_sendspace); SCTP_BASE_VAR(first_time) = 0; SCTP_BASE_VAR(sctp_pcb_initialized) = 0; sctp_pcb_init(); #if defined(SCTP_PACKET_LOGGING) SCTP_BASE_VAR(packet_log_writers) = 0; SCTP_BASE_VAR(packet_log_end) = 0; bzero(&SCTP_BASE_VAR(packet_log_buffer), SCTP_PACKET_LOG_SIZE); #endif } void sctp_finish(void) { sctp_pcb_finish(); } void sctp_pathmtu_adjustment(struct sctp_inpcb *inp, struct sctp_tcb *stcb, struct sctp_nets *net, uint16_t nxtsz) { struct sctp_tmit_chunk *chk; uint16_t overhead; /* Adjust that too */ stcb->asoc.smallest_mtu = nxtsz; /* now off to subtract IP_DF flag if needed */ overhead = IP_HDR_SIZE; if (sctp_auth_is_required_chunk(SCTP_DATA, stcb->asoc.peer_auth_chunks)) { overhead += sctp_get_auth_chunk_len(stcb->asoc.peer_hmac_id); } TAILQ_FOREACH(chk, &stcb->asoc.send_queue, sctp_next) { if ((chk->send_size + overhead) > nxtsz) { chk->flags |= CHUNK_FLAGS_FRAGMENT_OK; } } TAILQ_FOREACH(chk, &stcb->asoc.sent_queue, sctp_next) { if ((chk->send_size + overhead) > nxtsz) { /* * For this guy we also mark for immediate resend * since we sent to big of chunk */ chk->flags |= CHUNK_FLAGS_FRAGMENT_OK; if (chk->sent < SCTP_DATAGRAM_RESEND) { sctp_flight_size_decrease(chk); sctp_total_flight_decrease(stcb, chk); } if (chk->sent != SCTP_DATAGRAM_RESEND) { sctp_ucount_incr(stcb->asoc.sent_queue_retran_cnt); } chk->sent = SCTP_DATAGRAM_RESEND; chk->rec.data.doing_fast_retransmit = 0; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_FLIGHT_LOGGING_ENABLE) { sctp_misc_ints(SCTP_FLIGHT_LOG_DOWN_PMTU, chk->whoTo->flight_size, chk->book_size, (uintptr_t) chk->whoTo, chk->rec.data.TSN_seq); } /* Clear any time so NO RTT is being done */ chk->do_rtt = 0; } } } #ifdef INET static void sctp_notify_mbuf(struct sctp_inpcb *inp, struct sctp_tcb *stcb, struct sctp_nets *net, struct ip *ip, struct sctphdr *sh) { struct icmp *icmph; int totsz, tmr_stopped = 0; uint16_t nxtsz; /* protection */ if ((inp == NULL) || (stcb == NULL) || (net == NULL) || (ip == NULL) || (sh == NULL)) { if (stcb != NULL) { SCTP_TCB_UNLOCK(stcb); } return; } /* First job is to verify the vtag matches what I would send */ if (ntohl(sh->v_tag) != (stcb->asoc.peer_vtag)) { SCTP_TCB_UNLOCK(stcb); return; } icmph = (struct icmp *)((caddr_t)ip - (sizeof(struct icmp) - sizeof(struct ip))); if (icmph->icmp_type != ICMP_UNREACH) { /* We only care about unreachable */ SCTP_TCB_UNLOCK(stcb); return; } if (icmph->icmp_code != ICMP_UNREACH_NEEDFRAG) { /* not a unreachable message due to frag. */ SCTP_TCB_UNLOCK(stcb); return; } totsz = ip->ip_len; nxtsz = ntohs(icmph->icmp_nextmtu); if (nxtsz == 0) { /* * old type router that does not tell us what the next size * mtu is. Rats we will have to guess (in a educated fashion * of course) */ nxtsz = sctp_get_prev_mtu(totsz); } /* Stop any PMTU timer */ if (SCTP_OS_TIMER_PENDING(&net->pmtu_timer.timer)) { tmr_stopped = 1; sctp_timer_stop(SCTP_TIMER_TYPE_PATHMTURAISE, inp, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_1); } /* Adjust destination size limit */ if (net->mtu > nxtsz) { net->mtu = nxtsz; if (net->port) { net->mtu -= sizeof(struct udphdr); } } /* now what about the ep? */ if (stcb->asoc.smallest_mtu > nxtsz) { sctp_pathmtu_adjustment(inp, stcb, net, nxtsz); } if (tmr_stopped) sctp_timer_start(SCTP_TIMER_TYPE_PATHMTURAISE, inp, stcb, net); SCTP_TCB_UNLOCK(stcb); } #endif void sctp_notify(struct sctp_inpcb *inp, struct ip *ip, struct sctphdr *sh, struct sockaddr *to, struct sctp_tcb *stcb, struct sctp_nets *net) { #if defined (__APPLE__) || defined(SCTP_SO_LOCK_TESTING) struct socket *so; #endif /* protection */ int reason; struct icmp *icmph; if ((inp == NULL) || (stcb == NULL) || (net == NULL) || (sh == NULL) || (to == NULL)) { if (stcb) SCTP_TCB_UNLOCK(stcb); return; } /* First job is to verify the vtag matches what I would send */ if (ntohl(sh->v_tag) != (stcb->asoc.peer_vtag)) { SCTP_TCB_UNLOCK(stcb); return; } icmph = (struct icmp *)((caddr_t)ip - (sizeof(struct icmp) - sizeof(struct ip))); if (icmph->icmp_type != ICMP_UNREACH) { /* We only care about unreachable */ SCTP_TCB_UNLOCK(stcb); return; } if ((icmph->icmp_code == ICMP_UNREACH_NET) || (icmph->icmp_code == ICMP_UNREACH_HOST) || (icmph->icmp_code == ICMP_UNREACH_NET_UNKNOWN) || (icmph->icmp_code == ICMP_UNREACH_HOST_UNKNOWN) || (icmph->icmp_code == ICMP_UNREACH_ISOLATED) || (icmph->icmp_code == ICMP_UNREACH_NET_PROHIB) || (icmph->icmp_code == ICMP_UNREACH_HOST_PROHIB) || (icmph->icmp_code == ICMP_UNREACH_FILTER_PROHIB)) { /* * Hmm reachablity problems we must examine closely. If its * not reachable, we may have lost a network. Or if there is * NO protocol at the other end named SCTP. well we consider * it a OOTB abort. */ if (net->dest_state & SCTP_ADDR_REACHABLE) { /* Ok that destination is NOT reachable */ SCTP_PRINTF("ICMP (thresh %d/%d) takes interface %p down\n", net->error_count, net->failure_threshold, net); net->dest_state &= ~SCTP_ADDR_REACHABLE; net->dest_state |= SCTP_ADDR_NOT_REACHABLE; /* * JRS 5/14/07 - If a destination is unreachable, * the PF bit is turned off. This allows an * unambiguous use of the PF bit for destinations * that are reachable but potentially failed. If the * destination is set to the unreachable state, also * set the destination to the PF state. */ /* * Add debug message here if destination is not in * PF state. */ /* Stop any running T3 timers here? */ if ((stcb->asoc.sctp_cmt_on_off > 0) && (stcb->asoc.sctp_cmt_pf > 0)) { net->dest_state &= ~SCTP_ADDR_PF; SCTPDBG(SCTP_DEBUG_TIMER4, "Destination %p moved from PF to unreachable.\n", net); } net->error_count = net->failure_threshold + 1; sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_DOWN, stcb, SCTP_FAILED_THRESHOLD, (void *)net, SCTP_SO_NOT_LOCKED); } SCTP_TCB_UNLOCK(stcb); } else if ((icmph->icmp_code == ICMP_UNREACH_PROTOCOL) || (icmph->icmp_code == ICMP_UNREACH_PORT)) { /* * Here the peer is either playing tricks on us, including * an address that belongs to someone who does not support * SCTP OR was a userland implementation that shutdown and * now is dead. In either case treat it like a OOTB abort * with no TCB */ reason = SCTP_PEER_FAULTY; sctp_abort_notification(stcb, reason, SCTP_SO_NOT_LOCKED); #if defined (__APPLE__) || defined(SCTP_SO_LOCK_TESTING) so = SCTP_INP_SO(inp); atomic_add_int(&stcb->asoc.refcnt, 1); SCTP_TCB_UNLOCK(stcb); SCTP_SOCKET_LOCK(so, 1); SCTP_TCB_LOCK(stcb); atomic_subtract_int(&stcb->asoc.refcnt, 1); #endif (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_2); #if defined (__APPLE__) || defined(SCTP_SO_LOCK_TESTING) SCTP_SOCKET_UNLOCK(so, 1); /* SCTP_TCB_UNLOCK(stcb); MT: I think this is not needed. */ #endif /* no need to unlock here, since the TCB is gone */ } else { SCTP_TCB_UNLOCK(stcb); } } #ifdef INET void sctp_ctlinput(cmd, sa, vip) int cmd; struct sockaddr *sa; void *vip; { struct ip *ip = vip; struct sctphdr *sh; uint32_t vrf_id; /* FIX, for non-bsd is this right? */ vrf_id = SCTP_DEFAULT_VRFID; if (sa->sa_family != AF_INET || ((struct sockaddr_in *)sa)->sin_addr.s_addr == INADDR_ANY) { return; } if (PRC_IS_REDIRECT(cmd)) { ip = 0; } else if ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0) { return; } if (ip) { struct sctp_inpcb *inp = NULL; struct sctp_tcb *stcb = NULL; struct sctp_nets *net = NULL; struct sockaddr_in to, from; sh = (struct sctphdr *)((caddr_t)ip + (ip->ip_hl << 2)); bzero(&to, sizeof(to)); bzero(&from, sizeof(from)); from.sin_family = to.sin_family = AF_INET; from.sin_len = to.sin_len = sizeof(to); from.sin_port = sh->src_port; from.sin_addr = ip->ip_src; to.sin_port = sh->dest_port; to.sin_addr = ip->ip_dst; /* * 'to' holds the dest of the packet that failed to be sent. * 'from' holds our local endpoint address. Thus we reverse * the to and the from in the lookup. */ stcb = sctp_findassociation_addr_sa((struct sockaddr *)&from, (struct sockaddr *)&to, &inp, &net, 1, vrf_id); if (stcb != NULL && inp && (inp->sctp_socket != NULL)) { if (cmd != PRC_MSGSIZE) { sctp_notify(inp, ip, sh, (struct sockaddr *)&to, stcb, net); } else { /* handle possible ICMP size messages */ sctp_notify_mbuf(inp, stcb, net, ip, sh); } } else { if ((stcb == NULL) && (inp != NULL)) { /* reduce ref-count */ SCTP_INP_WLOCK(inp); SCTP_INP_DECR_REF(inp); SCTP_INP_WUNLOCK(inp); } if (stcb) { SCTP_TCB_UNLOCK(stcb); } } } return; } #endif static int sctp_getcred(SYSCTL_HANDLER_ARGS) { struct xucred xuc; struct sockaddr_in addrs[2]; struct sctp_inpcb *inp; struct sctp_nets *net; struct sctp_tcb *stcb; int error; uint32_t vrf_id; /* FIX, for non-bsd is this right? */ vrf_id = SCTP_DEFAULT_VRFID; error = priv_check(req->td, PRIV_NETINET_GETCRED); if (error) return (error); error = SYSCTL_IN(req, addrs, sizeof(addrs)); if (error) return (error); stcb = sctp_findassociation_addr_sa(sintosa(&addrs[0]), sintosa(&addrs[1]), &inp, &net, 1, vrf_id); if (stcb == NULL || inp == NULL || inp->sctp_socket == NULL) { if ((inp != NULL) && (stcb == NULL)) { /* reduce ref-count */ SCTP_INP_WLOCK(inp); SCTP_INP_DECR_REF(inp); goto cred_can_cont; } SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); error = ENOENT; goto out; } SCTP_TCB_UNLOCK(stcb); /* * We use the write lock here, only since in the error leg we need * it. If we used RLOCK, then we would have to * wlock/decr/unlock/rlock. Which in theory could create a hole. * Better to use higher wlock. */ SCTP_INP_WLOCK(inp); cred_can_cont: error = cr_canseesocket(req->td->td_ucred, inp->sctp_socket); if (error) { SCTP_INP_WUNLOCK(inp); goto out; } cru2x(inp->sctp_socket->so_cred, &xuc); SCTP_INP_WUNLOCK(inp); error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred)); out: return (error); } SYSCTL_PROC(_net_inet_sctp, OID_AUTO, getcred, CTLTYPE_OPAQUE | CTLFLAG_RW, 0, 0, sctp_getcred, "S,ucred", "Get the ucred of a SCTP connection"); #ifdef INET static void sctp_abort(struct socket *so) { struct sctp_inpcb *inp; uint32_t flags; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == 0) { return; } sctp_must_try_again: flags = inp->sctp_flags; #ifdef SCTP_LOG_CLOSING sctp_log_closing(inp, NULL, 17); #endif if (((flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) && (atomic_cmpset_int(&inp->sctp_flags, flags, (flags | SCTP_PCB_FLAGS_SOCKET_GONE | SCTP_PCB_FLAGS_CLOSE_IP)))) { #ifdef SCTP_LOG_CLOSING sctp_log_closing(inp, NULL, 16); #endif sctp_inpcb_free(inp, SCTP_FREE_SHOULD_USE_ABORT, SCTP_CALLED_AFTER_CMPSET_OFCLOSE); SOCK_LOCK(so); SCTP_SB_CLEAR(so->so_snd); /* * same for the rcv ones, they are only here for the * accounting/select. */ SCTP_SB_CLEAR(so->so_rcv); /* Now null out the reference, we are completely detached. */ so->so_pcb = NULL; SOCK_UNLOCK(so); } else { flags = inp->sctp_flags; if ((flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) { goto sctp_must_try_again; } } return; } static int sctp_attach(struct socket *so, int proto, struct thread *p) { struct sctp_inpcb *inp; struct inpcb *ip_inp; int error; uint32_t vrf_id = SCTP_DEFAULT_VRFID; #ifdef IPSEC uint32_t flags; #endif inp = (struct sctp_inpcb *)so->so_pcb; if (inp != 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return EINVAL; } if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { error = SCTP_SORESERVE(so, SCTP_BASE_SYSCTL(sctp_sendspace), SCTP_BASE_SYSCTL(sctp_recvspace)); if (error) { return error; } } error = sctp_inpcb_alloc(so, vrf_id); if (error) { return error; } inp = (struct sctp_inpcb *)so->so_pcb; SCTP_INP_WLOCK(inp); inp->sctp_flags &= ~SCTP_PCB_FLAGS_BOUND_V6; /* I'm not v6! */ ip_inp = &inp->ip_inp.inp; ip_inp->inp_vflag |= INP_IPV4; ip_inp->inp_ip_ttl = MODULE_GLOBAL(ip_defttl); #ifdef IPSEC error = ipsec_init_policy(so, &ip_inp->inp_sp); #ifdef SCTP_LOG_CLOSING sctp_log_closing(inp, NULL, 17); #endif if (error != 0) { try_again: flags = inp->sctp_flags; if (((flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) && (atomic_cmpset_int(&inp->sctp_flags, flags, (flags | SCTP_PCB_FLAGS_SOCKET_GONE | SCTP_PCB_FLAGS_CLOSE_IP)))) { #ifdef SCTP_LOG_CLOSING sctp_log_closing(inp, NULL, 15); #endif SCTP_INP_WUNLOCK(inp); sctp_inpcb_free(inp, SCTP_FREE_SHOULD_USE_ABORT, SCTP_CALLED_AFTER_CMPSET_OFCLOSE); } else { flags = inp->sctp_flags; if ((flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) { goto try_again; } else { SCTP_INP_WUNLOCK(inp); } } return error; } #endif /* IPSEC */ SCTP_INP_WUNLOCK(inp); return 0; } static int sctp_bind(struct socket *so, struct sockaddr *addr, struct thread *p) { struct sctp_inpcb *inp = NULL; int error; #ifdef INET6 if (addr && addr->sa_family != AF_INET) { /* must be a v4 address! */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return EINVAL; } #endif /* INET6 */ if (addr && (addr->sa_len != sizeof(struct sockaddr_in))) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return EINVAL; } inp = (struct sctp_inpcb *)so->so_pcb; if (inp == 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return EINVAL; } error = sctp_inpcb_bind(so, addr, NULL, p); return error; } #endif void sctp_close(struct socket *so) { struct sctp_inpcb *inp; uint32_t flags; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == 0) return; /* * Inform all the lower layer assoc that we are done. */ sctp_must_try_again: flags = inp->sctp_flags; #ifdef SCTP_LOG_CLOSING sctp_log_closing(inp, NULL, 17); #endif if (((flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) && (atomic_cmpset_int(&inp->sctp_flags, flags, (flags | SCTP_PCB_FLAGS_SOCKET_GONE | SCTP_PCB_FLAGS_CLOSE_IP)))) { if (((so->so_options & SO_LINGER) && (so->so_linger == 0)) || (so->so_rcv.sb_cc > 0)) { #ifdef SCTP_LOG_CLOSING sctp_log_closing(inp, NULL, 13); #endif sctp_inpcb_free(inp, SCTP_FREE_SHOULD_USE_ABORT, SCTP_CALLED_AFTER_CMPSET_OFCLOSE); } else { #ifdef SCTP_LOG_CLOSING sctp_log_closing(inp, NULL, 14); #endif sctp_inpcb_free(inp, SCTP_FREE_SHOULD_USE_GRACEFUL_CLOSE, SCTP_CALLED_AFTER_CMPSET_OFCLOSE); } /* * The socket is now detached, no matter what the state of * the SCTP association. */ SOCK_LOCK(so); SCTP_SB_CLEAR(so->so_snd); /* * same for the rcv ones, they are only here for the * accounting/select. */ SCTP_SB_CLEAR(so->so_rcv); /* Now null out the reference, we are completely detached. */ so->so_pcb = NULL; SOCK_UNLOCK(so); } else { flags = inp->sctp_flags; if ((flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) { goto sctp_must_try_again; } } return; } int sctp_sendm(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr, struct mbuf *control, struct thread *p); int sctp_sendm(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr, struct mbuf *control, struct thread *p) { struct sctp_inpcb *inp; int error; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == 0) { if (control) { sctp_m_freem(control); control = NULL; } SCTP_LTRACE_ERR_RET_PKT(m, inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); sctp_m_freem(m); return EINVAL; } /* Got to have an to address if we are NOT a connected socket */ if ((addr == NULL) && ((inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) || (inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE)) ) { goto connected_type; } else if (addr == NULL) { SCTP_LTRACE_ERR_RET_PKT(m, inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EDESTADDRREQ); error = EDESTADDRREQ; sctp_m_freem(m); if (control) { sctp_m_freem(control); control = NULL; } return (error); } #ifdef INET6 if (addr->sa_family != AF_INET) { /* must be a v4 address! */ SCTP_LTRACE_ERR_RET_PKT(m, inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EDESTADDRREQ); sctp_m_freem(m); if (control) { sctp_m_freem(control); control = NULL; } error = EDESTADDRREQ; return EDESTADDRREQ; } #endif /* INET6 */ connected_type: /* now what about control */ if (control) { if (inp->control) { SCTP_PRINTF("huh? control set?\n"); sctp_m_freem(inp->control); inp->control = NULL; } inp->control = control; } /* Place the data */ if (inp->pkt) { SCTP_BUF_NEXT(inp->pkt_last) = m; inp->pkt_last = m; } else { inp->pkt_last = inp->pkt = m; } if ( /* FreeBSD uses a flag passed */ ((flags & PRUS_MORETOCOME) == 0) ) { /* * note with the current version this code will only be used * by OpenBSD-- NetBSD, FreeBSD, and MacOS have methods for * re-defining sosend to use the sctp_sosend. One can * optionally switch back to this code (by changing back the * definitions) but this is not advisable. This code is used * by FreeBSD when sending a file with sendfile() though. */ int ret; ret = sctp_output(inp, inp->pkt, addr, inp->control, p, flags); inp->pkt = NULL; inp->control = NULL; return (ret); } else { return (0); } } int sctp_disconnect(struct socket *so) { struct sctp_inpcb *inp; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOTCONN); return (ENOTCONN); } SCTP_INP_RLOCK(inp); if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) { if (LIST_EMPTY(&inp->sctp_asoc_list)) { /* No connection */ SCTP_INP_RUNLOCK(inp); return (0); } else { struct sctp_association *asoc; struct sctp_tcb *stcb; stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb == NULL) { SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } SCTP_TCB_LOCK(stcb); asoc = &stcb->asoc; if (stcb->asoc.state & SCTP_STATE_ABOUT_TO_BE_FREED) { /* We are about to be freed, out of here */ SCTP_TCB_UNLOCK(stcb); SCTP_INP_RUNLOCK(inp); return (0); } if (((so->so_options & SO_LINGER) && (so->so_linger == 0)) || (so->so_rcv.sb_cc > 0)) { if (SCTP_GET_STATE(asoc) != SCTP_STATE_COOKIE_WAIT) { /* Left with Data unread */ struct mbuf *err; err = sctp_get_mbuf_for_msg(sizeof(struct sctp_paramhdr), 0, M_DONTWAIT, 1, MT_DATA); if (err) { /* * Fill in the user * initiated abort */ struct sctp_paramhdr *ph; ph = mtod(err, struct sctp_paramhdr *); SCTP_BUF_LEN(err) = sizeof(struct sctp_paramhdr); ph->param_type = htons(SCTP_CAUSE_USER_INITIATED_ABT); ph->param_length = htons(SCTP_BUF_LEN(err)); } #if defined(SCTP_PANIC_ON_ABORT) panic("disconnect does an abort"); #endif sctp_send_abort_tcb(stcb, err, SCTP_SO_LOCKED); SCTP_STAT_INCR_COUNTER32(sctps_aborted); } SCTP_INP_RUNLOCK(inp); if ((SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_OPEN) || (SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_SHUTDOWN_RECEIVED)) { SCTP_STAT_DECR_GAUGE32(sctps_currestab); } (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_3); /* No unlock tcb assoc is gone */ return (0); } if (TAILQ_EMPTY(&asoc->send_queue) && TAILQ_EMPTY(&asoc->sent_queue) && (asoc->stream_queue_cnt == 0)) { /* there is nothing queued to send, so done */ if (asoc->locked_on_sending) { goto abort_anyway; } if ((SCTP_GET_STATE(asoc) != SCTP_STATE_SHUTDOWN_SENT) && (SCTP_GET_STATE(asoc) != SCTP_STATE_SHUTDOWN_ACK_SENT)) { /* only send SHUTDOWN 1st time thru */ sctp_stop_timers_for_shutdown(stcb); sctp_send_shutdown(stcb, stcb->asoc.primary_destination); sctp_chunk_output(stcb->sctp_ep, stcb, SCTP_OUTPUT_FROM_T3, SCTP_SO_LOCKED); if ((SCTP_GET_STATE(asoc) == SCTP_STATE_OPEN) || (SCTP_GET_STATE(asoc) == SCTP_STATE_SHUTDOWN_RECEIVED)) { SCTP_STAT_DECR_GAUGE32(sctps_currestab); } SCTP_SET_STATE(asoc, SCTP_STATE_SHUTDOWN_SENT); SCTP_CLEAR_SUBSTATE(asoc, SCTP_STATE_SHUTDOWN_PENDING); sctp_timer_start(SCTP_TIMER_TYPE_SHUTDOWN, stcb->sctp_ep, stcb, asoc->primary_destination); sctp_timer_start(SCTP_TIMER_TYPE_SHUTDOWNGUARD, stcb->sctp_ep, stcb, asoc->primary_destination); } } else { /* * we still got (or just got) data to send, * so set SHUTDOWN_PENDING */ /* * XXX sockets draft says that SCTP_EOF * should be sent with no data. currently, * we will allow user data to be sent first * and move to SHUTDOWN-PENDING */ asoc->state |= SCTP_STATE_SHUTDOWN_PENDING; sctp_timer_start(SCTP_TIMER_TYPE_SHUTDOWNGUARD, stcb->sctp_ep, stcb, asoc->primary_destination); if (asoc->locked_on_sending) { /* Locked to send out the data */ struct sctp_stream_queue_pending *sp; sp = TAILQ_LAST(&asoc->locked_on_sending->outqueue, sctp_streamhead); if (sp == NULL) { SCTP_PRINTF("Error, sp is NULL, locked on sending is non-null strm:%d\n", asoc->locked_on_sending->stream_no); } else { if ((sp->length == 0) && (sp->msg_is_complete == 0)) asoc->state |= SCTP_STATE_PARTIAL_MSG_LEFT; } } if (TAILQ_EMPTY(&asoc->send_queue) && TAILQ_EMPTY(&asoc->sent_queue) && (asoc->state & SCTP_STATE_PARTIAL_MSG_LEFT)) { struct mbuf *op_err; abort_anyway: op_err = sctp_get_mbuf_for_msg((sizeof(struct sctp_paramhdr) + sizeof(uint32_t)), 0, M_DONTWAIT, 1, MT_DATA); if (op_err) { /* * Fill in the user * initiated abort */ struct sctp_paramhdr *ph; uint32_t *ippp; SCTP_BUF_LEN(op_err) = (sizeof(struct sctp_paramhdr) + sizeof(uint32_t)); ph = mtod(op_err, struct sctp_paramhdr *); ph->param_type = htons( SCTP_CAUSE_USER_INITIATED_ABT); ph->param_length = htons(SCTP_BUF_LEN(op_err)); ippp = (uint32_t *) (ph + 1); *ippp = htonl(SCTP_FROM_SCTP_USRREQ + SCTP_LOC_4); } #if defined(SCTP_PANIC_ON_ABORT) panic("disconnect does an abort"); #endif stcb->sctp_ep->last_abort_code = SCTP_FROM_SCTP_USRREQ + SCTP_LOC_4; sctp_send_abort_tcb(stcb, op_err, SCTP_SO_LOCKED); SCTP_STAT_INCR_COUNTER32(sctps_aborted); if ((SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_OPEN) || (SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_SHUTDOWN_RECEIVED)) { SCTP_STAT_DECR_GAUGE32(sctps_currestab); } SCTP_INP_RUNLOCK(inp); (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_5); return (0); } else { sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_CLOSING, SCTP_SO_LOCKED); } } soisdisconnecting(so); SCTP_TCB_UNLOCK(stcb); SCTP_INP_RUNLOCK(inp); return (0); } /* not reached */ } else { /* UDP model does not support this */ SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); return EOPNOTSUPP; } } int sctp_flush(struct socket *so, int how) { /* * We will just clear out the values and let subsequent close clear * out the data, if any. Note if the user did a shutdown(SHUT_RD) * they will not be able to read the data, the socket will block * that from happening. */ struct sctp_inpcb *inp; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return EINVAL; } SCTP_INP_RLOCK(inp); /* For the 1 to many model this does nothing */ if (inp->sctp_flags & SCTP_PCB_FLAGS_UDPTYPE) { SCTP_INP_RUNLOCK(inp); return (0); } SCTP_INP_RUNLOCK(inp); if ((how == PRU_FLUSH_RD) || (how == PRU_FLUSH_RDWR)) { /* * First make sure the sb will be happy, we don't use these * except maybe the count */ SCTP_INP_WLOCK(inp); SCTP_INP_READ_LOCK(inp); inp->sctp_flags |= SCTP_PCB_FLAGS_SOCKET_CANT_READ; SCTP_INP_READ_UNLOCK(inp); SCTP_INP_WUNLOCK(inp); so->so_rcv.sb_cc = 0; so->so_rcv.sb_mbcnt = 0; so->so_rcv.sb_mb = NULL; } if ((how == PRU_FLUSH_WR) || (how == PRU_FLUSH_RDWR)) { /* * First make sure the sb will be happy, we don't use these * except maybe the count */ so->so_snd.sb_cc = 0; so->so_snd.sb_mbcnt = 0; so->so_snd.sb_mb = NULL; } return (0); } int sctp_shutdown(struct socket *so) { struct sctp_inpcb *inp; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return EINVAL; } SCTP_INP_RLOCK(inp); /* For UDP model this is a invalid call */ if (inp->sctp_flags & SCTP_PCB_FLAGS_UDPTYPE) { /* Restore the flags that the soshutdown took away. */ SOCKBUF_LOCK(&so->so_rcv); so->so_rcv.sb_state &= ~SBS_CANTRCVMORE; SOCKBUF_UNLOCK(&so->so_rcv); /* This proc will wakeup for read and do nothing (I hope) */ SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); return (EOPNOTSUPP); } /* * Ok if we reach here its the TCP model and it is either a SHUT_WR * or SHUT_RDWR. This means we put the shutdown flag against it. */ { struct sctp_tcb *stcb; struct sctp_association *asoc; if ((so->so_state & (SS_ISCONNECTED | SS_ISCONNECTING | SS_ISDISCONNECTING)) == 0) { SCTP_INP_RUNLOCK(inp); return (ENOTCONN); } socantsendmore(so); stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb == NULL) { /* * Ok we hit the case that the shutdown call was * made after an abort or something. Nothing to do * now. */ SCTP_INP_RUNLOCK(inp); return (0); } SCTP_TCB_LOCK(stcb); asoc = &stcb->asoc; if (TAILQ_EMPTY(&asoc->send_queue) && TAILQ_EMPTY(&asoc->sent_queue) && (asoc->stream_queue_cnt == 0)) { if (asoc->locked_on_sending) { goto abort_anyway; } /* there is nothing queued to send, so I'm done... */ if (SCTP_GET_STATE(asoc) != SCTP_STATE_SHUTDOWN_SENT) { /* only send SHUTDOWN the first time through */ sctp_stop_timers_for_shutdown(stcb); sctp_send_shutdown(stcb, stcb->asoc.primary_destination); sctp_chunk_output(stcb->sctp_ep, stcb, SCTP_OUTPUT_FROM_T3, SCTP_SO_LOCKED); if ((SCTP_GET_STATE(asoc) == SCTP_STATE_OPEN) || (SCTP_GET_STATE(asoc) == SCTP_STATE_SHUTDOWN_RECEIVED)) { SCTP_STAT_DECR_GAUGE32(sctps_currestab); } SCTP_SET_STATE(asoc, SCTP_STATE_SHUTDOWN_SENT); SCTP_CLEAR_SUBSTATE(asoc, SCTP_STATE_SHUTDOWN_PENDING); sctp_timer_start(SCTP_TIMER_TYPE_SHUTDOWN, stcb->sctp_ep, stcb, asoc->primary_destination); sctp_timer_start(SCTP_TIMER_TYPE_SHUTDOWNGUARD, stcb->sctp_ep, stcb, asoc->primary_destination); } } else { /* * we still got (or just got) data to send, so set * SHUTDOWN_PENDING */ asoc->state |= SCTP_STATE_SHUTDOWN_PENDING; sctp_timer_start(SCTP_TIMER_TYPE_SHUTDOWNGUARD, stcb->sctp_ep, stcb, asoc->primary_destination); if (asoc->locked_on_sending) { /* Locked to send out the data */ struct sctp_stream_queue_pending *sp; sp = TAILQ_LAST(&asoc->locked_on_sending->outqueue, sctp_streamhead); if (sp == NULL) { SCTP_PRINTF("Error, sp is NULL, locked on sending is non-null strm:%d\n", asoc->locked_on_sending->stream_no); } else { if ((sp->length == 0) && (sp->msg_is_complete == 0)) { asoc->state |= SCTP_STATE_PARTIAL_MSG_LEFT; } } } if (TAILQ_EMPTY(&asoc->send_queue) && TAILQ_EMPTY(&asoc->sent_queue) && (asoc->state & SCTP_STATE_PARTIAL_MSG_LEFT)) { struct mbuf *op_err; abort_anyway: op_err = sctp_get_mbuf_for_msg((sizeof(struct sctp_paramhdr) + sizeof(uint32_t)), 0, M_DONTWAIT, 1, MT_DATA); if (op_err) { /* Fill in the user initiated abort */ struct sctp_paramhdr *ph; uint32_t *ippp; SCTP_BUF_LEN(op_err) = sizeof(struct sctp_paramhdr) + sizeof(uint32_t); ph = mtod(op_err, struct sctp_paramhdr *); ph->param_type = htons( SCTP_CAUSE_USER_INITIATED_ABT); ph->param_length = htons(SCTP_BUF_LEN(op_err)); ippp = (uint32_t *) (ph + 1); *ippp = htonl(SCTP_FROM_SCTP_USRREQ + SCTP_LOC_6); } #if defined(SCTP_PANIC_ON_ABORT) panic("shutdown does an abort"); #endif stcb->sctp_ep->last_abort_code = SCTP_FROM_SCTP_USRREQ + SCTP_LOC_6; sctp_abort_an_association(stcb->sctp_ep, stcb, SCTP_RESPONSE_TO_USER_REQ, op_err, SCTP_SO_LOCKED); goto skip_unlock; } else { sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_CLOSING, SCTP_SO_LOCKED); } } SCTP_TCB_UNLOCK(stcb); } skip_unlock: SCTP_INP_RUNLOCK(inp); return 0; } /* * copies a "user" presentable address and removes embedded scope, etc. * returns 0 on success, 1 on error */ static uint32_t sctp_fill_user_address(struct sockaddr_storage *ss, struct sockaddr *sa) { #ifdef INET6 struct sockaddr_in6 lsa6; sa = (struct sockaddr *)sctp_recover_scope((struct sockaddr_in6 *)sa, &lsa6); #endif memcpy(ss, sa, sa->sa_len); return (0); } /* * NOTE: assumes addr lock is held */ static size_t sctp_fill_up_addresses_vrf(struct sctp_inpcb *inp, struct sctp_tcb *stcb, size_t limit, struct sockaddr_storage *sas, uint32_t vrf_id) { struct sctp_ifn *sctp_ifn; struct sctp_ifa *sctp_ifa; int loopback_scope, ipv4_local_scope, local_scope, site_scope; size_t actual; int ipv4_addr_legal, ipv6_addr_legal; struct sctp_vrf *vrf; actual = 0; if (limit <= 0) return (actual); if (stcb) { /* Turn on all the appropriate scope */ loopback_scope = stcb->asoc.loopback_scope; ipv4_local_scope = stcb->asoc.ipv4_local_scope; local_scope = stcb->asoc.local_scope; site_scope = stcb->asoc.site_scope; } else { /* Turn on ALL scope, since we look at the EP */ loopback_scope = ipv4_local_scope = local_scope = site_scope = 1; } ipv4_addr_legal = ipv6_addr_legal = 0; if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { ipv6_addr_legal = 1; if (SCTP_IPV6_V6ONLY(inp) == 0) { ipv4_addr_legal = 1; } } else { ipv4_addr_legal = 1; } vrf = sctp_find_vrf(vrf_id); if (vrf == NULL) { return (0); } if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) { LIST_FOREACH(sctp_ifn, &vrf->ifnlist, next_ifn) { if ((loopback_scope == 0) && SCTP_IFN_IS_IFT_LOOP(sctp_ifn)) { /* Skip loopback if loopback_scope not set */ continue; } LIST_FOREACH(sctp_ifa, &sctp_ifn->ifalist, next_ifa) { if (stcb) { /* * For the BOUND-ALL case, the list * associated with a TCB is Always * considered a reverse list.. i.e. * it lists addresses that are NOT * part of the association. If this * is one of those we must skip it. */ if (sctp_is_addr_restricted(stcb, sctp_ifa)) { continue; } } switch (sctp_ifa->address.sa.sa_family) { #ifdef INET case AF_INET: if (ipv4_addr_legal) { struct sockaddr_in *sin; sin = (struct sockaddr_in *)&sctp_ifa->address.sa; if (sin->sin_addr.s_addr == 0) { /* * we skip * unspecifed * addresses */ continue; } if ((ipv4_local_scope == 0) && (IN4_ISPRIVATE_ADDRESS(&sin->sin_addr))) { continue; } #ifdef INET6 if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4)) { in6_sin_2_v4mapsin6(sin, (struct sockaddr_in6 *)sas); ((struct sockaddr_in6 *)sas)->sin6_port = inp->sctp_lport; sas = (struct sockaddr_storage *)((caddr_t)sas + sizeof(struct sockaddr_in6)); actual += sizeof(struct sockaddr_in6); } else { #endif memcpy(sas, sin, sizeof(*sin)); ((struct sockaddr_in *)sas)->sin_port = inp->sctp_lport; sas = (struct sockaddr_storage *)((caddr_t)sas + sizeof(*sin)); actual += sizeof(*sin); #ifdef INET6 } #endif if (actual >= limit) { return (actual); } } else { continue; } break; #endif #ifdef INET6 case AF_INET6: if (ipv6_addr_legal) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)&sctp_ifa->address.sa; if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { /* * we skip * unspecifed * addresses */ continue; } if (IN6_IS_ADDR_LINKLOCAL(&sin6->sin6_addr)) { if (local_scope == 0) continue; if (sin6->sin6_scope_id == 0) { if (sa6_recoverscope(sin6) != 0) /* * * bad * * li * nk * * loc * al * * add * re * ss * */ continue; } } if ((site_scope == 0) && (IN6_IS_ADDR_SITELOCAL(&sin6->sin6_addr))) { continue; } memcpy(sas, sin6, sizeof(*sin6)); ((struct sockaddr_in6 *)sas)->sin6_port = inp->sctp_lport; sas = (struct sockaddr_storage *)((caddr_t)sas + sizeof(*sin6)); actual += sizeof(*sin6); if (actual >= limit) { return (actual); } } else { continue; } break; #endif default: /* TSNH */ break; } } } } else { struct sctp_laddr *laddr; LIST_FOREACH(laddr, &inp->sctp_addr_list, sctp_nxt_addr) { if (stcb) { if (sctp_is_addr_restricted(stcb, laddr->ifa)) { continue; } } if (sctp_fill_user_address(sas, &laddr->ifa->address.sa)) continue; ((struct sockaddr_in6 *)sas)->sin6_port = inp->sctp_lport; sas = (struct sockaddr_storage *)((caddr_t)sas + laddr->ifa->address.sa.sa_len); actual += laddr->ifa->address.sa.sa_len; if (actual >= limit) { return (actual); } } } return (actual); } static size_t sctp_fill_up_addresses(struct sctp_inpcb *inp, struct sctp_tcb *stcb, size_t limit, struct sockaddr_storage *sas) { size_t size = 0; SCTP_IPI_ADDR_RLOCK(); /* fill up addresses for the endpoint's default vrf */ size = sctp_fill_up_addresses_vrf(inp, stcb, limit, sas, inp->def_vrf_id); SCTP_IPI_ADDR_RUNLOCK(); return (size); } /* * NOTE: assumes addr lock is held */ static int sctp_count_max_addresses_vrf(struct sctp_inpcb *inp, uint32_t vrf_id) { int cnt = 0; struct sctp_vrf *vrf = NULL; /* * In both sub-set bound an bound_all cases we return the MAXIMUM * number of addresses that you COULD get. In reality the sub-set * bound may have an exclusion list for a given TCB OR in the * bound-all case a TCB may NOT include the loopback or other * addresses as well. */ vrf = sctp_find_vrf(vrf_id); if (vrf == NULL) { return (0); } if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) { struct sctp_ifn *sctp_ifn; struct sctp_ifa *sctp_ifa; LIST_FOREACH(sctp_ifn, &vrf->ifnlist, next_ifn) { LIST_FOREACH(sctp_ifa, &sctp_ifn->ifalist, next_ifa) { /* Count them if they are the right type */ switch (sctp_ifa->address.sa.sa_family) { #ifdef INET case AF_INET: if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4)) cnt += sizeof(struct sockaddr_in6); else cnt += sizeof(struct sockaddr_in); break; #endif #ifdef INET6 case AF_INET6: cnt += sizeof(struct sockaddr_in6); break; #endif default: break; } } } } else { struct sctp_laddr *laddr; LIST_FOREACH(laddr, &inp->sctp_addr_list, sctp_nxt_addr) { switch (laddr->ifa->address.sa.sa_family) { #ifdef INET case AF_INET: if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4)) cnt += sizeof(struct sockaddr_in6); else cnt += sizeof(struct sockaddr_in); break; #endif #ifdef INET6 case AF_INET6: cnt += sizeof(struct sockaddr_in6); break; #endif default: break; } } } return (cnt); } static int sctp_count_max_addresses(struct sctp_inpcb *inp) { int cnt = 0; SCTP_IPI_ADDR_RLOCK(); /* count addresses for the endpoint's default VRF */ cnt = sctp_count_max_addresses_vrf(inp, inp->def_vrf_id); SCTP_IPI_ADDR_RUNLOCK(); return (cnt); } static int sctp_do_connect_x(struct socket *so, struct sctp_inpcb *inp, void *optval, size_t optsize, void *p, int delay) { int error = 0; int creat_lock_on = 0; struct sctp_tcb *stcb = NULL; struct sockaddr *sa; int num_v6 = 0, num_v4 = 0, *totaddrp, totaddr; int added = 0; uint32_t vrf_id; int bad_addresses = 0; sctp_assoc_t *a_id; SCTPDBG(SCTP_DEBUG_PCB1, "Connectx called\n"); if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) && (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED)) { /* We are already connected AND the TCP model */ SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, EADDRINUSE); return (EADDRINUSE); } if ((inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) && (sctp_is_feature_off(inp, SCTP_PCB_FLAGS_PORTREUSE))) { SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } if (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) { SCTP_INP_RLOCK(inp); stcb = LIST_FIRST(&inp->sctp_asoc_list); SCTP_INP_RUNLOCK(inp); } if (stcb) { SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, EALREADY); return (EALREADY); } SCTP_INP_INCR_REF(inp); SCTP_ASOC_CREATE_LOCK(inp); creat_lock_on = 1; if ((inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE) || (inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE)) { SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, EFAULT); error = EFAULT; goto out_now; } totaddrp = (int *)optval; totaddr = *totaddrp; sa = (struct sockaddr *)(totaddrp + 1); stcb = sctp_connectx_helper_find(inp, sa, &totaddr, &num_v4, &num_v6, &error, (optsize - sizeof(int)), &bad_addresses); if ((stcb != NULL) || bad_addresses) { /* Already have or am bring up an association */ SCTP_ASOC_CREATE_UNLOCK(inp); creat_lock_on = 0; if (stcb) SCTP_TCB_UNLOCK(stcb); if (bad_addresses == 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EALREADY); error = EALREADY; } goto out_now; } #ifdef INET6 if (((inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) == 0) && (num_v6 > 0)) { error = EINVAL; goto out_now; } if ((inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) && (num_v4 > 0)) { struct in6pcb *inp6; inp6 = (struct in6pcb *)inp; if (SCTP_IPV6_V6ONLY(inp6)) { /* * if IPV6_V6ONLY flag, ignore connections destined * to a v4 addr or v4-mapped addr */ SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto out_now; } } #endif /* INET6 */ if ((inp->sctp_flags & SCTP_PCB_FLAGS_UNBOUND) == SCTP_PCB_FLAGS_UNBOUND) { /* Bind a ephemeral port */ error = sctp_inpcb_bind(so, NULL, NULL, p); if (error) { goto out_now; } } /* FIX ME: do we want to pass in a vrf on the connect call? */ vrf_id = inp->def_vrf_id; /* We are GOOD to go */ stcb = sctp_aloc_assoc(inp, sa, &error, 0, vrf_id, (struct thread *)p ); if (stcb == NULL) { /* Gak! no memory */ goto out_now; } SCTP_SET_STATE(&stcb->asoc, SCTP_STATE_COOKIE_WAIT); /* move to second address */ switch (sa->sa_family) { #ifdef INET case AF_INET: sa = (struct sockaddr *)((caddr_t)sa + sizeof(struct sockaddr_in)); break; #endif #ifdef INET6 case AF_INET6: sa = (struct sockaddr *)((caddr_t)sa + sizeof(struct sockaddr_in6)); break; #endif default: break; } error = 0; added = sctp_connectx_helper_add(stcb, sa, (totaddr - 1), &error); /* Fill in the return id */ if (error) { (void)sctp_free_assoc(inp, stcb, SCTP_PCBFREE_FORCE, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_6); goto out_now; } a_id = (sctp_assoc_t *) optval; *a_id = sctp_get_associd(stcb); /* initialize authentication parameters for the assoc */ sctp_initialize_auth_params(inp, stcb); if (delay) { /* doing delayed connection */ stcb->asoc.delayed_connection = 1; sctp_timer_start(SCTP_TIMER_TYPE_INIT, inp, stcb, stcb->asoc.primary_destination); } else { (void)SCTP_GETTIME_TIMEVAL(&stcb->asoc.time_entered); sctp_send_initiate(inp, stcb, SCTP_SO_LOCKED); } SCTP_TCB_UNLOCK(stcb); if (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) { stcb->sctp_ep->sctp_flags |= SCTP_PCB_FLAGS_CONNECTED; /* Set the connected flag so we can queue data */ soisconnecting(so); } out_now: if (creat_lock_on) { SCTP_ASOC_CREATE_UNLOCK(inp); } SCTP_INP_DECR_REF(inp); return error; } #define SCTP_FIND_STCB(inp, stcb, assoc_id) { \ if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) ||\ (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) { \ SCTP_INP_RLOCK(inp); \ stcb = LIST_FIRST(&inp->sctp_asoc_list); \ if (stcb) { \ SCTP_TCB_LOCK(stcb); \ } \ SCTP_INP_RUNLOCK(inp); \ } else if (assoc_id != 0) { \ stcb = sctp_findassociation_ep_asocid(inp, assoc_id, 1); \ if (stcb == NULL) { \ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); \ error = ENOENT; \ break; \ } \ } else { \ stcb = NULL; \ } \ } #define SCTP_CHECK_AND_CAST(destp, srcp, type, size) {\ if (size < sizeof(type)) { \ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); \ error = EINVAL; \ break; \ } else { \ destp = (type *)srcp; \ } \ } static int sctp_getopt(struct socket *so, int optname, void *optval, size_t *optsize, void *p) { struct sctp_inpcb *inp = NULL; int error, val = 0; struct sctp_tcb *stcb = NULL; if (optval == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } inp = (struct sctp_inpcb *)so->so_pcb; if (inp == 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return EINVAL; } error = 0; switch (optname) { case SCTP_NODELAY: case SCTP_AUTOCLOSE: case SCTP_EXPLICIT_EOR: case SCTP_AUTO_ASCONF: case SCTP_DISABLE_FRAGMENTS: case SCTP_I_WANT_MAPPED_V4_ADDR: case SCTP_USE_EXT_RCVINFO: SCTP_INP_RLOCK(inp); switch (optname) { case SCTP_DISABLE_FRAGMENTS: val = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NO_FRAGMENT); break; case SCTP_I_WANT_MAPPED_V4_ADDR: val = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4); break; case SCTP_AUTO_ASCONF: if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) { /* only valid for bound all sockets */ val = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_AUTO_ASCONF); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto flags_out; } break; case SCTP_EXPLICIT_EOR: val = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_EXPLICIT_EOR); break; case SCTP_NODELAY: val = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NODELAY); break; case SCTP_USE_EXT_RCVINFO: val = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_EXT_RCVINFO); break; case SCTP_AUTOCLOSE: if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_AUTOCLOSE)) val = TICKS_TO_SEC(inp->sctp_ep.auto_close_time); else val = 0; break; default: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOPROTOOPT); error = ENOPROTOOPT; } /* end switch (sopt->sopt_name) */ if (optname != SCTP_AUTOCLOSE) { /* make it an "on/off" value */ val = (val != 0); } if (*optsize < sizeof(val)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } flags_out: SCTP_INP_RUNLOCK(inp); if (error == 0) { /* return the option value */ *(int *)optval = val; *optsize = sizeof(val); } break; case SCTP_GET_PACKET_LOG: { #ifdef SCTP_PACKET_LOGGING uint8_t *target; int ret; SCTP_CHECK_AND_CAST(target, optval, uint8_t, *optsize); ret = sctp_copy_out_packet_log(target, (int)*optsize); *optsize = ret; #else SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); error = EOPNOTSUPP; #endif break; } case SCTP_REUSE_PORT: { uint32_t *value; if ((inp->sctp_flags & SCTP_PCB_FLAGS_UDPTYPE)) { /* Can't do this for a 1-m socket */ error = EINVAL; break; } SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); *value = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_PORTREUSE); *optsize = sizeof(uint32_t); } break; case SCTP_PARTIAL_DELIVERY_POINT: { uint32_t *value; SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); *value = inp->partial_delivery_point; *optsize = sizeof(uint32_t); } break; case SCTP_FRAGMENT_INTERLEAVE: { uint32_t *value; SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_FRAG_INTERLEAVE)) { if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_INTERLEAVE_STRMS)) { *value = SCTP_FRAG_LEVEL_2; } else { *value = SCTP_FRAG_LEVEL_1; } } else { *value = SCTP_FRAG_LEVEL_0; } *optsize = sizeof(uint32_t); } break; case SCTP_CMT_ON_OFF: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.sctp_cmt_on_off; SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_RLOCK(inp); av->assoc_value = inp->sctp_cmt_on_off; SCTP_INP_RUNLOCK(inp); } *optsize = sizeof(*av); } break; /* JRS - Get socket option for pluggable congestion control */ case SCTP_PLUGGABLE_CC: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.congestion_control_module; SCTP_TCB_UNLOCK(stcb); } else { av->assoc_value = inp->sctp_ep.sctp_default_cc_module; } *optsize = sizeof(*av); } break; case SCTP_CC_OPTION: { struct sctp_cc_option *cc_opt; SCTP_CHECK_AND_CAST(cc_opt, optval, struct sctp_cc_option, *optsize); SCTP_FIND_STCB(inp, stcb, cc_opt->aid_value.assoc_id); if (stcb == NULL) { error = EINVAL; } else { if (stcb->asoc.cc_functions.sctp_cwnd_socket_option == NULL) { error = ENOTSUP; } else { error = (*stcb->asoc.cc_functions.sctp_cwnd_socket_option) (stcb, 0, cc_opt); *optsize = sizeof(*cc_opt); } SCTP_TCB_UNLOCK(stcb); } } break; /* RS - Get socket option for pluggable stream scheduling */ case SCTP_PLUGGABLE_SS: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.stream_scheduling_module; SCTP_TCB_UNLOCK(stcb); } else { av->assoc_value = inp->sctp_ep.sctp_default_ss_module; } *optsize = sizeof(*av); } break; case SCTP_SS_VALUE: { struct sctp_stream_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_stream_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { if (stcb->asoc.ss_functions.sctp_ss_get_value(stcb, &stcb->asoc, &stcb->asoc.strmout[av->stream_id], &av->stream_value) < 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } else { *optsize = sizeof(*av); } SCTP_TCB_UNLOCK(stcb); } else { /* * Can't get stream value without * association */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; case SCTP_GET_ADDR_LEN: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); error = EINVAL; #ifdef INET if (av->assoc_value == AF_INET) { av->assoc_value = sizeof(struct sockaddr_in); error = 0; } #endif #ifdef INET6 if (av->assoc_value == AF_INET6) { av->assoc_value = sizeof(struct sockaddr_in6); error = 0; } #endif if (error) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); } *optsize = sizeof(*av); } break; case SCTP_GET_ASSOC_NUMBER: { uint32_t *value, cnt; SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); cnt = 0; SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { cnt++; } SCTP_INP_RUNLOCK(inp); *value = cnt; *optsize = sizeof(uint32_t); } break; case SCTP_GET_ASSOC_ID_LIST: { struct sctp_assoc_ids *ids; unsigned int at, limit; SCTP_CHECK_AND_CAST(ids, optval, struct sctp_assoc_ids, *optsize); at = 0; limit = (*optsize - sizeof(uint32_t)) / sizeof(sctp_assoc_t); SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { if (at < limit) { ids->gaids_assoc_id[at++] = sctp_get_associd(stcb); } else { error = EINVAL; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } SCTP_INP_RUNLOCK(inp); ids->gaids_number_of_ids = at; *optsize = ((at * sizeof(sctp_assoc_t)) + sizeof(uint32_t)); } break; case SCTP_CONTEXT: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.context; SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_RLOCK(inp); av->assoc_value = inp->sctp_context; SCTP_INP_RUNLOCK(inp); } *optsize = sizeof(*av); } break; case SCTP_VRF_ID: { uint32_t *default_vrfid; SCTP_CHECK_AND_CAST(default_vrfid, optval, uint32_t, *optsize); *default_vrfid = inp->def_vrf_id; break; } case SCTP_GET_ASOC_VRF: { struct sctp_assoc_value *id; SCTP_CHECK_AND_CAST(id, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, id->assoc_id); if (stcb == NULL) { error = EINVAL; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } id->assoc_value = stcb->asoc.vrf_id; break; } case SCTP_GET_VRF_IDS: { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); error = EOPNOTSUPP; break; } case SCTP_GET_NONCE_VALUES: { struct sctp_get_nonce_values *gnv; SCTP_CHECK_AND_CAST(gnv, optval, struct sctp_get_nonce_values, *optsize); SCTP_FIND_STCB(inp, stcb, gnv->gn_assoc_id); if (stcb) { gnv->gn_peers_tag = stcb->asoc.peer_vtag; gnv->gn_local_tag = stcb->asoc.my_vtag; SCTP_TCB_UNLOCK(stcb); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOTCONN); error = ENOTCONN; } *optsize = sizeof(*gnv); } break; case SCTP_DELAYED_SACK: { struct sctp_sack_info *sack; SCTP_CHECK_AND_CAST(sack, optval, struct sctp_sack_info, *optsize); SCTP_FIND_STCB(inp, stcb, sack->sack_assoc_id); if (stcb) { sack->sack_delay = stcb->asoc.delayed_ack; sack->sack_freq = stcb->asoc.sack_freq; SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_RLOCK(inp); sack->sack_delay = TICKS_TO_MSEC(inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_RECV]); sack->sack_freq = inp->sctp_ep.sctp_sack_freq; SCTP_INP_RUNLOCK(inp); } *optsize = sizeof(*sack); } break; case SCTP_GET_SNDBUF_USE: { struct sctp_sockstat *ss; SCTP_CHECK_AND_CAST(ss, optval, struct sctp_sockstat, *optsize); SCTP_FIND_STCB(inp, stcb, ss->ss_assoc_id); if (stcb) { ss->ss_total_sndbuf = stcb->asoc.total_output_queue_size; ss->ss_total_recv_buf = (stcb->asoc.size_on_reasm_queue + stcb->asoc.size_on_all_streams); SCTP_TCB_UNLOCK(stcb); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOTCONN); error = ENOTCONN; } *optsize = sizeof(struct sctp_sockstat); } break; case SCTP_MAX_BURST: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.max_burst; SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_RLOCK(inp); av->assoc_value = inp->sctp_ep.max_burst; SCTP_INP_RUNLOCK(inp); } *optsize = sizeof(struct sctp_assoc_value); } break; case SCTP_MAXSEG: { struct sctp_assoc_value *av; int ovh; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = sctp_get_frag_point(stcb, &stcb->asoc); SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_RLOCK(inp); if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { ovh = SCTP_MED_OVERHEAD; } else { ovh = SCTP_MED_V4_OVERHEAD; } if (inp->sctp_frag_point >= SCTP_DEFAULT_MAXSEGMENT) av->assoc_value = 0; else av->assoc_value = inp->sctp_frag_point - ovh; SCTP_INP_RUNLOCK(inp); } *optsize = sizeof(struct sctp_assoc_value); } break; case SCTP_GET_STAT_LOG: error = sctp_fill_stat_log(optval, optsize); break; case SCTP_EVENTS: { struct sctp_event_subscribe *events; SCTP_CHECK_AND_CAST(events, optval, struct sctp_event_subscribe, *optsize); memset(events, 0, sizeof(*events)); SCTP_INP_RLOCK(inp); if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVDATAIOEVNT)) events->sctp_data_io_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVASSOCEVNT)) events->sctp_association_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVPADDREVNT)) events->sctp_address_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVSENDFAILEVNT)) events->sctp_send_failure_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVPEERERR)) events->sctp_peer_error_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVSHUTDOWNEVNT)) events->sctp_shutdown_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_PDAPIEVNT)) events->sctp_partial_delivery_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_ADAPTATIONEVNT)) events->sctp_adaptation_layer_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_AUTHEVNT)) events->sctp_authentication_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_DRYEVNT)) events->sctp_sender_dry_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_STREAM_RESETEVNT)) events->sctp_stream_reset_event = 1; SCTP_INP_RUNLOCK(inp); *optsize = sizeof(struct sctp_event_subscribe); } break; case SCTP_ADAPTATION_LAYER: { uint32_t *value; SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); SCTP_INP_RLOCK(inp); *value = inp->sctp_ep.adaptation_layer_indicator; SCTP_INP_RUNLOCK(inp); *optsize = sizeof(uint32_t); } break; case SCTP_SET_INITIAL_DBG_SEQ: { uint32_t *value; SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); SCTP_INP_RLOCK(inp); *value = inp->sctp_ep.initial_sequence_debug; SCTP_INP_RUNLOCK(inp); *optsize = sizeof(uint32_t); } break; case SCTP_GET_LOCAL_ADDR_SIZE: { uint32_t *value; SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); SCTP_INP_RLOCK(inp); *value = sctp_count_max_addresses(inp); SCTP_INP_RUNLOCK(inp); *optsize = sizeof(uint32_t); } break; case SCTP_GET_REMOTE_ADDR_SIZE: { uint32_t *value; size_t size; struct sctp_nets *net; SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); /* FIXME MT: change to sctp_assoc_value? */ SCTP_FIND_STCB(inp, stcb, (sctp_assoc_t) * value); if (stcb) { size = 0; /* Count the sizes */ TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4)) { size += sizeof(struct sockaddr_in6); } else { switch (((struct sockaddr *)&net->ro._l_addr)->sa_family) { #ifdef INET case AF_INET: size += sizeof(struct sockaddr_in); break; #endif #ifdef INET6 case AF_INET6: size += sizeof(struct sockaddr_in6); break; #endif default: break; } } } SCTP_TCB_UNLOCK(stcb); *value = (uint32_t) size; } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOTCONN); error = ENOTCONN; } *optsize = sizeof(uint32_t); } break; case SCTP_GET_PEER_ADDRESSES: /* * Get the address information, an array is passed in to * fill up we pack it. */ { size_t cpsz, left; struct sockaddr_storage *sas; struct sctp_nets *net; struct sctp_getaddresses *saddr; SCTP_CHECK_AND_CAST(saddr, optval, struct sctp_getaddresses, *optsize); SCTP_FIND_STCB(inp, stcb, saddr->sget_assoc_id); if (stcb) { left = (*optsize) - sizeof(struct sctp_getaddresses); *optsize = sizeof(struct sctp_getaddresses); sas = (struct sockaddr_storage *)&saddr->addr[0]; TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4)) { cpsz = sizeof(struct sockaddr_in6); } else { switch (((struct sockaddr *)&net->ro._l_addr)->sa_family) { #ifdef INET case AF_INET: cpsz = sizeof(struct sockaddr_in); break; #endif #ifdef INET6 case AF_INET6: cpsz = sizeof(struct sockaddr_in6); break; #endif default: cpsz = 0; break; } } if (cpsz == 0) { break; } if (left < cpsz) { /* not enough room. */ break; } #if defined(INET) && defined(INET6) if ((sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4)) && (((struct sockaddr *)&net->ro._l_addr)->sa_family == AF_INET)) { /* Must map the address */ in6_sin_2_v4mapsin6((struct sockaddr_in *)&net->ro._l_addr, (struct sockaddr_in6 *)sas); } else { #endif memcpy(sas, &net->ro._l_addr, cpsz); #if defined(INET) && defined(INET6) } #endif ((struct sockaddr_in *)sas)->sin_port = stcb->rport; sas = (struct sockaddr_storage *)((caddr_t)sas + cpsz); left -= cpsz; *optsize += cpsz; } SCTP_TCB_UNLOCK(stcb); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); error = ENOENT; } } break; case SCTP_GET_LOCAL_ADDRESSES: { size_t limit, actual; struct sockaddr_storage *sas; struct sctp_getaddresses *saddr; SCTP_CHECK_AND_CAST(saddr, optval, struct sctp_getaddresses, *optsize); SCTP_FIND_STCB(inp, stcb, saddr->sget_assoc_id); sas = (struct sockaddr_storage *)&saddr->addr[0]; limit = *optsize - sizeof(sctp_assoc_t); actual = sctp_fill_up_addresses(inp, stcb, limit, sas); if (stcb) { SCTP_TCB_UNLOCK(stcb); } *optsize = sizeof(struct sockaddr_storage) + actual; } break; case SCTP_PEER_ADDR_PARAMS: { struct sctp_paddrparams *paddrp; struct sctp_nets *net; SCTP_CHECK_AND_CAST(paddrp, optval, struct sctp_paddrparams, *optsize); SCTP_FIND_STCB(inp, stcb, paddrp->spp_assoc_id); net = NULL; if (stcb) { net = sctp_findnet(stcb, (struct sockaddr *)&paddrp->spp_address); } else { /* * We increment here since * sctp_findassociation_ep_addr() wil do a * decrement if it finds the stcb as long as * the locked tcb (last argument) is NOT a * TCB.. aka NULL. */ SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, (struct sockaddr *)&paddrp->spp_address, &net, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } } if (stcb && (net == NULL)) { struct sockaddr *sa; sa = (struct sockaddr *)&paddrp->spp_address; #ifdef INET if (sa->sa_family == AF_INET) { struct sockaddr_in *sin; sin = (struct sockaddr_in *)sa; if (sin->sin_addr.s_addr) { error = EINVAL; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif #ifdef INET6 if (sa->sa_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)sa; if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { error = EINVAL; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif { error = EAFNOSUPPORT; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } if (stcb) { /* Applys to the specific association */ paddrp->spp_flags = 0; if (net) { int ovh; if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { ovh = SCTP_MED_OVERHEAD; } else { ovh = SCTP_MED_V4_OVERHEAD; } paddrp->spp_pathmaxrxt = net->failure_threshold; paddrp->spp_pathmtu = net->mtu - ovh; /* get flags for HB */ if (net->dest_state & SCTP_ADDR_NOHB) paddrp->spp_flags |= SPP_HB_DISABLE; else paddrp->spp_flags |= SPP_HB_ENABLE; /* get flags for PMTU */ if (SCTP_OS_TIMER_PENDING(&net->pmtu_timer.timer)) { paddrp->spp_flags |= SPP_PMTUD_ENABLE; } else { paddrp->spp_flags |= SPP_PMTUD_DISABLE; } #ifdef INET if (net->ro._l_addr.sin.sin_family == AF_INET) { paddrp->spp_ipv4_tos = net->tos_flowlabel & 0x000000fc; paddrp->spp_flags |= SPP_IPV4_TOS; } #endif #ifdef INET6 if (net->ro._l_addr.sin6.sin6_family == AF_INET6) { paddrp->spp_ipv6_flowlabel = net->tos_flowlabel; paddrp->spp_flags |= SPP_IPV6_FLOWLABEL; } #endif } else { /* * No destination so return default * value */ int cnt = 0; paddrp->spp_pathmaxrxt = stcb->asoc.def_net_failure; paddrp->spp_pathmtu = sctp_get_frag_point(stcb, &stcb->asoc); #ifdef INET paddrp->spp_ipv4_tos = stcb->asoc.default_tos & 0x000000fc; paddrp->spp_flags |= SPP_IPV4_TOS; #endif #ifdef INET6 paddrp->spp_ipv6_flowlabel = stcb->asoc.default_flowlabel; paddrp->spp_flags |= SPP_IPV6_FLOWLABEL; #endif /* default settings should be these */ if (stcb->asoc.hb_is_disabled == 0) { paddrp->spp_flags |= SPP_HB_ENABLE; } else { paddrp->spp_flags |= SPP_HB_DISABLE; } TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { if (SCTP_OS_TIMER_PENDING(&net->pmtu_timer.timer)) { cnt++; } } if (cnt) { paddrp->spp_flags |= SPP_PMTUD_ENABLE; } } paddrp->spp_hbinterval = stcb->asoc.heart_beat_delay; paddrp->spp_assoc_id = sctp_get_associd(stcb); SCTP_TCB_UNLOCK(stcb); } else { /* Use endpoint defaults */ SCTP_INP_RLOCK(inp); paddrp->spp_pathmaxrxt = inp->sctp_ep.def_net_failure; paddrp->spp_hbinterval = TICKS_TO_MSEC(inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_HEARTBEAT]); paddrp->spp_assoc_id = (sctp_assoc_t) 0; /* get inp's default */ #ifdef INET paddrp->spp_ipv4_tos = inp->ip_inp.inp.inp_ip_tos; paddrp->spp_flags |= SPP_IPV4_TOS; #endif #ifdef INET6 if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { paddrp->spp_ipv6_flowlabel = ((struct in6pcb *)inp)->in6p_flowinfo; paddrp->spp_flags |= SPP_IPV6_FLOWLABEL; } #endif /* can't return this */ paddrp->spp_pathmtu = 0; /* default behavior, no stcb */ paddrp->spp_flags = SPP_PMTUD_ENABLE; if (sctp_is_feature_off(inp, SCTP_PCB_FLAGS_DONOT_HEARTBEAT)) { paddrp->spp_flags |= SPP_HB_ENABLE; } else { paddrp->spp_flags |= SPP_HB_DISABLE; } SCTP_INP_RUNLOCK(inp); } *optsize = sizeof(struct sctp_paddrparams); } break; case SCTP_GET_PEER_ADDR_INFO: { struct sctp_paddrinfo *paddri; struct sctp_nets *net; SCTP_CHECK_AND_CAST(paddri, optval, struct sctp_paddrinfo, *optsize); SCTP_FIND_STCB(inp, stcb, paddri->spinfo_assoc_id); net = NULL; if (stcb) { net = sctp_findnet(stcb, (struct sockaddr *)&paddri->spinfo_address); } else { /* * We increment here since * sctp_findassociation_ep_addr() wil do a * decrement if it finds the stcb as long as * the locked tcb (last argument) is NOT a * TCB.. aka NULL. */ SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, (struct sockaddr *)&paddri->spinfo_address, &net, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } } if ((stcb) && (net)) { if (net->dest_state & SCTP_ADDR_UNCONFIRMED) { /* It's unconfirmed */ paddri->spinfo_state = SCTP_UNCONFIRMED; } else if (net->dest_state & SCTP_ADDR_REACHABLE) { /* It's active */ paddri->spinfo_state = SCTP_ACTIVE; } else { /* It's inactive */ paddri->spinfo_state = SCTP_INACTIVE; } paddri->spinfo_cwnd = net->cwnd; paddri->spinfo_srtt = net->lastsa >> SCTP_RTT_SHIFT; paddri->spinfo_rto = net->RTO; paddri->spinfo_assoc_id = sctp_get_associd(stcb); SCTP_TCB_UNLOCK(stcb); } else { if (stcb) { SCTP_TCB_UNLOCK(stcb); } SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); error = ENOENT; } *optsize = sizeof(struct sctp_paddrinfo); } break; case SCTP_PCB_STATUS: { struct sctp_pcbinfo *spcb; SCTP_CHECK_AND_CAST(spcb, optval, struct sctp_pcbinfo, *optsize); sctp_fill_pcbinfo(spcb); *optsize = sizeof(struct sctp_pcbinfo); } break; case SCTP_STATUS: { struct sctp_nets *net; struct sctp_status *sstat; SCTP_CHECK_AND_CAST(sstat, optval, struct sctp_status, *optsize); SCTP_FIND_STCB(inp, stcb, sstat->sstat_assoc_id); if (stcb == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); error = EINVAL; break; } /* * I think passing the state is fine since * sctp_constants.h will be available to the user * land. */ sstat->sstat_state = stcb->asoc.state; sstat->sstat_assoc_id = sctp_get_associd(stcb); sstat->sstat_rwnd = stcb->asoc.peers_rwnd; sstat->sstat_unackdata = stcb->asoc.sent_queue_cnt; /* * We can't include chunks that have been passed to * the socket layer. Only things in queue. */ sstat->sstat_penddata = (stcb->asoc.cnt_on_reasm_queue + stcb->asoc.cnt_on_all_streams); sstat->sstat_instrms = stcb->asoc.streamincnt; sstat->sstat_outstrms = stcb->asoc.streamoutcnt; sstat->sstat_fragmentation_point = sctp_get_frag_point(stcb, &stcb->asoc); memcpy(&sstat->sstat_primary.spinfo_address, &stcb->asoc.primary_destination->ro._l_addr, ((struct sockaddr *)(&stcb->asoc.primary_destination->ro._l_addr))->sa_len); net = stcb->asoc.primary_destination; ((struct sockaddr_in *)&sstat->sstat_primary.spinfo_address)->sin_port = stcb->rport; /* * Again the user can get info from sctp_constants.h * for what the state of the network is. */ if (net->dest_state & SCTP_ADDR_UNCONFIRMED) { /* It's unconfirmed */ sstat->sstat_primary.spinfo_state = SCTP_UNCONFIRMED; } else if (net->dest_state & SCTP_ADDR_REACHABLE) { /* It's active */ sstat->sstat_primary.spinfo_state = SCTP_ACTIVE; } else { /* It's inactive */ sstat->sstat_primary.spinfo_state = SCTP_INACTIVE; } sstat->sstat_primary.spinfo_cwnd = net->cwnd; sstat->sstat_primary.spinfo_srtt = net->lastsa >> SCTP_RTT_SHIFT; sstat->sstat_primary.spinfo_rto = net->RTO; sstat->sstat_primary.spinfo_mtu = net->mtu; sstat->sstat_primary.spinfo_assoc_id = sctp_get_associd(stcb); SCTP_TCB_UNLOCK(stcb); *optsize = sizeof(*sstat); } break; case SCTP_RTOINFO: { struct sctp_rtoinfo *srto; SCTP_CHECK_AND_CAST(srto, optval, struct sctp_rtoinfo, *optsize); SCTP_FIND_STCB(inp, stcb, srto->srto_assoc_id); if (stcb) { srto->srto_initial = stcb->asoc.initial_rto; srto->srto_max = stcb->asoc.maxrto; srto->srto_min = stcb->asoc.minrto; SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_RLOCK(inp); srto->srto_initial = inp->sctp_ep.initial_rto; srto->srto_max = inp->sctp_ep.sctp_maxrto; srto->srto_min = inp->sctp_ep.sctp_minrto; SCTP_INP_RUNLOCK(inp); } *optsize = sizeof(*srto); } break; case SCTP_TIMEOUTS: { struct sctp_timeouts *stimo; SCTP_CHECK_AND_CAST(stimo, optval, struct sctp_timeouts, *optsize); SCTP_FIND_STCB(inp, stcb, stimo->stimo_assoc_id); if (stcb) { stimo->stimo_init = stcb->asoc.timoinit; stimo->stimo_data = stcb->asoc.timodata; stimo->stimo_sack = stcb->asoc.timosack; stimo->stimo_shutdown = stcb->asoc.timoshutdown; stimo->stimo_heartbeat = stcb->asoc.timoheartbeat; stimo->stimo_cookie = stcb->asoc.timocookie; stimo->stimo_shutdownack = stcb->asoc.timoshutdownack; SCTP_TCB_UNLOCK(stcb); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); error = EINVAL; } *optsize = sizeof(*stimo); } break; case SCTP_ASSOCINFO: { struct sctp_assocparams *sasoc; uint32_t oldval; SCTP_CHECK_AND_CAST(sasoc, optval, struct sctp_assocparams, *optsize); SCTP_FIND_STCB(inp, stcb, sasoc->sasoc_assoc_id); if (stcb) { oldval = sasoc->sasoc_cookie_life; sasoc->sasoc_cookie_life = TICKS_TO_MSEC(stcb->asoc.cookie_life); sasoc->sasoc_asocmaxrxt = stcb->asoc.max_send_times; sasoc->sasoc_number_peer_destinations = stcb->asoc.numnets; sasoc->sasoc_peer_rwnd = stcb->asoc.peers_rwnd; sasoc->sasoc_local_rwnd = stcb->asoc.my_rwnd; SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_RLOCK(inp); sasoc->sasoc_cookie_life = TICKS_TO_MSEC(inp->sctp_ep.def_cookie_life); sasoc->sasoc_asocmaxrxt = inp->sctp_ep.max_send_times; sasoc->sasoc_number_peer_destinations = 0; sasoc->sasoc_peer_rwnd = 0; sasoc->sasoc_local_rwnd = sbspace(&inp->sctp_socket->so_rcv); SCTP_INP_RUNLOCK(inp); } *optsize = sizeof(*sasoc); } break; case SCTP_DEFAULT_SEND_PARAM: { struct sctp_sndrcvinfo *s_info; SCTP_CHECK_AND_CAST(s_info, optval, struct sctp_sndrcvinfo, *optsize); SCTP_FIND_STCB(inp, stcb, s_info->sinfo_assoc_id); if (stcb) { memcpy(s_info, &stcb->asoc.def_send, sizeof(stcb->asoc.def_send)); SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_RLOCK(inp); memcpy(s_info, &inp->def_send, sizeof(inp->def_send)); SCTP_INP_RUNLOCK(inp); } *optsize = sizeof(*s_info); } break; case SCTP_INITMSG: { struct sctp_initmsg *sinit; SCTP_CHECK_AND_CAST(sinit, optval, struct sctp_initmsg, *optsize); SCTP_INP_RLOCK(inp); sinit->sinit_num_ostreams = inp->sctp_ep.pre_open_stream_count; sinit->sinit_max_instreams = inp->sctp_ep.max_open_streams_intome; sinit->sinit_max_attempts = inp->sctp_ep.max_init_times; sinit->sinit_max_init_timeo = inp->sctp_ep.initial_init_rto_max; SCTP_INP_RUNLOCK(inp); *optsize = sizeof(*sinit); } break; case SCTP_PRIMARY_ADDR: /* we allow a "get" operation on this */ { struct sctp_setprim *ssp; SCTP_CHECK_AND_CAST(ssp, optval, struct sctp_setprim, *optsize); SCTP_FIND_STCB(inp, stcb, ssp->ssp_assoc_id); if (stcb) { /* simply copy out the sockaddr_storage... */ int len; len = *optsize; if (len > stcb->asoc.primary_destination->ro._l_addr.sa.sa_len) len = stcb->asoc.primary_destination->ro._l_addr.sa.sa_len; memcpy(&ssp->ssp_addr, &stcb->asoc.primary_destination->ro._l_addr, len); SCTP_TCB_UNLOCK(stcb); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); error = EINVAL; } *optsize = sizeof(*ssp); } break; case SCTP_HMAC_IDENT: { struct sctp_hmacalgo *shmac; sctp_hmaclist_t *hmaclist; uint32_t size; int i; SCTP_CHECK_AND_CAST(shmac, optval, struct sctp_hmacalgo, *optsize); SCTP_INP_RLOCK(inp); hmaclist = inp->sctp_ep.local_hmacs; if (hmaclist == NULL) { /* no HMACs to return */ *optsize = sizeof(*shmac); SCTP_INP_RUNLOCK(inp); break; } /* is there room for all of the hmac ids? */ size = sizeof(*shmac) + (hmaclist->num_algo * sizeof(shmac->shmac_idents[0])); if ((size_t)(*optsize) < size) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); error = EINVAL; SCTP_INP_RUNLOCK(inp); break; } /* copy in the list */ shmac->shmac_number_of_idents = hmaclist->num_algo; for (i = 0; i < hmaclist->num_algo; i++) { shmac->shmac_idents[i] = hmaclist->hmac[i]; } SCTP_INP_RUNLOCK(inp); *optsize = size; break; } case SCTP_AUTH_ACTIVE_KEY: { struct sctp_authkeyid *scact; SCTP_CHECK_AND_CAST(scact, optval, struct sctp_authkeyid, *optsize); SCTP_FIND_STCB(inp, stcb, scact->scact_assoc_id); if (stcb) { /* get the active key on the assoc */ scact->scact_keynumber = stcb->asoc.authinfo.active_keyid; SCTP_TCB_UNLOCK(stcb); } else { /* get the endpoint active key */ SCTP_INP_RLOCK(inp); scact->scact_keynumber = inp->sctp_ep.default_keyid; SCTP_INP_RUNLOCK(inp); } *optsize = sizeof(*scact); break; } case SCTP_LOCAL_AUTH_CHUNKS: { struct sctp_authchunks *sac; sctp_auth_chklist_t *chklist = NULL; size_t size = 0; SCTP_CHECK_AND_CAST(sac, optval, struct sctp_authchunks, *optsize); SCTP_FIND_STCB(inp, stcb, sac->gauth_assoc_id); if (stcb) { /* get off the assoc */ chklist = stcb->asoc.local_auth_chunks; /* is there enough space? */ size = sctp_auth_get_chklist_size(chklist); if (*optsize < (sizeof(struct sctp_authchunks) + size)) { error = EINVAL; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); } else { /* copy in the chunks */ (void)sctp_serialize_auth_chunks(chklist, sac->gauth_chunks); } SCTP_TCB_UNLOCK(stcb); } else { /* get off the endpoint */ SCTP_INP_RLOCK(inp); chklist = inp->sctp_ep.local_auth_chunks; /* is there enough space? */ size = sctp_auth_get_chklist_size(chklist); if (*optsize < (sizeof(struct sctp_authchunks) + size)) { error = EINVAL; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); } else { /* copy in the chunks */ (void)sctp_serialize_auth_chunks(chklist, sac->gauth_chunks); } SCTP_INP_RUNLOCK(inp); } *optsize = sizeof(struct sctp_authchunks) + size; break; } case SCTP_PEER_AUTH_CHUNKS: { struct sctp_authchunks *sac; sctp_auth_chklist_t *chklist = NULL; size_t size = 0; SCTP_CHECK_AND_CAST(sac, optval, struct sctp_authchunks, *optsize); SCTP_FIND_STCB(inp, stcb, sac->gauth_assoc_id); if (stcb) { /* get off the assoc */ chklist = stcb->asoc.peer_auth_chunks; /* is there enough space? */ size = sctp_auth_get_chklist_size(chklist); if (*optsize < (sizeof(struct sctp_authchunks) + size)) { error = EINVAL; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); } else { /* copy in the chunks */ (void)sctp_serialize_auth_chunks(chklist, sac->gauth_chunks); } SCTP_TCB_UNLOCK(stcb); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); error = ENOENT; } *optsize = sizeof(struct sctp_authchunks) + size; break; } default: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOPROTOOPT); error = ENOPROTOOPT; *optsize = 0; break; } /* end switch (sopt->sopt_name) */ return (error); } static int sctp_setopt(struct socket *so, int optname, void *optval, size_t optsize, void *p) { int error, set_opt; uint32_t *mopt; struct sctp_tcb *stcb = NULL; struct sctp_inpcb *inp = NULL; uint32_t vrf_id; if (optval == NULL) { SCTP_PRINTF("optval is NULL\n"); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } inp = (struct sctp_inpcb *)so->so_pcb; if (inp == 0) { SCTP_PRINTF("inp is NULL?\n"); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return EINVAL; } vrf_id = inp->def_vrf_id; error = 0; switch (optname) { case SCTP_NODELAY: case SCTP_AUTOCLOSE: case SCTP_AUTO_ASCONF: case SCTP_EXPLICIT_EOR: case SCTP_DISABLE_FRAGMENTS: case SCTP_USE_EXT_RCVINFO: case SCTP_I_WANT_MAPPED_V4_ADDR: /* copy in the option value */ SCTP_CHECK_AND_CAST(mopt, optval, uint32_t, optsize); set_opt = 0; if (error) break; switch (optname) { case SCTP_DISABLE_FRAGMENTS: set_opt = SCTP_PCB_FLAGS_NO_FRAGMENT; break; case SCTP_AUTO_ASCONF: /* * NOTE: we don't really support this flag */ if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) { /* only valid for bound all sockets */ set_opt = SCTP_PCB_FLAGS_AUTO_ASCONF; } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } break; case SCTP_EXPLICIT_EOR: set_opt = SCTP_PCB_FLAGS_EXPLICIT_EOR; break; case SCTP_USE_EXT_RCVINFO: set_opt = SCTP_PCB_FLAGS_EXT_RCVINFO; break; case SCTP_I_WANT_MAPPED_V4_ADDR: if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { set_opt = SCTP_PCB_FLAGS_NEEDS_MAPPED_V4; } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } break; case SCTP_NODELAY: set_opt = SCTP_PCB_FLAGS_NODELAY; break; case SCTP_AUTOCLOSE: if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } set_opt = SCTP_PCB_FLAGS_AUTOCLOSE; /* * The value is in ticks. Note this does not effect * old associations, only new ones. */ inp->sctp_ep.auto_close_time = SEC_TO_TICKS(*mopt); break; } SCTP_INP_WLOCK(inp); if (*mopt != 0) { sctp_feature_on(inp, set_opt); } else { sctp_feature_off(inp, set_opt); } SCTP_INP_WUNLOCK(inp); break; case SCTP_REUSE_PORT: { SCTP_CHECK_AND_CAST(mopt, optval, uint32_t, optsize); if ((inp->sctp_flags & SCTP_PCB_FLAGS_UNBOUND) == 0) { /* Can't set it after we are bound */ error = EINVAL; break; } if ((inp->sctp_flags & SCTP_PCB_FLAGS_UDPTYPE)) { /* Can't do this for a 1-m socket */ error = EINVAL; break; } if (optval) sctp_feature_on(inp, SCTP_PCB_FLAGS_PORTREUSE); else sctp_feature_off(inp, SCTP_PCB_FLAGS_PORTREUSE); } break; case SCTP_PARTIAL_DELIVERY_POINT: { uint32_t *value; SCTP_CHECK_AND_CAST(value, optval, uint32_t, optsize); if (*value > SCTP_SB_LIMIT_RCV(so)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } inp->partial_delivery_point = *value; } break; case SCTP_FRAGMENT_INTERLEAVE: /* not yet until we re-write sctp_recvmsg() */ { uint32_t *level; SCTP_CHECK_AND_CAST(level, optval, uint32_t, optsize); if (*level == SCTP_FRAG_LEVEL_2) { sctp_feature_on(inp, SCTP_PCB_FLAGS_FRAG_INTERLEAVE); sctp_feature_on(inp, SCTP_PCB_FLAGS_INTERLEAVE_STRMS); } else if (*level == SCTP_FRAG_LEVEL_1) { sctp_feature_on(inp, SCTP_PCB_FLAGS_FRAG_INTERLEAVE); sctp_feature_off(inp, SCTP_PCB_FLAGS_INTERLEAVE_STRMS); } else if (*level == SCTP_FRAG_LEVEL_0) { sctp_feature_off(inp, SCTP_PCB_FLAGS_FRAG_INTERLEAVE); sctp_feature_off(inp, SCTP_PCB_FLAGS_INTERLEAVE_STRMS); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; case SCTP_CMT_ON_OFF: if (SCTP_BASE_SYSCTL(sctp_cmt_on_off)) { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { - stcb->asoc.sctp_cmt_on_off = av->assoc_value; - if (stcb->asoc.sctp_cmt_on_off > 2) { - stcb->asoc.sctp_cmt_on_off = 2; + if (av->assoc_value > SCTP_CMT_MAX) { + SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); + error = EINVAL; + } else { + stcb->asoc.sctp_cmt_on_off = av->assoc_value; } SCTP_TCB_UNLOCK(stcb); } else { - SCTP_INP_WLOCK(inp); - inp->sctp_cmt_on_off = av->assoc_value; - if (inp->sctp_cmt_on_off > 2) { - inp->sctp_cmt_on_off = 2; + if (av->assoc_value > SCTP_CMT_MAX) { + SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); + error = EINVAL; + } else { + SCTP_INP_WLOCK(inp); + inp->sctp_cmt_on_off = av->assoc_value; + SCTP_INP_WUNLOCK(inp); } - SCTP_INP_WUNLOCK(inp); } } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOPROTOOPT); error = ENOPROTOOPT; } break; /* JRS - Set socket option for pluggable congestion control */ case SCTP_PLUGGABLE_CC: { struct sctp_assoc_value *av; struct sctp_nets *net; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { switch (av->assoc_value) { case SCTP_CC_RFC2581: case SCTP_CC_HSTCP: case SCTP_CC_HTCP: case SCTP_CC_RTCC: stcb->asoc.cc_functions = sctp_cc_functions[av->assoc_value]; stcb->asoc.congestion_control_module = av->assoc_value; if (stcb->asoc.cc_functions.sctp_set_initial_cc_param != NULL) { TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { stcb->asoc.cc_functions.sctp_set_initial_cc_param(stcb, net); } } break; default: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } SCTP_TCB_UNLOCK(stcb); } else { switch (av->assoc_value) { case SCTP_CC_RFC2581: case SCTP_CC_HSTCP: case SCTP_CC_HTCP: case SCTP_CC_RTCC: SCTP_INP_WLOCK(inp); inp->sctp_ep.sctp_default_cc_module = av->assoc_value; SCTP_INP_WUNLOCK(inp); break; default: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } } } break; case SCTP_CC_OPTION: { struct sctp_cc_option *cc_opt; SCTP_CHECK_AND_CAST(cc_opt, optval, struct sctp_cc_option, optsize); SCTP_FIND_STCB(inp, stcb, cc_opt->aid_value.assoc_id); if (stcb == NULL) { error = EINVAL; } else { if (stcb->asoc.cc_functions.sctp_cwnd_socket_option == NULL) { error = ENOTSUP; } else { error = (*stcb->asoc.cc_functions.sctp_cwnd_socket_option) (stcb, 1, cc_opt); } SCTP_TCB_UNLOCK(stcb); } } break; /* RS - Set socket option for pluggable stream scheduling */ case SCTP_PLUGGABLE_SS: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { switch (av->assoc_value) { case SCTP_SS_DEFAULT: case SCTP_SS_ROUND_ROBIN: case SCTP_SS_ROUND_ROBIN_PACKET: case SCTP_SS_PRIORITY: case SCTP_SS_FAIR_BANDWITH: case SCTP_SS_FIRST_COME: stcb->asoc.ss_functions.sctp_ss_clear(stcb, &stcb->asoc, 1, 1); stcb->asoc.ss_functions = sctp_ss_functions[av->assoc_value]; stcb->asoc.stream_scheduling_module = av->assoc_value; stcb->asoc.ss_functions.sctp_ss_init(stcb, &stcb->asoc, 1); break; default: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } SCTP_TCB_UNLOCK(stcb); } else { switch (av->assoc_value) { case SCTP_SS_DEFAULT: case SCTP_SS_ROUND_ROBIN: case SCTP_SS_ROUND_ROBIN_PACKET: case SCTP_SS_PRIORITY: case SCTP_SS_FAIR_BANDWITH: case SCTP_SS_FIRST_COME: SCTP_INP_WLOCK(inp); inp->sctp_ep.sctp_default_ss_module = av->assoc_value; SCTP_INP_WUNLOCK(inp); break; default: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } } } break; case SCTP_SS_VALUE: { struct sctp_stream_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_stream_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { if (stcb->asoc.ss_functions.sctp_ss_set_value(stcb, &stcb->asoc, &stcb->asoc.strmout[av->stream_id], av->stream_value) < 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_TCB_UNLOCK(stcb); } else { /* * Can't set stream value without * association */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; case SCTP_CLR_STAT_LOG: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); error = EOPNOTSUPP; break; case SCTP_CONTEXT: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { stcb->asoc.context = av->assoc_value; SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_WLOCK(inp); inp->sctp_context = av->assoc_value; SCTP_INP_WUNLOCK(inp); } } break; case SCTP_VRF_ID: { uint32_t *default_vrfid; SCTP_CHECK_AND_CAST(default_vrfid, optval, uint32_t, optsize); if (*default_vrfid > SCTP_MAX_VRF_ID) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } inp->def_vrf_id = *default_vrfid; break; } case SCTP_DEL_VRF_ID: { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); error = EOPNOTSUPP; break; } case SCTP_ADD_VRF_ID: { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); error = EOPNOTSUPP; break; } case SCTP_DELAYED_SACK: { struct sctp_sack_info *sack; SCTP_CHECK_AND_CAST(sack, optval, struct sctp_sack_info, optsize); SCTP_FIND_STCB(inp, stcb, sack->sack_assoc_id); if (sack->sack_delay) { if (sack->sack_delay > SCTP_MAX_SACK_DELAY) sack->sack_delay = SCTP_MAX_SACK_DELAY; } if (stcb) { if (sack->sack_delay) { if (MSEC_TO_TICKS(sack->sack_delay) < 1) { sack->sack_delay = TICKS_TO_MSEC(1); } stcb->asoc.delayed_ack = sack->sack_delay; } if (sack->sack_freq) { stcb->asoc.sack_freq = sack->sack_freq; } SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_WLOCK(inp); if (sack->sack_delay) { if (MSEC_TO_TICKS(sack->sack_delay) < 1) { sack->sack_delay = TICKS_TO_MSEC(1); } inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_RECV] = MSEC_TO_TICKS(sack->sack_delay); } if (sack->sack_freq) { inp->sctp_ep.sctp_sack_freq = sack->sack_freq; } SCTP_INP_WUNLOCK(inp); } break; } case SCTP_AUTH_CHUNK: { struct sctp_authchunk *sauth; SCTP_CHECK_AND_CAST(sauth, optval, struct sctp_authchunk, optsize); SCTP_INP_WLOCK(inp); if (sctp_auth_add_chunk(sauth->sauth_chunk, inp->sctp_ep.local_auth_chunks)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_INP_WUNLOCK(inp); break; } case SCTP_AUTH_KEY: { struct sctp_authkey *sca; struct sctp_keyhead *shared_keys; sctp_sharedkey_t *shared_key; sctp_key_t *key = NULL; size_t size; SCTP_CHECK_AND_CAST(sca, optval, struct sctp_authkey, optsize); SCTP_FIND_STCB(inp, stcb, sca->sca_assoc_id); size = optsize - sizeof(*sca); if (stcb) { /* set it on the assoc */ shared_keys = &stcb->asoc.shared_keys; /* clear the cached keys for this key id */ sctp_clear_cachedkeys(stcb, sca->sca_keynumber); /* * create the new shared key and * insert/replace it */ if (size > 0) { key = sctp_set_key(sca->sca_key, (uint32_t) size); if (key == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOMEM); error = ENOMEM; SCTP_TCB_UNLOCK(stcb); break; } } shared_key = sctp_alloc_sharedkey(); if (shared_key == NULL) { sctp_free_key(key); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOMEM); error = ENOMEM; SCTP_TCB_UNLOCK(stcb); break; } shared_key->key = key; shared_key->keyid = sca->sca_keynumber; error = sctp_insert_sharedkey(shared_keys, shared_key); SCTP_TCB_UNLOCK(stcb); } else { /* set it on the endpoint */ SCTP_INP_WLOCK(inp); shared_keys = &inp->sctp_ep.shared_keys; /* * clear the cached keys on all assocs for * this key id */ sctp_clear_cachedkeys_ep(inp, sca->sca_keynumber); /* * create the new shared key and * insert/replace it */ if (size > 0) { key = sctp_set_key(sca->sca_key, (uint32_t) size); if (key == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOMEM); error = ENOMEM; SCTP_INP_WUNLOCK(inp); break; } } shared_key = sctp_alloc_sharedkey(); if (shared_key == NULL) { sctp_free_key(key); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOMEM); error = ENOMEM; SCTP_INP_WUNLOCK(inp); break; } shared_key->key = key; shared_key->keyid = sca->sca_keynumber; error = sctp_insert_sharedkey(shared_keys, shared_key); SCTP_INP_WUNLOCK(inp); } break; } case SCTP_HMAC_IDENT: { struct sctp_hmacalgo *shmac; sctp_hmaclist_t *hmaclist; uint16_t hmacid; uint32_t i; size_t found; SCTP_CHECK_AND_CAST(shmac, optval, struct sctp_hmacalgo, optsize); if (optsize < sizeof(struct sctp_hmacalgo) + shmac->shmac_number_of_idents * sizeof(uint16_t)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } hmaclist = sctp_alloc_hmaclist(shmac->shmac_number_of_idents); if (hmaclist == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOMEM); error = ENOMEM; break; } for (i = 0; i < shmac->shmac_number_of_idents; i++) { hmacid = shmac->shmac_idents[i]; if (sctp_auth_add_hmacid(hmaclist, hmacid)) { /* invalid HMACs were found */ ; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; sctp_free_hmaclist(hmaclist); goto sctp_set_hmac_done; } } found = 0; for (i = 0; i < hmaclist->num_algo; i++) { if (hmaclist->hmac[i] == SCTP_AUTH_HMAC_ID_SHA1) { /* already in list */ found = 1; } } if (!found) { sctp_free_hmaclist(hmaclist); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } /* set it on the endpoint */ SCTP_INP_WLOCK(inp); if (inp->sctp_ep.local_hmacs) sctp_free_hmaclist(inp->sctp_ep.local_hmacs); inp->sctp_ep.local_hmacs = hmaclist; SCTP_INP_WUNLOCK(inp); sctp_set_hmac_done: break; } case SCTP_AUTH_ACTIVE_KEY: { struct sctp_authkeyid *scact; SCTP_CHECK_AND_CAST(scact, optval, struct sctp_authkeyid, optsize); SCTP_FIND_STCB(inp, stcb, scact->scact_assoc_id); /* set the active key on the right place */ if (stcb) { /* set the active key on the assoc */ if (sctp_auth_setactivekey(stcb, scact->scact_keynumber)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_TCB_UNLOCK(stcb); } else { /* set the active key on the endpoint */ SCTP_INP_WLOCK(inp); if (sctp_auth_setactivekey_ep(inp, scact->scact_keynumber)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_INP_WUNLOCK(inp); } break; } case SCTP_AUTH_DELETE_KEY: { struct sctp_authkeyid *scdel; SCTP_CHECK_AND_CAST(scdel, optval, struct sctp_authkeyid, optsize); SCTP_FIND_STCB(inp, stcb, scdel->scact_assoc_id); /* delete the key from the right place */ if (stcb) { if (sctp_delete_sharedkey(stcb, scdel->scact_keynumber)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_WLOCK(inp); if (sctp_delete_sharedkey_ep(inp, scdel->scact_keynumber)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_INP_WUNLOCK(inp); } break; } case SCTP_AUTH_DEACTIVATE_KEY: { struct sctp_authkeyid *keyid; SCTP_CHECK_AND_CAST(keyid, optval, struct sctp_authkeyid, optsize); SCTP_FIND_STCB(inp, stcb, keyid->scact_assoc_id); /* deactivate the key from the right place */ if (stcb) { if (sctp_deact_sharedkey(stcb, keyid->scact_keynumber)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_WLOCK(inp); if (sctp_deact_sharedkey_ep(inp, keyid->scact_keynumber)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_INP_WUNLOCK(inp); } break; } case SCTP_RESET_STREAMS: { struct sctp_stream_reset *strrst; uint8_t send_in = 0, send_tsn = 0, send_out = 0, addstream = 0; uint16_t addstrmcnt = 0; int i; SCTP_CHECK_AND_CAST(strrst, optval, struct sctp_stream_reset, optsize); SCTP_FIND_STCB(inp, stcb, strrst->strrst_assoc_id); if (stcb == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); error = ENOENT; break; } if (stcb->asoc.peer_supports_strreset == 0) { /* * Peer does not support it, we return * protocol not supported since this is true * for this feature and this peer, not the * socket request in general. */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EPROTONOSUPPORT); error = EPROTONOSUPPORT; SCTP_TCB_UNLOCK(stcb); break; } if (stcb->asoc.stream_reset_outstanding) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EALREADY); error = EALREADY; SCTP_TCB_UNLOCK(stcb); break; } if (strrst->strrst_flags == SCTP_RESET_LOCAL_RECV) { send_in = 1; } else if (strrst->strrst_flags == SCTP_RESET_LOCAL_SEND) { send_out = 1; } else if (strrst->strrst_flags == SCTP_RESET_BOTH) { send_in = 1; send_out = 1; } else if (strrst->strrst_flags == SCTP_RESET_TSN) { send_tsn = 1; } else if (strrst->strrst_flags == SCTP_RESET_ADD_STREAMS) { if (send_tsn || send_in || send_out) { /* We can't do that and add streams */ error = EINVAL; goto skip_stuff; } if (stcb->asoc.stream_reset_outstanding) { error = EBUSY; goto skip_stuff; } addstream = 1; /* We allocate here */ addstrmcnt = strrst->strrst_num_streams; if ((int)(addstrmcnt + stcb->asoc.streamoutcnt) > 0xffff) { /* You can't have more than 64k */ error = EINVAL; goto skip_stuff; } if ((stcb->asoc.strm_realoutsize - stcb->asoc.streamoutcnt) < addstrmcnt) { /* Need to allocate more */ struct sctp_stream_out *oldstream; struct sctp_stream_queue_pending *sp, *nsp; oldstream = stcb->asoc.strmout; /* get some more */ SCTP_MALLOC(stcb->asoc.strmout, struct sctp_stream_out *, ((stcb->asoc.streamoutcnt + addstrmcnt) * sizeof(struct sctp_stream_out)), SCTP_M_STRMO); if (stcb->asoc.strmout == NULL) { stcb->asoc.strmout = oldstream; error = ENOMEM; goto skip_stuff; } /* * Ok now we proceed with copying * the old out stuff and * initializing the new stuff. */ SCTP_TCB_SEND_LOCK(stcb); stcb->asoc.ss_functions.sctp_ss_clear(stcb, &stcb->asoc, 0, 1); for (i = 0; i < stcb->asoc.streamoutcnt; i++) { TAILQ_INIT(&stcb->asoc.strmout[i].outqueue); stcb->asoc.strmout[i].next_sequence_sent = oldstream[i].next_sequence_sent; stcb->asoc.strmout[i].last_msg_incomplete = oldstream[i].last_msg_incomplete; stcb->asoc.strmout[i].stream_no = i; stcb->asoc.ss_functions.sctp_ss_init_stream(&stcb->asoc.strmout[i], &oldstream[i]); /* * now anything on those * queues? */ TAILQ_FOREACH_SAFE(sp, &oldstream[i].outqueue, next, nsp) { TAILQ_REMOVE(&oldstream[i].outqueue, sp, next); TAILQ_INSERT_TAIL(&stcb->asoc.strmout[i].outqueue, sp, next); } /* * Now move assoc pointers * too */ if (stcb->asoc.last_out_stream == &oldstream[i]) { stcb->asoc.last_out_stream = &stcb->asoc.strmout[i]; } if (stcb->asoc.locked_on_sending == &oldstream[i]) { stcb->asoc.locked_on_sending = &stcb->asoc.strmout[i]; } } /* now the new streams */ stcb->asoc.ss_functions.sctp_ss_init(stcb, &stcb->asoc, 1); for (i = stcb->asoc.streamoutcnt; i < (stcb->asoc.streamoutcnt + addstrmcnt); i++) { stcb->asoc.strmout[i].next_sequence_sent = 0x0; TAILQ_INIT(&stcb->asoc.strmout[i].outqueue); stcb->asoc.strmout[i].stream_no = i; stcb->asoc.strmout[i].last_msg_incomplete = 0; stcb->asoc.ss_functions.sctp_ss_init_stream(&stcb->asoc.strmout[i], NULL); } stcb->asoc.strm_realoutsize = stcb->asoc.streamoutcnt + addstrmcnt; SCTP_FREE(oldstream, SCTP_M_STRMO); } SCTP_TCB_SEND_UNLOCK(stcb); goto skip_stuff; } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; SCTP_TCB_UNLOCK(stcb); break; } for (i = 0; i < strrst->strrst_num_streams; i++) { if ((send_in) && (strrst->strrst_list[i] > stcb->asoc.streamincnt)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto get_out; } if ((send_out) && (strrst->strrst_list[i] > stcb->asoc.streamoutcnt)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto get_out; } } skip_stuff: if (error) { get_out: SCTP_TCB_UNLOCK(stcb); break; } error = sctp_send_str_reset_req(stcb, strrst->strrst_num_streams, strrst->strrst_list, send_out, (stcb->asoc.str_reset_seq_in - 3), send_in, send_tsn, addstream, addstrmcnt); sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_STRRST_REQ, SCTP_SO_LOCKED); SCTP_TCB_UNLOCK(stcb); } break; case SCTP_CONNECT_X: if (optsize < (sizeof(int) + sizeof(struct sockaddr_in))) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } error = sctp_do_connect_x(so, inp, optval, optsize, p, 0); break; case SCTP_CONNECT_X_DELAYED: if (optsize < (sizeof(int) + sizeof(struct sockaddr_in))) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } error = sctp_do_connect_x(so, inp, optval, optsize, p, 1); break; case SCTP_CONNECT_X_COMPLETE: { struct sockaddr *sa; struct sctp_nets *net; /* FIXME MT: check correct? */ SCTP_CHECK_AND_CAST(sa, optval, struct sockaddr, optsize); /* find tcb */ if (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) { SCTP_INP_RLOCK(inp); stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb) { SCTP_TCB_LOCK(stcb); net = sctp_findnet(stcb, sa); } SCTP_INP_RUNLOCK(inp); } else { /* * We increment here since * sctp_findassociation_ep_addr() wil do a * decrement if it finds the stcb as long as * the locked tcb (last argument) is NOT a * TCB.. aka NULL. */ SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, sa, &net, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } } if (stcb == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); error = ENOENT; break; } if (stcb->asoc.delayed_connection == 1) { stcb->asoc.delayed_connection = 0; (void)SCTP_GETTIME_TIMEVAL(&stcb->asoc.time_entered); sctp_timer_stop(SCTP_TIMER_TYPE_INIT, inp, stcb, stcb->asoc.primary_destination, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_9); sctp_send_initiate(inp, stcb, SCTP_SO_LOCKED); } else { /* * already expired or did not use delayed * connectx */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EALREADY); error = EALREADY; } SCTP_TCB_UNLOCK(stcb); } break; case SCTP_MAX_BURST: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { stcb->asoc.max_burst = av->assoc_value; SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_WLOCK(inp); inp->sctp_ep.max_burst = av->assoc_value; SCTP_INP_WUNLOCK(inp); } } break; case SCTP_MAXSEG: { struct sctp_assoc_value *av; int ovh; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { ovh = SCTP_MED_OVERHEAD; } else { ovh = SCTP_MED_V4_OVERHEAD; } if (stcb) { if (av->assoc_value) { stcb->asoc.sctp_frag_point = (av->assoc_value + ovh); } else { stcb->asoc.sctp_frag_point = SCTP_DEFAULT_MAXSEGMENT; } SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_WLOCK(inp); /* * FIXME MT: I think this is not in tune * with the API ID */ if (av->assoc_value) { inp->sctp_frag_point = (av->assoc_value + ovh); } else { inp->sctp_frag_point = SCTP_DEFAULT_MAXSEGMENT; } SCTP_INP_WUNLOCK(inp); } } break; case SCTP_EVENTS: { struct sctp_event_subscribe *events; SCTP_CHECK_AND_CAST(events, optval, struct sctp_event_subscribe, optsize); SCTP_INP_WLOCK(inp); if (events->sctp_data_io_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_RECVDATAIOEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_RECVDATAIOEVNT); } if (events->sctp_association_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_RECVASSOCEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_RECVASSOCEVNT); } if (events->sctp_address_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_RECVPADDREVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_RECVPADDREVNT); } if (events->sctp_send_failure_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_RECVSENDFAILEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_RECVSENDFAILEVNT); } if (events->sctp_peer_error_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_RECVPEERERR); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_RECVPEERERR); } if (events->sctp_shutdown_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_RECVSHUTDOWNEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_RECVSHUTDOWNEVNT); } if (events->sctp_partial_delivery_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_PDAPIEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_PDAPIEVNT); } if (events->sctp_adaptation_layer_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_ADAPTATIONEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_ADAPTATIONEVNT); } if (events->sctp_authentication_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_AUTHEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_AUTHEVNT); } if (events->sctp_sender_dry_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_DRYEVNT); if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) { stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb) { SCTP_TCB_LOCK(stcb); } if (stcb && TAILQ_EMPTY(&stcb->asoc.send_queue) && TAILQ_EMPTY(&stcb->asoc.sent_queue) && (stcb->asoc.stream_queue_cnt == 0)) { sctp_ulp_notify(SCTP_NOTIFY_SENDER_DRY, stcb, 0, NULL, SCTP_SO_LOCKED); } if (stcb) { SCTP_TCB_UNLOCK(stcb); } } } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_DRYEVNT); } if (events->sctp_stream_reset_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_STREAM_RESETEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_STREAM_RESETEVNT); } SCTP_INP_WUNLOCK(inp); } break; case SCTP_ADAPTATION_LAYER: { struct sctp_setadaptation *adap_bits; SCTP_CHECK_AND_CAST(adap_bits, optval, struct sctp_setadaptation, optsize); SCTP_INP_WLOCK(inp); inp->sctp_ep.adaptation_layer_indicator = adap_bits->ssb_adaptation_ind; SCTP_INP_WUNLOCK(inp); } break; #ifdef SCTP_DEBUG case SCTP_SET_INITIAL_DBG_SEQ: { uint32_t *vvv; SCTP_CHECK_AND_CAST(vvv, optval, uint32_t, optsize); SCTP_INP_WLOCK(inp); inp->sctp_ep.initial_sequence_debug = *vvv; SCTP_INP_WUNLOCK(inp); } break; #endif case SCTP_DEFAULT_SEND_PARAM: { struct sctp_sndrcvinfo *s_info; SCTP_CHECK_AND_CAST(s_info, optval, struct sctp_sndrcvinfo, optsize); SCTP_FIND_STCB(inp, stcb, s_info->sinfo_assoc_id); if (stcb) { if (s_info->sinfo_stream <= stcb->asoc.streamoutcnt) { memcpy(&stcb->asoc.def_send, s_info, min(optsize, sizeof(stcb->asoc.def_send))); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_WLOCK(inp); memcpy(&inp->def_send, s_info, min(optsize, sizeof(inp->def_send))); SCTP_INP_WUNLOCK(inp); } } break; case SCTP_PEER_ADDR_PARAMS: /* Applys to the specific association */ { struct sctp_paddrparams *paddrp; struct sctp_nets *net; SCTP_CHECK_AND_CAST(paddrp, optval, struct sctp_paddrparams, optsize); SCTP_FIND_STCB(inp, stcb, paddrp->spp_assoc_id); net = NULL; if (stcb) { net = sctp_findnet(stcb, (struct sockaddr *)&paddrp->spp_address); } else { /* * We increment here since * sctp_findassociation_ep_addr() wil do a * decrement if it finds the stcb as long as * the locked tcb (last argument) is NOT a * TCB.. aka NULL. */ SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, (struct sockaddr *)&paddrp->spp_address, &net, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } } if (stcb && (net == NULL)) { struct sockaddr *sa; sa = (struct sockaddr *)&paddrp->spp_address; #ifdef INET if (sa->sa_family == AF_INET) { struct sockaddr_in *sin; sin = (struct sockaddr_in *)sa; if (sin->sin_addr.s_addr) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); SCTP_TCB_UNLOCK(stcb); error = EINVAL; break; } } else #endif #ifdef INET6 if (sa->sa_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)sa; if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); SCTP_TCB_UNLOCK(stcb); error = EINVAL; break; } } else #endif { error = EAFNOSUPPORT; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } /* sanity checks */ if ((paddrp->spp_flags & SPP_HB_ENABLE) && (paddrp->spp_flags & SPP_HB_DISABLE)) { if (stcb) SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } if ((paddrp->spp_flags & SPP_PMTUD_ENABLE) && (paddrp->spp_flags & SPP_PMTUD_DISABLE)) { if (stcb) SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } if (stcb) { /************************TCB SPECIFIC SET ******************/ /* * do we change the timer for HB, we run * only one? */ int ovh = 0; if (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { ovh = SCTP_MED_OVERHEAD; } else { ovh = SCTP_MED_V4_OVERHEAD; } if (paddrp->spp_hbinterval) stcb->asoc.heart_beat_delay = paddrp->spp_hbinterval; else if (paddrp->spp_flags & SPP_HB_TIME_IS_ZERO) stcb->asoc.heart_beat_delay = 0; /* network sets ? */ if (net) { /************************NET SPECIFIC SET ******************/ if (paddrp->spp_flags & SPP_HB_DEMAND) { /* on demand HB */ if (sctp_send_hb(stcb, 1, net) < 0) { /* asoc destroyed */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } } if (paddrp->spp_flags & SPP_HB_DISABLE) { net->dest_state |= SCTP_ADDR_NOHB; } if (paddrp->spp_flags & SPP_HB_ENABLE) { net->dest_state &= ~SCTP_ADDR_NOHB; } if ((paddrp->spp_flags & SPP_PMTUD_DISABLE) && (paddrp->spp_pathmtu >= SCTP_SMALLEST_PMTU)) { if (SCTP_OS_TIMER_PENDING(&net->pmtu_timer.timer)) { sctp_timer_stop(SCTP_TIMER_TYPE_PATHMTURAISE, inp, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_10); } if (paddrp->spp_pathmtu > SCTP_DEFAULT_MINSEGMENT) { net->mtu = paddrp->spp_pathmtu + ovh; if (net->mtu < stcb->asoc.smallest_mtu) { sctp_pathmtu_adjustment(inp, stcb, net, net->mtu); } } } if (paddrp->spp_flags & SPP_PMTUD_ENABLE) { if (SCTP_OS_TIMER_PENDING(&net->pmtu_timer.timer)) { sctp_timer_start(SCTP_TIMER_TYPE_PATHMTURAISE, inp, stcb, net); } } if (paddrp->spp_pathmaxrxt) net->failure_threshold = paddrp->spp_pathmaxrxt; #ifdef INET if (paddrp->spp_flags & SPP_IPV4_TOS) { if (net->ro._l_addr.sin.sin_family == AF_INET) { net->tos_flowlabel = paddrp->spp_ipv4_tos & 0x000000fc; } } #endif #ifdef INET6 if (paddrp->spp_flags & SPP_IPV6_FLOWLABEL) { if (net->ro._l_addr.sin6.sin6_family == AF_INET6) { net->tos_flowlabel = paddrp->spp_ipv6_flowlabel; } } #endif } else { /************************ASSOC ONLY -- NO NET SPECIFIC SET ******************/ if (paddrp->spp_pathmaxrxt) stcb->asoc.def_net_failure = paddrp->spp_pathmaxrxt; if (paddrp->spp_flags & SPP_HB_ENABLE) { /* Turn back on the timer */ stcb->asoc.hb_is_disabled = 0; sctp_timer_start(SCTP_TIMER_TYPE_HEARTBEAT, inp, stcb, net); } if ((paddrp->spp_flags & SPP_PMTUD_DISABLE) && (paddrp->spp_pathmtu >= SCTP_SMALLEST_PMTU)) { TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { if (SCTP_OS_TIMER_PENDING(&net->pmtu_timer.timer)) { sctp_timer_stop(SCTP_TIMER_TYPE_PATHMTURAISE, inp, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_10); } if (paddrp->spp_pathmtu > SCTP_DEFAULT_MINSEGMENT) { net->mtu = paddrp->spp_pathmtu + ovh; if (net->mtu < stcb->asoc.smallest_mtu) { sctp_pathmtu_adjustment(inp, stcb, net, net->mtu); } } } } if (paddrp->spp_flags & SPP_PMTUD_ENABLE) { TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { if (SCTP_OS_TIMER_PENDING(&net->pmtu_timer.timer)) { sctp_timer_start(SCTP_TIMER_TYPE_PATHMTURAISE, inp, stcb, net); } } } if (paddrp->spp_flags & SPP_HB_DISABLE) { int cnt_of_unconf = 0; struct sctp_nets *lnet; stcb->asoc.hb_is_disabled = 1; TAILQ_FOREACH(lnet, &stcb->asoc.nets, sctp_next) { if (lnet->dest_state & SCTP_ADDR_UNCONFIRMED) { cnt_of_unconf++; } } /* * stop the timer ONLY if we * have no unconfirmed * addresses */ if (cnt_of_unconf == 0) { TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { sctp_timer_stop(SCTP_TIMER_TYPE_HEARTBEAT, inp, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_11); } } } if (paddrp->spp_flags & SPP_HB_ENABLE) { /* start up the timer. */ TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { sctp_timer_start(SCTP_TIMER_TYPE_HEARTBEAT, inp, stcb, net); } } #ifdef INET if (paddrp->spp_flags & SPP_IPV4_TOS) stcb->asoc.default_tos = paddrp->spp_ipv4_tos & 0x000000fc; #endif #ifdef INET6 if (paddrp->spp_flags & SPP_IPV6_FLOWLABEL) stcb->asoc.default_flowlabel = paddrp->spp_ipv6_flowlabel; #endif } SCTP_TCB_UNLOCK(stcb); } else { /************************NO TCB, SET TO default stuff ******************/ SCTP_INP_WLOCK(inp); /* * For the TOS/FLOWLABEL stuff you set it * with the options on the socket */ if (paddrp->spp_pathmaxrxt) { inp->sctp_ep.def_net_failure = paddrp->spp_pathmaxrxt; } if (paddrp->spp_flags & SPP_HB_TIME_IS_ZERO) inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_HEARTBEAT] = 0; else if (paddrp->spp_hbinterval) { if (paddrp->spp_hbinterval > SCTP_MAX_HB_INTERVAL) paddrp->spp_hbinterval = SCTP_MAX_HB_INTERVAL; inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_HEARTBEAT] = MSEC_TO_TICKS(paddrp->spp_hbinterval); } if (paddrp->spp_flags & SPP_HB_ENABLE) { sctp_feature_off(inp, SCTP_PCB_FLAGS_DONOT_HEARTBEAT); } else if (paddrp->spp_flags & SPP_HB_DISABLE) { sctp_feature_on(inp, SCTP_PCB_FLAGS_DONOT_HEARTBEAT); } SCTP_INP_WUNLOCK(inp); } } break; case SCTP_RTOINFO: { struct sctp_rtoinfo *srto; uint32_t new_init, new_min, new_max; SCTP_CHECK_AND_CAST(srto, optval, struct sctp_rtoinfo, optsize); SCTP_FIND_STCB(inp, stcb, srto->srto_assoc_id); if (stcb) { if (srto->srto_initial) new_init = srto->srto_initial; else new_init = stcb->asoc.initial_rto; if (srto->srto_max) new_max = srto->srto_max; else new_max = stcb->asoc.maxrto; if (srto->srto_min) new_min = srto->srto_min; else new_min = stcb->asoc.minrto; if ((new_min <= new_init) && (new_init <= new_max)) { stcb->asoc.initial_rto = new_init; stcb->asoc.maxrto = new_max; stcb->asoc.minrto = new_min; } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_WLOCK(inp); if (srto->srto_initial) new_init = srto->srto_initial; else new_init = inp->sctp_ep.initial_rto; if (srto->srto_max) new_max = srto->srto_max; else new_max = inp->sctp_ep.sctp_maxrto; if (srto->srto_min) new_min = srto->srto_min; else new_min = inp->sctp_ep.sctp_minrto; if ((new_min <= new_init) && (new_init <= new_max)) { inp->sctp_ep.initial_rto = new_init; inp->sctp_ep.sctp_maxrto = new_max; inp->sctp_ep.sctp_minrto = new_min; } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_INP_WUNLOCK(inp); } } break; case SCTP_ASSOCINFO: { struct sctp_assocparams *sasoc; SCTP_CHECK_AND_CAST(sasoc, optval, struct sctp_assocparams, optsize); SCTP_FIND_STCB(inp, stcb, sasoc->sasoc_assoc_id); if (sasoc->sasoc_cookie_life) { /* boundary check the cookie life */ if (sasoc->sasoc_cookie_life < 1000) sasoc->sasoc_cookie_life = 1000; if (sasoc->sasoc_cookie_life > SCTP_MAX_COOKIE_LIFE) { sasoc->sasoc_cookie_life = SCTP_MAX_COOKIE_LIFE; } } if (stcb) { if (sasoc->sasoc_asocmaxrxt) stcb->asoc.max_send_times = sasoc->sasoc_asocmaxrxt; sasoc->sasoc_number_peer_destinations = stcb->asoc.numnets; sasoc->sasoc_peer_rwnd = 0; sasoc->sasoc_local_rwnd = 0; if (sasoc->sasoc_cookie_life) { stcb->asoc.cookie_life = MSEC_TO_TICKS(sasoc->sasoc_cookie_life); } SCTP_TCB_UNLOCK(stcb); } else { SCTP_INP_WLOCK(inp); if (sasoc->sasoc_asocmaxrxt) inp->sctp_ep.max_send_times = sasoc->sasoc_asocmaxrxt; sasoc->sasoc_number_peer_destinations = 0; sasoc->sasoc_peer_rwnd = 0; sasoc->sasoc_local_rwnd = 0; if (sasoc->sasoc_cookie_life) { inp->sctp_ep.def_cookie_life = MSEC_TO_TICKS(sasoc->sasoc_cookie_life); } SCTP_INP_WUNLOCK(inp); } } break; case SCTP_INITMSG: { struct sctp_initmsg *sinit; SCTP_CHECK_AND_CAST(sinit, optval, struct sctp_initmsg, optsize); SCTP_INP_WLOCK(inp); if (sinit->sinit_num_ostreams) inp->sctp_ep.pre_open_stream_count = sinit->sinit_num_ostreams; if (sinit->sinit_max_instreams) inp->sctp_ep.max_open_streams_intome = sinit->sinit_max_instreams; if (sinit->sinit_max_attempts) inp->sctp_ep.max_init_times = sinit->sinit_max_attempts; if (sinit->sinit_max_init_timeo) inp->sctp_ep.initial_init_rto_max = sinit->sinit_max_init_timeo; SCTP_INP_WUNLOCK(inp); } break; case SCTP_PRIMARY_ADDR: { struct sctp_setprim *spa; struct sctp_nets *net, *lnet; SCTP_CHECK_AND_CAST(spa, optval, struct sctp_setprim, optsize); SCTP_FIND_STCB(inp, stcb, spa->ssp_assoc_id); net = NULL; if (stcb) { net = sctp_findnet(stcb, (struct sockaddr *)&spa->ssp_addr); } else { /* * We increment here since * sctp_findassociation_ep_addr() wil do a * decrement if it finds the stcb as long as * the locked tcb (last argument) is NOT a * TCB.. aka NULL. */ SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, (struct sockaddr *)&spa->ssp_addr, &net, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } } if ((stcb) && (net)) { if ((net != stcb->asoc.primary_destination) && (!(net->dest_state & SCTP_ADDR_UNCONFIRMED))) { /* Ok we need to set it */ lnet = stcb->asoc.primary_destination; if (sctp_set_primary_addr(stcb, (struct sockaddr *)NULL, net) == 0) { if (net->dest_state & SCTP_ADDR_SWITCH_PRIMARY) { net->dest_state |= SCTP_ADDR_DOUBLE_SWITCH; } net->dest_state |= SCTP_ADDR_SWITCH_PRIMARY; } } } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } if (stcb) { SCTP_TCB_UNLOCK(stcb); } } break; case SCTP_SET_DYNAMIC_PRIMARY: { union sctp_sockstore *ss; error = priv_check(curthread, PRIV_NETINET_RESERVEDPORT); if (error) break; SCTP_CHECK_AND_CAST(ss, optval, union sctp_sockstore, optsize); /* SUPER USER CHECK? */ error = sctp_dynamic_set_primary(&ss->sa, vrf_id); } break; case SCTP_SET_PEER_PRIMARY_ADDR: { struct sctp_setpeerprim *sspp; SCTP_CHECK_AND_CAST(sspp, optval, struct sctp_setpeerprim, optsize); SCTP_FIND_STCB(inp, stcb, sspp->sspp_assoc_id); if (stcb != NULL) { struct sctp_ifa *ifa; ifa = sctp_find_ifa_by_addr((struct sockaddr *)&sspp->sspp_addr, stcb->asoc.vrf_id, SCTP_ADDR_NOT_LOCKED); if (ifa == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto out_of_it; } if ((inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) == 0) { /* * Must validate the ifa found is in * our ep */ struct sctp_laddr *laddr; int found = 0; LIST_FOREACH(laddr, &inp->sctp_addr_list, sctp_nxt_addr) { if (laddr->ifa == NULL) { SCTPDBG(SCTP_DEBUG_OUTPUT1, "%s: NULL ifa\n", __FUNCTION__); continue; } if (laddr->ifa == ifa) { found = 1; break; } } if (!found) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto out_of_it; } } if (sctp_set_primary_ip_address_sa(stcb, (struct sockaddr *)&sspp->sspp_addr) != 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } out_of_it: SCTP_TCB_UNLOCK(stcb); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; case SCTP_BINDX_ADD_ADDR: { struct sctp_getaddresses *addrs; size_t sz; struct thread *td; td = (struct thread *)p; SCTP_CHECK_AND_CAST(addrs, optval, struct sctp_getaddresses, optsize); #ifdef INET if (addrs->addr->sa_family == AF_INET) { sz = sizeof(struct sctp_getaddresses) - sizeof(struct sockaddr) + sizeof(struct sockaddr_in); if (optsize < sz) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } if (td != NULL && (error = prison_local_ip4(td->td_ucred, &(((struct sockaddr_in *)(addrs->addr))->sin_addr)))) { SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif #ifdef INET6 if (addrs->addr->sa_family == AF_INET6) { sz = sizeof(struct sctp_getaddresses) - sizeof(struct sockaddr) + sizeof(struct sockaddr_in6); if (optsize < sz) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } if (td != NULL && (error = prison_local_ip6(td->td_ucred, &(((struct sockaddr_in6 *)(addrs->addr))->sin6_addr), (SCTP_IPV6_V6ONLY(inp) != 0))) != 0) { SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif { error = EAFNOSUPPORT; break; } sctp_bindx_add_address(so, inp, addrs->addr, addrs->sget_assoc_id, vrf_id, &error, p); } break; case SCTP_BINDX_REM_ADDR: { struct sctp_getaddresses *addrs; size_t sz; struct thread *td; td = (struct thread *)p; SCTP_CHECK_AND_CAST(addrs, optval, struct sctp_getaddresses, optsize); #ifdef INET if (addrs->addr->sa_family == AF_INET) { sz = sizeof(struct sctp_getaddresses) - sizeof(struct sockaddr) + sizeof(struct sockaddr_in); if (optsize < sz) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } if (td != NULL && (error = prison_local_ip4(td->td_ucred, &(((struct sockaddr_in *)(addrs->addr))->sin_addr)))) { SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif #ifdef INET6 if (addrs->addr->sa_family == AF_INET6) { sz = sizeof(struct sctp_getaddresses) - sizeof(struct sockaddr) + sizeof(struct sockaddr_in6); if (optsize < sz) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } if (td != NULL && (error = prison_local_ip6(td->td_ucred, &(((struct sockaddr_in6 *)(addrs->addr))->sin6_addr), (SCTP_IPV6_V6ONLY(inp) != 0))) != 0) { SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif { error = EAFNOSUPPORT; break; } sctp_bindx_delete_address(so, inp, addrs->addr, addrs->sget_assoc_id, vrf_id, &error); } break; default: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOPROTOOPT); error = ENOPROTOOPT; break; } /* end switch (opt) */ return (error); } int sctp_ctloutput(struct socket *so, struct sockopt *sopt) { void *optval = NULL; size_t optsize = 0; struct sctp_inpcb *inp; void *p; int error = 0; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == 0) { /* I made the same as TCP since we are not setup? */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (ECONNRESET); } if (sopt->sopt_level != IPPROTO_SCTP) { /* wrong proto level... send back up to IP */ #ifdef INET6 if (INP_CHECK_SOCKAF(so, AF_INET6)) error = ip6_ctloutput(so, sopt); #endif /* INET6 */ #if defined(INET) && defined (INET6) else #endif #ifdef INET error = ip_ctloutput(so, sopt); #endif return (error); } optsize = sopt->sopt_valsize; if (optsize) { SCTP_MALLOC(optval, void *, optsize, SCTP_M_SOCKOPT); if (optval == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOBUFS); return (ENOBUFS); } error = sooptcopyin(sopt, optval, optsize, optsize); if (error) { SCTP_FREE(optval, SCTP_M_SOCKOPT); goto out; } } p = (void *)sopt->sopt_td; if (sopt->sopt_dir == SOPT_SET) { error = sctp_setopt(so, sopt->sopt_name, optval, optsize, p); } else if (sopt->sopt_dir == SOPT_GET) { error = sctp_getopt(so, sopt->sopt_name, optval, &optsize, p); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } if ((error == 0) && (optval != NULL)) { error = sooptcopyout(sopt, optval, optsize); SCTP_FREE(optval, SCTP_M_SOCKOPT); } else if (optval != NULL) { SCTP_FREE(optval, SCTP_M_SOCKOPT); } out: return (error); } #ifdef INET static int sctp_connect(struct socket *so, struct sockaddr *addr, struct thread *p) { int error = 0; int create_lock_on = 0; uint32_t vrf_id; struct sctp_inpcb *inp; struct sctp_tcb *stcb = NULL; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == 0) { /* I made the same as TCP since we are not setup? */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (ECONNRESET); } if (addr == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return EINVAL; } switch (addr->sa_family) { #ifdef INET6 case AF_INET6: { struct sockaddr_in6 *sin6p; if (addr->sa_len != sizeof(struct sockaddr_in6)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } sin6p = (struct sockaddr_in6 *)addr; if (p != NULL && (error = prison_remote_ip6(p->td_ucred, &sin6p->sin6_addr)) != 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); return (error); } break; } #endif #ifdef INET case AF_INET: { struct sockaddr_in *sinp; if (addr->sa_len != sizeof(struct sockaddr_in)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } sinp = (struct sockaddr_in *)addr; if (p != NULL && (error = prison_remote_ip4(p->td_ucred, &sinp->sin_addr)) != 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); return (error); } break; } #endif default: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EAFNOSUPPORT); return (EAFNOSUPPORT); } SCTP_INP_INCR_REF(inp); SCTP_ASOC_CREATE_LOCK(inp); create_lock_on = 1; if ((inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE) || (inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE)) { /* Should I really unlock ? */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EFAULT); error = EFAULT; goto out_now; } #ifdef INET6 if (((inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) == 0) && (addr->sa_family == AF_INET6)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto out_now; } #endif /* INET6 */ if ((inp->sctp_flags & SCTP_PCB_FLAGS_UNBOUND) == SCTP_PCB_FLAGS_UNBOUND) { /* Bind a ephemeral port */ error = sctp_inpcb_bind(so, NULL, NULL, p); if (error) { goto out_now; } } /* Now do we connect? */ if ((inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) && (sctp_is_feature_off(inp, SCTP_PCB_FLAGS_PORTREUSE))) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto out_now; } if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) && (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED)) { /* We are already connected AND the TCP model */ SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, EADDRINUSE); error = EADDRINUSE; goto out_now; } if (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) { SCTP_INP_RLOCK(inp); stcb = LIST_FIRST(&inp->sctp_asoc_list); SCTP_INP_RUNLOCK(inp); } else { /* * We increment here since sctp_findassociation_ep_addr() * will do a decrement if it finds the stcb as long as the * locked tcb (last argument) is NOT a TCB.. aka NULL. */ SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, addr, NULL, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } else { SCTP_TCB_UNLOCK(stcb); } } if (stcb != NULL) { /* Already have or am bring up an association */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EALREADY); error = EALREADY; goto out_now; } vrf_id = inp->def_vrf_id; /* We are GOOD to go */ stcb = sctp_aloc_assoc(inp, addr, &error, 0, vrf_id, p); if (stcb == NULL) { /* Gak! no memory */ goto out_now; } if (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) { stcb->sctp_ep->sctp_flags |= SCTP_PCB_FLAGS_CONNECTED; /* Set the connected flag so we can queue data */ SOCKBUF_LOCK(&so->so_rcv); so->so_rcv.sb_state &= ~SBS_CANTRCVMORE; SOCKBUF_UNLOCK(&so->so_rcv); SOCKBUF_LOCK(&so->so_snd); so->so_snd.sb_state &= ~SBS_CANTSENDMORE; SOCKBUF_UNLOCK(&so->so_snd); SOCK_LOCK(so); so->so_state &= ~SS_ISDISCONNECTING; SOCK_UNLOCK(so); soisconnecting(so); } SCTP_SET_STATE(&stcb->asoc, SCTP_STATE_COOKIE_WAIT); (void)SCTP_GETTIME_TIMEVAL(&stcb->asoc.time_entered); /* initialize authentication parameters for the assoc */ sctp_initialize_auth_params(inp, stcb); sctp_send_initiate(inp, stcb, SCTP_SO_LOCKED); SCTP_TCB_UNLOCK(stcb); out_now: if (create_lock_on) { SCTP_ASOC_CREATE_UNLOCK(inp); } SCTP_INP_DECR_REF(inp); return error; } #endif int sctp_listen(struct socket *so, int backlog, struct thread *p) { /* * Note this module depends on the protocol processing being called * AFTER any socket level flags and backlog are applied to the * socket. The traditional way that the socket flags are applied is * AFTER protocol processing. We have made a change to the * sys/kern/uipc_socket.c module to reverse this but this MUST be in * place if the socket API for SCTP is to work properly. */ int error = 0; struct sctp_inpcb *inp; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == 0) { /* I made the same as TCP since we are not setup? */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (ECONNRESET); } if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_PORTREUSE)) { /* See if we have a listener */ struct sctp_inpcb *tinp; union sctp_sockstore store, *sp; sp = &store; if ((inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) == 0) { /* not bound all */ struct sctp_laddr *laddr; LIST_FOREACH(laddr, &inp->sctp_addr_list, sctp_nxt_addr) { memcpy(&store, &laddr->ifa->address, sizeof(store)); switch (sp->sa.sa_family) { #ifdef INET case AF_INET: sp->sin.sin_port = inp->sctp_lport; break; #endif #ifdef INET6 case AF_INET6: sp->sin6.sin6_port = inp->sctp_lport; break; #endif default: break; } tinp = sctp_pcb_findep(&sp->sa, 0, 0, inp->def_vrf_id); if (tinp && (tinp != inp) && ((tinp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE) == 0) && ((tinp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) && (tinp->sctp_socket->so_qlimit)) { /* * we have a listener already and * its not this inp. */ SCTP_INP_DECR_REF(tinp); return (EADDRINUSE); } else if (tinp) { SCTP_INP_DECR_REF(tinp); } } } else { /* Setup a local addr bound all */ memset(&store, 0, sizeof(store)); switch (sp->sa.sa_family) { #ifdef INET case AF_INET: store.sin.sin_port = inp->sctp_lport; break; #endif #ifdef INET6 case AF_INET6: sp->sin6.sin6_port = inp->sctp_lport; break; #endif default: break; } #ifdef INET6 if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { store.sa.sa_family = AF_INET6; store.sa.sa_len = sizeof(struct sockaddr_in6); } #endif #ifdef INET if ((inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) == 0) { store.sa.sa_family = AF_INET; store.sa.sa_len = sizeof(struct sockaddr_in); } #endif tinp = sctp_pcb_findep(&sp->sa, 0, 0, inp->def_vrf_id); if (tinp && (tinp != inp) && ((tinp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE) == 0) && ((tinp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) && (tinp->sctp_socket->so_qlimit)) { /* * we have a listener already and its not * this inp. */ SCTP_INP_DECR_REF(tinp); return (EADDRINUSE); } else if (tinp) { SCTP_INP_DECR_REF(inp); } } } SCTP_INP_RLOCK(inp); #ifdef SCTP_LOCK_LOGGING if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_LOCK_LOGGING_ENABLE) { sctp_log_lock(inp, (struct sctp_tcb *)NULL, SCTP_LOG_LOCK_SOCK); } #endif SOCK_LOCK(so); error = solisten_proto_check(so); if (error) { SOCK_UNLOCK(so); SCTP_INP_RUNLOCK(inp); return (error); } if ((sctp_is_feature_on(inp, SCTP_PCB_FLAGS_PORTREUSE)) && (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) { /* * The unlucky case - We are in the tcp pool with this guy. * - Someone else is in the main inp slot. - We must move * this guy (the listener) to the main slot - We must then * move the guy that was listener to the TCP Pool. */ if (sctp_swap_inpcb_for_listen(inp)) { goto in_use; } } if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) && (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED)) { /* We are already connected AND the TCP model */ in_use: SCTP_INP_RUNLOCK(inp); SOCK_UNLOCK(so); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EADDRINUSE); return (EADDRINUSE); } SCTP_INP_RUNLOCK(inp); if (inp->sctp_flags & SCTP_PCB_FLAGS_UNBOUND) { /* We must do a bind. */ SOCK_UNLOCK(so); if ((error = sctp_inpcb_bind(so, NULL, NULL, p))) { /* bind error, probably perm */ return (error); } SOCK_LOCK(so); } /* It appears for 7.0 and on, we must always call this. */ solisten_proto(so, backlog); if (inp->sctp_flags & SCTP_PCB_FLAGS_UDPTYPE) { /* remove the ACCEPTCONN flag for one-to-many sockets */ so->so_options &= ~SO_ACCEPTCONN; } if (backlog == 0) { /* turning off listen */ so->so_options &= ~SO_ACCEPTCONN; } SOCK_UNLOCK(so); return (error); } static int sctp_defered_wakeup_cnt = 0; int sctp_accept(struct socket *so, struct sockaddr **addr) { struct sctp_tcb *stcb; struct sctp_inpcb *inp; union sctp_sockstore store; #ifdef INET6 int error; #endif inp = (struct sctp_inpcb *)so->so_pcb; if (inp == 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (ECONNRESET); } SCTP_INP_RLOCK(inp); if (inp->sctp_flags & SCTP_PCB_FLAGS_UDPTYPE) { SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); return (EOPNOTSUPP); } if (so->so_state & SS_ISDISCONNECTED) { SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ECONNABORTED); return (ECONNABORTED); } stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb == NULL) { SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (ECONNRESET); } SCTP_TCB_LOCK(stcb); SCTP_INP_RUNLOCK(inp); store = stcb->asoc.primary_destination->ro._l_addr; stcb->asoc.state &= ~SCTP_STATE_IN_ACCEPT_QUEUE; SCTP_TCB_UNLOCK(stcb); switch (store.sa.sa_family) { #ifdef INET case AF_INET: { struct sockaddr_in *sin; SCTP_MALLOC_SONAME(sin, struct sockaddr_in *, sizeof *sin); if (sin == NULL) return (ENOMEM); sin->sin_family = AF_INET; sin->sin_len = sizeof(*sin); sin->sin_port = ((struct sockaddr_in *)&store)->sin_port; sin->sin_addr = ((struct sockaddr_in *)&store)->sin_addr; *addr = (struct sockaddr *)sin; break; } #endif #ifdef INET6 case AF_INET6: { struct sockaddr_in6 *sin6; SCTP_MALLOC_SONAME(sin6, struct sockaddr_in6 *, sizeof *sin6); if (sin6 == NULL) return (ENOMEM); sin6->sin6_family = AF_INET6; sin6->sin6_len = sizeof(*sin6); sin6->sin6_port = ((struct sockaddr_in6 *)&store)->sin6_port; sin6->sin6_addr = ((struct sockaddr_in6 *)&store)->sin6_addr; if ((error = sa6_recoverscope(sin6)) != 0) { SCTP_FREE_SONAME(sin6); return (error); } *addr = (struct sockaddr *)sin6; break; } #endif default: /* TSNH */ break; } /* Wake any delayed sleep action */ if (inp->sctp_flags & SCTP_PCB_FLAGS_DONT_WAKE) { SCTP_INP_WLOCK(inp); inp->sctp_flags &= ~SCTP_PCB_FLAGS_DONT_WAKE; if (inp->sctp_flags & SCTP_PCB_FLAGS_WAKEOUTPUT) { inp->sctp_flags &= ~SCTP_PCB_FLAGS_WAKEOUTPUT; SCTP_INP_WUNLOCK(inp); SOCKBUF_LOCK(&inp->sctp_socket->so_snd); if (sowriteable(inp->sctp_socket)) { sowwakeup_locked(inp->sctp_socket); } else { SOCKBUF_UNLOCK(&inp->sctp_socket->so_snd); } SCTP_INP_WLOCK(inp); } if (inp->sctp_flags & SCTP_PCB_FLAGS_WAKEINPUT) { inp->sctp_flags &= ~SCTP_PCB_FLAGS_WAKEINPUT; SCTP_INP_WUNLOCK(inp); SOCKBUF_LOCK(&inp->sctp_socket->so_rcv); if (soreadable(inp->sctp_socket)) { sctp_defered_wakeup_cnt++; sorwakeup_locked(inp->sctp_socket); } else { SOCKBUF_UNLOCK(&inp->sctp_socket->so_rcv); } SCTP_INP_WLOCK(inp); } SCTP_INP_WUNLOCK(inp); } if (stcb->asoc.state & SCTP_STATE_ABOUT_TO_BE_FREED) { SCTP_TCB_LOCK(stcb); sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_7); } return (0); } #ifdef INET int sctp_ingetaddr(struct socket *so, struct sockaddr **addr) { struct sockaddr_in *sin; uint32_t vrf_id; struct sctp_inpcb *inp; struct sctp_ifa *sctp_ifa; /* * Do the malloc first in case it blocks. */ SCTP_MALLOC_SONAME(sin, struct sockaddr_in *, sizeof *sin); if (sin == NULL) return (ENOMEM); sin->sin_family = AF_INET; sin->sin_len = sizeof(*sin); inp = (struct sctp_inpcb *)so->so_pcb; if (!inp) { SCTP_FREE_SONAME(sin); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return ECONNRESET; } SCTP_INP_RLOCK(inp); sin->sin_port = inp->sctp_lport; if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) { if (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) { struct sctp_tcb *stcb; struct sockaddr_in *sin_a; struct sctp_nets *net; int fnd; stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb == NULL) { goto notConn; } fnd = 0; sin_a = NULL; SCTP_TCB_LOCK(stcb); TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { sin_a = (struct sockaddr_in *)&net->ro._l_addr; if (sin_a == NULL) /* this will make coverity happy */ continue; if (sin_a->sin_family == AF_INET) { fnd = 1; break; } } if ((!fnd) || (sin_a == NULL)) { /* punt */ SCTP_TCB_UNLOCK(stcb); goto notConn; } vrf_id = inp->def_vrf_id; sctp_ifa = sctp_source_address_selection(inp, stcb, (sctp_route_t *) & net->ro, net, 0, vrf_id); if (sctp_ifa) { sin->sin_addr = sctp_ifa->address.sin.sin_addr; sctp_free_ifa(sctp_ifa); } SCTP_TCB_UNLOCK(stcb); } else { /* For the bound all case you get back 0 */ notConn: sin->sin_addr.s_addr = 0; } } else { /* Take the first IPv4 address in the list */ struct sctp_laddr *laddr; int fnd = 0; LIST_FOREACH(laddr, &inp->sctp_addr_list, sctp_nxt_addr) { if (laddr->ifa->address.sa.sa_family == AF_INET) { struct sockaddr_in *sin_a; sin_a = (struct sockaddr_in *)&laddr->ifa->address.sa; sin->sin_addr = sin_a->sin_addr; fnd = 1; break; } } if (!fnd) { SCTP_FREE_SONAME(sin); SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); return ENOENT; } } SCTP_INP_RUNLOCK(inp); (*addr) = (struct sockaddr *)sin; return (0); } int sctp_peeraddr(struct socket *so, struct sockaddr **addr) { struct sockaddr_in *sin = (struct sockaddr_in *)*addr; int fnd; struct sockaddr_in *sin_a; struct sctp_inpcb *inp; struct sctp_tcb *stcb; struct sctp_nets *net; /* Do the malloc first in case it blocks. */ inp = (struct sctp_inpcb *)so->so_pcb; if ((inp == NULL) || ((inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) == 0)) { /* UDP type and listeners will drop out here */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOTCONN); return (ENOTCONN); } SCTP_MALLOC_SONAME(sin, struct sockaddr_in *, sizeof *sin); if (sin == NULL) return (ENOMEM); sin->sin_family = AF_INET; sin->sin_len = sizeof(*sin); /* We must recapture incase we blocked */ inp = (struct sctp_inpcb *)so->so_pcb; if (!inp) { SCTP_FREE_SONAME(sin); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return ECONNRESET; } SCTP_INP_RLOCK(inp); stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb) { SCTP_TCB_LOCK(stcb); } SCTP_INP_RUNLOCK(inp); if (stcb == NULL) { SCTP_FREE_SONAME(sin); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return ECONNRESET; } fnd = 0; TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { sin_a = (struct sockaddr_in *)&net->ro._l_addr; if (sin_a->sin_family == AF_INET) { fnd = 1; sin->sin_port = stcb->rport; sin->sin_addr = sin_a->sin_addr; break; } } SCTP_TCB_UNLOCK(stcb); if (!fnd) { /* No IPv4 address */ SCTP_FREE_SONAME(sin); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); return ENOENT; } (*addr) = (struct sockaddr *)sin; return (0); } #ifdef INET struct pr_usrreqs sctp_usrreqs = { .pru_abort = sctp_abort, .pru_accept = sctp_accept, .pru_attach = sctp_attach, .pru_bind = sctp_bind, .pru_connect = sctp_connect, .pru_control = in_control, .pru_close = sctp_close, .pru_detach = sctp_close, .pru_sopoll = sopoll_generic, .pru_flush = sctp_flush, .pru_disconnect = sctp_disconnect, .pru_listen = sctp_listen, .pru_peeraddr = sctp_peeraddr, .pru_send = sctp_sendm, .pru_shutdown = sctp_shutdown, .pru_sockaddr = sctp_ingetaddr, .pru_sosend = sctp_sosend, .pru_soreceive = sctp_soreceive }; #endif #endif Index: projects/largeSMP/sys/nfs/nfs_nfssvc.c =================================================================== --- projects/largeSMP/sys/nfs/nfs_nfssvc.c (revision 221494) +++ projects/largeSMP/sys/nfs/nfs_nfssvc.c (revision 221495) @@ -1,153 +1,156 @@ /*- * Copyright (c) 1989, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Rick Macklem at The University of Guelph. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * */ #include __FBSDID("$FreeBSD$"); #include "opt_nfs.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include static int nfssvc_offset = SYS_nfssvc; static struct sysent nfssvc_prev_sysent; MAKE_SYSENT(nfssvc); /* * This tiny module simply handles the nfssvc() system call. The other * nfs modules that use the system call register themselves by setting * the nfsd_call_xxx function pointers non-NULL. */ int (*nfsd_call_nfsserver)(struct thread *, struct nfssvc_args *) = NULL; int (*nfsd_call_nfscommon)(struct thread *, struct nfssvc_args *) = NULL; int (*nfsd_call_nfscl)(struct thread *, struct nfssvc_args *) = NULL; int (*nfsd_call_nfsd)(struct thread *, struct nfssvc_args *) = NULL; /* * Nfs server psuedo system call for the nfsd's */ int nfssvc(struct thread *td, struct nfssvc_args *uap) { int error; KASSERT(!mtx_owned(&Giant), ("nfssvc(): called with Giant")); AUDIT_ARG_CMD(uap->flag); - error = priv_check(td, PRIV_NFS_DAEMON); - if (error) - return (error); + /* Allow anyone to get the stats. */ + if ((uap->flag & ~NFSSVC_GETSTATS) != 0) { + error = priv_check(td, PRIV_NFS_DAEMON); + if (error != 0) + return (error); + } error = EINVAL; if ((uap->flag & (NFSSVC_ADDSOCK | NFSSVC_OLDNFSD | NFSSVC_NFSD)) && nfsd_call_nfsserver != NULL) error = (*nfsd_call_nfsserver)(td, uap); else if ((uap->flag & (NFSSVC_CBADDSOCK | NFSSVC_NFSCBD)) && nfsd_call_nfscl != NULL) error = (*nfsd_call_nfscl)(td, uap); else if ((uap->flag & (NFSSVC_IDNAME | NFSSVC_GETSTATS | NFSSVC_GSSDADDPORT | NFSSVC_GSSDADDFIRST | NFSSVC_GSSDDELETEALL | NFSSVC_NFSUSERDPORT | NFSSVC_NFSUSERDDELPORT)) && nfsd_call_nfscommon != NULL) error = (*nfsd_call_nfscommon)(td, uap); else if ((uap->flag & (NFSSVC_NFSDNFSD | NFSSVC_NFSDADDSOCK | NFSSVC_PUBLICFH | NFSSVC_V4ROOTEXPORT | NFSSVC_NOPUBLICFH | NFSSVC_STABLERESTART | NFSSVC_ADMINREVOKE | NFSSVC_DUMPCLIENTS | NFSSVC_DUMPLOCKS | NFSSVC_BACKUPSTABLE)) && nfsd_call_nfsd != NULL) error = (*nfsd_call_nfsd)(td, uap); if (error == EINTR || error == ERESTART) error = 0; return (error); } /* * Called once to initialize data structures... */ static int nfssvc_modevent(module_t mod, int type, void *data) { static int registered; int error = 0; switch (type) { case MOD_LOAD: error = syscall_register(&nfssvc_offset, &nfssvc_sysent, &nfssvc_prev_sysent); if (error) break; registered = 1; break; case MOD_UNLOAD: if (nfsd_call_nfsserver != NULL || nfsd_call_nfscommon != NULL || nfsd_call_nfscl != NULL || nfsd_call_nfsd != NULL) { error = EBUSY; break; } if (registered) syscall_deregister(&nfssvc_offset, &nfssvc_prev_sysent); registered = 0; break; default: error = EOPNOTSUPP; break; } return error; } static moduledata_t nfssvc_mod = { "nfssvc", nfssvc_modevent, NULL, }; DECLARE_MODULE(nfssvc, nfssvc_mod, SI_SUB_VFS, SI_ORDER_ANY); /* So that loader and kldload(2) can find us, wherever we are.. */ MODULE_VERSION(nfssvc, 1); Index: projects/largeSMP/sys/sys/stddef.h =================================================================== --- projects/largeSMP/sys/sys/stddef.h (revision 221494) +++ projects/largeSMP/sys/sys/stddef.h (revision 221495) @@ -1,40 +1,43 @@ /*- * Copyright (c) 2002 Maxime Henrion * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #ifndef _SYS_STDDEF_H_ #define _SYS_STDDEF_H_ #include #include #include +#ifndef _PTRDIFF_T_DECLARED typedef __ptrdiff_t ptrdiff_t; +#define _PTRDIFF_T_DECLARED +#endif #define offsetof(type, field) __offsetof(type, field) #endif /* !_SYS_STDDEF_H_ */ Index: projects/largeSMP/sys =================================================================== --- projects/largeSMP/sys (revision 221494) +++ projects/largeSMP/sys (revision 221495) Property changes on: projects/largeSMP/sys ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys:r221444-221493 Index: projects/largeSMP/tools/regression/bin/sh/errors/bad-parm-exp6.2 =================================================================== --- projects/largeSMP/tools/regression/bin/sh/errors/bad-parm-exp6.2 (nonexistent) +++ projects/largeSMP/tools/regression/bin/sh/errors/bad-parm-exp6.2 (revision 221495) @@ -0,0 +1,2 @@ +# $FreeBSD$ +${#foo^} Property changes on: projects/largeSMP/tools/regression/bin/sh/errors/bad-parm-exp6.2 ___________________________________________________________________ Added: svn:eol-style ## -0,0 +1 ## +native \ No newline at end of property Added: svn:keywords ## -0,0 +1 ## +FreeBSD=%H \ No newline at end of property Added: svn:mime-type ## -0,0 +1 ## +text/plain \ No newline at end of property Index: projects/largeSMP/tools/regression/bin/sh/errors/bad-parm-exp6.2.stderr =================================================================== --- projects/largeSMP/tools/regression/bin/sh/errors/bad-parm-exp6.2.stderr (nonexistent) +++ projects/largeSMP/tools/regression/bin/sh/errors/bad-parm-exp6.2.stderr (revision 221495) @@ -0,0 +1 @@ +./errors/bad-parm-exp6.2: ${foo...}: Bad substitution Property changes on: projects/largeSMP/tools/regression/bin/sh/errors/bad-parm-exp6.2.stderr ___________________________________________________________________ Added: fbsd:nokeywords ## -0,0 +1 ## +y \ No newline at end of property Index: projects/largeSMP/tools/regression/bin/sh/expansion/set-u3.0 =================================================================== --- projects/largeSMP/tools/regression/bin/sh/expansion/set-u3.0 (nonexistent) +++ projects/largeSMP/tools/regression/bin/sh/expansion/set-u3.0 (revision 221495) @@ -0,0 +1,6 @@ +# $FreeBSD$ + +set -u +unset x +v=$( (eval ': $((x))') 2>&1 >/dev/null) +[ $? -ne 0 ] && [ -n "$v" ] Property changes on: projects/largeSMP/tools/regression/bin/sh/expansion/set-u3.0 ___________________________________________________________________ Added: svn:keywords ## -0,0 +1 ## +FreeBSD=%H \ No newline at end of property Index: projects/largeSMP/usr.bin/calendar =================================================================== --- projects/largeSMP/usr.bin/calendar (revision 221494) +++ projects/largeSMP/usr.bin/calendar (revision 221495) Property changes on: projects/largeSMP/usr.bin/calendar ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/usr.bin/calendar:r221444-221493 Index: projects/largeSMP/usr.bin/csup =================================================================== --- projects/largeSMP/usr.bin/csup (revision 221494) +++ projects/largeSMP/usr.bin/csup (revision 221495) Property changes on: projects/largeSMP/usr.bin/csup ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/usr.bin/csup:r221444-221493 Index: projects/largeSMP/usr.bin/nfsstat/nfsstat.1 =================================================================== --- projects/largeSMP/usr.bin/nfsstat/nfsstat.1 (revision 221494) +++ projects/largeSMP/usr.bin/nfsstat/nfsstat.1 (revision 221495) @@ -1,106 +1,112 @@ .\" Copyright (c) 1989, 1990, 1993 .\" The Regents of the University of California. All rights reserved. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" 4. Neither the name of the University nor the names of its contributors .\" may be used to endorse or promote products derived from this software .\" without specific prior written permission. .\" .\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" .\" From: @(#)nfsstat.1 8.1 (Berkeley) 6/6/93 .\" $FreeBSD$ .\" -.Dd May 1, 2011 +.Dd May 4, 2011 .Dt NFSSTAT 1 .Os .Sh NAME .Nm nfsstat .Nd display .Tn NFS statistics .Sh SYNOPSIS .Nm -.Op Fl ceszW +.Op Fl ceoszW .Op Fl M Ar core .Op Fl N Ar system .Op Fl w Ar wait .Sh DESCRIPTION The .Nm command displays statistics kept about .Tn NFS client and server activity. .Pp The options are as follows: .Bl -tag -width indent .It Fl c Only display client side statistics. +.It Fl e +Report the extra statistics collected by the new NFS client and +server for NFSv4. +This option is incompatible with +.Fl o . .It Fl M Extract values associated with the name list from the specified core instead of the default .Pa /dev/kmem . .It Fl N Extract the name list from the specified system instead of the default .Pa /boot/kernel/kernel . +.It Fl o +Report statistics for the old NFS client and/or server. +Without this +option statistics for the new NFS client and/or server will be reported. .It Fl s Only display server side statistics. .It Fl W Use wide format with interval short summary. This option is especially useful when combined with .Fl c or .Fl s and a time delay. .It Fl w Display a shorter summary of .Tn NFS activity for both the client and server at .Ar wait second intervals. .It Fl z Reset statistics after displaying them. -.It Fl e -Gather statistics from the experimental nfs subsystem that includes -support for NFSv4 instead of the regular nfs subsystem. .El .Sh FILES .Bl -tag -width ".Pa /boot/kernel/kernel" -compact .It Pa /boot/kernel/kernel default kernel namelist .It Pa /dev/kmem default memory file .El .Sh SEE ALSO .Xr fstat 1 , .Xr netstat 1 , .Xr ps 1 , .Xr systat 1 , .Xr sysctl 3 , .Xr iostat 8 , .Xr nfsdumpstate 8 , .Xr pstat 8 , .Xr vmstat 8 .Sh HISTORY The .Nm command appeared in .Bx 4.4 . Index: projects/largeSMP/usr.bin/nfsstat/nfsstat.c =================================================================== --- projects/largeSMP/usr.bin/nfsstat/nfsstat.c (revision 221494) +++ projects/largeSMP/usr.bin/nfsstat/nfsstat.c (revision 221495) @@ -1,880 +1,1029 @@ /* * Copyright (c) 1983, 1989, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Rick Macklem at The University of Guelph. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef lint static const char copyright[] = "@(#) Copyright (c) 1983, 1989, 1993\n\ The Regents of the University of California. All rights reserved.\n"; #endif /* not lint */ #ifndef lint #if 0 static char sccsid[] = "@(#)nfsstat.c 8.2 (Berkeley) 3/31/95"; #endif static const char rcsid[] = "$FreeBSD$"; #endif /* not lint */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include struct nlist nl[] = { #define N_NFSSTAT 0 { .n_name = "nfsstats" }, #define N_NFSRVSTAT 1 { .n_name = "nfsrvstats" }, { .n_name = NULL }, }; kvm_t *kd; static int deadkernel = 0; static int widemode = 0; static int zflag = 0; -static int run_v4 = 0; +static int run_v4 = 1; static int printtitle = 1; static struct ext_nfsstats ext_nfsstats; +static int extra_output = 0; void intpr(int, int); void printhdr(int, int); void sidewaysintpr(u_int, int, int); void usage(void); char *sperc1(int, int); char *sperc2(int, int); void exp_intpr(int, int); void exp_sidewaysintpr(u_int, int, int); #define DELTA(field) (nfsstats.field - lastst.field) int main(int argc, char **argv) { u_int interval; int clientOnly = -1; int serverOnly = -1; int ch; char *memf, *nlistf; char errbuf[_POSIX2_LINE_MAX]; interval = 0; memf = nlistf = NULL; - while ((ch = getopt(argc, argv, "cesWM:N:w:z")) != -1) + while ((ch = getopt(argc, argv, "cesWM:N:ow:z")) != -1) switch(ch) { case 'M': memf = optarg; break; case 'N': nlistf = optarg; break; case 'W': widemode = 1; break; case 'w': interval = atoi(optarg); break; case 'c': clientOnly = 1; if (serverOnly < 0) serverOnly = 0; break; case 's': serverOnly = 1; if (clientOnly < 0) clientOnly = 0; break; case 'z': zflag = 1; break; + case 'o': + if (extra_output != 0) + err(1, "-o incompatible with -e"); + run_v4 = 0; + break; case 'e': - run_v4 = 1; + if (run_v4 == 0) + err(1, "-e incompatible with -o"); + extra_output = 1; break; case '?': default: usage(); } argc -= optind; argv += optind; #define BACKWARD_COMPATIBILITY #ifdef BACKWARD_COMPATIBILITY if (*argv) { interval = atoi(*argv); if (*++argv) { nlistf = *argv; if (*++argv) memf = *argv; } } #endif if (run_v4 != 0 && modfind("nfscommon") < 0) - errx(1, "experimental client/server not loaded"); + errx(1, "new client/server not loaded"); if (run_v4 == 0 && (nlistf != NULL || memf != NULL)) { deadkernel = 1; if ((kd = kvm_openfiles(nlistf, memf, NULL, O_RDONLY, errbuf)) == 0) { errx(1, "kvm_openfiles: %s", errbuf); } if (kvm_nlist(kd, nl) != 0) { errx(1, "kvm_nlist: can't get names"); } } if (interval) { if (run_v4 > 0) exp_sidewaysintpr(interval, clientOnly, serverOnly); else sidewaysintpr(interval, clientOnly, serverOnly); } else { - if (run_v4 > 0) + if (extra_output != 0) exp_intpr(clientOnly, serverOnly); else intpr(clientOnly, serverOnly); } exit(0); } /* * Read the nfs stats using sysctl(3) for live kernels, or kvm_read * for dead ones. */ static void readstats(struct nfsstats **stp, struct nfsrvstats **srvstp, int zero) { union { struct nfsstats client; struct nfsrvstats server; } zerostat; size_t buflen; if (deadkernel) { if (*stp != NULL && kvm_read(kd, (u_long)nl[N_NFSSTAT].n_value, *stp, sizeof(struct nfsstats)) < 0) { *stp = NULL; } if (*srvstp != NULL && kvm_read(kd, (u_long)nl[N_NFSRVSTAT].n_value, *srvstp, sizeof(struct nfsrvstats)) < 0) { *srvstp = NULL; } } else { if (zero) bzero(&zerostat, sizeof(zerostat)); buflen = sizeof(struct nfsstats); if (*stp != NULL && sysctlbyname("vfs.nfs.nfsstats", *stp, &buflen, zero ? &zerostat : NULL, zero ? buflen : 0) < 0) { if (errno != ENOENT) err(1, "sysctl: vfs.nfs.nfsstats"); *stp = NULL; } buflen = sizeof(struct nfsrvstats); if (*srvstp != NULL && sysctlbyname("vfs.nfsrv.nfsrvstats", *srvstp, &buflen, zero ? &zerostat : NULL, zero ? buflen : 0) < 0) { if (errno != ENOENT) err(1, "sysctl: vfs.nfsrv.nfsrvstats"); *srvstp = NULL; } } } /* * Print a description of the nfs stats. */ void intpr(int clientOnly, int serverOnly) { struct nfsstats nfsstats, *nfsstatsp; struct nfsrvstats nfsrvstats, *nfsrvstatsp; + int nfssvc_flag; - /* - * Only read the stats we are going to display to avoid zeroing - * stats the user didn't request. - */ - if (clientOnly) - nfsstatsp = &nfsstats; - else - nfsstatsp = NULL; - if (serverOnly) - nfsrvstatsp = &nfsrvstats; - else - nfsrvstatsp = NULL; - - readstats(&nfsstatsp, &nfsrvstatsp, zflag); - - if (clientOnly && !nfsstatsp) { - printf("Client not present!\n"); - clientOnly = 0; + if (run_v4 == 0) { + /* + * Only read the stats we are going to display to avoid zeroing + * stats the user didn't request. + */ + if (clientOnly) + nfsstatsp = &nfsstats; + else + nfsstatsp = NULL; + if (serverOnly) + nfsrvstatsp = &nfsrvstats; + else + nfsrvstatsp = NULL; + + readstats(&nfsstatsp, &nfsrvstatsp, zflag); + + if (clientOnly && !nfsstatsp) { + printf("Client not present!\n"); + clientOnly = 0; + } + } else { + nfssvc_flag = NFSSVC_GETSTATS; + if (zflag != 0) { + if (clientOnly != 0) + nfssvc_flag |= NFSSVC_ZEROCLTSTATS; + if (serverOnly != 0) + nfssvc_flag |= NFSSVC_ZEROSRVSTATS; + } + if (nfssvc(nfssvc_flag, &ext_nfsstats) < 0) + err(1, "Can't get stats"); } if (clientOnly) { printf("Client Info:\n"); printf("Rpc Counts:\n"); printf("%9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s\n", "Getattr", "Setattr", "Lookup", "Readlink", "Read", "Write", "Create", "Remove"); - printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", - nfsstats.rpccnt[NFSPROC_GETATTR], - nfsstats.rpccnt[NFSPROC_SETATTR], - nfsstats.rpccnt[NFSPROC_LOOKUP], - nfsstats.rpccnt[NFSPROC_READLINK], - nfsstats.rpccnt[NFSPROC_READ], - nfsstats.rpccnt[NFSPROC_WRITE], - nfsstats.rpccnt[NFSPROC_CREATE], - nfsstats.rpccnt[NFSPROC_REMOVE]); + if (run_v4 == 0) + printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", + nfsstats.rpccnt[NFSPROC_GETATTR], + nfsstats.rpccnt[NFSPROC_SETATTR], + nfsstats.rpccnt[NFSPROC_LOOKUP], + nfsstats.rpccnt[NFSPROC_READLINK], + nfsstats.rpccnt[NFSPROC_READ], + nfsstats.rpccnt[NFSPROC_WRITE], + nfsstats.rpccnt[NFSPROC_CREATE], + nfsstats.rpccnt[NFSPROC_REMOVE]); + else + printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", + ext_nfsstats.rpccnt[NFSPROC_GETATTR], + ext_nfsstats.rpccnt[NFSPROC_SETATTR], + ext_nfsstats.rpccnt[NFSPROC_LOOKUP], + ext_nfsstats.rpccnt[NFSPROC_READLINK], + ext_nfsstats.rpccnt[NFSPROC_READ], + ext_nfsstats.rpccnt[NFSPROC_WRITE], + ext_nfsstats.rpccnt[NFSPROC_CREATE], + ext_nfsstats.rpccnt[NFSPROC_REMOVE]); printf("%9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s\n", "Rename", "Link", "Symlink", "Mkdir", "Rmdir", "Readdir", "RdirPlus", "Access"); - printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", - nfsstats.rpccnt[NFSPROC_RENAME], - nfsstats.rpccnt[NFSPROC_LINK], - nfsstats.rpccnt[NFSPROC_SYMLINK], - nfsstats.rpccnt[NFSPROC_MKDIR], - nfsstats.rpccnt[NFSPROC_RMDIR], - nfsstats.rpccnt[NFSPROC_READDIR], - nfsstats.rpccnt[NFSPROC_READDIRPLUS], - nfsstats.rpccnt[NFSPROC_ACCESS]); + if (run_v4 == 0) + printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", + nfsstats.rpccnt[NFSPROC_RENAME], + nfsstats.rpccnt[NFSPROC_LINK], + nfsstats.rpccnt[NFSPROC_SYMLINK], + nfsstats.rpccnt[NFSPROC_MKDIR], + nfsstats.rpccnt[NFSPROC_RMDIR], + nfsstats.rpccnt[NFSPROC_READDIR], + nfsstats.rpccnt[NFSPROC_READDIRPLUS], + nfsstats.rpccnt[NFSPROC_ACCESS]); + else + printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", + ext_nfsstats.rpccnt[NFSPROC_RENAME], + ext_nfsstats.rpccnt[NFSPROC_LINK], + ext_nfsstats.rpccnt[NFSPROC_SYMLINK], + ext_nfsstats.rpccnt[NFSPROC_MKDIR], + ext_nfsstats.rpccnt[NFSPROC_RMDIR], + ext_nfsstats.rpccnt[NFSPROC_READDIR], + ext_nfsstats.rpccnt[NFSPROC_READDIRPLUS], + ext_nfsstats.rpccnt[NFSPROC_ACCESS]); printf("%9.9s %9.9s %9.9s %9.9s %9.9s\n", "Mknod", "Fsstat", "Fsinfo", "PathConf", "Commit"); - printf("%9d %9d %9d %9d %9d\n", - nfsstats.rpccnt[NFSPROC_MKNOD], - nfsstats.rpccnt[NFSPROC_FSSTAT], - nfsstats.rpccnt[NFSPROC_FSINFO], - nfsstats.rpccnt[NFSPROC_PATHCONF], - nfsstats.rpccnt[NFSPROC_COMMIT]); + if (run_v4 == 0) + printf("%9d %9d %9d %9d %9d\n", + nfsstats.rpccnt[NFSPROC_MKNOD], + nfsstats.rpccnt[NFSPROC_FSSTAT], + nfsstats.rpccnt[NFSPROC_FSINFO], + nfsstats.rpccnt[NFSPROC_PATHCONF], + nfsstats.rpccnt[NFSPROC_COMMIT]); + else + printf("%9d %9d %9d %9d %9d\n", + ext_nfsstats.rpccnt[NFSPROC_MKNOD], + ext_nfsstats.rpccnt[NFSPROC_FSSTAT], + ext_nfsstats.rpccnt[NFSPROC_FSINFO], + ext_nfsstats.rpccnt[NFSPROC_PATHCONF], + ext_nfsstats.rpccnt[NFSPROC_COMMIT]); printf("Rpc Info:\n"); printf("%9.9s %9.9s %9.9s %9.9s %9.9s\n", "TimedOut", "Invalid", "X Replies", "Retries", "Requests"); - printf("%9d %9d %9d %9d %9d\n", - nfsstats.rpctimeouts, - nfsstats.rpcinvalid, - nfsstats.rpcunexpected, - nfsstats.rpcretries, - nfsstats.rpcrequests); + if (run_v4 == 0) + printf("%9d %9d %9d %9d %9d\n", + nfsstats.rpctimeouts, + nfsstats.rpcinvalid, + nfsstats.rpcunexpected, + nfsstats.rpcretries, + nfsstats.rpcrequests); + else + printf("%9d %9d %9d %9d %9d\n", + ext_nfsstats.rpctimeouts, + ext_nfsstats.rpcinvalid, + ext_nfsstats.rpcunexpected, + ext_nfsstats.rpcretries, + ext_nfsstats.rpcrequests); printf("Cache Info:\n"); printf("%9.9s %9.9s %9.9s %9.9s", "Attr Hits", "Misses", "Lkup Hits", "Misses"); printf(" %9.9s %9.9s %9.9s %9.9s\n", "BioR Hits", "Misses", "BioW Hits", "Misses"); - printf("%9d %9d %9d %9d", - nfsstats.attrcache_hits, nfsstats.attrcache_misses, - nfsstats.lookupcache_hits, nfsstats.lookupcache_misses); - printf(" %9d %9d %9d %9d\n", - nfsstats.biocache_reads-nfsstats.read_bios, - nfsstats.read_bios, - nfsstats.biocache_writes-nfsstats.write_bios, - nfsstats.write_bios); + if (run_v4 == 0) { + printf("%9d %9d %9d %9d", + nfsstats.attrcache_hits, + nfsstats.attrcache_misses, + nfsstats.lookupcache_hits, + nfsstats.lookupcache_misses); + printf(" %9d %9d %9d %9d\n", + nfsstats.biocache_reads-nfsstats.read_bios, + nfsstats.read_bios, + nfsstats.biocache_writes-nfsstats.write_bios, + nfsstats.write_bios); + } else { + printf("%9d %9d %9d %9d", + ext_nfsstats.attrcache_hits, + ext_nfsstats.attrcache_misses, + ext_nfsstats.lookupcache_hits, + ext_nfsstats.lookupcache_misses); + printf(" %9d %9d %9d %9d\n", + ext_nfsstats.biocache_reads - + ext_nfsstats.read_bios, + ext_nfsstats.read_bios, + ext_nfsstats.biocache_writes - + ext_nfsstats.write_bios, + ext_nfsstats.write_bios); + } printf("%9.9s %9.9s %9.9s %9.9s", "BioRLHits", "Misses", "BioD Hits", "Misses"); printf(" %9.9s %9.9s %9.9s %9.9s\n", "DirE Hits", "Misses", "Accs Hits", "Misses"); - printf("%9d %9d %9d %9d", - nfsstats.biocache_readlinks-nfsstats.readlink_bios, - nfsstats.readlink_bios, - nfsstats.biocache_readdirs-nfsstats.readdir_bios, - nfsstats.readdir_bios); - printf(" %9d %9d %9d %9d\n", - nfsstats.direofcache_hits, nfsstats.direofcache_misses, - nfsstats.accesscache_hits, nfsstats.accesscache_misses); + if (run_v4 == 0) { + printf("%9d %9d %9d %9d", + nfsstats.biocache_readlinks - + nfsstats.readlink_bios, + nfsstats.readlink_bios, + nfsstats.biocache_readdirs - + nfsstats.readdir_bios, + nfsstats.readdir_bios); + printf(" %9d %9d %9d %9d\n", + nfsstats.direofcache_hits, + nfsstats.direofcache_misses, + nfsstats.accesscache_hits, + nfsstats.accesscache_misses); + } else { + printf("%9d %9d %9d %9d", + ext_nfsstats.biocache_readlinks - + ext_nfsstats.readlink_bios, + ext_nfsstats.readlink_bios, + ext_nfsstats.biocache_readdirs - + ext_nfsstats.readdir_bios, + ext_nfsstats.readdir_bios); + printf(" %9d %9d %9d %9d\n", + ext_nfsstats.direofcache_hits, + ext_nfsstats.direofcache_misses, + ext_nfsstats.accesscache_hits, + ext_nfsstats.accesscache_misses); + } } - if (serverOnly && !nfsrvstatsp) { + if (run_v4 == 0 && serverOnly && !nfsrvstatsp) { printf("Server not present!\n"); serverOnly = 0; } if (serverOnly) { printf("\nServer Info:\n"); printf("%9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s\n", "Getattr", "Setattr", "Lookup", "Readlink", "Read", "Write", "Create", "Remove"); - printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", - nfsrvstats.srvrpccnt[NFSPROC_GETATTR], - nfsrvstats.srvrpccnt[NFSPROC_SETATTR], - nfsrvstats.srvrpccnt[NFSPROC_LOOKUP], - nfsrvstats.srvrpccnt[NFSPROC_READLINK], - nfsrvstats.srvrpccnt[NFSPROC_READ], - nfsrvstats.srvrpccnt[NFSPROC_WRITE], - nfsrvstats.srvrpccnt[NFSPROC_CREATE], - nfsrvstats.srvrpccnt[NFSPROC_REMOVE]); + if (run_v4 == 0) + printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", + nfsrvstats.srvrpccnt[NFSPROC_GETATTR], + nfsrvstats.srvrpccnt[NFSPROC_SETATTR], + nfsrvstats.srvrpccnt[NFSPROC_LOOKUP], + nfsrvstats.srvrpccnt[NFSPROC_READLINK], + nfsrvstats.srvrpccnt[NFSPROC_READ], + nfsrvstats.srvrpccnt[NFSPROC_WRITE], + nfsrvstats.srvrpccnt[NFSPROC_CREATE], + nfsrvstats.srvrpccnt[NFSPROC_REMOVE]); + else + printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", + ext_nfsstats.srvrpccnt[NFSPROC_GETATTR], + ext_nfsstats.srvrpccnt[NFSPROC_SETATTR], + ext_nfsstats.srvrpccnt[NFSPROC_LOOKUP], + ext_nfsstats.srvrpccnt[NFSPROC_READLINK], + ext_nfsstats.srvrpccnt[NFSPROC_READ], + ext_nfsstats.srvrpccnt[NFSPROC_WRITE], + ext_nfsstats.srvrpccnt[NFSPROC_CREATE], + ext_nfsstats.srvrpccnt[NFSPROC_REMOVE]); printf("%9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s\n", "Rename", "Link", "Symlink", "Mkdir", "Rmdir", "Readdir", "RdirPlus", "Access"); - printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", - nfsrvstats.srvrpccnt[NFSPROC_RENAME], - nfsrvstats.srvrpccnt[NFSPROC_LINK], - nfsrvstats.srvrpccnt[NFSPROC_SYMLINK], - nfsrvstats.srvrpccnt[NFSPROC_MKDIR], - nfsrvstats.srvrpccnt[NFSPROC_RMDIR], - nfsrvstats.srvrpccnt[NFSPROC_READDIR], - nfsrvstats.srvrpccnt[NFSPROC_READDIRPLUS], - nfsrvstats.srvrpccnt[NFSPROC_ACCESS]); + if (run_v4 == 0) + printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", + nfsrvstats.srvrpccnt[NFSPROC_RENAME], + nfsrvstats.srvrpccnt[NFSPROC_LINK], + nfsrvstats.srvrpccnt[NFSPROC_SYMLINK], + nfsrvstats.srvrpccnt[NFSPROC_MKDIR], + nfsrvstats.srvrpccnt[NFSPROC_RMDIR], + nfsrvstats.srvrpccnt[NFSPROC_READDIR], + nfsrvstats.srvrpccnt[NFSPROC_READDIRPLUS], + nfsrvstats.srvrpccnt[NFSPROC_ACCESS]); + else + printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", + ext_nfsstats.srvrpccnt[NFSPROC_RENAME], + ext_nfsstats.srvrpccnt[NFSPROC_LINK], + ext_nfsstats.srvrpccnt[NFSPROC_SYMLINK], + ext_nfsstats.srvrpccnt[NFSPROC_MKDIR], + ext_nfsstats.srvrpccnt[NFSPROC_RMDIR], + ext_nfsstats.srvrpccnt[NFSPROC_READDIR], + ext_nfsstats.srvrpccnt[NFSPROC_READDIRPLUS], + ext_nfsstats.srvrpccnt[NFSPROC_ACCESS]); printf("%9.9s %9.9s %9.9s %9.9s %9.9s\n", "Mknod", "Fsstat", "Fsinfo", "PathConf", "Commit"); - printf("%9d %9d %9d %9d %9d\n", - nfsrvstats.srvrpccnt[NFSPROC_MKNOD], - nfsrvstats.srvrpccnt[NFSPROC_FSSTAT], - nfsrvstats.srvrpccnt[NFSPROC_FSINFO], - nfsrvstats.srvrpccnt[NFSPROC_PATHCONF], - nfsrvstats.srvrpccnt[NFSPROC_COMMIT]); + if (run_v4 == 0) + printf("%9d %9d %9d %9d %9d\n", + nfsrvstats.srvrpccnt[NFSPROC_MKNOD], + nfsrvstats.srvrpccnt[NFSPROC_FSSTAT], + nfsrvstats.srvrpccnt[NFSPROC_FSINFO], + nfsrvstats.srvrpccnt[NFSPROC_PATHCONF], + nfsrvstats.srvrpccnt[NFSPROC_COMMIT]); + else + printf("%9d %9d %9d %9d %9d\n", + ext_nfsstats.srvrpccnt[NFSPROC_MKNOD], + ext_nfsstats.srvrpccnt[NFSPROC_FSSTAT], + ext_nfsstats.srvrpccnt[NFSPROC_FSINFO], + ext_nfsstats.srvrpccnt[NFSPROC_PATHCONF], + ext_nfsstats.srvrpccnt[NFSPROC_COMMIT]); printf("Server Ret-Failed\n"); - printf("%17d\n", nfsrvstats.srvrpc_errs); + if (run_v4 == 0) + printf("%17d\n", nfsrvstats.srvrpc_errs); + else + printf("%17d\n", ext_nfsstats.srvrpc_errs); printf("Server Faults\n"); - printf("%13d\n", nfsrvstats.srv_errs); + if (run_v4 == 0) + printf("%13d\n", nfsrvstats.srv_errs); + else + printf("%13d\n", ext_nfsstats.srv_errs); printf("Server Cache Stats:\n"); printf("%9.9s %9.9s %9.9s %9.9s\n", "Inprog", "Idem", "Non-idem", "Misses"); - printf("%9d %9d %9d %9d\n", - nfsrvstats.srvcache_inproghits, - nfsrvstats.srvcache_idemdonehits, - nfsrvstats.srvcache_nonidemdonehits, - nfsrvstats.srvcache_misses); + if (run_v4 == 0) + printf("%9d %9d %9d %9d\n", + nfsrvstats.srvcache_inproghits, + nfsrvstats.srvcache_idemdonehits, + nfsrvstats.srvcache_nonidemdonehits, + nfsrvstats.srvcache_misses); + else + printf("%9d %9d %9d %9d\n", + ext_nfsstats.srvcache_inproghits, + ext_nfsstats.srvcache_idemdonehits, + ext_nfsstats.srvcache_nonidemdonehits, + ext_nfsstats.srvcache_misses); printf("Server Write Gathering:\n"); printf("%9.9s %9.9s %9.9s\n", "WriteOps", "WriteRPC", "Opsaved"); - printf("%9d %9d %9d\n", - nfsrvstats.srvvop_writes, - nfsrvstats.srvrpccnt[NFSPROC_WRITE], - nfsrvstats.srvrpccnt[NFSPROC_WRITE] - - nfsrvstats.srvvop_writes); + if (run_v4 == 0) + printf("%9d %9d %9d\n", + nfsrvstats.srvvop_writes, + nfsrvstats.srvrpccnt[NFSPROC_WRITE], + nfsrvstats.srvrpccnt[NFSPROC_WRITE] - + nfsrvstats.srvvop_writes); + else + /* + * The new client doesn't do write gathering. It was + * only useful for NFSv2. + */ + printf("%9d %9d %9d\n", + ext_nfsstats.srvrpccnt[NFSPROC_WRITE], + ext_nfsstats.srvrpccnt[NFSPROC_WRITE], 0); } } u_char signalled; /* set if alarm goes off "early" */ /* * Print a running summary of nfs statistics. * Repeat display every interval seconds, showing statistics * collected over that interval. Assumes that interval is non-zero. * First line printed at top of screen is always cumulative. */ void sidewaysintpr(u_int interval, int clientOnly, int serverOnly) { struct nfsstats nfsstats, lastst, *nfsstatsp; struct nfsrvstats nfsrvstats, lastsrvst, *nfsrvstatsp; int hdrcnt = 1; nfsstatsp = &lastst; nfsrvstatsp = &lastsrvst; readstats(&nfsstatsp, &nfsrvstatsp, 0); if (clientOnly && !nfsstatsp) { printf("Client not present!\n"); clientOnly = 0; } if (serverOnly && !nfsrvstatsp) { printf("Server not present!\n"); serverOnly = 0; } sleep(interval); for (;;) { nfsstatsp = &nfsstats; nfsrvstatsp = &nfsrvstats; readstats(&nfsstatsp, &nfsrvstatsp, 0); if (--hdrcnt == 0) { printhdr(clientOnly, serverOnly); if (clientOnly && serverOnly) hdrcnt = 10; else hdrcnt = 20; } if (clientOnly) { printf("%s %6d %6d %6d %6d %6d %6d %6d %6d", ((clientOnly && serverOnly) ? "Client:" : ""), DELTA(attrcache_hits) + DELTA(attrcache_misses), DELTA(lookupcache_hits) + DELTA(lookupcache_misses), DELTA(biocache_readlinks), DELTA(biocache_reads), DELTA(biocache_writes), nfsstats.rpccnt[NFSPROC_RENAME]-lastst.rpccnt[NFSPROC_RENAME], DELTA(accesscache_hits) + DELTA(accesscache_misses), DELTA(biocache_readdirs) ); if (widemode) { printf(" %s %s %s %s %s %s", sperc1(DELTA(attrcache_hits), DELTA(attrcache_misses)), sperc1(DELTA(lookupcache_hits), DELTA(lookupcache_misses)), sperc2(DELTA(biocache_reads), DELTA(read_bios)), sperc2(DELTA(biocache_writes), DELTA(write_bios)), sperc1(DELTA(accesscache_hits), DELTA(accesscache_misses)), sperc2(DELTA(biocache_readdirs), DELTA(readdir_bios)) ); } printf("\n"); lastst = nfsstats; } if (serverOnly) { printf("%s %6d %6d %6d %6d %6d %6d %6d %6d", ((clientOnly && serverOnly) ? "Server:" : ""), nfsrvstats.srvrpccnt[NFSPROC_GETATTR]-lastsrvst.srvrpccnt[NFSPROC_GETATTR], nfsrvstats.srvrpccnt[NFSPROC_LOOKUP]-lastsrvst.srvrpccnt[NFSPROC_LOOKUP], nfsrvstats.srvrpccnt[NFSPROC_READLINK]-lastsrvst.srvrpccnt[NFSPROC_READLINK], nfsrvstats.srvrpccnt[NFSPROC_READ]-lastsrvst.srvrpccnt[NFSPROC_READ], nfsrvstats.srvrpccnt[NFSPROC_WRITE]-lastsrvst.srvrpccnt[NFSPROC_WRITE], nfsrvstats.srvrpccnt[NFSPROC_RENAME]-lastsrvst.srvrpccnt[NFSPROC_RENAME], nfsrvstats.srvrpccnt[NFSPROC_ACCESS]-lastsrvst.srvrpccnt[NFSPROC_ACCESS], (nfsrvstats.srvrpccnt[NFSPROC_READDIR]-lastsrvst.srvrpccnt[NFSPROC_READDIR]) +(nfsrvstats.srvrpccnt[NFSPROC_READDIRPLUS]-lastsrvst.srvrpccnt[NFSPROC_READDIRPLUS])); printf("\n"); lastsrvst = nfsrvstats; } fflush(stdout); sleep(interval); } /*NOTREACHED*/ } void printhdr(int clientOnly, int serverOnly) { printf("%s%6.6s %6.6s %6.6s %6.6s %6.6s %6.6s %6.6s %6.6s", ((serverOnly && clientOnly) ? " " : " "), "GtAttr", "Lookup", "Rdlink", "Read", "Write", "Rename", "Access", "Rddir"); if (widemode && clientOnly) { printf(" Attr Lkup BioR BioW Accs BioD"); } printf("\n"); fflush(stdout); } void usage(void) { (void)fprintf(stderr, - "usage: nfsstat [-ceszW] [-M core] [-N system] [-w interval]\n"); + "usage: nfsstat [-ceoszW] [-M core] [-N system] [-w wait]\n"); exit(1); } static char SPBuf[64][8]; static int SPIndex; char * sperc1(int hits, int misses) { char *p = SPBuf[SPIndex]; if (hits + misses) { sprintf(p, "%3d%%", (int)(char)((quad_t)hits * 100 / (hits + misses))); } else { sprintf(p, " -"); } SPIndex = (SPIndex + 1) & 63; return(p); } char * sperc2(int ttl, int misses) { char *p = SPBuf[SPIndex]; if (ttl) { sprintf(p, "%3d%%", (int)(char)((quad_t)(ttl - misses) * 100 / ttl)); } else { sprintf(p, " -"); } SPIndex = (SPIndex + 1) & 63; return(p); } /* * Print a description of the nfs stats for the experimental client/server. */ void exp_intpr(int clientOnly, int serverOnly) { int nfssvc_flag; nfssvc_flag = NFSSVC_GETSTATS; if (zflag != 0) { if (clientOnly != 0) nfssvc_flag |= NFSSVC_ZEROCLTSTATS; if (serverOnly != 0) nfssvc_flag |= NFSSVC_ZEROSRVSTATS; } if (nfssvc(nfssvc_flag, &ext_nfsstats) < 0) err(1, "Can't get stats"); if (clientOnly != 0) { if (printtitle) { printf("Client Info:\n"); printf("Rpc Counts:\n"); printf( "%9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s\n" , "Getattr", "Setattr", "Lookup", "Readlink", "Read", "Write", "Create", "Remove"); } printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", ext_nfsstats.rpccnt[NFSPROC_GETATTR], ext_nfsstats.rpccnt[NFSPROC_SETATTR], ext_nfsstats.rpccnt[NFSPROC_LOOKUP], ext_nfsstats.rpccnt[NFSPROC_READLINK], ext_nfsstats.rpccnt[NFSPROC_READ], ext_nfsstats.rpccnt[NFSPROC_WRITE], ext_nfsstats.rpccnt[NFSPROC_CREATE], ext_nfsstats.rpccnt[NFSPROC_REMOVE]); if (printtitle) printf( "%9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s\n" , "Rename", "Link", "Symlink", "Mkdir", "Rmdir", "Readdir", "RdirPlus", "Access"); printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", ext_nfsstats.rpccnt[NFSPROC_RENAME], ext_nfsstats.rpccnt[NFSPROC_LINK], ext_nfsstats.rpccnt[NFSPROC_SYMLINK], ext_nfsstats.rpccnt[NFSPROC_MKDIR], ext_nfsstats.rpccnt[NFSPROC_RMDIR], ext_nfsstats.rpccnt[NFSPROC_READDIR], ext_nfsstats.rpccnt[NFSPROC_READDIRPLUS], ext_nfsstats.rpccnt[NFSPROC_ACCESS]); if (printtitle) printf( "%9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s\n" , "Mknod", "Fsstat", "Fsinfo", "PathConf", "Commit", "SetClId", "SetClIdCf", "Lock"); printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", ext_nfsstats.rpccnt[NFSPROC_MKNOD], ext_nfsstats.rpccnt[NFSPROC_FSSTAT], ext_nfsstats.rpccnt[NFSPROC_FSINFO], ext_nfsstats.rpccnt[NFSPROC_PATHCONF], ext_nfsstats.rpccnt[NFSPROC_COMMIT], ext_nfsstats.rpccnt[NFSPROC_SETCLIENTID], ext_nfsstats.rpccnt[NFSPROC_SETCLIENTIDCFRM], ext_nfsstats.rpccnt[NFSPROC_LOCK]); if (printtitle) printf("%9.9s %9.9s %9.9s %9.9s\n", "LockT", "LockU", "Open", "OpenCfr"); printf("%9d %9d %9d %9d\n", ext_nfsstats.rpccnt[NFSPROC_LOCKT], ext_nfsstats.rpccnt[NFSPROC_LOCKU], ext_nfsstats.rpccnt[NFSPROC_OPEN], ext_nfsstats.rpccnt[NFSPROC_OPENCONFIRM]); if (printtitle) printf( "%9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s\n" , "OpenOwner", "Opens", "LockOwner", "Locks", "Delegs", "LocalOwn", "LocalOpen", "LocalLOwn"); printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", ext_nfsstats.clopenowners, ext_nfsstats.clopens, ext_nfsstats.cllockowners, ext_nfsstats.cllocks, ext_nfsstats.cldelegates, ext_nfsstats.cllocalopenowners, ext_nfsstats.cllocalopens, ext_nfsstats.cllocallockowners); if (printtitle) printf("%9.9s\n", "LocalLock"); printf("%9d\n", ext_nfsstats.cllocallocks); if (printtitle) { printf("Rpc Info:\n"); printf("%9.9s %9.9s %9.9s %9.9s %9.9s\n", "TimedOut", "Invalid", "X Replies", "Retries", "Requests"); } printf("%9d %9d %9d %9d %9d\n", ext_nfsstats.rpctimeouts, ext_nfsstats.rpcinvalid, ext_nfsstats.rpcunexpected, ext_nfsstats.rpcretries, ext_nfsstats.rpcrequests); if (printtitle) { printf("Cache Info:\n"); printf("%9.9s %9.9s %9.9s %9.9s", "Attr Hits", "Misses", "Lkup Hits", "Misses"); printf(" %9.9s %9.9s %9.9s %9.9s\n", "BioR Hits", "Misses", "BioW Hits", "Misses"); } printf("%9d %9d %9d %9d", ext_nfsstats.attrcache_hits, ext_nfsstats.attrcache_misses, ext_nfsstats.lookupcache_hits, ext_nfsstats.lookupcache_misses); printf(" %9d %9d %9d %9d\n", - ext_nfsstats.biocache_reads, + ext_nfsstats.biocache_reads - ext_nfsstats.read_bios, ext_nfsstats.read_bios, - ext_nfsstats.biocache_writes, + ext_nfsstats.biocache_writes - ext_nfsstats.write_bios, ext_nfsstats.write_bios); if (printtitle) { printf("%9.9s %9.9s %9.9s %9.9s", "BioRLHits", "Misses", "BioD Hits", "Misses"); printf(" %9.9s %9.9s\n", "DirE Hits", "Misses"); } printf("%9d %9d %9d %9d", - ext_nfsstats.biocache_readlinks, + ext_nfsstats.biocache_readlinks - ext_nfsstats.readlink_bios, - ext_nfsstats.biocache_readdirs, + ext_nfsstats.readlink_bios, + ext_nfsstats.biocache_readdirs - + ext_nfsstats.readdir_bios, ext_nfsstats.readdir_bios); printf(" %9d %9d\n", ext_nfsstats.direofcache_hits, ext_nfsstats.direofcache_misses); } if (serverOnly != 0) { if (printtitle) { printf("\nServer Info:\n"); printf( "%9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s\n" , "Getattr", "Setattr", "Lookup", "Readlink", "Read", "Write", "Create", "Remove"); } printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", ext_nfsstats.srvrpccnt[NFSV4OP_GETATTR], ext_nfsstats.srvrpccnt[NFSV4OP_SETATTR], ext_nfsstats.srvrpccnt[NFSV4OP_LOOKUP], ext_nfsstats.srvrpccnt[NFSV4OP_READLINK], ext_nfsstats.srvrpccnt[NFSV4OP_READ], ext_nfsstats.srvrpccnt[NFSV4OP_WRITE], ext_nfsstats.srvrpccnt[NFSV4OP_V3CREATE], ext_nfsstats.srvrpccnt[NFSV4OP_REMOVE]); if (printtitle) printf( "%9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s\n" , "Rename", "Link", "Symlink", "Mkdir", "Rmdir", "Readdir", "RdirPlus", "Access"); printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", ext_nfsstats.srvrpccnt[NFSV4OP_RENAME], ext_nfsstats.srvrpccnt[NFSV4OP_LINK], ext_nfsstats.srvrpccnt[NFSV4OP_SYMLINK], ext_nfsstats.srvrpccnt[NFSV4OP_MKDIR], ext_nfsstats.srvrpccnt[NFSV4OP_RMDIR], ext_nfsstats.srvrpccnt[NFSV4OP_READDIR], ext_nfsstats.srvrpccnt[NFSV4OP_READDIRPLUS], ext_nfsstats.srvrpccnt[NFSV4OP_ACCESS]); if (printtitle) printf( "%9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s\n" , "Mknod", "Fsstat", "Fsinfo", "PathConf", "Commit", "LookupP", "SetClId", "SetClIdCf"); printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", ext_nfsstats.srvrpccnt[NFSV4OP_MKNOD], ext_nfsstats.srvrpccnt[NFSV4OP_FSSTAT], ext_nfsstats.srvrpccnt[NFSV4OP_FSINFO], ext_nfsstats.srvrpccnt[NFSV4OP_PATHCONF], ext_nfsstats.srvrpccnt[NFSV4OP_COMMIT], ext_nfsstats.srvrpccnt[NFSV4OP_LOOKUPP], ext_nfsstats.srvrpccnt[NFSV4OP_SETCLIENTID], ext_nfsstats.srvrpccnt[NFSV4OP_SETCLIENTIDCFRM]); if (printtitle) printf( "%9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s\n" , "Open", "OpenAttr", "OpenDwnGr", "OpenCfrm", "DelePurge", "DeleRet", "GetFH", "Lock"); printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", ext_nfsstats.srvrpccnt[NFSV4OP_OPEN], ext_nfsstats.srvrpccnt[NFSV4OP_OPENATTR], ext_nfsstats.srvrpccnt[NFSV4OP_OPENDOWNGRADE], ext_nfsstats.srvrpccnt[NFSV4OP_OPENCONFIRM], ext_nfsstats.srvrpccnt[NFSV4OP_DELEGPURGE], ext_nfsstats.srvrpccnt[NFSV4OP_DELEGRETURN], ext_nfsstats.srvrpccnt[NFSV4OP_GETFH], ext_nfsstats.srvrpccnt[NFSV4OP_LOCK]); if (printtitle) printf( "%9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s %9.9s\n" , "LockT", "LockU", "Close", "Verify", "NVerify", "PutFH", "PutPubFH", "PutRootFH"); printf("%9d %9d %9d %9d %9d %9d %9d %9d\n", ext_nfsstats.srvrpccnt[NFSV4OP_LOCKT], ext_nfsstats.srvrpccnt[NFSV4OP_LOCKU], ext_nfsstats.srvrpccnt[NFSV4OP_CLOSE], ext_nfsstats.srvrpccnt[NFSV4OP_VERIFY], ext_nfsstats.srvrpccnt[NFSV4OP_NVERIFY], ext_nfsstats.srvrpccnt[NFSV4OP_PUTFH], ext_nfsstats.srvrpccnt[NFSV4OP_PUTPUBFH], ext_nfsstats.srvrpccnt[NFSV4OP_PUTROOTFH]); if (printtitle) printf("%9.9s %9.9s %9.9s %9.9s %9.9s %9.9s\n", "Renew", "RestoreFH", "SaveFH", "Secinfo", "RelLckOwn", "V4Create"); printf("%9d %9d %9d %9d %9d %9d\n", ext_nfsstats.srvrpccnt[NFSV4OP_RENEW], ext_nfsstats.srvrpccnt[NFSV4OP_RESTOREFH], ext_nfsstats.srvrpccnt[NFSV4OP_SAVEFH], ext_nfsstats.srvrpccnt[NFSV4OP_SECINFO], ext_nfsstats.srvrpccnt[NFSV4OP_RELEASELCKOWN], ext_nfsstats.srvrpccnt[NFSV4OP_CREATE]); if (printtitle) { printf("Server:\n"); printf("%9.9s %9.9s %9.9s\n", "Retfailed", "Faults", "Clients"); } printf("%9d %9d %9d\n", ext_nfsstats.srv_errs, ext_nfsstats.srvrpc_errs, ext_nfsstats.srvclients); if (printtitle) printf("%9.9s %9.9s %9.9s %9.9s %9.9s \n", "OpenOwner", "Opens", "LockOwner", "Locks", "Delegs"); printf("%9d %9d %9d %9d %9d \n", ext_nfsstats.srvopenowners, ext_nfsstats.srvopens, ext_nfsstats.srvlockowners, ext_nfsstats.srvlocks, ext_nfsstats.srvdelegates); if (printtitle) { printf("Server Cache Stats:\n"); printf("%9.9s %9.9s %9.9s %9.9s %9.9s %9.9s\n", "Inprog", "Idem", "Non-idem", "Misses", "CacheSize", "TCPPeak"); } printf("%9d %9d %9d %9d %9d %9d\n", ext_nfsstats.srvcache_inproghits, ext_nfsstats.srvcache_idemdonehits, ext_nfsstats.srvcache_nonidemdonehits, ext_nfsstats.srvcache_misses, ext_nfsstats.srvcache_size, ext_nfsstats.srvcache_tcppeak); } } /* * Print a running summary of nfs statistics for the experimental client and/or * server. * Repeat display every interval seconds, showing statistics * collected over that interval. Assumes that interval is non-zero. * First line printed at top of screen is always cumulative. */ void exp_sidewaysintpr(u_int interval, int clientOnly, int serverOnly) { struct ext_nfsstats nfsstats, lastst, *ext_nfsstatsp; int hdrcnt = 1; ext_nfsstatsp = &lastst; if (nfssvc(NFSSVC_GETSTATS, ext_nfsstatsp) < 0) err(1, "Can't get stats"); sleep(interval); for (;;) { ext_nfsstatsp = &nfsstats; if (nfssvc(NFSSVC_GETSTATS, ext_nfsstatsp) < 0) err(1, "Can't get stats"); if (--hdrcnt == 0) { printhdr(clientOnly, serverOnly); if (clientOnly && serverOnly) hdrcnt = 10; else hdrcnt = 20; } if (clientOnly) { printf("%s %6d %6d %6d %6d %6d %6d %6d %6d", ((clientOnly && serverOnly) ? "Client:" : ""), DELTA(attrcache_hits) + DELTA(attrcache_misses), DELTA(lookupcache_hits) + DELTA(lookupcache_misses), DELTA(biocache_readlinks), DELTA(biocache_reads), DELTA(biocache_writes), nfsstats.rpccnt[NFSPROC_RENAME] - lastst.rpccnt[NFSPROC_RENAME], DELTA(accesscache_hits) + DELTA(accesscache_misses), DELTA(biocache_readdirs) ); if (widemode) { printf(" %s %s %s %s %s %s", sperc1(DELTA(attrcache_hits), DELTA(attrcache_misses)), sperc1(DELTA(lookupcache_hits), DELTA(lookupcache_misses)), sperc2(DELTA(biocache_reads), DELTA(read_bios)), sperc2(DELTA(biocache_writes), DELTA(write_bios)), sperc1(DELTA(accesscache_hits), DELTA(accesscache_misses)), sperc2(DELTA(biocache_readdirs), DELTA(readdir_bios)) ); } printf("\n"); lastst = nfsstats; } if (serverOnly) { printf("%s %6d %6d %6d %6d %6d %6d %6d %6d", ((clientOnly && serverOnly) ? "Server:" : ""), nfsstats.srvrpccnt[NFSPROC_GETATTR] - lastst.srvrpccnt[NFSPROC_GETATTR], nfsstats.srvrpccnt[NFSPROC_LOOKUP] - lastst.srvrpccnt[NFSPROC_LOOKUP], nfsstats.srvrpccnt[NFSPROC_READLINK] - lastst.srvrpccnt[NFSPROC_READLINK], nfsstats.srvrpccnt[NFSPROC_READ] - lastst.srvrpccnt[NFSPROC_READ], nfsstats.srvrpccnt[NFSPROC_WRITE] - lastst.srvrpccnt[NFSPROC_WRITE], nfsstats.srvrpccnt[NFSPROC_RENAME] - lastst.srvrpccnt[NFSPROC_RENAME], nfsstats.srvrpccnt[NFSPROC_ACCESS] - lastst.srvrpccnt[NFSPROC_ACCESS], (nfsstats.srvrpccnt[NFSPROC_READDIR] - lastst.srvrpccnt[NFSPROC_READDIR]) + (nfsstats.srvrpccnt[NFSPROC_READDIRPLUS] - lastst.srvrpccnt[NFSPROC_READDIRPLUS])); printf("\n"); lastst = nfsstats; } fflush(stdout); sleep(interval); } /*NOTREACHED*/ } Index: projects/largeSMP/usr.bin/procstat =================================================================== --- projects/largeSMP/usr.bin/procstat (revision 221494) +++ projects/largeSMP/usr.bin/procstat (revision 221495) Property changes on: projects/largeSMP/usr.bin/procstat ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/usr.bin/procstat:r221444-221493 Index: projects/largeSMP/usr.sbin/bluetooth/bthidd/hid.c =================================================================== --- projects/largeSMP/usr.sbin/bluetooth/bthidd/hid.c (revision 221494) +++ projects/largeSMP/usr.sbin/bluetooth/bthidd/hid.c (revision 221495) @@ -1,397 +1,408 @@ /* * hid.c */ /*- * Copyright (c) 2006 Maksim Yevmenkin * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $Id: hid.c,v 1.5 2006/09/07 21:06:53 max Exp $ * $FreeBSD$ */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include "bthid_config.h" #include "bthidd.h" #include "kbd.h" #undef min #define min(x, y) (((x) < (y))? (x) : (y)) #undef ASIZE #define ASIZE(a) (sizeof(a)/sizeof(a[0])) /* * Process data from control channel */ int32_t hid_control(bthid_session_p s, uint8_t *data, int32_t len) { assert(s != NULL); assert(data != NULL); assert(len > 0); switch (data[0] >> 4) { case 0: /* Handshake (response to command) */ if (data[0] & 0xf) syslog(LOG_ERR, "Got handshake message with error " \ "response 0x%x from %s", data[0], bt_ntoa(&s->bdaddr, NULL)); break; case 1: /* HID Control */ switch (data[0] & 0xf) { case 0: /* NOP */ break; case 1: /* Hard reset */ case 2: /* Soft reset */ syslog(LOG_WARNING, "Device %s requested %s reset", bt_ntoa(&s->bdaddr, NULL), ((data[0] & 0xf) == 1)? "hard" : "soft"); break; case 3: /* Suspend */ syslog(LOG_NOTICE, "Device %s requested Suspend", bt_ntoa(&s->bdaddr, NULL)); break; case 4: /* Exit suspend */ syslog(LOG_NOTICE, "Device %s requested Exit Suspend", bt_ntoa(&s->bdaddr, NULL)); break; case 5: /* Virtual cable unplug */ syslog(LOG_NOTICE, "Device %s unplugged virtual cable", bt_ntoa(&s->bdaddr, NULL)); session_close(s); break; default: syslog(LOG_WARNING, "Device %s sent unknown " \ "HID_Control message 0x%x", bt_ntoa(&s->bdaddr, NULL), data[0]); break; } break; default: syslog(LOG_WARNING, "Got unexpected message 0x%x on Control " \ "channel from %s", data[0], bt_ntoa(&s->bdaddr, NULL)); break; } return (0); } /* * Process data from the interrupt channel */ int32_t hid_interrupt(bthid_session_p s, uint8_t *data, int32_t len) { hid_device_p hid_device; hid_data_t d; hid_item_t h; int32_t report_id, usage, page, val, mouse_x, mouse_y, mouse_z, mouse_butt, mevents, kevents, i; assert(s != NULL); assert(s->srv != NULL); assert(data != NULL); if (len < 3) { syslog(LOG_ERR, "Got short message (%d bytes) on Interrupt " \ "channel from %s", len, bt_ntoa(&s->bdaddr, NULL)); return (-1); } if (data[0] != 0xa1) { syslog(LOG_ERR, "Got unexpected message 0x%x on " \ "Interrupt channel from %s", data[0], bt_ntoa(&s->bdaddr, NULL)); return (-1); } report_id = data[1]; data ++; len --; hid_device = get_hid_device(&s->bdaddr); assert(hid_device != NULL); mouse_x = mouse_y = mouse_z = mouse_butt = mevents = kevents = 0; for (d = hid_start_parse(hid_device->desc, 1 << hid_input, -1); hid_get_item(d, &h) > 0; ) { if ((h.flags & HIO_CONST) || (h.report_ID != report_id) || (h.kind != hid_input)) continue; page = HID_PAGE(h.usage); usage = HID_USAGE(h.usage); val = hid_get_data(data, &h); switch (page) { case HUP_GENERIC_DESKTOP: switch (usage) { case HUG_X: mouse_x = val; mevents ++; break; case HUG_Y: mouse_y = val; mevents ++; break; case HUG_WHEEL: mouse_z = -val; mevents ++; break; case HUG_SYSTEM_SLEEP: if (val) syslog(LOG_NOTICE, "Sleep button pressed"); break; } break; case HUP_KEYBOARD: kevents ++; if (h.flags & HIO_VARIABLE) { if (val && usage < kbd_maxkey()) bit_set(s->keys1, usage); } else { if (val && val < kbd_maxkey()) bit_set(s->keys1, val); for (i = 1; i < h.report_count; i++) { h.pos += h.report_size; val = hid_get_data(data, &h); if (val && val < kbd_maxkey()) bit_set(s->keys1, val); } } break; case HUP_BUTTON: if (usage != 0) { if (usage == 2) usage = 3; else if (usage == 3) usage = 2; mouse_butt |= (val << (usage - 1)); mevents ++; } break; case HUP_CONSUMER: if (!val) break; switch (usage) { + case HUC_AC_PAN: + /* Horizontal scroll */ + if (val < 0) + mouse_butt |= (1 << 5); + else + mouse_butt |= (1 << 6); + + mevents ++; + val = 0; + break; + case 0xb5: /* Scan Next Track */ val = 0x19; break; case 0xb6: /* Scan Previous Track */ val = 0x10; break; case 0xb7: /* Stop */ val = 0x24; break; case 0xcd: /* Play/Pause */ val = 0x22; break; case 0xe2: /* Mute */ val = 0x20; break; case 0xe9: /* Volume Up */ val = 0x30; break; case 0xea: /* Volume Down */ val = 0x2E; break; case 0x183: /* Media Select */ val = 0x6D; break; case 0x018a: /* Mail */ val = 0x6C; break; case 0x192: /* Calculator */ val = 0x21; break; case 0x194: /* My Computer */ val = 0x6B; break; case 0x221: /* WWW Search */ val = 0x65; break; case 0x223: /* WWW Home */ val = 0x32; break; case 0x224: /* WWW Back */ val = 0x6A; break; case 0x225: /* WWW Forward */ val = 0x69; break; case 0x226: /* WWW Stop */ val = 0x68; break; case 0x227: /* WWW Refresh */ val = 0x67; break; case 0x22a: /* WWW Favorites */ val = 0x66; break; default: val = 0; break; } /* XXX FIXME - UGLY HACK */ if (val != 0) { if (hid_device->keyboard) { int32_t buf[4] = { 0xe0, val, 0xe0, val|0x80 }; assert(s->vkbd != -1); write(s->vkbd, buf, sizeof(buf)); } else syslog(LOG_ERR, "Keyboard events " \ "received from non-keyboard " \ "device %s. Please report", bt_ntoa(&s->bdaddr, NULL)); } break; case HUP_MICROSOFT: switch (usage) { case 0xfe01: if (!hid_device->battery_power) break; switch (val) { case 1: syslog(LOG_INFO, "Battery is OK on %s", bt_ntoa(&s->bdaddr, NULL)); break; case 2: syslog(LOG_NOTICE, "Low battery on %s", bt_ntoa(&s->bdaddr, NULL)); break; case 3: syslog(LOG_WARNING, "Very low battery "\ "on %s", bt_ntoa(&s->bdaddr, NULL)); break; } break; } break; } } hid_end_parse(d); /* * XXX FIXME Feed keyboard events into kernel. * The code below works, bit host also needs to track * and handle repeat. * * Key repeat currently works in X, but not in console. */ if (kevents > 0) { if (hid_device->keyboard) { assert(s->vkbd != -1); kbd_process_keys(s); } else syslog(LOG_ERR, "Keyboard events received from " \ "non-keyboard device %s. Please report", bt_ntoa(&s->bdaddr, NULL)); } /* * XXX FIXME Feed mouse events into kernel. * The code block below works, but it is not good enough. * Need to track double-clicks etc. * * Double click currently works in X, but not in console. */ if (mevents > 0) { struct mouse_info mi; mi.operation = MOUSE_ACTION; mi.u.data.x = mouse_x; mi.u.data.y = mouse_y; mi.u.data.z = mouse_z; mi.u.data.buttons = mouse_butt; if (ioctl(s->srv->cons, CONS_MOUSECTL, &mi) < 0) syslog(LOG_ERR, "Could not process mouse events from " \ "%s. %s (%d)", bt_ntoa(&s->bdaddr, NULL), strerror(errno), errno); } return (0); } Index: projects/largeSMP/usr.sbin/makefs/cd9660/cd9660_eltorito.c =================================================================== --- projects/largeSMP/usr.sbin/makefs/cd9660/cd9660_eltorito.c (revision 221494) +++ projects/largeSMP/usr.sbin/makefs/cd9660/cd9660_eltorito.c (revision 221495) @@ -1,618 +1,637 @@ /* $NetBSD: cd9660_eltorito.c,v 1.14 2010/10/27 18:51:35 christos Exp $ */ /* * Copyright (c) 2005 Daniel Watt, Walter Deignan, Ryan Gabrys, Alan * Perez-Rathke and Ram Vedam. All rights reserved. * * This code was written by Daniel Watt, Walter Deignan, Ryan Gabrys, * Alan Perez-Rathke and Ram Vedam. * * Redistribution and use in source and binary forms, with or * without modification, are permitted provided that the following * conditions are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above * copyright notice, this list of conditions and the following * disclaimer in the documentation and/or other materials provided * with the distribution. * * THIS SOFTWARE IS PROVIDED BY DANIEL WATT, WALTER DEIGNAN, RYAN * GABRYS, ALAN PEREZ-RATHKE AND RAM VEDAM ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE * DISCLAIMED. IN NO EVENT SHALL DANIEL WATT, WALTER DEIGNAN, RYAN * GABRYS, ALAN PEREZ-RATHKE AND RAM VEDAM BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF * USE,DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY * OF SUCH DAMAGE. */ #include #include "cd9660.h" #include "cd9660_eltorito.h" #include __FBSDID("$FreeBSD$"); #ifdef DEBUG #define ELTORITO_DPRINTF(__x) printf __x #else #define ELTORITO_DPRINTF(__x) #endif static struct boot_catalog_entry *cd9660_init_boot_catalog_entry(void); static struct boot_catalog_entry *cd9660_boot_setup_validation_entry(char); static struct boot_catalog_entry *cd9660_boot_setup_default_entry( struct cd9660_boot_image *); static struct boot_catalog_entry *cd9660_boot_setup_section_head(char); static struct boot_catalog_entry *cd9660_boot_setup_validation_entry(char); #if 0 static u_char cd9660_boot_get_system_type(struct cd9660_boot_image *); #endif int cd9660_add_boot_disk(const char *boot_info) { struct stat stbuf; const char *mode_msg; char *temp; char *sysname; char *filename; struct cd9660_boot_image *new_image, *tmp_image; assert(boot_info != NULL); if (*boot_info == '\0') { warnx("Error: Boot disk information must be in the " "format 'system;filename'"); return 0; } /* First decode the boot information */ if ((temp = strdup(boot_info)) == NULL) { warn("%s: strdup", __func__); return 0; } sysname = temp; filename = strchr(sysname, ';'); if (filename == NULL) { warnx("supply boot disk information in the format " "'system;filename'"); free(temp); return 0; } *filename++ = '\0'; if (diskStructure.verbose_level > 0) { printf("Found bootdisk with system %s, and filename %s\n", sysname, filename); } if ((new_image = malloc(sizeof(*new_image))) == NULL) { warn("%s: malloc", __func__); free(temp); return 0; } (void)memset(new_image, 0, sizeof(*new_image)); new_image->loadSegment = 0; /* default for now */ /* Decode System */ if (strcmp(sysname, "i386") == 0) new_image->system = ET_SYS_X86; else if (strcmp(sysname, "powerpc") == 0) new_image->system = ET_SYS_PPC; else if (strcmp(sysname, "macppc") == 0 || strcmp(sysname, "mac68k") == 0) new_image->system = ET_SYS_MAC; else { warnx("boot disk system must be " "i386, powerpc, macppc, or mac68k"); free(temp); free(new_image); return 0; } if ((new_image->filename = strdup(filename)) == NULL) { warn("%s: strdup", __func__); free(temp); free(new_image); return 0; } free(temp); /* Get information about the file */ if (lstat(new_image->filename, &stbuf) == -1) err(EXIT_FAILURE, "%s: lstat(\"%s\")", __func__, new_image->filename); switch (stbuf.st_size) { case 1440 * 1024: new_image->targetMode = ET_MEDIA_144FDD; mode_msg = "Assigned boot image to 1.44 emulation mode"; break; case 1200 * 1024: new_image->targetMode = ET_MEDIA_12FDD; mode_msg = "Assigned boot image to 1.2 emulation mode"; break; case 2880 * 1024: new_image->targetMode = ET_MEDIA_288FDD; mode_msg = "Assigned boot image to 2.88 emulation mode"; break; default: new_image->targetMode = ET_MEDIA_NOEM; mode_msg = "Assigned boot image to no emulation mode"; break; } if (diskStructure.verbose_level > 0) printf("%s\n", mode_msg); new_image->size = stbuf.st_size; new_image->num_sectors = howmany(new_image->size, diskStructure.sectorSize) * howmany(diskStructure.sectorSize, 512); if (diskStructure.verbose_level > 0) { printf("New image has size %d, uses %d 512-byte sectors\n", new_image->size, new_image->num_sectors); } new_image->sector = -1; /* Bootable by default */ new_image->bootable = ET_BOOTABLE; /* Add boot disk */ /* Group images for the same platform together. */ TAILQ_FOREACH(tmp_image, &diskStructure.boot_images, image_list) { if (tmp_image->system != new_image->system) break; } if (tmp_image == NULL) { TAILQ_INSERT_HEAD(&diskStructure.boot_images, new_image, image_list); } else TAILQ_INSERT_BEFORE(tmp_image, new_image, image_list); new_image->serialno = diskStructure.image_serialno++; /* TODO : Need to do anything about the boot image in the tree? */ diskStructure.is_bootable = 1; return 1; } int cd9660_eltorito_add_boot_option(const char *option_string, const char *value) { char *eptr; struct cd9660_boot_image *image; assert(option_string != NULL); /* Find the last image added */ TAILQ_FOREACH(image, &diskStructure.boot_images, image_list) { if (image->serialno + 1 == diskStructure.image_serialno) break; } if (image == NULL) errx(EXIT_FAILURE, "Attempted to add boot option, " "but no boot images have been specified"); if (strcmp(option_string, "no-emul-boot") == 0) { image->targetMode = ET_MEDIA_NOEM; } else if (strcmp(option_string, "no-boot") == 0) { image->bootable = ET_NOT_BOOTABLE; } else if (strcmp(option_string, "hard-disk-boot") == 0) { image->targetMode = ET_MEDIA_HDD; } else if (strcmp(option_string, "boot-load-segment") == 0) { image->loadSegment = strtoul(value, &eptr, 16); if (eptr == value || *eptr != '\0' || errno != ERANGE) { warn("%s: strtoul", __func__); return 0; } } else { return 0; } return 1; } static struct boot_catalog_entry * cd9660_init_boot_catalog_entry(void) { struct boot_catalog_entry *temp; if ((temp = malloc(sizeof(*temp))) == NULL) return NULL; return memset(temp, 0, sizeof(*temp)); } static struct boot_catalog_entry * cd9660_boot_setup_validation_entry(char sys) { struct boot_catalog_entry *entry; boot_catalog_validation_entry *ve; int16_t checksum; unsigned char *csptr; int i; entry = cd9660_init_boot_catalog_entry(); if (entry == NULL) { warnx("Error: memory allocation failed in " "cd9660_boot_setup_validation_entry"); return 0; } ve = &entry->entry_data.VE; ve->header_id[0] = 1; ve->platform_id[0] = sys; ve->key[0] = 0x55; ve->key[1] = 0xAA; /* Calculate checksum */ checksum = 0; cd9660_721(0, ve->checksum); csptr = (unsigned char*)ve; for (i = 0; i < sizeof(*ve); i += 2) { checksum += (int16_t)csptr[i]; checksum += 256 * (int16_t)csptr[i + 1]; } checksum = -checksum; cd9660_721(checksum, ve->checksum); ELTORITO_DPRINTF(("%s: header_id %d, platform_id %d, key[0] %d, key[1] %d, " "checksum %04x\n", __func__, ve->header_id[0], ve->platform_id[0], ve->key[0], ve->key[1], checksum)); return entry; } static struct boot_catalog_entry * cd9660_boot_setup_default_entry(struct cd9660_boot_image *disk) { struct boot_catalog_entry *default_entry; boot_catalog_initial_entry *ie; default_entry = cd9660_init_boot_catalog_entry(); if (default_entry == NULL) return NULL; ie = &default_entry->entry_data.IE; ie->boot_indicator[0] = disk->bootable; ie->media_type[0] = disk->targetMode; cd9660_721(disk->loadSegment, ie->load_segment); ie->system_type[0] = disk->system; cd9660_721(disk->num_sectors, ie->sector_count); cd9660_731(disk->sector, ie->load_rba); ELTORITO_DPRINTF(("%s: boot indicator %d, media type %d, " "load segment %04x, system type %d, sector count %d, " "load rba %d\n", __func__, ie->boot_indicator[0], ie->media_type[0], disk->loadSegment, ie->system_type[0], disk->num_sectors, disk->sector)); return default_entry; } static struct boot_catalog_entry * cd9660_boot_setup_section_head(char platform) { struct boot_catalog_entry *entry; boot_catalog_section_header *sh; entry = cd9660_init_boot_catalog_entry(); if (entry == NULL) return NULL; sh = &entry->entry_data.SH; /* More by default. The last one will manually be set to 0x91 */ sh->header_indicator[0] = ET_SECTION_HEADER_MORE; sh->platform_id[0] = platform; sh->num_section_entries[0] = 0; return entry; } static struct boot_catalog_entry * cd9660_boot_setup_section_entry(struct cd9660_boot_image *disk) { struct boot_catalog_entry *entry; boot_catalog_section_entry *se; if ((entry = cd9660_init_boot_catalog_entry()) == NULL) return NULL; se = &entry->entry_data.SE; se->boot_indicator[0] = ET_BOOTABLE; se->media_type[0] = disk->targetMode; cd9660_721(disk->loadSegment, se->load_segment); cd9660_721(disk->num_sectors, se->sector_count); cd9660_731(disk->sector, se->load_rba); return entry; } #if 0 static u_char cd9660_boot_get_system_type(struct cd9660_boot_image *disk) { /* For hard drive booting, we need to examine the MBR to figure out what the partition type is */ return 0; } #endif /* * Set up the BVD, Boot catalog, and the boot entries, but do no writing */ int cd9660_setup_boot(int first_sector) { int sector; int used_sectors; int num_entries = 0; int catalog_sectors; struct boot_catalog_entry *x86_head, *mac_head, *ppc_head, *valid_entry, *default_entry, *temp, *head, **headp, *next; struct cd9660_boot_image *tmp_disk; headp = NULL; x86_head = mac_head = ppc_head = NULL; /* If there are no boot disks, don't bother building boot information */ if (TAILQ_EMPTY(&diskStructure.boot_images)) return 0; /* Point to catalog: For now assume it consumes one sector */ ELTORITO_DPRINTF(("Boot catalog will go in sector %d\n", first_sector)); diskStructure.boot_catalog_sector = first_sector; cd9660_bothendian_dword(first_sector, diskStructure.boot_descriptor->boot_catalog_pointer); /* Step 1: Generate boot catalog */ /* Step 1a: Validation entry */ valid_entry = cd9660_boot_setup_validation_entry(ET_SYS_X86); if (valid_entry == NULL) return -1; /* * Count how many boot images there are, * and how many sectors they consume. */ num_entries = 1; used_sectors = 0; TAILQ_FOREACH(tmp_disk, &diskStructure.boot_images, image_list) { used_sectors += tmp_disk->num_sectors; /* One default entry per image */ num_entries++; } catalog_sectors = howmany(num_entries * 0x20, diskStructure.sectorSize); used_sectors += catalog_sectors; if (diskStructure.verbose_level > 0) { printf("%s: there will be %i entries consuming %i sectors. " "Catalog is %i sectors\n", __func__, num_entries, used_sectors, catalog_sectors); } /* Populate sector numbers */ sector = first_sector + catalog_sectors; TAILQ_FOREACH(tmp_disk, &diskStructure.boot_images, image_list) { tmp_disk->sector = sector; sector += tmp_disk->num_sectors; } LIST_INSERT_HEAD(&diskStructure.boot_entries, valid_entry, ll_struct); /* Step 1b: Initial/default entry */ /* TODO : PARAM */ tmp_disk = TAILQ_FIRST(&diskStructure.boot_images); default_entry = cd9660_boot_setup_default_entry(tmp_disk); if (default_entry == NULL) { warnx("Error: memory allocation failed in cd9660_setup_boot"); return -1; } LIST_INSERT_AFTER(valid_entry, default_entry, ll_struct); /* Todo: multiple default entries? */ tmp_disk = TAILQ_NEXT(tmp_disk, image_list); temp = default_entry; /* If multiple boot images are given : */ while (tmp_disk != NULL) { /* Step 2: Section header */ switch (tmp_disk->system) { case ET_SYS_X86: headp = &x86_head; break; case ET_SYS_PPC: headp = &ppc_head; break; case ET_SYS_MAC: headp = &mac_head; break; default: warnx("%s: internal error: unknown system type", __func__); return -1; } if (*headp == NULL) { head = cd9660_boot_setup_section_head(tmp_disk->system); if (head == NULL) { warnx("Error: memory allocation failed in " "cd9660_setup_boot"); return -1; } LIST_INSERT_AFTER(default_entry, head, ll_struct); *headp = head; } else head = *headp; head->entry_data.SH.num_section_entries[0]++; /* Step 2a: Section entry and extensions */ temp = cd9660_boot_setup_section_entry(tmp_disk); if (temp == NULL) { warn("%s: cd9660_boot_setup_section_entry", __func__); return -1; } while ((next = LIST_NEXT(head, ll_struct)) != NULL && next->entry_type == ET_ENTRY_SE) head = next; LIST_INSERT_AFTER(head, temp, ll_struct); tmp_disk = TAILQ_NEXT(tmp_disk, image_list); } /* TODO: Remaining boot disks when implemented */ return first_sector + used_sectors; } int cd9660_setup_boot_volume_descriptor(volume_descriptor *bvd) { boot_volume_descriptor *bvdData = (boot_volume_descriptor*)bvd->volumeDescriptorData; bvdData->boot_record_indicator[0] = ISO_VOLUME_DESCRIPTOR_BOOT; memcpy(bvdData->identifier, ISO_VOLUME_DESCRIPTOR_STANDARD_ID, 5); bvdData->version[0] = 1; memcpy(bvdData->boot_system_identifier, ET_ID, 23); memcpy(bvdData->identifier, ISO_VOLUME_DESCRIPTOR_STANDARD_ID, 5); diskStructure.boot_descriptor = (boot_volume_descriptor*) bvd->volumeDescriptorData; return 1; } static int cd9660_write_apm_partition_entry(FILE *fd, int index, int total_partitions, off_t sector_start, off_t nsectors, off_t sector_size, const char *part_name, const char *part_type) { uint32_t apm32; uint16_t apm16; fseeko(fd, (off_t)(index + 1) * sector_size, SEEK_SET); /* Signature */ apm16 = htons(0x504d); fwrite(&apm16, sizeof(apm16), 1, fd); apm16 = 0; fwrite(&apm16, sizeof(apm16), 1, fd); /* Total number of partitions */ apm32 = htonl(total_partitions); fwrite(&apm32, sizeof(apm32), 1, fd); /* Bounds */ apm32 = htonl(sector_start); fwrite(&apm32, sizeof(apm32), 1, fd); apm32 = htonl(nsectors); fwrite(&apm32, sizeof(apm32), 1, fd); fwrite(part_name, strlen(part_name) + 1, 1, fd); fseek(fd, 32 - strlen(part_name) - 1, SEEK_CUR); fwrite(part_type, strlen(part_type) + 1, 1, fd); + if (sector_size > 512) { + /* + * Some old broken software looks at 512-byte boundaries for + * partition table entries instead of sector boundaries. We + * can fit 3 entries into the first 2048-byte block, so use + * that to humor old code. + */ + + int n_512_parts = (sector_size / 512) - 1; + if (n_512_parts > total_partitions) + n_512_parts = total_partitions; + + if (index < n_512_parts) + cd9660_write_apm_partition_entry(fd, index, n_512_parts, + sector_start * (sector_size / 512), + nsectors * (sector_size / 512), 512, part_name, + part_type); + } + return 0; } int cd9660_write_boot(FILE *fd) { struct boot_catalog_entry *e; struct cd9660_boot_image *t; int apm_partitions = 0; /* write boot catalog */ if (fseeko(fd, (off_t)diskStructure.boot_catalog_sector * diskStructure.sectorSize, SEEK_SET) == -1) err(1, "fseeko"); if (diskStructure.verbose_level > 0) { printf("Writing boot catalog to sector %" PRId64 "\n", diskStructure.boot_catalog_sector); } LIST_FOREACH(e, &diskStructure.boot_entries, ll_struct) { if (diskStructure.verbose_level > 0) { printf("Writing catalog entry of type %d\n", e->entry_type); } /* * It doesnt matter which one gets written * since they are the same size */ fwrite(&(e->entry_data.VE), 1, 32, fd); } if (diskStructure.verbose_level > 0) printf("Finished writing boot catalog\n"); /* copy boot images */ TAILQ_FOREACH(t, &diskStructure.boot_images, image_list) { if (diskStructure.verbose_level > 0) { printf("Writing boot image from %s to sectors %d\n", t->filename, t->sector); } cd9660_copy_file(fd, t->sector, t->filename); if (t->system == ET_SYS_MAC) apm_partitions++; } if (apm_partitions > 0) { /* Write DDR and global APM info */ uint32_t apm32; uint16_t apm16; int total_parts; fseek(fd, 0, SEEK_SET); apm16 = htons(0x4552); fwrite(&apm16, sizeof(apm16), 1, fd); apm16 = htons(diskStructure.sectorSize); fwrite(&apm16, sizeof(apm16), 1, fd); apm32 = htonl(diskStructure.totalSectors); fwrite(&apm32, sizeof(apm32), 1, fd); /* Count total needed entries */ total_parts = 2 + apm_partitions; /* Self + ISO9660 */ /* Write self-descriptor */ cd9660_write_apm_partition_entry(fd, 0, total_parts, 1, total_parts, diskStructure.sectorSize, "Apple", "Apple_partition_map"); /* Write ISO9660 descriptor, enclosing the whole disk */ cd9660_write_apm_partition_entry(fd, 1, total_parts, 0, diskStructure.totalSectors, diskStructure.sectorSize, "", "CD_ROM_Mode_1"); /* Write all partition entries */ apm_partitions = 0; TAILQ_FOREACH(t, &diskStructure.boot_images, image_list) { if (t->system != ET_SYS_MAC) continue; cd9660_write_apm_partition_entry(fd, 2 + apm_partitions++, total_parts, t->sector, t->num_sectors, diskStructure.sectorSize, "CD Boot", "Apple_Bootstrap"); } } return 0; } Index: projects/largeSMP/usr.sbin/ndiscvt =================================================================== --- projects/largeSMP/usr.sbin/ndiscvt (revision 221494) +++ projects/largeSMP/usr.sbin/ndiscvt (revision 221495) Property changes on: projects/largeSMP/usr.sbin/ndiscvt ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/usr.sbin/ndiscvt:r221444-221493 Index: projects/largeSMP/usr.sbin/zic =================================================================== --- projects/largeSMP/usr.sbin/zic (revision 221494) +++ projects/largeSMP/usr.sbin/zic (revision 221495) Property changes on: projects/largeSMP/usr.sbin/zic ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/usr.sbin/zic:r221444-221493 Index: projects/largeSMP =================================================================== --- projects/largeSMP (revision 221494) +++ projects/largeSMP (revision 221495) Property changes on: projects/largeSMP ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head:r221444-221493