Index: projects/clang1100-import/Makefile.inc1 =================================================================== --- projects/clang1100-import/Makefile.inc1 (revision 364034) +++ projects/clang1100-import/Makefile.inc1 (revision 364035) @@ -1,3346 +1,3342 @@ # # $FreeBSD$ # # Make command line options: # -DNO_CLEANDIR run ${MAKE} clean, instead of ${MAKE} cleandir # -DNO_CLEAN do not clean at all # -DDB_FROM_SRC use the user/group databases in src/etc instead of # the system database when installing. # -DNO_SHARE do not go into share subdir # -DKERNFAST define NO_KERNEL{CONFIG,CLEAN,OBJ} # -DNO_KERNELCONFIG do not run config in ${MAKE} buildkernel # -DNO_KERNELCLEAN do not run ${MAKE} clean in ${MAKE} buildkernel # -DNO_KERNELOBJ do not run ${MAKE} obj in ${MAKE} buildkernel # -DNO_PORTSUPDATE do not update ports in ${MAKE} update # -DNO_ROOT install without using root privilege # -DNO_DOCUPDATE do not update doc in ${MAKE} update # -DWITHOUT_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 # LOCAL_ITOOLS="list of tools" to add additional tools to the ITOOLS list # LOCAL_LIB_DIRS="list of dirs" to add additional dirs to libraries target # LOCAL_MTREE="list of mtree files" to process to allow local directories # to be created before files are installed # LOCAL_TOOL_DIRS="list of dirs" to add additional dirs to the build-tools # list # LOCAL_XTOOL_DIRS="list of dirs" to add additional dirs to the # cross-tools target # METALOG="path to metadata log" to write permission and ownership # when NO_ROOT is set. (default: ${DESTDIR}/METALOG) # TARGET="machine" to crossbuild world for a different machine type # TARGET_ARCH= may be required when a TARGET supports multiple endians # BUILDENV_SHELL= shell to launch for the buildenv target (def:${SHELL}) # WORLD_FLAGS= additional flags to pass to make(1) during buildworld # KERNEL_FLAGS= additional flags to pass to make(1) during buildkernel # SUBDIR_OVERRIDE="list of dirs" to build rather than everything. # All libraries and includes, and some build tools will still build. # # The intended user-driven targets are: # buildworld - rebuild *everything*, including glue to help do upgrades # installworld- install everything built by "buildworld" # checkworld - run test suite on installed world # doxygen - build API documentation of the kernel # update - convenient way to update your source tree (eg: svn/svnup) # # Standard targets (not defined here) are documented in the makefiles in # /usr/share/mk. These include: # obj depend all install clean cleandepend cleanobj .if !defined(TARGET) || !defined(TARGET_ARCH) .error "Both TARGET and TARGET_ARCH must be defined." .endif .if make(showconfig) || make(test-system-*) _MKSHOWCONFIG= t .endif SRCDIR?= ${.CURDIR} LOCALBASE?= /usr/local .include "share/mk/src.tools.mk" # Cross toolchain changes must be in effect before bsd.compiler.mk # so that gets the right CC, and pass CROSS_TOOLCHAIN to submakes. .if defined(CROSS_TOOLCHAIN) .if exists(${LOCALBASE}/share/toolchains/${CROSS_TOOLCHAIN}.mk) .include "${LOCALBASE}/share/toolchains/${CROSS_TOOLCHAIN}.mk" .elif exists(/usr/share/toolchains/${CROSS_TOOLCHAIN}.mk) .include "/usr/share/toolchains/${CROSS_TOOLCHAIN}.mk" .elif exists(${CROSS_TOOLCHAIN}) .include "${CROSS_TOOLCHAIN}" .else .error CROSS_TOOLCHAIN ${CROSS_TOOLCHAIN} not found .endif CROSSENV+=CROSS_TOOLCHAIN="${CROSS_TOOLCHAIN}" .endif .if defined(CROSS_TOOLCHAIN_PREFIX) CROSS_COMPILER_PREFIX?=${CROSS_TOOLCHAIN_PREFIX} .endif XCOMPILERS= CC CXX CPP .for COMPILER in ${XCOMPILERS} .if defined(CROSS_COMPILER_PREFIX) X${COMPILER}?= ${CROSS_COMPILER_PREFIX}${${COMPILER}} .else X${COMPILER}?= ${${COMPILER}} .endif .endfor # If a full path to an external cross compiler is given, don't build # a cross compiler. .if ${XCC:N${CCACHE_BIN}:M/*} MK_CLANG_BOOTSTRAP= no .endif # Pull in compiler metadata from buildworld/toolchain if possible to avoid # running CC from bsd.compiler.mk. .if make(installworld) || make(install) || make(distributeworld) || \ make(stageworld) .-include "${OBJTOP}/toolchain-metadata.mk" .if !defined(_LOADED_TOOLCHAIN_METADATA) .error A build is required first. You may have the wrong MAKEOBJDIRPREFIX set. .endif .endif # Pull in COMPILER_TYPE and COMPILER_FREEBSD_VERSION early. Pull it from the # tree to be friendlier to foreign OS builds. It's safe to do so unconditionally # here since we will always have the right make, unlike in src/Makefile # Don't include bsd.linker.mk yet until XBINUTILS is handled (after src.opts.mk) _NO_INCLUDE_LINKERMK= t # We also want the X_COMPILER* variables if we are using an external toolchain. _WANT_TOOLCHAIN_CROSS_VARS= t .include "share/mk/bsd.compiler.mk" .undef _NO_INCLUDE_LINKERMK .undef _WANT_TOOLCHAIN_CROSS_VARS # src.opts.mk depends on COMPILER_FEATURES .include "share/mk/src.opts.mk" .if ${TARGET} == ${MACHINE} TARGET_CPUTYPE?=${CPUTYPE} .else TARGET_CPUTYPE?= .endif .if !empty(TARGET_CPUTYPE) _TARGET_CPUTYPE=${TARGET_CPUTYPE} .else _TARGET_CPUTYPE=dummy .endif .if ${TARGET} == "arm" .if ${TARGET_ARCH:Marmv[67]*} != "" && ${TARGET_CPUTYPE:M*soft*} == "" TARGET_ABI= gnueabihf .else TARGET_ABI= gnueabi .endif .endif MACHINE_ABI?= unknown MACHINE_TRIPLE?=${MACHINE_ARCH:S/amd64/x86_64/:C/[hs]f$//:S/mipsn32/mips64/}-${MACHINE_ABI}-freebsd13.0 TARGET_ABI?= unknown TARGET_TRIPLE?= ${TARGET_ARCH:S/amd64/x86_64/:C/[hs]f$//:S/mipsn32/mips64/}-${TARGET_ABI}-freebsd13.0 KNOWN_ARCHES?= aarch64/arm64 \ amd64 \ armv6/arm \ armv7/arm \ i386 \ mips \ mipsel/mips \ mips64el/mips \ mipsn32el/mips \ mips64/mips \ mipsn32/mips \ mipshf/mips \ mipselhf/mips \ mips64elhf/mips \ mips64hf/mips \ powerpc \ powerpc64/powerpc \ powerpcspe/powerpc \ riscv64/riscv \ riscv64sf/riscv .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 all targets are disabled for system llvm then don't expect it to work # for cross-builds. .if !defined(TOOLS_PREFIX) && ${MK_LLVM_TARGET_ALL} == "no" && \ ${MACHINE} != ${TARGET} && ${MACHINE_ARCH} != ${TARGET_ARCH} && \ !make(showconfig) MK_SYSTEM_COMPILER= no MK_SYSTEM_LINKER= no .endif # Handle external binutils. .if defined(CROSS_TOOLCHAIN_PREFIX) CROSS_BINUTILS_PREFIX?=${CROSS_TOOLCHAIN_PREFIX} .endif XBINUTILS= AS AR LD NM OBJCOPY RANLIB SIZE STRINGS .for BINUTIL in ${XBINUTILS} .if defined(CROSS_BINUTILS_PREFIX) && \ exists(${CROSS_BINUTILS_PREFIX}/${${BINUTIL}}) X${BINUTIL}?= ${CROSS_BINUTILS_PREFIX:C,/*$,,}/${${BINUTIL}} .else X${BINUTIL}?= ${${BINUTIL}} .endif .endfor # If a full path to an external linker is given, don't build lld. .if ${XLD:M/*} MK_LLD_BOOTSTRAP= no .endif # We also want the X_LINKER* variables if we are using an external toolchain. _WANT_TOOLCHAIN_CROSS_VARS= t .include "share/mk/bsd.linker.mk" .undef _WANT_TOOLCHAIN_CROSS_VARS # Begin WITH_SYSTEM_COMPILER / WITH_SYSTEM_LD # WITH_SYSTEM_COMPILER - Pull in needed values and make a decision. # Check if there is a local compiler that can satisfy as an external compiler. # Which compiler is expected to be used? .if ${MK_CLANG_BOOTSTRAP} == "yes" WANT_COMPILER_TYPE= clang .else WANT_COMPILER_TYPE= .endif .if !defined(WANT_COMPILER_FREEBSD_VERSION) && !make(showconfig) && \ !make(test-system-linker) .if ${WANT_COMPILER_TYPE} == "clang" WANT_COMPILER_FREEBSD_VERSION_FILE= lib/clang/freebsd_cc_version.h WANT_COMPILER_FREEBSD_VERSION!= \ awk '$$2 == "FREEBSD_CC_VERSION" {printf("%d\n", $$3)}' \ ${SRCDIR}/${WANT_COMPILER_FREEBSD_VERSION_FILE} || echo unknown WANT_COMPILER_VERSION_FILE= lib/clang/include/clang/Basic/Version.inc WANT_COMPILER_VERSION!= \ awk '$$2 == "CLANG_VERSION" {split($$3, a, "."); print a[1] * 10000 + a[2] * 100 + a[3]}' \ ${SRCDIR}/${WANT_COMPILER_VERSION_FILE} || echo unknown .endif .export WANT_COMPILER_FREEBSD_VERSION WANT_COMPILER_VERSION .endif # !defined(WANT_COMPILER_FREEBSD_VERSION) # It needs to be the same revision as we would build for the bootstrap. # If the expected vs CC is different then we can't skip. # GCC cannot be used for cross-arch yet. For clang we pass -target later if # TARGET_ARCH!=MACHINE_ARCH. .if ${MK_SYSTEM_COMPILER} == "yes" && \ defined(WANT_COMPILER_FREEBSD_VERSION) && \ ${MK_CLANG_BOOTSTRAP} == "yes" && \ !make(xdev*) && \ ${X_COMPILER_TYPE} == ${WANT_COMPILER_TYPE} && \ (${X_COMPILER_TYPE} == "clang" || ${TARGET_ARCH} == ${MACHINE_ARCH}) && \ ${X_COMPILER_VERSION} == ${WANT_COMPILER_VERSION} && \ ${X_COMPILER_FREEBSD_VERSION} == ${WANT_COMPILER_FREEBSD_VERSION} # Everything matches, disable the bootstrap compiler. MK_CLANG_BOOTSTRAP= no USING_SYSTEM_COMPILER= yes .endif # ${WANT_COMPILER_TYPE} == ${COMPILER_TYPE} # WITH_SYSTEM_LD - Pull in needed values and make a decision. # Check if there is a local linker that can satisfy as an external linker. # Which linker is expected to be used? .if ${MK_LLD_BOOTSTRAP} == "yes" WANT_LINKER_TYPE= lld .else WANT_LINKER_TYPE= .endif .if !defined(WANT_LINKER_FREEBSD_VERSION) && !make(showconfig) && \ !make(test-system-compiler) .if ${WANT_LINKER_TYPE} == "lld" WANT_LINKER_FREEBSD_VERSION_FILE= lib/clang/include/VCSVersion.inc _WANT_LINKER_FREEBSD_VERSION!= \ awk '$$2 == "LLD_REVISION" {gsub(/"/, "", $$3); print $$3}' \ ${SRCDIR}/${WANT_LINKER_FREEBSD_VERSION_FILE} || echo unknown WANT_LINKER_FREEBSD_VERSION=${_WANT_LINKER_FREEBSD_VERSION:C/.*-(.*)/\1/} WANT_LINKER_VERSION_FILE= lib/clang/include/lld/Common/Version.inc WANT_LINKER_VERSION!= \ awk '$$2 == "LLD_VERSION" {split($$3, a, "."); print a[1] * 10000 + a[2] * 100 + a[3]}' \ ${SRCDIR}/${WANT_LINKER_VERSION_FILE} || echo unknown .else WANT_LINKER_FREEBSD_VERSION_FILE= WANT_LINKER_FREEBSD_VERSION= .endif .export WANT_LINKER_FREEBSD_VERSION WANT_LINKER_VERSION .endif # !defined(WANT_LINKER_FREEBSD_VERSION) .if ${MK_SYSTEM_LINKER} == "yes" && \ defined(WANT_LINKER_FREEBSD_VERSION) && \ (${MK_LLD_BOOTSTRAP} == "yes") && \ !make(xdev*) && \ ${X_LINKER_TYPE} == ${WANT_LINKER_TYPE} && \ ${X_LINKER_VERSION} == ${WANT_LINKER_VERSION} && \ ${X_LINKER_FREEBSD_VERSION} == ${WANT_LINKER_FREEBSD_VERSION} # Everything matches, disable the bootstrap linker. MK_LLD_BOOTSTRAP= no USING_SYSTEM_LINKER= yes .endif # ${WANT_LINKER_TYPE} == ${LINKER_TYPE} # WITH_SYSTEM_COMPILER / WITH_SYSTEM_LINKER - Handle defaults and debug. USING_SYSTEM_COMPILER?= no USING_SYSTEM_LINKER?= no TEST_SYSTEM_COMPILER_VARS= \ USING_SYSTEM_COMPILER MK_SYSTEM_COMPILER \ MK_CROSS_COMPILER MK_CLANG_BOOTSTRAP \ WANT_COMPILER_TYPE WANT_COMPILER_VERSION WANT_COMPILER_VERSION_FILE \ WANT_COMPILER_FREEBSD_VERSION WANT_COMPILER_FREEBSD_VERSION_FILE \ CC COMPILER_TYPE COMPILER_FEATURES COMPILER_VERSION \ COMPILER_FREEBSD_VERSION \ XCC X_COMPILER_TYPE X_COMPILER_FEATURES X_COMPILER_VERSION \ X_COMPILER_FREEBSD_VERSION TEST_SYSTEM_LINKER_VARS= \ USING_SYSTEM_LINKER MK_SYSTEM_LINKER \ MK_LLD_BOOTSTRAP \ WANT_LINKER_TYPE WANT_LINKER_VERSION WANT_LINKER_VERSION_FILE \ WANT_LINKER_FREEBSD_VERSION WANT_LINKER_FREEBSD_VERSION_FILE \ LD LINKER_TYPE LINKER_FEATURES LINKER_VERSION \ LINKER_FREEBSD_VERSION \ XLD X_LINKER_TYPE X_LINKER_FEATURES X_LINKER_VERSION \ X_LINKER_FREEBSD_VERSION .for _t in compiler linker test-system-${_t}: .PHONY .for v in ${TEST_SYSTEM_${_t:tu}_VARS} ${_+_}@printf "%-35s= %s\n" "${v}" "${${v}}" .endfor .endfor .if (make(buildworld) || make(buildkernel) || make(kernel-toolchain) || \ make(toolchain) || make(_cross-tools)) .if ${USING_SYSTEM_COMPILER} == "yes" .info SYSTEM_COMPILER: Determined that CC=${CC} matches the source tree. Not bootstrapping a cross-compiler. .elif ${MK_CLANG_BOOTSTRAP} == "yes" .info SYSTEM_COMPILER: libclang will be built for bootstrapping a cross-compiler. .endif .if ${USING_SYSTEM_LINKER} == "yes" .info SYSTEM_LINKER: Determined that LD=${LD} matches the source tree. Not bootstrapping a cross-linker. .elif ${MK_LLD_BOOTSTRAP} == "yes" .info SYSTEM_LINKER: libclang will be built for bootstrapping a cross-linker. .endif .endif # End WITH_SYSTEM_COMPILER / WITH_SYSTEM_LD # Store some compiler metadata for use in installworld where we don't # want to invoke CC at all. _TOOLCHAIN_METADATA_VARS= COMPILER_VERSION \ COMPILER_TYPE \ COMPILER_FEATURES \ COMPILER_FREEBSD_VERSION \ COMPILER_RESOURCE_DIR \ LINKER_VERSION \ LINKER_FEATURES \ LINKER_TYPE \ LINKER_FREEBSD_VERSION toolchain-metadata.mk: .PHONY .META @: > ${.TARGET} @echo ".info Using cached toolchain metadata from build at $$(hostname) on $$(date)" \ > ${.TARGET} @echo "_LOADED_TOOLCHAIN_METADATA=t" >> ${.TARGET} .for v in ${_TOOLCHAIN_METADATA_VARS} @echo "${v}=${${v}}" >> ${.TARGET} @echo "X_${v}=${X_${v}}" >> ${.TARGET} .endfor @echo ".export ${_TOOLCHAIN_METADATA_VARS}" >> ${.TARGET} @echo ".export ${_TOOLCHAIN_METADATA_VARS:C,^,X_,}" >> ${.TARGET} # We must do lib/ and libexec/ before bin/ in case of a mid-install error to # keep the users system reasonably usable. For static->dynamic root upgrades, # we don't want to install a dynamic binary without rtld and the needed # libraries. More commonly, for dynamic root, we don't want to install a # binary that requires a newer library version that hasn't been installed yet. # This ordering is not a guarantee though. The only guarantee of a working # system here would require fine-grained ordering of all components based # on their dependencies. .if !empty(SUBDIR_OVERRIDE) SUBDIR= ${SUBDIR_OVERRIDE} .else SUBDIR= lib libexec # Add LOCAL_LIB_DIRS, but only if they will not be picked up as a SUBDIR # of a LOCAL_DIRS directory. This allows LOCAL_DIRS=foo and # LOCAL_LIB_DIRS=foo/lib to behave as expected. .for _DIR in ${LOCAL_DIRS:M*/} ${LOCAL_DIRS:N*/:S|$|/|} _REDUNDANT_LIB_DIRS+= ${LOCAL_LIB_DIRS:M${_DIR}*} .endfor .for _DIR in ${LOCAL_LIB_DIRS} .if ${_DIR} == ".WAIT" || (empty(_REDUNDANT_LIB_DIRS:M${_DIR}) && exists(${.CURDIR}/${_DIR}/Makefile)) SUBDIR+= ${_DIR} .endif .endfor .if !defined(NO_ROOT) && (make(installworld) || make(install)) # Ensure libraries are installed before progressing. SUBDIR+=.WAIT .endif SUBDIR+=bin .if ${MK_CDDL} != "no" SUBDIR+=cddl .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 .if ${MK_BOOT} != "no" SUBDIR+=stand .endif SUBDIR+=sys usr.bin usr.sbin .if ${MK_TESTS} != "no" SUBDIR+= tests .endif # Local directories are built in parallel with the base system directories. # Users may insert a .WAIT directive at the beginning or elsewhere within # the LOCAL_DIRS and LOCAL_LIB_DIRS lists as needed. .for _DIR in ${LOCAL_DIRS} .if ${_DIR} == ".WAIT" || exists(${.CURDIR}/${_DIR}/Makefile) SUBDIR+= ${_DIR} .endif .endfor # We must do etc/ last as it hooks into building the man whatis file # by calling 'makedb' in share/man. This is only relevant for # install/distribute so they build the whatis file after every manpage is # installed. .if make(installworld) || make(install) SUBDIR+=.WAIT .endif SUBDIR+=etc .endif # !empty(SUBDIR_OVERRIDE) .if defined(NOCLEAN) .warning NOCLEAN option is deprecated. Use NO_CLEAN instead. NO_CLEAN= ${NOCLEAN} .endif .if defined(NO_CLEANDIR) CLEANDIR= clean cleandepend .else CLEANDIR= cleandir .endif .if defined(WORLDFAST) NO_CLEAN= t NO_OBJWALK= t .endif .if ${MK_META_MODE} == "yes" # If filemon is used then we can rely on the build being incremental-safe. # The .meta files will also track the build command and rebuild should # it change. .if empty(.MAKE.MODE:Mnofilemon) NO_CLEAN= t .endif .endif .if defined(NO_OBJWALK) || ${MK_AUTO_OBJ} == "yes" NO_OBJWALK= t NO_KERNELOBJ= t .endif .if !defined(NO_OBJWALK) _obj= obj .endif LOCAL_TOOL_DIRS?= PACKAGEDIR?= ${DESTDIR}/${DISTDIR} .if empty(SHELL:M*csh*) BUILDENV_SHELL?=${SHELL} .else BUILDENV_SHELL?=/bin/sh .endif .if !defined(_MKSHOWCONFIG) .if !defined(SVN_CMD) || empty(SVN_CMD) . for _P in /usr/bin /usr/local/bin . for _S in svn svnlite . if exists(${_P}/${_S}) SVN_CMD= ${_P}/${_S} . endif . endfor . endfor .export SVN_CMD .endif SVNFLAGS?= -r HEAD .if !defined(VCS_REVISION) || empty(VCS_REVISION) .if !defined(SVNVERSION_CMD) || empty(SVNVERSION_CMD) . for _D in ${PATH:S,:, ,g} . if exists(${_D}/svnversion) SVNVERSION_CMD?=${_D}/svnversion . endif . if exists(${_D}/svnliteversion) SVNVERSION_CMD?=${_D}/svnliteversion . endif . endfor .endif _VCS_REVISION?= $$(eval ${SVNVERSION_CMD} ${SRCDIR}) . if !empty(_VCS_REVISION) VCS_REVISION= $$(echo r${_VCS_REVISION}) . endif .export VCS_REVISION .endif .if !defined(OSRELDATE) .if exists(/usr/include/osreldate.h) OSRELDATE!= awk '/^\#define[[:space:]]*__FreeBSD_version/ { print $$3 }' \ /usr/include/osreldate.h .else OSRELDATE= 0 .endif .export OSRELDATE .endif # Set VERSION for CTFMERGE to use via the default CTFFLAGS=-L VERSION. .if !defined(_REVISION) _REVISION!= ${MAKE} -C ${SRCDIR}/release MK_AUTO_OBJ=no -V REVISION .export _REVISION .endif .if !defined(_BRANCH) _BRANCH!= ${MAKE} -C ${SRCDIR}/release MK_AUTO_OBJ=no -V BRANCH .export _BRANCH .endif .if !defined(SRCRELDATE) SRCRELDATE!= awk '/^\#define[[:space:]]*__FreeBSD_version/ { print $$3 }' \ ${SRCDIR}/sys/sys/param.h .export SRCRELDATE .endif .if !defined(VERSION) VERSION= FreeBSD ${_REVISION}-${_BRANCH:C/-p[0-9]+$//} ${TARGET_ARCH} ${SRCRELDATE} .export VERSION .endif .if !defined(PKG_VERSION) .if ${_BRANCH:MSTABLE*} || ${_BRANCH:MCURRENT*} TIMENOW= %Y%m%d%H%M%S EXTRA_REVISION= .s${TIMENOW:gmtime} .elif ${_BRANCH:MALPHA*} EXTRA_REVISION= _${_BRANCH:C/-ALPHA/.a/} .elif ${_BRANCH:MBETA*} EXTRA_REVISION= _${_BRANCH:C/-BETA/.b/} .elif ${_BRANCH:MRC*} EXTRA_REVISION= _${_BRANCH:C/-RC/.r/} .elif ${_BRANCH:MPRERELEASE*} EXTRA_REVISION= _${_BRANCH:C/-PRERELEASE/.p/} .elif ${_BRANCH:M*-p*} EXTRA_REVISION= _${_BRANCH:C/.*-p([0-9]+$)/\1/} .endif PKG_VERSION= ${_REVISION}${EXTRA_REVISION} .endif .endif # !defined(PKG_VERSION) .if !defined(PKG_TIMESTAMP) TIMEEPOCHNOW= %s SOURCE_DATE_EPOCH= ${TIMEEPOCHNOW:gmtime} .else SOURCE_DATE_EPOCH= ${PKG_TIMESTAMP} .endif .if !defined(_MKSHOWCONFIG) _CPUTYPE!= MAKEFLAGS= CPUTYPE=${_TARGET_CPUTYPE} ${MAKE} -f /dev/null \ -m ${.CURDIR}/share/mk MK_AUTO_OBJ=no -V CPUTYPE .if ${_CPUTYPE} != ${_TARGET_CPUTYPE} .error CPUTYPE global should be set with ?=. .endif .endif .if make(buildworld) BUILD_ARCH!= uname -p .if ${MACHINE_ARCH} != ${BUILD_ARCH} .error To cross-build, set TARGET_ARCH. .endif .endif WORLDTMP?= ${OBJTOP}/tmp BPATH= ${CCACHE_WRAPPER_PATH_PFX}${WORLDTMP}/legacy/usr/sbin:${WORLDTMP}/legacy/usr/bin:${WORLDTMP}/legacy/bin:${WORLDTMP}/legacy/usr/libexec XPATH= ${WORLDTMP}/usr/sbin:${WORLDTMP}/usr/bin # When building we want to find the cross tools before the host tools in ${BPATH}. # We also need to add UNIVERSE_TOOLCHAIN_PATH so that we can find the shared # toolchain files (clang, lld, etc.) during make universe/tinderbox STRICTTMPPATH= ${XPATH}:${BPATH}:${UNIVERSE_TOOLCHAIN_PATH} # We should not be using tools from /usr/bin accidentally since this could cause # the build to break on other systems that don't have that tool. For now we # still allow using the old behaviour (inheriting $PATH) if # BUILD_WITH_STRICT_TMPPATH is set to 0 but this will eventually be removed. # Currently strict $PATH can cause build failures and does not work yet with # USING_SYSTEM_LINKER/USING_SYSTEM_COMPILER. Once these issues have been # resolved it will be turned on by default. BUILD_WITH_STRICT_TMPPATH?=0 .if ${BUILD_WITH_STRICT_TMPPATH} != 0 TMPPATH= ${STRICTTMPPATH} .else TMPPATH= ${STRICTTMPPATH}:${PATH} .endif # # 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) || make(stageworld) .if ${BUILD_WITH_STRICT_TMPPATH} != 0 MKTEMP=${WORLDTMP}/legacy/usr/bin/mktemp .if !exists(${MKTEMP}) .error "mktemp binary doesn't exist in expected location: ${MKTEMP}" .endif .else MKTEMP=mktemp .endif INSTALLTMP!= ${MKTEMP} -d -u -t install .endif .if make(stagekernel) || make(distributekernel) TAGS+= kernel PACKAGE= kernel .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. These are generally # APIs that tools from one of those three stages need to # build that aren't present on the host. # 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. Some programs are listed during # this phase because they build binaries to generate # files needed to build these programs. This stage also # builds the 'build-tools' target rather than 'all'. # 3. cross-tools stage [XMAKE] # This stage is responsible for creating any tools that # are needed for building the system. A cross-compiler is one # of them. This differs from build tools in two ways: # 1. the 'all' target is built rather than 'build-tools' # 2. these tools are installed into TMPPATH for stage 4. # 4. world stage [WMAKE] # This stage actually builds the world. # 5. install stage (optional) [IMAKE] # This stage installs a previously built world. # BOOTSTRAPPING?= 0 # Keep these in sync MINIMUM_SUPPORTED_OSREL?= 1002501 MINIMUM_SUPPORTED_REL?= 10.3 # Common environment for world related stages CROSSENV+= \ MACHINE_ARCH=${TARGET_ARCH} \ MACHINE=${TARGET} \ CPUTYPE=${TARGET_CPUTYPE} .if ${MK_META_MODE} != "no" # Don't rebuild build-tools targets during normal build. CROSSENV+= BUILD_TOOLS_META=.NOMETA .endif .if defined(TARGET_CFLAGS) CROSSENV+= ${TARGET_CFLAGS} .endif .if (${TARGET} != ${MACHINE} && !defined(WITH_LOCAL_MODULES)) || \ defined(WITHOUT_LOCAL_MODULES) CROSSENV+= LOCAL_MODULES= .endif BOOTSTRAPPING_OSRELDATE?=${OSRELDATE} # bootstrap-tools stage BMAKEENV= INSTALL="sh ${.CURDIR}/tools/install.sh" \ TOOLS_PREFIX=${TOOLS_PREFIX_UNDEF:U${WORLDTMP}} \ PATH=${BPATH}:${PATH} \ WORLDTMP=${WORLDTMP} \ MAKEFLAGS="-m ${.CURDIR}/tools/build/mk ${.MAKEFLAGS}" # need to keep this in sync with targets/pseudo/bootstrap-tools/Makefile BSARGS= DESTDIR= \ OBJTOP='${WORLDTMP}/obj-tools' \ OBJROOT='$${OBJTOP}/' \ MAKEOBJDIRPREFIX= \ BOOTSTRAPPING=${BOOTSTRAPPING_OSRELDATE} \ BWPHASE=${.TARGET:C,^_,,} \ SSP_CFLAGS= \ MK_HTML=no NO_LINT=yes MK_MAN=no \ -DNO_PIC MK_PROFILE=no -DNO_SHARED \ -DNO_CPU_CFLAGS MK_WARNS=no MK_CTF=no \ MK_CLANG_EXTRAS=no MK_CLANG_FORMAT=no MK_CLANG_FULL=no \ MK_LLDB=no MK_RETPOLINE=no MK_TESTS=no \ MK_INCLUDES=yes BMAKE= \ ${BMAKEENV} ${MAKE} ${WORLD_FLAGS} -f Makefile.inc1 \ ${BSARGS} .if empty(.MAKEOVERRIDES:MMK_LLVM_TARGET_ALL) BMAKE+= MK_LLVM_TARGET_ALL=no .endif # build-tools stage TMAKE= \ ${BMAKEENV} ${MAKE} ${WORLD_FLAGS} -f Makefile.inc1 \ TARGET=${TARGET} TARGET_ARCH=${TARGET_ARCH} \ DESTDIR= \ BOOTSTRAPPING=${BOOTSTRAPPING_OSRELDATE} \ BWPHASE=${.TARGET:C,^_,,} \ SSP_CFLAGS= \ -DNO_LINT \ -DNO_CPU_CFLAGS MK_WARNS=no MK_CTF=no \ MK_CLANG_EXTRAS=no MK_CLANG_FORMAT=no MK_CLANG_FULL=no \ MK_LLDB=no MK_RETPOLINE=no MK_TESTS=no # cross-tools stage # TOOLS_PREFIX set in BMAKE XMAKE= ${BMAKE} \ TARGET=${TARGET} TARGET_ARCH=${TARGET_ARCH} \ MK_CLANG_IS_CC=${MK_CLANG_BOOTSTRAP} \ MK_GDB=no MK_TESTS=no # kernel-tools stage KTMAKEENV= INSTALL="sh ${.CURDIR}/tools/install.sh" \ PATH=${BPATH}:${PATH} \ WORLDTMP=${WORLDTMP} KTMAKE= \ TOOLS_PREFIX=${TOOLS_PREFIX_UNDEF:U${WORLDTMP}} \ ${KTMAKEENV} ${MAKE} ${WORLD_FLAGS} -f Makefile.inc1 \ DESTDIR= \ OBJTOP='${WORLDTMP}/obj-kernel-tools' \ OBJROOT='$${OBJTOP}/' \ MAKEOBJDIRPREFIX= \ BOOTSTRAPPING=${BOOTSTRAPPING_OSRELDATE} \ SSP_CFLAGS= \ MK_HTML=no -DNO_LINT MK_MAN=no \ -DNO_PIC MK_PROFILE=no -DNO_SHARED \ -DNO_CPU_CFLAGS MK_RETPOLINE=no MK_WARNS=no MK_CTF=no # world stage WMAKEENV= ${CROSSENV} \ INSTALL="sh ${.CURDIR}/tools/install.sh" \ PATH=${TMPPATH} \ SYSROOT=${WORLDTMP} # make hierarchy HMAKE= PATH=${TMPPATH} ${MAKE} LOCAL_MTREE=${LOCAL_MTREE:Q} .if defined(NO_ROOT) HMAKE+= PATH=${TMPPATH} METALOG=${METALOG} -DNO_ROOT .endif CROSSENV+= CC="${XCC} ${XCFLAGS}" CXX="${XCXX} ${XCXXFLAGS} ${XCFLAGS}" \ CPP="${XCPP} ${XCFLAGS}" \ AS="${XAS}" AR="${XAR}" LD="${XLD}" LLVM_LINK="${XLLVM_LINK}" \ NM=${XNM} OBJCOPY="${XOBJCOPY}" \ RANLIB=${XRANLIB} STRINGS=${XSTRINGS} \ SIZE="${XSIZE}" .if defined(CROSS_BINUTILS_PREFIX) && exists(${CROSS_BINUTILS_PREFIX}) # In the case of xdev-build tools, CROSS_BINUTILS_PREFIX won't be a # directory, but the compiler will look in the right place for its # tools so we don't need to tell it where to look. BFLAGS+= -B${CROSS_BINUTILS_PREFIX} .endif # The internal bootstrap compiler has a default sysroot set by TOOLS_PREFIX # and target set by TARGET/TARGET_ARCH. However, there are several needs to # always pass an explicit --sysroot and -target. # - External compiler needs sysroot and target flags. # - External ld needs sysroot. # - To be clear about the use of a sysroot when using the internal compiler. # - Easier debugging. # - Allowing WITH_SYSTEM_COMPILER+WITH_META_MODE to work together due to # the flip-flopping build command when sometimes using external and # sometimes using internal. # - Allow using lld which has no support for default paths. .if !defined(CROSS_BINUTILS_PREFIX) || !exists(${CROSS_BINUTILS_PREFIX}) BFLAGS+= -B${WORLDTMP}/usr/bin .endif .if ${WANT_COMPILER_TYPE} == gcc || \ (defined(X_COMPILER_TYPE) && ${X_COMPILER_TYPE} == gcc) .elif ${WANT_COMPILER_TYPE} == clang || \ (defined(X_COMPILER_TYPE) && ${X_COMPILER_TYPE} == clang) XCFLAGS+= -target ${TARGET_TRIPLE} .endif XCFLAGS+= --sysroot=${WORLDTMP} .if !empty(BFLAGS) XCFLAGS+= ${BFLAGS} .endif .if ${MK_LIB32} == "yes" _LIBCOMPAT= 32 .include "Makefile.libcompat" .elif ${MK_LIBSOFT} == "yes" _LIBCOMPAT= SOFT .include "Makefile.libcompat" .endif # META_MODE normally ignores host file changes since every build updates # timestamps (see NO_META_IGNORE_HOST in sys.mk). There are known times # when the ABI breaks though that we want to force rebuilding WORLDTMP # to get updated host tools. .if ${MK_META_MODE} == "yes" && defined(NO_CLEAN) && \ !defined(NO_META_IGNORE_HOST) && !defined(NO_META_IGNORE_HOST_HEADERS) && \ !defined(_MKSHOWCONFIG) # r318736 - ino64 major ABI breakage META_MODE_BAD_ABI_VERS+= 1200031 .if !defined(OBJDIR_HOST_OSRELDATE) .if exists(${OBJTOP}/host-osreldate.h) OBJDIR_HOST_OSRELDATE!= \ awk '/^\#define[[:space:]]*__FreeBSD_version/ { print $$3 }' \ ${OBJTOP}/host-osreldate.h .elif exists(${WORLDTMP}/usr/include/osreldate.h) OBJDIR_HOST_OSRELDATE= 0 .endif .export OBJDIR_HOST_OSRELDATE .endif # Note that this logic is the opposite of normal BOOTSTRAP handling. We want # to compare the WORLDTMP's OSRELDATE to the host's OSRELDATE. If the WORLDTMP # is older than the ABI-breakage OSRELDATE of the HOST then we rebuild. .if defined(OBJDIR_HOST_OSRELDATE) .for _ver in ${META_MODE_BAD_ABI_VERS} .if ${OSRELDATE} >= ${_ver} && ${OBJDIR_HOST_OSRELDATE} < ${_ver} _meta_mode_need_rebuild= ${_ver} .endif .endfor .if defined(_meta_mode_need_rebuild) .info META_MODE: Rebuilding host tools due to ABI breakage in __FreeBSD_version ${_meta_mode_need_rebuild}. NO_META_IGNORE_HOST_HEADERS= 1 .export NO_META_IGNORE_HOST_HEADERS .endif # defined(_meta_mode_need_rebuild) .endif # defined(OBJDIR_HOST_OSRELDATE) .endif # ${MK_META_MODE} == "yes" && defined(NO_CLEAN) ... # This is only used for META_MODE+filemon to track what the oldest # __FreeBSD_version is in WORLDTMP. This purposely does NOT have # a make dependency on /usr/include/osreldate.h as the file should # only be copied when it is missing or meta mode determines it has changed. # Since host files are normally ignored without NO_META_IGNORE_HOST # the file will never be updated unless that flag is specified. This # allows tracking the oldest osreldate to force rebuilds via # META_MODE_BADABI_REVS above. host-osreldate.h: # DO NOT ADD /usr/include/osreldate.h here @cp -f /usr/include/osreldate.h ${.TARGET} WMAKE= ${WMAKEENV} ${MAKE} ${WORLD_FLAGS} -f Makefile.inc1 \ BWPHASE=${.TARGET:C,^_,,} \ DESTDIR=${WORLDTMP} IMAKEENV= ${CROSSENV} IMAKE= ${IMAKEENV} ${MAKE} -f Makefile.inc1 \ ${IMAKE_INSTALL} ${IMAKE_MTREE} .if empty(.MAKEFLAGS:M-n) IMAKEENV+= PATH=${STRICTTMPPATH}:${INSTALLTMP} \ LD_LIBRARY_PATH=${INSTALLTMP} \ PATH_LOCALE=${INSTALLTMP}/locale IMAKE+= __MAKE_SHELL=${INSTALLTMP}/sh .else IMAKEENV+= PATH=${TMPPATH}:${INSTALLTMP} .endif # When generating install media, do not allow user and group information from # the build host to affect the contents of the distribution. .if make(distributeworld) || make(distrib-dirs) || make(distribution) || \ make(stageworld) DB_FROM_SRC= yes .endif .if defined(DB_FROM_SRC) INSTALLFLAGS+= -N ${.CURDIR}/etc MTREEFLAGS+= -N ${.CURDIR}/etc .endif _INSTALL_DDIR= ${DESTDIR}/${DISTDIR} INSTALL_DDIR= ${_INSTALL_DDIR:S://:/:g:C:/$::} .if defined(NO_ROOT) METALOG?= ${DESTDIR}/${DISTDIR}/METALOG METALOG:= ${METALOG:C,//+,/,g} IMAKE+= -DNO_ROOT METALOG=${METALOG} INSTALLFLAGS+= -U -M ${METALOG} -D ${INSTALL_DDIR} MTREEFLAGS+= -W .endif .if defined(BUILD_PKGS) INSTALLFLAGS+= -h sha256 .endif .if defined(DB_FROM_SRC) || defined(NO_ROOT) IMAKE_INSTALL= INSTALL="${INSTALL_CMD} ${INSTALLFLAGS}" IMAKE_MTREE= MTREE_CMD="${MTREE_CMD} ${MTREEFLAGS}" .endif DESTDIR_MTREEFLAGS= -deU # When creating worldtmp we don't need to set the directories as owned by root # so we also pass -W WORLDTMP_MTREEFLAGS= -deUW .if defined(NO_ROOT) # When building with -DNO_ROOT we shouldn't be changing the directories # that are created by mtree to be owned by root/wheel. DESTDIR_MTREEFLAGS+= -W .endif DISTR_MTREE= ${MTREE_CMD} .if ${BUILD_WITH_STRICT_TMPPATH} != 0 DISTR_MTREE= ${WORLDTMP}/legacy/usr/sbin/mtree .endif WORLDTMP_MTREE= ${DISTR_MTREE} ${WORLDTMP_MTREEFLAGS} DESTDIR_MTREE= ${DISTR_MTREE} ${DESTDIR_MTREEFLAGS} # kernel stage KMAKEENV= ${WMAKEENV:NSYSROOT=*} KMAKE= ${KMAKEENV} ${MAKE} ${.MAKEFLAGS} ${KERNEL_FLAGS} KERNEL=${INSTKERNNAME} # # buildworld # # Attempt to rebuild the entire system, with reasonable chance of # success, regardless of how old your existing system is. # _sanity_check: .PHONY .MAKE .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 .elif ${.CURDIR:M*\:*} != "" # Using ':' leaks into PATH and breaks finding cross-tools. @echo @echo "*** Error: path to source tree contains a colon ':'" @echo @false .endif # Our current approach to dependency tracking cannot cope with certain source # tree changes, particularly with respect to removing source files and # replacing generated files. Handle these cases here in an ad-hoc fashion. _cleanobj_fast_depend_hack: .PHONY @echo ">>> Deleting stale dependencies..."; sh ${.CURDIR}/tools/build/depend-cleanup.sh ${OBJTOP} # Date SVN Rev Syscalls/Changes # Syscall stubs rewritten in C and obsolete MD assembly implementations # 20191009 r353340 removal of opensolaris_atomic.S (also r353381) .if ${MACHINE} != i386 .for f in opensolaris_atomic @if [ -e "${OBJTOP}/cddl/lib/libzpool/.depend.${f}.o" ] && \ egrep -qw 'opensolaris_atomic\.S' ${OBJTOP}/cddl/lib/libzpool/.depend.${f}.o; then \ echo "Removing stale dependencies for opensolaris_atomic"; \ rm -f ${OBJTOP}/cddl/lib/libzpool/.depend.${f}.* \ ${LIBCOMPAT:D${LIBCOMPAT_OBJTOP}/cddl/lib/libzpool/.depend.${f}.*}; \ fi .endfor .endif # 20190925 r352689 removal of obsolete i386 memchr.S .for f in memchr @if [ -e "${OBJTOP}/lib/libc/.depend.${f}.o" ] && \ egrep -qw 'i386/string/memchr\.S' ${OBJTOP}/lib/libc/.depend.${f}.o; then \ echo "Removing stale dependencies for memchr"; \ rm -f ${OBJTOP}/lib/libc/.depend.${f}.*; \ fi .if defined(_LIBCOMPAT) @if [ -e "${LIBCOMPAT_OBJTOP}/lib/libc/.depend.${f}.o" ] && \ egrep -qw 'i386/string/memchr\.S' ${LIBCOMPAT_OBJTOP}/lib/libc/.depend.${f}.o; then \ echo "Removing stale dependencies for memchr"; \ rm -f ${LIBCOMPAT_OBJTOP}/lib/libc/.depend.${f}.*; \ fi .endif .endfor _worldtmp: .PHONY @echo @echo "--------------------------------------------------------------" @echo ">>> Rebuilding the temporary build tree" @echo "--------------------------------------------------------------" .if !defined(NO_CLEAN) rm -rf ${WORLDTMP} .else # Note: for delete-old we need to set $PATH to also include the host $PATH # since otherwise a partial build with missing symlinks in ${WORLDTMP}/legacy/ # will fail to run due to missing binaries. $WMAKE sets PATH to only ${TMPPATH} # so we remove that assingnment from $WMAKE and prepend the new $PATH ${_+_}@if [ -e "${WORLDTMP}" ]; then \ echo ">>> Deleting stale files in build tree..."; \ cd ${.CURDIR}; env PATH=${TMPPATH}:${PATH} ${WMAKE:NPATH=*} \ _NO_INCLUDE_COMPILERMK=t -DBATCH_DELETE_OLD_FILES delete-old \ delete-old-libs >/dev/null; \ fi rm -rf ${WORLDTMP}/legacy/usr/include .if ${USING_SYSTEM_COMPILER} == "yes" .for cc in cc c++ if [ -x ${WORLDTMP}/usr/bin/${cc} ]; then \ inum=$$(stat -f %i ${WORLDTMP}/usr/bin/${cc}); \ find ${WORLDTMP}/usr/bin -inum $${inum} -delete; \ fi .endfor .endif # ${USING_SYSTEM_COMPILER} == "yes" .if ${USING_SYSTEM_LINKER} == "yes" @rm -f ${WORLDTMP}/usr/bin/ld ${WORLDTMP}/usr/bin/ld.lld .endif # ${USING_SYSTEM_LINKER} == "yes" .endif # !defined(NO_CLEAN) @mkdir -p ${WORLDTMP} @touch ${WORLDTMP}/${.TARGET} # We can't use mtree to create the worldtmp directories since it may not be # available on the target system (this happens e.g. when building on non-FreeBSD) cd ${.CURDIR}/tools/build; \ ${MAKE} DIRPRFX=tools/build/ DESTDIR=${WORLDTMP}/legacy installdirs # In order to build without inheriting $PATH we need to add symlinks to the host # tools in $WORLDTMP for the tools that we don't build during bootstrap-tools cd ${.CURDIR}/tools/build; \ ${MAKE} DIRPRFX=tools/build/ DESTDIR=${WORLDTMP}/legacy host-symlinks _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 mkdir -p ${WORLDTMP}/usr ${WORLDTMP}/lib/casper ${WORLDTMP}/lib/geom ${WORLDTMP_MTREE} -f ${.CURDIR}/etc/mtree/BSD.usr.dist \ -p ${WORLDTMP}/usr >/dev/null ${WORLDTMP_MTREE} -f ${.CURDIR}/etc/mtree/BSD.include.dist \ -p ${WORLDTMP}/usr/include >/dev/null ln -sf ${.CURDIR}/sys ${WORLDTMP} .if ${MK_DEBUG_FILES} != "no" ${WORLDTMP_MTREE} -f ${.CURDIR}/etc/mtree/BSD.debug.dist \ -p ${WORLDTMP}/usr/lib >/dev/null .endif .for _mtree in ${LOCAL_MTREE} ${WORLDTMP_MTREE} -f ${.CURDIR}/${_mtree} -p ${WORLDTMP} > /dev/null .endfor _cleanobj: .if !defined(NO_CLEAN) @echo @echo "--------------------------------------------------------------" @echo ">>> stage 2.1: cleaning up the object tree" @echo "--------------------------------------------------------------" # Avoid including bsd.compiler.mk in clean and obj with _NO_INCLUDE_COMPILERMK # since the restricted $PATH might not contain a valid cc binary ${_+_}cd ${.CURDIR}; ${WMAKE} _NO_INCLUDE_COMPILERMK=t ${CLEANDIR} .if defined(_LIBCOMPAT) ${_+_}cd ${.CURDIR}; ${LIBCOMPATWMAKE} _NO_INCLUDE_COMPILERMK=t -f Makefile.inc1 ${CLEANDIR} .endif .else ${_+_}cd ${.CURDIR}; ${WMAKE} _NO_INCLUDE_COMPILERMK=t _cleanobj_fast_depend_hack .endif # !defined(NO_CLEAN) _obj: @echo @echo "--------------------------------------------------------------" @echo ">>> stage 2.2: rebuilding the object tree" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${WMAKE} _NO_INCLUDE_COMPILERMK=t 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 "--------------------------------------------------------------" @rm -f ${OBJTOP}/toolchain-metadata.mk ${_+_}cd ${.CURDIR}; ${XMAKE} cross-tools ${_+_}cd ${.CURDIR}; ${XMAKE} kernel-tools _build-metadata: @echo @echo "--------------------------------------------------------------" @echo ">>> stage 3.1: recording build metadata" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${WMAKE} toolchain-metadata.mk ${_+_}cd ${.CURDIR}; ${WMAKE} host-osreldate.h _includes: @echo @echo "--------------------------------------------------------------" @echo ">>> stage 4.1: building includes" @echo "--------------------------------------------------------------" # Special handling for SUBDIR_OVERRIDE in buildworld as they most likely need # headers from default SUBDIR. Do SUBDIR_OVERRIDE includes last. ${_+_}cd ${.CURDIR}; ${WMAKE} SUBDIR_OVERRIDE= SHARED=symlinks \ MK_INCLUDES=yes includes .if !empty(SUBDIR_OVERRIDE) && make(buildworld) ${_+_}cd ${.CURDIR}; ${WMAKE} MK_INCLUDES=yes SHARED=symlinks includes .endif _libraries: @echo @echo "--------------------------------------------------------------" @echo ">>> stage 4.2: building libraries" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; \ ${WMAKE} -DNO_FSCHG MK_HTML=no -DNO_LINT MK_MAN=no \ MK_PROFILE=no MK_TESTS=no MK_TESTS_SUPPORT=${MK_TESTS_SUPPORT} \ libraries everything: .PHONY @echo @echo "--------------------------------------------------------------" @echo ">>> stage 4.4: building everything" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; _PARALLEL_SUBDIR_OK=1 ${WMAKE} all WMAKE_TGTS= .if !defined(WORLDFAST) WMAKE_TGTS+= _sanity_check _worldtmp _legacy .if empty(SUBDIR_OVERRIDE) WMAKE_TGTS+= _bootstrap-tools .endif WMAKE_TGTS+= _cleanobj .if !defined(NO_OBJWALK) WMAKE_TGTS+= _obj .endif WMAKE_TGTS+= _build-tools _cross-tools WMAKE_TGTS+= _build-metadata WMAKE_TGTS+= _includes .endif .if !defined(NO_LIBS) WMAKE_TGTS+= _libraries .endif .if defined(_LIBCOMPAT) && empty(SUBDIR_OVERRIDE) WMAKE_TGTS+= build${libcompat} .endif WMAKE_TGTS+= everything # record buildworld time in seconds .if make(buildworld) _BUILDWORLD_START!= date '+%s' .export _BUILDWORLD_START .endif buildworld: buildworld_prologue ${WMAKE_TGTS} buildworld_epilogue .PHONY .ORDER: buildworld_prologue ${WMAKE_TGTS} buildworld_epilogue buildworld_prologue: .PHONY @echo "--------------------------------------------------------------" @echo ">>> World build started on `LC_ALL=C date`" @echo "--------------------------------------------------------------" buildworld_epilogue: .PHONY @echo @echo "--------------------------------------------------------------" @echo ">>> World build completed on `LC_ALL=C date`" @seconds=$$(($$(date '+%s') - ${_BUILDWORLD_START})); \ echo -n ">>> World built in $$seconds seconds, "; \ echo "ncpu: $$(sysctl -n hw.ncpu)${.MAKE.JOBS:S/^/, make -j/}" @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: .PHONY @echo ${WMAKEENV:Q} ${.MAKE.EXPORTED:@v@$v=\"${$v}\"@} .if ${.TARGETS:Mbuildenv} .if ${.MAKEFLAGS:M-j} .error The buildenv target is incompatible with -j .endif .endif BUILDENV_DIR?= ${.CURDIR} # # Note: make will report any errors the shell reports. This can # be odd if the last command in an interactive shell generates an # error or is terminated by SIGINT. These reported errors look bad, # but are harmless. Allowing them also allows BUIDLENV_SHELL to # be a complex command whose status will be returned to the caller. # Some scripts in tools rely on this behavior to report build errors. # buildenv: .PHONY @echo Entering world for ${TARGET_ARCH}:${TARGET} .if ${BUILDENV_SHELL:M*zsh*} @echo For ZSH you must run: export CPUTYPE=${TARGET_CPUTYPE} .endif @cd ${BUILDENV_DIR} && env ${WMAKEENV} BUILDENV=1 ${BUILDENV_SHELL} TOOLCHAIN_TGTS= ${WMAKE_TGTS:Neverything:Nbuild${libcompat}} toolchain: ${TOOLCHAIN_TGTS} .PHONY KERNEL_TOOLCHAIN_TGTS= ${TOOLCHAIN_TGTS:N_obj:N_cleanobj:N_includes:N_libraries} .if make(kernel-toolchain) .ORDER: ${KERNEL_TOOLCHAIN_TGTS} .endif kernel-toolchain: ${KERNEL_TOOLCHAIN_TGTS} .PHONY # # installcheck # # Checks to be sure system is ready for installworld/installkernel. # installcheck: _installcheck_world _installcheck_kernel .PHONY _installcheck_world: .PHONY @echo "--------------------------------------------------------------" @echo ">>> Install check world" @echo "--------------------------------------------------------------" _installcheck_kernel: .PHONY @echo "--------------------------------------------------------------" @echo ">>> Install check kernel" @echo "--------------------------------------------------------------" # # Require DESTDIR to be set if installing for a different architecture or # using the user/group database in the source tree. # .if ${TARGET_ARCH} != ${MACHINE_ARCH} || ${TARGET} != ${MACHINE} || \ defined(DB_FROM_SRC) .if !make(distributeworld) _installcheck_world: __installcheck_DESTDIR _installcheck_kernel: __installcheck_DESTDIR __installcheck_DESTDIR: .PHONY .if !defined(DESTDIR) || empty(DESTDIR) @echo "ERROR: Please set DESTDIR!"; \ false .endif .endif .endif .if !defined(DB_FROM_SRC) # # Check for missing UIDs/GIDs. # CHECK_UIDS= auditdistd CHECK_GIDS= audit CHECK_UIDS+= ntpd CHECK_GIDS+= ntpd CHECK_UIDS+= proxy CHECK_GIDS+= proxy authpf CHECK_UIDS+= smmsp CHECK_GIDS+= smmsp CHECK_UIDS+= unbound CHECK_GIDS+= unbound _installcheck_world: __installcheck_UGID __installcheck_UGID: .PHONY .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 .endif # # If installing over the running system (DESTDIR is / or unset) and the install # includes rescue, try running rescue from the objdir as a sanity check. If # rescue is not functional (e.g., because it depends on a system call not # supported by the currently running kernel), abort the installation. # .if !make(distributeworld) && ${MK_RESCUE} != "no" && \ (empty(DESTDIR) || ${DESTDIR} == "/") && empty(BYPASS_INSTALLCHECK_SH) _installcheck_world: __installcheck_sh_check __installcheck_sh_check: .PHONY @if [ "`${OBJTOP}/rescue/rescue/rescue sh -c 'echo OK'`" != \ OK ]; then \ echo "rescue/sh check failed, installation aborted" >&2; \ false; \ fi .endif # # Required install tools to be saved in a scratch dir for safety. # .if ${MK_ZONEINFO} != "no" _zoneinfo= zic tzsetup .endif ITOOLS= [ awk cap_mkdb cat chflags chmod chown cmp cp \ date echo egrep find grep id install ${_install-info} \ ln make mkdir mtree mv pwd_mkdb \ rm sed services_mkdb sh sort strip sysctl test true uname wc ${_zoneinfo} \ ${LOCAL_ITOOLS} # Needed for share/man .if ${MK_MAN_UTILS} != "no" ITOOLS+=makewhatis .endif # # distributeworld # # Distributes everything compiled by a `buildworld'. # # installworld # # Installs everything compiled by a 'buildworld'. # # Non-base distributions produced by the base system EXTRA_DISTRIBUTIONS= .if defined(_LIBCOMPAT) EXTRA_DISTRIBUTIONS+= lib${libcompat} .endif .if ${MK_TESTS} != "no" EXTRA_DISTRIBUTIONS+= tests .endif DEBUG_DISTRIBUTIONS= .if ${MK_DEBUG_FILES} != "no" DEBUG_DISTRIBUTIONS+= base ${EXTRA_DISTRIBUTIONS:S,tests,,} .endif MTREE_MAGIC?= mtree 2.0 distributeworld installworld stageworld: _installcheck_world .PHONY 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 defined(NO_ROOT) -mkdir -p ${METALOG:H} echo "#${MTREE_MAGIC}" > ${METALOG} .endif .if make(distributeworld) .for dist in ${EXTRA_DISTRIBUTIONS} -mkdir ${DESTDIR}/${DISTDIR}/${dist} ${DESTDIR_MTREE} -f ${.CURDIR}/etc/mtree/BSD.root.dist \ -p ${DESTDIR}/${DISTDIR}/${dist} >/dev/null ${DESTDIR_MTREE} -f ${.CURDIR}/etc/mtree/BSD.usr.dist \ -p ${DESTDIR}/${DISTDIR}/${dist}/usr >/dev/null ${DESTDIR_MTREE} -f ${.CURDIR}/etc/mtree/BSD.include.dist \ -p ${DESTDIR}/${DISTDIR}/${dist}/usr/include >/dev/null .if ${MK_DEBUG_FILES} != "no" ${DESTDIR_MTREE} -f ${.CURDIR}/etc/mtree/BSD.debug.dist \ -p ${DESTDIR}/${DISTDIR}/${dist}/usr/lib >/dev/null .endif .if defined(_LIBCOMPAT) ${DESTDIR_MTREE} -f ${.CURDIR}/etc/mtree/BSD.lib${libcompat}.dist \ -p ${DESTDIR}/${DISTDIR}/${dist}/usr >/dev/null .if ${MK_DEBUG_FILES} != "no" ${DESTDIR_MTREE} -f ${.CURDIR}/etc/mtree/BSD.lib${libcompat}.dist \ -p ${DESTDIR}/${DISTDIR}/${dist}/usr/lib/debug/usr >/dev/null .endif .endif .if ${MK_TESTS} != "no" && ${dist} == "tests" -mkdir -p ${DESTDIR}/${DISTDIR}/${dist}${TESTSBASE} ${DESTDIR_MTREE} -f ${.CURDIR}/etc/mtree/BSD.tests.dist \ -p ${DESTDIR}/${DISTDIR}/${dist}${TESTSBASE} >/dev/null .if ${MK_DEBUG_FILES} != "no" ${DESTDIR_MTREE} -f ${.CURDIR}/etc/mtree/BSD.tests.dist \ -p ${DESTDIR}/${DISTDIR}/${dist}/usr/lib/debug/${TESTSBASE} >/dev/null .endif .endif .if defined(NO_ROOT) ${IMAKEENV} ${DISTR_MTREE} -C -f ${.CURDIR}/etc/mtree/BSD.root.dist | \ sed -e 's#^\./#./${dist}/#' >> ${METALOG} ${IMAKEENV} ${DISTR_MTREE} -C -f ${.CURDIR}/etc/mtree/BSD.usr.dist | \ sed -e 's#^\./#./${dist}/usr/#' >> ${METALOG} ${IMAKEENV} ${DISTR_MTREE} -C -f ${.CURDIR}/etc/mtree/BSD.include.dist | \ sed -e 's#^\./#./${dist}/usr/include/#' >> ${METALOG} .if defined(_LIBCOMPAT) ${IMAKEENV} ${DISTR_MTREE} -C -f ${.CURDIR}/etc/mtree/BSD.lib${libcompat}.dist | \ sed -e 's#^\./#./${dist}/usr/#' >> ${METALOG} .endif .endif .endfor -mkdir ${DESTDIR}/${DISTDIR}/base ${_+_}cd ${.CURDIR}/etc; ${CROSSENV} PATH=${TMPPATH} ${MAKE} \ METALOG=${METALOG} ${IMAKE_INSTALL} ${IMAKE_MTREE} \ DISTBASE=/base DESTDIR=${DESTDIR}/${DISTDIR}/base \ LOCAL_MTREE=${LOCAL_MTREE:Q} distrib-dirs ${INSTALL_SYMLINK} ${INSTALLFLAGS} usr/src/sys ${INSTALL_DDIR}/base/sys .endif ${_+_}cd ${.CURDIR}; ${IMAKE} re${.TARGET:S/world$//}; \ ${IMAKEENV} rm -rf ${INSTALLTMP} .if make(distributeworld) .for dist in ${EXTRA_DISTRIBUTIONS} find ${DESTDIR}/${DISTDIR}/${dist} -mindepth 1 -type d -empty -delete .endfor .if defined(NO_ROOT) .for dist in base ${EXTRA_DISTRIBUTIONS} @# For each file that exists in this dist, print the corresponding @# line from the METALOG. This relies on the fact that @# a line containing only the filename will sort immediately before @# the relevant mtree line. cd ${DESTDIR}/${DISTDIR}; \ find ./${dist} | sort -u ${METALOG} - | \ awk 'BEGIN { print "#${MTREE_MAGIC}" } !/ type=/ { file = $$1 } / type=/ { if ($$1 == file) { sub(/^\.\/${dist}\//, "./"); print } }' > \ ${DESTDIR}/${DISTDIR}/${dist}.meta .endfor .for dist in ${DEBUG_DISTRIBUTIONS} @# For each file that exists in this dist, print the corresponding @# line from the METALOG. This relies on the fact that @# a line containing only the filename will sort immediately before @# the relevant mtree line. cd ${DESTDIR}/${DISTDIR}; \ find ./${dist}/usr/lib/debug | sort -u ${METALOG} - | \ awk 'BEGIN { print "#${MTREE_MAGIC}" } !/ type=/ { file = $$1 } / type=/ { if ($$1 == file) { sub(/^\.\/${dist}\//, "./"); print } }' > \ ${DESTDIR}/${DISTDIR}/${dist}.debug.meta .endfor .endif .endif packageworld: .PHONY .for dist in base ${EXTRA_DISTRIBUTIONS} .if defined(NO_ROOT) ${_+_}cd ${DESTDIR}/${DISTDIR}/${dist}; \ tar cvf - --exclude usr/lib/debug \ @${DESTDIR}/${DISTDIR}/${dist}.meta | \ ${XZ_CMD} > ${PACKAGEDIR}/${dist}.txz .else ${_+_}cd ${DESTDIR}/${DISTDIR}/${dist}; \ tar cvf - --exclude usr/lib/debug . | \ ${XZ_CMD} > ${PACKAGEDIR}/${dist}.txz .endif .endfor .for dist in ${DEBUG_DISTRIBUTIONS} . if defined(NO_ROOT) ${_+_}cd ${DESTDIR}/${DISTDIR}/${dist}; \ tar cvf - @${DESTDIR}/${DISTDIR}/${dist}.debug.meta | \ ${XZ_CMD} > ${PACKAGEDIR}/${dist}-dbg.txz . else ${_+_}cd ${DESTDIR}/${DISTDIR}/${dist}; \ tar cvLf - usr/lib/debug | \ ${XZ_CMD} > ${PACKAGEDIR}/${dist}-dbg.txz . endif .endfor makeman: .PHONY ${_+_}cd ${.CURDIR}/tools/build/options; sh makeman > \ ${.CURDIR}/share/man/man5/src.conf.5 # We can't assume here that ${TMPPATH} will include ${PATH} or /usr/libexec # because we may be building with a STRICTTMPPATH, so we explicitly include # /usr/libexec here for flua. ${TMPPATH} still usefully includes anything else # we may need to function. _sysent_PATH= ${TMPPATH}:/usr/libexec _sysent_dirs= sys/kern _sysent_dirs+= sys/compat/freebsd32 _sysent_dirs+= sys/compat/cloudabi32 \ sys/compat/cloudabi64 _sysent_dirs+= sys/amd64/linux \ sys/amd64/linux32 \ sys/arm/linux \ sys/arm64/linux \ sys/i386/linux sysent: .PHONY .for _dir in ${_sysent_dirs} @echo "${MAKE} -C ${.CURDIR}/${_dir} sysent" ${_+_}@env PATH=${_sysent_PATH} ${MAKE} -C ${.CURDIR}/${_dir} sysent .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. # restage reinstall: .MAKE .PHONY @echo "--------------------------------------------------------------" @echo ">>> Making hierarchy" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${MAKE} -f Makefile.inc1 \ LOCAL_MTREE=${LOCAL_MTREE:Q} hierarchy .if make(restage) @echo "--------------------------------------------------------------" @echo ">>> Making distribution" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${MAKE} -f Makefile.inc1 \ LOCAL_MTREE=${LOCAL_MTREE:Q} distribution .endif @echo @echo "--------------------------------------------------------------" @echo ">>> Installing everything started on `LC_ALL=C date`" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${MAKE} -f Makefile.inc1 install .if defined(_LIBCOMPAT) ${_+_}cd ${.CURDIR}; ${MAKE} -f Makefile.inc1 install${libcompat} .endif @echo "--------------------------------------------------------------" @echo ">>> Installing everything completed on `LC_ALL=C date`" @echo "--------------------------------------------------------------" redistribute: .MAKE .PHONY @echo "--------------------------------------------------------------" @echo ">>> Distributing everything" @echo "--------------------------------------------------------------" ${_+_}cd ${.CURDIR}; ${MAKE} -f Makefile.inc1 distribute .if defined(_LIBCOMPAT) ${_+_}cd ${.CURDIR}; ${MAKE} -f Makefile.inc1 distribute${libcompat} \ DISTRIBUTION=lib${libcompat} .endif distrib-dirs distribution: .MAKE .PHONY ${_+_}cd ${.CURDIR}/etc; ${CROSSENV} PATH=${TMPPATH} ${MAKE} \ ${IMAKE_INSTALL} ${IMAKE_MTREE} METALOG=${METALOG} ${.TARGET} .if make(distribution) ${_+_}cd ${.CURDIR}; ${CROSSENV} PATH=${TMPPATH} \ ${MAKE} -f Makefile.inc1 ${IMAKE_INSTALL} \ METALOG=${METALOG} MK_TESTS=no \ MK_TESTS_SUPPORT=${MK_TESTS_SUPPORT} installconfig .endif # # 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_KERNELOBJ= t # Shortcut for KERNCONF=Blah -DKERNFAST is now KERNFAST=Blah .if !defined(KERNCONF) && ${KERNFAST} != "1" KERNCONF=${KERNFAST} .endif .endif .if ${TARGET_ARCH} == "powerpc64" KERNCONF?= GENERIC64 .else KERNCONF?= GENERIC .endif INSTKERNNAME?= kernel KERNSRCDIR?= ${.CURDIR}/sys KRNLCONFDIR= ${KERNSRCDIR}/${TARGET}/conf KRNLOBJDIR= ${OBJTOP}${KERNSRCDIR:C,^${.CURDIR},,} KERNCONFDIR?= ${KRNLCONFDIR} BUILDKERNELS= INSTALLKERNEL= .if defined(NO_INSTALLKERNEL) # All of the BUILDKERNELS loops start at index 1. BUILDKERNELS+= dummy .endif .for _kernel in ${KERNCONF} .if !defined(_MKSHOWCONFIG) && exists(${KERNCONFDIR}/${_kernel}) BUILDKERNELS+= ${_kernel} .if empty(INSTALLKERNEL) && !defined(NO_INSTALLKERNEL) INSTALLKERNEL= ${_kernel} .endif .else .if make(buildkernel) .error Missing KERNCONF ${KERNCONFDIR}/${_kernel} .endif .endif .endfor _cleankernobj_fast_depend_hack: .PHONY # 20191009 r353340 removal of opensolaris_atomic.S (also r353381) .if ${MACHINE} != i386 .for f in opensolaris_atomic .for m in opensolaris zfs @if [ -e "${KRNLOBJDIR}/${KERNCONF}/modules${SRCTOP}/sys/modules/${m}/.depend.${f}.o" ] && \ grep -q ${f}.S "${KRNLOBJDIR}/${KERNCONF}/modules${SRCTOP}/sys/modules/${m}/.depend.${f}.o"; then \ echo "Removing stale dependencies for opensolaris_atomic"; \ rm -f ${KRNLOBJDIR}/${KERNCONF}/modules${SRCTOP}/sys/modules/${m}/.depend.${f}.*; \ fi .endfor .endfor .endif ${WMAKE_TGTS:N_worldtmp:Nbuild${libcompat}} ${.ALLTARGETS:M_*:N_worldtmp}: .MAKE .PHONY # record kernel(s) build time in seconds .if make(buildkernel) _BUILDKERNEL_START!= date '+%s' .endif # # buildkernel # # Builds all kernels defined by BUILDKERNELS. # buildkernel: .MAKE .PHONY .if empty(BUILDKERNELS:Ndummy) @echo "ERROR: Missing kernel configuration file(s) (${KERNCONF})."; \ false .endif @echo .for _kernel in ${BUILDKERNELS:Ndummy} @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} \ -I '${KERNCONFDIR}' -I '${KRNLCONFDIR}' \ '${KERNCONFDIR}/${_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} .else ${_+_}cd ${.CURDIR}; ${WMAKE} _cleankernobj_fast_depend_hack .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 ${.CURDIR}; ${KTMAKE} kernel-tools @echo @echo "--------------------------------------------------------------" @echo ">>> stage 3.1: 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 @seconds=$$(($$(date '+%s') - ${_BUILDKERNEL_START})); \ echo -n ">>> Kernel(s) ${BUILDKERNELS} built in $$seconds seconds, "; \ echo "ncpu: $$(sysctl -n hw.ncpu)${.MAKE.JOBS:S/^/, make -j/}" @echo "--------------------------------------------------------------" NO_INSTALLEXTRAKERNELS?= yes # # installkernel, etc. # # Install the kernel defined by INSTALLKERNEL # installkernel installkernel.debug \ reinstallkernel reinstallkernel.debug: _installcheck_kernel .PHONY .if !defined(NO_INSTALLKERNEL) .if empty(INSTALLKERNEL) @echo "ERROR: No kernel \"${KERNCONF}\" to install."; \ false .endif @echo "--------------------------------------------------------------" @echo ">>> Installing kernel ${INSTALLKERNEL} on $$(LC_ALL=C date)" @echo "--------------------------------------------------------------" ${_+_}cd ${KRNLOBJDIR}/${INSTALLKERNEL}; \ ${CROSSENV} PATH=${TMPPATH} \ ${MAKE} ${IMAKE_INSTALL} KERNEL=${INSTKERNNAME} ${.TARGET:S/kernel//} @echo "--------------------------------------------------------------" @echo ">>> Installing kernel ${INSTALLKERNEL} completed on $$(LC_ALL=C date)" @echo "--------------------------------------------------------------" .endif .if ${BUILDKERNELS:[#]} > 1 && ${NO_INSTALLEXTRAKERNELS} != "yes" .for _kernel in ${BUILDKERNELS:[2..-1]} @echo "--------------------------------------------------------------" @echo ">>> Installing kernel ${_kernel} $$(LC_ALL=C date)" @echo "--------------------------------------------------------------" ${_+_}cd ${KRNLOBJDIR}/${_kernel}; \ ${CROSSENV} PATH=${TMPPATH} \ ${MAKE} ${IMAKE_INSTALL} KERNEL=${INSTKERNNAME}.${_kernel} ${.TARGET:S/kernel//} @echo "--------------------------------------------------------------" @echo ">>> Installing kernel ${_kernel} completed on $$(LC_ALL=C date)" @echo "--------------------------------------------------------------" .endfor .endif distributekernel distributekernel.debug: .PHONY .if !defined(NO_INSTALLKERNEL) .if empty(INSTALLKERNEL) @echo "ERROR: No kernel \"${KERNCONF}\" to install."; \ false .endif mkdir -p ${DESTDIR}/${DISTDIR} .if defined(NO_ROOT) @echo "#${MTREE_MAGIC}" > ${DESTDIR}/${DISTDIR}/kernel.premeta .endif ${_+_}cd ${KRNLOBJDIR}/${INSTALLKERNEL}; \ ${IMAKEENV} ${IMAKE_INSTALL:S/METALOG/kernel.premeta/} \ ${IMAKE_MTREE} PATH=${TMPPATH} ${MAKE} KERNEL=${INSTKERNNAME} \ DESTDIR=${INSTALL_DDIR}/kernel \ ${.TARGET:S/distributekernel/install/} .if defined(NO_ROOT) @sed -e 's|^./kernel|.|' ${DESTDIR}/${DISTDIR}/kernel.premeta > \ ${DESTDIR}/${DISTDIR}/kernel.meta .endif .endif .if ${BUILDKERNELS:[#]} > 1 && ${NO_INSTALLEXTRAKERNELS} != "yes" .for _kernel in ${BUILDKERNELS:[2..-1]} .if defined(NO_ROOT) @echo "#${MTREE_MAGIC}" > ${DESTDIR}/${DISTDIR}/kernel.${_kernel}.premeta .endif ${_+_}cd ${KRNLOBJDIR}/${_kernel}; \ ${IMAKEENV} ${IMAKE_INSTALL:S/METALOG/kernel.${_kernel}.premeta/} \ ${IMAKE_MTREE} PATH=${TMPPATH} ${MAKE} \ KERNEL=${INSTKERNNAME}.${_kernel} \ DESTDIR=${INSTALL_DDIR}/kernel.${_kernel} \ ${.TARGET:S/distributekernel/install/} .if defined(NO_ROOT) @sed -e "s|^./kernel.${_kernel}|.|" \ ${DESTDIR}/${DISTDIR}/kernel.${_kernel}.premeta > \ ${DESTDIR}/${DISTDIR}/kernel.${_kernel}.meta .endif .endfor .endif packagekernel: .PHONY .if defined(NO_ROOT) .if !defined(NO_INSTALLKERNEL) cd ${DESTDIR}/${DISTDIR}/kernel; \ tar cvf - --exclude '*.debug' \ @${DESTDIR}/${DISTDIR}/kernel.meta | \ ${XZ_CMD} > ${PACKAGEDIR}/kernel.txz .endif .if ${MK_DEBUG_FILES} != "no" cd ${DESTDIR}/${DISTDIR}/kernel; \ tar cvf - --include '*/*/*.debug' \ @${DESTDIR}/${DISTDIR}/kernel.meta | \ ${XZ_CMD} > ${DESTDIR}/${DISTDIR}/kernel-dbg.txz .endif .if ${BUILDKERNELS:[#]} > 1 && ${NO_INSTALLEXTRAKERNELS} != "yes" .for _kernel in ${BUILDKERNELS:[2..-1]} cd ${DESTDIR}/${DISTDIR}/kernel.${_kernel}; \ tar cvf - --exclude '*.debug' \ @${DESTDIR}/${DISTDIR}/kernel.${_kernel}.meta | \ ${XZ_CMD} > ${PACKAGEDIR}/kernel.${_kernel}.txz .if ${MK_DEBUG_FILES} != "no" cd ${DESTDIR}/${DISTDIR}/kernel.${_kernel}; \ tar cvf - --include '*/*/*.debug' \ @${DESTDIR}/${DISTDIR}/kernel.${_kernel}.meta | \ ${XZ_CMD} > ${DESTDIR}/${DISTDIR}/kernel.${_kernel}-dbg.txz .endif .endfor .endif .else .if !defined(NO_INSTALLKERNEL) cd ${DESTDIR}/${DISTDIR}/kernel; \ tar cvf - --exclude '*.debug' . | \ ${XZ_CMD} > ${PACKAGEDIR}/kernel.txz .endif .if ${MK_DEBUG_FILES} != "no" cd ${DESTDIR}/${DISTDIR}/kernel; \ tar cvf - --include '*/*/*.debug' $$(eval find .) | \ ${XZ_CMD} > ${DESTDIR}/${DISTDIR}/kernel-dbg.txz .endif .if ${BUILDKERNELS:[#]} > 1 && ${NO_INSTALLEXTRAKERNELS} != "yes" .for _kernel in ${BUILDKERNELS:[2..-1]} cd ${DESTDIR}/${DISTDIR}/kernel.${_kernel}; \ tar cvf - --exclude '*.debug' . | \ ${XZ_CMD} > ${PACKAGEDIR}/kernel.${_kernel}.txz .if ${MK_DEBUG_FILES} != "no" cd ${DESTDIR}/${DISTDIR}/kernel.${_kernel}; \ tar cvf - --include '*/*/*.debug' $$(eval find .) | \ ${XZ_CMD} > ${DESTDIR}/${DISTDIR}/kernel.${_kernel}-dbg.txz .endif .endfor .endif .endif stagekernel: .PHONY ${_+_}${MAKE} -C ${.CURDIR} ${.MAKEFLAGS} distributekernel PORTSDIR?= /usr/ports WSTAGEDIR?= ${OBJTOP}/worldstage KSTAGEDIR?= ${OBJTOP}/kernelstage REPODIR?= ${OBJROOT}repo PKG_FORMAT?= txz PKGSIGNKEY?= # empty PKG_OUTPUT_DIR?= ${PKG_VERSION} .ORDER: stage-packages create-packages .ORDER: create-packages create-world-packages .ORDER: create-packages create-kernel-packages .ORDER: create-packages sign-packages _pkgbootstrap: .PHONY .if make(*package*) && !exists(${LOCALBASE}/sbin/pkg) @env ASSUME_ALWAYS_YES=YES pkg bootstrap .endif packages: .PHONY ${_+_}${MAKE} -C ${.CURDIR} PKG_VERSION=${PKG_VERSION} real-packages package-pkg: .PHONY rm -rf /tmp/ports.${TARGET} || : env ${WMAKEENV:Q} SRCDIR=${.CURDIR} PORTSDIR=${PORTSDIR} REVISION=${_REVISION} \ PKG_CMD=${PKG_CMD} PKG_VERSION=${PKG_VERSION} REPODIR=${REPODIR} \ WSTAGEDIR=${WSTAGEDIR} \ sh ${.CURDIR}/release/scripts/make-pkg-package.sh real-packages: stage-packages create-packages sign-packages .PHONY stage-packages-world: .PHONY @mkdir -p ${WSTAGEDIR} ${_+_}@cd ${.CURDIR}; \ ${MAKE} DESTDIR=${WSTAGEDIR} -DNO_ROOT stageworld stage-packages-kernel: .PHONY @mkdir -p ${KSTAGEDIR} ${_+_}@cd ${.CURDIR}; \ ${MAKE} DESTDIR=${KSTAGEDIR} -DNO_ROOT stagekernel stage-packages: .PHONY stage-packages-world stage-packages-kernel _repodir: .PHONY @mkdir -p ${REPODIR} create-packages-world: _pkgbootstrap _repodir .PHONY ${_+_}@cd ${.CURDIR}; \ ${MAKE} -f Makefile.inc1 \ DESTDIR=${WSTAGEDIR} \ PKG_VERSION=${PKG_VERSION} create-world-packages create-packages-kernel: _pkgbootstrap _repodir .PHONY ${_+_}@cd ${.CURDIR}; \ ${MAKE} -f Makefile.inc1 \ DESTDIR=${KSTAGEDIR} \ PKG_VERSION=${PKG_VERSION} DISTDIR=kernel \ SOURCE_DATE_EPOCH=${SOURCE_DATE_EPOCH} \ create-kernel-packages create-packages: .PHONY create-packages-world create-packages-kernel create-world-packages: _pkgbootstrap .PHONY @rm -f ${WSTAGEDIR}/*.plist 2>/dev/null || : @cd ${WSTAGEDIR} ; \ env -i LC_COLLATE=C sort ${WSTAGEDIR}/${DISTDIR}/METALOG | \ awk -f ${SRCDIR}/release/scripts/mtree-to-plist.awk @for plist in ${WSTAGEDIR}/*.plist; do \ plist=$${plist##*/} ; \ pkgname=$${plist%.plist} ; \ echo "_PKGS+= $${pkgname}" ; \ done > ${WSTAGEDIR}/packages.mk ${_+_}@cd ${.CURDIR}; \ ${MAKE} -f Makefile.inc1 create-world-packages-jobs \ SOURCE_DATE_EPOCH=${SOURCE_DATE_EPOCH} \ .MAKE.JOB.PREFIX= .if make(create-world-packages-jobs) .include "${WSTAGEDIR}/packages.mk" .endif .if make(create-world-packages-jobs) || make(create-kernel-packages*) PKG_ABI!=${PKG_CMD} -o ABI_FILE=${WSTAGEDIR}/usr/bin/uname config ABI .endif create-world-packages-jobs: .PHONY .for pkgname in ${_PKGS} create-world-packages-jobs: create-world-package-${pkgname} create-world-package-${pkgname}: .PHONY @sh ${SRCDIR}/release/packages/generate-ucl.sh -o ${pkgname} \ -s ${SRCDIR} -u ${WSTAGEDIR}/${pkgname}.ucl @awk -F\" ' \ /^name/ { printf("===> Creating %s-", $$2); next } \ /^version/ { print $$2; next } \ ' ${WSTAGEDIR}/${pkgname}.ucl @if [ "${pkgname}" == "runtime" ]; then \ sed -i '' -e "s/%VCS_REVISION%/${VCS_REVISION}/" ${WSTAGEDIR}/${pkgname}.ucl ; \ fi ${PKG_CMD} -o ABI_FILE=${WSTAGEDIR}/usr/bin/uname -o ALLOW_BASE_SHLIBS=yes \ create -f ${PKG_FORMAT} -M ${WSTAGEDIR}/${pkgname}.ucl \ -p ${WSTAGEDIR}/${pkgname}.plist \ -r ${WSTAGEDIR} \ -o ${REPODIR}/${PKG_ABI}/${PKG_OUTPUT_DIR} .endfor _default_flavor= -default .if make(*package*) && exists(${KSTAGEDIR}/kernel.meta) . if ${MK_DEBUG_FILES} != "no" _debug=-dbg . endif create-kernel-packages: .PHONY . for flavor in "" ${_debug} create-kernel-packages: create-kernel-packages-flavor${flavor:C,^""$,${_default_flavor},} create-kernel-packages-flavor${flavor:C,^""$,${_default_flavor},}: _pkgbootstrap .PHONY @cd ${KSTAGEDIR}/${DISTDIR} ; \ env -i LC_COLLATE=C sort ${KSTAGEDIR}/kernel.meta | \ awk -f ${SRCDIR}/release/scripts/mtree-to-plist.awk \ -v kernel=yes -v _kernconf=${INSTALLKERNEL} ; \ sed -e "s/%VERSION%/${PKG_VERSION}/" \ -e "s/%PKGNAME%/kernel-${INSTALLKERNEL:tl}${flavor}/" \ -e "s/%KERNELDIR%/kernel/" \ -e "s/%COMMENT%/FreeBSD ${INSTALLKERNEL} kernel ${flavor}/" \ -e "s/%DESC%/FreeBSD ${INSTALLKERNEL} kernel ${flavor}/" \ -e "s/ %VCS_REVISION%/${VCS_REVISION}/" \ ${SRCDIR}/release/packages/kernel.ucl \ > ${KSTAGEDIR}/${DISTDIR}/kernel.${INSTALLKERNEL}${flavor}.ucl ; \ awk -F\" ' \ /name/ { printf("===> Creating %s-", $$2); next } \ /version/ {print $$2; next } ' \ ${KSTAGEDIR}/${DISTDIR}/kernel.${INSTALLKERNEL}${flavor}.ucl ; \ ${PKG_CMD} -o ABI_FILE=${WSTAGEDIR}/usr/bin/uname -o ALLOW_BASE_SHLIBS=yes \ create -f ${PKG_FORMAT} \ -M ${KSTAGEDIR}/${DISTDIR}/kernel.${INSTALLKERNEL}${flavor}.ucl \ -p ${KSTAGEDIR}/${DISTDIR}/kernel.${INSTALLKERNEL}${flavor}.plist \ -r ${KSTAGEDIR}/${DISTDIR} \ -o ${REPODIR}/${PKG_ABI}/${PKG_OUTPUT_DIR} . endfor .endif .if ${BUILDKERNELS:[#]} > 1 && ${NO_INSTALLEXTRAKERNELS} != "yes" . for _kernel in ${BUILDKERNELS:[2..-1]} . if exists(${KSTAGEDIR}/kernel.${_kernel}.meta) . if ${MK_DEBUG_FILES} != "no" _debug=-dbg . endif . for flavor in "" ${_debug} create-kernel-packages: create-kernel-packages-extra-flavor${flavor:C,^""$,${_default_flavor},}-${_kernel} create-kernel-packages-extra-flavor${flavor:C,^""$,${_default_flavor},}-${_kernel}: _pkgbootstrap .PHONY @cd ${KSTAGEDIR}/kernel.${_kernel} ; \ env -i LC_COLLATE=C sort ${KSTAGEDIR}/kernel.${_kernel}.meta | \ awk -f ${SRCDIR}/release/scripts/mtree-to-plist.awk \ -v kernel=yes -v _kernconf=${_kernel} ; \ sed -e "s/%VERSION%/${PKG_VERSION}/" \ -e "s/%PKGNAME%/kernel-${_kernel:tl}${flavor}/" \ -e "s/%KERNELDIR%/kernel.${_kernel}/" \ -e "s/%COMMENT%/FreeBSD ${_kernel} kernel ${flavor}/" \ -e "s/%DESC%/FreeBSD ${_kernel} kernel ${flavor}/" \ -e "s/ %VCS_REVISION%/${VCS_REVISION}/" \ ${SRCDIR}/release/packages/kernel.ucl \ > ${KSTAGEDIR}/kernel.${_kernel}/kernel.${_kernel}${flavor}.ucl ; \ awk -F\" ' \ /name/ { printf("===> Creating %s-", $$2); next } \ /version/ {print $$2; next } ' \ ${KSTAGEDIR}/kernel.${_kernel}/kernel.${_kernel}${flavor}.ucl ; \ ${PKG_CMD} -o ABI_FILE=${WSTAGEDIR}/usr/bin/uname -o ALLOW_BASE_SHLIBS=yes \ create -f ${PKG_FORMAT} \ -M ${KSTAGEDIR}/kernel.${_kernel}/kernel.${_kernel}${flavor}.ucl \ -p ${KSTAGEDIR}/kernel.${_kernel}/kernel.${_kernel}${flavor}.plist \ -r ${KSTAGEDIR}/kernel.${_kernel} \ -o ${REPODIR}/${PKG_ABI}/${PKG_OUTPUT_DIR} . endfor . endif . endfor .endif sign-packages: _pkgbootstrap .PHONY printf "version = 2;\npacking_format = \"${PKG_FORMAT}\";\n" > ${WSTAGEDIR}/meta @[ -L "${REPODIR}/$$(${PKG_CMD} -o ABI_FILE=${WSTAGEDIR}/usr/bin/uname config ABI)/latest" ] && \ unlink ${REPODIR}/$$(${PKG_CMD} -o ABI_FILE=${WSTAGEDIR}/usr/bin/uname config ABI)/latest ; \ ${PKG_CMD} -o ABI_FILE=${WSTAGEDIR}/usr/bin/uname repo \ -m ${WSTAGEDIR}/meta \ -o ${REPODIR}/$$(${PKG_CMD} -o ABI_FILE=${WSTAGEDIR}/usr/bin/uname config ABI)/${PKG_VERSION} \ ${REPODIR}/$$(${PKG_CMD} -o ABI_FILE=${WSTAGEDIR}/usr/bin/uname config ABI)/${PKG_VERSION} \ ${PKGSIGNKEY} ; \ cd ${REPODIR}/$$(${PKG_CMD} -o ABI_FILE=${WSTAGEDIR}/usr/bin/uname config ABI); \ ln -s ${PKG_OUTPUT_DIR} latest # # # checkworld # # Run test suite on installed world. # checkworld: .PHONY @if [ ! -x "${LOCALBASE}/bin/kyua" ] && [ ! -x "/usr/bin/kyua" ]; then \ echo "You need kyua (devel/kyua) to run the test suite." | /usr/bin/fmt; \ exit 1; \ fi ${_+_}PATH="$$PATH:${LOCALBASE}/bin" kyua test -k ${TESTSBASE}/Kyuafile # # # doxygen # # Build the API documentation with doxygen # doxygen: .PHONY @if [ ! -x "${LOCALBASE}/bin/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(s), by running svn/svnup to update to the # latest copy. # update: .PHONY .if defined(SVN_UPDATE) @echo "--------------------------------------------------------------" @echo ">>> Updating ${.CURDIR} using Subversion" @echo "--------------------------------------------------------------" @(cd ${.CURDIR}; ${SVN_CMD} update ${SVNFLAGS}) .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. This is a # minimal set of tools and shims necessary to compensate for older systems # which don't have the APIs required by the targets built in bootstrap-tools, # build-tools or cross-tools. # # libnv is a requirement for config(8), which is an unconditional # bootstrap-tool. _config_deps= lib/libnv legacy: .PHONY .if ${BOOTSTRAPPING} < ${MINIMUM_SUPPORTED_OSREL} && ${BOOTSTRAPPING} != 0 @echo "ERROR: Source upgrades from versions prior to ${MINIMUM_SUPPORTED_REL} are not supported."; \ false .endif .for _tool in tools/build ${_config_deps} ${_+_}@${ECHODIR} "===> ${_tool} (obj,includes,all,install)"; \ cd ${.CURDIR}/${_tool}; \ if [ -z "${NO_OBJWALK}" ]; then ${MAKE} DIRPRFX=${_tool}/ obj; fi; \ ${MAKE} DIRPRFX=${_tool}/ DESTDIR=${WORLDTMP}/legacy includes; \ ${MAKE} DIRPRFX=${_tool}/ MK_INCLUDES=no all; \ ${MAKE} DIRPRFX=${_tool}/ MK_INCLUDES=no \ DESTDIR=${WORLDTMP}/legacy install .endfor # # bootstrap-tools: Build tools needed for compatibility. These are binaries that # are built to build other binaries in the system. However, the focus of these # binaries is usually quite narrow. Bootstrap tools use the host's compiler and # libraries, augmented by -legacy, in addition to the libraries built during # bootstrap-tools. # _bt= _bootstrap-tools # We want to run the build with only ${WORLDTMP} in $PATH to ensure we don't # accidentally run tools that are incompatible but happen to be in $PATH. # This is especially important when building on Linux/MacOS where many of the # programs used during the build accept different flags or generate different # output. On those platforms we only symlink the tools known to be compatible # (e.g. basic utilities such as mkdir) into ${WORLDTMP} and build all others # from the FreeBSD sources during the bootstrap-tools stage. # We want to build without the user's $PATH starting in the bootstrap-tools # phase so the tools used in that phase (ln, cp, etc) must have already been # linked to $WORLDTMP. The tools are listed in the _host_tools_to_symlink # variable in tools/build/Makefile and are linked during the legacy phase. # Since they could be Linux or MacOS binaries, too we must only use flags that # are portable across operating systems. # If BOOTSTRAP_ALL_TOOLS is set we will build all the required tools from the # current source tree. Otherwise we create a symlink to the version found in # $PATH during the bootstrap-tools stage. .if defined(BOOTSTRAP_ALL_TOOLS) # BOOTSTRAPPING will be set on the command line so we can't override it here. # Instead set BOOTSTRAPPING_OSRELDATE so that the value 0 is set ${BSARGS} BOOTSTRAPPING_OSRELDATE:= 0 .endif .if ${MK_GAMES} != "no" _strfile= usr.bin/fortune/strfile .endif .if ${MK_VT} != "no" _vtfontcvt= usr.bin/vtfontcvt .endif # If we are not building the bootstrap because BOOTSTRAPPING is sufficient # we symlink the host version to $WORLDTMP instead. By doing this we can also # detect when a bootstrap tool is being used without the required MK_FOO. # If you add a new bootstrap tool where we could also use the host version, # please ensure that you also add a .else case where you add the tool to the # _bootstrap_tools_links variable. .if ${BOOTSTRAPPING} < 1000033 _m4= usr.bin/m4 _lex= usr.bin/lex # Note: lex needs m4 to build but m4 also depends on lex. However, lex can be # bootstrapped so we build lex first. ${_bt}-usr.bin/m4: ${_bt}-lib/libopenbsd ${_bt}-usr.bin/yacc ${_bt}-${_lex} _bt_m4_depend=${_bt}-${_m4} _bt_lex_depend=${_bt}-${_lex} ${_bt_m4_depend} .else _bootstrap_tools_links+=m4 lex .endif # ELF Tool Chain libraries are needed for ELF tools and dtrace tools. # r296685 fix cross-endian objcopy # r310724 fixed PR 215350, a crash in libdwarf with objects built by GCC 6.2. # r334881 added libdwarf constants used by ctfconvert. # r338478 fixed a crash in objcopy for mips64el objects # r339083 libelf: correct mips64el test to use ELF header # r348347 Add missing powerpc64 relocation support to libdwarf .if ${BOOTSTRAPPING} < 1300030 _elftoolchain_libs= lib/libelf lib/libdwarf lib/libzstd ${_bt}-lib/libelf: ${_bt_m4_depend} ${_bt}-lib/libdwarf: ${_bt_m4_depend} .endif # flua is required to regenerate syscall files. It first appeared during the # 13.0-CURRENT cycle, thus needs to be built on -older releases and stable # branches. .if ${BOOTSTRAPPING} < 1300059 ${_bt}-libexec/flua: ${_bt}-lib/liblua ${_bt}-lib/libucl _flua= lib/liblua lib/libucl libexec/flua .endif # r245440 mtree -N support added # r313404 requires sha384.h for libnetbsd, added to libmd in r292782 .if ${BOOTSTRAPPING} < 1100093 _nmtree= lib/libmd \ lib/libnetbsd \ usr.sbin/nmtree ${_bt}-lib/libnetbsd: ${_bt}-lib/libmd ${_bt}-usr.sbin/nmtree: ${_bt}-lib/libnetbsd .else _bootstrap_tools_links+=mtree .endif # r246097: log addition login.conf.db, passwd, pwd.db, and spwd.db with cat -l .if ${BOOTSTRAPPING} < 1000027 _cat= bin/cat .else _bootstrap_tools_links+=cat .endif # r277259 crunchide: Correct 64-bit section header offset # r281674 crunchide: always include both 32- and 64-bit ELF support .if ${BOOTSTRAPPING} < 1100078 _crunchide= usr.sbin/crunch/crunchide .else _bootstrap_tools_links+=crunchide .endif # r285986 crunchen: use STRIPBIN rather than STRIP # 1100113: Support MK_AUTO_OBJ # 1200006: META_MODE fixes .if ${BOOTSTRAPPING} < 1100078 || \ (${MK_AUTO_OBJ} == "yes" && ${BOOTSTRAPPING} < 1100114) || \ (${MK_META_MODE} == "yes" && ${BOOTSTRAPPING} < 1200006) _crunchgen= usr.sbin/crunch/crunchgen .else _bootstrap_tools_links+=crunchgen .endif # r296926 -P keymap search path, MFC to stable/10 in r298297 .if ${BOOTSTRAPPING} < 1003501 || \ (${BOOTSTRAPPING} >= 1100000 && ${BOOTSTRAPPING} < 1100103) _kbdcontrol= usr.sbin/kbdcontrol .else _bootstrap_tools_links+=kbdcontrol .endif _yacc= usr.bin/yacc .if ${MK_BSNMP} != "no" _gensnmptree= usr.sbin/bsnmpd/gensnmptree .endif -.if ${MK_LOCALES} != "no" -_localedef= usr.bin/localedef -.endif # We need to build tblgen when we're building clang or lld, either as # bootstrap tools, or as the part of the normal build. .if ${MK_CLANG_BOOTSTRAP} != "no" || ${MK_CLANG} != "no" || \ ${MK_LLD_BOOTSTRAP} != "no" || ${MK_LLD} != "no" _clang_tblgen= \ lib/clang/libllvmminimal \ usr.bin/clang/llvm-tblgen \ usr.bin/clang/clang-tblgen \ usr.bin/clang/lldb-tblgen # XXX: lldb-tblgen is not needed, if top-level MK_LLDB=no ${_bt}-usr.bin/clang/clang-tblgen: ${_bt}-lib/clang/libllvmminimal ${_bt}-usr.bin/clang/llvm-tblgen: ${_bt}-lib/clang/libllvmminimal ${_bt}-usr.bin/clang/lldb-tblgen: ${_bt}-lib/clang/libllvmminimal .endif .if ${MK_LOCALES} != "no" _localedef= usr.bin/localedef .endif .if ${MK_KERBEROS} != "no" _kerberos5_bootstrap_tools= \ kerberos5/tools/make-roken \ kerberos5/lib/libroken \ kerberos5/lib/libvers \ kerberos5/tools/asn1_compile \ kerberos5/tools/slc \ usr.bin/compile_et .ORDER: ${_kerberos5_bootstrap_tools:C/^/${_bt}-/g} .for _tool in ${_kerberos5_bootstrap_tools} ${_bt}-${_tool}: ${_bt}-usr.bin/yacc ${_bt_lex_depend} .endfor .endif ${_bt}-usr.bin/mandoc: ${_bt}-lib/libopenbsd # The tools listed in _basic_bootstrap_tools will generally not be # bootstrapped unless BOOTSTRAP_ALL_TOOL is set. However, when building on a # Linux or MacOS host the host versions are incompatible so we need to build # them from the source tree. Usually the link name will be the same as the subdir, # but some directories such as grep or test install multiple binaries. In that # case we use the _basic_bootstrap_tools_multilink variable which is a list of # subdirectory and comma-separated list of files. _basic_bootstrap_tools_multilink=usr.bin/grep grep,egrep,fgrep _basic_bootstrap_tools_multilink+=bin/test test,[ # bootstrap tools needed by buildworld: _basic_bootstrap_tools=usr.bin/awk usr.bin/cut bin/expr usr.bin/gencat \ usr.bin/join usr.bin/mktemp bin/rmdir usr.bin/sed usr.bin/sort \ usr.bin/truncate usr.bin/tsort # file2c is required for building usr.sbin/config: _basic_bootstrap_tools+=usr.bin/file2c # uuencode/uudecode required for share/tabset _basic_bootstrap_tools+=usr.bin/uuencode usr.bin/uudecode # xargs is required by mkioctls _basic_bootstrap_tools+=usr.bin/xargs # cap_mkdb is required for share/termcap: _basic_bootstrap_tools+=usr.bin/cap_mkdb # ldd is required for installcheck (TODO: just always use /usr/bin/ldd instead?) _basic_bootstrap_tools+=usr.bin/ldd # services_mkdb/pwd_mkdb are required for installworld: _basic_bootstrap_tools+=usr.sbin/services_mkdb usr.sbin/pwd_mkdb # sysctl/chflags are required for installkernel: _basic_bootstrap_tools+=sbin/sysctl bin/chflags # mkfifo is used by sys/conf/newvers.sh _basic_bootstrap_tools+=usr.bin/mkfifo .if ${MK_BOOT} != "no" _basic_bootstrap_tools+=bin/dd # xz/unxz is used by EFI _basic_bootstrap_tools_multilink+=usr.bin/xz xz,unxz # md5 is used by boot/beri (and possibly others) _basic_bootstrap_tools+=sbin/md5 .if defined(BOOTSTRAP_ALL_TOOLS) ${_bt}-sbin/md5: ${_bt}-lib/libmd .endif .endif .if ${MK_ZONEINFO} != "no" _basic_bootstrap_tools+=usr.sbin/zic usr.sbin/tzsetup .endif .if defined(BOOTSTRAP_ALL_TOOLS) _other_bootstrap_tools+=${_basic_bootstrap_tools} .for _subdir _links in ${_basic_bootstrap_tools_multilink} _other_bootstrap_tools+=${_subdir} .endfor ${_bt}-usr.bin/awk: ${_bt_lex_depend} ${_bt}-usr.bin/yacc ${_bt}-bin/expr: ${_bt_lex_depend} ${_bt}-usr.bin/yacc # If we are bootstrapping file2c, we have to build it before config: ${_bt}-usr.sbin/config: ${_bt}-usr.bin/file2c ${_bt_lex_depend} # Note: no symlink to make/bmake in the !BOOTSTRAP_ALL_TOOLS case here since # the links to make/bmake make links will have already have been created in the # `make legacy` step. Not adding a link to make is important on non-FreeBSD # since "make" will usually point to GNU make there. _other_bootstrap_tools+=usr.bin/bmake .else # All tools in _basic_bootstrap_tools have the same name as the subdirectory # so we can use :T to get the name of the symlinks that we need to create. _bootstrap_tools_links+=${_basic_bootstrap_tools:T} .for _subdir _links in ${_basic_bootstrap_tools_multilink} _bootstrap_tools_links+=${_links:S/,/ /g} .endfor .endif # defined(BOOTSTRAP_ALL_TOOLS) # Link the tools that we need for building but don't need to bootstrap because # the host version is known to be compatible into ${WORLDTMP}/legacy # We do this before building any of the bootstrap tools in case they depend on # the presence of any of the links (e.g. as m4/lex/awk) ${_bt}-links: .PHONY .for _tool in ${_bootstrap_tools_links} ${_bt}-link-${_tool}: .PHONY .MAKE - @if [ ! -e "${WORLDTMP}/legacy/bin/${_tool}" ]; then \ - source_path=`which ${_tool}`; \ - if [ ! -e "$${source_path}" ] ; then \ - echo "Cannot find host tool '${_tool}'"; false; \ - fi; \ - ln -sfnv "$${source_path}" "${WORLDTMP}/legacy/bin/${_tool}"; \ - fi + @rm -f "${WORLDTMP}/legacy/bin/${_tool}"; \ + source_path=`which ${_tool}`; \ + if [ ! -e "$${source_path}" ] ; then \ + echo "Cannot find host tool '${_tool}'"; false; \ + fi; \ + cp -f "$${source_path}" "${WORLDTMP}/legacy/bin/${_tool}" ${_bt}-links: ${_bt}-link-${_tool} .endfor bootstrap-tools: ${_bt}-links .PHONY # Please document (add comment) why something is in 'bootstrap-tools'. # Try to bound the building of the bootstrap-tool to just the # FreeBSD versions that need the tool built at this stage of the build. .for _tool in \ ${_clang_tblgen} \ ${_kerberos5_bootstrap_tools} \ ${_strfile} \ usr.bin/dtc \ ${_cat} \ ${_kbdcontrol} \ ${_elftoolchain_libs} \ usr.bin/lorder \ lib/libopenbsd \ usr.bin/mandoc \ usr.bin/rpcgen \ ${_yacc} \ ${_m4} \ ${_lex} \ ${_other_bootstrap_tools} \ usr.bin/xinstall \ ${_gensnmptree} \ usr.sbin/config \ ${_flua} \ ${_crunchide} \ ${_crunchgen} \ ${_nmtree} \ ${_vtfontcvt} \ ${_localedef} ${_bt}-${_tool}: ${_bt}-links .PHONY .MAKE ${_+_}@${ECHODIR} "===> ${_tool} (obj,all,install)"; \ cd ${.CURDIR}/${_tool}; \ if [ -z "${NO_OBJWALK}" ]; then ${MAKE} DIRPRFX=${_tool}/ obj; fi; \ if [ "${_tool}" = "usr.bin/lex" ]; then \ ${MAKE} DIRPRFX=${_tool}/ bootstrap; \ fi; \ ${MAKE} DIRPRFX=${_tool}/ all; \ ${MAKE} DIRPRFX=${_tool}/ DESTDIR=${WORLDTMP}/legacy install bootstrap-tools: ${_bt}-${_tool} .endfor # # build-tools: Build special purpose build tools # .if !defined(NO_SHARE) && ${MK_SYSCONS} != "no" _share= share/syscons/scrnmaps .endif .if ${MK_RESCUE} != "no" # rescue includes programs that have build-tools targets _rescue=rescue/rescue .endif .if ${MK_TCSH} != "no" _tcsh=bin/csh .endif .if ${MK_FILE} != "no" _libmagic=lib/libmagic .endif .if ${MK_PMC} != "no" && \ (${TARGET_ARCH} == "aarch64" || ${TARGET_ARCH} == "amd64" || \ ${TARGET_ARCH} == "i386") _jevents=lib/libpmc/pmu-events .endif # kernel-toolchain skips _cleanobj, so handle cleaning up previous # build-tools directories if needed. .if !defined(NO_CLEAN) && make(kernel-toolchain) _bt_clean= ${CLEANDIR} .endif .for _tool in \ ${_tcsh} \ bin/sh \ ${LOCAL_TOOL_DIRS} \ ${_jevents} \ lib/ncurses/ncurses \ lib/ncurses/ncursesw \ ${_rescue} \ ${_share} \ usr.bin/awk \ ${_libmagic} \ usr.bin/mkesdb_static \ usr.bin/mkcsmapper_static \ usr.bin/vi/catalog build-tools_${_tool}: .PHONY ${_+_}@${ECHODIR} "===> ${_tool} (${_bt_clean:D${_bt_clean},}obj,build-tools)"; \ cd ${.CURDIR}/${_tool}; \ if [ -n "${_bt_clean}" ]; then ${MAKE} DIRPRFX=${_tool}/ ${_bt_clean}; fi; \ if [ -z "${NO_OBJWALK}" ]; then ${MAKE} DIRPRFX=${_tool}/ obj; fi; \ ${MAKE} DIRPRFX=${_tool}/ build-tools build-tools: build-tools_${_tool} .endfor # # kernel-tools: Build kernel-building tools # kernel-tools: .PHONY mkdir -p ${WORLDTMP}/usr ${WORLDTMP_MTREE} -f ${.CURDIR}/etc/mtree/BSD.usr.dist \ -p ${WORLDTMP}/usr >/dev/null # # cross-tools: All the tools needed to build the rest of the system after # we get done with the earlier stages. It is the last set of tools needed # to begin building the target binaries. # .if ${TARGET_ARCH} != ${MACHINE_ARCH} || ${BUILD_WITH_STRICT_TMPPATH} != 0 .if ${TARGET_ARCH} == "amd64" || ${TARGET_ARCH} == "i386" _btxld= usr.sbin/btxld .endif .endif # Rebuild ctfconvert and ctfmerge to avoid difficult-to-diagnose failures # resulting from missing bug fixes or ELF Toolchain updates. .if ${MK_CDDL} != "no" _dtrace_tools= cddl/lib/libctf cddl/usr.bin/ctfconvert \ cddl/usr.bin/ctfmerge .endif # If we're given an XAS, don't build binutils. .if ${XAS:M/*} == "" .if ${MK_ELFTOOLCHAIN_BOOTSTRAP} != "no" _elftctools= lib/libelftc \ lib/libpe \ usr.bin/objcopy \ usr.bin/nm \ usr.bin/size \ usr.bin/strings # These are not required by the build, but can be useful for developers who # cross-build on a FreeBSD 10 host: _elftctools+= usr.bin/addr2line .endif .elif ${TARGET_ARCH} != ${MACHINE_ARCH} && ${MK_ELFTOOLCHAIN_BOOTSTRAP} != "no" # If cross-building with an external binutils we still need to build strip for # the target (for at least crunchide). _elftctools= lib/libelftc \ lib/libpe \ usr.bin/objcopy .endif .if ${MK_CLANG_BOOTSTRAP} != "no" _clang= usr.bin/clang .endif .if ${MK_LLD_BOOTSTRAP} != "no" _lld= usr.bin/clang/lld .endif .if ${MK_CLANG_BOOTSTRAP} != "no" || ${MK_LLD_BOOTSTRAP} != "no" _clang_libs= lib/clang .endif .if ${MK_USB} != "no" _usb_tools= stand/usb/tools .endif .if ${BUILD_WITH_STRICT_TMPPATH} != 0 || defined(BOOTSTRAP_ALL_TOOLS) _ar=usr.bin/ar .endif cross-tools: .MAKE .PHONY .for _tool in \ ${LOCAL_XTOOL_DIRS} \ ${_ar} \ ${_clang_libs} \ ${_clang} \ ${_lld} \ ${_elftctools} \ ${_dtrace_tools} \ ${_btxld} \ ${_usb_tools} ${_+_}@${ECHODIR} "===> ${_tool} (obj,all,install)"; \ cd ${.CURDIR}/${_tool}; \ if [ -z "${NO_OBJWALK}" ]; then ${MAKE} DIRPRFX=${_tool}/ obj; fi; \ ${MAKE} DIRPRFX=${_tool}/ all; \ ${MAKE} DIRPRFX=${_tool}/ DESTDIR=${WORLDTMP} install .endfor # # native-xtools is the current target for qemu-user cross builds of ports # via poudriere and the imgact_binmisc kernel module. # This target merely builds a toolchan/sysroot, then builds the tools it wants # with the options it wants in a special MAKEOBJDIRPREFIX, using the toolchain # already built. It then installs the static tools to NXBDESTDIR for Poudriere # to pickup. # NXBOBJROOT= ${OBJROOT}${MACHINE}.${MACHINE_ARCH}/nxb/ NXBOBJTOP= ${NXBOBJROOT}${NXB_TARGET}.${NXB_TARGET_ARCH} NXTP?= /nxb-bin .if ${NXTP:N/*} .error NXTP variable should be an absolute path .endif NXBDESTDIR?= ${DESTDIR}${NXTP} # This is the list of tools to be built/installed as static and where # appropriate to build for the given TARGET.TARGET_ARCH. NXBDIRS+= \ bin/cat \ bin/chmod \ bin/cp \ ${_tcsh} \ bin/echo \ bin/expr \ bin/hostname \ bin/ln \ bin/ls \ bin/mkdir \ bin/mv \ bin/ps \ bin/realpath \ bin/rm \ bin/rmdir \ bin/sh \ bin/sleep \ sbin/md5 \ sbin/sysctl \ usr.bin/addr2line \ usr.bin/ar \ usr.bin/awk \ usr.bin/basename \ usr.bin/bmake \ usr.bin/bzip2 \ usr.bin/cmp \ usr.bin/diff \ usr.bin/dirname \ usr.bin/objcopy \ usr.bin/env \ usr.bin/fetch \ usr.bin/find \ usr.bin/grep \ usr.bin/gzip \ usr.bin/head \ usr.bin/id \ usr.bin/lex \ usr.bin/limits \ usr.bin/lorder \ usr.bin/mandoc \ usr.bin/mktemp \ usr.bin/mt \ usr.bin/nm \ usr.bin/patch \ usr.bin/readelf \ usr.bin/sed \ usr.bin/size \ usr.bin/sort \ usr.bin/strings \ usr.bin/tar \ usr.bin/touch \ usr.bin/tr \ usr.bin/true \ usr.bin/uniq \ usr.bin/unzip \ usr.bin/wc \ usr.bin/xargs \ usr.bin/xinstall \ usr.bin/xz \ usr.bin/yacc \ usr.sbin/chown SUBDIR_DEPEND_usr.bin/clang= lib/clang .if ${MK_CLANG} != "no" NXBDIRS+= lib/clang NXBDIRS+= usr.bin/clang .endif # XXX: native-xtools passes along ${NXBDIRS} in SUBDIR_OVERRIDE that needs # to be evaluated after NXBDIRS is set. .if make(install) && !empty(SUBDIR_OVERRIDE) SUBDIR= ${SUBDIR_OVERRIDE} .endif NXBMAKEARGS+= \ OBJTOP=${NXBOBJTOP:Q} \ OBJROOT=${NXBOBJROOT:Q} \ MAKEOBJDIRPREFIX= \ -DNO_SHARED \ -DNO_CPU_CFLAGS \ -DNO_PIC \ SSP_CFLAGS= \ MK_CASPER=no \ MK_CLANG_EXTRAS=no \ MK_CLANG_FORMAT=no \ MK_CLANG_FULL=no \ MK_CTF=no \ MK_DEBUG_FILES=no \ MK_GDB=no \ MK_HTML=no \ MK_LLDB=no \ MK_MAN=no \ MK_MAN_UTILS=yes \ MK_OFED=no \ MK_OPENSSH=no \ MK_PROFILE=no \ MK_RETPOLINE=no \ MK_SENDMAIL=no \ MK_SVNLITE=no \ MK_TESTS=no \ MK_WARNS=no \ MK_ZFS=no .if make(native-xtools*) && \ (!defined(NXB_TARGET) || !defined(NXB_TARGET_ARCH)) .error Missing NXB_TARGET / NXB_TARGET_ARCH .endif # For 'toolchain' we want to produce native binaries that themselves generate # native binaries. NXBTMAKE= ${NXBMAKEENV} ${MAKE} ${NXBMAKEARGS:N-DNO_PIC:N-DNO_SHARED} \ TARGET=${MACHINE} TARGET_ARCH=${MACHINE_ARCH} # For 'everything' we want to produce native binaries (hence -target to # be MACHINE) that themselves generate TARGET.TARGET_ARCH binaries. # TARGET/TARGET_ARCH are still passed along from user. # # Use the toolchain we create as an external toolchain. .if ${USING_SYSTEM_COMPILER} == "yes" || ${XCC:N${CCACHE_BIN}:M/*} NXBMAKE+= XCC="${XCC}" \ XCXX="${XCXX}" \ XCPP="${XCPP}" .else NXBMAKE+= XCC="${NXBOBJTOP}/tmp/usr/bin/cc" \ XCXX="${NXBOBJTOP}/tmp/usr/bin/c++" \ XCPP="${NXBOBJTOP}/tmp/usr/bin/cpp" .endif NXBMAKE+= ${NXBMAKEENV} ${MAKE} -f Makefile.inc1 ${NXBMAKEARGS} \ MACHINE=${MACHINE} MACHINE_ARCH=${MACHINE_ARCH} \ TARGET=${NXB_TARGET} TARGET_ARCH=${NXB_TARGET_ARCH} \ TARGET_TRIPLE=${MACHINE_TRIPLE:Q} # NXBDIRS is improperly based on MACHINE rather than NXB_TARGET. Need to # invoke a sub-make to reevaluate MK_CLANG, etc, for NXBDIRS. NXBMAKE+= SUBDIR_OVERRIDE='$${NXBDIRS:M*}' # Need to avoid the -isystem logic when using clang as an external toolchain # even if the TARGET being built for wants GCC. NXBMAKE+= WANT_COMPILER_TYPE='$${X_COMPILER_TYPE}' native-xtools: .PHONY ${_+_}cd ${.CURDIR}; ${NXBTMAKE} _cleanobj # Build the bootstrap/host/cross tools that produce native binaries ${_+_}cd ${.CURDIR}; ${NXBTMAKE} kernel-toolchain # Populate includes/libraries sysroot that produce native binaries. # This is split out from 'toolchain' above mostly so that target LLVM # libraries have a proper LLVM_DEFAULT_TARGET_TRIPLE without # polluting the cross-compiler build. The LLVM/GCC libs are skipped # here to avoid the problem but are kept in 'toolchain' so that # needed build tools are built. ${_+_}cd ${.CURDIR}; ${NXBTMAKE} _includes MK_CLANG=no ${_+_}cd ${.CURDIR}; ${NXBTMAKE} _libraries MK_CLANG=no .if !defined(NO_OBJWALK) ${_+_}cd ${.CURDIR}; ${NXBMAKE} _obj .endif ${_+_}cd ${.CURDIR}; ${NXBMAKE} everything @echo ">> native-xtools done. Use 'make native-xtools-install' to install to a given DESTDIR" native-xtools-install: .PHONY mkdir -p ${NXBDESTDIR}/bin ${NXBDESTDIR}/sbin ${NXBDESTDIR}/usr ${DESTDIR_MTREE} -f ${.CURDIR}/etc/mtree/BSD.usr.dist \ -p ${NXBDESTDIR}/usr >/dev/null ${DESTDIR_MTREE} -f ${.CURDIR}/etc/mtree/BSD.include.dist \ -p ${NXBDESTDIR}/usr/include >/dev/null ${_+_}cd ${.CURDIR}; ${NXBMAKE} \ DESTDIR=${NXBDESTDIR} \ -DNO_ROOT \ install # # hierarchy - ensure that all the needed directories are present # hierarchy hier: .MAKE .PHONY ${_+_}cd ${.CURDIR}/etc; ${HMAKE} 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: .MAKE .PHONY ${_+_}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= lib/libcompiler_rt .if ${MK_SSP} != "no" _prereq_libs+= lib/libssp_nonshared .endif # These dependencies are not automatically generated: # # lib/csu and lib/libc must be built before # all shared libraries for ELF. # _startup_libs= lib/csu _startup_libs+= lib/libc _startup_libs+= lib/libc_nonshared .if ${MK_LIBCPLUSPLUS} != "no" _startup_libs+= lib/libcxxrt .endif _prereq_libs+= lib/libgcc_eh lib/libgcc_s _startup_libs+= lib/libgcc_eh lib/libgcc_s lib/libgcc_s__L: lib/libc__L lib/libgcc_s__L: lib/libc_nonshared__L .if ${MK_LIBCPLUSPLUS} != "no" lib/libcxxrt__L: lib/libgcc_s__L .endif _prebuild_libs= ${_kerberos5_lib_libasn1} \ ${_kerberos5_lib_libhdb} \ ${_kerberos5_lib_libheimbase} \ ${_kerberos5_lib_libheimntlm} \ ${_libsqlite3} \ ${_kerberos5_lib_libheimipcc} \ ${_kerberos5_lib_libhx509} ${_kerberos5_lib_libkrb5} \ ${_kerberos5_lib_libroken} \ ${_kerberos5_lib_libwind} \ lib/libbz2 ${_libcom_err} lib/libcrypt \ lib/libelf lib/libexpat \ lib/libfigpar \ ${_lib_libgssapi} \ lib/libkiconv lib/libkvm lib/liblzma lib/libmd lib/libnv \ lib/libzstd \ ${_lib_casper} \ lib/ncurses/ncurses lib/ncurses/ncursesw \ lib/libopie lib/libpam/libpam ${_lib_libthr} \ ${_lib_libradius} lib/libsbuf lib/libtacplus \ lib/libgeom \ ${_cddl_lib_libumem} ${_cddl_lib_libnvpair} \ ${_cddl_lib_libuutil} \ ${_cddl_lib_libavl} \ ${_cddl_lib_libzfs_core} ${_cddl_lib_libzfs} \ ${_cddl_lib_libctf} \ lib/libufs \ lib/libutil lib/libpjdlog ${_lib_libypclnt} lib/libz lib/msun \ ${_secure_lib_libcrypto} ${_secure_lib_libssl} \ ${_lib_libldns} ${_secure_lib_libssh} .if ${MK_DIALOG} != "no" _prebuild_libs+= gnu/lib/libdialog gnu/lib/libdialog__L: lib/msun__L lib/ncurses/ncursesw__L .endif .if ${MK_GOOGLETEST} != "no" _prebuild_libs+= lib/libregex .endif .if ${MK_LIBCPLUSPLUS} != "no" _prebuild_libs+= lib/libc++ .endif lib/libgeom__L: lib/libexpat__L lib/libsbuf__L lib/libkvm__L: lib/libelf__L .if ${MK_LIBTHR} != "no" _lib_libthr= lib/libthr .endif .if ${MK_RADIUS_SUPPORT} != "no" _lib_libradius= lib/libradius .endif .if ${MK_OFED} != "no" _prebuild_libs+= \ lib/ofed/libibverbs \ lib/ofed/libibmad \ lib/ofed/libibumad \ lib/ofed/complib \ lib/ofed/libmlx5 lib/ofed/libibmad__L: lib/ofed/libibumad__L lib/ofed/complib__L: lib/libthr__L lib/ofed/libmlx5__L: lib/ofed/libibverbs__L lib/libthr__L .endif .if ${MK_CASPER} != "no" _lib_casper= lib/libcasper .endif lib/libpjdlog__L: lib/libutil__L lib/libcasper__L: lib/libnv__L lib/liblzma__L: lib/libmd__L lib/libthr__L lib/libzstd__L: lib/libthr__L _generic_libs= ${_cddl_lib} gnu/lib ${_kerberos5_lib} lib ${_secure_lib} .if ${MK_IPFILTER} != "no" _generic_libs+= sbin/ipf/libipf .endif .for _DIR in ${LOCAL_LIB_DIRS} .if ${_DIR} == ".WAIT" || (empty(_generic_libs:M${_DIR}) && exists(${.CURDIR}/${_DIR}/Makefile)) _generic_libs+= ${_DIR} .endif .endfor lib/libopie__L lib/libtacplus__L: lib/libmd__L .if ${MK_CDDL} != "no" _cddl_lib_libumem= cddl/lib/libumem _cddl_lib_libnvpair= cddl/lib/libnvpair _cddl_lib_libavl= cddl/lib/libavl _cddl_lib_libuutil= cddl/lib/libuutil .if ${MK_ZFS} != "no" _cddl_lib_libzfs_core= cddl/lib/libzfs_core _cddl_lib_libzfs= cddl/lib/libzfs cddl/lib/libzfs_core__L: cddl/lib/libnvpair__L cddl/lib/libzfs__L: cddl/lib/libzfs_core__L lib/msun__L lib/libutil__L cddl/lib/libzfs__L: lib/libthr__L lib/libmd__L lib/libz__L cddl/lib/libumem__L cddl/lib/libzfs__L: cddl/lib/libuutil__L cddl/lib/libavl__L lib/libgeom__L lib/libbe__L: cddl/lib/libzfs__L .endif _cddl_lib_libctf= cddl/lib/libctf _cddl_lib= cddl/lib cddl/lib/libctf__L: lib/libz__L .endif # cddl/lib/libdtrace requires lib/libproc and lib/librtld_db _prebuild_libs+= lib/libprocstat lib/libproc lib/librtld_db lib/libprocstat__L: lib/libelf__L lib/libkvm__L lib/libutil__L lib/libproc__L: lib/libprocstat__L lib/librtld_db__L: lib/libprocstat__L .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 secure/lib/libcrypto__L: lib/libthr__L .if ${MK_LDNS} != "no" _lib_libldns= lib/libldns lib/libldns__L: secure/lib/libssl__L .endif .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_LDNS} != "no" secure/lib/libssh__L: lib/libldns__L .endif .if ${MK_GSSAPI} != "no" && ${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_KERBEROS} != "no" kerberos5/lib/libasn1__L: lib/libcom_err__L kerberos5/lib/libroken__L kerberos5/lib/libhdb__L: kerberos5/lib/libasn1__L lib/libcom_err__L \ kerberos5/lib/libkrb5__L kerberos5/lib/libroken__L \ kerberos5/lib/libwind__L lib/libsqlite3__L kerberos5/lib/libheimntlm__L: secure/lib/libcrypto__L kerberos5/lib/libkrb5__L \ kerberos5/lib/libroken__L lib/libcom_err__L kerberos5/lib/libhx509__L: kerberos5/lib/libasn1__L lib/libcom_err__L \ secure/lib/libcrypto__L kerberos5/lib/libroken__L kerberos5/lib/libwind__L kerberos5/lib/libkrb5__L: kerberos5/lib/libasn1__L lib/libcom_err__L \ lib/libcrypt__L secure/lib/libcrypto__L kerberos5/lib/libhx509__L \ kerberos5/lib/libroken__L kerberos5/lib/libwind__L \ kerberos5/lib/libheimbase__L kerberos5/lib/libheimipcc__L kerberos5/lib/libroken__L: lib/libcrypt__L kerberos5/lib/libwind__L: kerberos5/lib/libroken__L lib/libcom_err__L kerberos5/lib/libheimbase__L: lib/libthr__L kerberos5/lib/libheimipcc__L: kerberos5/lib/libroken__L kerberos5/lib/libheimbase__L lib/libthr__L .endif lib/libsqlite3__L: lib/libthr__L .if ${MK_GSSAPI} != "no" _lib_libgssapi= lib/libgssapi .endif .if ${MK_KERBEROS} != "no" _kerberos5_lib= kerberos5/lib _kerberos5_lib_libasn1= kerberos5/lib/libasn1 _kerberos5_lib_libhdb= kerberos5/lib/libhdb _kerberos5_lib_libheimbase= kerberos5/lib/libheimbase _kerberos5_lib_libkrb5= kerberos5/lib/libkrb5 _kerberos5_lib_libhx509= kerberos5/lib/libhx509 _kerberos5_lib_libroken= kerberos5/lib/libroken _kerberos5_lib_libheimntlm= kerberos5/lib/libheimntlm _libsqlite3= lib/libsqlite3 _kerberos5_lib_libheimipcc= kerberos5/lib/libheimipcc _kerberos5_lib_libwind= kerberos5/lib/libwind _libcom_err= lib/libcom_err .endif .if ${MK_NIS} != "no" _lib_libypclnt= lib/libypclnt .endif .if ${MK_OPENSSL} == "no" lib/libradius__L: lib/libmd__L .endif lib/libproc__L: \ ${_cddl_lib_libctf:D${_cddl_lib_libctf}__L} lib/libelf__L lib/librtld_db__L lib/libutil__L .if ${MK_CXX} != "no" && ${MK_LIBCPLUSPLUS} != "no" lib/libproc__L: lib/libcxxrt__L .endif .for _lib in ${_prereq_libs} ${_lib}__PL: .PHONY .MAKE .if !defined(_MKSHOWCONFIG) && exists(${.CURDIR}/${_lib}) ${_+_}@${ECHODIR} "===> ${_lib} (obj,all,install)"; \ cd ${.CURDIR}/${_lib}; \ if [ -z "${NO_OBJWALK}" ]; then ${MAKE} MK_TESTS=no DIRPRFX=${_lib}/ obj; fi; \ ${MAKE} MK_TESTS=no MK_PROFILE=no -DNO_PIC \ DIRPRFX=${_lib}/ all; \ ${MAKE} MK_TESTS=no MK_PROFILE=no -DNO_PIC \ DIRPRFX=${_lib}/ install .endif .endfor .for _lib in ${_startup_libs} ${_prebuild_libs} ${_generic_libs} ${_lib}__L: .PHONY .MAKE .if !defined(_MKSHOWCONFIG) && exists(${.CURDIR}/${_lib}) ${_+_}@${ECHODIR} "===> ${_lib} (obj,all,install)"; \ cd ${.CURDIR}/${_lib}; \ if [ -z "${NO_OBJWALK}" ]; then ${MAKE} MK_TESTS=no DIRPRFX=${_lib}/ obj; fi; \ ${MAKE} MK_TESTS=no DIRPRFX=${_lib}/ all; \ ${MAKE} MK_TESTS=no DIRPRFX=${_lib}/ install .endif .endfor _prereq_libs: ${_prereq_libs:S/$/__PL/} _startup_libs: ${_startup_libs:S/$/__L/} _prebuild_libs: ${_prebuild_libs:S/$/__L/} _generic_libs: ${_generic_libs:S/$/__L/} # Enable SUBDIR_PARALLEL when not calling 'make all', unless called from # 'everything' with _PARALLEL_SUBDIR_OK set. This is because it is unlikely # that running 'make all' from the top-level, especially with a SUBDIR_OVERRIDE # or LOCAL_DIRS set, will have a reliable build if SUBDIRs are built in # parallel. This is safe for the world stage of buildworld though since it has # already built libraries in a proper order and installed includes into # WORLDTMP. Special handling is done for SUBDIR ordering for 'install*' to # avoid trashing a system if it crashes mid-install. .if !make(all) || defined(_PARALLEL_SUBDIR_OK) SUBDIR_PARALLEL= .endif .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: .PHONY @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; \ cd ${.CURDIR}; \ ${MAKE} -f ${.CURDIR}/Makefile.inc1 ${.MAKEFLAGS} ${.TARGET} \ -V OLD_FILES -V "OLD_FILES:Musr/share/*.gz:R" | xargs -n1 | sort | \ 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}/$${file}" <&3; \ fi; \ for ext in debug symbols; do \ if ! [ -e "${DESTDIR}/$${file}" ] && [ -f \ "${DESTDIR}${DEBUGDIR}/$${file}.$${ext}" ]; then \ rm ${RM_I} "${DESTDIR}${DEBUGDIR}/$${file}.$${ext}" \ <&3; \ fi; \ done; \ done # Remove catpages without corresponding manpages. @exec 3<&0; \ find ${DESTDIR}/usr/share/man/cat* ! -type d 2>/dev/null | sort | \ sed -ep -e's:${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: .PHONY @echo ">>> Checking for old files" @cd ${.CURDIR}; \ ${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}/$${file}"; \ fi; \ for ext in debug symbols; do \ if [ -f "${DESTDIR}${DEBUGDIR}/$${file}.$${ext}" ]; then \ echo "${DESTDIR}${DEBUGDIR}/$${file}.$${ext}"; \ fi; \ done; \ done | sort # Check for catpages without corresponding manpages. @find ${DESTDIR}/usr/share/man/cat* ! -type d 2>/dev/null | \ sed -ep -e's:${DESTDIR}/usr/share/man/cat:${DESTDIR}/usr/share/man/man:' | \ while read catpage; do \ read manpage; \ if [ ! -e "$${manpage}" ]; then \ echo $${catpage}; \ fi; \ done | sort delete-old-libs: .PHONY @echo ">>> Removing old libraries" @echo "${OLD_LIBS_MESSAGE}" | fmt @exec 3<&0; \ cd ${.CURDIR}; \ ${MAKE} -f ${.CURDIR}/Makefile.inc1 ${.MAKEFLAGS} ${.TARGET} \ -V OLD_LIBS | xargs -n1 | sort | \ while read file; do \ if [ -f "${DESTDIR}/$${file}" -o -L "${DESTDIR}/$${file}" ]; then \ chflags noschg "${DESTDIR}/$${file}" 2>/dev/null || true; \ rm ${RM_I} "${DESTDIR}/$${file}" <&3; \ fi; \ for ext in debug symbols; do \ if ! [ -e "${DESTDIR}/$${file}" ] && [ -f \ "${DESTDIR}${DEBUGDIR}/$${file}.$${ext}" ]; then \ rm ${RM_I} "${DESTDIR}${DEBUGDIR}/$${file}.$${ext}" \ <&3; \ fi; \ done; \ done @echo ">>> Old libraries removed" check-old-libs: .PHONY @echo ">>> Checking for old libraries" @cd ${.CURDIR}; \ ${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}/$${file}"; \ fi; \ for ext in debug symbols; do \ if [ -f "${DESTDIR}${DEBUGDIR}/$${file}.$${ext}" ]; then \ echo "${DESTDIR}${DEBUGDIR}/$${file}.$${ext}"; \ fi; \ done; \ done | sort delete-old-dirs: .PHONY @echo ">>> Removing old directories" @cd ${.CURDIR}; \ ${MAKE} -f ${.CURDIR}/Makefile.inc1 ${.MAKEFLAGS} ${.TARGET} \ -V OLD_DIRS | xargs -n1 | sort -r | \ while read dir; do \ if [ -d "${DESTDIR}/$${dir}" ]; then \ rmdir -v "${DESTDIR}/$${dir}" || true; \ elif [ -L "${DESTDIR}/$${dir}" ]; then \ echo "${DESTDIR}/$${dir} is a link, please remove everything manually."; \ fi; \ if [ -d "${DESTDIR}${DEBUGDIR}/$${dir}" ]; then \ rmdir -v "${DESTDIR}${DEBUGDIR}/$${dir}" || true; \ elif [ -L "${DESTDIR}${DEBUGDIR}/$${dir}" ]; then \ echo "${DESTDIR}${DEBUGDIR}/$${dir} is a link, please remove everything manually."; \ fi; \ done @echo ">>> Old directories removed" check-old-dirs: .PHONY @echo ">>> Checking for old directories" @cd ${.CURDIR}; \ ${MAKE} -f ${.CURDIR}/Makefile.inc1 ${.MAKEFLAGS} ${.TARGET} \ -V OLD_DIRS | xargs -n1 | sort -r | \ while read dir; do \ if [ -d "${DESTDIR}/$${dir}" ]; then \ echo "${DESTDIR}/$${dir}"; \ elif [ -L "${DESTDIR}/$${dir}" ]; then \ echo "${DESTDIR}/$${dir} is a link, please remove everything manually."; \ fi; \ if [ -d "${DESTDIR}${DEBUGDIR}/$${dir}" ]; then \ echo "${DESTDIR}${DEBUGDIR}/$${dir}"; \ elif [ -L "${DESTDIR}${DEBUGDIR}/$${dir}" ]; then \ echo "${DESTDIR}${DEBUGDIR}/$${dir} is a link, please remove everything manually."; \ fi; \ done delete-old: delete-old-files delete-old-dirs .PHONY @echo "To remove old libraries run '${MAKE_CMD} delete-old-libs'." check-old: check-old-files check-old-libs check-old-dirs .PHONY @echo "To remove old files and directories run '${MAKE_CMD} delete-old'." @echo "To remove old libraries run '${MAKE_CMD} delete-old-libs'." .endif # # showconfig - show build configuration. # showconfig: .PHONY @(${MAKE} -n -f ${.CURDIR}/sys/conf/kern.opts.mk -V dummy -dg1 UPDATE_DEPENDFILE=no NO_OBJ=yes; \ ${MAKE} -n -f ${.CURDIR}/share/mk/src.opts.mk -V dummy -dg1 UPDATE_DEPENDFILE=no NO_OBJ=yes) 2>&1 | grep ^MK_ | sort -u .if !empty(KRNLOBJDIR) && !empty(KERNCONF) DTBOUTPUTPATH= ${KRNLOBJDIR}/${KERNCONF}/ .if !defined(FDT_DTS_FILE) || empty(FDT_DTS_FILE) .if !defined(_MKSHOWCONFIG) && exists(${KERNCONFDIR}/${KERNCONF}) FDT_DTS_FILE!= awk 'BEGIN {FS="="} /^makeoptions[[:space:]]+FDT_DTS_FILE/ {print $$2}' \ '${KERNCONFDIR}/${KERNCONF}' ; echo .endif .endif .endif .if !defined(DTBOUTPUTPATH) || !exists(${DTBOUTPUTPATH}) DTBOUTPUTPATH= ${.CURDIR} .endif # # Build 'standalone' Device Tree Blob # builddtb: .PHONY @PATH=${TMPPATH} MACHINE=${TARGET} \ ${.CURDIR}/sys/tools/fdt/make_dtb.sh ${.CURDIR}/sys \ "${FDT_DTS_FILE}" ${DTBOUTPUTPATH} ############### # cleanworld # In the following, the first 'rm' in a series will usually remove all # files and directories. If it does not, then there are probably some # files with file flags set, so this unsets them and tries the 'rm' a # second time. There are situations where this target will be cleaning # some directories via more than one method, but that duplication is # needed to correctly handle all the possible situations. Removing all # files without file flags set in the first 'rm' instance saves time, # because 'chflags' will need to operate on fewer files afterwards. # # It is expected that BW_CANONICALOBJDIR == the CANONICALOBJDIR as would be # created by bsd.obj.mk, except that we don't want to .include that file # in this makefile. We don't do a cleandir walk if MK_AUTO_OBJ is yes # since it is not possible for files to land in the wrong place. # .if make(cleanworld) BW_CANONICALOBJDIR:=${OBJTOP}/ .elif make(cleanuniverse) BW_CANONICALOBJDIR:=${OBJROOT} .if ${MK_UNIFIED_OBJDIR} == "no" .error ${.TARGETS} only supported with WITH_UNIFIED_OBJDIR enabled. .endif .endif cleanworld cleanuniverse: .PHONY .if !empty(BW_CANONICALOBJDIR) && exists(${BW_CANONICALOBJDIR}) && \ ${.CURDIR:tA} != ${BW_CANONICALOBJDIR:tA} -rm -rf ${BW_CANONICALOBJDIR}* -chflags -R 0 ${BW_CANONICALOBJDIR} rm -rf ${BW_CANONICALOBJDIR}* .endif .if make(cleanworld) && ${MK_AUTO_OBJ} == "no" && \ (empty(BW_CANONICALOBJDIR) || ${.CURDIR:tA} == ${BW_CANONICALOBJDIR:tA}) .if ${.CURDIR} == ${.OBJDIR} || ${.CURDIR}/obj == ${.OBJDIR} # To be safe in this case, fall back to a 'make cleandir' ${_+_}@cd ${.CURDIR}; ${MAKE} cleandir .endif .endif .if ${TARGET} == ${MACHINE} && ${TARGET_ARCH} == ${MACHINE_ARCH} XDEV_CPUTYPE?=${CPUTYPE} .else XDEV_CPUTYPE?=${TARGET_CPUTYPE} .endif NOFUN=-DNO_FSCHG MK_HTML=no -DNO_LINT \ MK_MAN=no MK_NLS=no MK_PROFILE=no \ MK_KERBEROS=no MK_RESCUE=no MK_TESTS=no MK_WARNS=no \ TARGET=${TARGET} TARGET_ARCH=${TARGET_ARCH} \ CPUTYPE=${XDEV_CPUTYPE} XDDIR=${TARGET_ARCH}-freebsd XDTP?=/usr/${XDDIR} .if ${XDTP:N/*} .error XDTP variable should be an absolute path .endif CDBOBJROOT= ${OBJROOT}${MACHINE}.${MACHINE_ARCH}/xdev/ CDBOBJTOP= ${CDBOBJROOT}${XDDIR} CDBENV= \ INSTALL="sh ${.CURDIR}/tools/install.sh" CDENV= ${CDBENV} \ TOOLS_PREFIX=${XDTP} CDMAKEARGS= \ OBJTOP=${CDBOBJTOP:Q} \ OBJROOT=${CDBOBJROOT:Q} CD2MAKEARGS= ${CDMAKEARGS} .if ${WANT_COMPILER_TYPE} == gcc || \ (defined(X_COMPILER_TYPE) && ${X_COMPILER_TYPE} == gcc) # GCC requires -isystem and -L when using a cross-compiler. --sysroot # won't set header path and -L is used to ensure the base library path # is added before the port PREFIX library path. CD2CFLAGS+= -isystem ${XDDESTDIR}/usr/include -L${XDDESTDIR}/usr/lib # GCC requires -B to find /usr/lib/crti.o when using a cross-compiler # combined with --sysroot. CD2CFLAGS+= -B${XDDESTDIR}/usr/lib # Force using libc++ for external GCC. .if defined(X_COMPILER_TYPE) && \ ${X_COMPILER_TYPE} == gcc && ${X_COMPILER_VERSION} >= 40800 CD2CXXFLAGS+= -isystem ${XDDESTDIR}/usr/include/c++/v1 -std=c++11 \ -nostdinc++ .endif .endif CD2CFLAGS+= --sysroot=${XDDESTDIR}/ CD2ENV=${CDENV} CC="${CC} ${CD2CFLAGS}" CXX="${CXX} ${CD2CXXFLAGS} ${CD2CFLAGS}" \ CPP="${CPP} ${CD2CFLAGS}" \ MACHINE=${TARGET} MACHINE_ARCH=${TARGET_ARCH} CDTMP= ${OBJTOP}/${XDDIR}/tmp CDMAKE=${CDENV} PATH=${CDTMP}/usr/bin:${PATH} ${MAKE} ${CDMAKEARGS} ${NOFUN} CD2MAKE=${CD2ENV} PATH=${CDTMP}/usr/bin:${XDDESTDIR}/usr/bin:${PATH} \ ${MAKE} ${CD2MAKEARGS} ${NOFUN} .if ${MK_META_MODE} != "no" # Don't rebuild build-tools targets during normal build. CD2MAKE+= BUILD_TOOLS_META=.NOMETA .endif XDDESTDIR=${DESTDIR}${XDTP} .ORDER: xdev-build xdev-install xdev-links xdev: xdev-build xdev-install .PHONY .ORDER: _xb-worldtmp _xb-bootstrap-tools _xb-build-tools _xb-cross-tools xdev-build: _xb-worldtmp _xb-bootstrap-tools _xb-build-tools _xb-cross-tools .PHONY _xb-worldtmp: .PHONY mkdir -p ${CDTMP}/usr ${WORLDTMP_MTREE} -f ${.CURDIR}/etc/mtree/BSD.usr.dist \ -p ${CDTMP}/usr >/dev/null _xb-bootstrap-tools: .PHONY .for _tool in \ ${_clang_tblgen} \ ${_yacc} ${_+_}@${ECHODIR} "===> ${_tool} (obj,all,install)"; \ cd ${.CURDIR}/${_tool}; \ if [ -z "${NO_OBJWALK}" ]; then ${CDMAKE} DIRPRFX=${_tool}/ obj; fi; \ ${CDMAKE} DIRPRFX=${_tool}/ all; \ ${CDMAKE} DIRPRFX=${_tool}/ DESTDIR=${CDTMP} install .endfor _xb-build-tools: .PHONY ${_+_}@cd ${.CURDIR}; \ ${CDBENV} ${MAKE} ${CDMAKEARGS} -f Makefile.inc1 ${NOFUN} build-tools XDEVDIRS= \ ${_clang_libs} \ ${_lld} \ ${_elftctools} \ usr.bin/ar \ ${_clang} _xb-cross-tools: .PHONY .for _tool in ${XDEVDIRS} ${_+_}@${ECHODIR} "===> xdev ${_tool} (obj,all)"; \ cd ${.CURDIR}/${_tool}; \ if [ -z "${NO_OBJWALK}" ]; then ${CDMAKE} DIRPRFX=${_tool}/ obj; fi; \ ${CDMAKE} DIRPRFX=${_tool}/ all .endfor _xi-mtree: .PHONY ${_+_}@${ECHODIR} "mtree populating ${XDDESTDIR}" mkdir -p ${XDDESTDIR} ${DESTDIR_MTREE} -f ${.CURDIR}/etc/mtree/BSD.root.dist \ -p ${XDDESTDIR} >/dev/null ${DESTDIR_MTREE} -f ${.CURDIR}/etc/mtree/BSD.usr.dist \ -p ${XDDESTDIR}/usr >/dev/null ${DESTDIR_MTREE} -f ${.CURDIR}/etc/mtree/BSD.include.dist \ -p ${XDDESTDIR}/usr/include >/dev/null .if defined(_LIBCOMPAT) ${DESTDIR_MTREE} -f ${.CURDIR}/etc/mtree/BSD.lib${libcompat}.dist \ -p ${XDDESTDIR}/usr >/dev/null .endif .if ${MK_TESTS} != "no" mkdir -p ${XDDESTDIR}${TESTSBASE} ${DESTDIR_MTREE} -f ${.CURDIR}/etc/mtree/BSD.tests.dist \ -p ${XDDESTDIR}${TESTSBASE} >/dev/null .endif .ORDER: xdev-build _xi-mtree _xi-cross-tools _xi-includes _xi-libraries xdev-install: xdev-build _xi-mtree _xi-cross-tools _xi-includes _xi-libraries .PHONY _xi-cross-tools: .PHONY @echo "_xi-cross-tools" .for _tool in ${XDEVDIRS} ${_+_}@${ECHODIR} "===> xdev ${_tool} (install)"; \ cd ${.CURDIR}/${_tool}; \ ${CDMAKE} DIRPRFX=${_tool}/ install DESTDIR=${XDDESTDIR} .endfor _xi-includes: .PHONY .if !defined(NO_OBJWALK) ${_+_}cd ${.CURDIR}; ${CD2MAKE} -f Makefile.inc1 _obj \ DESTDIR=${XDDESTDIR} .endif ${_+_}cd ${.CURDIR}; ${CD2MAKE} -f Makefile.inc1 includes \ DESTDIR=${XDDESTDIR} _xi-libraries: .PHONY ${_+_}cd ${.CURDIR}; ${CD2MAKE} -f Makefile.inc1 libraries \ DESTDIR=${XDDESTDIR} xdev-links: .PHONY ${_+_}cd ${XDDESTDIR}/usr/bin; \ mkdir -p ../../../../usr/bin; \ for i in *; do \ ln -sf ../../${XDTP}/usr/bin/$$i \ ../../../../usr/bin/${XDDIR}-$$i; \ ln -sf ../../${XDTP}/usr/bin/$$i \ ../../../../usr/bin/${XDDIR}${_REVISION}-$$i; \ done Index: projects/clang1100-import/UPDATING =================================================================== --- projects/clang1100-import/UPDATING (revision 364034) +++ projects/clang1100-import/UPDATING (revision 364035) @@ -1,2317 +1,2323 @@ Updating Information for FreeBSD current users. This file is maintained and copyrighted by M. Warner Losh . See end of file for further details. For commonly done items, please see the COMMON ITEMS: section later in the file. These instructions assume that you basically know what you are doing. If not, then please consult the FreeBSD handbook: https://www.freebsd.org/doc/en_US.ISO8859-1/books/handbook/makeworld.html Items affecting the ports and packages system can be found in /usr/ports/UPDATING. Please read that file before running portupgrade. NOTE TO PEOPLE WHO THINK THAT FreeBSD 13.x IS SLOW: FreeBSD 13.x has many debugging features turned on, in both the kernel and userland. These features attempt to detect incorrect use of system primitives, and encourage loud failure through extra sanity checking and fail stop semantics. They also substantially impact system performance. If you want to do performance measurement, benchmarking, and optimization, you'll want to turn them off. This includes various WITNESS- related kernel options, INVARIANTS, malloc debugging flags in userland, and various verbose features in the kernel. Many developers choose to disable these features on build machines to maximize performance. (To completely disable malloc debugging, define MALLOC_PRODUCTION in /etc/make.conf, or to merely disable the most expensive debugging functionality run "ln -s 'abort:false,junk:false' /etc/malloc.conf".) 2020mmdd: Clang, llvm, lld, lldb, compiler-rt, libc++, libunwind and openmp have been upgraded to 11.0.0. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. +20200807: + Makefile.inc has been updated to work around the issue documented in + 20200729. It was a case where the optimization of using symbolic links + to point to binaries created a situation where we'd run new binaries + with old libraries starting midway through the installworld process. + 20200729: r363679 has redefined some undefined behavior in regcomp(3); notably, extraneous escapes of most ordinary characters will no longer be accepted. An exp-run has identified all of the problems with this in ports, but other non-ports software may need extra escapes removed to continue to function. Because of this change, installworld may encounter the following error from rtld: Undefined symbol "regcomp@FBSD_1.6" -- It is imperative that you do not halt installworld. Instead, let it run to completion (whether successful or not) and run installworld once more. 20200627: A new implementation of bc and dc has been imorted in r362681. This implementation corrects non-conformant behavior of the previous bc and adds GNU bc compatible options. It offers a number of extensions, is much faster on large values, and has support for message catalogs (a number of languages are already supported, contributions of further languages welcome). The option WITHOUT_GH_BC can be used to build the world with the previous versions of bc and dc. 20200625: r362639 changed the internal API used between the NFS kernel modules. As such, they all need to be rebuilt from sources. 20200613: r362158 changed the arguments for VFS_CHECKEXP(). As such, any out of tree file systems need to be modified and rebuilt. Also, any file systems that are modules must be rebuilt. 20200604: read(2) of a directory fd is now rejected by default. root may re-enable it for system root only on non-ZFS filesystems with the security.bsd.allow_read_dir sysctl(8) MIB if security.bsd.suser_enabled=1. It may be advised to setup aliases for grep to default to `-d skip` if commonly non-recursively grepping a list that includes directories and the potential for the resulting stderr output is not tolerable. Example aliases are now installed, commented out, in /root/.cshrc and /root/.shrc. 20200523: Clang, llvm, lld, lldb, compiler-rt, libc++, libunwind and openmp have been upgraded to 10.0.1. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. 20200512: Support for obsolete compilers has been removed from the build system. Clang 6 and GCC 6.4 are the minimum supported versions. 20200424: closefrom(2) has been moved under COMPAT12, and replaced in libc with a stub that calls close_range(2). If using a custom kernel configuration, you may want to ensure that the COMPAT_FREEBSD12 option is included, as a slightly older -CURRENT userland and older FreeBSD userlands may not be functional without closefrom(2). 20200414: Upstream DTS from Linux 5.6 was merged and they now have the SID and THS (Secure ID controller and THermal Sensor) node present. The DTB overlays have now been removed from the tree for the H3/H5 and A64 SoCs and the aw_sid and aw_thermal driver have been updated to deal with upstream DTS. If you are using those overlays you need to remove them from loader.conf and update the DTBs on the FAT partition. 20200310: Clang, llvm, lld, lldb, compiler-rt, libc++, libunwind and openmp have been upgraded to 10.0.0. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. 20200309: The amd(8) automount daemon has been removed from the source tree. As of FreeBSD 10.1 autofs(5) is the preferred tool for automounting. amd is still available in the sysutils/am-utils port. 20200301: Removed brooktree driver (bktr.4) from the tree. 20200229: The WITH_GPL_DTC option has been removed. The BSD-licenced device tree compiler in usr.bin/dtc is used on all architectures which use dtc, and the GPL dtc is available (if needed) from the sysutils/dtc port. 20200229: The WITHOUT_LLVM_LIBUNWIND option has been removed. LLVM's libunwind is used by all supported CPU architectures. 20200229: GCC 4.2.1 has been removed from the tree. The WITH_GCC, WITH_GCC_BOOTSTRAP, and WITH_GNUCXX options are no longer available. Users who wish to build FreeBSD with GCC must use the external toolchain ports or packages. 20200220: ncurses has been updated to a newer version (6.2-20200215). Given the ABI has changed, users will have to rebuild all the ports that are linked to ncurses. 20200217: The size of struct vnet and the magic cookie have changed. Users need to recompile libkvm and all modules using VIMAGE together with their new kernel. 20200212: Defining the long deprecated NO_CTF, NO_DEBUG_FILES, NO_INSTALLLIB, NO_MAN, NO_PROFILE, and NO_WARNS variables is now an error. Update your Makefiles and scripts to define MK_=no instead as required. One exception to this is that program or library Makefiles should define MAN to empty rather than setting MK_MAN=no. 20200108: Clang/LLVM is now the default compiler and LLD the default linker for riscv64. 20200107: make universe no longer uses GCC 4.2.1 on any architectures. Architectures not supported by in-tree Clang/LLVM require an external toolchain package. 20200104: GCC 4.2.1 is now not built by default, as part of the GCC 4.2.1 retirement plan. Specifically, the GCC, GCC_BOOTSTRAP, and GNUCXX options default to off for all supported CPU architectures. As a short-term transition aid they may be enabled via WITH_* options. GCC 4.2.1 is expected to be removed from the tree on 2020-03-31. 20200102: Support for armv5 has been disconnected and is being removed. The machine combination MACHINE=arm MACHINE_ARCH=arm is no longer valid. You must now use a MACHINE_ARCH of armv6 or armv7. The default MACHINE_ARCH for MACHINE=arm is now armv7. 20191226: Clang/LLVM is now the default compiler for all powerpc architectures. LLD is now the default linker for powerpc64. The change for powerpc64 also includes a change to the ELFv2 ABI, incompatible with the existing ABI. 20191226: Kernel-loadable random(4) modules are no longer unloadable. 20191222: Clang, llvm, lld, lldb, compiler-rt, libc++, libunwind and openmp have been upgraded to 9.0.1. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. 20191212: r355677 has modified the internal interface used between the NFS modules in the kernel. As such, they must all be upgraded simultaneously. I will do a version bump for this. 20191205: The root certificates of the Mozilla CA Certificate Store have been imported into the base system and can be managed with the certctl(8) utility. If you have installed the security/ca_root_nss port or package with the ETCSYMLINK option (the default), be advised that there may be differences between those included in the port and those included in base due to differences in nss branch used as well as general update frequency. Note also that certctl(8) cannot manage certs in the format used by the security/ca_root_nss port. 20191120: The amd(8) automount daemon has been disabled by default, and will be removed in the future. As of FreeBSD 10.1 the autofs(5) is available for automounting. 20191107: The nctgpio and wbwd drivers have been moved to the superio bus. If you have one of these drivers in a kernel configuration, then you should add device superio to it. If you use one of these drivers as a module and you compile a custom set of modules, then you should add superio to the set. 20191021: KPIs for network drivers to access interface addresses have changed. Users need to recompile NIC driver modules together with kernel. 20191021: The net.link.tap.user_open sysctl no longer prevents user opening of already created /dev/tapNN devices. Access is still controlled by node permissions, just like tun devices. The net.link.tap.user_open sysctl is now used only to allow users to perform devfs cloning of tap devices, and the subsequent open may not succeed if the user is not in the appropriate group. This sysctl may be deprecated/removed completely in the future. 20191009: mips, powerpc, and sparc64 are no longer built as part of universe / tinderbox unless MAKE_OBSOLETE_GCC is defined. If not defined, mips, powerpc, and sparc64 builds will look for the xtoolchain binaries and if installed use them for universe builds. As llvm 9.0 becomes vetted for these architectures, they will be removed from the list. 20191009: Clang, llvm, lld, lldb, compiler-rt, libc++, libunwind and openmp have been upgraded to 9.0.0. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. 20191003: The hpt27xx, hptmv, hptnr, and hptrr drivers have been removed from GENERIC. They are available as modules and can be loaded by adding to /boot/loader.conf hpt27xx_load="YES", hptmv_load="YES", hptnr_load="YES", or hptrr_load="YES", respectively. 20190913: ntpd no longer by default locks its pages in memory, allowing them to be paged out by the kernel. Use rlimit memlock to restore historic BSD behaviour. For example, add "rlimit memlock 32" to ntp.conf to lock up to 32 MB of ntpd address space in memory. 20190823: Several of ping6's options have been renamed for better consistency with ping. If you use any of -ARWXaghmrtwx, you must update your scripts. See ping6(8) for details. 20190727: The vfs.fusefs.sync_unmount and vfs.fusefs.init_backgrounded sysctls and the "-o sync_unmount" and "-o init_backgrounded" mount options have been removed from mount_fusefs(8). You can safely remove them from your scripts, because they had no effect. The vfs.fusefs.fix_broken_io, vfs.fusefs.sync_resize, vfs.fusefs.refresh_size, vfs.fusefs.mmap_enable, vfs.fusefs.reclaim_revoked, and vfs.fusefs.data_cache_invalidate sysctls have been removed. If you felt the need to set any of them to a non-default value, please tell asomers@FreeBSD.org why. 20190713: Default permissions on the /var/account/acct file (and copies of it rotated by periodic daily scripts) are changed from 0644 to 0640 because the file contains sensitive information that should not be world-readable. If the /var/account directory must be created by rc.d/accounting, the mode used is now 0750. Admins who use the accounting feature are encouraged to change the mode of an existing /var/account directory to 0750 or 0700. 20190620: Entropy collection and the /dev/random device are no longer optional components. The "device random" option has been removed. Implementations of distilling algorithms can still be made loadable with "options RANDOM_LOADABLE" (e.g., random_fortuna.ko). 20190612: Clang, llvm, lld, lldb, compiler-rt, libc++, libunwind and openmp have been upgraded to 8.0.1. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. 20190608: A fix was applied to i386 kernel modules to avoid panics with dpcpu or vnet. Users need to recompile i386 kernel modules having pcpu or vnet sections or they will refuse to load. 20190513: User-wired pages now have their own counter, vm.stats.vm.v_user_wire_count. The vm.max_wired sysctl was renamed to vm.max_user_wired and changed from an unsigned int to an unsigned long. bhyve VMs wired with the -S are now subject to the user wiring limit; the vm.max_user_wired sysctl may need to be tuned to avoid running into the limit. 20190507: The IPSEC option has been removed from GENERIC. Users requiring ipsec(4) must now load the ipsec(4) kernel module. 20190507: The tap(4) driver has been folded into tun(4), and the module has been renamed to tuntap. You should update any kld_list="if_tap" or kld_list="if_tun" entries in /etc/rc.conf, if_tap_load="YES" or if_tun_load="YES" entries in /boot/loader.conf to load the if_tuntap module instead, and "device tap" or "device tun" entries in kernel config files to select the tuntap device instead. 20190418: The following knobs have been added related to tradeoffs between safe use of the random device and availability in the absence of entropy: kern.random.initial_seeding.bypass_before_seeding: tunable; set non-zero to bypass the random device prior to seeding, or zero to block random requests until the random device is initially seeded. For now, set to 1 (unsafe) by default to restore pre-r346250 boot availability properties. kern.random.initial_seeding.read_random_bypassed_before_seeding: read-only diagnostic sysctl that is set when bypass is enabled and read_random(9) is bypassed, to enable programmatic handling of this initial condition, if desired. kern.random.initial_seeding.arc4random_bypassed_before_seeding: Similar to the above, but for for arc4random(9) initial seeding. kern.random.initial_seeding.disable_bypass_warnings: tunable; set non-zero to disable warnings in dmesg when the same conditions are met as for the diagnostic sysctls above. Defaults to zero, i.e., produce warnings in dmesg when the conditions are met. 20190416: The loadable random module KPI has changed; the random_infra_init() routine now requires a 3rd function pointer for a bool (*)(void) method that returns true if the random device is seeded (and therefore unblocked). 20190404: r345895 reverts r320698. This implies that an nfsuserd(8) daemon built from head sources between r320757 (July 6, 2017) and r338192 (Aug. 22, 2018) will not work unless the "-use-udpsock" is added to the command line. nfsuserd daemons built from head sources that are post-r338192 are not affected and should continue to work. 20190320: The fuse(4) module has been renamed to fusefs(4) for consistency with other filesystems. You should update any kld_load="fuse" entries in /etc/rc.conf, fuse_load="YES" entries in /boot/loader.conf, and "options FUSE" entries in kernel config files. 20190304: Clang, llvm, lld, lldb, compiler-rt and libc++ have been upgraded to 8.0.0. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. 20190226: geom_uzip(4) depends on the new module xz. If geom_uzip is statically compiled into your custom kernel, add 'device xz' statement to the kernel config. 20190219: drm and drm2 have been removed from the tree. Please see https://wiki.freebsd.org/Graphics for the latest information on migrating to the drm ports. 20190131: Iflib is no longer unconditionally compiled into the kernel. Drivers using iflib and statically compiled into the kernel, now require the 'device iflib' config option. For the same drivers loaded as modules on kernels not having 'device iflib', the iflib.ko module is loaded automatically. 20190125: The IEEE80211_AMPDU_AGE and AH_SUPPORT_AR5416 kernel configuration options no longer exist since r343219 and r343427 respectively; nothing uses them, so they should be just removed from custom kernel config files. 20181230: r342635 changes the way efibootmgr(8) works by requiring users to add the -b (bootnum) parameter for commands where the bootnum was previously specified with each option. For example 'efibootmgr -B 0001' is now 'efibootmgr -B -b 0001'. 20181220: r342286 modifies the NFSv4 server so that it obeys vfs.nfsd.nfs_privport in the same as it is applied to NFSv2 and 3. This implies that NFSv4 servers that have vfs.nfsd.nfs_privport set will only allow mounts from clients using a reserved port#. Since both the FreeBSD and Linux NFSv4 clients use reserved port#s by default, this should not affect most NFSv4 mounts. 20181219: The XLP config has been removed. We can't support 64-bit atomics in this kernel because it is running in 32-bit mode. XLP users must transition to running a 64-bit kernel (XLP64 or XLPN32). The mips GXEMUL support has been removed from FreeBSD. MALTA* + qemu is the preferred emulator today and we don't need two different ones. The old sibyte / swarm / Broadcom BCM1250 support has been removed from the mips port. 20181211: Clang, llvm, lld, lldb, compiler-rt and libc++ have been upgraded to 7.0.1. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. 20181211: Remove the timed and netdate programs from the base tree. Setting the time with these daemons has been obsolete for over a decade. 20181126: On amd64, arm64 and armv7 (architectures that install LLVM's ld.lld linker as /usr/bin/ld) GNU ld is no longer installed as ld.bfd, as it produces broken binaries when ifuncs are in use. Users needing GNU ld should install the binutils port or package. 20181123: The BSD crtbegin and crtend code has been enabled by default. It has had extensive testing on amd64, arm64, and i386. It can be disabled by building a world with -DWITHOUT_BSD_CRTBEGIN. 20181115: The set of CTM commands (ctm, ctm_smail, ctm_rmail, ctm_dequeue) has been converted to a port (misc/ctm) and will be removed from FreeBSD-13. It is available as a package (ctm) for all supported FreeBSD versions. 20181110: The default newsyslog.conf(5) file has been changed to only include files in /etc/newsyslog.conf.d/ and /usr/local/etc/newsyslog.conf.d/ if the filenames end in '.conf' and do not begin with a '.'. You should check the configuration files in these two directories match this naming convention. You can verify which configuration files are being included using the command: $ newsyslog -Nrv 20181015: Ports for the DRM modules have been simplified. Now, amd64 users should just install the drm-kmod port. All others should install drm-legacy-kmod. Graphics hardware that's newer than about 2010 usually works with drm-kmod. For hardware older than 2013, however, some users will need to use drm-legacy-kmod if drm-kmod doesn't work for them. Hardware older than 2008 usually only works in drm-legacy-kmod. The graphics team can only commit to hardware made since 2013 due to the complexity of the market and difficulty to test all the older cards effectively. If you have hardware supported by drm-kmod, you are strongly encouraged to use that as you will get better support. Other than KPI chasing, drm-legacy-kmod will not be updated. As outlined elsewhere, the drm and drm2 modules will be eliminated from the src base soon (with a limited exception for arm). Please update to the package asap and report any issues to x11@freebsd.org. Generally, anybody using the drm*-kmod packages should add WITHOUT_DRM_MODULE=t and WITHOUT_DRM2_MODULE=t to avoid nasty cross-threading surprises, especially with automatic driver loading from X11 startup. These will become the defaults in 13-current shortly. 20181012: The ixlv(4) driver has been renamed to iavf(4). As a consequence, custom kernel and module loading configuration files must be updated accordingly. Moreover, interfaces previous presented as ixlvN to the system are now exposed as iavfN and network configuration files must be adjusted as necessary. 20181009: OpenSSL has been updated to version 1.1.1. This update included additional various API changes throughout the base system. It is important to rebuild third-party software after upgrading. The value of __FreeBSD_version has been bumped accordingly. 20181006: The legacy DRM modules and drivers have now been added to the loader's module blacklist, in favor of loading them with kld_list in rc.conf(5). The module blacklist may be overridden with the loader.conf(5) 'module_blacklist' variable, but loading them via rc.conf(5) is strongly encouraged. 20181002: The cam(4) based nda(4) driver will be used over nvd(4) by default on powerpc64. You may set 'options NVME_USE_NVD=1' in your kernel conf or loader tunable 'hw.nvme.use_nvd=1' if you wish to use the existing driver. Make sure to edit /boot/etc/kboot.conf and fstab to use the nda device name. 20180913: Reproducible build mode is now on by default, in preparation for FreeBSD 12.0. This eliminates build metadata such as the user, host, and time from the kernel (and uname), unless the working tree corresponds to a modified checkout from a version control system. The previous behavior can be obtained by setting the /etc/src.conf knob WITHOUT_REPRODUCIBLE_BUILD. 20180826: The Yarrow CSPRNG has been removed from the kernel as it has not been supported by its designers since at least 2003. Fortuna has been the default since FreeBSD-11. 20180822: devctl freeze/thaw have gone into the tree, the rc scripts have been updated to use them and devmatch has been changed. You should update kernel, userland and rc scripts all at the same time. 20180818: The default interpreter has been switched from 4th to Lua. LOADER_DEFAULT_INTERP, documented in build(7), will override the default interpreter. If you have custom FORTH code you will need to set LOADER_DEFAULT_INTERP=4th (valid values are 4th, lua or simp) in src.conf for the build. This will create default hard links between loader and loader_4th instead of loader and loader_lua, the new default. If you are using UEFI it will create the proper hard link to loader.efi. bhyve uses userboot.so. It remains 4th-only until some issues are solved regarding coexisting with multiple versions of FreeBSD are resolved. 20180815: ls(1) now respects the COLORTERM environment variable used in other systems and software to indicate that a colored terminal is both supported and desired. If ls(1) is suddenly emitting colors, they may be disabled again by either removing the unwanted COLORTERM from your environment, or using `ls --color=never`. The ls(1) specific CLICOLOR may not be observed in a future release. 20180808: The default pager for most commands has been changed to "less". To restore the old behavior, set PAGER="more" and MANPAGER="more -s" in your environment. 20180731: The jedec_ts(4) driver has been removed. A superset of its functionality is available in the jedec_dimm(4) driver, and the manpage for that driver includes migration instructions. If you have "device jedec_ts" in your kernel configuration file, it must be removed. 20180730: amd64/GENERIC now has EFI runtime services, EFIRT, enabled by default. This should have no effect if the kernel is booted via BIOS/legacy boot. EFIRT may be disabled via a loader tunable, efi.rt.disabled, if a system has a buggy firmware that prevents a successful boot due to use of runtime services. 20180727: Atmel AT91RM9200 and AT91SAM9, Cavium CNS 11xx and XScale support has been removed from the tree. These ports were obsolete and/or known to be broken for many years. 20180723: loader.efi has been augmented to participate more fully in the UEFI boot manager protocol. loader.efi will now look at the BootXXXX environment variable to determine if a specific kernel or root partition was specified. XXXX is derived from BootCurrent. efibootmgr(8) manages these standard UEFI variables. 20180720: zfsloader's functionality has now been folded into loader. zfsloader is no longer necessary once you've updated your boot blocks. For a transition period, we will install a hardlink for zfsloader to loader to allow a smooth transition until the boot blocks can be updated (hard link because old zfs boot blocks don't understand symlinks). 20180719: ARM64 now have efifb support, if you want to have serial console on your arm64 board when an screen is connected and the bootloader setup a frame buffer for us to use, just add : boot_serial=YES boot_multicons=YES in /boot/loader.conf For Raspberry Pi 3 (RPI) users, this is needed even if you don't have an screen connected as the firmware will setup a frame buffer are that u-boot will expose as an EFI frame buffer. 20180719: New uid:gid added, ntpd:ntpd (123:123). Be sure to run mergemaster or take steps to update /etc/passwd before doing installworld on existing systems. Do not skip the "mergemaster -Fp" step before installworld, as described in the update procedures near the bottom of this document. Also, rc.d/ntpd now starts ntpd(8) as user ntpd if the new mac_ntpd(4) policy is available, unless ntpd_flags or the ntp config file contain options that change file/dir locations. When such options (e.g., "statsdir" or "crypto") are used, ntpd can still be run as non-root by setting ntpd_user=ntpd in rc.conf, after taking steps to ensure that all required files/dirs are accessible by the ntpd user. 20180717: Big endian arm support has been removed. 20180711: The static environment setup in kernel configs is no longer mutually exclusive with the loader(8) environment by default. In order to restore the previous default behavior of disabling the loader(8) environment if a static environment is present, you must specify loader_env.disabled=1 in the static environment. 20180705: The ABI of syscalls used by management tools like sockstat and netstat has been broken to allow 32-bit binaries to work on 64-bit kernels without modification. These programs will need to match the kernel in order to function. External programs may require minor modifications to accommodate a change of type in structures from pointers to 64-bit virtual addresses. 20180702: On i386 and amd64 atomics are now inlined. Out of tree modules using atomics will need to be rebuilt. 20180701: The '%I' format in the kern.corefile sysctl limits the number of core files that a process can generate to the number stored in the debug.ncores sysctl. The '%I' format is replaced by the single digit index. Previously, if all indexes were taken the kernel would overwrite only a core file with the highest index in a filename. Currently the system will create a new core file if there is a free index or if all slots are taken it will overwrite the oldest one. 20180630: Clang, llvm, lld, lldb, compiler-rt and libc++ have been upgraded to 6.0.1. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. 20180628: r335753 introduced a new quoting method. However, etc/devd/devmatch.conf needed to be changed to work with it. This change was made with r335763 and requires a mergemaster / etcupdate / etc to update the installed file. 20180612: r334930 changed the interface between the NFS modules, so they all need to be rebuilt. r335018 did a __FreeBSD_version bump for this. 20180530: As of r334391 lld is the default amd64 system linker; it is installed as /usr/bin/ld. Kernel build workarounds (see 20180510 entry) are no longer necessary. 20180530: The kernel / userland interface for devinfo changed, so you'll need a new kernel and userland as a pair for it to work (rebuilding lib/libdevinfo is all that's required). devinfo and devmatch will not work, but everything else will when there's a mismatch. 20180523: The on-disk format for hwpmc callchain records has changed to include threadid corresponding to a given record. This changes the field offsets and thus requires that libpmcstat be rebuilt before using a kernel later than r334108. 20180517: The vxge(4) driver has been removed. This driver was introduced into HEAD one week before the Exar left the Ethernet market and is not known to be used. If you have device vxge in your kernel config file it must be removed. 20180510: The amd64 kernel now requires a ld that supports ifunc to produce a working kernel, either lld or a newer binutils. lld is built by default on amd64, and the 'buildkernel' target uses it automatically. However, it is not the default linker, so building the kernel the traditional way requires LD=ld.lld on the command line (or LD=/usr/local/bin/ld for binutils port/package). lld will soon be default, and this requirement will go away. NOTE: As of r334391 lld is the default system linker on amd64, and no workaround is necessary. 20180508: The nxge(4) driver has been removed. This driver was for PCI-X 10g cards made by s2io/Neterion. The company was acquired by Exar and no longer sells or supports Ethernet products. If you have device nxge in your kernel config file it must be removed. 20180504: The tz database (tzdb) has been updated to 2018e. This version more correctly models time stamps in time zones with negative DST such as Europe/Dublin (from 1971 on), Europe/Prague (1946/7), and Africa/Windhoek (1994/2017). This does not affect the UT offsets, only time zone abbreviations and the tm_isdst flag. 20180502: The ixgb(4) driver has been removed. This driver was for an early and uncommon legacy PCI 10GbE for a single ASIC, Intel 82597EX. Intel quickly shifted to the long lived ixgbe family. If you have device ixgb in your kernel config file it must be removed. 20180501: The lmc(4) driver has been removed. This was a WAN interface card that was already reportedly rare in 2003, and had an ambiguous license. If you have device lmc in your kernel config file it must be removed. 20180413: Support for Arcnet networks has been removed. If you have device arcnet or device cm in your kernel config file they must be removed. 20180411: Support for FDDI networks has been removed. If you have device fddi or device fpa in your kernel config file they must be removed. 20180406: In addition to supporting RFC 3164 formatted messages, the syslogd(8) service is now capable of parsing RFC 5424 formatted log messages. The main benefit of using RFC 5424 is that clients may now send log messages with timestamps containing year numbers, microseconds and time zone offsets. Similarly, the syslog(3) C library function has been altered to send RFC 5424 formatted messages to the local system logging daemon. On systems using syslogd(8), this change should have no negative impact, as long as syslogd(8) and the C library are updated at the same time. On systems using a different system logging daemon, it may be necessary to make configuration adjustments, depending on the software used. When using syslog-ng, add the 'syslog-protocol' flag to local input sources to enable parsing of RFC 5424 formatted messages: source src { unix-dgram("/var/run/log" flags(syslog-protocol)); } When using rsyslog, disable the 'SysSock.UseSpecialParser' option of the 'imuxsock' module to let messages be processed by the regular RFC 3164/5424 parsing pipeline: module(load="imuxsock" SysSock.UseSpecialParser="off") Do note that these changes only affect communication between local applications and syslogd(8). The format that syslogd(8) uses to store messages on disk or forward messages to other systems remains unchanged. syslogd(8) still uses RFC 3164 for these purposes. Options to customize this behaviour will be added in the future. Utilities that process log files stored in /var/log are thus expected to continue to function as before. __FreeBSD_version has been incremented to 1200061 to denote this change. 20180328: Support for token ring networks has been removed. If you have "device token" in your kernel config you should remove it. No device drivers supported token ring. 20180323: makefs was modified to be able to tag ISO9660 El Torito boot catalog entries as EFI instead of overloading the i386 tag as done previously. The amd64 mkisoimages.sh script used to build amd64 ISO images for release was updated to use this. This may mean that makefs must be updated before "make cdrom" can be run in the release directory. This should be as simple as: $ cd $SRCDIR/usr.sbin/makefs $ make depend all install 20180212: FreeBSD boot loader enhanced with Lua scripting. It's purely opt-in for now by building WITH_LOADER_LUA and WITHOUT_FORTH in /etc/src.conf. Co-existence for the transition period will come shortly. Booting is a complex environment and test coverage for Lua-enabled loaders has been thin, so it would be prudent to assume it might not work and make provisions for backup boot methods. 20180211: devmatch functionality has been turned on in devd. It will automatically load drivers for unattached devices. This may cause unexpected drivers to be loaded. Please report any problems to current@ and imp@freebsd.org. 20180114: Clang, llvm, lld, lldb, compiler-rt and libc++ have been upgraded to 6.0.0. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. 20180110: LLVM's lld linker is now used as the FreeBSD/amd64 bootstrap linker. This means it is used to link the kernel and userland libraries and executables, but is not yet installed as /usr/bin/ld by default. To revert to ld.bfd as the bootstrap linker, in /etc/src.conf set WITHOUT_LLD_BOOTSTRAP=yes 20180110: On i386, pmtimer has been removed. Its functionality has been folded into apm. It was a no-op on ACPI in current for a while now (but was still needed on i386 in FreeBSD 11 and earlier). Users may need to remove it from kernel config files. 20180104: The use of RSS hash from the network card aka flowid has been disabled by default for lagg(4) as it's currently incompatible with the lacp and loadbalance protocols. This can be re-enabled by setting the following in loader.conf: net.link.lagg.default_use_flowid="1" 20180102: The SW_WATCHDOG option is no longer necessary to enable the hardclock-based software watchdog if no hardware watchdog is configured. As before, SW_WATCHDOG will cause the software watchdog to be enabled even if a hardware watchdog is configured. 20171215: r326887 fixes the issue described in the 20171214 UPDATING entry. r326888 flips the switch back to building GELI support always. 20171214: r362593 broke ZFS + GELI support for reasons unknown. However, it also broke ZFS support generally, so GELI has been turned off by default as the lesser evil in r326857. If you boot off ZFS and/or GELI, it might not be a good time to update. 20171125: PowerPC users must update loader(8) by rebuilding world before installing a new kernel, as the protocol connecting them has changed. Without the update, loader metadata will not be passed successfully to the kernel and users will have to enter their root partition at the kernel mountroot prompt to continue booting. Newer versions of loader can boot old kernels without issue. 20171110: The LOADER_FIREWIRE_SUPPORT build variable as been renamed to WITH/OUT_LOADER_FIREWIRE. LOADER_{NO_,}GELI_SUPPORT has been renamed to WITH/OUT_LOADER_GELI. 20171106: The naive and non-compliant support of posix_fallocate(2) in ZFS has been removed as of r325320. The system call now returns EINVAL when used on a ZFS file. Although the new behavior complies with the standard, some consumers are not prepared to cope with it. One known victim is lld prior to r325420. 20171102: Building in a FreeBSD src checkout will automatically create object directories now rather than store files in the current directory if 'make obj' was not ran. Calling 'make obj' is no longer necessary. This feature can be disabled by setting WITHOUT_AUTO_OBJ=yes in /etc/src-env.conf (not /etc/src.conf), or passing the option in the environment. 20171101: The default MAKEOBJDIR has changed from /usr/obj/ for native builds, and /usr/obj// for cross-builds, to a unified /usr/obj//. This behavior can be changed to the old format by setting WITHOUT_UNIFIED_OBJDIR=yes in /etc/src-env.conf, the environment, or with -DWITHOUT_UNIFIED_OBJDIR when building. The UNIFIED_OBJDIR option is a transitional feature that will be removed for 12.0 release; please migrate to the new format for any tools by looking up the OBJDIR used by 'make -V .OBJDIR' means rather than hardcoding paths. 20171028: The native-xtools target no longer installs the files by default to the OBJDIR. Use the native-xtools-install target with a DESTDIR to install to ${DESTDIR}/${NXTP} where NXTP defaults to /nxb-bin. 20171021: As part of the boot loader infrastructure cleanup, LOADER_*_SUPPORT options are changing from controlling the build if defined / undefined to controlling the build with explicit 'yes' or 'no' values. They will shift to WITH/WITHOUT options to match other options in the system. 20171010: libstand has turned into a private library for sys/boot use only. It is no longer supported as a public interface outside of sys/boot. 20171005: The arm port has split armv6 into armv6 and armv7. armv7 is now a valid TARGET_ARCH/MACHINE_ARCH setting. If you have an armv7 system and are running a kernel from before r324363, you will need to add MACHINE_ARCH=armv7 to 'make buildworld' to do a native build. 20171003: When building multiple kernels using KERNCONF, non-existent KERNCONF files will produce an error and buildkernel will fail. Previously missing KERNCONF files silently failed giving no indication as to why, only to subsequently discover during installkernel that the desired kernel was never built in the first place. 20170912: The default serial number format for CTL LUNs has changed. This will affect users who use /dev/diskid/* device nodes, or whose FibreChannel or iSCSI clients care about their LUNs' serial numbers. Users who require serial number stability should hardcode serial numbers in /etc/ctl.conf . 20170912: For 32-bit arm compiled for hard-float support, soft-floating point binaries now always get their shared libraries from LD_SOFT_LIBRARY_PATH (in the past, this was only used if /usr/libsoft also existed). Only users with a hard-float ld.so, but soft-float everything else should be affected. 20170826: The geli password typed at boot is now hidden. To restore the previous behavior, see geli(8) for configuration options. 20170825: Move PMTUD blackhole counters to TCPSTATS and remove them from bare sysctl values. Minor nit, but requires a rebuild of both world/kernel to complete. 20170814: "make check" behavior (made in ^/head@r295380) has been changed to execute from a limited sandbox, as opposed to executing from ${TESTSDIR}. Behavioral changes: - The "beforecheck" and "aftercheck" targets are now specified. - ${CHECKDIR} (added in commit noted above) has been removed. - Legacy behavior can be enabled by setting WITHOUT_MAKE_CHECK_USE_SANDBOX in src.conf(5) or the environment. If the limited sandbox mode is enabled, "make check" will execute "make distribution", then install, execute the tests, and clean up the sandbox if successful. The "make distribution" and "make install" targets are typically run as root to set appropriate permissions and ownership at installation time. The end-user should set "WITH_INSTALL_AS_USER" in src.conf(5) or the environment if executing "make check" with limited sandbox mode using an unprivileged user. 20170808: Since the switch to GPT disk labels, fsck for UFS/FFS has been unable to automatically find alternate superblocks. As of r322297, the information needed to find alternate superblocks has been moved to the end of the area reserved for the boot block. Filesystems created with a newfs of this vintage or later will create the recovery information. If you have a filesystem created prior to this change and wish to have a recovery block created for your filesystem, you can do so by running fsck in foreground mode (i.e., do not use the -p or -y options). As it starts, fsck will ask ``SAVE DATA TO FIND ALTERNATE SUPERBLOCKS'' to which you should answer yes. 20170728: As of r321665, an NFSv4 server configuration that services Kerberos mounts or clients that do not support the uid/gid in owner/owner_group string capability, must explicitly enable the nfsuserd daemon by adding nfsuserd_enable="YES" to the machine's /etc/rc.conf file. 20170722: Clang, llvm, lldb, compiler-rt and libc++ have been upgraded to 5.0.0. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. 20170701: WITHOUT_RCMDS is now the default. Set WITH_RCMDS if you need the r-commands (rlogin, rsh, etc.) to be built with the base system. 20170625: The FreeBSD/powerpc platform now uses a 64-bit type for time_t. This is a very major ABI incompatible change, so users of FreeBSD/powerpc must be careful when performing source upgrades. It is best to run 'make installworld' from an alternate root system, either a live CD/memory stick, or a temporary root partition. Additionally, all ports must be recompiled. powerpc64 is largely unaffected, except in the case of 32-bit compatibility. All 32-bit binaries will be affected. 20170623: Forward compatibility for the "ino64" project have been committed. This will allow most new binaries to run on older kernels in a limited fashion. This prevents many of the common foot-shooting actions in the upgrade as well as the limited ability to roll back the kernel across the ino64 upgrade. Complicated use cases may not work properly, though enough simpler ones work to allow recovery in most situations. 20170620: Switch back to the BSDL dtc (Device Tree Compiler). Set WITH_GPL_DTC if you require the GPL compiler. 20170618: The internal ABI used for communication between the NFS kernel modules was changed by r320085, so __FreeBSD_version was bumped to ensure all the NFS related modules are updated together. 20170617: The ABI of struct event was changed by extending the data member to 64bit and adding ext fields. For upgrade, same precautions as for the entry 20170523 "ino64" must be followed. 20170531: The GNU roff toolchain has been removed from base. To render manpages which are not supported by mandoc(1), man(1) can fallback on GNU roff from ports (and recommends to install it). To render roff(7) documents, consider using GNU roff from ports or the heirloom doctools roff toolchain from ports via pkg install groff or via pkg install heirloom-doctools. 20170524: The ath(4) and ath_hal(4) modules now build piecemeal to allow for smaller runtime footprint builds. This is useful for embedded systems which only require one chipset support. If you load it as a module, make sure this is in /boot/loader.conf: if_ath_load="YES" This will load the HAL, all chip/RF backends and if_ath_pci. If you have if_ath_pci in /boot/loader.conf, ensure it is after if_ath or it will not load any HAL chipset support. If you want to selectively load things (eg on ye cheape ARM/MIPS platforms where RAM is at a premium) you should: * load ath_hal * load the chip modules in question * load ath_rate, ath_dfs * load ath_main * load if_ath_pci and/or if_ath_ahb depending upon your particular bus bind type - this is where probe/attach is done. For further comments/feedback, poke adrian@ . 20170523: The "ino64" 64-bit inode project has been committed, which extends a number of types to 64 bits. Upgrading in place requires care and adherence to the documented upgrade procedure. If using a custom kernel configuration ensure that the COMPAT_FREEBSD11 option is included (as during the upgrade the system will be running the ino64 kernel with the existing world). For the safest in-place upgrade begin by removing previous build artifacts via "rm -rf /usr/obj/*". Then, carefully follow the full procedure documented below under the heading "To rebuild everything and install it on the current system." Specifically, a reboot is required after installing the new kernel before installing world. While an installworld normally works by accident from multiuser after rebooting the proper kernel, there are many cases where this will fail across this upgrade and installworld from single user is required. 20170424: The NATM framework including the en(4), fatm(4), hatm(4), and patm(4) devices has been removed. Consumers should plan a migration before the end-of-life date for FreeBSD 11. 20170420: GNU diff has been replaced by a BSD licensed diff. Some features of GNU diff has not been implemented, if those are needed a newer version of GNU diff is available via the diffutils package under the gdiff name. 20170413: As of r316810 for ipfilter, keep frags is no longer assumed when keep state is specified in a rule. r316810 aligns ipfilter with documentation in man pages separating keep frags from keep state. This allows keep state to be specified without forcing keep frags and allows keep frags to be specified independently of keep state. To maintain previous behaviour, also specify keep frags with keep state (as documented in ipf.conf.5). 20170407: arm64 builds now use the base system LLD 4.0.0 linker by default, instead of requiring that the aarch64-binutils port or package be installed. To continue using aarch64-binutils, set CROSS_BINUTILS_PREFIX=/usr/local/aarch64-freebsd/bin . 20170405: The UDP optimization in entry 20160818 that added the sysctl net.inet.udp.require_l2_bcast has been reverted. L2 broadcast packets will no longer be treated as L3 broadcast packets. 20170331: Binds and sends to the loopback addresses, IPv6 and IPv4, will now use any explicitly assigned loopback address available in the jail instead of using the first assigned address of the jail. 20170329: The ctl.ko module no longer implements the iSCSI target frontend: cfiscsi.ko does instead. If building cfiscsi.ko as a kernel module, the module can be loaded via one of the following methods: - `cfiscsi_load="YES"` in loader.conf(5). - Add `cfiscsi` to `$kld_list` in rc.conf(5). - ctladm(8)/ctld(8), when compiled with iSCSI support (`WITH_ISCSI=yes` in src.conf(5)) Please see cfiscsi(4) for more details. 20170316: The mmcsd.ko module now additionally depends on geom_flashmap.ko. Also, mmc.ko and mmcsd.ko need to be a matching pair built from the same source (previously, the dependency of mmcsd.ko on mmc.ko was missing, but mmcsd.ko now will refuse to load if it is incompatible with mmc.ko). 20170315: The syntax of ipfw(8) named states was changed to avoid ambiguity. If you have used named states in the firewall rules, you need to modify them after installworld and before rebooting. Now named states must be prefixed with colon. 20170311: The old drm (sys/dev/drm/) drivers for i915 and radeon have been removed as the userland we provide cannot use them. The KMS version (sys/dev/drm2) supports the same hardware. 20170302: Clang, llvm, lldb, compiler-rt and libc++ have been upgraded to 4.0.0. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. 20170221: The code that provides support for ZFS .zfs/ directory functionality has been reimplemented. It's not possible now to create a snapshot by mkdir under .zfs/snapshot/. That should be the only user visible change. 20170216: EISA bus support has been removed. The WITH_EISA option is no longer valid. 20170215: MCA bus support has been removed. 20170127: The WITH_LLD_AS_LD / WITHOUT_LLD_AS_LD build knobs have been renamed WITH_LLD_IS_LD / WITHOUT_LLD_IS_LD, for consistency with CLANG_IS_CC. 20170112: The EM_MULTIQUEUE kernel configuration option is deprecated now that the em(4) driver conforms to iflib specifications. 20170109: The igb(4), em(4) and lem(4) ethernet drivers are now implemented via IFLIB. If you have a custom kernel configuration that excludes em(4) but you use igb(4), you need to re-add em(4) to your custom configuration. 20161217: Clang, llvm, lldb, compiler-rt and libc++ have been upgraded to 3.9.1. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. 20161124: Clang, llvm, lldb, compiler-rt and libc++ have been upgraded to 3.9.0. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. 20161119: The layout of the pmap structure has changed for powerpc to put the pmap statistics at the front for all CPU variations. libkvm(3) and all tools that link against it need to be recompiled. 20161030: isl(4) and cyapa(4) drivers now require a new driver, chromebook_platform(4), to work properly on Chromebook-class hardware. On other types of hardware the drivers may need to be configured using device hints. Please see the corresponding manual pages for details. 20161017: The urtwn(4) driver was merged into rtwn(4) and now consists of rtwn(4) main module + rtwn_usb(4) and rtwn_pci(4) bus-specific parts. Also, firmware for RTL8188CE was renamed due to possible name conflict (rtwnrtl8192cU(B) -> rtwnrtl8192cE(B)) 20161015: GNU rcs has been removed from base. It is available as packages: - rcs: Latest GPLv3 GNU rcs version. - rcs57: Copy of the latest version of GNU rcs (GPLv2) before it was removed from base. 20161008: Use of the cc_cdg, cc_chd, cc_hd, or cc_vegas congestion control modules now requires that the kernel configuration contain the TCP_HHOOK option. (This option is included in the GENERIC kernel.) 20161003: The WITHOUT_ELFCOPY_AS_OBJCOPY src.conf(5) knob has been retired. ELF Tool Chain's elfcopy is always installed as /usr/bin/objcopy. 20160924: Relocatable object files with the extension of .So have been renamed to use an extension of .pico instead. The purpose of this change is to avoid a name clash with shared libraries on case-insensitive file systems. On those file systems, foo.So is the same file as foo.so. 20160918: GNU rcs has been turned off by default. It can (temporarily) be built again by adding WITH_RCS knob in src.conf. Otherwise, GNU rcs is available from packages: - rcs: Latest GPLv3 GNU rcs version. - rcs57: Copy of the latest version of GNU rcs (GPLv2) from base. 20160918: The backup_uses_rcs functionality has been removed from rc.subr. 20160908: The queue(3) debugging macro, QUEUE_MACRO_DEBUG, has been split into two separate components, QUEUE_MACRO_DEBUG_TRACE and QUEUE_MACRO_DEBUG_TRASH. Define both for the original QUEUE_MACRO_DEBUG behavior. 20160824: r304787 changed some ioctl interfaces between the iSCSI userspace programs and the kernel. ctladm, ctld, iscsictl, and iscsid must be rebuilt to work with new kernels. __FreeBSD_version has been bumped to 1200005. 20160818: The UDP receive code has been updated to only treat incoming UDP packets that were addressed to an L2 broadcast address as L3 broadcast packets. It is not expected that this will affect any standards-conforming UDP application. The new behaviour can be disabled by setting the sysctl net.inet.udp.require_l2_bcast to 0. 20160818: Remove the openbsd_poll system call. __FreeBSD_version has been bumped because of this. 20160708: The stable/11 branch has been created from head@r302406. 20160622: The libc stub for the pipe(2) system call has been replaced with a wrapper that calls the pipe2(2) system call and the pipe(2) system call is now only implemented by the kernels that include "options COMPAT_FREEBSD10" in their config file (this is the default). Users should ensure that this option is enabled in their kernel or upgrade userspace to r302092 before upgrading their kernel. 20160527: CAM will now strip leading spaces from SCSI disks' serial numbers. This will affect users who create UFS filesystems on SCSI disks using those disk's diskid device nodes. For example, if /etc/fstab previously contained a line like "/dev/diskid/DISK-%20%20%20%20%20%20%20ABCDEFG0123456", you should change it to "/dev/diskid/DISK-ABCDEFG0123456". Users of geom transforms like gmirror may also be affected. ZFS users should generally be fine. 20160523: The bitstring(3) API has been updated with new functionality and improved performance. But it is binary-incompatible with the old API. Objects built with the new headers may not be linked against objects built with the old headers. 20160520: The brk and sbrk functions have been removed from libc on arm64. Binutils from ports has been updated to not link to these functions and should be updated to the latest version before installing a new libc. 20160517: The armv6 port now defaults to hard float ABI. Limited support for running both hardfloat and soft float on the same system is available using the libraries installed with -DWITH_LIBSOFT. This has only been tested as an upgrade path for installworld and packages may fail or need manual intervention to run. New packages will be needed. To update an existing self-hosted armv6hf system, you must add TARGET_ARCH=armv6 on the make command line for both the build and the install steps. 20160510: Kernel modules compiled outside of a kernel build now default to installing to /boot/modules instead of /boot/kernel. Many kernel modules built this way (such as those in ports) already overrode KMODDIR explicitly to install into /boot/modules. However, manually building and installing a module from /sys/modules will now install to /boot/modules instead of /boot/kernel. 20160414: The CAM I/O scheduler has been committed to the kernel. There should be no user visible impact. This does enable NCQ Trim on ada SSDs. While the list of known rogues that claim support for this but actually corrupt data is believed to be complete, be on the lookout for data corruption. The known rogue list is believed to be complete: o Crucial MX100, M550 drives with MU01 firmware. o Micron M510 and M550 drives with MU01 firmware. o Micron M500 prior to MU07 firmware o Samsung 830, 840, and 850 all firmwares o FCCT M500 all firmwares Crucial has firmware http://www.crucial.com/usa/en/support-ssd-firmware with working NCQ TRIM. For Micron branded drives, see your sales rep for updated firmware. Black listed drives will work correctly because these drives work correctly so long as no NCQ TRIMs are sent to them. Given this list is the same as found in Linux, it's believed there are no other rogues in the market place. All other models from the above vendors work. To be safe, if you are at all concerned, you can quirk each of your drives to prevent NCQ from being sent by setting: kern.cam.ada.X.quirks="0x2" in loader.conf. If the drive requires the 4k sector quirk, set the quirks entry to 0x3. 20160330: The FAST_DEPEND build option has been removed and its functionality is now the one true way. The old mkdep(1) style of 'make depend' has been removed. See 20160311 for further details. 20160317: Resource range types have grown from unsigned long to uintmax_t. All drivers, and anything using libdevinfo, need to be recompiled. 20160311: WITH_FAST_DEPEND is now enabled by default for in-tree and out-of-tree builds. It no longer runs mkdep(1) during 'make depend', and the 'make depend' stage can safely be skipped now as it is auto ran when building 'make all' and will generate all SRCS and DPSRCS before building anything else. Dependencies are gathered at compile time with -MF flags kept in separate .depend files per object file. Users should run 'make cleandepend' once if using -DNO_CLEAN to clean out older stale .depend files. 20160306: On amd64, clang 3.8.0 can now insert sections of type AMD64_UNWIND into kernel modules. Therefore, if you load any kernel modules at boot time, please install the boot loaders after you install the kernel, but before rebooting, e.g.: make buildworld make buildkernel KERNCONF=YOUR_KERNEL_HERE make installkernel KERNCONF=YOUR_KERNEL_HERE make -C sys/boot install Then follow the usual steps, described in the General Notes section, below. 20160305: Clang, llvm, lldb and compiler-rt have been upgraded to 3.8.0. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. 20160301: The AIO subsystem is now a standard part of the kernel. The VFS_AIO kernel option and aio.ko kernel module have been removed. Due to stability concerns, asynchronous I/O requests are only permitted on sockets and raw disks by default. To enable asynchronous I/O requests on all file types, set the vfs.aio.enable_unsafe sysctl to a non-zero value. 20160226: The ELF object manipulation tool objcopy is now provided by the ELF Tool Chain project rather than by GNU binutils. It should be a drop-in replacement, with the addition of arm64 support. The (temporary) src.conf knob WITHOUT_ELFCOPY_AS_OBJCOPY knob may be set to obtain the GNU version if necessary. 20160129: Building ZFS pools on top of zvols is prohibited by default. That feature has never worked safely; it's always been prone to deadlocks. Using a zvol as the backing store for a VM guest's virtual disk will still work, even if the guest is using ZFS. Legacy behavior can be restored by setting vfs.zfs.vol.recursive=1. 20160119: The NONE and HPN patches has been removed from OpenSSH. They are still available in the security/openssh-portable port. 20160113: With the addition of ypldap(8), a new _ypldap user is now required during installworld. "mergemaster -p" can be used to add the user prior to installworld, as documented in the handbook. 20151216: The tftp loader (pxeboot) now uses the option root-path directive. As a consequence it no longer looks for a pxeboot.4th file on the tftp server. Instead it uses the regular /boot infrastructure as with the other loaders. 20151211: The code to start recording plug and play data into the modules has been committed. While the old tools will properly build a new kernel, a number of warnings about "unknown metadata record 4" will be produced for an older kldxref. To avoid such warnings, make sure to rebuild the kernel toolchain (or world). Make sure that you have r292078 or later when trying to build 292077 or later before rebuilding. 20151207: Debug data files are now built by default with 'make buildworld' and installed with 'make installworld'. This facilitates debugging but requires more disk space both during the build and for the installed world. Debug files may be disabled by setting WITHOUT_DEBUG_FILES=yes in src.conf(5). 20151130: r291527 changed the internal interface between the nfsd.ko and nfscommon.ko modules. As such, they must both be upgraded to-gether. __FreeBSD_version has been bumped because of this. 20151108: Add support for unicode collation strings leads to a change of order of files listed by ls(1) for example. To get back to the old behaviour, set LC_COLLATE environment variable to "C". Databases administrators will need to reindex their databases given collation results will be different. Due to a bug in install(1) it is recommended to remove the ancient locales before running make installworld. rm -rf /usr/share/locale/* 20151030: The OpenSSL has been upgraded to 1.0.2d. Any binaries requiring libcrypto.so.7 or libssl.so.7 must be recompiled. 20151020: Qlogic 24xx/25xx firmware images were updated from 5.5.0 to 7.3.0. Kernel modules isp_2400_multi and isp_2500_multi were removed and should be replaced with isp_2400 and isp_2500 modules respectively. 20151017: The build previously allowed using 'make -n' to not recurse into sub-directories while showing what commands would be executed, and 'make -n -n' to recursively show commands. Now 'make -n' will recurse and 'make -N' will not. 20151012: If you specify SENDMAIL_MC or SENDMAIL_CF in make.conf, mergemaster and etcupdate will now use this file. A custom sendmail.cf is now updated via this mechanism rather than via installworld. If you had excluded sendmail.cf in mergemaster.rc or etcupdate.conf, you may want to remove the exclusion or change it to "always install". /etc/mail/sendmail.cf is now managed the same way regardless of whether SENDMAIL_MC/SENDMAIL_CF is used. If you are not using SENDMAIL_MC/SENDMAIL_CF there should be no change in behavior. 20151011: Compatibility shims for legacy ATA device names have been removed. It includes ATA_STATIC_ID kernel option, kern.cam.ada.legacy_aliases and kern.geom.raid.legacy_aliases loader tunables, kern.devalias.* environment variables, /dev/ad* and /dev/ar* symbolic links. 20151006: Clang, llvm, lldb, compiler-rt and libc++ have been upgraded to 3.7.0. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using clang 3.5.0 or higher. 20150924: Kernel debug files have been moved to /usr/lib/debug/boot/kernel/, and renamed from .symbols to .debug. This reduces the size requirements on the boot partition or file system and provides consistency with userland debug files. When using the supported kernel installation method the /usr/lib/debug/boot/kernel directory will be renamed (to kernel.old) as is done with /boot/kernel. Developers wishing to maintain the historical behavior of installing debug files in /boot/kernel/ can set KERN_DEBUGDIR="" in src.conf(5). 20150827: The wireless drivers had undergone changes that remove the 'parent interface' from the ifconfig -l output. The rc.d network scripts used to check presence of a parent interface in the list, so old scripts would fail to start wireless networking. Thus, etcupdate(3) or mergemaster(8) run is required after kernel update, to update your rc.d scripts in /etc. 20150827: pf no longer supports 'scrub fragment crop' or 'scrub fragment drop-ovl' These configurations are now automatically interpreted as 'scrub fragment reassemble'. 20150817: Kernel-loadable modules for the random(4) device are back. To use them, the kernel must have device random options RANDOM_LOADABLE kldload(8) can then be used to load random_fortuna.ko or random_yarrow.ko. Please note that due to the indirect function calls that the loadable modules need to provide, the build-in variants will be slightly more efficient. The random(4) kernel option RANDOM_DUMMY has been retired due to unpopularity. It was not all that useful anyway. 20150813: The WITHOUT_ELFTOOLCHAIN_TOOLS src.conf(5) knob has been retired. Control over building the ELF Tool Chain tools is now provided by the WITHOUT_TOOLCHAIN knob. 20150810: The polarity of Pulse Per Second (PPS) capture events with the uart(4) driver has been corrected. Prior to this change the PPS "assert" event corresponded to the trailing edge of a positive PPS pulse and the "clear" event was the leading edge of the next pulse. As the width of a PPS pulse in a typical GPS receiver is on the order of 1 millisecond, most users will not notice any significant difference with this change. Anyone who has compensated for the historical polarity reversal by configuring a negative offset equal to the pulse width will need to remove that workaround. 20150809: The default group assigned to /dev/dri entries has been changed from 'wheel' to 'video' with the id of '44'. If you want to have access to the dri devices please add yourself to the video group with: # pw groupmod video -m $USER 20150806: The menu.rc and loader.rc files will now be replaced during upgrades. Please migrate local changes to menu.rc.local and loader.rc.local instead. 20150805: GNU Binutils versions of addr2line, c++filt, nm, readelf, size, strings and strip have been removed. The src.conf(5) knob WITHOUT_ELFTOOLCHAIN_TOOLS no longer provides the binutils tools. 20150728: As ZFS requires more kernel stack pages than is the default on some architectures e.g. i386, it now warns if KSTACK_PAGES is less than ZFS_MIN_KSTACK_PAGES (which is 4 at the time of writing). Please consider using 'options KSTACK_PAGES=X' where X is greater than or equal to ZFS_MIN_KSTACK_PAGES i.e. 4 in such configurations. 20150706: sendmail has been updated to 8.15.2. Starting with FreeBSD 11.0 and sendmail 8.15, sendmail uses uncompressed IPv6 addresses by default, i.e., they will not contain "::". For example, instead of ::1, it will be 0:0:0:0:0:0:0:1. This permits a zero subnet to have a more specific match, such as different map entries for IPv6:0:0 vs IPv6:0. This change requires that configuration data (including maps, files, classes, custom ruleset, etc.) must use the same format, so make certain such configuration data is upgrading. As a very simple check search for patterns like 'IPv6:[0-9a-fA-F:]*::' and 'IPv6::'. To return to the old behavior, set the m4 option confUSE_COMPRESSED_IPV6_ADDRESSES or the cf option UseCompressedIPv6Addresses. 20150630: The default kernel entropy-processing algorithm is now Fortuna, replacing Yarrow. Assuming you have 'device random' in your kernel config file, the configurations allow a kernel option to override this default. You may choose *ONE* of: options RANDOM_YARROW # Legacy /dev/random algorithm. options RANDOM_DUMMY # Blocking-only driver. If you have neither, you get Fortuna. For most people, read no further, Fortuna will give a /dev/random that works like it always used to, and the difference will be irrelevant. If you remove 'device random', you get *NO* kernel-processed entropy at all. This may be acceptable to folks building embedded systems, but has complications. Carry on reading, and it is assumed you know what you need. *PLEASE* read random(4) and random(9) if you are in the habit of tweaking kernel configs, and/or if you are a member of the embedded community, wanting specific and not-usual behaviour from your security subsystems. NOTE!! If you use RANDOM_DUMMY and/or have no 'device random', you will NOT have a functioning /dev/random, and many cryptographic features will not work, including SSH. You may also find strange behaviour from the random(3) set of library functions, in particular sranddev(3), srandomdev(3) and arc4random(3). The reason for this is that the KERN_ARND sysctl only returns entropy if it thinks it has some to share, and with RANDOM_DUMMY or no 'device random' this will never happen. 20150623: An additional fix for the issue described in the 20150614 sendmail entry below has been committed in revision 284717. 20150616: FreeBSD's old make (fmake) has been removed from the system. It is available as the devel/fmake port or via pkg install fmake. 20150615: The fix for the issue described in the 20150614 sendmail entry below has been committed in revision 284436. The work around described in that entry is no longer needed unless the default setting is overridden by a confDH_PARAMETERS configuration setting of '5' or pointing to a 512 bit DH parameter file. 20150614: ALLOW_DEPRECATED_ATF_TOOLS/ATFFILE support has been removed from atf.test.mk (included from bsd.test.mk). Please upgrade devel/atf and devel/kyua to version 0.20+ and adjust any calling code to work with Kyuafile and kyua. 20150614: The import of openssl to address the FreeBSD-SA-15:10.openssl security advisory includes a change which rejects handshakes with DH parameters below 768 bits. sendmail releases prior to 8.15.2 (not yet released), defaulted to a 512 bit DH parameter setting for client connections. To work around this interoperability, sendmail can be configured to use a 2048 bit DH parameter by: 1. Edit /etc/mail/`hostname`.mc 2. If a setting for confDH_PARAMETERS does not exist or exists and is set to a string beginning with '5', replace it with '2'. 3. If a setting for confDH_PARAMETERS exists and is set to a file path, create a new file with: openssl dhparam -out /path/to/file 2048 4. Rebuild the .cf file: cd /etc/mail/; make; make install 5. Restart sendmail: cd /etc/mail/; make restart A sendmail patch is coming, at which time this file will be updated. 20150604: Generation of legacy formatted entries have been disabled by default in pwd_mkdb(8), as all base system consumers of the legacy formatted entries were converted to use the new format by default when the new, machine independent format have been added and supported since FreeBSD 5.x. Please see the pwd_mkdb(8) manual page for further details. 20150525: Clang and llvm have been upgraded to 3.6.1 release. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using 3.5.0 or higher. 20150521: TI platform code switched to using vendor DTS files and this update may break existing systems running on Beaglebone, Beaglebone Black, and Pandaboard: - dtb files should be regenerated/reinstalled. Filenames are the same but content is different now - GPIO addressing was changed, now each GPIO bank (32 pins per bank) has its own /dev/gpiocX device, e.g. pin 121 on /dev/gpioc0 in old addressing scheme is now pin 25 on /dev/gpioc3. - Pandaboard: /etc/ttys should be updated, serial console device is now /dev/ttyu2, not /dev/ttyu0 20150501: soelim(1) from gnu/usr.bin/groff has been replaced by usr.bin/soelim. If you need the GNU extension from groff soelim(1), install groff from package: pkg install groff, or via ports: textproc/groff. 20150423: chmod, chflags, chown and chgrp now affect symlinks in -R mode as defined in symlink(7); previously symlinks were silently ignored. 20150415: The const qualifier has been removed from iconv(3) to comply with POSIX. The ports tree is aware of this from r384038 onwards. 20150416: Libraries specified by LIBADD in Makefiles must have a corresponding DPADD_ variable to ensure correct dependencies. This is now enforced in src.libnames.mk. 20150324: From legacy ata(4) driver was removed support for SATA controllers supported by more functional drivers ahci(4), siis(4) and mvs(4). Kernel modules ataahci and ataadaptec were removed completely, replaced by ahci and mvs modules respectively. 20150315: Clang, llvm and lldb have been upgraded to 3.6.0 release. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using 3.5.0 or higher. 20150307: The 32-bit PowerPC kernel has been changed to a position-independent executable. This can only be booted with a version of loader(8) newer than January 31, 2015, so make sure to update both world and kernel before rebooting. 20150217: If you are running a -CURRENT kernel since r273872 (Oct 30th, 2014), but before r278950, the RNG was not seeded properly. Immediately upgrade the kernel to r278950 or later and regenerate any keys (e.g. ssh keys or openssl keys) that were generated w/ a kernel from that range. This does not affect programs that directly used /dev/random or /dev/urandom. All userland uses of arc4random(3) are affected. 20150210: The autofs(4) ABI was changed in order to restore binary compatibility with 10.1-RELEASE. The automountd(8) daemon needs to be rebuilt to work with the new kernel. 20150131: The powerpc64 kernel has been changed to a position-independent executable. This can only be booted with a new version of loader(8), so make sure to update both world and kernel before rebooting. 20150118: Clang and llvm have been upgraded to 3.5.1 release. This is a bugfix only release, no new features have been added. Please see the 20141231 entry below for information about prerequisites and upgrading, if you are not already using 3.5.0. 20150107: ELF tools addr2line, elfcopy (strip), nm, size, and strings are now taken from the ELF Tool Chain project rather than GNU binutils. They should be drop-in replacements, with the addition of arm64 support. The WITHOUT_ELFTOOLCHAIN_TOOLS= knob may be used to obtain the binutils tools, if necessary. See 20150805 for updated information. 20150105: The default Unbound configuration now enables remote control using a local socket. Users who have already enabled the local_unbound service should regenerate their configuration by running "service local_unbound setup" as root. 20150102: The GNU texinfo and GNU info pages have been removed. To be able to view GNU info pages please install texinfo from ports. 20141231: Clang, llvm and lldb have been upgraded to 3.5.0 release. As of this release, a prerequisite for building clang, llvm and lldb is a C++11 capable compiler and C++11 standard library. This means that to be able to successfully build the cross-tools stage of buildworld, with clang as the bootstrap compiler, your system compiler or cross compiler should either be clang 3.3 or later, or gcc 4.8 or later, and your system C++ library should be libc++, or libdstdc++ from gcc 4.8 or later. On any standard FreeBSD 10.x or 11.x installation, where clang and libc++ are on by default (that is, on x86 or arm), this should work out of the box. On 9.x installations where clang is enabled by default, e.g. on x86 and powerpc, libc++ will not be enabled by default, so libc++ should be built (with clang) and installed first. If both clang and libc++ are missing, build clang first, then use it to build libc++. On 8.x and earlier installations, upgrade to 9.x first, and then follow the instructions for 9.x above. Sparc64 and mips users are unaffected, as they still use gcc 4.2.1 by default, and do not build clang. Many embedded systems are resource constrained, and will not be able to build clang in a reasonable time, or in some cases at all. In those cases, cross building bootable systems on amd64 is a workaround. This new version of clang introduces a number of new warnings, of which the following are most likely to appear: -Wabsolute-value This warns in two cases, for both C and C++: * When the code is trying to take the absolute value of an unsigned quantity, which is effectively a no-op, and almost never what was intended. The code should be fixed, if at all possible. If you are sure that the unsigned quantity can be safely cast to signed, without loss of information or undefined behavior, you can add an explicit cast, or disable the warning. * When the code is trying to take an absolute value, but the called abs() variant is for the wrong type, which can lead to truncation. If you want to disable the warning instead of fixing the code, please make sure that truncation will not occur, or it might lead to unwanted side-effects. -Wtautological-undefined-compare and -Wundefined-bool-conversion These warn when C++ code is trying to compare 'this' against NULL, while 'this' should never be NULL in well-defined C++ code. However, there is some legacy (pre C++11) code out there, which actively abuses this feature, which was less strictly defined in previous C++ versions. Squid and openjdk do this, for example. The warning can be turned off for C++98 and earlier, but compiling the code in C++11 mode might result in unexpected behavior; for example, the parts of the program that are unreachable could be optimized away. 20141222: The old NFS client and server (kernel options NFSCLIENT, NFSSERVER) kernel sources have been removed. The .h files remain, since some utilities include them. This will need to be fixed later. If "mount -t oldnfs ..." is attempted, it will fail. If the "-o" option on mountd(8), nfsd(8) or nfsstat(1) is used, the utilities will report errors. 20141121: The handling of LOCAL_LIB_DIRS has been altered to skip addition of directories to top level SUBDIR variable when their parent directory is included in LOCAL_DIRS. Users with build systems with such hierarchies and without SUBDIR entries in the parent directory Makefiles should add them or add the directories to LOCAL_DIRS. 20141109: faith(4) and faithd(8) have been removed from the base system. Faith has been obsolete for a very long time. 20141104: vt(4), the new console driver, is enabled by default. It brings support for Unicode and double-width characters, as well as support for UEFI and integration with the KMS kernel video drivers. You may need to update your console settings in /etc/rc.conf, most probably the keymap. During boot, /etc/rc.d/syscons will indicate what you need to do. vt(4) still has issues and lacks some features compared to syscons(4). See the wiki for up-to-date information: https://wiki.freebsd.org/Newcons If you want to keep using syscons(4), you can do so by adding the following line to /boot/loader.conf: kern.vty=sc 20141102: pjdfstest has been integrated into kyua as an opt-in test suite. Please see share/doc/pjdfstest/README for more details on how to execute it. 20141009: gperf has been removed from the base system for architectures that use clang. Ports that require gperf will obtain it from the devel/gperf port. 20140923: pjdfstest has been moved from tools/regression/pjdfstest to contrib/pjdfstest . 20140922: At svn r271982, The default linux compat kernel ABI has been adjusted to 2.6.18 in support of the linux-c6 compat ports infrastructure update. If you wish to continue using the linux-f10 compat ports, add compat.linux.osrelease=2.6.16 to your local sysctl.conf. Users are encouraged to update their linux-compat packages to linux-c6 during their next update cycle. 20140729: The ofwfb driver, used to provide a graphics console on PowerPC when using vt(4), no longer allows mmap() of all physical memory. This will prevent Xorg on PowerPC with some ATI graphics cards from initializing properly unless x11-servers/xorg-server is updated to 1.12.4_8 or newer. 20140723: The xdev targets have been converted to using TARGET and TARGET_ARCH instead of XDEV and XDEV_ARCH. 20140719: The default unbound configuration has been modified to address issues with reverse lookups on networks that use private address ranges. If you use the local_unbound service, run "service local_unbound setup" as root to regenerate your configuration, then "service local_unbound reload" to load the new configuration. 20140709: The GNU texinfo and GNU info pages are not built and installed anymore, WITH_INFO knob has been added to allow to built and install them again. UPDATE: see 20150102 entry on texinfo's removal 20140708: The GNU readline library is now an INTERNALLIB - that is, it is statically linked into consumers (GDB and variants) in the base system, and the shared library is no longer installed. The devel/readline port is available for third party software that requires readline. 20140702: The Itanium architecture (ia64) has been removed from the list of known architectures. This is the first step in the removal of the architecture. 20140701: Commit r268115 has added NFSv4.1 server support, merged from projects/nfsv4.1-server. Since this includes changes to the internal interfaces between the NFS related modules, a full build of the kernel and modules will be necessary. __FreeBSD_version has been bumped. 20140629: The WITHOUT_VT_SUPPORT kernel config knob has been renamed WITHOUT_VT. (The other _SUPPORT knobs have a consistent meaning which differs from the behaviour controlled by this knob.) 20140619: Maximal length of the serial number in CTL was increased from 16 to 64 chars, that breaks ABI. All CTL-related tools, such as ctladm and ctld, need to be rebuilt to work with a new kernel. 20140606: The libatf-c and libatf-c++ major versions were downgraded to 0 and 1 respectively to match the upstream numbers. They were out of sync because, when they were originally added to FreeBSD, the upstream versions were not respected. These libraries are private and not yet built by default, so renumbering them should be a non-issue. However, unclean source trees will yield broken test programs once the operator executes "make delete-old-libs" after a "make installworld". Additionally, the atf-sh binary was made private by moving it into /usr/libexec/. Already-built shell test programs will keep the path to the old binary so they will break after "make delete-old" is run. If you are using WITH_TESTS=yes (not the default), wipe the object tree and rebuild from scratch to prevent spurious test failures. This is only needed once: the misnumbered libraries and misplaced binaries have been added to OptionalObsoleteFiles.inc so they will be removed during a clean upgrade. 20140512: Clang and llvm have been upgraded to 3.4.1 release. 20140508: We bogusly installed src.opts.mk in /usr/share/mk. This file should be removed to avoid issues in the future (and has been added to ObsoleteFiles.inc). 20140505: /etc/src.conf now affects only builds of the FreeBSD src tree. In the past, it affected all builds that used the bsd.*.mk files. The old behavior was a bug, but people may have relied upon it. To get this behavior back, you can .include /etc/src.conf from /etc/make.conf (which is still global and isn't changed). This also changes the behavior of incremental builds inside the tree of individual directories. Set MAKESYSPATH to ".../share/mk" to do that. Although this has survived make universe and some upgrade scenarios, other upgrade scenarios may have broken. At least one form of temporary breakage was fixed with MAKESYSPATH settings for buildworld as well... In cases where MAKESYSPATH isn't working with this setting, you'll need to set it to the full path to your tree. One side effect of all this cleaning up is that bsd.compiler.mk is no longer implicitly included by bsd.own.mk. If you wish to use COMPILER_TYPE, you must now explicitly include bsd.compiler.mk as well. 20140430: The lindev device has been removed since /dev/full has been made a standard device. __FreeBSD_version has been bumped. 20140424: The knob WITHOUT_VI was added to the base system, which controls building ex(1), vi(1), etc. Older releases of FreeBSD required ex(1) in order to reorder files share/termcap and didn't build ex(1) as a build tool, so building/installing with WITH_VI is highly advised for build hosts for older releases. This issue has been fixed in stable/9 and stable/10 in r277022 and r276991, respectively. 20140418: The YES_HESIOD knob has been removed. It has been obsolete for a decade. Please move to using WITH_HESIOD instead or your builds will silently lack HESIOD. 20140405: The uart(4) driver has been changed with respect to its handling of the low-level console. Previously the uart(4) driver prevented any process from changing the baudrate or the CLOCAL and HUPCL control flags. By removing the restrictions, operators can make changes to the serial console port without having to reboot. However, when getty(8) is started on the serial device that is associated with the low-level console, a misconfigured terminal line in /etc/ttys will now have a real impact. Before upgrading the kernel, make sure that /etc/ttys has the serial console device configured as 3wire without baudrate to preserve the previous behaviour. E.g: ttyu0 "/usr/libexec/getty 3wire" vt100 on secure 20140306: Support for libwrap (TCP wrappers) in rpcbind was disabled by default to improve performance. To re-enable it, if needed, run rpcbind with command line option -W. 20140226: Switched back to the GPL dtc compiler due to updates in the upstream dts files not being supported by the BSDL dtc compiler. You will need to rebuild your kernel toolchain to pick up the new compiler. Core dumps may result while building dtb files during a kernel build if you fail to do so. Set WITHOUT_GPL_DTC if you require the BSDL compiler. 20140216: Clang and llvm have been upgraded to 3.4 release. 20140216: The nve(4) driver has been removed. Please use the nfe(4) driver for NVIDIA nForce MCP Ethernet adapters instead. 20140212: An ABI incompatibility crept into the libc++ 3.4 import in r261283. This could cause certain C++ applications using shared libraries built against the previous version of libc++ to crash. The incompatibility has now been fixed, but any C++ applications or shared libraries built between r261283 and r261801 should be recompiled. 20140204: OpenSSH will now ignore errors caused by kernel lacking of Capsicum capability mode support. Please note that enabling the feature in kernel is still highly recommended. 20140131: OpenSSH is now built with sandbox support, and will use sandbox as the default privilege separation method. This requires Capsicum capability mode support in kernel. 20140128: The libelf and libdwarf libraries have been updated to newer versions from upstream. Shared library version numbers for these two libraries were bumped. Any ports or binaries requiring these two libraries should be recompiled. __FreeBSD_version is bumped to 1100006. 20140110: If a Makefile in a tests/ directory was auto-generating a Kyuafile instead of providing an explicit one, this would prevent such Makefile from providing its own Kyuafile in the future during NO_CLEAN builds. This has been fixed in the Makefiles but manual intervention is needed to clean an objdir if you use NO_CLEAN: # find /usr/obj -name Kyuafile | xargs rm -f 20131213: The behavior of gss_pseudo_random() for the krb5 mechanism has changed, for applications requesting a longer random string than produced by the underlying enctype's pseudo-random() function. In particular, the random string produced from a session key of enctype aes256-cts-hmac-sha1-96 or aes256-cts-hmac-sha1-96 will be different at the 17th octet and later, after this change. The counter used in the PRF+ construction is now encoded as a big-endian integer in accordance with RFC 4402. __FreeBSD_version is bumped to 1100004. 20131108: The WITHOUT_ATF build knob has been removed and its functionality has been subsumed into the more generic WITHOUT_TESTS. If you were using the former to disable the build of the ATF libraries, you should change your settings to use the latter. 20131025: The default version of mtree is nmtree which is obtained from NetBSD. The output is generally the same, but may vary slightly. If you found you need identical output adding "-F freebsd9" to the command line should do the trick. For the time being, the old mtree is available as fmtree. 20131014: libbsdyml has been renamed to libyaml and moved to /usr/lib/private. This will break ports-mgmt/pkg. Rebuild the port, or upgrade to pkg 1.1.4_8 and verify bsdyml not linked in, before running "make delete-old-libs": # make -C /usr/ports/ports-mgmt/pkg build deinstall install clean or # pkg install pkg; ldd /usr/local/sbin/pkg | grep bsdyml 20131010: The stable/10 branch has been created in subversion from head revision r256279. COMMON ITEMS: General Notes ------------- Sometimes, obscure build problems are the result of environment poisoning. This can happen because the make utility reads its environment when searching for values for global variables. To run your build attempts in an "environmental clean room", prefix all make commands with 'env -i '. See the env(1) manual page for more details. Occasionally a build failure will occur with "make -j" due to a race condition. If this happens try building again without -j, and please report a bug if it happens consistently. When upgrading from one major version to another it is generally best to upgrade to the latest code in the currently installed branch first, then do an upgrade to the new branch. This is the best-tested upgrade path, and has the highest probability of being successful. Please try this approach if you encounter problems with a major version upgrade. Since the stable 4.x branch point, one has generally been able to upgrade from anywhere in the most recent stable branch to head / current (or even the last couple of stable branches). See the top of this file when there's an exception. The update process will emit an error on an attempt to perform a build or install from a FreeBSD version below the earliest supported version. When updating from an older version the update should be performed one major release at a time, including running `make delete-old` at each step. When upgrading a live system, having a root shell around before installing anything can help undo problems. Not having a root shell around can lead to problems if pam has changed too much from your starting point to allow continued authentication after the upgrade. This file should be read as a log of events. When a later event changes information of a prior event, the prior event should not be deleted. Instead, a pointer to the entry with the new information should be placed in the old entry. Readers of this file should also sanity check older entries before relying on them blindly. Authors of new entries should write them with this in mind. ZFS notes --------- When upgrading the boot ZFS pool to a new version, always follow these two steps: 1.) recompile and reinstall the ZFS boot loader and boot block (this is part of "make buildworld" and "make installworld") 2.) update the ZFS boot block on your boot drive The following example updates the ZFS boot block on the first partition (freebsd-boot) of a GPT partitioned drive ada0: "gpart bootcode -p /boot/gptzfsboot -i 1 ada0" Non-boot pools do not need these updates. To build a kernel ----------------- If you are updating from a prior version of FreeBSD (even one just a few days old), you should follow this procedure. It is the most failsafe as it uses a /usr/obj tree with a fresh mini-buildworld, make kernel-toolchain make -DALWAYS_CHECK_MAKE buildkernel KERNCONF=YOUR_KERNEL_HERE make -DALWAYS_CHECK_MAKE installkernel KERNCONF=YOUR_KERNEL_HERE To test a kernel once --------------------- If you just want to boot a kernel once (because you are not sure if it works, or if you want to boot a known bad kernel to provide debugging information) run make installkernel KERNCONF=YOUR_KERNEL_HERE KODIR=/boot/testkernel nextboot -k testkernel To rebuild everything and install it on the current system. ----------------------------------------------------------- # Note: sometimes if you are running current you gotta do more than # is listed here if you are upgrading from a really old current. make buildworld make buildkernel KERNCONF=YOUR_KERNEL_HERE make installkernel KERNCONF=YOUR_KERNEL_HERE [1] [3] mergemaster -Fp [5] make installworld mergemaster -Fi [4] make delete-old [6] To cross-install current onto a separate partition -------------------------------------------------- # In this approach we use a separate partition to hold # current's root, 'usr', and 'var' directories. A partition # holding "/", "/usr" and "/var" should be about 2GB in # size. make buildworld make buildkernel KERNCONF=YOUR_KERNEL_HERE make installworld DESTDIR=${CURRENT_ROOT} -DDB_FROM_SRC make distribution DESTDIR=${CURRENT_ROOT} # if newfs'd make installkernel KERNCONF=YOUR_KERNEL_HERE DESTDIR=${CURRENT_ROOT} cp /etc/fstab ${CURRENT_ROOT}/etc/fstab # if newfs'd To upgrade in-place from stable to current ---------------------------------------------- make buildworld [9] make buildkernel KERNCONF=YOUR_KERNEL_HERE [8] make installkernel KERNCONF=YOUR_KERNEL_HERE [1] [3] mergemaster -Fp [5] make installworld mergemaster -Fi [4] make delete-old [6] Make sure that you've read the UPDATING file to understand the tweaks to various things you need. At this point in the life cycle of current, things change often and you are on your own to cope. The defaults can also change, so please read ALL of the UPDATING entries. Also, if you are tracking -current, you must be subscribed to freebsd-current@freebsd.org. Make sure that before you update your sources that you have read and understood all the recent messages there. If in doubt, please track -stable which has much fewer pitfalls. [1] If you have third party modules, such as vmware, you should disable them at this point so they don't crash your system on reboot. [3] From the bootblocks, boot -s, and then do fsck -p mount -u / mount -a sh /etc/rc.d/zfs start # mount zfs filesystem, if needed cd src # full path to source adjkerntz -i # if CMOS is wall time Also, when doing a major release upgrade, it is required that you boot into single user mode to do the installworld. [4] Note: This step is non-optional. Failure to do this step can result in a significant reduction in the functionality of the system. Attempting to do it by hand is not recommended and those that pursue this avenue should read this file carefully, as well as the archives of freebsd-current and freebsd-hackers mailing lists for potential gotchas. The -U option is also useful to consider. See mergemaster(8) for more information. [5] Usually this step is a no-op. However, from time to time you may need to do this if you get unknown user in the following step. It never hurts to do it all the time. You may need to install a new mergemaster (cd src/usr.sbin/mergemaster && make install) after the buildworld before this step if you last updated from current before 20130425 or from -stable before 20130430. [6] This only deletes old files and directories. Old libraries can be deleted by "make delete-old-libs", but you have to make sure that no program is using those libraries anymore. [8] The new kernel must be able to run existing binaries used by an installworld. When upgrading across major versions, the new kernel's configuration must include the correct COMPAT_FREEBSD option for existing binaries (e.g. COMPAT_FREEBSD11 to run 11.x binaries). Failure to do so may leave you with a system that is hard to boot to recover. A GENERIC kernel will include suitable compatibility options to run binaries from older branches. Note that the ability to run binaries from unsupported branches is not guaranteed. Make sure that you merge any new devices from GENERIC since the last time you updated your kernel config file. Options also change over time, so you may need to adjust your custom kernels for these as well. [9] If CPUTYPE is defined in your /etc/make.conf, make sure to use the "?=" instead of the "=" assignment operator, so that buildworld can override the CPUTYPE if it needs to. MAKEOBJDIRPREFIX must be defined in an environment variable, and not on the command line, or in /etc/make.conf. buildworld will warn if it is improperly defined. FORMAT: This file contains a list, in reverse chronological order, of major breakages in tracking -current. It is not guaranteed to be a complete list of such breakages, and only contains entries since September 23, 2011. If you need to see UPDATING entries from before that date, you will need to fetch an UPDATING file from an older FreeBSD release. Copyright information: Copyright 1998-2009 M. Warner Losh Redistribution, publication, translation and use, with or without modification, in full or in part, in any form or format of this document are permitted without further permission from the author. THIS DOCUMENT IS PROVIDED BY WARNER LOSH ``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 WARNER LOSH 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. Contact Warner Losh if you have any questions about your use of this document. $FreeBSD$ Index: projects/clang1100-import/cddl/contrib/opensolaris/lib/libctf/common/ctf_lib.c =================================================================== --- projects/clang1100-import/cddl/contrib/opensolaris/lib/libctf/common/ctf_lib.c (revision 364034) +++ projects/clang1100-import/cddl/contrib/opensolaris/lib/libctf/common/ctf_lib.c (revision 364035) @@ -1,528 +1,529 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License, Version 1.0 only * (the "License"). You may not use this file except in compliance * with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2003 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #pragma ident "%Z%%M% %I% %E% SMI" #include +#include #include #include #include #include #include #include #include #ifdef illumos #include #else #include #endif #include #ifdef illumos #ifdef _LP64 static const char *_libctf_zlib = "/usr/lib/64/libz.so"; #else static const char *_libctf_zlib = "/usr/lib/libz.so"; #endif #endif static struct { int (*z_uncompress)(uchar_t *, ulong_t *, const uchar_t *, ulong_t); const char *(*z_error)(int); void *z_dlp; } zlib; static size_t _PAGESIZE; static size_t _PAGEMASK; #ifdef illumos #pragma init(_libctf_init) #else void _libctf_init(void) __attribute__ ((constructor)); #endif void _libctf_init(void) { #ifdef illumos const char *p = getenv("LIBCTF_DECOMPRESSOR"); if (p != NULL) _libctf_zlib = p; /* use alternate decompression library */ #endif _libctf_debug = getenv("LIBCTF_DEBUG") != NULL; _PAGESIZE = getpagesize(); _PAGEMASK = ~(_PAGESIZE - 1); } /* * Attempt to dlopen the decompression library and locate the symbols of * interest that we will need to call. This information in cached so * that multiple calls to ctf_bufopen() do not need to reopen the library. */ void * ctf_zopen(int *errp) { #ifdef illumos ctf_dprintf("decompressing CTF data using %s\n", _libctf_zlib); if (zlib.z_dlp != NULL) return (zlib.z_dlp); /* library is already loaded */ if (access(_libctf_zlib, R_OK) == -1) return (ctf_set_open_errno(errp, ECTF_ZMISSING)); if ((zlib.z_dlp = dlopen(_libctf_zlib, RTLD_LAZY | RTLD_LOCAL)) == NULL) return (ctf_set_open_errno(errp, ECTF_ZINIT)); zlib.z_uncompress = (int (*)(uchar_t *, ulong_t *, const uchar_t *, ulong_t)) dlsym(zlib.z_dlp, "uncompress"); zlib.z_error = (const char *(*)(int)) dlsym(zlib.z_dlp, "zError"); if (zlib.z_uncompress == NULL || zlib.z_error == NULL) { (void) dlclose(zlib.z_dlp); bzero(&zlib, sizeof (zlib)); return (ctf_set_open_errno(errp, ECTF_ZINIT)); } #else zlib.z_uncompress = uncompress; zlib.z_error = zError; /* Dummy return variable as 'no error' */ zlib.z_dlp = (void *) (uintptr_t) 1; #endif return (zlib.z_dlp); } /* * The ctf_bufopen() routine calls these subroutines, defined by , * which we then patch through to the functions in the decompression library. */ int z_uncompress(void *dst, size_t *dstlen, const void *src, size_t srclen) { return (zlib.z_uncompress(dst, (ulong_t *)dstlen, src, srclen)); } const char * z_strerror(int err) { return (zlib.z_error(err)); } /* * Convert a 32-bit ELF file header into GElf. */ static void ehdr_to_gelf(const Elf32_Ehdr *src, GElf_Ehdr *dst) { bcopy(src->e_ident, dst->e_ident, EI_NIDENT); dst->e_type = src->e_type; dst->e_machine = src->e_machine; dst->e_version = src->e_version; dst->e_entry = (Elf64_Addr)src->e_entry; dst->e_phoff = (Elf64_Off)src->e_phoff; dst->e_shoff = (Elf64_Off)src->e_shoff; dst->e_flags = src->e_flags; dst->e_ehsize = src->e_ehsize; dst->e_phentsize = src->e_phentsize; dst->e_phnum = src->e_phnum; dst->e_shentsize = src->e_shentsize; dst->e_shnum = src->e_shnum; dst->e_shstrndx = src->e_shstrndx; } /* * Convert a 32-bit ELF section header into GElf. */ static void shdr_to_gelf(const Elf32_Shdr *src, GElf_Shdr *dst) { dst->sh_name = src->sh_name; dst->sh_type = src->sh_type; dst->sh_flags = src->sh_flags; dst->sh_addr = src->sh_addr; dst->sh_offset = src->sh_offset; dst->sh_size = src->sh_size; dst->sh_link = src->sh_link; dst->sh_info = src->sh_info; dst->sh_addralign = src->sh_addralign; dst->sh_entsize = src->sh_entsize; } /* * In order to mmap a section from the ELF file, we must round down sh_offset * to the previous page boundary, and mmap the surrounding page. We store * the pointer to the start of the actual section data back into sp->cts_data. */ const void * ctf_sect_mmap(ctf_sect_t *sp, int fd) { size_t pageoff = sp->cts_offset & ~_PAGEMASK; caddr_t base = mmap64(NULL, sp->cts_size + pageoff, PROT_READ, MAP_PRIVATE, fd, sp->cts_offset & _PAGEMASK); if (base != MAP_FAILED) sp->cts_data = base + pageoff; return (base); } /* * Since sp->cts_data has the adjusted offset, we have to again round down * to get the actual mmap address and round up to get the size. */ void ctf_sect_munmap(const ctf_sect_t *sp) { uintptr_t addr = (uintptr_t)sp->cts_data; uintptr_t pageoff = addr & ~_PAGEMASK; (void) munmap((void *)(addr - pageoff), sp->cts_size + pageoff); } /* * Open the specified file descriptor and return a pointer to a CTF container. * The file can be either an ELF file or raw CTF file. The caller is * responsible for closing the file descriptor when it is no longer needed. */ ctf_file_t * ctf_fdopen(int fd, int *errp) { ctf_sect_t ctfsect, symsect, strsect; ctf_file_t *fp = NULL; size_t shstrndx, shnum; struct stat64 st; ssize_t nbytes; union { ctf_preamble_t ctf; Elf32_Ehdr e32; GElf_Ehdr e64; } hdr; bzero(&ctfsect, sizeof (ctf_sect_t)); bzero(&symsect, sizeof (ctf_sect_t)); bzero(&strsect, sizeof (ctf_sect_t)); bzero(&hdr, sizeof (hdr)); if (fstat64(fd, &st) == -1) return (ctf_set_open_errno(errp, errno)); if ((nbytes = pread64(fd, &hdr, sizeof (hdr), 0)) <= 0) return (ctf_set_open_errno(errp, nbytes < 0? errno : ECTF_FMT)); /* * If we have read enough bytes to form a CTF header and the magic * string matches, attempt to interpret the file as raw CTF. */ if (nbytes >= (ssize_t) sizeof (ctf_preamble_t) && hdr.ctf.ctp_magic == CTF_MAGIC) { if (hdr.ctf.ctp_version > CTF_VERSION) return (ctf_set_open_errno(errp, ECTF_CTFVERS)); ctfsect.cts_data = mmap64(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0); if (ctfsect.cts_data == MAP_FAILED) return (ctf_set_open_errno(errp, errno)); ctfsect.cts_name = _CTF_SECTION; ctfsect.cts_type = SHT_PROGBITS; ctfsect.cts_flags = SHF_ALLOC; ctfsect.cts_size = (size_t)st.st_size; ctfsect.cts_entsize = 1; ctfsect.cts_offset = 0; if ((fp = ctf_bufopen(&ctfsect, NULL, NULL, errp)) == NULL) ctf_sect_munmap(&ctfsect); return (fp); } /* * If we have read enough bytes to form an ELF header and the magic * string matches, attempt to interpret the file as an ELF file. We * do our own largefile ELF processing, and convert everything to * GElf structures so that clients can operate on any data model. */ if (nbytes >= (ssize_t) sizeof (Elf32_Ehdr) && bcmp(&hdr.e32.e_ident[EI_MAG0], ELFMAG, SELFMAG) == 0) { #if BYTE_ORDER == _BIG_ENDIAN uchar_t order = ELFDATA2MSB; #else uchar_t order = ELFDATA2LSB; #endif GElf_Shdr *sp; void *strs_map; size_t strs_mapsz, i; char *strs; if (hdr.e32.e_ident[EI_DATA] != order) return (ctf_set_open_errno(errp, ECTF_ENDIAN)); if (hdr.e32.e_version != EV_CURRENT) return (ctf_set_open_errno(errp, ECTF_ELFVERS)); if (hdr.e32.e_ident[EI_CLASS] == ELFCLASS64) { if (nbytes < (ssize_t) sizeof (GElf_Ehdr)) return (ctf_set_open_errno(errp, ECTF_FMT)); } else { Elf32_Ehdr e32 = hdr.e32; ehdr_to_gelf(&e32, &hdr.e64); } shnum = hdr.e64.e_shnum; shstrndx = hdr.e64.e_shstrndx; /* Extended ELF sections */ if ((shstrndx == SHN_XINDEX) || (shnum == 0)) { if (hdr.e32.e_ident[EI_CLASS] == ELFCLASS32) { Elf32_Shdr x32; if (pread64(fd, &x32, sizeof (x32), hdr.e64.e_shoff) != sizeof (x32)) return (ctf_set_open_errno(errp, errno)); shnum = x32.sh_size; shstrndx = x32.sh_link; } else { Elf64_Shdr x64; if (pread64(fd, &x64, sizeof (x64), hdr.e64.e_shoff) != sizeof (x64)) return (ctf_set_open_errno(errp, errno)); shnum = x64.sh_size; shstrndx = x64.sh_link; } } if (shstrndx >= shnum) return (ctf_set_open_errno(errp, ECTF_CORRUPT)); nbytes = sizeof (GElf_Shdr) * shnum; if ((sp = malloc(nbytes)) == NULL) return (ctf_set_open_errno(errp, errno)); /* * Read in and convert to GElf the array of Shdr structures * from e_shoff so we can locate sections of interest. */ if (hdr.e32.e_ident[EI_CLASS] == ELFCLASS32) { Elf32_Shdr *sp32; nbytes = sizeof (Elf32_Shdr) * shnum; if ((sp32 = malloc(nbytes)) == NULL || pread64(fd, sp32, nbytes, hdr.e64.e_shoff) != nbytes) { free(sp); free(sp32); return (ctf_set_open_errno(errp, errno)); } for (i = 0; i < shnum; i++) shdr_to_gelf(&sp32[i], &sp[i]); free(sp32); } else if (pread64(fd, sp, nbytes, hdr.e64.e_shoff) != nbytes) { free(sp); return (ctf_set_open_errno(errp, errno)); } /* * Now mmap the section header strings section so that we can * perform string comparison on the section names. */ strs_mapsz = sp[shstrndx].sh_size + (sp[shstrndx].sh_offset & ~_PAGEMASK); strs_map = mmap64(NULL, strs_mapsz, PROT_READ, MAP_PRIVATE, fd, sp[shstrndx].sh_offset & _PAGEMASK); strs = (char *)strs_map + (sp[shstrndx].sh_offset & ~_PAGEMASK); if (strs_map == MAP_FAILED) { free(sp); return (ctf_set_open_errno(errp, ECTF_MMAP)); } /* * Iterate over the section header array looking for the CTF * section and symbol table. The strtab is linked to symtab. */ for (i = 0; i < shnum; i++) { const GElf_Shdr *shp = &sp[i]; const GElf_Shdr *lhp = &sp[shp->sh_link]; if (shp->sh_link >= shnum) continue; /* corrupt sh_link field */ if (shp->sh_name >= sp[shstrndx].sh_size || lhp->sh_name >= sp[shstrndx].sh_size) continue; /* corrupt sh_name field */ if (shp->sh_type == SHT_PROGBITS && strcmp(strs + shp->sh_name, _CTF_SECTION) == 0) { ctfsect.cts_name = strs + shp->sh_name; ctfsect.cts_type = shp->sh_type; ctfsect.cts_flags = shp->sh_flags; ctfsect.cts_size = shp->sh_size; ctfsect.cts_entsize = shp->sh_entsize; ctfsect.cts_offset = (off64_t)shp->sh_offset; } else if (shp->sh_type == SHT_SYMTAB) { symsect.cts_name = strs + shp->sh_name; symsect.cts_type = shp->sh_type; symsect.cts_flags = shp->sh_flags; symsect.cts_size = shp->sh_size; symsect.cts_entsize = shp->sh_entsize; symsect.cts_offset = (off64_t)shp->sh_offset; strsect.cts_name = strs + lhp->sh_name; strsect.cts_type = lhp->sh_type; strsect.cts_flags = lhp->sh_flags; strsect.cts_size = lhp->sh_size; strsect.cts_entsize = lhp->sh_entsize; strsect.cts_offset = (off64_t)lhp->sh_offset; } } free(sp); /* free section header array */ if (ctfsect.cts_type == SHT_NULL) { (void) munmap(strs_map, strs_mapsz); return (ctf_set_open_errno(errp, ECTF_NOCTFDATA)); } /* * Now mmap the CTF data, symtab, and strtab sections and * call ctf_bufopen() to do the rest of the work. */ if (ctf_sect_mmap(&ctfsect, fd) == MAP_FAILED) { (void) munmap(strs_map, strs_mapsz); return (ctf_set_open_errno(errp, ECTF_MMAP)); } if (symsect.cts_type != SHT_NULL && strsect.cts_type != SHT_NULL) { if (ctf_sect_mmap(&symsect, fd) == MAP_FAILED || ctf_sect_mmap(&strsect, fd) == MAP_FAILED) { (void) ctf_set_open_errno(errp, ECTF_MMAP); goto bad; /* unmap all and abort */ } fp = ctf_bufopen(&ctfsect, &symsect, &strsect, errp); } else fp = ctf_bufopen(&ctfsect, NULL, NULL, errp); bad: if (fp == NULL) { ctf_sect_munmap(&ctfsect); ctf_sect_munmap(&symsect); ctf_sect_munmap(&strsect); } else fp->ctf_flags |= LCTF_MMAP; (void) munmap(strs_map, strs_mapsz); return (fp); } return (ctf_set_open_errno(errp, ECTF_FMT)); } /* * Open the specified file and return a pointer to a CTF container. The file * can be either an ELF file or raw CTF file. This is just a convenient * wrapper around ctf_fdopen() for callers. */ ctf_file_t * ctf_open(const char *filename, int *errp) { ctf_file_t *fp; int fd; if ((fd = open64(filename, O_RDONLY)) == -1) { if (errp != NULL) *errp = errno; return (NULL); } fp = ctf_fdopen(fd, errp); (void) close(fd); return (fp); } /* * Write the uncompressed CTF data stream to the specified file descriptor. * This is useful for saving the results of dynamic CTF containers. */ int ctf_write(ctf_file_t *fp, int fd) { const uchar_t *buf = fp->ctf_base; ssize_t resid = fp->ctf_size; ssize_t len; while (resid != 0) { if ((len = write(fd, buf, resid)) <= 0) return (ctf_set_errno(fp, errno)); resid -= len; buf += len; } return (0); } /* * Set the CTF library client version to the specified version. If version is * zero, we just return the default library version number. */ int ctf_version(int version) { if (version < 0) { errno = EINVAL; return (-1); } if (version > 0) { if (version > CTF_VERSION) { errno = ENOTSUP; return (-1); } ctf_dprintf("ctf_version: client using version %d\n", version); _libctf_version = version; } return (_libctf_version); } Index: projects/clang1100-import/cddl/contrib/opensolaris/tools/ctf/cvt/ctfmerge.c =================================================================== --- projects/clang1100-import/cddl/contrib/opensolaris/tools/ctf/cvt/ctfmerge.c (revision 364034) +++ projects/clang1100-import/cddl/contrib/opensolaris/tools/ctf/cvt/ctfmerge.c (revision 364035) @@ -1,1024 +1,1025 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2008 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #pragma ident "%Z%%M% %I% %E% SMI" /* * Given several files containing CTF data, merge and uniquify that data into * a single CTF section in an output file. * * Merges can proceed independently. As such, we perform the merges in parallel * using a worker thread model. A given glob of CTF data (either all of the CTF * data from a single input file, or the result of one or more merges) can only * be involved in a single merge at any given time, so the process decreases in * parallelism, especially towards the end, as more and more files are * consolidated, finally resulting in a single merge of two large CTF graphs. * Unfortunately, the last merge is also the slowest, as the two graphs being * merged are each the product of merges of half of the input files. * * The algorithm consists of two phases, described in detail below. The first * phase entails the merging of CTF data in groups of eight. The second phase * takes the results of Phase I, and merges them two at a time. This disparity * is due to an observation that the merge time increases at least quadratically * with the size of the CTF data being merged. As such, merges of CTF graphs * newly read from input files are much faster than merges of CTF graphs that * are themselves the results of prior merges. * * A further complication is the need to ensure the repeatability of CTF merges. * That is, a merge should produce the same output every time, given the same * input. In both phases, this consistency requirement is met by imposing an * ordering on the merge process, thus ensuring that a given set of input files * are merged in the same order every time. * * Phase I * * The main thread reads the input files one by one, transforming the CTF * data they contain into tdata structures. When a given file has been read * and parsed, it is placed on the work queue for retrieval by worker threads. * * Central to Phase I is the Work In Progress (wip) array, which is used to * merge batches of files in a predictable order. Files are read by the main * thread, and are merged into wip array elements in round-robin order. When * the number of files merged into a given array slot equals the batch size, * the merged CTF graph in that array is added to the done slot in order by * array slot. * * For example, consider a case where we have five input files, a batch size * of two, a wip array size of two, and two worker threads (T1 and T2). * * 1. The wip array elements are assigned initial batch numbers 0 and 1. * 2. T1 reads an input file from the input queue (wq_queue). This is the * first input file, so it is placed into wip[0]. The second file is * similarly read and placed into wip[1]. The wip array slots now contain * one file each (wip_nmerged == 1). * 3. T1 reads the third input file, which it merges into wip[0]. The * number of files in wip[0] is equal to the batch size. * 4. T2 reads the fourth input file, which it merges into wip[1]. wip[1] * is now full too. * 5. T2 attempts to place the contents of wip[1] on the done queue * (wq_done_queue), but it can't, since the batch ID for wip[1] is 1. * Batch 0 needs to be on the done queue before batch 1 can be added, so * T2 blocks on wip[1]'s cv. * 6. T1 attempts to place the contents of wip[0] on the done queue, and * succeeds, updating wq_lastdonebatch to 0. It clears wip[0], and sets * its batch ID to 2. T1 then signals wip[1]'s cv to awaken T2. * 7. T2 wakes up, notices that wq_lastdonebatch is 0, which means that * batch 1 can now be added. It adds wip[1] to the done queue, clears * wip[1], and sets its batch ID to 3. It signals wip[0]'s cv, and * restarts. * * The above process continues until all input files have been consumed. At * this point, a pair of barriers are used to allow a single thread to move * any partial batches from the wip array to the done array in batch ID order. * When this is complete, wq_done_queue is moved to wq_queue, and Phase II * begins. * * Locking Semantics (Phase I) * * The input queue (wq_queue) and the done queue (wq_done_queue) are * protected by separate mutexes - wq_queue_lock and wq_done_queue. wip * array slots are protected by their own mutexes, which must be grabbed * before releasing the input queue lock. The wip array lock is dropped * when the thread restarts the loop. If the array slot was full, the * array lock will be held while the slot contents are added to the done * queue. The done queue lock is used to protect the wip slot cv's. * * The pow number is protected by the queue lock. The master batch ID * and last completed batch (wq_lastdonebatch) counters are protected *in * Phase I* by the done queue lock. * * Phase II * * When Phase II begins, the queue consists of the merged batches from the * first phase. Assume we have five batches: * * Q: a b c d e * * Using the same batch ID mechanism we used in Phase I, but without the wip * array, worker threads remove two entries at a time from the beginning of * the queue. These two entries are merged, and are added back to the tail * of the queue, as follows: * * Q: a b c d e # start * Q: c d e ab # a, b removed, merged, added to end * Q: e ab cd # c, d removed, merged, added to end * Q: cd eab # e, ab removed, merged, added to end * Q: cdeab # cd, eab removed, merged, added to end * * When one entry remains on the queue, with no merges outstanding, Phase II * finishes. We pre-determine the stopping point by pre-calculating the * number of nodes that will appear on the list. In the example above, the * number (wq_ninqueue) is 9. When ninqueue is 1, we conclude Phase II by * signaling the main thread via wq_done_cv. * * Locking Semantics (Phase II) * * The queue (wq_queue), ninqueue, and the master batch ID and last * completed batch counters are protected by wq_queue_lock. The done * queue and corresponding lock are unused in Phase II as is the wip array. * * Uniquification * * We want the CTF data that goes into a given module to be as small as * possible. For example, we don't want it to contain any type data that may * be present in another common module. As such, after creating the master * tdata_t for a given module, we can, if requested by the user, uniquify it * against the tdata_t from another module (genunix in the case of the SunOS * kernel). We perform a merge between the tdata_t for this module and the * tdata_t from genunix. Nodes found in this module that are not present in * genunix are added to a third tdata_t - the uniquified tdata_t. * * Additive Merges * * In some cases, for example if we are issuing a new version of a common * module in a patch, we need to make sure that the CTF data already present * in that module does not change. Changes to this data would void the CTF * data in any module that uniquified against the common module. To preserve * the existing data, we can perform what is known as an additive merge. In * this case, a final uniquification is performed against the CTF data in the * previous version of the module. The result will be the placement of new * and changed data after the existing data, thus preserving the existing type * ID space. * * Saving the result * * When the merges are complete, the resulting tdata_t is placed into the * output file, replacing the .SUNW_ctf section (if any) already in that file. * * The person who changes the merging thread code in this file without updating * this comment will not live to see the stock hit five. */ #include #include #include #include #include #ifdef illumos #include #endif #include #include #include #include #ifdef illumos #include #endif #include #include #include #ifdef illumos #include #endif #include "ctf_headers.h" #include "ctftools.h" #include "ctfmerge.h" #include "traverse.h" #include "memory.h" #include "fifo.h" #include "barrier.h" #pragma init(bigheap) #define MERGE_PHASE1_BATCH_SIZE 8 #define MERGE_PHASE1_MAX_SLOTS 5 #define MERGE_INPUT_THROTTLE_LEN 10 const char *progname; static char *outfile = NULL; static char *tmpname = NULL; static int dynsym; int debug_level = DEBUG_LEVEL; static size_t maxpgsize = 0x400000; void usage(void) { (void) fprintf(stderr, "Usage: %s [-fgstv] -l label | -L labelenv -o outfile file ...\n" " %s [-fgstv] -l label | -L labelenv -o outfile -d uniqfile\n" " %*s [-g] [-D uniqlabel] file ...\n" " %s [-fgstv] -l label | -L labelenv -o outfile -w withfile " "file ...\n" " %s [-g] -c srcfile destfile\n" "\n" " Note: if -L labelenv is specified and labelenv is not set in\n" " the environment, a default value is used.\n", progname, progname, (int)strlen(progname), " ", progname, progname); } #ifdef illumos static void bigheap(void) { size_t big, *size; int sizes; struct memcntl_mha mha; /* * First, get the available pagesizes. */ if ((sizes = getpagesizes(NULL, 0)) == -1) return; if (sizes == 1 || (size = alloca(sizeof (size_t) * sizes)) == NULL) return; if (getpagesizes(size, sizes) == -1) return; while (size[sizes - 1] > maxpgsize) sizes--; /* set big to the largest allowed page size */ big = size[sizes - 1]; if (big & (big - 1)) { /* * The largest page size is not a power of two for some * inexplicable reason; return. */ return; } /* * Now, align our break to the largest page size. */ if (brk((void *)((((uintptr_t)sbrk(0) - 1) & ~(big - 1)) + big)) != 0) return; /* * set the preferred page size for the heap */ mha.mha_cmd = MHA_MAPSIZE_BSSBRK; mha.mha_flags = 0; mha.mha_pagesize = big; (void) memcntl(NULL, 0, MC_HAT_ADVISE, (caddr_t)&mha, 0, 0); } #endif /* illumos */ static void finalize_phase_one(workqueue_t *wq) { int startslot, i; /* * wip slots are cleared out only when maxbatchsz td's have been merged * into them. We're not guaranteed that the number of files we're * merging is a multiple of maxbatchsz, so there will be some partial * groups in the wip array. Move them to the done queue in batch ID * order, starting with the slot containing the next batch that would * have been placed on the done queue, followed by the others. * One thread will be doing this while the others wait at the barrier * back in worker_thread(), so we don't need to worry about pesky things * like locks. */ for (startslot = -1, i = 0; i < wq->wq_nwipslots; i++) { if (wq->wq_wip[i].wip_batchid == wq->wq_lastdonebatch + 1) { startslot = i; break; } } assert(startslot != -1); for (i = startslot; i < startslot + wq->wq_nwipslots; i++) { int slotnum = i % wq->wq_nwipslots; wip_t *wipslot = &wq->wq_wip[slotnum]; if (wipslot->wip_td != NULL) { debug(2, "clearing slot %d (%d) (saving %d)\n", slotnum, i, wipslot->wip_nmerged); } else debug(2, "clearing slot %d (%d)\n", slotnum, i); if (wipslot->wip_td != NULL) { fifo_add(wq->wq_donequeue, wipslot->wip_td); wq->wq_wip[slotnum].wip_td = NULL; } } wq->wq_lastdonebatch = wq->wq_next_batchid++; debug(2, "phase one done: donequeue has %d items\n", fifo_len(wq->wq_donequeue)); } static void init_phase_two(workqueue_t *wq) { int num; /* * We're going to continually merge the first two entries on the queue, * placing the result on the end, until there's nothing left to merge. * At that point, everything will have been merged into one. The * initial value of ninqueue needs to be equal to the total number of * entries that will show up on the queue, both at the start of the * phase and as generated by merges during the phase. */ wq->wq_ninqueue = num = fifo_len(wq->wq_donequeue); while (num != 1) { wq->wq_ninqueue += num / 2; num = num / 2 + num % 2; } /* * Move the done queue to the work queue. We won't be using the done * queue in phase 2. */ assert(fifo_len(wq->wq_queue) == 0); fifo_free(wq->wq_queue, NULL); wq->wq_queue = wq->wq_donequeue; } static void wip_save_work(workqueue_t *wq, wip_t *slot, int slotnum) { pthread_mutex_lock(&wq->wq_donequeue_lock); while (wq->wq_lastdonebatch + 1 < slot->wip_batchid) pthread_cond_wait(&slot->wip_cv, &wq->wq_donequeue_lock); assert(wq->wq_lastdonebatch + 1 == slot->wip_batchid); fifo_add(wq->wq_donequeue, slot->wip_td); wq->wq_lastdonebatch++; pthread_cond_signal(&wq->wq_wip[(slotnum + 1) % wq->wq_nwipslots].wip_cv); /* reset the slot for next use */ slot->wip_td = NULL; slot->wip_batchid = wq->wq_next_batchid++; pthread_mutex_unlock(&wq->wq_donequeue_lock); } static void wip_add_work(wip_t *slot, tdata_t *pow) { if (slot->wip_td == NULL) { slot->wip_td = pow; slot->wip_nmerged = 1; } else { debug(2, "%d: merging %p into %p\n", pthread_self(), (void *)pow, (void *)slot->wip_td); merge_into_master(pow, slot->wip_td, NULL, 0); tdata_free(pow); slot->wip_nmerged++; } } static void worker_runphase1(workqueue_t *wq) { wip_t *wipslot; tdata_t *pow; int wipslotnum, pownum; for (;;) { pthread_mutex_lock(&wq->wq_queue_lock); while (fifo_empty(wq->wq_queue)) { if (wq->wq_nomorefiles == 1) { pthread_cond_broadcast(&wq->wq_work_avail); pthread_mutex_unlock(&wq->wq_queue_lock); /* on to phase 2 ... */ return; } pthread_cond_wait(&wq->wq_work_avail, &wq->wq_queue_lock); } /* there's work to be done! */ pow = fifo_remove(wq->wq_queue); pownum = wq->wq_nextpownum++; pthread_cond_broadcast(&wq->wq_work_removed); assert(pow != NULL); /* merge it into the right slot */ wipslotnum = pownum % wq->wq_nwipslots; wipslot = &wq->wq_wip[wipslotnum]; pthread_mutex_lock(&wipslot->wip_lock); pthread_mutex_unlock(&wq->wq_queue_lock); wip_add_work(wipslot, pow); if (wipslot->wip_nmerged == wq->wq_maxbatchsz) wip_save_work(wq, wipslot, wipslotnum); pthread_mutex_unlock(&wipslot->wip_lock); } } static void worker_runphase2(workqueue_t *wq) { tdata_t *pow1, *pow2; int batchid; for (;;) { pthread_mutex_lock(&wq->wq_queue_lock); if (wq->wq_ninqueue == 1) { pthread_cond_broadcast(&wq->wq_work_avail); pthread_mutex_unlock(&wq->wq_queue_lock); debug(2, "%d: entering p2 completion barrier\n", pthread_self()); if (barrier_wait(&wq->wq_bar1)) { pthread_mutex_lock(&wq->wq_queue_lock); wq->wq_alldone = 1; pthread_cond_signal(&wq->wq_alldone_cv); pthread_mutex_unlock(&wq->wq_queue_lock); } return; } if (fifo_len(wq->wq_queue) < 2) { pthread_cond_wait(&wq->wq_work_avail, &wq->wq_queue_lock); pthread_mutex_unlock(&wq->wq_queue_lock); continue; } /* there's work to be done! */ pow1 = fifo_remove(wq->wq_queue); pow2 = fifo_remove(wq->wq_queue); wq->wq_ninqueue -= 2; batchid = wq->wq_next_batchid++; pthread_mutex_unlock(&wq->wq_queue_lock); debug(2, "%d: merging %p into %p\n", pthread_self(), (void *)pow1, (void *)pow2); merge_into_master(pow1, pow2, NULL, 0); tdata_free(pow1); /* * merging is complete. place at the tail of the queue in * proper order. */ pthread_mutex_lock(&wq->wq_queue_lock); while (wq->wq_lastdonebatch + 1 != batchid) { pthread_cond_wait(&wq->wq_done_cv, &wq->wq_queue_lock); } wq->wq_lastdonebatch = batchid; fifo_add(wq->wq_queue, pow2); debug(2, "%d: added %p to queue, len now %d, ninqueue %d\n", pthread_self(), (void *)pow2, fifo_len(wq->wq_queue), wq->wq_ninqueue); pthread_cond_broadcast(&wq->wq_done_cv); pthread_cond_signal(&wq->wq_work_avail); pthread_mutex_unlock(&wq->wq_queue_lock); } } /* * Main loop for worker threads. */ static void worker_thread(workqueue_t *wq) { worker_runphase1(wq); debug(2, "%d: entering first barrier\n", pthread_self()); if (barrier_wait(&wq->wq_bar1)) { debug(2, "%d: doing work in first barrier\n", pthread_self()); finalize_phase_one(wq); init_phase_two(wq); debug(2, "%d: ninqueue is %d, %d on queue\n", pthread_self(), wq->wq_ninqueue, fifo_len(wq->wq_queue)); } debug(2, "%d: entering second barrier\n", pthread_self()); (void) barrier_wait(&wq->wq_bar2); debug(2, "%d: phase 1 complete\n", pthread_self()); worker_runphase2(wq); } /* * Pass a tdata_t tree, built from an input file, off to the work queue for * consumption by worker threads. */ static int merge_ctf_cb(tdata_t *td, char *name, void *arg) { workqueue_t *wq = arg; debug(3, "Adding tdata %p for processing\n", (void *)td); pthread_mutex_lock(&wq->wq_queue_lock); while (fifo_len(wq->wq_queue) > wq->wq_ithrottle) { debug(2, "Throttling input (len = %d, throttle = %d)\n", fifo_len(wq->wq_queue), wq->wq_ithrottle); pthread_cond_wait(&wq->wq_work_removed, &wq->wq_queue_lock); } fifo_add(wq->wq_queue, td); debug(1, "Thread %d announcing %s\n", pthread_self(), name); pthread_cond_broadcast(&wq->wq_work_avail); pthread_mutex_unlock(&wq->wq_queue_lock); return (1); } /* * This program is intended to be invoked from a Makefile, as part of the build. * As such, in the event of a failure or user-initiated interrupt (^C), we need * to ensure that a subsequent re-make will cause ctfmerge to be executed again. * Unfortunately, ctfmerge will usually be invoked directly after (and as part * of the same Makefile rule as) a link, and will operate on the linked file * in place. If we merely exit upon receipt of a SIGINT, a subsequent make * will notice that the *linked* file is newer than the object files, and thus * will not reinvoke ctfmerge. The only way to ensure that a subsequent make * reinvokes ctfmerge, is to remove the file to which we are adding CTF * data (confusingly named the output file). This means that the link will need * to happen again, but links are generally fast, and we can't allow the merge * to be skipped. * * Another possibility would be to block SIGINT entirely - to always run to * completion. The run time of ctfmerge can, however, be measured in minutes * in some cases, so this is not a valid option. */ static void handle_sig(int sig) { terminate("Caught signal %d - exiting\n", sig); } static void terminate_cleanup(void) { int dounlink = getenv("CTFMERGE_TERMINATE_NO_UNLINK") ? 0 : 1; if (tmpname != NULL && dounlink) unlink(tmpname); if (outfile == NULL) return; #if !defined(__FreeBSD__) if (dounlink) { fprintf(stderr, "Removing %s\n", outfile); unlink(outfile); } #endif } static void copy_ctf_data(char *srcfile, char *destfile, int keep_stabs) { tdata_t *srctd; if (read_ctf(&srcfile, 1, NULL, read_ctf_save_cb, &srctd, 1) == 0) terminate("No CTF data found in source file %s\n", srcfile); tmpname = mktmpname(destfile, ".ctf"); write_ctf(srctd, destfile, tmpname, CTF_COMPRESS | CTF_SWAP_BYTES | keep_stabs); if (rename(tmpname, destfile) != 0) { terminate("Couldn't rename temp file %s to %s", tmpname, destfile); } free(tmpname); tdata_free(srctd); } static void wq_init(workqueue_t *wq, int nfiles) { int throttle, nslots, i; if (getenv("CTFMERGE_MAX_SLOTS")) nslots = atoi(getenv("CTFMERGE_MAX_SLOTS")); else nslots = MERGE_PHASE1_MAX_SLOTS; if (getenv("CTFMERGE_PHASE1_BATCH_SIZE")) wq->wq_maxbatchsz = atoi(getenv("CTFMERGE_PHASE1_BATCH_SIZE")); else wq->wq_maxbatchsz = MERGE_PHASE1_BATCH_SIZE; nslots = MIN(nslots, (nfiles + wq->wq_maxbatchsz - 1) / wq->wq_maxbatchsz); wq->wq_wip = xcalloc(sizeof (wip_t) * nslots); wq->wq_nwipslots = nslots; wq->wq_nthreads = MIN(sysconf(_SC_NPROCESSORS_ONLN) * 3 / 2, nslots); wq->wq_thread = xmalloc(sizeof (pthread_t) * wq->wq_nthreads); if (getenv("CTFMERGE_INPUT_THROTTLE")) throttle = atoi(getenv("CTFMERGE_INPUT_THROTTLE")); else throttle = MERGE_INPUT_THROTTLE_LEN; wq->wq_ithrottle = throttle * wq->wq_nthreads; debug(1, "Using %d slots, %d threads\n", wq->wq_nwipslots, wq->wq_nthreads); wq->wq_next_batchid = 0; for (i = 0; i < nslots; i++) { pthread_mutex_init(&wq->wq_wip[i].wip_lock, NULL); + pthread_cond_init(&wq->wq_wip[i].wip_cv, NULL); wq->wq_wip[i].wip_batchid = wq->wq_next_batchid++; } pthread_mutex_init(&wq->wq_queue_lock, NULL); wq->wq_queue = fifo_new(); pthread_cond_init(&wq->wq_work_avail, NULL); pthread_cond_init(&wq->wq_work_removed, NULL); wq->wq_ninqueue = nfiles; wq->wq_nextpownum = 0; pthread_mutex_init(&wq->wq_donequeue_lock, NULL); wq->wq_donequeue = fifo_new(); wq->wq_lastdonebatch = -1; pthread_cond_init(&wq->wq_done_cv, NULL); pthread_cond_init(&wq->wq_alldone_cv, NULL); wq->wq_alldone = 0; barrier_init(&wq->wq_bar1, wq->wq_nthreads); barrier_init(&wq->wq_bar2, wq->wq_nthreads); wq->wq_nomorefiles = 0; } static void start_threads(workqueue_t *wq) { sigset_t sets; int i; sigemptyset(&sets); sigaddset(&sets, SIGINT); sigaddset(&sets, SIGQUIT); sigaddset(&sets, SIGTERM); pthread_sigmask(SIG_BLOCK, &sets, NULL); for (i = 0; i < wq->wq_nthreads; i++) { pthread_create(&wq->wq_thread[i], NULL, (void *(*)(void *))worker_thread, wq); } #ifdef illumos sigset(SIGINT, handle_sig); sigset(SIGQUIT, handle_sig); sigset(SIGTERM, handle_sig); #else signal(SIGINT, handle_sig); signal(SIGQUIT, handle_sig); signal(SIGTERM, handle_sig); #endif pthread_sigmask(SIG_UNBLOCK, &sets, NULL); } static void join_threads(workqueue_t *wq) { int i; for (i = 0; i < wq->wq_nthreads; i++) { pthread_join(wq->wq_thread[i], NULL); } } static int strcompare(const void *p1, const void *p2) { char *s1 = *((char **)p1); char *s2 = *((char **)p2); return (strcmp(s1, s2)); } /* * Core work queue structure; passed to worker threads on thread creation * as the main point of coordination. Allocate as a static structure; we * could have put this into a local variable in main, but passing a pointer * into your stack to another thread is fragile at best and leads to some * hard-to-debug failure modes. */ static workqueue_t wq; int main(int argc, char **argv) { tdata_t *mstrtd, *savetd; char *uniqfile = NULL, *uniqlabel = NULL; char *withfile = NULL; char *label = NULL; char **ifiles, **tifiles; int verbose = 0, docopy = 0; int write_fuzzy_match = 0; int keep_stabs = 0; int require_ctf = 0; int nifiles, nielems; int c, i, idx, tidx, err; progname = basename(argv[0]); if (getenv("CTFMERGE_DEBUG_LEVEL")) debug_level = atoi(getenv("CTFMERGE_DEBUG_LEVEL")); err = 0; while ((c = getopt(argc, argv, ":cd:D:fgl:L:o:tvw:s")) != EOF) { switch (c) { case 'c': docopy = 1; break; case 'd': /* Uniquify against `uniqfile' */ uniqfile = optarg; break; case 'D': /* Uniquify against label `uniqlabel' in `uniqfile' */ uniqlabel = optarg; break; case 'f': write_fuzzy_match = CTF_FUZZY_MATCH; break; case 'g': keep_stabs = CTF_KEEP_STABS; break; case 'l': /* Label merged types with `label' */ label = optarg; break; case 'L': /* Label merged types with getenv(`label`) */ if ((label = getenv(optarg)) == NULL) label = CTF_DEFAULT_LABEL; break; case 'o': /* Place merged types in CTF section in `outfile' */ outfile = optarg; break; case 't': /* Insist *all* object files built from C have CTF */ require_ctf = 1; break; case 'v': /* More debugging information */ verbose = 1; break; case 'w': /* Additive merge with data from `withfile' */ withfile = optarg; break; case 's': /* use the dynsym rather than the symtab */ dynsym = CTF_USE_DYNSYM; break; default: usage(); exit(2); } } /* Validate arguments */ if (docopy) { if (uniqfile != NULL || uniqlabel != NULL || label != NULL || outfile != NULL || withfile != NULL || dynsym != 0) err++; if (argc - optind != 2) err++; } else { if (uniqfile != NULL && withfile != NULL) err++; if (uniqlabel != NULL && uniqfile == NULL) err++; if (outfile == NULL || label == NULL) err++; if (argc - optind == 0) err++; } if (err) { usage(); exit(2); } if (getenv("STRIPSTABS_KEEP_STABS") != NULL) keep_stabs = CTF_KEEP_STABS; if (uniqfile && access(uniqfile, R_OK) != 0) { warning("Uniquification file %s couldn't be opened and " "will be ignored.\n", uniqfile); uniqfile = NULL; } if (withfile && access(withfile, R_OK) != 0) { warning("With file %s couldn't be opened and will be " "ignored.\n", withfile); withfile = NULL; } if (outfile && access(outfile, R_OK|W_OK) != 0) terminate("Cannot open output file %s for r/w", outfile); /* * This is ugly, but we don't want to have to have a separate tool * (yet) just for copying an ELF section with our specific requirements, * so we shoe-horn a copier into ctfmerge. */ if (docopy) { copy_ctf_data(argv[optind], argv[optind + 1], keep_stabs); exit(0); } set_terminate_cleanup(terminate_cleanup); /* Sort the input files and strip out duplicates */ nifiles = argc - optind; ifiles = xmalloc(sizeof (char *) * nifiles); tifiles = xmalloc(sizeof (char *) * nifiles); for (i = 0; i < nifiles; i++) tifiles[i] = argv[optind + i]; qsort(tifiles, nifiles, sizeof (char *), (int (*)())strcompare); ifiles[0] = tifiles[0]; for (idx = 0, tidx = 1; tidx < nifiles; tidx++) { if (strcmp(ifiles[idx], tifiles[tidx]) != 0) ifiles[++idx] = tifiles[tidx]; } nifiles = idx + 1; /* Make sure they all exist */ if ((nielems = count_files(ifiles, nifiles)) < 0) terminate("Some input files were inaccessible\n"); /* Prepare for the merge */ wq_init(&wq, nielems); start_threads(&wq); /* * Start the merge * * We're reading everything from each of the object files, so we * don't need to specify labels. */ if (read_ctf(ifiles, nifiles, NULL, merge_ctf_cb, &wq, require_ctf) == 0) { /* * If we're verifying that C files have CTF, it's safe to * assume that in this case, we're building only from assembly * inputs. */ if (require_ctf) exit(0); terminate("No ctf sections found to merge\n"); } pthread_mutex_lock(&wq.wq_queue_lock); wq.wq_nomorefiles = 1; pthread_cond_broadcast(&wq.wq_work_avail); pthread_mutex_unlock(&wq.wq_queue_lock); pthread_mutex_lock(&wq.wq_queue_lock); while (wq.wq_alldone == 0) pthread_cond_wait(&wq.wq_alldone_cv, &wq.wq_queue_lock); pthread_mutex_unlock(&wq.wq_queue_lock); join_threads(&wq); /* * All requested files have been merged, with the resulting tree in * mstrtd. savetd is the tree that will be placed into the output file. * * Regardless of whether we're doing a normal uniquification or an * additive merge, we need a type tree that has been uniquified * against uniqfile or withfile, as appropriate. * * If we're doing a uniquification, we stuff the resulting tree into * outfile. Otherwise, we add the tree to the tree already in withfile. */ assert(fifo_len(wq.wq_queue) == 1); mstrtd = fifo_remove(wq.wq_queue); if (verbose || debug_level) { debug(2, "Statistics for td %p\n", (void *)mstrtd); iidesc_stats(mstrtd->td_iihash); } if (uniqfile != NULL || withfile != NULL) { char *reffile, *reflabel = NULL; tdata_t *reftd; if (uniqfile != NULL) { reffile = uniqfile; reflabel = uniqlabel; } else reffile = withfile; if (read_ctf(&reffile, 1, reflabel, read_ctf_save_cb, &reftd, require_ctf) == 0) { terminate("No CTF data found in reference file %s\n", reffile); } savetd = tdata_new(); if (CTF_TYPE_ISCHILD(reftd->td_nextid)) terminate("No room for additional types in master\n"); savetd->td_nextid = withfile ? reftd->td_nextid : CTF_INDEX_TO_TYPE(1, TRUE); merge_into_master(mstrtd, reftd, savetd, 0); tdata_label_add(savetd, label, CTF_LABEL_LASTIDX); if (withfile) { /* * savetd holds the new data to be added to the withfile */ tdata_t *withtd = reftd; tdata_merge(withtd, savetd); savetd = withtd; } else { char uniqname[MAXPATHLEN]; labelent_t *parle; parle = tdata_label_top(reftd); savetd->td_parlabel = xstrdup(parle->le_name); strncpy(uniqname, reffile, sizeof (uniqname)); uniqname[MAXPATHLEN - 1] = '\0'; savetd->td_parname = xstrdup(basename(uniqname)); } } else { /* * No post processing. Write the merged tree as-is into the * output file. */ tdata_label_free(mstrtd); tdata_label_add(mstrtd, label, CTF_LABEL_LASTIDX); savetd = mstrtd; } tmpname = mktmpname(outfile, ".ctf"); write_ctf(savetd, outfile, tmpname, CTF_COMPRESS | CTF_SWAP_BYTES | write_fuzzy_match | dynsym | keep_stabs); if (rename(tmpname, outfile) != 0) terminate("Couldn't rename output temp file %s", tmpname); free(tmpname); return (0); } Index: projects/clang1100-import/cddl/contrib/opensolaris/tools/ctf/cvt/ctftools.h =================================================================== --- projects/clang1100-import/cddl/contrib/opensolaris/tools/ctf/cvt/ctftools.h (revision 364034) +++ projects/clang1100-import/cddl/contrib/opensolaris/tools/ctf/cvt/ctftools.h (revision 364035) @@ -1,454 +1,464 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2006 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #ifndef _CTFTOOLS_H #define _CTFTOOLS_H /* * Functions and data structures used in the manipulation of stabs and CTF data */ #include #include #include #include #include #include #include +#include #ifdef __cplusplus extern "C" { #endif #include "list.h" #include "hash.h" #ifndef DEBUG_LEVEL #define DEBUG_LEVEL 0 #endif #ifndef DEBUG_PARSE #define DEBUG_PARSE 0 #endif #ifndef DEBUG_STREAM #define DEBUG_STREAM stderr #endif #ifndef MAX #define MAX(a, b) ((a) < (b) ? (b) : (a)) #endif #ifndef MIN #define MIN(a, b) ((a) > (b) ? (b) : (a)) +#endif + +/* Sanity check for cross-build bootstrap tools */ +#if !defined(BYTE_ORDER) +#error "Missing BYTE_ORDER defines" +#elif !defined(_LITTLE_ENDIAN) +#error "Missing _LITTLE_ENDIAN defines" +#elif !defined(_BIG_ENDIAN) +#error "Missing _BIG_ENDIAN defines" #endif #define TRUE 1 #define FALSE 0 #define CTF_ELF_SCN_NAME ".SUNW_ctf" #define CTF_LABEL_LASTIDX -1 #define CTF_DEFAULT_LABEL "*** No Label Provided ***" /* * Default hash sizes */ #define TDATA_LAYOUT_HASH_SIZE 8191 /* A tdesc hash based on layout */ #define TDATA_ID_HASH_SIZE 997 /* A tdesc hash based on type id */ #define IIDESC_HASH_SIZE 8191 /* Hash of iidesc's */ /* * The default function argument array size. We'll realloc the array larger * if we need to, but we want a default value that will allow us to avoid * reallocation in the common case. */ #define FUNCARG_DEF 5 extern const char *progname; extern int debug_level; extern int debug_parse; extern char *curhdr; /* * This is a partial copy of the stab.h that DevPro includes with their * compiler. */ typedef struct stab { uint32_t n_strx; uint8_t n_type; int8_t n_other; int16_t n_desc; uint32_t n_value; } stab_t; #define N_GSYM 0x20 /* global symbol: name,,0,type,0 */ #define N_FUN 0x24 /* procedure: name,,0,linenumber,0 */ #define N_STSYM 0x26 /* static symbol: name,,0,type,0 or section relative */ #define N_LCSYM 0x28 /* .lcomm symbol: name,,0,type,0 or section relative */ #define N_ROSYM 0x2c /* ro_data: name,,0,type,0 or section relative */ #define N_OPT 0x3c /* compiler options */ #define N_RSYM 0x40 /* register sym: name,,0,type,register */ #define N_SO 0x64 /* source file name: name,,0,0,0 */ #define N_LSYM 0x80 /* local sym: name,,0,type,offset */ #define N_SOL 0x84 /* #included file name: name,,0,0,0 */ #define N_PSYM 0xa0 /* parameter: name,,0,type,offset */ #define N_LBRAC 0xc0 /* left bracket: 0,,0,nesting level,function relative */ #define N_RBRAC 0xe0 /* right bracket: 0,,0,nesting level,func relative */ #define N_BINCL 0x82 /* header file: name,,0,0,0 */ #define N_EINCL 0xa2 /* end of include file */ /* * Nodes in the type tree * * Each node consists of a single tdesc_t, with one of several auxiliary * structures linked in via the `data' union. */ /* The type of tdesc_t node */ typedef enum stabtype { STABTYPE_FIRST, /* do not use */ INTRINSIC, POINTER, ARRAY, FUNCTION, STRUCT, UNION, ENUM, FORWARD, TYPEDEF, TYPEDEF_UNRES, VOLATILE, CONST, RESTRICT, STABTYPE_LAST /* do not use */ } stabtype_t; typedef struct tdesc tdesc_t; /* Auxiliary structure for array tdesc_t */ typedef struct ardef { tdesc_t *ad_contents; tdesc_t *ad_idxtype; uint_t ad_nelems; } ardef_t; /* Auxiliary structure for structure/union tdesc_t */ typedef struct mlist { int ml_offset; /* Offset from start of structure (in bits) */ int ml_size; /* Member size (in bits) */ char *ml_name; /* Member name */ struct tdesc *ml_type; /* Member type */ struct mlist *ml_next; /* Next member */ } mlist_t; /* Auxiliary structure for enum tdesc_t */ typedef struct elist { char *el_name; int el_number; struct elist *el_next; } elist_t; /* Auxiliary structure for intrinsics (integers and reals) */ typedef enum { INTR_INT, INTR_REAL } intrtype_t; typedef struct intr { intrtype_t intr_type; int intr_signed; union { char _iformat; int _fformat; } _u; int intr_offset; int intr_nbits; } intr_t; #define intr_iformat _u._iformat #define intr_fformat _u._fformat typedef struct fnarg { char *fna_name; struct tdesc *fna_type; } fnarg_t; #define FN_F_GLOBAL 0x1 #define FN_F_VARARGS 0x2 typedef struct fndef { struct tdesc *fn_ret; uint_t fn_nargs; tdesc_t **fn_args; uint_t fn_vargs; } fndef_t; typedef int32_t tid_t; /* * The tdesc_t (Type DESCription) is the basic node type used in the stabs data * structure. Each data node gets a tdesc structure. Each node is linked into * a directed graph (think of it as a tree with multiple roots and multiple * leaves), with the root nodes at the top, and intrinsics at the bottom. The * root nodes, which are pointed to by iidesc nodes, correspond to the types, * globals, and statics defined by the stabs. */ struct tdesc { char *t_name; tdesc_t *t_next; /* Name hash next pointer */ tid_t t_id; tdesc_t *t_hash; /* ID hash next pointer */ stabtype_t t_type; int t_size; /* Size in bytes of object represented by this node */ union { intr_t *intr; /* int, real */ tdesc_t *tdesc; /* ptr, typedef, vol, const, restr */ ardef_t *ardef; /* array */ mlist_t *members; /* struct, union */ elist_t *emem; /* enum */ fndef_t *fndef; /* function - first is return type */ } t_data; int t_flags; int t_vgen; /* Visitation generation (see traverse.c) */ int t_emark; /* Equality mark (see equiv_cb() in merge.c) */ }; #define t_intr t_data.intr #define t_tdesc t_data.tdesc #define t_ardef t_data.ardef #define t_members t_data.members #define t_emem t_data.emem #define t_fndef t_data.fndef #define TDESC_F_ISROOT 0x1 /* Has an iidesc_t (see below) */ #define TDESC_F_GLOBAL 0x2 #define TDESC_F_RESOLVED 0x4 /* * iidesc_t (Interesting Item DESCription) nodes point to tdesc_t nodes that * correspond to "interesting" stabs. A stab is interesting if it defines a * global or static variable, a global or static function, or a data type. */ typedef enum iitype { II_NOT = 0, II_GFUN, /* Global function */ II_SFUN, /* Static function */ II_GVAR, /* Global variable */ II_SVAR, /* Static variable */ II_PSYM, /* Function argument */ II_SOU, /* Struct or union */ II_TYPE /* Type (typedef) */ } iitype_t; typedef struct iidesc { iitype_t ii_type; char *ii_name; tdesc_t *ii_dtype; char *ii_owner; /* File that defined this node */ int ii_flags; /* Function arguments (if any) */ int ii_nargs; tdesc_t **ii_args; int ii_vargs; /* Function uses varargs */ } iidesc_t; #define IIDESC_F_USED 0x1 /* Write this iidesc out */ /* * labelent_t nodes identify labels and corresponding type ranges associated * with them. The label in a given labelent_t is associated with types with * ids <= le_idx. */ typedef struct labelent { char *le_name; int le_idx; } labelent_t; /* * The tdata_t (Type DATA) structure contains or references all type data for * a given file or, during merging, several files. */ typedef struct tdata { int td_curemark; /* Equality mark (see merge.c) */ int td_curvgen; /* Visitation generation (see traverse.c) */ int td_nextid; /* The ID for the next tdesc_t created */ hash_t *td_iihash; /* The iidesc_t nodes for this file */ hash_t *td_layouthash; /* The tdesc nodes, hashed by structure */ hash_t *td_idhash; /* The tdesc nodes, hashed by type id */ list_t *td_fwdlist; /* All forward declaration tdesc nodes */ char *td_parlabel; /* Top label uniq'd against in parent */ char *td_parname; /* Basename of parent */ list_t *td_labels; /* Labels and their type ranges */ pthread_mutex_t td_mergelock; int td_ref; } tdata_t; /* * By design, the iidesc hash is heterogeneous. The CTF emitter, on the * other hand, needs to be able to access the elements of the list by type, * and in a specific sorted order. An iiburst holds these elements in that * order. (A burster is a machine that separates carbon-copy forms) */ typedef struct iiburst { int iib_nfuncs; int iib_curfunc; iidesc_t **iib_funcs; int iib_nobjts; int iib_curobjt; iidesc_t **iib_objts; list_t *iib_types; int iib_maxtypeid; tdata_t *iib_td; struct tdtrav_data *iib_tdtd; /* tdtrav_data_t */ } iiburst_t; typedef struct ctf_buf ctf_buf_t; typedef struct symit_data symit_data_t; /* fixup_tdescs.c */ void cvt_fixstabs(tdata_t *); void cvt_fixups(tdata_t *, size_t); /* ctf.c */ caddr_t ctf_gen(iiburst_t *, size_t *, int); tdata_t *ctf_load(char *, caddr_t, size_t, symit_data_t *, char *); /* iidesc.c */ iidesc_t *iidesc_new(char *); int iidesc_hash(int, void *); void iter_iidescs_by_name(tdata_t *, const char *, int (*)(void *, void *), void *); iidesc_t *iidesc_dup(iidesc_t *); iidesc_t *iidesc_dup_rename(iidesc_t *, char const *, char const *); void iidesc_add(hash_t *, iidesc_t *); void iidesc_free(void *, void *); int iidesc_count_type(void *, void *); void iidesc_stats(hash_t *); int iidesc_dump(iidesc_t *); /* input.c */ typedef enum source_types { SOURCE_NONE = 0, SOURCE_UNKNOWN = 1, SOURCE_C = 2, SOURCE_S = 4 } source_types_t; source_types_t built_source_types(Elf *, const char *); int count_files(char **, int); int read_ctf(char **, int, char *, int (*)(tdata_t *, char *, void *), void *, int); int read_ctf_save_cb(tdata_t *, char *, void *); symit_data_t *symit_new(Elf *, const char *); void symit_reset(symit_data_t *); char *symit_curfile(symit_data_t *); GElf_Sym *symit_next(symit_data_t *, int); char *symit_name(symit_data_t *); void symit_free(symit_data_t *); /* merge.c */ void merge_into_master(tdata_t *, tdata_t *, tdata_t *, int); /* output.c */ #define CTF_FUZZY_MATCH 0x1 /* match local symbols to global CTF */ #define CTF_USE_DYNSYM 0x2 /* use .dynsym not .symtab */ #define CTF_COMPRESS 0x4 /* compress CTF output */ #define CTF_KEEP_STABS 0x8 /* keep .stabs sections */ #define CTF_SWAP_BYTES 0x10 /* target byte order is different from host */ void write_ctf(tdata_t *, const char *, const char *, int); /* parse.c */ void parse_init(tdata_t *); void parse_finish(tdata_t *); int parse_stab(stab_t *, char *, iidesc_t **); tdesc_t *lookup(int); tdesc_t *lookupname(const char *); void check_hash(void); void resolve_typed_bitfields(void); /* stabs.c */ int stabs_read(tdata_t *, Elf *, char *); /* dwarf.c */ int dw_read(tdata_t *, Elf *, char *); const char *dw_tag2str(uint_t); /* tdata.c */ tdata_t *tdata_new(void); void tdata_free(tdata_t *); void tdata_build_hashes(tdata_t *td); const char *tdesc_name(tdesc_t *); int tdesc_idhash(int, void *); int tdesc_idcmp(void *, void *); int tdesc_namehash(int, void *); int tdesc_namecmp(void *, void *); int tdesc_layouthash(int, void *); int tdesc_layoutcmp(void *, void *); void tdesc_free(tdesc_t *); void tdata_label_add(tdata_t *, const char *, int); labelent_t *tdata_label_top(tdata_t *); int tdata_label_find(tdata_t *, char *); void tdata_label_free(tdata_t *); void tdata_merge(tdata_t *, tdata_t *); void tdata_label_newmax(tdata_t *, int); /* util.c */ int streq(const char *, const char *); int findelfsecidx(Elf *, const char *, const char *); size_t elf_ptrsz(Elf *); char *mktmpname(const char *, const char *); void terminate(const char *, ...) __NORETURN; void aborterr(const char *, ...) __NORETURN; void set_terminate_cleanup(void (*)(void)); void elfterminate(const char *, const char *, ...); void warning(const char *, ...); void vadebug(int, const char *, va_list); void debug(int, const char *, ...); void watch_dump(int); void watch_set(void *, int); #ifdef __cplusplus } #endif #endif /* _CTFTOOLS_H */ Index: projects/clang1100-import/cddl/contrib/opensolaris =================================================================== --- projects/clang1100-import/cddl/contrib/opensolaris (revision 364034) +++ projects/clang1100-import/cddl/contrib/opensolaris (revision 364035) Property changes on: projects/clang1100-import/cddl/contrib/opensolaris ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/cddl/contrib/opensolaris:r363987-364034 Index: projects/clang1100-import/cddl =================================================================== --- projects/clang1100-import/cddl (revision 364034) +++ projects/clang1100-import/cddl (revision 364035) Property changes on: projects/clang1100-import/cddl ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/cddl:r363987-364034 Index: projects/clang1100-import/contrib/telnet/telnet/telnet.1 =================================================================== --- projects/clang1100-import/contrib/telnet/telnet/telnet.1 (revision 364034) +++ projects/clang1100-import/contrib/telnet/telnet/telnet.1 (revision 364035) @@ -1,1472 +1,1480 @@ .\" Copyright (c) 1983, 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. .\" 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. .\" .\" @(#)telnet.1 8.6 (Berkeley) 6/1/94 .\" $FreeBSD$ .\" -.Dd September 18, 2006 +.Dd August 7, 2020 .Dt TELNET 1 .Os .Sh NAME .Nm telnet .Nd user interface to the .Tn TELNET protocol .Sh SYNOPSIS .Nm .Op Fl 468EFKLNacdfruxy .Op Fl B Ar baudrate .Op Fl S Ar tos .Op Fl X Ar authtype .Op Fl e Ar escapechar .Op Fl k Ar realm .Op Fl l Ar user .Op Fl n Ar tracefile .Op Fl s Ar src_addr +.Op Fl P Ar policy .Oo .Ar host .Op Ar port .Oc .Sh DESCRIPTION The .Nm command is used to communicate with another host using the .Tn TELNET protocol. If .Nm is invoked without the .Ar host argument, it enters command mode, indicated by its prompt .Pq Dq Li telnet\&> . In this mode, it accepts and executes the commands listed below. If it is invoked with arguments, it performs an .Ic open command with those arguments. .Pp Options: .Bl -tag -width indent .It Fl 4 Forces .Nm to use IPv4 addresses only. .It Fl 6 Forces .Nm to use IPv6 addresses only. .It Fl 8 Specifies an 8-bit data path. This causes an attempt to negotiate the .Dv TELNET BINARY option on both input and output. .It Fl B Ar baudrate Sets the baud rate to .Ar baudrate . .It Fl E Stops any character from being recognized as an escape character. .It Fl F If Kerberos V5 authentication is being used, the .Fl F option allows the local credentials to be forwarded to the remote system, including any credentials that have already been forwarded into the local environment. .It Fl K Specifies no automatic login to the remote system. .It Fl L Specifies an 8-bit data path on output. This causes the .Dv BINARY option to be negotiated on output. .It Fl N Prevents IP address to name lookup when destination host is given as an IP address. .It Fl S Ar tos Sets the IP type-of-service (TOS) option for the telnet connection to the value .Ar tos , which can be a numeric TOS value or, on systems that support it, a symbolic TOS name found in the .Pa /etc/iptos file. .It Fl X Ar atype Disables the .Ar atype type of authentication. .It Fl a Attempt automatic login. This is now the default, so this option is ignored. Currently, this sends the user name via the .Ev USER variable of the .Ev ENVIRON option if supported by the remote system. The name used is that of the current user as returned by .Xr getlogin 2 if it agrees with the current user ID, otherwise it is the name associated with the user ID. .It Fl c Disables the reading of the user's .Pa \&.telnetrc file. (See the .Ic toggle skiprc command on this man page.) .It Fl d Sets the initial value of the .Ic debug toggle to .Dv TRUE . .It Fl e Ar escapechar Sets the initial .Nm escape character to .Ar escapechar . If .Ar escapechar is omitted, then there will be no escape character. .It Fl f If Kerberos V5 authentication is being used, the .Fl f option allows the local credentials to be forwarded to the remote system. .It Fl k Ar realm If Kerberos authentication is being used, the .Fl k option requests that .Nm obtain tickets for the remote host in realm .Ar realm instead of the remote host's realm, as determined by .Xr krb_realmofhost 3 . .It Fl l Ar user When connecting to the remote system, if the remote system understands the .Ev ENVIRON option, then .Ar user will be sent to the remote system as the value for the variable .Ev USER . This option implies the .Fl a option. This option may also be used with the .Ic open command. .It Fl n Ar tracefile Opens .Ar tracefile for recording trace information. See the .Ic set tracefile command below. +.It Fl P Ar policy +Use IPsec policy specification string +.Ar policy , +for the connections. +See +.Xr ipsec_set_policy 3 +for details. .It Fl r Specifies a user interface similar to .Xr rlogin 1 . In this mode, the escape character is set to the tilde (~) character, unless modified by the .Fl e option. .It Fl s Ar src_addr Set the source IP address for the .Nm connection to .Ar src_addr , which can be an IP address or a host name. .It Fl u Forces .Nm to use .Dv AF_UNIX addresses only (e.g., .Ux domain sockets, accessed with a file path). .It Fl x Turns on encryption of the data stream if possible. This is now the default, so this option is ignored. .It Fl y Suppresses encryption of the data stream. .It Ar host Indicates the official name, an alias, or the Internet address of a remote host. If .Ar host starts with a .Ql / , .Nm establishes a connection to the corresponding named socket. .It Ar port Indicates a port number (address of an application). If a number is not specified, the default .Nm port is used. .El .Pp When in rlogin mode, a line of the form ~.\& disconnects from the remote host; ~ is the .Nm escape character. Similarly, the line ~^Z suspends the .Nm session. The line ~^] escapes to the normal .Nm escape prompt. .Pp Once a connection has been opened, .Nm will attempt to enable the .Dv TELNET LINEMODE option. If this fails, then .Nm will revert to one of two input modes: either \*(Lqcharacter at a time\*(Rq or \*(Lqold line by line\*(Rq depending on what the remote system supports. .Pp When .Dv LINEMODE is enabled, character processing is done on the local system, under the control of the remote system. When input editing or character echoing is to be disabled, the remote system will relay that information. The remote system will also relay changes to any special characters that happen on the remote system, so that they can take effect on the local system. .Pp In \*(Lqcharacter at a time\*(Rq mode, most text typed is immediately sent to the remote host for processing. .Pp In \*(Lqold line by line\*(Rq mode, all text is echoed locally, and (normally) only completed lines are sent to the remote host. The \*(Lqlocal echo character\*(Rq (initially \*(Lq^E\*(Rq) may be used to turn off and on the local echo (this would mostly be used to enter passwords without the password being echoed). .Pp If the .Dv LINEMODE option is enabled, or if the .Ic localchars toggle is .Dv TRUE (the default for \*(Lqold line by line\*(Rq; see below), the user's .Ic quit , .Ic intr , and .Ic flush characters are trapped locally, and sent as .Tn TELNET protocol sequences to the remote side. If .Dv LINEMODE has ever been enabled, then the user's .Ic susp and .Ic eof are also sent as .Tn TELNET protocol sequences, and .Ic quit is sent as a .Dv TELNET ABORT instead of .Dv BREAK . There are options (see .Ic toggle .Ic autoflush and .Ic toggle .Ic autosynch below) which cause this action to flush subsequent output to the terminal (until the remote host acknowledges the .Tn TELNET sequence) and flush previous terminal input (in the case of .Ic quit and .Ic intr ) . .Pp While connected to a remote host, .Nm command mode may be entered by typing the .Nm \*(Lqescape character\*(Rq (initially \*(Lq^]\*(Rq). When in command mode, the normal terminal editing conventions are available. .Pp The following .Nm commands are available. Only enough of each command to uniquely identify it need be typed (this is also true for arguments to the .Ic mode , .Ic set , .Ic toggle , .Ic unset , .Ic slc , .Ic environ , and .Ic display commands). .Bl -tag -width "mode type" .It Ic auth Ar argument ... The auth command manipulates the information sent through the .Dv TELNET AUTHENTICATE option. Valid arguments for the .Ic auth command are: .Bl -tag -width "disable type" .It Ic disable Ar type Disables the specified type of authentication. To obtain a list of available types, use the .Ic auth disable ?\& command. .It Ic enable Ar type Enables the specified type of authentication. To obtain a list of available types, use the .Ic auth enable ?\& command. .It Ic status Lists the current status of the various types of authentication. .El .It Ic close Close a .Tn TELNET session and return to command mode. .It Ic display Ar argument ... Displays all, or some, of the .Ic set and .Ic toggle values (see below). .It Ic encrypt Ar argument ... The encrypt command manipulates the information sent through the .Dv TELNET ENCRYPT option. .Pp Valid arguments for the .Ic encrypt command are: .Bl -tag -width Ar .It Ic disable Ar type Xo .Op Cm input | output .Xc Disables the specified type of encryption. If you omit the input and output, both input and output are disabled. To obtain a list of available types, use the .Ic encrypt disable ?\& command. .It Ic enable Ar type Xo .Op Cm input | output .Xc Enables the specified type of encryption. If you omit input and output, both input and output are enabled. To obtain a list of available types, use the .Ic encrypt enable ?\& command. .It Ic input This is the same as the .Ic encrypt start input command. .It Ic -input This is the same as the .Ic encrypt stop input command. .It Ic output This is the same as the .Ic encrypt start output command. .It Ic -output This is the same as the .Ic encrypt stop output command. .It Ic start Op Cm input | output Attempts to start encryption. If you omit .Ic input and .Ic output , both input and output are enabled. To obtain a list of available types, use the .Ic encrypt enable ?\& command. .It Ic status Lists the current status of encryption. .It Ic stop Op Cm input | output Stops encryption. If you omit input and output, encryption is on both input and output. .It Ic type Ar type Sets the default type of encryption to be used with later .Ic encrypt start or .Ic encrypt stop commands. .El .It Ic environ Ar arguments ... The .Ic environ command is used to manipulate the variables that may be sent through the .Dv TELNET ENVIRON option. The initial set of variables is taken from the users environment, with only the .Ev DISPLAY and .Ev PRINTER variables being exported by default. The .Ev USER variable is also exported if the .Fl a or .Fl l options are used. .Pp Valid arguments for the .Ic environ command are: .Bl -tag -width Fl .It Ic define Ar variable value Define the variable .Ar variable to have a value of .Ar value . Any variables defined by this command are automatically exported. The .Ar value may be enclosed in single or double quotes so that tabs and spaces may be included. .It Ic undefine Ar variable Remove .Ar variable from the list of environment variables. .It Ic export Ar variable Mark the variable .Ar variable to be exported to the remote side. .It Ic unexport Ar variable Mark the variable .Ar variable to not be exported unless explicitly asked for by the remote side. .It Ic list List the current set of environment variables. Those marked with a .Cm * will be sent automatically, other variables will only be sent if explicitly requested. .It Ic ?\& Prints out help information for the .Ic environ command. .El .It Ic logout Sends the .Dv TELNET LOGOUT option to the remote side. This command is similar to a .Ic close command; however, if the remote side does not support the .Dv LOGOUT option, nothing happens. If, however, the remote side does support the .Dv LOGOUT option, this command should cause the remote side to close the .Tn TELNET connection. If the remote side also supports the concept of suspending a user's session for later reattachment, the logout argument indicates that you should terminate the session immediately. .It Ic mode Ar type .Ar Type is one of several options, depending on the state of the .Tn TELNET session. The remote host is asked for permission to go into the requested mode. If the remote host is capable of entering that mode, the requested mode will be entered. .Bl -tag -width Ar .It Ic character Disable the .Dv TELNET LINEMODE option, or, if the remote side does not understand the .Dv LINEMODE option, then enter \*(Lqcharacter at a time\*(Rq mode. .It Ic line Enable the .Dv TELNET LINEMODE option, or, if the remote side does not understand the .Dv LINEMODE option, then attempt to enter \*(Lqold-line-by-line\*(Rq mode. .It Ic isig Pq Ic \-isig Attempt to enable (disable) the .Dv TRAPSIG mode of the .Dv LINEMODE option. This requires that the .Dv LINEMODE option be enabled. .It Ic edit Pq Ic \-edit Attempt to enable (disable) the .Dv EDIT mode of the .Dv LINEMODE option. This requires that the .Dv LINEMODE option be enabled. .It Ic softtabs Pq Ic \-softtabs Attempt to enable (disable) the .Dv SOFT_TAB mode of the .Dv LINEMODE option. This requires that the .Dv LINEMODE option be enabled. .It Ic litecho Pq Ic \-litecho Attempt to enable (disable) the .Dv LIT_ECHO mode of the .Dv LINEMODE option. This requires that the .Dv LINEMODE option be enabled. .It Ic ?\& Prints out help information for the .Ic mode command. .El .It Xo .Ic open .Op Fl l Ar user .Op Ar host .Op Oo Fl /+ Oc Ns Ar port .Xc Open a connection to the named host. If no port number is specified, .Nm will attempt to contact a .Tn TELNET server at the default port. The host specification may be either a host name (see .Xr hosts 5 ) , an Internet address specified in the \*(Lqdot notation\*(Rq (see .Xr inet 3 ) , or IPv6 host name or IPv6 coloned-hexadecimal addreess. The .Fl l option may be used to specify the user name to be passed to the remote system via the .Ev ENVIRON option. When connecting to a non-standard port, .Nm omits any automatic initiation of .Tn TELNET options. When the port number is preceded by a minus sign, the initial option negotiation is done. When, however, the port number is preceded by a plus sign, any option negotiation and understanding is prohibited, making telnet dumb client for POP3/SMTP/NNTP/HTTP-like protocols with any data including .Tn TELNET IAC character (0xff). After establishing a connection, the file .Pa \&.telnetrc in the users home directory is opened. Lines beginning with a # are comment lines. Blank lines are ignored. Lines that begin without white space are the start of a machine entry. The first thing on the line is the name of the machine that is being connected to. It may be the hostname or numeric address specified as the argument .Ar host , the canonical name of that string as determined by .Xr getaddrinfo 3 , or the string .Dq Li DEFAULT indicating all hosts. The rest of the line, and successive lines that begin with white space are assumed to be .Nm commands and are processed as if they had been typed in manually to the .Nm command prompt. .It Ic quit Close any open .Tn TELNET session and exit .Nm . An end of file (in command mode) will also close a session and exit. .It Ic send Ar arguments Sends one or more special character sequences to the remote host. The following are the arguments which may be specified (more than one argument may be specified at a time): .Bl -tag -width escape .It Ic abort Sends the .Dv TELNET ABORT (Abort processes) sequence. .It Ic ao Sends the .Dv TELNET AO (Abort Output) sequence, which should cause the remote system to flush all output .Em from the remote system .Em to the user's terminal. .It Ic ayt Sends the .Dv TELNET AYT (Are You There) sequence, to which the remote system may or may not choose to respond. .It Ic brk Sends the .Dv TELNET BRK (Break) sequence, which may have significance to the remote system. .It Ic ec Sends the .Dv TELNET EC (Erase Character) sequence, which should cause the remote system to erase the last character entered. .It Ic el Sends the .Dv TELNET EL (Erase Line) sequence, which should cause the remote system to erase the line currently being entered. .It Ic eof Sends the .Dv TELNET EOF (End Of File) sequence. .It Ic eor Sends the .Dv TELNET EOR (End of Record) sequence. .It Ic escape Sends the current .Nm escape character (initially \*(Lq^\*(Rq). .It Ic ga Sends the .Dv TELNET GA (Go Ahead) sequence, which likely has no significance to the remote system. .It Ic getstatus If the remote side supports the .Dv TELNET STATUS command, .Ic getstatus will send the subnegotiation to request that the server send its current option status. .It Ic ip Sends the .Dv TELNET IP (Interrupt Process) sequence, which should cause the remote system to abort the currently running process. .It Ic nop Sends the .Dv TELNET NOP (No OPeration) sequence. .It Ic susp Sends the .Dv TELNET SUSP (SUSPend process) sequence. .It Ic synch Sends the .Dv TELNET SYNCH sequence. This sequence causes the remote system to discard all previously typed (but not yet read) input. This sequence is sent as .Tn TCP urgent data (and may not work if the remote system is a .Bx 4.2 system -- if it doesn't work, a lower case \*(Lqr\*(Rq may be echoed on the terminal). .It Ic do Ar cmd .It Ic dont Ar cmd .It Ic will Ar cmd .It Ic wont Ar cmd Sends the .Dv TELNET DO .Ar cmd sequence. .Ar Cmd can be either a decimal number between 0 and 255, or a symbolic name for a specific .Dv TELNET command. .Ar Cmd can also be either .Ic help or .Ic ?\& to print out help information, including a list of known symbolic names. .It Ic ?\& Prints out help information for the .Ic send command. .El .It Ic set Ar argument value .It Ic unset Ar argument value The .Ic set command will set any one of a number of .Nm variables to a specific value or to .Dv TRUE . The special value .Ic off turns off the function associated with the variable, this is equivalent to using the .Ic unset command. The .Ic unset command will disable or set to .Dv FALSE any of the specified functions. The values of variables may be interrogated with the .Ic display command. The variables which may be set or unset, but not toggled, are listed here. In addition, any of the variables for the .Ic toggle command may be explicitly set or unset using the .Ic set and .Ic unset commands. .Bl -tag -width escape .It Ic ayt If .Tn TELNET is in localchars mode, or .Dv LINEMODE is enabled, and the status character is typed, a .Dv TELNET AYT sequence (see .Ic send ayt preceding) is sent to the remote host. The initial value for the \*(LqAre You There\*(Rq character is the terminal's status character. .It Ic echo This is the value (initially \*(Lq^E\*(Rq) which, when in \*(Lqline by line\*(Rq mode, toggles between doing local echoing of entered characters (for normal processing), and suppressing echoing of entered characters (for entering, say, a password). .It Ic eof If .Nm is operating in .Dv LINEMODE or \*(Lqold line by line\*(Rq mode, entering this character as the first character on a line will cause this character to be sent to the remote system. The initial value of the eof character is taken to be the terminal's .Ic eof character. .It Ic erase If .Nm is in .Ic localchars mode (see .Ic toggle .Ic localchars below), .Sy and if .Nm is operating in \*(Lqcharacter at a time\*(Rq mode, then when this character is typed, a .Dv TELNET EC sequence (see .Ic send .Ic ec above) is sent to the remote system. The initial value for the erase character is taken to be the terminal's .Ic erase character. .It Ic escape This is the .Nm escape character (initially \*(Lq^[\*(Rq) which causes entry into .Nm command mode (when connected to a remote system). .It Ic flushoutput If .Nm is in .Ic localchars mode (see .Ic toggle .Ic localchars below) and the .Ic flushoutput character is typed, a .Dv TELNET AO sequence (see .Ic send .Ic ao above) is sent to the remote host. The initial value for the flush character is taken to be the terminal's .Ic flush character. .It Ic forw1 .It Ic forw2 If .Nm is operating in .Dv LINEMODE , these are the characters that, when typed, cause partial lines to be forwarded to the remote system. The initial value for the forwarding characters are taken from the terminal's eol and eol2 characters. .It Ic interrupt If .Nm is in .Ic localchars mode (see .Ic toggle .Ic localchars below) and the .Ic interrupt character is typed, a .Dv TELNET IP sequence (see .Ic send .Ic ip above) is sent to the remote host. The initial value for the interrupt character is taken to be the terminal's .Ic intr character. .It Ic kill If .Nm is in .Ic localchars mode (see .Ic toggle .Ic localchars below), .Ic and if .Nm is operating in \*(Lqcharacter at a time\*(Rq mode, then when this character is typed, a .Dv TELNET EL sequence (see .Ic send .Ic el above) is sent to the remote system. The initial value for the kill character is taken to be the terminal's .Ic kill character. .It Ic lnext If .Nm is operating in .Dv LINEMODE or \*(Lqold line by line\*(Rq mode, then this character is taken to be the terminal's .Ic lnext character. The initial value for the lnext character is taken to be the terminal's .Ic lnext character. .It Ic quit If .Nm is in .Ic localchars mode (see .Ic toggle .Ic localchars below) and the .Ic quit character is typed, a .Dv TELNET BRK sequence (see .Ic send .Ic brk above) is sent to the remote host. The initial value for the quit character is taken to be the terminal's .Ic quit character. .It Ic reprint If .Nm is operating in .Dv LINEMODE or \*(Lqold line by line\*(Rq mode, then this character is taken to be the terminal's .Ic reprint character. The initial value for the reprint character is taken to be the terminal's .Ic reprint character. .It Ic rlogin This is the rlogin escape character. If set, the normal .Nm escape character is ignored unless it is preceded by this character at the beginning of a line. This character, at the beginning of a line followed by a "." closes the connection; when followed by a ^Z it suspends the .Nm command. The initial state is to disable the .Nm rlogin escape character. .It Ic start If the .Dv TELNET TOGGLE-FLOW-CONTROL option has been enabled, then this character is taken to be the terminal's .Ic start character. The initial value for the start character is taken to be the terminal's .Ic start character. .It Ic stop If the .Dv TELNET TOGGLE-FLOW-CONTROL option has been enabled, then this character is taken to be the terminal's .Ic stop character. The initial value for the stop character is taken to be the terminal's .Ic stop character. .It Ic susp If .Nm is in .Ic localchars mode, or .Dv LINEMODE is enabled, and the .Ic suspend character is typed, a .Dv TELNET SUSP sequence (see .Ic send .Ic susp above) is sent to the remote host. The initial value for the suspend character is taken to be the terminal's .Ic suspend character. .It Ic tracefile This is the file to which the output, caused by .Ic netdata or .Ic option tracing being .Dv TRUE , will be written. If it is set to .Dq Fl , then tracing information will be written to standard output (the default). .It Ic worderase If .Nm is operating in .Dv LINEMODE or \*(Lqold line by line\*(Rq mode, then this character is taken to be the terminal's .Ic worderase character. The initial value for the worderase character is taken to be the terminal's .Ic worderase character. .It Ic ?\& Displays the legal .Ic set .Pq Ic unset commands. .El .It Ic opie Ar sequence challenge The .Ic opie command computes a response to the OPIE challenge. .It Ic slc Ar state The .Ic slc command (Set Local Characters) is used to set or change the state of the special characters when the .Dv TELNET LINEMODE option has been enabled. Special characters are characters that get mapped to .Tn TELNET commands sequences (like .Ic ip or .Ic quit ) or line editing characters (like .Ic erase and .Ic kill ) . By default, the local special characters are exported. .Bl -tag -width Fl .It Ic check Verify the current settings for the current special characters. The remote side is requested to send all the current special character settings, and if there are any discrepancies with the local side, the local side will switch to the remote value. .It Ic export Switch to the local defaults for the special characters. The local default characters are those of the local terminal at the time when .Nm was started. .It Ic import Switch to the remote defaults for the special characters. The remote default characters are those of the remote system at the time when the .Tn TELNET connection was established. .It Ic ?\& Prints out help information for the .Ic slc command. .El .It Ic status Show the current status of .Nm . This includes the peer one is connected to, as well as the current mode. .It Ic toggle Ar arguments ... Toggle (between .Dv TRUE and .Dv FALSE ) various flags that control how .Nm responds to events. These flags may be set explicitly to .Dv TRUE or .Dv FALSE using the .Ic set and .Ic unset commands listed above. More than one argument may be specified. The state of these flags may be interrogated with the .Ic display command. Valid arguments are: .Bl -tag -width Ar .It Ic authdebug Turns on debugging information for the authentication code. .It Ic autoflush If .Ic autoflush and .Ic localchars are both .Dv TRUE , then when the .Ic ao , or .Ic quit characters are recognized (and transformed into .Tn TELNET sequences; see .Ic set above for details), .Nm refuses to display any data on the user's terminal until the remote system acknowledges (via a .Dv TELNET TIMING MARK option) that it has processed those .Tn TELNET sequences. The initial value for this toggle is .Dv TRUE if the terminal user had not done an "stty noflsh", otherwise .Dv FALSE (see .Xr stty 1 ) . .It Ic autodecrypt When the .Dv TELNET ENCRYPT option is negotiated, by default the actual encryption (decryption) of the data stream does not start automatically. The autoencrypt (autodecrypt) command states that encryption of the output (input) stream should be enabled as soon as possible. .It Ic autologin If the remote side supports the .Dv TELNET AUTHENTICATION option .Nm attempts to use it to perform automatic authentication. If the .Dv AUTHENTICATION option is not supported, the user's login name are propagated through the .Dv TELNET ENVIRON option. This command is the same as specifying .Fl a option on the .Ic open command. .It Ic autosynch If .Ic autosynch and .Ic localchars are both .Dv TRUE , then when either the .Ic intr or .Ic quit characters is typed (see .Ic set above for descriptions of the .Ic intr and .Ic quit characters), the resulting .Tn TELNET sequence sent is followed by the .Dv TELNET SYNCH sequence. This procedure .Ic should cause the remote system to begin throwing away all previously typed input until both of the .Tn TELNET sequences have been read and acted upon. The initial value of this toggle is .Dv FALSE . .It Ic binary Enable or disable the .Dv TELNET BINARY option on both input and output. .It Ic inbinary Enable or disable the .Dv TELNET BINARY option on input. .It Ic outbinary Enable or disable the .Dv TELNET BINARY option on output. .It Ic crlf If this is .Dv TRUE , then carriage returns will be sent as .Li . If this is .Dv FALSE , then carriage returns will be send as .Li . The initial value for this toggle is .Dv FALSE . .It Ic crmod Toggle carriage return mode. When this mode is enabled, most carriage return characters received from the remote host will be mapped into a carriage return followed by a line feed. This mode does not affect those characters typed by the user, only those received from the remote host. This mode is not very useful unless the remote host only sends carriage return, but never line feed. The initial value for this toggle is .Dv FALSE . .It Ic debug Toggles socket level debugging (useful only to the .Ic super user ) . The initial value for this toggle is .Dv FALSE . .It Ic encdebug Turns on debugging information for the encryption code. .It Ic localchars If this is .Dv TRUE , then the .Ic flush , .Ic interrupt , .Ic quit , .Ic erase , and .Ic kill characters (see .Ic set above) are recognized locally, and transformed into (hopefully) appropriate .Tn TELNET control sequences (respectively .Ic ao , .Ic ip , .Ic brk , .Ic ec , and .Ic el ; see .Ic send above). The initial value for this toggle is .Dv TRUE in \*(Lqold line by line\*(Rq mode, and .Dv FALSE in \*(Lqcharacter at a time\*(Rq mode. When the .Dv LINEMODE option is enabled, the value of .Ic localchars is ignored, and assumed to always be .Dv TRUE . If .Dv LINEMODE has ever been enabled, then .Ic quit is sent as .Ic abort , and .Ic eof and .Ic suspend are sent as .Ic eof and .Ic susp (see .Ic send above). .It Ic netdata Toggles the display of all network data (in hexadecimal format). The initial value for this toggle is .Dv FALSE . .It Ic options Toggles the display of some internal .Nm protocol processing (having to do with .Tn TELNET options). The initial value for this toggle is .Dv FALSE . .It Ic prettydump When the .Ic netdata toggle is enabled, if .Ic prettydump is enabled the output from the .Ic netdata command will be formatted in a more user readable format. Spaces are put between each character in the output, and the beginning of any .Nm escape sequence is preceded by a '*' to aid in locating them. .It Ic skiprc When the skiprc toggle is .Dv TRUE , .Nm skips the reading of the .Pa \&.telnetrc file in the users home directory when connections are opened. The initial value for this toggle is .Dv FALSE . .It Ic termdata Toggles the display of all terminal data (in hexadecimal format). The initial value for this toggle is .Dv FALSE . .It Ic verbose_encrypt When the .Ic verbose_encrypt toggle is .Dv TRUE , .Nm prints out a message each time encryption is enabled or disabled. The initial value for this toggle is .Dv FALSE . .It Ic ?\& Displays the legal .Ic toggle commands. .El .It Ic z Suspend .Nm . This command only works when the user is using the .Xr csh 1 . .It Ic \&! Op Ar command Execute a single command in a subshell on the local system. If .Ar command is omitted, then an interactive subshell is invoked. .It Ic ?\& Op Ar command Get help. With no arguments, .Nm prints a help summary. If .Ar command is specified, .Nm will print the help information for just that command. .El .Sh ENVIRONMENT .Nm uses at least the .Ev HOME , .Ev SHELL , .Ev DISPLAY , and .Ev TERM environment variables. Other environment variables may be propagated to the other side via the .Dv TELNET ENVIRON option. .Sh FILES .Bl -tag -width ~/.telnetrc -compact .It Pa ~/.telnetrc user customized telnet startup values .El .Sh SEE ALSO .Xr rlogin 1 , .Xr rsh 1 , .Xr hosts 5 , .Xr nologin 5 , .Xr telnetd 8 .Sh HISTORY The .Nm command appeared in .Bx 4.2 . .Pp IPv6 support was added by WIDE/KAME project. .Sh NOTES On some remote systems, echo has to be turned off manually when in \*(Lqold line by line\*(Rq mode. .Pp In \*(Lqold line by line\*(Rq mode or .Dv LINEMODE the terminal's .Ic eof character is only recognized (and sent to the remote system) when it is the first character on a line. Index: projects/clang1100-import/lib/lib80211/lib80211_regdomain.h =================================================================== --- projects/clang1100-import/lib/lib80211/lib80211_regdomain.h (revision 364034) +++ projects/clang1100-import/lib/lib80211/lib80211_regdomain.h (revision 364035) @@ -1,123 +1,123 @@ /*- * Copyright (c) 2007-2008 Sam Leffler, Errno Consulting * 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. * * $FreeBSD$ */ #ifndef _LIB80211_REGDOMAIN_H_ #define _LIB80211_REGDOMAIN_H_ #include #include #include __BEGIN_DECLS struct freqband { uint16_t freqStart; /* starting frequency (MHz) */ uint16_t freqEnd; /* ending frequency (MHz) */ uint8_t chanWidth; /* channel width (MHz) */ uint8_t chanSep; /* channel sepaaration (MHz) */ uint32_t flags; /* common operational constraints */ const void *id; LIST_ENTRY(freqband) next; }; /* private flags, don't pass to os */ #define REQ_ECM 0x1 /* enable if ECM set */ #define REQ_INDOOR 0x2 /* enable only for indoor operation */ #define REQ_OUTDOOR 0x4 /* enable only for outdoor operation */ #define REQ_FLAGS (REQ_ECM|REQ_INDOOR|REQ_OUTDOOR) struct netband { const struct freqband *band; /* channel list description */ uint8_t maxPower; /* regulatory cap on tx power (dBm) */ uint8_t maxPowerDFS; /* regulatory cap w/ DFS (dBm) */ uint8_t maxAntGain; /* max allowed antenna gain (.5 dBm) */ uint32_t flags; /* net80211 channel flags */ LIST_ENTRY(netband) next; }; typedef LIST_HEAD(, netband) netband_head; struct country; struct regdomain { enum RegdomainCode sku; /* regdomain code/SKU */ const char *name; /* printable name */ const struct country *cc; /* country code for 1-1/default map */ netband_head bands_11b; /* 11b operation */ netband_head bands_11g; /* 11g operation */ netband_head bands_11a; /* 11a operation */ - netband_head bands_11ng;/* 11ng operation */ - netband_head bands_11na;/* 11na operation */ - netband_head bands_11ac;/* 11ac 5GHz operation */ - netband_head bands_11acg;/* 11ac 2GHz operation */ + netband_head bands_11ng; /* 11ng operation */ + netband_head bands_11na; /* 11na operation */ + netband_head bands_11ac; /* 11ac 5GHz operation */ + netband_head bands_11acg; /* 11ac 2GHz operation */ LIST_ENTRY(regdomain) next; }; struct country { enum ISOCountryCode code; #define NO_COUNTRY 0xffff const struct regdomain *rd; const char* isoname; const char* name; LIST_ENTRY(country) next; }; struct ident; struct regdata { LIST_HEAD(, country) countries; /* country code table */ LIST_HEAD(, regdomain) domains; /* regulatory domains */ LIST_HEAD(, freqband) freqbands; /* frequency band table */ struct ident *ident; /* identifier table */ }; #define _PATH_REGDOMAIN "/etc/regdomain.xml" struct regdata *lib80211_alloc_regdata(void); void lib80211_free_regdata(struct regdata *); int lib80211_regdomain_readconfig(struct regdata *, const void *, size_t); void lib80211_regdomain_cleanup(struct regdata *); const struct regdomain *lib80211_regdomain_findbysku(const struct regdata *, enum RegdomainCode); const struct regdomain *lib80211_regdomain_findbyname(const struct regdata *, const char *); const struct country *lib80211_country_findbycc(const struct regdata *, enum ISOCountryCode); const struct country *lib80211_country_findbyname(const struct regdata *, const char *); __END_DECLS #endif /* _LIB80211_REGDOMAIN_H_ */ Index: projects/clang1100-import/lib/libc/locale/mbsrtowcs.3 =================================================================== --- projects/clang1100-import/lib/libc/locale/mbsrtowcs.3 (revision 364034) +++ projects/clang1100-import/lib/libc/locale/mbsrtowcs.3 (revision 364035) @@ -1,135 +1,149 @@ .\" Copyright (c) 2002-2004 Tim J. Robbins .\" 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 July 21, 2004 +.Dd August 7, 2020 .Dt MBSRTOWCS 3 .Os .Sh NAME .Nm mbsrtowcs , .Nm mbsnrtowcs .Nd "convert a character string to a wide-character string (restartable)" .Sh LIBRARY .Lb libc .Sh SYNOPSIS .In wchar.h .Ft size_t .Fo mbsrtowcs .Fa "wchar_t * restrict dst" "const char ** restrict src" "size_t len" .Fa "mbstate_t * restrict ps" .Fc .Ft size_t .Fo mbsnrtowcs .Fa "wchar_t * restrict dst" "const char ** restrict src" "size_t nms" .Fa "size_t len" "mbstate_t * restrict ps" .Fc .Sh DESCRIPTION The .Fn mbsrtowcs function converts a sequence of multibyte characters pointed to indirectly by .Fa src into a sequence of corresponding wide characters and stores at most .Fa len of them in the .Vt wchar_t array pointed to by .Fa dst , until it encounters a terminating null character .Pq Li '\e0' . .Pp If .Fa dst is .Dv NULL , no characters are stored. .Pp If .Fa dst is not .Dv NULL , the pointer pointed to by .Fa src is updated to point to the character after the one that conversion stopped at. If conversion stops because a null character is encountered, .Fa *src is set to .Dv NULL . .Pp The .Vt mbstate_t argument, .Fa ps , is used to keep track of the shift state. If it is .Dv NULL , .Fn mbsrtowcs uses an internal, static .Vt mbstate_t object, which is initialized to the initial conversion state at program startup. .Pp The .Fn mbsnrtowcs function behaves identically to .Fn mbsrtowcs , except that conversion stops after reading at most .Fa nms bytes from the buffer pointed to by .Fa src . .Sh RETURN VALUES -The +If successful, and +.Fa dst +is not NULL, the .Fn mbsrtowcs and .Fn mbsnrtowcs functions return the number of wide characters stored in the array pointed to by +.Fa dst . +.Pp +If .Fa dst -if successful, otherwise it returns -.Po Vt size_t Pc Ns \-1 . +was NULL then the functions +.Fn mbsrtowcs +and +.Fn mbsnrtowcs +return the number of wide characters that would have been stored where +.Fa dst +points to an infinitely large array. +.Pp +If either one of the functions is not successful then +.Po Vt size_t Pc Ns \-1 +is returned. .Sh ERRORS The .Fn mbsrtowcs and .Fn mbsnrtowcs functions will fail if: .Bl -tag -width Er .It Bq Er EILSEQ An invalid multibyte character sequence was encountered. .It Bq Er EINVAL The conversion state is invalid. .El .Sh SEE ALSO .Xr mbrtowc 3 , .Xr mbstowcs 3 , .Xr multibyte 3 , .Xr wcsrtombs 3 .Sh STANDARDS The .Fn mbsrtowcs function conforms to .St -isoC-99 . .Pp The .Fn mbsnrtowcs function is an extension to the standard. Index: projects/clang1100-import/lib/libc/locale/setlocale.3 =================================================================== --- projects/clang1100-import/lib/libc/locale/setlocale.3 (revision 364034) +++ projects/clang1100-import/lib/libc/locale/setlocale.3 (revision 364035) @@ -1,188 +1,211 @@ .\" Copyright (c) 1993 .\" The Regents of the University of California. All rights reserved. .\" .\" This code is derived from software contributed to Berkeley by .\" Donn Seeley at BSDI. .\" .\" 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. .\" .\" @(#)setlocale.3 8.1 (Berkeley) 6/9/93 .\" $FreeBSD$ .\" -.Dd September 9, 2019 +.Dd August 7, 2020 .Dt SETLOCALE 3 .Os .Sh NAME .Nm setlocale .Nd natural language formatting for C .Sh LIBRARY .Lb libc .Sh SYNOPSIS .In locale.h .Ft char * .Fn setlocale "int category" "const char *locale" .Sh DESCRIPTION The .Fn setlocale function sets the C library's notion of natural language formatting style for particular sets of routines. Each such style is called a .Sq locale and is invoked using an appropriate name passed as a C string. .Pp The .Fn setlocale function recognizes several categories of routines. These are the categories and the sets of routines they select: .Bl -tag -width LC_MONETARY .It Dv LC_ALL Set the entire locale generically. .It Dv LC_COLLATE Set a locale for string collation routines. This controls alphabetic ordering in .Fn strcoll and .Fn strxfrm . .It Dv LC_CTYPE Set a locale for the .Xr ctype 3 and .Xr multibyte 3 functions. This controls recognition of upper and lower case, alphabetic or non-alphabetic characters, and so on. .It Dv LC_MESSAGES Set a locale for message catalogs, see .Xr catopen 3 function. .It Dv LC_MONETARY Set a locale for formatting monetary values; this affects the .Fn localeconv function. .It Dv LC_NUMERIC Set a locale for formatting numbers. This controls the formatting of decimal points in input and output of floating point numbers in functions such as .Fn printf and .Fn scanf , as well as values returned by .Fn localeconv . .It Dv LC_TIME Set a locale for formatting dates and times using the .Fn strftime function. +.It Dv LANG +Sets the generic locale category for native language, local customs +and coded character set in the absence of more specific locale +variables. .El .Pp Only three locales are defined by default, the empty string .Li \&"\|" which denotes the native environment, and the .Li \&"C" and .Li \&"POSIX" locales, which denote the C language environment. A .Fa locale argument of .Dv NULL causes .Fn setlocale to return the current locale. .Pp The option .Fl a to the .Xr locale 1 command can be used to display all further possible names for the .Fa locale argument that are recognized. Specifying any unrecognized value for .Fa locale makes .Fn setlocale fail. .Pp By default, C programs start in the .Li \&"C" locale. .Pp The only function in the library that sets the locale is .Fn setlocale ; the locale is never changed as a side effect of some other routine. .Sh RETURN VALUES Upon successful completion, .Fn setlocale returns the string associated with the specified .Fa category for the requested .Fa locale . The .Fn setlocale function returns .Dv NULL and fails to change the locale if the given combination of .Fa category and .Fa locale makes no sense. +.Sh EXAMPLES +The following code illustrates how a program can initialize the +international environment for one language, while selectively +modifying the program's locale such that regular expressions and +string operations can be applied to text recorded in a different +language: +.Bd -literal + setlocale(LC_ALL, "de"); + setlocale(LC_COLLATE, "fr"); +.Ed +.Pp +When a process is started, its current locale is set to the C or POSIX +locale. +An internationalized program that depends on locale data not defined in +the C or POSIX locale must invoke the setlocale subroutine in the +following manner before using any of the locale-specific information: +.Bd -literal + setlocale(LC_ALL, ""); +.Ed .Sh FILES .Bl -tag -width /usr/share/locale/locale/category -compact .It Pa $PATH_LOCALE/ Ns Em locale/category .It Pa /usr/share/locale/ Ns Em locale/category locale file for the locale .Em locale and the category .Em category . .El .Sh ERRORS No errors are defined. .Sh SEE ALSO .Xr locale 1 , .Xr localedef 1 , .Xr catopen 3 , .Xr ctype 3 , .Xr localeconv 3 , .Xr multibyte 3 , .Xr strcoll 3 , .Xr strxfrm 3 , .Xr euc 5 , .Xr utf8 5 , .Xr environ 7 .Sh STANDARDS The .Fn setlocale function conforms to .St -isoC-99 . .Sh HISTORY The .Fn setlocale function first appeared in .Bx 4.4 . Index: projects/clang1100-import/lib/libc/stdio/tmpnam.3 =================================================================== --- projects/clang1100-import/lib/libc/stdio/tmpnam.3 (revision 364034) +++ projects/clang1100-import/lib/libc/stdio/tmpnam.3 (revision 364035) @@ -1,246 +1,248 @@ .\" Copyright (c) 1988, 1991, 1993 .\" The Regents of the University of California. All rights reserved. .\" .\" This code is derived from software contributed to Berkeley by .\" the American National Standards Committee X3, on Information .\" Processing Systems. .\" .\" 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. .\" .\" @(#)tmpnam.3 8.2 (Berkeley) 11/17/93 .\" $FreeBSD$ .\" -.Dd March 18, 2007 +.Dd August 7, 2020 .Dt TMPFILE 3 .Os .Sh NAME .Nm tempnam , .Nm tmpfile , .Nm tmpnam .Nd temporary file routines .Sh LIBRARY .Lb libc .Sh SYNOPSIS .In stdio.h .Ft FILE * .Fn tmpfile void .Ft char * .Fn tmpnam "char *str" .Ft char * .Fn tempnam "const char *tmpdir" "const char *prefix" .Sh DESCRIPTION The .Fn tmpfile function returns a pointer to a stream associated with a file descriptor returned by the routine .Xr mkstemp 3 . The created file is unlinked before .Fn tmpfile returns, causing the file to be automatically deleted when the last reference to it is closed. The file is opened with the access value .Ql w+ . The file is created in the directory determined by the environment variable .Ev TMPDIR if set. The default location if .Ev TMPDIR is not set is .Pa /tmp . .Pp The .Fn tmpnam function returns a pointer to a file name, in the .Dv P_tmpdir directory, which did not reference an existing file at some indeterminate point in the past. .Dv P_tmpdir is defined in the include file .In stdio.h . If the argument .Fa str is .Pf non- Dv NULL , the file name is copied to the buffer it references. Otherwise, the file name is copied to a static buffer. In either case, .Fn tmpnam returns a pointer to the file name. .Pp The buffer referenced by .Fa str is expected to be at least .Dv L_tmpnam bytes in length. .Dv L_tmpnam is defined in the include file .In stdio.h . .Pp The .Fn tempnam function is similar to .Fn tmpnam , but provides the ability to specify the directory which will contain the temporary file and the file name prefix. .Pp The environment variable .Ev TMPDIR (if set), the argument .Fa tmpdir (if .Pf non- Dv NULL ) , the directory .Dv P_tmpdir , and the directory .Pa /tmp are tried, in the listed order, as directories in which to store the temporary file. .Pp The argument .Fa prefix , if .Pf non- Dv NULL , is used to specify a file name prefix, which will be the first part of the created file name. The .Fn tempnam function allocates memory in which to store the file name; the returned pointer may be used as a subsequent argument to .Xr free 3 . .Sh RETURN VALUES The .Fn tmpfile function returns a pointer to an open file stream on success, and a .Dv NULL pointer on error. .Pp The .Fn tmpnam and .Fn tempfile functions return a pointer to a file name on success, and a .Dv NULL pointer on error. .Sh ENVIRONMENT .Bl -tag -width Ds .It Ev TMPDIR .Pf [ Fn tempnam +and +.Fn tmpfile only] If set, the directory in which the temporary file is stored. .Ev TMPDIR is ignored for processes for which .Xr issetugid 2 is true. .El .Sh COMPATIBILITY These interfaces are provided from System V and .Tn ANSI compatibility only. .Pp Most historic implementations of these functions provide only a limited number of possible temporary file names (usually 26) before file names will start being recycled. System V implementations of these functions (and of .Xr mktemp 3 ) use the .Xr access 2 system call to determine whether or not the temporary file may be created. This has obvious ramifications for setuid or setgid programs, complicating the portable use of these interfaces in such programs. .Pp The .Fn tmpfile interface should not be used in software expected to be used on other systems if there is any possibility that the user does not wish the temporary file to be publicly readable and writable. .Sh ERRORS The .Fn tmpfile function may fail and set the global variable .Va errno for any of the errors specified for the library functions .Xr fdopen 3 or .Xr mkstemp 3 . .Pp The .Fn tmpnam function may fail and set .Va errno for any of the errors specified for the library function .Xr mktemp 3 . .Pp The .Fn tempnam function may fail and set .Va errno for any of the errors specified for the library functions .Xr malloc 3 or .Xr mktemp 3 . .Sh SEE ALSO .Xr mkstemp 3 , .Xr mktemp 3 .Sh STANDARDS The .Fn tmpfile and .Fn tmpnam functions conform to .St -isoC . .Sh SECURITY CONSIDERATIONS The .Fn tmpnam and .Fn tempnam functions are susceptible to a race condition occurring between the selection of the file name and the creation of the file, which allows malicious users to potentially overwrite arbitrary files in the system, depending on the level of privilege of the running program. Additionally, there is no means by which file permissions may be specified. It is strongly suggested that .Xr mkstemp 3 be used in place of these functions. Index: projects/clang1100-import/release/packages/generate-ucl.sh =================================================================== --- projects/clang1100-import/release/packages/generate-ucl.sh (revision 364034) +++ projects/clang1100-import/release/packages/generate-ucl.sh (revision 364035) @@ -1,151 +1,153 @@ #!/bin/sh # # $FreeBSD$ # main() { desc= comment= debug= uclsource= while getopts "do:s:u:" arg; do case ${arg} in d) debug=1 ;; o) outname="${OPTARG}" origname="${OPTARG}" ;; s) srctree="${OPTARG}" ;; u) uclfile="${OPTARG}" ;; *) echo "Unknown argument" ;; esac done shift $(( ${OPTIND} - 1 )) outname="$(echo ${outname} | tr '-' '_')" case "${outname}" in clibs) # clibs should not have any dependencies or anything # else imposed on it. ;; caroot) pkgdeps="utilities" ;; runtime) outname="runtime" uclfile="${uclfile}" ;; runtime_manuals) outname="${origname}" pkgdeps="runtime" ;; runtime_*) outname="${origname}" uclfile="${outname##*}${uclfile}" pkgdeps="runtime" _descr="$(make -C ${srctree}/release/packages -f Makefile.package -V ${outname}_DESCR)" ;; jail_*) outname="${origname}" uclfile="${outname##*}${uclfile}" pkgdeps="runtime" _descr="$(make -C ${srctree}/release/packages -f Makefile.package -V ${outname}_DESCR)" ;; *_lib32_dev) outname="${outname%%_lib32_dev}" _descr="32-bit Libraries, Development Files" pkgdeps="${outname}" ;; *_lib32_dbg) outname="${outname%%_lib32_dbg}" _descr="32-bit Libraries, Debugging Symbols" pkgdeps="${outname}" ;; *_lib32) outname="${outname%%_lib32}" _descr="32-bit Libraries" pkgdeps="${outname}" ;; *_dev) outname="${outname%%_dev}" _descr="Development Files" pkgdeps="${outname}" ;; *_dbg) outname="${outname%%_dbg}" _descr="Debugging Symbols" pkgdeps="${outname}" ;; ${origname}) pkgdeps="runtime" ;; *) uclfile="${outname##*}${origname}" outname="${outname##*}${origname}" ;; esac outname="${outname%%_*}" + pkgdeps="$(echo ${pkgdeps} | tr '_' '-')" + desc="$(make -C ${srctree}/release/packages -f Makefile.package -V ${outname}_DESC)" comment="$(make -C ${srctree}/release/packages -f Makefile.package -V ${outname}_COMMENT)" uclsource="${srctree}/release/packages/${outname}.ucl" if [ ! -e "${uclsource}" ]; then uclsource="${srctree}/release/packages/template.ucl" fi if [ ! -z "${debug}" ]; then echo "" echo "===============================================================" echo "DEBUG:" echo "_descr=${_descr}" echo "outname=${outname}" echo "origname=${origname}" echo "srctree=${srctree}" echo "uclfile=${uclfile}" echo "desc=${desc}" echo "comment=${comment}" echo "cp ${uclsource} -> ${uclfile}" echo "===============================================================" echo "" echo "" echo "" fi [ -z "${comment}" ] && comment="${outname} package" [ ! -z "${_descr}" ] && comment="${comment} (${_descr})" [ -z "${desc}" ] && desc="${outname} package" cp "${uclsource}" "${uclfile}" if [ ! -z "${pkgdeps}" ]; then cat <> ${uclfile} deps: { FreeBSD-${pkgdeps}: { origin: "base", version: "${PKG_VERSION}" } } EOF fi cap_arg="$( make -f ${srctree}/share/mk/bsd.endian.mk -VCAP_MKDB_ENDIAN )" sed -i '' -e "s/%VERSION%/${PKG_VERSION}/" \ -e "s/%PKGNAME%/${origname}/" \ -e "s/%COMMENT%/${comment}/" \ -e "s/%DESC%/${desc}/" \ -e "s/%CAP_MKDB_ENDIAN%/${cap_arg}/g" \ ${uclfile} return 0 } main "${@}" Index: projects/clang1100-import/sbin/ifconfig/ifieee80211.c =================================================================== --- projects/clang1100-import/sbin/ifconfig/ifieee80211.c (revision 364034) +++ projects/clang1100-import/sbin/ifconfig/ifieee80211.c (revision 364035) @@ -1,5996 +1,6019 @@ /*- * SPDX-License-Identifier: BSD-3-Clause * * Copyright 2001 The Aerospace Corporation. 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. The name of The Aerospace Corporation may not be used to endorse or * promote products derived from this software. * * THIS SOFTWARE IS PROVIDED BY THE AEROSPACE CORPORATION ``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 AEROSPACE CORPORATION 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. * * $FreeBSD$ */ /*- * Copyright (c) 1997, 1998, 2000 The NetBSD Foundation, Inc. * All rights reserved. * * This code is derived from software contributed to The NetBSD Foundation * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility, * NASA Ames Research Center. * * 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 NETBSD FOUNDATION, INC. 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 FOUNDATION 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 #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* NB: for offsetof */ #include #include #include "ifconfig.h" #include #include #ifndef IEEE80211_FIXED_RATE_NONE #define IEEE80211_FIXED_RATE_NONE 0xff #endif /* XXX need these publicly defined or similar */ #ifndef IEEE80211_NODE_AUTH #define IEEE80211_NODE_AUTH 0x000001 /* authorized for data */ #define IEEE80211_NODE_QOS 0x000002 /* QoS enabled */ #define IEEE80211_NODE_ERP 0x000004 /* ERP enabled */ #define IEEE80211_NODE_PWR_MGT 0x000010 /* power save mode enabled */ #define IEEE80211_NODE_AREF 0x000020 /* authentication ref held */ #define IEEE80211_NODE_HT 0x000040 /* HT enabled */ #define IEEE80211_NODE_HTCOMPAT 0x000080 /* HT setup w/ vendor OUI's */ #define IEEE80211_NODE_WPS 0x000100 /* WPS association */ #define IEEE80211_NODE_TSN 0x000200 /* TSN association */ #define IEEE80211_NODE_AMPDU_RX 0x000400 /* AMPDU rx enabled */ #define IEEE80211_NODE_AMPDU_TX 0x000800 /* AMPDU tx enabled */ #define IEEE80211_NODE_MIMO_PS 0x001000 /* MIMO power save enabled */ #define IEEE80211_NODE_MIMO_RTS 0x002000 /* send RTS in MIMO PS */ #define IEEE80211_NODE_RIFS 0x004000 /* RIFS enabled */ #define IEEE80211_NODE_SGI20 0x008000 /* Short GI in HT20 enabled */ #define IEEE80211_NODE_SGI40 0x010000 /* Short GI in HT40 enabled */ #define IEEE80211_NODE_ASSOCID 0x020000 /* xmit requires associd */ #define IEEE80211_NODE_AMSDU_RX 0x040000 /* AMSDU rx enabled */ #define IEEE80211_NODE_AMSDU_TX 0x080000 /* AMSDU tx enabled */ #define IEEE80211_NODE_VHT 0x100000 /* VHT enabled */ #define IEEE80211_NODE_LDPC 0x200000 /* LDPC enabled */ #define IEEE80211_NODE_UAPSD 0x400000 /* UAPSD enabled */ #endif #define MAXCHAN 1536 /* max 1.5K channels */ #define MAXCOL 78 static int col; static char spacer; static void LINE_INIT(char c); static void LINE_BREAK(void); static void LINE_CHECK(const char *fmt, ...); static const char *modename[IEEE80211_MODE_MAX] = { [IEEE80211_MODE_AUTO] = "auto", [IEEE80211_MODE_11A] = "11a", [IEEE80211_MODE_11B] = "11b", [IEEE80211_MODE_11G] = "11g", [IEEE80211_MODE_FH] = "fh", [IEEE80211_MODE_TURBO_A] = "turboA", [IEEE80211_MODE_TURBO_G] = "turboG", [IEEE80211_MODE_STURBO_A] = "sturbo", [IEEE80211_MODE_11NA] = "11na", [IEEE80211_MODE_11NG] = "11ng", [IEEE80211_MODE_HALF] = "half", [IEEE80211_MODE_QUARTER] = "quarter", [IEEE80211_MODE_VHT_2GHZ] = "11acg", [IEEE80211_MODE_VHT_5GHZ] = "11ac", }; static void set80211(int s, int type, int val, int len, void *data); static int get80211(int s, int type, void *data, int len); static int get80211len(int s, int type, void *data, int len, int *plen); static int get80211val(int s, int type, int *val); static const char *get_string(const char *val, const char *sep, u_int8_t *buf, int *lenp); static void print_string(const u_int8_t *buf, int len); static void print_regdomain(const struct ieee80211_regdomain *, int); static void print_channels(int, const struct ieee80211req_chaninfo *, int allchans, int verbose); static void regdomain_makechannels(struct ieee80211_regdomain_req *, const struct ieee80211_devcaps_req *); static const char *mesh_linkstate_string(uint8_t state); static struct ieee80211req_chaninfo *chaninfo; static struct ieee80211_regdomain regdomain; static int gotregdomain = 0; static struct ieee80211_roamparams_req roamparams; static int gotroam = 0; static struct ieee80211_txparams_req txparams; static int gottxparams = 0; static struct ieee80211_channel curchan; static int gotcurchan = 0; static struct ifmediareq *ifmr; static int htconf = 0; static int gothtconf = 0; static void gethtconf(int s) { if (gothtconf) return; if (get80211val(s, IEEE80211_IOC_HTCONF, &htconf) < 0) warn("unable to get HT configuration information"); gothtconf = 1; } /* VHT */ static int vhtconf = 0; static int gotvhtconf = 0; static void getvhtconf(int s) { if (gotvhtconf) return; if (get80211val(s, IEEE80211_IOC_VHTCONF, &vhtconf) < 0) warn("unable to get VHT configuration information"); gotvhtconf = 1; } /* * Collect channel info from the kernel. We use this (mostly) * to handle mapping between frequency and IEEE channel number. */ static void getchaninfo(int s) { if (chaninfo != NULL) return; chaninfo = malloc(IEEE80211_CHANINFO_SIZE(MAXCHAN)); if (chaninfo == NULL) errx(1, "no space for channel list"); if (get80211(s, IEEE80211_IOC_CHANINFO, chaninfo, IEEE80211_CHANINFO_SIZE(MAXCHAN)) < 0) err(1, "unable to get channel information"); ifmr = ifmedia_getstate(s); gethtconf(s); getvhtconf(s); } static struct regdata * getregdata(void) { static struct regdata *rdp = NULL; if (rdp == NULL) { rdp = lib80211_alloc_regdata(); if (rdp == NULL) errx(-1, "missing or corrupted regdomain database"); } return rdp; } /* * Given the channel at index i with attributes from, * check if there is a channel with attributes to in * the channel table. With suitable attributes this * allows the caller to look for promotion; e.g. from * 11b > 11g. */ static int canpromote(int i, int from, int to) { const struct ieee80211_channel *fc = &chaninfo->ic_chans[i]; u_int j; if ((fc->ic_flags & from) != from) return i; /* NB: quick check exploiting ordering of chans w/ same frequency */ if (i+1 < chaninfo->ic_nchans && chaninfo->ic_chans[i+1].ic_freq == fc->ic_freq && (chaninfo->ic_chans[i+1].ic_flags & to) == to) return i+1; /* brute force search in case channel list is not ordered */ for (j = 0; j < chaninfo->ic_nchans; j++) { const struct ieee80211_channel *tc = &chaninfo->ic_chans[j]; if (j != i && tc->ic_freq == fc->ic_freq && (tc->ic_flags & to) == to) return j; } return i; } /* * Handle channel promotion. When a channel is specified with * only a frequency we want to promote it to the ``best'' channel * available. The channel list has separate entries for 11b, 11g, * 11a, and 11n[ga] channels so specifying a frequency w/o any * attributes requires we upgrade, e.g. from 11b -> 11g. This * gets complicated when the channel is specified on the same * command line with a media request that constrains the available * channe list (e.g. mode 11a); we want to honor that to avoid * confusing behaviour. */ /* * XXX VHT */ static int promote(int i) { /* * Query the current mode of the interface in case it's * constrained (e.g. to 11a). We must do this carefully * as there may be a pending ifmedia request in which case * asking the kernel will give us the wrong answer. This * is an unfortunate side-effect of the way ifconfig is * structure for modularity (yech). * * NB: ifmr is actually setup in getchaninfo (above); we * assume it's called coincident with to this call so * we have a ``current setting''; otherwise we must pass * the socket descriptor down to here so we can make * the ifmedia_getstate call ourselves. */ int chanmode = ifmr != NULL ? IFM_MODE(ifmr->ifm_current) : IFM_AUTO; /* when ambiguous promote to ``best'' */ /* NB: we abitrarily pick HT40+ over HT40- */ if (chanmode != IFM_IEEE80211_11B) i = canpromote(i, IEEE80211_CHAN_B, IEEE80211_CHAN_G); if (chanmode != IFM_IEEE80211_11G && (htconf & 1)) { i = canpromote(i, IEEE80211_CHAN_G, IEEE80211_CHAN_G | IEEE80211_CHAN_HT20); if (htconf & 2) { i = canpromote(i, IEEE80211_CHAN_G, IEEE80211_CHAN_G | IEEE80211_CHAN_HT40D); i = canpromote(i, IEEE80211_CHAN_G, IEEE80211_CHAN_G | IEEE80211_CHAN_HT40U); } } if (chanmode != IFM_IEEE80211_11A && (htconf & 1)) { i = canpromote(i, IEEE80211_CHAN_A, IEEE80211_CHAN_A | IEEE80211_CHAN_HT20); if (htconf & 2) { i = canpromote(i, IEEE80211_CHAN_A, IEEE80211_CHAN_A | IEEE80211_CHAN_HT40D); i = canpromote(i, IEEE80211_CHAN_A, IEEE80211_CHAN_A | IEEE80211_CHAN_HT40U); } } return i; } static void mapfreq(struct ieee80211_channel *chan, int freq, int flags) { u_int i; for (i = 0; i < chaninfo->ic_nchans; i++) { const struct ieee80211_channel *c = &chaninfo->ic_chans[i]; if (c->ic_freq == freq && (c->ic_flags & flags) == flags) { if (flags == 0) { /* when ambiguous promote to ``best'' */ c = &chaninfo->ic_chans[promote(i)]; } *chan = *c; return; } } errx(1, "unknown/undefined frequency %u/0x%x", freq, flags); } static void mapchan(struct ieee80211_channel *chan, int ieee, int flags) { u_int i; for (i = 0; i < chaninfo->ic_nchans; i++) { const struct ieee80211_channel *c = &chaninfo->ic_chans[i]; if (c->ic_ieee == ieee && (c->ic_flags & flags) == flags) { if (flags == 0) { /* when ambiguous promote to ``best'' */ c = &chaninfo->ic_chans[promote(i)]; } *chan = *c; return; } } errx(1, "unknown/undefined channel number %d flags 0x%x", ieee, flags); } static const struct ieee80211_channel * getcurchan(int s) { if (gotcurchan) return &curchan; if (get80211(s, IEEE80211_IOC_CURCHAN, &curchan, sizeof(curchan)) < 0) { int val; /* fall back to legacy ioctl */ if (get80211val(s, IEEE80211_IOC_CHANNEL, &val) < 0) err(-1, "cannot figure out current channel"); getchaninfo(s); mapchan(&curchan, val, 0); } gotcurchan = 1; return &curchan; } static enum ieee80211_phymode chan2mode(const struct ieee80211_channel *c) { if (IEEE80211_IS_CHAN_VHTA(c)) return IEEE80211_MODE_VHT_5GHZ; if (IEEE80211_IS_CHAN_VHTG(c)) return IEEE80211_MODE_VHT_2GHZ; if (IEEE80211_IS_CHAN_HTA(c)) return IEEE80211_MODE_11NA; if (IEEE80211_IS_CHAN_HTG(c)) return IEEE80211_MODE_11NG; if (IEEE80211_IS_CHAN_108A(c)) return IEEE80211_MODE_TURBO_A; if (IEEE80211_IS_CHAN_108G(c)) return IEEE80211_MODE_TURBO_G; if (IEEE80211_IS_CHAN_ST(c)) return IEEE80211_MODE_STURBO_A; if (IEEE80211_IS_CHAN_FHSS(c)) return IEEE80211_MODE_FH; if (IEEE80211_IS_CHAN_HALF(c)) return IEEE80211_MODE_HALF; if (IEEE80211_IS_CHAN_QUARTER(c)) return IEEE80211_MODE_QUARTER; if (IEEE80211_IS_CHAN_A(c)) return IEEE80211_MODE_11A; if (IEEE80211_IS_CHAN_ANYG(c)) return IEEE80211_MODE_11G; if (IEEE80211_IS_CHAN_B(c)) return IEEE80211_MODE_11B; return IEEE80211_MODE_AUTO; } static void getroam(int s) { if (gotroam) return; if (get80211(s, IEEE80211_IOC_ROAM, &roamparams, sizeof(roamparams)) < 0) err(1, "unable to get roaming parameters"); gotroam = 1; } static void setroam_cb(int s, void *arg) { struct ieee80211_roamparams_req *roam = arg; set80211(s, IEEE80211_IOC_ROAM, 0, sizeof(*roam), roam); } static void gettxparams(int s) { if (gottxparams) return; if (get80211(s, IEEE80211_IOC_TXPARAMS, &txparams, sizeof(txparams)) < 0) err(1, "unable to get transmit parameters"); gottxparams = 1; } static void settxparams_cb(int s, void *arg) { struct ieee80211_txparams_req *txp = arg; set80211(s, IEEE80211_IOC_TXPARAMS, 0, sizeof(*txp), txp); } static void getregdomain(int s) { if (gotregdomain) return; if (get80211(s, IEEE80211_IOC_REGDOMAIN, ®domain, sizeof(regdomain)) < 0) err(1, "unable to get regulatory domain info"); gotregdomain = 1; } static void getdevcaps(int s, struct ieee80211_devcaps_req *dc) { if (get80211(s, IEEE80211_IOC_DEVCAPS, dc, IEEE80211_DEVCAPS_SPACE(dc)) < 0) err(1, "unable to get device capabilities"); } static void setregdomain_cb(int s, void *arg) { struct ieee80211_regdomain_req *req; struct ieee80211_regdomain *rd = arg; struct ieee80211_devcaps_req *dc; struct regdata *rdp = getregdata(); if (rd->country != NO_COUNTRY) { const struct country *cc; /* * Check current country seting to make sure it's * compatible with the new regdomain. If not, then * override it with any default country for this * SKU. If we cannot arrange a match, then abort. */ cc = lib80211_country_findbycc(rdp, rd->country); if (cc == NULL) errx(1, "unknown ISO country code %d", rd->country); if (cc->rd->sku != rd->regdomain) { const struct regdomain *rp; /* * Check if country is incompatible with regdomain. * To enable multiple regdomains for a country code * we permit a mismatch between the regdomain and * the country's associated regdomain when the * regdomain is setup w/o a default country. For * example, US is bound to the FCC regdomain but * we allow US to be combined with FCC3 because FCC3 * has not default country. This allows bogus * combinations like FCC3+DK which are resolved when * constructing the channel list by deferring to the * regdomain to construct the channel list. */ rp = lib80211_regdomain_findbysku(rdp, rd->regdomain); if (rp == NULL) errx(1, "country %s (%s) is not usable with " "regdomain %d", cc->isoname, cc->name, rd->regdomain); else if (rp->cc != NULL && rp->cc != cc) errx(1, "country %s (%s) is not usable with " "regdomain %s", cc->isoname, cc->name, rp->name); } } /* * Fetch the device capabilities and calculate the * full set of netbands for which we request a new * channel list be constructed. Once that's done we * push the regdomain info + channel list to the kernel. */ dc = malloc(IEEE80211_DEVCAPS_SIZE(MAXCHAN)); if (dc == NULL) errx(1, "no space for device capabilities"); dc->dc_chaninfo.ic_nchans = MAXCHAN; getdevcaps(s, dc); #if 0 if (verbose) { printf("drivercaps: 0x%x\n", dc->dc_drivercaps); printf("cryptocaps: 0x%x\n", dc->dc_cryptocaps); printf("htcaps : 0x%x\n", dc->dc_htcaps); printf("vhtcaps : 0x%x\n", dc->dc_vhtcaps); #if 0 memcpy(chaninfo, &dc->dc_chaninfo, IEEE80211_CHANINFO_SPACE(&dc->dc_chaninfo)); print_channels(s, &dc->dc_chaninfo, 1/*allchans*/, 1/*verbose*/); #endif } #endif req = malloc(IEEE80211_REGDOMAIN_SIZE(dc->dc_chaninfo.ic_nchans)); if (req == NULL) errx(1, "no space for regdomain request"); req->rd = *rd; regdomain_makechannels(req, dc); if (verbose) { LINE_INIT(':'); print_regdomain(rd, 1/*verbose*/); LINE_BREAK(); /* blech, reallocate channel list for new data */ if (chaninfo != NULL) free(chaninfo); chaninfo = malloc(IEEE80211_CHANINFO_SPACE(&req->chaninfo)); if (chaninfo == NULL) errx(1, "no space for channel list"); memcpy(chaninfo, &req->chaninfo, IEEE80211_CHANINFO_SPACE(&req->chaninfo)); print_channels(s, &req->chaninfo, 1/*allchans*/, 1/*verbose*/); } if (req->chaninfo.ic_nchans == 0) errx(1, "no channels calculated"); set80211(s, IEEE80211_IOC_REGDOMAIN, 0, IEEE80211_REGDOMAIN_SPACE(req), req); free(req); free(dc); } static int ieee80211_mhz2ieee(int freq, int flags) { struct ieee80211_channel chan; mapfreq(&chan, freq, flags); return chan.ic_ieee; } static int isanyarg(const char *arg) { return (strncmp(arg, "-", 1) == 0 || strncasecmp(arg, "any", 3) == 0 || strncasecmp(arg, "off", 3) == 0); } static void set80211ssid(const char *val, int d, int s, const struct afswtch *rafp) { int ssid; int len; u_int8_t data[IEEE80211_NWID_LEN]; ssid = 0; len = strlen(val); if (len > 2 && isdigit((int)val[0]) && val[1] == ':') { ssid = atoi(val)-1; val += 2; } bzero(data, sizeof(data)); len = sizeof(data); if (get_string(val, NULL, data, &len) == NULL) exit(1); set80211(s, IEEE80211_IOC_SSID, ssid, len, data); } static void set80211meshid(const char *val, int d, int s, const struct afswtch *rafp) { int len; u_int8_t data[IEEE80211_NWID_LEN]; memset(data, 0, sizeof(data)); len = sizeof(data); if (get_string(val, NULL, data, &len) == NULL) exit(1); set80211(s, IEEE80211_IOC_MESH_ID, 0, len, data); } static void set80211stationname(const char *val, int d, int s, const struct afswtch *rafp) { int len; u_int8_t data[33]; bzero(data, sizeof(data)); len = sizeof(data); get_string(val, NULL, data, &len); set80211(s, IEEE80211_IOC_STATIONNAME, 0, len, data); } /* * Parse a channel specification for attributes/flags. * The syntax is: * freq/xx channel width (5,10,20,40,40+,40-) * freq:mode channel mode (a,b,g,h,n,t,s,d) * * These can be combined in either order; e.g. 2437:ng/40. * Modes are case insensitive. * * The result is not validated here; it's assumed to be * checked against the channel table fetched from the kernel. */ static int getchannelflags(const char *val, int freq) { #define _CHAN_HT 0x80000000 const char *cp; int flags; int is_vht = 0; flags = 0; cp = strchr(val, ':'); if (cp != NULL) { for (cp++; isalpha((int) *cp); cp++) { /* accept mixed case */ int c = *cp; if (isupper(c)) c = tolower(c); switch (c) { case 'a': /* 802.11a */ flags |= IEEE80211_CHAN_A; break; case 'b': /* 802.11b */ flags |= IEEE80211_CHAN_B; break; case 'g': /* 802.11g */ flags |= IEEE80211_CHAN_G; break; case 'v': /* vht: 802.11ac */ is_vht = 1; /* Fallthrough */ case 'h': /* ht = 802.11n */ case 'n': /* 802.11n */ flags |= _CHAN_HT; /* NB: private */ break; case 'd': /* dt = Atheros Dynamic Turbo */ flags |= IEEE80211_CHAN_TURBO; break; case 't': /* ht, dt, st, t */ /* dt and unadorned t specify Dynamic Turbo */ if ((flags & (IEEE80211_CHAN_STURBO|_CHAN_HT)) == 0) flags |= IEEE80211_CHAN_TURBO; break; case 's': /* st = Atheros Static Turbo */ flags |= IEEE80211_CHAN_STURBO; break; default: errx(-1, "%s: Invalid channel attribute %c\n", val, *cp); } } } cp = strchr(val, '/'); if (cp != NULL) { char *ep; u_long cw = strtoul(cp+1, &ep, 10); switch (cw) { case 5: flags |= IEEE80211_CHAN_QUARTER; break; case 10: flags |= IEEE80211_CHAN_HALF; break; case 20: /* NB: this may be removed below */ flags |= IEEE80211_CHAN_HT20; break; case 40: case 80: case 160: /* Handle the 80/160 VHT flag */ if (cw == 80) flags |= IEEE80211_CHAN_VHT80; else if (cw == 160) flags |= IEEE80211_CHAN_VHT160; /* Fallthrough */ if (ep != NULL && *ep == '+') flags |= IEEE80211_CHAN_HT40U; else if (ep != NULL && *ep == '-') flags |= IEEE80211_CHAN_HT40D; break; default: errx(-1, "%s: Invalid channel width\n", val); } } /* * Cleanup specifications. */ if ((flags & _CHAN_HT) == 0) { /* * If user specified freq/20 or freq/40 quietly remove * HT cw attributes depending on channel use. To give * an explicit 20/40 width for an HT channel you must * indicate it is an HT channel since all HT channels * are also usable for legacy operation; e.g. freq:n/40. */ flags &= ~IEEE80211_CHAN_HT; flags &= ~IEEE80211_CHAN_VHT; } else { /* * Remove private indicator that this is an HT channel * and if no explicit channel width has been given * provide the default settings. */ flags &= ~_CHAN_HT; if ((flags & IEEE80211_CHAN_HT) == 0) { struct ieee80211_channel chan; /* * Consult the channel list to see if we can use * HT40+ or HT40- (if both the map routines choose). */ if (freq > 255) mapfreq(&chan, freq, 0); else mapchan(&chan, freq, 0); flags |= (chan.ic_flags & IEEE80211_CHAN_HT); } /* * If VHT is enabled, then also set the VHT flag and the * relevant channel up/down. */ if (is_vht && (flags & IEEE80211_CHAN_HT)) { /* * XXX yes, maybe we should just have VHT, and reuse * HT20/HT40U/HT40D */ if (flags & IEEE80211_CHAN_VHT80) ; else if (flags & IEEE80211_CHAN_HT20) flags |= IEEE80211_CHAN_VHT20; else if (flags & IEEE80211_CHAN_HT40U) flags |= IEEE80211_CHAN_VHT40U; else if (flags & IEEE80211_CHAN_HT40D) flags |= IEEE80211_CHAN_VHT40D; } } return flags; #undef _CHAN_HT } static void getchannel(int s, struct ieee80211_channel *chan, const char *val) { int v, flags; char *eptr; memset(chan, 0, sizeof(*chan)); if (isanyarg(val)) { chan->ic_freq = IEEE80211_CHAN_ANY; return; } getchaninfo(s); errno = 0; v = strtol(val, &eptr, 10); if (val[0] == '\0' || val == eptr || errno == ERANGE || /* channel may be suffixed with nothing, :flag, or /width */ (eptr[0] != '\0' && eptr[0] != ':' && eptr[0] != '/')) errx(1, "invalid channel specification%s", errno == ERANGE ? " (out of range)" : ""); flags = getchannelflags(val, v); if (v > 255) { /* treat as frequency */ mapfreq(chan, v, flags); } else { mapchan(chan, v, flags); } } static void set80211channel(const char *val, int d, int s, const struct afswtch *rafp) { struct ieee80211_channel chan; getchannel(s, &chan, val); set80211(s, IEEE80211_IOC_CURCHAN, 0, sizeof(chan), &chan); } static void set80211chanswitch(const char *val, int d, int s, const struct afswtch *rafp) { struct ieee80211_chanswitch_req csr; getchannel(s, &csr.csa_chan, val); csr.csa_mode = 1; csr.csa_count = 5; set80211(s, IEEE80211_IOC_CHANSWITCH, 0, sizeof(csr), &csr); } static void set80211authmode(const char *val, int d, int s, const struct afswtch *rafp) { int mode; if (strcasecmp(val, "none") == 0) { mode = IEEE80211_AUTH_NONE; } else if (strcasecmp(val, "open") == 0) { mode = IEEE80211_AUTH_OPEN; } else if (strcasecmp(val, "shared") == 0) { mode = IEEE80211_AUTH_SHARED; } else if (strcasecmp(val, "8021x") == 0) { mode = IEEE80211_AUTH_8021X; } else if (strcasecmp(val, "wpa") == 0) { mode = IEEE80211_AUTH_WPA; } else { errx(1, "unknown authmode"); } set80211(s, IEEE80211_IOC_AUTHMODE, mode, 0, NULL); } static void set80211powersavemode(const char *val, int d, int s, const struct afswtch *rafp) { int mode; if (strcasecmp(val, "off") == 0) { mode = IEEE80211_POWERSAVE_OFF; } else if (strcasecmp(val, "on") == 0) { mode = IEEE80211_POWERSAVE_ON; } else if (strcasecmp(val, "cam") == 0) { mode = IEEE80211_POWERSAVE_CAM; } else if (strcasecmp(val, "psp") == 0) { mode = IEEE80211_POWERSAVE_PSP; } else if (strcasecmp(val, "psp-cam") == 0) { mode = IEEE80211_POWERSAVE_PSP_CAM; } else { errx(1, "unknown powersavemode"); } set80211(s, IEEE80211_IOC_POWERSAVE, mode, 0, NULL); } static void set80211powersave(const char *val, int d, int s, const struct afswtch *rafp) { if (d == 0) set80211(s, IEEE80211_IOC_POWERSAVE, IEEE80211_POWERSAVE_OFF, 0, NULL); else set80211(s, IEEE80211_IOC_POWERSAVE, IEEE80211_POWERSAVE_ON, 0, NULL); } static void set80211powersavesleep(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_POWERSAVESLEEP, atoi(val), 0, NULL); } static void set80211wepmode(const char *val, int d, int s, const struct afswtch *rafp) { int mode; if (strcasecmp(val, "off") == 0) { mode = IEEE80211_WEP_OFF; } else if (strcasecmp(val, "on") == 0) { mode = IEEE80211_WEP_ON; } else if (strcasecmp(val, "mixed") == 0) { mode = IEEE80211_WEP_MIXED; } else { errx(1, "unknown wep mode"); } set80211(s, IEEE80211_IOC_WEP, mode, 0, NULL); } static void set80211wep(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_WEP, d, 0, NULL); } static int isundefarg(const char *arg) { return (strcmp(arg, "-") == 0 || strncasecmp(arg, "undef", 5) == 0); } static void set80211weptxkey(const char *val, int d, int s, const struct afswtch *rafp) { if (isundefarg(val)) set80211(s, IEEE80211_IOC_WEPTXKEY, IEEE80211_KEYIX_NONE, 0, NULL); else set80211(s, IEEE80211_IOC_WEPTXKEY, atoi(val)-1, 0, NULL); } static void set80211wepkey(const char *val, int d, int s, const struct afswtch *rafp) { int key = 0; int len; u_int8_t data[IEEE80211_KEYBUF_SIZE]; if (isdigit((int)val[0]) && val[1] == ':') { key = atoi(val)-1; val += 2; } bzero(data, sizeof(data)); len = sizeof(data); get_string(val, NULL, data, &len); set80211(s, IEEE80211_IOC_WEPKEY, key, len, data); } /* * This function is purely a NetBSD compatibility interface. The NetBSD * interface is too inflexible, but it's there so we'll support it since * it's not all that hard. */ static void set80211nwkey(const char *val, int d, int s, const struct afswtch *rafp) { int txkey; int i, len; u_int8_t data[IEEE80211_KEYBUF_SIZE]; set80211(s, IEEE80211_IOC_WEP, IEEE80211_WEP_ON, 0, NULL); if (isdigit((int)val[0]) && val[1] == ':') { txkey = val[0]-'0'-1; val += 2; for (i = 0; i < 4; i++) { bzero(data, sizeof(data)); len = sizeof(data); val = get_string(val, ",", data, &len); if (val == NULL) exit(1); set80211(s, IEEE80211_IOC_WEPKEY, i, len, data); } } else { bzero(data, sizeof(data)); len = sizeof(data); get_string(val, NULL, data, &len); txkey = 0; set80211(s, IEEE80211_IOC_WEPKEY, 0, len, data); bzero(data, sizeof(data)); for (i = 1; i < 4; i++) set80211(s, IEEE80211_IOC_WEPKEY, i, 0, data); } set80211(s, IEEE80211_IOC_WEPTXKEY, txkey, 0, NULL); } static void set80211rtsthreshold(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_RTSTHRESHOLD, isundefarg(val) ? IEEE80211_RTS_MAX : atoi(val), 0, NULL); } static void set80211protmode(const char *val, int d, int s, const struct afswtch *rafp) { int mode; if (strcasecmp(val, "off") == 0) { mode = IEEE80211_PROTMODE_OFF; } else if (strcasecmp(val, "cts") == 0) { mode = IEEE80211_PROTMODE_CTS; } else if (strncasecmp(val, "rtscts", 3) == 0) { mode = IEEE80211_PROTMODE_RTSCTS; } else { errx(1, "unknown protection mode"); } set80211(s, IEEE80211_IOC_PROTMODE, mode, 0, NULL); } static void set80211htprotmode(const char *val, int d, int s, const struct afswtch *rafp) { int mode; if (strcasecmp(val, "off") == 0) { mode = IEEE80211_PROTMODE_OFF; } else if (strncasecmp(val, "rts", 3) == 0) { mode = IEEE80211_PROTMODE_RTSCTS; } else { errx(1, "unknown protection mode"); } set80211(s, IEEE80211_IOC_HTPROTMODE, mode, 0, NULL); } static void set80211txpower(const char *val, int d, int s, const struct afswtch *rafp) { double v = atof(val); int txpow; txpow = (int) (2*v); if (txpow != 2*v) errx(-1, "invalid tx power (must be .5 dBm units)"); set80211(s, IEEE80211_IOC_TXPOWER, txpow, 0, NULL); } #define IEEE80211_ROAMING_DEVICE 0 #define IEEE80211_ROAMING_AUTO 1 #define IEEE80211_ROAMING_MANUAL 2 static void set80211roaming(const char *val, int d, int s, const struct afswtch *rafp) { int mode; if (strcasecmp(val, "device") == 0) { mode = IEEE80211_ROAMING_DEVICE; } else if (strcasecmp(val, "auto") == 0) { mode = IEEE80211_ROAMING_AUTO; } else if (strcasecmp(val, "manual") == 0) { mode = IEEE80211_ROAMING_MANUAL; } else { errx(1, "unknown roaming mode"); } set80211(s, IEEE80211_IOC_ROAMING, mode, 0, NULL); } static void set80211wme(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_WME, d, 0, NULL); } static void set80211hidessid(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_HIDESSID, d, 0, NULL); } static void set80211apbridge(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_APBRIDGE, d, 0, NULL); } static void set80211fastframes(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_FF, d, 0, NULL); } static void set80211dturbo(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_TURBOP, d, 0, NULL); } static void set80211chanlist(const char *val, int d, int s, const struct afswtch *rafp) { struct ieee80211req_chanlist chanlist; char *temp, *cp, *tp; temp = malloc(strlen(val) + 1); if (temp == NULL) errx(1, "malloc failed"); strcpy(temp, val); memset(&chanlist, 0, sizeof(chanlist)); cp = temp; for (;;) { int first, last, f, c; tp = strchr(cp, ','); if (tp != NULL) *tp++ = '\0'; switch (sscanf(cp, "%u-%u", &first, &last)) { case 1: if (first > IEEE80211_CHAN_MAX) errx(-1, "channel %u out of range, max %u", first, IEEE80211_CHAN_MAX); setbit(chanlist.ic_channels, first); break; case 2: if (first > IEEE80211_CHAN_MAX) errx(-1, "channel %u out of range, max %u", first, IEEE80211_CHAN_MAX); if (last > IEEE80211_CHAN_MAX) errx(-1, "channel %u out of range, max %u", last, IEEE80211_CHAN_MAX); if (first > last) errx(-1, "void channel range, %u > %u", first, last); for (f = first; f <= last; f++) setbit(chanlist.ic_channels, f); break; } if (tp == NULL) break; c = *tp; while (isspace(c)) tp++; if (!isdigit(c)) break; cp = tp; } set80211(s, IEEE80211_IOC_CHANLIST, 0, sizeof(chanlist), &chanlist); free(temp); } static void set80211bssid(const char *val, int d, int s, const struct afswtch *rafp) { if (!isanyarg(val)) { char *temp; struct sockaddr_dl sdl; temp = malloc(strlen(val) + 2); /* ':' and '\0' */ if (temp == NULL) errx(1, "malloc failed"); temp[0] = ':'; strcpy(temp + 1, val); sdl.sdl_len = sizeof(sdl); link_addr(temp, &sdl); free(temp); if (sdl.sdl_alen != IEEE80211_ADDR_LEN) errx(1, "malformed link-level address"); set80211(s, IEEE80211_IOC_BSSID, 0, IEEE80211_ADDR_LEN, LLADDR(&sdl)); } else { uint8_t zerobssid[IEEE80211_ADDR_LEN]; memset(zerobssid, 0, sizeof(zerobssid)); set80211(s, IEEE80211_IOC_BSSID, 0, IEEE80211_ADDR_LEN, zerobssid); } } static int getac(const char *ac) { if (strcasecmp(ac, "ac_be") == 0 || strcasecmp(ac, "be") == 0) return WME_AC_BE; if (strcasecmp(ac, "ac_bk") == 0 || strcasecmp(ac, "bk") == 0) return WME_AC_BK; if (strcasecmp(ac, "ac_vi") == 0 || strcasecmp(ac, "vi") == 0) return WME_AC_VI; if (strcasecmp(ac, "ac_vo") == 0 || strcasecmp(ac, "vo") == 0) return WME_AC_VO; errx(1, "unknown wme access class %s", ac); } static DECL_CMD_FUNC2(set80211cwmin, ac, val) { set80211(s, IEEE80211_IOC_WME_CWMIN, atoi(val), getac(ac), NULL); } static DECL_CMD_FUNC2(set80211cwmax, ac, val) { set80211(s, IEEE80211_IOC_WME_CWMAX, atoi(val), getac(ac), NULL); } static DECL_CMD_FUNC2(set80211aifs, ac, val) { set80211(s, IEEE80211_IOC_WME_AIFS, atoi(val), getac(ac), NULL); } static DECL_CMD_FUNC2(set80211txoplimit, ac, val) { set80211(s, IEEE80211_IOC_WME_TXOPLIMIT, atoi(val), getac(ac), NULL); } static DECL_CMD_FUNC(set80211acm, ac, d) { set80211(s, IEEE80211_IOC_WME_ACM, 1, getac(ac), NULL); } static DECL_CMD_FUNC(set80211noacm, ac, d) { set80211(s, IEEE80211_IOC_WME_ACM, 0, getac(ac), NULL); } static DECL_CMD_FUNC(set80211ackpolicy, ac, d) { set80211(s, IEEE80211_IOC_WME_ACKPOLICY, 1, getac(ac), NULL); } static DECL_CMD_FUNC(set80211noackpolicy, ac, d) { set80211(s, IEEE80211_IOC_WME_ACKPOLICY, 0, getac(ac), NULL); } static DECL_CMD_FUNC2(set80211bsscwmin, ac, val) { set80211(s, IEEE80211_IOC_WME_CWMIN, atoi(val), getac(ac)|IEEE80211_WMEPARAM_BSS, NULL); } static DECL_CMD_FUNC2(set80211bsscwmax, ac, val) { set80211(s, IEEE80211_IOC_WME_CWMAX, atoi(val), getac(ac)|IEEE80211_WMEPARAM_BSS, NULL); } static DECL_CMD_FUNC2(set80211bssaifs, ac, val) { set80211(s, IEEE80211_IOC_WME_AIFS, atoi(val), getac(ac)|IEEE80211_WMEPARAM_BSS, NULL); } static DECL_CMD_FUNC2(set80211bsstxoplimit, ac, val) { set80211(s, IEEE80211_IOC_WME_TXOPLIMIT, atoi(val), getac(ac)|IEEE80211_WMEPARAM_BSS, NULL); } static DECL_CMD_FUNC(set80211dtimperiod, val, d) { set80211(s, IEEE80211_IOC_DTIM_PERIOD, atoi(val), 0, NULL); } static DECL_CMD_FUNC(set80211bintval, val, d) { set80211(s, IEEE80211_IOC_BEACON_INTERVAL, atoi(val), 0, NULL); } static void set80211macmac(int s, int op, const char *val) { char *temp; struct sockaddr_dl sdl; temp = malloc(strlen(val) + 2); /* ':' and '\0' */ if (temp == NULL) errx(1, "malloc failed"); temp[0] = ':'; strcpy(temp + 1, val); sdl.sdl_len = sizeof(sdl); link_addr(temp, &sdl); free(temp); if (sdl.sdl_alen != IEEE80211_ADDR_LEN) errx(1, "malformed link-level address"); set80211(s, op, 0, IEEE80211_ADDR_LEN, LLADDR(&sdl)); } static DECL_CMD_FUNC(set80211addmac, val, d) { set80211macmac(s, IEEE80211_IOC_ADDMAC, val); } static DECL_CMD_FUNC(set80211delmac, val, d) { set80211macmac(s, IEEE80211_IOC_DELMAC, val); } static DECL_CMD_FUNC(set80211kickmac, val, d) { char *temp; struct sockaddr_dl sdl; struct ieee80211req_mlme mlme; temp = malloc(strlen(val) + 2); /* ':' and '\0' */ if (temp == NULL) errx(1, "malloc failed"); temp[0] = ':'; strcpy(temp + 1, val); sdl.sdl_len = sizeof(sdl); link_addr(temp, &sdl); free(temp); if (sdl.sdl_alen != IEEE80211_ADDR_LEN) errx(1, "malformed link-level address"); memset(&mlme, 0, sizeof(mlme)); mlme.im_op = IEEE80211_MLME_DEAUTH; mlme.im_reason = IEEE80211_REASON_AUTH_EXPIRE; memcpy(mlme.im_macaddr, LLADDR(&sdl), IEEE80211_ADDR_LEN); set80211(s, IEEE80211_IOC_MLME, 0, sizeof(mlme), &mlme); } static DECL_CMD_FUNC(set80211maccmd, val, d) { set80211(s, IEEE80211_IOC_MACCMD, d, 0, NULL); } static void set80211meshrtmac(int s, int req, const char *val) { char *temp; struct sockaddr_dl sdl; temp = malloc(strlen(val) + 2); /* ':' and '\0' */ if (temp == NULL) errx(1, "malloc failed"); temp[0] = ':'; strcpy(temp + 1, val); sdl.sdl_len = sizeof(sdl); link_addr(temp, &sdl); free(temp); if (sdl.sdl_alen != IEEE80211_ADDR_LEN) errx(1, "malformed link-level address"); set80211(s, IEEE80211_IOC_MESH_RTCMD, req, IEEE80211_ADDR_LEN, LLADDR(&sdl)); } static DECL_CMD_FUNC(set80211addmeshrt, val, d) { set80211meshrtmac(s, IEEE80211_MESH_RTCMD_ADD, val); } static DECL_CMD_FUNC(set80211delmeshrt, val, d) { set80211meshrtmac(s, IEEE80211_MESH_RTCMD_DELETE, val); } static DECL_CMD_FUNC(set80211meshrtcmd, val, d) { set80211(s, IEEE80211_IOC_MESH_RTCMD, d, 0, NULL); } static DECL_CMD_FUNC(set80211hwmprootmode, val, d) { int mode; if (strcasecmp(val, "normal") == 0) mode = IEEE80211_HWMP_ROOTMODE_NORMAL; else if (strcasecmp(val, "proactive") == 0) mode = IEEE80211_HWMP_ROOTMODE_PROACTIVE; else if (strcasecmp(val, "rann") == 0) mode = IEEE80211_HWMP_ROOTMODE_RANN; else mode = IEEE80211_HWMP_ROOTMODE_DISABLED; set80211(s, IEEE80211_IOC_HWMP_ROOTMODE, mode, 0, NULL); } static DECL_CMD_FUNC(set80211hwmpmaxhops, val, d) { set80211(s, IEEE80211_IOC_HWMP_MAXHOPS, atoi(val), 0, NULL); } static void set80211pureg(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_PUREG, d, 0, NULL); } static void set80211quiet(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_QUIET, d, 0, NULL); } static DECL_CMD_FUNC(set80211quietperiod, val, d) { set80211(s, IEEE80211_IOC_QUIET_PERIOD, atoi(val), 0, NULL); } static DECL_CMD_FUNC(set80211quietcount, val, d) { set80211(s, IEEE80211_IOC_QUIET_COUNT, atoi(val), 0, NULL); } static DECL_CMD_FUNC(set80211quietduration, val, d) { set80211(s, IEEE80211_IOC_QUIET_DUR, atoi(val), 0, NULL); } static DECL_CMD_FUNC(set80211quietoffset, val, d) { set80211(s, IEEE80211_IOC_QUIET_OFFSET, atoi(val), 0, NULL); } static void set80211bgscan(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_BGSCAN, d, 0, NULL); } static DECL_CMD_FUNC(set80211bgscanidle, val, d) { set80211(s, IEEE80211_IOC_BGSCAN_IDLE, atoi(val), 0, NULL); } static DECL_CMD_FUNC(set80211bgscanintvl, val, d) { set80211(s, IEEE80211_IOC_BGSCAN_INTERVAL, atoi(val), 0, NULL); } static DECL_CMD_FUNC(set80211scanvalid, val, d) { set80211(s, IEEE80211_IOC_SCANVALID, atoi(val), 0, NULL); } /* * Parse an optional trailing specification of which netbands * to apply a parameter to. This is basically the same syntax * as used for channels but you can concatenate to specify * multiple. For example: * 14:abg apply to 11a, 11b, and 11g * 6:ht apply to 11na and 11ng * We don't make a big effort to catch silly things; this is * really a convenience mechanism. */ static int getmodeflags(const char *val) { const char *cp; int flags; flags = 0; cp = strchr(val, ':'); if (cp != NULL) { for (cp++; isalpha((int) *cp); cp++) { /* accept mixed case */ int c = *cp; if (isupper(c)) c = tolower(c); switch (c) { case 'a': /* 802.11a */ flags |= IEEE80211_CHAN_A; break; case 'b': /* 802.11b */ flags |= IEEE80211_CHAN_B; break; case 'g': /* 802.11g */ flags |= IEEE80211_CHAN_G; break; case 'n': /* 802.11n */ flags |= IEEE80211_CHAN_HT; break; case 'd': /* dt = Atheros Dynamic Turbo */ flags |= IEEE80211_CHAN_TURBO; break; case 't': /* ht, dt, st, t */ /* dt and unadorned t specify Dynamic Turbo */ if ((flags & (IEEE80211_CHAN_STURBO|IEEE80211_CHAN_HT)) == 0) flags |= IEEE80211_CHAN_TURBO; break; case 's': /* st = Atheros Static Turbo */ flags |= IEEE80211_CHAN_STURBO; break; case 'h': /* 1/2-width channels */ flags |= IEEE80211_CHAN_HALF; break; case 'q': /* 1/4-width channels */ flags |= IEEE80211_CHAN_QUARTER; break; case 'v': /* XXX set HT too? */ flags |= IEEE80211_CHAN_VHT; break; default: errx(-1, "%s: Invalid mode attribute %c\n", val, *cp); } } } return flags; } #define _APPLY(_flags, _base, _param, _v) do { \ if (_flags & IEEE80211_CHAN_HT) { \ if ((_flags & (IEEE80211_CHAN_5GHZ|IEEE80211_CHAN_2GHZ)) == 0) {\ _base.params[IEEE80211_MODE_11NA]._param = _v; \ _base.params[IEEE80211_MODE_11NG]._param = _v; \ } else if (_flags & IEEE80211_CHAN_5GHZ) \ _base.params[IEEE80211_MODE_11NA]._param = _v; \ else \ _base.params[IEEE80211_MODE_11NG]._param = _v; \ } \ if (_flags & IEEE80211_CHAN_TURBO) { \ if ((_flags & (IEEE80211_CHAN_5GHZ|IEEE80211_CHAN_2GHZ)) == 0) {\ _base.params[IEEE80211_MODE_TURBO_A]._param = _v; \ _base.params[IEEE80211_MODE_TURBO_G]._param = _v; \ } else if (_flags & IEEE80211_CHAN_5GHZ) \ _base.params[IEEE80211_MODE_TURBO_A]._param = _v; \ else \ _base.params[IEEE80211_MODE_TURBO_G]._param = _v; \ } \ if (_flags & IEEE80211_CHAN_STURBO) \ _base.params[IEEE80211_MODE_STURBO_A]._param = _v; \ if ((_flags & IEEE80211_CHAN_A) == IEEE80211_CHAN_A) \ _base.params[IEEE80211_MODE_11A]._param = _v; \ if ((_flags & IEEE80211_CHAN_G) == IEEE80211_CHAN_G) \ _base.params[IEEE80211_MODE_11G]._param = _v; \ if ((_flags & IEEE80211_CHAN_B) == IEEE80211_CHAN_B) \ _base.params[IEEE80211_MODE_11B]._param = _v; \ if (_flags & IEEE80211_CHAN_HALF) \ _base.params[IEEE80211_MODE_HALF]._param = _v; \ if (_flags & IEEE80211_CHAN_QUARTER) \ _base.params[IEEE80211_MODE_QUARTER]._param = _v; \ } while (0) #define _APPLY1(_flags, _base, _param, _v) do { \ if (_flags & IEEE80211_CHAN_HT) { \ if (_flags & IEEE80211_CHAN_5GHZ) \ _base.params[IEEE80211_MODE_11NA]._param = _v; \ else \ _base.params[IEEE80211_MODE_11NG]._param = _v; \ } else if ((_flags & IEEE80211_CHAN_108A) == IEEE80211_CHAN_108A) \ _base.params[IEEE80211_MODE_TURBO_A]._param = _v; \ else if ((_flags & IEEE80211_CHAN_108G) == IEEE80211_CHAN_108G) \ _base.params[IEEE80211_MODE_TURBO_G]._param = _v; \ else if ((_flags & IEEE80211_CHAN_ST) == IEEE80211_CHAN_ST) \ _base.params[IEEE80211_MODE_STURBO_A]._param = _v; \ else if (_flags & IEEE80211_CHAN_HALF) \ _base.params[IEEE80211_MODE_HALF]._param = _v; \ else if (_flags & IEEE80211_CHAN_QUARTER) \ _base.params[IEEE80211_MODE_QUARTER]._param = _v; \ else if ((_flags & IEEE80211_CHAN_A) == IEEE80211_CHAN_A) \ _base.params[IEEE80211_MODE_11A]._param = _v; \ else if ((_flags & IEEE80211_CHAN_G) == IEEE80211_CHAN_G) \ _base.params[IEEE80211_MODE_11G]._param = _v; \ else if ((_flags & IEEE80211_CHAN_B) == IEEE80211_CHAN_B) \ _base.params[IEEE80211_MODE_11B]._param = _v; \ } while (0) #define _APPLY_RATE(_flags, _base, _param, _v) do { \ if (_flags & IEEE80211_CHAN_HT) { \ (_v) = (_v / 2) | IEEE80211_RATE_MCS; \ } \ _APPLY(_flags, _base, _param, _v); \ } while (0) #define _APPLY_RATE1(_flags, _base, _param, _v) do { \ if (_flags & IEEE80211_CHAN_HT) { \ (_v) = (_v / 2) | IEEE80211_RATE_MCS; \ } \ _APPLY1(_flags, _base, _param, _v); \ } while (0) static DECL_CMD_FUNC(set80211roamrssi, val, d) { double v = atof(val); int rssi, flags; rssi = (int) (2*v); if (rssi != 2*v) errx(-1, "invalid rssi (must be .5 dBm units)"); flags = getmodeflags(val); getroam(s); if (flags == 0) { /* NB: no flags => current channel */ flags = getcurchan(s)->ic_flags; _APPLY1(flags, roamparams, rssi, rssi); } else _APPLY(flags, roamparams, rssi, rssi); callback_register(setroam_cb, &roamparams); } static int getrate(const char *val, const char *tag) { double v = atof(val); int rate; rate = (int) (2*v); if (rate != 2*v) errx(-1, "invalid %s rate (must be .5 Mb/s units)", tag); return rate; /* NB: returns 2x the specified value */ } static DECL_CMD_FUNC(set80211roamrate, val, d) { int rate, flags; rate = getrate(val, "roam"); flags = getmodeflags(val); getroam(s); if (flags == 0) { /* NB: no flags => current channel */ flags = getcurchan(s)->ic_flags; _APPLY_RATE1(flags, roamparams, rate, rate); } else _APPLY_RATE(flags, roamparams, rate, rate); callback_register(setroam_cb, &roamparams); } static DECL_CMD_FUNC(set80211mcastrate, val, d) { int rate, flags; rate = getrate(val, "mcast"); flags = getmodeflags(val); gettxparams(s); if (flags == 0) { /* NB: no flags => current channel */ flags = getcurchan(s)->ic_flags; _APPLY_RATE1(flags, txparams, mcastrate, rate); } else _APPLY_RATE(flags, txparams, mcastrate, rate); callback_register(settxparams_cb, &txparams); } static DECL_CMD_FUNC(set80211mgtrate, val, d) { int rate, flags; rate = getrate(val, "mgmt"); flags = getmodeflags(val); gettxparams(s); if (flags == 0) { /* NB: no flags => current channel */ flags = getcurchan(s)->ic_flags; _APPLY_RATE1(flags, txparams, mgmtrate, rate); } else _APPLY_RATE(flags, txparams, mgmtrate, rate); callback_register(settxparams_cb, &txparams); } static DECL_CMD_FUNC(set80211ucastrate, val, d) { int flags; gettxparams(s); flags = getmodeflags(val); if (isanyarg(val)) { if (flags == 0) { /* NB: no flags => current channel */ flags = getcurchan(s)->ic_flags; _APPLY1(flags, txparams, ucastrate, IEEE80211_FIXED_RATE_NONE); } else _APPLY(flags, txparams, ucastrate, IEEE80211_FIXED_RATE_NONE); } else { int rate = getrate(val, "ucast"); if (flags == 0) { /* NB: no flags => current channel */ flags = getcurchan(s)->ic_flags; _APPLY_RATE1(flags, txparams, ucastrate, rate); } else _APPLY_RATE(flags, txparams, ucastrate, rate); } callback_register(settxparams_cb, &txparams); } static DECL_CMD_FUNC(set80211maxretry, val, d) { int v = atoi(val), flags; flags = getmodeflags(val); gettxparams(s); if (flags == 0) { /* NB: no flags => current channel */ flags = getcurchan(s)->ic_flags; _APPLY1(flags, txparams, maxretry, v); } else _APPLY(flags, txparams, maxretry, v); callback_register(settxparams_cb, &txparams); } #undef _APPLY_RATE #undef _APPLY static DECL_CMD_FUNC(set80211fragthreshold, val, d) { set80211(s, IEEE80211_IOC_FRAGTHRESHOLD, isundefarg(val) ? IEEE80211_FRAG_MAX : atoi(val), 0, NULL); } static DECL_CMD_FUNC(set80211bmissthreshold, val, d) { set80211(s, IEEE80211_IOC_BMISSTHRESHOLD, isundefarg(val) ? IEEE80211_HWBMISS_MAX : atoi(val), 0, NULL); } static void set80211burst(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_BURST, d, 0, NULL); } static void set80211doth(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_DOTH, d, 0, NULL); } static void set80211dfs(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_DFS, d, 0, NULL); } static void set80211shortgi(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_SHORTGI, d ? (IEEE80211_HTCAP_SHORTGI20 | IEEE80211_HTCAP_SHORTGI40) : 0, 0, NULL); } /* XXX 11ac density/size is different */ static void set80211ampdu(const char *val, int d, int s, const struct afswtch *rafp) { int ampdu; if (get80211val(s, IEEE80211_IOC_AMPDU, &du) < 0) errx(-1, "cannot set AMPDU setting"); if (d < 0) { d = -d; ampdu &= ~d; } else ampdu |= d; set80211(s, IEEE80211_IOC_AMPDU, ampdu, 0, NULL); } static void set80211stbc(const char *val, int d, int s, const struct afswtch *rafp) { int stbc; if (get80211val(s, IEEE80211_IOC_STBC, &stbc) < 0) errx(-1, "cannot set STBC setting"); if (d < 0) { d = -d; stbc &= ~d; } else stbc |= d; set80211(s, IEEE80211_IOC_STBC, stbc, 0, NULL); } static void set80211ldpc(const char *val, int d, int s, const struct afswtch *rafp) { int ldpc; if (get80211val(s, IEEE80211_IOC_LDPC, &ldpc) < 0) errx(-1, "cannot set LDPC setting"); if (d < 0) { d = -d; ldpc &= ~d; } else ldpc |= d; set80211(s, IEEE80211_IOC_LDPC, ldpc, 0, NULL); } static void set80211uapsd(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_UAPSD, d, 0, NULL); } static DECL_CMD_FUNC(set80211ampdulimit, val, d) { int v; switch (atoi(val)) { case 8: case 8*1024: v = IEEE80211_HTCAP_MAXRXAMPDU_8K; break; case 16: case 16*1024: v = IEEE80211_HTCAP_MAXRXAMPDU_16K; break; case 32: case 32*1024: v = IEEE80211_HTCAP_MAXRXAMPDU_32K; break; case 64: case 64*1024: v = IEEE80211_HTCAP_MAXRXAMPDU_64K; break; default: errx(-1, "invalid A-MPDU limit %s", val); } set80211(s, IEEE80211_IOC_AMPDU_LIMIT, v, 0, NULL); } /* XXX 11ac density/size is different */ static DECL_CMD_FUNC(set80211ampdudensity, val, d) { int v; if (isanyarg(val) || strcasecmp(val, "na") == 0) v = IEEE80211_HTCAP_MPDUDENSITY_NA; else switch ((int)(atof(val)*4)) { case 0: v = IEEE80211_HTCAP_MPDUDENSITY_NA; break; case 1: v = IEEE80211_HTCAP_MPDUDENSITY_025; break; case 2: v = IEEE80211_HTCAP_MPDUDENSITY_05; break; case 4: v = IEEE80211_HTCAP_MPDUDENSITY_1; break; case 8: v = IEEE80211_HTCAP_MPDUDENSITY_2; break; case 16: v = IEEE80211_HTCAP_MPDUDENSITY_4; break; case 32: v = IEEE80211_HTCAP_MPDUDENSITY_8; break; case 64: v = IEEE80211_HTCAP_MPDUDENSITY_16; break; default: errx(-1, "invalid A-MPDU density %s", val); } set80211(s, IEEE80211_IOC_AMPDU_DENSITY, v, 0, NULL); } static void set80211amsdu(const char *val, int d, int s, const struct afswtch *rafp) { int amsdu; if (get80211val(s, IEEE80211_IOC_AMSDU, &amsdu) < 0) err(-1, "cannot get AMSDU setting"); if (d < 0) { d = -d; amsdu &= ~d; } else amsdu |= d; set80211(s, IEEE80211_IOC_AMSDU, amsdu, 0, NULL); } static DECL_CMD_FUNC(set80211amsdulimit, val, d) { set80211(s, IEEE80211_IOC_AMSDU_LIMIT, atoi(val), 0, NULL); } static void set80211puren(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_PUREN, d, 0, NULL); } static void set80211htcompat(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_HTCOMPAT, d, 0, NULL); } static void set80211htconf(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_HTCONF, d, 0, NULL); htconf = d; } static void set80211dwds(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_DWDS, d, 0, NULL); } static void set80211inact(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_INACTIVITY, d, 0, NULL); } static void set80211tsn(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_TSN, d, 0, NULL); } static void set80211dotd(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_DOTD, d, 0, NULL); } static void set80211smps(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_SMPS, d, 0, NULL); } static void set80211rifs(const char *val, int d, int s, const struct afswtch *rafp) { set80211(s, IEEE80211_IOC_RIFS, d, 0, NULL); } static void set80211vhtconf(const char *val, int d, int s, const struct afswtch *rafp) { if (get80211val(s, IEEE80211_IOC_VHTCONF, &vhtconf) < 0) errx(-1, "cannot set VHT setting"); printf("%s: vhtconf=0x%08x, d=%d\n", __func__, vhtconf, d); if (d < 0) { d = -d; vhtconf &= ~d; } else vhtconf |= d; printf("%s: vhtconf is now 0x%08x\n", __func__, vhtconf); set80211(s, IEEE80211_IOC_VHTCONF, vhtconf, 0, NULL); } static DECL_CMD_FUNC(set80211tdmaslot, val, d) { set80211(s, IEEE80211_IOC_TDMA_SLOT, atoi(val), 0, NULL); } static DECL_CMD_FUNC(set80211tdmaslotcnt, val, d) { set80211(s, IEEE80211_IOC_TDMA_SLOTCNT, atoi(val), 0, NULL); } static DECL_CMD_FUNC(set80211tdmaslotlen, val, d) { set80211(s, IEEE80211_IOC_TDMA_SLOTLEN, atoi(val), 0, NULL); } static DECL_CMD_FUNC(set80211tdmabintval, val, d) { set80211(s, IEEE80211_IOC_TDMA_BINTERVAL, atoi(val), 0, NULL); } static DECL_CMD_FUNC(set80211meshttl, val, d) { set80211(s, IEEE80211_IOC_MESH_TTL, atoi(val), 0, NULL); } static DECL_CMD_FUNC(set80211meshforward, val, d) { set80211(s, IEEE80211_IOC_MESH_FWRD, d, 0, NULL); } static DECL_CMD_FUNC(set80211meshgate, val, d) { set80211(s, IEEE80211_IOC_MESH_GATE, d, 0, NULL); } static DECL_CMD_FUNC(set80211meshpeering, val, d) { set80211(s, IEEE80211_IOC_MESH_AP, d, 0, NULL); } static DECL_CMD_FUNC(set80211meshmetric, val, d) { char v[12]; memcpy(v, val, sizeof(v)); set80211(s, IEEE80211_IOC_MESH_PR_METRIC, 0, 0, v); } static DECL_CMD_FUNC(set80211meshpath, val, d) { char v[12]; memcpy(v, val, sizeof(v)); set80211(s, IEEE80211_IOC_MESH_PR_PATH, 0, 0, v); } static int regdomain_sort(const void *a, const void *b) { #define CHAN_ALL \ (IEEE80211_CHAN_ALLTURBO|IEEE80211_CHAN_HALF|IEEE80211_CHAN_QUARTER) const struct ieee80211_channel *ca = a; const struct ieee80211_channel *cb = b; return ca->ic_freq == cb->ic_freq ? (ca->ic_flags & CHAN_ALL) - (cb->ic_flags & CHAN_ALL) : ca->ic_freq - cb->ic_freq; #undef CHAN_ALL } static const struct ieee80211_channel * chanlookup(const struct ieee80211_channel chans[], int nchans, int freq, int flags) { int i; flags &= IEEE80211_CHAN_ALLTURBO; for (i = 0; i < nchans; i++) { const struct ieee80211_channel *c = &chans[i]; if (c->ic_freq == freq && (c->ic_flags & IEEE80211_CHAN_ALLTURBO) == flags) return c; } return NULL; } static int chanfind(const struct ieee80211_channel chans[], int nchans, int flags) { int i; for (i = 0; i < nchans; i++) { const struct ieee80211_channel *c = &chans[i]; if ((c->ic_flags & flags) == flags) return 1; } return 0; } /* * Check channel compatibility. */ static int checkchan(const struct ieee80211req_chaninfo *avail, int freq, int flags) { flags &= ~REQ_FLAGS; /* * Check if exact channel is in the calibration table; * everything below is to deal with channels that we * want to include but that are not explicitly listed. */ if (chanlookup(avail->ic_chans, avail->ic_nchans, freq, flags) != NULL) return 1; if (flags & IEEE80211_CHAN_GSM) { /* * XXX GSM frequency mapping is handled in the kernel * so we cannot find them in the calibration table; * just accept the channel and the kernel will reject * the channel list if it's wrong. */ return 1; } /* * If this is a 1/2 or 1/4 width channel allow it if a full * width channel is present for this frequency, and the device * supports fractional channels on this band. This is a hack * that avoids bloating the calibration table; it may be better * by per-band attributes though (we are effectively calculating * this attribute by scanning the channel list ourself). */ if ((flags & (IEEE80211_CHAN_HALF | IEEE80211_CHAN_QUARTER)) == 0) return 0; if (chanlookup(avail->ic_chans, avail->ic_nchans, freq, flags &~ (IEEE80211_CHAN_HALF | IEEE80211_CHAN_QUARTER)) == NULL) return 0; if (flags & IEEE80211_CHAN_HALF) { return chanfind(avail->ic_chans, avail->ic_nchans, IEEE80211_CHAN_HALF | (flags & (IEEE80211_CHAN_2GHZ | IEEE80211_CHAN_5GHZ))); } else { return chanfind(avail->ic_chans, avail->ic_nchans, IEEE80211_CHAN_QUARTER | (flags & (IEEE80211_CHAN_2GHZ | IEEE80211_CHAN_5GHZ))); } } static void regdomain_addchans(struct ieee80211req_chaninfo *ci, const netband_head *bands, const struct ieee80211_regdomain *reg, uint32_t chanFlags, const struct ieee80211req_chaninfo *avail) { const struct netband *nb; const struct freqband *b; struct ieee80211_channel *c, *prev; int freq, hi_adj, lo_adj, channelSep; uint32_t flags; hi_adj = (chanFlags & IEEE80211_CHAN_HT40U) ? -20 : 0; lo_adj = (chanFlags & IEEE80211_CHAN_HT40D) ? 20 : 0; channelSep = (chanFlags & IEEE80211_CHAN_2GHZ) ? 0 : 40; LIST_FOREACH(nb, bands, next) { b = nb->band; if (verbose) { printf("%s:", __func__); printb(" chanFlags", chanFlags, IEEE80211_CHAN_BITS); printb(" bandFlags", nb->flags | b->flags, IEEE80211_CHAN_BITS); putchar('\n'); } prev = NULL; for (freq = b->freqStart + lo_adj; freq <= b->freqEnd + hi_adj; freq += b->chanSep) { /* * Construct flags for the new channel. We take * the attributes from the band descriptions except * for HT40 which is enabled generically (i.e. +/- * extension channel) in the band description and * then constrained according by channel separation. */ flags = nb->flags | b->flags; /* * VHT first - HT is a subset. * * XXX TODO: VHT80p80, VHT160 is not yet done. */ if (flags & IEEE80211_CHAN_VHT) { if ((chanFlags & IEEE80211_CHAN_VHT20) && (flags & IEEE80211_CHAN_VHT20) == 0) { if (verbose) printf("%u: skip, not a " "VHT20 channel\n", freq); continue; } if ((chanFlags & IEEE80211_CHAN_VHT40) && (flags & IEEE80211_CHAN_VHT40) == 0) { if (verbose) printf("%u: skip, not a " "VHT40 channel\n", freq); continue; } if ((chanFlags & IEEE80211_CHAN_VHT80) && (flags & IEEE80211_CHAN_VHT80) == 0) { if (verbose) printf("%u: skip, not a " "VHT80 channel\n", freq); continue; } flags &= ~IEEE80211_CHAN_VHT; flags |= chanFlags & IEEE80211_CHAN_VHT; } /* Now, constrain HT */ if (flags & IEEE80211_CHAN_HT) { /* * HT channels are generated specially; we're * called to add HT20, HT40+, and HT40- chan's * so we need to expand only band specs for * the HT channel type being added. */ if ((chanFlags & IEEE80211_CHAN_HT20) && (flags & IEEE80211_CHAN_HT20) == 0) { if (verbose) printf("%u: skip, not an " "HT20 channel\n", freq); continue; } if ((chanFlags & IEEE80211_CHAN_HT40) && (flags & IEEE80211_CHAN_HT40) == 0) { if (verbose) printf("%u: skip, not an " "HT40 channel\n", freq); continue; } /* NB: HT attribute comes from caller */ flags &= ~IEEE80211_CHAN_HT; flags |= chanFlags & IEEE80211_CHAN_HT; } /* * Check if device can operate on this frequency. */ if (!checkchan(avail, freq, flags)) { if (verbose) { printf("%u: skip, ", freq); printb("flags", flags, IEEE80211_CHAN_BITS); printf(" not available\n"); } continue; } if ((flags & REQ_ECM) && !reg->ecm) { if (verbose) printf("%u: skip, ECM channel\n", freq); continue; } if ((flags & REQ_INDOOR) && reg->location == 'O') { if (verbose) printf("%u: skip, indoor channel\n", freq); continue; } if ((flags & REQ_OUTDOOR) && reg->location == 'I') { if (verbose) printf("%u: skip, outdoor channel\n", freq); continue; } if ((flags & IEEE80211_CHAN_HT40) && prev != NULL && (freq - prev->ic_freq) < channelSep) { if (verbose) printf("%u: skip, only %u channel " "separation, need %d\n", freq, freq - prev->ic_freq, channelSep); continue; } if (ci->ic_nchans == IEEE80211_CHAN_MAX) { if (verbose) printf("%u: skip, channel table full\n", freq); break; } c = &ci->ic_chans[ci->ic_nchans++]; memset(c, 0, sizeof(*c)); c->ic_freq = freq; c->ic_flags = flags; if (c->ic_flags & IEEE80211_CHAN_DFS) c->ic_maxregpower = nb->maxPowerDFS; else c->ic_maxregpower = nb->maxPower; if (verbose) { printf("[%3d] add freq %u ", ci->ic_nchans-1, c->ic_freq); printb("flags", c->ic_flags, IEEE80211_CHAN_BITS); printf(" power %u\n", c->ic_maxregpower); } /* NB: kernel fills in other fields */ prev = c; } } } static void regdomain_makechannels( struct ieee80211_regdomain_req *req, const struct ieee80211_devcaps_req *dc) { struct regdata *rdp = getregdata(); const struct country *cc; const struct ieee80211_regdomain *reg = &req->rd; struct ieee80211req_chaninfo *ci = &req->chaninfo; const struct regdomain *rd; /* * Locate construction table for new channel list. We treat * the regdomain/SKU as definitive so a country can be in * multiple with different properties (e.g. US in FCC+FCC3). * If no regdomain is specified then we fallback on the country * code to find the associated regdomain since countries always * belong to at least one regdomain. */ if (reg->regdomain == 0) { cc = lib80211_country_findbycc(rdp, reg->country); if (cc == NULL) errx(1, "internal error, country %d not found", reg->country); rd = cc->rd; } else rd = lib80211_regdomain_findbysku(rdp, reg->regdomain); if (rd == NULL) errx(1, "internal error, regdomain %d not found", reg->regdomain); if (rd->sku != SKU_DEBUG) { /* * regdomain_addchans incrememnts the channel count for * each channel it adds so initialize ic_nchans to zero. * Note that we know we have enough space to hold all possible * channels because the devcaps list size was used to * allocate our request. */ ci->ic_nchans = 0; if (!LIST_EMPTY(&rd->bands_11b)) regdomain_addchans(ci, &rd->bands_11b, reg, IEEE80211_CHAN_B, &dc->dc_chaninfo); if (!LIST_EMPTY(&rd->bands_11g)) regdomain_addchans(ci, &rd->bands_11g, reg, IEEE80211_CHAN_G, &dc->dc_chaninfo); if (!LIST_EMPTY(&rd->bands_11a)) regdomain_addchans(ci, &rd->bands_11a, reg, IEEE80211_CHAN_A, &dc->dc_chaninfo); if (!LIST_EMPTY(&rd->bands_11na) && dc->dc_htcaps != 0) { regdomain_addchans(ci, &rd->bands_11na, reg, IEEE80211_CHAN_A | IEEE80211_CHAN_HT20, &dc->dc_chaninfo); if (dc->dc_htcaps & IEEE80211_HTCAP_CHWIDTH40) { regdomain_addchans(ci, &rd->bands_11na, reg, IEEE80211_CHAN_A | IEEE80211_CHAN_HT40U, &dc->dc_chaninfo); regdomain_addchans(ci, &rd->bands_11na, reg, IEEE80211_CHAN_A | IEEE80211_CHAN_HT40D, &dc->dc_chaninfo); } } if (!LIST_EMPTY(&rd->bands_11ac) && dc->dc_vhtcaps != 0) { regdomain_addchans(ci, &rd->bands_11ac, reg, IEEE80211_CHAN_A | IEEE80211_CHAN_HT20 | IEEE80211_CHAN_VHT20, &dc->dc_chaninfo); /* VHT40 is a function of HT40.. */ if (dc->dc_htcaps & IEEE80211_HTCAP_CHWIDTH40) { regdomain_addchans(ci, &rd->bands_11ac, reg, IEEE80211_CHAN_A | IEEE80211_CHAN_HT40U | IEEE80211_CHAN_VHT40U, &dc->dc_chaninfo); regdomain_addchans(ci, &rd->bands_11ac, reg, IEEE80211_CHAN_A | IEEE80211_CHAN_HT40D | IEEE80211_CHAN_VHT40D, &dc->dc_chaninfo); } /* VHT80 */ /* XXX dc_vhtcap? */ if (1) { regdomain_addchans(ci, &rd->bands_11ac, reg, IEEE80211_CHAN_A | IEEE80211_CHAN_HT40U | IEEE80211_CHAN_VHT80, &dc->dc_chaninfo); regdomain_addchans(ci, &rd->bands_11ac, reg, IEEE80211_CHAN_A | IEEE80211_CHAN_HT40D | IEEE80211_CHAN_VHT80, &dc->dc_chaninfo); } /* XXX TODO: VHT80_80, VHT160 */ } if (!LIST_EMPTY(&rd->bands_11ng) && dc->dc_htcaps != 0) { regdomain_addchans(ci, &rd->bands_11ng, reg, IEEE80211_CHAN_G | IEEE80211_CHAN_HT20, &dc->dc_chaninfo); if (dc->dc_htcaps & IEEE80211_HTCAP_CHWIDTH40) { regdomain_addchans(ci, &rd->bands_11ng, reg, IEEE80211_CHAN_G | IEEE80211_CHAN_HT40U, &dc->dc_chaninfo); regdomain_addchans(ci, &rd->bands_11ng, reg, IEEE80211_CHAN_G | IEEE80211_CHAN_HT40D, &dc->dc_chaninfo); } } qsort(ci->ic_chans, ci->ic_nchans, sizeof(ci->ic_chans[0]), regdomain_sort); } else memcpy(ci, &dc->dc_chaninfo, IEEE80211_CHANINFO_SPACE(&dc->dc_chaninfo)); } static void list_countries(void) { struct regdata *rdp = getregdata(); const struct country *cp; const struct regdomain *dp; int i; i = 0; printf("\nCountry codes:\n"); LIST_FOREACH(cp, &rdp->countries, next) { printf("%2s %-15.15s%s", cp->isoname, cp->name, ((i+1)%4) == 0 ? "\n" : " "); i++; } i = 0; printf("\nRegulatory domains:\n"); LIST_FOREACH(dp, &rdp->domains, next) { printf("%-15.15s%s", dp->name, ((i+1)%4) == 0 ? "\n" : " "); i++; } printf("\n"); } static void defaultcountry(const struct regdomain *rd) { struct regdata *rdp = getregdata(); const struct country *cc; cc = lib80211_country_findbycc(rdp, rd->cc->code); if (cc == NULL) errx(1, "internal error, ISO country code %d not " "defined for regdomain %s", rd->cc->code, rd->name); regdomain.country = cc->code; regdomain.isocc[0] = cc->isoname[0]; regdomain.isocc[1] = cc->isoname[1]; } static DECL_CMD_FUNC(set80211regdomain, val, d) { struct regdata *rdp = getregdata(); const struct regdomain *rd; rd = lib80211_regdomain_findbyname(rdp, val); if (rd == NULL) { char *eptr; long sku = strtol(val, &eptr, 0); if (eptr != val) rd = lib80211_regdomain_findbysku(rdp, sku); if (eptr == val || rd == NULL) errx(1, "unknown regdomain %s", val); } getregdomain(s); regdomain.regdomain = rd->sku; if (regdomain.country == 0 && rd->cc != NULL) { /* * No country code setup and there's a default * one for this regdomain fill it in. */ defaultcountry(rd); } callback_register(setregdomain_cb, ®domain); } static DECL_CMD_FUNC(set80211country, val, d) { struct regdata *rdp = getregdata(); const struct country *cc; cc = lib80211_country_findbyname(rdp, val); if (cc == NULL) { char *eptr; long code = strtol(val, &eptr, 0); if (eptr != val) cc = lib80211_country_findbycc(rdp, code); if (eptr == val || cc == NULL) errx(1, "unknown ISO country code %s", val); } getregdomain(s); regdomain.regdomain = cc->rd->sku; regdomain.country = cc->code; regdomain.isocc[0] = cc->isoname[0]; regdomain.isocc[1] = cc->isoname[1]; callback_register(setregdomain_cb, ®domain); } static void set80211location(const char *val, int d, int s, const struct afswtch *rafp) { getregdomain(s); regdomain.location = d; callback_register(setregdomain_cb, ®domain); } static void set80211ecm(const char *val, int d, int s, const struct afswtch *rafp) { getregdomain(s); regdomain.ecm = d; callback_register(setregdomain_cb, ®domain); } static void LINE_INIT(char c) { spacer = c; if (c == '\t') col = 8; else col = 1; } static void LINE_BREAK(void) { if (spacer != '\t') { printf("\n"); spacer = '\t'; } col = 8; /* 8-col tab */ } static void LINE_CHECK(const char *fmt, ...) { char buf[80]; va_list ap; int n; va_start(ap, fmt); n = vsnprintf(buf+1, sizeof(buf)-1, fmt, ap); va_end(ap); col += 1+n; if (col > MAXCOL) { LINE_BREAK(); col += n; } buf[0] = spacer; printf("%s", buf); spacer = ' '; } static int getmaxrate(const uint8_t rates[15], uint8_t nrates) { int i, maxrate = -1; for (i = 0; i < nrates; i++) { int rate = rates[i] & IEEE80211_RATE_VAL; if (rate > maxrate) maxrate = rate; } return maxrate / 2; } static const char * getcaps(int capinfo) { static char capstring[32]; char *cp = capstring; if (capinfo & IEEE80211_CAPINFO_ESS) *cp++ = 'E'; if (capinfo & IEEE80211_CAPINFO_IBSS) *cp++ = 'I'; if (capinfo & IEEE80211_CAPINFO_CF_POLLABLE) *cp++ = 'c'; if (capinfo & IEEE80211_CAPINFO_CF_POLLREQ) *cp++ = 'C'; if (capinfo & IEEE80211_CAPINFO_PRIVACY) *cp++ = 'P'; if (capinfo & IEEE80211_CAPINFO_SHORT_PREAMBLE) *cp++ = 'S'; if (capinfo & IEEE80211_CAPINFO_PBCC) *cp++ = 'B'; if (capinfo & IEEE80211_CAPINFO_CHNL_AGILITY) *cp++ = 'A'; if (capinfo & IEEE80211_CAPINFO_SHORT_SLOTTIME) *cp++ = 's'; if (capinfo & IEEE80211_CAPINFO_RSN) *cp++ = 'R'; if (capinfo & IEEE80211_CAPINFO_DSSSOFDM) *cp++ = 'D'; *cp = '\0'; return capstring; } static const char * getflags(int flags) { static char flagstring[32]; char *cp = flagstring; if (flags & IEEE80211_NODE_AUTH) *cp++ = 'A'; if (flags & IEEE80211_NODE_QOS) *cp++ = 'Q'; if (flags & IEEE80211_NODE_ERP) *cp++ = 'E'; if (flags & IEEE80211_NODE_PWR_MGT) *cp++ = 'P'; if (flags & IEEE80211_NODE_HT) { *cp++ = 'H'; if (flags & IEEE80211_NODE_HTCOMPAT) *cp++ = '+'; } if (flags & IEEE80211_NODE_VHT) *cp++ = 'V'; if (flags & IEEE80211_NODE_WPS) *cp++ = 'W'; if (flags & IEEE80211_NODE_TSN) *cp++ = 'N'; if (flags & IEEE80211_NODE_AMPDU_TX) *cp++ = 'T'; if (flags & IEEE80211_NODE_AMPDU_RX) *cp++ = 'R'; if (flags & IEEE80211_NODE_MIMO_PS) { *cp++ = 'M'; if (flags & IEEE80211_NODE_MIMO_RTS) *cp++ = '+'; } if (flags & IEEE80211_NODE_RIFS) *cp++ = 'I'; if (flags & IEEE80211_NODE_SGI40) { *cp++ = 'S'; if (flags & IEEE80211_NODE_SGI20) *cp++ = '+'; } else if (flags & IEEE80211_NODE_SGI20) *cp++ = 's'; if (flags & IEEE80211_NODE_AMSDU_TX) *cp++ = 't'; if (flags & IEEE80211_NODE_AMSDU_RX) *cp++ = 'r'; if (flags & IEEE80211_NODE_UAPSD) *cp++ = 'U'; if (flags & IEEE80211_NODE_LDPC) *cp++ = 'L'; *cp = '\0'; return flagstring; } static void printie(const char* tag, const uint8_t *ie, size_t ielen, int maxlen) { printf("%s", tag); if (verbose) { maxlen -= strlen(tag)+2; if (2*ielen > maxlen) maxlen--; printf("<"); for (; ielen > 0; ie++, ielen--) { if (maxlen-- <= 0) break; printf("%02x", *ie); } if (ielen != 0) printf("-"); printf(">"); } } #define LE_READ_2(p) \ ((u_int16_t) \ ((((const u_int8_t *)(p))[0] ) | \ (((const u_int8_t *)(p))[1] << 8))) #define LE_READ_4(p) \ ((u_int32_t) \ ((((const u_int8_t *)(p))[0] ) | \ (((const u_int8_t *)(p))[1] << 8) | \ (((const u_int8_t *)(p))[2] << 16) | \ (((const u_int8_t *)(p))[3] << 24))) /* * NB: The decoding routines assume a properly formatted ie * which should be safe as the kernel only retains them * if they parse ok. */ static void printwmeparam(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { #define MS(_v, _f) (((_v) & _f) >> _f##_S) static const char *acnames[] = { "BE", "BK", "VO", "VI" }; const struct ieee80211_wme_param *wme = (const struct ieee80211_wme_param *) ie; int i; printf("%s", tag); if (!verbose) return; printf("param_qosInfo); ie += offsetof(struct ieee80211_wme_param, params_acParams); for (i = 0; i < WME_NUM_AC; i++) { const struct ieee80211_wme_acparams *ac = &wme->params_acParams[i]; printf(" %s[%saifsn %u cwmin %u cwmax %u txop %u]" , acnames[i] , MS(ac->acp_aci_aifsn, WME_PARAM_ACM) ? "acm " : "" , MS(ac->acp_aci_aifsn, WME_PARAM_AIFSN) , MS(ac->acp_logcwminmax, WME_PARAM_LOGCWMIN) , MS(ac->acp_logcwminmax, WME_PARAM_LOGCWMAX) , LE_READ_2(&ac->acp_txop) ); } printf(">"); #undef MS } static void printwmeinfo(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { printf("%s", tag); if (verbose) { const struct ieee80211_wme_info *wme = (const struct ieee80211_wme_info *) ie; printf("", wme->wme_version, wme->wme_info); } } static void printvhtcap(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { printf("%s", tag); if (verbose) { const struct ieee80211_ie_vhtcap *vhtcap = (const struct ieee80211_ie_vhtcap *) ie; uint32_t vhtcap_info = LE_READ_4(&vhtcap->vht_cap_info); printf("supp_mcs.rx_mcs_map)); printf(" rx_highest %d", LE_READ_2(&vhtcap->supp_mcs.rx_highest) & 0x1fff); printf(" tx_mcs_map 0x%x", LE_READ_2(&vhtcap->supp_mcs.tx_mcs_map)); printf(" tx_highest %d", LE_READ_2(&vhtcap->supp_mcs.tx_highest) & 0x1fff); printf(">"); } } static void printvhtinfo(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { printf("%s", tag); if (verbose) { const struct ieee80211_ie_vht_operation *vhtinfo = (const struct ieee80211_ie_vht_operation *) ie; printf("", vhtinfo->chan_width, vhtinfo->center_freq_seg1_idx, vhtinfo->center_freq_seg2_idx, LE_READ_2(&vhtinfo->basic_mcs_set)); } } static void printvhtpwrenv(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { printf("%s", tag); static const char *txpwrmap[] = { "20", "40", "80", "160", }; if (verbose) { const struct ieee80211_ie_vht_txpwrenv *vhtpwr = (const struct ieee80211_ie_vht_txpwrenv *) ie; int i, n; const char *sep = ""; /* Get count; trim at ielen */ n = (vhtpwr->tx_info & IEEE80211_VHT_TXPWRENV_INFO_COUNT_MASK) + 1; /* Trim at ielen */ if (n > ielen - 3) n = ielen - 3; printf("tx_info); for (i = 0; i < n; i++) { printf("%s%s:%.2f", sep, txpwrmap[i], ((float) ((int8_t) ie[i+3])) / 2.0); sep = " "; } printf("]>"); } } static void printhtcap(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { printf("%s", tag); if (verbose) { const struct ieee80211_ie_htcap *htcap = (const struct ieee80211_ie_htcap *) ie; const char *sep; int i, j; printf("hc_cap), htcap->hc_param); printf(" mcsset["); sep = ""; for (i = 0; i < IEEE80211_HTRATE_MAXSIZE; i++) if (isset(htcap->hc_mcsset, i)) { for (j = i+1; j < IEEE80211_HTRATE_MAXSIZE; j++) if (isclr(htcap->hc_mcsset, j)) break; j--; if (i == j) printf("%s%u", sep, i); else printf("%s%u-%u", sep, i, j); i += j-i; sep = ","; } printf("] extcap 0x%x txbf 0x%x antenna 0x%x>", LE_READ_2(&htcap->hc_extcap), LE_READ_4(&htcap->hc_txbf), htcap->hc_antenna); } } static void printhtinfo(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { printf("%s", tag); if (verbose) { const struct ieee80211_ie_htinfo *htinfo = (const struct ieee80211_ie_htinfo *) ie; const char *sep; int i, j; printf("hi_ctrlchannel, htinfo->hi_byte1, htinfo->hi_byte2, htinfo->hi_byte3, LE_READ_2(&htinfo->hi_byte45)); printf(" basicmcs["); sep = ""; for (i = 0; i < IEEE80211_HTRATE_MAXSIZE; i++) if (isset(htinfo->hi_basicmcsset, i)) { for (j = i+1; j < IEEE80211_HTRATE_MAXSIZE; j++) if (isclr(htinfo->hi_basicmcsset, j)) break; j--; if (i == j) printf("%s%u", sep, i); else printf("%s%u-%u", sep, i, j); i += j-i; sep = ","; } printf("]>"); } } static void printathie(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { printf("%s", tag); if (verbose) { const struct ieee80211_ath_ie *ath = (const struct ieee80211_ath_ie *)ie; printf("<"); if (ath->ath_capability & ATHEROS_CAP_TURBO_PRIME) printf("DTURBO,"); if (ath->ath_capability & ATHEROS_CAP_COMPRESSION) printf("COMP,"); if (ath->ath_capability & ATHEROS_CAP_FAST_FRAME) printf("FF,"); if (ath->ath_capability & ATHEROS_CAP_XR) printf("XR,"); if (ath->ath_capability & ATHEROS_CAP_AR) printf("AR,"); if (ath->ath_capability & ATHEROS_CAP_BURST) printf("BURST,"); if (ath->ath_capability & ATHEROS_CAP_WME) printf("WME,"); if (ath->ath_capability & ATHEROS_CAP_BOOST) printf("BOOST,"); printf("0x%x>", LE_READ_2(ath->ath_defkeyix)); } } static void printmeshconf(const char *tag, const uint8_t *ie, size_t ielen, int maxlen) { printf("%s", tag); if (verbose) { const struct ieee80211_meshconf_ie *mconf = (const struct ieee80211_meshconf_ie *)ie; printf("conf_pselid == IEEE80211_MESHCONF_PATH_HWMP) printf("HWMP"); else printf("UNKNOWN"); printf(" LINK:"); if (mconf->conf_pmetid == IEEE80211_MESHCONF_METRIC_AIRTIME) printf("AIRTIME"); else printf("UNKNOWN"); printf(" CONGESTION:"); if (mconf->conf_ccid == IEEE80211_MESHCONF_CC_DISABLED) printf("DISABLED"); else printf("UNKNOWN"); printf(" SYNC:"); if (mconf->conf_syncid == IEEE80211_MESHCONF_SYNC_NEIGHOFF) printf("NEIGHOFF"); else printf("UNKNOWN"); printf(" AUTH:"); if (mconf->conf_authid == IEEE80211_MESHCONF_AUTH_DISABLED) printf("DISABLED"); else printf("UNKNOWN"); printf(" FORM:0x%x CAPS:0x%x>", mconf->conf_form, mconf->conf_cap); } } static void printbssload(const char *tag, const uint8_t *ie, size_t ielen, int maxlen) { printf("%s", tag); if (verbose) { const struct ieee80211_bss_load_ie *bssload = (const struct ieee80211_bss_load_ie *) ie; printf("", LE_READ_2(&bssload->sta_count), bssload->chan_load, bssload->aac); } } static void printapchanrep(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { printf("%s", tag); if (verbose) { const struct ieee80211_ap_chan_report_ie *ap = (const struct ieee80211_ap_chan_report_ie *) ie; const char *sep = ""; int i; printf("i_class); for (i = 3; i < ielen; i++) { printf("%s%u", sep, ie[i]); sep = ","; } printf("]>"); } } static const char * wpa_cipher(const u_int8_t *sel) { #define WPA_SEL(x) (((x)<<24)|WPA_OUI) u_int32_t w = LE_READ_4(sel); switch (w) { case WPA_SEL(WPA_CSE_NULL): return "NONE"; case WPA_SEL(WPA_CSE_WEP40): return "WEP40"; case WPA_SEL(WPA_CSE_WEP104): return "WEP104"; case WPA_SEL(WPA_CSE_TKIP): return "TKIP"; case WPA_SEL(WPA_CSE_CCMP): return "AES-CCMP"; } return "?"; /* NB: so 1<< is discarded */ #undef WPA_SEL } static const char * wpa_keymgmt(const u_int8_t *sel) { #define WPA_SEL(x) (((x)<<24)|WPA_OUI) u_int32_t w = LE_READ_4(sel); switch (w) { case WPA_SEL(WPA_ASE_8021X_UNSPEC): return "8021X-UNSPEC"; case WPA_SEL(WPA_ASE_8021X_PSK): return "8021X-PSK"; case WPA_SEL(WPA_ASE_NONE): return "NONE"; } return "?"; #undef WPA_SEL } static void printwpaie(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { u_int8_t len = ie[1]; printf("%s", tag); if (verbose) { const char *sep; int n; ie += 6, len -= 4; /* NB: len is payload only */ printf(" 0; n--) { printf("%s%s", sep, wpa_cipher(ie)); ie += 4, len -= 4; sep = "+"; } /* key management algorithms */ n = LE_READ_2(ie); ie += 2, len -= 2; sep = " km:"; for (; n > 0; n--) { printf("%s%s", sep, wpa_keymgmt(ie)); ie += 4, len -= 4; sep = "+"; } if (len > 2) /* optional capabilities */ printf(", caps 0x%x", LE_READ_2(ie)); printf(">"); } } static const char * rsn_cipher(const u_int8_t *sel) { #define RSN_SEL(x) (((x)<<24)|RSN_OUI) u_int32_t w = LE_READ_4(sel); switch (w) { case RSN_SEL(RSN_CSE_NULL): return "NONE"; case RSN_SEL(RSN_CSE_WEP40): return "WEP40"; case RSN_SEL(RSN_CSE_WEP104): return "WEP104"; case RSN_SEL(RSN_CSE_TKIP): return "TKIP"; case RSN_SEL(RSN_CSE_CCMP): return "AES-CCMP"; case RSN_SEL(RSN_CSE_WRAP): return "AES-OCB"; } return "?"; #undef WPA_SEL } static const char * rsn_keymgmt(const u_int8_t *sel) { #define RSN_SEL(x) (((x)<<24)|RSN_OUI) u_int32_t w = LE_READ_4(sel); switch (w) { case RSN_SEL(RSN_ASE_8021X_UNSPEC): return "8021X-UNSPEC"; case RSN_SEL(RSN_ASE_8021X_PSK): return "8021X-PSK"; case RSN_SEL(RSN_ASE_NONE): return "NONE"; } return "?"; #undef RSN_SEL } static void printrsnie(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { printf("%s", tag); if (verbose) { const char *sep; int n; ie += 2, ielen -= 2; printf(" 0; n--) { printf("%s%s", sep, rsn_cipher(ie)); ie += 4, ielen -= 4; sep = "+"; } /* key management algorithms */ n = LE_READ_2(ie); ie += 2, ielen -= 2; sep = " km:"; for (; n > 0; n--) { printf("%s%s", sep, rsn_keymgmt(ie)); ie += 4, ielen -= 4; sep = "+"; } if (ielen > 2) /* optional capabilities */ printf(", caps 0x%x", LE_READ_2(ie)); /* XXXPMKID */ printf(">"); } } #define BE_READ_2(p) \ ((u_int16_t) \ ((((const u_int8_t *)(p))[1] ) | \ (((const u_int8_t *)(p))[0] << 8))) static void printwpsie(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { u_int8_t len = ie[1]; printf("%s", tag); if (verbose) { static const char *dev_pass_id[] = { "D", /* Default (PIN) */ "U", /* User-specified */ "M", /* Machine-specified */ "K", /* Rekey */ "P", /* PushButton */ "R" /* Registrar-specified */ }; int n; int f; ie +=6, len -= 4; /* NB: len is payload only */ /* WPS IE in Beacon and Probe Resp frames have different fields */ printf("<"); while (len) { uint16_t tlv_type = BE_READ_2(ie); uint16_t tlv_len = BE_READ_2(ie + 2); uint16_t cfg_mthd; /* some devices broadcast invalid WPS frames */ if (tlv_len > len) { printf("bad frame length tlv_type=0x%02x " "tlv_len=%d len=%d", tlv_type, tlv_len, len); break; } ie += 4, len -= 4; switch (tlv_type) { case IEEE80211_WPS_ATTR_VERSION: printf("v:%d.%d", *ie >> 4, *ie & 0xf); break; case IEEE80211_WPS_ATTR_AP_SETUP_LOCKED: printf(" ap_setup:%s", *ie ? "locked" : "unlocked"); break; case IEEE80211_WPS_ATTR_CONFIG_METHODS: case IEEE80211_WPS_ATTR_SELECTED_REGISTRAR_CONFIG_METHODS: if (tlv_type == IEEE80211_WPS_ATTR_SELECTED_REGISTRAR_CONFIG_METHODS) printf(" sel_reg_cfg_mthd:"); else printf(" cfg_mthd:" ); cfg_mthd = BE_READ_2(ie); f = 0; for (n = 15; n >= 0; n--) { if (f) { printf(","); f = 0; } switch (cfg_mthd & (1 << n)) { case 0: break; case IEEE80211_WPS_CONFIG_USBA: printf("usba"); f++; break; case IEEE80211_WPS_CONFIG_ETHERNET: printf("ethernet"); f++; break; case IEEE80211_WPS_CONFIG_LABEL: printf("label"); f++; break; case IEEE80211_WPS_CONFIG_DISPLAY: if (!(cfg_mthd & (IEEE80211_WPS_CONFIG_VIRT_DISPLAY | IEEE80211_WPS_CONFIG_PHY_DISPLAY))) { printf("display"); f++; } break; case IEEE80211_WPS_CONFIG_EXT_NFC_TOKEN: printf("ext_nfc_tokenk"); f++; break; case IEEE80211_WPS_CONFIG_INT_NFC_TOKEN: printf("int_nfc_token"); f++; break; case IEEE80211_WPS_CONFIG_NFC_INTERFACE: printf("nfc_interface"); f++; break; case IEEE80211_WPS_CONFIG_PUSHBUTTON: if (!(cfg_mthd & (IEEE80211_WPS_CONFIG_VIRT_PUSHBUTTON | IEEE80211_WPS_CONFIG_PHY_PUSHBUTTON))) { printf("push_button"); f++; } break; case IEEE80211_WPS_CONFIG_KEYPAD: printf("keypad"); f++; break; case IEEE80211_WPS_CONFIG_VIRT_PUSHBUTTON: printf("virtual_push_button"); f++; break; case IEEE80211_WPS_CONFIG_PHY_PUSHBUTTON: printf("physical_push_button"); f++; break; case IEEE80211_WPS_CONFIG_P2PS: printf("p2ps"); f++; break; case IEEE80211_WPS_CONFIG_VIRT_DISPLAY: printf("virtual_display"); f++; break; case IEEE80211_WPS_CONFIG_PHY_DISPLAY: printf("physical_display"); f++; break; default: printf("unknown_wps_config<%04x>", cfg_mthd & (1 << n)); f++; break; } } break; case IEEE80211_WPS_ATTR_DEV_NAME: printf(" device_name:<%.*s>", tlv_len, ie); break; case IEEE80211_WPS_ATTR_DEV_PASSWORD_ID: n = LE_READ_2(ie); if (n < nitems(dev_pass_id)) printf(" dpi:%s", dev_pass_id[n]); break; case IEEE80211_WPS_ATTR_MANUFACTURER: printf(" manufacturer:<%.*s>", tlv_len, ie); break; case IEEE80211_WPS_ATTR_MODEL_NAME: printf(" model_name:<%.*s>", tlv_len, ie); break; case IEEE80211_WPS_ATTR_MODEL_NUMBER: printf(" model_number:<%.*s>", tlv_len, ie); break; case IEEE80211_WPS_ATTR_PRIMARY_DEV_TYPE: printf(" prim_dev:"); for (n = 0; n < tlv_len; n++) printf("%02x", ie[n]); break; case IEEE80211_WPS_ATTR_RF_BANDS: printf(" rf:"); f = 0; for (n = 7; n >= 0; n--) { if (f) { printf(","); f = 0; } switch (*ie & (1 << n)) { case 0: break; case IEEE80211_WPS_RF_BAND_24GHZ: printf("2.4Ghz"); f++; break; case IEEE80211_WPS_RF_BAND_50GHZ: printf("5Ghz"); f++; break; case IEEE80211_WPS_RF_BAND_600GHZ: printf("60Ghz"); f++; break; default: printf("unknown<%02x>", *ie & (1 << n)); f++; break; } } break; case IEEE80211_WPS_ATTR_RESPONSE_TYPE: printf(" resp_type:0x%02x", *ie); break; case IEEE80211_WPS_ATTR_SELECTED_REGISTRAR: printf(" sel:%s", *ie ? "T" : "F"); break; case IEEE80211_WPS_ATTR_SERIAL_NUMBER: printf(" serial_number:<%.*s>", tlv_len, ie); break; case IEEE80211_WPS_ATTR_UUID_E: printf(" uuid-e:"); for (n = 0; n < (tlv_len - 1); n++) printf("%02x-", ie[n]); printf("%02x", ie[n]); break; case IEEE80211_WPS_ATTR_VENDOR_EXT: printf(" vendor:"); for (n = 0; n < tlv_len; n++) printf("%02x", ie[n]); break; case IEEE80211_WPS_ATTR_WPS_STATE: switch (*ie) { case IEEE80211_WPS_STATE_NOT_CONFIGURED: printf(" state:N"); break; case IEEE80211_WPS_STATE_CONFIGURED: printf(" state:C"); break; default: printf(" state:B<%02x>", *ie); break; } break; default: printf(" unknown_wps_attr:0x%x", tlv_type); break; } ie += tlv_len, len -= tlv_len; } printf(">"); } } static void printtdmaie(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { printf("%s", tag); if (verbose && ielen >= sizeof(struct ieee80211_tdma_param)) { const struct ieee80211_tdma_param *tdma = (const struct ieee80211_tdma_param *) ie; /* XXX tstamp */ printf("", tdma->tdma_version, tdma->tdma_slot, tdma->tdma_slotcnt, LE_READ_2(&tdma->tdma_slotlen), tdma->tdma_bintval, tdma->tdma_inuse[0]); } } /* * Copy the ssid string contents into buf, truncating to fit. If the * ssid is entirely printable then just copy intact. Otherwise convert * to hexadecimal. If the result is truncated then replace the last * three characters with "...". */ static int copy_essid(char buf[], size_t bufsize, const u_int8_t *essid, size_t essid_len) { const u_int8_t *p; size_t maxlen; u_int i; if (essid_len > bufsize) maxlen = bufsize; else maxlen = essid_len; /* determine printable or not */ for (i = 0, p = essid; i < maxlen; i++, p++) { if (*p < ' ' || *p > 0x7e) break; } if (i != maxlen) { /* not printable, print as hex */ if (bufsize < 3) return 0; strlcpy(buf, "0x", bufsize); bufsize -= 2; p = essid; for (i = 0; i < maxlen && bufsize >= 2; i++) { sprintf(&buf[2+2*i], "%02x", p[i]); bufsize -= 2; } if (i != essid_len) memcpy(&buf[2+2*i-3], "...", 3); } else { /* printable, truncate as needed */ memcpy(buf, essid, maxlen); if (maxlen != essid_len) memcpy(&buf[maxlen-3], "...", 3); } return maxlen; } static void printssid(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { char ssid[2*IEEE80211_NWID_LEN+1]; printf("%s<%.*s>", tag, copy_essid(ssid, maxlen, ie+2, ie[1]), ssid); } static void printrates(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { const char *sep; int i; printf("%s", tag); sep = "<"; for (i = 2; i < ielen; i++) { printf("%s%s%d", sep, ie[i] & IEEE80211_RATE_BASIC ? "B" : "", ie[i] & IEEE80211_RATE_VAL); sep = ","; } printf(">"); } static void printcountry(const char *tag, const u_int8_t *ie, size_t ielen, int maxlen) { const struct ieee80211_country_ie *cie = (const struct ieee80211_country_ie *) ie; int i, nbands, schan, nchan; printf("%s<%c%c%c", tag, cie->cc[0], cie->cc[1], cie->cc[2]); nbands = (cie->len - 3) / sizeof(cie->band[0]); for (i = 0; i < nbands; i++) { schan = cie->band[i].schan; nchan = cie->band[i].nchan; if (nchan != 1) printf(" %u-%u,%u", schan, schan + nchan-1, cie->band[i].maxtxpwr); else printf(" %u,%u", schan, cie->band[i].maxtxpwr); } printf(">"); } static __inline int iswpaoui(const u_int8_t *frm) { return frm[1] > 3 && LE_READ_4(frm+2) == ((WPA_OUI_TYPE<<24)|WPA_OUI); } static __inline int iswmeinfo(const u_int8_t *frm) { return frm[1] > 5 && LE_READ_4(frm+2) == ((WME_OUI_TYPE<<24)|WME_OUI) && frm[6] == WME_INFO_OUI_SUBTYPE; } static __inline int iswmeparam(const u_int8_t *frm) { return frm[1] > 5 && LE_READ_4(frm+2) == ((WME_OUI_TYPE<<24)|WME_OUI) && frm[6] == WME_PARAM_OUI_SUBTYPE; } static __inline int isatherosoui(const u_int8_t *frm) { return frm[1] > 3 && LE_READ_4(frm+2) == ((ATH_OUI_TYPE<<24)|ATH_OUI); } static __inline int istdmaoui(const uint8_t *frm) { return frm[1] > 3 && LE_READ_4(frm+2) == ((TDMA_OUI_TYPE<<24)|TDMA_OUI); } static __inline int iswpsoui(const uint8_t *frm) { return frm[1] > 3 && LE_READ_4(frm+2) == ((WPS_OUI_TYPE<<24)|WPA_OUI); } static const char * iename(int elemid) { static char iename_buf[64]; switch (elemid) { case IEEE80211_ELEMID_FHPARMS: return " FHPARMS"; case IEEE80211_ELEMID_CFPARMS: return " CFPARMS"; case IEEE80211_ELEMID_TIM: return " TIM"; case IEEE80211_ELEMID_IBSSPARMS:return " IBSSPARMS"; case IEEE80211_ELEMID_BSSLOAD: return " BSSLOAD"; case IEEE80211_ELEMID_CHALLENGE:return " CHALLENGE"; case IEEE80211_ELEMID_PWRCNSTR: return " PWRCNSTR"; case IEEE80211_ELEMID_PWRCAP: return " PWRCAP"; case IEEE80211_ELEMID_TPCREQ: return " TPCREQ"; case IEEE80211_ELEMID_TPCREP: return " TPCREP"; case IEEE80211_ELEMID_SUPPCHAN: return " SUPPCHAN"; case IEEE80211_ELEMID_CSA: return " CSA"; case IEEE80211_ELEMID_MEASREQ: return " MEASREQ"; case IEEE80211_ELEMID_MEASREP: return " MEASREP"; case IEEE80211_ELEMID_QUIET: return " QUIET"; case IEEE80211_ELEMID_IBSSDFS: return " IBSSDFS"; case IEEE80211_ELEMID_RESERVED_47: return " RESERVED_47"; case IEEE80211_ELEMID_MOBILITY_DOMAIN: return " MOBILITY_DOMAIN"; case IEEE80211_ELEMID_RRM_ENACAPS: return " RRM_ENCAPS"; case IEEE80211_ELEMID_OVERLAP_BSS_SCAN_PARAM: return " OVERLAP_BSS"; case IEEE80211_ELEMID_TPC: return " TPC"; case IEEE80211_ELEMID_CCKM: return " CCKM"; case IEEE80211_ELEMID_EXTCAP: return " EXTCAP"; } snprintf(iename_buf, sizeof(iename_buf), " UNKNOWN_ELEMID_%d", elemid); return (const char *) iename_buf; } static void printies(const u_int8_t *vp, int ielen, int maxcols) { while (ielen > 0) { switch (vp[0]) { case IEEE80211_ELEMID_SSID: if (verbose) printssid(" SSID", vp, 2+vp[1], maxcols); break; case IEEE80211_ELEMID_RATES: case IEEE80211_ELEMID_XRATES: if (verbose) printrates(vp[0] == IEEE80211_ELEMID_RATES ? " RATES" : " XRATES", vp, 2+vp[1], maxcols); break; case IEEE80211_ELEMID_DSPARMS: if (verbose) printf(" DSPARMS<%u>", vp[2]); break; case IEEE80211_ELEMID_COUNTRY: if (verbose) printcountry(" COUNTRY", vp, 2+vp[1], maxcols); break; case IEEE80211_ELEMID_ERP: if (verbose) printf(" ERP<0x%x>", vp[2]); break; case IEEE80211_ELEMID_VENDOR: if (iswpaoui(vp)) printwpaie(" WPA", vp, 2+vp[1], maxcols); else if (iswmeinfo(vp)) printwmeinfo(" WME", vp, 2+vp[1], maxcols); else if (iswmeparam(vp)) printwmeparam(" WME", vp, 2+vp[1], maxcols); else if (isatherosoui(vp)) printathie(" ATH", vp, 2+vp[1], maxcols); else if (iswpsoui(vp)) printwpsie(" WPS", vp, 2+vp[1], maxcols); else if (istdmaoui(vp)) printtdmaie(" TDMA", vp, 2+vp[1], maxcols); else if (verbose) printie(" VEN", vp, 2+vp[1], maxcols); break; case IEEE80211_ELEMID_RSN: printrsnie(" RSN", vp, 2+vp[1], maxcols); break; case IEEE80211_ELEMID_HTCAP: printhtcap(" HTCAP", vp, 2+vp[1], maxcols); break; case IEEE80211_ELEMID_HTINFO: if (verbose) printhtinfo(" HTINFO", vp, 2+vp[1], maxcols); break; case IEEE80211_ELEMID_MESHID: if (verbose) printssid(" MESHID", vp, 2+vp[1], maxcols); break; case IEEE80211_ELEMID_MESHCONF: printmeshconf(" MESHCONF", vp, 2+vp[1], maxcols); break; case IEEE80211_ELEMID_VHT_CAP: printvhtcap(" VHTCAP", vp, 2+vp[1], maxcols); break; case IEEE80211_ELEMID_VHT_OPMODE: printvhtinfo(" VHTOPMODE", vp, 2+vp[1], maxcols); break; case IEEE80211_ELEMID_VHT_PWR_ENV: printvhtpwrenv(" VHTPWRENV", vp, 2+vp[1], maxcols); break; case IEEE80211_ELEMID_BSSLOAD: printbssload(" BSSLOAD", vp, 2+vp[1], maxcols); break; case IEEE80211_ELEMID_APCHANREP: printapchanrep(" APCHANREP", vp, 2+vp[1], maxcols); break; default: if (verbose) printie(iename(vp[0]), vp, 2+vp[1], maxcols); break; } ielen -= 2+vp[1]; vp += 2+vp[1]; } } static void printmimo(const struct ieee80211_mimo_info *mi) { int i; int r = 0; for (i = 0; i < IEEE80211_MAX_CHAINS; i++) { if (mi->ch[i].rssi != 0) { r = 1; break; } } /* NB: don't muddy display unless there's something to show */ if (r == 0) return; /* XXX TODO: ignore EVM; secondary channels for now */ printf(" (rssi %.1f:%.1f:%.1f:%.1f nf %d:%d:%d:%d)", mi->ch[0].rssi[0] / 2.0, mi->ch[1].rssi[0] / 2.0, mi->ch[2].rssi[0] / 2.0, mi->ch[3].rssi[0] / 2.0, mi->ch[0].noise[0], mi->ch[1].noise[0], mi->ch[2].noise[0], mi->ch[3].noise[0]); } static void list_scan(int s) { uint8_t buf[24*1024]; char ssid[IEEE80211_NWID_LEN+1]; const uint8_t *cp; int len, idlen; if (get80211len(s, IEEE80211_IOC_SCAN_RESULTS, buf, sizeof(buf), &len) < 0) errx(1, "unable to get scan results"); if (len < sizeof(struct ieee80211req_scan_result)) return; getchaninfo(s); printf("%-*.*s %-17.17s %4s %4s %-7s %3s %4s\n" , IEEE80211_NWID_LEN, IEEE80211_NWID_LEN, "SSID/MESH ID" , "BSSID" , "CHAN" , "RATE" , " S:N" , "INT" , "CAPS" ); cp = buf; do { const struct ieee80211req_scan_result *sr; const uint8_t *vp, *idp; sr = (const struct ieee80211req_scan_result *) cp; vp = cp + sr->isr_ie_off; if (sr->isr_meshid_len) { idp = vp + sr->isr_ssid_len; idlen = sr->isr_meshid_len; } else { idp = vp; idlen = sr->isr_ssid_len; } printf("%-*.*s %s %3d %3dM %4d:%-4d %4d %-4.4s" , IEEE80211_NWID_LEN , copy_essid(ssid, IEEE80211_NWID_LEN, idp, idlen) , ssid , ether_ntoa((const struct ether_addr *) sr->isr_bssid) , ieee80211_mhz2ieee(sr->isr_freq, sr->isr_flags) , getmaxrate(sr->isr_rates, sr->isr_nrates) , (sr->isr_rssi/2)+sr->isr_noise, sr->isr_noise , sr->isr_intval , getcaps(sr->isr_capinfo) ); printies(vp + sr->isr_ssid_len + sr->isr_meshid_len, sr->isr_ie_len, 24); printf("\n"); cp += sr->isr_len, len -= sr->isr_len; } while (len >= sizeof(struct ieee80211req_scan_result)); } static void scan_and_wait(int s) { struct ieee80211_scan_req sr; struct ieee80211req ireq; int sroute; sroute = socket(PF_ROUTE, SOCK_RAW, 0); if (sroute < 0) { perror("socket(PF_ROUTE,SOCK_RAW)"); return; } (void) memset(&ireq, 0, sizeof(ireq)); (void) strlcpy(ireq.i_name, name, sizeof(ireq.i_name)); ireq.i_type = IEEE80211_IOC_SCAN_REQ; memset(&sr, 0, sizeof(sr)); sr.sr_flags = IEEE80211_IOC_SCAN_ACTIVE | IEEE80211_IOC_SCAN_BGSCAN | IEEE80211_IOC_SCAN_NOPICK | IEEE80211_IOC_SCAN_ONCE; sr.sr_duration = IEEE80211_IOC_SCAN_FOREVER; sr.sr_nssid = 0; ireq.i_data = &sr; ireq.i_len = sizeof(sr); /* * NB: only root can trigger a scan so ignore errors. Also ignore * possible errors from net80211, even if no new scan could be * started there might still be a valid scan cache. */ if (ioctl(s, SIOCS80211, &ireq) == 0) { char buf[2048]; struct if_announcemsghdr *ifan; struct rt_msghdr *rtm; do { if (read(sroute, buf, sizeof(buf)) < 0) { perror("read(PF_ROUTE)"); break; } rtm = (struct rt_msghdr *) buf; if (rtm->rtm_version != RTM_VERSION) break; ifan = (struct if_announcemsghdr *) rtm; } while (rtm->rtm_type != RTM_IEEE80211 || ifan->ifan_what != RTM_IEEE80211_SCAN); } close(sroute); } static DECL_CMD_FUNC(set80211scan, val, d) { scan_and_wait(s); list_scan(s); } static enum ieee80211_opmode get80211opmode(int s); static int gettxseq(const struct ieee80211req_sta_info *si) { int i, txseq; if ((si->isi_state & IEEE80211_NODE_QOS) == 0) return si->isi_txseqs[0]; /* XXX not right but usually what folks want */ txseq = 0; for (i = 0; i < IEEE80211_TID_SIZE; i++) if (si->isi_txseqs[i] > txseq) txseq = si->isi_txseqs[i]; return txseq; } static int getrxseq(const struct ieee80211req_sta_info *si) { int i, rxseq; if ((si->isi_state & IEEE80211_NODE_QOS) == 0) return si->isi_rxseqs[0]; /* XXX not right but usually what folks want */ rxseq = 0; for (i = 0; i < IEEE80211_TID_SIZE; i++) if (si->isi_rxseqs[i] > rxseq) rxseq = si->isi_rxseqs[i]; return rxseq; } static void list_stations(int s) { union { struct ieee80211req_sta_req req; uint8_t buf[24*1024]; } u; enum ieee80211_opmode opmode = get80211opmode(s); const uint8_t *cp; int len; /* broadcast address =>'s get all stations */ (void) memset(u.req.is_u.macaddr, 0xff, IEEE80211_ADDR_LEN); if (opmode == IEEE80211_M_STA) { /* * Get information about the associated AP. */ (void) get80211(s, IEEE80211_IOC_BSSID, u.req.is_u.macaddr, IEEE80211_ADDR_LEN); } if (get80211len(s, IEEE80211_IOC_STA_INFO, &u, sizeof(u), &len) < 0) errx(1, "unable to get station information"); if (len < sizeof(struct ieee80211req_sta_info)) return; getchaninfo(s); if (opmode == IEEE80211_M_MBSS) printf("%-17.17s %4s %5s %5s %7s %4s %4s %4s %6s %6s\n" , "ADDR" , "CHAN" , "LOCAL" , "PEER" , "STATE" , "RATE" , "RSSI" , "IDLE" , "TXSEQ" , "RXSEQ" ); else printf("%-17.17s %4s %4s %4s %4s %4s %6s %6s %4s %-12s\n" , "ADDR" , "AID" , "CHAN" , "RATE" , "RSSI" , "IDLE" , "TXSEQ" , "RXSEQ" , "CAPS" , "FLAG" ); cp = (const uint8_t *) u.req.info; do { const struct ieee80211req_sta_info *si; si = (const struct ieee80211req_sta_info *) cp; if (si->isi_len < sizeof(*si)) break; if (opmode == IEEE80211_M_MBSS) printf("%s %4d %5x %5x %7.7s %3dM %4.1f %4d %6d %6d" , ether_ntoa((const struct ether_addr*) si->isi_macaddr) , ieee80211_mhz2ieee(si->isi_freq, si->isi_flags) , si->isi_localid , si->isi_peerid , mesh_linkstate_string(si->isi_peerstate) , si->isi_txmbps/2 , si->isi_rssi/2. , si->isi_inact , gettxseq(si) , getrxseq(si) ); else printf("%s %4u %4d %3dM %4.1f %4d %6d %6d %-4.4s %-12.12s" , ether_ntoa((const struct ether_addr*) si->isi_macaddr) , IEEE80211_AID(si->isi_associd) , ieee80211_mhz2ieee(si->isi_freq, si->isi_flags) , si->isi_txmbps/2 , si->isi_rssi/2. , si->isi_inact , gettxseq(si) , getrxseq(si) , getcaps(si->isi_capinfo) , getflags(si->isi_state) ); printies(cp + si->isi_ie_off, si->isi_ie_len, 24); printmimo(&si->isi_mimo); printf("\n"); cp += si->isi_len, len -= si->isi_len; } while (len >= sizeof(struct ieee80211req_sta_info)); } static const char * mesh_linkstate_string(uint8_t state) { static const char *state_names[] = { [0] = "IDLE", [1] = "OPEN-TX", [2] = "OPEN-RX", [3] = "CONF-RX", [4] = "ESTAB", [5] = "HOLDING", }; if (state >= nitems(state_names)) { static char buf[10]; snprintf(buf, sizeof(buf), "#%u", state); return buf; } else return state_names[state]; } static const char * get_chaninfo(const struct ieee80211_channel *c, int precise, char buf[], size_t bsize) { buf[0] = '\0'; if (IEEE80211_IS_CHAN_FHSS(c)) strlcat(buf, " FHSS", bsize); if (IEEE80211_IS_CHAN_A(c)) strlcat(buf, " 11a", bsize); else if (IEEE80211_IS_CHAN_ANYG(c)) strlcat(buf, " 11g", bsize); else if (IEEE80211_IS_CHAN_B(c)) strlcat(buf, " 11b", bsize); if (IEEE80211_IS_CHAN_HALF(c)) strlcat(buf, "/10MHz", bsize); if (IEEE80211_IS_CHAN_QUARTER(c)) strlcat(buf, "/5MHz", bsize); if (IEEE80211_IS_CHAN_TURBO(c)) strlcat(buf, " Turbo", bsize); if (precise) { /* XXX should make VHT80U, VHT80D */ if (IEEE80211_IS_CHAN_VHT80(c) && IEEE80211_IS_CHAN_HT40D(c)) strlcat(buf, " vht/80-", bsize); else if (IEEE80211_IS_CHAN_VHT80(c) && IEEE80211_IS_CHAN_HT40U(c)) strlcat(buf, " vht/80+", bsize); else if (IEEE80211_IS_CHAN_VHT80(c)) strlcat(buf, " vht/80", bsize); else if (IEEE80211_IS_CHAN_VHT40D(c)) strlcat(buf, " vht/40-", bsize); else if (IEEE80211_IS_CHAN_VHT40U(c)) strlcat(buf, " vht/40+", bsize); else if (IEEE80211_IS_CHAN_VHT20(c)) strlcat(buf, " vht/20", bsize); else if (IEEE80211_IS_CHAN_HT20(c)) strlcat(buf, " ht/20", bsize); else if (IEEE80211_IS_CHAN_HT40D(c)) strlcat(buf, " ht/40-", bsize); else if (IEEE80211_IS_CHAN_HT40U(c)) strlcat(buf, " ht/40+", bsize); } else { if (IEEE80211_IS_CHAN_VHT(c)) strlcat(buf, " vht", bsize); else if (IEEE80211_IS_CHAN_HT(c)) strlcat(buf, " ht", bsize); } return buf; } static void print_chaninfo(const struct ieee80211_channel *c, int verb) { char buf[14]; if (verb) printf("Channel %3u : %u%c%c%c%c%c MHz%-14.14s", ieee80211_mhz2ieee(c->ic_freq, c->ic_flags), c->ic_freq, IEEE80211_IS_CHAN_PASSIVE(c) ? '*' : ' ', IEEE80211_IS_CHAN_DFS(c) ? 'D' : ' ', IEEE80211_IS_CHAN_RADAR(c) ? 'R' : ' ', IEEE80211_IS_CHAN_CWINT(c) ? 'I' : ' ', IEEE80211_IS_CHAN_CACDONE(c) ? 'C' : ' ', get_chaninfo(c, verb, buf, sizeof(buf))); else printf("Channel %3u : %u%c MHz%-14.14s", ieee80211_mhz2ieee(c->ic_freq, c->ic_flags), c->ic_freq, IEEE80211_IS_CHAN_PASSIVE(c) ? '*' : ' ', get_chaninfo(c, verb, buf, sizeof(buf))); } static int chanpref(const struct ieee80211_channel *c) { if (IEEE80211_IS_CHAN_VHT160(c)) return 80; if (IEEE80211_IS_CHAN_VHT80_80(c)) return 75; if (IEEE80211_IS_CHAN_VHT80(c)) return 70; if (IEEE80211_IS_CHAN_VHT40(c)) return 60; if (IEEE80211_IS_CHAN_VHT20(c)) return 50; if (IEEE80211_IS_CHAN_HT40(c)) return 40; if (IEEE80211_IS_CHAN_HT20(c)) return 30; if (IEEE80211_IS_CHAN_HALF(c)) return 10; if (IEEE80211_IS_CHAN_QUARTER(c)) return 5; if (IEEE80211_IS_CHAN_TURBO(c)) return 25; if (IEEE80211_IS_CHAN_A(c)) return 20; if (IEEE80211_IS_CHAN_G(c)) return 20; if (IEEE80211_IS_CHAN_B(c)) return 15; if (IEEE80211_IS_CHAN_PUREG(c)) return 15; return 0; } static void print_channels(int s, const struct ieee80211req_chaninfo *chans, int allchans, int verb) { struct ieee80211req_chaninfo *achans; uint8_t reported[IEEE80211_CHAN_BYTES]; const struct ieee80211_channel *c; int i, half; achans = malloc(IEEE80211_CHANINFO_SPACE(chans)); if (achans == NULL) errx(1, "no space for active channel list"); achans->ic_nchans = 0; memset(reported, 0, sizeof(reported)); if (!allchans) { struct ieee80211req_chanlist active; if (get80211(s, IEEE80211_IOC_CHANLIST, &active, sizeof(active)) < 0) errx(1, "unable to get active channel list"); for (i = 0; i < chans->ic_nchans; i++) { c = &chans->ic_chans[i]; if (!isset(active.ic_channels, c->ic_ieee)) continue; /* * Suppress compatible duplicates unless * verbose. The kernel gives us it's * complete channel list which has separate * entries for 11g/11b and 11a/turbo. */ if (isset(reported, c->ic_ieee) && !verb) { /* XXX we assume duplicates are adjacent */ achans->ic_chans[achans->ic_nchans-1] = *c; } else { achans->ic_chans[achans->ic_nchans++] = *c; setbit(reported, c->ic_ieee); } } } else { for (i = 0; i < chans->ic_nchans; i++) { c = &chans->ic_chans[i]; /* suppress duplicates as above */ if (isset(reported, c->ic_ieee) && !verb) { /* XXX we assume duplicates are adjacent */ struct ieee80211_channel *a = &achans->ic_chans[achans->ic_nchans-1]; if (chanpref(c) > chanpref(a)) *a = *c; } else { achans->ic_chans[achans->ic_nchans++] = *c; setbit(reported, c->ic_ieee); } } } half = achans->ic_nchans / 2; if (achans->ic_nchans % 2) half++; for (i = 0; i < achans->ic_nchans / 2; i++) { print_chaninfo(&achans->ic_chans[i], verb); print_chaninfo(&achans->ic_chans[half+i], verb); printf("\n"); } if (achans->ic_nchans % 2) { print_chaninfo(&achans->ic_chans[i], verb); printf("\n"); } free(achans); } static void list_channels(int s, int allchans) { getchaninfo(s); print_channels(s, chaninfo, allchans, verbose); } static void print_txpow(const struct ieee80211_channel *c) { printf("Channel %3u : %u MHz %3.1f reg %2d ", c->ic_ieee, c->ic_freq, c->ic_maxpower/2., c->ic_maxregpower); } static void print_txpow_verbose(const struct ieee80211_channel *c) { print_chaninfo(c, 1); printf("min %4.1f dBm max %3.1f dBm reg %2d dBm", c->ic_minpower/2., c->ic_maxpower/2., c->ic_maxregpower); /* indicate where regulatory cap limits power use */ if (c->ic_maxpower > 2*c->ic_maxregpower) printf(" <"); } static void list_txpow(int s) { struct ieee80211req_chaninfo *achans; uint8_t reported[IEEE80211_CHAN_BYTES]; struct ieee80211_channel *c, *prev; int i, half; getchaninfo(s); achans = malloc(IEEE80211_CHANINFO_SPACE(chaninfo)); if (achans == NULL) errx(1, "no space for active channel list"); achans->ic_nchans = 0; memset(reported, 0, sizeof(reported)); for (i = 0; i < chaninfo->ic_nchans; i++) { c = &chaninfo->ic_chans[i]; /* suppress duplicates as above */ if (isset(reported, c->ic_ieee) && !verbose) { /* XXX we assume duplicates are adjacent */ assert(achans->ic_nchans > 0); prev = &achans->ic_chans[achans->ic_nchans-1]; /* display highest power on channel */ if (c->ic_maxpower > prev->ic_maxpower) *prev = *c; } else { achans->ic_chans[achans->ic_nchans++] = *c; setbit(reported, c->ic_ieee); } } if (!verbose) { half = achans->ic_nchans / 2; if (achans->ic_nchans % 2) half++; for (i = 0; i < achans->ic_nchans / 2; i++) { print_txpow(&achans->ic_chans[i]); print_txpow(&achans->ic_chans[half+i]); printf("\n"); } if (achans->ic_nchans % 2) { print_txpow(&achans->ic_chans[i]); printf("\n"); } } else { for (i = 0; i < achans->ic_nchans; i++) { print_txpow_verbose(&achans->ic_chans[i]); printf("\n"); } } free(achans); } static void list_keys(int s) { } static void list_capabilities(int s) { struct ieee80211_devcaps_req *dc; if (verbose) dc = malloc(IEEE80211_DEVCAPS_SIZE(MAXCHAN)); else dc = malloc(IEEE80211_DEVCAPS_SIZE(1)); if (dc == NULL) errx(1, "no space for device capabilities"); dc->dc_chaninfo.ic_nchans = verbose ? MAXCHAN : 1; getdevcaps(s, dc); printb("drivercaps", dc->dc_drivercaps, IEEE80211_C_BITS); if (dc->dc_cryptocaps != 0 || verbose) { putchar('\n'); printb("cryptocaps", dc->dc_cryptocaps, IEEE80211_CRYPTO_BITS); } if (dc->dc_htcaps != 0 || verbose) { putchar('\n'); printb("htcaps", dc->dc_htcaps, IEEE80211_HTCAP_BITS); } if (dc->dc_vhtcaps != 0 || verbose) { putchar('\n'); printb("vhtcaps", dc->dc_vhtcaps, IEEE80211_VHTCAP_BITS); } putchar('\n'); if (verbose) { chaninfo = &dc->dc_chaninfo; /* XXX */ print_channels(s, &dc->dc_chaninfo, 1/*allchans*/, verbose); } free(dc); } static int get80211wme(int s, int param, int ac, int *val) { struct ieee80211req ireq; (void) memset(&ireq, 0, sizeof(ireq)); (void) strlcpy(ireq.i_name, name, sizeof(ireq.i_name)); ireq.i_type = param; ireq.i_len = ac; if (ioctl(s, SIOCG80211, &ireq) < 0) { warn("cannot get WME parameter %d, ac %d%s", param, ac & IEEE80211_WMEPARAM_VAL, ac & IEEE80211_WMEPARAM_BSS ? " (BSS)" : ""); return -1; } *val = ireq.i_val; return 0; } static void list_wme_aci(int s, const char *tag, int ac) { int val; printf("\t%s", tag); /* show WME BSS parameters */ if (get80211wme(s, IEEE80211_IOC_WME_CWMIN, ac, &val) != -1) printf(" cwmin %2u", val); if (get80211wme(s, IEEE80211_IOC_WME_CWMAX, ac, &val) != -1) printf(" cwmax %2u", val); if (get80211wme(s, IEEE80211_IOC_WME_AIFS, ac, &val) != -1) printf(" aifs %2u", val); if (get80211wme(s, IEEE80211_IOC_WME_TXOPLIMIT, ac, &val) != -1) printf(" txopLimit %3u", val); if (get80211wme(s, IEEE80211_IOC_WME_ACM, ac, &val) != -1) { if (val) printf(" acm"); else if (verbose) printf(" -acm"); } /* !BSS only */ if ((ac & IEEE80211_WMEPARAM_BSS) == 0) { if (get80211wme(s, IEEE80211_IOC_WME_ACKPOLICY, ac, &val) != -1) { if (!val) printf(" -ack"); else if (verbose) printf(" ack"); } } printf("\n"); } static void list_wme(int s) { static const char *acnames[] = { "AC_BE", "AC_BK", "AC_VI", "AC_VO" }; int ac; if (verbose) { /* display both BSS and local settings */ for (ac = WME_AC_BE; ac <= WME_AC_VO; ac++) { again: if (ac & IEEE80211_WMEPARAM_BSS) list_wme_aci(s, " ", ac); else list_wme_aci(s, acnames[ac], ac); if ((ac & IEEE80211_WMEPARAM_BSS) == 0) { ac |= IEEE80211_WMEPARAM_BSS; goto again; } else ac &= ~IEEE80211_WMEPARAM_BSS; } } else { /* display only channel settings */ for (ac = WME_AC_BE; ac <= WME_AC_VO; ac++) list_wme_aci(s, acnames[ac], ac); } } static void list_roam(int s) { const struct ieee80211_roamparam *rp; int mode; getroam(s); for (mode = IEEE80211_MODE_11A; mode < IEEE80211_MODE_MAX; mode++) { rp = &roamparams.params[mode]; if (rp->rssi == 0 && rp->rate == 0) continue; if (mode == IEEE80211_MODE_11NA || mode == IEEE80211_MODE_11NG || mode == IEEE80211_MODE_VHT_2GHZ || mode == IEEE80211_MODE_VHT_5GHZ) { if (rp->rssi & 1) LINE_CHECK("roam:%-7.7s rssi %2u.5dBm MCS %2u ", modename[mode], rp->rssi/2, rp->rate &~ IEEE80211_RATE_MCS); else LINE_CHECK("roam:%-7.7s rssi %4udBm MCS %2u ", modename[mode], rp->rssi/2, rp->rate &~ IEEE80211_RATE_MCS); } else { if (rp->rssi & 1) LINE_CHECK("roam:%-7.7s rssi %2u.5dBm rate %2u Mb/s", modename[mode], rp->rssi/2, rp->rate/2); else LINE_CHECK("roam:%-7.7s rssi %4udBm rate %2u Mb/s", modename[mode], rp->rssi/2, rp->rate/2); } } } /* XXX TODO: rate-to-string method... */ static const char* get_mcs_mbs_rate_str(uint8_t rate) { return (rate & IEEE80211_RATE_MCS) ? "MCS " : "Mb/s"; } static uint8_t get_rate_value(uint8_t rate) { if (rate & IEEE80211_RATE_MCS) return (rate &~ IEEE80211_RATE_MCS); return (rate / 2); } static void list_txparams(int s) { const struct ieee80211_txparam *tp; int mode; gettxparams(s); for (mode = IEEE80211_MODE_11A; mode < IEEE80211_MODE_MAX; mode++) { tp = &txparams.params[mode]; if (tp->mgmtrate == 0 && tp->mcastrate == 0) continue; if (mode == IEEE80211_MODE_11NA || mode == IEEE80211_MODE_11NG || mode == IEEE80211_MODE_VHT_2GHZ || mode == IEEE80211_MODE_VHT_5GHZ) { if (tp->ucastrate == IEEE80211_FIXED_RATE_NONE) LINE_CHECK("%-7.7s ucast NONE mgmt %2u %s " "mcast %2u %s maxretry %u", modename[mode], get_rate_value(tp->mgmtrate), get_mcs_mbs_rate_str(tp->mgmtrate), get_rate_value(tp->mcastrate), get_mcs_mbs_rate_str(tp->mcastrate), tp->maxretry); else LINE_CHECK("%-7.7s ucast %2u MCS mgmt %2u %s " "mcast %2u %s maxretry %u", modename[mode], tp->ucastrate &~ IEEE80211_RATE_MCS, get_rate_value(tp->mgmtrate), get_mcs_mbs_rate_str(tp->mgmtrate), get_rate_value(tp->mcastrate), get_mcs_mbs_rate_str(tp->mcastrate), tp->maxretry); } else { if (tp->ucastrate == IEEE80211_FIXED_RATE_NONE) LINE_CHECK("%-7.7s ucast NONE mgmt %2u Mb/s " "mcast %2u Mb/s maxretry %u", modename[mode], tp->mgmtrate/2, tp->mcastrate/2, tp->maxretry); else LINE_CHECK("%-7.7s ucast %2u Mb/s mgmt %2u Mb/s " "mcast %2u Mb/s maxretry %u", modename[mode], tp->ucastrate/2, tp->mgmtrate/2, tp->mcastrate/2, tp->maxretry); } } } static void printpolicy(int policy) { switch (policy) { case IEEE80211_MACCMD_POLICY_OPEN: printf("policy: open\n"); break; case IEEE80211_MACCMD_POLICY_ALLOW: printf("policy: allow\n"); break; case IEEE80211_MACCMD_POLICY_DENY: printf("policy: deny\n"); break; case IEEE80211_MACCMD_POLICY_RADIUS: printf("policy: radius\n"); break; default: printf("policy: unknown (%u)\n", policy); break; } } static void list_mac(int s) { struct ieee80211req ireq; struct ieee80211req_maclist *acllist; int i, nacls, policy, len; uint8_t *data; char c; (void) memset(&ireq, 0, sizeof(ireq)); (void) strlcpy(ireq.i_name, name, sizeof(ireq.i_name)); /* XXX ?? */ ireq.i_type = IEEE80211_IOC_MACCMD; ireq.i_val = IEEE80211_MACCMD_POLICY; if (ioctl(s, SIOCG80211, &ireq) < 0) { if (errno == EINVAL) { printf("No acl policy loaded\n"); return; } err(1, "unable to get mac policy"); } policy = ireq.i_val; if (policy == IEEE80211_MACCMD_POLICY_OPEN) { c = '*'; } else if (policy == IEEE80211_MACCMD_POLICY_ALLOW) { c = '+'; } else if (policy == IEEE80211_MACCMD_POLICY_DENY) { c = '-'; } else if (policy == IEEE80211_MACCMD_POLICY_RADIUS) { c = 'r'; /* NB: should never have entries */ } else { printf("policy: unknown (%u)\n", policy); c = '?'; } if (verbose || c == '?') printpolicy(policy); ireq.i_val = IEEE80211_MACCMD_LIST; ireq.i_len = 0; if (ioctl(s, SIOCG80211, &ireq) < 0) err(1, "unable to get mac acl list size"); if (ireq.i_len == 0) { /* NB: no acls */ if (!(verbose || c == '?')) printpolicy(policy); return; } len = ireq.i_len; data = malloc(len); if (data == NULL) err(1, "out of memory for acl list"); ireq.i_data = data; if (ioctl(s, SIOCG80211, &ireq) < 0) err(1, "unable to get mac acl list"); nacls = len / sizeof(*acllist); acllist = (struct ieee80211req_maclist *) data; for (i = 0; i < nacls; i++) printf("%c%s\n", c, ether_ntoa( (const struct ether_addr *) acllist[i].ml_macaddr)); free(data); } static void print_regdomain(const struct ieee80211_regdomain *reg, int verb) { if ((reg->regdomain != 0 && reg->regdomain != reg->country) || verb) { const struct regdomain *rd = lib80211_regdomain_findbysku(getregdata(), reg->regdomain); if (rd == NULL) LINE_CHECK("regdomain %d", reg->regdomain); else LINE_CHECK("regdomain %s", rd->name); } if (reg->country != 0 || verb) { const struct country *cc = lib80211_country_findbycc(getregdata(), reg->country); if (cc == NULL) LINE_CHECK("country %d", reg->country); else LINE_CHECK("country %s", cc->isoname); } if (reg->location == 'I') LINE_CHECK("indoor"); else if (reg->location == 'O') LINE_CHECK("outdoor"); else if (verb) LINE_CHECK("anywhere"); if (reg->ecm) LINE_CHECK("ecm"); else if (verb) LINE_CHECK("-ecm"); } static void list_regdomain(int s, int channelsalso) { getregdomain(s); if (channelsalso) { getchaninfo(s); spacer = ':'; print_regdomain(®domain, 1); LINE_BREAK(); print_channels(s, chaninfo, 1/*allchans*/, 1/*verbose*/); } else print_regdomain(®domain, verbose); } static void list_mesh(int s) { struct ieee80211req ireq; struct ieee80211req_mesh_route routes[128]; struct ieee80211req_mesh_route *rt; (void) memset(&ireq, 0, sizeof(ireq)); (void) strlcpy(ireq.i_name, name, sizeof(ireq.i_name)); ireq.i_type = IEEE80211_IOC_MESH_RTCMD; ireq.i_val = IEEE80211_MESH_RTCMD_LIST; ireq.i_data = &routes; ireq.i_len = sizeof(routes); if (ioctl(s, SIOCG80211, &ireq) < 0) err(1, "unable to get the Mesh routing table"); printf("%-17.17s %-17.17s %4s %4s %4s %6s %s\n" , "DEST" , "NEXT HOP" , "HOPS" , "METRIC" , "LIFETIME" , "MSEQ" , "FLAGS"); for (rt = &routes[0]; rt - &routes[0] < ireq.i_len / sizeof(*rt); rt++){ printf("%s ", ether_ntoa((const struct ether_addr *)rt->imr_dest)); printf("%s %4u %4u %6u %6u %c%c\n", ether_ntoa((const struct ether_addr *)rt->imr_nexthop), rt->imr_nhops, rt->imr_metric, rt->imr_lifetime, rt->imr_lastmseq, (rt->imr_flags & IEEE80211_MESHRT_FLAGS_DISCOVER) ? 'D' : (rt->imr_flags & IEEE80211_MESHRT_FLAGS_VALID) ? 'V' : '!', (rt->imr_flags & IEEE80211_MESHRT_FLAGS_PROXY) ? 'P' : (rt->imr_flags & IEEE80211_MESHRT_FLAGS_GATE) ? 'G' :' '); } } static DECL_CMD_FUNC(set80211list, arg, d) { #define iseq(a,b) (strncasecmp(a,b,sizeof(b)-1) == 0) LINE_INIT('\t'); if (iseq(arg, "sta")) list_stations(s); else if (iseq(arg, "scan") || iseq(arg, "ap")) list_scan(s); else if (iseq(arg, "chan") || iseq(arg, "freq")) list_channels(s, 1); else if (iseq(arg, "active")) list_channels(s, 0); else if (iseq(arg, "keys")) list_keys(s); else if (iseq(arg, "caps")) list_capabilities(s); else if (iseq(arg, "wme") || iseq(arg, "wmm")) list_wme(s); else if (iseq(arg, "mac")) list_mac(s); else if (iseq(arg, "txpow")) list_txpow(s); else if (iseq(arg, "roam")) list_roam(s); else if (iseq(arg, "txparam") || iseq(arg, "txparm")) list_txparams(s); else if (iseq(arg, "regdomain")) list_regdomain(s, 1); else if (iseq(arg, "countries")) list_countries(); else if (iseq(arg, "mesh")) list_mesh(s); else errx(1, "Don't know how to list %s for %s", arg, name); LINE_BREAK(); #undef iseq } static enum ieee80211_opmode get80211opmode(int s) { struct ifmediareq ifmr; (void) memset(&ifmr, 0, sizeof(ifmr)); (void) strlcpy(ifmr.ifm_name, name, sizeof(ifmr.ifm_name)); if (ioctl(s, SIOCGIFMEDIA, (caddr_t)&ifmr) >= 0) { if (ifmr.ifm_current & IFM_IEEE80211_ADHOC) { if (ifmr.ifm_current & IFM_FLAG0) return IEEE80211_M_AHDEMO; else return IEEE80211_M_IBSS; } if (ifmr.ifm_current & IFM_IEEE80211_HOSTAP) return IEEE80211_M_HOSTAP; if (ifmr.ifm_current & IFM_IEEE80211_IBSS) return IEEE80211_M_IBSS; if (ifmr.ifm_current & IFM_IEEE80211_MONITOR) return IEEE80211_M_MONITOR; if (ifmr.ifm_current & IFM_IEEE80211_MBSS) return IEEE80211_M_MBSS; } return IEEE80211_M_STA; } #if 0 static void printcipher(int s, struct ieee80211req *ireq, int keylenop) { switch (ireq->i_val) { case IEEE80211_CIPHER_WEP: ireq->i_type = keylenop; if (ioctl(s, SIOCG80211, ireq) != -1) printf("WEP-%s", ireq->i_len <= 5 ? "40" : ireq->i_len <= 13 ? "104" : "128"); else printf("WEP"); break; case IEEE80211_CIPHER_TKIP: printf("TKIP"); break; case IEEE80211_CIPHER_AES_OCB: printf("AES-OCB"); break; case IEEE80211_CIPHER_AES_CCM: printf("AES-CCM"); break; case IEEE80211_CIPHER_CKIP: printf("CKIP"); break; case IEEE80211_CIPHER_NONE: printf("NONE"); break; default: printf("UNKNOWN (0x%x)", ireq->i_val); break; } } #endif static void printkey(const struct ieee80211req_key *ik) { static const uint8_t zerodata[IEEE80211_KEYBUF_SIZE]; u_int keylen = ik->ik_keylen; int printcontents; printcontents = printkeys && (memcmp(ik->ik_keydata, zerodata, keylen) != 0 || verbose); if (printcontents) LINE_BREAK(); switch (ik->ik_type) { case IEEE80211_CIPHER_WEP: /* compatibility */ LINE_CHECK("wepkey %u:%s", ik->ik_keyix+1, keylen <= 5 ? "40-bit" : keylen <= 13 ? "104-bit" : "128-bit"); break; case IEEE80211_CIPHER_TKIP: if (keylen > 128/8) keylen -= 128/8; /* ignore MIC for now */ LINE_CHECK("TKIP %u:%u-bit", ik->ik_keyix+1, 8*keylen); break; case IEEE80211_CIPHER_AES_OCB: LINE_CHECK("AES-OCB %u:%u-bit", ik->ik_keyix+1, 8*keylen); break; case IEEE80211_CIPHER_AES_CCM: LINE_CHECK("AES-CCM %u:%u-bit", ik->ik_keyix+1, 8*keylen); break; case IEEE80211_CIPHER_CKIP: LINE_CHECK("CKIP %u:%u-bit", ik->ik_keyix+1, 8*keylen); break; case IEEE80211_CIPHER_NONE: LINE_CHECK("NULL %u:%u-bit", ik->ik_keyix+1, 8*keylen); break; default: LINE_CHECK("UNKNOWN (0x%x) %u:%u-bit", ik->ik_type, ik->ik_keyix+1, 8*keylen); break; } if (printcontents) { u_int i; printf(" <"); for (i = 0; i < keylen; i++) printf("%02x", ik->ik_keydata[i]); printf(">"); if (ik->ik_type != IEEE80211_CIPHER_WEP && (ik->ik_keyrsc != 0 || verbose)) printf(" rsc %ju", (uintmax_t)ik->ik_keyrsc); if (ik->ik_type != IEEE80211_CIPHER_WEP && (ik->ik_keytsc != 0 || verbose)) printf(" tsc %ju", (uintmax_t)ik->ik_keytsc); if (ik->ik_flags != 0 && verbose) { const char *sep = " "; if (ik->ik_flags & IEEE80211_KEY_XMIT) printf("%stx", sep), sep = "+"; if (ik->ik_flags & IEEE80211_KEY_RECV) printf("%srx", sep), sep = "+"; if (ik->ik_flags & IEEE80211_KEY_DEFAULT) printf("%sdef", sep), sep = "+"; } LINE_BREAK(); } } static void printrate(const char *tag, int v, int defrate, int defmcs) { if ((v & IEEE80211_RATE_MCS) == 0) { if (v != defrate) { if (v & 1) LINE_CHECK("%s %d.5", tag, v/2); else LINE_CHECK("%s %d", tag, v/2); } } else { if (v != defmcs) LINE_CHECK("%s %d", tag, v &~ 0x80); } } static int getid(int s, int ix, void *data, size_t len, int *plen, int mesh) { struct ieee80211req ireq; (void) memset(&ireq, 0, sizeof(ireq)); (void) strlcpy(ireq.i_name, name, sizeof(ireq.i_name)); ireq.i_type = (!mesh) ? IEEE80211_IOC_SSID : IEEE80211_IOC_MESH_ID; ireq.i_val = ix; ireq.i_data = data; ireq.i_len = len; if (ioctl(s, SIOCG80211, &ireq) < 0) return -1; *plen = ireq.i_len; return 0; } +static int +getdevicename(int s, void *data, size_t len, int *plen) +{ + struct ieee80211req ireq; + + (void) memset(&ireq, 0, sizeof(ireq)); + (void) strlcpy(ireq.i_name, name, sizeof(ireq.i_name)); + ireq.i_type = IEEE80211_IOC_IC_NAME; + ireq.i_val = -1; + ireq.i_data = data; + ireq.i_len = len; + if (ioctl(s, SIOCG80211, &ireq) < 0) + return (-1); + *plen = ireq.i_len; + return (0); +} + static void ieee80211_status(int s) { static const uint8_t zerobssid[IEEE80211_ADDR_LEN]; enum ieee80211_opmode opmode = get80211opmode(s); int i, num, wpa, wme, bgscan, bgscaninterval, val, len, wepmode; uint8_t data[32]; const struct ieee80211_channel *c; const struct ieee80211_roamparam *rp; const struct ieee80211_txparam *tp; if (getid(s, -1, data, sizeof(data), &len, 0) < 0) { /* If we can't get the SSID, this isn't an 802.11 device. */ return; } /* * Invalidate cached state so printing status for multiple * if's doesn't reuse the first interfaces' cached state. */ gotcurchan = 0; gotroam = 0; gottxparams = 0; gothtconf = 0; gotregdomain = 0; printf("\t"); if (opmode == IEEE80211_M_MBSS) { printf("meshid "); getid(s, 0, data, sizeof(data), &len, 1); print_string(data, len); } else { if (get80211val(s, IEEE80211_IOC_NUMSSIDS, &num) < 0) num = 0; printf("ssid "); if (num > 1) { for (i = 0; i < num; i++) { if (getid(s, i, data, sizeof(data), &len, 0) >= 0 && len > 0) { printf(" %d:", i + 1); print_string(data, len); } } } else print_string(data, len); } c = getcurchan(s); if (c->ic_freq != IEEE80211_CHAN_ANY) { char buf[14]; printf(" channel %d (%u MHz%s)", c->ic_ieee, c->ic_freq, get_chaninfo(c, 1, buf, sizeof(buf))); } else if (verbose) printf(" channel UNDEF"); if (get80211(s, IEEE80211_IOC_BSSID, data, IEEE80211_ADDR_LEN) >= 0 && (memcmp(data, zerobssid, sizeof(zerobssid)) != 0 || verbose)) printf(" bssid %s", ether_ntoa((struct ether_addr *)data)); if (get80211len(s, IEEE80211_IOC_STATIONNAME, data, sizeof(data), &len) != -1) { printf("\n\tstationname "); print_string(data, len); } spacer = ' '; /* force first break */ LINE_BREAK(); list_regdomain(s, 0); wpa = 0; if (get80211val(s, IEEE80211_IOC_AUTHMODE, &val) != -1) { switch (val) { case IEEE80211_AUTH_NONE: LINE_CHECK("authmode NONE"); break; case IEEE80211_AUTH_OPEN: LINE_CHECK("authmode OPEN"); break; case IEEE80211_AUTH_SHARED: LINE_CHECK("authmode SHARED"); break; case IEEE80211_AUTH_8021X: LINE_CHECK("authmode 802.1x"); break; case IEEE80211_AUTH_WPA: if (get80211val(s, IEEE80211_IOC_WPA, &wpa) < 0) wpa = 1; /* default to WPA1 */ switch (wpa) { case 2: LINE_CHECK("authmode WPA2/802.11i"); break; case 3: LINE_CHECK("authmode WPA1+WPA2/802.11i"); break; default: LINE_CHECK("authmode WPA"); break; } break; case IEEE80211_AUTH_AUTO: LINE_CHECK("authmode AUTO"); break; default: LINE_CHECK("authmode UNKNOWN (0x%x)", val); break; } } if (wpa || verbose) { if (get80211val(s, IEEE80211_IOC_WPS, &val) != -1) { if (val) LINE_CHECK("wps"); else if (verbose) LINE_CHECK("-wps"); } if (get80211val(s, IEEE80211_IOC_TSN, &val) != -1) { if (val) LINE_CHECK("tsn"); else if (verbose) LINE_CHECK("-tsn"); } if (ioctl(s, IEEE80211_IOC_COUNTERMEASURES, &val) != -1) { if (val) LINE_CHECK("countermeasures"); else if (verbose) LINE_CHECK("-countermeasures"); } #if 0 /* XXX not interesting with WPA done in user space */ ireq.i_type = IEEE80211_IOC_KEYMGTALGS; if (ioctl(s, SIOCG80211, &ireq) != -1) { } ireq.i_type = IEEE80211_IOC_MCASTCIPHER; if (ioctl(s, SIOCG80211, &ireq) != -1) { LINE_CHECK("mcastcipher "); printcipher(s, &ireq, IEEE80211_IOC_MCASTKEYLEN); spacer = ' '; } ireq.i_type = IEEE80211_IOC_UCASTCIPHER; if (ioctl(s, SIOCG80211, &ireq) != -1) { LINE_CHECK("ucastcipher "); printcipher(s, &ireq, IEEE80211_IOC_UCASTKEYLEN); } if (wpa & 2) { ireq.i_type = IEEE80211_IOC_RSNCAPS; if (ioctl(s, SIOCG80211, &ireq) != -1) { LINE_CHECK("RSN caps 0x%x", ireq.i_val); spacer = ' '; } } ireq.i_type = IEEE80211_IOC_UCASTCIPHERS; if (ioctl(s, SIOCG80211, &ireq) != -1) { } #endif } if (get80211val(s, IEEE80211_IOC_WEP, &wepmode) != -1 && wepmode != IEEE80211_WEP_NOSUP) { switch (wepmode) { case IEEE80211_WEP_OFF: LINE_CHECK("privacy OFF"); break; case IEEE80211_WEP_ON: LINE_CHECK("privacy ON"); break; case IEEE80211_WEP_MIXED: LINE_CHECK("privacy MIXED"); break; default: LINE_CHECK("privacy UNKNOWN (0x%x)", wepmode); break; } /* * If we get here then we've got WEP support so we need * to print WEP status. */ if (get80211val(s, IEEE80211_IOC_WEPTXKEY, &val) < 0) { warn("WEP support, but no tx key!"); goto end; } if (val != -1) LINE_CHECK("deftxkey %d", val+1); else if (wepmode != IEEE80211_WEP_OFF || verbose) LINE_CHECK("deftxkey UNDEF"); if (get80211val(s, IEEE80211_IOC_NUMWEPKEYS, &num) < 0) { warn("WEP support, but no NUMWEPKEYS support!"); goto end; } for (i = 0; i < num; i++) { struct ieee80211req_key ik; memset(&ik, 0, sizeof(ik)); ik.ik_keyix = i; if (get80211(s, IEEE80211_IOC_WPAKEY, &ik, sizeof(ik)) < 0) { warn("WEP support, but can get keys!"); goto end; } if (ik.ik_keylen != 0) { if (verbose) LINE_BREAK(); printkey(&ik); } } end: ; } if (get80211val(s, IEEE80211_IOC_POWERSAVE, &val) != -1 && val != IEEE80211_POWERSAVE_NOSUP ) { if (val != IEEE80211_POWERSAVE_OFF || verbose) { switch (val) { case IEEE80211_POWERSAVE_OFF: LINE_CHECK("powersavemode OFF"); break; case IEEE80211_POWERSAVE_CAM: LINE_CHECK("powersavemode CAM"); break; case IEEE80211_POWERSAVE_PSP: LINE_CHECK("powersavemode PSP"); break; case IEEE80211_POWERSAVE_PSP_CAM: LINE_CHECK("powersavemode PSP-CAM"); break; } if (get80211val(s, IEEE80211_IOC_POWERSAVESLEEP, &val) != -1) LINE_CHECK("powersavesleep %d", val); } } if (get80211val(s, IEEE80211_IOC_TXPOWER, &val) != -1) { if (val & 1) LINE_CHECK("txpower %d.5", val/2); else LINE_CHECK("txpower %d", val/2); } if (verbose) { if (get80211val(s, IEEE80211_IOC_TXPOWMAX, &val) != -1) LINE_CHECK("txpowmax %.1f", val/2.); } if (get80211val(s, IEEE80211_IOC_DOTD, &val) != -1) { if (val) LINE_CHECK("dotd"); else if (verbose) LINE_CHECK("-dotd"); } if (get80211val(s, IEEE80211_IOC_RTSTHRESHOLD, &val) != -1) { if (val != IEEE80211_RTS_MAX || verbose) LINE_CHECK("rtsthreshold %d", val); } if (get80211val(s, IEEE80211_IOC_FRAGTHRESHOLD, &val) != -1) { if (val != IEEE80211_FRAG_MAX || verbose) LINE_CHECK("fragthreshold %d", val); } if (opmode == IEEE80211_M_STA || verbose) { if (get80211val(s, IEEE80211_IOC_BMISSTHRESHOLD, &val) != -1) { if (val != IEEE80211_HWBMISS_MAX || verbose) LINE_CHECK("bmiss %d", val); } } if (!verbose) { gettxparams(s); tp = &txparams.params[chan2mode(c)]; printrate("ucastrate", tp->ucastrate, IEEE80211_FIXED_RATE_NONE, IEEE80211_FIXED_RATE_NONE); printrate("mcastrate", tp->mcastrate, 2*1, IEEE80211_RATE_MCS|0); printrate("mgmtrate", tp->mgmtrate, 2*1, IEEE80211_RATE_MCS|0); if (tp->maxretry != 6) /* XXX */ LINE_CHECK("maxretry %d", tp->maxretry); } else { LINE_BREAK(); list_txparams(s); } bgscaninterval = -1; (void) get80211val(s, IEEE80211_IOC_BGSCAN_INTERVAL, &bgscaninterval); if (get80211val(s, IEEE80211_IOC_SCANVALID, &val) != -1) { if (val != bgscaninterval || verbose) LINE_CHECK("scanvalid %u", val); } bgscan = 0; if (get80211val(s, IEEE80211_IOC_BGSCAN, &bgscan) != -1) { if (bgscan) LINE_CHECK("bgscan"); else if (verbose) LINE_CHECK("-bgscan"); } if (bgscan || verbose) { if (bgscaninterval != -1) LINE_CHECK("bgscanintvl %u", bgscaninterval); if (get80211val(s, IEEE80211_IOC_BGSCAN_IDLE, &val) != -1) LINE_CHECK("bgscanidle %u", val); if (!verbose) { getroam(s); rp = &roamparams.params[chan2mode(c)]; if (rp->rssi & 1) LINE_CHECK("roam:rssi %u.5", rp->rssi/2); else LINE_CHECK("roam:rssi %u", rp->rssi/2); LINE_CHECK("roam:rate %s%u", (rp->rate & IEEE80211_RATE_MCS) ? "MCS " : "", get_rate_value(rp->rate)); } else { LINE_BREAK(); list_roam(s); LINE_BREAK(); } } if (IEEE80211_IS_CHAN_ANYG(c) || verbose) { if (get80211val(s, IEEE80211_IOC_PUREG, &val) != -1) { if (val) LINE_CHECK("pureg"); else if (verbose) LINE_CHECK("-pureg"); } if (get80211val(s, IEEE80211_IOC_PROTMODE, &val) != -1) { switch (val) { case IEEE80211_PROTMODE_OFF: LINE_CHECK("protmode OFF"); break; case IEEE80211_PROTMODE_CTS: LINE_CHECK("protmode CTS"); break; case IEEE80211_PROTMODE_RTSCTS: LINE_CHECK("protmode RTSCTS"); break; default: LINE_CHECK("protmode UNKNOWN (0x%x)", val); break; } } } if (IEEE80211_IS_CHAN_HT(c) || verbose) { gethtconf(s); switch (htconf & 3) { case 0: case 2: LINE_CHECK("-ht"); break; case 1: LINE_CHECK("ht20"); break; case 3: if (verbose) LINE_CHECK("ht"); break; } if (get80211val(s, IEEE80211_IOC_HTCOMPAT, &val) != -1) { if (!val) LINE_CHECK("-htcompat"); else if (verbose) LINE_CHECK("htcompat"); } if (get80211val(s, IEEE80211_IOC_AMPDU, &val) != -1) { switch (val) { case 0: LINE_CHECK("-ampdu"); break; case 1: LINE_CHECK("ampdutx -ampdurx"); break; case 2: LINE_CHECK("-ampdutx ampdurx"); break; case 3: if (verbose) LINE_CHECK("ampdu"); break; } } /* XXX 11ac density/size is different */ if (get80211val(s, IEEE80211_IOC_AMPDU_LIMIT, &val) != -1) { switch (val) { case IEEE80211_HTCAP_MAXRXAMPDU_8K: LINE_CHECK("ampdulimit 8k"); break; case IEEE80211_HTCAP_MAXRXAMPDU_16K: LINE_CHECK("ampdulimit 16k"); break; case IEEE80211_HTCAP_MAXRXAMPDU_32K: LINE_CHECK("ampdulimit 32k"); break; case IEEE80211_HTCAP_MAXRXAMPDU_64K: LINE_CHECK("ampdulimit 64k"); break; } } /* XXX 11ac density/size is different */ if (get80211val(s, IEEE80211_IOC_AMPDU_DENSITY, &val) != -1) { switch (val) { case IEEE80211_HTCAP_MPDUDENSITY_NA: if (verbose) LINE_CHECK("ampdudensity NA"); break; case IEEE80211_HTCAP_MPDUDENSITY_025: LINE_CHECK("ampdudensity .25"); break; case IEEE80211_HTCAP_MPDUDENSITY_05: LINE_CHECK("ampdudensity .5"); break; case IEEE80211_HTCAP_MPDUDENSITY_1: LINE_CHECK("ampdudensity 1"); break; case IEEE80211_HTCAP_MPDUDENSITY_2: LINE_CHECK("ampdudensity 2"); break; case IEEE80211_HTCAP_MPDUDENSITY_4: LINE_CHECK("ampdudensity 4"); break; case IEEE80211_HTCAP_MPDUDENSITY_8: LINE_CHECK("ampdudensity 8"); break; case IEEE80211_HTCAP_MPDUDENSITY_16: LINE_CHECK("ampdudensity 16"); break; } } if (get80211val(s, IEEE80211_IOC_AMSDU, &val) != -1) { switch (val) { case 0: LINE_CHECK("-amsdu"); break; case 1: LINE_CHECK("amsdutx -amsdurx"); break; case 2: LINE_CHECK("-amsdutx amsdurx"); break; case 3: if (verbose) LINE_CHECK("amsdu"); break; } } /* XXX amsdu limit */ if (get80211val(s, IEEE80211_IOC_SHORTGI, &val) != -1) { if (val) LINE_CHECK("shortgi"); else if (verbose) LINE_CHECK("-shortgi"); } if (get80211val(s, IEEE80211_IOC_HTPROTMODE, &val) != -1) { if (val == IEEE80211_PROTMODE_OFF) LINE_CHECK("htprotmode OFF"); else if (val != IEEE80211_PROTMODE_RTSCTS) LINE_CHECK("htprotmode UNKNOWN (0x%x)", val); else if (verbose) LINE_CHECK("htprotmode RTSCTS"); } if (get80211val(s, IEEE80211_IOC_PUREN, &val) != -1) { if (val) LINE_CHECK("puren"); else if (verbose) LINE_CHECK("-puren"); } if (get80211val(s, IEEE80211_IOC_SMPS, &val) != -1) { if (val == IEEE80211_HTCAP_SMPS_DYNAMIC) LINE_CHECK("smpsdyn"); else if (val == IEEE80211_HTCAP_SMPS_ENA) LINE_CHECK("smps"); else if (verbose) LINE_CHECK("-smps"); } if (get80211val(s, IEEE80211_IOC_RIFS, &val) != -1) { if (val) LINE_CHECK("rifs"); else if (verbose) LINE_CHECK("-rifs"); } /* XXX VHT STBC? */ if (get80211val(s, IEEE80211_IOC_STBC, &val) != -1) { switch (val) { case 0: LINE_CHECK("-stbc"); break; case 1: LINE_CHECK("stbctx -stbcrx"); break; case 2: LINE_CHECK("-stbctx stbcrx"); break; case 3: if (verbose) LINE_CHECK("stbc"); break; } } if (get80211val(s, IEEE80211_IOC_LDPC, &val) != -1) { switch (val) { case 0: LINE_CHECK("-ldpc"); break; case 1: LINE_CHECK("ldpctx -ldpcrx"); break; case 2: LINE_CHECK("-ldpctx ldpcrx"); break; case 3: if (verbose) LINE_CHECK("ldpc"); break; } } if (get80211val(s, IEEE80211_IOC_UAPSD, &val) != -1) { switch (val) { case 0: LINE_CHECK("-uapsd"); break; case 1: LINE_CHECK("uapsd"); break; } } } if (IEEE80211_IS_CHAN_VHT(c) || verbose) { getvhtconf(s); if (vhtconf & 0x1) LINE_CHECK("vht"); else LINE_CHECK("-vht"); if (vhtconf & 0x2) LINE_CHECK("vht40"); else LINE_CHECK("-vht40"); if (vhtconf & 0x4) LINE_CHECK("vht80"); else LINE_CHECK("-vht80"); if (vhtconf & 0x8) LINE_CHECK("vht80p80"); else LINE_CHECK("-vht80p80"); if (vhtconf & 0x10) LINE_CHECK("vht160"); else LINE_CHECK("-vht160"); } if (get80211val(s, IEEE80211_IOC_WME, &wme) != -1) { if (wme) LINE_CHECK("wme"); else if (verbose) LINE_CHECK("-wme"); } else wme = 0; if (get80211val(s, IEEE80211_IOC_BURST, &val) != -1) { if (val) LINE_CHECK("burst"); else if (verbose) LINE_CHECK("-burst"); } if (get80211val(s, IEEE80211_IOC_FF, &val) != -1) { if (val) LINE_CHECK("ff"); else if (verbose) LINE_CHECK("-ff"); } if (get80211val(s, IEEE80211_IOC_TURBOP, &val) != -1) { if (val) LINE_CHECK("dturbo"); else if (verbose) LINE_CHECK("-dturbo"); } if (get80211val(s, IEEE80211_IOC_DWDS, &val) != -1) { if (val) LINE_CHECK("dwds"); else if (verbose) LINE_CHECK("-dwds"); } if (opmode == IEEE80211_M_HOSTAP) { if (get80211val(s, IEEE80211_IOC_HIDESSID, &val) != -1) { if (val) LINE_CHECK("hidessid"); else if (verbose) LINE_CHECK("-hidessid"); } if (get80211val(s, IEEE80211_IOC_APBRIDGE, &val) != -1) { if (!val) LINE_CHECK("-apbridge"); else if (verbose) LINE_CHECK("apbridge"); } if (get80211val(s, IEEE80211_IOC_DTIM_PERIOD, &val) != -1) LINE_CHECK("dtimperiod %u", val); if (get80211val(s, IEEE80211_IOC_DOTH, &val) != -1) { if (!val) LINE_CHECK("-doth"); else if (verbose) LINE_CHECK("doth"); } if (get80211val(s, IEEE80211_IOC_DFS, &val) != -1) { if (!val) LINE_CHECK("-dfs"); else if (verbose) LINE_CHECK("dfs"); } if (get80211val(s, IEEE80211_IOC_INACTIVITY, &val) != -1) { if (!val) LINE_CHECK("-inact"); else if (verbose) LINE_CHECK("inact"); } } else { if (get80211val(s, IEEE80211_IOC_ROAMING, &val) != -1) { if (val != IEEE80211_ROAMING_AUTO || verbose) { switch (val) { case IEEE80211_ROAMING_DEVICE: LINE_CHECK("roaming DEVICE"); break; case IEEE80211_ROAMING_AUTO: LINE_CHECK("roaming AUTO"); break; case IEEE80211_ROAMING_MANUAL: LINE_CHECK("roaming MANUAL"); break; default: LINE_CHECK("roaming UNKNOWN (0x%x)", val); break; } } } } if (opmode == IEEE80211_M_AHDEMO) { if (get80211val(s, IEEE80211_IOC_TDMA_SLOT, &val) != -1) LINE_CHECK("tdmaslot %u", val); if (get80211val(s, IEEE80211_IOC_TDMA_SLOTCNT, &val) != -1) LINE_CHECK("tdmaslotcnt %u", val); if (get80211val(s, IEEE80211_IOC_TDMA_SLOTLEN, &val) != -1) LINE_CHECK("tdmaslotlen %u", val); if (get80211val(s, IEEE80211_IOC_TDMA_BINTERVAL, &val) != -1) LINE_CHECK("tdmabintval %u", val); } else if (get80211val(s, IEEE80211_IOC_BEACON_INTERVAL, &val) != -1) { /* XXX default define not visible */ if (val != 100 || verbose) LINE_CHECK("bintval %u", val); } if (wme && verbose) { LINE_BREAK(); list_wme(s); } if (opmode == IEEE80211_M_MBSS) { if (get80211val(s, IEEE80211_IOC_MESH_TTL, &val) != -1) { LINE_CHECK("meshttl %u", val); } if (get80211val(s, IEEE80211_IOC_MESH_AP, &val) != -1) { if (val) LINE_CHECK("meshpeering"); else LINE_CHECK("-meshpeering"); } if (get80211val(s, IEEE80211_IOC_MESH_FWRD, &val) != -1) { if (val) LINE_CHECK("meshforward"); else LINE_CHECK("-meshforward"); } if (get80211val(s, IEEE80211_IOC_MESH_GATE, &val) != -1) { if (val) LINE_CHECK("meshgate"); else LINE_CHECK("-meshgate"); } if (get80211len(s, IEEE80211_IOC_MESH_PR_METRIC, data, 12, &len) != -1) { data[len] = '\0'; LINE_CHECK("meshmetric %s", data); } if (get80211len(s, IEEE80211_IOC_MESH_PR_PATH, data, 12, &len) != -1) { data[len] = '\0'; LINE_CHECK("meshpath %s", data); } if (get80211val(s, IEEE80211_IOC_HWMP_ROOTMODE, &val) != -1) { switch (val) { case IEEE80211_HWMP_ROOTMODE_DISABLED: LINE_CHECK("hwmprootmode DISABLED"); break; case IEEE80211_HWMP_ROOTMODE_NORMAL: LINE_CHECK("hwmprootmode NORMAL"); break; case IEEE80211_HWMP_ROOTMODE_PROACTIVE: LINE_CHECK("hwmprootmode PROACTIVE"); break; case IEEE80211_HWMP_ROOTMODE_RANN: LINE_CHECK("hwmprootmode RANN"); break; default: LINE_CHECK("hwmprootmode UNKNOWN(%d)", val); break; } } if (get80211val(s, IEEE80211_IOC_HWMP_MAXHOPS, &val) != -1) { LINE_CHECK("hwmpmaxhops %u", val); } } + + LINE_BREAK(); + + if (getdevicename(s, data, sizeof(data), &len) < 0) + return; + LINE_CHECK("parent interface: %s", data); LINE_BREAK(); } static int get80211(int s, int type, void *data, int len) { return (lib80211_get80211(s, name, type, data, len)); } static int get80211len(int s, int type, void *data, int len, int *plen) { return (lib80211_get80211len(s, name, type, data, len, plen)); } static int get80211val(int s, int type, int *val) { return (lib80211_get80211val(s, name, type, val)); } static void set80211(int s, int type, int val, int len, void *data) { int ret; ret = lib80211_set80211(s, name, type, val, len, data); if (ret < 0) err(1, "SIOCS80211"); } static const char * get_string(const char *val, const char *sep, u_int8_t *buf, int *lenp) { int len; int hexstr; u_int8_t *p; len = *lenp; p = buf; hexstr = (val[0] == '0' && tolower((u_char)val[1]) == 'x'); if (hexstr) val += 2; for (;;) { if (*val == '\0') break; if (sep != NULL && strchr(sep, *val) != NULL) { val++; break; } if (hexstr) { if (!isxdigit((u_char)val[0])) { warnx("bad hexadecimal digits"); return NULL; } if (!isxdigit((u_char)val[1])) { warnx("odd count hexadecimal digits"); return NULL; } } if (p >= buf + len) { if (hexstr) warnx("hexadecimal digits too long"); else warnx("string too long"); return NULL; } if (hexstr) { #define tohex(x) (isdigit(x) ? (x) - '0' : tolower(x) - 'a' + 10) *p++ = (tohex((u_char)val[0]) << 4) | tohex((u_char)val[1]); #undef tohex val += 2; } else *p++ = *val++; } len = p - buf; /* The string "-" is treated as the empty string. */ if (!hexstr && len == 1 && buf[0] == '-') { len = 0; memset(buf, 0, *lenp); } else if (len < *lenp) memset(p, 0, *lenp - len); *lenp = len; return val; } static void print_string(const u_int8_t *buf, int len) { int i; int hasspc; int utf8; i = 0; hasspc = 0; setlocale(LC_CTYPE, ""); utf8 = strncmp("UTF-8", nl_langinfo(CODESET), 5) == 0; for (; i < len; i++) { if (!isprint(buf[i]) && buf[i] != '\0' && !utf8) break; if (isspace(buf[i])) hasspc++; } if (i == len || utf8) { if (hasspc || len == 0 || buf[0] == '\0') printf("\"%.*s\"", len, buf); else printf("%.*s", len, buf); } else { printf("0x"); for (i = 0; i < len; i++) printf("%02x", buf[i]); } } static void setdefregdomain(int s) { struct regdata *rdp = getregdata(); const struct regdomain *rd; /* Check if regdomain/country was already set by a previous call. */ /* XXX is it possible? */ if (regdomain.regdomain != 0 || regdomain.country != CTRY_DEFAULT) return; getregdomain(s); /* Check if it was already set by the driver. */ if (regdomain.regdomain != 0 || regdomain.country != CTRY_DEFAULT) return; /* Set FCC/US as default. */ rd = lib80211_regdomain_findbysku(rdp, SKU_FCC); if (rd == NULL) errx(1, "FCC regdomain was not found"); regdomain.regdomain = rd->sku; if (rd->cc != NULL) defaultcountry(rd); /* Send changes to net80211. */ setregdomain_cb(s, ®domain); /* Cleanup (so it can be overriden by subsequent parameters). */ regdomain.regdomain = 0; regdomain.country = CTRY_DEFAULT; regdomain.isocc[0] = 0; regdomain.isocc[1] = 0; } /* * Virtual AP cloning support. */ static struct ieee80211_clone_params params = { .icp_opmode = IEEE80211_M_STA, /* default to station mode */ }; static void wlan_create(int s, struct ifreq *ifr) { static const uint8_t zerobssid[IEEE80211_ADDR_LEN]; char orig_name[IFNAMSIZ]; if (params.icp_parent[0] == '\0') errx(1, "must specify a parent device (wlandev) when creating " "a wlan device"); if (params.icp_opmode == IEEE80211_M_WDS && memcmp(params.icp_bssid, zerobssid, sizeof(zerobssid)) == 0) errx(1, "no bssid specified for WDS (use wlanbssid)"); ifr->ifr_data = (caddr_t) ¶ms; if (ioctl(s, SIOCIFCREATE2, ifr) < 0) err(1, "SIOCIFCREATE2"); /* XXX preserve original name for ifclonecreate(). */ strlcpy(orig_name, name, sizeof(orig_name)); strlcpy(name, ifr->ifr_name, sizeof(name)); setdefregdomain(s); strlcpy(name, orig_name, sizeof(name)); } static DECL_CMD_FUNC(set80211clone_wlandev, arg, d) { strlcpy(params.icp_parent, arg, IFNAMSIZ); } static DECL_CMD_FUNC(set80211clone_wlanbssid, arg, d) { const struct ether_addr *ea; ea = ether_aton(arg); if (ea == NULL) errx(1, "%s: cannot parse bssid", arg); memcpy(params.icp_bssid, ea->octet, IEEE80211_ADDR_LEN); } static DECL_CMD_FUNC(set80211clone_wlanaddr, arg, d) { const struct ether_addr *ea; ea = ether_aton(arg); if (ea == NULL) errx(1, "%s: cannot parse address", arg); memcpy(params.icp_macaddr, ea->octet, IEEE80211_ADDR_LEN); params.icp_flags |= IEEE80211_CLONE_MACADDR; } static DECL_CMD_FUNC(set80211clone_wlanmode, arg, d) { #define iseq(a,b) (strncasecmp(a,b,sizeof(b)-1) == 0) if (iseq(arg, "sta")) params.icp_opmode = IEEE80211_M_STA; else if (iseq(arg, "ahdemo") || iseq(arg, "adhoc-demo")) params.icp_opmode = IEEE80211_M_AHDEMO; else if (iseq(arg, "ibss") || iseq(arg, "adhoc")) params.icp_opmode = IEEE80211_M_IBSS; else if (iseq(arg, "ap") || iseq(arg, "host")) params.icp_opmode = IEEE80211_M_HOSTAP; else if (iseq(arg, "wds")) params.icp_opmode = IEEE80211_M_WDS; else if (iseq(arg, "monitor")) params.icp_opmode = IEEE80211_M_MONITOR; else if (iseq(arg, "tdma")) { params.icp_opmode = IEEE80211_M_AHDEMO; params.icp_flags |= IEEE80211_CLONE_TDMA; } else if (iseq(arg, "mesh") || iseq(arg, "mp")) /* mesh point */ params.icp_opmode = IEEE80211_M_MBSS; else errx(1, "Don't know to create %s for %s", arg, name); #undef iseq } static void set80211clone_beacons(const char *val, int d, int s, const struct afswtch *rafp) { /* NB: inverted sense */ if (d) params.icp_flags &= ~IEEE80211_CLONE_NOBEACONS; else params.icp_flags |= IEEE80211_CLONE_NOBEACONS; } static void set80211clone_bssid(const char *val, int d, int s, const struct afswtch *rafp) { if (d) params.icp_flags |= IEEE80211_CLONE_BSSID; else params.icp_flags &= ~IEEE80211_CLONE_BSSID; } static void set80211clone_wdslegacy(const char *val, int d, int s, const struct afswtch *rafp) { if (d) params.icp_flags |= IEEE80211_CLONE_WDSLEGACY; else params.icp_flags &= ~IEEE80211_CLONE_WDSLEGACY; } static struct cmd ieee80211_cmds[] = { DEF_CMD_ARG("ssid", set80211ssid), DEF_CMD_ARG("nwid", set80211ssid), DEF_CMD_ARG("meshid", set80211meshid), DEF_CMD_ARG("stationname", set80211stationname), DEF_CMD_ARG("station", set80211stationname), /* BSD/OS */ DEF_CMD_ARG("channel", set80211channel), DEF_CMD_ARG("authmode", set80211authmode), DEF_CMD_ARG("powersavemode", set80211powersavemode), DEF_CMD("powersave", 1, set80211powersave), DEF_CMD("-powersave", 0, set80211powersave), DEF_CMD_ARG("powersavesleep", set80211powersavesleep), DEF_CMD_ARG("wepmode", set80211wepmode), DEF_CMD("wep", 1, set80211wep), DEF_CMD("-wep", 0, set80211wep), DEF_CMD_ARG("deftxkey", set80211weptxkey), DEF_CMD_ARG("weptxkey", set80211weptxkey), DEF_CMD_ARG("wepkey", set80211wepkey), DEF_CMD_ARG("nwkey", set80211nwkey), /* NetBSD */ DEF_CMD("-nwkey", 0, set80211wep), /* NetBSD */ DEF_CMD_ARG("rtsthreshold", set80211rtsthreshold), DEF_CMD_ARG("protmode", set80211protmode), DEF_CMD_ARG("txpower", set80211txpower), DEF_CMD_ARG("roaming", set80211roaming), DEF_CMD("wme", 1, set80211wme), DEF_CMD("-wme", 0, set80211wme), DEF_CMD("wmm", 1, set80211wme), DEF_CMD("-wmm", 0, set80211wme), DEF_CMD("hidessid", 1, set80211hidessid), DEF_CMD("-hidessid", 0, set80211hidessid), DEF_CMD("apbridge", 1, set80211apbridge), DEF_CMD("-apbridge", 0, set80211apbridge), DEF_CMD_ARG("chanlist", set80211chanlist), DEF_CMD_ARG("bssid", set80211bssid), DEF_CMD_ARG("ap", set80211bssid), DEF_CMD("scan", 0, set80211scan), DEF_CMD_ARG("list", set80211list), DEF_CMD_ARG2("cwmin", set80211cwmin), DEF_CMD_ARG2("cwmax", set80211cwmax), DEF_CMD_ARG2("aifs", set80211aifs), DEF_CMD_ARG2("txoplimit", set80211txoplimit), DEF_CMD_ARG("acm", set80211acm), DEF_CMD_ARG("-acm", set80211noacm), DEF_CMD_ARG("ack", set80211ackpolicy), DEF_CMD_ARG("-ack", set80211noackpolicy), DEF_CMD_ARG2("bss:cwmin", set80211bsscwmin), DEF_CMD_ARG2("bss:cwmax", set80211bsscwmax), DEF_CMD_ARG2("bss:aifs", set80211bssaifs), DEF_CMD_ARG2("bss:txoplimit", set80211bsstxoplimit), DEF_CMD_ARG("dtimperiod", set80211dtimperiod), DEF_CMD_ARG("bintval", set80211bintval), DEF_CMD("mac:open", IEEE80211_MACCMD_POLICY_OPEN, set80211maccmd), DEF_CMD("mac:allow", IEEE80211_MACCMD_POLICY_ALLOW, set80211maccmd), DEF_CMD("mac:deny", IEEE80211_MACCMD_POLICY_DENY, set80211maccmd), DEF_CMD("mac:radius", IEEE80211_MACCMD_POLICY_RADIUS, set80211maccmd), DEF_CMD("mac:flush", IEEE80211_MACCMD_FLUSH, set80211maccmd), DEF_CMD("mac:detach", IEEE80211_MACCMD_DETACH, set80211maccmd), DEF_CMD_ARG("mac:add", set80211addmac), DEF_CMD_ARG("mac:del", set80211delmac), DEF_CMD_ARG("mac:kick", set80211kickmac), DEF_CMD("pureg", 1, set80211pureg), DEF_CMD("-pureg", 0, set80211pureg), DEF_CMD("ff", 1, set80211fastframes), DEF_CMD("-ff", 0, set80211fastframes), DEF_CMD("dturbo", 1, set80211dturbo), DEF_CMD("-dturbo", 0, set80211dturbo), DEF_CMD("bgscan", 1, set80211bgscan), DEF_CMD("-bgscan", 0, set80211bgscan), DEF_CMD_ARG("bgscanidle", set80211bgscanidle), DEF_CMD_ARG("bgscanintvl", set80211bgscanintvl), DEF_CMD_ARG("scanvalid", set80211scanvalid), DEF_CMD("quiet", 1, set80211quiet), DEF_CMD("-quiet", 0, set80211quiet), DEF_CMD_ARG("quiet_count", set80211quietcount), DEF_CMD_ARG("quiet_period", set80211quietperiod), DEF_CMD_ARG("quiet_duration", set80211quietduration), DEF_CMD_ARG("quiet_offset", set80211quietoffset), DEF_CMD_ARG("roam:rssi", set80211roamrssi), DEF_CMD_ARG("roam:rate", set80211roamrate), DEF_CMD_ARG("mcastrate", set80211mcastrate), DEF_CMD_ARG("ucastrate", set80211ucastrate), DEF_CMD_ARG("mgtrate", set80211mgtrate), DEF_CMD_ARG("mgmtrate", set80211mgtrate), DEF_CMD_ARG("maxretry", set80211maxretry), DEF_CMD_ARG("fragthreshold", set80211fragthreshold), DEF_CMD("burst", 1, set80211burst), DEF_CMD("-burst", 0, set80211burst), DEF_CMD_ARG("bmiss", set80211bmissthreshold), DEF_CMD_ARG("bmissthreshold", set80211bmissthreshold), DEF_CMD("shortgi", 1, set80211shortgi), DEF_CMD("-shortgi", 0, set80211shortgi), DEF_CMD("ampdurx", 2, set80211ampdu), DEF_CMD("-ampdurx", -2, set80211ampdu), DEF_CMD("ampdutx", 1, set80211ampdu), DEF_CMD("-ampdutx", -1, set80211ampdu), DEF_CMD("ampdu", 3, set80211ampdu), /* NB: tx+rx */ DEF_CMD("-ampdu", -3, set80211ampdu), DEF_CMD_ARG("ampdulimit", set80211ampdulimit), DEF_CMD_ARG("ampdudensity", set80211ampdudensity), DEF_CMD("amsdurx", 2, set80211amsdu), DEF_CMD("-amsdurx", -2, set80211amsdu), DEF_CMD("amsdutx", 1, set80211amsdu), DEF_CMD("-amsdutx", -1, set80211amsdu), DEF_CMD("amsdu", 3, set80211amsdu), /* NB: tx+rx */ DEF_CMD("-amsdu", -3, set80211amsdu), DEF_CMD_ARG("amsdulimit", set80211amsdulimit), DEF_CMD("stbcrx", 2, set80211stbc), DEF_CMD("-stbcrx", -2, set80211stbc), DEF_CMD("stbctx", 1, set80211stbc), DEF_CMD("-stbctx", -1, set80211stbc), DEF_CMD("stbc", 3, set80211stbc), /* NB: tx+rx */ DEF_CMD("-stbc", -3, set80211stbc), DEF_CMD("ldpcrx", 2, set80211ldpc), DEF_CMD("-ldpcrx", -2, set80211ldpc), DEF_CMD("ldpctx", 1, set80211ldpc), DEF_CMD("-ldpctx", -1, set80211ldpc), DEF_CMD("ldpc", 3, set80211ldpc), /* NB: tx+rx */ DEF_CMD("-ldpc", -3, set80211ldpc), DEF_CMD("uapsd", 1, set80211uapsd), DEF_CMD("-uapsd", 0, set80211uapsd), DEF_CMD("puren", 1, set80211puren), DEF_CMD("-puren", 0, set80211puren), DEF_CMD("doth", 1, set80211doth), DEF_CMD("-doth", 0, set80211doth), DEF_CMD("dfs", 1, set80211dfs), DEF_CMD("-dfs", 0, set80211dfs), DEF_CMD("htcompat", 1, set80211htcompat), DEF_CMD("-htcompat", 0, set80211htcompat), DEF_CMD("dwds", 1, set80211dwds), DEF_CMD("-dwds", 0, set80211dwds), DEF_CMD("inact", 1, set80211inact), DEF_CMD("-inact", 0, set80211inact), DEF_CMD("tsn", 1, set80211tsn), DEF_CMD("-tsn", 0, set80211tsn), DEF_CMD_ARG("regdomain", set80211regdomain), DEF_CMD_ARG("country", set80211country), DEF_CMD("indoor", 'I', set80211location), DEF_CMD("-indoor", 'O', set80211location), DEF_CMD("outdoor", 'O', set80211location), DEF_CMD("-outdoor", 'I', set80211location), DEF_CMD("anywhere", ' ', set80211location), DEF_CMD("ecm", 1, set80211ecm), DEF_CMD("-ecm", 0, set80211ecm), DEF_CMD("dotd", 1, set80211dotd), DEF_CMD("-dotd", 0, set80211dotd), DEF_CMD_ARG("htprotmode", set80211htprotmode), DEF_CMD("ht20", 1, set80211htconf), DEF_CMD("-ht20", 0, set80211htconf), DEF_CMD("ht40", 3, set80211htconf), /* NB: 20+40 */ DEF_CMD("-ht40", 0, set80211htconf), DEF_CMD("ht", 3, set80211htconf), /* NB: 20+40 */ DEF_CMD("-ht", 0, set80211htconf), DEF_CMD("vht", 1, set80211vhtconf), DEF_CMD("-vht", 0, set80211vhtconf), DEF_CMD("vht40", 2, set80211vhtconf), DEF_CMD("-vht40", -2, set80211vhtconf), DEF_CMD("vht80", 4, set80211vhtconf), DEF_CMD("-vht80", -4, set80211vhtconf), DEF_CMD("vht80p80", 8, set80211vhtconf), DEF_CMD("-vht80p80", -8, set80211vhtconf), DEF_CMD("vht160", 16, set80211vhtconf), DEF_CMD("-vht160", -16, set80211vhtconf), DEF_CMD("rifs", 1, set80211rifs), DEF_CMD("-rifs", 0, set80211rifs), DEF_CMD("smps", IEEE80211_HTCAP_SMPS_ENA, set80211smps), DEF_CMD("smpsdyn", IEEE80211_HTCAP_SMPS_DYNAMIC, set80211smps), DEF_CMD("-smps", IEEE80211_HTCAP_SMPS_OFF, set80211smps), /* XXX for testing */ DEF_CMD_ARG("chanswitch", set80211chanswitch), DEF_CMD_ARG("tdmaslot", set80211tdmaslot), DEF_CMD_ARG("tdmaslotcnt", set80211tdmaslotcnt), DEF_CMD_ARG("tdmaslotlen", set80211tdmaslotlen), DEF_CMD_ARG("tdmabintval", set80211tdmabintval), DEF_CMD_ARG("meshttl", set80211meshttl), DEF_CMD("meshforward", 1, set80211meshforward), DEF_CMD("-meshforward", 0, set80211meshforward), DEF_CMD("meshgate", 1, set80211meshgate), DEF_CMD("-meshgate", 0, set80211meshgate), DEF_CMD("meshpeering", 1, set80211meshpeering), DEF_CMD("-meshpeering", 0, set80211meshpeering), DEF_CMD_ARG("meshmetric", set80211meshmetric), DEF_CMD_ARG("meshpath", set80211meshpath), DEF_CMD("meshrt:flush", IEEE80211_MESH_RTCMD_FLUSH, set80211meshrtcmd), DEF_CMD_ARG("meshrt:add", set80211addmeshrt), DEF_CMD_ARG("meshrt:del", set80211delmeshrt), DEF_CMD_ARG("hwmprootmode", set80211hwmprootmode), DEF_CMD_ARG("hwmpmaxhops", set80211hwmpmaxhops), /* vap cloning support */ DEF_CLONE_CMD_ARG("wlanaddr", set80211clone_wlanaddr), DEF_CLONE_CMD_ARG("wlanbssid", set80211clone_wlanbssid), DEF_CLONE_CMD_ARG("wlandev", set80211clone_wlandev), DEF_CLONE_CMD_ARG("wlanmode", set80211clone_wlanmode), DEF_CLONE_CMD("beacons", 1, set80211clone_beacons), DEF_CLONE_CMD("-beacons", 0, set80211clone_beacons), DEF_CLONE_CMD("bssid", 1, set80211clone_bssid), DEF_CLONE_CMD("-bssid", 0, set80211clone_bssid), DEF_CLONE_CMD("wdslegacy", 1, set80211clone_wdslegacy), DEF_CLONE_CMD("-wdslegacy", 0, set80211clone_wdslegacy), }; static struct afswtch af_ieee80211 = { .af_name = "af_ieee80211", .af_af = AF_UNSPEC, .af_other_status = ieee80211_status, }; static __constructor void ieee80211_ctor(void) { int i; for (i = 0; i < nitems(ieee80211_cmds); i++) cmd_register(&ieee80211_cmds[i]); af_register(&af_ieee80211); clone_setdefcallback("wlan", wlan_create); } Index: projects/clang1100-import/sbin/iscontrol/Makefile =================================================================== --- projects/clang1100-import/sbin/iscontrol/Makefile (revision 364034) +++ projects/clang1100-import/sbin/iscontrol/Makefile (revision 364035) @@ -1,14 +1,14 @@ # $FreeBSD$ -PACKAGE=iscsi_legacy +PACKAGE=iscsilegacy SRCS= iscontrol.c pdu.c fsm.c config.c login.c auth_subr.c misc.c PROG= iscontrol LIBADD= cam md S= ${SRCTOP}/sys WARNS?= 3 CFLAGS+= -I$S MAN= iscontrol.8 .include Index: projects/clang1100-import/sbin/mount/mount.c =================================================================== --- projects/clang1100-import/sbin/mount/mount.c (revision 364034) +++ projects/clang1100-import/sbin/mount/mount.c (revision 364035) @@ -1,984 +1,984 @@ /*- * SPDX-License-Identifier: BSD-3-Clause * * Copyright (c) 1980, 1989, 1993, 1994 * 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. * 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. */ #ifndef lint static const char copyright[] = "@(#) Copyright (c) 1980, 1989, 1993, 1994\n\ The Regents of the University of California. All rights reserved.\n"; #if 0 static char sccsid[] = "@(#)mount.c 8.25 (Berkeley) 5/8/95"; #endif #endif /* not lint */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "extern.h" #include "mntopts.h" #include "pathnames.h" /* `meta' options */ #define MOUNT_META_OPTION_FSTAB "fstab" #define MOUNT_META_OPTION_CURRENT "current" static int debug, fstab_style, verbose; struct cpa { char **a; ssize_t sz; int c; }; char *catopt(char *, const char *); struct statfs *getmntpt(const char *); int hasopt(const char *, const char *); int ismounted(struct fstab *, struct statfs *, int); int isremountable(const char *); void mangle(char *, struct cpa *); char *update_options(char *, char *, int); int mountfs(const char *, const char *, const char *, int, const char *, const char *); void remopt(char *, const char *); void prmount(struct statfs *); void putfsent(struct statfs *); void usage(void); char *flags2opts(int); /* Map from mount options to printable formats. */ static struct opt { uint64_t o_opt; const char *o_name; } optnames[] = { { MNT_ASYNC, "asynchronous" }, { MNT_EXPORTED, "NFS exported" }, { MNT_LOCAL, "local" }, { MNT_NOATIME, "noatime" }, { MNT_NOEXEC, "noexec" }, { MNT_NOSUID, "nosuid" }, { MNT_NOSYMFOLLOW, "nosymfollow" }, { MNT_QUOTA, "with quotas" }, { MNT_RDONLY, "read-only" }, { MNT_SYNCHRONOUS, "synchronous" }, { MNT_UNION, "union" }, { MNT_NOCLUSTERR, "noclusterr" }, { MNT_NOCLUSTERW, "noclusterw" }, { MNT_SUIDDIR, "suiddir" }, { MNT_SOFTDEP, "soft-updates" }, { MNT_SUJ, "journaled soft-updates" }, { MNT_MULTILABEL, "multilabel" }, { MNT_ACLS, "acls" }, { MNT_NFS4ACLS, "nfsv4acls" }, { MNT_GJOURNAL, "gjournal" }, { MNT_AUTOMOUNTED, "automounted" }, { MNT_VERIFIED, "verified" }, { MNT_UNTRUSTED, "untrusted" }, { MNT_NOCOVER, "nocover" }, { MNT_EMPTYDIR, "emptydir" }, { 0, NULL } }; /* * List of VFS types that can be remounted without becoming mounted on top * of each other. * XXX Is this list correct? */ static const char * remountable_fs_names[] = { "ufs", "ffs", "ext2fs", 0 }; static const char userquotaeq[] = "userquota="; static const char groupquotaeq[] = "groupquota="; static char *mountprog = NULL; static int use_mountprog(const char *vfstype) { /* XXX: We need to get away from implementing external mount * programs for every filesystem, and move towards having * each filesystem properly implement the nmount() system call. */ unsigned int i; const char *fs[] = { "cd9660", "mfs", "msdosfs", "nfs", "nullfs", "smbfs", "udf", "unionfs", NULL }; if (mountprog != NULL) return (1); for (i = 0; fs[i] != NULL; ++i) { if (strcmp(vfstype, fs[i]) == 0) return (1); } return (0); } static int exec_mountprog(const char *name, const char *execname, char *const argv[]) { pid_t pid; int status; switch (pid = fork()) { case -1: /* Error. */ warn("fork"); exit (1); case 0: /* Child. */ /* Go find an executable. */ execvP(execname, _PATH_SYSPATH, argv); if (errno == ENOENT) { warn("exec %s not found", execname); if (execname[0] != '/') { warnx("in path: %s", _PATH_SYSPATH); } } exit(1); default: /* Parent. */ if (waitpid(pid, &status, 0) < 0) { warn("waitpid"); return (1); } if (WIFEXITED(status)) { if (WEXITSTATUS(status) != 0) return (WEXITSTATUS(status)); } else if (WIFSIGNALED(status)) { warnx("%s: %s", name, sys_siglist[WTERMSIG(status)]); return (1); } break; } return (0); } static int specified_ro(const char *arg) { char *optbuf, *opt; int ret = 0; optbuf = strdup(arg); if (optbuf == NULL) err(1, NULL); for (opt = optbuf; (opt = strtok(opt, ",")) != NULL; opt = NULL) { if (strcmp(opt, "ro") == 0) { ret = 1; break; } } free(optbuf); return (ret); } static void restart_mountd(void) { struct pidfh *pfh; pid_t mountdpid; mountdpid = 0; pfh = pidfile_open(_PATH_MOUNTDPID, 0600, &mountdpid); if (pfh != NULL) { /* Mountd is not running. */ pidfile_remove(pfh); return; } if (errno != EEXIST) { /* Cannot open pidfile for some reason. */ return; } /* * Refuse to send broadcast or group signals, this has * happened due to the bugs in pidfile(3). */ if (mountdpid <= 0) { warnx("mountd pid %d, refusing to send SIGHUP", mountdpid); return; } /* We have mountd(8) PID in mountdpid varible, let's signal it. */ if (kill(mountdpid, SIGHUP) == -1) err(1, "signal mountd"); } int main(int argc, char *argv[]) { const char *mntfromname, **vfslist, *vfstype; struct fstab *fs; struct statfs *mntbuf; int all, ch, i, init_flags, late, failok, mntsize, rval, have_fstab, ro; int onlylate; char *cp, *ep, *options; all = init_flags = late = onlylate = 0; ro = 0; options = NULL; vfslist = NULL; vfstype = "ufs"; while ((ch = getopt(argc, argv, "adF:fLlno:prt:uvw")) != -1) switch (ch) { case 'a': all = 1; break; case 'd': debug = 1; break; case 'F': setfstab(optarg); break; case 'f': init_flags |= MNT_FORCE; break; case 'L': onlylate = 1; late = 1; break; case 'l': late = 1; break; case 'n': /* For compatibility with the Linux version of mount. */ break; case 'o': if (*optarg) { options = catopt(options, optarg); if (specified_ro(optarg)) ro = 1; } break; case 'p': fstab_style = 1; verbose = 1; break; case 'r': options = catopt(options, "ro"); ro = 1; break; case 't': if (vfslist != NULL) errx(1, "only one -t option may be specified"); vfslist = makevfslist(optarg); vfstype = optarg; break; case 'u': init_flags |= MNT_UPDATE; break; case 'v': verbose = 1; break; case 'w': options = catopt(options, "noro"); break; case '?': default: usage(); /* NOTREACHED */ } argc -= optind; argv += optind; #define BADTYPE(type) \ (strcmp(type, FSTAB_RO) && \ strcmp(type, FSTAB_RW) && strcmp(type, FSTAB_RQ)) if ((init_flags & MNT_UPDATE) && (ro == 0)) options = catopt(options, "noro"); rval = 0; switch (argc) { case 0: if ((mntsize = getmntinfo(&mntbuf, verbose ? MNT_WAIT : MNT_NOWAIT)) == 0) err(1, "getmntinfo"); if (all) { while ((fs = getfsent()) != NULL) { if (BADTYPE(fs->fs_type)) continue; if (checkvfsname(fs->fs_vfstype, vfslist)) continue; if (hasopt(fs->fs_mntops, "noauto")) continue; if (!hasopt(fs->fs_mntops, "late") && onlylate) continue; if (hasopt(fs->fs_mntops, "late") && !late) continue; if (hasopt(fs->fs_mntops, "failok")) failok = 1; else failok = 0; if (!(init_flags & MNT_UPDATE) && !hasopt(fs->fs_mntops, "update") && ismounted(fs, mntbuf, mntsize)) continue; options = update_options(options, fs->fs_mntops, mntbuf->f_flags); if (mountfs(fs->fs_vfstype, fs->fs_spec, fs->fs_file, init_flags, options, fs->fs_mntops) && !failok) rval = 1; } } else if (fstab_style) { for (i = 0; i < mntsize; i++) { if (checkvfsname(mntbuf[i].f_fstypename, vfslist)) continue; putfsent(&mntbuf[i]); } } else { for (i = 0; i < mntsize; i++) { if (checkvfsname(mntbuf[i].f_fstypename, vfslist)) continue; if (!verbose && (mntbuf[i].f_flags & MNT_IGNORE) != 0) continue; prmount(&mntbuf[i]); } } exit(rval); case 1: if (vfslist != NULL) usage(); rmslashes(*argv, *argv); if (init_flags & MNT_UPDATE) { mntfromname = NULL; have_fstab = 0; if ((mntbuf = getmntpt(*argv)) == NULL) errx(1, "not currently mounted %s", *argv); /* * Only get the mntflags from fstab if both mntpoint * and mntspec are identical. Also handle the special * case where just '/' is mounted and 'spec' is not * identical with the one from fstab ('/dev' is missing * in the spec-string at boot-time). */ if ((fs = getfsfile(mntbuf->f_mntonname)) != NULL) { if (strcmp(fs->fs_spec, mntbuf->f_mntfromname) == 0 && strcmp(fs->fs_file, mntbuf->f_mntonname) == 0) { have_fstab = 1; mntfromname = mntbuf->f_mntfromname; } else if (argv[0][0] == '/' && argv[0][1] == '\0' && strcmp(fs->fs_vfstype, mntbuf->f_fstypename) == 0) { fs = getfsfile("/"); have_fstab = 1; mntfromname = fs->fs_spec; } } if (have_fstab) { options = update_options(options, fs->fs_mntops, mntbuf->f_flags); } else { mntfromname = mntbuf->f_mntfromname; options = update_options(options, NULL, mntbuf->f_flags); } rval = mountfs(mntbuf->f_fstypename, mntfromname, mntbuf->f_mntonname, init_flags, options, 0); break; } if ((fs = getfsfile(*argv)) == NULL && (fs = getfsspec(*argv)) == NULL) errx(1, "%s: unknown special file or file system", *argv); if (BADTYPE(fs->fs_type)) errx(1, "%s has unknown file system type", *argv); rval = mountfs(fs->fs_vfstype, fs->fs_spec, fs->fs_file, init_flags, options, fs->fs_mntops); break; case 2: /* * If -t flag has not been specified, the path cannot be * found, spec contains either a ':' or a '@', then assume * that an NFS file system is being specified ala Sun. * Check if the hostname contains only allowed characters * to reduce false positives. IPv6 addresses containing * ':' will be correctly parsed only if the separator is '@'. * The definition of a valid hostname is taken from RFC 1034. */ if (vfslist == NULL && ((ep = strchr(argv[0], '@')) != NULL || (ep = strchr(argv[0], ':')) != NULL)) { if (*ep == '@') { cp = ep + 1; ep = cp + strlen(cp); } else cp = argv[0]; while (cp != ep) { if (!isdigit(*cp) && !isalpha(*cp) && *cp != '.' && *cp != '-' && *cp != ':') break; cp++; } if (cp == ep) vfstype = "nfs"; } rval = mountfs(vfstype, argv[0], argv[1], init_flags, options, NULL); break; default: usage(); /* NOTREACHED */ } /* * If the mount was successfully, and done by root, tell mountd the * good news. */ if (rval == 0 && getuid() == 0) restart_mountd(); exit(rval); } int ismounted(struct fstab *fs, struct statfs *mntbuf, int mntsize) { char realfsfile[PATH_MAX]; int i; if (fs->fs_file[0] == '/' && fs->fs_file[1] == '\0') /* the root file system can always be remounted */ return (0); /* The user may have specified a symlink in fstab, resolve the path */ if (realpath(fs->fs_file, realfsfile) == NULL) { /* Cannot resolve the path, use original one */ strlcpy(realfsfile, fs->fs_file, sizeof(realfsfile)); } /* * Consider the filesystem to be mounted if: * It has the same mountpoint as a mounted filesytem, and * It has the same type as that same mounted filesystem, and * It has the same device name as that same mounted filesystem, OR * It is a nonremountable filesystem */ for (i = mntsize - 1; i >= 0; --i) if (strcmp(realfsfile, mntbuf[i].f_mntonname) == 0 && strcmp(fs->fs_vfstype, mntbuf[i].f_fstypename) == 0 && (!isremountable(fs->fs_vfstype) || (strcmp(fs->fs_spec, mntbuf[i].f_mntfromname) == 0))) return (1); return (0); } int isremountable(const char *vfsname) { const char **cp; for (cp = remountable_fs_names; *cp; cp++) if (strcmp(*cp, vfsname) == 0) return (1); return (0); } int hasopt(const char *mntopts, const char *option) { int negative, found; char *opt, *optbuf; if (option[0] == 'n' && option[1] == 'o') { negative = 1; option += 2; } else negative = 0; optbuf = strdup(mntopts); found = 0; for (opt = optbuf; (opt = strtok(opt, ",")) != NULL; opt = NULL) { if (opt[0] == 'n' && opt[1] == 'o') { if (!strcasecmp(opt + 2, option)) found = negative; } else if (!strcasecmp(opt, option)) found = !negative; } free(optbuf); return (found); } static void append_arg(struct cpa *sa, char *arg) { if (sa->c + 1 == sa->sz) { sa->sz = sa->sz == 0 ? 8 : sa->sz * 2; sa->a = realloc(sa->a, sizeof(*sa->a) * sa->sz); if (sa->a == NULL) errx(1, "realloc failed"); } sa->a[++sa->c] = arg; } int mountfs(const char *vfstype, const char *spec, const char *name, int flags, const char *options, const char *mntopts) { struct statfs sf; int i, ret; char *optbuf, execname[PATH_MAX], mntpath[PATH_MAX]; static struct cpa mnt_argv; /* resolve the mountpoint with realpath(3) */ if (checkpath(name, mntpath) != 0) { warn("%s", mntpath); return (1); } name = mntpath; if (mntopts == NULL) mntopts = ""; optbuf = catopt(strdup(mntopts), options); if (strcmp(name, "/") == 0) flags |= MNT_UPDATE; if (flags & MNT_FORCE) optbuf = catopt(optbuf, "force"); if (flags & MNT_RDONLY) optbuf = catopt(optbuf, "ro"); /* * XXX * The mount_mfs (newfs) command uses -o to select the * optimization mode. We don't pass the default "-o rw" * for that reason. */ if (flags & MNT_UPDATE) optbuf = catopt(optbuf, "update"); /* Compatibility glue. */ if (strcmp(vfstype, "msdos") == 0) vfstype = "msdosfs"; /* Construct the name of the appropriate mount command */ (void)snprintf(execname, sizeof(execname), "mount_%s", vfstype); mnt_argv.c = -1; append_arg(&mnt_argv, execname); mangle(optbuf, &mnt_argv); if (mountprog != NULL) strlcpy(execname, mountprog, sizeof(execname)); append_arg(&mnt_argv, strdup(spec)); append_arg(&mnt_argv, strdup(name)); append_arg(&mnt_argv, NULL); if (debug) { if (use_mountprog(vfstype)) printf("exec: %s", execname); else printf("mount -t %s", vfstype); for (i = 1; i < mnt_argv.c; i++) (void)printf(" %s", mnt_argv.a[i]); (void)printf("\n"); free(optbuf); free(mountprog); mountprog = NULL; return (0); } if (use_mountprog(vfstype)) { ret = exec_mountprog(name, execname, mnt_argv.a); } else { ret = mount_fs(vfstype, mnt_argv.c, mnt_argv.a); } free(optbuf); free(mountprog); mountprog = NULL; if (verbose) { if (statfs(name, &sf) < 0) { warn("statfs %s", name); return (1); } if (fstab_style) putfsent(&sf); else prmount(&sf); } return (ret); } void prmount(struct statfs *sfp) { uint64_t flags; unsigned int i; struct opt *o; struct passwd *pw; (void)printf("%s on %s (%s", sfp->f_mntfromname, sfp->f_mntonname, sfp->f_fstypename); flags = sfp->f_flags & MNT_VISFLAGMASK; for (o = optnames; flags != 0 && o->o_opt != 0; o++) if (flags & o->o_opt) { (void)printf(", %s", o->o_name); flags &= ~o->o_opt; } /* * Inform when file system is mounted by an unprivileged user * or privileged non-root user. */ if ((flags & MNT_USER) != 0 || sfp->f_owner != 0) { (void)printf(", mounted by "); if ((pw = getpwuid(sfp->f_owner)) != NULL) (void)printf("%s", pw->pw_name); else (void)printf("%d", sfp->f_owner); } if (verbose) { if (sfp->f_syncwrites != 0 || sfp->f_asyncwrites != 0) (void)printf(", writes: sync %ju async %ju", (uintmax_t)sfp->f_syncwrites, (uintmax_t)sfp->f_asyncwrites); if (sfp->f_syncreads != 0 || sfp->f_asyncreads != 0) (void)printf(", reads: sync %ju async %ju", (uintmax_t)sfp->f_syncreads, (uintmax_t)sfp->f_asyncreads); if (sfp->f_fsid.val[0] != 0 || sfp->f_fsid.val[1] != 0) { - printf(", fsid "); + (void)printf(", fsid "); for (i = 0; i < sizeof(sfp->f_fsid); i++) - printf("%02x", ((u_char *)&sfp->f_fsid)[i]); + (void)printf("%02x", ((u_char *)&sfp->f_fsid)[i]); } } (void)printf(")\n"); } struct statfs * getmntpt(const char *name) { struct statfs *mntbuf; int i, mntsize; mntsize = getmntinfo(&mntbuf, MNT_NOWAIT); for (i = mntsize - 1; i >= 0; i--) { if (strcmp(mntbuf[i].f_mntfromname, name) == 0 || strcmp(mntbuf[i].f_mntonname, name) == 0) return (&mntbuf[i]); } return (NULL); } char * catopt(char *s0, const char *s1) { char *cp; if (s1 == NULL || *s1 == '\0') return (s0); if (s0 && *s0) { if (asprintf(&cp, "%s,%s", s0, s1) == -1) errx(1, "asprintf failed"); } else cp = strdup(s1); if (s0) free(s0); return (cp); } void mangle(char *options, struct cpa *a) { char *p, *s, *val; for (s = options; (p = strsep(&s, ",")) != NULL;) if (*p != '\0') { if (strcmp(p, "noauto") == 0) { /* * Do not pass noauto option to nmount(). * or external mount program. noauto is * only used to prevent mounting a filesystem * when 'mount -a' is specified, and is * not a real mount option. */ continue; } else if (strcmp(p, "late") == 0) { /* * "late" is used to prevent certain file * systems from being mounted before late * in the boot cycle; for instance, * loopback NFS mounts can't be mounted * before mountd starts. */ continue; } else if (strcmp(p, "failok") == 0) { /* * "failok" is used to prevent certain file * systems from being causing the system to * drop into single user mode in the boot * cycle, and is not a real mount option. */ continue; } else if (strncmp(p, "mountprog", 9) == 0) { /* * "mountprog" is used to force the use of * userland mount programs. */ val = strchr(p, '='); if (val != NULL) { ++val; if (*val != '\0') mountprog = strdup(val); } if (mountprog == NULL) { errx(1, "Need value for -o mountprog"); } continue; } else if (strcmp(p, "userquota") == 0) { continue; } else if (strncmp(p, userquotaeq, sizeof(userquotaeq) - 1) == 0) { continue; } else if (strcmp(p, "groupquota") == 0) { continue; } else if (strncmp(p, groupquotaeq, sizeof(groupquotaeq) - 1) == 0) { continue; } else if (*p == '-') { append_arg(a, p); p = strchr(p, '='); if (p != NULL) { *p = '\0'; append_arg(a, p + 1); } } else { append_arg(a, strdup("-o")); append_arg(a, p); } } } char * update_options(char *opts, char *fstab, int curflags) { char *o, *p; char *cur; char *expopt, *newopt, *tmpopt; if (opts == NULL) return (strdup("")); /* remove meta options from list */ remopt(fstab, MOUNT_META_OPTION_FSTAB); remopt(fstab, MOUNT_META_OPTION_CURRENT); cur = flags2opts(curflags); /* * Expand all meta-options passed to us first. */ expopt = NULL; for (p = opts; (o = strsep(&p, ",")) != NULL;) { if (strcmp(MOUNT_META_OPTION_FSTAB, o) == 0) expopt = catopt(expopt, fstab); else if (strcmp(MOUNT_META_OPTION_CURRENT, o) == 0) expopt = catopt(expopt, cur); else expopt = catopt(expopt, o); } free(cur); free(opts); /* * Remove previous contradictory arguments. Given option "foo" we * remove all the "nofoo" options. Given "nofoo" we remove "nonofoo" * and "foo" - so we can deal with possible options like "notice". */ newopt = NULL; for (p = expopt; (o = strsep(&p, ",")) != NULL;) { if ((tmpopt = malloc( strlen(o) + 2 + 1 )) == NULL) errx(1, "malloc failed"); strcpy(tmpopt, "no"); strcat(tmpopt, o); remopt(newopt, tmpopt); free(tmpopt); if (strncmp("no", o, 2) == 0) remopt(newopt, o+2); newopt = catopt(newopt, o); } free(expopt); return (newopt); } void remopt(char *string, const char *opt) { char *o, *p, *r; if (string == NULL || *string == '\0' || opt == NULL || *opt == '\0') return; r = string; for (p = string; (o = strsep(&p, ",")) != NULL;) { if (strcmp(opt, o) != 0) { if (*r == ',' && *o != '\0') r++; while ((*r++ = *o++) != '\0') ; *--r = ','; } } *r = '\0'; } void usage(void) { (void)fprintf(stderr, "%s\n%s\n%s\n", "usage: mount [-adflpruvw] [-F fstab] [-o options] [-t ufs | external_type]", " mount [-dfpruvw] special | node", " mount [-dfpruvw] [-o options] [-t ufs | external_type] special node"); exit(1); } void putfsent(struct statfs *ent) { struct fstab *fst; char *opts, *rw; int l; opts = NULL; /* flags2opts() doesn't return the "rw" option. */ if ((ent->f_flags & MNT_RDONLY) != 0) rw = NULL; else rw = catopt(NULL, "rw"); opts = flags2opts(ent->f_flags); opts = catopt(rw, opts); if (strncmp(ent->f_mntfromname, "", 7) == 0 || strncmp(ent->f_mntfromname, "", 7) == 0) { strlcpy(ent->f_mntfromname, (strnstr(ent->f_mntfromname, ":", 8) +1), sizeof(ent->f_mntfromname)); } l = strlen(ent->f_mntfromname); printf("%s%s%s%s", ent->f_mntfromname, l < 8 ? "\t" : "", l < 16 ? "\t" : "", l < 24 ? "\t" : " "); l = strlen(ent->f_mntonname); printf("%s%s%s%s", ent->f_mntonname, l < 8 ? "\t" : "", l < 16 ? "\t" : "", l < 24 ? "\t" : " "); printf("%s\t", ent->f_fstypename); l = strlen(opts); printf("%s%s", opts, l < 8 ? "\t" : " "); free(opts); if ((fst = getfsspec(ent->f_mntfromname))) printf("\t%u %u\n", fst->fs_freq, fst->fs_passno); else if ((fst = getfsfile(ent->f_mntonname))) printf("\t%u %u\n", fst->fs_freq, fst->fs_passno); else if (strcmp(ent->f_fstypename, "ufs") == 0) { if (strcmp(ent->f_mntonname, "/") == 0) printf("\t1 1\n"); else printf("\t2 2\n"); } else printf("\t0 0\n"); } char * flags2opts(int flags) { char *res; res = NULL; if (flags & MNT_RDONLY) res = catopt(res, "ro"); if (flags & MNT_SYNCHRONOUS) res = catopt(res, "sync"); if (flags & MNT_NOEXEC) res = catopt(res, "noexec"); if (flags & MNT_NOSUID) res = catopt(res, "nosuid"); if (flags & MNT_UNION) res = catopt(res, "union"); if (flags & MNT_ASYNC) res = catopt(res, "async"); if (flags & MNT_NOATIME) res = catopt(res, "noatime"); if (flags & MNT_NOCLUSTERR) res = catopt(res, "noclusterr"); if (flags & MNT_NOCLUSTERW) res = catopt(res, "noclusterw"); if (flags & MNT_NOSYMFOLLOW) res = catopt(res, "nosymfollow"); if (flags & MNT_SUIDDIR) res = catopt(res, "suiddir"); if (flags & MNT_MULTILABEL) res = catopt(res, "multilabel"); if (flags & MNT_ACLS) res = catopt(res, "acls"); if (flags & MNT_NFS4ACLS) res = catopt(res, "nfsv4acls"); if (flags & MNT_UNTRUSTED) res = catopt(res, "untrusted"); if (flags & MNT_NOCOVER) res = catopt(res, "nocover"); if (flags & MNT_EMPTYDIR) res = catopt(res, "emptydir"); return (res); } Index: projects/clang1100-import/share/man/man4/net80211.4 =================================================================== --- projects/clang1100-import/share/man/man4/net80211.4 (revision 364034) +++ projects/clang1100-import/share/man/man4/net80211.4 (revision 364035) @@ -1,1318 +1,1328 @@ .\"- .\" Copyright (c) 2009 Sam Leffler, Errno Consulting .\" 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, .\" without modification. .\" 2. Redistributions in binary form must reproduce at minimum a disclaimer .\" similar to the "NO WARRANTY" disclaimer below ("Disclaimer") and any .\" redistribution must be conditioned upon including a substantially .\" similar Disclaimer requirement for further binary redistribution. .\" .\" NO WARRANTY .\" 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 NONINFRINGEMENT, MERCHANTIBILITY .\" AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL .\" THE COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR 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 DAMAGES. .\" .\" $FreeBSD$ .\" -.Dd July 10, 2009 +.Dd August 7, 2020 .Dt NET80211 4 .Os .Sh NAME .Nm net80211 .Nd standard interface to IEEE 802.11 devices .Sh SYNOPSIS .In sys/types.h .In sys/socket.h .In net/if.h .In net/ethernet.h .In net80211/ieee80211_ioctl.h .Sh DESCRIPTION This section describes the standard programming interface to configure and retrieve status information for IEEE 802.11 devices that depend on the .Xr wlan 4 module for operation. The interface is via one of the following .Xr ioctl 2 calls on a socket: .Bl -tag -width ".Dv SIOCG80211" .It Dv SIOCG80211 Get configuration or status information. .It Dv SIOCS80211 Set configuration information. .El .Pp These requests are made via a modified .Vt ifreq structure. This structure is defined as follows: .Bd -literal struct ieee80211req { char i_name[IFNAMSIZ]; /* if_name, e.g. "wi0" */ uint16_t i_type; /* req type */ int16_t i_val; /* Index or simple value */ int16_t i_len; /* Index or simple value */ void *i_data; /* Extra data */ }; .Ed .Pp Requests that are not supported by the underlying device return -1 and set the global variable errno to .Er EOPNOTSUPP . .Dv SIOCG80211 requests that return data to an application place small values in .Va i_val or in a user-specified buffer pointed to by .Va i_data . When an indirect buffer is used .Va i_len specifies how large the indirect buffer is and on return it is set by the system to the actual amount of data returned. .Dv SIOCS80211 requests use a similar scheme with data passed to the system taken either from .Va i_val or an indirect buffer pointed to by .Va i_data . .Pp For .Dv SIOCG80211 the following values of .Va i_type are valid: .Bl -tag -width indent .It Dv IEEE80211_IOC_AMPDU Return whether or not AMPDU is enabled in .Va i_val . AMPDU is an aggregation scheme that is part of the 802.11n specification and is used only when operating on an HT channel. The value returned is one of: 0 (AMPDU disabled), 1 (AMPDU enabled for transmit), 2 (AMPDU enabled for receive), and 3 (AMPDU enabled for transmit and receive). The 802.11n specification says a compliant station must receive AMPDU but may not support transmitting AMPDU frames. Disabling AMPDU receive is mainly useful for testing and working around bugs. .It Dv IEEE80211_IOC_AMPDU_DENSITY Return the minimum density for bursting AMPDU frames in .Va i_val . The value returned is one of: 0 (no time restriction), 1 (1/4 usec), 2 (1/2 usec), 3 (1 usec), 4 (2 usec), 5 (4 usec), 6 (8 usec), and 7 (16 usec). .It Dv IEEE80211_IOC_AMPDU_LIMIT Return the limit on the size of AMPDU frames in .Va i_val . The value returned is one of: 0 (8 kilobytes), 1 (16 kilobytes), 2 (32 kilobytes), and 3 (64 kilobytes). .It Dv IEEE80211_IOC_AMSDU Return whether or not AMSDU is enabled in .Va i_val . AMSDU is an aggregation scheme that is part of the 802.11n specification and is used only when operating on an HT channel. The value returned is one of: 0 (AMSDU disabled), 1 (AMSDU enabled for transmit), 2 (AMSDU enabled for receive), and 3 (AMSDU enabled for transmit and receive). The 802.11n specification says a compliant station must receive AMSDU but may not support transmitting AMSDU frames. Disabling AMSDU receive is mainly useful for testing and working around bugs. .It Dv IEEE80211_IOC_AMSDU_LIMIT Return the limit on the size of AMSDU frames in .Va i_val . The value returned is one of: 3839 (bytes) and 7935 (bytes). Note these values are specified by 802.11n; arbitrary values are not allowed. .It Dv IEEE80211_IOC_APBRIDGE Return whether AP bridging is enabled in .Va i_val . Normally packets sent between stations associated to the same access point are delivered without going through system layers that do packet filtering; when AP bridging is disabled packets are passed up the system to be forwarded using some other mechanism. This value will be non-zero when AP bridging is enabled and otherwise zero. .It Dv IEEE80211_IOC_APPIE Return an application information element via .Va i_data . Application IE's are maintained for many 802.11 frames; the request must identify the frame to return an IE for in .Va i_val . For example, to retrieve the IE sent in each Beacon frame .Va i_val would be set to .Va IEEE80211_FC0_SUBTYPE_BEACON | IEEE80211_FC0_TYPE_MGT . If no information element is installed then .Er EINVAL is returned. If the data buffer for returning data is too small to hold the information element the value is truncated; this permits querying the presence of data by requesting zero bytes of data be returned. .It Dv IEEE80211_IOC_AUTHMODE Return the current authentication mode in .Va i_val . Valid values are .Dv IEEE80211_AUTH_NONE (no authentication), .Dv IEEE80211_AUTH_OPEN (open authentication), .Dv IEEE80211_AUTH_SHARED (shared key authentication), .Dv IEEE80211_AUTH_8021X (802.1x only authentication), and .Dv IEEE80211_AUTH_WPA (WPA/802.11i/802.1x authentication). .It Dv IEEE80211_IOC_BEACON_INTERVAL Return the time between Beacon frames (in TU) in .Va i_val . .It Dv IEEE80211_IOC_BGSCAN Return whether background scanning is enabled in .Va i_val . When this value is non-zero and operating in station mode the station will periodically leave the current channel and scan for neighboring stations. See also .Dv IEEE80211_IOC_BGSCAN_IDLE and .Dv IEEE80211_IOC_BGSCAN_INTERVAL . .It Dv IEEE80211_IOC_BGSCAN_IDLE Return in .Va i_val the minimum time (msecs) a station must be idle (i.e. not transmitting or receiving frames) before it will do a background scan. See also .Dv IEEE80211_IOC_BGSCAN_INTERVAL . .It Dv IEEE80211_IOC_BGSCAN_INTERVAL Return in .Va i_val the minimum time (seconds) between background scan operations. See also .Dv IEEE80211_IOC_BGSCAN_IDLE . .It Dv IEEE80211_IOC_BMISSTHRESHOLD Return in .Va i_val the number of consecutive missed Beacon frames before the system will attempt to roam to a different/better access point. .It Dv IEEE80211_IOC_BSSID Return the MAC address for the current BSS identifier via .Va i_data . When the interface is running, the bssid is either the value configured locally (e.g. for an IBSS network started by the local station) or the value adopted when joining an existing network. For WDS interfaces this value is the address of the remote station. When the interface is not running, the bssid returned is the desired bssid, if any, that has been configured. .It Dv IEEE80211_IOC_BURST Return whether or not packet bursting is enabled in .Va i_val . If this value is non-zero then the system will try to send packets closely spaced to improve throughput. .It Dv IEEE80211_IOC_CHANINFO Return the set of available channels via .Va i_data . Note this data should be used by user applications to map between channel specifications (frequency and attributes) and IEEE channel numbers as user applications may not have the necessary information to do this directly (e.g. for 900MHz radios, operation in the Public Safety Band). .It Dv IEEE80211_IOC_CHANLIST Return the current list of usable channels via .Va i_data . The channel list is returned as a bit vector with bit N set to 1 if IEEE channel number N is available for use. .It Dv IEEE80211_IOC_CHANNEL Return the IEEE channel number of the current channel in .Va i_val . Note this request is deprecated; use .Dv IEEE80211_IOC_CURCHAN instead. .It Dv IEEE80211_IOC_COUNTERMEASURES Return whether TKIP Countermeasures are enabled in .Va i_val . This value will be non-zero when Countermeasures are enabled and otherwise zero. .It Dv IEEE80211_IOC_CURCHAN Return information for the current channel via .Va i_data . This information includes the IEEE channel number, the frequency, and attributes that describe the operating constraints (e.g. Passive Scan, DFS, usage restrictions). .It Dv IEEE80211_IOC_DEVCAPS Return device capabilities in the data buffer pointed at by .Va i_data . The buffer must be large enough to return the number of available channels but otherwise may be made small to limit how much information is returned. .It Dv IEEE80211_IOC_DFS Return whether or not Dynamic Frequency Selection (DFS) is enabled in .Va i_val . DFS embodies several facilities including detection of overlapping radar signals, dynamic transmit power control, and channel selection according to a least-congested criteria. DFS support is mandatory for some 5GHz frequencies in certain locales (e.g. ETSI). By default DFS is enabled according to the regulatory definitions and the current country code, regdomain, and channel. .It Dv IEEE80211_IOC_DOTD Return whether or not 802.11d support is enabled in .Va i_val . When 802.11d is enabled in station mode, Beacon frames that advertise a country code different than the currently configured country code will cause an event to be dispatched to user applications. This event can be used by the station to adopt that country code and operate according to the associated regulatory constraints. When operating as an access point with 802.11d enabled the Beacon and ProbeResponse frames transmitted will advertise the current regulatory domain settings. .It Dv IEEE80211_IOC_DOTH Return whether 802.11h support is enabled in .Va i_val . When 802.11h is enabled Beacon and ProbeResponse frames will have the SpectrumMgt bit set in the capabilities field and country and power constraint information elements will be present. 802.11h support also includes handling Channel Switch Announcements (CSA) which are a mechanism to coordinate channel changes by an access point. By default 802.11h is enabled if the device is capable. .It Dv IEEE80211_IOC_DROPUNENCRYPTED Return, in .Va i_val , whether unencrypted packets transmit/received should be discarded. This value will be zero if unencrypted packets will be accepted and non-zero if they are to be discarded. .It Dv IEEE80211_IOC_DTIM_PERIOD Return the period (in beacon intervals) between DTIM events in .Va i_val . .It Dv IEEE80211_IOC_DWDS Return, in .Va i_val , whether or not Dynamic WDS support is enabled. Dynamic WDS is a facility by which packets may be tunneled over normal Infrastructure BSS associations using 4-address (WDS) frames. .It Dv IEEE80211_IOC_FF Return, in .Va i_val , whether Atheros fast-frames support is enabled. Fast-frames is a non-standard protocol extension that aggregates multiple frames to improve throughput. Note that enabling fast-frames support does not guarantee use; the client and access point must negotiate its use. .It Dv IEEE80211_IOC_FRAGTHRESHOLD Return, in .Va i_val , the threshold (in bytes) for enabling fragmentation frames. Packets larger than this value will automatically be split into multiple fragmented frames that are sent one after the other. .It Dv IEEE80211_IOC_GREENFIELD Return, in .Va i_val , whether or not Greenfield preamble use is enabled. This setting is meaningful only when operating with 802.11n on an HT channel. .It Dv IEEE80211_IOC_HIDESSID Return, in .Va i_val , whether SSID hiding/cloaking is enabled. SSID hiding is only meaningful when operating as an access point. When this is enabled Beacon frames do not include the SSID and ProbeRequest frames are not answered unless they include the AP's SSID. This value will be non-zero when SSID hiding is enabled and otherwise zero. .It Dv IEEE80211_IOC_HTCOMPAT Return, in .Va i_val , whether or not 802.11n compatibility support is enabled. The 802.11n protocol specification went through several incompatible iterations. Some vendors implemented 11n support to older specifications that will not interoperate with a purely 11n-compliant station. In particular the information elements included in management frames for old devices are different. When compatibility support is enabled both standard and compatible data will be provided and/or accepted. .It Dv IEEE80211_IOC_HTCONF Return the setting for automatic promotion of HT channels in .Va i_val . Promotion happens when the system must select a channel and may choose between legacy, HT20, and HT40 operation (e.g. when scanning). Valid values are: 0 (do not promote, use legacy), 1 (promote to HT20), and 2 (promote to HT40). .It Dv IEEE80211_IOC_HTPROTMODE Return, in .Va i_val , the technique used to protect HT frames in a mixed 802.11n network. Valid values are: .Dv IEEE80211_PROTMODE_OFF (no protection enabled) and .Dv IEEE80211_PROTMODE_RTSCTS (send RTS and wait for CTS). .It Dv IEEE80211_IOC_HWMP_MAXHOPS Return the maximum acceptable hop count in an HWMP path in .Va i_val . .It Dv IEEE80211_IOC_HWMP_ROOTMODE Return the setting for Mesh root mode operation in .Va i_val . Valid values are: .Dv IEEE80211_HWMP_ROOTMODE_DISABLED (root mode is disabled), .Dv IEEE80211_HWMP_ROOTMODE_NORMAL (send broadcast Path Request frames), .Dv IEEE80211_HWMP_ROOTMODE_PROACTIVE (send broadcast Path Request frames and force replies) and .Dv IEEE80211_HWMP_ROOTMODE_RANN (send broadcast Root Announcement (RANN) frames). +.It Dv IEEE80211_IOC_IC_NAME +Return the underlying hardware +.Xr device 9 +name in the buffer pointed to by +.Va i_data +and the name length including terminating NUL character in +.Va i_len . +If the buffer length is too small to hold the full name +.Er EINVAL +will be returned. .It Dv IEEE80211_IOC_INACTIVITY Return whether or not the system handles inactivity processing in .Va i_val . When inactivity processing is enabled the system will track stations that have not transmitted frames and periodically probe them to check if they are still present. Stations that are inactive and do not respond to probes are dropped. .It Dv IEEE80211_IOC_MACCMD Return information about the state of the MAC address access control list (ACL) system. There are two requests supported: .Dv IEEE80211_MACCMD_POLICY (to retrieve the current policy in .Va i_val ), and .Dv IEEE80211_MACCMD_LIST to retrieve the list installed/active ACL's via .Va i_data . The .Xr wlan_acl 4 module must be installed and enabled or .Er EINVAL will be returned. .It Dv IEEE80211_IOC_MESH_AP Return whether or not Mesh AP support is enabled in .Va i_val . .It Dv IEEE80211_IOC_MESH_ID Return the Mesh ID in the buffer pointed to by .Va i_data . .It Dv IEEE80211_IOC_MESH_FWRD Return whether or not packet forwarding support is enabled in .Va i_val . .It Dv IEEE80211_IOC_MESH_PP_METRIC Return the link metric protocol in the buffer pointed to by .Va i_data . .It Dv IEEE80211_IOC_MESH_PP_PATH Return the path selection protocol in the buffer pointed to by .Va i_data . .It Dv IEEE80211_IOC_MESH_RTCMD Return information about the state of the Mesh routing tables. One request is supported: .Dv IEEE80211_MESH_RTCMD_LIST to retrieve the contents of the routing table in the buffer pointed to by .Va i_data . .It Dv IEEE80211_IOC_MESH_TTL Return, in .Va i_val , the Mesh Time To Live (TTL) setting installed in packets transmitted by this mesh node. .It Dv IEEE80211_IOC_NUMSSIDS Return the number of SSIDs supported in .Va i_val . .It Dv IEEE80211_IOC_NUMWEPKEYS Return the number of WEP keys supported in .Va i_val . .It Dv IEEE80211_IOC_POWERSAVE Return the current powersaving mode in .Va i_val . Valid values are .Dv IEEE80211_POWERSAVE_OFF (power save operation is disabled) and .Dv IEEE80211_POWERSAVE_ON (power save operation is enabled). .It Dv IEEE80211_IOC_POWERSAVESLEEP Return the powersave sleep time in TU in .Va i_val . This value is also termed the listen interval and represents the maximum time a station will sleep before waking to retrieve packets buffered by an access point. .It Dv IEEE80211_IOC_PRIVACY Return the current MLME setting for PRIVACY in .Va i_val . When PRIVACY is enabled all data packets must be encrypted. This value will be zero if PRIVACY is disabled and non-zero when PRIVACY is enabled. .It Dv IEEE80211_IOC_PROTMODE Return the current 802.11g protection mode in .Va i_val . Protection is the mechanism used to safeguard 802.11b stations operating on an 802.11g network. Valid values are .Dv IEEE80211_PROTMODE_OFF (no protection enabled), .Dv IEEE80211_PROTMODE_CTS (send CTS to yourself), and .Dv IEEE80211_PROTMODE_RTSCTS (send RTS and wait for CTS). .It Dv IEEE80211_IOC_PUREG Return whether ``pure 11g'' mode is enabled in .Va i_val . This setting is meaningful only for access point operation; when non-zero, 802.11b stations will not be allowed to associate. .It Dv IEEE80211_IOC_PUREN Return whether ``pure 11n'' mode is enabled in .Va i_val . This setting is meaningful only for access point operation; when non-zero, legacy (non-11n capable) stations will not be allowed to associate. .It Dv IEEE80211_IOC_REGDOMAIN Return the regulatory state in the buffer pointed to by .Va i_data . .It Dv IEEE80211_IOC_RIFS Return whether or not Reduced InterFrame Spacing (RIFS) is enabled in .Va i_val . This setting is meaningful only when operating with 802.11n on an HT channel. .It Dv IEEE80211_IOC_ROAM Return station roaming parameters in the buffer pointed to by .Va i_data . .It Dv IEEE80211_IOC_ROAMING Return the current roaming mode in .Va i_val . Roaming mode specifies which entity controls operation of the MLME state machine when operating as a station in an Infrastructure BSS. Valid values are: .Dv IEEE80211_ROAMING_DEVICE (driver/firmware is in control), .Dv IEEE80211_ROAMING_AUTO (host 802.11 layer is in control), and .Dv IEEE80211_ROAMING_MANUAL (application is in control). .It Dv IEEE80211_IOC_RTSTHRESHOLD Return the threshold (in bytes) for enabling transmission of RTS frames in .Va i_val . Packets larger than this value will automatically have an RTS frame sent preceding it to reduce the likelihood of packet loss. .It Dv IEEE80211_IOC_SCAN_RESULTS Return the current contents of the scan cache in the data area pointed to by .Va i_data . .It Dv IEEE80211_IOC_SCANVALID Return in .Va i_val how long (in seconds) results from a scan operation will be considered valid. When scan results are no longer valid and they are needed (e.g. to roam) the system will initiate a scan operation to replenish the scan cache. .It Dv IEEE80211_IOC_SHORTGI Return whether or not Short Guard Interval (SGI) is enabled in .Va i_val . Note SGI is only used when operating with 802.11n on an HT channel. .It Dv IEEE80211_IOC_SMPS Return the Spatial Multiplexing Power Save (SMPS) setting in .Va i_val . This setting is meaningful only when operating with 802.11n on an HT channel. Valid values are: .Dv IEEE80211_HTCAP_SMPS_DYNAMIC (Dynamic SMPS is enabled), .Dv IEEE80211_HTCAP_SMPS_ENA (Static SMPS is enabled), and .Dv IEEE80211_HTCAP_SMPS_OFF (SMPS is disabled). .It Dv IEEE80211_IOC_SSID Return the requested SSID in the buffer pointed to by .Va i_data . If .Va i_val is \(>= 0 then the request refers to the configured value for that slot. Generally, 0 is the only valid value, but some interfaces support more SSIDs. .It Dv IEEE80211_IOC_STA_INFO Return information about the current state of the specified station(s) via .Va i_data . The MAC address of a single station may be passed in or, if the broadcast address is supplied, information about all stations will be returned. If a single station is requested and the MAC address is unknown then .Er ENOENT will be returned. .It Dv IEEE80211_IOC_STA_STATS Return collected statistics for the specified station via .Va i_data . The MAC address of the desired station is passed in; if it is unknown .Er ENOENT will be returned. .It Dv IEEE80211_IOC_STA_VLAN Return any VLAN tag assigned to a station via .Va i_data . .It Dv IEEE80211_IOC_TDMA_SLOT Return the slot number for the station in .Va i_val . Slot number zero is the master station in a TDMA network. .It Dv IEEE80211_IOC_TDMA_SLOTCNT Return the count of slots in the TDMA network in .Va i_val . .It Dv IEEE80211_IOC_TDMA_SLOTLEN Return the length (in usecs) of the TDMA slot assigned to each station in the network in .Va i_val . .It Dv IEEE80211_IOC_TDMA_BINTERVAL Return the number of superframes between Beacon frames in .Va i_val . A TDMA network with N slots and slot length T has a superframe of NxT. .It Dv IEEE80211_IOC_TSN Return whether or not Transitional Security Network (TSN) is enabled in .Va i_val . .It Dv IEEE80211_IOC_TURBOP Return whether Atheros Dynamic Turbo mode is enabled in .Va i_val . Dynamic Turbo mode is a non-standard protocol extension available only on Atheros devices where channel width is dynamically changed between 20MHz and 40MHz. Note that enabling Dynamic Turbo mode support does not guarantee use; both client and access point must use Atheros devices and support must be enabled on both sides. .It Dv IEEE80211_IOC_TXPARAMS Return transmit parameters in the buffer pointed to by .Va i_data . .It Dv IEEE80211_IOC_TXPOWER Return the transmit power limit in .5 dBm units in .Va i_val . This value represents the effective maximum and is calculated according to the maximum power allowed by local regulations, any user-specified power limit, and the maximum power the device is capable of. .It Dv IEEE80211_IOC_TXPOWMAX Return the user-specified maximum transmit power in .5 dBm units in .Va i_val . The maximum setting is applied after any regulatory cap. .It Dv IEEE80211_IOC_WEP Return the current WEP status in .Va i_val . Valid values are: .Dv IEEE80211_WEP_ON (enabled for all packets sent and received), .Dv IEEE80211_WEP_OFF (disabled), and .Dv IEEE80211_WEP_MIXED (enabled for transmit and receive but also willing to receive unencrypted frames). This request is deprecated; use .Dv IEEE80211_IOC_PRIVACY and .Dv IEEE80211_IOC_UNENCRYPTED instead. .It Dv IEEE80211_IOC_WEPKEY Return the requested WEP key via .Va i_data . The key number is specified in .Va i_val and may be 0-3. If the device does not support returning the WEP key or the user is not root then the key may be returned as all zeros. This request is deprecated in favor of .Dv IEEE80211_IOC_WPAKEY . .It Dv IEEE80211_IOC_WEPTXKEY Return the number of the WEP key used for transmission in .Va i_val . .It Dv IEEE80211_IOC_WME Return whether 802.11e/WME/WMM support is enabled in .Va i_val . This value will be non-zero when support is enabled and otherwise zero. .It Dv IEEE80211_IOC_WME_CWMIN Return the WME CWmin setting (log2) for the specified Access Class (AC) in .Va i_val . The AC is passed in through .Va i_len together with an optional IEEE80211_WMEPARAM_BSS flag to indicate if the parameter for the BSS or the channel is desired. If WME is not supported then .Er EINVAL will be returned. .It Dv IEEE80211_IOC_WME_CWMAX Return the WME CWmax setting (log2) for the specified Access Class (AC) in .Va i_val . See .Dv IEEE80211_IOC_WME_CWMIN above for more details. .It Dv IEEE80211_IOC_WME_AIFS Return the WME AIFS setting for the specified Access Class (AC) in .Va i_val . See .Dv IEEE80211_IOC_WME_CWMIN above for more details. .It Dv IEEE80211_IOC_WME_TXOPLIMIT Return the WME TxOpLimit (msec) for the specified Access Class (AC) in .Va i_val . See .Dv IEEE80211_IOC_WME_CWMIN above for more details. .It Dv IEEE80211_IOC_WME_ACM Return the WME Admission Control Mechanism (ACM) setting for the specified Access Class (AC) in .Va i_val . This value is meaningful only for the BSS (not channel). See .Dv IEEE80211_IOC_WME_CWMIN above for more details. .It Dv IEEE80211_IOC_WME_ACKPOLICY Return the WME ACK Policy setting for the specified Access Class (AC) in .Va i_val . When this value is zero frames will be transmitted without waiting for an Acknowledgement. This value is meaningful only for the channel (not BSS). See .Dv IEEE80211_IOC_WME_CWMIN above for more details. .It Dv IEEE80211_IOC_WPA Return the WPA configuration in .Va i_val . Valid values are 0 (WPA is not enabled), 1 (WPA1 is enabled), 2 (WPA2/802.11i is enabled), and 3 (WPA1 and WPA2/802.11i are both enabled). .It Dv IEEE80211_IOC_WPAIE Return any WPA information element for an associated station via .Va i_data . The request passed in through .Va i_data identifies the MAC address of the desired station. If an RSN (802.11i) element is present it is returned; otherwise any WPA element is returned. Note this request is deprecated; use .Dv IEEE80211_IOC_WPAIE2 instead. .It Dv IEEE80211_IOC_WPAIE2 Return any WPA information elements for an associated station via .Va i_data . The request passed in through .Va i_data identifies the MAC address of the desired station. One or both of RSN (802.11i) and WPA elements may be returned. .It Dv IEEE80211_IOC_WPAKEY Return the requested cryptographic key in the buffer pointed to by .Va i_data . The key number is specified in .Va i_val and may be 0-3. A key number of zero is used to retrieve a station's unicast cipher key when operating with WPA enabled. If the user is not root then the key data returned is all zeros. .It Dv IEEE80211_IOC_WPS Return whether or not Wi-FI Protected Setup (WPS) is enabled in .Va i_val . .El .Pp For .Dv SIOCS80211 the following values of .Va i_type are valid. Note that changing a value on an interface that is running may cause the interface to be .Sq reset . Resets may be handled without altering the state if the parameter does not affect the MLME state (e.g. RTS threshold), but in some cases the interface may need to scan for a new network or clear state (including any associated stations); in that case the interface is said to be .Sq restarted (it is equivalent to marking the interface down and back up). The information below identifies whether changing a value affects the state of a running interface. .Bl -tag -width indent .It Dv IEEE80211_IOC_ADDMAC Add an entry to the MAC address Access Control List (ACL) database using the value pointed to by .Va i_data . The .Xr wlan_acl 4 module must be installed and enabled or .Er EINVAL will be returned. .It Dv IEEE80211_IOC_AMPDU Set whether or not AMPDU is enabled for transmit and/or receive using the value in .Va i_val . This request causes a running interface operating on an HT channel to be reset. See .Dv IEEE80211_IOC_AMPDU above for details. .It Dv IEEE80211_IOC_AMPDU_DENSITY Set the minimum density for bursting AMPDU frames to the value in .Va i_val . This request causes a running interface to be reset. See .Dv IEEE80211_IOC_AMPDU_DENSITY above for details. .It Dv IEEE80211_IOC_AMPDU_LIMIT Set the limit on the size of AMPDU frames to the value in .Va i_val . This request causes a running interface to be reset. See .Dv IEEE80211_IOC_AMPDU_LIMIT above for details. .It Dv IEEE80211_IOC_AMSDU Set whether or not AMSDU is enabled for transmit and/or receive using the value in .Va i_val . This request causes a running interface operating on an HT channel to be reset. See .Dv IEEE80211_IOC_AMSDU above for details. .It Dv IEEE80211_IOC_AMSDU_LIMIT Set the limit on the size of AMSDU frames to the value in .Va i_val . This request causes a running interface to be reset. See .Dv IEEE80211_IOC_AMSDU_LIMIT above for details. .It Dv IEEE80211_IOC_APBRIDGE Set whether AP bridging is enabled using the value in .Va i_val . See .Dv IEEE80211_IOC_APBRIDGE above for details. .It Dv IEEE80211_IOC_APPIE Set an application information element using the data pointed to by .Va i_data . This request causes a running interface to be restarted if the WPA information element is changed. See .Dv IEEE80211_IOC_APPIE above for details. .It Dv IEEE80211_IOC_AUTHMODE Set the current authentication mode using the value in .Va i_val . This request causes a running interface to be restarted. See .Dv IEEE80211_IOC_AUTHMODE above for details. This request causes a running interface to be restarted. .It Dv IEEE80211_IOC_BEACON_INTERVAL Set the time between Beacon frames (in TU) to the value in .Va i_val . This request causes a running interface to be restarted. .It Dv IEEE80211_IOC_BGSCAN Set whether background scanning is enabled using the value in .Va i_val . .It Dv IEEE80211_IOC_BGSCAN_IDLE Set the minimum time (in msecs) a station must be idle before it will do a background scan to the value in .Va i_val . .It Dv IEEE80211_IOC_BGSCAN_INTERVAL Set the minimum time (seconds) between background scan operations to the value in .Va i_val . .It Dv IEEE80211_IOC_BMISSTHRESHOLD Set the number of consecutive missed Beacon frames before the system will attempt to roam to the value in .Va i_val . This request causes a running interface to be reset. .It Dv IEEE80211_IOC_BSSID Set the 802.11 MAC address for the desired BSS identifier according to .Va i_data . This request causes a running interface to be restarted. .It Dv IEEE80211_IOC_BURST Set whether or not packet bursting is enabled using the value in .Va i_val . This request causes a running interface to be reset. .It Dv IEEE80211_IOC_CHANNEL Set the desired/current channel to the value given by .Va i_val . This request causes a running interface to immediately change to the specified channel if possible; otherwise the interface will be restarted. Note this request is deprecated; use .Dv IEEE80211_IOC_CURCHAN instead. .It Dv IEEE80211_IOC_CHANLIST Set the list of available channels using the channel list pointed to by .Va i_data . The channel list is a bit vector with bit N set to 1 if IEEE channel number N is available for use. The specified channel list is checked against the set of supported channels and any channels not supported are silently ignored. If the intersection of the channel list and the supported channels is empty .Er EINVAL is returned. Note the current channel may be marked invalid after installing a new channel list. This request causes a running interface to be restarted. .It Dv IEEE80211_IOC_COUNTERMEASURES Set whether TKIP Countermeasures are enabled using the value in .Va i_val . This request can only be used when the authentication mode is set WPA; otherwise .Er EOPNOTSUPP will be returned. .It Dv IEEE80211_IOC_CURCHAN Set the current channel using the information referenced by .Va i_data . This request causes a running interface to immediately change to the specified channel if possible; otherwise the interface will be restarted. .It Dv IEEE80211_IOC_DELKEY Delete the key specified by the information referenced by .Va i_data . .It Dv IEEE80211_IOC_DELMAC Remove an entry in the MAC address Access Control List (ACL) database using the value pointed to by .Va i_data . The .Xr wlan_acl 4 module must be installed and enabled or .Er EINVAL will be returned. .It Dv IEEE80211_IOC_DFS Set whether or not Dynamic Frequency Selection (DFS) is enabled using the value in .Va i_val . This request will fail with .Er EINVAL if 802.11h support is not enabled. See .Dv IEEE80211_IOC_DFS above for details. .It Dv IEEE80211_IOC_DOTD Set whether or not 802.11d support is enabled using the value in .Va i_val . This request causes a running interface to be restarted. See .Dv IEEE80211_IOC_DOTD above for details. .It Dv IEEE80211_IOC_DOTH Return whether 802.11h support is enabled using the value in .Va i_val . See .Dv IEEE80211_IOC_DOTH above for details. .It Dv IEEE80211_IOC_DROPUNENCRYPTED Set whether unencrypted packets transmit/received should be discarded using the value in .Va i_val . .It Dv IEEE80211_IOC_DTIM_PERIOD Set the period (in beacon intervals) between DTIM events to the value in .Va i_val . This request causes a running interface to be restarted. .It Dv IEEE80211_IOC_DWDS Set whether or not Dynamic WDS support is enabled using the value in .Va i_val . See .Dv IEEE80211_IOC_DWDS above for details. .It Dv IEEE80211_IOC_FF Set whether Atheros fast-frames support is enabled using the value in .Va i_val . This request causes a running interface to be restarted. See .Dv IEEE80211_IOC_FF above for details. .It Dv IEEE80211_IOC_FRAGTHRESHOLD Set the threshold (in bytes) for enabling fragmentation frames using the value in .Va i_val . This request causes a running interface to be reset. See .Dv IEEE80211_IOC_FRAGTHRESHOLD above for details. .It Dv IEEE80211_IOC_GREENFIELD Set whether or not Greenfield preamble use is enabled using the value in .Va i_val . This request causes a running interface to be reset. See .Dv IEEE80211_IOC_GREENFIELD above for details. .It Dv IEEE80211_IOC_HIDESSID Set whether SSID hiding/cloaking is enabled using the value in .Va i_val . This request causes a running interface to be reset. See .Dv IEEE80211_IOC_HIDESSID above for details. .It Dv IEEE80211_IOC_HTCOMPAT Set whether or not 802.11n compatibility support is enabled using the value in .Va i_val . This request causes a running interface to be reset if operating on HT channel. See .Dv IEEE80211_IOC_HTCOMPAT above for details. .It Dv IEEE80211_IOC_HTCONF Set automatic promotion of HT channels using the value in .Va i_val . This request causes a running interface to be restarted. See .Dv IEEE80211_IOC_HTCONF above for details. .It Dv IEEE80211_IOC_HTPROTMODE Set the technique used to protect HT frames in a mixed 802.11n network using the value in .Va i_val . This request causes a running interface to be reset. See .Dv IEEE80211_IOC_HTPROTMODE above for details. .It Dv IEEE80211_IOC_HWMP_MAXHOPS Set the maximum acceptable hop count in an HWMP path according to .Va i_val . Values must be in the range [0-255]. .It Dv IEEE80211_IOC_HWMP_ROOTMODE Set the Mesh root mode operation according to .Va i_val . Valid values are .Dv IEEE80211_HWMP_ROOTMODE_DISABLED (root mode is disabled), .Dv IEEE80211_HWMP_ROOTMODE_NORMAL (send broadcast Path Request frames), .Dv IEEE80211_HWMP_ROOTMODE_PROACTIVE (send broadcast Path Request frames and force replies) and .Dv IEEE80211_HWMP_ROOTMODE_RANN (send broadcast Root Announcement (RANN) frames). .It Dv IEEE80211_IOC_INACTIVITY Set whether or not the system handles inactivity processing using the value in .Va i_val . When inactivity processing is enabled the system will track stations that have not transmitted frames and periodically probe them to check if they are still present. Stations that are inactive and do not respond to probes are dropped. .It Dv IEEE80211_IOC_MACCMD Change the state of the MAC address Access Control List (ACL) system. There are several requests supported: .Dv IEEE80211_MACCMD_POLICY_OPEN (set the current policy to disable ACL use), .Dv IEEE80211_MACCMD_POLICY_ALLOW (set the current policy to allow only addresses listed in the database), .Dv IEEE80211_MACCMD_POLICY_DENY (set the current policy to deny addresses listed in the database), .Dv IEEE80211_MACCMD_POLICY_RADUS (set the current policy to enable use of a RADIUS backend), .Dv IEEE80211_MACCMD_FLUSH (flush all addresses from the database), and .Dv IEEE80211_MACCMD_DETACH (detach the ACL subsystem, disabling it). The .Xr wlan_acl 4 module must be installed or .Er EINVAL will be returned. .It Dv IEEE80211_IOC_MESH_AP Set whether or not Mesh AP support is enabled using .Va i_val . .It Dv IEEE80211_IOC_MESH_FWRD Set whether or not packet forwarding support is enabled using .Va i_val . .It Dv IEEE80211_IOC_MESH_ID Set the Mesh ID using the value pointed to by .Va i_data . A Mesh ID can be up to .Dv IEEE80211_MESHID_LEN bytes long. .It Dv IEEE80211_IOC_MESH_PP_METRIC Set the link metric protocol using the value pointed to by .Va i_data . .It Dv IEEE80211_IOC_MESH_PP_PATH Set the path selection protocol using the value pointed to by .Va i_data . .It Dv IEEE80211_IOC_MESH_RTCMD Manipulate the state of the Mesh routing tables. Several requests are supported: .Dv IEEE80211_MESH_RTCMD_FLUSH (flush the contents of the routing table), .Dv IEEE80211_MESH_RTCMD_ADD (add an entry for the MAC address specified in .Va i_data and start the Peer discovery process), and .Dv IEEE80211_MESH_RTCMD_DELETE (delete the entry corresponding to the MAC address specified in .Va i_data ). .It Dv IEEE80211_IOC_MESH_TTL Set the Mesh Time To Live (TTL) setting installed in packets transmitted by this mesh node using .Va i_val . .It Dv IEEE80211_IOC_MLME Explicitly control the MLME state machine for a station using the MLME request pointed to by .Va i_data . There are several MLME operations supported: .Dv IEEE80211_MLME_ASSOC (request association to an access point), .Dv IEEE80211_MLME_DIASSOC (diassociate the specified station), .Dv IEEE80211_MLME_DEAUTH (deauthenticate the specified station), .Dv IEEE80211_MLME_AUHORIZE (mark the specified station authorized to pass data frames), .Dv IEEE80211_MLME_UNAUTHORIZE (revoke the specified station's ability to pass data frames), and .Dv IEEE80211_MLME_AUTH (request authentication to an access point). Note when this facility is used for stations operating in infrastructure mode the roaming mode should be set to manual. .It Dv IEEE80211_IOC_POWERSAVE Set the current powersaving mode to the value in .Va i_val . See .Dv IEEE80211_IOC_POWERSAVE above for valid values. This request causes a running interface to be reset. .It Dv IEEE80211_IOC_POWERSAVESLEEP Set the powersave sleep time in TU to the value in .Va i_val . This request causes a running interface to be reset. .It Dv IEEE80211_IOC_PRIVACY Set the current MLME setting for PRIVACY using the value in .Va i_val . See .Dv IEEE80211_IOC_PRIVACY above for details. .It Dv IEEE80211_IOC_PROTMODE Set the current 802.11g protection mode to the value in .Va i_val . This request causes a running interface to be reset. See .Dv IEEE80211_IOC_PROTMODE above for details. This request causes a running interface to be reset. .It Dv IEEE80211_IOC_PUREG Set whether ``pure 11g'' mode is enabled using the value in .Va i_val . This request causes a running interface to be restarted. See .Dv IEEE80211_IOC_PUREG above for details. .It Dv IEEE80211_IOC_PUREN Set whether ``pure 11n'' mode is enabled using the value in .Va i_val . This request causes a running interface to be restarted. See .Dv IEEE80211_IOC_PUREN above for details. .It Dv IEEE80211_IOC_REGDOMAIN Set the regulatory state using the data referenced by .Va i_data . This request can only be issued when all interfaces cloned from the underlying physical device are marked down; otherwise .Er EBUSY is returned. Note the new regulatory data may invalidate any desired channel. .It Dv IEEE80211_IOC_RIFS Set whether or not Reduced InterFrame Spacing (RIFS) is enabled using the value in .Va i_val . This setting is meaningful only when operating with 802.11n on an HT channel. This request causes a running interface to be reset. .It Dv IEEE80211_IOC_ROAM Set station roaming parameters using the data pointed to by .Va i_data . .It Dv IEEE80211_IOC_ROAMING Set the current roaming mode to the value in .Va i_val . See .Dv IEEE80211_IOC_ROAMING above for details. .It Dv IEEE80211_IOC_RTSTHRESHOLD Set the threshold (in bytes) for enabling transmission of RTS frames to the value in .Va i_val . This request causes a running interface to be reset. See .Dv IEEE80211_IOC_RTSTHRESHOLD above for details. .It Dv IEEE80211_IOC_SCANVALID Set the age (in seconds) that results from a scan operation will be considered valid. When scan results are no longer valid and they are needed (e.g. to roam) the system will initiate a scan operation to replenish the scan cache. .It Dv IEEE80211_IOC_SCAN_REQ Request a scan operation using the parameters pointed to by .Va i_val . The underlying device must be running or .Er ENXIO will be returned. Values for .Va sr_duration , .Va sr_mindwell , and .Va sr_maxdwell shorter than 1 clock tick are rounded up to a tick. If more SSID's are supplied than the system is capable of handling the extra ones are silently ignored. If a scan operation is already in progress the request will be (silently) ignored. .It Dv IEEE80211_IOC_SCAN_CANCEL Cancel any pending/active scan operation. .It Dv IEEE80211_IOC_SHORTGI Set whether or not Short Guard Interval (SGI) is enabled using the value in .Va i_val . Note SGI is only used when operating on an HT (802.11n) channel. This request causes a running interface to be reset. .It Dv IEEE80211_IOC_SMPS Set the Spatial Multiplexing Power Save (SMPS) setting to the value in .Va i_val . This request causes a running interface to be reset. See .Dv IEEE80211_IOC_SMPS above for details. .It Dv IEEE80211_IOC_SSID Set the desired SSID using the value pointed to by .Va i_data . The string may be at most IEEE80211_NWID_LEN bytes. This request causes a running interface to be restarted. .It Dv IEEE80211_IOC_STA_STATS Clear accumulated statistics for the specified station. .It Dv IEEE80211_IOC_STA_VLAN Set the VLAN tag for the specified station using the information pointed to by .Va i_data . .It Dv IEEE80211_IOC_TDMA_BINTERVAL Set the interval between Beacon frames to the value in .Va i_val . Values must be positive. This request causes a running interface to be reset. .It Dv IEEE80211_IOC_TDMA_SLOT Set the current TDMA slot to the value in .Va i_val . Values must be in the range [0-slotcnt]. Slot 0 identifies the master in the TDMA network; if it running it will immediately start sending Beacon frames. .It Dv IEEE80211_IOC_TDMA_SLOTCNT Set the number of slots in the TDMA network to the value in .Va i_val . This request causes a running interface to be reset. .It Dv IEEE80211_IOC_TDMA_SLOTLEN Set the length of the TDMA slot assigned to each station in the network to the value in .Va i_val . Slot lengths must be in the range 200 usecs to 1024 milliseconds (though values outside the range 1-200ms are unlikely to work well). This request causes a running interface to be reset. .It Dv IEEE80211_IOC_TSN Set whether or not Transitional Security Network (TSN) is enabled using the value in .Va i_val . .It Dv IEEE80211_IOC_TURBOP Set whether Atheros Dynamic Turbo mode is enabled using the value in .Va i_val . This request causes a running interface to be restarted. .It Dv IEEE80211_IOC_TXPARAMS Set transmit parameters using the data pointed to be .Va i_data . This request causes a running interface to be restarted. .It Dv IEEE80211_IOC_TXPOWER Set the maximum transmit power limit in .5 dBm units to the value in .Va i_val . This request causes a running interface to be reset. .It Dv IEEE80211_IOC_WEP Set the current WEP mode to the value in .Va i_val . See .Dv IEEE80211_IOC_WEP above for valid values. This request causes a running interface to be restarted. Note this request is deprecated; use .Dv IEEE80211_IOC_PRIVACY and .Dv IEEE80211_IOC_DROPUNENCRYPTED instead. .It Dv IEEE80211_IOC_WEPKEY Set the WEP key indicated by .Va i_val using the data pointed to by .Va i_data . Note this request is deprecated; use .Dv IEEE80211_IOC_WPAKEY instead. .It Dv IEEE80211_IOC_WEPTXKEY Set the default transmit key used for transmission to the value in .Va i_val . .It Dv IEEE80211_IOC_WME Set whether or not WME/WMM support is enabled using the value in .Va i_val . This request causes a running interface to be reset. .It Dv IEEE80211_IOC_WME_ACKPOLICY Set the WME ACK Policy for the Access Class (AC) specified in .Va i_len using the value in .Va i_val . .It Dv IEEE80211_IOC_WME_ACM Set the WME Admission Control Mechanism for the Access Class (AC) specified in .Va i_len using the value in .Va i_val . .It Dv IEEE80211_IOC_WME_AIFS Set the WME AIFS parameter for the Access Class (AC) specified in .Va i_len using the value in .Va i_val . .It Dv IEEE80211_IOC_WME_CWMAX Set the WME CWmax parameter for the Access Class (AC) specified in .Va i_len using the value in .Va i_val . .It Dv IEEE80211_IOC_WME_CWMIN Set the WME CWmin parameter for the Access Class (AC) specified in .Va i_len using the value in .Va i_val . .It Dv IEEE80211_IOC_WME_TXOPLIMIT Set the WME TxOpLimit parameter for the Access Class (AC) specified in .Va i_len using the value in .Va i_val . .It Dv IEEE80211_IOC_WPA Set the WPA configuration using the value in .Va i_val . This request causes a running interface to be reset. See .Dv IEEE80211_IOC_WPA above for details. .It Dv IEEE80211_IOC_WPAKEY Set the requested cryptographic key using data in the buffer pointed to by .Va i_data . See .Dv IEEE80211_IOC_WPAKEY for details. .It Dv IEEE80211_IOC_WPS Set whether or not Wi-FI Protected Setup (WPS) is enabled using the value in .Va i_val . .El .Sh SEE ALSO .Xr ioctl 2 , .Xr wlan 4 , .Xr wlan_acl 4 , .Xr wlan_xauth 4 , .Xr hostapd 8 , .Xr ifconfig 8 , .Xr wpa_supplicant 8 Index: projects/clang1100-import/share/man/man9/seqc.9 =================================================================== --- projects/clang1100-import/share/man/man9/seqc.9 (revision 364034) +++ projects/clang1100-import/share/man/man9/seqc.9 (revision 364035) @@ -1,138 +1,139 @@ .\" .\" Copyright (C) 2019 Mariusz Zaborski .\" .\" 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(s), this list of conditions and the following disclaimer as .\" the first lines of this file unmodified other than the possible .\" addition of one or more copyright notices. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice(s), 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 COPYRIGHT HOLDER(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 COPYRIGHT HOLDER(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. .\" .\" $FreeBSD$ .\" .Dd July 29, 2019 .Dt SEQC 9 .Os .Sh NAME .Nm seqc_consistent , .Nm seqc_read , .Nm seqc_write_begin , .Nm seqc_write_end .Nd "lockless read algorithm" .Sh SYNOPSIS .In sys/seqc.h .Ft void .Fn seqc_write_begin "seqc_t *seqcp" .Ft void .Fn seqc_write_end "seqc_t *seqcp" .Ft seqc_t .Fn seqc_read "seqc_t *seqcp" .Ft seqc_t .Fn seqc_consistent "const seqc_t *seqcp" "seqc_t oldseqc" .Sh DESCRIPTION The .Nm seqc allows zero or more readers and zero or one writer to concurrently access an object, providing a consistent snapshot of the object for readers. No mutual exclusion between readers and writers is required, but readers may be starved indefinitely by writers. .Pp The functions .Fn seqc_write_begin and .Fn seqc_write_end are used to create a transaction for writer, and notify the readers that the object will be modified. .Pp The .Fn seqc_read function returns the current sequence number. If a writer has started a transaction, this function will spin until the transaction has ended. .Pp The .Fn seqc_consistent function compares the sequence number with a previously fetched value. The .Fa oldseqc variable should contain a sequence number from the beginning of read transaction. .Pp The reader at the end of a transaction checks if the sequence number has changed. If the sequence number didn't change the object wasn't modified, and fetched variables are valid. If the sequence number changed the object was modified and the fetch should be repeated. In case when sequence number is odd the object change is in progress and the reader will wait until the write will the sequence number will become even. .Sh EXAMPLES The following example for a writer changees the .Va var1 and .Va var2 variables in the .Va obj structure: .Bd -literal lock_exclusive(&obj->lock); seqc_write_begin(&obj->seqc); obj->var1 = 1; obj->var2 = 2; seqc_write_end(&obj->seqc); unlock_exclusive(&obj->lock); .Ed +.Pp The following example for a reader reads the .Va var1 and .Va var2 variables from the .Va obj structure. In the case where the sequence number was changed it restarts the whole process. .Bd -literal int var1, var2; seqc_t seqc; for (;;) { seqc = seqc_read(&obj->seqc); var1 = obj->var1; var2 = obj->var2; if (seqc_consistent(&obj->seqc, seqc)) break; } .Ed .Sh AUTHORS The .Nm seqc functions was implemented by .An Mateusz Guzik Aq Mt mjg@FreeBSD.org . This manual page was written by .An Mariusz Zaborski Aq Mt oshogbo@FreeBSD.org . .Sh CAVEATS There is no guarantee of progress for readers. In case when there are a lot of writers the reader can be starved. This concern may be solved by returning error after a few attempts. .Pp Theoretically if reading takes a very long time, and when there are many writers the counter may overflow and wrap around to the same value. In that case the reader will not notice that the object was changed. Given that this needs 4 billion transactional writes across a single contended reader, it is unlikely to ever happen. This could be avoided by extending the interface to allow 64-bit counters. Index: projects/clang1100-import/stand/defs.mk =================================================================== --- projects/clang1100-import/stand/defs.mk (revision 364034) +++ projects/clang1100-import/stand/defs.mk (revision 364035) @@ -1,242 +1,242 @@ # $FreeBSD$ .if !defined(__BOOT_DEFS_MK__) __BOOT_DEFS_MK__=${MFILE} # We need to define all the MK_ options before including src.opts.mk # because it includes bsd.own.mk which needs the right MK_ values, # espeically MK_CTF. MK_CTF= no MK_SSP= no MK_PROFILE= no MAN= .if !defined(PIC) NO_PIC= INTERNALLIB= .endif .include .include WARNS?= 1 BOOTSRC= ${SRCTOP}/stand EFISRC= ${BOOTSRC}/efi EFIINC= ${EFISRC}/include EFIINCMD= ${EFIINC}/${MACHINE} FDTSRC= ${BOOTSRC}/fdt FICLSRC= ${BOOTSRC}/ficl LDRSRC= ${BOOTSRC}/common LIBLUASRC= ${BOOTSRC}/liblua LIBOFWSRC= ${BOOTSRC}/libofw LUASRC= ${SRCTOP}/contrib/lua/src SASRC= ${BOOTSRC}/libsa SYSDIR= ${SRCTOP}/sys UBOOTSRC= ${BOOTSRC}/uboot ZFSSRC= ${SASRC}/zfs LIBCSRC= ${SRCTOP}/lib/libc BOOTOBJ= ${OBJTOP}/stand # BINDIR is where we install BINDIR?= /boot # LUAPATH is where we search for and install lua scripts. LUAPATH?= /boot/lua FLUASRC?= ${SRCTOP}/libexec/flua LIBSA= ${BOOTOBJ}/libsa/libsa.a .if ${MACHINE} == "i386" LIBSA32= ${LIBSA} .else LIBSA32= ${BOOTOBJ}/libsa32/libsa32.a .endif # Standard options: CFLAGS+= -nostdinc .if ${MACHINE_ARCH} == "amd64" && ${DO32:U0} == 1 CFLAGS+= -I${BOOTOBJ}/libsa32 .else CFLAGS+= -I${BOOTOBJ}/libsa .endif CFLAGS+= -I${SASRC} -D_STANDALONE CFLAGS+= -I${SYSDIR} # Spike the floating point interfaces CFLAGS+= -Ddouble=jagged-little-pill -Dfloat=floaty-mcfloatface .if ${MACHINE_ARCH} == "i386" || ${MACHINE_ARCH} == "amd64" # Slim down the image. This saves about 15% in size with clang 6 on x86 # Our most constrained /boot/loader env is BIOS booting on x86, where # our text + data + BTX have to fit into 640k below the ISA hole. # Experience has shown that problems arise between ~520k to ~530k. CFLAGS.clang+= -Oz CFLAGS.gcc+= -Os CFLAGS+= -ffunction-sections -fdata-sections .endif # GELI Support, with backward compat hooks (mostly) .if defined(LOADER_NO_GELI_SUPPORT) MK_LOADER_GELI=no .warning "Please move from LOADER_NO_GELI_SUPPORT to WITHOUT_LOADER_GELI" .endif .if defined(LOADER_GELI_SUPPORT) MK_LOADER_GELI=yes .warning "Please move from LOADER_GELI_SUPPORT to WITH_LOADER_GELI" .endif .if ${MK_LOADER_GELI} == "yes" CFLAGS+= -DLOADER_GELI_SUPPORT CFLAGS+= -I${SASRC}/geli .endif # MK_LOADER_GELI # These should be confined to loader.mk, but can't because uboot/lib # also uses it. It's part of loader, but isn't a loader so we can't # just include loader.mk .if ${LOADER_DISK_SUPPORT:Uyes} == "yes" CFLAGS+= -DLOADER_DISK_SUPPORT .endif # Machine specific flags for all builds here # All PowerPC builds are 32 bit. We have no 64-bit loaders on powerpc # or powerpc64. .if ${MACHINE_ARCH} == "powerpc64" CFLAGS+= -m32 -mcpu=powerpc .endif # For amd64, there's a bit of mixed bag. Some of the tree (i386, lib*32) is # build 32-bit and some 64-bit (lib*, efi). Centralize all the 32-bit magic here # and activate it when DO32 is explicitly defined to be 1. .if ${MACHINE_ARCH} == "amd64" && ${DO32:U0} == 1 CFLAGS+= -m32 # LD_FLAGS is passed directly to ${LD}, not via ${CC}: LD_FLAGS+= -m elf_i386_fbsd AFLAGS+= --32 .endif SSP_CFLAGS= # Add in the no float / no SIMD stuff and announce we're freestanding # aarch64 and riscv don't have -msoft-float, but all others do. CFLAGS+= -ffreestanding ${CFLAGS_NO_SIMD} .if ${MACHINE_CPUARCH} == "aarch64" CFLAGS+= -mgeneral-regs-only -ffixed-x18 -fPIC .elif ${MACHINE_CPUARCH} == "riscv" CFLAGS+= -march=rv64imac -mabi=lp64 -fPIC CFLAGS.clang+= -mcmodel=medium CFLAGS.gcc+= -mcmodel=medany .else CFLAGS+= -msoft-float .endif # -msoft-float seems to be insufficient for powerpcspe .if ${MACHINE_ARCH} == "powerpcspe" CFLAGS+= -mno-spe .endif .if ${MACHINE_CPUARCH} == "i386" || (${MACHINE_CPUARCH} == "amd64" && ${DO32:U0} == 1) CFLAGS+= -march=i386 CFLAGS.gcc+= -mpreferred-stack-boundary=2 .endif .if ${MACHINE_CPUARCH} == "amd64" && ${DO32:U0} == 0 CFLAGS+= -fPIC -mno-red-zone .endif .if ${MACHINE_CPUARCH} == "arm" # Do not generate movt/movw, because the relocation fixup for them does not # translate to the -Bsymbolic -pie format required by self_reloc() in loader(8). # Also, the fpu is not available in a standalone environment. CFLAGS.clang+= -mno-movt CFLAGS.clang+= -mfpu=none CFLAGS+= -fPIC .endif # Some RISC-V linkers have support for relaxations, while some (lld) do not # yet. If this is the case we inhibit the compiler from emitting relaxations. .if ${LINKER_FEATURES:Mriscv-relaxations} == "" CFLAGS+= -mno-relax .endif # The boot loader build uses dd status=none, where possible, for reproducible # build output (since performance varies from run to run). Trouble is that # option was recently (10.3) added to FreeBSD and is non-standard. Only use it # when this test succeeds rather than require dd to be a bootstrap tool. DD_NOSTATUS!=(dd status=none count=0 2> /dev/null && echo status=none) || true DD=dd ${DD_NOSTATUS} .if ${MACHINE_CPUARCH} == "mips" CFLAGS+= -G0 -fno-pic -mno-abicalls .endif .if ${MK_LOADER_FORCE_LE} != "no" .if ${MACHINE_ARCH} == "powerpc64" CFLAGS+= -mlittle-endian .endif .endif # # Have a sensible default # .if ${MK_LOADER_LUA} == "yes" LOADER_DEFAULT_INTERP?=lua .elif ${MK_FORTH} == "yes" LOADER_DEFAULT_INTERP?=4th .else LOADER_DEFAULT_INTERP?=simp .endif LOADER_INTERP?=${LOADER_DEFAULT_INTERP} # Make sure we use the machine link we're about to create CFLAGS+=-I. all: ${PROG} CLEANFILES+= teken_state.h teken.c: teken_state.h teken_state.h: ${SYSDIR}/teken/sequences awk -f ${SYSDIR}/teken/gensequences \ ${SYSDIR}/teken/sequences > teken_state.h .if !defined(NO_OBJ) _ILINKS=include/machine .if ${MACHINE} != ${MACHINE_CPUARCH} && ${MACHINE} != "arm64" _ILINKS+=include/${MACHINE_CPUARCH} .endif .if ${MACHINE_CPUARCH} == "i386" || ${MACHINE_CPUARCH} == "amd64" _ILINKS+=include/x86 .endif CFLAGS+= -Iinclude CLEANDIRS+= include beforedepend: ${_ILINKS} beforebuild: ${_ILINKS} # Ensure that the links exist without depending on it when it exists which # causes all the modules to be rebuilt when the directory pointed to changes. .for _link in ${_ILINKS} .if !exists(${.OBJDIR}/${_link}) ${OBJS}: ${_link} .endif # _link exists .endfor .NOPATH: ${_ILINKS} ${_ILINKS}: .NOMETA @case ${.TARGET:T} in \ machine) \ if [ ${DO32:U0} -eq 0 ]; then \ path=${SYSDIR}/${MACHINE}/include ; \ else \ path=${SYSDIR}/${MACHINE:C/amd64/i386/}/include ; \ fi ;; \ *) \ path=${SYSDIR}/${.TARGET:T}/include ;; \ esac ; \ case ${.TARGET} in \ */*) mkdir -p ${.TARGET:H};; \ esac ; \ path=`(cd $$path && /bin/pwd)` ; \ ${ECHO} ${.TARGET} "->" $$path ; \ - ln -fhs $$path ${.TARGET} + ln -fns $$path ${.TARGET} .endif # !NO_OBJ .endif # __BOOT_DEFS_MK__ Index: projects/clang1100-import/sys/cddl/compat/opensolaris/sys/assfail.h =================================================================== --- projects/clang1100-import/sys/cddl/compat/opensolaris/sys/assfail.h (revision 364034) +++ projects/clang1100-import/sys/cddl/compat/opensolaris/sys/assfail.h (revision 364035) @@ -1,88 +1,86 @@ /*- * Copyright (c) 2012 Martin Matuska * 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. * * $FreeBSD$ */ #ifndef _OPENSOLARIS_SYS_ASSFAIL_H_ #define _OPENSOLARIS_SYS_ASSFAIL_H_ #include #ifndef _KERNEL #include #include #endif #ifdef __cplusplus extern "C" { #endif #ifdef _KERNEL int assfail(const char *, const char *, int); void assfail3(const char *, uintmax_t, const char *, uintmax_t, const char *, int); #else /* !defined(_KERNEL) */ #ifndef HAVE_ASSFAIL extern int aok; -__inline int __assfail(const char *expr, const char *file, int line); - -__inline int +static __inline int __assfail(const char *expr, const char *file, int line) { (void)fprintf(stderr, "Assertion failed: (%s), file %s, line %d.\n", expr, file, line); if (!aok) abort(); return (0); } #define assfail __assfail #endif #ifndef HAVE_ASSFAIL3 extern int aok; static __inline void __assfail3(const char *expr, uintmax_t lv, const char *op, uintmax_t rv, const char *file, int line) { (void)fprintf(stderr, "Assertion failed: %s (0x%jx %s 0x%jx), file %s, line %d.\n", expr, lv, op, rv, file, line); if (!aok) abort(); } #define assfail3 __assfail3 #endif #endif /* !defined(_KERNEL) */ #ifdef __cplusplus } #endif #endif /* _OPENSOLARIS_SYS_ASSFAIL_H_ */ Index: projects/clang1100-import/sys/cddl/compat/opensolaris/sys/stat.h =================================================================== --- projects/clang1100-import/sys/cddl/compat/opensolaris/sys/stat.h (revision 364034) +++ projects/clang1100-import/sys/cddl/compat/opensolaris/sys/stat.h (revision 364035) @@ -1,56 +1,65 @@ /* * Copyright (C) 2007 John Birrell * 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 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 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 _COMPAT_OPENSOLARIS_SYS_STAT_H_ #define _COMPAT_OPENSOLARIS_SYS_STAT_H_ #include_next +/* + * When bootstrapping on Linux a stat64/fstat64 functions exists in both + * glibc and musl libc. To avoid compilation errors, use those functions instead + * of redefining them to stat/fstat. + * Similarly, macOS provides (deprecated) stat64 functions that we can use + * for now. + */ +#if !defined(__linux__) && !defined(__APPLE__) #define stat64 stat #define MAXOFFSET_T OFF_MAX -#ifndef _KERNEL +#if !defined(_KERNEL) #include static __inline int fstat64(int fd, struct stat *sb) { int ret; ret = fstat(fd, sb); if (ret == 0) { if (S_ISCHR(sb->st_mode)) (void)ioctl(fd, DIOCGMEDIASIZE, &sb->st_size); } return (ret); } -#endif +#endif /* !defined(_KERNEL) */ +#endif /* !defined(__linux__) && !defined(__APPLE__) */ #endif /* !_COMPAT_OPENSOLARIS_SYS_STAT_H_ */ Index: projects/clang1100-import/sys/cddl/compat/opensolaris/sys/time.h =================================================================== --- projects/clang1100-import/sys/cddl/compat/opensolaris/sys/time.h (revision 364034) +++ projects/clang1100-import/sys/cddl/compat/opensolaris/sys/time.h (revision 364035) @@ -1,94 +1,95 @@ /*- * Copyright (c) 2007 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. * * $FreeBSD$ */ #ifndef _OPENSOLARIS_SYS_TIME_H_ #define _OPENSOLARIS_SYS_TIME_H_ +#include #include_next #define SEC 1 #define MILLISEC 1000 #define MICROSEC 1000000 #define NANOSEC 1000000000 #define TIME_MAX LLONG_MAX #define MSEC2NSEC(m) ((hrtime_t)(m) * (NANOSEC / MILLISEC)) #define NSEC2MSEC(n) ((n) / (NANOSEC / MILLISEC)) #define USEC2NSEC(m) ((hrtime_t)(m) * (NANOSEC / MICROSEC)) #define NSEC2USEC(n) ((n) / (NANOSEC / MICROSEC)) #define NSEC2SEC(n) ((n) / (NANOSEC / SEC)) #define SEC2NSEC(m) ((hrtime_t)(m) * (NANOSEC / SEC)) typedef longlong_t hrtime_t; #if defined(__i386__) || defined(__powerpc__) #define TIMESPEC_OVERFLOW(ts) \ ((ts)->tv_sec < INT32_MIN || (ts)->tv_sec > INT32_MAX) #else #define TIMESPEC_OVERFLOW(ts) \ ((ts)->tv_sec < INT64_MIN || (ts)->tv_sec > INT64_MAX) #endif #define SEC_TO_TICK(sec) ((sec) * hz) #define NSEC_TO_TICK(nsec) ((nsec) / (NANOSEC / hz)) #ifdef _KERNEL static __inline hrtime_t gethrtime(void) { struct timespec ts; hrtime_t nsec; getnanouptime(&ts); nsec = (hrtime_t)ts.tv_sec * NANOSEC + ts.tv_nsec; return (nsec); } #define gethrestime_sec() (time_second) #define gethrestime(ts) getnanotime(ts) #define gethrtime_waitfree() gethrtime() extern int nsec_per_tick; /* nanoseconds per clock tick */ #define ddi_get_lbolt64() \ (int64_t)(((getsbinuptime() >> 16) * hz) >> 16) #define ddi_get_lbolt() (clock_t)ddi_get_lbolt64() #else static __inline hrtime_t gethrtime(void) { struct timespec ts; clock_gettime(CLOCK_UPTIME,&ts); return (((u_int64_t) ts.tv_sec) * NANOSEC + ts.tv_nsec); } #endif /* _KERNEL */ #endif /* !_OPENSOLARIS_SYS_TIME_H_ */ Index: projects/clang1100-import/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dbuf.c =================================================================== --- projects/clang1100-import/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dbuf.c (revision 364034) +++ projects/clang1100-import/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dbuf.c (revision 364035) @@ -1,4255 +1,4248 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2012, 2018 by Delphix. All rights reserved. * Copyright (c) 2013 by Saso Kiselkov. All rights reserved. * Copyright (c) 2013, Joyent, Inc. All rights reserved. * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved. * Copyright (c) 2014 Integros [integros.com] */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include kstat_t *dbuf_ksp; typedef struct dbuf_stats { /* * Various statistics about the size of the dbuf cache. */ kstat_named_t cache_count; kstat_named_t cache_size_bytes; kstat_named_t cache_size_bytes_max; /* * Statistics regarding the bounds on the dbuf cache size. */ kstat_named_t cache_target_bytes; kstat_named_t cache_lowater_bytes; kstat_named_t cache_hiwater_bytes; /* * Total number of dbuf cache evictions that have occurred. */ kstat_named_t cache_total_evicts; /* * The distribution of dbuf levels in the dbuf cache and * the total size of all dbufs at each level. */ kstat_named_t cache_levels[DN_MAX_LEVELS]; kstat_named_t cache_levels_bytes[DN_MAX_LEVELS]; /* * Statistics about the dbuf hash table. */ kstat_named_t hash_hits; kstat_named_t hash_misses; kstat_named_t hash_collisions; kstat_named_t hash_elements; kstat_named_t hash_elements_max; /* * Number of sublists containing more than one dbuf in the dbuf * hash table. Keep track of the longest hash chain. */ kstat_named_t hash_chains; kstat_named_t hash_chain_max; /* * Number of times a dbuf_create() discovers that a dbuf was * already created and in the dbuf hash table. */ kstat_named_t hash_insert_race; /* * Statistics about the size of the metadata dbuf cache. */ kstat_named_t metadata_cache_count; kstat_named_t metadata_cache_size_bytes; kstat_named_t metadata_cache_size_bytes_max; /* * For diagnostic purposes, this is incremented whenever we can't add * something to the metadata cache because it's full, and instead put * the data in the regular dbuf cache. */ kstat_named_t metadata_cache_overflow; } dbuf_stats_t; dbuf_stats_t dbuf_stats = { { "cache_count", KSTAT_DATA_UINT64 }, { "cache_size_bytes", KSTAT_DATA_UINT64 }, { "cache_size_bytes_max", KSTAT_DATA_UINT64 }, { "cache_target_bytes", KSTAT_DATA_UINT64 }, { "cache_lowater_bytes", KSTAT_DATA_UINT64 }, { "cache_hiwater_bytes", KSTAT_DATA_UINT64 }, { "cache_total_evicts", KSTAT_DATA_UINT64 }, { { "cache_levels_N", KSTAT_DATA_UINT64 } }, { { "cache_levels_bytes_N", KSTAT_DATA_UINT64 } }, { "hash_hits", KSTAT_DATA_UINT64 }, { "hash_misses", KSTAT_DATA_UINT64 }, { "hash_collisions", KSTAT_DATA_UINT64 }, { "hash_elements", KSTAT_DATA_UINT64 }, { "hash_elements_max", KSTAT_DATA_UINT64 }, { "hash_chains", KSTAT_DATA_UINT64 }, { "hash_chain_max", KSTAT_DATA_UINT64 }, { "hash_insert_race", KSTAT_DATA_UINT64 }, { "metadata_cache_count", KSTAT_DATA_UINT64 }, { "metadata_cache_size_bytes", KSTAT_DATA_UINT64 }, { "metadata_cache_size_bytes_max", KSTAT_DATA_UINT64 }, { "metadata_cache_overflow", KSTAT_DATA_UINT64 } }; #define DBUF_STAT_INCR(stat, val) \ atomic_add_64(&dbuf_stats.stat.value.ui64, (val)); #define DBUF_STAT_DECR(stat, val) \ DBUF_STAT_INCR(stat, -(val)); #define DBUF_STAT_BUMP(stat) \ DBUF_STAT_INCR(stat, 1); #define DBUF_STAT_BUMPDOWN(stat) \ DBUF_STAT_INCR(stat, -1); #define DBUF_STAT_MAX(stat, v) { \ uint64_t _m; \ while ((v) > (_m = dbuf_stats.stat.value.ui64) && \ (_m != atomic_cas_64(&dbuf_stats.stat.value.ui64, _m, (v))))\ continue; \ } struct dbuf_hold_impl_data { /* Function arguments */ dnode_t *dh_dn; uint8_t dh_level; uint64_t dh_blkid; boolean_t dh_fail_sparse; boolean_t dh_fail_uncached; void *dh_tag; dmu_buf_impl_t **dh_dbp; /* Local variables */ dmu_buf_impl_t *dh_db; dmu_buf_impl_t *dh_parent; blkptr_t *dh_bp; int dh_err; dbuf_dirty_record_t *dh_dr; int dh_depth; }; static void __dbuf_hold_impl_init(struct dbuf_hold_impl_data *dh, dnode_t *dn, uint8_t level, uint64_t blkid, boolean_t fail_sparse, boolean_t fail_uncached, void *tag, dmu_buf_impl_t **dbp, int depth); static int __dbuf_hold_impl(struct dbuf_hold_impl_data *dh); static boolean_t dbuf_undirty(dmu_buf_impl_t *db, dmu_tx_t *tx); static void dbuf_write(dbuf_dirty_record_t *dr, arc_buf_t *data, dmu_tx_t *tx); -#ifndef __lint -extern inline void dmu_buf_init_user(dmu_buf_user_t *dbu, - dmu_buf_evict_func_t *evict_func_sync, - dmu_buf_evict_func_t *evict_func_async, - dmu_buf_t **clear_on_evict_dbufp); -#endif /* ! __lint */ - /* * Global data structures and functions for the dbuf cache. */ static kmem_cache_t *dbuf_kmem_cache; static taskq_t *dbu_evict_taskq; static kthread_t *dbuf_cache_evict_thread; static kmutex_t dbuf_evict_lock; static kcondvar_t dbuf_evict_cv; static boolean_t dbuf_evict_thread_exit; /* * There are two dbuf caches; each dbuf can only be in one of them at a time. * * 1. Cache of metadata dbufs, to help make read-heavy administrative commands * from /sbin/zfs run faster. The "metadata cache" specifically stores dbufs * that represent the metadata that describes filesystems/snapshots/ * bookmarks/properties/etc. We only evict from this cache when we export a * pool, to short-circuit as much I/O as possible for all administrative * commands that need the metadata. There is no eviction policy for this * cache, because we try to only include types in it which would occupy a * very small amount of space per object but create a large impact on the * performance of these commands. Instead, after it reaches a maximum size * (which should only happen on very small memory systems with a very large * number of filesystem objects), we stop taking new dbufs into the * metadata cache, instead putting them in the normal dbuf cache. * * 2. LRU cache of dbufs. The dbuf cache maintains a list of dbufs that * are not currently held but have been recently released. These dbufs * are not eligible for arc eviction until they are aged out of the cache. * Dbufs that are aged out of the cache will be immediately destroyed and * become eligible for arc eviction. * * Dbufs are added to these caches once the last hold is released. If a dbuf is * later accessed and still exists in the dbuf cache, then it will be removed * from the cache and later re-added to the head of the cache. * * If a given dbuf meets the requirements for the metadata cache, it will go * there, otherwise it will be considered for the generic LRU dbuf cache. The * caches and the refcounts tracking their sizes are stored in an array indexed * by those caches' matching enum values (from dbuf_cached_state_t). */ typedef struct dbuf_cache { multilist_t *cache; zfs_refcount_t size; } dbuf_cache_t; dbuf_cache_t dbuf_caches[DB_CACHE_MAX]; /* Size limits for the caches */ uint64_t dbuf_cache_max_bytes = 0; uint64_t dbuf_metadata_cache_max_bytes = 0; /* Set the default sizes of the caches to log2 fraction of arc size */ int dbuf_cache_shift = 5; int dbuf_metadata_cache_shift = 6; /* * For diagnostic purposes, this is incremented whenever we can't add * something to the metadata cache because it's full, and instead put * the data in the regular dbuf cache. */ uint64_t dbuf_metadata_cache_overflow; /* * The LRU dbuf cache uses a three-stage eviction policy: * - A low water marker designates when the dbuf eviction thread * should stop evicting from the dbuf cache. * - When we reach the maximum size (aka mid water mark), we * signal the eviction thread to run. * - The high water mark indicates when the eviction thread * is unable to keep up with the incoming load and eviction must * happen in the context of the calling thread. * * The dbuf cache: * (max size) * low water mid water hi water * +----------------------------------------+----------+----------+ * | | | | * | | | | * | | | | * | | | | * +----------------------------------------+----------+----------+ * stop signal evict * evicting eviction directly * thread * * The high and low water marks indicate the operating range for the eviction * thread. The low water mark is, by default, 90% of the total size of the * cache and the high water mark is at 110% (both of these percentages can be * changed by setting dbuf_cache_lowater_pct and dbuf_cache_hiwater_pct, * respectively). The eviction thread will try to ensure that the cache remains * within this range by waking up every second and checking if the cache is * above the low water mark. The thread can also be woken up by callers adding * elements into the cache if the cache is larger than the mid water (i.e max * cache size). Once the eviction thread is woken up and eviction is required, * it will continue evicting buffers until it's able to reduce the cache size * to the low water mark. If the cache size continues to grow and hits the high * water mark, then callers adding elments to the cache will begin to evict * directly from the cache until the cache is no longer above the high water * mark. */ /* * The percentage above and below the maximum cache size. */ uint_t dbuf_cache_hiwater_pct = 10; uint_t dbuf_cache_lowater_pct = 10; SYSCTL_DECL(_vfs_zfs); SYSCTL_QUAD(_vfs_zfs, OID_AUTO, dbuf_cache_max_bytes, CTLFLAG_RWTUN, &dbuf_cache_max_bytes, 0, "dbuf cache size in bytes"); SYSCTL_QUAD(_vfs_zfs, OID_AUTO, dbuf_metadata_cache_max_bytes, CTLFLAG_RWTUN, &dbuf_metadata_cache_max_bytes, 0, "dbuf metadata cache size in bytes"); SYSCTL_INT(_vfs_zfs, OID_AUTO, dbuf_cache_shift, CTLFLAG_RDTUN, &dbuf_cache_shift, 0, "dbuf cache size as log2 fraction of ARC"); SYSCTL_INT(_vfs_zfs, OID_AUTO, dbuf_metadata_cache_shift, CTLFLAG_RDTUN, &dbuf_metadata_cache_shift, 0, "dbuf metadata cache size as log2 fraction of ARC"); SYSCTL_QUAD(_vfs_zfs, OID_AUTO, dbuf_metadata_cache_overflow, CTLFLAG_RD, &dbuf_metadata_cache_overflow, 0, "dbuf metadata cache overflow"); SYSCTL_UINT(_vfs_zfs, OID_AUTO, dbuf_cache_hiwater_pct, CTLFLAG_RWTUN, &dbuf_cache_hiwater_pct, 0, "max percents above the dbuf cache size"); SYSCTL_UINT(_vfs_zfs, OID_AUTO, dbuf_cache_lowater_pct, CTLFLAG_RWTUN, &dbuf_cache_lowater_pct, 0, "max percents below the dbuf cache size"); /* ARGSUSED */ static int dbuf_cons(void *vdb, void *unused, int kmflag) { dmu_buf_impl_t *db = vdb; bzero(db, sizeof (dmu_buf_impl_t)); mutex_init(&db->db_mtx, NULL, MUTEX_DEFAULT, NULL); cv_init(&db->db_changed, NULL, CV_DEFAULT, NULL); multilist_link_init(&db->db_cache_link); zfs_refcount_create(&db->db_holds); return (0); } /* ARGSUSED */ static void dbuf_dest(void *vdb, void *unused) { dmu_buf_impl_t *db = vdb; mutex_destroy(&db->db_mtx); cv_destroy(&db->db_changed); ASSERT(!multilist_link_active(&db->db_cache_link)); zfs_refcount_destroy(&db->db_holds); } /* * dbuf hash table routines */ static dbuf_hash_table_t dbuf_hash_table; static uint64_t dbuf_hash_count; /* * We use Cityhash for this. It's fast, and has good hash properties without * requiring any large static buffers. */ static uint64_t dbuf_hash(void *os, uint64_t obj, uint8_t lvl, uint64_t blkid) { return (cityhash4((uintptr_t)os, obj, (uint64_t)lvl, blkid)); } #define DBUF_EQUAL(dbuf, os, obj, level, blkid) \ ((dbuf)->db.db_object == (obj) && \ (dbuf)->db_objset == (os) && \ (dbuf)->db_level == (level) && \ (dbuf)->db_blkid == (blkid)) dmu_buf_impl_t * dbuf_find(objset_t *os, uint64_t obj, uint8_t level, uint64_t blkid) { dbuf_hash_table_t *h = &dbuf_hash_table; uint64_t hv = dbuf_hash(os, obj, level, blkid); uint64_t idx = hv & h->hash_table_mask; dmu_buf_impl_t *db; mutex_enter(DBUF_HASH_MUTEX(h, idx)); for (db = h->hash_table[idx]; db != NULL; db = db->db_hash_next) { if (DBUF_EQUAL(db, os, obj, level, blkid)) { mutex_enter(&db->db_mtx); if (db->db_state != DB_EVICTING) { mutex_exit(DBUF_HASH_MUTEX(h, idx)); return (db); } mutex_exit(&db->db_mtx); } } mutex_exit(DBUF_HASH_MUTEX(h, idx)); return (NULL); } static dmu_buf_impl_t * dbuf_find_bonus(objset_t *os, uint64_t object) { dnode_t *dn; dmu_buf_impl_t *db = NULL; if (dnode_hold(os, object, FTAG, &dn) == 0) { rw_enter(&dn->dn_struct_rwlock, RW_READER); if (dn->dn_bonus != NULL) { db = dn->dn_bonus; mutex_enter(&db->db_mtx); } rw_exit(&dn->dn_struct_rwlock); dnode_rele(dn, FTAG); } return (db); } /* * Insert an entry into the hash table. If there is already an element * equal to elem in the hash table, then the already existing element * will be returned and the new element will not be inserted. * Otherwise returns NULL. */ static dmu_buf_impl_t * dbuf_hash_insert(dmu_buf_impl_t *db) { dbuf_hash_table_t *h = &dbuf_hash_table; objset_t *os = db->db_objset; uint64_t obj = db->db.db_object; int level = db->db_level; uint64_t blkid, hv, idx; dmu_buf_impl_t *dbf; uint32_t i; blkid = db->db_blkid; hv = dbuf_hash(os, obj, level, blkid); idx = hv & h->hash_table_mask; mutex_enter(DBUF_HASH_MUTEX(h, idx)); for (dbf = h->hash_table[idx], i = 0; dbf != NULL; dbf = dbf->db_hash_next, i++) { if (DBUF_EQUAL(dbf, os, obj, level, blkid)) { mutex_enter(&dbf->db_mtx); if (dbf->db_state != DB_EVICTING) { mutex_exit(DBUF_HASH_MUTEX(h, idx)); return (dbf); } mutex_exit(&dbf->db_mtx); } } if (i > 0) { DBUF_STAT_BUMP(hash_collisions); if (i == 1) DBUF_STAT_BUMP(hash_chains); DBUF_STAT_MAX(hash_chain_max, i); } mutex_enter(&db->db_mtx); db->db_hash_next = h->hash_table[idx]; h->hash_table[idx] = db; mutex_exit(DBUF_HASH_MUTEX(h, idx)); atomic_inc_64(&dbuf_hash_count); DBUF_STAT_MAX(hash_elements_max, dbuf_hash_count); return (NULL); } /* * Remove an entry from the hash table. It must be in the EVICTING state. */ static void dbuf_hash_remove(dmu_buf_impl_t *db) { dbuf_hash_table_t *h = &dbuf_hash_table; uint64_t hv, idx; dmu_buf_impl_t *dbf, **dbp; hv = dbuf_hash(db->db_objset, db->db.db_object, db->db_level, db->db_blkid); idx = hv & h->hash_table_mask; /* * We mustn't hold db_mtx to maintain lock ordering: * DBUF_HASH_MUTEX > db_mtx. */ ASSERT(zfs_refcount_is_zero(&db->db_holds)); ASSERT(db->db_state == DB_EVICTING); ASSERT(!MUTEX_HELD(&db->db_mtx)); mutex_enter(DBUF_HASH_MUTEX(h, idx)); dbp = &h->hash_table[idx]; while ((dbf = *dbp) != db) { dbp = &dbf->db_hash_next; ASSERT(dbf != NULL); } *dbp = db->db_hash_next; db->db_hash_next = NULL; if (h->hash_table[idx] && h->hash_table[idx]->db_hash_next == NULL) DBUF_STAT_BUMPDOWN(hash_chains); mutex_exit(DBUF_HASH_MUTEX(h, idx)); atomic_dec_64(&dbuf_hash_count); } typedef enum { DBVU_EVICTING, DBVU_NOT_EVICTING } dbvu_verify_type_t; static void dbuf_verify_user(dmu_buf_impl_t *db, dbvu_verify_type_t verify_type) { #ifdef ZFS_DEBUG int64_t holds; if (db->db_user == NULL) return; /* Only data blocks support the attachment of user data. */ ASSERT(db->db_level == 0); /* Clients must resolve a dbuf before attaching user data. */ ASSERT(db->db.db_data != NULL); ASSERT3U(db->db_state, ==, DB_CACHED); holds = zfs_refcount_count(&db->db_holds); if (verify_type == DBVU_EVICTING) { /* * Immediate eviction occurs when holds == dirtycnt. * For normal eviction buffers, holds is zero on * eviction, except when dbuf_fix_old_data() calls * dbuf_clear_data(). However, the hold count can grow * during eviction even though db_mtx is held (see * dmu_bonus_hold() for an example), so we can only * test the generic invariant that holds >= dirtycnt. */ ASSERT3U(holds, >=, db->db_dirtycnt); } else { if (db->db_user_immediate_evict == TRUE) ASSERT3U(holds, >=, db->db_dirtycnt); else ASSERT3U(holds, >, 0); } #endif } static void dbuf_evict_user(dmu_buf_impl_t *db) { dmu_buf_user_t *dbu = db->db_user; ASSERT(MUTEX_HELD(&db->db_mtx)); if (dbu == NULL) return; dbuf_verify_user(db, DBVU_EVICTING); db->db_user = NULL; #ifdef ZFS_DEBUG if (dbu->dbu_clear_on_evict_dbufp != NULL) *dbu->dbu_clear_on_evict_dbufp = NULL; #endif /* * There are two eviction callbacks - one that we call synchronously * and one that we invoke via a taskq. The async one is useful for * avoiding lock order reversals and limiting stack depth. * * Note that if we have a sync callback but no async callback, * it's likely that the sync callback will free the structure * containing the dbu. In that case we need to take care to not * dereference dbu after calling the sync evict func. */ boolean_t has_async = (dbu->dbu_evict_func_async != NULL); if (dbu->dbu_evict_func_sync != NULL) dbu->dbu_evict_func_sync(dbu); if (has_async) { taskq_dispatch_ent(dbu_evict_taskq, dbu->dbu_evict_func_async, dbu, 0, &dbu->dbu_tqent); } } boolean_t dbuf_is_metadata(dmu_buf_impl_t *db) { if (db->db_level > 0) { return (B_TRUE); } else { boolean_t is_metadata; DB_DNODE_ENTER(db); is_metadata = DMU_OT_IS_METADATA(DB_DNODE(db)->dn_type); DB_DNODE_EXIT(db); return (is_metadata); } } /* * This returns whether this dbuf should be stored in the metadata cache, which * is based on whether it's from one of the dnode types that store data related * to traversing dataset hierarchies. */ static boolean_t dbuf_include_in_metadata_cache(dmu_buf_impl_t *db) { DB_DNODE_ENTER(db); dmu_object_type_t type = DB_DNODE(db)->dn_type; DB_DNODE_EXIT(db); /* Check if this dbuf is one of the types we care about */ if (DMU_OT_IS_METADATA_CACHED(type)) { /* If we hit this, then we set something up wrong in dmu_ot */ ASSERT(DMU_OT_IS_METADATA(type)); /* * Sanity check for small-memory systems: don't allocate too * much memory for this purpose. */ if (zfs_refcount_count( &dbuf_caches[DB_DBUF_METADATA_CACHE].size) > dbuf_metadata_cache_max_bytes) { dbuf_metadata_cache_overflow++; DTRACE_PROBE1(dbuf__metadata__cache__overflow, dmu_buf_impl_t *, db); return (B_FALSE); } return (B_TRUE); } return (B_FALSE); } /* * This function *must* return indices evenly distributed between all * sublists of the multilist. This is needed due to how the dbuf eviction * code is laid out; dbuf_evict_thread() assumes dbufs are evenly * distributed between all sublists and uses this assumption when * deciding which sublist to evict from and how much to evict from it. */ unsigned int dbuf_cache_multilist_index_func(multilist_t *ml, void *obj) { dmu_buf_impl_t *db = obj; /* * The assumption here, is the hash value for a given * dmu_buf_impl_t will remain constant throughout it's lifetime * (i.e. it's objset, object, level and blkid fields don't change). * Thus, we don't need to store the dbuf's sublist index * on insertion, as this index can be recalculated on removal. * * Also, the low order bits of the hash value are thought to be * distributed evenly. Otherwise, in the case that the multilist * has a power of two number of sublists, each sublists' usage * would not be evenly distributed. */ return (dbuf_hash(db->db_objset, db->db.db_object, db->db_level, db->db_blkid) % multilist_get_num_sublists(ml)); } static inline unsigned long dbuf_cache_target_bytes(void) { return MIN(dbuf_cache_max_bytes, arc_max_bytes() >> dbuf_cache_shift); } static inline uint64_t dbuf_cache_hiwater_bytes(void) { uint64_t dbuf_cache_target = dbuf_cache_target_bytes(); return (dbuf_cache_target + (dbuf_cache_target * dbuf_cache_hiwater_pct) / 100); } static inline uint64_t dbuf_cache_lowater_bytes(void) { uint64_t dbuf_cache_target = dbuf_cache_target_bytes(); return (dbuf_cache_target - (dbuf_cache_target * dbuf_cache_lowater_pct) / 100); } static inline boolean_t dbuf_cache_above_lowater(void) { return (zfs_refcount_count(&dbuf_caches[DB_DBUF_CACHE].size) > dbuf_cache_lowater_bytes()); } /* * Evict the oldest eligible dbuf from the dbuf cache. */ static void dbuf_evict_one(void) { int idx = multilist_get_random_index(dbuf_caches[DB_DBUF_CACHE].cache); multilist_sublist_t *mls = multilist_sublist_lock( dbuf_caches[DB_DBUF_CACHE].cache, idx); ASSERT(!MUTEX_HELD(&dbuf_evict_lock)); dmu_buf_impl_t *db = multilist_sublist_tail(mls); while (db != NULL && mutex_tryenter(&db->db_mtx) == 0) { db = multilist_sublist_prev(mls, db); } DTRACE_PROBE2(dbuf__evict__one, dmu_buf_impl_t *, db, multilist_sublist_t *, mls); if (db != NULL) { multilist_sublist_remove(mls, db); multilist_sublist_unlock(mls); (void) zfs_refcount_remove_many( &dbuf_caches[DB_DBUF_CACHE].size, db->db.db_size, db); DBUF_STAT_BUMPDOWN(cache_levels[db->db_level]); DBUF_STAT_BUMPDOWN(cache_count); DBUF_STAT_DECR(cache_levels_bytes[db->db_level], db->db.db_size); ASSERT3U(db->db_caching_status, ==, DB_DBUF_CACHE); db->db_caching_status = DB_NO_CACHE; dbuf_destroy(db); DBUF_STAT_BUMP(cache_total_evicts); } else { multilist_sublist_unlock(mls); } } /* * The dbuf evict thread is responsible for aging out dbufs from the * cache. Once the cache has reached it's maximum size, dbufs are removed * and destroyed. The eviction thread will continue running until the size * of the dbuf cache is at or below the maximum size. Once the dbuf is aged * out of the cache it is destroyed and becomes eligible for arc eviction. */ /* ARGSUSED */ static void dbuf_evict_thread(void *unused __unused) { callb_cpr_t cpr; CALLB_CPR_INIT(&cpr, &dbuf_evict_lock, callb_generic_cpr, FTAG); mutex_enter(&dbuf_evict_lock); while (!dbuf_evict_thread_exit) { while (!dbuf_cache_above_lowater() && !dbuf_evict_thread_exit) { CALLB_CPR_SAFE_BEGIN(&cpr); (void) cv_timedwait_hires(&dbuf_evict_cv, &dbuf_evict_lock, SEC2NSEC(1), MSEC2NSEC(1), 0); CALLB_CPR_SAFE_END(&cpr, &dbuf_evict_lock); #ifdef __FreeBSD__ if (dbuf_ksp != NULL) dbuf_ksp->ks_update(dbuf_ksp, KSTAT_READ); #endif } mutex_exit(&dbuf_evict_lock); /* * Keep evicting as long as we're above the low water mark * for the cache. We do this without holding the locks to * minimize lock contention. */ while (dbuf_cache_above_lowater() && !dbuf_evict_thread_exit) { dbuf_evict_one(); } mutex_enter(&dbuf_evict_lock); } dbuf_evict_thread_exit = B_FALSE; cv_broadcast(&dbuf_evict_cv); CALLB_CPR_EXIT(&cpr); /* drops dbuf_evict_lock */ thread_exit(); } /* * Wake up the dbuf eviction thread if the dbuf cache is at its max size. * If the dbuf cache is at its high water mark, then evict a dbuf from the * dbuf cache using the callers context. */ static void dbuf_evict_notify(uint64_t size) { /* * We check if we should evict without holding the dbuf_evict_lock, * because it's OK to occasionally make the wrong decision here, * and grabbing the lock results in massive lock contention. */ if (size > dbuf_cache_max_bytes) { if (size > dbuf_cache_hiwater_bytes()) dbuf_evict_one(); cv_signal(&dbuf_evict_cv); } } static int dbuf_kstat_update(kstat_t *ksp, int rw) { dbuf_stats_t *ds = ksp->ks_data; if (rw == KSTAT_WRITE) { return (SET_ERROR(EACCES)); } else { ds->metadata_cache_size_bytes.value.ui64 = zfs_refcount_count(&dbuf_caches[DB_DBUF_METADATA_CACHE].size); ds->cache_size_bytes.value.ui64 = zfs_refcount_count(&dbuf_caches[DB_DBUF_CACHE].size); ds->cache_target_bytes.value.ui64 = dbuf_cache_target_bytes(); ds->cache_hiwater_bytes.value.ui64 = dbuf_cache_hiwater_bytes(); ds->cache_lowater_bytes.value.ui64 = dbuf_cache_lowater_bytes(); ds->hash_elements.value.ui64 = dbuf_hash_count; } return (0); } void dbuf_init(void) { uint64_t hsize = 1ULL << 16; dbuf_hash_table_t *h = &dbuf_hash_table; int i; /* * The hash table is big enough to fill all of physical memory * with an average 4K block size. The table will take up * totalmem*sizeof(void*)/4K (i.e. 2MB/GB with 8-byte pointers). */ while (hsize * 4096 < (uint64_t)physmem * PAGESIZE) hsize <<= 1; retry: h->hash_table_mask = hsize - 1; h->hash_table = kmem_zalloc(hsize * sizeof (void *), KM_NOSLEEP); if (h->hash_table == NULL) { /* XXX - we should really return an error instead of assert */ ASSERT(hsize > (1ULL << 10)); hsize >>= 1; goto retry; } dbuf_kmem_cache = kmem_cache_create("dmu_buf_impl_t", sizeof (dmu_buf_impl_t), 0, dbuf_cons, dbuf_dest, NULL, NULL, NULL, 0); for (i = 0; i < DBUF_MUTEXES; i++) mutex_init(&h->hash_mutexes[i], NULL, MUTEX_DEFAULT, NULL); dbuf_stats_init(h); /* * Setup the parameters for the dbuf caches. We set the sizes of the * dbuf cache and the metadata cache to 1/32nd and 1/16th (default) * of the size of the ARC, respectively. If the values are set in * /etc/system and they're not greater than the size of the ARC, then * we honor that value. */ if (dbuf_cache_max_bytes == 0 || dbuf_cache_max_bytes >= arc_max_bytes()) { dbuf_cache_max_bytes = arc_max_bytes() >> dbuf_cache_shift; } if (dbuf_metadata_cache_max_bytes == 0 || dbuf_metadata_cache_max_bytes >= arc_max_bytes()) { dbuf_metadata_cache_max_bytes = arc_max_bytes() >> dbuf_metadata_cache_shift; } /* * All entries are queued via taskq_dispatch_ent(), so min/maxalloc * configuration is not required. */ dbu_evict_taskq = taskq_create("dbu_evict", 1, minclsyspri, 0, 0, 0); for (dbuf_cached_state_t dcs = 0; dcs < DB_CACHE_MAX; dcs++) { dbuf_caches[dcs].cache = multilist_create(sizeof (dmu_buf_impl_t), offsetof(dmu_buf_impl_t, db_cache_link), dbuf_cache_multilist_index_func); zfs_refcount_create(&dbuf_caches[dcs].size); } dbuf_evict_thread_exit = B_FALSE; mutex_init(&dbuf_evict_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&dbuf_evict_cv, NULL, CV_DEFAULT, NULL); dbuf_cache_evict_thread = thread_create(NULL, 0, dbuf_evict_thread, NULL, 0, &p0, TS_RUN, minclsyspri); dbuf_ksp = kstat_create("zfs", 0, "dbufstats", "misc", KSTAT_TYPE_NAMED, sizeof (dbuf_stats) / sizeof (kstat_named_t), KSTAT_FLAG_VIRTUAL); if (dbuf_ksp != NULL) { for (i = 0; i < DN_MAX_LEVELS; i++) { snprintf(dbuf_stats.cache_levels[i].name, KSTAT_STRLEN, "cache_level_%d", i); dbuf_stats.cache_levels[i].data_type = KSTAT_DATA_UINT64; snprintf(dbuf_stats.cache_levels_bytes[i].name, KSTAT_STRLEN, "cache_level_%d_bytes", i); dbuf_stats.cache_levels_bytes[i].data_type = KSTAT_DATA_UINT64; } dbuf_ksp->ks_data = &dbuf_stats; dbuf_ksp->ks_update = dbuf_kstat_update; kstat_install(dbuf_ksp); } } void dbuf_fini(void) { dbuf_hash_table_t *h = &dbuf_hash_table; int i; dbuf_stats_destroy(); for (i = 0; i < DBUF_MUTEXES; i++) mutex_destroy(&h->hash_mutexes[i]); kmem_free(h->hash_table, (h->hash_table_mask + 1) * sizeof (void *)); kmem_cache_destroy(dbuf_kmem_cache); taskq_destroy(dbu_evict_taskq); mutex_enter(&dbuf_evict_lock); dbuf_evict_thread_exit = B_TRUE; while (dbuf_evict_thread_exit) { cv_signal(&dbuf_evict_cv); cv_wait(&dbuf_evict_cv, &dbuf_evict_lock); } mutex_exit(&dbuf_evict_lock); mutex_destroy(&dbuf_evict_lock); cv_destroy(&dbuf_evict_cv); for (dbuf_cached_state_t dcs = 0; dcs < DB_CACHE_MAX; dcs++) { zfs_refcount_destroy(&dbuf_caches[dcs].size); multilist_destroy(dbuf_caches[dcs].cache); } if (dbuf_ksp != NULL) { kstat_delete(dbuf_ksp); dbuf_ksp = NULL; } } /* * Other stuff. */ #ifdef ZFS_DEBUG static void dbuf_verify(dmu_buf_impl_t *db) { dnode_t *dn; dbuf_dirty_record_t *dr; ASSERT(MUTEX_HELD(&db->db_mtx)); if (!(zfs_flags & ZFS_DEBUG_DBUF_VERIFY)) return; ASSERT(db->db_objset != NULL); DB_DNODE_ENTER(db); dn = DB_DNODE(db); if (dn == NULL) { ASSERT(db->db_parent == NULL); ASSERT(db->db_blkptr == NULL); } else { ASSERT3U(db->db.db_object, ==, dn->dn_object); ASSERT3P(db->db_objset, ==, dn->dn_objset); ASSERT3U(db->db_level, <, dn->dn_nlevels); ASSERT(db->db_blkid == DMU_BONUS_BLKID || db->db_blkid == DMU_SPILL_BLKID || !avl_is_empty(&dn->dn_dbufs)); } if (db->db_blkid == DMU_BONUS_BLKID) { ASSERT(dn != NULL); ASSERT3U(db->db.db_size, >=, dn->dn_bonuslen); ASSERT3U(db->db.db_offset, ==, DMU_BONUS_BLKID); } else if (db->db_blkid == DMU_SPILL_BLKID) { ASSERT(dn != NULL); ASSERT0(db->db.db_offset); } else { ASSERT3U(db->db.db_offset, ==, db->db_blkid * db->db.db_size); } for (dr = db->db_data_pending; dr != NULL; dr = dr->dr_next) ASSERT(dr->dr_dbuf == db); for (dr = db->db_last_dirty; dr != NULL; dr = dr->dr_next) ASSERT(dr->dr_dbuf == db); /* * We can't assert that db_size matches dn_datablksz because it * can be momentarily different when another thread is doing * dnode_set_blksz(). */ if (db->db_level == 0 && db->db.db_object == DMU_META_DNODE_OBJECT) { dr = db->db_data_pending; /* * It should only be modified in syncing context, so * make sure we only have one copy of the data. */ ASSERT(dr == NULL || dr->dt.dl.dr_data == db->db_buf); } /* verify db->db_blkptr */ if (db->db_blkptr) { if (db->db_parent == dn->dn_dbuf) { /* db is pointed to by the dnode */ /* ASSERT3U(db->db_blkid, <, dn->dn_nblkptr); */ if (DMU_OBJECT_IS_SPECIAL(db->db.db_object)) ASSERT(db->db_parent == NULL); else ASSERT(db->db_parent != NULL); if (db->db_blkid != DMU_SPILL_BLKID) ASSERT3P(db->db_blkptr, ==, &dn->dn_phys->dn_blkptr[db->db_blkid]); } else { /* db is pointed to by an indirect block */ int epb = db->db_parent->db.db_size >> SPA_BLKPTRSHIFT; ASSERT3U(db->db_parent->db_level, ==, db->db_level+1); ASSERT3U(db->db_parent->db.db_object, ==, db->db.db_object); /* * dnode_grow_indblksz() can make this fail if we don't * have the struct_rwlock. XXX indblksz no longer * grows. safe to do this now? */ if (RW_WRITE_HELD(&dn->dn_struct_rwlock)) { ASSERT3P(db->db_blkptr, ==, ((blkptr_t *)db->db_parent->db.db_data + db->db_blkid % epb)); } } } if ((db->db_blkptr == NULL || BP_IS_HOLE(db->db_blkptr)) && (db->db_buf == NULL || db->db_buf->b_data) && db->db.db_data && db->db_blkid != DMU_BONUS_BLKID && db->db_state != DB_FILL && !dn->dn_free_txg) { /* * If the blkptr isn't set but they have nonzero data, * it had better be dirty, otherwise we'll lose that * data when we evict this buffer. * * There is an exception to this rule for indirect blocks; in * this case, if the indirect block is a hole, we fill in a few * fields on each of the child blocks (importantly, birth time) * to prevent hole birth times from being lost when you * partially fill in a hole. */ if (db->db_dirtycnt == 0) { if (db->db_level == 0) { uint64_t *buf = db->db.db_data; int i; for (i = 0; i < db->db.db_size >> 3; i++) { ASSERT(buf[i] == 0); } } else { blkptr_t *bps = db->db.db_data; ASSERT3U(1 << DB_DNODE(db)->dn_indblkshift, ==, db->db.db_size); /* * We want to verify that all the blkptrs in the * indirect block are holes, but we may have * automatically set up a few fields for them. * We iterate through each blkptr and verify * they only have those fields set. */ for (int i = 0; i < db->db.db_size / sizeof (blkptr_t); i++) { blkptr_t *bp = &bps[i]; ASSERT(ZIO_CHECKSUM_IS_ZERO( &bp->blk_cksum)); ASSERT( DVA_IS_EMPTY(&bp->blk_dva[0]) && DVA_IS_EMPTY(&bp->blk_dva[1]) && DVA_IS_EMPTY(&bp->blk_dva[2])); ASSERT0(bp->blk_fill); ASSERT0(bp->blk_pad[0]); ASSERT0(bp->blk_pad[1]); ASSERT(!BP_IS_EMBEDDED(bp)); ASSERT(BP_IS_HOLE(bp)); ASSERT0(bp->blk_phys_birth); } } } } DB_DNODE_EXIT(db); } #endif static void dbuf_clear_data(dmu_buf_impl_t *db) { ASSERT(MUTEX_HELD(&db->db_mtx)); dbuf_evict_user(db); ASSERT3P(db->db_buf, ==, NULL); db->db.db_data = NULL; if (db->db_state != DB_NOFILL) db->db_state = DB_UNCACHED; } static void dbuf_set_data(dmu_buf_impl_t *db, arc_buf_t *buf) { ASSERT(MUTEX_HELD(&db->db_mtx)); ASSERT(buf != NULL); db->db_buf = buf; ASSERT(buf->b_data != NULL); db->db.db_data = buf->b_data; } /* * Loan out an arc_buf for read. Return the loaned arc_buf. */ arc_buf_t * dbuf_loan_arcbuf(dmu_buf_impl_t *db) { arc_buf_t *abuf; ASSERT(db->db_blkid != DMU_BONUS_BLKID); mutex_enter(&db->db_mtx); if (arc_released(db->db_buf) || zfs_refcount_count(&db->db_holds) > 1) { int blksz = db->db.db_size; spa_t *spa = db->db_objset->os_spa; mutex_exit(&db->db_mtx); abuf = arc_loan_buf(spa, B_FALSE, blksz); bcopy(db->db.db_data, abuf->b_data, blksz); } else { abuf = db->db_buf; arc_loan_inuse_buf(abuf, db); db->db_buf = NULL; dbuf_clear_data(db); mutex_exit(&db->db_mtx); } return (abuf); } /* * Calculate which level n block references the data at the level 0 offset * provided. */ uint64_t dbuf_whichblock(dnode_t *dn, int64_t level, uint64_t offset) { if (dn->dn_datablkshift != 0 && dn->dn_indblkshift != 0) { /* * The level n blkid is equal to the level 0 blkid divided by * the number of level 0s in a level n block. * * The level 0 blkid is offset >> datablkshift = * offset / 2^datablkshift. * * The number of level 0s in a level n is the number of block * pointers in an indirect block, raised to the power of level. * This is 2^(indblkshift - SPA_BLKPTRSHIFT)^level = * 2^(level*(indblkshift - SPA_BLKPTRSHIFT)). * * Thus, the level n blkid is: offset / * ((2^datablkshift)*(2^(level*(indblkshift - SPA_BLKPTRSHIFT))) * = offset / 2^(datablkshift + level * * (indblkshift - SPA_BLKPTRSHIFT)) * = offset >> (datablkshift + level * * (indblkshift - SPA_BLKPTRSHIFT)) */ return (offset >> (dn->dn_datablkshift + level * (dn->dn_indblkshift - SPA_BLKPTRSHIFT))); } else { ASSERT3U(offset, <, dn->dn_datablksz); return (0); } } static void dbuf_read_done(zio_t *zio, const zbookmark_phys_t *zb, const blkptr_t *bp, arc_buf_t *buf, void *vdb) { dmu_buf_impl_t *db = vdb; mutex_enter(&db->db_mtx); ASSERT3U(db->db_state, ==, DB_READ); /* * All reads are synchronous, so we must have a hold on the dbuf */ ASSERT(zfs_refcount_count(&db->db_holds) > 0); ASSERT(db->db_buf == NULL); ASSERT(db->db.db_data == NULL); if (buf == NULL) { /* i/o error */ ASSERT(zio == NULL || zio->io_error != 0); ASSERT(db->db_blkid != DMU_BONUS_BLKID); ASSERT3P(db->db_buf, ==, NULL); db->db_state = DB_UNCACHED; } else if (db->db_level == 0 && db->db_freed_in_flight) { /* freed in flight */ ASSERT(zio == NULL || zio->io_error == 0); if (buf == NULL) { buf = arc_alloc_buf(db->db_objset->os_spa, db, DBUF_GET_BUFC_TYPE(db), db->db.db_size); } arc_release(buf, db); bzero(buf->b_data, db->db.db_size); arc_buf_freeze(buf); db->db_freed_in_flight = FALSE; dbuf_set_data(db, buf); db->db_state = DB_CACHED; } else { /* success */ ASSERT(zio == NULL || zio->io_error == 0); dbuf_set_data(db, buf); db->db_state = DB_CACHED; } cv_broadcast(&db->db_changed); dbuf_rele_and_unlock(db, NULL, B_FALSE); } static void dbuf_read_impl(dmu_buf_impl_t *db, zio_t *zio, uint32_t flags) { dnode_t *dn; zbookmark_phys_t zb; arc_flags_t aflags = ARC_FLAG_NOWAIT; DB_DNODE_ENTER(db); dn = DB_DNODE(db); ASSERT(!zfs_refcount_is_zero(&db->db_holds)); /* We need the struct_rwlock to prevent db_blkptr from changing. */ ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock)); ASSERT(MUTEX_HELD(&db->db_mtx)); ASSERT(db->db_state == DB_UNCACHED); ASSERT(db->db_buf == NULL); if (db->db_blkid == DMU_BONUS_BLKID) { /* * The bonus length stored in the dnode may be less than * the maximum available space in the bonus buffer. */ int bonuslen = MIN(dn->dn_bonuslen, dn->dn_phys->dn_bonuslen); int max_bonuslen = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots); ASSERT3U(bonuslen, <=, db->db.db_size); db->db.db_data = zio_buf_alloc(max_bonuslen); arc_space_consume(max_bonuslen, ARC_SPACE_BONUS); if (bonuslen < max_bonuslen) bzero(db->db.db_data, max_bonuslen); if (bonuslen) bcopy(DN_BONUS(dn->dn_phys), db->db.db_data, bonuslen); DB_DNODE_EXIT(db); db->db_state = DB_CACHED; mutex_exit(&db->db_mtx); return; } /* * Recheck BP_IS_HOLE() after dnode_block_freed() in case dnode_sync() * processes the delete record and clears the bp while we are waiting * for the dn_mtx (resulting in a "no" from block_freed). */ if (db->db_blkptr == NULL || BP_IS_HOLE(db->db_blkptr) || (db->db_level == 0 && (dnode_block_freed(dn, db->db_blkid) || BP_IS_HOLE(db->db_blkptr)))) { arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db); dbuf_set_data(db, arc_alloc_buf(db->db_objset->os_spa, db, type, db->db.db_size)); bzero(db->db.db_data, db->db.db_size); if (db->db_blkptr != NULL && db->db_level > 0 && BP_IS_HOLE(db->db_blkptr) && db->db_blkptr->blk_birth != 0) { blkptr_t *bps = db->db.db_data; for (int i = 0; i < ((1 << DB_DNODE(db)->dn_indblkshift) / sizeof (blkptr_t)); i++) { blkptr_t *bp = &bps[i]; ASSERT3U(BP_GET_LSIZE(db->db_blkptr), ==, 1 << dn->dn_indblkshift); BP_SET_LSIZE(bp, BP_GET_LEVEL(db->db_blkptr) == 1 ? dn->dn_datablksz : BP_GET_LSIZE(db->db_blkptr)); BP_SET_TYPE(bp, BP_GET_TYPE(db->db_blkptr)); BP_SET_LEVEL(bp, BP_GET_LEVEL(db->db_blkptr) - 1); BP_SET_BIRTH(bp, db->db_blkptr->blk_birth, 0); } } DB_DNODE_EXIT(db); db->db_state = DB_CACHED; mutex_exit(&db->db_mtx); return; } DB_DNODE_EXIT(db); db->db_state = DB_READ; mutex_exit(&db->db_mtx); if (DBUF_IS_L2CACHEABLE(db)) aflags |= ARC_FLAG_L2CACHE; SET_BOOKMARK(&zb, db->db_objset->os_dsl_dataset ? db->db_objset->os_dsl_dataset->ds_object : DMU_META_OBJSET, db->db.db_object, db->db_level, db->db_blkid); dbuf_add_ref(db, NULL); (void) arc_read(zio, db->db_objset->os_spa, db->db_blkptr, dbuf_read_done, db, ZIO_PRIORITY_SYNC_READ, (flags & DB_RF_CANFAIL) ? ZIO_FLAG_CANFAIL : ZIO_FLAG_MUSTSUCCEED, &aflags, &zb); } /* * This is our just-in-time copy function. It makes a copy of buffers that * have been modified in a previous transaction group before we access them in * the current active group. * * This function is used in three places: when we are dirtying a buffer for the * first time in a txg, when we are freeing a range in a dnode that includes * this buffer, and when we are accessing a buffer which was received compressed * and later referenced in a WRITE_BYREF record. * * Note that when we are called from dbuf_free_range() we do not put a hold on * the buffer, we just traverse the active dbuf list for the dnode. */ static void dbuf_fix_old_data(dmu_buf_impl_t *db, uint64_t txg) { dbuf_dirty_record_t *dr = db->db_last_dirty; ASSERT(MUTEX_HELD(&db->db_mtx)); ASSERT(db->db.db_data != NULL); ASSERT(db->db_level == 0); ASSERT(db->db.db_object != DMU_META_DNODE_OBJECT); if (dr == NULL || (dr->dt.dl.dr_data != ((db->db_blkid == DMU_BONUS_BLKID) ? db->db.db_data : db->db_buf))) return; /* * If the last dirty record for this dbuf has not yet synced * and its referencing the dbuf data, either: * reset the reference to point to a new copy, * or (if there a no active holders) * just null out the current db_data pointer. */ ASSERT(dr->dr_txg >= txg - 2); if (db->db_blkid == DMU_BONUS_BLKID) { /* Note that the data bufs here are zio_bufs */ dnode_t *dn = DB_DNODE(db); int bonuslen = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots); dr->dt.dl.dr_data = zio_buf_alloc(bonuslen); arc_space_consume(bonuslen, ARC_SPACE_BONUS); bcopy(db->db.db_data, dr->dt.dl.dr_data, bonuslen); } else if (zfs_refcount_count(&db->db_holds) > db->db_dirtycnt) { int size = arc_buf_size(db->db_buf); arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db); spa_t *spa = db->db_objset->os_spa; enum zio_compress compress_type = arc_get_compression(db->db_buf); if (compress_type == ZIO_COMPRESS_OFF) { dr->dt.dl.dr_data = arc_alloc_buf(spa, db, type, size); } else { ASSERT3U(type, ==, ARC_BUFC_DATA); dr->dt.dl.dr_data = arc_alloc_compressed_buf(spa, db, size, arc_buf_lsize(db->db_buf), compress_type); } bcopy(db->db.db_data, dr->dt.dl.dr_data->b_data, size); } else { db->db_buf = NULL; dbuf_clear_data(db); } } int dbuf_read(dmu_buf_impl_t *db, zio_t *zio, uint32_t flags) { int err = 0; boolean_t prefetch; dnode_t *dn; /* * We don't have to hold the mutex to check db_state because it * can't be freed while we have a hold on the buffer. */ ASSERT(!zfs_refcount_is_zero(&db->db_holds)); if (db->db_state == DB_NOFILL) return (SET_ERROR(EIO)); DB_DNODE_ENTER(db); dn = DB_DNODE(db); if ((flags & DB_RF_HAVESTRUCT) == 0) rw_enter(&dn->dn_struct_rwlock, RW_READER); prefetch = db->db_level == 0 && db->db_blkid != DMU_BONUS_BLKID && (flags & DB_RF_NOPREFETCH) == 0 && dn != NULL && DBUF_IS_CACHEABLE(db); mutex_enter(&db->db_mtx); if (db->db_state == DB_CACHED) { /* * If the arc buf is compressed, we need to decompress it to * read the data. This could happen during the "zfs receive" of * a stream which is compressed and deduplicated. */ if (db->db_buf != NULL && arc_get_compression(db->db_buf) != ZIO_COMPRESS_OFF) { dbuf_fix_old_data(db, spa_syncing_txg(dmu_objset_spa(db->db_objset))); err = arc_decompress(db->db_buf); dbuf_set_data(db, db->db_buf); } mutex_exit(&db->db_mtx); if (prefetch) dmu_zfetch(&dn->dn_zfetch, db->db_blkid, 1, B_TRUE); if ((flags & DB_RF_HAVESTRUCT) == 0) rw_exit(&dn->dn_struct_rwlock); DB_DNODE_EXIT(db); DBUF_STAT_BUMP(hash_hits); } else if (db->db_state == DB_UNCACHED) { spa_t *spa = dn->dn_objset->os_spa; boolean_t need_wait = B_FALSE; if (zio == NULL && db->db_blkptr != NULL && !BP_IS_HOLE(db->db_blkptr)) { zio = zio_root(spa, NULL, NULL, ZIO_FLAG_CANFAIL); need_wait = B_TRUE; } dbuf_read_impl(db, zio, flags); /* dbuf_read_impl has dropped db_mtx for us */ if (prefetch) dmu_zfetch(&dn->dn_zfetch, db->db_blkid, 1, B_TRUE); if ((flags & DB_RF_HAVESTRUCT) == 0) rw_exit(&dn->dn_struct_rwlock); DB_DNODE_EXIT(db); DBUF_STAT_BUMP(hash_misses); if (need_wait) err = zio_wait(zio); } else { /* * Another reader came in while the dbuf was in flight * between UNCACHED and CACHED. Either a writer will finish * writing the buffer (sending the dbuf to CACHED) or the * first reader's request will reach the read_done callback * and send the dbuf to CACHED. Otherwise, a failure * occurred and the dbuf went to UNCACHED. */ mutex_exit(&db->db_mtx); if (prefetch) dmu_zfetch(&dn->dn_zfetch, db->db_blkid, 1, B_TRUE); if ((flags & DB_RF_HAVESTRUCT) == 0) rw_exit(&dn->dn_struct_rwlock); DB_DNODE_EXIT(db); DBUF_STAT_BUMP(hash_misses); /* Skip the wait per the caller's request. */ mutex_enter(&db->db_mtx); if ((flags & DB_RF_NEVERWAIT) == 0) { while (db->db_state == DB_READ || db->db_state == DB_FILL) { ASSERT(db->db_state == DB_READ || (flags & DB_RF_HAVESTRUCT) == 0); DTRACE_PROBE2(blocked__read, dmu_buf_impl_t *, db, zio_t *, zio); cv_wait(&db->db_changed, &db->db_mtx); } if (db->db_state == DB_UNCACHED) err = SET_ERROR(EIO); } mutex_exit(&db->db_mtx); } return (err); } static void dbuf_noread(dmu_buf_impl_t *db) { ASSERT(!zfs_refcount_is_zero(&db->db_holds)); ASSERT(db->db_blkid != DMU_BONUS_BLKID); mutex_enter(&db->db_mtx); while (db->db_state == DB_READ || db->db_state == DB_FILL) cv_wait(&db->db_changed, &db->db_mtx); if (db->db_state == DB_UNCACHED) { arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db); spa_t *spa = db->db_objset->os_spa; ASSERT(db->db_buf == NULL); ASSERT(db->db.db_data == NULL); dbuf_set_data(db, arc_alloc_buf(spa, db, type, db->db.db_size)); db->db_state = DB_FILL; } else if (db->db_state == DB_NOFILL) { dbuf_clear_data(db); } else { ASSERT3U(db->db_state, ==, DB_CACHED); } mutex_exit(&db->db_mtx); } void dbuf_unoverride(dbuf_dirty_record_t *dr) { dmu_buf_impl_t *db = dr->dr_dbuf; blkptr_t *bp = &dr->dt.dl.dr_overridden_by; uint64_t txg = dr->dr_txg; ASSERT(MUTEX_HELD(&db->db_mtx)); /* * This assert is valid because dmu_sync() expects to be called by * a zilog's get_data while holding a range lock. This call only * comes from dbuf_dirty() callers who must also hold a range lock. */ ASSERT(dr->dt.dl.dr_override_state != DR_IN_DMU_SYNC); ASSERT(db->db_level == 0); if (db->db_blkid == DMU_BONUS_BLKID || dr->dt.dl.dr_override_state == DR_NOT_OVERRIDDEN) return; ASSERT(db->db_data_pending != dr); /* free this block */ if (!BP_IS_HOLE(bp) && !dr->dt.dl.dr_nopwrite) zio_free(db->db_objset->os_spa, txg, bp); dr->dt.dl.dr_override_state = DR_NOT_OVERRIDDEN; dr->dt.dl.dr_nopwrite = B_FALSE; /* * Release the already-written buffer, so we leave it in * a consistent dirty state. Note that all callers are * modifying the buffer, so they will immediately do * another (redundant) arc_release(). Therefore, leave * the buf thawed to save the effort of freezing & * immediately re-thawing it. */ arc_release(dr->dt.dl.dr_data, db); } /* * Evict (if its unreferenced) or clear (if its referenced) any level-0 * data blocks in the free range, so that any future readers will find * empty blocks. */ void dbuf_free_range(dnode_t *dn, uint64_t start_blkid, uint64_t end_blkid, dmu_tx_t *tx) { dmu_buf_impl_t db_search; dmu_buf_impl_t *db, *db_next; uint64_t txg = tx->tx_txg; avl_index_t where; if (end_blkid > dn->dn_maxblkid && !(start_blkid == DMU_SPILL_BLKID || end_blkid == DMU_SPILL_BLKID)) end_blkid = dn->dn_maxblkid; dprintf_dnode(dn, "start=%llu end=%llu\n", start_blkid, end_blkid); db_search.db_level = 0; db_search.db_blkid = start_blkid; db_search.db_state = DB_SEARCH; mutex_enter(&dn->dn_dbufs_mtx); db = avl_find(&dn->dn_dbufs, &db_search, &where); ASSERT3P(db, ==, NULL); db = avl_nearest(&dn->dn_dbufs, where, AVL_AFTER); for (; db != NULL; db = db_next) { db_next = AVL_NEXT(&dn->dn_dbufs, db); ASSERT(db->db_blkid != DMU_BONUS_BLKID); if (db->db_level != 0 || db->db_blkid > end_blkid) { break; } ASSERT3U(db->db_blkid, >=, start_blkid); /* found a level 0 buffer in the range */ mutex_enter(&db->db_mtx); if (dbuf_undirty(db, tx)) { /* mutex has been dropped and dbuf destroyed */ continue; } if (db->db_state == DB_UNCACHED || db->db_state == DB_NOFILL || db->db_state == DB_EVICTING) { ASSERT(db->db.db_data == NULL); mutex_exit(&db->db_mtx); continue; } if (db->db_state == DB_READ || db->db_state == DB_FILL) { /* will be handled in dbuf_read_done or dbuf_rele */ db->db_freed_in_flight = TRUE; mutex_exit(&db->db_mtx); continue; } if (zfs_refcount_count(&db->db_holds) == 0) { ASSERT(db->db_buf); dbuf_destroy(db); continue; } /* The dbuf is referenced */ if (db->db_last_dirty != NULL) { dbuf_dirty_record_t *dr = db->db_last_dirty; if (dr->dr_txg == txg) { /* * This buffer is "in-use", re-adjust the file * size to reflect that this buffer may * contain new data when we sync. */ if (db->db_blkid != DMU_SPILL_BLKID && db->db_blkid > dn->dn_maxblkid) dn->dn_maxblkid = db->db_blkid; dbuf_unoverride(dr); } else { /* * This dbuf is not dirty in the open context. * Either uncache it (if its not referenced in * the open context) or reset its contents to * empty. */ dbuf_fix_old_data(db, txg); } } /* clear the contents if its cached */ if (db->db_state == DB_CACHED) { ASSERT(db->db.db_data != NULL); arc_release(db->db_buf, db); bzero(db->db.db_data, db->db.db_size); arc_buf_freeze(db->db_buf); } mutex_exit(&db->db_mtx); } mutex_exit(&dn->dn_dbufs_mtx); } void dbuf_new_size(dmu_buf_impl_t *db, int size, dmu_tx_t *tx) { arc_buf_t *buf, *obuf; int osize = db->db.db_size; arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db); dnode_t *dn; ASSERT(db->db_blkid != DMU_BONUS_BLKID); DB_DNODE_ENTER(db); dn = DB_DNODE(db); /* XXX does *this* func really need the lock? */ ASSERT(RW_WRITE_HELD(&dn->dn_struct_rwlock)); /* * This call to dmu_buf_will_dirty() with the dn_struct_rwlock held * is OK, because there can be no other references to the db * when we are changing its size, so no concurrent DB_FILL can * be happening. */ /* * XXX we should be doing a dbuf_read, checking the return * value and returning that up to our callers */ dmu_buf_will_dirty(&db->db, tx); /* create the data buffer for the new block */ buf = arc_alloc_buf(dn->dn_objset->os_spa, db, type, size); /* copy old block data to the new block */ obuf = db->db_buf; bcopy(obuf->b_data, buf->b_data, MIN(osize, size)); /* zero the remainder */ if (size > osize) bzero((uint8_t *)buf->b_data + osize, size - osize); mutex_enter(&db->db_mtx); dbuf_set_data(db, buf); arc_buf_destroy(obuf, db); db->db.db_size = size; if (db->db_level == 0) { ASSERT3U(db->db_last_dirty->dr_txg, ==, tx->tx_txg); db->db_last_dirty->dt.dl.dr_data = buf; } mutex_exit(&db->db_mtx); dmu_objset_willuse_space(dn->dn_objset, size - osize, tx); DB_DNODE_EXIT(db); } void dbuf_release_bp(dmu_buf_impl_t *db) { objset_t *os = db->db_objset; ASSERT(dsl_pool_sync_context(dmu_objset_pool(os))); ASSERT(arc_released(os->os_phys_buf) || list_link_active(&os->os_dsl_dataset->ds_synced_link)); ASSERT(db->db_parent == NULL || arc_released(db->db_parent->db_buf)); (void) arc_release(db->db_buf, db); } /* * We already have a dirty record for this TXG, and we are being * dirtied again. */ static void dbuf_redirty(dbuf_dirty_record_t *dr) { dmu_buf_impl_t *db = dr->dr_dbuf; ASSERT(MUTEX_HELD(&db->db_mtx)); if (db->db_level == 0 && db->db_blkid != DMU_BONUS_BLKID) { /* * If this buffer has already been written out, * we now need to reset its state. */ dbuf_unoverride(dr); if (db->db.db_object != DMU_META_DNODE_OBJECT && db->db_state != DB_NOFILL) { /* Already released on initial dirty, so just thaw. */ ASSERT(arc_released(db->db_buf)); arc_buf_thaw(db->db_buf); } } } dbuf_dirty_record_t * dbuf_dirty(dmu_buf_impl_t *db, dmu_tx_t *tx) { dnode_t *dn; objset_t *os; dbuf_dirty_record_t **drp, *dr; int drop_struct_lock = FALSE; int txgoff = tx->tx_txg & TXG_MASK; ASSERT(tx->tx_txg != 0); ASSERT(!zfs_refcount_is_zero(&db->db_holds)); DMU_TX_DIRTY_BUF(tx, db); DB_DNODE_ENTER(db); dn = DB_DNODE(db); /* * Shouldn't dirty a regular buffer in syncing context. Private * objects may be dirtied in syncing context, but only if they * were already pre-dirtied in open context. */ #ifdef DEBUG if (dn->dn_objset->os_dsl_dataset != NULL) { rrw_enter(&dn->dn_objset->os_dsl_dataset->ds_bp_rwlock, RW_READER, FTAG); } ASSERT(!dmu_tx_is_syncing(tx) || BP_IS_HOLE(dn->dn_objset->os_rootbp) || DMU_OBJECT_IS_SPECIAL(dn->dn_object) || dn->dn_objset->os_dsl_dataset == NULL); if (dn->dn_objset->os_dsl_dataset != NULL) rrw_exit(&dn->dn_objset->os_dsl_dataset->ds_bp_rwlock, FTAG); #endif /* * We make this assert for private objects as well, but after we * check if we're already dirty. They are allowed to re-dirty * in syncing context. */ ASSERT(dn->dn_object == DMU_META_DNODE_OBJECT || dn->dn_dirtyctx == DN_UNDIRTIED || dn->dn_dirtyctx == (dmu_tx_is_syncing(tx) ? DN_DIRTY_SYNC : DN_DIRTY_OPEN)); mutex_enter(&db->db_mtx); /* * XXX make this true for indirects too? The problem is that * transactions created with dmu_tx_create_assigned() from * syncing context don't bother holding ahead. */ ASSERT(db->db_level != 0 || db->db_state == DB_CACHED || db->db_state == DB_FILL || db->db_state == DB_NOFILL); mutex_enter(&dn->dn_mtx); /* * Don't set dirtyctx to SYNC if we're just modifying this as we * initialize the objset. */ if (dn->dn_dirtyctx == DN_UNDIRTIED) { if (dn->dn_objset->os_dsl_dataset != NULL) { rrw_enter(&dn->dn_objset->os_dsl_dataset->ds_bp_rwlock, RW_READER, FTAG); } if (!BP_IS_HOLE(dn->dn_objset->os_rootbp)) { dn->dn_dirtyctx = (dmu_tx_is_syncing(tx) ? DN_DIRTY_SYNC : DN_DIRTY_OPEN); ASSERT(dn->dn_dirtyctx_firstset == NULL); dn->dn_dirtyctx_firstset = kmem_alloc(1, KM_SLEEP); } if (dn->dn_objset->os_dsl_dataset != NULL) { rrw_exit(&dn->dn_objset->os_dsl_dataset->ds_bp_rwlock, FTAG); } } if (tx->tx_txg > dn->dn_dirty_txg) dn->dn_dirty_txg = tx->tx_txg; mutex_exit(&dn->dn_mtx); if (db->db_blkid == DMU_SPILL_BLKID) dn->dn_have_spill = B_TRUE; /* * If this buffer is already dirty, we're done. */ drp = &db->db_last_dirty; ASSERT(*drp == NULL || (*drp)->dr_txg <= tx->tx_txg || db->db.db_object == DMU_META_DNODE_OBJECT); while ((dr = *drp) != NULL && dr->dr_txg > tx->tx_txg) drp = &dr->dr_next; if (dr && dr->dr_txg == tx->tx_txg) { DB_DNODE_EXIT(db); dbuf_redirty(dr); mutex_exit(&db->db_mtx); return (dr); } /* * Only valid if not already dirty. */ ASSERT(dn->dn_object == 0 || dn->dn_dirtyctx == DN_UNDIRTIED || dn->dn_dirtyctx == (dmu_tx_is_syncing(tx) ? DN_DIRTY_SYNC : DN_DIRTY_OPEN)); ASSERT3U(dn->dn_nlevels, >, db->db_level); /* * We should only be dirtying in syncing context if it's the * mos or we're initializing the os or it's a special object. * However, we are allowed to dirty in syncing context provided * we already dirtied it in open context. Hence we must make * this assertion only if we're not already dirty. */ os = dn->dn_objset; VERIFY3U(tx->tx_txg, <=, spa_final_dirty_txg(os->os_spa)); #ifdef DEBUG if (dn->dn_objset->os_dsl_dataset != NULL) rrw_enter(&os->os_dsl_dataset->ds_bp_rwlock, RW_READER, FTAG); ASSERT(!dmu_tx_is_syncing(tx) || DMU_OBJECT_IS_SPECIAL(dn->dn_object) || os->os_dsl_dataset == NULL || BP_IS_HOLE(os->os_rootbp)); if (dn->dn_objset->os_dsl_dataset != NULL) rrw_exit(&os->os_dsl_dataset->ds_bp_rwlock, FTAG); #endif ASSERT(db->db.db_size != 0); dprintf_dbuf(db, "size=%llx\n", (u_longlong_t)db->db.db_size); if (db->db_blkid != DMU_BONUS_BLKID) { dmu_objset_willuse_space(os, db->db.db_size, tx); } /* * If this buffer is dirty in an old transaction group we need * to make a copy of it so that the changes we make in this * transaction group won't leak out when we sync the older txg. */ dr = kmem_zalloc(sizeof (dbuf_dirty_record_t), KM_SLEEP); list_link_init(&dr->dr_dirty_node); if (db->db_level == 0) { void *data_old = db->db_buf; if (db->db_state != DB_NOFILL) { if (db->db_blkid == DMU_BONUS_BLKID) { dbuf_fix_old_data(db, tx->tx_txg); data_old = db->db.db_data; } else if (db->db.db_object != DMU_META_DNODE_OBJECT) { /* * Release the data buffer from the cache so * that we can modify it without impacting * possible other users of this cached data * block. Note that indirect blocks and * private objects are not released until the * syncing state (since they are only modified * then). */ arc_release(db->db_buf, db); dbuf_fix_old_data(db, tx->tx_txg); data_old = db->db_buf; } ASSERT(data_old != NULL); } dr->dt.dl.dr_data = data_old; } else { mutex_init(&dr->dt.di.dr_mtx, NULL, MUTEX_DEFAULT, NULL); list_create(&dr->dt.di.dr_children, sizeof (dbuf_dirty_record_t), offsetof(dbuf_dirty_record_t, dr_dirty_node)); } if (db->db_blkid != DMU_BONUS_BLKID && os->os_dsl_dataset != NULL) dr->dr_accounted = db->db.db_size; dr->dr_dbuf = db; dr->dr_txg = tx->tx_txg; dr->dr_next = *drp; *drp = dr; /* * We could have been freed_in_flight between the dbuf_noread * and dbuf_dirty. We win, as though the dbuf_noread() had * happened after the free. */ if (db->db_level == 0 && db->db_blkid != DMU_BONUS_BLKID && db->db_blkid != DMU_SPILL_BLKID) { mutex_enter(&dn->dn_mtx); if (dn->dn_free_ranges[txgoff] != NULL) { range_tree_clear(dn->dn_free_ranges[txgoff], db->db_blkid, 1); } mutex_exit(&dn->dn_mtx); db->db_freed_in_flight = FALSE; } /* * This buffer is now part of this txg */ dbuf_add_ref(db, (void *)(uintptr_t)tx->tx_txg); db->db_dirtycnt += 1; ASSERT3U(db->db_dirtycnt, <=, 3); mutex_exit(&db->db_mtx); if (db->db_blkid == DMU_BONUS_BLKID || db->db_blkid == DMU_SPILL_BLKID) { mutex_enter(&dn->dn_mtx); ASSERT(!list_link_active(&dr->dr_dirty_node)); list_insert_tail(&dn->dn_dirty_records[txgoff], dr); mutex_exit(&dn->dn_mtx); dnode_setdirty(dn, tx); DB_DNODE_EXIT(db); return (dr); } /* * The dn_struct_rwlock prevents db_blkptr from changing * due to a write from syncing context completing * while we are running, so we want to acquire it before * looking at db_blkptr. */ if (!RW_WRITE_HELD(&dn->dn_struct_rwlock)) { rw_enter(&dn->dn_struct_rwlock, RW_READER); drop_struct_lock = TRUE; } /* * We need to hold the dn_struct_rwlock to make this assertion, * because it protects dn_phys / dn_next_nlevels from changing. */ ASSERT((dn->dn_phys->dn_nlevels == 0 && db->db_level == 0) || dn->dn_phys->dn_nlevels > db->db_level || dn->dn_next_nlevels[txgoff] > db->db_level || dn->dn_next_nlevels[(tx->tx_txg-1) & TXG_MASK] > db->db_level || dn->dn_next_nlevels[(tx->tx_txg-2) & TXG_MASK] > db->db_level); /* * If we are overwriting a dedup BP, then unless it is snapshotted, * when we get to syncing context we will need to decrement its * refcount in the DDT. Prefetch the relevant DDT block so that * syncing context won't have to wait for the i/o. */ ddt_prefetch(os->os_spa, db->db_blkptr); if (db->db_level == 0) { dnode_new_blkid(dn, db->db_blkid, tx, drop_struct_lock); ASSERT(dn->dn_maxblkid >= db->db_blkid); } if (db->db_level+1 < dn->dn_nlevels) { dmu_buf_impl_t *parent = db->db_parent; dbuf_dirty_record_t *di; int parent_held = FALSE; if (db->db_parent == NULL || db->db_parent == dn->dn_dbuf) { int epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT; parent = dbuf_hold_level(dn, db->db_level+1, db->db_blkid >> epbs, FTAG); ASSERT(parent != NULL); parent_held = TRUE; } if (drop_struct_lock) rw_exit(&dn->dn_struct_rwlock); ASSERT3U(db->db_level+1, ==, parent->db_level); di = dbuf_dirty(parent, tx); if (parent_held) dbuf_rele(parent, FTAG); mutex_enter(&db->db_mtx); /* * Since we've dropped the mutex, it's possible that * dbuf_undirty() might have changed this out from under us. */ if (db->db_last_dirty == dr || dn->dn_object == DMU_META_DNODE_OBJECT) { mutex_enter(&di->dt.di.dr_mtx); ASSERT3U(di->dr_txg, ==, tx->tx_txg); ASSERT(!list_link_active(&dr->dr_dirty_node)); list_insert_tail(&di->dt.di.dr_children, dr); mutex_exit(&di->dt.di.dr_mtx); dr->dr_parent = di; } mutex_exit(&db->db_mtx); } else { ASSERT(db->db_level+1 == dn->dn_nlevels); ASSERT(db->db_blkid < dn->dn_nblkptr); ASSERT(db->db_parent == NULL || db->db_parent == dn->dn_dbuf); mutex_enter(&dn->dn_mtx); ASSERT(!list_link_active(&dr->dr_dirty_node)); list_insert_tail(&dn->dn_dirty_records[txgoff], dr); mutex_exit(&dn->dn_mtx); if (drop_struct_lock) rw_exit(&dn->dn_struct_rwlock); } dnode_setdirty(dn, tx); DB_DNODE_EXIT(db); return (dr); } /* * Undirty a buffer in the transaction group referenced by the given * transaction. Return whether this evicted the dbuf. */ static boolean_t dbuf_undirty(dmu_buf_impl_t *db, dmu_tx_t *tx) { dnode_t *dn; uint64_t txg = tx->tx_txg; dbuf_dirty_record_t *dr, **drp; ASSERT(txg != 0); /* * Due to our use of dn_nlevels below, this can only be called * in open context, unless we are operating on the MOS. * From syncing context, dn_nlevels may be different from the * dn_nlevels used when dbuf was dirtied. */ ASSERT(db->db_objset == dmu_objset_pool(db->db_objset)->dp_meta_objset || txg != spa_syncing_txg(dmu_objset_spa(db->db_objset))); ASSERT(db->db_blkid != DMU_BONUS_BLKID); ASSERT0(db->db_level); ASSERT(MUTEX_HELD(&db->db_mtx)); /* * If this buffer is not dirty, we're done. */ for (drp = &db->db_last_dirty; (dr = *drp) != NULL; drp = &dr->dr_next) if (dr->dr_txg <= txg) break; if (dr == NULL || dr->dr_txg < txg) return (B_FALSE); ASSERT(dr->dr_txg == txg); ASSERT(dr->dr_dbuf == db); DB_DNODE_ENTER(db); dn = DB_DNODE(db); dprintf_dbuf(db, "size=%llx\n", (u_longlong_t)db->db.db_size); ASSERT(db->db.db_size != 0); dsl_pool_undirty_space(dmu_objset_pool(dn->dn_objset), dr->dr_accounted, txg); *drp = dr->dr_next; /* * Note that there are three places in dbuf_dirty() * where this dirty record may be put on a list. * Make sure to do a list_remove corresponding to * every one of those list_insert calls. */ if (dr->dr_parent) { mutex_enter(&dr->dr_parent->dt.di.dr_mtx); list_remove(&dr->dr_parent->dt.di.dr_children, dr); mutex_exit(&dr->dr_parent->dt.di.dr_mtx); } else if (db->db_blkid == DMU_SPILL_BLKID || db->db_level + 1 == dn->dn_nlevels) { ASSERT(db->db_blkptr == NULL || db->db_parent == dn->dn_dbuf); mutex_enter(&dn->dn_mtx); list_remove(&dn->dn_dirty_records[txg & TXG_MASK], dr); mutex_exit(&dn->dn_mtx); } DB_DNODE_EXIT(db); if (db->db_state != DB_NOFILL) { dbuf_unoverride(dr); ASSERT(db->db_buf != NULL); ASSERT(dr->dt.dl.dr_data != NULL); if (dr->dt.dl.dr_data != db->db_buf) arc_buf_destroy(dr->dt.dl.dr_data, db); } kmem_free(dr, sizeof (dbuf_dirty_record_t)); ASSERT(db->db_dirtycnt > 0); db->db_dirtycnt -= 1; if (zfs_refcount_remove(&db->db_holds, (void *)(uintptr_t)txg) == 0) { ASSERT(db->db_state == DB_NOFILL || arc_released(db->db_buf)); dbuf_destroy(db); return (B_TRUE); } return (B_FALSE); } void dmu_buf_will_dirty(dmu_buf_t *db_fake, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; int rf = DB_RF_MUST_SUCCEED | DB_RF_NOPREFETCH; ASSERT(tx->tx_txg != 0); ASSERT(!zfs_refcount_is_zero(&db->db_holds)); /* * Quick check for dirtyness. For already dirty blocks, this * reduces runtime of this function by >90%, and overall performance * by 50% for some workloads (e.g. file deletion with indirect blocks * cached). */ mutex_enter(&db->db_mtx); dbuf_dirty_record_t *dr; for (dr = db->db_last_dirty; dr != NULL && dr->dr_txg >= tx->tx_txg; dr = dr->dr_next) { /* * It's possible that it is already dirty but not cached, * because there are some calls to dbuf_dirty() that don't * go through dmu_buf_will_dirty(). */ if (dr->dr_txg == tx->tx_txg && db->db_state == DB_CACHED) { /* This dbuf is already dirty and cached. */ dbuf_redirty(dr); mutex_exit(&db->db_mtx); return; } } mutex_exit(&db->db_mtx); DB_DNODE_ENTER(db); if (RW_WRITE_HELD(&DB_DNODE(db)->dn_struct_rwlock)) rf |= DB_RF_HAVESTRUCT; DB_DNODE_EXIT(db); (void) dbuf_read(db, NULL, rf); (void) dbuf_dirty(db, tx); } void dmu_buf_will_not_fill(dmu_buf_t *db_fake, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; db->db_state = DB_NOFILL; dmu_buf_will_fill(db_fake, tx); } void dmu_buf_will_fill(dmu_buf_t *db_fake, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; ASSERT(db->db_blkid != DMU_BONUS_BLKID); ASSERT(tx->tx_txg != 0); ASSERT(db->db_level == 0); ASSERT(!zfs_refcount_is_zero(&db->db_holds)); ASSERT(db->db.db_object != DMU_META_DNODE_OBJECT || dmu_tx_private_ok(tx)); dbuf_noread(db); (void) dbuf_dirty(db, tx); } #pragma weak dmu_buf_fill_done = dbuf_fill_done /* ARGSUSED */ void dbuf_fill_done(dmu_buf_impl_t *db, dmu_tx_t *tx) { mutex_enter(&db->db_mtx); DBUF_VERIFY(db); if (db->db_state == DB_FILL) { if (db->db_level == 0 && db->db_freed_in_flight) { ASSERT(db->db_blkid != DMU_BONUS_BLKID); /* we were freed while filling */ /* XXX dbuf_undirty? */ bzero(db->db.db_data, db->db.db_size); db->db_freed_in_flight = FALSE; } db->db_state = DB_CACHED; cv_broadcast(&db->db_changed); } mutex_exit(&db->db_mtx); } void dmu_buf_write_embedded(dmu_buf_t *dbuf, void *data, bp_embedded_type_t etype, enum zio_compress comp, int uncompressed_size, int compressed_size, int byteorder, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)dbuf; struct dirty_leaf *dl; dmu_object_type_t type; if (etype == BP_EMBEDDED_TYPE_DATA) { ASSERT(spa_feature_is_active(dmu_objset_spa(db->db_objset), SPA_FEATURE_EMBEDDED_DATA)); } DB_DNODE_ENTER(db); type = DB_DNODE(db)->dn_type; DB_DNODE_EXIT(db); ASSERT0(db->db_level); ASSERT(db->db_blkid != DMU_BONUS_BLKID); dmu_buf_will_not_fill(dbuf, tx); ASSERT3U(db->db_last_dirty->dr_txg, ==, tx->tx_txg); dl = &db->db_last_dirty->dt.dl; encode_embedded_bp_compressed(&dl->dr_overridden_by, data, comp, uncompressed_size, compressed_size); BPE_SET_ETYPE(&dl->dr_overridden_by, etype); BP_SET_TYPE(&dl->dr_overridden_by, type); BP_SET_LEVEL(&dl->dr_overridden_by, 0); BP_SET_BYTEORDER(&dl->dr_overridden_by, byteorder); dl->dr_override_state = DR_OVERRIDDEN; dl->dr_overridden_by.blk_birth = db->db_last_dirty->dr_txg; } /* * Directly assign a provided arc buf to a given dbuf if it's not referenced * by anybody except our caller. Otherwise copy arcbuf's contents to dbuf. */ void dbuf_assign_arcbuf(dmu_buf_impl_t *db, arc_buf_t *buf, dmu_tx_t *tx) { ASSERT(!zfs_refcount_is_zero(&db->db_holds)); ASSERT(db->db_blkid != DMU_BONUS_BLKID); ASSERT(db->db_level == 0); ASSERT3U(dbuf_is_metadata(db), ==, arc_is_metadata(buf)); ASSERT(buf != NULL); ASSERT(arc_buf_lsize(buf) == db->db.db_size); ASSERT(tx->tx_txg != 0); arc_return_buf(buf, db); ASSERT(arc_released(buf)); mutex_enter(&db->db_mtx); while (db->db_state == DB_READ || db->db_state == DB_FILL) cv_wait(&db->db_changed, &db->db_mtx); ASSERT(db->db_state == DB_CACHED || db->db_state == DB_UNCACHED); if (db->db_state == DB_CACHED && zfs_refcount_count(&db->db_holds) - 1 > db->db_dirtycnt) { mutex_exit(&db->db_mtx); (void) dbuf_dirty(db, tx); bcopy(buf->b_data, db->db.db_data, db->db.db_size); arc_buf_destroy(buf, db); xuio_stat_wbuf_copied(); return; } xuio_stat_wbuf_nocopy(); if (db->db_state == DB_CACHED) { dbuf_dirty_record_t *dr = db->db_last_dirty; ASSERT(db->db_buf != NULL); if (dr != NULL && dr->dr_txg == tx->tx_txg) { ASSERT(dr->dt.dl.dr_data == db->db_buf); if (!arc_released(db->db_buf)) { ASSERT(dr->dt.dl.dr_override_state == DR_OVERRIDDEN); arc_release(db->db_buf, db); } dr->dt.dl.dr_data = buf; arc_buf_destroy(db->db_buf, db); } else if (dr == NULL || dr->dt.dl.dr_data != db->db_buf) { arc_release(db->db_buf, db); arc_buf_destroy(db->db_buf, db); } db->db_buf = NULL; } ASSERT(db->db_buf == NULL); dbuf_set_data(db, buf); db->db_state = DB_FILL; mutex_exit(&db->db_mtx); (void) dbuf_dirty(db, tx); dmu_buf_fill_done(&db->db, tx); } void dbuf_destroy(dmu_buf_impl_t *db) { dnode_t *dn; dmu_buf_impl_t *parent = db->db_parent; dmu_buf_impl_t *dndb; ASSERT(MUTEX_HELD(&db->db_mtx)); ASSERT(zfs_refcount_is_zero(&db->db_holds)); if (db->db_buf != NULL) { arc_buf_destroy(db->db_buf, db); db->db_buf = NULL; } if (db->db_blkid == DMU_BONUS_BLKID) { int slots = DB_DNODE(db)->dn_num_slots; int bonuslen = DN_SLOTS_TO_BONUSLEN(slots); if (db->db.db_data != NULL) { zio_buf_free(db->db.db_data, bonuslen); arc_space_return(bonuslen, ARC_SPACE_BONUS); db->db_state = DB_UNCACHED; } } dbuf_clear_data(db); if (multilist_link_active(&db->db_cache_link)) { ASSERT(db->db_caching_status == DB_DBUF_CACHE || db->db_caching_status == DB_DBUF_METADATA_CACHE); multilist_remove(dbuf_caches[db->db_caching_status].cache, db); (void) zfs_refcount_remove_many( &dbuf_caches[db->db_caching_status].size, db->db.db_size, db); if (db->db_caching_status == DB_DBUF_METADATA_CACHE) { DBUF_STAT_BUMPDOWN(metadata_cache_count); } else { DBUF_STAT_BUMPDOWN(cache_levels[db->db_level]); DBUF_STAT_BUMPDOWN(cache_count); DBUF_STAT_DECR(cache_levels_bytes[db->db_level], db->db.db_size); } db->db_caching_status = DB_NO_CACHE; } ASSERT(db->db_state == DB_UNCACHED || db->db_state == DB_NOFILL); ASSERT(db->db_data_pending == NULL); db->db_state = DB_EVICTING; db->db_blkptr = NULL; /* * Now that db_state is DB_EVICTING, nobody else can find this via * the hash table. We can now drop db_mtx, which allows us to * acquire the dn_dbufs_mtx. */ mutex_exit(&db->db_mtx); DB_DNODE_ENTER(db); dn = DB_DNODE(db); dndb = dn->dn_dbuf; if (db->db_blkid != DMU_BONUS_BLKID) { boolean_t needlock = !MUTEX_HELD(&dn->dn_dbufs_mtx); if (needlock) mutex_enter(&dn->dn_dbufs_mtx); avl_remove(&dn->dn_dbufs, db); membar_producer(); DB_DNODE_EXIT(db); if (needlock) mutex_exit(&dn->dn_dbufs_mtx); /* * Decrementing the dbuf count means that the hold corresponding * to the removed dbuf is no longer discounted in dnode_move(), * so the dnode cannot be moved until after we release the hold. * The membar_producer() ensures visibility of the decremented * value in dnode_move(), since DB_DNODE_EXIT doesn't actually * release any lock. */ mutex_enter(&dn->dn_mtx); dnode_rele_and_unlock(dn, db, B_TRUE); db->db_dnode_handle = NULL; dbuf_hash_remove(db); } else { DB_DNODE_EXIT(db); } ASSERT(zfs_refcount_is_zero(&db->db_holds)); db->db_parent = NULL; ASSERT(db->db_buf == NULL); ASSERT(db->db.db_data == NULL); ASSERT(db->db_hash_next == NULL); ASSERT(db->db_blkptr == NULL); ASSERT(db->db_data_pending == NULL); ASSERT3U(db->db_caching_status, ==, DB_NO_CACHE); ASSERT(!multilist_link_active(&db->db_cache_link)); kmem_cache_free(dbuf_kmem_cache, db); arc_space_return(sizeof (dmu_buf_impl_t), ARC_SPACE_DBUF); /* * If this dbuf is referenced from an indirect dbuf, * decrement the ref count on the indirect dbuf. */ if (parent && parent != dndb) { mutex_enter(&parent->db_mtx); dbuf_rele_and_unlock(parent, db, B_TRUE); } } /* * Note: While bpp will always be updated if the function returns success, * parentp will not be updated if the dnode does not have dn_dbuf filled in; * this happens when the dnode is the meta-dnode, or a userused or groupused * object. */ __attribute__((always_inline)) static inline int dbuf_findbp(dnode_t *dn, int level, uint64_t blkid, int fail_sparse, dmu_buf_impl_t **parentp, blkptr_t **bpp, struct dbuf_hold_impl_data *dh) { *parentp = NULL; *bpp = NULL; ASSERT(blkid != DMU_BONUS_BLKID); if (blkid == DMU_SPILL_BLKID) { mutex_enter(&dn->dn_mtx); if (dn->dn_have_spill && (dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR)) *bpp = DN_SPILL_BLKPTR(dn->dn_phys); else *bpp = NULL; dbuf_add_ref(dn->dn_dbuf, NULL); *parentp = dn->dn_dbuf; mutex_exit(&dn->dn_mtx); return (0); } int nlevels = (dn->dn_phys->dn_nlevels == 0) ? 1 : dn->dn_phys->dn_nlevels; int epbs = dn->dn_indblkshift - SPA_BLKPTRSHIFT; ASSERT3U(level * epbs, <, 64); ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock)); /* * This assertion shouldn't trip as long as the max indirect block size * is less than 1M. The reason for this is that up to that point, * the number of levels required to address an entire object with blocks * of size SPA_MINBLOCKSIZE satisfies nlevels * epbs + 1 <= 64. In * other words, if N * epbs + 1 > 64, then if (N-1) * epbs + 1 > 55 * (i.e. we can address the entire object), objects will all use at most * N-1 levels and the assertion won't overflow. However, once epbs is * 13, 4 * 13 + 1 = 53, but 5 * 13 + 1 = 66. Then, 4 levels will not be * enough to address an entire object, so objects will have 5 levels, * but then this assertion will overflow. * * All this is to say that if we ever increase DN_MAX_INDBLKSHIFT, we * need to redo this logic to handle overflows. */ ASSERT(level >= nlevels || ((nlevels - level - 1) * epbs) + highbit64(dn->dn_phys->dn_nblkptr) <= 64); if (level >= nlevels || blkid >= ((uint64_t)dn->dn_phys->dn_nblkptr << ((nlevels - level - 1) * epbs)) || (fail_sparse && blkid > (dn->dn_phys->dn_maxblkid >> (level * epbs)))) { /* the buffer has no parent yet */ return (SET_ERROR(ENOENT)); } else if (level < nlevels-1) { /* this block is referenced from an indirect block */ int err; if (dh == NULL) { err = dbuf_hold_impl(dn, level+1, blkid >> epbs, fail_sparse, FALSE, NULL, parentp); } else { __dbuf_hold_impl_init(dh + 1, dn, dh->dh_level + 1, blkid >> epbs, fail_sparse, FALSE, NULL, parentp, dh->dh_depth + 1); err = __dbuf_hold_impl(dh + 1); } if (err) return (err); err = dbuf_read(*parentp, NULL, (DB_RF_HAVESTRUCT | DB_RF_NOPREFETCH | DB_RF_CANFAIL)); if (err) { dbuf_rele(*parentp, NULL); *parentp = NULL; return (err); } *bpp = ((blkptr_t *)(*parentp)->db.db_data) + (blkid & ((1ULL << epbs) - 1)); if (blkid > (dn->dn_phys->dn_maxblkid >> (level * epbs))) ASSERT(BP_IS_HOLE(*bpp)); return (0); } else { /* the block is referenced from the dnode */ ASSERT3U(level, ==, nlevels-1); ASSERT(dn->dn_phys->dn_nblkptr == 0 || blkid < dn->dn_phys->dn_nblkptr); if (dn->dn_dbuf) { dbuf_add_ref(dn->dn_dbuf, NULL); *parentp = dn->dn_dbuf; } *bpp = &dn->dn_phys->dn_blkptr[blkid]; return (0); } } static dmu_buf_impl_t * dbuf_create(dnode_t *dn, uint8_t level, uint64_t blkid, dmu_buf_impl_t *parent, blkptr_t *blkptr) { objset_t *os = dn->dn_objset; dmu_buf_impl_t *db, *odb; ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock)); ASSERT(dn->dn_type != DMU_OT_NONE); db = kmem_cache_alloc(dbuf_kmem_cache, KM_SLEEP); db->db_objset = os; db->db.db_object = dn->dn_object; db->db_level = level; db->db_blkid = blkid; db->db_last_dirty = NULL; db->db_dirtycnt = 0; db->db_dnode_handle = dn->dn_handle; db->db_parent = parent; db->db_blkptr = blkptr; db->db_user = NULL; db->db_user_immediate_evict = FALSE; db->db_freed_in_flight = FALSE; db->db_pending_evict = FALSE; if (blkid == DMU_BONUS_BLKID) { ASSERT3P(parent, ==, dn->dn_dbuf); db->db.db_size = DN_SLOTS_TO_BONUSLEN(dn->dn_num_slots) - (dn->dn_nblkptr-1) * sizeof (blkptr_t); ASSERT3U(db->db.db_size, >=, dn->dn_bonuslen); db->db.db_offset = DMU_BONUS_BLKID; db->db_state = DB_UNCACHED; db->db_caching_status = DB_NO_CACHE; /* the bonus dbuf is not placed in the hash table */ arc_space_consume(sizeof (dmu_buf_impl_t), ARC_SPACE_DBUF); return (db); } else if (blkid == DMU_SPILL_BLKID) { db->db.db_size = (blkptr != NULL) ? BP_GET_LSIZE(blkptr) : SPA_MINBLOCKSIZE; db->db.db_offset = 0; } else { int blocksize = db->db_level ? 1 << dn->dn_indblkshift : dn->dn_datablksz; db->db.db_size = blocksize; db->db.db_offset = db->db_blkid * blocksize; } /* * Hold the dn_dbufs_mtx while we get the new dbuf * in the hash table *and* added to the dbufs list. * This prevents a possible deadlock with someone * trying to look up this dbuf before its added to the * dn_dbufs list. */ mutex_enter(&dn->dn_dbufs_mtx); db->db_state = DB_EVICTING; if ((odb = dbuf_hash_insert(db)) != NULL) { /* someone else inserted it first */ kmem_cache_free(dbuf_kmem_cache, db); mutex_exit(&dn->dn_dbufs_mtx); DBUF_STAT_BUMP(hash_insert_race); return (odb); } avl_add(&dn->dn_dbufs, db); db->db_state = DB_UNCACHED; db->db_caching_status = DB_NO_CACHE; mutex_exit(&dn->dn_dbufs_mtx); arc_space_consume(sizeof (dmu_buf_impl_t), ARC_SPACE_DBUF); if (parent && parent != dn->dn_dbuf) dbuf_add_ref(parent, db); ASSERT(dn->dn_object == DMU_META_DNODE_OBJECT || zfs_refcount_count(&dn->dn_holds) > 0); (void) zfs_refcount_add(&dn->dn_holds, db); dprintf_dbuf(db, "db=%p\n", db); return (db); } typedef struct dbuf_prefetch_arg { spa_t *dpa_spa; /* The spa to issue the prefetch in. */ zbookmark_phys_t dpa_zb; /* The target block to prefetch. */ int dpa_epbs; /* Entries (blkptr_t's) Per Block Shift. */ int dpa_curlevel; /* The current level that we're reading */ dnode_t *dpa_dnode; /* The dnode associated with the prefetch */ zio_priority_t dpa_prio; /* The priority I/Os should be issued at. */ zio_t *dpa_zio; /* The parent zio_t for all prefetches. */ arc_flags_t dpa_aflags; /* Flags to pass to the final prefetch. */ } dbuf_prefetch_arg_t; /* * Actually issue the prefetch read for the block given. */ static void dbuf_issue_final_prefetch(dbuf_prefetch_arg_t *dpa, blkptr_t *bp) { if (BP_IS_HOLE(bp) || BP_IS_EMBEDDED(bp)) return; arc_flags_t aflags = dpa->dpa_aflags | ARC_FLAG_NOWAIT | ARC_FLAG_PREFETCH; ASSERT3U(dpa->dpa_curlevel, ==, BP_GET_LEVEL(bp)); ASSERT3U(dpa->dpa_curlevel, ==, dpa->dpa_zb.zb_level); ASSERT(dpa->dpa_zio != NULL); (void) arc_read(dpa->dpa_zio, dpa->dpa_spa, bp, NULL, NULL, dpa->dpa_prio, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &aflags, &dpa->dpa_zb); } /* * Called when an indirect block above our prefetch target is read in. This * will either read in the next indirect block down the tree or issue the actual * prefetch if the next block down is our target. */ static void dbuf_prefetch_indirect_done(zio_t *zio, const zbookmark_phys_t *zb, const blkptr_t *iobp, arc_buf_t *abuf, void *private) { dbuf_prefetch_arg_t *dpa = private; ASSERT3S(dpa->dpa_zb.zb_level, <, dpa->dpa_curlevel); ASSERT3S(dpa->dpa_curlevel, >, 0); if (abuf == NULL) { ASSERT(zio == NULL || zio->io_error != 0); kmem_free(dpa, sizeof (*dpa)); return; } ASSERT(zio == NULL || zio->io_error == 0); /* * The dpa_dnode is only valid if we are called with a NULL * zio. This indicates that the arc_read() returned without * first calling zio_read() to issue a physical read. Once * a physical read is made the dpa_dnode must be invalidated * as the locks guarding it may have been dropped. If the * dpa_dnode is still valid, then we want to add it to the dbuf * cache. To do so, we must hold the dbuf associated with the block * we just prefetched, read its contents so that we associate it * with an arc_buf_t, and then release it. */ if (zio != NULL) { ASSERT3S(BP_GET_LEVEL(zio->io_bp), ==, dpa->dpa_curlevel); if (zio->io_flags & ZIO_FLAG_RAW) { ASSERT3U(BP_GET_PSIZE(zio->io_bp), ==, zio->io_size); } else { ASSERT3U(BP_GET_LSIZE(zio->io_bp), ==, zio->io_size); } ASSERT3P(zio->io_spa, ==, dpa->dpa_spa); dpa->dpa_dnode = NULL; } else if (dpa->dpa_dnode != NULL) { uint64_t curblkid = dpa->dpa_zb.zb_blkid >> (dpa->dpa_epbs * (dpa->dpa_curlevel - dpa->dpa_zb.zb_level)); dmu_buf_impl_t *db = dbuf_hold_level(dpa->dpa_dnode, dpa->dpa_curlevel, curblkid, FTAG); (void) dbuf_read(db, NULL, DB_RF_MUST_SUCCEED | DB_RF_NOPREFETCH | DB_RF_HAVESTRUCT); dbuf_rele(db, FTAG); } if (abuf == NULL) { kmem_free(dpa, sizeof(*dpa)); return; } dpa->dpa_curlevel--; uint64_t nextblkid = dpa->dpa_zb.zb_blkid >> (dpa->dpa_epbs * (dpa->dpa_curlevel - dpa->dpa_zb.zb_level)); blkptr_t *bp = ((blkptr_t *)abuf->b_data) + P2PHASE(nextblkid, 1ULL << dpa->dpa_epbs); if (BP_IS_HOLE(bp)) { kmem_free(dpa, sizeof (*dpa)); } else if (dpa->dpa_curlevel == dpa->dpa_zb.zb_level) { ASSERT3U(nextblkid, ==, dpa->dpa_zb.zb_blkid); dbuf_issue_final_prefetch(dpa, bp); kmem_free(dpa, sizeof (*dpa)); } else { arc_flags_t iter_aflags = ARC_FLAG_NOWAIT; zbookmark_phys_t zb; /* flag if L2ARC eligible, l2arc_noprefetch then decides */ if (dpa->dpa_aflags & ARC_FLAG_L2CACHE) iter_aflags |= ARC_FLAG_L2CACHE; ASSERT3U(dpa->dpa_curlevel, ==, BP_GET_LEVEL(bp)); SET_BOOKMARK(&zb, dpa->dpa_zb.zb_objset, dpa->dpa_zb.zb_object, dpa->dpa_curlevel, nextblkid); (void) arc_read(dpa->dpa_zio, dpa->dpa_spa, bp, dbuf_prefetch_indirect_done, dpa, dpa->dpa_prio, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &iter_aflags, &zb); } arc_buf_destroy(abuf, private); } /* * Issue prefetch reads for the given block on the given level. If the indirect * blocks above that block are not in memory, we will read them in * asynchronously. As a result, this call never blocks waiting for a read to * complete. */ void dbuf_prefetch(dnode_t *dn, int64_t level, uint64_t blkid, zio_priority_t prio, arc_flags_t aflags) { blkptr_t bp; int epbs, nlevels, curlevel; uint64_t curblkid; ASSERT(blkid != DMU_BONUS_BLKID); ASSERT(RW_LOCK_HELD(&dn->dn_struct_rwlock)); if (blkid > dn->dn_maxblkid) return; if (dnode_block_freed(dn, blkid)) return; /* * This dnode hasn't been written to disk yet, so there's nothing to * prefetch. */ nlevels = dn->dn_phys->dn_nlevels; if (level >= nlevels || dn->dn_phys->dn_nblkptr == 0) return; epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT; if (dn->dn_phys->dn_maxblkid < blkid << (epbs * level)) return; dmu_buf_impl_t *db = dbuf_find(dn->dn_objset, dn->dn_object, level, blkid); if (db != NULL) { mutex_exit(&db->db_mtx); /* * This dbuf already exists. It is either CACHED, or * (we assume) about to be read or filled. */ return; } /* * Find the closest ancestor (indirect block) of the target block * that is present in the cache. In this indirect block, we will * find the bp that is at curlevel, curblkid. */ curlevel = level; curblkid = blkid; while (curlevel < nlevels - 1) { int parent_level = curlevel + 1; uint64_t parent_blkid = curblkid >> epbs; dmu_buf_impl_t *db; if (dbuf_hold_impl(dn, parent_level, parent_blkid, FALSE, TRUE, FTAG, &db) == 0) { blkptr_t *bpp = db->db_buf->b_data; bp = bpp[P2PHASE(curblkid, 1 << epbs)]; dbuf_rele(db, FTAG); break; } curlevel = parent_level; curblkid = parent_blkid; } if (curlevel == nlevels - 1) { /* No cached indirect blocks found. */ ASSERT3U(curblkid, <, dn->dn_phys->dn_nblkptr); bp = dn->dn_phys->dn_blkptr[curblkid]; } if (BP_IS_HOLE(&bp)) return; ASSERT3U(curlevel, ==, BP_GET_LEVEL(&bp)); zio_t *pio = zio_root(dmu_objset_spa(dn->dn_objset), NULL, NULL, ZIO_FLAG_CANFAIL); dbuf_prefetch_arg_t *dpa = kmem_zalloc(sizeof (*dpa), KM_SLEEP); dsl_dataset_t *ds = dn->dn_objset->os_dsl_dataset; SET_BOOKMARK(&dpa->dpa_zb, ds != NULL ? ds->ds_object : DMU_META_OBJSET, dn->dn_object, level, blkid); dpa->dpa_curlevel = curlevel; dpa->dpa_prio = prio; dpa->dpa_aflags = aflags; dpa->dpa_spa = dn->dn_objset->os_spa; dpa->dpa_dnode = dn; dpa->dpa_epbs = epbs; dpa->dpa_zio = pio; /* flag if L2ARC eligible, l2arc_noprefetch then decides */ if (DNODE_LEVEL_IS_L2CACHEABLE(dn, level)) dpa->dpa_aflags |= ARC_FLAG_L2CACHE; /* * If we have the indirect just above us, no need to do the asynchronous * prefetch chain; we'll just run the last step ourselves. If we're at * a higher level, though, we want to issue the prefetches for all the * indirect blocks asynchronously, so we can go on with whatever we were * doing. */ if (curlevel == level) { ASSERT3U(curblkid, ==, blkid); dbuf_issue_final_prefetch(dpa, &bp); kmem_free(dpa, sizeof (*dpa)); } else { arc_flags_t iter_aflags = ARC_FLAG_NOWAIT; zbookmark_phys_t zb; /* flag if L2ARC eligible, l2arc_noprefetch then decides */ if (DNODE_LEVEL_IS_L2CACHEABLE(dn, level)) iter_aflags |= ARC_FLAG_L2CACHE; SET_BOOKMARK(&zb, ds != NULL ? ds->ds_object : DMU_META_OBJSET, dn->dn_object, curlevel, curblkid); (void) arc_read(dpa->dpa_zio, dpa->dpa_spa, &bp, dbuf_prefetch_indirect_done, dpa, prio, ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE, &iter_aflags, &zb); } /* * We use pio here instead of dpa_zio since it's possible that * dpa may have already been freed. */ zio_nowait(pio); } #define DBUF_HOLD_IMPL_MAX_DEPTH 20 /* * Helper function for __dbuf_hold_impl() to copy a buffer. Handles * the case of encrypted, compressed and uncompressed buffers by * allocating the new buffer, respectively, with arc_alloc_raw_buf(), * arc_alloc_compressed_buf() or arc_alloc_buf().* * * NOTE: Declared noinline to avoid stack bloat in __dbuf_hold_impl(). */ noinline static void dbuf_hold_copy(struct dbuf_hold_impl_data *dh) { dnode_t *dn = dh->dh_dn; dmu_buf_impl_t *db = dh->dh_db; dbuf_dirty_record_t *dr = dh->dh_dr; arc_buf_t *data = dr->dt.dl.dr_data; enum zio_compress compress_type = arc_get_compression(data); if (compress_type != ZIO_COMPRESS_OFF) { dbuf_set_data(db, arc_alloc_compressed_buf( dn->dn_objset->os_spa, db, arc_buf_size(data), arc_buf_lsize(data), compress_type)); } else { dbuf_set_data(db, arc_alloc_buf(dn->dn_objset->os_spa, db, DBUF_GET_BUFC_TYPE(db), db->db.db_size)); } bcopy(data->b_data, db->db.db_data, arc_buf_size(data)); } /* * Returns with db_holds incremented, and db_mtx not held. * Note: dn_struct_rwlock must be held. */ static int __dbuf_hold_impl(struct dbuf_hold_impl_data *dh) { ASSERT3S(dh->dh_depth, <, DBUF_HOLD_IMPL_MAX_DEPTH); dh->dh_parent = NULL; ASSERT(dh->dh_blkid != DMU_BONUS_BLKID); ASSERT(RW_LOCK_HELD(&dh->dh_dn->dn_struct_rwlock)); ASSERT3U(dh->dh_dn->dn_nlevels, >, dh->dh_level); *(dh->dh_dbp) = NULL; /* dbuf_find() returns with db_mtx held */ dh->dh_db = dbuf_find(dh->dh_dn->dn_objset, dh->dh_dn->dn_object, dh->dh_level, dh->dh_blkid); if (dh->dh_db == NULL) { dh->dh_bp = NULL; if (dh->dh_fail_uncached) return (SET_ERROR(ENOENT)); ASSERT3P(dh->dh_parent, ==, NULL); dh->dh_err = dbuf_findbp(dh->dh_dn, dh->dh_level, dh->dh_blkid, dh->dh_fail_sparse, &dh->dh_parent, &dh->dh_bp, dh); if (dh->dh_fail_sparse) { if (dh->dh_err == 0 && dh->dh_bp && BP_IS_HOLE(dh->dh_bp)) dh->dh_err = SET_ERROR(ENOENT); if (dh->dh_err) { if (dh->dh_parent) dbuf_rele(dh->dh_parent, NULL); return (dh->dh_err); } } if (dh->dh_err && dh->dh_err != ENOENT) return (dh->dh_err); dh->dh_db = dbuf_create(dh->dh_dn, dh->dh_level, dh->dh_blkid, dh->dh_parent, dh->dh_bp); } if (dh->dh_fail_uncached && dh->dh_db->db_state != DB_CACHED) { mutex_exit(&dh->dh_db->db_mtx); return (SET_ERROR(ENOENT)); } if (dh->dh_db->db_buf != NULL) { arc_buf_access(dh->dh_db->db_buf); ASSERT3P(dh->dh_db->db.db_data, ==, dh->dh_db->db_buf->b_data); } ASSERT(dh->dh_db->db_buf == NULL || arc_referenced(dh->dh_db->db_buf)); /* * If this buffer is currently syncing out, and we are are * still referencing it from db_data, we need to make a copy * of it in case we decide we want to dirty it again in this txg. */ if (dh->dh_db->db_level == 0 && dh->dh_db->db_blkid != DMU_BONUS_BLKID && dh->dh_dn->dn_object != DMU_META_DNODE_OBJECT && dh->dh_db->db_state == DB_CACHED && dh->dh_db->db_data_pending) { dh->dh_dr = dh->dh_db->db_data_pending; if (dh->dh_dr->dt.dl.dr_data == dh->dh_db->db_buf) dbuf_hold_copy(dh); } if (multilist_link_active(&dh->dh_db->db_cache_link)) { ASSERT(zfs_refcount_is_zero(&dh->dh_db->db_holds)); ASSERT(dh->dh_db->db_caching_status == DB_DBUF_CACHE || dh->dh_db->db_caching_status == DB_DBUF_METADATA_CACHE); multilist_remove( dbuf_caches[dh->dh_db->db_caching_status].cache, dh->dh_db); (void) zfs_refcount_remove_many( &dbuf_caches[dh->dh_db->db_caching_status].size, dh->dh_db->db.db_size, dh->dh_db); if (dh->dh_db->db_caching_status == DB_DBUF_METADATA_CACHE) { DBUF_STAT_BUMPDOWN(metadata_cache_count); } else { DBUF_STAT_BUMPDOWN(cache_levels[dh->dh_db->db_level]); DBUF_STAT_BUMPDOWN(cache_count); DBUF_STAT_DECR(cache_levels_bytes[dh->dh_db->db_level], dh->dh_db->db.db_size); } dh->dh_db->db_caching_status = DB_NO_CACHE; } (void) zfs_refcount_add(&dh->dh_db->db_holds, dh->dh_tag); DBUF_VERIFY(dh->dh_db); mutex_exit(&dh->dh_db->db_mtx); /* NOTE: we can't rele the parent until after we drop the db_mtx */ if (dh->dh_parent) dbuf_rele(dh->dh_parent, NULL); ASSERT3P(DB_DNODE(dh->dh_db), ==, dh->dh_dn); ASSERT3U(dh->dh_db->db_blkid, ==, dh->dh_blkid); ASSERT3U(dh->dh_db->db_level, ==, dh->dh_level); *(dh->dh_dbp) = dh->dh_db; return (0); } /* * The following code preserves the recursive function dbuf_hold_impl() * but moves the local variables AND function arguments to the heap to * minimize the stack frame size. Enough space is initially allocated * on the stack for 20 levels of recursion. */ int dbuf_hold_impl(dnode_t *dn, uint8_t level, uint64_t blkid, boolean_t fail_sparse, boolean_t fail_uncached, void *tag, dmu_buf_impl_t **dbp) { struct dbuf_hold_impl_data *dh; int error; dh = kmem_alloc(sizeof (struct dbuf_hold_impl_data) * DBUF_HOLD_IMPL_MAX_DEPTH, KM_SLEEP); __dbuf_hold_impl_init(dh, dn, level, blkid, fail_sparse, fail_uncached, tag, dbp, 0); error = __dbuf_hold_impl(dh); kmem_free(dh, sizeof (struct dbuf_hold_impl_data) * DBUF_HOLD_IMPL_MAX_DEPTH); return (error); } static void __dbuf_hold_impl_init(struct dbuf_hold_impl_data *dh, dnode_t *dn, uint8_t level, uint64_t blkid, boolean_t fail_sparse, boolean_t fail_uncached, void *tag, dmu_buf_impl_t **dbp, int depth) { dh->dh_dn = dn; dh->dh_level = level; dh->dh_blkid = blkid; dh->dh_fail_sparse = fail_sparse; dh->dh_fail_uncached = fail_uncached; dh->dh_tag = tag; dh->dh_dbp = dbp; dh->dh_db = NULL; dh->dh_parent = NULL; dh->dh_bp = NULL; dh->dh_err = 0; dh->dh_dr = NULL; dh->dh_depth = depth; } dmu_buf_impl_t * dbuf_hold(dnode_t *dn, uint64_t blkid, void *tag) { return (dbuf_hold_level(dn, 0, blkid, tag)); } dmu_buf_impl_t * dbuf_hold_level(dnode_t *dn, int level, uint64_t blkid, void *tag) { dmu_buf_impl_t *db; int err = dbuf_hold_impl(dn, level, blkid, FALSE, FALSE, tag, &db); return (err ? NULL : db); } void dbuf_create_bonus(dnode_t *dn) { ASSERT(RW_WRITE_HELD(&dn->dn_struct_rwlock)); ASSERT(dn->dn_bonus == NULL); dn->dn_bonus = dbuf_create(dn, 0, DMU_BONUS_BLKID, dn->dn_dbuf, NULL); } int dbuf_spill_set_blksz(dmu_buf_t *db_fake, uint64_t blksz, dmu_tx_t *tx) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dnode_t *dn; if (db->db_blkid != DMU_SPILL_BLKID) return (SET_ERROR(ENOTSUP)); if (blksz == 0) blksz = SPA_MINBLOCKSIZE; ASSERT3U(blksz, <=, spa_maxblocksize(dmu_objset_spa(db->db_objset))); blksz = P2ROUNDUP(blksz, SPA_MINBLOCKSIZE); DB_DNODE_ENTER(db); dn = DB_DNODE(db); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); dbuf_new_size(db, blksz, tx); rw_exit(&dn->dn_struct_rwlock); DB_DNODE_EXIT(db); return (0); } void dbuf_rm_spill(dnode_t *dn, dmu_tx_t *tx) { dbuf_free_range(dn, DMU_SPILL_BLKID, DMU_SPILL_BLKID, tx); } #pragma weak dmu_buf_add_ref = dbuf_add_ref void dbuf_add_ref(dmu_buf_impl_t *db, void *tag) { int64_t holds = zfs_refcount_add(&db->db_holds, tag); ASSERT3S(holds, >, 1); } #pragma weak dmu_buf_try_add_ref = dbuf_try_add_ref boolean_t dbuf_try_add_ref(dmu_buf_t *db_fake, objset_t *os, uint64_t obj, uint64_t blkid, void *tag) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dmu_buf_impl_t *found_db; boolean_t result = B_FALSE; if (db->db_blkid == DMU_BONUS_BLKID) found_db = dbuf_find_bonus(os, obj); else found_db = dbuf_find(os, obj, 0, blkid); if (found_db != NULL) { if (db == found_db && dbuf_refcount(db) > db->db_dirtycnt) { (void) zfs_refcount_add(&db->db_holds, tag); result = B_TRUE; } mutex_exit(&db->db_mtx); } return (result); } /* * If you call dbuf_rele() you had better not be referencing the dnode handle * unless you have some other direct or indirect hold on the dnode. (An indirect * hold is a hold on one of the dnode's dbufs, including the bonus buffer.) * Without that, the dbuf_rele() could lead to a dnode_rele() followed by the * dnode's parent dbuf evicting its dnode handles. */ void dbuf_rele(dmu_buf_impl_t *db, void *tag) { mutex_enter(&db->db_mtx); dbuf_rele_and_unlock(db, tag, B_FALSE); } void dmu_buf_rele(dmu_buf_t *db, void *tag) { dbuf_rele((dmu_buf_impl_t *)db, tag); } /* * dbuf_rele() for an already-locked dbuf. This is necessary to allow * db_dirtycnt and db_holds to be updated atomically. The 'evicting' * argument should be set if we are already in the dbuf-evicting code * path, in which case we don't want to recursively evict. This allows us to * avoid deeply nested stacks that would have a call flow similar to this: * * dbuf_rele()-->dbuf_rele_and_unlock()-->dbuf_evict_notify() * ^ | * | | * +-----dbuf_destroy()<--dbuf_evict_one()<--------+ * */ void dbuf_rele_and_unlock(dmu_buf_impl_t *db, void *tag, boolean_t evicting) { int64_t holds; uint64_t size; ASSERT(MUTEX_HELD(&db->db_mtx)); DBUF_VERIFY(db); /* * Remove the reference to the dbuf before removing its hold on the * dnode so we can guarantee in dnode_move() that a referenced bonus * buffer has a corresponding dnode hold. */ holds = zfs_refcount_remove(&db->db_holds, tag); ASSERT(holds >= 0); /* * We can't freeze indirects if there is a possibility that they * may be modified in the current syncing context. */ if (db->db_buf != NULL && holds == (db->db_level == 0 ? db->db_dirtycnt : 0)) { arc_buf_freeze(db->db_buf); } if (holds == db->db_dirtycnt && db->db_level == 0 && db->db_user_immediate_evict) dbuf_evict_user(db); if (holds == 0) { if (db->db_blkid == DMU_BONUS_BLKID) { dnode_t *dn; boolean_t evict_dbuf = db->db_pending_evict; /* * If the dnode moves here, we cannot cross this * barrier until the move completes. */ DB_DNODE_ENTER(db); dn = DB_DNODE(db); atomic_dec_32(&dn->dn_dbufs_count); /* * Decrementing the dbuf count means that the bonus * buffer's dnode hold is no longer discounted in * dnode_move(). The dnode cannot move until after * the dnode_rele() below. */ DB_DNODE_EXIT(db); /* * Do not reference db after its lock is dropped. * Another thread may evict it. */ mutex_exit(&db->db_mtx); if (evict_dbuf) dnode_evict_bonus(dn); dnode_rele(dn, db); } else if (db->db_buf == NULL) { /* * This is a special case: we never associated this * dbuf with any data allocated from the ARC. */ ASSERT(db->db_state == DB_UNCACHED || db->db_state == DB_NOFILL); dbuf_destroy(db); } else if (arc_released(db->db_buf)) { /* * This dbuf has anonymous data associated with it. */ dbuf_destroy(db); } else { boolean_t do_arc_evict = B_FALSE; blkptr_t bp; spa_t *spa = dmu_objset_spa(db->db_objset); if (!DBUF_IS_CACHEABLE(db) && db->db_blkptr != NULL && !BP_IS_HOLE(db->db_blkptr) && !BP_IS_EMBEDDED(db->db_blkptr)) { do_arc_evict = B_TRUE; bp = *db->db_blkptr; } if (!DBUF_IS_CACHEABLE(db) || db->db_pending_evict) { dbuf_destroy(db); } else if (!multilist_link_active(&db->db_cache_link)) { ASSERT3U(db->db_caching_status, ==, DB_NO_CACHE); dbuf_cached_state_t dcs = dbuf_include_in_metadata_cache(db) ? DB_DBUF_METADATA_CACHE : DB_DBUF_CACHE; db->db_caching_status = dcs; multilist_insert(dbuf_caches[dcs].cache, db); size = zfs_refcount_add_many( &dbuf_caches[dcs].size, db->db.db_size, db); if (dcs == DB_DBUF_METADATA_CACHE) { DBUF_STAT_BUMP(metadata_cache_count); DBUF_STAT_MAX( metadata_cache_size_bytes_max, size); } else { DBUF_STAT_BUMP( cache_levels[db->db_level]); DBUF_STAT_BUMP(cache_count); DBUF_STAT_INCR( cache_levels_bytes[db->db_level], db->db.db_size); DBUF_STAT_MAX(cache_size_bytes_max, size); } mutex_exit(&db->db_mtx); if (dcs == DB_DBUF_CACHE && !evicting) dbuf_evict_notify(size); } if (do_arc_evict) arc_freed(spa, &bp); } } else { mutex_exit(&db->db_mtx); } } #pragma weak dmu_buf_refcount = dbuf_refcount uint64_t dbuf_refcount(dmu_buf_impl_t *db) { return (zfs_refcount_count(&db->db_holds)); } void * dmu_buf_replace_user(dmu_buf_t *db_fake, dmu_buf_user_t *old_user, dmu_buf_user_t *new_user) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; mutex_enter(&db->db_mtx); dbuf_verify_user(db, DBVU_NOT_EVICTING); if (db->db_user == old_user) db->db_user = new_user; else old_user = db->db_user; dbuf_verify_user(db, DBVU_NOT_EVICTING); mutex_exit(&db->db_mtx); return (old_user); } void * dmu_buf_set_user(dmu_buf_t *db_fake, dmu_buf_user_t *user) { return (dmu_buf_replace_user(db_fake, NULL, user)); } void * dmu_buf_set_user_ie(dmu_buf_t *db_fake, dmu_buf_user_t *user) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; db->db_user_immediate_evict = TRUE; return (dmu_buf_set_user(db_fake, user)); } void * dmu_buf_remove_user(dmu_buf_t *db_fake, dmu_buf_user_t *user) { return (dmu_buf_replace_user(db_fake, user, NULL)); } void * dmu_buf_get_user(dmu_buf_t *db_fake) { dmu_buf_impl_t *db = (dmu_buf_impl_t *)db_fake; dbuf_verify_user(db, DBVU_NOT_EVICTING); return (db->db_user); } void dmu_buf_user_evict_wait() { taskq_wait(dbu_evict_taskq); } blkptr_t * dmu_buf_get_blkptr(dmu_buf_t *db) { dmu_buf_impl_t *dbi = (dmu_buf_impl_t *)db; return (dbi->db_blkptr); } objset_t * dmu_buf_get_objset(dmu_buf_t *db) { dmu_buf_impl_t *dbi = (dmu_buf_impl_t *)db; return (dbi->db_objset); } dnode_t * dmu_buf_dnode_enter(dmu_buf_t *db) { dmu_buf_impl_t *dbi = (dmu_buf_impl_t *)db; DB_DNODE_ENTER(dbi); return (DB_DNODE(dbi)); } void dmu_buf_dnode_exit(dmu_buf_t *db) { dmu_buf_impl_t *dbi = (dmu_buf_impl_t *)db; DB_DNODE_EXIT(dbi); } static void dbuf_check_blkptr(dnode_t *dn, dmu_buf_impl_t *db) { /* ASSERT(dmu_tx_is_syncing(tx) */ ASSERT(MUTEX_HELD(&db->db_mtx)); if (db->db_blkptr != NULL) return; if (db->db_blkid == DMU_SPILL_BLKID) { db->db_blkptr = DN_SPILL_BLKPTR(dn->dn_phys); BP_ZERO(db->db_blkptr); return; } if (db->db_level == dn->dn_phys->dn_nlevels-1) { /* * This buffer was allocated at a time when there was * no available blkptrs from the dnode, or it was * inappropriate to hook it in (i.e., nlevels mis-match). */ ASSERT(db->db_blkid < dn->dn_phys->dn_nblkptr); ASSERT(db->db_parent == NULL); db->db_parent = dn->dn_dbuf; db->db_blkptr = &dn->dn_phys->dn_blkptr[db->db_blkid]; DBUF_VERIFY(db); } else { dmu_buf_impl_t *parent = db->db_parent; int epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT; ASSERT(dn->dn_phys->dn_nlevels > 1); if (parent == NULL) { mutex_exit(&db->db_mtx); rw_enter(&dn->dn_struct_rwlock, RW_READER); parent = dbuf_hold_level(dn, db->db_level + 1, db->db_blkid >> epbs, db); rw_exit(&dn->dn_struct_rwlock); mutex_enter(&db->db_mtx); db->db_parent = parent; } db->db_blkptr = (blkptr_t *)parent->db.db_data + (db->db_blkid & ((1ULL << epbs) - 1)); DBUF_VERIFY(db); } } /* * dbuf_sync_indirect() is called recursively from dbuf_sync_list() so it * is critical the we not allow the compiler to inline this function in to * dbuf_sync_list() thereby drastically bloating the stack usage. */ noinline static void dbuf_sync_indirect(dbuf_dirty_record_t *dr, dmu_tx_t *tx) { dmu_buf_impl_t *db = dr->dr_dbuf; dnode_t *dn; zio_t *zio; ASSERT(dmu_tx_is_syncing(tx)); dprintf_dbuf_bp(db, db->db_blkptr, "blkptr=%p", db->db_blkptr); mutex_enter(&db->db_mtx); ASSERT(db->db_level > 0); DBUF_VERIFY(db); /* Read the block if it hasn't been read yet. */ if (db->db_buf == NULL) { mutex_exit(&db->db_mtx); (void) dbuf_read(db, NULL, DB_RF_MUST_SUCCEED); mutex_enter(&db->db_mtx); } ASSERT3U(db->db_state, ==, DB_CACHED); ASSERT(db->db_buf != NULL); DB_DNODE_ENTER(db); dn = DB_DNODE(db); /* Indirect block size must match what the dnode thinks it is. */ ASSERT3U(db->db.db_size, ==, 1<dn_phys->dn_indblkshift); dbuf_check_blkptr(dn, db); DB_DNODE_EXIT(db); /* Provide the pending dirty record to child dbufs */ db->db_data_pending = dr; mutex_exit(&db->db_mtx); dbuf_write(dr, db->db_buf, tx); zio = dr->dr_zio; mutex_enter(&dr->dt.di.dr_mtx); dbuf_sync_list(&dr->dt.di.dr_children, db->db_level - 1, tx); ASSERT(list_head(&dr->dt.di.dr_children) == NULL); mutex_exit(&dr->dt.di.dr_mtx); zio_nowait(zio); } /* * dbuf_sync_leaf() is called recursively from dbuf_sync_list() so it is * critical the we not allow the compiler to inline this function in to * dbuf_sync_list() thereby drastically bloating the stack usage. */ noinline static void dbuf_sync_leaf(dbuf_dirty_record_t *dr, dmu_tx_t *tx) { arc_buf_t **datap = &dr->dt.dl.dr_data; dmu_buf_impl_t *db = dr->dr_dbuf; dnode_t *dn; objset_t *os; uint64_t txg = tx->tx_txg; ASSERT(dmu_tx_is_syncing(tx)); dprintf_dbuf_bp(db, db->db_blkptr, "blkptr=%p", db->db_blkptr); mutex_enter(&db->db_mtx); /* * To be synced, we must be dirtied. But we * might have been freed after the dirty. */ if (db->db_state == DB_UNCACHED) { /* This buffer has been freed since it was dirtied */ ASSERT(db->db.db_data == NULL); } else if (db->db_state == DB_FILL) { /* This buffer was freed and is now being re-filled */ ASSERT(db->db.db_data != dr->dt.dl.dr_data); } else { ASSERT(db->db_state == DB_CACHED || db->db_state == DB_NOFILL); } DBUF_VERIFY(db); DB_DNODE_ENTER(db); dn = DB_DNODE(db); if (db->db_blkid == DMU_SPILL_BLKID) { mutex_enter(&dn->dn_mtx); if (!(dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR)) { /* * In the previous transaction group, the bonus buffer * was entirely used to store the attributes for the * dnode which overrode the dn_spill field. However, * when adding more attributes to the file a spill * block was required to hold the extra attributes. * * Make sure to clear the garbage left in the dn_spill * field from the previous attributes in the bonus * buffer. Otherwise, after writing out the spill * block to the new allocated dva, it will free * the old block pointed to by the invalid dn_spill. */ db->db_blkptr = NULL; } dn->dn_phys->dn_flags |= DNODE_FLAG_SPILL_BLKPTR; mutex_exit(&dn->dn_mtx); } /* * If this is a bonus buffer, simply copy the bonus data into the * dnode. It will be written out when the dnode is synced (and it * will be synced, since it must have been dirty for dbuf_sync to * be called). */ if (db->db_blkid == DMU_BONUS_BLKID) { dbuf_dirty_record_t **drp; ASSERT(*datap != NULL); ASSERT0(db->db_level); ASSERT3U(DN_MAX_BONUS_LEN(dn->dn_phys), <=, DN_SLOTS_TO_BONUSLEN(dn->dn_phys->dn_extra_slots + 1)); bcopy(*datap, DN_BONUS(dn->dn_phys), DN_MAX_BONUS_LEN(dn->dn_phys)); DB_DNODE_EXIT(db); if (*datap != db->db.db_data) { int slots = DB_DNODE(db)->dn_num_slots; int bonuslen = DN_SLOTS_TO_BONUSLEN(slots); zio_buf_free(*datap, bonuslen); arc_space_return(bonuslen, ARC_SPACE_BONUS); } db->db_data_pending = NULL; drp = &db->db_last_dirty; while (*drp != dr) drp = &(*drp)->dr_next; ASSERT(dr->dr_next == NULL); ASSERT(dr->dr_dbuf == db); *drp = dr->dr_next; if (dr->dr_dbuf->db_level != 0) { mutex_destroy(&dr->dt.di.dr_mtx); list_destroy(&dr->dt.di.dr_children); } kmem_free(dr, sizeof (dbuf_dirty_record_t)); ASSERT(db->db_dirtycnt > 0); db->db_dirtycnt -= 1; dbuf_rele_and_unlock(db, (void *)(uintptr_t)txg, B_FALSE); return; } os = dn->dn_objset; /* * This function may have dropped the db_mtx lock allowing a dmu_sync * operation to sneak in. As a result, we need to ensure that we * don't check the dr_override_state until we have returned from * dbuf_check_blkptr. */ dbuf_check_blkptr(dn, db); /* * If this buffer is in the middle of an immediate write, * wait for the synchronous IO to complete. */ while (dr->dt.dl.dr_override_state == DR_IN_DMU_SYNC) { ASSERT(dn->dn_object != DMU_META_DNODE_OBJECT); cv_wait(&db->db_changed, &db->db_mtx); ASSERT(dr->dt.dl.dr_override_state != DR_NOT_OVERRIDDEN); } if (db->db_state != DB_NOFILL && dn->dn_object != DMU_META_DNODE_OBJECT && zfs_refcount_count(&db->db_holds) > 1 && dr->dt.dl.dr_override_state != DR_OVERRIDDEN && *datap == db->db_buf) { /* * If this buffer is currently "in use" (i.e., there * are active holds and db_data still references it), * then make a copy before we start the write so that * any modifications from the open txg will not leak * into this write. * * NOTE: this copy does not need to be made for * objects only modified in the syncing context (e.g. * DNONE_DNODE blocks). */ int psize = arc_buf_size(*datap); arc_buf_contents_t type = DBUF_GET_BUFC_TYPE(db); enum zio_compress compress_type = arc_get_compression(*datap); if (compress_type == ZIO_COMPRESS_OFF) { *datap = arc_alloc_buf(os->os_spa, db, type, psize); } else { ASSERT3U(type, ==, ARC_BUFC_DATA); int lsize = arc_buf_lsize(*datap); *datap = arc_alloc_compressed_buf(os->os_spa, db, psize, lsize, compress_type); } bcopy(db->db.db_data, (*datap)->b_data, psize); } db->db_data_pending = dr; mutex_exit(&db->db_mtx); dbuf_write(dr, *datap, tx); ASSERT(!list_link_active(&dr->dr_dirty_node)); if (dn->dn_object == DMU_META_DNODE_OBJECT) { list_insert_tail(&dn->dn_dirty_records[txg&TXG_MASK], dr); DB_DNODE_EXIT(db); } else { /* * Although zio_nowait() does not "wait for an IO", it does * initiate the IO. If this is an empty write it seems plausible * that the IO could actually be completed before the nowait * returns. We need to DB_DNODE_EXIT() first in case * zio_nowait() invalidates the dbuf. */ DB_DNODE_EXIT(db); zio_nowait(dr->dr_zio); } } void dbuf_sync_list(list_t *list, int level, dmu_tx_t *tx) { dbuf_dirty_record_t *dr; while (dr = list_head(list)) { if (dr->dr_zio != NULL) { /* * If we find an already initialized zio then we * are processing the meta-dnode, and we have finished. * The dbufs for all dnodes are put back on the list * during processing, so that we can zio_wait() * these IOs after initiating all child IOs. */ ASSERT3U(dr->dr_dbuf->db.db_object, ==, DMU_META_DNODE_OBJECT); break; } if (dr->dr_dbuf->db_blkid != DMU_BONUS_BLKID && dr->dr_dbuf->db_blkid != DMU_SPILL_BLKID) { VERIFY3U(dr->dr_dbuf->db_level, ==, level); } list_remove(list, dr); if (dr->dr_dbuf->db_level > 0) dbuf_sync_indirect(dr, tx); else dbuf_sync_leaf(dr, tx); } } /* ARGSUSED */ static void dbuf_write_ready(zio_t *zio, arc_buf_t *buf, void *vdb) { dmu_buf_impl_t *db = vdb; dnode_t *dn; blkptr_t *bp = zio->io_bp; blkptr_t *bp_orig = &zio->io_bp_orig; spa_t *spa = zio->io_spa; int64_t delta; uint64_t fill = 0; int i; ASSERT3P(db->db_blkptr, !=, NULL); ASSERT3P(&db->db_data_pending->dr_bp_copy, ==, bp); DB_DNODE_ENTER(db); dn = DB_DNODE(db); delta = bp_get_dsize_sync(spa, bp) - bp_get_dsize_sync(spa, bp_orig); dnode_diduse_space(dn, delta - zio->io_prev_space_delta); zio->io_prev_space_delta = delta; if (bp->blk_birth != 0) { ASSERT((db->db_blkid != DMU_SPILL_BLKID && BP_GET_TYPE(bp) == dn->dn_type) || (db->db_blkid == DMU_SPILL_BLKID && BP_GET_TYPE(bp) == dn->dn_bonustype) || BP_IS_EMBEDDED(bp)); ASSERT(BP_GET_LEVEL(bp) == db->db_level); } mutex_enter(&db->db_mtx); #ifdef ZFS_DEBUG if (db->db_blkid == DMU_SPILL_BLKID) { ASSERT(dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR); ASSERT(!(BP_IS_HOLE(bp)) && db->db_blkptr == DN_SPILL_BLKPTR(dn->dn_phys)); } #endif if (db->db_level == 0) { mutex_enter(&dn->dn_mtx); if (db->db_blkid > dn->dn_phys->dn_maxblkid && db->db_blkid != DMU_SPILL_BLKID) dn->dn_phys->dn_maxblkid = db->db_blkid; mutex_exit(&dn->dn_mtx); if (dn->dn_type == DMU_OT_DNODE) { i = 0; while (i < db->db.db_size) { dnode_phys_t *dnp = (void *)(((char *)db->db.db_data) + i); i += DNODE_MIN_SIZE; if (dnp->dn_type != DMU_OT_NONE) { fill++; i += dnp->dn_extra_slots * DNODE_MIN_SIZE; } } } else { if (BP_IS_HOLE(bp)) { fill = 0; } else { fill = 1; } } } else { blkptr_t *ibp = db->db.db_data; ASSERT3U(db->db.db_size, ==, 1<dn_phys->dn_indblkshift); for (i = db->db.db_size >> SPA_BLKPTRSHIFT; i > 0; i--, ibp++) { if (BP_IS_HOLE(ibp)) continue; fill += BP_GET_FILL(ibp); } } DB_DNODE_EXIT(db); if (!BP_IS_EMBEDDED(bp)) bp->blk_fill = fill; mutex_exit(&db->db_mtx); rw_enter(&dn->dn_struct_rwlock, RW_WRITER); *db->db_blkptr = *bp; rw_exit(&dn->dn_struct_rwlock); } /* ARGSUSED */ /* * This function gets called just prior to running through the compression * stage of the zio pipeline. If we're an indirect block comprised of only * holes, then we want this indirect to be compressed away to a hole. In * order to do that we must zero out any information about the holes that * this indirect points to prior to before we try to compress it. */ static void dbuf_write_children_ready(zio_t *zio, arc_buf_t *buf, void *vdb) { dmu_buf_impl_t *db = vdb; dnode_t *dn; blkptr_t *bp; unsigned int epbs, i; ASSERT3U(db->db_level, >, 0); DB_DNODE_ENTER(db); dn = DB_DNODE(db); epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT; ASSERT3U(epbs, <, 31); /* Determine if all our children are holes */ for (i = 0, bp = db->db.db_data; i < 1 << epbs; i++, bp++) { if (!BP_IS_HOLE(bp)) break; } /* * If all the children are holes, then zero them all out so that * we may get compressed away. */ if (i == 1 << epbs) { /* * We only found holes. Grab the rwlock to prevent * anybody from reading the blocks we're about to * zero out. */ rw_enter(&dn->dn_struct_rwlock, RW_WRITER); bzero(db->db.db_data, db->db.db_size); rw_exit(&dn->dn_struct_rwlock); } DB_DNODE_EXIT(db); } /* * The SPA will call this callback several times for each zio - once * for every physical child i/o (zio->io_phys_children times). This * allows the DMU to monitor the progress of each logical i/o. For example, * there may be 2 copies of an indirect block, or many fragments of a RAID-Z * block. There may be a long delay before all copies/fragments are completed, * so this callback allows us to retire dirty space gradually, as the physical * i/os complete. */ /* ARGSUSED */ static void dbuf_write_physdone(zio_t *zio, arc_buf_t *buf, void *arg) { dmu_buf_impl_t *db = arg; objset_t *os = db->db_objset; dsl_pool_t *dp = dmu_objset_pool(os); dbuf_dirty_record_t *dr; int delta = 0; dr = db->db_data_pending; ASSERT3U(dr->dr_txg, ==, zio->io_txg); /* * The callback will be called io_phys_children times. Retire one * portion of our dirty space each time we are called. Any rounding * error will be cleaned up by dsl_pool_sync()'s call to * dsl_pool_undirty_space(). */ delta = dr->dr_accounted / zio->io_phys_children; dsl_pool_undirty_space(dp, delta, zio->io_txg); } /* ARGSUSED */ static void dbuf_write_done(zio_t *zio, arc_buf_t *buf, void *vdb) { dmu_buf_impl_t *db = vdb; blkptr_t *bp_orig = &zio->io_bp_orig; blkptr_t *bp = db->db_blkptr; objset_t *os = db->db_objset; dmu_tx_t *tx = os->os_synctx; dbuf_dirty_record_t **drp, *dr; ASSERT0(zio->io_error); ASSERT(db->db_blkptr == bp); /* * For nopwrites and rewrites we ensure that the bp matches our * original and bypass all the accounting. */ if (zio->io_flags & (ZIO_FLAG_IO_REWRITE | ZIO_FLAG_NOPWRITE)) { ASSERT(BP_EQUAL(bp, bp_orig)); } else { dsl_dataset_t *ds = os->os_dsl_dataset; (void) dsl_dataset_block_kill(ds, bp_orig, tx, B_TRUE); dsl_dataset_block_born(ds, bp, tx); } mutex_enter(&db->db_mtx); DBUF_VERIFY(db); drp = &db->db_last_dirty; while ((dr = *drp) != db->db_data_pending) drp = &dr->dr_next; ASSERT(!list_link_active(&dr->dr_dirty_node)); ASSERT(dr->dr_dbuf == db); ASSERT(dr->dr_next == NULL); *drp = dr->dr_next; #ifdef ZFS_DEBUG if (db->db_blkid == DMU_SPILL_BLKID) { dnode_t *dn; DB_DNODE_ENTER(db); dn = DB_DNODE(db); ASSERT(dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR); ASSERT(!(BP_IS_HOLE(db->db_blkptr)) && db->db_blkptr == DN_SPILL_BLKPTR(dn->dn_phys)); DB_DNODE_EXIT(db); } #endif if (db->db_level == 0) { ASSERT(db->db_blkid != DMU_BONUS_BLKID); ASSERT(dr->dt.dl.dr_override_state == DR_NOT_OVERRIDDEN); if (db->db_state != DB_NOFILL) { if (dr->dt.dl.dr_data != db->db_buf) arc_buf_destroy(dr->dt.dl.dr_data, db); } } else { dnode_t *dn; DB_DNODE_ENTER(db); dn = DB_DNODE(db); ASSERT(list_head(&dr->dt.di.dr_children) == NULL); ASSERT3U(db->db.db_size, ==, 1 << dn->dn_phys->dn_indblkshift); if (!BP_IS_HOLE(db->db_blkptr)) { int epbs = dn->dn_phys->dn_indblkshift - SPA_BLKPTRSHIFT; ASSERT3U(db->db_blkid, <=, dn->dn_phys->dn_maxblkid >> (db->db_level * epbs)); ASSERT3U(BP_GET_LSIZE(db->db_blkptr), ==, db->db.db_size); } DB_DNODE_EXIT(db); mutex_destroy(&dr->dt.di.dr_mtx); list_destroy(&dr->dt.di.dr_children); } kmem_free(dr, sizeof (dbuf_dirty_record_t)); cv_broadcast(&db->db_changed); ASSERT(db->db_dirtycnt > 0); db->db_dirtycnt -= 1; db->db_data_pending = NULL; dbuf_rele_and_unlock(db, (void *)(uintptr_t)tx->tx_txg, B_FALSE); } static void dbuf_write_nofill_ready(zio_t *zio) { dbuf_write_ready(zio, NULL, zio->io_private); } static void dbuf_write_nofill_done(zio_t *zio) { dbuf_write_done(zio, NULL, zio->io_private); } static void dbuf_write_override_ready(zio_t *zio) { dbuf_dirty_record_t *dr = zio->io_private; dmu_buf_impl_t *db = dr->dr_dbuf; dbuf_write_ready(zio, NULL, db); } static void dbuf_write_override_done(zio_t *zio) { dbuf_dirty_record_t *dr = zio->io_private; dmu_buf_impl_t *db = dr->dr_dbuf; blkptr_t *obp = &dr->dt.dl.dr_overridden_by; mutex_enter(&db->db_mtx); if (!BP_EQUAL(zio->io_bp, obp)) { if (!BP_IS_HOLE(obp)) dsl_free(spa_get_dsl(zio->io_spa), zio->io_txg, obp); arc_release(dr->dt.dl.dr_data, db); } mutex_exit(&db->db_mtx); dbuf_write_done(zio, NULL, db); if (zio->io_abd != NULL) abd_put(zio->io_abd); } typedef struct dbuf_remap_impl_callback_arg { objset_t *drica_os; uint64_t drica_blk_birth; dmu_tx_t *drica_tx; } dbuf_remap_impl_callback_arg_t; static void dbuf_remap_impl_callback(uint64_t vdev, uint64_t offset, uint64_t size, void *arg) { dbuf_remap_impl_callback_arg_t *drica = arg; objset_t *os = drica->drica_os; spa_t *spa = dmu_objset_spa(os); dmu_tx_t *tx = drica->drica_tx; ASSERT(dsl_pool_sync_context(spa_get_dsl(spa))); if (os == spa_meta_objset(spa)) { spa_vdev_indirect_mark_obsolete(spa, vdev, offset, size, tx); } else { dsl_dataset_block_remapped(dmu_objset_ds(os), vdev, offset, size, drica->drica_blk_birth, tx); } } static void dbuf_remap_impl(dnode_t *dn, blkptr_t *bp, dmu_tx_t *tx) { blkptr_t bp_copy = *bp; spa_t *spa = dmu_objset_spa(dn->dn_objset); dbuf_remap_impl_callback_arg_t drica; ASSERT(dsl_pool_sync_context(spa_get_dsl(spa))); drica.drica_os = dn->dn_objset; drica.drica_blk_birth = bp->blk_birth; drica.drica_tx = tx; if (spa_remap_blkptr(spa, &bp_copy, dbuf_remap_impl_callback, &drica)) { /* * The struct_rwlock prevents dbuf_read_impl() from * dereferencing the BP while we are changing it. To * avoid lock contention, only grab it when we are actually * changing the BP. */ rw_enter(&dn->dn_struct_rwlock, RW_WRITER); *bp = bp_copy; rw_exit(&dn->dn_struct_rwlock); } } /* * Returns true if a dbuf_remap would modify the dbuf. We do this by attempting * to remap a copy of every bp in the dbuf. */ boolean_t dbuf_can_remap(const dmu_buf_impl_t *db) { spa_t *spa = dmu_objset_spa(db->db_objset); blkptr_t *bp = db->db.db_data; boolean_t ret = B_FALSE; ASSERT3U(db->db_level, >, 0); ASSERT3S(db->db_state, ==, DB_CACHED); ASSERT(spa_feature_is_active(spa, SPA_FEATURE_DEVICE_REMOVAL)); spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); for (int i = 0; i < db->db.db_size >> SPA_BLKPTRSHIFT; i++) { blkptr_t bp_copy = bp[i]; if (spa_remap_blkptr(spa, &bp_copy, NULL, NULL)) { ret = B_TRUE; break; } } spa_config_exit(spa, SCL_VDEV, FTAG); return (ret); } boolean_t dnode_needs_remap(const dnode_t *dn) { spa_t *spa = dmu_objset_spa(dn->dn_objset); boolean_t ret = B_FALSE; if (dn->dn_phys->dn_nlevels == 0) { return (B_FALSE); } ASSERT(spa_feature_is_active(spa, SPA_FEATURE_DEVICE_REMOVAL)); spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); for (int j = 0; j < dn->dn_phys->dn_nblkptr; j++) { blkptr_t bp_copy = dn->dn_phys->dn_blkptr[j]; if (spa_remap_blkptr(spa, &bp_copy, NULL, NULL)) { ret = B_TRUE; break; } } spa_config_exit(spa, SCL_VDEV, FTAG); return (ret); } /* * Remap any existing BP's to concrete vdevs, if possible. */ static void dbuf_remap(dnode_t *dn, dmu_buf_impl_t *db, dmu_tx_t *tx) { spa_t *spa = dmu_objset_spa(db->db_objset); ASSERT(dsl_pool_sync_context(spa_get_dsl(spa))); if (!spa_feature_is_active(spa, SPA_FEATURE_DEVICE_REMOVAL)) return; if (db->db_level > 0) { blkptr_t *bp = db->db.db_data; for (int i = 0; i < db->db.db_size >> SPA_BLKPTRSHIFT; i++) { dbuf_remap_impl(dn, &bp[i], tx); } } else if (db->db.db_object == DMU_META_DNODE_OBJECT) { dnode_phys_t *dnp = db->db.db_data; ASSERT3U(db->db_dnode_handle->dnh_dnode->dn_type, ==, DMU_OT_DNODE); for (int i = 0; i < db->db.db_size >> DNODE_SHIFT; i += dnp[i].dn_extra_slots + 1) { for (int j = 0; j < dnp[i].dn_nblkptr; j++) { dbuf_remap_impl(dn, &dnp[i].dn_blkptr[j], tx); } } } } /* Issue I/O to commit a dirty buffer to disk. */ static void dbuf_write(dbuf_dirty_record_t *dr, arc_buf_t *data, dmu_tx_t *tx) { dmu_buf_impl_t *db = dr->dr_dbuf; dnode_t *dn; objset_t *os; dmu_buf_impl_t *parent = db->db_parent; uint64_t txg = tx->tx_txg; zbookmark_phys_t zb; zio_prop_t zp; zio_t *zio; int wp_flag = 0; ASSERT(dmu_tx_is_syncing(tx)); DB_DNODE_ENTER(db); dn = DB_DNODE(db); os = dn->dn_objset; if (db->db_state != DB_NOFILL) { if (db->db_level > 0 || dn->dn_type == DMU_OT_DNODE) { /* * Private object buffers are released here rather * than in dbuf_dirty() since they are only modified * in the syncing context and we don't want the * overhead of making multiple copies of the data. */ if (BP_IS_HOLE(db->db_blkptr)) { arc_buf_thaw(data); } else { dbuf_release_bp(db); } dbuf_remap(dn, db, tx); } } if (parent != dn->dn_dbuf) { /* Our parent is an indirect block. */ /* We have a dirty parent that has been scheduled for write. */ ASSERT(parent && parent->db_data_pending); /* Our parent's buffer is one level closer to the dnode. */ ASSERT(db->db_level == parent->db_level-1); /* * We're about to modify our parent's db_data by modifying * our block pointer, so the parent must be released. */ ASSERT(arc_released(parent->db_buf)); zio = parent->db_data_pending->dr_zio; } else { /* Our parent is the dnode itself. */ ASSERT((db->db_level == dn->dn_phys->dn_nlevels-1 && db->db_blkid != DMU_SPILL_BLKID) || (db->db_blkid == DMU_SPILL_BLKID && db->db_level == 0)); if (db->db_blkid != DMU_SPILL_BLKID) ASSERT3P(db->db_blkptr, ==, &dn->dn_phys->dn_blkptr[db->db_blkid]); zio = dn->dn_zio; } ASSERT(db->db_level == 0 || data == db->db_buf); ASSERT3U(db->db_blkptr->blk_birth, <=, txg); ASSERT(zio); SET_BOOKMARK(&zb, os->os_dsl_dataset ? os->os_dsl_dataset->ds_object : DMU_META_OBJSET, db->db.db_object, db->db_level, db->db_blkid); if (db->db_blkid == DMU_SPILL_BLKID) wp_flag = WP_SPILL; wp_flag |= (db->db_state == DB_NOFILL) ? WP_NOFILL : 0; dmu_write_policy(os, dn, db->db_level, wp_flag, &zp); DB_DNODE_EXIT(db); /* * We copy the blkptr now (rather than when we instantiate the dirty * record), because its value can change between open context and * syncing context. We do not need to hold dn_struct_rwlock to read * db_blkptr because we are in syncing context. */ dr->dr_bp_copy = *db->db_blkptr; if (db->db_level == 0 && dr->dt.dl.dr_override_state == DR_OVERRIDDEN) { /* * The BP for this block has been provided by open context * (by dmu_sync() or dmu_buf_write_embedded()). */ abd_t *contents = (data != NULL) ? abd_get_from_buf(data->b_data, arc_buf_size(data)) : NULL; dr->dr_zio = zio_write(zio, os->os_spa, txg, &dr->dr_bp_copy, contents, db->db.db_size, db->db.db_size, &zp, dbuf_write_override_ready, NULL, NULL, dbuf_write_override_done, dr, ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_MUSTSUCCEED, &zb); mutex_enter(&db->db_mtx); dr->dt.dl.dr_override_state = DR_NOT_OVERRIDDEN; zio_write_override(dr->dr_zio, &dr->dt.dl.dr_overridden_by, dr->dt.dl.dr_copies, dr->dt.dl.dr_nopwrite); mutex_exit(&db->db_mtx); } else if (db->db_state == DB_NOFILL) { ASSERT(zp.zp_checksum == ZIO_CHECKSUM_OFF || zp.zp_checksum == ZIO_CHECKSUM_NOPARITY); dr->dr_zio = zio_write(zio, os->os_spa, txg, &dr->dr_bp_copy, NULL, db->db.db_size, db->db.db_size, &zp, dbuf_write_nofill_ready, NULL, NULL, dbuf_write_nofill_done, db, ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_MUSTSUCCEED | ZIO_FLAG_NODATA, &zb); } else { ASSERT(arc_released(data)); /* * For indirect blocks, we want to setup the children * ready callback so that we can properly handle an indirect * block that only contains holes. */ arc_write_done_func_t *children_ready_cb = NULL; if (db->db_level != 0) children_ready_cb = dbuf_write_children_ready; dr->dr_zio = arc_write(zio, os->os_spa, txg, &dr->dr_bp_copy, data, DBUF_IS_L2CACHEABLE(db), &zp, dbuf_write_ready, children_ready_cb, dbuf_write_physdone, dbuf_write_done, db, ZIO_PRIORITY_ASYNC_WRITE, ZIO_FLAG_MUSTSUCCEED, &zb); } } Index: projects/clang1100-import/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dmu.h =================================================================== --- projects/clang1100-import/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dmu.h (revision 364034) +++ projects/clang1100-import/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dmu.h (revision 364035) @@ -1,1028 +1,1028 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2018 by Delphix. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2012, Joyent, Inc. All rights reserved. * Copyright 2013 DEY Storage Systems, Inc. * Copyright 2014 HybridCluster. All rights reserved. * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved. * Copyright 2013 Saso Kiselkov. All rights reserved. * Copyright (c) 2017, Intel Corporation. * Copyright (c) 2014 Integros [integros.com] */ /* Portions Copyright 2010 Robert Milkowski */ #ifndef _SYS_DMU_H #define _SYS_DMU_H /* * This file describes the interface that the DMU provides for its * consumers. * * The DMU also interacts with the SPA. That interface is described in * dmu_spa.h. */ #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif struct uio; struct xuio; struct page; struct vnode; struct spa; struct zilog; struct zio; struct blkptr; struct zap_cursor; struct dsl_dataset; struct dsl_pool; struct dnode; struct drr_begin; struct drr_end; struct zbookmark_phys; struct spa; struct nvlist; struct arc_buf; struct zio_prop; struct sa_handle; struct file; struct locked_range; typedef struct objset objset_t; typedef struct dmu_tx dmu_tx_t; typedef struct dsl_dir dsl_dir_t; typedef struct dnode dnode_t; typedef enum dmu_object_byteswap { DMU_BSWAP_UINT8, DMU_BSWAP_UINT16, DMU_BSWAP_UINT32, DMU_BSWAP_UINT64, DMU_BSWAP_ZAP, DMU_BSWAP_DNODE, DMU_BSWAP_OBJSET, DMU_BSWAP_ZNODE, DMU_BSWAP_OLDACL, DMU_BSWAP_ACL, /* * Allocating a new byteswap type number makes the on-disk format * incompatible with any other format that uses the same number. * * Data can usually be structured to work with one of the * DMU_BSWAP_UINT* or DMU_BSWAP_ZAP types. */ DMU_BSWAP_NUMFUNCS } dmu_object_byteswap_t; #define DMU_OT_NEWTYPE 0x80 #define DMU_OT_METADATA 0x40 #define DMU_OT_BYTESWAP_MASK 0x3f /* * Defines a uint8_t object type. Object types specify if the data * in the object is metadata (boolean) and how to byteswap the data * (dmu_object_byteswap_t). All of the types created by this method * are cached in the dbuf metadata cache. */ #define DMU_OT(byteswap, metadata) \ (DMU_OT_NEWTYPE | \ ((metadata) ? DMU_OT_METADATA : 0) | \ ((byteswap) & DMU_OT_BYTESWAP_MASK)) #define DMU_OT_IS_VALID(ot) (((ot) & DMU_OT_NEWTYPE) ? \ ((ot) & DMU_OT_BYTESWAP_MASK) < DMU_BSWAP_NUMFUNCS : \ (ot) < DMU_OT_NUMTYPES) #define DMU_OT_IS_METADATA(ot) (((ot) & DMU_OT_NEWTYPE) ? \ ((ot) & DMU_OT_METADATA) : \ dmu_ot[(ot)].ot_metadata) #define DMU_OT_IS_DDT(ot) \ ((ot) == DMU_OT_DDT_ZAP) #define DMU_OT_IS_ZIL(ot) \ ((ot) == DMU_OT_INTENT_LOG) /* Note: ztest uses DMU_OT_UINT64_OTHER as a proxy for file blocks */ #define DMU_OT_IS_FILE(ot) \ ((ot) == DMU_OT_PLAIN_FILE_CONTENTS || (ot) == DMU_OT_UINT64_OTHER) #define DMU_OT_IS_METADATA_CACHED(ot) (((ot) & DMU_OT_NEWTYPE) ? \ B_TRUE : dmu_ot[(ot)].ot_dbuf_metadata_cache) /* * These object types use bp_fill != 1 for their L0 bp's. Therefore they can't * have their data embedded (i.e. use a BP_IS_EMBEDDED() bp), because bp_fill * is repurposed for embedded BPs. */ #define DMU_OT_HAS_FILL(ot) \ ((ot) == DMU_OT_DNODE || (ot) == DMU_OT_OBJSET) #define DMU_OT_BYTESWAP(ot) (((ot) & DMU_OT_NEWTYPE) ? \ ((ot) & DMU_OT_BYTESWAP_MASK) : \ dmu_ot[(ot)].ot_byteswap) typedef enum dmu_object_type { DMU_OT_NONE, /* general: */ DMU_OT_OBJECT_DIRECTORY, /* ZAP */ DMU_OT_OBJECT_ARRAY, /* UINT64 */ DMU_OT_PACKED_NVLIST, /* UINT8 (XDR by nvlist_pack/unpack) */ DMU_OT_PACKED_NVLIST_SIZE, /* UINT64 */ DMU_OT_BPOBJ, /* UINT64 */ DMU_OT_BPOBJ_HDR, /* UINT64 */ /* spa: */ DMU_OT_SPACE_MAP_HEADER, /* UINT64 */ DMU_OT_SPACE_MAP, /* UINT64 */ /* zil: */ DMU_OT_INTENT_LOG, /* UINT64 */ /* dmu: */ DMU_OT_DNODE, /* DNODE */ DMU_OT_OBJSET, /* OBJSET */ /* dsl: */ DMU_OT_DSL_DIR, /* UINT64 */ DMU_OT_DSL_DIR_CHILD_MAP, /* ZAP */ DMU_OT_DSL_DS_SNAP_MAP, /* ZAP */ DMU_OT_DSL_PROPS, /* ZAP */ DMU_OT_DSL_DATASET, /* UINT64 */ /* zpl: */ DMU_OT_ZNODE, /* ZNODE */ DMU_OT_OLDACL, /* Old ACL */ DMU_OT_PLAIN_FILE_CONTENTS, /* UINT8 */ DMU_OT_DIRECTORY_CONTENTS, /* ZAP */ DMU_OT_MASTER_NODE, /* ZAP */ DMU_OT_UNLINKED_SET, /* ZAP */ /* zvol: */ DMU_OT_ZVOL, /* UINT8 */ DMU_OT_ZVOL_PROP, /* ZAP */ /* other; for testing only! */ DMU_OT_PLAIN_OTHER, /* UINT8 */ DMU_OT_UINT64_OTHER, /* UINT64 */ DMU_OT_ZAP_OTHER, /* ZAP */ /* new object types: */ DMU_OT_ERROR_LOG, /* ZAP */ DMU_OT_SPA_HISTORY, /* UINT8 */ DMU_OT_SPA_HISTORY_OFFSETS, /* spa_his_phys_t */ DMU_OT_POOL_PROPS, /* ZAP */ DMU_OT_DSL_PERMS, /* ZAP */ DMU_OT_ACL, /* ACL */ DMU_OT_SYSACL, /* SYSACL */ DMU_OT_FUID, /* FUID table (Packed NVLIST UINT8) */ DMU_OT_FUID_SIZE, /* FUID table size UINT64 */ DMU_OT_NEXT_CLONES, /* ZAP */ DMU_OT_SCAN_QUEUE, /* ZAP */ DMU_OT_USERGROUP_USED, /* ZAP */ DMU_OT_USERGROUP_QUOTA, /* ZAP */ DMU_OT_USERREFS, /* ZAP */ DMU_OT_DDT_ZAP, /* ZAP */ DMU_OT_DDT_STATS, /* ZAP */ DMU_OT_SA, /* System attr */ DMU_OT_SA_MASTER_NODE, /* ZAP */ DMU_OT_SA_ATTR_REGISTRATION, /* ZAP */ DMU_OT_SA_ATTR_LAYOUTS, /* ZAP */ DMU_OT_SCAN_XLATE, /* ZAP */ DMU_OT_DEDUP, /* fake dedup BP from ddt_bp_create() */ DMU_OT_DEADLIST, /* ZAP */ DMU_OT_DEADLIST_HDR, /* UINT64 */ DMU_OT_DSL_CLONES, /* ZAP */ DMU_OT_BPOBJ_SUBOBJ, /* UINT64 */ /* * Do not allocate new object types here. Doing so makes the on-disk * format incompatible with any other format that uses the same object * type number. * * When creating an object which does not have one of the above types * use the DMU_OTN_* type with the correct byteswap and metadata * values. * * The DMU_OTN_* types do not have entries in the dmu_ot table, * use the DMU_OT_IS_METDATA() and DMU_OT_BYTESWAP() macros instead * use the DMU_OT_IS_METADATA() and DMU_OT_BYTESWAP() macros instead * of indexing into dmu_ot directly (this works for both DMU_OT_* types * and DMU_OTN_* types). */ DMU_OT_NUMTYPES, /* * Names for valid types declared with DMU_OT(). */ DMU_OTN_UINT8_DATA = DMU_OT(DMU_BSWAP_UINT8, B_FALSE), DMU_OTN_UINT8_METADATA = DMU_OT(DMU_BSWAP_UINT8, B_TRUE), DMU_OTN_UINT16_DATA = DMU_OT(DMU_BSWAP_UINT16, B_FALSE), DMU_OTN_UINT16_METADATA = DMU_OT(DMU_BSWAP_UINT16, B_TRUE), DMU_OTN_UINT32_DATA = DMU_OT(DMU_BSWAP_UINT32, B_FALSE), DMU_OTN_UINT32_METADATA = DMU_OT(DMU_BSWAP_UINT32, B_TRUE), DMU_OTN_UINT64_DATA = DMU_OT(DMU_BSWAP_UINT64, B_FALSE), DMU_OTN_UINT64_METADATA = DMU_OT(DMU_BSWAP_UINT64, B_TRUE), DMU_OTN_ZAP_DATA = DMU_OT(DMU_BSWAP_ZAP, B_FALSE), DMU_OTN_ZAP_METADATA = DMU_OT(DMU_BSWAP_ZAP, B_TRUE), } dmu_object_type_t; /* * These flags are intended to be used to specify the "txg_how" * parameter when calling the dmu_tx_assign() function. See the comment * above dmu_tx_assign() for more details on the meaning of these flags. */ #define TXG_NOWAIT (0ULL) #define TXG_WAIT (1ULL<<0) #define TXG_NOTHROTTLE (1ULL<<1) void byteswap_uint64_array(void *buf, size_t size); void byteswap_uint32_array(void *buf, size_t size); void byteswap_uint16_array(void *buf, size_t size); void byteswap_uint8_array(void *buf, size_t size); void zap_byteswap(void *buf, size_t size); void zfs_oldacl_byteswap(void *buf, size_t size); void zfs_acl_byteswap(void *buf, size_t size); void zfs_znode_byteswap(void *buf, size_t size); #define DS_FIND_SNAPSHOTS (1<<0) #define DS_FIND_CHILDREN (1<<1) #define DS_FIND_SERIALIZE (1<<2) /* * The maximum number of bytes that can be accessed as part of one * operation, including metadata. */ #define DMU_MAX_ACCESS (32 * 1024 * 1024) /* 32MB */ #define DMU_MAX_DELETEBLKCNT (20480) /* ~5MB of indirect blocks */ #define DMU_USERUSED_OBJECT (-1ULL) #define DMU_GROUPUSED_OBJECT (-2ULL) /* * artificial blkids for bonus buffer and spill blocks */ #define DMU_BONUS_BLKID (-1ULL) #define DMU_SPILL_BLKID (-2ULL) /* * Public routines to create, destroy, open, and close objsets. */ int dmu_objset_hold(const char *name, void *tag, objset_t **osp); int dmu_objset_own(const char *name, dmu_objset_type_t type, boolean_t readonly, void *tag, objset_t **osp); void dmu_objset_rele(objset_t *os, void *tag); void dmu_objset_disown(objset_t *os, void *tag); int dmu_objset_open_ds(struct dsl_dataset *ds, objset_t **osp); void dmu_objset_evict_dbufs(objset_t *os); int dmu_objset_create(const char *name, dmu_objset_type_t type, uint64_t flags, void (*func)(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx), void *arg); int dmu_get_recursive_snaps_nvl(char *fsname, const char *snapname, struct nvlist *snaps); int dmu_objset_clone(const char *name, const char *origin); int dsl_destroy_snapshots_nvl(struct nvlist *snaps, boolean_t defer, struct nvlist *errlist); int dmu_objset_snapshot_one(const char *fsname, const char *snapname); int dmu_objset_snapshot_tmp(const char *, const char *, int); int dmu_objset_find(char *name, int func(const char *, void *), void *arg, int flags); void dmu_objset_byteswap(void *buf, size_t size); int dsl_dataset_rename_snapshot(const char *fsname, const char *oldsnapname, const char *newsnapname, boolean_t recursive); int dmu_objset_remap_indirects(const char *fsname); typedef struct dmu_buf { uint64_t db_object; /* object that this buffer is part of */ uint64_t db_offset; /* byte offset in this object */ uint64_t db_size; /* size of buffer in bytes */ void *db_data; /* data in buffer */ } dmu_buf_t; /* * The names of zap entries in the DIRECTORY_OBJECT of the MOS. */ #define DMU_POOL_DIRECTORY_OBJECT 1 #define DMU_POOL_CONFIG "config" #define DMU_POOL_FEATURES_FOR_WRITE "features_for_write" #define DMU_POOL_FEATURES_FOR_READ "features_for_read" #define DMU_POOL_FEATURE_DESCRIPTIONS "feature_descriptions" #define DMU_POOL_FEATURE_ENABLED_TXG "feature_enabled_txg" #define DMU_POOL_ROOT_DATASET "root_dataset" #define DMU_POOL_SYNC_BPOBJ "sync_bplist" #define DMU_POOL_ERRLOG_SCRUB "errlog_scrub" #define DMU_POOL_ERRLOG_LAST "errlog_last" #define DMU_POOL_SPARES "spares" #define DMU_POOL_DEFLATE "deflate" #define DMU_POOL_HISTORY "history" #define DMU_POOL_PROPS "pool_props" #define DMU_POOL_L2CACHE "l2cache" #define DMU_POOL_TMP_USERREFS "tmp_userrefs" #define DMU_POOL_DDT "DDT-%s-%s-%s" #define DMU_POOL_DDT_STATS "DDT-statistics" #define DMU_POOL_CREATION_VERSION "creation_version" #define DMU_POOL_SCAN "scan" #define DMU_POOL_FREE_BPOBJ "free_bpobj" #define DMU_POOL_BPTREE_OBJ "bptree_obj" #define DMU_POOL_EMPTY_BPOBJ "empty_bpobj" #define DMU_POOL_CHECKSUM_SALT "org.illumos:checksum_salt" #define DMU_POOL_VDEV_ZAP_MAP "com.delphix:vdev_zap_map" #define DMU_POOL_REMOVING "com.delphix:removing" #define DMU_POOL_OBSOLETE_BPOBJ "com.delphix:obsolete_bpobj" #define DMU_POOL_CONDENSING_INDIRECT "com.delphix:condensing_indirect" #define DMU_POOL_ZPOOL_CHECKPOINT "com.delphix:zpool_checkpoint" /* * Allocate an object from this objset. The range of object numbers * available is (0, DN_MAX_OBJECT). Object 0 is the meta-dnode. * * The transaction must be assigned to a txg. The newly allocated * object will be "held" in the transaction (ie. you can modify the * newly allocated object in this transaction). * * dmu_object_alloc() chooses an object and returns it in *objectp. * * dmu_object_claim() allocates a specific object number. If that * number is already allocated, it fails and returns EEXIST. * * Return 0 on success, or ENOSPC or EEXIST as specified above. */ uint64_t dmu_object_alloc(objset_t *os, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); uint64_t dmu_object_alloc_ibs(objset_t *os, dmu_object_type_t ot, int blocksize, int indirect_blockshift, dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *tx); uint64_t dmu_object_alloc_dnsize(objset_t *os, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonus_type, int bonus_len, int dnodesize, dmu_tx_t *tx); int dmu_object_claim_dnsize(objset_t *os, uint64_t object, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonus_type, int bonus_len, int dnodesize, dmu_tx_t *tx); int dmu_object_reclaim_dnsize(objset_t *os, uint64_t object, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonustype, int bonuslen, int dnodesize, dmu_tx_t *txp); int dmu_object_claim(objset_t *os, uint64_t object, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonus_type, int bonus_len, dmu_tx_t *tx); int dmu_object_reclaim(objset_t *os, uint64_t object, dmu_object_type_t ot, int blocksize, dmu_object_type_t bonustype, int bonuslen, dmu_tx_t *txp); /* * Free an object from this objset. * * The object's data will be freed as well (ie. you don't need to call * dmu_free(object, 0, -1, tx)). * * The object need not be held in the transaction. * * If there are any holds on this object's buffers (via dmu_buf_hold()), * or tx holds on the object (via dmu_tx_hold_object()), you can not * free it; it fails and returns EBUSY. * * If the object is not allocated, it fails and returns ENOENT. * * Return 0 on success, or EBUSY or ENOENT as specified above. */ int dmu_object_free(objset_t *os, uint64_t object, dmu_tx_t *tx); /* * Find the next allocated or free object. * * The objectp parameter is in-out. It will be updated to be the next * object which is allocated. Ignore objects which have not been * modified since txg. * * XXX Can only be called on a objset with no dirty data. * * Returns 0 on success, or ENOENT if there are no more objects. */ int dmu_object_next(objset_t *os, uint64_t *objectp, boolean_t hole, uint64_t txg); /* * Set the data blocksize for an object. * * The object cannot have any blocks allcated beyond the first. If * the first block is allocated already, the new size must be greater * than the current block size. If these conditions are not met, * ENOTSUP will be returned. * * Returns 0 on success, or EBUSY if there are any holds on the object * contents, or ENOTSUP as described above. */ int dmu_object_set_blocksize(objset_t *os, uint64_t object, uint64_t size, int ibs, dmu_tx_t *tx); /* * Set the checksum property on a dnode. The new checksum algorithm will * apply to all newly written blocks; existing blocks will not be affected. */ void dmu_object_set_checksum(objset_t *os, uint64_t object, uint8_t checksum, dmu_tx_t *tx); /* * Set the compress property on a dnode. The new compression algorithm will * apply to all newly written blocks; existing blocks will not be affected. */ void dmu_object_set_compress(objset_t *os, uint64_t object, uint8_t compress, dmu_tx_t *tx); int dmu_object_remap_indirects(objset_t *os, uint64_t object, uint64_t txg); void dmu_write_embedded(objset_t *os, uint64_t object, uint64_t offset, void *data, uint8_t etype, uint8_t comp, int uncompressed_size, int compressed_size, int byteorder, dmu_tx_t *tx); /* * Decide how to write a block: checksum, compression, number of copies, etc. */ #define WP_NOFILL 0x1 #define WP_DMU_SYNC 0x2 #define WP_SPILL 0x4 void dmu_write_policy(objset_t *os, dnode_t *dn, int level, int wp, struct zio_prop *zp); /* * The bonus data is accessed more or less like a regular buffer. * You must dmu_bonus_hold() to get the buffer, which will give you a * dmu_buf_t with db_offset==-1ULL, and db_size = the size of the bonus * data. As with any normal buffer, you must call dmu_buf_will_dirty() * before modifying it, and the * object must be held in an assigned transaction before calling * dmu_buf_will_dirty. You may use dmu_buf_set_user() on the bonus * buffer as well. You must release your hold with dmu_buf_rele(). * * Returns ENOENT, EIO, or 0. */ int dmu_bonus_hold(objset_t *os, uint64_t object, void *tag, dmu_buf_t **); int dmu_bonus_max(void); int dmu_set_bonus(dmu_buf_t *, int, dmu_tx_t *); int dmu_set_bonustype(dmu_buf_t *, dmu_object_type_t, dmu_tx_t *); dmu_object_type_t dmu_get_bonustype(dmu_buf_t *); int dmu_rm_spill(objset_t *, uint64_t, dmu_tx_t *); /* * Special spill buffer support used by "SA" framework */ int dmu_spill_hold_by_bonus(dmu_buf_t *bonus, void *tag, dmu_buf_t **dbp); int dmu_spill_hold_by_dnode(dnode_t *dn, uint32_t flags, void *tag, dmu_buf_t **dbp); int dmu_spill_hold_existing(dmu_buf_t *bonus, void *tag, dmu_buf_t **dbp); /* * Obtain the DMU buffer from the specified object which contains the * specified offset. dmu_buf_hold() puts a "hold" on the buffer, so * that it will remain in memory. You must release the hold with * dmu_buf_rele(). You musn't access the dmu_buf_t after releasing your * hold. You must have a hold on any dmu_buf_t* you pass to the DMU. * * You must call dmu_buf_read, dmu_buf_will_dirty, or dmu_buf_will_fill * on the returned buffer before reading or writing the buffer's * db_data. The comments for those routines describe what particular * operations are valid after calling them. * * The object number must be a valid, allocated object number. */ int dmu_buf_hold(objset_t *os, uint64_t object, uint64_t offset, void *tag, dmu_buf_t **, int flags); int dmu_buf_hold_by_dnode(dnode_t *dn, uint64_t offset, void *tag, dmu_buf_t **dbp, int flags); /* * Add a reference to a dmu buffer that has already been held via * dmu_buf_hold() in the current context. */ void dmu_buf_add_ref(dmu_buf_t *db, void* tag); /* * Attempt to add a reference to a dmu buffer that is in an unknown state, * using a pointer that may have been invalidated by eviction processing. * The request will succeed if the passed in dbuf still represents the * same os/object/blkid, is ineligible for eviction, and has at least * one hold by a user other than the syncer. */ boolean_t dmu_buf_try_add_ref(dmu_buf_t *, objset_t *os, uint64_t object, uint64_t blkid, void *tag); void dmu_buf_rele(dmu_buf_t *db, void *tag); uint64_t dmu_buf_refcount(dmu_buf_t *db); /* * dmu_buf_hold_array holds the DMU buffers which contain all bytes in a * range of an object. A pointer to an array of dmu_buf_t*'s is * returned (in *dbpp). * * dmu_buf_rele_array releases the hold on an array of dmu_buf_t*'s, and * frees the array. The hold on the array of buffers MUST be released * with dmu_buf_rele_array. You can NOT release the hold on each buffer * individually with dmu_buf_rele. */ int dmu_buf_hold_array_by_bonus(dmu_buf_t *db, uint64_t offset, uint64_t length, boolean_t read, void *tag, int *numbufsp, dmu_buf_t ***dbpp); int dmu_buf_hold_array_by_dnode(dnode_t *dn, uint64_t offset, uint64_t length, boolean_t read, void *tag, int *numbufsp, dmu_buf_t ***dbpp, uint32_t flags); void dmu_buf_rele_array(dmu_buf_t **, int numbufs, void *tag); typedef void dmu_buf_evict_func_t(void *user_ptr); /* * A DMU buffer user object may be associated with a dbuf for the * duration of its lifetime. This allows the user of a dbuf (client) * to attach private data to a dbuf (e.g. in-core only data such as a * dnode_children_t, zap_t, or zap_leaf_t) and be optionally notified * when that dbuf has been evicted. Clients typically respond to the * eviction notification by freeing their private data, thus ensuring * the same lifetime for both dbuf and private data. * * The mapping from a dmu_buf_user_t to any client private data is the * client's responsibility. All current consumers of the API with private * data embed a dmu_buf_user_t as the first member of the structure for * their private data. This allows conversions between the two types * with a simple cast. Since the DMU buf user API never needs access * to the private data, other strategies can be employed if necessary * or convenient for the client (e.g. using container_of() to do the * conversion for private data that cannot have the dmu_buf_user_t as * its first member). * * Eviction callbacks are executed without the dbuf mutex held or any * other type of mechanism to guarantee that the dbuf is still available. * For this reason, users must assume the dbuf has already been freed * and not reference the dbuf from the callback context. * * Users requesting "immediate eviction" are notified as soon as the dbuf * is only referenced by dirty records (dirties == holds). Otherwise the * notification occurs after eviction processing for the dbuf begins. */ typedef struct dmu_buf_user { /* * Asynchronous user eviction callback state. */ taskq_ent_t dbu_tqent; /* * This instance's eviction function pointers. * * dbu_evict_func_sync is called synchronously and then * dbu_evict_func_async is executed asynchronously on a taskq. */ dmu_buf_evict_func_t *dbu_evict_func_sync; dmu_buf_evict_func_t *dbu_evict_func_async; #ifdef ZFS_DEBUG /* * Pointer to user's dbuf pointer. NULL for clients that do * not associate a dbuf with their user data. * * The dbuf pointer is cleared upon eviction so as to catch * use-after-evict bugs in clients. */ dmu_buf_t **dbu_clear_on_evict_dbufp; #endif } dmu_buf_user_t; /* * Initialize the given dmu_buf_user_t instance with the eviction function * evict_func, to be called when the user is evicted. * * NOTE: This function should only be called once on a given dmu_buf_user_t. * To allow enforcement of this, dbu must already be zeroed on entry. */ /*ARGSUSED*/ -inline void +static inline void dmu_buf_init_user(dmu_buf_user_t *dbu, dmu_buf_evict_func_t *evict_func_sync, dmu_buf_evict_func_t *evict_func_async, dmu_buf_t **clear_on_evict_dbufp) { ASSERT(dbu->dbu_evict_func_sync == NULL); ASSERT(dbu->dbu_evict_func_async == NULL); /* must have at least one evict func */ IMPLY(evict_func_sync == NULL, evict_func_async != NULL); dbu->dbu_evict_func_sync = evict_func_sync; dbu->dbu_evict_func_async = evict_func_async; #ifdef ZFS_DEBUG dbu->dbu_clear_on_evict_dbufp = clear_on_evict_dbufp; #endif } /* * Attach user data to a dbuf and mark it for normal (when the dbuf's * data is cleared or its reference count goes to zero) eviction processing. * * Returns NULL on success, or the existing user if another user currently * owns the buffer. */ void *dmu_buf_set_user(dmu_buf_t *db, dmu_buf_user_t *user); /* * Attach user data to a dbuf and mark it for immediate (its dirty and * reference counts are equal) eviction processing. * * Returns NULL on success, or the existing user if another user currently * owns the buffer. */ void *dmu_buf_set_user_ie(dmu_buf_t *db, dmu_buf_user_t *user); /* * Replace the current user of a dbuf. * * If given the current user of a dbuf, replaces the dbuf's user with * "new_user" and returns the user data pointer that was replaced. * Otherwise returns the current, and unmodified, dbuf user pointer. */ void *dmu_buf_replace_user(dmu_buf_t *db, dmu_buf_user_t *old_user, dmu_buf_user_t *new_user); /* * Remove the specified user data for a DMU buffer. * * Returns the user that was removed on success, or the current user if * another user currently owns the buffer. */ void *dmu_buf_remove_user(dmu_buf_t *db, dmu_buf_user_t *user); /* * Returns the user data (dmu_buf_user_t *) associated with this dbuf. */ void *dmu_buf_get_user(dmu_buf_t *db); objset_t *dmu_buf_get_objset(dmu_buf_t *db); dnode_t *dmu_buf_dnode_enter(dmu_buf_t *db); void dmu_buf_dnode_exit(dmu_buf_t *db); /* Block until any in-progress dmu buf user evictions complete. */ void dmu_buf_user_evict_wait(void); /* * Returns the blkptr associated with this dbuf, or NULL if not set. */ struct blkptr *dmu_buf_get_blkptr(dmu_buf_t *db); /* * Indicate that you are going to modify the buffer's data (db_data). * * The transaction (tx) must be assigned to a txg (ie. you've called * dmu_tx_assign()). The buffer's object must be held in the tx * (ie. you've called dmu_tx_hold_object(tx, db->db_object)). */ void dmu_buf_will_dirty(dmu_buf_t *db, dmu_tx_t *tx); /* * You must create a transaction, then hold the objects which you will * (or might) modify as part of this transaction. Then you must assign * the transaction to a transaction group. Once the transaction has * been assigned, you can modify buffers which belong to held objects as * part of this transaction. You can't modify buffers before the * transaction has been assigned; you can't modify buffers which don't * belong to objects which this transaction holds; you can't hold * objects once the transaction has been assigned. You may hold an * object which you are going to free (with dmu_object_free()), but you * don't have to. * * You can abort the transaction before it has been assigned. * * Note that you may hold buffers (with dmu_buf_hold) at any time, * regardless of transaction state. */ #define DMU_NEW_OBJECT (-1ULL) #define DMU_OBJECT_END (-1ULL) dmu_tx_t *dmu_tx_create(objset_t *os); void dmu_tx_hold_write(dmu_tx_t *tx, uint64_t object, uint64_t off, int len); void dmu_tx_hold_write_by_dnode(dmu_tx_t *tx, dnode_t *dn, uint64_t off, int len); void dmu_tx_hold_free(dmu_tx_t *tx, uint64_t object, uint64_t off, uint64_t len); void dmu_tx_hold_free_by_dnode(dmu_tx_t *tx, dnode_t *dn, uint64_t off, uint64_t len); void dmu_tx_hold_remap_l1indirect(dmu_tx_t *tx, uint64_t object); void dmu_tx_hold_zap(dmu_tx_t *tx, uint64_t object, int add, const char *name); void dmu_tx_hold_zap_by_dnode(dmu_tx_t *tx, dnode_t *dn, int add, const char *name); void dmu_tx_hold_bonus(dmu_tx_t *tx, uint64_t object); void dmu_tx_hold_bonus_by_dnode(dmu_tx_t *tx, dnode_t *dn); void dmu_tx_hold_spill(dmu_tx_t *tx, uint64_t object); void dmu_tx_hold_sa(dmu_tx_t *tx, struct sa_handle *hdl, boolean_t may_grow); void dmu_tx_hold_sa_create(dmu_tx_t *tx, int total_size); void dmu_tx_abort(dmu_tx_t *tx); int dmu_tx_assign(dmu_tx_t *tx, uint64_t txg_how); void dmu_tx_wait(dmu_tx_t *tx); void dmu_tx_commit(dmu_tx_t *tx); void dmu_tx_mark_netfree(dmu_tx_t *tx); /* * To register a commit callback, dmu_tx_callback_register() must be called. * * dcb_data is a pointer to caller private data that is passed on as a * callback parameter. The caller is responsible for properly allocating and * freeing it. * * When registering a callback, the transaction must be already created, but * it cannot be committed or aborted. It can be assigned to a txg or not. * * The callback will be called after the transaction has been safely written * to stable storage and will also be called if the dmu_tx is aborted. * If there is any error which prevents the transaction from being committed to * disk, the callback will be called with a value of error != 0. */ typedef void dmu_tx_callback_func_t(void *dcb_data, int error); void dmu_tx_callback_register(dmu_tx_t *tx, dmu_tx_callback_func_t *dcb_func, void *dcb_data); /* * Free up the data blocks for a defined range of a file. If size is * -1, the range from offset to end-of-file is freed. */ int dmu_free_range(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, dmu_tx_t *tx); int dmu_free_long_range(objset_t *os, uint64_t object, uint64_t offset, uint64_t size); int dmu_free_long_object(objset_t *os, uint64_t object); /* * Convenience functions. * * Canfail routines will return 0 on success, or an errno if there is a * nonrecoverable I/O error. */ #define DMU_READ_PREFETCH 0 /* prefetch */ #define DMU_READ_NO_PREFETCH 1 /* don't prefetch */ int dmu_read(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, void *buf, uint32_t flags); int dmu_read_by_dnode(dnode_t *dn, uint64_t offset, uint64_t size, void *buf, uint32_t flags); void dmu_write(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, const void *buf, dmu_tx_t *tx); void dmu_write_by_dnode(dnode_t *dn, uint64_t offset, uint64_t size, const void *buf, dmu_tx_t *tx); void dmu_prealloc(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, dmu_tx_t *tx); int dmu_read_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size); int dmu_read_uio_dbuf(dmu_buf_t *zdb, struct uio *uio, uint64_t size); int dmu_read_uio_dnode(dnode_t *dn, struct uio *uio, uint64_t size); int dmu_write_uio(objset_t *os, uint64_t object, struct uio *uio, uint64_t size, dmu_tx_t *tx); int dmu_write_uio_dbuf(dmu_buf_t *zdb, struct uio *uio, uint64_t size, dmu_tx_t *tx); int dmu_write_uio_dnode(dnode_t *dn, struct uio *uio, uint64_t size, dmu_tx_t *tx); #ifdef _KERNEL #ifdef illumos int dmu_write_pages(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, struct page *pp, dmu_tx_t *tx); #else int dmu_write_pages(objset_t *os, uint64_t object, uint64_t offset, uint64_t size, struct vm_page **ppa, dmu_tx_t *tx); int dmu_read_pages(objset_t *os, uint64_t object, vm_page_t *ma, int count, int *rbehind, int *rahead, int last_size); #endif #endif struct arc_buf *dmu_request_arcbuf(dmu_buf_t *handle, int size); void dmu_return_arcbuf(struct arc_buf *buf); void dmu_assign_arcbuf_dnode(dnode_t *handle, uint64_t offset, struct arc_buf *buf, dmu_tx_t *tx); void dmu_assign_arcbuf(dmu_buf_t *handle, uint64_t offset, struct arc_buf *buf, dmu_tx_t *tx); int dmu_xuio_init(struct xuio *uio, int niov); void dmu_xuio_fini(struct xuio *uio); int dmu_xuio_add(struct xuio *uio, struct arc_buf *abuf, offset_t off, size_t n); int dmu_xuio_cnt(struct xuio *uio); struct arc_buf *dmu_xuio_arcbuf(struct xuio *uio, int i); void dmu_xuio_clear(struct xuio *uio, int i); void xuio_stat_wbuf_copied(void); void xuio_stat_wbuf_nocopy(void); extern boolean_t zfs_prefetch_disable; extern int zfs_max_recordsize; /* * Asynchronously try to read in the data. */ void dmu_prefetch(objset_t *os, uint64_t object, int64_t level, uint64_t offset, uint64_t len, enum zio_priority pri); typedef struct dmu_object_info { /* All sizes are in bytes unless otherwise indicated. */ uint32_t doi_data_block_size; uint32_t doi_metadata_block_size; dmu_object_type_t doi_type; dmu_object_type_t doi_bonus_type; uint64_t doi_bonus_size; uint8_t doi_indirection; /* 2 = dnode->indirect->data */ uint8_t doi_checksum; uint8_t doi_compress; uint8_t doi_nblkptr; int8_t doi_pad[4]; uint64_t doi_dnodesize; uint64_t doi_physical_blocks_512; /* data + metadata, 512b blks */ uint64_t doi_max_offset; uint64_t doi_fill_count; /* number of non-empty blocks */ } dmu_object_info_t; typedef void arc_byteswap_func_t(void *buf, size_t size); typedef struct dmu_object_type_info { dmu_object_byteswap_t ot_byteswap; boolean_t ot_metadata; boolean_t ot_dbuf_metadata_cache; char *ot_name; } dmu_object_type_info_t; typedef struct dmu_object_byteswap_info { arc_byteswap_func_t *ob_func; char *ob_name; } dmu_object_byteswap_info_t; extern const dmu_object_type_info_t dmu_ot[DMU_OT_NUMTYPES]; extern const dmu_object_byteswap_info_t dmu_ot_byteswap[DMU_BSWAP_NUMFUNCS]; /* * Get information on a DMU object. * * Return 0 on success or ENOENT if object is not allocated. * * If doi is NULL, just indicates whether the object exists. */ int dmu_object_info(objset_t *os, uint64_t object, dmu_object_info_t *doi); void __dmu_object_info_from_dnode(struct dnode *dn, dmu_object_info_t *doi); /* Like dmu_object_info, but faster if you have a held dnode in hand. */ void dmu_object_info_from_dnode(dnode_t *dn, dmu_object_info_t *doi); /* Like dmu_object_info, but faster if you have a held dbuf in hand. */ void dmu_object_info_from_db(dmu_buf_t *db, dmu_object_info_t *doi); /* * Like dmu_object_info_from_db, but faster still when you only care about * the size. This is specifically optimized for zfs_getattr(). */ void dmu_object_size_from_db(dmu_buf_t *db, uint32_t *blksize, u_longlong_t *nblk512); void dmu_object_dnsize_from_db(dmu_buf_t *db, int *dnsize); typedef struct dmu_objset_stats { uint64_t dds_num_clones; /* number of clones of this */ uint64_t dds_creation_txg; uint64_t dds_guid; dmu_objset_type_t dds_type; uint8_t dds_is_snapshot; uint8_t dds_inconsistent; char dds_origin[ZFS_MAX_DATASET_NAME_LEN]; } dmu_objset_stats_t; /* * Get stats on a dataset. */ void dmu_objset_fast_stat(objset_t *os, dmu_objset_stats_t *stat); /* * Add entries to the nvlist for all the objset's properties. See * zfs_prop_table[] and zfs(1m) for details on the properties. */ void dmu_objset_stats(objset_t *os, struct nvlist *nv); /* * Get the space usage statistics for statvfs(). * * refdbytes is the amount of space "referenced" by this objset. * availbytes is the amount of space available to this objset, taking * into account quotas & reservations, assuming that no other objsets * use the space first. These values correspond to the 'referenced' and * 'available' properties, described in the zfs(1m) manpage. * * usedobjs and availobjs are the number of objects currently allocated, * and available. */ void dmu_objset_space(objset_t *os, uint64_t *refdbytesp, uint64_t *availbytesp, uint64_t *usedobjsp, uint64_t *availobjsp); /* * The fsid_guid is a 56-bit ID that can change to avoid collisions. * (Contrast with the ds_guid which is a 64-bit ID that will never * change, so there is a small probability that it will collide.) */ uint64_t dmu_objset_fsid_guid(objset_t *os); /* * Get the [cm]time for an objset's snapshot dir */ timestruc_t dmu_objset_snap_cmtime(objset_t *os); int dmu_objset_is_snapshot(objset_t *os); extern struct spa *dmu_objset_spa(objset_t *os); extern struct zilog *dmu_objset_zil(objset_t *os); extern struct dsl_pool *dmu_objset_pool(objset_t *os); extern struct dsl_dataset *dmu_objset_ds(objset_t *os); extern void dmu_objset_name(objset_t *os, char *buf); extern dmu_objset_type_t dmu_objset_type(objset_t *os); extern uint64_t dmu_objset_id(objset_t *os); extern uint64_t dmu_objset_dnodesize(objset_t *os); extern zfs_sync_type_t dmu_objset_syncprop(objset_t *os); extern zfs_logbias_op_t dmu_objset_logbias(objset_t *os); extern int dmu_snapshot_list_next(objset_t *os, int namelen, char *name, uint64_t *id, uint64_t *offp, boolean_t *case_conflict); extern int dmu_snapshot_realname(objset_t *os, char *name, char *real, int maxlen, boolean_t *conflict); extern int dmu_dir_list_next(objset_t *os, int namelen, char *name, uint64_t *idp, uint64_t *offp); typedef int objset_used_cb_t(dmu_object_type_t bonustype, void *bonus, uint64_t *userp, uint64_t *groupp); extern void dmu_objset_register_type(dmu_objset_type_t ost, objset_used_cb_t *cb); extern void dmu_objset_set_user(objset_t *os, void *user_ptr); extern void *dmu_objset_get_user(objset_t *os); /* * Return the txg number for the given assigned transaction. */ uint64_t dmu_tx_get_txg(dmu_tx_t *tx); /* * Synchronous write. * If a parent zio is provided this function initiates a write on the * provided buffer as a child of the parent zio. * In the absence of a parent zio, the write is completed synchronously. * At write completion, blk is filled with the bp of the written block. * Note that while the data covered by this function will be on stable * storage when the write completes this new data does not become a * permanent part of the file until the associated transaction commits. */ /* * {zfs,zvol,ztest}_get_done() args */ typedef struct zgd { struct lwb *zgd_lwb; struct blkptr *zgd_bp; dmu_buf_t *zgd_db; struct locked_range *zgd_lr; void *zgd_private; } zgd_t; typedef void dmu_sync_cb_t(zgd_t *arg, int error); int dmu_sync(struct zio *zio, uint64_t txg, dmu_sync_cb_t *done, zgd_t *zgd); /* * Find the next hole or data block in file starting at *off * Return found offset in *off. Return ESRCH for end of file. */ int dmu_offset_next(objset_t *os, uint64_t object, boolean_t hole, uint64_t *off); /* * Check if a DMU object has any dirty blocks. If so, sync out * all pending transaction groups. Otherwise, this function * does not alter DMU state. This could be improved to only sync * out the necessary transaction groups for this particular * object. */ int dmu_object_wait_synced(objset_t *os, uint64_t object); /* * Initial setup and final teardown. */ extern void dmu_init(void); extern void dmu_fini(void); typedef void (*dmu_traverse_cb_t)(objset_t *os, void *arg, struct blkptr *bp, uint64_t object, uint64_t offset, int len); void dmu_traverse_objset(objset_t *os, uint64_t txg_start, dmu_traverse_cb_t cb, void *arg); int dmu_diff(const char *tosnap_name, const char *fromsnap_name, struct file *fp, offset_t *offp); /* CRC64 table */ #define ZFS_CRC64_POLY 0xC96C5795D7870F42ULL /* ECMA-182, reflected form */ extern uint64_t zfs_crc64_table[256]; extern int zfs_mdcomp_disable; #ifdef __cplusplus } #endif #endif /* _SYS_DMU_H */ Index: projects/clang1100-import/sys/cddl/contrib/opensolaris/uts/common/sys/sysmacros.h =================================================================== --- projects/clang1100-import/sys/cddl/contrib/opensolaris/uts/common/sys/sysmacros.h (revision 364034) +++ projects/clang1100-import/sys/cddl/contrib/opensolaris/uts/common/sys/sysmacros.h (revision 364035) @@ -1,459 +1,466 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* Copyright (c) 1984, 1986, 1987, 1988, 1989 AT&T */ /* All Rights Reserved */ /* * Copyright 2008 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #ifndef _SYS_SYSMACROS_H #define _SYS_SYSMACROS_H #include #include #if defined(__FreeBSD__) && defined(_KERNEL) #include #endif #ifdef __cplusplus extern "C" { #endif /* * Some macros for units conversion */ /* * Disk blocks (sectors) and bytes. */ +#ifndef dtob #define dtob(DD) ((DD) << DEV_BSHIFT) +#endif +#ifndef btod #define btod(BB) (((BB) + DEV_BSIZE - 1) >> DEV_BSHIFT) +#endif #define btodt(BB) ((BB) >> DEV_BSHIFT) #define lbtod(BB) (((offset_t)(BB) + DEV_BSIZE - 1) >> DEV_BSHIFT) /* common macros */ #ifndef MIN #define MIN(a, b) ((a) < (b) ? (a) : (b)) #endif #ifndef MAX #define MAX(a, b) ((a) < (b) ? (b) : (a)) #endif #ifndef ABS #define ABS(a) ((a) < 0 ? -(a) : (a)) #endif #ifndef SIGNOF #define SIGNOF(a) ((a) < 0 ? -1 : (a) > 0) #endif #ifdef _KERNEL /* * Convert a single byte to/from binary-coded decimal (BCD). */ extern unsigned char byte_to_bcd[256]; extern unsigned char bcd_to_byte[256]; #define BYTE_TO_BCD(x) byte_to_bcd[(x) & 0xff] #define BCD_TO_BYTE(x) bcd_to_byte[(x) & 0xff] #endif /* _KERNEL */ /* * WARNING: The device number macros defined here should not be used by device * drivers or user software. Device drivers should use the device functions * defined in the DDI/DKI interface (see also ddi.h). Application software * should make use of the library routines available in makedev(3). A set of * new device macros are provided to operate on the expanded device number * format supported in SVR4. Macro versions of the DDI device functions are * provided for use by kernel proper routines only. Macro routines bmajor(), * major(), minor(), emajor(), eminor(), and makedev() will be removed or * their definitions changed at the next major release following SVR4. */ #define O_BITSMAJOR 7 /* # of SVR3 major device bits */ #define O_BITSMINOR 8 /* # of SVR3 minor device bits */ #define O_MAXMAJ 0x7f /* SVR3 max major value */ #define O_MAXMIN 0xff /* SVR3 max minor value */ #define L_BITSMAJOR32 14 /* # of SVR4 major device bits */ #define L_BITSMINOR32 18 /* # of SVR4 minor device bits */ #define L_MAXMAJ32 0x3fff /* SVR4 max major value */ #define L_MAXMIN32 0x3ffff /* MAX minor for 3b2 software drivers. */ /* For 3b2 hardware devices the minor is */ /* restricted to 256 (0-255) */ #ifdef _LP64 #define L_BITSMAJOR 32 /* # of major device bits in 64-bit Solaris */ #define L_BITSMINOR 32 /* # of minor device bits in 64-bit Solaris */ #define L_MAXMAJ 0xfffffffful /* max major value */ #define L_MAXMIN 0xfffffffful /* max minor value */ #else #define L_BITSMAJOR L_BITSMAJOR32 #define L_BITSMINOR L_BITSMINOR32 #define L_MAXMAJ L_MAXMAJ32 #define L_MAXMIN L_MAXMIN32 #endif #ifdef illumos #ifdef _KERNEL /* major part of a device internal to the kernel */ #define major(x) (major_t)((((unsigned)(x)) >> O_BITSMINOR) & O_MAXMAJ) #define bmajor(x) (major_t)((((unsigned)(x)) >> O_BITSMINOR) & O_MAXMAJ) /* get internal major part of expanded device number */ #define getmajor(x) (major_t)((((dev_t)(x)) >> L_BITSMINOR) & L_MAXMAJ) /* minor part of a device internal to the kernel */ #define minor(x) (minor_t)((x) & O_MAXMIN) /* get internal minor part of expanded device number */ #define getminor(x) (minor_t)((x) & L_MAXMIN) #else /* major part of a device external from the kernel (same as emajor below) */ #define major(x) (major_t)((((unsigned)(x)) >> O_BITSMINOR) & O_MAXMAJ) /* minor part of a device external from the kernel (same as eminor below) */ #define minor(x) (minor_t)((x) & O_MAXMIN) #endif /* _KERNEL */ /* create old device number */ #define makedev(x, y) (unsigned short)(((x) << O_BITSMINOR) | ((y) & O_MAXMIN)) /* make an new device number */ #define makedevice(x, y) (dev_t)(((dev_t)(x) << L_BITSMINOR) | ((y) & L_MAXMIN)) /* * emajor() allows kernel/driver code to print external major numbers * eminor() allows kernel/driver code to print external minor numbers */ #define emajor(x) \ (major_t)(((unsigned int)(x) >> O_BITSMINOR) > O_MAXMAJ) ? \ NODEV : (((unsigned int)(x) >> O_BITSMINOR) & O_MAXMAJ) #define eminor(x) \ (minor_t)((x) & O_MAXMIN) /* * get external major and minor device * components from expanded device number */ #define getemajor(x) (major_t)((((dev_t)(x) >> L_BITSMINOR) > L_MAXMAJ) ? \ NODEV : (((dev_t)(x) >> L_BITSMINOR) & L_MAXMAJ)) #define geteminor(x) (minor_t)((x) & L_MAXMIN) #endif /* illumos */ /* * These are versions of the kernel routines for compressing and * expanding long device numbers that don't return errors. */ #if (L_BITSMAJOR32 == L_BITSMAJOR) && (L_BITSMINOR32 == L_BITSMINOR) #define DEVCMPL(x) (x) #define DEVEXPL(x) (x) #else #define DEVCMPL(x) \ (dev32_t)((((x) >> L_BITSMINOR) > L_MAXMAJ32 || \ ((x) & L_MAXMIN) > L_MAXMIN32) ? NODEV32 : \ ((((x) >> L_BITSMINOR) << L_BITSMINOR32) | ((x) & L_MAXMIN32))) #define DEVEXPL(x) \ (((x) == NODEV32) ? NODEV : \ makedevice(((x) >> L_BITSMINOR32) & L_MAXMAJ32, (x) & L_MAXMIN32)) #endif /* L_BITSMAJOR32 ... */ /* convert to old (SVR3.2) dev format */ #define cmpdev(x) \ (o_dev_t)((((x) >> L_BITSMINOR) > O_MAXMAJ || \ ((x) & L_MAXMIN) > O_MAXMIN) ? NODEV : \ ((((x) >> L_BITSMINOR) << O_BITSMINOR) | ((x) & O_MAXMIN))) /* convert to new (SVR4) dev format */ #define expdev(x) \ (dev_t)(((dev_t)(((x) >> O_BITSMINOR) & O_MAXMAJ) << L_BITSMINOR) | \ ((x) & O_MAXMIN)) /* * Macro for checking power of 2 address alignment. */ #define IS_P2ALIGNED(v, a) ((((uintptr_t)(v)) & ((uintptr_t)(a) - 1)) == 0) /* * Macros for counting and rounding. */ +#ifndef howmany #define howmany(x, y) (((x)+((y)-1))/(y)) +#endif +#ifndef roundup #define roundup(x, y) ((((x)+((y)-1))/(y))*(y)) - +#endif /* * Macro to determine if value is a power of 2 */ #define ISP2(x) (((x) & ((x) - 1)) == 0) /* * Macros for various sorts of alignment and rounding. The "align" must * be a power of 2. Often times it is a block, sector, or page. */ /* * return x rounded down to an align boundary * eg, P2ALIGN(1200, 1024) == 1024 (1*align) * eg, P2ALIGN(1024, 1024) == 1024 (1*align) * eg, P2ALIGN(0x1234, 0x100) == 0x1200 (0x12*align) * eg, P2ALIGN(0x5600, 0x100) == 0x5600 (0x56*align) */ #define P2ALIGN(x, align) ((x) & -(align)) /* * return x % (mod) align * eg, P2PHASE(0x1234, 0x100) == 0x34 (x-0x12*align) * eg, P2PHASE(0x5600, 0x100) == 0x00 (x-0x56*align) */ #define P2PHASE(x, align) ((x) & ((align) - 1)) /* * return how much space is left in this block (but if it's perfectly * aligned, return 0). * eg, P2NPHASE(0x1234, 0x100) == 0xcc (0x13*align-x) * eg, P2NPHASE(0x5600, 0x100) == 0x00 (0x56*align-x) */ #define P2NPHASE(x, align) (-(x) & ((align) - 1)) /* * return x rounded up to an align boundary * eg, P2ROUNDUP(0x1234, 0x100) == 0x1300 (0x13*align) * eg, P2ROUNDUP(0x5600, 0x100) == 0x5600 (0x56*align) */ #define P2ROUNDUP(x, align) (-(-(x) & -(align))) /* * return the ending address of the block that x is in * eg, P2END(0x1234, 0x100) == 0x12ff (0x13*align - 1) * eg, P2END(0x5600, 0x100) == 0x56ff (0x57*align - 1) */ #define P2END(x, align) (-(~(x) & -(align))) /* * return x rounded up to the next phase (offset) within align. * phase should be < align. * eg, P2PHASEUP(0x1234, 0x100, 0x10) == 0x1310 (0x13*align + phase) * eg, P2PHASEUP(0x5600, 0x100, 0x10) == 0x5610 (0x56*align + phase) */ #define P2PHASEUP(x, align, phase) ((phase) - (((phase) - (x)) & -(align))) /* * return TRUE if adding len to off would cause it to cross an align * boundary. * eg, P2BOUNDARY(0x1234, 0xe0, 0x100) == TRUE (0x1234 + 0xe0 == 0x1314) * eg, P2BOUNDARY(0x1234, 0x50, 0x100) == FALSE (0x1234 + 0x50 == 0x1284) */ #define P2BOUNDARY(off, len, align) \ (((off) ^ ((off) + (len) - 1)) > (align) - 1) /* * Return TRUE if they have the same highest bit set. * eg, P2SAMEHIGHBIT(0x1234, 0x1001) == TRUE (the high bit is 0x1000) * eg, P2SAMEHIGHBIT(0x1234, 0x3010) == FALSE (high bit of 0x3010 is 0x2000) */ #define P2SAMEHIGHBIT(x, y) (((x) ^ (y)) < ((x) & (y))) /* * Typed version of the P2* macros. These macros should be used to ensure * that the result is correctly calculated based on the data type of (x), * which is passed in as the last argument, regardless of the data * type of the alignment. For example, if (x) is of type uint64_t, * and we want to round it up to a page boundary using "PAGESIZE" as * the alignment, we can do either * P2ROUNDUP(x, (uint64_t)PAGESIZE) * or * P2ROUNDUP_TYPED(x, PAGESIZE, uint64_t) */ #define P2ALIGN_TYPED(x, align, type) \ ((type)(x) & -(type)(align)) #define P2PHASE_TYPED(x, align, type) \ ((type)(x) & ((type)(align) - 1)) #define P2NPHASE_TYPED(x, align, type) \ (-(type)(x) & ((type)(align) - 1)) #define P2ROUNDUP_TYPED(x, align, type) \ (-(-(type)(x) & -(type)(align))) #define P2END_TYPED(x, align, type) \ (-(~(type)(x) & -(type)(align))) #define P2PHASEUP_TYPED(x, align, phase, type) \ ((type)(phase) - (((type)(phase) - (type)(x)) & -(type)(align))) #define P2CROSS_TYPED(x, y, align, type) \ (((type)(x) ^ (type)(y)) > (type)(align) - 1) #define P2SAMEHIGHBIT_TYPED(x, y, type) \ (((type)(x) ^ (type)(y)) < ((type)(x) & (type)(y))) /* * Macros to atomically increment/decrement a variable. mutex and var * must be pointers. */ #define INCR_COUNT(var, mutex) mutex_enter(mutex), (*(var))++, mutex_exit(mutex) #define DECR_COUNT(var, mutex) mutex_enter(mutex), (*(var))--, mutex_exit(mutex) /* * Macros to declare bitfields - the order in the parameter list is * Low to High - that is, declare bit 0 first. We only support 8-bit bitfields * because if a field crosses a byte boundary it's not likely to be meaningful * without reassembly in its nonnative endianness. */ #if defined(_BIT_FIELDS_LTOH) #define DECL_BITFIELD2(_a, _b) \ uint8_t _a, _b #define DECL_BITFIELD3(_a, _b, _c) \ uint8_t _a, _b, _c #define DECL_BITFIELD4(_a, _b, _c, _d) \ uint8_t _a, _b, _c, _d #define DECL_BITFIELD5(_a, _b, _c, _d, _e) \ uint8_t _a, _b, _c, _d, _e #define DECL_BITFIELD6(_a, _b, _c, _d, _e, _f) \ uint8_t _a, _b, _c, _d, _e, _f #define DECL_BITFIELD7(_a, _b, _c, _d, _e, _f, _g) \ uint8_t _a, _b, _c, _d, _e, _f, _g #define DECL_BITFIELD8(_a, _b, _c, _d, _e, _f, _g, _h) \ uint8_t _a, _b, _c, _d, _e, _f, _g, _h #elif defined(_BIT_FIELDS_HTOL) #define DECL_BITFIELD2(_a, _b) \ uint8_t _b, _a #define DECL_BITFIELD3(_a, _b, _c) \ uint8_t _c, _b, _a #define DECL_BITFIELD4(_a, _b, _c, _d) \ uint8_t _d, _c, _b, _a #define DECL_BITFIELD5(_a, _b, _c, _d, _e) \ uint8_t _e, _d, _c, _b, _a #define DECL_BITFIELD6(_a, _b, _c, _d, _e, _f) \ uint8_t _f, _e, _d, _c, _b, _a #define DECL_BITFIELD7(_a, _b, _c, _d, _e, _f, _g) \ uint8_t _g, _f, _e, _d, _c, _b, _a #define DECL_BITFIELD8(_a, _b, _c, _d, _e, _f, _g, _h) \ uint8_t _h, _g, _f, _e, _d, _c, _b, _a #else #error One of _BIT_FIELDS_LTOH or _BIT_FIELDS_HTOL must be defined #endif /* _BIT_FIELDS_LTOH */ #if defined(_KERNEL) && !defined(_KMEMUSER) && !defined(offsetof) /* avoid any possibility of clashing with version */ #define offsetof(s, m) ((size_t)(&(((s *)0)->m))) #endif /* * Find highest one bit set. * Returns bit number + 1 of highest bit that is set, otherwise returns 0. * High order bit is 31 (or 63 in _LP64 kernel). */ static __inline int highbit(unsigned long i) { #if defined(__FreeBSD__) && defined(_KERNEL) && defined(HAVE_INLINE_FLSL) return (flsl(i)); #else int h = 1; if (i == 0) return (0); #ifdef _LP64 if (i & 0xffffffff00000000ul) { h += 32; i >>= 32; } #endif if (i & 0xffff0000) { h += 16; i >>= 16; } if (i & 0xff00) { h += 8; i >>= 8; } if (i & 0xf0) { h += 4; i >>= 4; } if (i & 0xc) { h += 2; i >>= 2; } if (i & 0x2) { h += 1; } return (h); #endif } /* * Find highest one bit set. * Returns bit number + 1 of highest bit that is set, otherwise returns 0. */ static __inline int highbit64(uint64_t i) { #if defined(__FreeBSD__) && defined(_KERNEL) && defined(HAVE_INLINE_FLSLL) return (flsll(i)); #else int h = 1; if (i == 0) return (0); if (i & 0xffffffff00000000ULL) { h += 32; i >>= 32; } if (i & 0xffff0000) { h += 16; i >>= 16; } if (i & 0xff00) { h += 8; i >>= 8; } if (i & 0xf0) { h += 4; i >>= 4; } if (i & 0xc) { h += 2; i >>= 2; } if (i & 0x2) { h += 1; } return (h); #endif } #ifdef __cplusplus } #endif #endif /* _SYS_SYSMACROS_H */ Index: projects/clang1100-import/sys/cddl/contrib/opensolaris =================================================================== --- projects/clang1100-import/sys/cddl/contrib/opensolaris (revision 364034) +++ projects/clang1100-import/sys/cddl/contrib/opensolaris (revision 364035) Property changes on: projects/clang1100-import/sys/cddl/contrib/opensolaris ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/cddl/contrib/opensolaris:r363987-364034 Index: projects/clang1100-import/sys/compat/linuxkpi/common/include/linux/radix-tree.h =================================================================== --- projects/clang1100-import/sys/compat/linuxkpi/common/include/linux/radix-tree.h (revision 364034) +++ projects/clang1100-import/sys/compat/linuxkpi/common/include/linux/radix-tree.h (revision 364035) @@ -1,84 +1,85 @@ /*- * Copyright (c) 2010 Isilon Systems, Inc. * Copyright (c) 2010 iX Systems, Inc. * Copyright (c) 2010 Panasas, Inc. * Copyright (c) 2013-2018 Mellanox Technologies, Ltd. * 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 unmodified, 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. * * $FreeBSD$ */ #ifndef _LINUX_RADIX_TREE_H_ #define _LINUX_RADIX_TREE_H_ #include #define RADIX_TREE_MAP_SHIFT 6 #define RADIX_TREE_MAP_SIZE (1UL << RADIX_TREE_MAP_SHIFT) #define RADIX_TREE_MAP_MASK (RADIX_TREE_MAP_SIZE - 1UL) #define RADIX_TREE_MAX_HEIGHT \ howmany(sizeof(long) * NBBY, RADIX_TREE_MAP_SHIFT) #define RADIX_TREE_ENTRY_MASK 3UL #define RADIX_TREE_EXCEPTIONAL_ENTRY 2UL #define RADIX_TREE_EXCEPTIONAL_SHIFT 2 struct radix_tree_node { void *slots[RADIX_TREE_MAP_SIZE]; int count; }; struct radix_tree_root { struct radix_tree_node *rnode; gfp_t gfp_mask; int height; }; struct radix_tree_iter { unsigned long index; }; #define RADIX_TREE_INIT(mask) \ { .rnode = NULL, .gfp_mask = mask, .height = 0 }; #define INIT_RADIX_TREE(root, mask) \ { (root)->rnode = NULL; (root)->gfp_mask = mask; (root)->height = 0; } #define RADIX_TREE(name, mask) \ struct radix_tree_root name = RADIX_TREE_INIT(mask) #define radix_tree_for_each_slot(slot, root, iter, start) \ for ((iter)->index = (start); \ radix_tree_iter_find(root, iter, &(slot)); (iter)->index++) static inline int radix_tree_exception(void *arg) { return ((uintptr_t)arg & RADIX_TREE_ENTRY_MASK); } void *radix_tree_lookup(struct radix_tree_root *, unsigned long); void *radix_tree_delete(struct radix_tree_root *, unsigned long); int radix_tree_insert(struct radix_tree_root *, unsigned long, void *); +int radix_tree_store(struct radix_tree_root *, unsigned long, void **); bool radix_tree_iter_find(struct radix_tree_root *, struct radix_tree_iter *, void ***); void radix_tree_iter_delete(struct radix_tree_root *, struct radix_tree_iter *, void **); #endif /* _LINUX_RADIX_TREE_H_ */ Index: projects/clang1100-import/sys/compat/linuxkpi/common/src/linux_radix.c =================================================================== --- projects/clang1100-import/sys/compat/linuxkpi/common/src/linux_radix.c (revision 364034) +++ projects/clang1100-import/sys/compat/linuxkpi/common/src/linux_radix.c (revision 364035) @@ -1,264 +1,387 @@ /*- * Copyright (c) 2010 Isilon Systems, Inc. * Copyright (c) 2010 iX Systems, Inc. * Copyright (c) 2010 Panasas, Inc. - * Copyright (c) 2013-2018 Mellanox Technologies, Ltd. + * Copyright (c) 2013-2020 Mellanox Technologies, Ltd. * 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 unmodified, 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. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include static MALLOC_DEFINE(M_RADIX, "radix", "Linux radix compat"); static inline unsigned long radix_max(struct radix_tree_root *root) { return ((1UL << (root->height * RADIX_TREE_MAP_SHIFT)) - 1UL); } static inline int radix_pos(long id, int height) { return (id >> (RADIX_TREE_MAP_SHIFT * height)) & RADIX_TREE_MAP_MASK; } +static void +radix_tree_clean_root_node(struct radix_tree_root *root) +{ + /* Check if the root node should be freed */ + if (root->rnode->count == 0) { + free(root->rnode, M_RADIX); + root->rnode = NULL; + root->height = 0; + } +} + void * radix_tree_lookup(struct radix_tree_root *root, unsigned long index) { struct radix_tree_node *node; void *item; int height; item = NULL; node = root->rnode; height = root->height - 1; if (index > radix_max(root)) goto out; while (height && node) node = node->slots[radix_pos(index, height--)]; if (node) item = node->slots[radix_pos(index, 0)]; out: return (item); } bool radix_tree_iter_find(struct radix_tree_root *root, struct radix_tree_iter *iter, void ***pppslot) { struct radix_tree_node *node; unsigned long index = iter->index; int height; restart: node = root->rnode; if (node == NULL) return (false); height = root->height - 1; if (height == -1 || index > radix_max(root)) return (false); do { unsigned long mask = RADIX_TREE_MAP_MASK << (RADIX_TREE_MAP_SHIFT * height); unsigned long step = 1UL << (RADIX_TREE_MAP_SHIFT * height); int pos = radix_pos(index, height); struct radix_tree_node *next; /* track last slot */ *pppslot = node->slots + pos; next = node->slots[pos]; if (next == NULL) { index += step; index &= -step; if ((index & mask) == 0) goto restart; } else { node = next; height--; } } while (height != -1); iter->index = index; return (true); } void * radix_tree_delete(struct radix_tree_root *root, unsigned long index) { struct radix_tree_node *stack[RADIX_TREE_MAX_HEIGHT]; struct radix_tree_node *node; void *item; int height; int idx; item = NULL; node = root->rnode; height = root->height - 1; if (index > radix_max(root)) goto out; /* * Find the node and record the path in stack. */ while (height && node) { stack[height] = node; node = node->slots[radix_pos(index, height--)]; } idx = radix_pos(index, 0); if (node) item = node->slots[idx]; /* * If we removed something reduce the height of the tree. */ if (item) for (;;) { node->slots[idx] = NULL; node->count--; if (node->count > 0) break; free(node, M_RADIX); if (node == root->rnode) { root->rnode = NULL; root->height = 0; break; } height++; node = stack[height]; idx = radix_pos(index, height); } out: return (item); } void radix_tree_iter_delete(struct radix_tree_root *root, struct radix_tree_iter *iter, void **slot) { radix_tree_delete(root, iter->index); } int radix_tree_insert(struct radix_tree_root *root, unsigned long index, void *item) { struct radix_tree_node *node; struct radix_tree_node *temp[RADIX_TREE_MAX_HEIGHT - 1]; int height; int idx; /* bail out upon insertion of a NULL item */ if (item == NULL) return (-EINVAL); /* get root node, if any */ node = root->rnode; /* allocate root node, if any */ if (node == NULL) { node = malloc(sizeof(*node), M_RADIX, root->gfp_mask | M_ZERO); if (node == NULL) return (-ENOMEM); root->rnode = node; root->height++; } /* expand radix tree as needed */ while (radix_max(root) < index) { /* check if the radix tree is getting too big */ - if (root->height == RADIX_TREE_MAX_HEIGHT) + if (root->height == RADIX_TREE_MAX_HEIGHT) { + radix_tree_clean_root_node(root); return (-E2BIG); + } /* * If the root radix level is not empty, we need to * allocate a new radix level: */ if (node->count != 0) { node = malloc(sizeof(*node), M_RADIX, root->gfp_mask | M_ZERO); - if (node == NULL) + if (node == NULL) { + /* + * Freeing the already allocated radix + * levels, if any, will be handled by + * the radix_tree_delete() function. + * This code path can only happen when + * the tree is not empty. + */ return (-ENOMEM); + } node->slots[0] = root->rnode; node->count++; root->rnode = node; } root->height++; } /* get radix tree height index */ height = root->height - 1; /* walk down the tree until the first missing node, if any */ for ( ; height != 0; height--) { idx = radix_pos(index, height); if (node->slots[idx] == NULL) break; node = node->slots[idx]; } /* allocate the missing radix levels, if any */ for (idx = 0; idx != height; idx++) { temp[idx] = malloc(sizeof(*node), M_RADIX, root->gfp_mask | M_ZERO); if (temp[idx] == NULL) { - while(idx--) + while (idx--) free(temp[idx], M_RADIX); - /* Check if we should free the root node as well. */ - if (root->rnode->count == 0) { - free(root->rnode, M_RADIX); - root->rnode = NULL; - root->height = 0; - } + radix_tree_clean_root_node(root); return (-ENOMEM); } } /* setup new radix levels, if any */ for ( ; height != 0; height--) { idx = radix_pos(index, height); node->slots[idx] = temp[height - 1]; node->count++; node = node->slots[idx]; } /* * Insert and adjust count if the item does not already exist. */ idx = radix_pos(index, 0); if (node->slots[idx]) return (-EEXIST); node->slots[idx] = item; node->count++; + return (0); +} + +int +radix_tree_store(struct radix_tree_root *root, unsigned long index, void **ppitem) +{ + struct radix_tree_node *node; + struct radix_tree_node *temp[RADIX_TREE_MAX_HEIGHT - 1]; + void *pitem; + int height; + int idx; + + /* + * Inserting a NULL item means delete it. The old pointer is + * stored at the location pointed to by "ppitem". + */ + if (*ppitem == NULL) { + *ppitem = radix_tree_delete(root, index); + return (0); + } + + /* get root node, if any */ + node = root->rnode; + + /* allocate root node, if any */ + if (node == NULL) { + node = malloc(sizeof(*node), M_RADIX, root->gfp_mask | M_ZERO); + if (node == NULL) + return (-ENOMEM); + root->rnode = node; + root->height++; + } + + /* expand radix tree as needed */ + while (radix_max(root) < index) { + + /* check if the radix tree is getting too big */ + if (root->height == RADIX_TREE_MAX_HEIGHT) { + radix_tree_clean_root_node(root); + return (-E2BIG); + } + + /* + * If the root radix level is not empty, we need to + * allocate a new radix level: + */ + if (node->count != 0) { + node = malloc(sizeof(*node), M_RADIX, root->gfp_mask | M_ZERO); + if (node == NULL) { + /* + * Freeing the already allocated radix + * levels, if any, will be handled by + * the radix_tree_delete() function. + * This code path can only happen when + * the tree is not empty. + */ + return (-ENOMEM); + } + node->slots[0] = root->rnode; + node->count++; + root->rnode = node; + } + root->height++; + } + + /* get radix tree height index */ + height = root->height - 1; + + /* walk down the tree until the first missing node, if any */ + for ( ; height != 0; height--) { + idx = radix_pos(index, height); + if (node->slots[idx] == NULL) + break; + node = node->slots[idx]; + } + + /* allocate the missing radix levels, if any */ + for (idx = 0; idx != height; idx++) { + temp[idx] = malloc(sizeof(*node), M_RADIX, + root->gfp_mask | M_ZERO); + if (temp[idx] == NULL) { + while (idx--) + free(temp[idx], M_RADIX); + radix_tree_clean_root_node(root); + return (-ENOMEM); + } + } + + /* setup new radix levels, if any */ + for ( ; height != 0; height--) { + idx = radix_pos(index, height); + node->slots[idx] = temp[height - 1]; + node->count++; + node = node->slots[idx]; + } + + /* + * Insert and adjust count if the item does not already exist. + */ + idx = radix_pos(index, 0); + /* swap */ + pitem = node->slots[idx]; + node->slots[idx] = *ppitem; + *ppitem = pitem; + + if (pitem == NULL) + node->count++; return (0); } Index: projects/clang1100-import/sys/dev/acpica/acpi_apei.c =================================================================== --- projects/clang1100-import/sys/dev/acpica/acpi_apei.c (revision 364034) +++ projects/clang1100-import/sys/dev/acpica/acpi_apei.c (revision 364035) @@ -1,684 +1,688 @@ /*- * SPDX-License-Identifier: BSD-2-Clause-FreeBSD * * Copyright (c) 2020 Alexander Motin * * 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_acpi.h" +#include "opt_pci.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include struct apei_ge { union { ACPI_HEST_GENERIC v1; ACPI_HEST_GENERIC_V2 v2; }; int res_type; int res_rid; struct resource *res; int res2_type; int res2_rid; struct resource *res2; uint8_t *buf, *copybuf; TAILQ_ENTRY(apei_ge) link; struct callout poll; void *swi_ih; } *apei_nmi_ge; struct apei_softc { ACPI_TABLE_HEST *hest; TAILQ_HEAD(, apei_ge) ges; }; struct apei_mem_error { uint64_t ValidationBits; uint64_t ErrorStatus; uint64_t PhysicalAddress; uint64_t PhysicalAddressMask; uint16_t Node; uint16_t Card; uint16_t Module; uint16_t Bank; uint16_t Device; uint16_t Row; uint16_t Column; uint16_t BitPosition; uint64_t RequesterID; uint64_t ResponderID; uint64_t TargetID; uint8_t MemoryErrorType; uint8_t Extended; uint16_t RankNumber; uint16_t CardHandle; uint16_t ModuleHandle; }; struct apei_pcie_error { uint64_t ValidationBits; uint32_t PortType; uint32_t Version; uint32_t CommandStatus; uint32_t Reserved; uint8_t DeviceID[16]; uint8_t DeviceSerialNumber[8]; uint8_t BridgeControlStatus[4]; uint8_t CapabilityStructure[60]; uint8_t AERInfo[96]; }; #ifdef __i386__ static __inline uint64_t apei_bus_read_8(struct resource *res, bus_size_t offset) { return (bus_read_4(res, offset) | ((uint64_t)bus_read_4(res, offset + 4)) << 32); } static __inline void apei_bus_write_8(struct resource *res, bus_size_t offset, uint64_t val) { bus_write_4(res, offset, val); bus_write_4(res, offset + 4, val >> 32); } #define READ8(r, o) apei_bus_read_8((r), (o)) #define WRITE8(r, o, v) apei_bus_write_8((r), (o), (v)) #else #define READ8(r, o) bus_read_8((r), (o)) #define WRITE8(r, o, v) bus_write_8((r), (o), (v)) #endif int apei_nmi_handler(void); static const char * apei_severity(uint32_t s) { switch (s) { case ACPI_HEST_GEN_ERROR_RECOVERABLE: return ("Recoverable"); case ACPI_HEST_GEN_ERROR_FATAL: return ("Fatal"); case ACPI_HEST_GEN_ERROR_CORRECTED: return ("Corrected"); case ACPI_HEST_GEN_ERROR_NONE: return ("Informational"); } return ("???"); } static int apei_mem_handler(ACPI_HEST_GENERIC_DATA *ged) { struct apei_mem_error *p = (struct apei_mem_error *)(ged + 1); printf("APEI %s Memory Error:\n", apei_severity(ged->ErrorSeverity)); if (p->ValidationBits & 0x01) printf(" Error Status: 0x%jx\n", p->ErrorStatus); if (p->ValidationBits & 0x02) printf(" Physical Address: 0x%jx\n", p->PhysicalAddress); if (p->ValidationBits & 0x04) printf(" Physical Address Mask: 0x%jx\n", p->PhysicalAddressMask); if (p->ValidationBits & 0x08) printf(" Node: %u\n", p->Node); if (p->ValidationBits & 0x10) printf(" Card: %u\n", p->Card); if (p->ValidationBits & 0x20) printf(" Module: %u\n", p->Module); if (p->ValidationBits & 0x40) printf(" Bank: %u\n", p->Bank); if (p->ValidationBits & 0x80) printf(" Device: %u\n", p->Device); if (p->ValidationBits & 0x100) printf(" Row: %u\n", p->Row); if (p->ValidationBits & 0x200) printf(" Column: %u\n", p->Column); if (p->ValidationBits & 0x400) printf(" Bit Position: %u\n", p->BitPosition); if (p->ValidationBits & 0x800) printf(" Requester ID: 0x%jx\n", p->RequesterID); if (p->ValidationBits & 0x1000) printf(" Responder ID: 0x%jx\n", p->ResponderID); if (p->ValidationBits & 0x2000) printf(" Target ID: 0x%jx\n", p->TargetID); if (p->ValidationBits & 0x4000) printf(" Memory Error Type: %u\n", p->MemoryErrorType); if (p->ValidationBits & 0x8000) printf(" Rank Number: %u\n", p->RankNumber); if (p->ValidationBits & 0x10000) printf(" Card Handle: 0x%x\n", p->CardHandle); if (p->ValidationBits & 0x20000) printf(" Module Handle: 0x%x\n", p->ModuleHandle); if (p->ValidationBits & 0x40000) printf(" Extended Row: %u\n", (uint32_t)(p->Extended & 0x3) << 16 | p->Row); if (p->ValidationBits & 0x80000) printf(" Bank Group: %u\n", p->Bank >> 8); if (p->ValidationBits & 0x100000) printf(" Bank Address: %u\n", p->Bank & 0xff); if (p->ValidationBits & 0x200000) printf(" Chip Identification: %u\n", (p->Extended >> 5) & 0x7); return (0); } static int apei_pcie_handler(ACPI_HEST_GENERIC_DATA *ged) { struct apei_pcie_error *p = (struct apei_pcie_error *)(ged + 1); + int h = 0, off; +#ifdef DEV_PCI device_t dev; - int h = 0, off, sev; + int sev; if ((p->ValidationBits & 0x8) == 0x8) { mtx_lock(&Giant); dev = pci_find_dbsf((uint32_t)p->DeviceID[10] << 8 | p->DeviceID[9], p->DeviceID[11], p->DeviceID[8], p->DeviceID[7]); if (dev != NULL) { switch (ged->ErrorSeverity) { case ACPI_HEST_GEN_ERROR_FATAL: sev = PCIEM_STA_FATAL_ERROR; break; case ACPI_HEST_GEN_ERROR_RECOVERABLE: sev = PCIEM_STA_NON_FATAL_ERROR; break; default: sev = PCIEM_STA_CORRECTABLE_ERROR; break; } pcie_apei_error(dev, sev, (p->ValidationBits & 0x80) ? p->AERInfo : NULL); h = 1; } mtx_unlock(&Giant); } if (h) return (h); +#endif printf("APEI %s PCIe Error:\n", apei_severity(ged->ErrorSeverity)); if (p->ValidationBits & 0x01) printf(" Port Type: %u\n", p->PortType); if (p->ValidationBits & 0x02) printf(" Version: %x\n", p->Version); if (p->ValidationBits & 0x04) printf(" Command Status: 0x%08x\n", p->CommandStatus); if (p->ValidationBits & 0x08) { printf(" DeviceID:"); for (off = 0; off < sizeof(p->DeviceID); off++) printf(" %02x", p->DeviceID[off]); printf("\n"); } if (p->ValidationBits & 0x10) { printf(" Device Serial Number:"); for (off = 0; off < sizeof(p->DeviceSerialNumber); off++) printf(" %02x", p->DeviceSerialNumber[off]); printf("\n"); } if (p->ValidationBits & 0x20) { printf(" Bridge Control Status:"); for (off = 0; off < sizeof(p->BridgeControlStatus); off++) printf(" %02x", p->BridgeControlStatus[off]); printf("\n"); } if (p->ValidationBits & 0x40) { printf(" Capability Structure:\n"); for (off = 0; off < sizeof(p->CapabilityStructure); off++) { printf(" %02x", p->CapabilityStructure[off]); if ((off % 16) == 15 || off + 1 == sizeof(p->CapabilityStructure)) printf("\n"); } } if (p->ValidationBits & 0x80) { printf(" AER Info:\n"); for (off = 0; off < sizeof(p->AERInfo); off++) { printf(" %02x", p->AERInfo[off]); if ((off % 16) == 15 || off + 1 == sizeof(p->AERInfo)) printf("\n"); } } return (h); } static void apei_ged_handler(ACPI_HEST_GENERIC_DATA *ged) { ACPI_HEST_GENERIC_DATA_V300 *ged3 = (ACPI_HEST_GENERIC_DATA_V300 *)ged; /* A5BC1114-6F64-4EDE-B863-3E83ED7C83B1 */ static uint8_t mem_uuid[ACPI_UUID_LENGTH] = { 0x14, 0x11, 0xBC, 0xA5, 0x64, 0x6F, 0xDE, 0x4E, 0xB8, 0x63, 0x3E, 0x83, 0xED, 0x7C, 0x83, 0xB1 }; /* D995E954-BBC1-430F-AD91-B44DCB3C6F35 */ static uint8_t pcie_uuid[ACPI_UUID_LENGTH] = { 0x54, 0xE9, 0x95, 0xD9, 0xC1, 0xBB, 0x0F, 0x43, 0xAD, 0x91, 0xB4, 0x4D, 0xCB, 0x3C, 0x6F, 0x35 }; uint8_t *t; int h = 0, off; if (memcmp(mem_uuid, ged->SectionType, ACPI_UUID_LENGTH) == 0) { h = apei_mem_handler(ged); } else if (memcmp(pcie_uuid, ged->SectionType, ACPI_UUID_LENGTH) == 0) { h = apei_pcie_handler(ged); } else { t = ged->SectionType; printf("APEI %s Error %02x%02x%02x%02x-%02x%02x-" "%02x%02x-%02x%02x-%02x%02x%02x%02x%02x%02x:\n", apei_severity(ged->ErrorSeverity), t[3], t[2], t[1], t[0], t[5], t[4], t[7], t[6], t[8], t[9], t[10], t[11], t[12], t[13], t[14], t[15]); printf(" Error Data:\n"); t = (uint8_t *)(ged + 1); for (off = 0; off < ged->ErrorDataLength; off++) { printf(" %02x", t[off]); if ((off % 16) == 15 || off + 1 == ged->ErrorDataLength) printf("\n"); } } if (h) return; printf(" Flags: 0x%x\n", ged->Flags); if (ged->ValidationBits & ACPI_HEST_GEN_VALID_FRU_ID) { t = ged->FruId; printf(" FRU Id: %02x%02x%02x%02x-%02x%02x-%02x%02x-" "%02x%02x-%02x%02x%02x%02x%02x%02x\n", t[3], t[2], t[1], t[0], t[5], t[4], t[7], t[6], t[8], t[9], t[10], t[11], t[12], t[13], t[14], t[15]); } if (ged->ValidationBits & ACPI_HEST_GEN_VALID_FRU_STRING) printf(" FRU Text: %.20s", ged->FruText); if (ged->Revision == 0x300 && ged->ValidationBits & ACPI_HEST_GEN_VALID_TIMESTAMP) printf(" Timestamp: %016jx", ged3->TimeStamp); } static int apei_ge_handler(struct apei_ge *ge, bool copy) { uint8_t *buf = copy ? ge->copybuf : ge->buf; ACPI_HEST_GENERIC_STATUS *ges = (ACPI_HEST_GENERIC_STATUS *)buf; ACPI_HEST_GENERIC_DATA *ged; uint32_t sev; int i, c, off; if (ges->BlockStatus == 0) return (0); c = (ges->BlockStatus >> 4) & 0x3ff; sev = ges->ErrorSeverity; /* Process error entries. */ for (off = i = 0; i < c && off + sizeof(*ged) <= ges->DataLength; i++) { ged = (ACPI_HEST_GENERIC_DATA *)&buf[sizeof(*ges) + off]; apei_ged_handler(ged); off += sizeof(*ged) + ged->ErrorDataLength; } /* Acknowledge the error has been processed. */ ges->BlockStatus = 0; if (!copy && ge->v1.Header.Type == ACPI_HEST_TYPE_GENERIC_ERROR_V2) { uint64_t val = READ8(ge->res2, 0); val &= ge->v2.ReadAckPreserve; val |= ge->v2.ReadAckWrite; WRITE8(ge->res2, 0, val); } /* If ACPI told the error is fatal -- make it so. */ if (sev == ACPI_HEST_GEN_ERROR_FATAL) panic("APEI Fatal Hardware Error!"); return (1); } static void apei_nmi_swi(void *arg) { struct apei_ge *ge = arg; apei_ge_handler(ge, true); } int apei_nmi_handler(void) { struct apei_ge *ge = apei_nmi_ge; ACPI_HEST_GENERIC_STATUS *ges, *gesc; if (ge == NULL) return (0); ges = (ACPI_HEST_GENERIC_STATUS *)ge->buf; if (ges->BlockStatus == 0) return (0); /* If ACPI told the error is fatal -- make it so. */ if (ges->ErrorSeverity == ACPI_HEST_GEN_ERROR_FATAL) panic("APEI Fatal Hardware Error!"); /* Copy the buffer for later processing. */ gesc = (ACPI_HEST_GENERIC_STATUS *)ge->copybuf; if (gesc->BlockStatus == 0) memcpy(ge->copybuf, ge->buf, ge->v1.ErrorBlockLength); /* Acknowledge the error has been processed. */ ges->BlockStatus = 0; if (ge->v1.Header.Type == ACPI_HEST_TYPE_GENERIC_ERROR_V2) { uint64_t val = READ8(ge->res2, 0); val &= ge->v2.ReadAckPreserve; val |= ge->v2.ReadAckWrite; WRITE8(ge->res2, 0, val); } /* Schedule SWI for real handling. */ swi_sched(ge->swi_ih, SWI_FROMNMI); return (1); } static void apei_callout_handler(void *context) { struct apei_ge *ge = context; apei_ge_handler(ge, false); callout_schedule(&ge->poll, ge->v1.Notify.PollInterval * hz / 1000); } static void apei_notify_handler(ACPI_HANDLE h, UINT32 notify, void *context) { device_t dev = context; struct apei_softc *sc = device_get_softc(dev); struct apei_ge *ge; TAILQ_FOREACH(ge, &sc->ges, link) { if (ge->v1.Notify.Type == ACPI_HEST_NOTIFY_SCI || ge->v1.Notify.Type == ACPI_HEST_NOTIFY_GPIO || ge->v1.Notify.Type == ACPI_HEST_NOTIFY_GSIV) apei_ge_handler(ge, false); } } static int hest_parse_structure(struct apei_softc *sc, void *addr, int remaining) { ACPI_HEST_HEADER *hdr = addr; struct apei_ge *ge; if (remaining < (int)sizeof(ACPI_HEST_HEADER)) return (-1); switch (hdr->Type) { case ACPI_HEST_TYPE_IA32_CHECK: { ACPI_HEST_IA_MACHINE_CHECK *s = addr; return (sizeof(*s) + s->NumHardwareBanks * sizeof(ACPI_HEST_IA_ERROR_BANK)); } case ACPI_HEST_TYPE_IA32_CORRECTED_CHECK: { ACPI_HEST_IA_CORRECTED *s = addr; return (sizeof(*s) + s->NumHardwareBanks * sizeof(ACPI_HEST_IA_ERROR_BANK)); } case ACPI_HEST_TYPE_IA32_NMI: { ACPI_HEST_IA_NMI *s = addr; return (sizeof(*s)); } case ACPI_HEST_TYPE_AER_ROOT_PORT: { ACPI_HEST_AER_ROOT *s = addr; return (sizeof(*s)); } case ACPI_HEST_TYPE_AER_ENDPOINT: { ACPI_HEST_AER *s = addr; return (sizeof(*s)); } case ACPI_HEST_TYPE_AER_BRIDGE: { ACPI_HEST_AER_BRIDGE *s = addr; return (sizeof(*s)); } case ACPI_HEST_TYPE_GENERIC_ERROR: { ACPI_HEST_GENERIC *s = addr; ge = malloc(sizeof(*ge), M_DEVBUF, M_WAITOK | M_ZERO); ge->v1 = *s; TAILQ_INSERT_TAIL(&sc->ges, ge, link); return (sizeof(*s)); } case ACPI_HEST_TYPE_GENERIC_ERROR_V2: { ACPI_HEST_GENERIC_V2 *s = addr; ge = malloc(sizeof(*ge), M_DEVBUF, M_WAITOK | M_ZERO); ge->v2 = *s; TAILQ_INSERT_TAIL(&sc->ges, ge, link); return (sizeof(*s)); } case ACPI_HEST_TYPE_IA32_DEFERRED_CHECK: { ACPI_HEST_IA_DEFERRED_CHECK *s = addr; return (sizeof(*s) + s->NumHardwareBanks * sizeof(ACPI_HEST_IA_ERROR_BANK)); } default: return (-1); } } static void hest_parse_table(struct apei_softc *sc) { ACPI_TABLE_HEST *hest = sc->hest; char *cp; int remaining, consumed; remaining = hest->Header.Length - sizeof(ACPI_TABLE_HEST); while (remaining > 0) { cp = (char *)hest + hest->Header.Length - remaining; consumed = hest_parse_structure(sc, cp, remaining); if (consumed <= 0) break; else remaining -= consumed; } } static char *apei_ids[] = { "PNP0C33", NULL }; static devclass_t apei_devclass; static ACPI_STATUS apei_find(ACPI_HANDLE handle, UINT32 level, void *context, void **status) { int *found = (int *)status; char **ids; for (ids = apei_ids; *ids != NULL; ids++) { if (acpi_MatchHid(handle, *ids)) { *found = 1; break; } } return (AE_OK); } static void apei_identify(driver_t *driver, device_t parent) { device_t child; int found; if (acpi_disabled("apei")) return; if (acpi_find_table(ACPI_SIG_HEST) == 0) return; /* Only one APEI device can exist. */ if (devclass_get_device(apei_devclass, 0)) return; /* Search for ACPI error device to be used. */ found = 0; AcpiWalkNamespace(ACPI_TYPE_DEVICE, ACPI_ROOT_OBJECT, 100, apei_find, NULL, NULL, (void *)&found); if (found) return; /* If not found - create a fake one. */ child = BUS_ADD_CHILD(parent, 2, "apei", 0); if (child == NULL) printf("%s: can't add child\n", __func__); } static int apei_probe(device_t dev) { int rv; if (acpi_disabled("apei")) return (ENXIO); if (acpi_find_table(ACPI_SIG_HEST) == 0) return (ENXIO); if (acpi_get_handle(dev) != NULL) rv = ACPI_ID_PROBE(device_get_parent(dev), dev, apei_ids, NULL); else rv = 0; if (rv <= 0) device_set_desc(dev, "Platform Error Interface"); return (rv); } static int apei_attach(device_t dev) { struct apei_softc *sc = device_get_softc(dev); struct apei_ge *ge; ACPI_STATUS status; int rid; TAILQ_INIT(&sc->ges); /* Search and parse HEST table. */ status = AcpiGetTable(ACPI_SIG_HEST, 0, (ACPI_TABLE_HEADER **)&sc->hest); if (ACPI_FAILURE(status)) return (ENXIO); hest_parse_table(sc); AcpiPutTable((ACPI_TABLE_HEADER *)sc->hest); rid = 0; TAILQ_FOREACH(ge, &sc->ges, link) { ge->res_rid = rid++; acpi_bus_alloc_gas(dev, &ge->res_type, &ge->res_rid, &ge->v1.ErrorStatusAddress, &ge->res, 0); if (ge->v1.Header.Type == ACPI_HEST_TYPE_GENERIC_ERROR_V2) { ge->res2_rid = rid++; acpi_bus_alloc_gas(dev, &ge->res2_type, &ge->res2_rid, &ge->v2.ReadAckRegister, &ge->res2, 0); } ge->buf = pmap_mapdev_attr(READ8(ge->res, 0), ge->v1.ErrorBlockLength, VM_MEMATTR_WRITE_COMBINING); if (ge->v1.Notify.Type == ACPI_HEST_NOTIFY_POLLED) { callout_init(&ge->poll, 1); callout_reset(&ge->poll, ge->v1.Notify.PollInterval * hz / 1000, apei_callout_handler, ge); } else if (ge->v1.Notify.Type == ACPI_HEST_NOTIFY_NMI) { ge->copybuf = malloc(ge->v1.ErrorBlockLength, M_DEVBUF, M_WAITOK | M_ZERO); swi_add(&clk_intr_event, "apei", apei_nmi_swi, ge, SWI_CLOCK, INTR_MPSAFE, &ge->swi_ih); apei_nmi_ge = ge; apei_nmi = apei_nmi_handler; } } if (acpi_get_handle(dev) != NULL) { AcpiInstallNotifyHandler(acpi_get_handle(dev), ACPI_DEVICE_NOTIFY, apei_notify_handler, dev); } return (0); } static int apei_detach(device_t dev) { struct apei_softc *sc = device_get_softc(dev); struct apei_ge *ge; apei_nmi = NULL; apei_nmi_ge = NULL; if (acpi_get_handle(dev) != NULL) { AcpiRemoveNotifyHandler(acpi_get_handle(dev), ACPI_DEVICE_NOTIFY, apei_notify_handler); } while ((ge = TAILQ_FIRST(&sc->ges)) != NULL) { TAILQ_REMOVE(&sc->ges, ge, link); bus_release_resource(dev, ge->res_type, ge->res_rid, ge->res); if (ge->res2) { bus_release_resource(dev, ge->res2_type, ge->res2_rid, ge->res2); } if (ge->v1.Notify.Type == ACPI_HEST_NOTIFY_POLLED) { callout_drain(&ge->poll); } else if (ge->v1.Notify.Type == ACPI_HEST_NOTIFY_NMI) { swi_remove(&ge->swi_ih); free(ge->copybuf, M_DEVBUF); } pmap_unmapdev((vm_offset_t)ge->buf, ge->v1.ErrorBlockLength); free(ge, M_DEVBUF); } return (0); } static device_method_t apei_methods[] = { /* Device interface */ DEVMETHOD(device_identify, apei_identify), DEVMETHOD(device_probe, apei_probe), DEVMETHOD(device_attach, apei_attach), DEVMETHOD(device_detach, apei_detach), DEVMETHOD_END }; static driver_t apei_driver = { "apei", apei_methods, sizeof(struct apei_softc), }; DRIVER_MODULE(apei, acpi, apei_driver, apei_devclass, 0, 0); MODULE_DEPEND(apei, acpi, 1, 1, 1); Index: projects/clang1100-import/sys/dev/e1000/if_em.c =================================================================== --- projects/clang1100-import/sys/dev/e1000/if_em.c (revision 364034) +++ projects/clang1100-import/sys/dev/e1000/if_em.c (revision 364035) @@ -1,4636 +1,4641 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2016 Nicole Graziano * 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. */ /* $FreeBSD$ */ #include "if_em.h" #include #include #define em_mac_min e1000_82547 #define igb_mac_min e1000_82575 /********************************************************************* * Driver version: *********************************************************************/ char em_driver_version[] = "7.6.1-k"; /********************************************************************* * PCI Device ID Table * * Used by probe to select devices to load on * Last field stores an index into e1000_strings * Last entry must be all 0s * * { Vendor ID, Device ID, SubVendor ID, SubDevice ID, String Index } *********************************************************************/ static pci_vendor_info_t em_vendor_info_array[] = { /* Intel(R) PRO/1000 Network Connection - Legacy em*/ PVID(0x8086, E1000_DEV_ID_82540EM, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82540EM_LOM, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82540EP, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82540EP_LOM, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82540EP_LP, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82541EI, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82541ER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82541ER_LOM, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82541EI_MOBILE, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82541GI, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82541GI_LF, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82541GI_MOBILE, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82542, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82543GC_FIBER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82543GC_COPPER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82544EI_COPPER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82544EI_FIBER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82544GC_COPPER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82544GC_LOM, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82545EM_COPPER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82545EM_FIBER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82545GM_COPPER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82545GM_FIBER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82545GM_SERDES, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82546EB_COPPER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82546EB_FIBER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82546EB_QUAD_COPPER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82546GB_COPPER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82546GB_FIBER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82546GB_SERDES, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82546GB_PCIE, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82546GB_QUAD_COPPER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82546GB_QUAD_COPPER_KSP3, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82547EI, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82547EI_MOBILE, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82547GI, "Intel(R) PRO/1000 Network Connection"), /* Intel(R) PRO/1000 Network Connection - em */ PVID(0x8086, E1000_DEV_ID_82571EB_COPPER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82571EB_FIBER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82571EB_SERDES, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82571EB_SERDES_DUAL, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82571EB_SERDES_QUAD, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82571EB_QUAD_COPPER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82571EB_QUAD_COPPER_LP, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82571EB_QUAD_FIBER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82571PT_QUAD_COPPER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82572EI, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82572EI_COPPER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82572EI_FIBER, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82572EI_SERDES, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82573E, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82573E_IAMT, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82573L, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82583V, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_80003ES2LAN_COPPER_SPT, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_80003ES2LAN_SERDES_SPT, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_80003ES2LAN_COPPER_DPT, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_80003ES2LAN_SERDES_DPT, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH8_IGP_M_AMT, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH8_IGP_AMT, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH8_IGP_C, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH8_IFE, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH8_IFE_GT, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH8_IFE_G, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH8_IGP_M, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH8_82567V_3, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH9_IGP_M_AMT, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH9_IGP_AMT, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH9_IGP_C, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH9_IGP_M, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH9_IGP_M_V, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH9_IFE, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH9_IFE_GT, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH9_IFE_G, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH9_BM, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82574L, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_82574LA, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH10_R_BM_LM, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH10_R_BM_LF, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH10_R_BM_V, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH10_D_BM_LM, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH10_D_BM_LF, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_ICH10_D_BM_V, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_M_HV_LM, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_M_HV_LC, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_D_HV_DM, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_D_HV_DC, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH2_LV_LM, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH2_LV_V, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_LPT_I217_LM, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_LPT_I217_V, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_LPTLP_I218_LM, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_LPTLP_I218_V, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_I218_LM2, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_I218_V2, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_I218_LM3, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_I218_V3, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_SPT_I219_LM, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_SPT_I219_V, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_SPT_I219_LM2, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_SPT_I219_V2, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_LBG_I219_LM3, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_SPT_I219_LM4, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_SPT_I219_V4, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_SPT_I219_LM5, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_SPT_I219_V5, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_CNP_I219_LM6, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_CNP_I219_V6, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_CNP_I219_LM7, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_CNP_I219_V7, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_ICP_I219_LM8, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_ICP_I219_V8, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_ICP_I219_LM9, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_ICP_I219_V9, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_CMP_I219_LM10, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_CMP_I219_V10, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_CMP_I219_LM11, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_CMP_I219_V11, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_CMP_I219_LM12, "Intel(R) PRO/1000 Network Connection"), PVID(0x8086, E1000_DEV_ID_PCH_CMP_I219_V12, "Intel(R) PRO/1000 Network Connection"), /* required last entry */ PVID_END }; static pci_vendor_info_t igb_vendor_info_array[] = { /* Intel(R) PRO/1000 Network Connection - igb */ PVID(0x8086, E1000_DEV_ID_82575EB_COPPER, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82575EB_FIBER_SERDES, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82575GB_QUAD_COPPER, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82576, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82576_NS, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82576_NS_SERDES, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82576_FIBER, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82576_SERDES, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82576_SERDES_QUAD, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82576_QUAD_COPPER, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82576_QUAD_COPPER_ET2, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82576_VF, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82580_COPPER, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82580_FIBER, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82580_SERDES, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82580_SGMII, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82580_COPPER_DUAL, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_82580_QUAD_FIBER, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_DH89XXCC_SERDES, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_DH89XXCC_SGMII, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_DH89XXCC_SFP, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_DH89XXCC_BACKPLANE, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I350_COPPER, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I350_FIBER, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I350_SERDES, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I350_SGMII, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I350_VF, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I210_COPPER, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I210_COPPER_IT, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I210_COPPER_OEM1, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I210_COPPER_FLASHLESS, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I210_SERDES_FLASHLESS, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I210_FIBER, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I210_SERDES, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I210_SGMII, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I211_COPPER, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I354_BACKPLANE_1GBPS, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I354_BACKPLANE_2_5GBPS, "Intel(R) PRO/1000 PCI-Express Network Driver"), PVID(0x8086, E1000_DEV_ID_I354_SGMII, "Intel(R) PRO/1000 PCI-Express Network Driver"), /* required last entry */ PVID_END }; /********************************************************************* * Function prototypes *********************************************************************/ static void *em_register(device_t dev); static void *igb_register(device_t dev); static int em_if_attach_pre(if_ctx_t ctx); static int em_if_attach_post(if_ctx_t ctx); static int em_if_detach(if_ctx_t ctx); static int em_if_shutdown(if_ctx_t ctx); static int em_if_suspend(if_ctx_t ctx); static int em_if_resume(if_ctx_t ctx); static int em_if_tx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs, int ntxqs, int ntxqsets); static int em_if_rx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs, int nrxqs, int nrxqsets); static void em_if_queues_free(if_ctx_t ctx); static uint64_t em_if_get_counter(if_ctx_t, ift_counter); static void em_if_init(if_ctx_t ctx); static void em_if_stop(if_ctx_t ctx); static void em_if_media_status(if_ctx_t, struct ifmediareq *); static int em_if_media_change(if_ctx_t ctx); static int em_if_mtu_set(if_ctx_t ctx, uint32_t mtu); static void em_if_timer(if_ctx_t ctx, uint16_t qid); static void em_if_vlan_register(if_ctx_t ctx, u16 vtag); static void em_if_vlan_unregister(if_ctx_t ctx, u16 vtag); static void em_if_watchdog_reset(if_ctx_t ctx); static bool em_if_needs_restart(if_ctx_t ctx, enum iflib_restart_event event); static void em_identify_hardware(if_ctx_t ctx); static int em_allocate_pci_resources(if_ctx_t ctx); static void em_free_pci_resources(if_ctx_t ctx); static void em_reset(if_ctx_t ctx); static int em_setup_interface(if_ctx_t ctx); static int em_setup_msix(if_ctx_t ctx); static void em_initialize_transmit_unit(if_ctx_t ctx); static void em_initialize_receive_unit(if_ctx_t ctx); static void em_if_intr_enable(if_ctx_t ctx); static void em_if_intr_disable(if_ctx_t ctx); static void igb_if_intr_enable(if_ctx_t ctx); static void igb_if_intr_disable(if_ctx_t ctx); static int em_if_rx_queue_intr_enable(if_ctx_t ctx, uint16_t rxqid); static int em_if_tx_queue_intr_enable(if_ctx_t ctx, uint16_t txqid); static int igb_if_rx_queue_intr_enable(if_ctx_t ctx, uint16_t rxqid); static int igb_if_tx_queue_intr_enable(if_ctx_t ctx, uint16_t txqid); static void em_if_multi_set(if_ctx_t ctx); static void em_if_update_admin_status(if_ctx_t ctx); static void em_if_debug(if_ctx_t ctx); static void em_update_stats_counters(struct adapter *); static void em_add_hw_stats(struct adapter *adapter); static int em_if_set_promisc(if_ctx_t ctx, int flags); static void em_setup_vlan_hw_support(struct adapter *); static int em_sysctl_nvm_info(SYSCTL_HANDLER_ARGS); static void em_print_nvm_info(struct adapter *); static int em_sysctl_debug_info(SYSCTL_HANDLER_ARGS); static int em_get_rs(SYSCTL_HANDLER_ARGS); static void em_print_debug_info(struct adapter *); static int em_is_valid_ether_addr(u8 *); static int em_sysctl_int_delay(SYSCTL_HANDLER_ARGS); static void em_add_int_delay_sysctl(struct adapter *, const char *, const char *, struct em_int_delay_info *, int, int); /* Management and WOL Support */ static void em_init_manageability(struct adapter *); static void em_release_manageability(struct adapter *); static void em_get_hw_control(struct adapter *); static void em_release_hw_control(struct adapter *); static void em_get_wakeup(if_ctx_t ctx); static void em_enable_wakeup(if_ctx_t ctx); static int em_enable_phy_wakeup(struct adapter *); static void em_disable_aspm(struct adapter *); int em_intr(void *arg); static void em_disable_promisc(if_ctx_t ctx); /* MSI-X handlers */ static int em_if_msix_intr_assign(if_ctx_t, int); static int em_msix_link(void *); static void em_handle_link(void *context); static void em_enable_vectors_82574(if_ctx_t); static int em_set_flowcntl(SYSCTL_HANDLER_ARGS); static int em_sysctl_eee(SYSCTL_HANDLER_ARGS); static void em_if_led_func(if_ctx_t ctx, int onoff); static int em_get_regs(SYSCTL_HANDLER_ARGS); static void lem_smartspeed(struct adapter *adapter); static void igb_configure_queues(struct adapter *adapter); /********************************************************************* * FreeBSD Device Interface Entry Points *********************************************************************/ static device_method_t em_methods[] = { /* Device interface */ DEVMETHOD(device_register, em_register), DEVMETHOD(device_probe, iflib_device_probe), DEVMETHOD(device_attach, iflib_device_attach), DEVMETHOD(device_detach, iflib_device_detach), DEVMETHOD(device_shutdown, iflib_device_shutdown), DEVMETHOD(device_suspend, iflib_device_suspend), DEVMETHOD(device_resume, iflib_device_resume), DEVMETHOD_END }; static device_method_t igb_methods[] = { /* Device interface */ DEVMETHOD(device_register, igb_register), DEVMETHOD(device_probe, iflib_device_probe), DEVMETHOD(device_attach, iflib_device_attach), DEVMETHOD(device_detach, iflib_device_detach), DEVMETHOD(device_shutdown, iflib_device_shutdown), DEVMETHOD(device_suspend, iflib_device_suspend), DEVMETHOD(device_resume, iflib_device_resume), DEVMETHOD_END }; static driver_t em_driver = { "em", em_methods, sizeof(struct adapter), }; static devclass_t em_devclass; DRIVER_MODULE(em, pci, em_driver, em_devclass, 0, 0); MODULE_DEPEND(em, pci, 1, 1, 1); MODULE_DEPEND(em, ether, 1, 1, 1); MODULE_DEPEND(em, iflib, 1, 1, 1); IFLIB_PNP_INFO(pci, em, em_vendor_info_array); static driver_t igb_driver = { "igb", igb_methods, sizeof(struct adapter), }; static devclass_t igb_devclass; DRIVER_MODULE(igb, pci, igb_driver, igb_devclass, 0, 0); MODULE_DEPEND(igb, pci, 1, 1, 1); MODULE_DEPEND(igb, ether, 1, 1, 1); MODULE_DEPEND(igb, iflib, 1, 1, 1); IFLIB_PNP_INFO(pci, igb, igb_vendor_info_array); static device_method_t em_if_methods[] = { DEVMETHOD(ifdi_attach_pre, em_if_attach_pre), DEVMETHOD(ifdi_attach_post, em_if_attach_post), DEVMETHOD(ifdi_detach, em_if_detach), DEVMETHOD(ifdi_shutdown, em_if_shutdown), DEVMETHOD(ifdi_suspend, em_if_suspend), DEVMETHOD(ifdi_resume, em_if_resume), DEVMETHOD(ifdi_init, em_if_init), DEVMETHOD(ifdi_stop, em_if_stop), DEVMETHOD(ifdi_msix_intr_assign, em_if_msix_intr_assign), DEVMETHOD(ifdi_intr_enable, em_if_intr_enable), DEVMETHOD(ifdi_intr_disable, em_if_intr_disable), DEVMETHOD(ifdi_tx_queues_alloc, em_if_tx_queues_alloc), DEVMETHOD(ifdi_rx_queues_alloc, em_if_rx_queues_alloc), DEVMETHOD(ifdi_queues_free, em_if_queues_free), DEVMETHOD(ifdi_update_admin_status, em_if_update_admin_status), DEVMETHOD(ifdi_multi_set, em_if_multi_set), DEVMETHOD(ifdi_media_status, em_if_media_status), DEVMETHOD(ifdi_media_change, em_if_media_change), DEVMETHOD(ifdi_mtu_set, em_if_mtu_set), DEVMETHOD(ifdi_promisc_set, em_if_set_promisc), DEVMETHOD(ifdi_timer, em_if_timer), DEVMETHOD(ifdi_watchdog_reset, em_if_watchdog_reset), DEVMETHOD(ifdi_vlan_register, em_if_vlan_register), DEVMETHOD(ifdi_vlan_unregister, em_if_vlan_unregister), DEVMETHOD(ifdi_get_counter, em_if_get_counter), DEVMETHOD(ifdi_led_func, em_if_led_func), DEVMETHOD(ifdi_rx_queue_intr_enable, em_if_rx_queue_intr_enable), DEVMETHOD(ifdi_tx_queue_intr_enable, em_if_tx_queue_intr_enable), DEVMETHOD(ifdi_debug, em_if_debug), DEVMETHOD(ifdi_needs_restart, em_if_needs_restart), DEVMETHOD_END }; static driver_t em_if_driver = { "em_if", em_if_methods, sizeof(struct adapter) }; static device_method_t igb_if_methods[] = { DEVMETHOD(ifdi_attach_pre, em_if_attach_pre), DEVMETHOD(ifdi_attach_post, em_if_attach_post), DEVMETHOD(ifdi_detach, em_if_detach), DEVMETHOD(ifdi_shutdown, em_if_shutdown), DEVMETHOD(ifdi_suspend, em_if_suspend), DEVMETHOD(ifdi_resume, em_if_resume), DEVMETHOD(ifdi_init, em_if_init), DEVMETHOD(ifdi_stop, em_if_stop), DEVMETHOD(ifdi_msix_intr_assign, em_if_msix_intr_assign), DEVMETHOD(ifdi_intr_enable, igb_if_intr_enable), DEVMETHOD(ifdi_intr_disable, igb_if_intr_disable), DEVMETHOD(ifdi_tx_queues_alloc, em_if_tx_queues_alloc), DEVMETHOD(ifdi_rx_queues_alloc, em_if_rx_queues_alloc), DEVMETHOD(ifdi_queues_free, em_if_queues_free), DEVMETHOD(ifdi_update_admin_status, em_if_update_admin_status), DEVMETHOD(ifdi_multi_set, em_if_multi_set), DEVMETHOD(ifdi_media_status, em_if_media_status), DEVMETHOD(ifdi_media_change, em_if_media_change), DEVMETHOD(ifdi_mtu_set, em_if_mtu_set), DEVMETHOD(ifdi_promisc_set, em_if_set_promisc), DEVMETHOD(ifdi_timer, em_if_timer), DEVMETHOD(ifdi_watchdog_reset, em_if_watchdog_reset), DEVMETHOD(ifdi_vlan_register, em_if_vlan_register), DEVMETHOD(ifdi_vlan_unregister, em_if_vlan_unregister), DEVMETHOD(ifdi_get_counter, em_if_get_counter), DEVMETHOD(ifdi_led_func, em_if_led_func), DEVMETHOD(ifdi_rx_queue_intr_enable, igb_if_rx_queue_intr_enable), DEVMETHOD(ifdi_tx_queue_intr_enable, igb_if_tx_queue_intr_enable), DEVMETHOD(ifdi_debug, em_if_debug), DEVMETHOD(ifdi_needs_restart, em_if_needs_restart), DEVMETHOD_END }; static driver_t igb_if_driver = { "igb_if", igb_if_methods, sizeof(struct adapter) }; /********************************************************************* * Tunable default values. *********************************************************************/ #define EM_TICKS_TO_USECS(ticks) ((1024 * (ticks) + 500) / 1000) #define EM_USECS_TO_TICKS(usecs) ((1000 * (usecs) + 512) / 1024) #define MAX_INTS_PER_SEC 8000 #define DEFAULT_ITR (1000000000/(MAX_INTS_PER_SEC * 256)) /* Allow common code without TSO */ #ifndef CSUM_TSO #define CSUM_TSO 0 #endif static SYSCTL_NODE(_hw, OID_AUTO, em, CTLFLAG_RD | CTLFLAG_MPSAFE, 0, "EM driver parameters"); static int em_disable_crc_stripping = 0; SYSCTL_INT(_hw_em, OID_AUTO, disable_crc_stripping, CTLFLAG_RDTUN, &em_disable_crc_stripping, 0, "Disable CRC Stripping"); static int em_tx_int_delay_dflt = EM_TICKS_TO_USECS(EM_TIDV); static int em_rx_int_delay_dflt = EM_TICKS_TO_USECS(EM_RDTR); SYSCTL_INT(_hw_em, OID_AUTO, tx_int_delay, CTLFLAG_RDTUN, &em_tx_int_delay_dflt, 0, "Default transmit interrupt delay in usecs"); SYSCTL_INT(_hw_em, OID_AUTO, rx_int_delay, CTLFLAG_RDTUN, &em_rx_int_delay_dflt, 0, "Default receive interrupt delay in usecs"); static int em_tx_abs_int_delay_dflt = EM_TICKS_TO_USECS(EM_TADV); static int em_rx_abs_int_delay_dflt = EM_TICKS_TO_USECS(EM_RADV); SYSCTL_INT(_hw_em, OID_AUTO, tx_abs_int_delay, CTLFLAG_RDTUN, &em_tx_abs_int_delay_dflt, 0, "Default transmit interrupt delay limit in usecs"); SYSCTL_INT(_hw_em, OID_AUTO, rx_abs_int_delay, CTLFLAG_RDTUN, &em_rx_abs_int_delay_dflt, 0, "Default receive interrupt delay limit in usecs"); static int em_smart_pwr_down = FALSE; SYSCTL_INT(_hw_em, OID_AUTO, smart_pwr_down, CTLFLAG_RDTUN, &em_smart_pwr_down, 0, "Set to true to leave smart power down enabled on newer adapters"); /* Controls whether promiscuous also shows bad packets */ static int em_debug_sbp = TRUE; SYSCTL_INT(_hw_em, OID_AUTO, sbp, CTLFLAG_RDTUN, &em_debug_sbp, 0, "Show bad packets in promiscuous mode"); /* How many packets rxeof tries to clean at a time */ static int em_rx_process_limit = 100; SYSCTL_INT(_hw_em, OID_AUTO, rx_process_limit, CTLFLAG_RDTUN, &em_rx_process_limit, 0, "Maximum number of received packets to process " "at a time, -1 means unlimited"); /* Energy efficient ethernet - default to OFF */ static int eee_setting = 1; SYSCTL_INT(_hw_em, OID_AUTO, eee_setting, CTLFLAG_RDTUN, &eee_setting, 0, "Enable Energy Efficient Ethernet"); /* ** Tuneable Interrupt rate */ static int em_max_interrupt_rate = 8000; SYSCTL_INT(_hw_em, OID_AUTO, max_interrupt_rate, CTLFLAG_RDTUN, &em_max_interrupt_rate, 0, "Maximum interrupts per second"); /* Global used in WOL setup with multiport cards */ static int global_quad_port_a = 0; extern struct if_txrx igb_txrx; extern struct if_txrx em_txrx; extern struct if_txrx lem_txrx; static struct if_shared_ctx em_sctx_init = { .isc_magic = IFLIB_MAGIC, .isc_q_align = PAGE_SIZE, .isc_tx_maxsize = EM_TSO_SIZE + sizeof(struct ether_vlan_header), .isc_tx_maxsegsize = PAGE_SIZE, .isc_tso_maxsize = EM_TSO_SIZE + sizeof(struct ether_vlan_header), .isc_tso_maxsegsize = EM_TSO_SEG_SIZE, .isc_rx_maxsize = MJUM9BYTES, .isc_rx_nsegments = 1, .isc_rx_maxsegsize = MJUM9BYTES, .isc_nfl = 1, .isc_nrxqs = 1, .isc_ntxqs = 1, .isc_admin_intrcnt = 1, .isc_vendor_info = em_vendor_info_array, .isc_driver_version = em_driver_version, .isc_driver = &em_if_driver, .isc_flags = IFLIB_NEED_SCRATCH | IFLIB_TSO_INIT_IP | IFLIB_NEED_ZERO_CSUM, .isc_nrxd_min = {EM_MIN_RXD}, .isc_ntxd_min = {EM_MIN_TXD}, .isc_nrxd_max = {EM_MAX_RXD}, .isc_ntxd_max = {EM_MAX_TXD}, .isc_nrxd_default = {EM_DEFAULT_RXD}, .isc_ntxd_default = {EM_DEFAULT_TXD}, }; if_shared_ctx_t em_sctx = &em_sctx_init; static struct if_shared_ctx igb_sctx_init = { .isc_magic = IFLIB_MAGIC, .isc_q_align = PAGE_SIZE, .isc_tx_maxsize = EM_TSO_SIZE + sizeof(struct ether_vlan_header), .isc_tx_maxsegsize = PAGE_SIZE, .isc_tso_maxsize = EM_TSO_SIZE + sizeof(struct ether_vlan_header), .isc_tso_maxsegsize = EM_TSO_SEG_SIZE, .isc_rx_maxsize = MJUM9BYTES, .isc_rx_nsegments = 1, .isc_rx_maxsegsize = MJUM9BYTES, .isc_nfl = 1, .isc_nrxqs = 1, .isc_ntxqs = 1, .isc_admin_intrcnt = 1, .isc_vendor_info = igb_vendor_info_array, .isc_driver_version = em_driver_version, .isc_driver = &igb_if_driver, .isc_flags = IFLIB_NEED_SCRATCH | IFLIB_TSO_INIT_IP | IFLIB_NEED_ZERO_CSUM, .isc_nrxd_min = {EM_MIN_RXD}, .isc_ntxd_min = {EM_MIN_TXD}, .isc_nrxd_max = {IGB_MAX_RXD}, .isc_ntxd_max = {IGB_MAX_TXD}, .isc_nrxd_default = {EM_DEFAULT_RXD}, .isc_ntxd_default = {EM_DEFAULT_TXD}, }; if_shared_ctx_t igb_sctx = &igb_sctx_init; /***************************************************************** * * Dump Registers * ****************************************************************/ #define IGB_REGS_LEN 739 static int em_get_regs(SYSCTL_HANDLER_ARGS) { struct adapter *adapter = (struct adapter *)arg1; struct e1000_hw *hw = &adapter->hw; struct sbuf *sb; u32 *regs_buff; int rc; regs_buff = malloc(sizeof(u32) * IGB_REGS_LEN, M_DEVBUF, M_WAITOK); memset(regs_buff, 0, IGB_REGS_LEN * sizeof(u32)); rc = sysctl_wire_old_buffer(req, 0); MPASS(rc == 0); if (rc != 0) { free(regs_buff, M_DEVBUF); return (rc); } sb = sbuf_new_for_sysctl(NULL, NULL, 32*400, req); MPASS(sb != NULL); if (sb == NULL) { free(regs_buff, M_DEVBUF); return (ENOMEM); } /* General Registers */ regs_buff[0] = E1000_READ_REG(hw, E1000_CTRL); regs_buff[1] = E1000_READ_REG(hw, E1000_STATUS); regs_buff[2] = E1000_READ_REG(hw, E1000_CTRL_EXT); regs_buff[3] = E1000_READ_REG(hw, E1000_ICR); regs_buff[4] = E1000_READ_REG(hw, E1000_RCTL); regs_buff[5] = E1000_READ_REG(hw, E1000_RDLEN(0)); regs_buff[6] = E1000_READ_REG(hw, E1000_RDH(0)); regs_buff[7] = E1000_READ_REG(hw, E1000_RDT(0)); regs_buff[8] = E1000_READ_REG(hw, E1000_RXDCTL(0)); regs_buff[9] = E1000_READ_REG(hw, E1000_RDBAL(0)); regs_buff[10] = E1000_READ_REG(hw, E1000_RDBAH(0)); regs_buff[11] = E1000_READ_REG(hw, E1000_TCTL); regs_buff[12] = E1000_READ_REG(hw, E1000_TDBAL(0)); regs_buff[13] = E1000_READ_REG(hw, E1000_TDBAH(0)); regs_buff[14] = E1000_READ_REG(hw, E1000_TDLEN(0)); regs_buff[15] = E1000_READ_REG(hw, E1000_TDH(0)); regs_buff[16] = E1000_READ_REG(hw, E1000_TDT(0)); regs_buff[17] = E1000_READ_REG(hw, E1000_TXDCTL(0)); regs_buff[18] = E1000_READ_REG(hw, E1000_TDFH); regs_buff[19] = E1000_READ_REG(hw, E1000_TDFT); regs_buff[20] = E1000_READ_REG(hw, E1000_TDFHS); regs_buff[21] = E1000_READ_REG(hw, E1000_TDFPC); sbuf_printf(sb, "General Registers\n"); sbuf_printf(sb, "\tCTRL\t %08x\n", regs_buff[0]); sbuf_printf(sb, "\tSTATUS\t %08x\n", regs_buff[1]); sbuf_printf(sb, "\tCTRL_EXIT\t %08x\n\n", regs_buff[2]); sbuf_printf(sb, "Interrupt Registers\n"); sbuf_printf(sb, "\tICR\t %08x\n\n", regs_buff[3]); sbuf_printf(sb, "RX Registers\n"); sbuf_printf(sb, "\tRCTL\t %08x\n", regs_buff[4]); sbuf_printf(sb, "\tRDLEN\t %08x\n", regs_buff[5]); sbuf_printf(sb, "\tRDH\t %08x\n", regs_buff[6]); sbuf_printf(sb, "\tRDT\t %08x\n", regs_buff[7]); sbuf_printf(sb, "\tRXDCTL\t %08x\n", regs_buff[8]); sbuf_printf(sb, "\tRDBAL\t %08x\n", regs_buff[9]); sbuf_printf(sb, "\tRDBAH\t %08x\n\n", regs_buff[10]); sbuf_printf(sb, "TX Registers\n"); sbuf_printf(sb, "\tTCTL\t %08x\n", regs_buff[11]); sbuf_printf(sb, "\tTDBAL\t %08x\n", regs_buff[12]); sbuf_printf(sb, "\tTDBAH\t %08x\n", regs_buff[13]); sbuf_printf(sb, "\tTDLEN\t %08x\n", regs_buff[14]); sbuf_printf(sb, "\tTDH\t %08x\n", regs_buff[15]); sbuf_printf(sb, "\tTDT\t %08x\n", regs_buff[16]); sbuf_printf(sb, "\tTXDCTL\t %08x\n", regs_buff[17]); sbuf_printf(sb, "\tTDFH\t %08x\n", regs_buff[18]); sbuf_printf(sb, "\tTDFT\t %08x\n", regs_buff[19]); sbuf_printf(sb, "\tTDFHS\t %08x\n", regs_buff[20]); sbuf_printf(sb, "\tTDFPC\t %08x\n\n", regs_buff[21]); free(regs_buff, M_DEVBUF); #ifdef DUMP_DESCS { if_softc_ctx_t scctx = adapter->shared; struct rx_ring *rxr = &rx_que->rxr; struct tx_ring *txr = &tx_que->txr; int ntxd = scctx->isc_ntxd[0]; int nrxd = scctx->isc_nrxd[0]; int j; for (j = 0; j < nrxd; j++) { u32 staterr = le32toh(rxr->rx_base[j].wb.upper.status_error); u32 length = le32toh(rxr->rx_base[j].wb.upper.length); sbuf_printf(sb, "\tReceive Descriptor Address %d: %08" PRIx64 " Error:%d Length:%d\n", j, rxr->rx_base[j].read.buffer_addr, staterr, length); } for (j = 0; j < min(ntxd, 256); j++) { unsigned int *ptr = (unsigned int *)&txr->tx_base[j]; sbuf_printf(sb, "\tTXD[%03d] [0]: %08x [1]: %08x [2]: %08x [3]: %08x eop: %d DD=%d\n", j, ptr[0], ptr[1], ptr[2], ptr[3], buf->eop, buf->eop != -1 ? txr->tx_base[buf->eop].upper.fields.status & E1000_TXD_STAT_DD : 0); } } #endif rc = sbuf_finish(sb); sbuf_delete(sb); return(rc); } static void * em_register(device_t dev) { return (em_sctx); } static void * igb_register(device_t dev) { return (igb_sctx); } static int em_set_num_queues(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); int maxqueues; /* Sanity check based on HW */ switch (adapter->hw.mac.type) { case e1000_82576: case e1000_82580: case e1000_i350: case e1000_i354: maxqueues = 8; break; case e1000_i210: case e1000_82575: maxqueues = 4; break; case e1000_i211: case e1000_82574: maxqueues = 2; break; default: maxqueues = 1; break; } return (maxqueues); } #define LEM_CAPS \ IFCAP_HWCSUM | IFCAP_VLAN_MTU | IFCAP_VLAN_HWTAGGING | \ IFCAP_VLAN_HWCSUM | IFCAP_WOL | IFCAP_VLAN_HWFILTER #define EM_CAPS \ IFCAP_HWCSUM | IFCAP_VLAN_MTU | IFCAP_VLAN_HWTAGGING | \ IFCAP_VLAN_HWCSUM | IFCAP_WOL | IFCAP_VLAN_HWFILTER | IFCAP_TSO4 | \ IFCAP_LRO | IFCAP_VLAN_HWTSO #define IGB_CAPS \ IFCAP_HWCSUM | IFCAP_VLAN_MTU | IFCAP_VLAN_HWTAGGING | \ IFCAP_VLAN_HWCSUM | IFCAP_WOL | IFCAP_VLAN_HWFILTER | IFCAP_TSO4 | \ IFCAP_LRO | IFCAP_VLAN_HWTSO | IFCAP_JUMBO_MTU | IFCAP_HWCSUM_IPV6 |\ IFCAP_TSO6 /********************************************************************* * Device initialization routine * * The attach entry point is called when the driver is being loaded. * This routine identifies the type of hardware, allocates all resources * and initializes the hardware. * * return 0 on success, positive on failure *********************************************************************/ static int em_if_attach_pre(if_ctx_t ctx) { struct adapter *adapter; if_softc_ctx_t scctx; device_t dev; struct e1000_hw *hw; int error = 0; INIT_DEBUGOUT("em_if_attach_pre: begin"); dev = iflib_get_dev(ctx); adapter = iflib_get_softc(ctx); adapter->ctx = adapter->osdep.ctx = ctx; adapter->dev = adapter->osdep.dev = dev; scctx = adapter->shared = iflib_get_softc_ctx(ctx); adapter->media = iflib_get_media(ctx); hw = &adapter->hw; adapter->tx_process_limit = scctx->isc_ntxd[0]; /* SYSCTL stuff */ SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev), SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), OID_AUTO, "nvm", CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_NEEDGIANT, adapter, 0, em_sysctl_nvm_info, "I", "NVM Information"); SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev), SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), OID_AUTO, "debug", CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_NEEDGIANT, adapter, 0, em_sysctl_debug_info, "I", "Debug Information"); SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev), SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), OID_AUTO, "fc", CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_NEEDGIANT, adapter, 0, em_set_flowcntl, "I", "Flow Control"); SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev), SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), OID_AUTO, "reg_dump", CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_NEEDGIANT, adapter, 0, em_get_regs, "A", "Dump Registers"); SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev), SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), OID_AUTO, "rs_dump", CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_NEEDGIANT, adapter, 0, em_get_rs, "I", "Dump RS indexes"); /* Determine hardware and mac info */ em_identify_hardware(ctx); scctx->isc_tx_nsegments = EM_MAX_SCATTER; scctx->isc_nrxqsets_max = scctx->isc_ntxqsets_max = em_set_num_queues(ctx); if (bootverbose) device_printf(dev, "attach_pre capping queues at %d\n", scctx->isc_ntxqsets_max); if (adapter->hw.mac.type >= igb_mac_min) { scctx->isc_txqsizes[0] = roundup2(scctx->isc_ntxd[0] * sizeof(union e1000_adv_tx_desc), EM_DBA_ALIGN); scctx->isc_rxqsizes[0] = roundup2(scctx->isc_nrxd[0] * sizeof(union e1000_adv_rx_desc), EM_DBA_ALIGN); scctx->isc_txd_size[0] = sizeof(union e1000_adv_tx_desc); scctx->isc_rxd_size[0] = sizeof(union e1000_adv_rx_desc); scctx->isc_txrx = &igb_txrx; scctx->isc_tx_tso_segments_max = EM_MAX_SCATTER; scctx->isc_tx_tso_size_max = EM_TSO_SIZE; scctx->isc_tx_tso_segsize_max = EM_TSO_SEG_SIZE; scctx->isc_capabilities = scctx->isc_capenable = IGB_CAPS; scctx->isc_tx_csum_flags = CSUM_TCP | CSUM_UDP | CSUM_TSO | CSUM_IP6_TCP | CSUM_IP6_UDP; if (adapter->hw.mac.type != e1000_82575) scctx->isc_tx_csum_flags |= CSUM_SCTP | CSUM_IP6_SCTP; /* ** Some new devices, as with ixgbe, now may ** use a different BAR, so we need to keep ** track of which is used. */ scctx->isc_msix_bar = PCIR_BAR(EM_MSIX_BAR); if (pci_read_config(dev, scctx->isc_msix_bar, 4) == 0) scctx->isc_msix_bar += 4; } else if (adapter->hw.mac.type >= em_mac_min) { scctx->isc_txqsizes[0] = roundup2(scctx->isc_ntxd[0]* sizeof(struct e1000_tx_desc), EM_DBA_ALIGN); scctx->isc_rxqsizes[0] = roundup2(scctx->isc_nrxd[0] * sizeof(union e1000_rx_desc_extended), EM_DBA_ALIGN); scctx->isc_txd_size[0] = sizeof(struct e1000_tx_desc); scctx->isc_rxd_size[0] = sizeof(union e1000_rx_desc_extended); scctx->isc_txrx = &em_txrx; scctx->isc_tx_tso_segments_max = EM_MAX_SCATTER; scctx->isc_tx_tso_size_max = EM_TSO_SIZE; scctx->isc_tx_tso_segsize_max = EM_TSO_SEG_SIZE; scctx->isc_capabilities = scctx->isc_capenable = EM_CAPS; /* * For EM-class devices, don't enable IFCAP_{TSO4,VLAN_HWTSO} * by default as we don't have workarounds for all associated * silicon errata. E. g., with several MACs such as 82573E, * TSO only works at Gigabit speed and otherwise can cause the * hardware to hang (which also would be next to impossible to * work around given that already queued TSO-using descriptors * would need to be flushed and vlan(4) reconfigured at runtime * in case of a link speed change). Moreover, MACs like 82579 * still can hang at Gigabit even with all publicly documented * TSO workarounds implemented. Generally, the penality of * these workarounds is rather high and may involve copying * mbuf data around so advantages of TSO lapse. Still, TSO may * work for a few MACs of this class - at least when sticking * with Gigabit - in which case users may enable TSO manually. */ scctx->isc_capenable &= ~(IFCAP_TSO4 | IFCAP_VLAN_HWTSO); scctx->isc_tx_csum_flags = CSUM_TCP | CSUM_UDP | CSUM_IP_TSO; /* * We support MSI-X with 82574 only, but indicate to iflib(4) * that it shall give MSI at least a try with other devices. */ if (adapter->hw.mac.type == e1000_82574) { scctx->isc_msix_bar = PCIR_BAR(EM_MSIX_BAR); } else { scctx->isc_msix_bar = -1; scctx->isc_disable_msix = 1; } } else { scctx->isc_txqsizes[0] = roundup2((scctx->isc_ntxd[0] + 1) * sizeof(struct e1000_tx_desc), EM_DBA_ALIGN); scctx->isc_rxqsizes[0] = roundup2((scctx->isc_nrxd[0] + 1) * sizeof(struct e1000_rx_desc), EM_DBA_ALIGN); scctx->isc_txd_size[0] = sizeof(struct e1000_tx_desc); scctx->isc_rxd_size[0] = sizeof(struct e1000_rx_desc); scctx->isc_tx_csum_flags = CSUM_TCP | CSUM_UDP; scctx->isc_txrx = &lem_txrx; scctx->isc_capabilities = scctx->isc_capenable = LEM_CAPS; if (adapter->hw.mac.type < e1000_82543) scctx->isc_capenable &= ~(IFCAP_HWCSUM|IFCAP_VLAN_HWCSUM); /* INTx only */ scctx->isc_msix_bar = 0; } /* Setup PCI resources */ if (em_allocate_pci_resources(ctx)) { device_printf(dev, "Allocation of PCI resources failed\n"); error = ENXIO; goto err_pci; } /* ** For ICH8 and family we need to ** map the flash memory, and this ** must happen after the MAC is ** identified */ if ((hw->mac.type == e1000_ich8lan) || (hw->mac.type == e1000_ich9lan) || (hw->mac.type == e1000_ich10lan) || (hw->mac.type == e1000_pchlan) || (hw->mac.type == e1000_pch2lan) || (hw->mac.type == e1000_pch_lpt)) { int rid = EM_BAR_TYPE_FLASH; adapter->flash = bus_alloc_resource_any(dev, SYS_RES_MEMORY, &rid, RF_ACTIVE); if (adapter->flash == NULL) { device_printf(dev, "Mapping of Flash failed\n"); error = ENXIO; goto err_pci; } /* This is used in the shared code */ hw->flash_address = (u8 *)adapter->flash; adapter->osdep.flash_bus_space_tag = rman_get_bustag(adapter->flash); adapter->osdep.flash_bus_space_handle = rman_get_bushandle(adapter->flash); } /* ** In the new SPT device flash is not a ** separate BAR, rather it is also in BAR0, ** so use the same tag and an offset handle for the ** FLASH read/write macros in the shared code. */ else if (hw->mac.type >= e1000_pch_spt) { adapter->osdep.flash_bus_space_tag = adapter->osdep.mem_bus_space_tag; adapter->osdep.flash_bus_space_handle = adapter->osdep.mem_bus_space_handle + E1000_FLASH_BASE_ADDR; } /* Do Shared Code initialization */ error = e1000_setup_init_funcs(hw, TRUE); if (error) { device_printf(dev, "Setup of Shared code failed, error %d\n", error); error = ENXIO; goto err_pci; } em_setup_msix(ctx); e1000_get_bus_info(hw); /* Set up some sysctls for the tunable interrupt delays */ em_add_int_delay_sysctl(adapter, "rx_int_delay", "receive interrupt delay in usecs", &adapter->rx_int_delay, E1000_REGISTER(hw, E1000_RDTR), em_rx_int_delay_dflt); em_add_int_delay_sysctl(adapter, "tx_int_delay", "transmit interrupt delay in usecs", &adapter->tx_int_delay, E1000_REGISTER(hw, E1000_TIDV), em_tx_int_delay_dflt); em_add_int_delay_sysctl(adapter, "rx_abs_int_delay", "receive interrupt delay limit in usecs", &adapter->rx_abs_int_delay, E1000_REGISTER(hw, E1000_RADV), em_rx_abs_int_delay_dflt); em_add_int_delay_sysctl(adapter, "tx_abs_int_delay", "transmit interrupt delay limit in usecs", &adapter->tx_abs_int_delay, E1000_REGISTER(hw, E1000_TADV), em_tx_abs_int_delay_dflt); em_add_int_delay_sysctl(adapter, "itr", "interrupt delay limit in usecs/4", &adapter->tx_itr, E1000_REGISTER(hw, E1000_ITR), DEFAULT_ITR); hw->mac.autoneg = DO_AUTO_NEG; hw->phy.autoneg_wait_to_complete = FALSE; hw->phy.autoneg_advertised = AUTONEG_ADV_DEFAULT; if (adapter->hw.mac.type < em_mac_min) { e1000_init_script_state_82541(&adapter->hw, TRUE); e1000_set_tbi_compatibility_82543(&adapter->hw, TRUE); } /* Copper options */ if (hw->phy.media_type == e1000_media_type_copper) { hw->phy.mdix = AUTO_ALL_MODES; hw->phy.disable_polarity_correction = FALSE; hw->phy.ms_type = EM_MASTER_SLAVE; } /* * Set the frame limits assuming * standard ethernet sized frames. */ scctx->isc_max_frame_size = adapter->hw.mac.max_frame_size = ETHERMTU + ETHER_HDR_LEN + ETHERNET_FCS_SIZE; /* * This controls when hardware reports transmit completion * status. */ hw->mac.report_tx_early = 1; /* Allocate multicast array memory. */ adapter->mta = malloc(sizeof(u8) * ETHER_ADDR_LEN * MAX_NUM_MULTICAST_ADDRESSES, M_DEVBUF, M_NOWAIT); if (adapter->mta == NULL) { device_printf(dev, "Can not allocate multicast setup array\n"); error = ENOMEM; goto err_late; } /* Check SOL/IDER usage */ if (e1000_check_reset_block(hw)) device_printf(dev, "PHY reset is blocked" " due to SOL/IDER session.\n"); /* Sysctl for setting Energy Efficient Ethernet */ hw->dev_spec.ich8lan.eee_disable = eee_setting; SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev), SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), OID_AUTO, "eee_control", CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_NEEDGIANT, adapter, 0, em_sysctl_eee, "I", "Disable Energy Efficient Ethernet"); /* ** Start from a known state, this is ** important in reading the nvm and ** mac from that. */ e1000_reset_hw(hw); /* Make sure we have a good EEPROM before we read from it */ if (e1000_validate_nvm_checksum(hw) < 0) { /* ** Some PCI-E parts fail the first check due to ** the link being in sleep state, call it again, ** if it fails a second time its a real issue. */ if (e1000_validate_nvm_checksum(hw) < 0) { device_printf(dev, "The EEPROM Checksum Is Not Valid\n"); error = EIO; goto err_late; } } /* Copy the permanent MAC address out of the EEPROM */ if (e1000_read_mac_addr(hw) < 0) { device_printf(dev, "EEPROM read error while reading MAC" " address\n"); error = EIO; goto err_late; } if (!em_is_valid_ether_addr(hw->mac.addr)) { device_printf(dev, "Invalid MAC address\n"); error = EIO; goto err_late; } /* Disable ULP support */ e1000_disable_ulp_lpt_lp(hw, TRUE); /* * Get Wake-on-Lan and Management info for later use */ em_get_wakeup(ctx); /* Enable only WOL MAGIC by default */ scctx->isc_capenable &= ~IFCAP_WOL; if (adapter->wol != 0) scctx->isc_capenable |= IFCAP_WOL_MAGIC; iflib_set_mac(ctx, hw->mac.addr); return (0); err_late: em_release_hw_control(adapter); err_pci: em_free_pci_resources(ctx); free(adapter->mta, M_DEVBUF); return (error); } static int em_if_attach_post(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); struct e1000_hw *hw = &adapter->hw; int error = 0; /* Setup OS specific network interface */ error = em_setup_interface(ctx); if (error != 0) { goto err_late; } em_reset(ctx); /* Initialize statistics */ em_update_stats_counters(adapter); hw->mac.get_link_status = 1; em_if_update_admin_status(ctx); em_add_hw_stats(adapter); /* Non-AMT based hardware can now take control from firmware */ if (adapter->has_manage && !adapter->has_amt) em_get_hw_control(adapter); INIT_DEBUGOUT("em_if_attach_post: end"); return (error); err_late: em_release_hw_control(adapter); em_free_pci_resources(ctx); em_if_queues_free(ctx); free(adapter->mta, M_DEVBUF); return (error); } /********************************************************************* * Device removal routine * * The detach entry point is called when the driver is being removed. * This routine stops the adapter and deallocates all the resources * that were allocated for driver operation. * * return 0 on success, positive on failure *********************************************************************/ static int em_if_detach(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); INIT_DEBUGOUT("em_if_detach: begin"); e1000_phy_hw_reset(&adapter->hw); em_release_manageability(adapter); em_release_hw_control(adapter); em_free_pci_resources(ctx); return (0); } /********************************************************************* * * Shutdown entry point * **********************************************************************/ static int em_if_shutdown(if_ctx_t ctx) { return em_if_suspend(ctx); } /* * Suspend/resume device methods. */ static int em_if_suspend(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); em_release_manageability(adapter); em_release_hw_control(adapter); em_enable_wakeup(ctx); return (0); } static int em_if_resume(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); if (adapter->hw.mac.type == e1000_pch2lan) e1000_resume_workarounds_pchlan(&adapter->hw); em_if_init(ctx); em_init_manageability(adapter); return(0); } static int em_if_mtu_set(if_ctx_t ctx, uint32_t mtu) { int max_frame_size; struct adapter *adapter = iflib_get_softc(ctx); if_softc_ctx_t scctx = iflib_get_softc_ctx(ctx); IOCTL_DEBUGOUT("ioctl rcv'd: SIOCSIFMTU (Set Interface MTU)"); switch (adapter->hw.mac.type) { case e1000_82571: case e1000_82572: case e1000_ich9lan: case e1000_ich10lan: case e1000_pch2lan: case e1000_pch_lpt: case e1000_pch_spt: case e1000_pch_cnp: case e1000_82574: case e1000_82583: case e1000_80003es2lan: /* 9K Jumbo Frame size */ max_frame_size = 9234; break; case e1000_pchlan: max_frame_size = 4096; break; case e1000_82542: case e1000_ich8lan: /* Adapters that do not support jumbo frames */ max_frame_size = ETHER_MAX_LEN; break; default: if (adapter->hw.mac.type >= igb_mac_min) max_frame_size = 9234; else /* lem */ max_frame_size = MAX_JUMBO_FRAME_SIZE; } if (mtu > max_frame_size - ETHER_HDR_LEN - ETHER_CRC_LEN) { return (EINVAL); } scctx->isc_max_frame_size = adapter->hw.mac.max_frame_size = mtu + ETHER_HDR_LEN + ETHER_CRC_LEN; return (0); } /********************************************************************* * Init entry point * * This routine is used in two ways. It is used by the stack as * init entry point in network interface structure. It is also used * by the driver as a hw/sw initialization routine to get to a * consistent state. * **********************************************************************/ static void em_if_init(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); if_softc_ctx_t scctx = adapter->shared; struct ifnet *ifp = iflib_get_ifp(ctx); struct em_tx_queue *tx_que; int i; INIT_DEBUGOUT("em_if_init: begin"); /* Get the latest mac address, User can use a LAA */ bcopy(if_getlladdr(ifp), adapter->hw.mac.addr, ETHER_ADDR_LEN); /* Put the address into the Receive Address Array */ e1000_rar_set(&adapter->hw, adapter->hw.mac.addr, 0); /* * With the 82571 adapter, RAR[0] may be overwritten * when the other port is reset, we make a duplicate * in RAR[14] for that eventuality, this assures * the interface continues to function. */ if (adapter->hw.mac.type == e1000_82571) { e1000_set_laa_state_82571(&adapter->hw, TRUE); e1000_rar_set(&adapter->hw, adapter->hw.mac.addr, E1000_RAR_ENTRIES - 1); } /* Initialize the hardware */ em_reset(ctx); em_if_update_admin_status(ctx); for (i = 0, tx_que = adapter->tx_queues; i < adapter->tx_num_queues; i++, tx_que++) { struct tx_ring *txr = &tx_que->txr; txr->tx_rs_cidx = txr->tx_rs_pidx; /* Initialize the last processed descriptor to be the end of * the ring, rather than the start, so that we avoid an * off-by-one error when calculating how many descriptors are * done in the credits_update function. */ txr->tx_cidx_processed = scctx->isc_ntxd[0] - 1; } /* Setup VLAN support, basic and offload if available */ E1000_WRITE_REG(&adapter->hw, E1000_VET, ETHERTYPE_VLAN); /* Clear bad data from Rx FIFOs */ if (adapter->hw.mac.type >= igb_mac_min) e1000_rx_fifo_flush_82575(&adapter->hw); /* Configure for OS presence */ em_init_manageability(adapter); /* Prepare transmit descriptors and buffers */ em_initialize_transmit_unit(ctx); /* Setup Multicast table */ em_if_multi_set(ctx); adapter->rx_mbuf_sz = iflib_get_rx_mbuf_sz(ctx); em_initialize_receive_unit(ctx); /* Use real VLAN Filter support? */ if (if_getcapenable(ifp) & IFCAP_VLAN_HWTAGGING) { if (if_getcapenable(ifp) & IFCAP_VLAN_HWFILTER) /* Use real VLAN Filter support */ em_setup_vlan_hw_support(adapter); else { u32 ctrl; ctrl = E1000_READ_REG(&adapter->hw, E1000_CTRL); ctrl |= E1000_CTRL_VME; E1000_WRITE_REG(&adapter->hw, E1000_CTRL, ctrl); } + } else { + u32 ctrl; + ctrl = E1000_READ_REG(&adapter->hw, E1000_CTRL); + ctrl &= ~E1000_CTRL_VME; + E1000_WRITE_REG(&adapter->hw, E1000_CTRL, ctrl); } /* Don't lose promiscuous settings */ em_if_set_promisc(ctx, IFF_PROMISC); e1000_clear_hw_cntrs_base_generic(&adapter->hw); /* MSI-X configuration for 82574 */ if (adapter->hw.mac.type == e1000_82574) { int tmp = E1000_READ_REG(&adapter->hw, E1000_CTRL_EXT); tmp |= E1000_CTRL_EXT_PBA_CLR; E1000_WRITE_REG(&adapter->hw, E1000_CTRL_EXT, tmp); /* Set the IVAR - interrupt vector routing. */ E1000_WRITE_REG(&adapter->hw, E1000_IVAR, adapter->ivars); } else if (adapter->intr_type == IFLIB_INTR_MSIX) /* Set up queue routing */ igb_configure_queues(adapter); /* this clears any pending interrupts */ E1000_READ_REG(&adapter->hw, E1000_ICR); E1000_WRITE_REG(&adapter->hw, E1000_ICS, E1000_ICS_LSC); /* AMT based hardware can now take control from firmware */ if (adapter->has_manage && adapter->has_amt) em_get_hw_control(adapter); /* Set Energy Efficient Ethernet */ if (adapter->hw.mac.type >= igb_mac_min && adapter->hw.phy.media_type == e1000_media_type_copper) { if (adapter->hw.mac.type == e1000_i354) e1000_set_eee_i354(&adapter->hw, TRUE, TRUE); else e1000_set_eee_i350(&adapter->hw, TRUE, TRUE); } } /********************************************************************* * * Fast Legacy/MSI Combined Interrupt Service routine * *********************************************************************/ int em_intr(void *arg) { struct adapter *adapter = arg; if_ctx_t ctx = adapter->ctx; u32 reg_icr; reg_icr = E1000_READ_REG(&adapter->hw, E1000_ICR); /* Hot eject? */ if (reg_icr == 0xffffffff) return FILTER_STRAY; /* Definitely not our interrupt. */ if (reg_icr == 0x0) return FILTER_STRAY; /* * Starting with the 82571 chip, bit 31 should be used to * determine whether the interrupt belongs to us. */ if (adapter->hw.mac.type >= e1000_82571 && (reg_icr & E1000_ICR_INT_ASSERTED) == 0) return FILTER_STRAY; /* * Only MSI-X interrupts have one-shot behavior by taking advantage * of the EIAC register. Thus, explicitly disable interrupts. This * also works around the MSI message reordering errata on certain * systems. */ IFDI_INTR_DISABLE(ctx); /* Link status change */ if (reg_icr & (E1000_ICR_RXSEQ | E1000_ICR_LSC)) em_handle_link(ctx); if (reg_icr & E1000_ICR_RXO) adapter->rx_overruns++; return (FILTER_SCHEDULE_THREAD); } static int em_if_rx_queue_intr_enable(if_ctx_t ctx, uint16_t rxqid) { struct adapter *adapter = iflib_get_softc(ctx); struct em_rx_queue *rxq = &adapter->rx_queues[rxqid]; E1000_WRITE_REG(&adapter->hw, E1000_IMS, rxq->eims); return (0); } static int em_if_tx_queue_intr_enable(if_ctx_t ctx, uint16_t txqid) { struct adapter *adapter = iflib_get_softc(ctx); struct em_tx_queue *txq = &adapter->tx_queues[txqid]; E1000_WRITE_REG(&adapter->hw, E1000_IMS, txq->eims); return (0); } static int igb_if_rx_queue_intr_enable(if_ctx_t ctx, uint16_t rxqid) { struct adapter *adapter = iflib_get_softc(ctx); struct em_rx_queue *rxq = &adapter->rx_queues[rxqid]; E1000_WRITE_REG(&adapter->hw, E1000_EIMS, rxq->eims); return (0); } static int igb_if_tx_queue_intr_enable(if_ctx_t ctx, uint16_t txqid) { struct adapter *adapter = iflib_get_softc(ctx); struct em_tx_queue *txq = &adapter->tx_queues[txqid]; E1000_WRITE_REG(&adapter->hw, E1000_EIMS, txq->eims); return (0); } /********************************************************************* * * MSI-X RX Interrupt Service routine * **********************************************************************/ static int em_msix_que(void *arg) { struct em_rx_queue *que = arg; ++que->irqs; return (FILTER_SCHEDULE_THREAD); } /********************************************************************* * * MSI-X Link Fast Interrupt Service routine * **********************************************************************/ static int em_msix_link(void *arg) { struct adapter *adapter = arg; u32 reg_icr; ++adapter->link_irq; MPASS(adapter->hw.back != NULL); reg_icr = E1000_READ_REG(&adapter->hw, E1000_ICR); if (reg_icr & E1000_ICR_RXO) adapter->rx_overruns++; if (reg_icr & (E1000_ICR_RXSEQ | E1000_ICR_LSC)) { em_handle_link(adapter->ctx); } else if (adapter->hw.mac.type == e1000_82574) { /* Only re-arm 82574 if em_if_update_admin_status() won't. */ E1000_WRITE_REG(&adapter->hw, E1000_IMS, EM_MSIX_LINK | E1000_IMS_LSC); } if (adapter->hw.mac.type == e1000_82574) { /* * Because we must read the ICR for this interrupt it may * clear other causes using autoclear, for this reason we * simply create a soft interrupt for all these vectors. */ if (reg_icr) E1000_WRITE_REG(&adapter->hw, E1000_ICS, adapter->ims); } else { /* Re-arm unconditionally */ E1000_WRITE_REG(&adapter->hw, E1000_IMS, E1000_IMS_LSC); E1000_WRITE_REG(&adapter->hw, E1000_EIMS, adapter->link_mask); } return (FILTER_HANDLED); } static void em_handle_link(void *context) { if_ctx_t ctx = context; struct adapter *adapter = iflib_get_softc(ctx); adapter->hw.mac.get_link_status = 1; iflib_admin_intr_deferred(ctx); } /********************************************************************* * * Media Ioctl callback * * This routine is called whenever the user queries the status of * the interface using ifconfig. * **********************************************************************/ static void em_if_media_status(if_ctx_t ctx, struct ifmediareq *ifmr) { struct adapter *adapter = iflib_get_softc(ctx); u_char fiber_type = IFM_1000_SX; INIT_DEBUGOUT("em_if_media_status: begin"); iflib_admin_intr_deferred(ctx); ifmr->ifm_status = IFM_AVALID; ifmr->ifm_active = IFM_ETHER; if (!adapter->link_active) { return; } ifmr->ifm_status |= IFM_ACTIVE; if ((adapter->hw.phy.media_type == e1000_media_type_fiber) || (adapter->hw.phy.media_type == e1000_media_type_internal_serdes)) { if (adapter->hw.mac.type == e1000_82545) fiber_type = IFM_1000_LX; ifmr->ifm_active |= fiber_type | IFM_FDX; } else { switch (adapter->link_speed) { case 10: ifmr->ifm_active |= IFM_10_T; break; case 100: ifmr->ifm_active |= IFM_100_TX; break; case 1000: ifmr->ifm_active |= IFM_1000_T; break; } if (adapter->link_duplex == FULL_DUPLEX) ifmr->ifm_active |= IFM_FDX; else ifmr->ifm_active |= IFM_HDX; } } /********************************************************************* * * Media Ioctl callback * * This routine is called when the user changes speed/duplex using * media/mediopt option with ifconfig. * **********************************************************************/ static int em_if_media_change(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); struct ifmedia *ifm = iflib_get_media(ctx); INIT_DEBUGOUT("em_if_media_change: begin"); if (IFM_TYPE(ifm->ifm_media) != IFM_ETHER) return (EINVAL); switch (IFM_SUBTYPE(ifm->ifm_media)) { case IFM_AUTO: adapter->hw.mac.autoneg = DO_AUTO_NEG; adapter->hw.phy.autoneg_advertised = AUTONEG_ADV_DEFAULT; break; case IFM_1000_LX: case IFM_1000_SX: case IFM_1000_T: adapter->hw.mac.autoneg = DO_AUTO_NEG; adapter->hw.phy.autoneg_advertised = ADVERTISE_1000_FULL; break; case IFM_100_TX: adapter->hw.mac.autoneg = FALSE; adapter->hw.phy.autoneg_advertised = 0; if ((ifm->ifm_media & IFM_GMASK) == IFM_FDX) adapter->hw.mac.forced_speed_duplex = ADVERTISE_100_FULL; else adapter->hw.mac.forced_speed_duplex = ADVERTISE_100_HALF; break; case IFM_10_T: adapter->hw.mac.autoneg = FALSE; adapter->hw.phy.autoneg_advertised = 0; if ((ifm->ifm_media & IFM_GMASK) == IFM_FDX) adapter->hw.mac.forced_speed_duplex = ADVERTISE_10_FULL; else adapter->hw.mac.forced_speed_duplex = ADVERTISE_10_HALF; break; default: device_printf(adapter->dev, "Unsupported media type\n"); } em_if_init(ctx); return (0); } static int em_if_set_promisc(if_ctx_t ctx, int flags) { struct adapter *adapter = iflib_get_softc(ctx); u32 reg_rctl; em_disable_promisc(ctx); reg_rctl = E1000_READ_REG(&adapter->hw, E1000_RCTL); if (flags & IFF_PROMISC) { reg_rctl |= (E1000_RCTL_UPE | E1000_RCTL_MPE); /* Turn this on if you want to see bad packets */ if (em_debug_sbp) reg_rctl |= E1000_RCTL_SBP; E1000_WRITE_REG(&adapter->hw, E1000_RCTL, reg_rctl); } else if (flags & IFF_ALLMULTI) { reg_rctl |= E1000_RCTL_MPE; reg_rctl &= ~E1000_RCTL_UPE; E1000_WRITE_REG(&adapter->hw, E1000_RCTL, reg_rctl); } return (0); } static void em_disable_promisc(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); struct ifnet *ifp = iflib_get_ifp(ctx); u32 reg_rctl; int mcnt = 0; reg_rctl = E1000_READ_REG(&adapter->hw, E1000_RCTL); reg_rctl &= (~E1000_RCTL_UPE); if (if_getflags(ifp) & IFF_ALLMULTI) mcnt = MAX_NUM_MULTICAST_ADDRESSES; else mcnt = if_llmaddr_count(ifp); /* Don't disable if in MAX groups */ if (mcnt < MAX_NUM_MULTICAST_ADDRESSES) reg_rctl &= (~E1000_RCTL_MPE); reg_rctl &= (~E1000_RCTL_SBP); E1000_WRITE_REG(&adapter->hw, E1000_RCTL, reg_rctl); } static u_int em_copy_maddr(void *arg, struct sockaddr_dl *sdl, u_int cnt) { u8 *mta = arg; if (cnt == MAX_NUM_MULTICAST_ADDRESSES) return (1); bcopy(LLADDR(sdl), &mta[cnt * ETHER_ADDR_LEN], ETHER_ADDR_LEN); return (1); } /********************************************************************* * Multicast Update * * This routine is called whenever multicast address list is updated. * **********************************************************************/ static void em_if_multi_set(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); struct ifnet *ifp = iflib_get_ifp(ctx); u32 reg_rctl = 0; u8 *mta; /* Multicast array memory */ int mcnt = 0; IOCTL_DEBUGOUT("em_set_multi: begin"); mta = adapter->mta; bzero(mta, sizeof(u8) * ETHER_ADDR_LEN * MAX_NUM_MULTICAST_ADDRESSES); if (adapter->hw.mac.type == e1000_82542 && adapter->hw.revision_id == E1000_REVISION_2) { reg_rctl = E1000_READ_REG(&adapter->hw, E1000_RCTL); if (adapter->hw.bus.pci_cmd_word & CMD_MEM_WRT_INVALIDATE) e1000_pci_clear_mwi(&adapter->hw); reg_rctl |= E1000_RCTL_RST; E1000_WRITE_REG(&adapter->hw, E1000_RCTL, reg_rctl); msec_delay(5); } mcnt = if_foreach_llmaddr(ifp, em_copy_maddr, mta); if (mcnt >= MAX_NUM_MULTICAST_ADDRESSES) { reg_rctl = E1000_READ_REG(&adapter->hw, E1000_RCTL); reg_rctl |= E1000_RCTL_MPE; E1000_WRITE_REG(&adapter->hw, E1000_RCTL, reg_rctl); } else e1000_update_mc_addr_list(&adapter->hw, mta, mcnt); if (adapter->hw.mac.type == e1000_82542 && adapter->hw.revision_id == E1000_REVISION_2) { reg_rctl = E1000_READ_REG(&adapter->hw, E1000_RCTL); reg_rctl &= ~E1000_RCTL_RST; E1000_WRITE_REG(&adapter->hw, E1000_RCTL, reg_rctl); msec_delay(5); if (adapter->hw.bus.pci_cmd_word & CMD_MEM_WRT_INVALIDATE) e1000_pci_set_mwi(&adapter->hw); } } /********************************************************************* * Timer routine * * This routine schedules em_if_update_admin_status() to check for * link status and to gather statistics as well as to perform some * controller-specific hardware patting. * **********************************************************************/ static void em_if_timer(if_ctx_t ctx, uint16_t qid) { if (qid != 0) return; iflib_admin_intr_deferred(ctx); } static void em_if_update_admin_status(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); struct e1000_hw *hw = &adapter->hw; device_t dev = iflib_get_dev(ctx); u32 link_check, thstat, ctrl; link_check = thstat = ctrl = 0; /* Get the cached link value or read phy for real */ switch (hw->phy.media_type) { case e1000_media_type_copper: if (hw->mac.get_link_status) { if (hw->mac.type == e1000_pch_spt) msec_delay(50); /* Do the work to read phy */ e1000_check_for_link(hw); link_check = !hw->mac.get_link_status; if (link_check) /* ESB2 fix */ e1000_cfg_on_link_up(hw); } else { link_check = TRUE; } break; case e1000_media_type_fiber: e1000_check_for_link(hw); link_check = (E1000_READ_REG(hw, E1000_STATUS) & E1000_STATUS_LU); break; case e1000_media_type_internal_serdes: e1000_check_for_link(hw); link_check = adapter->hw.mac.serdes_has_link; break; /* VF device is type_unknown */ case e1000_media_type_unknown: e1000_check_for_link(hw); link_check = !hw->mac.get_link_status; /* FALLTHROUGH */ default: break; } /* Check for thermal downshift or shutdown */ if (hw->mac.type == e1000_i350) { thstat = E1000_READ_REG(hw, E1000_THSTAT); ctrl = E1000_READ_REG(hw, E1000_CTRL_EXT); } /* Now check for a transition */ if (link_check && (adapter->link_active == 0)) { e1000_get_speed_and_duplex(hw, &adapter->link_speed, &adapter->link_duplex); /* Check if we must disable SPEED_MODE bit on PCI-E */ if ((adapter->link_speed != SPEED_1000) && ((hw->mac.type == e1000_82571) || (hw->mac.type == e1000_82572))) { int tarc0; tarc0 = E1000_READ_REG(hw, E1000_TARC(0)); tarc0 &= ~TARC_SPEED_MODE_BIT; E1000_WRITE_REG(hw, E1000_TARC(0), tarc0); } if (bootverbose) device_printf(dev, "Link is up %d Mbps %s\n", adapter->link_speed, ((adapter->link_duplex == FULL_DUPLEX) ? "Full Duplex" : "Half Duplex")); adapter->link_active = 1; adapter->smartspeed = 0; if ((ctrl & E1000_CTRL_EXT_LINK_MODE_MASK) == E1000_CTRL_EXT_LINK_MODE_GMII && (thstat & E1000_THSTAT_LINK_THROTTLE)) device_printf(dev, "Link: thermal downshift\n"); /* Delay Link Up for Phy update */ if (((hw->mac.type == e1000_i210) || (hw->mac.type == e1000_i211)) && (hw->phy.id == I210_I_PHY_ID)) msec_delay(I210_LINK_DELAY); /* Reset if the media type changed. */ if ((hw->dev_spec._82575.media_changed) && (adapter->hw.mac.type >= igb_mac_min)) { hw->dev_spec._82575.media_changed = false; adapter->flags |= IGB_MEDIA_RESET; em_reset(ctx); } iflib_link_state_change(ctx, LINK_STATE_UP, IF_Mbps(adapter->link_speed)); } else if (!link_check && (adapter->link_active == 1)) { adapter->link_speed = 0; adapter->link_duplex = 0; adapter->link_active = 0; iflib_link_state_change(ctx, LINK_STATE_DOWN, 0); } em_update_stats_counters(adapter); /* Reset LAA into RAR[0] on 82571 */ if (hw->mac.type == e1000_82571 && e1000_get_laa_state_82571(hw)) e1000_rar_set(hw, hw->mac.addr, 0); if (hw->mac.type < em_mac_min) lem_smartspeed(adapter); else if (hw->mac.type == e1000_82574 && adapter->intr_type == IFLIB_INTR_MSIX) E1000_WRITE_REG(&adapter->hw, E1000_IMS, EM_MSIX_LINK | E1000_IMS_LSC); } static void em_if_watchdog_reset(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); /* * Just count the event; iflib(4) will already trigger a * sufficient reset of the controller. */ adapter->watchdog_events++; } /********************************************************************* * * This routine disables all traffic on the adapter by issuing a * global reset on the MAC. * **********************************************************************/ static void em_if_stop(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); INIT_DEBUGOUT("em_if_stop: begin"); e1000_reset_hw(&adapter->hw); if (adapter->hw.mac.type >= e1000_82544) E1000_WRITE_REG(&adapter->hw, E1000_WUFC, 0); e1000_led_off(&adapter->hw); e1000_cleanup_led(&adapter->hw); } /********************************************************************* * * Determine hardware revision. * **********************************************************************/ static void em_identify_hardware(if_ctx_t ctx) { device_t dev = iflib_get_dev(ctx); struct adapter *adapter = iflib_get_softc(ctx); /* Make sure our PCI config space has the necessary stuff set */ adapter->hw.bus.pci_cmd_word = pci_read_config(dev, PCIR_COMMAND, 2); /* Save off the information about this board */ adapter->hw.vendor_id = pci_get_vendor(dev); adapter->hw.device_id = pci_get_device(dev); adapter->hw.revision_id = pci_read_config(dev, PCIR_REVID, 1); adapter->hw.subsystem_vendor_id = pci_read_config(dev, PCIR_SUBVEND_0, 2); adapter->hw.subsystem_device_id = pci_read_config(dev, PCIR_SUBDEV_0, 2); /* Do Shared Code Init and Setup */ if (e1000_set_mac_type(&adapter->hw)) { device_printf(dev, "Setup init failure\n"); return; } } static int em_allocate_pci_resources(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); device_t dev = iflib_get_dev(ctx); int rid, val; rid = PCIR_BAR(0); adapter->memory = bus_alloc_resource_any(dev, SYS_RES_MEMORY, &rid, RF_ACTIVE); if (adapter->memory == NULL) { device_printf(dev, "Unable to allocate bus resource: memory\n"); return (ENXIO); } adapter->osdep.mem_bus_space_tag = rman_get_bustag(adapter->memory); adapter->osdep.mem_bus_space_handle = rman_get_bushandle(adapter->memory); adapter->hw.hw_addr = (u8 *)&adapter->osdep.mem_bus_space_handle; /* Only older adapters use IO mapping */ if (adapter->hw.mac.type < em_mac_min && adapter->hw.mac.type > e1000_82543) { /* Figure our where our IO BAR is ? */ for (rid = PCIR_BAR(0); rid < PCIR_CIS;) { val = pci_read_config(dev, rid, 4); if (EM_BAR_TYPE(val) == EM_BAR_TYPE_IO) { break; } rid += 4; /* check for 64bit BAR */ if (EM_BAR_MEM_TYPE(val) == EM_BAR_MEM_TYPE_64BIT) rid += 4; } if (rid >= PCIR_CIS) { device_printf(dev, "Unable to locate IO BAR\n"); return (ENXIO); } adapter->ioport = bus_alloc_resource_any(dev, SYS_RES_IOPORT, &rid, RF_ACTIVE); if (adapter->ioport == NULL) { device_printf(dev, "Unable to allocate bus resource: " "ioport\n"); return (ENXIO); } adapter->hw.io_base = 0; adapter->osdep.io_bus_space_tag = rman_get_bustag(adapter->ioport); adapter->osdep.io_bus_space_handle = rman_get_bushandle(adapter->ioport); } adapter->hw.back = &adapter->osdep; return (0); } /********************************************************************* * * Set up the MSI-X Interrupt handlers * **********************************************************************/ static int em_if_msix_intr_assign(if_ctx_t ctx, int msix) { struct adapter *adapter = iflib_get_softc(ctx); struct em_rx_queue *rx_que = adapter->rx_queues; struct em_tx_queue *tx_que = adapter->tx_queues; int error, rid, i, vector = 0, rx_vectors; char buf[16]; /* First set up ring resources */ for (i = 0; i < adapter->rx_num_queues; i++, rx_que++, vector++) { rid = vector + 1; snprintf(buf, sizeof(buf), "rxq%d", i); error = iflib_irq_alloc_generic(ctx, &rx_que->que_irq, rid, IFLIB_INTR_RXTX, em_msix_que, rx_que, rx_que->me, buf); if (error) { device_printf(iflib_get_dev(ctx), "Failed to allocate que int %d err: %d", i, error); adapter->rx_num_queues = i + 1; goto fail; } rx_que->msix = vector; /* * Set the bit to enable interrupt * in E1000_IMS -- bits 20 and 21 * are for RX0 and RX1, note this has * NOTHING to do with the MSI-X vector */ if (adapter->hw.mac.type == e1000_82574) { rx_que->eims = 1 << (20 + i); adapter->ims |= rx_que->eims; adapter->ivars |= (8 | rx_que->msix) << (i * 4); } else if (adapter->hw.mac.type == e1000_82575) rx_que->eims = E1000_EICR_TX_QUEUE0 << vector; else rx_que->eims = 1 << vector; } rx_vectors = vector; vector = 0; for (i = 0; i < adapter->tx_num_queues; i++, tx_que++, vector++) { snprintf(buf, sizeof(buf), "txq%d", i); tx_que = &adapter->tx_queues[i]; iflib_softirq_alloc_generic(ctx, &adapter->rx_queues[i % adapter->rx_num_queues].que_irq, IFLIB_INTR_TX, tx_que, tx_que->me, buf); tx_que->msix = (vector % adapter->rx_num_queues); /* * Set the bit to enable interrupt * in E1000_IMS -- bits 22 and 23 * are for TX0 and TX1, note this has * NOTHING to do with the MSI-X vector */ if (adapter->hw.mac.type == e1000_82574) { tx_que->eims = 1 << (22 + i); adapter->ims |= tx_que->eims; adapter->ivars |= (8 | tx_que->msix) << (8 + (i * 4)); } else if (adapter->hw.mac.type == e1000_82575) { tx_que->eims = E1000_EICR_TX_QUEUE0 << i; } else { tx_que->eims = 1 << i; } } /* Link interrupt */ rid = rx_vectors + 1; error = iflib_irq_alloc_generic(ctx, &adapter->irq, rid, IFLIB_INTR_ADMIN, em_msix_link, adapter, 0, "aq"); if (error) { device_printf(iflib_get_dev(ctx), "Failed to register admin handler"); goto fail; } adapter->linkvec = rx_vectors; if (adapter->hw.mac.type < igb_mac_min) { adapter->ivars |= (8 | rx_vectors) << 16; adapter->ivars |= 0x80000000; } return (0); fail: iflib_irq_free(ctx, &adapter->irq); rx_que = adapter->rx_queues; for (int i = 0; i < adapter->rx_num_queues; i++, rx_que++) iflib_irq_free(ctx, &rx_que->que_irq); return (error); } static void igb_configure_queues(struct adapter *adapter) { struct e1000_hw *hw = &adapter->hw; struct em_rx_queue *rx_que; struct em_tx_queue *tx_que; u32 tmp, ivar = 0, newitr = 0; /* First turn on RSS capability */ if (adapter->hw.mac.type != e1000_82575) E1000_WRITE_REG(hw, E1000_GPIE, E1000_GPIE_MSIX_MODE | E1000_GPIE_EIAME | E1000_GPIE_PBA | E1000_GPIE_NSICR); /* Turn on MSI-X */ switch (adapter->hw.mac.type) { case e1000_82580: case e1000_i350: case e1000_i354: case e1000_i210: case e1000_i211: case e1000_vfadapt: case e1000_vfadapt_i350: /* RX entries */ for (int i = 0; i < adapter->rx_num_queues; i++) { u32 index = i >> 1; ivar = E1000_READ_REG_ARRAY(hw, E1000_IVAR0, index); rx_que = &adapter->rx_queues[i]; if (i & 1) { ivar &= 0xFF00FFFF; ivar |= (rx_que->msix | E1000_IVAR_VALID) << 16; } else { ivar &= 0xFFFFFF00; ivar |= rx_que->msix | E1000_IVAR_VALID; } E1000_WRITE_REG_ARRAY(hw, E1000_IVAR0, index, ivar); } /* TX entries */ for (int i = 0; i < adapter->tx_num_queues; i++) { u32 index = i >> 1; ivar = E1000_READ_REG_ARRAY(hw, E1000_IVAR0, index); tx_que = &adapter->tx_queues[i]; if (i & 1) { ivar &= 0x00FFFFFF; ivar |= (tx_que->msix | E1000_IVAR_VALID) << 24; } else { ivar &= 0xFFFF00FF; ivar |= (tx_que->msix | E1000_IVAR_VALID) << 8; } E1000_WRITE_REG_ARRAY(hw, E1000_IVAR0, index, ivar); adapter->que_mask |= tx_que->eims; } /* And for the link interrupt */ ivar = (adapter->linkvec | E1000_IVAR_VALID) << 8; adapter->link_mask = 1 << adapter->linkvec; E1000_WRITE_REG(hw, E1000_IVAR_MISC, ivar); break; case e1000_82576: /* RX entries */ for (int i = 0; i < adapter->rx_num_queues; i++) { u32 index = i & 0x7; /* Each IVAR has two entries */ ivar = E1000_READ_REG_ARRAY(hw, E1000_IVAR0, index); rx_que = &adapter->rx_queues[i]; if (i < 8) { ivar &= 0xFFFFFF00; ivar |= rx_que->msix | E1000_IVAR_VALID; } else { ivar &= 0xFF00FFFF; ivar |= (rx_que->msix | E1000_IVAR_VALID) << 16; } E1000_WRITE_REG_ARRAY(hw, E1000_IVAR0, index, ivar); adapter->que_mask |= rx_que->eims; } /* TX entries */ for (int i = 0; i < adapter->tx_num_queues; i++) { u32 index = i & 0x7; /* Each IVAR has two entries */ ivar = E1000_READ_REG_ARRAY(hw, E1000_IVAR0, index); tx_que = &adapter->tx_queues[i]; if (i < 8) { ivar &= 0xFFFF00FF; ivar |= (tx_que->msix | E1000_IVAR_VALID) << 8; } else { ivar &= 0x00FFFFFF; ivar |= (tx_que->msix | E1000_IVAR_VALID) << 24; } E1000_WRITE_REG_ARRAY(hw, E1000_IVAR0, index, ivar); adapter->que_mask |= tx_que->eims; } /* And for the link interrupt */ ivar = (adapter->linkvec | E1000_IVAR_VALID) << 8; adapter->link_mask = 1 << adapter->linkvec; E1000_WRITE_REG(hw, E1000_IVAR_MISC, ivar); break; case e1000_82575: /* enable MSI-X support*/ tmp = E1000_READ_REG(hw, E1000_CTRL_EXT); tmp |= E1000_CTRL_EXT_PBA_CLR; /* Auto-Mask interrupts upon ICR read. */ tmp |= E1000_CTRL_EXT_EIAME; tmp |= E1000_CTRL_EXT_IRCA; E1000_WRITE_REG(hw, E1000_CTRL_EXT, tmp); /* Queues */ for (int i = 0; i < adapter->rx_num_queues; i++) { rx_que = &adapter->rx_queues[i]; tmp = E1000_EICR_RX_QUEUE0 << i; tmp |= E1000_EICR_TX_QUEUE0 << i; rx_que->eims = tmp; E1000_WRITE_REG_ARRAY(hw, E1000_MSIXBM(0), i, rx_que->eims); adapter->que_mask |= rx_que->eims; } /* Link */ E1000_WRITE_REG(hw, E1000_MSIXBM(adapter->linkvec), E1000_EIMS_OTHER); adapter->link_mask |= E1000_EIMS_OTHER; default: break; } /* Set the starting interrupt rate */ if (em_max_interrupt_rate > 0) newitr = (4000000 / em_max_interrupt_rate) & 0x7FFC; if (hw->mac.type == e1000_82575) newitr |= newitr << 16; else newitr |= E1000_EITR_CNT_IGNR; for (int i = 0; i < adapter->rx_num_queues; i++) { rx_que = &adapter->rx_queues[i]; E1000_WRITE_REG(hw, E1000_EITR(rx_que->msix), newitr); } return; } static void em_free_pci_resources(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); struct em_rx_queue *que = adapter->rx_queues; device_t dev = iflib_get_dev(ctx); /* Release all MSI-X queue resources */ if (adapter->intr_type == IFLIB_INTR_MSIX) iflib_irq_free(ctx, &adapter->irq); for (int i = 0; i < adapter->rx_num_queues; i++, que++) { iflib_irq_free(ctx, &que->que_irq); } if (adapter->memory != NULL) { bus_release_resource(dev, SYS_RES_MEMORY, rman_get_rid(adapter->memory), adapter->memory); adapter->memory = NULL; } if (adapter->flash != NULL) { bus_release_resource(dev, SYS_RES_MEMORY, rman_get_rid(adapter->flash), adapter->flash); adapter->flash = NULL; } if (adapter->ioport != NULL) { bus_release_resource(dev, SYS_RES_IOPORT, rman_get_rid(adapter->ioport), adapter->ioport); adapter->ioport = NULL; } } /* Set up MSI or MSI-X */ static int em_setup_msix(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); if (adapter->hw.mac.type == e1000_82574) { em_enable_vectors_82574(ctx); } return (0); } /********************************************************************* * * Workaround for SmartSpeed on 82541 and 82547 controllers * **********************************************************************/ static void lem_smartspeed(struct adapter *adapter) { u16 phy_tmp; if (adapter->link_active || (adapter->hw.phy.type != e1000_phy_igp) || adapter->hw.mac.autoneg == 0 || (adapter->hw.phy.autoneg_advertised & ADVERTISE_1000_FULL) == 0) return; if (adapter->smartspeed == 0) { /* If Master/Slave config fault is asserted twice, * we assume back-to-back */ e1000_read_phy_reg(&adapter->hw, PHY_1000T_STATUS, &phy_tmp); if (!(phy_tmp & SR_1000T_MS_CONFIG_FAULT)) return; e1000_read_phy_reg(&adapter->hw, PHY_1000T_STATUS, &phy_tmp); if (phy_tmp & SR_1000T_MS_CONFIG_FAULT) { e1000_read_phy_reg(&adapter->hw, PHY_1000T_CTRL, &phy_tmp); if(phy_tmp & CR_1000T_MS_ENABLE) { phy_tmp &= ~CR_1000T_MS_ENABLE; e1000_write_phy_reg(&adapter->hw, PHY_1000T_CTRL, phy_tmp); adapter->smartspeed++; if(adapter->hw.mac.autoneg && !e1000_copper_link_autoneg(&adapter->hw) && !e1000_read_phy_reg(&adapter->hw, PHY_CONTROL, &phy_tmp)) { phy_tmp |= (MII_CR_AUTO_NEG_EN | MII_CR_RESTART_AUTO_NEG); e1000_write_phy_reg(&adapter->hw, PHY_CONTROL, phy_tmp); } } } return; } else if(adapter->smartspeed == EM_SMARTSPEED_DOWNSHIFT) { /* If still no link, perhaps using 2/3 pair cable */ e1000_read_phy_reg(&adapter->hw, PHY_1000T_CTRL, &phy_tmp); phy_tmp |= CR_1000T_MS_ENABLE; e1000_write_phy_reg(&adapter->hw, PHY_1000T_CTRL, phy_tmp); if(adapter->hw.mac.autoneg && !e1000_copper_link_autoneg(&adapter->hw) && !e1000_read_phy_reg(&adapter->hw, PHY_CONTROL, &phy_tmp)) { phy_tmp |= (MII_CR_AUTO_NEG_EN | MII_CR_RESTART_AUTO_NEG); e1000_write_phy_reg(&adapter->hw, PHY_CONTROL, phy_tmp); } } /* Restart process after EM_SMARTSPEED_MAX iterations */ if(adapter->smartspeed++ == EM_SMARTSPEED_MAX) adapter->smartspeed = 0; } /********************************************************************* * * Initialize the DMA Coalescing feature * **********************************************************************/ static void igb_init_dmac(struct adapter *adapter, u32 pba) { device_t dev = adapter->dev; struct e1000_hw *hw = &adapter->hw; u32 dmac, reg = ~E1000_DMACR_DMAC_EN; u16 hwm; u16 max_frame_size; if (hw->mac.type == e1000_i211) return; max_frame_size = adapter->shared->isc_max_frame_size; if (hw->mac.type > e1000_82580) { if (adapter->dmac == 0) { /* Disabling it */ E1000_WRITE_REG(hw, E1000_DMACR, reg); return; } else device_printf(dev, "DMA Coalescing enabled\n"); /* Set starting threshold */ E1000_WRITE_REG(hw, E1000_DMCTXTH, 0); hwm = 64 * pba - max_frame_size / 16; if (hwm < 64 * (pba - 6)) hwm = 64 * (pba - 6); reg = E1000_READ_REG(hw, E1000_FCRTC); reg &= ~E1000_FCRTC_RTH_COAL_MASK; reg |= ((hwm << E1000_FCRTC_RTH_COAL_SHIFT) & E1000_FCRTC_RTH_COAL_MASK); E1000_WRITE_REG(hw, E1000_FCRTC, reg); dmac = pba - max_frame_size / 512; if (dmac < pba - 10) dmac = pba - 10; reg = E1000_READ_REG(hw, E1000_DMACR); reg &= ~E1000_DMACR_DMACTHR_MASK; reg |= ((dmac << E1000_DMACR_DMACTHR_SHIFT) & E1000_DMACR_DMACTHR_MASK); /* transition to L0x or L1 if available..*/ reg |= (E1000_DMACR_DMAC_EN | E1000_DMACR_DMAC_LX_MASK); /* Check if status is 2.5Gb backplane connection * before configuration of watchdog timer, which is * in msec values in 12.8usec intervals * watchdog timer= msec values in 32usec intervals * for non 2.5Gb connection */ if (hw->mac.type == e1000_i354) { int status = E1000_READ_REG(hw, E1000_STATUS); if ((status & E1000_STATUS_2P5_SKU) && (!(status & E1000_STATUS_2P5_SKU_OVER))) reg |= ((adapter->dmac * 5) >> 6); else reg |= (adapter->dmac >> 5); } else { reg |= (adapter->dmac >> 5); } E1000_WRITE_REG(hw, E1000_DMACR, reg); E1000_WRITE_REG(hw, E1000_DMCRTRH, 0); /* Set the interval before transition */ reg = E1000_READ_REG(hw, E1000_DMCTLX); if (hw->mac.type == e1000_i350) reg |= IGB_DMCTLX_DCFLUSH_DIS; /* ** in 2.5Gb connection, TTLX unit is 0.4 usec ** which is 0x4*2 = 0xA. But delay is still 4 usec */ if (hw->mac.type == e1000_i354) { int status = E1000_READ_REG(hw, E1000_STATUS); if ((status & E1000_STATUS_2P5_SKU) && (!(status & E1000_STATUS_2P5_SKU_OVER))) reg |= 0xA; else reg |= 0x4; } else { reg |= 0x4; } E1000_WRITE_REG(hw, E1000_DMCTLX, reg); /* free space in tx packet buffer to wake from DMA coal */ E1000_WRITE_REG(hw, E1000_DMCTXTH, (IGB_TXPBSIZE - (2 * max_frame_size)) >> 6); /* make low power state decision controlled by DMA coal */ reg = E1000_READ_REG(hw, E1000_PCIEMISC); reg &= ~E1000_PCIEMISC_LX_DECISION; E1000_WRITE_REG(hw, E1000_PCIEMISC, reg); } else if (hw->mac.type == e1000_82580) { u32 reg = E1000_READ_REG(hw, E1000_PCIEMISC); E1000_WRITE_REG(hw, E1000_PCIEMISC, reg & ~E1000_PCIEMISC_LX_DECISION); E1000_WRITE_REG(hw, E1000_DMACR, 0); } } /********************************************************************* * * Initialize the hardware to a configuration as specified by the * adapter structure. * **********************************************************************/ static void em_reset(if_ctx_t ctx) { device_t dev = iflib_get_dev(ctx); struct adapter *adapter = iflib_get_softc(ctx); struct ifnet *ifp = iflib_get_ifp(ctx); struct e1000_hw *hw = &adapter->hw; u16 rx_buffer_size; u32 pba; INIT_DEBUGOUT("em_reset: begin"); /* Let the firmware know the OS is in control */ em_get_hw_control(adapter); /* Set up smart power down as default off on newer adapters. */ if (!em_smart_pwr_down && (hw->mac.type == e1000_82571 || hw->mac.type == e1000_82572)) { u16 phy_tmp = 0; /* Speed up time to link by disabling smart power down. */ e1000_read_phy_reg(hw, IGP02E1000_PHY_POWER_MGMT, &phy_tmp); phy_tmp &= ~IGP02E1000_PM_SPD; e1000_write_phy_reg(hw, IGP02E1000_PHY_POWER_MGMT, phy_tmp); } /* * Packet Buffer Allocation (PBA) * Writing PBA sets the receive portion of the buffer * the remainder is used for the transmit buffer. */ switch (hw->mac.type) { /* Total Packet Buffer on these is 48K */ case e1000_82571: case e1000_82572: case e1000_80003es2lan: pba = E1000_PBA_32K; /* 32K for Rx, 16K for Tx */ break; case e1000_82573: /* 82573: Total Packet Buffer is 32K */ pba = E1000_PBA_12K; /* 12K for Rx, 20K for Tx */ break; case e1000_82574: case e1000_82583: pba = E1000_PBA_20K; /* 20K for Rx, 20K for Tx */ break; case e1000_ich8lan: pba = E1000_PBA_8K; break; case e1000_ich9lan: case e1000_ich10lan: /* Boost Receive side for jumbo frames */ if (adapter->hw.mac.max_frame_size > 4096) pba = E1000_PBA_14K; else pba = E1000_PBA_10K; break; case e1000_pchlan: case e1000_pch2lan: case e1000_pch_lpt: case e1000_pch_spt: case e1000_pch_cnp: pba = E1000_PBA_26K; break; case e1000_82575: pba = E1000_PBA_32K; break; case e1000_82576: case e1000_vfadapt: pba = E1000_READ_REG(hw, E1000_RXPBS); pba &= E1000_RXPBS_SIZE_MASK_82576; break; case e1000_82580: case e1000_i350: case e1000_i354: case e1000_vfadapt_i350: pba = E1000_READ_REG(hw, E1000_RXPBS); pba = e1000_rxpbs_adjust_82580(pba); break; case e1000_i210: case e1000_i211: pba = E1000_PBA_34K; break; default: if (adapter->hw.mac.max_frame_size > 8192) pba = E1000_PBA_40K; /* 40K for Rx, 24K for Tx */ else pba = E1000_PBA_48K; /* 48K for Rx, 16K for Tx */ } /* Special needs in case of Jumbo frames */ if ((hw->mac.type == e1000_82575) && (ifp->if_mtu > ETHERMTU)) { u32 tx_space, min_tx, min_rx; pba = E1000_READ_REG(hw, E1000_PBA); tx_space = pba >> 16; pba &= 0xffff; min_tx = (adapter->hw.mac.max_frame_size + sizeof(struct e1000_tx_desc) - ETHERNET_FCS_SIZE) * 2; min_tx = roundup2(min_tx, 1024); min_tx >>= 10; min_rx = adapter->hw.mac.max_frame_size; min_rx = roundup2(min_rx, 1024); min_rx >>= 10; if (tx_space < min_tx && ((min_tx - tx_space) < pba)) { pba = pba - (min_tx - tx_space); /* * if short on rx space, rx wins * and must trump tx adjustment */ if (pba < min_rx) pba = min_rx; } E1000_WRITE_REG(hw, E1000_PBA, pba); } if (hw->mac.type < igb_mac_min) E1000_WRITE_REG(&adapter->hw, E1000_PBA, pba); INIT_DEBUGOUT1("em_reset: pba=%dK",pba); /* * These parameters control the automatic generation (Tx) and * response (Rx) to Ethernet PAUSE frames. * - High water mark should allow for at least two frames to be * received after sending an XOFF. * - Low water mark works best when it is very near the high water mark. * This allows the receiver to restart by sending XON when it has * drained a bit. Here we use an arbitrary value of 1500 which will * restart after one full frame is pulled from the buffer. There * could be several smaller frames in the buffer and if so they will * not trigger the XON until their total number reduces the buffer * by 1500. * - The pause time is fairly large at 1000 x 512ns = 512 usec. */ rx_buffer_size = (pba & 0xffff) << 10; hw->fc.high_water = rx_buffer_size - roundup2(adapter->hw.mac.max_frame_size, 1024); hw->fc.low_water = hw->fc.high_water - 1500; if (adapter->fc) /* locally set flow control value? */ hw->fc.requested_mode = adapter->fc; else hw->fc.requested_mode = e1000_fc_full; if (hw->mac.type == e1000_80003es2lan) hw->fc.pause_time = 0xFFFF; else hw->fc.pause_time = EM_FC_PAUSE_TIME; hw->fc.send_xon = TRUE; /* Device specific overrides/settings */ switch (hw->mac.type) { case e1000_pchlan: /* Workaround: no TX flow ctrl for PCH */ hw->fc.requested_mode = e1000_fc_rx_pause; hw->fc.pause_time = 0xFFFF; /* override */ if (if_getmtu(ifp) > ETHERMTU) { hw->fc.high_water = 0x3500; hw->fc.low_water = 0x1500; } else { hw->fc.high_water = 0x5000; hw->fc.low_water = 0x3000; } hw->fc.refresh_time = 0x1000; break; case e1000_pch2lan: case e1000_pch_lpt: case e1000_pch_spt: case e1000_pch_cnp: hw->fc.high_water = 0x5C20; hw->fc.low_water = 0x5048; hw->fc.pause_time = 0x0650; hw->fc.refresh_time = 0x0400; /* Jumbos need adjusted PBA */ if (if_getmtu(ifp) > ETHERMTU) E1000_WRITE_REG(hw, E1000_PBA, 12); else E1000_WRITE_REG(hw, E1000_PBA, 26); break; case e1000_82575: case e1000_82576: /* 8-byte granularity */ hw->fc.low_water = hw->fc.high_water - 8; break; case e1000_82580: case e1000_i350: case e1000_i354: case e1000_i210: case e1000_i211: case e1000_vfadapt: case e1000_vfadapt_i350: /* 16-byte granularity */ hw->fc.low_water = hw->fc.high_water - 16; break; case e1000_ich9lan: case e1000_ich10lan: if (if_getmtu(ifp) > ETHERMTU) { hw->fc.high_water = 0x2800; hw->fc.low_water = hw->fc.high_water - 8; break; } /* FALLTHROUGH */ default: if (hw->mac.type == e1000_80003es2lan) hw->fc.pause_time = 0xFFFF; break; } /* Issue a global reset */ e1000_reset_hw(hw); if (adapter->hw.mac.type >= igb_mac_min) { E1000_WRITE_REG(hw, E1000_WUC, 0); } else { E1000_WRITE_REG(hw, E1000_WUFC, 0); em_disable_aspm(adapter); } if (adapter->flags & IGB_MEDIA_RESET) { e1000_setup_init_funcs(hw, TRUE); e1000_get_bus_info(hw); adapter->flags &= ~IGB_MEDIA_RESET; } /* and a re-init */ if (e1000_init_hw(hw) < 0) { device_printf(dev, "Hardware Initialization Failed\n"); return; } if (adapter->hw.mac.type >= igb_mac_min) igb_init_dmac(adapter, pba); E1000_WRITE_REG(hw, E1000_VET, ETHERTYPE_VLAN); e1000_get_phy_info(hw); e1000_check_for_link(hw); } /* * Initialise the RSS mapping for NICs that support multiple transmit/ * receive rings. */ #define RSSKEYLEN 10 static void em_initialize_rss_mapping(struct adapter *adapter) { uint8_t rss_key[4 * RSSKEYLEN]; uint32_t reta = 0; struct e1000_hw *hw = &adapter->hw; int i; /* * Configure RSS key */ arc4rand(rss_key, sizeof(rss_key), 0); for (i = 0; i < RSSKEYLEN; ++i) { uint32_t rssrk = 0; rssrk = EM_RSSRK_VAL(rss_key, i); E1000_WRITE_REG(hw,E1000_RSSRK(i), rssrk); } /* * Configure RSS redirect table in following fashion: * (hash & ring_cnt_mask) == rdr_table[(hash & rdr_table_mask)] */ for (i = 0; i < sizeof(reta); ++i) { uint32_t q; q = (i % adapter->rx_num_queues) << 7; reta |= q << (8 * i); } for (i = 0; i < 32; ++i) E1000_WRITE_REG(hw, E1000_RETA(i), reta); E1000_WRITE_REG(hw, E1000_MRQC, E1000_MRQC_RSS_ENABLE_2Q | E1000_MRQC_RSS_FIELD_IPV4_TCP | E1000_MRQC_RSS_FIELD_IPV4 | E1000_MRQC_RSS_FIELD_IPV6_TCP_EX | E1000_MRQC_RSS_FIELD_IPV6_EX | E1000_MRQC_RSS_FIELD_IPV6); } static void igb_initialize_rss_mapping(struct adapter *adapter) { struct e1000_hw *hw = &adapter->hw; int i; int queue_id; u32 reta; u32 rss_key[10], mrqc, shift = 0; /* XXX? */ if (adapter->hw.mac.type == e1000_82575) shift = 6; /* * The redirection table controls which destination * queue each bucket redirects traffic to. * Each DWORD represents four queues, with the LSB * being the first queue in the DWORD. * * This just allocates buckets to queues using round-robin * allocation. * * NOTE: It Just Happens to line up with the default * RSS allocation method. */ /* Warning FM follows */ reta = 0; for (i = 0; i < 128; i++) { #ifdef RSS queue_id = rss_get_indirection_to_bucket(i); /* * If we have more queues than buckets, we'll * end up mapping buckets to a subset of the * queues. * * If we have more buckets than queues, we'll * end up instead assigning multiple buckets * to queues. * * Both are suboptimal, but we need to handle * the case so we don't go out of bounds * indexing arrays and such. */ queue_id = queue_id % adapter->rx_num_queues; #else queue_id = (i % adapter->rx_num_queues); #endif /* Adjust if required */ queue_id = queue_id << shift; /* * The low 8 bits are for hash value (n+0); * The next 8 bits are for hash value (n+1), etc. */ reta = reta >> 8; reta = reta | ( ((uint32_t) queue_id) << 24); if ((i & 3) == 3) { E1000_WRITE_REG(hw, E1000_RETA(i >> 2), reta); reta = 0; } } /* Now fill in hash table */ /* * MRQC: Multiple Receive Queues Command * Set queuing to RSS control, number depends on the device. */ mrqc = E1000_MRQC_ENABLE_RSS_8Q; #ifdef RSS /* XXX ew typecasting */ rss_getkey((uint8_t *) &rss_key); #else arc4rand(&rss_key, sizeof(rss_key), 0); #endif for (i = 0; i < 10; i++) E1000_WRITE_REG_ARRAY(hw, E1000_RSSRK(0), i, rss_key[i]); /* * Configure the RSS fields to hash upon. */ mrqc |= (E1000_MRQC_RSS_FIELD_IPV4 | E1000_MRQC_RSS_FIELD_IPV4_TCP); mrqc |= (E1000_MRQC_RSS_FIELD_IPV6 | E1000_MRQC_RSS_FIELD_IPV6_TCP); mrqc |=( E1000_MRQC_RSS_FIELD_IPV4_UDP | E1000_MRQC_RSS_FIELD_IPV6_UDP); mrqc |=( E1000_MRQC_RSS_FIELD_IPV6_UDP_EX | E1000_MRQC_RSS_FIELD_IPV6_TCP_EX); E1000_WRITE_REG(hw, E1000_MRQC, mrqc); } /********************************************************************* * * Setup networking device structure and register interface media. * **********************************************************************/ static int em_setup_interface(if_ctx_t ctx) { struct ifnet *ifp = iflib_get_ifp(ctx); struct adapter *adapter = iflib_get_softc(ctx); if_softc_ctx_t scctx = adapter->shared; INIT_DEBUGOUT("em_setup_interface: begin"); /* Single Queue */ if (adapter->tx_num_queues == 1) { if_setsendqlen(ifp, scctx->isc_ntxd[0] - 1); if_setsendqready(ifp); } /* * Specify the media types supported by this adapter and register * callbacks to update media and link information */ if ((adapter->hw.phy.media_type == e1000_media_type_fiber) || (adapter->hw.phy.media_type == e1000_media_type_internal_serdes)) { u_char fiber_type = IFM_1000_SX; /* default type */ if (adapter->hw.mac.type == e1000_82545) fiber_type = IFM_1000_LX; ifmedia_add(adapter->media, IFM_ETHER | fiber_type | IFM_FDX, 0, NULL); ifmedia_add(adapter->media, IFM_ETHER | fiber_type, 0, NULL); } else { ifmedia_add(adapter->media, IFM_ETHER | IFM_10_T, 0, NULL); ifmedia_add(adapter->media, IFM_ETHER | IFM_10_T | IFM_FDX, 0, NULL); ifmedia_add(adapter->media, IFM_ETHER | IFM_100_TX, 0, NULL); ifmedia_add(adapter->media, IFM_ETHER | IFM_100_TX | IFM_FDX, 0, NULL); if (adapter->hw.phy.type != e1000_phy_ife) { ifmedia_add(adapter->media, IFM_ETHER | IFM_1000_T | IFM_FDX, 0, NULL); ifmedia_add(adapter->media, IFM_ETHER | IFM_1000_T, 0, NULL); } } ifmedia_add(adapter->media, IFM_ETHER | IFM_AUTO, 0, NULL); ifmedia_set(adapter->media, IFM_ETHER | IFM_AUTO); return (0); } static int em_if_tx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs, int ntxqs, int ntxqsets) { struct adapter *adapter = iflib_get_softc(ctx); if_softc_ctx_t scctx = adapter->shared; int error = E1000_SUCCESS; struct em_tx_queue *que; int i, j; MPASS(adapter->tx_num_queues > 0); MPASS(adapter->tx_num_queues == ntxqsets); /* First allocate the top level queue structs */ if (!(adapter->tx_queues = (struct em_tx_queue *) malloc(sizeof(struct em_tx_queue) * adapter->tx_num_queues, M_DEVBUF, M_NOWAIT | M_ZERO))) { device_printf(iflib_get_dev(ctx), "Unable to allocate queue memory\n"); return(ENOMEM); } for (i = 0, que = adapter->tx_queues; i < adapter->tx_num_queues; i++, que++) { /* Set up some basics */ struct tx_ring *txr = &que->txr; txr->adapter = que->adapter = adapter; que->me = txr->me = i; /* Allocate report status array */ if (!(txr->tx_rsq = (qidx_t *) malloc(sizeof(qidx_t) * scctx->isc_ntxd[0], M_DEVBUF, M_NOWAIT | M_ZERO))) { device_printf(iflib_get_dev(ctx), "failed to allocate rs_idxs memory\n"); error = ENOMEM; goto fail; } for (j = 0; j < scctx->isc_ntxd[0]; j++) txr->tx_rsq[j] = QIDX_INVALID; /* get the virtual and physical address of the hardware queues */ txr->tx_base = (struct e1000_tx_desc *)vaddrs[i*ntxqs]; txr->tx_paddr = paddrs[i*ntxqs]; } if (bootverbose) device_printf(iflib_get_dev(ctx), "allocated for %d tx_queues\n", adapter->tx_num_queues); return (0); fail: em_if_queues_free(ctx); return (error); } static int em_if_rx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs, int nrxqs, int nrxqsets) { struct adapter *adapter = iflib_get_softc(ctx); int error = E1000_SUCCESS; struct em_rx_queue *que; int i; MPASS(adapter->rx_num_queues > 0); MPASS(adapter->rx_num_queues == nrxqsets); /* First allocate the top level queue structs */ if (!(adapter->rx_queues = (struct em_rx_queue *) malloc(sizeof(struct em_rx_queue) * adapter->rx_num_queues, M_DEVBUF, M_NOWAIT | M_ZERO))) { device_printf(iflib_get_dev(ctx), "Unable to allocate queue memory\n"); error = ENOMEM; goto fail; } for (i = 0, que = adapter->rx_queues; i < nrxqsets; i++, que++) { /* Set up some basics */ struct rx_ring *rxr = &que->rxr; rxr->adapter = que->adapter = adapter; rxr->que = que; que->me = rxr->me = i; /* get the virtual and physical address of the hardware queues */ rxr->rx_base = (union e1000_rx_desc_extended *)vaddrs[i*nrxqs]; rxr->rx_paddr = paddrs[i*nrxqs]; } if (bootverbose) device_printf(iflib_get_dev(ctx), "allocated for %d rx_queues\n", adapter->rx_num_queues); return (0); fail: em_if_queues_free(ctx); return (error); } static void em_if_queues_free(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); struct em_tx_queue *tx_que = adapter->tx_queues; struct em_rx_queue *rx_que = adapter->rx_queues; if (tx_que != NULL) { for (int i = 0; i < adapter->tx_num_queues; i++, tx_que++) { struct tx_ring *txr = &tx_que->txr; if (txr->tx_rsq == NULL) break; free(txr->tx_rsq, M_DEVBUF); txr->tx_rsq = NULL; } free(adapter->tx_queues, M_DEVBUF); adapter->tx_queues = NULL; } if (rx_que != NULL) { free(adapter->rx_queues, M_DEVBUF); adapter->rx_queues = NULL; } em_release_hw_control(adapter); if (adapter->mta != NULL) { free(adapter->mta, M_DEVBUF); } } /********************************************************************* * * Enable transmit unit. * **********************************************************************/ static void em_initialize_transmit_unit(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); if_softc_ctx_t scctx = adapter->shared; struct em_tx_queue *que; struct tx_ring *txr; struct e1000_hw *hw = &adapter->hw; u32 tctl, txdctl = 0, tarc, tipg = 0; INIT_DEBUGOUT("em_initialize_transmit_unit: begin"); for (int i = 0; i < adapter->tx_num_queues; i++, txr++) { u64 bus_addr; caddr_t offp, endp; que = &adapter->tx_queues[i]; txr = &que->txr; bus_addr = txr->tx_paddr; /* Clear checksum offload context. */ offp = (caddr_t)&txr->csum_flags; endp = (caddr_t)(txr + 1); bzero(offp, endp - offp); /* Base and Len of TX Ring */ E1000_WRITE_REG(hw, E1000_TDLEN(i), scctx->isc_ntxd[0] * sizeof(struct e1000_tx_desc)); E1000_WRITE_REG(hw, E1000_TDBAH(i), (u32)(bus_addr >> 32)); E1000_WRITE_REG(hw, E1000_TDBAL(i), (u32)bus_addr); /* Init the HEAD/TAIL indices */ E1000_WRITE_REG(hw, E1000_TDT(i), 0); E1000_WRITE_REG(hw, E1000_TDH(i), 0); HW_DEBUGOUT2("Base = %x, Length = %x\n", E1000_READ_REG(&adapter->hw, E1000_TDBAL(i)), E1000_READ_REG(&adapter->hw, E1000_TDLEN(i))); txdctl = 0; /* clear txdctl */ txdctl |= 0x1f; /* PTHRESH */ txdctl |= 1 << 8; /* HTHRESH */ txdctl |= 1 << 16;/* WTHRESH */ txdctl |= 1 << 22; /* Reserved bit 22 must always be 1 */ txdctl |= E1000_TXDCTL_GRAN; txdctl |= 1 << 25; /* LWTHRESH */ E1000_WRITE_REG(hw, E1000_TXDCTL(i), txdctl); } /* Set the default values for the Tx Inter Packet Gap timer */ switch (adapter->hw.mac.type) { case e1000_80003es2lan: tipg = DEFAULT_82543_TIPG_IPGR1; tipg |= DEFAULT_80003ES2LAN_TIPG_IPGR2 << E1000_TIPG_IPGR2_SHIFT; break; case e1000_82542: tipg = DEFAULT_82542_TIPG_IPGT; tipg |= DEFAULT_82542_TIPG_IPGR1 << E1000_TIPG_IPGR1_SHIFT; tipg |= DEFAULT_82542_TIPG_IPGR2 << E1000_TIPG_IPGR2_SHIFT; break; default: if ((adapter->hw.phy.media_type == e1000_media_type_fiber) || (adapter->hw.phy.media_type == e1000_media_type_internal_serdes)) tipg = DEFAULT_82543_TIPG_IPGT_FIBER; else tipg = DEFAULT_82543_TIPG_IPGT_COPPER; tipg |= DEFAULT_82543_TIPG_IPGR1 << E1000_TIPG_IPGR1_SHIFT; tipg |= DEFAULT_82543_TIPG_IPGR2 << E1000_TIPG_IPGR2_SHIFT; } E1000_WRITE_REG(&adapter->hw, E1000_TIPG, tipg); E1000_WRITE_REG(&adapter->hw, E1000_TIDV, adapter->tx_int_delay.value); if(adapter->hw.mac.type >= e1000_82540) E1000_WRITE_REG(&adapter->hw, E1000_TADV, adapter->tx_abs_int_delay.value); if ((adapter->hw.mac.type == e1000_82571) || (adapter->hw.mac.type == e1000_82572)) { tarc = E1000_READ_REG(&adapter->hw, E1000_TARC(0)); tarc |= TARC_SPEED_MODE_BIT; E1000_WRITE_REG(&adapter->hw, E1000_TARC(0), tarc); } else if (adapter->hw.mac.type == e1000_80003es2lan) { /* errata: program both queues to unweighted RR */ tarc = E1000_READ_REG(&adapter->hw, E1000_TARC(0)); tarc |= 1; E1000_WRITE_REG(&adapter->hw, E1000_TARC(0), tarc); tarc = E1000_READ_REG(&adapter->hw, E1000_TARC(1)); tarc |= 1; E1000_WRITE_REG(&adapter->hw, E1000_TARC(1), tarc); } else if (adapter->hw.mac.type == e1000_82574) { tarc = E1000_READ_REG(&adapter->hw, E1000_TARC(0)); tarc |= TARC_ERRATA_BIT; if ( adapter->tx_num_queues > 1) { tarc |= (TARC_COMPENSATION_MODE | TARC_MQ_FIX); E1000_WRITE_REG(&adapter->hw, E1000_TARC(0), tarc); E1000_WRITE_REG(&adapter->hw, E1000_TARC(1), tarc); } else E1000_WRITE_REG(&adapter->hw, E1000_TARC(0), tarc); } if (adapter->tx_int_delay.value > 0) adapter->txd_cmd |= E1000_TXD_CMD_IDE; /* Program the Transmit Control Register */ tctl = E1000_READ_REG(&adapter->hw, E1000_TCTL); tctl &= ~E1000_TCTL_CT; tctl |= (E1000_TCTL_PSP | E1000_TCTL_RTLC | E1000_TCTL_EN | (E1000_COLLISION_THRESHOLD << E1000_CT_SHIFT)); if (adapter->hw.mac.type >= e1000_82571) tctl |= E1000_TCTL_MULR; /* This write will effectively turn on the transmit unit. */ E1000_WRITE_REG(&adapter->hw, E1000_TCTL, tctl); /* SPT and KBL errata workarounds */ if (hw->mac.type == e1000_pch_spt) { u32 reg; reg = E1000_READ_REG(hw, E1000_IOSFPC); reg |= E1000_RCTL_RDMTS_HEX; E1000_WRITE_REG(hw, E1000_IOSFPC, reg); /* i218-i219 Specification Update 1.5.4.5 */ reg = E1000_READ_REG(hw, E1000_TARC(0)); reg &= ~E1000_TARC0_CB_MULTIQ_3_REQ; reg |= E1000_TARC0_CB_MULTIQ_2_REQ; E1000_WRITE_REG(hw, E1000_TARC(0), reg); } } /********************************************************************* * * Enable receive unit. * **********************************************************************/ static void em_initialize_receive_unit(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); if_softc_ctx_t scctx = adapter->shared; struct ifnet *ifp = iflib_get_ifp(ctx); struct e1000_hw *hw = &adapter->hw; struct em_rx_queue *que; int i; u32 rctl, rxcsum, rfctl; INIT_DEBUGOUT("em_initialize_receive_units: begin"); /* * Make sure receives are disabled while setting * up the descriptor ring */ rctl = E1000_READ_REG(hw, E1000_RCTL); /* Do not disable if ever enabled on this hardware */ if ((hw->mac.type != e1000_82574) && (hw->mac.type != e1000_82583)) E1000_WRITE_REG(hw, E1000_RCTL, rctl & ~E1000_RCTL_EN); /* Setup the Receive Control Register */ rctl &= ~(3 << E1000_RCTL_MO_SHIFT); rctl |= E1000_RCTL_EN | E1000_RCTL_BAM | E1000_RCTL_LBM_NO | E1000_RCTL_RDMTS_HALF | (hw->mac.mc_filter_type << E1000_RCTL_MO_SHIFT); /* Do not store bad packets */ rctl &= ~E1000_RCTL_SBP; /* Enable Long Packet receive */ if (if_getmtu(ifp) > ETHERMTU) rctl |= E1000_RCTL_LPE; else rctl &= ~E1000_RCTL_LPE; /* Strip the CRC */ if (!em_disable_crc_stripping) rctl |= E1000_RCTL_SECRC; if (adapter->hw.mac.type >= e1000_82540) { E1000_WRITE_REG(&adapter->hw, E1000_RADV, adapter->rx_abs_int_delay.value); /* * Set the interrupt throttling rate. Value is calculated * as DEFAULT_ITR = 1/(MAX_INTS_PER_SEC * 256ns) */ E1000_WRITE_REG(hw, E1000_ITR, DEFAULT_ITR); } E1000_WRITE_REG(&adapter->hw, E1000_RDTR, adapter->rx_int_delay.value); /* Use extended rx descriptor formats */ rfctl = E1000_READ_REG(hw, E1000_RFCTL); rfctl |= E1000_RFCTL_EXTEN; /* * When using MSI-X interrupts we need to throttle * using the EITR register (82574 only) */ if (hw->mac.type == e1000_82574) { for (int i = 0; i < 4; i++) E1000_WRITE_REG(hw, E1000_EITR_82574(i), DEFAULT_ITR); /* Disable accelerated acknowledge */ rfctl |= E1000_RFCTL_ACK_DIS; } E1000_WRITE_REG(hw, E1000_RFCTL, rfctl); rxcsum = E1000_READ_REG(hw, E1000_RXCSUM); if (if_getcapenable(ifp) & IFCAP_RXCSUM && adapter->hw.mac.type >= e1000_82543) { if (adapter->tx_num_queues > 1) { if (adapter->hw.mac.type >= igb_mac_min) { rxcsum |= E1000_RXCSUM_PCSD; if (hw->mac.type != e1000_82575) rxcsum |= E1000_RXCSUM_CRCOFL; } else rxcsum |= E1000_RXCSUM_TUOFL | E1000_RXCSUM_IPOFL | E1000_RXCSUM_PCSD; } else { if (adapter->hw.mac.type >= igb_mac_min) rxcsum |= E1000_RXCSUM_IPPCSE; else rxcsum |= E1000_RXCSUM_TUOFL | E1000_RXCSUM_IPOFL; if (adapter->hw.mac.type > e1000_82575) rxcsum |= E1000_RXCSUM_CRCOFL; } } else rxcsum &= ~E1000_RXCSUM_TUOFL; E1000_WRITE_REG(hw, E1000_RXCSUM, rxcsum); if (adapter->rx_num_queues > 1) { if (adapter->hw.mac.type >= igb_mac_min) igb_initialize_rss_mapping(adapter); else em_initialize_rss_mapping(adapter); } /* * XXX TEMPORARY WORKAROUND: on some systems with 82573 * long latencies are observed, like Lenovo X60. This * change eliminates the problem, but since having positive * values in RDTR is a known source of problems on other * platforms another solution is being sought. */ if (hw->mac.type == e1000_82573) E1000_WRITE_REG(hw, E1000_RDTR, 0x20); for (i = 0, que = adapter->rx_queues; i < adapter->rx_num_queues; i++, que++) { struct rx_ring *rxr = &que->rxr; /* Setup the Base and Length of the Rx Descriptor Ring */ u64 bus_addr = rxr->rx_paddr; #if 0 u32 rdt = adapter->rx_num_queues -1; /* default */ #endif E1000_WRITE_REG(hw, E1000_RDLEN(i), scctx->isc_nrxd[0] * sizeof(union e1000_rx_desc_extended)); E1000_WRITE_REG(hw, E1000_RDBAH(i), (u32)(bus_addr >> 32)); E1000_WRITE_REG(hw, E1000_RDBAL(i), (u32)bus_addr); /* Setup the Head and Tail Descriptor Pointers */ E1000_WRITE_REG(hw, E1000_RDH(i), 0); E1000_WRITE_REG(hw, E1000_RDT(i), 0); } /* * Set PTHRESH for improved jumbo performance * According to 10.2.5.11 of Intel 82574 Datasheet, * RXDCTL(1) is written whenever RXDCTL(0) is written. * Only write to RXDCTL(1) if there is a need for different * settings. */ if (((adapter->hw.mac.type == e1000_ich9lan) || (adapter->hw.mac.type == e1000_pch2lan) || (adapter->hw.mac.type == e1000_ich10lan)) && (if_getmtu(ifp) > ETHERMTU)) { u32 rxdctl = E1000_READ_REG(hw, E1000_RXDCTL(0)); E1000_WRITE_REG(hw, E1000_RXDCTL(0), rxdctl | 3); } else if (adapter->hw.mac.type == e1000_82574) { for (int i = 0; i < adapter->rx_num_queues; i++) { u32 rxdctl = E1000_READ_REG(hw, E1000_RXDCTL(i)); rxdctl |= 0x20; /* PTHRESH */ rxdctl |= 4 << 8; /* HTHRESH */ rxdctl |= 4 << 16;/* WTHRESH */ rxdctl |= 1 << 24; /* Switch to granularity */ E1000_WRITE_REG(hw, E1000_RXDCTL(i), rxdctl); } } else if (adapter->hw.mac.type >= igb_mac_min) { u32 psize, srrctl = 0; if (if_getmtu(ifp) > ETHERMTU) { /* Set maximum packet len */ if (adapter->rx_mbuf_sz <= 4096) { srrctl |= 4096 >> E1000_SRRCTL_BSIZEPKT_SHIFT; rctl |= E1000_RCTL_SZ_4096 | E1000_RCTL_BSEX; } else if (adapter->rx_mbuf_sz > 4096) { srrctl |= 8192 >> E1000_SRRCTL_BSIZEPKT_SHIFT; rctl |= E1000_RCTL_SZ_8192 | E1000_RCTL_BSEX; } psize = scctx->isc_max_frame_size; /* are we on a vlan? */ if (ifp->if_vlantrunk != NULL) psize += VLAN_TAG_SIZE; E1000_WRITE_REG(&adapter->hw, E1000_RLPML, psize); } else { srrctl |= 2048 >> E1000_SRRCTL_BSIZEPKT_SHIFT; rctl |= E1000_RCTL_SZ_2048; } /* * If TX flow control is disabled and there's >1 queue defined, * enable DROP. * * This drops frames rather than hanging the RX MAC for all queues. */ if ((adapter->rx_num_queues > 1) && (adapter->fc == e1000_fc_none || adapter->fc == e1000_fc_rx_pause)) { srrctl |= E1000_SRRCTL_DROP_EN; } /* Setup the Base and Length of the Rx Descriptor Rings */ for (i = 0, que = adapter->rx_queues; i < adapter->rx_num_queues; i++, que++) { struct rx_ring *rxr = &que->rxr; u64 bus_addr = rxr->rx_paddr; u32 rxdctl; #ifdef notyet /* Configure for header split? -- ignore for now */ rxr->hdr_split = igb_header_split; #else srrctl |= E1000_SRRCTL_DESCTYPE_ADV_ONEBUF; #endif E1000_WRITE_REG(hw, E1000_RDLEN(i), scctx->isc_nrxd[0] * sizeof(struct e1000_rx_desc)); E1000_WRITE_REG(hw, E1000_RDBAH(i), (uint32_t)(bus_addr >> 32)); E1000_WRITE_REG(hw, E1000_RDBAL(i), (uint32_t)bus_addr); E1000_WRITE_REG(hw, E1000_SRRCTL(i), srrctl); /* Enable this Queue */ rxdctl = E1000_READ_REG(hw, E1000_RXDCTL(i)); rxdctl |= E1000_RXDCTL_QUEUE_ENABLE; rxdctl &= 0xFFF00000; rxdctl |= IGB_RX_PTHRESH; rxdctl |= IGB_RX_HTHRESH << 8; rxdctl |= IGB_RX_WTHRESH << 16; E1000_WRITE_REG(hw, E1000_RXDCTL(i), rxdctl); } } else if (adapter->hw.mac.type >= e1000_pch2lan) { if (if_getmtu(ifp) > ETHERMTU) e1000_lv_jumbo_workaround_ich8lan(hw, TRUE); else e1000_lv_jumbo_workaround_ich8lan(hw, FALSE); } /* Make sure VLAN Filters are off */ rctl &= ~E1000_RCTL_VFE; if (adapter->hw.mac.type < igb_mac_min) { if (adapter->rx_mbuf_sz == MCLBYTES) rctl |= E1000_RCTL_SZ_2048; else if (adapter->rx_mbuf_sz == MJUMPAGESIZE) rctl |= E1000_RCTL_SZ_4096 | E1000_RCTL_BSEX; else if (adapter->rx_mbuf_sz > MJUMPAGESIZE) rctl |= E1000_RCTL_SZ_8192 | E1000_RCTL_BSEX; /* ensure we clear use DTYPE of 00 here */ rctl &= ~0x00000C00; } /* Write out the settings */ E1000_WRITE_REG(hw, E1000_RCTL, rctl); return; } static void em_if_vlan_register(if_ctx_t ctx, u16 vtag) { struct adapter *adapter = iflib_get_softc(ctx); u32 index, bit; index = (vtag >> 5) & 0x7F; bit = vtag & 0x1F; adapter->shadow_vfta[index] |= (1 << bit); ++adapter->num_vlans; } static void em_if_vlan_unregister(if_ctx_t ctx, u16 vtag) { struct adapter *adapter = iflib_get_softc(ctx); u32 index, bit; index = (vtag >> 5) & 0x7F; bit = vtag & 0x1F; adapter->shadow_vfta[index] &= ~(1 << bit); --adapter->num_vlans; } static void em_setup_vlan_hw_support(struct adapter *adapter) { struct e1000_hw *hw = &adapter->hw; u32 reg; /* * We get here thru init_locked, meaning * a soft reset, this has already cleared * the VFTA and other state, so if there * have been no vlan's registered do nothing. */ if (adapter->num_vlans == 0) return; /* * A soft reset zero's out the VFTA, so * we need to repopulate it now. */ for (int i = 0; i < EM_VFTA_SIZE; i++) if (adapter->shadow_vfta[i] != 0) E1000_WRITE_REG_ARRAY(hw, E1000_VFTA, i, adapter->shadow_vfta[i]); reg = E1000_READ_REG(hw, E1000_CTRL); reg |= E1000_CTRL_VME; E1000_WRITE_REG(hw, E1000_CTRL, reg); /* Enable the Filter Table */ reg = E1000_READ_REG(hw, E1000_RCTL); reg &= ~E1000_RCTL_CFIEN; reg |= E1000_RCTL_VFE; E1000_WRITE_REG(hw, E1000_RCTL, reg); } static void em_if_intr_enable(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); struct e1000_hw *hw = &adapter->hw; u32 ims_mask = IMS_ENABLE_MASK; if (hw->mac.type == e1000_82574) { E1000_WRITE_REG(hw, EM_EIAC, EM_MSIX_MASK); ims_mask |= adapter->ims; } E1000_WRITE_REG(hw, E1000_IMS, ims_mask); } static void em_if_intr_disable(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); struct e1000_hw *hw = &adapter->hw; if (hw->mac.type == e1000_82574) E1000_WRITE_REG(hw, EM_EIAC, 0); E1000_WRITE_REG(hw, E1000_IMC, 0xffffffff); } static void igb_if_intr_enable(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); struct e1000_hw *hw = &adapter->hw; u32 mask; if (__predict_true(adapter->intr_type == IFLIB_INTR_MSIX)) { mask = (adapter->que_mask | adapter->link_mask); E1000_WRITE_REG(hw, E1000_EIAC, mask); E1000_WRITE_REG(hw, E1000_EIAM, mask); E1000_WRITE_REG(hw, E1000_EIMS, mask); E1000_WRITE_REG(hw, E1000_IMS, E1000_IMS_LSC); } else E1000_WRITE_REG(hw, E1000_IMS, IMS_ENABLE_MASK); E1000_WRITE_FLUSH(hw); } static void igb_if_intr_disable(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); struct e1000_hw *hw = &adapter->hw; if (__predict_true(adapter->intr_type == IFLIB_INTR_MSIX)) { E1000_WRITE_REG(hw, E1000_EIMC, 0xffffffff); E1000_WRITE_REG(hw, E1000_EIAC, 0); } E1000_WRITE_REG(hw, E1000_IMC, 0xffffffff); E1000_WRITE_FLUSH(hw); } /* * Bit of a misnomer, what this really means is * to enable OS management of the system... aka * to disable special hardware management features */ static void em_init_manageability(struct adapter *adapter) { /* A shared code workaround */ #define E1000_82542_MANC2H E1000_MANC2H if (adapter->has_manage) { int manc2h = E1000_READ_REG(&adapter->hw, E1000_MANC2H); int manc = E1000_READ_REG(&adapter->hw, E1000_MANC); /* disable hardware interception of ARP */ manc &= ~(E1000_MANC_ARP_EN); /* enable receiving management packets to the host */ manc |= E1000_MANC_EN_MNG2HOST; #define E1000_MNG2HOST_PORT_623 (1 << 5) #define E1000_MNG2HOST_PORT_664 (1 << 6) manc2h |= E1000_MNG2HOST_PORT_623; manc2h |= E1000_MNG2HOST_PORT_664; E1000_WRITE_REG(&adapter->hw, E1000_MANC2H, manc2h); E1000_WRITE_REG(&adapter->hw, E1000_MANC, manc); } } /* * Give control back to hardware management * controller if there is one. */ static void em_release_manageability(struct adapter *adapter) { if (adapter->has_manage) { int manc = E1000_READ_REG(&adapter->hw, E1000_MANC); /* re-enable hardware interception of ARP */ manc |= E1000_MANC_ARP_EN; manc &= ~E1000_MANC_EN_MNG2HOST; E1000_WRITE_REG(&adapter->hw, E1000_MANC, manc); } } /* * em_get_hw_control sets the {CTRL_EXT|FWSM}:DRV_LOAD bit. * For ASF and Pass Through versions of f/w this means * that the driver is loaded. For AMT version type f/w * this means that the network i/f is open. */ static void em_get_hw_control(struct adapter *adapter) { u32 ctrl_ext, swsm; if (adapter->vf_ifp) return; if (adapter->hw.mac.type == e1000_82573) { swsm = E1000_READ_REG(&adapter->hw, E1000_SWSM); E1000_WRITE_REG(&adapter->hw, E1000_SWSM, swsm | E1000_SWSM_DRV_LOAD); return; } /* else */ ctrl_ext = E1000_READ_REG(&adapter->hw, E1000_CTRL_EXT); E1000_WRITE_REG(&adapter->hw, E1000_CTRL_EXT, ctrl_ext | E1000_CTRL_EXT_DRV_LOAD); } /* * em_release_hw_control resets {CTRL_EXT|FWSM}:DRV_LOAD bit. * For ASF and Pass Through versions of f/w this means that * the driver is no longer loaded. For AMT versions of the * f/w this means that the network i/f is closed. */ static void em_release_hw_control(struct adapter *adapter) { u32 ctrl_ext, swsm; if (!adapter->has_manage) return; if (adapter->hw.mac.type == e1000_82573) { swsm = E1000_READ_REG(&adapter->hw, E1000_SWSM); E1000_WRITE_REG(&adapter->hw, E1000_SWSM, swsm & ~E1000_SWSM_DRV_LOAD); return; } /* else */ ctrl_ext = E1000_READ_REG(&adapter->hw, E1000_CTRL_EXT); E1000_WRITE_REG(&adapter->hw, E1000_CTRL_EXT, ctrl_ext & ~E1000_CTRL_EXT_DRV_LOAD); return; } static int em_is_valid_ether_addr(u8 *addr) { char zero_addr[6] = { 0, 0, 0, 0, 0, 0 }; if ((addr[0] & 1) || (!bcmp(addr, zero_addr, ETHER_ADDR_LEN))) { return (FALSE); } return (TRUE); } /* ** Parse the interface capabilities with regard ** to both system management and wake-on-lan for ** later use. */ static void em_get_wakeup(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); device_t dev = iflib_get_dev(ctx); u16 eeprom_data = 0, device_id, apme_mask; adapter->has_manage = e1000_enable_mng_pass_thru(&adapter->hw); apme_mask = EM_EEPROM_APME; switch (adapter->hw.mac.type) { case e1000_82542: case e1000_82543: break; case e1000_82544: e1000_read_nvm(&adapter->hw, NVM_INIT_CONTROL2_REG, 1, &eeprom_data); apme_mask = EM_82544_APME; break; case e1000_82546: case e1000_82546_rev_3: if (adapter->hw.bus.func == 1) { e1000_read_nvm(&adapter->hw, NVM_INIT_CONTROL3_PORT_B, 1, &eeprom_data); break; } else e1000_read_nvm(&adapter->hw, NVM_INIT_CONTROL3_PORT_A, 1, &eeprom_data); break; case e1000_82573: case e1000_82583: adapter->has_amt = TRUE; /* FALLTHROUGH */ case e1000_82571: case e1000_82572: case e1000_80003es2lan: if (adapter->hw.bus.func == 1) { e1000_read_nvm(&adapter->hw, NVM_INIT_CONTROL3_PORT_B, 1, &eeprom_data); break; } else e1000_read_nvm(&adapter->hw, NVM_INIT_CONTROL3_PORT_A, 1, &eeprom_data); break; case e1000_ich8lan: case e1000_ich9lan: case e1000_ich10lan: case e1000_pchlan: case e1000_pch2lan: case e1000_pch_lpt: case e1000_pch_spt: case e1000_82575: /* listing all igb devices */ case e1000_82576: case e1000_82580: case e1000_i350: case e1000_i354: case e1000_i210: case e1000_i211: case e1000_vfadapt: case e1000_vfadapt_i350: apme_mask = E1000_WUC_APME; adapter->has_amt = TRUE; eeprom_data = E1000_READ_REG(&adapter->hw, E1000_WUC); break; default: e1000_read_nvm(&adapter->hw, NVM_INIT_CONTROL3_PORT_A, 1, &eeprom_data); break; } if (eeprom_data & apme_mask) adapter->wol = (E1000_WUFC_MAG | E1000_WUFC_MC); /* * We have the eeprom settings, now apply the special cases * where the eeprom may be wrong or the board won't support * wake on lan on a particular port */ device_id = pci_get_device(dev); switch (device_id) { case E1000_DEV_ID_82546GB_PCIE: adapter->wol = 0; break; case E1000_DEV_ID_82546EB_FIBER: case E1000_DEV_ID_82546GB_FIBER: /* Wake events only supported on port A for dual fiber * regardless of eeprom setting */ if (E1000_READ_REG(&adapter->hw, E1000_STATUS) & E1000_STATUS_FUNC_1) adapter->wol = 0; break; case E1000_DEV_ID_82546GB_QUAD_COPPER_KSP3: /* if quad port adapter, disable WoL on all but port A */ if (global_quad_port_a != 0) adapter->wol = 0; /* Reset for multiple quad port adapters */ if (++global_quad_port_a == 4) global_quad_port_a = 0; break; case E1000_DEV_ID_82571EB_FIBER: /* Wake events only supported on port A for dual fiber * regardless of eeprom setting */ if (E1000_READ_REG(&adapter->hw, E1000_STATUS) & E1000_STATUS_FUNC_1) adapter->wol = 0; break; case E1000_DEV_ID_82571EB_QUAD_COPPER: case E1000_DEV_ID_82571EB_QUAD_FIBER: case E1000_DEV_ID_82571EB_QUAD_COPPER_LP: /* if quad port adapter, disable WoL on all but port A */ if (global_quad_port_a != 0) adapter->wol = 0; /* Reset for multiple quad port adapters */ if (++global_quad_port_a == 4) global_quad_port_a = 0; break; } return; } /* * Enable PCI Wake On Lan capability */ static void em_enable_wakeup(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); device_t dev = iflib_get_dev(ctx); if_t ifp = iflib_get_ifp(ctx); int error = 0; u32 pmc, ctrl, ctrl_ext, rctl; u16 status; if (pci_find_cap(dev, PCIY_PMG, &pmc) != 0) return; /* * Determine type of Wakeup: note that wol * is set with all bits on by default. */ if ((if_getcapenable(ifp) & IFCAP_WOL_MAGIC) == 0) adapter->wol &= ~E1000_WUFC_MAG; if ((if_getcapenable(ifp) & IFCAP_WOL_UCAST) == 0) adapter->wol &= ~E1000_WUFC_EX; if ((if_getcapenable(ifp) & IFCAP_WOL_MCAST) == 0) adapter->wol &= ~E1000_WUFC_MC; else { rctl = E1000_READ_REG(&adapter->hw, E1000_RCTL); rctl |= E1000_RCTL_MPE; E1000_WRITE_REG(&adapter->hw, E1000_RCTL, rctl); } if (!(adapter->wol & (E1000_WUFC_EX | E1000_WUFC_MAG | E1000_WUFC_MC))) goto pme; /* Advertise the wakeup capability */ ctrl = E1000_READ_REG(&adapter->hw, E1000_CTRL); ctrl |= (E1000_CTRL_SWDPIN2 | E1000_CTRL_SWDPIN3); E1000_WRITE_REG(&adapter->hw, E1000_CTRL, ctrl); /* Keep the laser running on Fiber adapters */ if (adapter->hw.phy.media_type == e1000_media_type_fiber || adapter->hw.phy.media_type == e1000_media_type_internal_serdes) { ctrl_ext = E1000_READ_REG(&adapter->hw, E1000_CTRL_EXT); ctrl_ext |= E1000_CTRL_EXT_SDP3_DATA; E1000_WRITE_REG(&adapter->hw, E1000_CTRL_EXT, ctrl_ext); } if ((adapter->hw.mac.type == e1000_ich8lan) || (adapter->hw.mac.type == e1000_pchlan) || (adapter->hw.mac.type == e1000_ich9lan) || (adapter->hw.mac.type == e1000_ich10lan)) e1000_suspend_workarounds_ich8lan(&adapter->hw); if ( adapter->hw.mac.type >= e1000_pchlan) { error = em_enable_phy_wakeup(adapter); if (error) goto pme; } else { /* Enable wakeup by the MAC */ E1000_WRITE_REG(&adapter->hw, E1000_WUC, E1000_WUC_PME_EN); E1000_WRITE_REG(&adapter->hw, E1000_WUFC, adapter->wol); } if (adapter->hw.phy.type == e1000_phy_igp_3) e1000_igp3_phy_powerdown_workaround_ich8lan(&adapter->hw); pme: status = pci_read_config(dev, pmc + PCIR_POWER_STATUS, 2); status &= ~(PCIM_PSTAT_PME | PCIM_PSTAT_PMEENABLE); if (!error && (if_getcapenable(ifp) & IFCAP_WOL)) status |= PCIM_PSTAT_PME | PCIM_PSTAT_PMEENABLE; pci_write_config(dev, pmc + PCIR_POWER_STATUS, status, 2); return; } /* * WOL in the newer chipset interfaces (pchlan) * require thing to be copied into the phy */ static int em_enable_phy_wakeup(struct adapter *adapter) { struct e1000_hw *hw = &adapter->hw; u32 mreg, ret = 0; u16 preg; /* copy MAC RARs to PHY RARs */ e1000_copy_rx_addrs_to_phy_ich8lan(hw); /* copy MAC MTA to PHY MTA */ for (int i = 0; i < adapter->hw.mac.mta_reg_count; i++) { mreg = E1000_READ_REG_ARRAY(hw, E1000_MTA, i); e1000_write_phy_reg(hw, BM_MTA(i), (u16)(mreg & 0xFFFF)); e1000_write_phy_reg(hw, BM_MTA(i) + 1, (u16)((mreg >> 16) & 0xFFFF)); } /* configure PHY Rx Control register */ e1000_read_phy_reg(&adapter->hw, BM_RCTL, &preg); mreg = E1000_READ_REG(hw, E1000_RCTL); if (mreg & E1000_RCTL_UPE) preg |= BM_RCTL_UPE; if (mreg & E1000_RCTL_MPE) preg |= BM_RCTL_MPE; preg &= ~(BM_RCTL_MO_MASK); if (mreg & E1000_RCTL_MO_3) preg |= (((mreg & E1000_RCTL_MO_3) >> E1000_RCTL_MO_SHIFT) << BM_RCTL_MO_SHIFT); if (mreg & E1000_RCTL_BAM) preg |= BM_RCTL_BAM; if (mreg & E1000_RCTL_PMCF) preg |= BM_RCTL_PMCF; mreg = E1000_READ_REG(hw, E1000_CTRL); if (mreg & E1000_CTRL_RFCE) preg |= BM_RCTL_RFCE; e1000_write_phy_reg(&adapter->hw, BM_RCTL, preg); /* enable PHY wakeup in MAC register */ E1000_WRITE_REG(hw, E1000_WUC, E1000_WUC_PHY_WAKE | E1000_WUC_PME_EN | E1000_WUC_APME); E1000_WRITE_REG(hw, E1000_WUFC, adapter->wol); /* configure and enable PHY wakeup in PHY registers */ e1000_write_phy_reg(&adapter->hw, BM_WUFC, adapter->wol); e1000_write_phy_reg(&adapter->hw, BM_WUC, E1000_WUC_PME_EN); /* activate PHY wakeup */ ret = hw->phy.ops.acquire(hw); if (ret) { printf("Could not acquire PHY\n"); return ret; } e1000_write_phy_reg_mdic(hw, IGP01E1000_PHY_PAGE_SELECT, (BM_WUC_ENABLE_PAGE << IGP_PAGE_SHIFT)); ret = e1000_read_phy_reg_mdic(hw, BM_WUC_ENABLE_REG, &preg); if (ret) { printf("Could not read PHY page 769\n"); goto out; } preg |= BM_WUC_ENABLE_BIT | BM_WUC_HOST_WU_BIT; ret = e1000_write_phy_reg_mdic(hw, BM_WUC_ENABLE_REG, preg); if (ret) printf("Could not set PHY Host Wakeup bit\n"); out: hw->phy.ops.release(hw); return ret; } static void em_if_led_func(if_ctx_t ctx, int onoff) { struct adapter *adapter = iflib_get_softc(ctx); if (onoff) { e1000_setup_led(&adapter->hw); e1000_led_on(&adapter->hw); } else { e1000_led_off(&adapter->hw); e1000_cleanup_led(&adapter->hw); } } /* * Disable the L0S and L1 LINK states */ static void em_disable_aspm(struct adapter *adapter) { int base, reg; u16 link_cap,link_ctrl; device_t dev = adapter->dev; switch (adapter->hw.mac.type) { case e1000_82573: case e1000_82574: case e1000_82583: break; default: return; } if (pci_find_cap(dev, PCIY_EXPRESS, &base) != 0) return; reg = base + PCIER_LINK_CAP; link_cap = pci_read_config(dev, reg, 2); if ((link_cap & PCIEM_LINK_CAP_ASPM) == 0) return; reg = base + PCIER_LINK_CTL; link_ctrl = pci_read_config(dev, reg, 2); link_ctrl &= ~PCIEM_LINK_CTL_ASPMC; pci_write_config(dev, reg, link_ctrl, 2); return; } /********************************************************************** * * Update the board statistics counters. * **********************************************************************/ static void em_update_stats_counters(struct adapter *adapter) { u64 prev_xoffrxc = adapter->stats.xoffrxc; if(adapter->hw.phy.media_type == e1000_media_type_copper || (E1000_READ_REG(&adapter->hw, E1000_STATUS) & E1000_STATUS_LU)) { adapter->stats.symerrs += E1000_READ_REG(&adapter->hw, E1000_SYMERRS); adapter->stats.sec += E1000_READ_REG(&adapter->hw, E1000_SEC); } adapter->stats.crcerrs += E1000_READ_REG(&adapter->hw, E1000_CRCERRS); adapter->stats.mpc += E1000_READ_REG(&adapter->hw, E1000_MPC); adapter->stats.scc += E1000_READ_REG(&adapter->hw, E1000_SCC); adapter->stats.ecol += E1000_READ_REG(&adapter->hw, E1000_ECOL); adapter->stats.mcc += E1000_READ_REG(&adapter->hw, E1000_MCC); adapter->stats.latecol += E1000_READ_REG(&adapter->hw, E1000_LATECOL); adapter->stats.colc += E1000_READ_REG(&adapter->hw, E1000_COLC); adapter->stats.dc += E1000_READ_REG(&adapter->hw, E1000_DC); adapter->stats.rlec += E1000_READ_REG(&adapter->hw, E1000_RLEC); adapter->stats.xonrxc += E1000_READ_REG(&adapter->hw, E1000_XONRXC); adapter->stats.xontxc += E1000_READ_REG(&adapter->hw, E1000_XONTXC); adapter->stats.xoffrxc += E1000_READ_REG(&adapter->hw, E1000_XOFFRXC); /* ** For watchdog management we need to know if we have been ** paused during the last interval, so capture that here. */ if (adapter->stats.xoffrxc != prev_xoffrxc) adapter->shared->isc_pause_frames = 1; adapter->stats.xofftxc += E1000_READ_REG(&adapter->hw, E1000_XOFFTXC); adapter->stats.fcruc += E1000_READ_REG(&adapter->hw, E1000_FCRUC); adapter->stats.prc64 += E1000_READ_REG(&adapter->hw, E1000_PRC64); adapter->stats.prc127 += E1000_READ_REG(&adapter->hw, E1000_PRC127); adapter->stats.prc255 += E1000_READ_REG(&adapter->hw, E1000_PRC255); adapter->stats.prc511 += E1000_READ_REG(&adapter->hw, E1000_PRC511); adapter->stats.prc1023 += E1000_READ_REG(&adapter->hw, E1000_PRC1023); adapter->stats.prc1522 += E1000_READ_REG(&adapter->hw, E1000_PRC1522); adapter->stats.gprc += E1000_READ_REG(&adapter->hw, E1000_GPRC); adapter->stats.bprc += E1000_READ_REG(&adapter->hw, E1000_BPRC); adapter->stats.mprc += E1000_READ_REG(&adapter->hw, E1000_MPRC); adapter->stats.gptc += E1000_READ_REG(&adapter->hw, E1000_GPTC); /* For the 64-bit byte counters the low dword must be read first. */ /* Both registers clear on the read of the high dword */ adapter->stats.gorc += E1000_READ_REG(&adapter->hw, E1000_GORCL) + ((u64)E1000_READ_REG(&adapter->hw, E1000_GORCH) << 32); adapter->stats.gotc += E1000_READ_REG(&adapter->hw, E1000_GOTCL) + ((u64)E1000_READ_REG(&adapter->hw, E1000_GOTCH) << 32); adapter->stats.rnbc += E1000_READ_REG(&adapter->hw, E1000_RNBC); adapter->stats.ruc += E1000_READ_REG(&adapter->hw, E1000_RUC); adapter->stats.rfc += E1000_READ_REG(&adapter->hw, E1000_RFC); adapter->stats.roc += E1000_READ_REG(&adapter->hw, E1000_ROC); adapter->stats.rjc += E1000_READ_REG(&adapter->hw, E1000_RJC); adapter->stats.tor += E1000_READ_REG(&adapter->hw, E1000_TORH); adapter->stats.tot += E1000_READ_REG(&adapter->hw, E1000_TOTH); adapter->stats.tpr += E1000_READ_REG(&adapter->hw, E1000_TPR); adapter->stats.tpt += E1000_READ_REG(&adapter->hw, E1000_TPT); adapter->stats.ptc64 += E1000_READ_REG(&adapter->hw, E1000_PTC64); adapter->stats.ptc127 += E1000_READ_REG(&adapter->hw, E1000_PTC127); adapter->stats.ptc255 += E1000_READ_REG(&adapter->hw, E1000_PTC255); adapter->stats.ptc511 += E1000_READ_REG(&adapter->hw, E1000_PTC511); adapter->stats.ptc1023 += E1000_READ_REG(&adapter->hw, E1000_PTC1023); adapter->stats.ptc1522 += E1000_READ_REG(&adapter->hw, E1000_PTC1522); adapter->stats.mptc += E1000_READ_REG(&adapter->hw, E1000_MPTC); adapter->stats.bptc += E1000_READ_REG(&adapter->hw, E1000_BPTC); /* Interrupt Counts */ adapter->stats.iac += E1000_READ_REG(&adapter->hw, E1000_IAC); adapter->stats.icrxptc += E1000_READ_REG(&adapter->hw, E1000_ICRXPTC); adapter->stats.icrxatc += E1000_READ_REG(&adapter->hw, E1000_ICRXATC); adapter->stats.ictxptc += E1000_READ_REG(&adapter->hw, E1000_ICTXPTC); adapter->stats.ictxatc += E1000_READ_REG(&adapter->hw, E1000_ICTXATC); adapter->stats.ictxqec += E1000_READ_REG(&adapter->hw, E1000_ICTXQEC); adapter->stats.ictxqmtc += E1000_READ_REG(&adapter->hw, E1000_ICTXQMTC); adapter->stats.icrxdmtc += E1000_READ_REG(&adapter->hw, E1000_ICRXDMTC); adapter->stats.icrxoc += E1000_READ_REG(&adapter->hw, E1000_ICRXOC); if (adapter->hw.mac.type >= e1000_82543) { adapter->stats.algnerrc += E1000_READ_REG(&adapter->hw, E1000_ALGNERRC); adapter->stats.rxerrc += E1000_READ_REG(&adapter->hw, E1000_RXERRC); adapter->stats.tncrs += E1000_READ_REG(&adapter->hw, E1000_TNCRS); adapter->stats.cexterr += E1000_READ_REG(&adapter->hw, E1000_CEXTERR); adapter->stats.tsctc += E1000_READ_REG(&adapter->hw, E1000_TSCTC); adapter->stats.tsctfc += E1000_READ_REG(&adapter->hw, E1000_TSCTFC); } } static uint64_t em_if_get_counter(if_ctx_t ctx, ift_counter cnt) { struct adapter *adapter = iflib_get_softc(ctx); struct ifnet *ifp = iflib_get_ifp(ctx); switch (cnt) { case IFCOUNTER_COLLISIONS: return (adapter->stats.colc); case IFCOUNTER_IERRORS: return (adapter->dropped_pkts + adapter->stats.rxerrc + adapter->stats.crcerrs + adapter->stats.algnerrc + adapter->stats.ruc + adapter->stats.roc + adapter->stats.mpc + adapter->stats.cexterr); case IFCOUNTER_OERRORS: return (adapter->stats.ecol + adapter->stats.latecol + adapter->watchdog_events); default: return (if_get_counter_default(ifp, cnt)); } } /* em_if_needs_restart - Tell iflib when the driver needs to be reinitialized * @ctx: iflib context * @event: event code to check * * Defaults to returning true for unknown events. * * @returns true if iflib needs to reinit the interface */ static bool em_if_needs_restart(if_ctx_t ctx __unused, enum iflib_restart_event event) { switch (event) { case IFLIB_RESTART_VLAN_CONFIG: default: return (true); } } /* Export a single 32-bit register via a read-only sysctl. */ static int em_sysctl_reg_handler(SYSCTL_HANDLER_ARGS) { struct adapter *adapter; u_int val; adapter = oidp->oid_arg1; val = E1000_READ_REG(&adapter->hw, oidp->oid_arg2); return (sysctl_handle_int(oidp, &val, 0, req)); } /* * Add sysctl variables, one per statistic, to the system. */ static void em_add_hw_stats(struct adapter *adapter) { device_t dev = iflib_get_dev(adapter->ctx); struct em_tx_queue *tx_que = adapter->tx_queues; struct em_rx_queue *rx_que = adapter->rx_queues; struct sysctl_ctx_list *ctx = device_get_sysctl_ctx(dev); struct sysctl_oid *tree = device_get_sysctl_tree(dev); struct sysctl_oid_list *child = SYSCTL_CHILDREN(tree); struct e1000_hw_stats *stats = &adapter->stats; struct sysctl_oid *stat_node, *queue_node, *int_node; struct sysctl_oid_list *stat_list, *queue_list, *int_list; #define QUEUE_NAME_LEN 32 char namebuf[QUEUE_NAME_LEN]; /* Driver Statistics */ SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "dropped", CTLFLAG_RD, &adapter->dropped_pkts, "Driver dropped packets"); SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "link_irq", CTLFLAG_RD, &adapter->link_irq, "Link MSI-X IRQ Handled"); SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "rx_overruns", CTLFLAG_RD, &adapter->rx_overruns, "RX overruns"); SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "watchdog_timeouts", CTLFLAG_RD, &adapter->watchdog_events, "Watchdog timeouts"); SYSCTL_ADD_PROC(ctx, child, OID_AUTO, "device_control", CTLTYPE_UINT | CTLFLAG_RD | CTLFLAG_NEEDGIANT, adapter, E1000_CTRL, em_sysctl_reg_handler, "IU", "Device Control Register"); SYSCTL_ADD_PROC(ctx, child, OID_AUTO, "rx_control", CTLTYPE_UINT | CTLFLAG_RD | CTLFLAG_NEEDGIANT, adapter, E1000_RCTL, em_sysctl_reg_handler, "IU", "Receiver Control Register"); SYSCTL_ADD_UINT(ctx, child, OID_AUTO, "fc_high_water", CTLFLAG_RD, &adapter->hw.fc.high_water, 0, "Flow Control High Watermark"); SYSCTL_ADD_UINT(ctx, child, OID_AUTO, "fc_low_water", CTLFLAG_RD, &adapter->hw.fc.low_water, 0, "Flow Control Low Watermark"); for (int i = 0; i < adapter->tx_num_queues; i++, tx_que++) { struct tx_ring *txr = &tx_que->txr; snprintf(namebuf, QUEUE_NAME_LEN, "queue_tx_%d", i); queue_node = SYSCTL_ADD_NODE(ctx, child, OID_AUTO, namebuf, CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, "TX Queue Name"); queue_list = SYSCTL_CHILDREN(queue_node); SYSCTL_ADD_PROC(ctx, queue_list, OID_AUTO, "txd_head", CTLTYPE_UINT | CTLFLAG_RD | CTLFLAG_NEEDGIANT, adapter, E1000_TDH(txr->me), em_sysctl_reg_handler, "IU", "Transmit Descriptor Head"); SYSCTL_ADD_PROC(ctx, queue_list, OID_AUTO, "txd_tail", CTLTYPE_UINT | CTLFLAG_RD | CTLFLAG_NEEDGIANT, adapter, E1000_TDT(txr->me), em_sysctl_reg_handler, "IU", "Transmit Descriptor Tail"); SYSCTL_ADD_ULONG(ctx, queue_list, OID_AUTO, "tx_irq", CTLFLAG_RD, &txr->tx_irq, "Queue MSI-X Transmit Interrupts"); } for (int j = 0; j < adapter->rx_num_queues; j++, rx_que++) { struct rx_ring *rxr = &rx_que->rxr; snprintf(namebuf, QUEUE_NAME_LEN, "queue_rx_%d", j); queue_node = SYSCTL_ADD_NODE(ctx, child, OID_AUTO, namebuf, CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, "RX Queue Name"); queue_list = SYSCTL_CHILDREN(queue_node); SYSCTL_ADD_PROC(ctx, queue_list, OID_AUTO, "rxd_head", CTLTYPE_UINT | CTLFLAG_RD | CTLFLAG_NEEDGIANT, adapter, E1000_RDH(rxr->me), em_sysctl_reg_handler, "IU", "Receive Descriptor Head"); SYSCTL_ADD_PROC(ctx, queue_list, OID_AUTO, "rxd_tail", CTLTYPE_UINT | CTLFLAG_RD | CTLFLAG_NEEDGIANT, adapter, E1000_RDT(rxr->me), em_sysctl_reg_handler, "IU", "Receive Descriptor Tail"); SYSCTL_ADD_ULONG(ctx, queue_list, OID_AUTO, "rx_irq", CTLFLAG_RD, &rxr->rx_irq, "Queue MSI-X Receive Interrupts"); } /* MAC stats get their own sub node */ stat_node = SYSCTL_ADD_NODE(ctx, child, OID_AUTO, "mac_stats", CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, "Statistics"); stat_list = SYSCTL_CHILDREN(stat_node); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "excess_coll", CTLFLAG_RD, &stats->ecol, "Excessive collisions"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "single_coll", CTLFLAG_RD, &stats->scc, "Single collisions"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "multiple_coll", CTLFLAG_RD, &stats->mcc, "Multiple collisions"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "late_coll", CTLFLAG_RD, &stats->latecol, "Late collisions"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "collision_count", CTLFLAG_RD, &stats->colc, "Collision Count"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "symbol_errors", CTLFLAG_RD, &adapter->stats.symerrs, "Symbol Errors"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "sequence_errors", CTLFLAG_RD, &adapter->stats.sec, "Sequence Errors"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "defer_count", CTLFLAG_RD, &adapter->stats.dc, "Defer Count"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "missed_packets", CTLFLAG_RD, &adapter->stats.mpc, "Missed Packets"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_no_buff", CTLFLAG_RD, &adapter->stats.rnbc, "Receive No Buffers"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_undersize", CTLFLAG_RD, &adapter->stats.ruc, "Receive Undersize"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_fragmented", CTLFLAG_RD, &adapter->stats.rfc, "Fragmented Packets Received "); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_oversize", CTLFLAG_RD, &adapter->stats.roc, "Oversized Packets Received"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_jabber", CTLFLAG_RD, &adapter->stats.rjc, "Recevied Jabber"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_errs", CTLFLAG_RD, &adapter->stats.rxerrc, "Receive Errors"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "crc_errs", CTLFLAG_RD, &adapter->stats.crcerrs, "CRC errors"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "alignment_errs", CTLFLAG_RD, &adapter->stats.algnerrc, "Alignment Errors"); /* On 82575 these are collision counts */ SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "coll_ext_errs", CTLFLAG_RD, &adapter->stats.cexterr, "Collision/Carrier extension errors"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "xon_recvd", CTLFLAG_RD, &adapter->stats.xonrxc, "XON Received"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "xon_txd", CTLFLAG_RD, &adapter->stats.xontxc, "XON Transmitted"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "xoff_recvd", CTLFLAG_RD, &adapter->stats.xoffrxc, "XOFF Received"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "xoff_txd", CTLFLAG_RD, &adapter->stats.xofftxc, "XOFF Transmitted"); /* Packet Reception Stats */ SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "total_pkts_recvd", CTLFLAG_RD, &adapter->stats.tpr, "Total Packets Received "); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "good_pkts_recvd", CTLFLAG_RD, &adapter->stats.gprc, "Good Packets Received"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "bcast_pkts_recvd", CTLFLAG_RD, &adapter->stats.bprc, "Broadcast Packets Received"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "mcast_pkts_recvd", CTLFLAG_RD, &adapter->stats.mprc, "Multicast Packets Received"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_64", CTLFLAG_RD, &adapter->stats.prc64, "64 byte frames received "); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_65_127", CTLFLAG_RD, &adapter->stats.prc127, "65-127 byte frames received"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_128_255", CTLFLAG_RD, &adapter->stats.prc255, "128-255 byte frames received"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_256_511", CTLFLAG_RD, &adapter->stats.prc511, "256-511 byte frames received"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_512_1023", CTLFLAG_RD, &adapter->stats.prc1023, "512-1023 byte frames received"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_1024_1522", CTLFLAG_RD, &adapter->stats.prc1522, "1023-1522 byte frames received"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "good_octets_recvd", CTLFLAG_RD, &adapter->stats.gorc, "Good Octets Received"); /* Packet Transmission Stats */ SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "good_octets_txd", CTLFLAG_RD, &adapter->stats.gotc, "Good Octets Transmitted"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "total_pkts_txd", CTLFLAG_RD, &adapter->stats.tpt, "Total Packets Transmitted"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "good_pkts_txd", CTLFLAG_RD, &adapter->stats.gptc, "Good Packets Transmitted"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "bcast_pkts_txd", CTLFLAG_RD, &adapter->stats.bptc, "Broadcast Packets Transmitted"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "mcast_pkts_txd", CTLFLAG_RD, &adapter->stats.mptc, "Multicast Packets Transmitted"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_64", CTLFLAG_RD, &adapter->stats.ptc64, "64 byte frames transmitted "); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_65_127", CTLFLAG_RD, &adapter->stats.ptc127, "65-127 byte frames transmitted"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_128_255", CTLFLAG_RD, &adapter->stats.ptc255, "128-255 byte frames transmitted"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_256_511", CTLFLAG_RD, &adapter->stats.ptc511, "256-511 byte frames transmitted"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_512_1023", CTLFLAG_RD, &adapter->stats.ptc1023, "512-1023 byte frames transmitted"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_1024_1522", CTLFLAG_RD, &adapter->stats.ptc1522, "1024-1522 byte frames transmitted"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tso_txd", CTLFLAG_RD, &adapter->stats.tsctc, "TSO Contexts Transmitted"); SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tso_ctx_fail", CTLFLAG_RD, &adapter->stats.tsctfc, "TSO Contexts Failed"); /* Interrupt Stats */ int_node = SYSCTL_ADD_NODE(ctx, child, OID_AUTO, "interrupts", CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, "Interrupt Statistics"); int_list = SYSCTL_CHILDREN(int_node); SYSCTL_ADD_UQUAD(ctx, int_list, OID_AUTO, "asserts", CTLFLAG_RD, &adapter->stats.iac, "Interrupt Assertion Count"); SYSCTL_ADD_UQUAD(ctx, int_list, OID_AUTO, "rx_pkt_timer", CTLFLAG_RD, &adapter->stats.icrxptc, "Interrupt Cause Rx Pkt Timer Expire Count"); SYSCTL_ADD_UQUAD(ctx, int_list, OID_AUTO, "rx_abs_timer", CTLFLAG_RD, &adapter->stats.icrxatc, "Interrupt Cause Rx Abs Timer Expire Count"); SYSCTL_ADD_UQUAD(ctx, int_list, OID_AUTO, "tx_pkt_timer", CTLFLAG_RD, &adapter->stats.ictxptc, "Interrupt Cause Tx Pkt Timer Expire Count"); SYSCTL_ADD_UQUAD(ctx, int_list, OID_AUTO, "tx_abs_timer", CTLFLAG_RD, &adapter->stats.ictxatc, "Interrupt Cause Tx Abs Timer Expire Count"); SYSCTL_ADD_UQUAD(ctx, int_list, OID_AUTO, "tx_queue_empty", CTLFLAG_RD, &adapter->stats.ictxqec, "Interrupt Cause Tx Queue Empty Count"); SYSCTL_ADD_UQUAD(ctx, int_list, OID_AUTO, "tx_queue_min_thresh", CTLFLAG_RD, &adapter->stats.ictxqmtc, "Interrupt Cause Tx Queue Min Thresh Count"); SYSCTL_ADD_UQUAD(ctx, int_list, OID_AUTO, "rx_desc_min_thresh", CTLFLAG_RD, &adapter->stats.icrxdmtc, "Interrupt Cause Rx Desc Min Thresh Count"); SYSCTL_ADD_UQUAD(ctx, int_list, OID_AUTO, "rx_overrun", CTLFLAG_RD, &adapter->stats.icrxoc, "Interrupt Cause Receiver Overrun Count"); } /********************************************************************** * * This routine provides a way to dump out the adapter eeprom, * often a useful debug/service tool. This only dumps the first * 32 words, stuff that matters is in that extent. * **********************************************************************/ static int em_sysctl_nvm_info(SYSCTL_HANDLER_ARGS) { struct adapter *adapter = (struct adapter *)arg1; int error; int result; result = -1; error = sysctl_handle_int(oidp, &result, 0, req); if (error || !req->newptr) return (error); /* * This value will cause a hex dump of the * first 32 16-bit words of the EEPROM to * the screen. */ if (result == 1) em_print_nvm_info(adapter); return (error); } static void em_print_nvm_info(struct adapter *adapter) { u16 eeprom_data; int i, j, row = 0; /* Its a bit crude, but it gets the job done */ printf("\nInterface EEPROM Dump:\n"); printf("Offset\n0x0000 "); for (i = 0, j = 0; i < 32; i++, j++) { if (j == 8) { /* Make the offset block */ j = 0; ++row; printf("\n0x00%x0 ",row); } e1000_read_nvm(&adapter->hw, i, 1, &eeprom_data); printf("%04x ", eeprom_data); } printf("\n"); } static int em_sysctl_int_delay(SYSCTL_HANDLER_ARGS) { struct em_int_delay_info *info; struct adapter *adapter; u32 regval; int error, usecs, ticks; info = (struct em_int_delay_info *) arg1; usecs = info->value; error = sysctl_handle_int(oidp, &usecs, 0, req); if (error != 0 || req->newptr == NULL) return (error); if (usecs < 0 || usecs > EM_TICKS_TO_USECS(65535)) return (EINVAL); info->value = usecs; ticks = EM_USECS_TO_TICKS(usecs); if (info->offset == E1000_ITR) /* units are 256ns here */ ticks *= 4; adapter = info->adapter; regval = E1000_READ_OFFSET(&adapter->hw, info->offset); regval = (regval & ~0xffff) | (ticks & 0xffff); /* Handle a few special cases. */ switch (info->offset) { case E1000_RDTR: break; case E1000_TIDV: if (ticks == 0) { adapter->txd_cmd &= ~E1000_TXD_CMD_IDE; /* Don't write 0 into the TIDV register. */ regval++; } else adapter->txd_cmd |= E1000_TXD_CMD_IDE; break; } E1000_WRITE_OFFSET(&adapter->hw, info->offset, regval); return (0); } static void em_add_int_delay_sysctl(struct adapter *adapter, const char *name, const char *description, struct em_int_delay_info *info, int offset, int value) { info->adapter = adapter; info->offset = offset; info->value = value; SYSCTL_ADD_PROC(device_get_sysctl_ctx(adapter->dev), SYSCTL_CHILDREN(device_get_sysctl_tree(adapter->dev)), OID_AUTO, name, CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_NEEDGIANT, info, 0, em_sysctl_int_delay, "I", description); } /* * Set flow control using sysctl: * Flow control values: * 0 - off * 1 - rx pause * 2 - tx pause * 3 - full */ static int em_set_flowcntl(SYSCTL_HANDLER_ARGS) { int error; static int input = 3; /* default is full */ struct adapter *adapter = (struct adapter *) arg1; error = sysctl_handle_int(oidp, &input, 0, req); if ((error) || (req->newptr == NULL)) return (error); if (input == adapter->fc) /* no change? */ return (error); switch (input) { case e1000_fc_rx_pause: case e1000_fc_tx_pause: case e1000_fc_full: case e1000_fc_none: adapter->hw.fc.requested_mode = input; adapter->fc = input; break; default: /* Do nothing */ return (error); } adapter->hw.fc.current_mode = adapter->hw.fc.requested_mode; e1000_force_mac_fc(&adapter->hw); return (error); } /* * Manage Energy Efficient Ethernet: * Control values: * 0/1 - enabled/disabled */ static int em_sysctl_eee(SYSCTL_HANDLER_ARGS) { struct adapter *adapter = (struct adapter *) arg1; int error, value; value = adapter->hw.dev_spec.ich8lan.eee_disable; error = sysctl_handle_int(oidp, &value, 0, req); if (error || req->newptr == NULL) return (error); adapter->hw.dev_spec.ich8lan.eee_disable = (value != 0); em_if_init(adapter->ctx); return (0); } static int em_sysctl_debug_info(SYSCTL_HANDLER_ARGS) { struct adapter *adapter; int error; int result; result = -1; error = sysctl_handle_int(oidp, &result, 0, req); if (error || !req->newptr) return (error); if (result == 1) { adapter = (struct adapter *) arg1; em_print_debug_info(adapter); } return (error); } static int em_get_rs(SYSCTL_HANDLER_ARGS) { struct adapter *adapter = (struct adapter *) arg1; int error; int result; result = 0; error = sysctl_handle_int(oidp, &result, 0, req); if (error || !req->newptr || result != 1) return (error); em_dump_rs(adapter); return (error); } static void em_if_debug(if_ctx_t ctx) { em_dump_rs(iflib_get_softc(ctx)); } /* * This routine is meant to be fluid, add whatever is * needed for debugging a problem. -jfv */ static void em_print_debug_info(struct adapter *adapter) { device_t dev = iflib_get_dev(adapter->ctx); struct ifnet *ifp = iflib_get_ifp(adapter->ctx); struct tx_ring *txr = &adapter->tx_queues->txr; struct rx_ring *rxr = &adapter->rx_queues->rxr; if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) printf("Interface is RUNNING "); else printf("Interface is NOT RUNNING\n"); if (if_getdrvflags(ifp) & IFF_DRV_OACTIVE) printf("and INACTIVE\n"); else printf("and ACTIVE\n"); for (int i = 0; i < adapter->tx_num_queues; i++, txr++) { device_printf(dev, "TX Queue %d ------\n", i); device_printf(dev, "hw tdh = %d, hw tdt = %d\n", E1000_READ_REG(&adapter->hw, E1000_TDH(i)), E1000_READ_REG(&adapter->hw, E1000_TDT(i))); } for (int j=0; j < adapter->rx_num_queues; j++, rxr++) { device_printf(dev, "RX Queue %d ------\n", j); device_printf(dev, "hw rdh = %d, hw rdt = %d\n", E1000_READ_REG(&adapter->hw, E1000_RDH(j)), E1000_READ_REG(&adapter->hw, E1000_RDT(j))); } } /* * 82574 only: * Write a new value to the EEPROM increasing the number of MSI-X * vectors from 3 to 5, for proper multiqueue support. */ static void em_enable_vectors_82574(if_ctx_t ctx) { struct adapter *adapter = iflib_get_softc(ctx); struct e1000_hw *hw = &adapter->hw; device_t dev = iflib_get_dev(ctx); u16 edata; e1000_read_nvm(hw, EM_NVM_PCIE_CTRL, 1, &edata); if (bootverbose) device_printf(dev, "EM_NVM_PCIE_CTRL = %#06x\n", edata); if (((edata & EM_NVM_MSIX_N_MASK) >> EM_NVM_MSIX_N_SHIFT) != 4) { device_printf(dev, "Writing to eeprom: increasing " "reported MSI-X vectors from 3 to 5...\n"); edata &= ~(EM_NVM_MSIX_N_MASK); edata |= 4 << EM_NVM_MSIX_N_SHIFT; e1000_write_nvm(hw, EM_NVM_PCIE_CTRL, 1, &edata); e1000_update_nvm_checksum(hw); device_printf(dev, "Writing to eeprom: done\n"); } } Index: projects/clang1100-import/sys/kern/subr_epoch.c =================================================================== --- projects/clang1100-import/sys/kern/subr_epoch.c (revision 364034) +++ projects/clang1100-import/sys/kern/subr_epoch.c (revision 364035) @@ -1,848 +1,893 @@ /*- * SPDX-License-Identifier: BSD-2-Clause-FreeBSD * * Copyright (c) 2018, Matthew Macy * * 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 #include #include #include #ifdef EPOCH_TRACE #include #include #include #endif #include #include #include #include #include -static MALLOC_DEFINE(M_EPOCH, "epoch", "epoch based reclamation"); - #ifdef __amd64__ #define EPOCH_ALIGN CACHE_LINE_SIZE*2 #else #define EPOCH_ALIGN CACHE_LINE_SIZE #endif TAILQ_HEAD (epoch_tdlist, epoch_tracker); typedef struct epoch_record { ck_epoch_record_t er_record; struct epoch_context er_drain_ctx; struct epoch *er_parent; volatile struct epoch_tdlist er_tdlist; volatile uint32_t er_gen; uint32_t er_cpuid; } __aligned(EPOCH_ALIGN) *epoch_record_t; struct epoch { struct ck_epoch e_epoch __aligned(EPOCH_ALIGN); epoch_record_t e_pcpu_record; - int e_idx; + int e_in_use; int e_flags; struct sx e_drain_sx; struct mtx e_drain_mtx; volatile int e_drain_count; const char *e_name; }; /* arbitrary --- needs benchmarking */ #define MAX_ADAPTIVE_SPIN 100 #define MAX_EPOCHS 64 CTASSERT(sizeof(ck_epoch_entry_t) == sizeof(struct epoch_context)); SYSCTL_NODE(_kern, OID_AUTO, epoch, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, "epoch information"); SYSCTL_NODE(_kern_epoch, OID_AUTO, stats, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, "epoch stats"); /* Stats. */ static counter_u64_t block_count; SYSCTL_COUNTER_U64(_kern_epoch_stats, OID_AUTO, nblocked, CTLFLAG_RW, &block_count, "# of times a thread was in an epoch when epoch_wait was called"); static counter_u64_t migrate_count; SYSCTL_COUNTER_U64(_kern_epoch_stats, OID_AUTO, migrations, CTLFLAG_RW, &migrate_count, "# of times thread was migrated to another CPU in epoch_wait"); static counter_u64_t turnstile_count; SYSCTL_COUNTER_U64(_kern_epoch_stats, OID_AUTO, ncontended, CTLFLAG_RW, &turnstile_count, "# of times a thread was blocked on a lock in an epoch during an epoch_wait"); static counter_u64_t switch_count; SYSCTL_COUNTER_U64(_kern_epoch_stats, OID_AUTO, switches, CTLFLAG_RW, &switch_count, "# of times a thread voluntarily context switched in epoch_wait"); static counter_u64_t epoch_call_count; SYSCTL_COUNTER_U64(_kern_epoch_stats, OID_AUTO, epoch_calls, CTLFLAG_RW, &epoch_call_count, "# of times a callback was deferred"); static counter_u64_t epoch_call_task_count; SYSCTL_COUNTER_U64(_kern_epoch_stats, OID_AUTO, epoch_call_tasks, CTLFLAG_RW, &epoch_call_task_count, "# of times a callback task was run"); TAILQ_HEAD (threadlist, thread); CK_STACK_CONTAINER(struct ck_epoch_entry, stack_entry, ck_epoch_entry_container) -epoch_t allepochs[MAX_EPOCHS]; +static struct epoch epoch_array[MAX_EPOCHS]; DPCPU_DEFINE(struct grouptask, epoch_cb_task); DPCPU_DEFINE(int, epoch_cb_count); static __read_mostly int inited; -static __read_mostly int epoch_count; __read_mostly epoch_t global_epoch; __read_mostly epoch_t global_epoch_preempt; static void epoch_call_task(void *context __unused); static uma_zone_t pcpu_zone_record; +static struct sx epoch_sx; + +#define EPOCH_LOCK() sx_xlock(&epoch_sx) +#define EPOCH_UNLOCK() sx_xunlock(&epoch_sx) + #ifdef EPOCH_TRACE struct stackentry { RB_ENTRY(stackentry) se_node; struct stack se_stack; }; static int stackentry_compare(struct stackentry *a, struct stackentry *b) { if (a->se_stack.depth > b->se_stack.depth) return (1); if (a->se_stack.depth < b->se_stack.depth) return (-1); for (int i = 0; i < a->se_stack.depth; i++) { if (a->se_stack.pcs[i] > b->se_stack.pcs[i]) return (1); if (a->se_stack.pcs[i] < b->se_stack.pcs[i]) return (-1); } return (0); } RB_HEAD(stacktree, stackentry) epoch_stacks = RB_INITIALIZER(&epoch_stacks); RB_GENERATE_STATIC(stacktree, stackentry, se_node, stackentry_compare); static struct mtx epoch_stacks_lock; MTX_SYSINIT(epochstacks, &epoch_stacks_lock, "epoch_stacks", MTX_DEF); static bool epoch_trace_stack_print = true; SYSCTL_BOOL(_kern_epoch, OID_AUTO, trace_stack_print, CTLFLAG_RWTUN, &epoch_trace_stack_print, 0, "Print stack traces on epoch reports"); static void epoch_trace_report(const char *fmt, ...) __printflike(1, 2); static inline void epoch_trace_report(const char *fmt, ...) { va_list ap; struct stackentry se, *new; stack_zero(&se.se_stack); /* XXX: is it really needed? */ stack_save(&se.se_stack); /* Tree is never reduced - go lockless. */ if (RB_FIND(stacktree, &epoch_stacks, &se) != NULL) return; new = malloc(sizeof(*new), M_STACK, M_NOWAIT); if (new != NULL) { bcopy(&se.se_stack, &new->se_stack, sizeof(struct stack)); mtx_lock(&epoch_stacks_lock); new = RB_INSERT(stacktree, &epoch_stacks, new); mtx_unlock(&epoch_stacks_lock); if (new != NULL) free(new, M_STACK); } va_start(ap, fmt); (void)vprintf(fmt, ap); va_end(ap); if (epoch_trace_stack_print) stack_print_ddb(&se.se_stack); } static inline void epoch_trace_enter(struct thread *td, epoch_t epoch, epoch_tracker_t et, const char *file, int line) { epoch_tracker_t iet; SLIST_FOREACH(iet, &td->td_epochs, et_tlink) if (iet->et_epoch == epoch) epoch_trace_report("Recursively entering epoch %s " "at %s:%d, previously entered at %s:%d\n", epoch->e_name, file, line, iet->et_file, iet->et_line); et->et_epoch = epoch; et->et_file = file; et->et_line = line; SLIST_INSERT_HEAD(&td->td_epochs, et, et_tlink); } static inline void epoch_trace_exit(struct thread *td, epoch_t epoch, epoch_tracker_t et, const char *file, int line) { if (SLIST_FIRST(&td->td_epochs) != et) { epoch_trace_report("Exiting epoch %s in a not nested order " "at %s:%d. Most recently entered %s at %s:%d\n", epoch->e_name, file, line, SLIST_FIRST(&td->td_epochs)->et_epoch->e_name, SLIST_FIRST(&td->td_epochs)->et_file, SLIST_FIRST(&td->td_epochs)->et_line); /* This will panic if et is not anywhere on td_epochs. */ SLIST_REMOVE(&td->td_epochs, et, epoch_tracker, et_tlink); } else SLIST_REMOVE_HEAD(&td->td_epochs, et_tlink); } /* Used by assertions that check thread state before going to sleep. */ void epoch_trace_list(struct thread *td) { epoch_tracker_t iet; SLIST_FOREACH(iet, &td->td_epochs, et_tlink) printf("Epoch %s entered at %s:%d\n", iet->et_epoch->e_name, iet->et_file, iet->et_line); } #endif /* EPOCH_TRACE */ static void epoch_init(void *arg __unused) { int cpu; block_count = counter_u64_alloc(M_WAITOK); migrate_count = counter_u64_alloc(M_WAITOK); turnstile_count = counter_u64_alloc(M_WAITOK); switch_count = counter_u64_alloc(M_WAITOK); epoch_call_count = counter_u64_alloc(M_WAITOK); epoch_call_task_count = counter_u64_alloc(M_WAITOK); pcpu_zone_record = uma_zcreate("epoch_record pcpu", sizeof(struct epoch_record), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_PCPU); CPU_FOREACH(cpu) { GROUPTASK_INIT(DPCPU_ID_PTR(cpu, epoch_cb_task), 0, epoch_call_task, NULL); taskqgroup_attach_cpu(qgroup_softirq, DPCPU_ID_PTR(cpu, epoch_cb_task), NULL, cpu, NULL, NULL, "epoch call task"); } #ifdef EPOCH_TRACE SLIST_INIT(&thread0.td_epochs); #endif + sx_init(&epoch_sx, "epoch-sx"); inited = 1; global_epoch = epoch_alloc("Global", 0); global_epoch_preempt = epoch_alloc("Global preemptible", EPOCH_PREEMPT); } SYSINIT(epoch, SI_SUB_EPOCH, SI_ORDER_FIRST, epoch_init, NULL); #if !defined(EARLY_AP_STARTUP) static void epoch_init_smp(void *dummy __unused) { inited = 2; } SYSINIT(epoch_smp, SI_SUB_SMP + 1, SI_ORDER_FIRST, epoch_init_smp, NULL); #endif static void epoch_ctor(epoch_t epoch) { epoch_record_t er; int cpu; epoch->e_pcpu_record = uma_zalloc_pcpu(pcpu_zone_record, M_WAITOK); CPU_FOREACH(cpu) { er = zpcpu_get_cpu(epoch->e_pcpu_record, cpu); bzero(er, sizeof(*er)); ck_epoch_register(&epoch->e_epoch, &er->er_record, NULL); TAILQ_INIT((struct threadlist *)(uintptr_t)&er->er_tdlist); er->er_cpuid = cpu; er->er_parent = epoch; } } static void epoch_adjust_prio(struct thread *td, u_char prio) { thread_lock(td); sched_prio(td, prio); thread_unlock(td); } epoch_t epoch_alloc(const char *name, int flags) { epoch_t epoch; + int i; + MPASS(name != NULL); + if (__predict_false(!inited)) panic("%s called too early in boot", __func__); - epoch = malloc(sizeof(struct epoch), M_EPOCH, M_ZERO | M_WAITOK); + + EPOCH_LOCK(); + + /* + * Find a free index in the epoch array. If no free index is + * found, try to use the index after the last one. + */ + for (i = 0;; i++) { + /* + * If too many epochs are currently allocated, + * return NULL. + */ + if (i == MAX_EPOCHS) { + epoch = NULL; + goto done; + } + if (epoch_array[i].e_in_use == 0) + break; + } + + epoch = epoch_array + i; ck_epoch_init(&epoch->e_epoch); epoch_ctor(epoch); - MPASS(epoch_count < MAX_EPOCHS - 2); epoch->e_flags = flags; - epoch->e_idx = epoch_count; epoch->e_name = name; sx_init(&epoch->e_drain_sx, "epoch-drain-sx"); mtx_init(&epoch->e_drain_mtx, "epoch-drain-mtx", NULL, MTX_DEF); - allepochs[epoch_count++] = epoch; + + /* + * Set e_in_use last, because when this field is set the + * epoch_call_task() function will start scanning this epoch + * structure. + */ + atomic_store_rel_int(&epoch->e_in_use, 1); +done: + EPOCH_UNLOCK(); return (epoch); } void epoch_free(epoch_t epoch) { + EPOCH_LOCK(); + + MPASS(epoch->e_in_use != 0); + epoch_drain_callbacks(epoch); - allepochs[epoch->e_idx] = NULL; + + atomic_store_rel_int(&epoch->e_in_use, 0); + /* + * Make sure the epoch_call_task() function see e_in_use equal + * to zero, by calling epoch_wait() on the global_epoch: + */ epoch_wait(global_epoch); uma_zfree_pcpu(pcpu_zone_record, epoch->e_pcpu_record); mtx_destroy(&epoch->e_drain_mtx); sx_destroy(&epoch->e_drain_sx); - free(epoch, M_EPOCH); + memset(epoch, 0, sizeof(*epoch)); + + EPOCH_UNLOCK(); } static epoch_record_t epoch_currecord(epoch_t epoch) { return (zpcpu_get(epoch->e_pcpu_record)); } #define INIT_CHECK(epoch) \ do { \ if (__predict_false((epoch) == NULL)) \ return; \ } while (0) void _epoch_enter_preempt(epoch_t epoch, epoch_tracker_t et EPOCH_FILE_LINE) { struct epoch_record *er; struct thread *td; MPASS(cold || epoch != NULL); MPASS(epoch->e_flags & EPOCH_PREEMPT); td = curthread; MPASS((vm_offset_t)et >= td->td_kstack && (vm_offset_t)et + sizeof(struct epoch_tracker) <= td->td_kstack + td->td_kstack_pages * PAGE_SIZE); INIT_CHECK(epoch); #ifdef EPOCH_TRACE epoch_trace_enter(td, epoch, et, file, line); #endif et->et_td = td; THREAD_NO_SLEEPING(); critical_enter(); sched_pin(); td->td_pre_epoch_prio = td->td_priority; er = epoch_currecord(epoch); TAILQ_INSERT_TAIL(&er->er_tdlist, et, et_link); ck_epoch_begin(&er->er_record, &et->et_section); critical_exit(); } void epoch_enter(epoch_t epoch) { epoch_record_t er; MPASS(cold || epoch != NULL); INIT_CHECK(epoch); critical_enter(); er = epoch_currecord(epoch); ck_epoch_begin(&er->er_record, NULL); } void _epoch_exit_preempt(epoch_t epoch, epoch_tracker_t et EPOCH_FILE_LINE) { struct epoch_record *er; struct thread *td; INIT_CHECK(epoch); td = curthread; critical_enter(); sched_unpin(); THREAD_SLEEPING_OK(); er = epoch_currecord(epoch); MPASS(epoch->e_flags & EPOCH_PREEMPT); MPASS(et != NULL); MPASS(et->et_td == td); #ifdef INVARIANTS et->et_td = (void*)0xDEADBEEF; #endif ck_epoch_end(&er->er_record, &et->et_section); TAILQ_REMOVE(&er->er_tdlist, et, et_link); er->er_gen++; if (__predict_false(td->td_pre_epoch_prio != td->td_priority)) epoch_adjust_prio(td, td->td_pre_epoch_prio); critical_exit(); #ifdef EPOCH_TRACE epoch_trace_exit(td, epoch, et, file, line); #endif } void epoch_exit(epoch_t epoch) { epoch_record_t er; INIT_CHECK(epoch); er = epoch_currecord(epoch); ck_epoch_end(&er->er_record, NULL); critical_exit(); } /* * epoch_block_handler_preempt() is a callback from the CK code when another * thread is currently in an epoch section. */ static void epoch_block_handler_preempt(struct ck_epoch *global __unused, ck_epoch_record_t *cr, void *arg __unused) { epoch_record_t record; struct thread *td, *owner, *curwaittd; struct epoch_tracker *tdwait; struct turnstile *ts; struct lock_object *lock; int spincount, gen; int locksheld __unused; record = __containerof(cr, struct epoch_record, er_record); td = curthread; locksheld = td->td_locks; spincount = 0; counter_u64_add(block_count, 1); /* * We lost a race and there's no longer any threads * on the CPU in an epoch section. */ if (TAILQ_EMPTY(&record->er_tdlist)) return; if (record->er_cpuid != curcpu) { /* * If the head of the list is running, we can wait for it * to remove itself from the list and thus save us the * overhead of a migration */ gen = record->er_gen; thread_unlock(td); /* * We can't actually check if the waiting thread is running * so we simply poll for it to exit before giving up and * migrating. */ do { cpu_spinwait(); } while (!TAILQ_EMPTY(&record->er_tdlist) && gen == record->er_gen && spincount++ < MAX_ADAPTIVE_SPIN); thread_lock(td); /* * If the generation has changed we can poll again * otherwise we need to migrate. */ if (gen != record->er_gen) return; /* * Being on the same CPU as that of the record on which * we need to wait allows us access to the thread * list associated with that CPU. We can then examine the * oldest thread in the queue and wait on its turnstile * until it resumes and so on until a grace period * elapses. * */ counter_u64_add(migrate_count, 1); sched_bind(td, record->er_cpuid); /* * At this point we need to return to the ck code * to scan to see if a grace period has elapsed. * We can't move on to check the thread list, because * in the meantime new threads may have arrived that * in fact belong to a different epoch. */ return; } /* * Try to find a thread in an epoch section on this CPU * waiting on a turnstile. Otherwise find the lowest * priority thread (highest prio value) and drop our priority * to match to allow it to run. */ TAILQ_FOREACH(tdwait, &record->er_tdlist, et_link) { /* * Propagate our priority to any other waiters to prevent us * from starving them. They will have their original priority * restore on exit from epoch_wait(). */ curwaittd = tdwait->et_td; if (!TD_IS_INHIBITED(curwaittd) && curwaittd->td_priority > td->td_priority) { critical_enter(); thread_unlock(td); thread_lock(curwaittd); sched_prio(curwaittd, td->td_priority); thread_unlock(curwaittd); thread_lock(td); critical_exit(); } if (TD_IS_INHIBITED(curwaittd) && TD_ON_LOCK(curwaittd) && ((ts = curwaittd->td_blocked) != NULL)) { /* * We unlock td to allow turnstile_wait to reacquire * the thread lock. Before unlocking it we enter a * critical section to prevent preemption after we * reenable interrupts by dropping the thread lock in * order to prevent curwaittd from getting to run. */ critical_enter(); thread_unlock(td); if (turnstile_lock(ts, &lock, &owner)) { if (ts == curwaittd->td_blocked) { MPASS(TD_IS_INHIBITED(curwaittd) && TD_ON_LOCK(curwaittd)); critical_exit(); turnstile_wait(ts, owner, curwaittd->td_tsqueue); counter_u64_add(turnstile_count, 1); thread_lock(td); return; } turnstile_unlock(ts, lock); } thread_lock(td); critical_exit(); KASSERT(td->td_locks == locksheld, ("%d extra locks held", td->td_locks - locksheld)); } } /* * We didn't find any threads actually blocked on a lock * so we have nothing to do except context switch away. */ counter_u64_add(switch_count, 1); mi_switch(SW_VOL | SWT_RELINQUISH); /* * It is important the thread lock is dropped while yielding * to allow other threads to acquire the lock pointed to by * TDQ_LOCKPTR(td). Currently mi_switch() will unlock the * thread lock before returning. Else a deadlock like * situation might happen. */ thread_lock(td); } void epoch_wait_preempt(epoch_t epoch) { struct thread *td; int was_bound; int old_cpu; int old_pinned; u_char old_prio; int locks __unused; MPASS(cold || epoch != NULL); INIT_CHECK(epoch); td = curthread; #ifdef INVARIANTS locks = curthread->td_locks; MPASS(epoch->e_flags & EPOCH_PREEMPT); if ((epoch->e_flags & EPOCH_LOCKED) == 0) WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL, "epoch_wait() can be long running"); KASSERT(!in_epoch(epoch), ("epoch_wait_preempt() called in the middle " "of an epoch section of the same epoch")); #endif DROP_GIANT(); thread_lock(td); old_cpu = PCPU_GET(cpuid); old_pinned = td->td_pinned; old_prio = td->td_priority; was_bound = sched_is_bound(td); sched_unbind(td); td->td_pinned = 0; sched_bind(td, old_cpu); ck_epoch_synchronize_wait(&epoch->e_epoch, epoch_block_handler_preempt, NULL); /* restore CPU binding, if any */ if (was_bound != 0) { sched_bind(td, old_cpu); } else { /* get thread back to initial CPU, if any */ if (old_pinned != 0) sched_bind(td, old_cpu); sched_unbind(td); } /* restore pinned after bind */ td->td_pinned = old_pinned; /* restore thread priority */ sched_prio(td, old_prio); thread_unlock(td); PICKUP_GIANT(); KASSERT(td->td_locks == locks, ("%d residual locks held", td->td_locks - locks)); } static void epoch_block_handler(struct ck_epoch *g __unused, ck_epoch_record_t *c __unused, void *arg __unused) { cpu_spinwait(); } void epoch_wait(epoch_t epoch) { MPASS(cold || epoch != NULL); INIT_CHECK(epoch); MPASS(epoch->e_flags == 0); critical_enter(); ck_epoch_synchronize_wait(&epoch->e_epoch, epoch_block_handler, NULL); critical_exit(); } void epoch_call(epoch_t epoch, epoch_callback_t callback, epoch_context_t ctx) { epoch_record_t er; ck_epoch_entry_t *cb; cb = (void *)ctx; MPASS(callback); /* too early in boot to have epoch set up */ if (__predict_false(epoch == NULL)) goto boottime; #if !defined(EARLY_AP_STARTUP) if (__predict_false(inited < 2)) goto boottime; #endif critical_enter(); *DPCPU_PTR(epoch_cb_count) += 1; er = epoch_currecord(epoch); ck_epoch_call(&er->er_record, cb, (ck_epoch_cb_t *)callback); critical_exit(); return; boottime: callback(ctx); } static void epoch_call_task(void *arg __unused) { ck_stack_entry_t *cursor, *head, *next; ck_epoch_record_t *record; epoch_record_t er; epoch_t epoch; ck_stack_t cb_stack; int i, npending, total; ck_stack_init(&cb_stack); critical_enter(); epoch_enter(global_epoch); - for (total = i = 0; i < epoch_count; i++) { - if (__predict_false((epoch = allepochs[i]) == NULL)) + for (total = i = 0; i != MAX_EPOCHS; i++) { + epoch = epoch_array + i; + if (__predict_false( + atomic_load_acq_int(&epoch->e_in_use) == 0)) continue; er = epoch_currecord(epoch); record = &er->er_record; if ((npending = record->n_pending) == 0) continue; ck_epoch_poll_deferred(record, &cb_stack); total += npending - record->n_pending; } epoch_exit(global_epoch); *DPCPU_PTR(epoch_cb_count) -= total; critical_exit(); counter_u64_add(epoch_call_count, total); counter_u64_add(epoch_call_task_count, 1); head = ck_stack_batch_pop_npsc(&cb_stack); for (cursor = head; cursor != NULL; cursor = next) { struct ck_epoch_entry *entry = ck_epoch_entry_container(cursor); next = CK_STACK_NEXT(cursor); entry->function(entry); } } int in_epoch_verbose(epoch_t epoch, int dump_onfail) { struct epoch_tracker *tdwait; struct thread *td; epoch_record_t er; td = curthread; if (THREAD_CAN_SLEEP()) return (0); if (__predict_false((epoch) == NULL)) return (0); critical_enter(); er = epoch_currecord(epoch); TAILQ_FOREACH(tdwait, &er->er_tdlist, et_link) if (tdwait->et_td == td) { critical_exit(); return (1); } #ifdef INVARIANTS if (dump_onfail) { MPASS(td->td_pinned); printf("cpu: %d id: %d\n", curcpu, td->td_tid); TAILQ_FOREACH(tdwait, &er->er_tdlist, et_link) printf("td_tid: %d ", tdwait->et_td->td_tid); printf("\n"); } #endif critical_exit(); return (0); } int in_epoch(epoch_t epoch) { return (in_epoch_verbose(epoch, 0)); } static void epoch_drain_cb(struct epoch_context *ctx) { struct epoch *epoch = __containerof(ctx, struct epoch_record, er_drain_ctx)->er_parent; if (atomic_fetchadd_int(&epoch->e_drain_count, -1) == 1) { mtx_lock(&epoch->e_drain_mtx); wakeup(epoch); mtx_unlock(&epoch->e_drain_mtx); } } void epoch_drain_callbacks(epoch_t epoch) { epoch_record_t er; struct thread *td; int was_bound; int old_pinned; int old_cpu; int cpu; WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL, "epoch_drain_callbacks() may sleep!"); /* too early in boot to have epoch set up */ if (__predict_false(epoch == NULL)) return; #if !defined(EARLY_AP_STARTUP) if (__predict_false(inited < 2)) return; #endif DROP_GIANT(); sx_xlock(&epoch->e_drain_sx); mtx_lock(&epoch->e_drain_mtx); td = curthread; thread_lock(td); old_cpu = PCPU_GET(cpuid); old_pinned = td->td_pinned; was_bound = sched_is_bound(td); sched_unbind(td); td->td_pinned = 0; CPU_FOREACH(cpu) epoch->e_drain_count++; CPU_FOREACH(cpu) { er = zpcpu_get_cpu(epoch->e_pcpu_record, cpu); sched_bind(td, cpu); epoch_call(epoch, &epoch_drain_cb, &er->er_drain_ctx); } /* restore CPU binding, if any */ if (was_bound != 0) { sched_bind(td, old_cpu); } else { /* get thread back to initial CPU, if any */ if (old_pinned != 0) sched_bind(td, old_cpu); sched_unbind(td); } /* restore pinned after bind */ td->td_pinned = old_pinned; thread_unlock(td); while (epoch->e_drain_count != 0) msleep(epoch, &epoch->e_drain_mtx, PZERO, "EDRAIN", 0); mtx_unlock(&epoch->e_drain_mtx); sx_xunlock(&epoch->e_drain_sx); PICKUP_GIANT(); } Index: projects/clang1100-import/sys/net/iflib.c =================================================================== --- projects/clang1100-import/sys/net/iflib.c (revision 364034) +++ projects/clang1100-import/sys/net/iflib.c (revision 364035) @@ -1,6898 +1,6890 @@ /*- * Copyright (c) 2014-2018, Matthew Macy * 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. Neither the name of Matthew Macy 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("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include "opt_acpi.h" #include "opt_sched.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 #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 "ifdi_if.h" #ifdef PCI_IOV #include #endif #include /* * enable accounting of every mbuf as it comes in to and goes out of * iflib's software descriptor references */ #define MEMORY_LOGGING 0 /* * Enable mbuf vectors for compressing long mbuf chains */ /* * NB: * - Prefetching in tx cleaning should perhaps be a tunable. The distance ahead * we prefetch needs to be determined by the time spent in m_free vis a vis * the cost of a prefetch. This will of course vary based on the workload: * - NFLX's m_free path is dominated by vm-based M_EXT manipulation which * is quite expensive, thus suggesting very little prefetch. * - small packet forwarding which is just returning a single mbuf to * UMA will typically be very fast vis a vis the cost of a memory * access. */ /* * File organization: * - private structures * - iflib private utility functions * - ifnet functions * - vlan registry and other exported functions * - iflib public core functions * * */ MALLOC_DEFINE(M_IFLIB, "iflib", "ifnet library"); #define IFLIB_RXEOF_MORE (1U << 0) #define IFLIB_RXEOF_EMPTY (2U << 0) struct iflib_txq; typedef struct iflib_txq *iflib_txq_t; struct iflib_rxq; typedef struct iflib_rxq *iflib_rxq_t; struct iflib_fl; typedef struct iflib_fl *iflib_fl_t; struct iflib_ctx; static void iru_init(if_rxd_update_t iru, iflib_rxq_t rxq, uint8_t flid); static void iflib_timer(void *arg); typedef struct iflib_filter_info { driver_filter_t *ifi_filter; void *ifi_filter_arg; struct grouptask *ifi_task; void *ifi_ctx; } *iflib_filter_info_t; struct iflib_ctx { KOBJ_FIELDS; /* * Pointer to hardware driver's softc */ void *ifc_softc; device_t ifc_dev; if_t ifc_ifp; cpuset_t ifc_cpus; if_shared_ctx_t ifc_sctx; struct if_softc_ctx ifc_softc_ctx; struct sx ifc_ctx_sx; struct mtx ifc_state_mtx; iflib_txq_t ifc_txqs; iflib_rxq_t ifc_rxqs; uint32_t ifc_if_flags; uint32_t ifc_flags; uint32_t ifc_max_fl_buf_size; uint32_t ifc_rx_mbuf_sz; int ifc_link_state; int ifc_watchdog_events; struct cdev *ifc_led_dev; struct resource *ifc_msix_mem; struct if_irq ifc_legacy_irq; struct grouptask ifc_admin_task; struct grouptask ifc_vflr_task; struct iflib_filter_info ifc_filter_info; struct ifmedia ifc_media; struct ifmedia *ifc_mediap; struct sysctl_oid *ifc_sysctl_node; uint16_t ifc_sysctl_ntxqs; uint16_t ifc_sysctl_nrxqs; uint16_t ifc_sysctl_qs_eq_override; uint16_t ifc_sysctl_rx_budget; uint16_t ifc_sysctl_tx_abdicate; uint16_t ifc_sysctl_core_offset; #define CORE_OFFSET_UNSPECIFIED 0xffff uint8_t ifc_sysctl_separate_txrx; qidx_t ifc_sysctl_ntxds[8]; qidx_t ifc_sysctl_nrxds[8]; struct if_txrx ifc_txrx; #define isc_txd_encap ifc_txrx.ift_txd_encap #define isc_txd_flush ifc_txrx.ift_txd_flush #define isc_txd_credits_update ifc_txrx.ift_txd_credits_update #define isc_rxd_available ifc_txrx.ift_rxd_available #define isc_rxd_pkt_get ifc_txrx.ift_rxd_pkt_get #define isc_rxd_refill ifc_txrx.ift_rxd_refill #define isc_rxd_flush ifc_txrx.ift_rxd_flush #define isc_legacy_intr ifc_txrx.ift_legacy_intr eventhandler_tag ifc_vlan_attach_event; eventhandler_tag ifc_vlan_detach_event; struct ether_addr ifc_mac; }; void * iflib_get_softc(if_ctx_t ctx) { return (ctx->ifc_softc); } device_t iflib_get_dev(if_ctx_t ctx) { return (ctx->ifc_dev); } if_t iflib_get_ifp(if_ctx_t ctx) { return (ctx->ifc_ifp); } struct ifmedia * iflib_get_media(if_ctx_t ctx) { return (ctx->ifc_mediap); } uint32_t iflib_get_flags(if_ctx_t ctx) { return (ctx->ifc_flags); } void iflib_set_mac(if_ctx_t ctx, uint8_t mac[ETHER_ADDR_LEN]) { bcopy(mac, ctx->ifc_mac.octet, ETHER_ADDR_LEN); } if_softc_ctx_t iflib_get_softc_ctx(if_ctx_t ctx) { return (&ctx->ifc_softc_ctx); } if_shared_ctx_t iflib_get_sctx(if_ctx_t ctx) { return (ctx->ifc_sctx); } #define IP_ALIGNED(m) ((((uintptr_t)(m)->m_data) & 0x3) == 0x2) #define CACHE_PTR_INCREMENT (CACHE_LINE_SIZE/sizeof(void*)) #define CACHE_PTR_NEXT(ptr) ((void *)(((uintptr_t)(ptr)+CACHE_LINE_SIZE-1) & (CACHE_LINE_SIZE-1))) #define LINK_ACTIVE(ctx) ((ctx)->ifc_link_state == LINK_STATE_UP) #define CTX_IS_VF(ctx) ((ctx)->ifc_sctx->isc_flags & IFLIB_IS_VF) typedef struct iflib_sw_rx_desc_array { bus_dmamap_t *ifsd_map; /* bus_dma maps for packet */ struct mbuf **ifsd_m; /* pkthdr mbufs */ caddr_t *ifsd_cl; /* direct cluster pointer for rx */ bus_addr_t *ifsd_ba; /* bus addr of cluster for rx */ } iflib_rxsd_array_t; typedef struct iflib_sw_tx_desc_array { bus_dmamap_t *ifsd_map; /* bus_dma maps for packet */ bus_dmamap_t *ifsd_tso_map; /* bus_dma maps for TSO packet */ struct mbuf **ifsd_m; /* pkthdr mbufs */ } if_txsd_vec_t; /* magic number that should be high enough for any hardware */ #define IFLIB_MAX_TX_SEGS 128 #define IFLIB_RX_COPY_THRESH 128 #define IFLIB_MAX_RX_REFRESH 32 /* The minimum descriptors per second before we start coalescing */ #define IFLIB_MIN_DESC_SEC 16384 #define IFLIB_DEFAULT_TX_UPDATE_FREQ 16 #define IFLIB_QUEUE_IDLE 0 #define IFLIB_QUEUE_HUNG 1 #define IFLIB_QUEUE_WORKING 2 /* maximum number of txqs that can share an rx interrupt */ #define IFLIB_MAX_TX_SHARED_INTR 4 /* this should really scale with ring size - this is a fairly arbitrary value */ #define TX_BATCH_SIZE 32 #define IFLIB_RESTART_BUDGET 8 #define CSUM_OFFLOAD (CSUM_IP_TSO|CSUM_IP6_TSO|CSUM_IP| \ CSUM_IP_UDP|CSUM_IP_TCP|CSUM_IP_SCTP| \ CSUM_IP6_UDP|CSUM_IP6_TCP|CSUM_IP6_SCTP) struct iflib_txq { qidx_t ift_in_use; qidx_t ift_cidx; qidx_t ift_cidx_processed; qidx_t ift_pidx; uint8_t ift_gen; uint8_t ift_br_offset; uint16_t ift_npending; uint16_t ift_db_pending; uint16_t ift_rs_pending; /* implicit pad */ uint8_t ift_txd_size[8]; uint64_t ift_processed; uint64_t ift_cleaned; uint64_t ift_cleaned_prev; #if MEMORY_LOGGING uint64_t ift_enqueued; uint64_t ift_dequeued; #endif uint64_t ift_no_tx_dma_setup; uint64_t ift_no_desc_avail; uint64_t ift_mbuf_defrag_failed; uint64_t ift_mbuf_defrag; uint64_t ift_map_failed; uint64_t ift_txd_encap_efbig; uint64_t ift_pullups; uint64_t ift_last_timer_tick; struct mtx ift_mtx; struct mtx ift_db_mtx; /* constant values */ if_ctx_t ift_ctx; struct ifmp_ring *ift_br; struct grouptask ift_task; qidx_t ift_size; uint16_t ift_id; struct callout ift_timer; if_txsd_vec_t ift_sds; uint8_t ift_qstatus; uint8_t ift_closed; uint8_t ift_update_freq; struct iflib_filter_info ift_filter_info; bus_dma_tag_t ift_buf_tag; bus_dma_tag_t ift_tso_buf_tag; iflib_dma_info_t ift_ifdi; #define MTX_NAME_LEN 32 char ift_mtx_name[MTX_NAME_LEN]; bus_dma_segment_t ift_segs[IFLIB_MAX_TX_SEGS] __aligned(CACHE_LINE_SIZE); #ifdef IFLIB_DIAGNOSTICS uint64_t ift_cpu_exec_count[256]; #endif } __aligned(CACHE_LINE_SIZE); struct iflib_fl { qidx_t ifl_cidx; qidx_t ifl_pidx; qidx_t ifl_credits; uint8_t ifl_gen; uint8_t ifl_rxd_size; #if MEMORY_LOGGING uint64_t ifl_m_enqueued; uint64_t ifl_m_dequeued; uint64_t ifl_cl_enqueued; uint64_t ifl_cl_dequeued; #endif /* implicit pad */ bitstr_t *ifl_rx_bitmap; qidx_t ifl_fragidx; /* constant */ qidx_t ifl_size; uint16_t ifl_buf_size; uint16_t ifl_cltype; uma_zone_t ifl_zone; iflib_rxsd_array_t ifl_sds; iflib_rxq_t ifl_rxq; uint8_t ifl_id; bus_dma_tag_t ifl_buf_tag; iflib_dma_info_t ifl_ifdi; uint64_t ifl_bus_addrs[IFLIB_MAX_RX_REFRESH] __aligned(CACHE_LINE_SIZE); qidx_t ifl_rxd_idxs[IFLIB_MAX_RX_REFRESH]; } __aligned(CACHE_LINE_SIZE); static inline qidx_t get_inuse(int size, qidx_t cidx, qidx_t pidx, uint8_t gen) { qidx_t used; if (pidx > cidx) used = pidx - cidx; else if (pidx < cidx) used = size - cidx + pidx; else if (gen == 0 && pidx == cidx) used = 0; else if (gen == 1 && pidx == cidx) used = size; else panic("bad state"); return (used); } #define TXQ_AVAIL(txq) (txq->ift_size - get_inuse(txq->ift_size, txq->ift_cidx, txq->ift_pidx, txq->ift_gen)) #define IDXDIFF(head, tail, wrap) \ ((head) >= (tail) ? (head) - (tail) : (wrap) - (tail) + (head)) struct iflib_rxq { if_ctx_t ifr_ctx; iflib_fl_t ifr_fl; uint64_t ifr_rx_irq; struct pfil_head *pfil; /* * If there is a separate completion queue (IFLIB_HAS_RXCQ), this is * the command queue consumer index. Otherwise it's unused. */ qidx_t ifr_cq_cidx; uint16_t ifr_id; uint8_t ifr_nfl; uint8_t ifr_ntxqirq; uint8_t ifr_txqid[IFLIB_MAX_TX_SHARED_INTR]; uint8_t ifr_fl_offset; struct lro_ctrl ifr_lc; struct grouptask ifr_task; struct callout ifr_watchdog; struct iflib_filter_info ifr_filter_info; iflib_dma_info_t ifr_ifdi; /* dynamically allocate if any drivers need a value substantially larger than this */ struct if_rxd_frag ifr_frags[IFLIB_MAX_RX_SEGS] __aligned(CACHE_LINE_SIZE); #ifdef IFLIB_DIAGNOSTICS uint64_t ifr_cpu_exec_count[256]; #endif } __aligned(CACHE_LINE_SIZE); typedef struct if_rxsd { caddr_t *ifsd_cl; iflib_fl_t ifsd_fl; } *if_rxsd_t; /* multiple of word size */ #ifdef __LP64__ #define PKT_INFO_SIZE 6 #define RXD_INFO_SIZE 5 #define PKT_TYPE uint64_t #else #define PKT_INFO_SIZE 11 #define RXD_INFO_SIZE 8 #define PKT_TYPE uint32_t #endif #define PKT_LOOP_BOUND ((PKT_INFO_SIZE/3)*3) #define RXD_LOOP_BOUND ((RXD_INFO_SIZE/4)*4) typedef struct if_pkt_info_pad { PKT_TYPE pkt_val[PKT_INFO_SIZE]; } *if_pkt_info_pad_t; typedef struct if_rxd_info_pad { PKT_TYPE rxd_val[RXD_INFO_SIZE]; } *if_rxd_info_pad_t; CTASSERT(sizeof(struct if_pkt_info_pad) == sizeof(struct if_pkt_info)); CTASSERT(sizeof(struct if_rxd_info_pad) == sizeof(struct if_rxd_info)); static inline void pkt_info_zero(if_pkt_info_t pi) { if_pkt_info_pad_t pi_pad; pi_pad = (if_pkt_info_pad_t)pi; pi_pad->pkt_val[0] = 0; pi_pad->pkt_val[1] = 0; pi_pad->pkt_val[2] = 0; pi_pad->pkt_val[3] = 0; pi_pad->pkt_val[4] = 0; pi_pad->pkt_val[5] = 0; #ifndef __LP64__ pi_pad->pkt_val[6] = 0; pi_pad->pkt_val[7] = 0; pi_pad->pkt_val[8] = 0; pi_pad->pkt_val[9] = 0; pi_pad->pkt_val[10] = 0; #endif } static device_method_t iflib_pseudo_methods[] = { DEVMETHOD(device_attach, noop_attach), DEVMETHOD(device_detach, iflib_pseudo_detach), DEVMETHOD_END }; driver_t iflib_pseudodriver = { "iflib_pseudo", iflib_pseudo_methods, sizeof(struct iflib_ctx), }; static inline void rxd_info_zero(if_rxd_info_t ri) { if_rxd_info_pad_t ri_pad; int i; ri_pad = (if_rxd_info_pad_t)ri; for (i = 0; i < RXD_LOOP_BOUND; i += 4) { ri_pad->rxd_val[i] = 0; ri_pad->rxd_val[i+1] = 0; ri_pad->rxd_val[i+2] = 0; ri_pad->rxd_val[i+3] = 0; } #ifdef __LP64__ ri_pad->rxd_val[RXD_INFO_SIZE-1] = 0; #endif } /* * Only allow a single packet to take up most 1/nth of the tx ring */ #define MAX_SINGLE_PACKET_FRACTION 12 #define IF_BAD_DMA (bus_addr_t)-1 #define CTX_ACTIVE(ctx) ((if_getdrvflags((ctx)->ifc_ifp) & IFF_DRV_RUNNING)) #define CTX_LOCK_INIT(_sc) sx_init(&(_sc)->ifc_ctx_sx, "iflib ctx lock") #define CTX_LOCK(ctx) sx_xlock(&(ctx)->ifc_ctx_sx) #define CTX_UNLOCK(ctx) sx_xunlock(&(ctx)->ifc_ctx_sx) #define CTX_LOCK_DESTROY(ctx) sx_destroy(&(ctx)->ifc_ctx_sx) #define STATE_LOCK_INIT(_sc, _name) mtx_init(&(_sc)->ifc_state_mtx, _name, "iflib state lock", MTX_DEF) #define STATE_LOCK(ctx) mtx_lock(&(ctx)->ifc_state_mtx) #define STATE_UNLOCK(ctx) mtx_unlock(&(ctx)->ifc_state_mtx) #define STATE_LOCK_DESTROY(ctx) mtx_destroy(&(ctx)->ifc_state_mtx) #define CALLOUT_LOCK(txq) mtx_lock(&txq->ift_mtx) #define CALLOUT_UNLOCK(txq) mtx_unlock(&txq->ift_mtx) void iflib_set_detach(if_ctx_t ctx) { STATE_LOCK(ctx); ctx->ifc_flags |= IFC_IN_DETACH; STATE_UNLOCK(ctx); } /* Our boot-time initialization hook */ static int iflib_module_event_handler(module_t, int, void *); static moduledata_t iflib_moduledata = { "iflib", iflib_module_event_handler, NULL }; DECLARE_MODULE(iflib, iflib_moduledata, SI_SUB_INIT_IF, SI_ORDER_ANY); MODULE_VERSION(iflib, 1); MODULE_DEPEND(iflib, pci, 1, 1, 1); MODULE_DEPEND(iflib, ether, 1, 1, 1); TASKQGROUP_DEFINE(if_io_tqg, mp_ncpus, 1); TASKQGROUP_DEFINE(if_config_tqg, 1, 1); #ifndef IFLIB_DEBUG_COUNTERS #ifdef INVARIANTS #define IFLIB_DEBUG_COUNTERS 1 #else #define IFLIB_DEBUG_COUNTERS 0 #endif /* !INVARIANTS */ #endif static SYSCTL_NODE(_net, OID_AUTO, iflib, CTLFLAG_RD | CTLFLAG_MPSAFE, 0, "iflib driver parameters"); /* * XXX need to ensure that this can't accidentally cause the head to be moved backwards */ static int iflib_min_tx_latency = 0; SYSCTL_INT(_net_iflib, OID_AUTO, min_tx_latency, CTLFLAG_RW, &iflib_min_tx_latency, 0, "minimize transmit latency at the possible expense of throughput"); static int iflib_no_tx_batch = 0; SYSCTL_INT(_net_iflib, OID_AUTO, no_tx_batch, CTLFLAG_RW, &iflib_no_tx_batch, 0, "minimize transmit latency at the possible expense of throughput"); #if IFLIB_DEBUG_COUNTERS static int iflib_tx_seen; static int iflib_tx_sent; static int iflib_tx_encap; static int iflib_rx_allocs; static int iflib_fl_refills; static int iflib_fl_refills_large; static int iflib_tx_frees; SYSCTL_INT(_net_iflib, OID_AUTO, tx_seen, CTLFLAG_RD, &iflib_tx_seen, 0, "# TX mbufs seen"); SYSCTL_INT(_net_iflib, OID_AUTO, tx_sent, CTLFLAG_RD, &iflib_tx_sent, 0, "# TX mbufs sent"); SYSCTL_INT(_net_iflib, OID_AUTO, tx_encap, CTLFLAG_RD, &iflib_tx_encap, 0, "# TX mbufs encapped"); SYSCTL_INT(_net_iflib, OID_AUTO, tx_frees, CTLFLAG_RD, &iflib_tx_frees, 0, "# TX frees"); SYSCTL_INT(_net_iflib, OID_AUTO, rx_allocs, CTLFLAG_RD, &iflib_rx_allocs, 0, "# RX allocations"); SYSCTL_INT(_net_iflib, OID_AUTO, fl_refills, CTLFLAG_RD, &iflib_fl_refills, 0, "# refills"); SYSCTL_INT(_net_iflib, OID_AUTO, fl_refills_large, CTLFLAG_RD, &iflib_fl_refills_large, 0, "# large refills"); static int iflib_txq_drain_flushing; static int iflib_txq_drain_oactive; static int iflib_txq_drain_notready; SYSCTL_INT(_net_iflib, OID_AUTO, txq_drain_flushing, CTLFLAG_RD, &iflib_txq_drain_flushing, 0, "# drain flushes"); SYSCTL_INT(_net_iflib, OID_AUTO, txq_drain_oactive, CTLFLAG_RD, &iflib_txq_drain_oactive, 0, "# drain oactives"); SYSCTL_INT(_net_iflib, OID_AUTO, txq_drain_notready, CTLFLAG_RD, &iflib_txq_drain_notready, 0, "# drain notready"); static int iflib_encap_load_mbuf_fail; static int iflib_encap_pad_mbuf_fail; static int iflib_encap_txq_avail_fail; static int iflib_encap_txd_encap_fail; SYSCTL_INT(_net_iflib, OID_AUTO, encap_load_mbuf_fail, CTLFLAG_RD, &iflib_encap_load_mbuf_fail, 0, "# busdma load failures"); SYSCTL_INT(_net_iflib, OID_AUTO, encap_pad_mbuf_fail, CTLFLAG_RD, &iflib_encap_pad_mbuf_fail, 0, "# runt frame pad failures"); SYSCTL_INT(_net_iflib, OID_AUTO, encap_txq_avail_fail, CTLFLAG_RD, &iflib_encap_txq_avail_fail, 0, "# txq avail failures"); SYSCTL_INT(_net_iflib, OID_AUTO, encap_txd_encap_fail, CTLFLAG_RD, &iflib_encap_txd_encap_fail, 0, "# driver encap failures"); static int iflib_task_fn_rxs; static int iflib_rx_intr_enables; static int iflib_fast_intrs; static int iflib_rx_unavail; static int iflib_rx_ctx_inactive; static int iflib_rx_if_input; static int iflib_rxd_flush; static int iflib_verbose_debug; SYSCTL_INT(_net_iflib, OID_AUTO, task_fn_rx, CTLFLAG_RD, &iflib_task_fn_rxs, 0, "# task_fn_rx calls"); SYSCTL_INT(_net_iflib, OID_AUTO, rx_intr_enables, CTLFLAG_RD, &iflib_rx_intr_enables, 0, "# RX intr enables"); SYSCTL_INT(_net_iflib, OID_AUTO, fast_intrs, CTLFLAG_RD, &iflib_fast_intrs, 0, "# fast_intr calls"); SYSCTL_INT(_net_iflib, OID_AUTO, rx_unavail, CTLFLAG_RD, &iflib_rx_unavail, 0, "# times rxeof called with no available data"); SYSCTL_INT(_net_iflib, OID_AUTO, rx_ctx_inactive, CTLFLAG_RD, &iflib_rx_ctx_inactive, 0, "# times rxeof called with inactive context"); SYSCTL_INT(_net_iflib, OID_AUTO, rx_if_input, CTLFLAG_RD, &iflib_rx_if_input, 0, "# times rxeof called if_input"); SYSCTL_INT(_net_iflib, OID_AUTO, rxd_flush, CTLFLAG_RD, &iflib_rxd_flush, 0, "# times rxd_flush called"); SYSCTL_INT(_net_iflib, OID_AUTO, verbose_debug, CTLFLAG_RW, &iflib_verbose_debug, 0, "enable verbose debugging"); #define DBG_COUNTER_INC(name) atomic_add_int(&(iflib_ ## name), 1) static void iflib_debug_reset(void) { iflib_tx_seen = iflib_tx_sent = iflib_tx_encap = iflib_rx_allocs = iflib_fl_refills = iflib_fl_refills_large = iflib_tx_frees = iflib_txq_drain_flushing = iflib_txq_drain_oactive = iflib_txq_drain_notready = iflib_encap_load_mbuf_fail = iflib_encap_pad_mbuf_fail = iflib_encap_txq_avail_fail = iflib_encap_txd_encap_fail = iflib_task_fn_rxs = iflib_rx_intr_enables = iflib_fast_intrs = iflib_rx_unavail = iflib_rx_ctx_inactive = iflib_rx_if_input = iflib_rxd_flush = 0; } #else #define DBG_COUNTER_INC(name) static void iflib_debug_reset(void) {} #endif #define IFLIB_DEBUG 0 static void iflib_tx_structures_free(if_ctx_t ctx); static void iflib_rx_structures_free(if_ctx_t ctx); static int iflib_queues_alloc(if_ctx_t ctx); static int iflib_tx_credits_update(if_ctx_t ctx, iflib_txq_t txq); static int iflib_rxd_avail(if_ctx_t ctx, iflib_rxq_t rxq, qidx_t cidx, qidx_t budget); static int iflib_qset_structures_setup(if_ctx_t ctx); static int iflib_msix_init(if_ctx_t ctx); static int iflib_legacy_setup(if_ctx_t ctx, driver_filter_t filter, void *filterarg, int *rid, const char *str); static void iflib_txq_check_drain(iflib_txq_t txq, int budget); static uint32_t iflib_txq_can_drain(struct ifmp_ring *); #ifdef ALTQ static void iflib_altq_if_start(if_t ifp); static int iflib_altq_if_transmit(if_t ifp, struct mbuf *m); #endif static int iflib_register(if_ctx_t); static void iflib_deregister(if_ctx_t); static void iflib_unregister_vlan_handlers(if_ctx_t ctx); static uint16_t iflib_get_mbuf_size_for(unsigned int size); static void iflib_init_locked(if_ctx_t ctx); static void iflib_add_device_sysctl_pre(if_ctx_t ctx); static void iflib_add_device_sysctl_post(if_ctx_t ctx); static void iflib_ifmp_purge(iflib_txq_t txq); static void _iflib_pre_assert(if_softc_ctx_t scctx); static void iflib_if_init_locked(if_ctx_t ctx); static void iflib_free_intr_mem(if_ctx_t ctx); #ifndef __NO_STRICT_ALIGNMENT static struct mbuf * iflib_fixup_rx(struct mbuf *m); #endif static SLIST_HEAD(cpu_offset_list, cpu_offset) cpu_offsets = SLIST_HEAD_INITIALIZER(cpu_offsets); struct cpu_offset { SLIST_ENTRY(cpu_offset) entries; cpuset_t set; unsigned int refcount; uint16_t offset; }; static struct mtx cpu_offset_mtx; MTX_SYSINIT(iflib_cpu_offset, &cpu_offset_mtx, "iflib_cpu_offset lock", MTX_DEF); DEBUGNET_DEFINE(iflib); static int iflib_num_rx_descs(if_ctx_t ctx) { if_softc_ctx_t scctx = &ctx->ifc_softc_ctx; if_shared_ctx_t sctx = ctx->ifc_sctx; uint16_t first_rxq = (sctx->isc_flags & IFLIB_HAS_RXCQ) ? 1 : 0; return scctx->isc_nrxd[first_rxq]; } static int iflib_num_tx_descs(if_ctx_t ctx) { if_softc_ctx_t scctx = &ctx->ifc_softc_ctx; if_shared_ctx_t sctx = ctx->ifc_sctx; uint16_t first_txq = (sctx->isc_flags & IFLIB_HAS_TXCQ) ? 1 : 0; return scctx->isc_ntxd[first_txq]; } #ifdef DEV_NETMAP #include #include #include MODULE_DEPEND(iflib, netmap, 1, 1, 1); static int netmap_fl_refill(iflib_rxq_t rxq, struct netmap_kring *kring, uint32_t nm_i, bool init); /* * device-specific sysctl variables: * * iflib_crcstrip: 0: keep CRC in rx frames (default), 1: strip it. * During regular operations the CRC is stripped, but on some * hardware reception of frames not multiple of 64 is slower, * so using crcstrip=0 helps in benchmarks. * * iflib_rx_miss, iflib_rx_miss_bufs: * count packets that might be missed due to lost interrupts. */ SYSCTL_DECL(_dev_netmap); /* * The xl driver by default strips CRCs and we do not override it. */ int iflib_crcstrip = 1; SYSCTL_INT(_dev_netmap, OID_AUTO, iflib_crcstrip, CTLFLAG_RW, &iflib_crcstrip, 1, "strip CRC on RX frames"); int iflib_rx_miss, iflib_rx_miss_bufs; SYSCTL_INT(_dev_netmap, OID_AUTO, iflib_rx_miss, CTLFLAG_RW, &iflib_rx_miss, 0, "potentially missed RX intr"); SYSCTL_INT(_dev_netmap, OID_AUTO, iflib_rx_miss_bufs, CTLFLAG_RW, &iflib_rx_miss_bufs, 0, "potentially missed RX intr bufs"); /* * Register/unregister. We are already under netmap lock. * Only called on the first register or the last unregister. */ static int iflib_netmap_register(struct netmap_adapter *na, int onoff) { if_t ifp = na->ifp; if_ctx_t ctx = ifp->if_softc; int status; CTX_LOCK(ctx); IFDI_INTR_DISABLE(ctx); /* Tell the stack that the interface is no longer active */ ifp->if_drv_flags &= ~(IFF_DRV_RUNNING | IFF_DRV_OACTIVE); if (!CTX_IS_VF(ctx)) IFDI_CRCSTRIP_SET(ctx, onoff, iflib_crcstrip); iflib_stop(ctx); /* * Enable (or disable) netmap flags, and intercept (or restore) * ifp->if_transmit. This is done once the device has been stopped * to prevent race conditions. */ if (onoff) { nm_set_native_flags(na); } else { nm_clear_native_flags(na); } iflib_init_locked(ctx); IFDI_CRCSTRIP_SET(ctx, onoff, iflib_crcstrip); // XXX why twice ? status = ifp->if_drv_flags & IFF_DRV_RUNNING ? 0 : 1; if (status) nm_clear_native_flags(na); CTX_UNLOCK(ctx); return (status); } static int netmap_fl_refill(iflib_rxq_t rxq, struct netmap_kring *kring, uint32_t nm_i, bool init) { struct netmap_adapter *na = kring->na; u_int const lim = kring->nkr_num_slots - 1; u_int head = kring->rhead; struct netmap_ring *ring = kring->ring; bus_dmamap_t *map; struct if_rxd_update iru; if_ctx_t ctx = rxq->ifr_ctx; iflib_fl_t fl = &rxq->ifr_fl[0]; uint32_t refill_pidx, nic_i; #if IFLIB_DEBUG_COUNTERS int rf_count = 0; #endif if (nm_i == head && __predict_true(!init)) return 0; iru_init(&iru, rxq, 0 /* flid */); map = fl->ifl_sds.ifsd_map; refill_pidx = netmap_idx_k2n(kring, nm_i); /* * IMPORTANT: we must leave one free slot in the ring, * so move head back by one unit */ head = nm_prev(head, lim); nic_i = UINT_MAX; DBG_COUNTER_INC(fl_refills); while (nm_i != head) { #if IFLIB_DEBUG_COUNTERS if (++rf_count == 9) DBG_COUNTER_INC(fl_refills_large); #endif for (int tmp_pidx = 0; tmp_pidx < IFLIB_MAX_RX_REFRESH && nm_i != head; tmp_pidx++) { struct netmap_slot *slot = &ring->slot[nm_i]; void *addr = PNMB(na, slot, &fl->ifl_bus_addrs[tmp_pidx]); uint32_t nic_i_dma = refill_pidx; nic_i = netmap_idx_k2n(kring, nm_i); MPASS(tmp_pidx < IFLIB_MAX_RX_REFRESH); if (addr == NETMAP_BUF_BASE(na)) /* bad buf */ return netmap_ring_reinit(kring); if (__predict_false(init)) { netmap_load_map(na, fl->ifl_buf_tag, map[nic_i], addr); } else if (slot->flags & NS_BUF_CHANGED) { /* buffer has changed, reload map */ netmap_reload_map(na, fl->ifl_buf_tag, map[nic_i], addr); } slot->flags &= ~NS_BUF_CHANGED; nm_i = nm_next(nm_i, lim); fl->ifl_rxd_idxs[tmp_pidx] = nic_i = nm_next(nic_i, lim); if (nm_i != head && tmp_pidx < IFLIB_MAX_RX_REFRESH-1) continue; iru.iru_pidx = refill_pidx; iru.iru_count = tmp_pidx+1; ctx->isc_rxd_refill(ctx->ifc_softc, &iru); refill_pidx = nic_i; for (int n = 0; n < iru.iru_count; n++) { bus_dmamap_sync(fl->ifl_buf_tag, map[nic_i_dma], BUS_DMASYNC_PREREAD); /* XXX - change this to not use the netmap func*/ nic_i_dma = nm_next(nic_i_dma, lim); } } } kring->nr_hwcur = head; bus_dmamap_sync(fl->ifl_ifdi->idi_tag, fl->ifl_ifdi->idi_map, BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); if (__predict_true(nic_i != UINT_MAX)) { ctx->isc_rxd_flush(ctx->ifc_softc, rxq->ifr_id, fl->ifl_id, nic_i); DBG_COUNTER_INC(rxd_flush); } return (0); } /* * Reconcile kernel and user view of the transmit ring. * * All information is in the kring. * Userspace wants to send packets up to the one before kring->rhead, * kernel knows kring->nr_hwcur is the first unsent packet. * * Here we push packets out (as many as possible), and possibly * reclaim buffers from previously completed transmission. * * The caller (netmap) guarantees that there is only one instance * running at any time. Any interference with other driver * methods should be handled by the individual drivers. */ static int iflib_netmap_txsync(struct netmap_kring *kring, int flags) { struct netmap_adapter *na = kring->na; if_t ifp = na->ifp; struct netmap_ring *ring = kring->ring; u_int nm_i; /* index into the netmap kring */ u_int nic_i; /* index into the NIC ring */ u_int n; u_int const lim = kring->nkr_num_slots - 1; u_int const head = kring->rhead; struct if_pkt_info pi; /* * interrupts on every tx packet are expensive so request * them every half ring, or where NS_REPORT is set */ u_int report_frequency = kring->nkr_num_slots >> 1; /* device-specific */ if_ctx_t ctx = ifp->if_softc; iflib_txq_t txq = &ctx->ifc_txqs[kring->ring_id]; bus_dmamap_sync(txq->ift_ifdi->idi_tag, txq->ift_ifdi->idi_map, BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); /* * First part: process new packets to send. * nm_i is the current index in the netmap kring, * nic_i is the corresponding index in the NIC ring. * * If we have packets to send (nm_i != head) * iterate over the netmap ring, fetch length and update * the corresponding slot in the NIC ring. Some drivers also * need to update the buffer's physical address in the NIC slot * even NS_BUF_CHANGED is not set (PNMB computes the addresses). * * The netmap_reload_map() calls is especially expensive, * even when (as in this case) the tag is 0, so do only * when the buffer has actually changed. * * If possible do not set the report/intr bit on all slots, * but only a few times per ring or when NS_REPORT is set. * * Finally, on 10G and faster drivers, it might be useful * to prefetch the next slot and txr entry. */ nm_i = kring->nr_hwcur; if (nm_i != head) { /* we have new packets to send */ pkt_info_zero(&pi); pi.ipi_segs = txq->ift_segs; pi.ipi_qsidx = kring->ring_id; nic_i = netmap_idx_k2n(kring, nm_i); __builtin_prefetch(&ring->slot[nm_i]); __builtin_prefetch(&txq->ift_sds.ifsd_m[nic_i]); __builtin_prefetch(&txq->ift_sds.ifsd_map[nic_i]); for (n = 0; nm_i != head; n++) { struct netmap_slot *slot = &ring->slot[nm_i]; u_int len = slot->len; uint64_t paddr; void *addr = PNMB(na, slot, &paddr); int flags = (slot->flags & NS_REPORT || nic_i == 0 || nic_i == report_frequency) ? IPI_TX_INTR : 0; /* device-specific */ pi.ipi_len = len; pi.ipi_segs[0].ds_addr = paddr; pi.ipi_segs[0].ds_len = len; pi.ipi_nsegs = 1; pi.ipi_ndescs = 0; pi.ipi_pidx = nic_i; pi.ipi_flags = flags; /* Fill the slot in the NIC ring. */ ctx->isc_txd_encap(ctx->ifc_softc, &pi); DBG_COUNTER_INC(tx_encap); /* prefetch for next round */ __builtin_prefetch(&ring->slot[nm_i + 1]); __builtin_prefetch(&txq->ift_sds.ifsd_m[nic_i + 1]); __builtin_prefetch(&txq->ift_sds.ifsd_map[nic_i + 1]); NM_CHECK_ADDR_LEN(na, addr, len); if (slot->flags & NS_BUF_CHANGED) { /* buffer has changed, reload map */ netmap_reload_map(na, txq->ift_buf_tag, txq->ift_sds.ifsd_map[nic_i], addr); } /* make sure changes to the buffer are synced */ bus_dmamap_sync(txq->ift_buf_tag, txq->ift_sds.ifsd_map[nic_i], BUS_DMASYNC_PREWRITE); slot->flags &= ~(NS_REPORT | NS_BUF_CHANGED); nm_i = nm_next(nm_i, lim); nic_i = nm_next(nic_i, lim); } kring->nr_hwcur = nm_i; /* synchronize the NIC ring */ bus_dmamap_sync(txq->ift_ifdi->idi_tag, txq->ift_ifdi->idi_map, BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); /* (re)start the tx unit up to slot nic_i (excluded) */ ctx->isc_txd_flush(ctx->ifc_softc, txq->ift_id, nic_i); } /* * Second part: reclaim buffers for completed transmissions. * * If there are unclaimed buffers, attempt to reclaim them. * If none are reclaimed, and TX IRQs are not in use, do an initial * minimal delay, then trigger the tx handler which will spin in the * group task queue. */ if (kring->nr_hwtail != nm_prev(kring->nr_hwcur, lim)) { if (iflib_tx_credits_update(ctx, txq)) { /* some tx completed, increment avail */ nic_i = txq->ift_cidx_processed; kring->nr_hwtail = nm_prev(netmap_idx_n2k(kring, nic_i), lim); } } if (!(ctx->ifc_flags & IFC_NETMAP_TX_IRQ)) if (kring->nr_hwtail != nm_prev(kring->nr_hwcur, lim)) { callout_reset_on(&txq->ift_timer, hz < 2000 ? 1 : hz / 1000, iflib_timer, txq, txq->ift_timer.c_cpu); } return (0); } /* * Reconcile kernel and user view of the receive ring. * Same as for the txsync, this routine must be efficient. * The caller guarantees a single invocations, but races against * the rest of the driver should be handled here. * * On call, kring->rhead is the first packet that userspace wants * to keep, and kring->rcur is the wakeup point. * The kernel has previously reported packets up to kring->rtail. * * If (flags & NAF_FORCE_READ) also check for incoming packets irrespective * of whether or not we received an interrupt. */ static int iflib_netmap_rxsync(struct netmap_kring *kring, int flags) { struct netmap_adapter *na = kring->na; struct netmap_ring *ring = kring->ring; if_t ifp = na->ifp; - iflib_fl_t fl; uint32_t nm_i; /* index into the netmap ring */ uint32_t nic_i; /* index into the NIC ring */ - u_int i, n; + u_int n; u_int const lim = kring->nkr_num_slots - 1; - u_int const head = kring->rhead; int force_update = (flags & NAF_FORCE_READ) || kring->nr_kflags & NKR_PENDINTR; - struct if_rxd_info ri; if_ctx_t ctx = ifp->if_softc; iflib_rxq_t rxq = &ctx->ifc_rxqs[kring->ring_id]; - if (head > lim) - return netmap_ring_reinit(kring); + iflib_fl_t fl = &rxq->ifr_fl[0]; + struct if_rxd_info ri; /* - * XXX netmap_fl_refill() only ever (re)fills free list 0 so far. + * netmap only uses free list 0, to avoid out of order consumption + * of receive buffers */ - for (i = 0, fl = rxq->ifr_fl; i < rxq->ifr_nfl; i++, fl++) { - bus_dmamap_sync(fl->ifl_ifdi->idi_tag, fl->ifl_ifdi->idi_map, - BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); - } + bus_dmamap_sync(fl->ifl_ifdi->idi_tag, fl->ifl_ifdi->idi_map, + BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); /* * First part: import newly received packets. * * nm_i is the index of the next free slot in the netmap ring, * nic_i is the index of the next received packet in the NIC ring, * and they may differ in case if_init() has been called while * in netmap mode. For the receive ring we have * * nic_i = rxr->next_check; * nm_i = kring->nr_hwtail (previous) * and * nm_i == (nic_i + kring->nkr_hwofs) % ring_size * * rxr->next_check is set to 0 on a ring reinit */ if (netmap_no_pendintr || force_update) { uint32_t hwtail_lim = nm_prev(kring->nr_hwcur, lim); int crclen = iflib_crcstrip ? 0 : 4; int error, avail; - for (i = 0; i < rxq->ifr_nfl; i++) { - fl = &rxq->ifr_fl[i]; - nic_i = fl->ifl_cidx; - nm_i = netmap_idx_n2k(kring, nic_i); - avail = ctx->isc_rxd_available(ctx->ifc_softc, - rxq->ifr_id, nic_i, USHRT_MAX); - for (n = 0; avail > 0 && nm_i != hwtail_lim; n++, avail--) { - rxd_info_zero(&ri); - ri.iri_frags = rxq->ifr_frags; - ri.iri_qsidx = kring->ring_id; - ri.iri_ifp = ctx->ifc_ifp; - ri.iri_cidx = nic_i; + nic_i = fl->ifl_cidx; + nm_i = netmap_idx_n2k(kring, nic_i); + avail = ctx->isc_rxd_available(ctx->ifc_softc, + rxq->ifr_id, nic_i, USHRT_MAX); + for (n = 0; avail > 0 && nm_i != hwtail_lim; n++, avail--) { + rxd_info_zero(&ri); + ri.iri_frags = rxq->ifr_frags; + ri.iri_qsidx = kring->ring_id; + ri.iri_ifp = ctx->ifc_ifp; + ri.iri_cidx = nic_i; - error = ctx->isc_rxd_pkt_get(ctx->ifc_softc, &ri); - ring->slot[nm_i].len = error ? 0 : ri.iri_len - crclen; - ring->slot[nm_i].flags = 0; - bus_dmamap_sync(fl->ifl_buf_tag, - fl->ifl_sds.ifsd_map[nic_i], BUS_DMASYNC_POSTREAD); - nm_i = nm_next(nm_i, lim); - nic_i = nm_next(nic_i, lim); + error = ctx->isc_rxd_pkt_get(ctx->ifc_softc, &ri); + ring->slot[nm_i].len = error ? 0 : ri.iri_len - crclen; + ring->slot[nm_i].flags = 0; + bus_dmamap_sync(fl->ifl_buf_tag, + fl->ifl_sds.ifsd_map[nic_i], BUS_DMASYNC_POSTREAD); + nm_i = nm_next(nm_i, lim); + nic_i = nm_next(nic_i, lim); + } + if (n) { /* update the state variables */ + if (netmap_no_pendintr && !force_update) { + /* diagnostics */ + iflib_rx_miss ++; + iflib_rx_miss_bufs += n; } - if (n) { /* update the state variables */ - if (netmap_no_pendintr && !force_update) { - /* diagnostics */ - iflib_rx_miss ++; - iflib_rx_miss_bufs += n; - } - fl->ifl_cidx = nic_i; - kring->nr_hwtail = nm_i; - } - kring->nr_kflags &= ~NKR_PENDINTR; + fl->ifl_cidx = nic_i; + kring->nr_hwtail = nm_i; } + kring->nr_kflags &= ~NKR_PENDINTR; } /* * Second part: skip past packets that userspace has released. * (kring->nr_hwcur to head excluded), * and make the buffers available for reception. * As usual nm_i is the index in the netmap ring, * nic_i is the index in the NIC ring, and * nm_i == (nic_i + kring->nkr_hwofs) % ring_size */ - /* XXX not sure how this will work with multiple free lists */ nm_i = kring->nr_hwcur; return (netmap_fl_refill(rxq, kring, nm_i, false)); } static void iflib_netmap_intr(struct netmap_adapter *na, int onoff) { if_ctx_t ctx = na->ifp->if_softc; CTX_LOCK(ctx); if (onoff) { IFDI_INTR_ENABLE(ctx); } else { IFDI_INTR_DISABLE(ctx); } CTX_UNLOCK(ctx); } static int iflib_netmap_attach(if_ctx_t ctx) { struct netmap_adapter na; bzero(&na, sizeof(na)); na.ifp = ctx->ifc_ifp; na.na_flags = NAF_BDG_MAYSLEEP; MPASS(ctx->ifc_softc_ctx.isc_ntxqsets); MPASS(ctx->ifc_softc_ctx.isc_nrxqsets); na.num_tx_desc = iflib_num_tx_descs(ctx); na.num_rx_desc = iflib_num_rx_descs(ctx); na.nm_txsync = iflib_netmap_txsync; na.nm_rxsync = iflib_netmap_rxsync; na.nm_register = iflib_netmap_register; na.nm_intr = iflib_netmap_intr; na.num_tx_rings = ctx->ifc_softc_ctx.isc_ntxqsets; na.num_rx_rings = ctx->ifc_softc_ctx.isc_nrxqsets; return (netmap_attach(&na)); } static int iflib_netmap_txq_init(if_ctx_t ctx, iflib_txq_t txq) { struct netmap_adapter *na = NA(ctx->ifc_ifp); struct netmap_slot *slot; slot = netmap_reset(na, NR_TX, txq->ift_id, 0); if (slot == NULL) return (0); for (int i = 0; i < ctx->ifc_softc_ctx.isc_ntxd[0]; i++) { /* * In netmap mode, set the map for the packet buffer. * NOTE: Some drivers (not this one) also need to set * the physical buffer address in the NIC ring. * netmap_idx_n2k() maps a nic index, i, into the corresponding * netmap slot index, si */ int si = netmap_idx_n2k(na->tx_rings[txq->ift_id], i); netmap_load_map(na, txq->ift_buf_tag, txq->ift_sds.ifsd_map[i], NMB(na, slot + si)); } return (1); } static int iflib_netmap_rxq_init(if_ctx_t ctx, iflib_rxq_t rxq) { struct netmap_adapter *na = NA(ctx->ifc_ifp); struct netmap_kring *kring; struct netmap_slot *slot; uint32_t nm_i; slot = netmap_reset(na, NR_RX, rxq->ifr_id, 0); if (slot == NULL) return (0); kring = na->rx_rings[rxq->ifr_id]; nm_i = netmap_idx_n2k(kring, 0); netmap_fl_refill(rxq, kring, nm_i, true); return (1); } static void iflib_netmap_timer_adjust(if_ctx_t ctx, iflib_txq_t txq, uint32_t *reset_on) { struct netmap_kring *kring; uint16_t txqid; txqid = txq->ift_id; kring = netmap_kring_on(NA(ctx->ifc_ifp), txqid, NR_TX); if (kring == NULL) return; if (kring->nr_hwcur != nm_next(kring->nr_hwtail, kring->nkr_num_slots - 1)) { bus_dmamap_sync(txq->ift_ifdi->idi_tag, txq->ift_ifdi->idi_map, BUS_DMASYNC_POSTREAD); if (ctx->isc_txd_credits_update(ctx->ifc_softc, txqid, false)) netmap_tx_irq(ctx->ifc_ifp, txqid); if (!(ctx->ifc_flags & IFC_NETMAP_TX_IRQ)) { if (hz < 2000) *reset_on = 1; else *reset_on = hz / 1000; } } } #define iflib_netmap_detach(ifp) netmap_detach(ifp) #else #define iflib_netmap_txq_init(ctx, txq) (0) #define iflib_netmap_rxq_init(ctx, rxq) (0) #define iflib_netmap_detach(ifp) #define iflib_netmap_attach(ctx) (0) #define netmap_rx_irq(ifp, qid, budget) (0) #define netmap_tx_irq(ifp, qid) do {} while (0) #define iflib_netmap_timer_adjust(ctx, txq, reset_on) #endif #if defined(__i386__) || defined(__amd64__) static __inline void prefetch(void *x) { __asm volatile("prefetcht0 %0" :: "m" (*(unsigned long *)x)); } static __inline void prefetch2cachelines(void *x) { __asm volatile("prefetcht0 %0" :: "m" (*(unsigned long *)x)); #if (CACHE_LINE_SIZE < 128) __asm volatile("prefetcht0 %0" :: "m" (*(((unsigned long *)x)+CACHE_LINE_SIZE/(sizeof(unsigned long))))); #endif } #else #define prefetch(x) #define prefetch2cachelines(x) #endif static void iru_init(if_rxd_update_t iru, iflib_rxq_t rxq, uint8_t flid) { iflib_fl_t fl; fl = &rxq->ifr_fl[flid]; iru->iru_paddrs = fl->ifl_bus_addrs; iru->iru_idxs = fl->ifl_rxd_idxs; iru->iru_qsidx = rxq->ifr_id; iru->iru_buf_size = fl->ifl_buf_size; iru->iru_flidx = fl->ifl_id; } static void _iflib_dmamap_cb(void *arg, bus_dma_segment_t *segs, int nseg, int err) { if (err) return; *(bus_addr_t *) arg = segs[0].ds_addr; } int iflib_dma_alloc_align(if_ctx_t ctx, int size, int align, iflib_dma_info_t dma, int mapflags) { int err; device_t dev = ctx->ifc_dev; err = bus_dma_tag_create(bus_get_dma_tag(dev), /* parent */ align, 0, /* alignment, bounds */ BUS_SPACE_MAXADDR, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ NULL, NULL, /* filter, filterarg */ size, /* maxsize */ 1, /* nsegments */ size, /* maxsegsize */ BUS_DMA_ALLOCNOW, /* flags */ NULL, /* lockfunc */ NULL, /* lockarg */ &dma->idi_tag); if (err) { device_printf(dev, "%s: bus_dma_tag_create failed: %d\n", __func__, err); goto fail_0; } err = bus_dmamem_alloc(dma->idi_tag, (void**) &dma->idi_vaddr, BUS_DMA_NOWAIT | BUS_DMA_COHERENT | BUS_DMA_ZERO, &dma->idi_map); if (err) { device_printf(dev, "%s: bus_dmamem_alloc(%ju) failed: %d\n", __func__, (uintmax_t)size, err); goto fail_1; } dma->idi_paddr = IF_BAD_DMA; err = bus_dmamap_load(dma->idi_tag, dma->idi_map, dma->idi_vaddr, size, _iflib_dmamap_cb, &dma->idi_paddr, mapflags | BUS_DMA_NOWAIT); if (err || dma->idi_paddr == IF_BAD_DMA) { device_printf(dev, "%s: bus_dmamap_load failed: %d\n", __func__, err); goto fail_2; } dma->idi_size = size; return (0); fail_2: bus_dmamem_free(dma->idi_tag, dma->idi_vaddr, dma->idi_map); fail_1: bus_dma_tag_destroy(dma->idi_tag); fail_0: dma->idi_tag = NULL; return (err); } int iflib_dma_alloc(if_ctx_t ctx, int size, iflib_dma_info_t dma, int mapflags) { if_shared_ctx_t sctx = ctx->ifc_sctx; KASSERT(sctx->isc_q_align != 0, ("alignment value not initialized")); return (iflib_dma_alloc_align(ctx, size, sctx->isc_q_align, dma, mapflags)); } int iflib_dma_alloc_multi(if_ctx_t ctx, int *sizes, iflib_dma_info_t *dmalist, int mapflags, int count) { int i, err; iflib_dma_info_t *dmaiter; dmaiter = dmalist; for (i = 0; i < count; i++, dmaiter++) { if ((err = iflib_dma_alloc(ctx, sizes[i], *dmaiter, mapflags)) != 0) break; } if (err) iflib_dma_free_multi(dmalist, i); return (err); } void iflib_dma_free(iflib_dma_info_t dma) { if (dma->idi_tag == NULL) return; if (dma->idi_paddr != IF_BAD_DMA) { bus_dmamap_sync(dma->idi_tag, dma->idi_map, BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); bus_dmamap_unload(dma->idi_tag, dma->idi_map); dma->idi_paddr = IF_BAD_DMA; } if (dma->idi_vaddr != NULL) { bus_dmamem_free(dma->idi_tag, dma->idi_vaddr, dma->idi_map); dma->idi_vaddr = NULL; } bus_dma_tag_destroy(dma->idi_tag); dma->idi_tag = NULL; } void iflib_dma_free_multi(iflib_dma_info_t *dmalist, int count) { int i; iflib_dma_info_t *dmaiter = dmalist; for (i = 0; i < count; i++, dmaiter++) iflib_dma_free(*dmaiter); } static int iflib_fast_intr(void *arg) { iflib_filter_info_t info = arg; struct grouptask *gtask = info->ifi_task; int result; DBG_COUNTER_INC(fast_intrs); if (info->ifi_filter != NULL) { result = info->ifi_filter(info->ifi_filter_arg); if ((result & FILTER_SCHEDULE_THREAD) == 0) return (result); } GROUPTASK_ENQUEUE(gtask); return (FILTER_HANDLED); } static int iflib_fast_intr_rxtx(void *arg) { iflib_filter_info_t info = arg; struct grouptask *gtask = info->ifi_task; if_ctx_t ctx; iflib_rxq_t rxq = (iflib_rxq_t)info->ifi_ctx; iflib_txq_t txq; void *sc; int i, cidx, result; qidx_t txqid; bool intr_enable, intr_legacy; DBG_COUNTER_INC(fast_intrs); if (info->ifi_filter != NULL) { result = info->ifi_filter(info->ifi_filter_arg); if ((result & FILTER_SCHEDULE_THREAD) == 0) return (result); } ctx = rxq->ifr_ctx; sc = ctx->ifc_softc; intr_enable = false; intr_legacy = !!(ctx->ifc_flags & IFC_LEGACY); MPASS(rxq->ifr_ntxqirq); for (i = 0; i < rxq->ifr_ntxqirq; i++) { txqid = rxq->ifr_txqid[i]; txq = &ctx->ifc_txqs[txqid]; bus_dmamap_sync(txq->ift_ifdi->idi_tag, txq->ift_ifdi->idi_map, BUS_DMASYNC_POSTREAD); if (!ctx->isc_txd_credits_update(sc, txqid, false)) { if (intr_legacy) intr_enable = true; else IFDI_TX_QUEUE_INTR_ENABLE(ctx, txqid); continue; } GROUPTASK_ENQUEUE(&txq->ift_task); } if (ctx->ifc_sctx->isc_flags & IFLIB_HAS_RXCQ) cidx = rxq->ifr_cq_cidx; else cidx = rxq->ifr_fl[0].ifl_cidx; if (iflib_rxd_avail(ctx, rxq, cidx, 1)) GROUPTASK_ENQUEUE(gtask); else { if (intr_legacy) intr_enable = true; else IFDI_RX_QUEUE_INTR_ENABLE(ctx, rxq->ifr_id); DBG_COUNTER_INC(rx_intr_enables); } if (intr_enable) IFDI_INTR_ENABLE(ctx); return (FILTER_HANDLED); } static int iflib_fast_intr_ctx(void *arg) { iflib_filter_info_t info = arg; struct grouptask *gtask = info->ifi_task; int result; DBG_COUNTER_INC(fast_intrs); if (info->ifi_filter != NULL) { result = info->ifi_filter(info->ifi_filter_arg); if ((result & FILTER_SCHEDULE_THREAD) == 0) return (result); } GROUPTASK_ENQUEUE(gtask); return (FILTER_HANDLED); } static int _iflib_irq_alloc(if_ctx_t ctx, if_irq_t irq, int rid, driver_filter_t filter, driver_intr_t handler, void *arg, const char *name) { struct resource *res; void *tag = NULL; device_t dev = ctx->ifc_dev; int flags, i, rc; flags = RF_ACTIVE; if (ctx->ifc_flags & IFC_LEGACY) flags |= RF_SHAREABLE; MPASS(rid < 512); i = rid; res = bus_alloc_resource_any(dev, SYS_RES_IRQ, &i, flags); if (res == NULL) { device_printf(dev, "failed to allocate IRQ for rid %d, name %s.\n", rid, name); return (ENOMEM); } irq->ii_res = res; KASSERT(filter == NULL || handler == NULL, ("filter and handler can't both be non-NULL")); rc = bus_setup_intr(dev, res, INTR_MPSAFE | INTR_TYPE_NET, filter, handler, arg, &tag); if (rc != 0) { device_printf(dev, "failed to setup interrupt for rid %d, name %s: %d\n", rid, name ? name : "unknown", rc); return (rc); } else if (name) bus_describe_intr(dev, res, tag, "%s", name); irq->ii_tag = tag; return (0); } /********************************************************************* * * Allocate DMA resources for TX buffers as well as memory for the TX * mbuf map. TX DMA maps (non-TSO/TSO) and TX mbuf map are kept in a * iflib_sw_tx_desc_array structure, storing all the information that * is needed to transmit a packet on the wire. This is called only * once at attach, setup is done every reset. * **********************************************************************/ static int iflib_txsd_alloc(iflib_txq_t txq) { if_ctx_t ctx = txq->ift_ctx; if_shared_ctx_t sctx = ctx->ifc_sctx; if_softc_ctx_t scctx = &ctx->ifc_softc_ctx; device_t dev = ctx->ifc_dev; bus_size_t tsomaxsize; int err, nsegments, ntsosegments; bool tso; nsegments = scctx->isc_tx_nsegments; ntsosegments = scctx->isc_tx_tso_segments_max; tsomaxsize = scctx->isc_tx_tso_size_max; if (if_getcapabilities(ctx->ifc_ifp) & IFCAP_VLAN_MTU) tsomaxsize += sizeof(struct ether_vlan_header); MPASS(scctx->isc_ntxd[0] > 0); MPASS(scctx->isc_ntxd[txq->ift_br_offset] > 0); MPASS(nsegments > 0); if (if_getcapabilities(ctx->ifc_ifp) & IFCAP_TSO) { MPASS(ntsosegments > 0); MPASS(sctx->isc_tso_maxsize >= tsomaxsize); } /* * Set up DMA tags for TX buffers. */ if ((err = bus_dma_tag_create(bus_get_dma_tag(dev), 1, 0, /* alignment, bounds */ BUS_SPACE_MAXADDR, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ NULL, NULL, /* filter, filterarg */ sctx->isc_tx_maxsize, /* maxsize */ nsegments, /* nsegments */ sctx->isc_tx_maxsegsize, /* maxsegsize */ 0, /* flags */ NULL, /* lockfunc */ NULL, /* lockfuncarg */ &txq->ift_buf_tag))) { device_printf(dev,"Unable to allocate TX DMA tag: %d\n", err); device_printf(dev,"maxsize: %ju nsegments: %d maxsegsize: %ju\n", (uintmax_t)sctx->isc_tx_maxsize, nsegments, (uintmax_t)sctx->isc_tx_maxsegsize); goto fail; } tso = (if_getcapabilities(ctx->ifc_ifp) & IFCAP_TSO) != 0; if (tso && (err = bus_dma_tag_create(bus_get_dma_tag(dev), 1, 0, /* alignment, bounds */ BUS_SPACE_MAXADDR, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ NULL, NULL, /* filter, filterarg */ tsomaxsize, /* maxsize */ ntsosegments, /* nsegments */ sctx->isc_tso_maxsegsize,/* maxsegsize */ 0, /* flags */ NULL, /* lockfunc */ NULL, /* lockfuncarg */ &txq->ift_tso_buf_tag))) { device_printf(dev, "Unable to allocate TSO TX DMA tag: %d\n", err); goto fail; } /* Allocate memory for the TX mbuf map. */ if (!(txq->ift_sds.ifsd_m = (struct mbuf **) malloc(sizeof(struct mbuf *) * scctx->isc_ntxd[txq->ift_br_offset], M_IFLIB, M_NOWAIT | M_ZERO))) { device_printf(dev, "Unable to allocate TX mbuf map memory\n"); err = ENOMEM; goto fail; } /* * Create the DMA maps for TX buffers. */ if ((txq->ift_sds.ifsd_map = (bus_dmamap_t *)malloc( sizeof(bus_dmamap_t) * scctx->isc_ntxd[txq->ift_br_offset], M_IFLIB, M_NOWAIT | M_ZERO)) == NULL) { device_printf(dev, "Unable to allocate TX buffer DMA map memory\n"); err = ENOMEM; goto fail; } if (tso && (txq->ift_sds.ifsd_tso_map = (bus_dmamap_t *)malloc( sizeof(bus_dmamap_t) * scctx->isc_ntxd[txq->ift_br_offset], M_IFLIB, M_NOWAIT | M_ZERO)) == NULL) { device_printf(dev, "Unable to allocate TSO TX buffer map memory\n"); err = ENOMEM; goto fail; } for (int i = 0; i < scctx->isc_ntxd[txq->ift_br_offset]; i++) { err = bus_dmamap_create(txq->ift_buf_tag, 0, &txq->ift_sds.ifsd_map[i]); if (err != 0) { device_printf(dev, "Unable to create TX DMA map\n"); goto fail; } if (!tso) continue; err = bus_dmamap_create(txq->ift_tso_buf_tag, 0, &txq->ift_sds.ifsd_tso_map[i]); if (err != 0) { device_printf(dev, "Unable to create TSO TX DMA map\n"); goto fail; } } return (0); fail: /* We free all, it handles case where we are in the middle */ iflib_tx_structures_free(ctx); return (err); } static void iflib_txsd_destroy(if_ctx_t ctx, iflib_txq_t txq, int i) { bus_dmamap_t map; if (txq->ift_sds.ifsd_map != NULL) { map = txq->ift_sds.ifsd_map[i]; bus_dmamap_sync(txq->ift_buf_tag, map, BUS_DMASYNC_POSTWRITE); bus_dmamap_unload(txq->ift_buf_tag, map); bus_dmamap_destroy(txq->ift_buf_tag, map); txq->ift_sds.ifsd_map[i] = NULL; } if (txq->ift_sds.ifsd_tso_map != NULL) { map = txq->ift_sds.ifsd_tso_map[i]; bus_dmamap_sync(txq->ift_tso_buf_tag, map, BUS_DMASYNC_POSTWRITE); bus_dmamap_unload(txq->ift_tso_buf_tag, map); bus_dmamap_destroy(txq->ift_tso_buf_tag, map); txq->ift_sds.ifsd_tso_map[i] = NULL; } } static void iflib_txq_destroy(iflib_txq_t txq) { if_ctx_t ctx = txq->ift_ctx; for (int i = 0; i < txq->ift_size; i++) iflib_txsd_destroy(ctx, txq, i); if (txq->ift_br != NULL) { ifmp_ring_free(txq->ift_br); txq->ift_br = NULL; } mtx_destroy(&txq->ift_mtx); if (txq->ift_sds.ifsd_map != NULL) { free(txq->ift_sds.ifsd_map, M_IFLIB); txq->ift_sds.ifsd_map = NULL; } if (txq->ift_sds.ifsd_tso_map != NULL) { free(txq->ift_sds.ifsd_tso_map, M_IFLIB); txq->ift_sds.ifsd_tso_map = NULL; } if (txq->ift_sds.ifsd_m != NULL) { free(txq->ift_sds.ifsd_m, M_IFLIB); txq->ift_sds.ifsd_m = NULL; } if (txq->ift_buf_tag != NULL) { bus_dma_tag_destroy(txq->ift_buf_tag); txq->ift_buf_tag = NULL; } if (txq->ift_tso_buf_tag != NULL) { bus_dma_tag_destroy(txq->ift_tso_buf_tag); txq->ift_tso_buf_tag = NULL; } if (txq->ift_ifdi != NULL) { free(txq->ift_ifdi, M_IFLIB); } } static void iflib_txsd_free(if_ctx_t ctx, iflib_txq_t txq, int i) { struct mbuf **mp; mp = &txq->ift_sds.ifsd_m[i]; if (*mp == NULL) return; if (txq->ift_sds.ifsd_map != NULL) { bus_dmamap_sync(txq->ift_buf_tag, txq->ift_sds.ifsd_map[i], BUS_DMASYNC_POSTWRITE); bus_dmamap_unload(txq->ift_buf_tag, txq->ift_sds.ifsd_map[i]); } if (txq->ift_sds.ifsd_tso_map != NULL) { bus_dmamap_sync(txq->ift_tso_buf_tag, txq->ift_sds.ifsd_tso_map[i], BUS_DMASYNC_POSTWRITE); bus_dmamap_unload(txq->ift_tso_buf_tag, txq->ift_sds.ifsd_tso_map[i]); } m_free(*mp); DBG_COUNTER_INC(tx_frees); *mp = NULL; } static int iflib_txq_setup(iflib_txq_t txq) { if_ctx_t ctx = txq->ift_ctx; if_softc_ctx_t scctx = &ctx->ifc_softc_ctx; if_shared_ctx_t sctx = ctx->ifc_sctx; iflib_dma_info_t di; int i; /* Set number of descriptors available */ txq->ift_qstatus = IFLIB_QUEUE_IDLE; /* XXX make configurable */ txq->ift_update_freq = IFLIB_DEFAULT_TX_UPDATE_FREQ; /* Reset indices */ txq->ift_cidx_processed = 0; txq->ift_pidx = txq->ift_cidx = txq->ift_npending = 0; txq->ift_size = scctx->isc_ntxd[txq->ift_br_offset]; for (i = 0, di = txq->ift_ifdi; i < sctx->isc_ntxqs; i++, di++) bzero((void *)di->idi_vaddr, di->idi_size); IFDI_TXQ_SETUP(ctx, txq->ift_id); for (i = 0, di = txq->ift_ifdi; i < sctx->isc_ntxqs; i++, di++) bus_dmamap_sync(di->idi_tag, di->idi_map, BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); return (0); } /********************************************************************* * * Allocate DMA resources for RX buffers as well as memory for the RX * mbuf map, direct RX cluster pointer map and RX cluster bus address * map. RX DMA map, RX mbuf map, direct RX cluster pointer map and * RX cluster map are kept in a iflib_sw_rx_desc_array structure. * Since we use use one entry in iflib_sw_rx_desc_array per received * packet, the maximum number of entries we'll need is equal to the * number of hardware receive descriptors that we've allocated. * **********************************************************************/ static int iflib_rxsd_alloc(iflib_rxq_t rxq) { if_ctx_t ctx = rxq->ifr_ctx; if_shared_ctx_t sctx = ctx->ifc_sctx; if_softc_ctx_t scctx = &ctx->ifc_softc_ctx; device_t dev = ctx->ifc_dev; iflib_fl_t fl; int err; MPASS(scctx->isc_nrxd[0] > 0); MPASS(scctx->isc_nrxd[rxq->ifr_fl_offset] > 0); fl = rxq->ifr_fl; for (int i = 0; i < rxq->ifr_nfl; i++, fl++) { fl->ifl_size = scctx->isc_nrxd[rxq->ifr_fl_offset]; /* this isn't necessarily the same */ /* Set up DMA tag for RX buffers. */ err = bus_dma_tag_create(bus_get_dma_tag(dev), /* parent */ 1, 0, /* alignment, bounds */ BUS_SPACE_MAXADDR, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ NULL, NULL, /* filter, filterarg */ sctx->isc_rx_maxsize, /* maxsize */ sctx->isc_rx_nsegments, /* nsegments */ sctx->isc_rx_maxsegsize, /* maxsegsize */ 0, /* flags */ NULL, /* lockfunc */ NULL, /* lockarg */ &fl->ifl_buf_tag); if (err) { device_printf(dev, "Unable to allocate RX DMA tag: %d\n", err); goto fail; } /* Allocate memory for the RX mbuf map. */ if (!(fl->ifl_sds.ifsd_m = (struct mbuf **) malloc(sizeof(struct mbuf *) * scctx->isc_nrxd[rxq->ifr_fl_offset], M_IFLIB, M_NOWAIT | M_ZERO))) { device_printf(dev, "Unable to allocate RX mbuf map memory\n"); err = ENOMEM; goto fail; } /* Allocate memory for the direct RX cluster pointer map. */ if (!(fl->ifl_sds.ifsd_cl = (caddr_t *) malloc(sizeof(caddr_t) * scctx->isc_nrxd[rxq->ifr_fl_offset], M_IFLIB, M_NOWAIT | M_ZERO))) { device_printf(dev, "Unable to allocate RX cluster map memory\n"); err = ENOMEM; goto fail; } /* Allocate memory for the RX cluster bus address map. */ if (!(fl->ifl_sds.ifsd_ba = (bus_addr_t *) malloc(sizeof(bus_addr_t) * scctx->isc_nrxd[rxq->ifr_fl_offset], M_IFLIB, M_NOWAIT | M_ZERO))) { device_printf(dev, "Unable to allocate RX bus address map memory\n"); err = ENOMEM; goto fail; } /* * Create the DMA maps for RX buffers. */ if (!(fl->ifl_sds.ifsd_map = (bus_dmamap_t *) malloc(sizeof(bus_dmamap_t) * scctx->isc_nrxd[rxq->ifr_fl_offset], M_IFLIB, M_NOWAIT | M_ZERO))) { device_printf(dev, "Unable to allocate RX buffer DMA map memory\n"); err = ENOMEM; goto fail; } for (int i = 0; i < scctx->isc_nrxd[rxq->ifr_fl_offset]; i++) { err = bus_dmamap_create(fl->ifl_buf_tag, 0, &fl->ifl_sds.ifsd_map[i]); if (err != 0) { device_printf(dev, "Unable to create RX buffer DMA map\n"); goto fail; } } } return (0); fail: iflib_rx_structures_free(ctx); return (err); } /* * Internal service routines */ struct rxq_refill_cb_arg { int error; bus_dma_segment_t seg; int nseg; }; static void _rxq_refill_cb(void *arg, bus_dma_segment_t *segs, int nseg, int error) { struct rxq_refill_cb_arg *cb_arg = arg; cb_arg->error = error; cb_arg->seg = segs[0]; cb_arg->nseg = nseg; } /** * iflib_fl_refill - refill an rxq free-buffer list * @ctx: the iflib context * @fl: the free list to refill * @count: the number of new buffers to allocate * * (Re)populate an rxq free-buffer list with up to @count new packet buffers. * The caller must assure that @count does not exceed the queue's capacity. */ static uint8_t iflib_fl_refill(if_ctx_t ctx, iflib_fl_t fl, int count) { struct if_rxd_update iru; struct rxq_refill_cb_arg cb_arg; struct mbuf *m; caddr_t cl, *sd_cl; struct mbuf **sd_m; bus_dmamap_t *sd_map; bus_addr_t bus_addr, *sd_ba; int err, frag_idx, i, idx, n, pidx; qidx_t credits; sd_m = fl->ifl_sds.ifsd_m; sd_map = fl->ifl_sds.ifsd_map; sd_cl = fl->ifl_sds.ifsd_cl; sd_ba = fl->ifl_sds.ifsd_ba; pidx = fl->ifl_pidx; idx = pidx; frag_idx = fl->ifl_fragidx; credits = fl->ifl_credits; i = 0; n = count; MPASS(n > 0); MPASS(credits + n <= fl->ifl_size); if (pidx < fl->ifl_cidx) MPASS(pidx + n <= fl->ifl_cidx); if (pidx == fl->ifl_cidx && (credits < fl->ifl_size)) MPASS(fl->ifl_gen == 0); if (pidx > fl->ifl_cidx) MPASS(n <= fl->ifl_size - pidx + fl->ifl_cidx); DBG_COUNTER_INC(fl_refills); if (n > 8) DBG_COUNTER_INC(fl_refills_large); iru_init(&iru, fl->ifl_rxq, fl->ifl_id); while (n-- > 0) { /* * We allocate an uninitialized mbuf + cluster, mbuf is * initialized after rx. * * If the cluster is still set then we know a minimum sized * packet was received */ bit_ffc_at(fl->ifl_rx_bitmap, frag_idx, fl->ifl_size, &frag_idx); if (frag_idx < 0) bit_ffc(fl->ifl_rx_bitmap, fl->ifl_size, &frag_idx); MPASS(frag_idx >= 0); if ((cl = sd_cl[frag_idx]) == NULL) { cl = uma_zalloc(fl->ifl_zone, M_NOWAIT); if (__predict_false(cl == NULL)) break; cb_arg.error = 0; MPASS(sd_map != NULL); err = bus_dmamap_load(fl->ifl_buf_tag, sd_map[frag_idx], cl, fl->ifl_buf_size, _rxq_refill_cb, &cb_arg, BUS_DMA_NOWAIT); if (__predict_false(err != 0 || cb_arg.error)) { uma_zfree(fl->ifl_zone, cl); break; } sd_ba[frag_idx] = bus_addr = cb_arg.seg.ds_addr; sd_cl[frag_idx] = cl; #if MEMORY_LOGGING fl->ifl_cl_enqueued++; #endif } else { bus_addr = sd_ba[frag_idx]; } bus_dmamap_sync(fl->ifl_buf_tag, sd_map[frag_idx], BUS_DMASYNC_PREREAD); if (sd_m[frag_idx] == NULL) { m = m_gethdr(M_NOWAIT, MT_NOINIT); if (__predict_false(m == NULL)) break; sd_m[frag_idx] = m; } bit_set(fl->ifl_rx_bitmap, frag_idx); #if MEMORY_LOGGING fl->ifl_m_enqueued++; #endif DBG_COUNTER_INC(rx_allocs); fl->ifl_rxd_idxs[i] = frag_idx; fl->ifl_bus_addrs[i] = bus_addr; credits++; i++; MPASS(credits <= fl->ifl_size); if (++idx == fl->ifl_size) { #ifdef INVARIANTS fl->ifl_gen = 1; #endif idx = 0; } if (n == 0 || i == IFLIB_MAX_RX_REFRESH) { iru.iru_pidx = pidx; iru.iru_count = i; ctx->isc_rxd_refill(ctx->ifc_softc, &iru); fl->ifl_pidx = idx; fl->ifl_credits = credits; pidx = idx; i = 0; } } if (n < count - 1) { if (i != 0) { iru.iru_pidx = pidx; iru.iru_count = i; ctx->isc_rxd_refill(ctx->ifc_softc, &iru); fl->ifl_pidx = idx; fl->ifl_credits = credits; } DBG_COUNTER_INC(rxd_flush); if (fl->ifl_pidx == 0) pidx = fl->ifl_size - 1; else pidx = fl->ifl_pidx - 1; bus_dmamap_sync(fl->ifl_ifdi->idi_tag, fl->ifl_ifdi->idi_map, BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); ctx->isc_rxd_flush(ctx->ifc_softc, fl->ifl_rxq->ifr_id, fl->ifl_id, pidx); if (__predict_true(bit_test(fl->ifl_rx_bitmap, frag_idx))) { fl->ifl_fragidx = frag_idx + 1; if (fl->ifl_fragidx == fl->ifl_size) fl->ifl_fragidx = 0; } else { fl->ifl_fragidx = frag_idx; } } return (n == -1 ? 0 : IFLIB_RXEOF_EMPTY); } static inline uint8_t iflib_fl_refill_all(if_ctx_t ctx, iflib_fl_t fl) { /* we avoid allowing pidx to catch up with cidx as it confuses ixl */ int32_t reclaimable = fl->ifl_size - fl->ifl_credits - 1; #ifdef INVARIANTS int32_t delta = fl->ifl_size - get_inuse(fl->ifl_size, fl->ifl_cidx, fl->ifl_pidx, fl->ifl_gen) - 1; #endif MPASS(fl->ifl_credits <= fl->ifl_size); MPASS(reclaimable == delta); if (reclaimable > 0) return (iflib_fl_refill(ctx, fl, reclaimable)); return (0); } uint8_t iflib_in_detach(if_ctx_t ctx) { bool in_detach; STATE_LOCK(ctx); in_detach = !!(ctx->ifc_flags & IFC_IN_DETACH); STATE_UNLOCK(ctx); return (in_detach); } static void iflib_fl_bufs_free(iflib_fl_t fl) { iflib_dma_info_t idi = fl->ifl_ifdi; bus_dmamap_t sd_map; uint32_t i; for (i = 0; i < fl->ifl_size; i++) { struct mbuf **sd_m = &fl->ifl_sds.ifsd_m[i]; caddr_t *sd_cl = &fl->ifl_sds.ifsd_cl[i]; if (*sd_cl != NULL) { sd_map = fl->ifl_sds.ifsd_map[i]; bus_dmamap_sync(fl->ifl_buf_tag, sd_map, BUS_DMASYNC_POSTREAD); bus_dmamap_unload(fl->ifl_buf_tag, sd_map); uma_zfree(fl->ifl_zone, *sd_cl); *sd_cl = NULL; if (*sd_m != NULL) { m_init(*sd_m, M_NOWAIT, MT_DATA, 0); uma_zfree(zone_mbuf, *sd_m); *sd_m = NULL; } } else { MPASS(*sd_m == NULL); } #if MEMORY_LOGGING fl->ifl_m_dequeued++; fl->ifl_cl_dequeued++; #endif } #ifdef INVARIANTS for (i = 0; i < fl->ifl_size; i++) { MPASS(fl->ifl_sds.ifsd_cl[i] == NULL); MPASS(fl->ifl_sds.ifsd_m[i] == NULL); } #endif /* * Reset free list values */ fl->ifl_credits = fl->ifl_cidx = fl->ifl_pidx = fl->ifl_gen = fl->ifl_fragidx = 0; bzero(idi->idi_vaddr, idi->idi_size); } /********************************************************************* * * Initialize a free list and its buffers. * **********************************************************************/ static int iflib_fl_setup(iflib_fl_t fl) { iflib_rxq_t rxq = fl->ifl_rxq; if_ctx_t ctx = rxq->ifr_ctx; if_softc_ctx_t scctx = &ctx->ifc_softc_ctx; int qidx; bit_nclear(fl->ifl_rx_bitmap, 0, fl->ifl_size - 1); /* ** Free current RX buffer structs and their mbufs */ iflib_fl_bufs_free(fl); /* Now replenish the mbufs */ MPASS(fl->ifl_credits == 0); qidx = rxq->ifr_fl_offset + fl->ifl_id; if (scctx->isc_rxd_buf_size[qidx] != 0) fl->ifl_buf_size = scctx->isc_rxd_buf_size[qidx]; else fl->ifl_buf_size = ctx->ifc_rx_mbuf_sz; /* * ifl_buf_size may be a driver-supplied value, so pull it up * to the selected mbuf size. */ fl->ifl_buf_size = iflib_get_mbuf_size_for(fl->ifl_buf_size); if (fl->ifl_buf_size > ctx->ifc_max_fl_buf_size) ctx->ifc_max_fl_buf_size = fl->ifl_buf_size; fl->ifl_cltype = m_gettype(fl->ifl_buf_size); fl->ifl_zone = m_getzone(fl->ifl_buf_size); /* avoid pre-allocating zillions of clusters to an idle card * potentially speeding up attach */ (void)iflib_fl_refill(ctx, fl, min(128, fl->ifl_size)); MPASS(min(128, fl->ifl_size) == fl->ifl_credits); if (min(128, fl->ifl_size) != fl->ifl_credits) return (ENOBUFS); /* * handle failure */ MPASS(rxq != NULL); MPASS(fl->ifl_ifdi != NULL); bus_dmamap_sync(fl->ifl_ifdi->idi_tag, fl->ifl_ifdi->idi_map, BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); return (0); } /********************************************************************* * * Free receive ring data structures * **********************************************************************/ static void iflib_rx_sds_free(iflib_rxq_t rxq) { iflib_fl_t fl; int i, j; if (rxq->ifr_fl != NULL) { for (i = 0; i < rxq->ifr_nfl; i++) { fl = &rxq->ifr_fl[i]; if (fl->ifl_buf_tag != NULL) { if (fl->ifl_sds.ifsd_map != NULL) { for (j = 0; j < fl->ifl_size; j++) { bus_dmamap_sync( fl->ifl_buf_tag, fl->ifl_sds.ifsd_map[j], BUS_DMASYNC_POSTREAD); bus_dmamap_unload( fl->ifl_buf_tag, fl->ifl_sds.ifsd_map[j]); bus_dmamap_destroy( fl->ifl_buf_tag, fl->ifl_sds.ifsd_map[j]); } } bus_dma_tag_destroy(fl->ifl_buf_tag); fl->ifl_buf_tag = NULL; } free(fl->ifl_sds.ifsd_m, M_IFLIB); free(fl->ifl_sds.ifsd_cl, M_IFLIB); free(fl->ifl_sds.ifsd_ba, M_IFLIB); free(fl->ifl_sds.ifsd_map, M_IFLIB); fl->ifl_sds.ifsd_m = NULL; fl->ifl_sds.ifsd_cl = NULL; fl->ifl_sds.ifsd_ba = NULL; fl->ifl_sds.ifsd_map = NULL; } free(rxq->ifr_fl, M_IFLIB); rxq->ifr_fl = NULL; free(rxq->ifr_ifdi, M_IFLIB); rxq->ifr_ifdi = NULL; rxq->ifr_cq_cidx = 0; } } /* * Timer routine */ static void iflib_timer(void *arg) { iflib_txq_t txq = arg; if_ctx_t ctx = txq->ift_ctx; if_softc_ctx_t sctx = &ctx->ifc_softc_ctx; uint64_t this_tick = ticks; uint32_t reset_on = hz / 2; if (!(if_getdrvflags(ctx->ifc_ifp) & IFF_DRV_RUNNING)) return; /* ** Check on the state of the TX queue(s), this ** can be done without the lock because its RO ** and the HUNG state will be static if set. */ if (this_tick - txq->ift_last_timer_tick >= hz / 2) { txq->ift_last_timer_tick = this_tick; IFDI_TIMER(ctx, txq->ift_id); if ((txq->ift_qstatus == IFLIB_QUEUE_HUNG) && ((txq->ift_cleaned_prev == txq->ift_cleaned) || (sctx->isc_pause_frames == 0))) goto hung; if (txq->ift_qstatus != IFLIB_QUEUE_IDLE && ifmp_ring_is_stalled(txq->ift_br)) { KASSERT(ctx->ifc_link_state == LINK_STATE_UP, ("queue can't be marked as hung if interface is down")); txq->ift_qstatus = IFLIB_QUEUE_HUNG; } txq->ift_cleaned_prev = txq->ift_cleaned; } #ifdef DEV_NETMAP if (if_getcapenable(ctx->ifc_ifp) & IFCAP_NETMAP) iflib_netmap_timer_adjust(ctx, txq, &reset_on); #endif /* handle any laggards */ if (txq->ift_db_pending) GROUPTASK_ENQUEUE(&txq->ift_task); sctx->isc_pause_frames = 0; if (if_getdrvflags(ctx->ifc_ifp) & IFF_DRV_RUNNING) callout_reset_on(&txq->ift_timer, reset_on, iflib_timer, txq, txq->ift_timer.c_cpu); return; hung: device_printf(ctx->ifc_dev, "Watchdog timeout (TX: %d desc avail: %d pidx: %d) -- resetting\n", txq->ift_id, TXQ_AVAIL(txq), txq->ift_pidx); STATE_LOCK(ctx); if_setdrvflagbits(ctx->ifc_ifp, IFF_DRV_OACTIVE, IFF_DRV_RUNNING); ctx->ifc_flags |= (IFC_DO_WATCHDOG|IFC_DO_RESET); iflib_admin_intr_deferred(ctx); STATE_UNLOCK(ctx); } static uint16_t iflib_get_mbuf_size_for(unsigned int size) { if (size <= MCLBYTES) return (MCLBYTES); else return (MJUMPAGESIZE); } static void iflib_calc_rx_mbuf_sz(if_ctx_t ctx) { if_softc_ctx_t sctx = &ctx->ifc_softc_ctx; /* * XXX don't set the max_frame_size to larger * than the hardware can handle */ ctx->ifc_rx_mbuf_sz = iflib_get_mbuf_size_for(sctx->isc_max_frame_size); } uint32_t iflib_get_rx_mbuf_sz(if_ctx_t ctx) { return (ctx->ifc_rx_mbuf_sz); } static void iflib_init_locked(if_ctx_t ctx) { if_softc_ctx_t sctx = &ctx->ifc_softc_ctx; if_softc_ctx_t scctx = &ctx->ifc_softc_ctx; if_t ifp = ctx->ifc_ifp; iflib_fl_t fl; iflib_txq_t txq; iflib_rxq_t rxq; int i, j, tx_ip_csum_flags, tx_ip6_csum_flags; if_setdrvflagbits(ifp, IFF_DRV_OACTIVE, IFF_DRV_RUNNING); IFDI_INTR_DISABLE(ctx); tx_ip_csum_flags = scctx->isc_tx_csum_flags & (CSUM_IP | CSUM_TCP | CSUM_UDP | CSUM_SCTP); tx_ip6_csum_flags = scctx->isc_tx_csum_flags & (CSUM_IP6_TCP | CSUM_IP6_UDP | CSUM_IP6_SCTP); /* Set hardware offload abilities */ if_clearhwassist(ifp); if (if_getcapenable(ifp) & IFCAP_TXCSUM) if_sethwassistbits(ifp, tx_ip_csum_flags, 0); if (if_getcapenable(ifp) & IFCAP_TXCSUM_IPV6) if_sethwassistbits(ifp, tx_ip6_csum_flags, 0); if (if_getcapenable(ifp) & IFCAP_TSO4) if_sethwassistbits(ifp, CSUM_IP_TSO, 0); if (if_getcapenable(ifp) & IFCAP_TSO6) if_sethwassistbits(ifp, CSUM_IP6_TSO, 0); for (i = 0, txq = ctx->ifc_txqs; i < sctx->isc_ntxqsets; i++, txq++) { CALLOUT_LOCK(txq); callout_stop(&txq->ift_timer); CALLOUT_UNLOCK(txq); iflib_netmap_txq_init(ctx, txq); } /* * Calculate a suitable Rx mbuf size prior to calling IFDI_INIT, so * that drivers can use the value when setting up the hardware receive * buffers. */ iflib_calc_rx_mbuf_sz(ctx); #ifdef INVARIANTS i = if_getdrvflags(ifp); #endif IFDI_INIT(ctx); MPASS(if_getdrvflags(ifp) == i); for (i = 0, rxq = ctx->ifc_rxqs; i < sctx->isc_nrxqsets; i++, rxq++) { if (iflib_netmap_rxq_init(ctx, rxq) > 0) { /* This rxq is in netmap mode. Skip normal init. */ continue; } for (j = 0, fl = rxq->ifr_fl; j < rxq->ifr_nfl; j++, fl++) { if (iflib_fl_setup(fl)) { device_printf(ctx->ifc_dev, "setting up free list %d failed - " "check cluster settings\n", j); goto done; } } } done: if_setdrvflagbits(ctx->ifc_ifp, IFF_DRV_RUNNING, IFF_DRV_OACTIVE); IFDI_INTR_ENABLE(ctx); txq = ctx->ifc_txqs; for (i = 0; i < sctx->isc_ntxqsets; i++, txq++) callout_reset_on(&txq->ift_timer, hz/2, iflib_timer, txq, txq->ift_timer.c_cpu); } static int iflib_media_change(if_t ifp) { if_ctx_t ctx = if_getsoftc(ifp); int err; CTX_LOCK(ctx); if ((err = IFDI_MEDIA_CHANGE(ctx)) == 0) iflib_init_locked(ctx); CTX_UNLOCK(ctx); return (err); } static void iflib_media_status(if_t ifp, struct ifmediareq *ifmr) { if_ctx_t ctx = if_getsoftc(ifp); CTX_LOCK(ctx); IFDI_UPDATE_ADMIN_STATUS(ctx); IFDI_MEDIA_STATUS(ctx, ifmr); CTX_UNLOCK(ctx); } void iflib_stop(if_ctx_t ctx) { iflib_txq_t txq = ctx->ifc_txqs; iflib_rxq_t rxq = ctx->ifc_rxqs; if_softc_ctx_t scctx = &ctx->ifc_softc_ctx; if_shared_ctx_t sctx = ctx->ifc_sctx; iflib_dma_info_t di; iflib_fl_t fl; int i, j; /* Tell the stack that the interface is no longer active */ if_setdrvflagbits(ctx->ifc_ifp, IFF_DRV_OACTIVE, IFF_DRV_RUNNING); IFDI_INTR_DISABLE(ctx); DELAY(1000); IFDI_STOP(ctx); DELAY(1000); iflib_debug_reset(); /* Wait for current tx queue users to exit to disarm watchdog timer. */ for (i = 0; i < scctx->isc_ntxqsets; i++, txq++) { /* make sure all transmitters have completed before proceeding XXX */ CALLOUT_LOCK(txq); callout_stop(&txq->ift_timer); CALLOUT_UNLOCK(txq); /* clean any enqueued buffers */ iflib_ifmp_purge(txq); /* Free any existing tx buffers. */ for (j = 0; j < txq->ift_size; j++) { iflib_txsd_free(ctx, txq, j); } txq->ift_processed = txq->ift_cleaned = txq->ift_cidx_processed = 0; txq->ift_in_use = txq->ift_gen = txq->ift_cidx = txq->ift_pidx = txq->ift_no_desc_avail = 0; txq->ift_closed = txq->ift_mbuf_defrag = txq->ift_mbuf_defrag_failed = 0; txq->ift_no_tx_dma_setup = txq->ift_txd_encap_efbig = txq->ift_map_failed = 0; txq->ift_pullups = 0; ifmp_ring_reset_stats(txq->ift_br); for (j = 0, di = txq->ift_ifdi; j < sctx->isc_ntxqs; j++, di++) bzero((void *)di->idi_vaddr, di->idi_size); } for (i = 0; i < scctx->isc_nrxqsets; i++, rxq++) { /* make sure all transmitters have completed before proceeding XXX */ rxq->ifr_cq_cidx = 0; for (j = 0, di = rxq->ifr_ifdi; j < sctx->isc_nrxqs; j++, di++) bzero((void *)di->idi_vaddr, di->idi_size); /* also resets the free lists pidx/cidx */ for (j = 0, fl = rxq->ifr_fl; j < rxq->ifr_nfl; j++, fl++) iflib_fl_bufs_free(fl); } } static inline caddr_t calc_next_rxd(iflib_fl_t fl, int cidx) { qidx_t size; int nrxd; caddr_t start, end, cur, next; nrxd = fl->ifl_size; size = fl->ifl_rxd_size; start = fl->ifl_ifdi->idi_vaddr; if (__predict_false(size == 0)) return (start); cur = start + size*cidx; end = start + size*nrxd; next = CACHE_PTR_NEXT(cur); return (next < end ? next : start); } static inline void prefetch_pkts(iflib_fl_t fl, int cidx) { int nextptr; int nrxd = fl->ifl_size; caddr_t next_rxd; nextptr = (cidx + CACHE_PTR_INCREMENT) & (nrxd-1); prefetch(&fl->ifl_sds.ifsd_m[nextptr]); prefetch(&fl->ifl_sds.ifsd_cl[nextptr]); next_rxd = calc_next_rxd(fl, cidx); prefetch(next_rxd); prefetch(fl->ifl_sds.ifsd_m[(cidx + 1) & (nrxd-1)]); prefetch(fl->ifl_sds.ifsd_m[(cidx + 2) & (nrxd-1)]); prefetch(fl->ifl_sds.ifsd_m[(cidx + 3) & (nrxd-1)]); prefetch(fl->ifl_sds.ifsd_m[(cidx + 4) & (nrxd-1)]); prefetch(fl->ifl_sds.ifsd_cl[(cidx + 1) & (nrxd-1)]); prefetch(fl->ifl_sds.ifsd_cl[(cidx + 2) & (nrxd-1)]); prefetch(fl->ifl_sds.ifsd_cl[(cidx + 3) & (nrxd-1)]); prefetch(fl->ifl_sds.ifsd_cl[(cidx + 4) & (nrxd-1)]); } static struct mbuf * rxd_frag_to_sd(iflib_rxq_t rxq, if_rxd_frag_t irf, bool unload, if_rxsd_t sd, int *pf_rv, if_rxd_info_t ri) { bus_dmamap_t map; iflib_fl_t fl; caddr_t payload; struct mbuf *m; int flid, cidx, len, next; map = NULL; flid = irf->irf_flid; cidx = irf->irf_idx; fl = &rxq->ifr_fl[flid]; sd->ifsd_fl = fl; m = fl->ifl_sds.ifsd_m[cidx]; sd->ifsd_cl = &fl->ifl_sds.ifsd_cl[cidx]; fl->ifl_credits--; #if MEMORY_LOGGING fl->ifl_m_dequeued++; #endif if (rxq->ifr_ctx->ifc_flags & IFC_PREFETCH) prefetch_pkts(fl, cidx); next = (cidx + CACHE_PTR_INCREMENT) & (fl->ifl_size-1); prefetch(&fl->ifl_sds.ifsd_map[next]); map = fl->ifl_sds.ifsd_map[cidx]; bus_dmamap_sync(fl->ifl_buf_tag, map, BUS_DMASYNC_POSTREAD); if (rxq->pfil != NULL && PFIL_HOOKED_IN(rxq->pfil) && pf_rv != NULL && irf->irf_len != 0) { payload = *sd->ifsd_cl; payload += ri->iri_pad; len = ri->iri_len - ri->iri_pad; *pf_rv = pfil_run_hooks(rxq->pfil, payload, ri->iri_ifp, len | PFIL_MEMPTR | PFIL_IN, NULL); switch (*pf_rv) { case PFIL_DROPPED: case PFIL_CONSUMED: /* * The filter ate it. Everything is recycled. */ m = NULL; unload = 0; break; case PFIL_REALLOCED: /* * The filter copied it. Everything is recycled. */ m = pfil_mem2mbuf(payload); unload = 0; break; case PFIL_PASS: /* * Filter said it was OK, so receive like * normal */ fl->ifl_sds.ifsd_m[cidx] = NULL; break; default: MPASS(0); } } else { fl->ifl_sds.ifsd_m[cidx] = NULL; *pf_rv = PFIL_PASS; } if (unload && irf->irf_len != 0) bus_dmamap_unload(fl->ifl_buf_tag, map); fl->ifl_cidx = (fl->ifl_cidx + 1) & (fl->ifl_size-1); if (__predict_false(fl->ifl_cidx == 0)) fl->ifl_gen = 0; bit_clear(fl->ifl_rx_bitmap, cidx); return (m); } static struct mbuf * assemble_segments(iflib_rxq_t rxq, if_rxd_info_t ri, if_rxsd_t sd, int *pf_rv) { struct mbuf *m, *mh, *mt; caddr_t cl; int *pf_rv_ptr, flags, i, padlen; bool consumed; i = 0; mh = NULL; consumed = false; *pf_rv = PFIL_PASS; pf_rv_ptr = pf_rv; do { m = rxd_frag_to_sd(rxq, &ri->iri_frags[i], !consumed, sd, pf_rv_ptr, ri); MPASS(*sd->ifsd_cl != NULL); /* * Exclude zero-length frags & frags from * packets the filter has consumed or dropped */ if (ri->iri_frags[i].irf_len == 0 || consumed || *pf_rv == PFIL_CONSUMED || *pf_rv == PFIL_DROPPED) { if (mh == NULL) { /* everything saved here */ consumed = true; pf_rv_ptr = NULL; continue; } /* XXX we can save the cluster here, but not the mbuf */ m_init(m, M_NOWAIT, MT_DATA, 0); m_free(m); continue; } if (mh == NULL) { flags = M_PKTHDR|M_EXT; mh = mt = m; padlen = ri->iri_pad; } else { flags = M_EXT; mt->m_next = m; mt = m; /* assuming padding is only on the first fragment */ padlen = 0; } cl = *sd->ifsd_cl; *sd->ifsd_cl = NULL; /* Can these two be made one ? */ m_init(m, M_NOWAIT, MT_DATA, flags); m_cljset(m, cl, sd->ifsd_fl->ifl_cltype); /* * These must follow m_init and m_cljset */ m->m_data += padlen; ri->iri_len -= padlen; m->m_len = ri->iri_frags[i].irf_len; } while (++i < ri->iri_nfrags); return (mh); } /* * Process one software descriptor */ static struct mbuf * iflib_rxd_pkt_get(iflib_rxq_t rxq, if_rxd_info_t ri) { struct if_rxsd sd; struct mbuf *m; int pf_rv; /* should I merge this back in now that the two paths are basically duplicated? */ if (ri->iri_nfrags == 1 && ri->iri_frags[0].irf_len != 0 && ri->iri_frags[0].irf_len <= MIN(IFLIB_RX_COPY_THRESH, MHLEN)) { m = rxd_frag_to_sd(rxq, &ri->iri_frags[0], false, &sd, &pf_rv, ri); if (pf_rv != PFIL_PASS && pf_rv != PFIL_REALLOCED) return (m); if (pf_rv == PFIL_PASS) { m_init(m, M_NOWAIT, MT_DATA, M_PKTHDR); #ifndef __NO_STRICT_ALIGNMENT if (!IP_ALIGNED(m)) m->m_data += 2; #endif memcpy(m->m_data, *sd.ifsd_cl, ri->iri_len); m->m_len = ri->iri_frags[0].irf_len; } } else { m = assemble_segments(rxq, ri, &sd, &pf_rv); if (m == NULL) return (NULL); if (pf_rv != PFIL_PASS && pf_rv != PFIL_REALLOCED) return (m); } m->m_pkthdr.len = ri->iri_len; m->m_pkthdr.rcvif = ri->iri_ifp; m->m_flags |= ri->iri_flags; m->m_pkthdr.ether_vtag = ri->iri_vtag; m->m_pkthdr.flowid = ri->iri_flowid; M_HASHTYPE_SET(m, ri->iri_rsstype); m->m_pkthdr.csum_flags = ri->iri_csum_flags; m->m_pkthdr.csum_data = ri->iri_csum_data; return (m); } #if defined(INET6) || defined(INET) static void iflib_get_ip_forwarding(struct lro_ctrl *lc, bool *v4, bool *v6) { CURVNET_SET(lc->ifp->if_vnet); #if defined(INET6) *v6 = V_ip6_forwarding; #endif #if defined(INET) *v4 = V_ipforwarding; #endif CURVNET_RESTORE(); } /* * Returns true if it's possible this packet could be LROed. * if it returns false, it is guaranteed that tcp_lro_rx() * would not return zero. */ static bool iflib_check_lro_possible(struct mbuf *m, bool v4_forwarding, bool v6_forwarding) { struct ether_header *eh; eh = mtod(m, struct ether_header *); switch (eh->ether_type) { #if defined(INET6) case htons(ETHERTYPE_IPV6): return (!v6_forwarding); #endif #if defined (INET) case htons(ETHERTYPE_IP): return (!v4_forwarding); #endif } return false; } #else static void iflib_get_ip_forwarding(struct lro_ctrl *lc __unused, bool *v4 __unused, bool *v6 __unused) { } #endif static void _task_fn_rx_watchdog(void *context) { iflib_rxq_t rxq = context; GROUPTASK_ENQUEUE(&rxq->ifr_task); } static uint8_t iflib_rxeof(iflib_rxq_t rxq, qidx_t budget) { if_t ifp; if_ctx_t ctx = rxq->ifr_ctx; if_shared_ctx_t sctx = ctx->ifc_sctx; if_softc_ctx_t scctx = &ctx->ifc_softc_ctx; int avail, i; qidx_t *cidxp; struct if_rxd_info ri; int err, budget_left, rx_bytes, rx_pkts; iflib_fl_t fl; int lro_enabled; bool v4_forwarding, v6_forwarding, lro_possible; uint8_t retval = 0; /* * XXX early demux data packets so that if_input processing only handles * acks in interrupt context */ struct mbuf *m, *mh, *mt, *mf; NET_EPOCH_ASSERT(); lro_possible = v4_forwarding = v6_forwarding = false; ifp = ctx->ifc_ifp; mh = mt = NULL; MPASS(budget > 0); rx_pkts = rx_bytes = 0; if (sctx->isc_flags & IFLIB_HAS_RXCQ) cidxp = &rxq->ifr_cq_cidx; else cidxp = &rxq->ifr_fl[0].ifl_cidx; if ((avail = iflib_rxd_avail(ctx, rxq, *cidxp, budget)) == 0) { for (i = 0, fl = &rxq->ifr_fl[0]; i < sctx->isc_nfl; i++, fl++) retval |= iflib_fl_refill_all(ctx, fl); DBG_COUNTER_INC(rx_unavail); return (retval); } /* pfil needs the vnet to be set */ CURVNET_SET_QUIET(ifp->if_vnet); for (budget_left = budget; budget_left > 0 && avail > 0;) { if (__predict_false(!CTX_ACTIVE(ctx))) { DBG_COUNTER_INC(rx_ctx_inactive); break; } /* * Reset client set fields to their default values */ rxd_info_zero(&ri); ri.iri_qsidx = rxq->ifr_id; ri.iri_cidx = *cidxp; ri.iri_ifp = ifp; ri.iri_frags = rxq->ifr_frags; err = ctx->isc_rxd_pkt_get(ctx->ifc_softc, &ri); if (err) goto err; rx_pkts += 1; rx_bytes += ri.iri_len; if (sctx->isc_flags & IFLIB_HAS_RXCQ) { *cidxp = ri.iri_cidx; /* Update our consumer index */ /* XXX NB: shurd - check if this is still safe */ while (rxq->ifr_cq_cidx >= scctx->isc_nrxd[0]) rxq->ifr_cq_cidx -= scctx->isc_nrxd[0]; /* was this only a completion queue message? */ if (__predict_false(ri.iri_nfrags == 0)) continue; } MPASS(ri.iri_nfrags != 0); MPASS(ri.iri_len != 0); /* will advance the cidx on the corresponding free lists */ m = iflib_rxd_pkt_get(rxq, &ri); avail--; budget_left--; if (avail == 0 && budget_left) avail = iflib_rxd_avail(ctx, rxq, *cidxp, budget_left); if (__predict_false(m == NULL)) continue; /* imm_pkt: -- cxgb */ if (mh == NULL) mh = mt = m; else { mt->m_nextpkt = m; mt = m; } } CURVNET_RESTORE(); /* make sure that we can refill faster than drain */ for (i = 0, fl = &rxq->ifr_fl[0]; i < sctx->isc_nfl; i++, fl++) retval |= iflib_fl_refill_all(ctx, fl); lro_enabled = (if_getcapenable(ifp) & IFCAP_LRO); if (lro_enabled) iflib_get_ip_forwarding(&rxq->ifr_lc, &v4_forwarding, &v6_forwarding); mt = mf = NULL; while (mh != NULL) { m = mh; mh = mh->m_nextpkt; m->m_nextpkt = NULL; #ifndef __NO_STRICT_ALIGNMENT if (!IP_ALIGNED(m) && (m = iflib_fixup_rx(m)) == NULL) continue; #endif rx_bytes += m->m_pkthdr.len; rx_pkts++; #if defined(INET6) || defined(INET) if (lro_enabled) { if (!lro_possible) { lro_possible = iflib_check_lro_possible(m, v4_forwarding, v6_forwarding); if (lro_possible && mf != NULL) { ifp->if_input(ifp, mf); DBG_COUNTER_INC(rx_if_input); mt = mf = NULL; } } if ((m->m_pkthdr.csum_flags & (CSUM_L4_CALC|CSUM_L4_VALID)) == (CSUM_L4_CALC|CSUM_L4_VALID)) { if (lro_possible && tcp_lro_rx(&rxq->ifr_lc, m, 0) == 0) continue; } } #endif if (lro_possible) { ifp->if_input(ifp, m); DBG_COUNTER_INC(rx_if_input); continue; } if (mf == NULL) mf = m; if (mt != NULL) mt->m_nextpkt = m; mt = m; } if (mf != NULL) { ifp->if_input(ifp, mf); DBG_COUNTER_INC(rx_if_input); } if_inc_counter(ifp, IFCOUNTER_IBYTES, rx_bytes); if_inc_counter(ifp, IFCOUNTER_IPACKETS, rx_pkts); /* * Flush any outstanding LRO work */ #if defined(INET6) || defined(INET) tcp_lro_flush_all(&rxq->ifr_lc); #endif if (avail != 0 || iflib_rxd_avail(ctx, rxq, *cidxp, 1) != 0) retval |= IFLIB_RXEOF_MORE; return (retval); err: STATE_LOCK(ctx); ctx->ifc_flags |= IFC_DO_RESET; iflib_admin_intr_deferred(ctx); STATE_UNLOCK(ctx); return (0); } #define TXD_NOTIFY_COUNT(txq) (((txq)->ift_size / (txq)->ift_update_freq)-1) static inline qidx_t txq_max_db_deferred(iflib_txq_t txq, qidx_t in_use) { qidx_t notify_count = TXD_NOTIFY_COUNT(txq); qidx_t minthresh = txq->ift_size / 8; if (in_use > 4*minthresh) return (notify_count); if (in_use > 2*minthresh) return (notify_count >> 1); if (in_use > minthresh) return (notify_count >> 3); return (0); } static inline qidx_t txq_max_rs_deferred(iflib_txq_t txq) { qidx_t notify_count = TXD_NOTIFY_COUNT(txq); qidx_t minthresh = txq->ift_size / 8; if (txq->ift_in_use > 4*minthresh) return (notify_count); if (txq->ift_in_use > 2*minthresh) return (notify_count >> 1); if (txq->ift_in_use > minthresh) return (notify_count >> 2); return (2); } #define M_CSUM_FLAGS(m) ((m)->m_pkthdr.csum_flags) #define M_HAS_VLANTAG(m) (m->m_flags & M_VLANTAG) #define TXQ_MAX_DB_DEFERRED(txq, in_use) txq_max_db_deferred((txq), (in_use)) #define TXQ_MAX_RS_DEFERRED(txq) txq_max_rs_deferred(txq) #define TXQ_MAX_DB_CONSUMED(size) (size >> 4) /* forward compatibility for cxgb */ #define FIRST_QSET(ctx) 0 #define NTXQSETS(ctx) ((ctx)->ifc_softc_ctx.isc_ntxqsets) #define NRXQSETS(ctx) ((ctx)->ifc_softc_ctx.isc_nrxqsets) #define QIDX(ctx, m) ((((m)->m_pkthdr.flowid & ctx->ifc_softc_ctx.isc_rss_table_mask) % NTXQSETS(ctx)) + FIRST_QSET(ctx)) #define DESC_RECLAIMABLE(q) ((int)((q)->ift_processed - (q)->ift_cleaned - (q)->ift_ctx->ifc_softc_ctx.isc_tx_nsegments)) /* XXX we should be setting this to something other than zero */ #define RECLAIM_THRESH(ctx) ((ctx)->ifc_sctx->isc_tx_reclaim_thresh) #define MAX_TX_DESC(ctx) max((ctx)->ifc_softc_ctx.isc_tx_tso_segments_max, \ (ctx)->ifc_softc_ctx.isc_tx_nsegments) static inline bool iflib_txd_db_check(if_ctx_t ctx, iflib_txq_t txq, int ring, qidx_t in_use) { qidx_t dbval, max; bool rang; rang = false; max = TXQ_MAX_DB_DEFERRED(txq, in_use); if (ring || txq->ift_db_pending >= max) { dbval = txq->ift_npending ? txq->ift_npending : txq->ift_pidx; bus_dmamap_sync(txq->ift_ifdi->idi_tag, txq->ift_ifdi->idi_map, BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); ctx->isc_txd_flush(ctx->ifc_softc, txq->ift_id, dbval); txq->ift_db_pending = txq->ift_npending = 0; rang = true; } return (rang); } #ifdef PKT_DEBUG static void print_pkt(if_pkt_info_t pi) { printf("pi len: %d qsidx: %d nsegs: %d ndescs: %d flags: %x pidx: %d\n", pi->ipi_len, pi->ipi_qsidx, pi->ipi_nsegs, pi->ipi_ndescs, pi->ipi_flags, pi->ipi_pidx); printf("pi new_pidx: %d csum_flags: %lx tso_segsz: %d mflags: %x vtag: %d\n", pi->ipi_new_pidx, pi->ipi_csum_flags, pi->ipi_tso_segsz, pi->ipi_mflags, pi->ipi_vtag); printf("pi etype: %d ehdrlen: %d ip_hlen: %d ipproto: %d\n", pi->ipi_etype, pi->ipi_ehdrlen, pi->ipi_ip_hlen, pi->ipi_ipproto); } #endif #define IS_TSO4(pi) ((pi)->ipi_csum_flags & CSUM_IP_TSO) #define IS_TX_OFFLOAD4(pi) ((pi)->ipi_csum_flags & (CSUM_IP_TCP | CSUM_IP_TSO)) #define IS_TSO6(pi) ((pi)->ipi_csum_flags & CSUM_IP6_TSO) #define IS_TX_OFFLOAD6(pi) ((pi)->ipi_csum_flags & (CSUM_IP6_TCP | CSUM_IP6_TSO)) static int iflib_parse_header(iflib_txq_t txq, if_pkt_info_t pi, struct mbuf **mp) { if_shared_ctx_t sctx = txq->ift_ctx->ifc_sctx; struct ether_vlan_header *eh; struct mbuf *m; m = *mp; if ((sctx->isc_flags & IFLIB_NEED_SCRATCH) && M_WRITABLE(m) == 0) { if ((m = m_dup(m, M_NOWAIT)) == NULL) { return (ENOMEM); } else { m_freem(*mp); DBG_COUNTER_INC(tx_frees); *mp = m; } } /* * Determine where frame payload starts. * Jump over vlan headers if already present, * helpful for QinQ too. */ if (__predict_false(m->m_len < sizeof(*eh))) { txq->ift_pullups++; if (__predict_false((m = m_pullup(m, sizeof(*eh))) == NULL)) return (ENOMEM); } eh = mtod(m, struct ether_vlan_header *); if (eh->evl_encap_proto == htons(ETHERTYPE_VLAN)) { pi->ipi_etype = ntohs(eh->evl_proto); pi->ipi_ehdrlen = ETHER_HDR_LEN + ETHER_VLAN_ENCAP_LEN; } else { pi->ipi_etype = ntohs(eh->evl_encap_proto); pi->ipi_ehdrlen = ETHER_HDR_LEN; } switch (pi->ipi_etype) { #ifdef INET case ETHERTYPE_IP: { struct mbuf *n; struct ip *ip = NULL; struct tcphdr *th = NULL; int minthlen; minthlen = min(m->m_pkthdr.len, pi->ipi_ehdrlen + sizeof(*ip) + sizeof(*th)); if (__predict_false(m->m_len < minthlen)) { /* * if this code bloat is causing too much of a hit * move it to a separate function and mark it noinline */ if (m->m_len == pi->ipi_ehdrlen) { n = m->m_next; MPASS(n); if (n->m_len >= sizeof(*ip)) { ip = (struct ip *)n->m_data; if (n->m_len >= (ip->ip_hl << 2) + sizeof(*th)) th = (struct tcphdr *)((caddr_t)ip + (ip->ip_hl << 2)); } else { txq->ift_pullups++; if (__predict_false((m = m_pullup(m, minthlen)) == NULL)) return (ENOMEM); ip = (struct ip *)(m->m_data + pi->ipi_ehdrlen); } } else { txq->ift_pullups++; if (__predict_false((m = m_pullup(m, minthlen)) == NULL)) return (ENOMEM); ip = (struct ip *)(m->m_data + pi->ipi_ehdrlen); if (m->m_len >= (ip->ip_hl << 2) + sizeof(*th)) th = (struct tcphdr *)((caddr_t)ip + (ip->ip_hl << 2)); } } else { ip = (struct ip *)(m->m_data + pi->ipi_ehdrlen); if (m->m_len >= (ip->ip_hl << 2) + sizeof(*th)) th = (struct tcphdr *)((caddr_t)ip + (ip->ip_hl << 2)); } pi->ipi_ip_hlen = ip->ip_hl << 2; pi->ipi_ipproto = ip->ip_p; pi->ipi_flags |= IPI_TX_IPV4; /* TCP checksum offload may require TCP header length */ if (IS_TX_OFFLOAD4(pi)) { if (__predict_true(pi->ipi_ipproto == IPPROTO_TCP)) { if (__predict_false(th == NULL)) { txq->ift_pullups++; if (__predict_false((m = m_pullup(m, (ip->ip_hl << 2) + sizeof(*th))) == NULL)) return (ENOMEM); th = (struct tcphdr *)((caddr_t)ip + pi->ipi_ip_hlen); } pi->ipi_tcp_hflags = th->th_flags; pi->ipi_tcp_hlen = th->th_off << 2; pi->ipi_tcp_seq = th->th_seq; } if (IS_TSO4(pi)) { if (__predict_false(ip->ip_p != IPPROTO_TCP)) return (ENXIO); /* * TSO always requires hardware checksum offload. */ pi->ipi_csum_flags |= (CSUM_IP_TCP | CSUM_IP); th->th_sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, htons(IPPROTO_TCP)); pi->ipi_tso_segsz = m->m_pkthdr.tso_segsz; if (sctx->isc_flags & IFLIB_TSO_INIT_IP) { ip->ip_sum = 0; ip->ip_len = htons(pi->ipi_ip_hlen + pi->ipi_tcp_hlen + pi->ipi_tso_segsz); } } } if ((sctx->isc_flags & IFLIB_NEED_ZERO_CSUM) && (pi->ipi_csum_flags & CSUM_IP)) ip->ip_sum = 0; break; } #endif #ifdef INET6 case ETHERTYPE_IPV6: { struct ip6_hdr *ip6 = (struct ip6_hdr *)(m->m_data + pi->ipi_ehdrlen); struct tcphdr *th; pi->ipi_ip_hlen = sizeof(struct ip6_hdr); if (__predict_false(m->m_len < pi->ipi_ehdrlen + sizeof(struct ip6_hdr))) { txq->ift_pullups++; if (__predict_false((m = m_pullup(m, pi->ipi_ehdrlen + sizeof(struct ip6_hdr))) == NULL)) return (ENOMEM); } th = (struct tcphdr *)((caddr_t)ip6 + pi->ipi_ip_hlen); /* XXX-BZ this will go badly in case of ext hdrs. */ pi->ipi_ipproto = ip6->ip6_nxt; pi->ipi_flags |= IPI_TX_IPV6; /* TCP checksum offload may require TCP header length */ if (IS_TX_OFFLOAD6(pi)) { if (pi->ipi_ipproto == IPPROTO_TCP) { if (__predict_false(m->m_len < pi->ipi_ehdrlen + sizeof(struct ip6_hdr) + sizeof(struct tcphdr))) { txq->ift_pullups++; if (__predict_false((m = m_pullup(m, pi->ipi_ehdrlen + sizeof(struct ip6_hdr) + sizeof(struct tcphdr))) == NULL)) return (ENOMEM); } pi->ipi_tcp_hflags = th->th_flags; pi->ipi_tcp_hlen = th->th_off << 2; pi->ipi_tcp_seq = th->th_seq; } if (IS_TSO6(pi)) { if (__predict_false(ip6->ip6_nxt != IPPROTO_TCP)) return (ENXIO); /* * TSO always requires hardware checksum offload. */ pi->ipi_csum_flags |= CSUM_IP6_TCP; th->th_sum = in6_cksum_pseudo(ip6, 0, IPPROTO_TCP, 0); pi->ipi_tso_segsz = m->m_pkthdr.tso_segsz; } } break; } #endif default: pi->ipi_csum_flags &= ~CSUM_OFFLOAD; pi->ipi_ip_hlen = 0; break; } *mp = m; return (0); } /* * If dodgy hardware rejects the scatter gather chain we've handed it * we'll need to remove the mbuf chain from ifsg_m[] before we can add the * m_defrag'd mbufs */ static __noinline struct mbuf * iflib_remove_mbuf(iflib_txq_t txq) { int ntxd, pidx; struct mbuf *m, **ifsd_m; ifsd_m = txq->ift_sds.ifsd_m; ntxd = txq->ift_size; pidx = txq->ift_pidx & (ntxd - 1); ifsd_m = txq->ift_sds.ifsd_m; m = ifsd_m[pidx]; ifsd_m[pidx] = NULL; bus_dmamap_unload(txq->ift_buf_tag, txq->ift_sds.ifsd_map[pidx]); if (txq->ift_sds.ifsd_tso_map != NULL) bus_dmamap_unload(txq->ift_tso_buf_tag, txq->ift_sds.ifsd_tso_map[pidx]); #if MEMORY_LOGGING txq->ift_dequeued++; #endif return (m); } static inline caddr_t calc_next_txd(iflib_txq_t txq, int cidx, uint8_t qid) { qidx_t size; int ntxd; caddr_t start, end, cur, next; ntxd = txq->ift_size; size = txq->ift_txd_size[qid]; start = txq->ift_ifdi[qid].idi_vaddr; if (__predict_false(size == 0)) return (start); cur = start + size*cidx; end = start + size*ntxd; next = CACHE_PTR_NEXT(cur); return (next < end ? next : start); } /* * Pad an mbuf to ensure a minimum ethernet frame size. * min_frame_size is the frame size (less CRC) to pad the mbuf to */ static __noinline int iflib_ether_pad(device_t dev, struct mbuf **m_head, uint16_t min_frame_size) { /* * 18 is enough bytes to pad an ARP packet to 46 bytes, and * and ARP message is the smallest common payload I can think of */ static char pad[18]; /* just zeros */ int n; struct mbuf *new_head; if (!M_WRITABLE(*m_head)) { new_head = m_dup(*m_head, M_NOWAIT); if (new_head == NULL) { m_freem(*m_head); device_printf(dev, "cannot pad short frame, m_dup() failed"); DBG_COUNTER_INC(encap_pad_mbuf_fail); DBG_COUNTER_INC(tx_frees); return ENOMEM; } m_freem(*m_head); *m_head = new_head; } for (n = min_frame_size - (*m_head)->m_pkthdr.len; n > 0; n -= sizeof(pad)) if (!m_append(*m_head, min(n, sizeof(pad)), pad)) break; if (n > 0) { m_freem(*m_head); device_printf(dev, "cannot pad short frame\n"); DBG_COUNTER_INC(encap_pad_mbuf_fail); DBG_COUNTER_INC(tx_frees); return (ENOBUFS); } return 0; } static int iflib_encap(iflib_txq_t txq, struct mbuf **m_headp) { if_ctx_t ctx; if_shared_ctx_t sctx; if_softc_ctx_t scctx; bus_dma_tag_t buf_tag; bus_dma_segment_t *segs; struct mbuf *m_head, **ifsd_m; void *next_txd; bus_dmamap_t map; struct if_pkt_info pi; int remap = 0; int err, nsegs, ndesc, max_segs, pidx, cidx, next, ntxd; ctx = txq->ift_ctx; sctx = ctx->ifc_sctx; scctx = &ctx->ifc_softc_ctx; segs = txq->ift_segs; ntxd = txq->ift_size; m_head = *m_headp; map = NULL; /* * If we're doing TSO the next descriptor to clean may be quite far ahead */ cidx = txq->ift_cidx; pidx = txq->ift_pidx; if (ctx->ifc_flags & IFC_PREFETCH) { next = (cidx + CACHE_PTR_INCREMENT) & (ntxd-1); if (!(ctx->ifc_flags & IFLIB_HAS_TXCQ)) { next_txd = calc_next_txd(txq, cidx, 0); prefetch(next_txd); } /* prefetch the next cache line of mbuf pointers and flags */ prefetch(&txq->ift_sds.ifsd_m[next]); prefetch(&txq->ift_sds.ifsd_map[next]); next = (cidx + CACHE_LINE_SIZE) & (ntxd-1); } map = txq->ift_sds.ifsd_map[pidx]; ifsd_m = txq->ift_sds.ifsd_m; if (m_head->m_pkthdr.csum_flags & CSUM_TSO) { buf_tag = txq->ift_tso_buf_tag; max_segs = scctx->isc_tx_tso_segments_max; map = txq->ift_sds.ifsd_tso_map[pidx]; MPASS(buf_tag != NULL); MPASS(max_segs > 0); } else { buf_tag = txq->ift_buf_tag; max_segs = scctx->isc_tx_nsegments; map = txq->ift_sds.ifsd_map[pidx]; } if ((sctx->isc_flags & IFLIB_NEED_ETHER_PAD) && __predict_false(m_head->m_pkthdr.len < scctx->isc_min_frame_size)) { err = iflib_ether_pad(ctx->ifc_dev, m_headp, scctx->isc_min_frame_size); if (err) { DBG_COUNTER_INC(encap_txd_encap_fail); return err; } } m_head = *m_headp; pkt_info_zero(&pi); pi.ipi_mflags = (m_head->m_flags & (M_VLANTAG|M_BCAST|M_MCAST)); pi.ipi_pidx = pidx; pi.ipi_qsidx = txq->ift_id; pi.ipi_len = m_head->m_pkthdr.len; pi.ipi_csum_flags = m_head->m_pkthdr.csum_flags; pi.ipi_vtag = M_HAS_VLANTAG(m_head) ? m_head->m_pkthdr.ether_vtag : 0; /* deliberate bitwise OR to make one condition */ if (__predict_true((pi.ipi_csum_flags | pi.ipi_vtag))) { if (__predict_false((err = iflib_parse_header(txq, &pi, m_headp)) != 0)) { DBG_COUNTER_INC(encap_txd_encap_fail); return (err); } m_head = *m_headp; } retry: err = bus_dmamap_load_mbuf_sg(buf_tag, map, m_head, segs, &nsegs, BUS_DMA_NOWAIT); defrag: if (__predict_false(err)) { switch (err) { case EFBIG: /* try collapse once and defrag once */ if (remap == 0) { m_head = m_collapse(*m_headp, M_NOWAIT, max_segs); /* try defrag if collapsing fails */ if (m_head == NULL) remap++; } if (remap == 1) { txq->ift_mbuf_defrag++; m_head = m_defrag(*m_headp, M_NOWAIT); } /* * remap should never be >1 unless bus_dmamap_load_mbuf_sg * failed to map an mbuf that was run through m_defrag */ MPASS(remap <= 1); if (__predict_false(m_head == NULL || remap > 1)) goto defrag_failed; remap++; *m_headp = m_head; goto retry; break; case ENOMEM: txq->ift_no_tx_dma_setup++; break; default: txq->ift_no_tx_dma_setup++; m_freem(*m_headp); DBG_COUNTER_INC(tx_frees); *m_headp = NULL; break; } txq->ift_map_failed++; DBG_COUNTER_INC(encap_load_mbuf_fail); DBG_COUNTER_INC(encap_txd_encap_fail); return (err); } ifsd_m[pidx] = m_head; /* * XXX assumes a 1 to 1 relationship between segments and * descriptors - this does not hold true on all drivers, e.g. * cxgb */ if (__predict_false(nsegs + 2 > TXQ_AVAIL(txq))) { txq->ift_no_desc_avail++; bus_dmamap_unload(buf_tag, map); DBG_COUNTER_INC(encap_txq_avail_fail); DBG_COUNTER_INC(encap_txd_encap_fail); if ((txq->ift_task.gt_task.ta_flags & TASK_ENQUEUED) == 0) GROUPTASK_ENQUEUE(&txq->ift_task); return (ENOBUFS); } /* * On Intel cards we can greatly reduce the number of TX interrupts * we see by only setting report status on every Nth descriptor. * However, this also means that the driver will need to keep track * of the descriptors that RS was set on to check them for the DD bit. */ txq->ift_rs_pending += nsegs + 1; if (txq->ift_rs_pending > TXQ_MAX_RS_DEFERRED(txq) || iflib_no_tx_batch || (TXQ_AVAIL(txq) - nsegs) <= MAX_TX_DESC(ctx) + 2) { pi.ipi_flags |= IPI_TX_INTR; txq->ift_rs_pending = 0; } pi.ipi_segs = segs; pi.ipi_nsegs = nsegs; MPASS(pidx >= 0 && pidx < txq->ift_size); #ifdef PKT_DEBUG print_pkt(&pi); #endif if ((err = ctx->isc_txd_encap(ctx->ifc_softc, &pi)) == 0) { bus_dmamap_sync(buf_tag, map, BUS_DMASYNC_PREWRITE); DBG_COUNTER_INC(tx_encap); MPASS(pi.ipi_new_pidx < txq->ift_size); ndesc = pi.ipi_new_pidx - pi.ipi_pidx; if (pi.ipi_new_pidx < pi.ipi_pidx) { ndesc += txq->ift_size; txq->ift_gen = 1; } /* * drivers can need as many as * two sentinels */ MPASS(ndesc <= pi.ipi_nsegs + 2); MPASS(pi.ipi_new_pidx != pidx); MPASS(ndesc > 0); txq->ift_in_use += ndesc; /* * We update the last software descriptor again here because there may * be a sentinel and/or there may be more mbufs than segments */ txq->ift_pidx = pi.ipi_new_pidx; txq->ift_npending += pi.ipi_ndescs; } else { *m_headp = m_head = iflib_remove_mbuf(txq); if (err == EFBIG) { txq->ift_txd_encap_efbig++; if (remap < 2) { remap = 1; goto defrag; } } goto defrag_failed; } /* * err can't possibly be non-zero here, so we don't neet to test it * to see if we need to DBG_COUNTER_INC(encap_txd_encap_fail). */ return (err); defrag_failed: txq->ift_mbuf_defrag_failed++; txq->ift_map_failed++; m_freem(*m_headp); DBG_COUNTER_INC(tx_frees); *m_headp = NULL; DBG_COUNTER_INC(encap_txd_encap_fail); return (ENOMEM); } static void iflib_tx_desc_free(iflib_txq_t txq, int n) { uint32_t qsize, cidx, mask, gen; struct mbuf *m, **ifsd_m; bool do_prefetch; cidx = txq->ift_cidx; gen = txq->ift_gen; qsize = txq->ift_size; mask = qsize-1; ifsd_m = txq->ift_sds.ifsd_m; do_prefetch = (txq->ift_ctx->ifc_flags & IFC_PREFETCH); while (n-- > 0) { if (do_prefetch) { prefetch(ifsd_m[(cidx + 3) & mask]); prefetch(ifsd_m[(cidx + 4) & mask]); } if ((m = ifsd_m[cidx]) != NULL) { prefetch(&ifsd_m[(cidx + CACHE_PTR_INCREMENT) & mask]); if (m->m_pkthdr.csum_flags & CSUM_TSO) { bus_dmamap_sync(txq->ift_tso_buf_tag, txq->ift_sds.ifsd_tso_map[cidx], BUS_DMASYNC_POSTWRITE); bus_dmamap_unload(txq->ift_tso_buf_tag, txq->ift_sds.ifsd_tso_map[cidx]); } else { bus_dmamap_sync(txq->ift_buf_tag, txq->ift_sds.ifsd_map[cidx], BUS_DMASYNC_POSTWRITE); bus_dmamap_unload(txq->ift_buf_tag, txq->ift_sds.ifsd_map[cidx]); } /* XXX we don't support any drivers that batch packets yet */ MPASS(m->m_nextpkt == NULL); m_freem(m); ifsd_m[cidx] = NULL; #if MEMORY_LOGGING txq->ift_dequeued++; #endif DBG_COUNTER_INC(tx_frees); } if (__predict_false(++cidx == qsize)) { cidx = 0; gen = 0; } } txq->ift_cidx = cidx; txq->ift_gen = gen; } static __inline int iflib_completed_tx_reclaim(iflib_txq_t txq, int thresh) { int reclaim; if_ctx_t ctx = txq->ift_ctx; KASSERT(thresh >= 0, ("invalid threshold to reclaim")); MPASS(thresh /*+ MAX_TX_DESC(txq->ift_ctx) */ < txq->ift_size); /* * Need a rate-limiting check so that this isn't called every time */ iflib_tx_credits_update(ctx, txq); reclaim = DESC_RECLAIMABLE(txq); if (reclaim <= thresh /* + MAX_TX_DESC(txq->ift_ctx) */) { #ifdef INVARIANTS if (iflib_verbose_debug) { printf("%s processed=%ju cleaned=%ju tx_nsegments=%d reclaim=%d thresh=%d\n", __FUNCTION__, txq->ift_processed, txq->ift_cleaned, txq->ift_ctx->ifc_softc_ctx.isc_tx_nsegments, reclaim, thresh); } #endif return (0); } iflib_tx_desc_free(txq, reclaim); txq->ift_cleaned += reclaim; txq->ift_in_use -= reclaim; return (reclaim); } static struct mbuf ** _ring_peek_one(struct ifmp_ring *r, int cidx, int offset, int remaining) { int next, size; struct mbuf **items; size = r->size; next = (cidx + CACHE_PTR_INCREMENT) & (size-1); items = __DEVOLATILE(struct mbuf **, &r->items[0]); prefetch(items[(cidx + offset) & (size-1)]); if (remaining > 1) { prefetch2cachelines(&items[next]); prefetch2cachelines(items[(cidx + offset + 1) & (size-1)]); prefetch2cachelines(items[(cidx + offset + 2) & (size-1)]); prefetch2cachelines(items[(cidx + offset + 3) & (size-1)]); } return (__DEVOLATILE(struct mbuf **, &r->items[(cidx + offset) & (size-1)])); } static void iflib_txq_check_drain(iflib_txq_t txq, int budget) { ifmp_ring_check_drainage(txq->ift_br, budget); } static uint32_t iflib_txq_can_drain(struct ifmp_ring *r) { iflib_txq_t txq = r->cookie; if_ctx_t ctx = txq->ift_ctx; if (TXQ_AVAIL(txq) > MAX_TX_DESC(ctx) + 2) return (1); bus_dmamap_sync(txq->ift_ifdi->idi_tag, txq->ift_ifdi->idi_map, BUS_DMASYNC_POSTREAD); return (ctx->isc_txd_credits_update(ctx->ifc_softc, txq->ift_id, false)); } static uint32_t iflib_txq_drain(struct ifmp_ring *r, uint32_t cidx, uint32_t pidx) { iflib_txq_t txq = r->cookie; if_ctx_t ctx = txq->ift_ctx; if_t ifp = ctx->ifc_ifp; struct mbuf *m, **mp; int avail, bytes_sent, consumed, count, err, i, in_use_prev; int mcast_sent, pkt_sent, reclaimed, txq_avail; bool do_prefetch, rang, ring; if (__predict_false(!(if_getdrvflags(ifp) & IFF_DRV_RUNNING) || !LINK_ACTIVE(ctx))) { DBG_COUNTER_INC(txq_drain_notready); return (0); } reclaimed = iflib_completed_tx_reclaim(txq, RECLAIM_THRESH(ctx)); rang = iflib_txd_db_check(ctx, txq, reclaimed, txq->ift_in_use); avail = IDXDIFF(pidx, cidx, r->size); if (__predict_false(ctx->ifc_flags & IFC_QFLUSH)) { DBG_COUNTER_INC(txq_drain_flushing); for (i = 0; i < avail; i++) { if (__predict_true(r->items[(cidx + i) & (r->size-1)] != (void *)txq)) m_free(r->items[(cidx + i) & (r->size-1)]); r->items[(cidx + i) & (r->size-1)] = NULL; } return (avail); } if (__predict_false(if_getdrvflags(ctx->ifc_ifp) & IFF_DRV_OACTIVE)) { txq->ift_qstatus = IFLIB_QUEUE_IDLE; CALLOUT_LOCK(txq); callout_stop(&txq->ift_timer); CALLOUT_UNLOCK(txq); DBG_COUNTER_INC(txq_drain_oactive); return (0); } if (reclaimed) txq->ift_qstatus = IFLIB_QUEUE_IDLE; consumed = mcast_sent = bytes_sent = pkt_sent = 0; count = MIN(avail, TX_BATCH_SIZE); #ifdef INVARIANTS if (iflib_verbose_debug) printf("%s avail=%d ifc_flags=%x txq_avail=%d ", __FUNCTION__, avail, ctx->ifc_flags, TXQ_AVAIL(txq)); #endif do_prefetch = (ctx->ifc_flags & IFC_PREFETCH); txq_avail = TXQ_AVAIL(txq); err = 0; for (i = 0; i < count && txq_avail > MAX_TX_DESC(ctx) + 2; i++) { int rem = do_prefetch ? count - i : 0; mp = _ring_peek_one(r, cidx, i, rem); MPASS(mp != NULL && *mp != NULL); if (__predict_false(*mp == (struct mbuf *)txq)) { consumed++; continue; } in_use_prev = txq->ift_in_use; err = iflib_encap(txq, mp); if (__predict_false(err)) { /* no room - bail out */ if (err == ENOBUFS) break; consumed++; /* we can't send this packet - skip it */ continue; } consumed++; pkt_sent++; m = *mp; DBG_COUNTER_INC(tx_sent); bytes_sent += m->m_pkthdr.len; mcast_sent += !!(m->m_flags & M_MCAST); txq_avail = TXQ_AVAIL(txq); txq->ift_db_pending += (txq->ift_in_use - in_use_prev); ETHER_BPF_MTAP(ifp, m); if (__predict_false(!(ifp->if_drv_flags & IFF_DRV_RUNNING))) break; rang = iflib_txd_db_check(ctx, txq, false, in_use_prev); } /* deliberate use of bitwise or to avoid gratuitous short-circuit */ ring = rang ? false : (iflib_min_tx_latency | err) || (TXQ_AVAIL(txq) < MAX_TX_DESC(ctx)); iflib_txd_db_check(ctx, txq, ring, txq->ift_in_use); if_inc_counter(ifp, IFCOUNTER_OBYTES, bytes_sent); if_inc_counter(ifp, IFCOUNTER_OPACKETS, pkt_sent); if (mcast_sent) if_inc_counter(ifp, IFCOUNTER_OMCASTS, mcast_sent); #ifdef INVARIANTS if (iflib_verbose_debug) printf("consumed=%d\n", consumed); #endif return (consumed); } static uint32_t iflib_txq_drain_always(struct ifmp_ring *r) { return (1); } static uint32_t iflib_txq_drain_free(struct ifmp_ring *r, uint32_t cidx, uint32_t pidx) { int i, avail; struct mbuf **mp; iflib_txq_t txq; txq = r->cookie; txq->ift_qstatus = IFLIB_QUEUE_IDLE; CALLOUT_LOCK(txq); callout_stop(&txq->ift_timer); CALLOUT_UNLOCK(txq); avail = IDXDIFF(pidx, cidx, r->size); for (i = 0; i < avail; i++) { mp = _ring_peek_one(r, cidx, i, avail - i); if (__predict_false(*mp == (struct mbuf *)txq)) continue; m_freem(*mp); DBG_COUNTER_INC(tx_frees); } MPASS(ifmp_ring_is_stalled(r) == 0); return (avail); } static void iflib_ifmp_purge(iflib_txq_t txq) { struct ifmp_ring *r; r = txq->ift_br; r->drain = iflib_txq_drain_free; r->can_drain = iflib_txq_drain_always; ifmp_ring_check_drainage(r, r->size); r->drain = iflib_txq_drain; r->can_drain = iflib_txq_can_drain; } static void _task_fn_tx(void *context) { iflib_txq_t txq = context; if_ctx_t ctx = txq->ift_ctx; if_t ifp = ctx->ifc_ifp; int abdicate = ctx->ifc_sysctl_tx_abdicate; #ifdef IFLIB_DIAGNOSTICS txq->ift_cpu_exec_count[curcpu]++; #endif if (!(if_getdrvflags(ifp) & IFF_DRV_RUNNING)) return; #ifdef DEV_NETMAP if ((if_getcapenable(ifp) & IFCAP_NETMAP) && netmap_tx_irq(ifp, txq->ift_id)) goto skip_ifmp; #endif #ifdef ALTQ if (ALTQ_IS_ENABLED(&ifp->if_snd)) iflib_altq_if_start(ifp); #endif if (txq->ift_db_pending) ifmp_ring_enqueue(txq->ift_br, (void **)&txq, 1, TX_BATCH_SIZE, abdicate); else if (!abdicate) ifmp_ring_check_drainage(txq->ift_br, TX_BATCH_SIZE); /* * When abdicating, we always need to check drainage, not just when we don't enqueue */ if (abdicate) ifmp_ring_check_drainage(txq->ift_br, TX_BATCH_SIZE); #ifdef DEV_NETMAP skip_ifmp: #endif if (ctx->ifc_flags & IFC_LEGACY) IFDI_INTR_ENABLE(ctx); else IFDI_TX_QUEUE_INTR_ENABLE(ctx, txq->ift_id); } static void _task_fn_rx(void *context) { iflib_rxq_t rxq = context; if_ctx_t ctx = rxq->ifr_ctx; uint8_t more; uint16_t budget; #ifdef DEV_NETMAP u_int work = 0; int nmirq; #endif #ifdef IFLIB_DIAGNOSTICS rxq->ifr_cpu_exec_count[curcpu]++; #endif DBG_COUNTER_INC(task_fn_rxs); if (__predict_false(!(if_getdrvflags(ctx->ifc_ifp) & IFF_DRV_RUNNING))) return; #ifdef DEV_NETMAP nmirq = netmap_rx_irq(ctx->ifc_ifp, rxq->ifr_id, &work); if (nmirq != NM_IRQ_PASS) { more = (nmirq == NM_IRQ_RESCHED) ? IFLIB_RXEOF_MORE : 0; goto skip_rxeof; } #endif budget = ctx->ifc_sysctl_rx_budget; if (budget == 0) budget = 16; /* XXX */ more = iflib_rxeof(rxq, budget); #ifdef DEV_NETMAP skip_rxeof: #endif if ((more & IFLIB_RXEOF_MORE) == 0) { if (ctx->ifc_flags & IFC_LEGACY) IFDI_INTR_ENABLE(ctx); else IFDI_RX_QUEUE_INTR_ENABLE(ctx, rxq->ifr_id); DBG_COUNTER_INC(rx_intr_enables); } if (__predict_false(!(if_getdrvflags(ctx->ifc_ifp) & IFF_DRV_RUNNING))) return; if (more & IFLIB_RXEOF_MORE) GROUPTASK_ENQUEUE(&rxq->ifr_task); else if (more & IFLIB_RXEOF_EMPTY) callout_reset_curcpu(&rxq->ifr_watchdog, 1, &_task_fn_rx_watchdog, rxq); } static void _task_fn_admin(void *context) { if_ctx_t ctx = context; if_softc_ctx_t sctx = &ctx->ifc_softc_ctx; iflib_txq_t txq; int i; bool oactive, running, do_reset, do_watchdog, in_detach; uint32_t reset_on = hz / 2; STATE_LOCK(ctx); running = (if_getdrvflags(ctx->ifc_ifp) & IFF_DRV_RUNNING); oactive = (if_getdrvflags(ctx->ifc_ifp) & IFF_DRV_OACTIVE); do_reset = (ctx->ifc_flags & IFC_DO_RESET); do_watchdog = (ctx->ifc_flags & IFC_DO_WATCHDOG); in_detach = (ctx->ifc_flags & IFC_IN_DETACH); ctx->ifc_flags &= ~(IFC_DO_RESET|IFC_DO_WATCHDOG); STATE_UNLOCK(ctx); if ((!running && !oactive) && !(ctx->ifc_sctx->isc_flags & IFLIB_ADMIN_ALWAYS_RUN)) return; if (in_detach) return; CTX_LOCK(ctx); for (txq = ctx->ifc_txqs, i = 0; i < sctx->isc_ntxqsets; i++, txq++) { CALLOUT_LOCK(txq); callout_stop(&txq->ift_timer); CALLOUT_UNLOCK(txq); } if (do_watchdog) { ctx->ifc_watchdog_events++; IFDI_WATCHDOG_RESET(ctx); } IFDI_UPDATE_ADMIN_STATUS(ctx); for (txq = ctx->ifc_txqs, i = 0; i < sctx->isc_ntxqsets; i++, txq++) { #ifdef DEV_NETMAP reset_on = hz / 2; if (if_getcapenable(ctx->ifc_ifp) & IFCAP_NETMAP) iflib_netmap_timer_adjust(ctx, txq, &reset_on); #endif callout_reset_on(&txq->ift_timer, reset_on, iflib_timer, txq, txq->ift_timer.c_cpu); } IFDI_LINK_INTR_ENABLE(ctx); if (do_reset) iflib_if_init_locked(ctx); CTX_UNLOCK(ctx); if (LINK_ACTIVE(ctx) == 0) return; for (txq = ctx->ifc_txqs, i = 0; i < sctx->isc_ntxqsets; i++, txq++) iflib_txq_check_drain(txq, IFLIB_RESTART_BUDGET); } static void _task_fn_iov(void *context) { if_ctx_t ctx = context; if (!(if_getdrvflags(ctx->ifc_ifp) & IFF_DRV_RUNNING) && !(ctx->ifc_sctx->isc_flags & IFLIB_ADMIN_ALWAYS_RUN)) return; CTX_LOCK(ctx); IFDI_VFLR_HANDLE(ctx); CTX_UNLOCK(ctx); } static int iflib_sysctl_int_delay(SYSCTL_HANDLER_ARGS) { int err; if_int_delay_info_t info; if_ctx_t ctx; info = (if_int_delay_info_t)arg1; ctx = info->iidi_ctx; info->iidi_req = req; info->iidi_oidp = oidp; CTX_LOCK(ctx); err = IFDI_SYSCTL_INT_DELAY(ctx, info); CTX_UNLOCK(ctx); return (err); } /********************************************************************* * * IFNET FUNCTIONS * **********************************************************************/ static void iflib_if_init_locked(if_ctx_t ctx) { iflib_stop(ctx); iflib_init_locked(ctx); } static void iflib_if_init(void *arg) { if_ctx_t ctx = arg; CTX_LOCK(ctx); iflib_if_init_locked(ctx); CTX_UNLOCK(ctx); } static int iflib_if_transmit(if_t ifp, struct mbuf *m) { if_ctx_t ctx = if_getsoftc(ifp); iflib_txq_t txq; int err, qidx; int abdicate = ctx->ifc_sysctl_tx_abdicate; if (__predict_false((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0 || !LINK_ACTIVE(ctx))) { DBG_COUNTER_INC(tx_frees); m_freem(m); return (ENETDOWN); } MPASS(m->m_nextpkt == NULL); /* ALTQ-enabled interfaces always use queue 0. */ qidx = 0; if ((NTXQSETS(ctx) > 1) && M_HASHTYPE_GET(m) && !ALTQ_IS_ENABLED(&ifp->if_snd)) qidx = QIDX(ctx, m); /* * XXX calculate buf_ring based on flowid (divvy up bits?) */ txq = &ctx->ifc_txqs[qidx]; #ifdef DRIVER_BACKPRESSURE if (txq->ift_closed) { while (m != NULL) { next = m->m_nextpkt; m->m_nextpkt = NULL; m_freem(m); DBG_COUNTER_INC(tx_frees); m = next; } return (ENOBUFS); } #endif #ifdef notyet qidx = count = 0; mp = marr; next = m; do { count++; next = next->m_nextpkt; } while (next != NULL); if (count > nitems(marr)) if ((mp = malloc(count*sizeof(struct mbuf *), M_IFLIB, M_NOWAIT)) == NULL) { /* XXX check nextpkt */ m_freem(m); /* XXX simplify for now */ DBG_COUNTER_INC(tx_frees); return (ENOBUFS); } for (next = m, i = 0; next != NULL; i++) { mp[i] = next; next = next->m_nextpkt; mp[i]->m_nextpkt = NULL; } #endif DBG_COUNTER_INC(tx_seen); err = ifmp_ring_enqueue(txq->ift_br, (void **)&m, 1, TX_BATCH_SIZE, abdicate); if (abdicate) GROUPTASK_ENQUEUE(&txq->ift_task); if (err) { if (!abdicate) GROUPTASK_ENQUEUE(&txq->ift_task); /* support forthcoming later */ #ifdef DRIVER_BACKPRESSURE txq->ift_closed = TRUE; #endif ifmp_ring_check_drainage(txq->ift_br, TX_BATCH_SIZE); m_freem(m); DBG_COUNTER_INC(tx_frees); } return (err); } #ifdef ALTQ /* * The overall approach to integrating iflib with ALTQ is to continue to use * the iflib mp_ring machinery between the ALTQ queue(s) and the hardware * ring. Technically, when using ALTQ, queueing to an intermediate mp_ring * is redundant/unnecessary, but doing so minimizes the amount of * ALTQ-specific code required in iflib. It is assumed that the overhead of * redundantly queueing to an intermediate mp_ring is swamped by the * performance limitations inherent in using ALTQ. * * When ALTQ support is compiled in, all iflib drivers will use a transmit * routine, iflib_altq_if_transmit(), that checks if ALTQ is enabled for the * given interface. If ALTQ is enabled for an interface, then all * transmitted packets for that interface will be submitted to the ALTQ * subsystem via IFQ_ENQUEUE(). We don't use the legacy if_transmit() * implementation because it uses IFQ_HANDOFF(), which will duplicatively * update stats that the iflib machinery handles, and which is sensitve to * the disused IFF_DRV_OACTIVE flag. Additionally, iflib_altq_if_start() * will be installed as the start routine for use by ALTQ facilities that * need to trigger queue drains on a scheduled basis. * */ static void iflib_altq_if_start(if_t ifp) { struct ifaltq *ifq = &ifp->if_snd; struct mbuf *m; IFQ_LOCK(ifq); IFQ_DEQUEUE_NOLOCK(ifq, m); while (m != NULL) { iflib_if_transmit(ifp, m); IFQ_DEQUEUE_NOLOCK(ifq, m); } IFQ_UNLOCK(ifq); } static int iflib_altq_if_transmit(if_t ifp, struct mbuf *m) { int err; if (ALTQ_IS_ENABLED(&ifp->if_snd)) { IFQ_ENQUEUE(&ifp->if_snd, m, err); if (err == 0) iflib_altq_if_start(ifp); } else err = iflib_if_transmit(ifp, m); return (err); } #endif /* ALTQ */ static void iflib_if_qflush(if_t ifp) { if_ctx_t ctx = if_getsoftc(ifp); iflib_txq_t txq = ctx->ifc_txqs; int i; STATE_LOCK(ctx); ctx->ifc_flags |= IFC_QFLUSH; STATE_UNLOCK(ctx); for (i = 0; i < NTXQSETS(ctx); i++, txq++) while (!(ifmp_ring_is_idle(txq->ift_br) || ifmp_ring_is_stalled(txq->ift_br))) iflib_txq_check_drain(txq, 0); STATE_LOCK(ctx); ctx->ifc_flags &= ~IFC_QFLUSH; STATE_UNLOCK(ctx); /* * When ALTQ is enabled, this will also take care of purging the * ALTQ queue(s). */ if_qflush(ifp); } #define IFCAP_FLAGS (IFCAP_HWCSUM_IPV6 | IFCAP_HWCSUM | IFCAP_LRO | \ IFCAP_TSO | IFCAP_VLAN_HWTAGGING | IFCAP_HWSTATS | \ IFCAP_VLAN_MTU | IFCAP_VLAN_HWFILTER | \ IFCAP_VLAN_HWTSO | IFCAP_VLAN_HWCSUM | IFCAP_NOMAP) static int iflib_if_ioctl(if_t ifp, u_long command, caddr_t data) { if_ctx_t ctx = if_getsoftc(ifp); struct ifreq *ifr = (struct ifreq *)data; #if defined(INET) || defined(INET6) struct ifaddr *ifa = (struct ifaddr *)data; #endif bool avoid_reset = false; int err = 0, reinit = 0, bits; switch (command) { case SIOCSIFADDR: #ifdef INET if (ifa->ifa_addr->sa_family == AF_INET) avoid_reset = true; #endif #ifdef INET6 if (ifa->ifa_addr->sa_family == AF_INET6) avoid_reset = true; #endif /* ** Calling init results in link renegotiation, ** so we avoid doing it when possible. */ if (avoid_reset) { if_setflagbits(ifp, IFF_UP,0); if (!(if_getdrvflags(ifp) & IFF_DRV_RUNNING)) reinit = 1; #ifdef INET if (!(if_getflags(ifp) & IFF_NOARP)) arp_ifinit(ifp, ifa); #endif } else err = ether_ioctl(ifp, command, data); break; case SIOCSIFMTU: CTX_LOCK(ctx); if (ifr->ifr_mtu == if_getmtu(ifp)) { CTX_UNLOCK(ctx); break; } bits = if_getdrvflags(ifp); /* stop the driver and free any clusters before proceeding */ iflib_stop(ctx); if ((err = IFDI_MTU_SET(ctx, ifr->ifr_mtu)) == 0) { STATE_LOCK(ctx); if (ifr->ifr_mtu > ctx->ifc_max_fl_buf_size) ctx->ifc_flags |= IFC_MULTISEG; else ctx->ifc_flags &= ~IFC_MULTISEG; STATE_UNLOCK(ctx); err = if_setmtu(ifp, ifr->ifr_mtu); } iflib_init_locked(ctx); STATE_LOCK(ctx); if_setdrvflags(ifp, bits); STATE_UNLOCK(ctx); CTX_UNLOCK(ctx); break; case SIOCSIFFLAGS: CTX_LOCK(ctx); if (if_getflags(ifp) & IFF_UP) { if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) { if ((if_getflags(ifp) ^ ctx->ifc_if_flags) & (IFF_PROMISC | IFF_ALLMULTI)) { CTX_UNLOCK(ctx); err = IFDI_PROMISC_SET(ctx, if_getflags(ifp)); CTX_LOCK(ctx); } } else reinit = 1; } else if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) { iflib_stop(ctx); } ctx->ifc_if_flags = if_getflags(ifp); CTX_UNLOCK(ctx); break; case SIOCADDMULTI: case SIOCDELMULTI: if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) { CTX_LOCK(ctx); IFDI_INTR_DISABLE(ctx); IFDI_MULTI_SET(ctx); IFDI_INTR_ENABLE(ctx); CTX_UNLOCK(ctx); } break; case SIOCSIFMEDIA: CTX_LOCK(ctx); IFDI_MEDIA_SET(ctx); CTX_UNLOCK(ctx); /* FALLTHROUGH */ case SIOCGIFMEDIA: case SIOCGIFXMEDIA: err = ifmedia_ioctl(ifp, ifr, ctx->ifc_mediap, command); break; case SIOCGI2C: { struct ifi2creq i2c; err = copyin(ifr_data_get_ptr(ifr), &i2c, sizeof(i2c)); if (err != 0) break; if (i2c.dev_addr != 0xA0 && i2c.dev_addr != 0xA2) { err = EINVAL; break; } if (i2c.len > sizeof(i2c.data)) { err = EINVAL; break; } if ((err = IFDI_I2C_REQ(ctx, &i2c)) == 0) err = copyout(&i2c, ifr_data_get_ptr(ifr), sizeof(i2c)); break; } case SIOCSIFCAP: { int mask, setmask, oldmask; oldmask = if_getcapenable(ifp); mask = ifr->ifr_reqcap ^ oldmask; mask &= ctx->ifc_softc_ctx.isc_capabilities | IFCAP_NOMAP; setmask = 0; #ifdef TCP_OFFLOAD setmask |= mask & (IFCAP_TOE4|IFCAP_TOE6); #endif setmask |= (mask & IFCAP_FLAGS); setmask |= (mask & IFCAP_WOL); /* * If any RX csum has changed, change all the ones that * are supported by the driver. */ if (setmask & (IFCAP_RXCSUM | IFCAP_RXCSUM_IPV6)) { setmask |= ctx->ifc_softc_ctx.isc_capabilities & (IFCAP_RXCSUM | IFCAP_RXCSUM_IPV6); } /* * want to ensure that traffic has stopped before we change any of the flags */ if (setmask) { CTX_LOCK(ctx); bits = if_getdrvflags(ifp); if (bits & IFF_DRV_RUNNING && setmask & ~IFCAP_WOL) iflib_stop(ctx); STATE_LOCK(ctx); if_togglecapenable(ifp, setmask); STATE_UNLOCK(ctx); if (bits & IFF_DRV_RUNNING && setmask & ~IFCAP_WOL) iflib_init_locked(ctx); STATE_LOCK(ctx); if_setdrvflags(ifp, bits); STATE_UNLOCK(ctx); CTX_UNLOCK(ctx); } if_vlancap(ifp); break; } case SIOCGPRIVATE_0: case SIOCSDRVSPEC: case SIOCGDRVSPEC: CTX_LOCK(ctx); err = IFDI_PRIV_IOCTL(ctx, command, data); CTX_UNLOCK(ctx); break; default: err = ether_ioctl(ifp, command, data); break; } if (reinit) iflib_if_init(ctx); return (err); } static uint64_t iflib_if_get_counter(if_t ifp, ift_counter cnt) { if_ctx_t ctx = if_getsoftc(ifp); return (IFDI_GET_COUNTER(ctx, cnt)); } /********************************************************************* * * OTHER FUNCTIONS EXPORTED TO THE STACK * **********************************************************************/ static void iflib_vlan_register(void *arg, if_t ifp, uint16_t vtag) { if_ctx_t ctx = if_getsoftc(ifp); if ((void *)ctx != arg) return; if ((vtag == 0) || (vtag > 4095)) return; if (iflib_in_detach(ctx)) return; CTX_LOCK(ctx); /* Driver may need all untagged packets to be flushed */ if (IFDI_NEEDS_RESTART(ctx, IFLIB_RESTART_VLAN_CONFIG)) iflib_stop(ctx); IFDI_VLAN_REGISTER(ctx, vtag); /* Re-init to load the changes, if required */ if (IFDI_NEEDS_RESTART(ctx, IFLIB_RESTART_VLAN_CONFIG)) iflib_init_locked(ctx); CTX_UNLOCK(ctx); } static void iflib_vlan_unregister(void *arg, if_t ifp, uint16_t vtag) { if_ctx_t ctx = if_getsoftc(ifp); if ((void *)ctx != arg) return; if ((vtag == 0) || (vtag > 4095)) return; CTX_LOCK(ctx); /* Driver may need all tagged packets to be flushed */ if (IFDI_NEEDS_RESTART(ctx, IFLIB_RESTART_VLAN_CONFIG)) iflib_stop(ctx); IFDI_VLAN_UNREGISTER(ctx, vtag); /* Re-init to load the changes, if required */ if (IFDI_NEEDS_RESTART(ctx, IFLIB_RESTART_VLAN_CONFIG)) iflib_init_locked(ctx); CTX_UNLOCK(ctx); } static void iflib_led_func(void *arg, int onoff) { if_ctx_t ctx = arg; CTX_LOCK(ctx); IFDI_LED_FUNC(ctx, onoff); CTX_UNLOCK(ctx); } /********************************************************************* * * BUS FUNCTION DEFINITIONS * **********************************************************************/ int iflib_device_probe(device_t dev) { const pci_vendor_info_t *ent; if_shared_ctx_t sctx; uint16_t pci_device_id, pci_rev_id, pci_subdevice_id, pci_subvendor_id; uint16_t pci_vendor_id; if ((sctx = DEVICE_REGISTER(dev)) == NULL || sctx->isc_magic != IFLIB_MAGIC) return (ENOTSUP); pci_vendor_id = pci_get_vendor(dev); pci_device_id = pci_get_device(dev); pci_subvendor_id = pci_get_subvendor(dev); pci_subdevice_id = pci_get_subdevice(dev); pci_rev_id = pci_get_revid(dev); if (sctx->isc_parse_devinfo != NULL) sctx->isc_parse_devinfo(&pci_device_id, &pci_subvendor_id, &pci_subdevice_id, &pci_rev_id); ent = sctx->isc_vendor_info; while (ent->pvi_vendor_id != 0) { if (pci_vendor_id != ent->pvi_vendor_id) { ent++; continue; } if ((pci_device_id == ent->pvi_device_id) && ((pci_subvendor_id == ent->pvi_subvendor_id) || (ent->pvi_subvendor_id == 0)) && ((pci_subdevice_id == ent->pvi_subdevice_id) || (ent->pvi_subdevice_id == 0)) && ((pci_rev_id == ent->pvi_rev_id) || (ent->pvi_rev_id == 0))) { device_set_desc_copy(dev, ent->pvi_name); /* this needs to be changed to zero if the bus probing code * ever stops re-probing on best match because the sctx * may have its values over written by register calls * in subsequent probes */ return (BUS_PROBE_DEFAULT); } ent++; } return (ENXIO); } int iflib_device_probe_vendor(device_t dev) { int probe; probe = iflib_device_probe(dev); if (probe == BUS_PROBE_DEFAULT) return (BUS_PROBE_VENDOR); else return (probe); } static void iflib_reset_qvalues(if_ctx_t ctx) { if_softc_ctx_t scctx = &ctx->ifc_softc_ctx; if_shared_ctx_t sctx = ctx->ifc_sctx; device_t dev = ctx->ifc_dev; int i; if (ctx->ifc_sysctl_ntxqs != 0) scctx->isc_ntxqsets = ctx->ifc_sysctl_ntxqs; if (ctx->ifc_sysctl_nrxqs != 0) scctx->isc_nrxqsets = ctx->ifc_sysctl_nrxqs; for (i = 0; i < sctx->isc_ntxqs; i++) { if (ctx->ifc_sysctl_ntxds[i] != 0) scctx->isc_ntxd[i] = ctx->ifc_sysctl_ntxds[i]; else scctx->isc_ntxd[i] = sctx->isc_ntxd_default[i]; } for (i = 0; i < sctx->isc_nrxqs; i++) { if (ctx->ifc_sysctl_nrxds[i] != 0) scctx->isc_nrxd[i] = ctx->ifc_sysctl_nrxds[i]; else scctx->isc_nrxd[i] = sctx->isc_nrxd_default[i]; } for (i = 0; i < sctx->isc_nrxqs; i++) { if (scctx->isc_nrxd[i] < sctx->isc_nrxd_min[i]) { device_printf(dev, "nrxd%d: %d less than nrxd_min %d - resetting to min\n", i, scctx->isc_nrxd[i], sctx->isc_nrxd_min[i]); scctx->isc_nrxd[i] = sctx->isc_nrxd_min[i]; } if (scctx->isc_nrxd[i] > sctx->isc_nrxd_max[i]) { device_printf(dev, "nrxd%d: %d greater than nrxd_max %d - resetting to max\n", i, scctx->isc_nrxd[i], sctx->isc_nrxd_max[i]); scctx->isc_nrxd[i] = sctx->isc_nrxd_max[i]; } if (!powerof2(scctx->isc_nrxd[i])) { device_printf(dev, "nrxd%d: %d is not a power of 2 - using default value of %d\n", i, scctx->isc_nrxd[i], sctx->isc_nrxd_default[i]); scctx->isc_nrxd[i] = sctx->isc_nrxd_default[i]; } } for (i = 0; i < sctx->isc_ntxqs; i++) { if (scctx->isc_ntxd[i] < sctx->isc_ntxd_min[i]) { device_printf(dev, "ntxd%d: %d less than ntxd_min %d - resetting to min\n", i, scctx->isc_ntxd[i], sctx->isc_ntxd_min[i]); scctx->isc_ntxd[i] = sctx->isc_ntxd_min[i]; } if (scctx->isc_ntxd[i] > sctx->isc_ntxd_max[i]) { device_printf(dev, "ntxd%d: %d greater than ntxd_max %d - resetting to max\n", i, scctx->isc_ntxd[i], sctx->isc_ntxd_max[i]); scctx->isc_ntxd[i] = sctx->isc_ntxd_max[i]; } if (!powerof2(scctx->isc_ntxd[i])) { device_printf(dev, "ntxd%d: %d is not a power of 2 - using default value of %d\n", i, scctx->isc_ntxd[i], sctx->isc_ntxd_default[i]); scctx->isc_ntxd[i] = sctx->isc_ntxd_default[i]; } } } static void iflib_add_pfil(if_ctx_t ctx) { struct pfil_head *pfil; struct pfil_head_args pa; iflib_rxq_t rxq; int i; pa.pa_version = PFIL_VERSION; pa.pa_flags = PFIL_IN; pa.pa_type = PFIL_TYPE_ETHERNET; pa.pa_headname = ctx->ifc_ifp->if_xname; pfil = pfil_head_register(&pa); for (i = 0, rxq = ctx->ifc_rxqs; i < NRXQSETS(ctx); i++, rxq++) { rxq->pfil = pfil; } } static void iflib_rem_pfil(if_ctx_t ctx) { struct pfil_head *pfil; iflib_rxq_t rxq; int i; rxq = ctx->ifc_rxqs; pfil = rxq->pfil; for (i = 0; i < NRXQSETS(ctx); i++, rxq++) { rxq->pfil = NULL; } pfil_head_unregister(pfil); } static uint16_t get_ctx_core_offset(if_ctx_t ctx) { if_softc_ctx_t scctx = &ctx->ifc_softc_ctx; struct cpu_offset *op; uint16_t qc; uint16_t ret = ctx->ifc_sysctl_core_offset; if (ret != CORE_OFFSET_UNSPECIFIED) return (ret); if (ctx->ifc_sysctl_separate_txrx) qc = scctx->isc_ntxqsets + scctx->isc_nrxqsets; else qc = max(scctx->isc_ntxqsets, scctx->isc_nrxqsets); mtx_lock(&cpu_offset_mtx); SLIST_FOREACH(op, &cpu_offsets, entries) { if (CPU_CMP(&ctx->ifc_cpus, &op->set) == 0) { ret = op->offset; op->offset += qc; MPASS(op->refcount < UINT_MAX); op->refcount++; break; } } if (ret == CORE_OFFSET_UNSPECIFIED) { ret = 0; op = malloc(sizeof(struct cpu_offset), M_IFLIB, M_NOWAIT | M_ZERO); if (op == NULL) { device_printf(ctx->ifc_dev, "allocation for cpu offset failed.\n"); } else { op->offset = qc; op->refcount = 1; CPU_COPY(&ctx->ifc_cpus, &op->set); SLIST_INSERT_HEAD(&cpu_offsets, op, entries); } } mtx_unlock(&cpu_offset_mtx); return (ret); } static void unref_ctx_core_offset(if_ctx_t ctx) { struct cpu_offset *op, *top; mtx_lock(&cpu_offset_mtx); SLIST_FOREACH_SAFE(op, &cpu_offsets, entries, top) { if (CPU_CMP(&ctx->ifc_cpus, &op->set) == 0) { MPASS(op->refcount > 0); op->refcount--; if (op->refcount == 0) { SLIST_REMOVE(&cpu_offsets, op, cpu_offset, entries); free(op, M_IFLIB); } break; } } mtx_unlock(&cpu_offset_mtx); } int iflib_device_register(device_t dev, void *sc, if_shared_ctx_t sctx, if_ctx_t *ctxp) { if_ctx_t ctx; if_t ifp; if_softc_ctx_t scctx; kobjop_desc_t kobj_desc; kobj_method_t *kobj_method; int err, msix, rid; int num_txd, num_rxd; ctx = malloc(sizeof(* ctx), M_IFLIB, M_WAITOK|M_ZERO); if (sc == NULL) { sc = malloc(sctx->isc_driver->size, M_IFLIB, M_WAITOK|M_ZERO); device_set_softc(dev, ctx); ctx->ifc_flags |= IFC_SC_ALLOCATED; } ctx->ifc_sctx = sctx; ctx->ifc_dev = dev; ctx->ifc_softc = sc; if ((err = iflib_register(ctx)) != 0) { device_printf(dev, "iflib_register failed %d\n", err); goto fail_ctx_free; } iflib_add_device_sysctl_pre(ctx); scctx = &ctx->ifc_softc_ctx; ifp = ctx->ifc_ifp; iflib_reset_qvalues(ctx); CTX_LOCK(ctx); if ((err = IFDI_ATTACH_PRE(ctx)) != 0) { device_printf(dev, "IFDI_ATTACH_PRE failed %d\n", err); goto fail_unlock; } _iflib_pre_assert(scctx); ctx->ifc_txrx = *scctx->isc_txrx; if (sctx->isc_flags & IFLIB_DRIVER_MEDIA) ctx->ifc_mediap = scctx->isc_media; #ifdef INVARIANTS if (scctx->isc_capabilities & IFCAP_TXCSUM) MPASS(scctx->isc_tx_csum_flags); #endif if_setcapabilities(ifp, scctx->isc_capabilities | IFCAP_HWSTATS | IFCAP_NOMAP); if_setcapenable(ifp, scctx->isc_capenable | IFCAP_HWSTATS | IFCAP_NOMAP); if (scctx->isc_ntxqsets == 0 || (scctx->isc_ntxqsets_max && scctx->isc_ntxqsets_max < scctx->isc_ntxqsets)) scctx->isc_ntxqsets = scctx->isc_ntxqsets_max; if (scctx->isc_nrxqsets == 0 || (scctx->isc_nrxqsets_max && scctx->isc_nrxqsets_max < scctx->isc_nrxqsets)) scctx->isc_nrxqsets = scctx->isc_nrxqsets_max; num_txd = iflib_num_tx_descs(ctx); num_rxd = iflib_num_rx_descs(ctx); /* XXX change for per-queue sizes */ device_printf(dev, "Using %d TX descriptors and %d RX descriptors\n", num_txd, num_rxd); if (scctx->isc_tx_nsegments > num_txd / MAX_SINGLE_PACKET_FRACTION) scctx->isc_tx_nsegments = max(1, num_txd / MAX_SINGLE_PACKET_FRACTION); if (scctx->isc_tx_tso_segments_max > num_txd / MAX_SINGLE_PACKET_FRACTION) scctx->isc_tx_tso_segments_max = max(1, num_txd / MAX_SINGLE_PACKET_FRACTION); /* TSO parameters - dig these out of the data sheet - simply correspond to tag setup */ if (if_getcapabilities(ifp) & IFCAP_TSO) { /* * The stack can't handle a TSO size larger than IP_MAXPACKET, * but some MACs do. */ if_sethwtsomax(ifp, min(scctx->isc_tx_tso_size_max, IP_MAXPACKET)); /* * Take maximum number of m_pullup(9)'s in iflib_parse_header() * into account. In the worst case, each of these calls will * add another mbuf and, thus, the requirement for another DMA * segment. So for best performance, it doesn't make sense to * advertize a maximum of TSO segments that typically will * require defragmentation in iflib_encap(). */ if_sethwtsomaxsegcount(ifp, scctx->isc_tx_tso_segments_max - 3); if_sethwtsomaxsegsize(ifp, scctx->isc_tx_tso_segsize_max); } if (scctx->isc_rss_table_size == 0) scctx->isc_rss_table_size = 64; scctx->isc_rss_table_mask = scctx->isc_rss_table_size-1; GROUPTASK_INIT(&ctx->ifc_admin_task, 0, _task_fn_admin, ctx); /* XXX format name */ taskqgroup_attach(qgroup_if_config_tqg, &ctx->ifc_admin_task, ctx, NULL, NULL, "admin"); /* Set up cpu set. If it fails, use the set of all CPUs. */ if (bus_get_cpus(dev, INTR_CPUS, sizeof(ctx->ifc_cpus), &ctx->ifc_cpus) != 0) { device_printf(dev, "Unable to fetch CPU list\n"); CPU_COPY(&all_cpus, &ctx->ifc_cpus); } MPASS(CPU_COUNT(&ctx->ifc_cpus) > 0); /* ** Now set up MSI or MSI-X, should return us the number of supported ** vectors (will be 1 for a legacy interrupt and MSI). */ if (sctx->isc_flags & IFLIB_SKIP_MSIX) { msix = scctx->isc_vectors; } else if (scctx->isc_msix_bar != 0) /* * The simple fact that isc_msix_bar is not 0 does not mean we * we have a good value there that is known to work. */ msix = iflib_msix_init(ctx); else { scctx->isc_vectors = 1; scctx->isc_ntxqsets = 1; scctx->isc_nrxqsets = 1; scctx->isc_intr = IFLIB_INTR_LEGACY; msix = 0; } /* Get memory for the station queues */ if ((err = iflib_queues_alloc(ctx))) { device_printf(dev, "Unable to allocate queue memory\n"); goto fail_intr_free; } if ((err = iflib_qset_structures_setup(ctx))) goto fail_queues; /* * Now that we know how many queues there are, get the core offset. */ ctx->ifc_sysctl_core_offset = get_ctx_core_offset(ctx); if (msix > 1) { /* * When using MSI-X, ensure that ifdi_{r,t}x_queue_intr_enable * aren't the default NULL implementation. */ kobj_desc = &ifdi_rx_queue_intr_enable_desc; kobj_method = kobj_lookup_method(((kobj_t)ctx)->ops->cls, NULL, kobj_desc); if (kobj_method == &kobj_desc->deflt) { device_printf(dev, "MSI-X requires ifdi_rx_queue_intr_enable method"); err = EOPNOTSUPP; goto fail_queues; } kobj_desc = &ifdi_tx_queue_intr_enable_desc; kobj_method = kobj_lookup_method(((kobj_t)ctx)->ops->cls, NULL, kobj_desc); if (kobj_method == &kobj_desc->deflt) { device_printf(dev, "MSI-X requires ifdi_tx_queue_intr_enable method"); err = EOPNOTSUPP; goto fail_queues; } /* * Assign the MSI-X vectors. * Note that the default NULL ifdi_msix_intr_assign method will * fail here, too. */ err = IFDI_MSIX_INTR_ASSIGN(ctx, msix); if (err != 0) { device_printf(dev, "IFDI_MSIX_INTR_ASSIGN failed %d\n", err); goto fail_queues; } } else if (scctx->isc_intr != IFLIB_INTR_MSIX) { rid = 0; if (scctx->isc_intr == IFLIB_INTR_MSI) { MPASS(msix == 1); rid = 1; } if ((err = iflib_legacy_setup(ctx, ctx->isc_legacy_intr, ctx->ifc_softc, &rid, "irq0")) != 0) { device_printf(dev, "iflib_legacy_setup failed %d\n", err); goto fail_queues; } } else { device_printf(dev, "Cannot use iflib with only 1 MSI-X interrupt!\n"); err = ENODEV; goto fail_intr_free; } ether_ifattach(ctx->ifc_ifp, ctx->ifc_mac.octet); if ((err = IFDI_ATTACH_POST(ctx)) != 0) { device_printf(dev, "IFDI_ATTACH_POST failed %d\n", err); goto fail_detach; } /* * Tell the upper layer(s) if IFCAP_VLAN_MTU is supported. * This must appear after the call to ether_ifattach() because * ether_ifattach() sets if_hdrlen to the default value. */ if (if_getcapabilities(ifp) & IFCAP_VLAN_MTU) if_setifheaderlen(ifp, sizeof(struct ether_vlan_header)); if ((err = iflib_netmap_attach(ctx))) { device_printf(ctx->ifc_dev, "netmap attach failed: %d\n", err); goto fail_detach; } *ctxp = ctx; DEBUGNET_SET(ctx->ifc_ifp, iflib); if_setgetcounterfn(ctx->ifc_ifp, iflib_if_get_counter); iflib_add_device_sysctl_post(ctx); iflib_add_pfil(ctx); ctx->ifc_flags |= IFC_INIT_DONE; CTX_UNLOCK(ctx); return (0); fail_detach: ether_ifdetach(ctx->ifc_ifp); fail_intr_free: iflib_free_intr_mem(ctx); fail_queues: iflib_tx_structures_free(ctx); iflib_rx_structures_free(ctx); taskqgroup_detach(qgroup_if_config_tqg, &ctx->ifc_admin_task); IFDI_DETACH(ctx); fail_unlock: CTX_UNLOCK(ctx); iflib_deregister(ctx); fail_ctx_free: device_set_softc(ctx->ifc_dev, NULL); if (ctx->ifc_flags & IFC_SC_ALLOCATED) free(ctx->ifc_softc, M_IFLIB); free(ctx, M_IFLIB); return (err); } int iflib_pseudo_register(device_t dev, if_shared_ctx_t sctx, if_ctx_t *ctxp, struct iflib_cloneattach_ctx *clctx) { int num_txd, num_rxd; int err; if_ctx_t ctx; if_t ifp; if_softc_ctx_t scctx; int i; void *sc; ctx = malloc(sizeof(*ctx), M_IFLIB, M_WAITOK|M_ZERO); sc = malloc(sctx->isc_driver->size, M_IFLIB, M_WAITOK|M_ZERO); ctx->ifc_flags |= IFC_SC_ALLOCATED; if (sctx->isc_flags & (IFLIB_PSEUDO|IFLIB_VIRTUAL)) ctx->ifc_flags |= IFC_PSEUDO; ctx->ifc_sctx = sctx; ctx->ifc_softc = sc; ctx->ifc_dev = dev; if ((err = iflib_register(ctx)) != 0) { device_printf(dev, "%s: iflib_register failed %d\n", __func__, err); goto fail_ctx_free; } iflib_add_device_sysctl_pre(ctx); scctx = &ctx->ifc_softc_ctx; ifp = ctx->ifc_ifp; iflib_reset_qvalues(ctx); CTX_LOCK(ctx); if ((err = IFDI_ATTACH_PRE(ctx)) != 0) { device_printf(dev, "IFDI_ATTACH_PRE failed %d\n", err); goto fail_unlock; } if (sctx->isc_flags & IFLIB_GEN_MAC) ether_gen_addr(ifp, &ctx->ifc_mac); if ((err = IFDI_CLONEATTACH(ctx, clctx->cc_ifc, clctx->cc_name, clctx->cc_params)) != 0) { device_printf(dev, "IFDI_CLONEATTACH failed %d\n", err); goto fail_unlock; } #ifdef INVARIANTS if (scctx->isc_capabilities & IFCAP_TXCSUM) MPASS(scctx->isc_tx_csum_flags); #endif if_setcapabilities(ifp, scctx->isc_capabilities | IFCAP_HWSTATS | IFCAP_LINKSTATE); if_setcapenable(ifp, scctx->isc_capenable | IFCAP_HWSTATS | IFCAP_LINKSTATE); ifp->if_flags |= IFF_NOGROUP; if (sctx->isc_flags & IFLIB_PSEUDO) { ifmedia_add(ctx->ifc_mediap, IFM_ETHER | IFM_AUTO, 0, NULL); ifmedia_set(ctx->ifc_mediap, IFM_ETHER | IFM_AUTO); if (sctx->isc_flags & IFLIB_PSEUDO_ETHER) { ether_ifattach(ctx->ifc_ifp, ctx->ifc_mac.octet); } else { if_attach(ctx->ifc_ifp); bpfattach(ctx->ifc_ifp, DLT_NULL, sizeof(u_int32_t)); } if ((err = IFDI_ATTACH_POST(ctx)) != 0) { device_printf(dev, "IFDI_ATTACH_POST failed %d\n", err); goto fail_detach; } *ctxp = ctx; /* * Tell the upper layer(s) if IFCAP_VLAN_MTU is supported. * This must appear after the call to ether_ifattach() because * ether_ifattach() sets if_hdrlen to the default value. */ if (if_getcapabilities(ifp) & IFCAP_VLAN_MTU) if_setifheaderlen(ifp, sizeof(struct ether_vlan_header)); if_setgetcounterfn(ctx->ifc_ifp, iflib_if_get_counter); iflib_add_device_sysctl_post(ctx); ctx->ifc_flags |= IFC_INIT_DONE; CTX_UNLOCK(ctx); return (0); } ifmedia_add(ctx->ifc_mediap, IFM_ETHER | IFM_1000_T | IFM_FDX, 0, NULL); ifmedia_add(ctx->ifc_mediap, IFM_ETHER | IFM_AUTO, 0, NULL); ifmedia_set(ctx->ifc_mediap, IFM_ETHER | IFM_AUTO); _iflib_pre_assert(scctx); ctx->ifc_txrx = *scctx->isc_txrx; if (scctx->isc_ntxqsets == 0 || (scctx->isc_ntxqsets_max && scctx->isc_ntxqsets_max < scctx->isc_ntxqsets)) scctx->isc_ntxqsets = scctx->isc_ntxqsets_max; if (scctx->isc_nrxqsets == 0 || (scctx->isc_nrxqsets_max && scctx->isc_nrxqsets_max < scctx->isc_nrxqsets)) scctx->isc_nrxqsets = scctx->isc_nrxqsets_max; num_txd = iflib_num_tx_descs(ctx); num_rxd = iflib_num_rx_descs(ctx); /* XXX change for per-queue sizes */ device_printf(dev, "Using %d TX descriptors and %d RX descriptors\n", num_txd, num_rxd); if (scctx->isc_tx_nsegments > num_txd / MAX_SINGLE_PACKET_FRACTION) scctx->isc_tx_nsegments = max(1, num_txd / MAX_SINGLE_PACKET_FRACTION); if (scctx->isc_tx_tso_segments_max > num_txd / MAX_SINGLE_PACKET_FRACTION) scctx->isc_tx_tso_segments_max = max(1, num_txd / MAX_SINGLE_PACKET_FRACTION); /* TSO parameters - dig these out of the data sheet - simply correspond to tag setup */ if (if_getcapabilities(ifp) & IFCAP_TSO) { /* * The stack can't handle a TSO size larger than IP_MAXPACKET, * but some MACs do. */ if_sethwtsomax(ifp, min(scctx->isc_tx_tso_size_max, IP_MAXPACKET)); /* * Take maximum number of m_pullup(9)'s in iflib_parse_header() * into account. In the worst case, each of these calls will * add another mbuf and, thus, the requirement for another DMA * segment. So for best performance, it doesn't make sense to * advertize a maximum of TSO segments that typically will * require defragmentation in iflib_encap(). */ if_sethwtsomaxsegcount(ifp, scctx->isc_tx_tso_segments_max - 3); if_sethwtsomaxsegsize(ifp, scctx->isc_tx_tso_segsize_max); } if (scctx->isc_rss_table_size == 0) scctx->isc_rss_table_size = 64; scctx->isc_rss_table_mask = scctx->isc_rss_table_size-1; GROUPTASK_INIT(&ctx->ifc_admin_task, 0, _task_fn_admin, ctx); /* XXX format name */ taskqgroup_attach(qgroup_if_config_tqg, &ctx->ifc_admin_task, ctx, NULL, NULL, "admin"); /* XXX --- can support > 1 -- but keep it simple for now */ scctx->isc_intr = IFLIB_INTR_LEGACY; /* Get memory for the station queues */ if ((err = iflib_queues_alloc(ctx))) { device_printf(dev, "Unable to allocate queue memory\n"); goto fail_iflib_detach; } if ((err = iflib_qset_structures_setup(ctx))) { device_printf(dev, "qset structure setup failed %d\n", err); goto fail_queues; } /* * XXX What if anything do we want to do about interrupts? */ ether_ifattach(ctx->ifc_ifp, ctx->ifc_mac.octet); if ((err = IFDI_ATTACH_POST(ctx)) != 0) { device_printf(dev, "IFDI_ATTACH_POST failed %d\n", err); goto fail_detach; } /* * Tell the upper layer(s) if IFCAP_VLAN_MTU is supported. * This must appear after the call to ether_ifattach() because * ether_ifattach() sets if_hdrlen to the default value. */ if (if_getcapabilities(ifp) & IFCAP_VLAN_MTU) if_setifheaderlen(ifp, sizeof(struct ether_vlan_header)); /* XXX handle more than one queue */ for (i = 0; i < scctx->isc_nrxqsets; i++) IFDI_RX_CLSET(ctx, 0, i, ctx->ifc_rxqs[i].ifr_fl[0].ifl_sds.ifsd_cl); *ctxp = ctx; if_setgetcounterfn(ctx->ifc_ifp, iflib_if_get_counter); iflib_add_device_sysctl_post(ctx); ctx->ifc_flags |= IFC_INIT_DONE; CTX_UNLOCK(ctx); return (0); fail_detach: ether_ifdetach(ctx->ifc_ifp); fail_queues: iflib_tx_structures_free(ctx); iflib_rx_structures_free(ctx); fail_iflib_detach: IFDI_DETACH(ctx); fail_unlock: CTX_UNLOCK(ctx); iflib_deregister(ctx); fail_ctx_free: free(ctx->ifc_softc, M_IFLIB); free(ctx, M_IFLIB); return (err); } int iflib_pseudo_deregister(if_ctx_t ctx) { if_t ifp = ctx->ifc_ifp; if_shared_ctx_t sctx = ctx->ifc_sctx; iflib_txq_t txq; iflib_rxq_t rxq; int i, j; struct taskqgroup *tqg; iflib_fl_t fl; /* Unregister VLAN event handlers early */ iflib_unregister_vlan_handlers(ctx); if ((sctx->isc_flags & IFLIB_PSEUDO) && (sctx->isc_flags & IFLIB_PSEUDO_ETHER) == 0) { bpfdetach(ifp); if_detach(ifp); } else { ether_ifdetach(ifp); } /* XXX drain any dependent tasks */ tqg = qgroup_if_io_tqg; for (txq = ctx->ifc_txqs, i = 0; i < NTXQSETS(ctx); i++, txq++) { callout_drain(&txq->ift_timer); if (txq->ift_task.gt_uniq != NULL) taskqgroup_detach(tqg, &txq->ift_task); } for (i = 0, rxq = ctx->ifc_rxqs; i < NRXQSETS(ctx); i++, rxq++) { callout_drain(&rxq->ifr_watchdog); if (rxq->ifr_task.gt_uniq != NULL) taskqgroup_detach(tqg, &rxq->ifr_task); for (j = 0, fl = rxq->ifr_fl; j < rxq->ifr_nfl; j++, fl++) free(fl->ifl_rx_bitmap, M_IFLIB); } tqg = qgroup_if_config_tqg; if (ctx->ifc_admin_task.gt_uniq != NULL) taskqgroup_detach(tqg, &ctx->ifc_admin_task); if (ctx->ifc_vflr_task.gt_uniq != NULL) taskqgroup_detach(tqg, &ctx->ifc_vflr_task); iflib_tx_structures_free(ctx); iflib_rx_structures_free(ctx); iflib_deregister(ctx); if (ctx->ifc_flags & IFC_SC_ALLOCATED) free(ctx->ifc_softc, M_IFLIB); free(ctx, M_IFLIB); return (0); } int iflib_device_attach(device_t dev) { if_ctx_t ctx; if_shared_ctx_t sctx; if ((sctx = DEVICE_REGISTER(dev)) == NULL || sctx->isc_magic != IFLIB_MAGIC) return (ENOTSUP); pci_enable_busmaster(dev); return (iflib_device_register(dev, NULL, sctx, &ctx)); } int iflib_device_deregister(if_ctx_t ctx) { if_t ifp = ctx->ifc_ifp; iflib_txq_t txq; iflib_rxq_t rxq; device_t dev = ctx->ifc_dev; int i, j; struct taskqgroup *tqg; iflib_fl_t fl; /* Make sure VLANS are not using driver */ if (if_vlantrunkinuse(ifp)) { device_printf(dev, "Vlan in use, detach first\n"); return (EBUSY); } #ifdef PCI_IOV if (!CTX_IS_VF(ctx) && pci_iov_detach(dev) != 0) { device_printf(dev, "SR-IOV in use; detach first.\n"); return (EBUSY); } #endif STATE_LOCK(ctx); ctx->ifc_flags |= IFC_IN_DETACH; STATE_UNLOCK(ctx); /* Unregister VLAN handlers before calling iflib_stop() */ iflib_unregister_vlan_handlers(ctx); iflib_netmap_detach(ifp); ether_ifdetach(ifp); CTX_LOCK(ctx); iflib_stop(ctx); CTX_UNLOCK(ctx); iflib_rem_pfil(ctx); if (ctx->ifc_led_dev != NULL) led_destroy(ctx->ifc_led_dev); /* XXX drain any dependent tasks */ tqg = qgroup_if_io_tqg; for (txq = ctx->ifc_txqs, i = 0; i < NTXQSETS(ctx); i++, txq++) { callout_drain(&txq->ift_timer); if (txq->ift_task.gt_uniq != NULL) taskqgroup_detach(tqg, &txq->ift_task); } for (i = 0, rxq = ctx->ifc_rxqs; i < NRXQSETS(ctx); i++, rxq++) { if (rxq->ifr_task.gt_uniq != NULL) taskqgroup_detach(tqg, &rxq->ifr_task); for (j = 0, fl = rxq->ifr_fl; j < rxq->ifr_nfl; j++, fl++) free(fl->ifl_rx_bitmap, M_IFLIB); } tqg = qgroup_if_config_tqg; if (ctx->ifc_admin_task.gt_uniq != NULL) taskqgroup_detach(tqg, &ctx->ifc_admin_task); if (ctx->ifc_vflr_task.gt_uniq != NULL) taskqgroup_detach(tqg, &ctx->ifc_vflr_task); CTX_LOCK(ctx); IFDI_DETACH(ctx); CTX_UNLOCK(ctx); /* ether_ifdetach calls if_qflush - lock must be destroy afterwards*/ iflib_free_intr_mem(ctx); bus_generic_detach(dev); iflib_tx_structures_free(ctx); iflib_rx_structures_free(ctx); iflib_deregister(ctx); device_set_softc(ctx->ifc_dev, NULL); if (ctx->ifc_flags & IFC_SC_ALLOCATED) free(ctx->ifc_softc, M_IFLIB); unref_ctx_core_offset(ctx); free(ctx, M_IFLIB); return (0); } static void iflib_free_intr_mem(if_ctx_t ctx) { if (ctx->ifc_softc_ctx.isc_intr != IFLIB_INTR_MSIX) { iflib_irq_free(ctx, &ctx->ifc_legacy_irq); } if (ctx->ifc_softc_ctx.isc_intr != IFLIB_INTR_LEGACY) { pci_release_msi(ctx->ifc_dev); } if (ctx->ifc_msix_mem != NULL) { bus_release_resource(ctx->ifc_dev, SYS_RES_MEMORY, rman_get_rid(ctx->ifc_msix_mem), ctx->ifc_msix_mem); ctx->ifc_msix_mem = NULL; } } int iflib_device_detach(device_t dev) { if_ctx_t ctx = device_get_softc(dev); return (iflib_device_deregister(ctx)); } int iflib_device_suspend(device_t dev) { if_ctx_t ctx = device_get_softc(dev); CTX_LOCK(ctx); IFDI_SUSPEND(ctx); CTX_UNLOCK(ctx); return bus_generic_suspend(dev); } int iflib_device_shutdown(device_t dev) { if_ctx_t ctx = device_get_softc(dev); CTX_LOCK(ctx); IFDI_SHUTDOWN(ctx); CTX_UNLOCK(ctx); return bus_generic_suspend(dev); } int iflib_device_resume(device_t dev) { if_ctx_t ctx = device_get_softc(dev); iflib_txq_t txq = ctx->ifc_txqs; CTX_LOCK(ctx); IFDI_RESUME(ctx); iflib_if_init_locked(ctx); CTX_UNLOCK(ctx); for (int i = 0; i < NTXQSETS(ctx); i++, txq++) iflib_txq_check_drain(txq, IFLIB_RESTART_BUDGET); return (bus_generic_resume(dev)); } int iflib_device_iov_init(device_t dev, uint16_t num_vfs, const nvlist_t *params) { int error; if_ctx_t ctx = device_get_softc(dev); CTX_LOCK(ctx); error = IFDI_IOV_INIT(ctx, num_vfs, params); CTX_UNLOCK(ctx); return (error); } void iflib_device_iov_uninit(device_t dev) { if_ctx_t ctx = device_get_softc(dev); CTX_LOCK(ctx); IFDI_IOV_UNINIT(ctx); CTX_UNLOCK(ctx); } int iflib_device_iov_add_vf(device_t dev, uint16_t vfnum, const nvlist_t *params) { int error; if_ctx_t ctx = device_get_softc(dev); CTX_LOCK(ctx); error = IFDI_IOV_VF_ADD(ctx, vfnum, params); CTX_UNLOCK(ctx); return (error); } /********************************************************************* * * MODULE FUNCTION DEFINITIONS * **********************************************************************/ /* * - Start a fast taskqueue thread for each core * - Start a taskqueue for control operations */ static int iflib_module_init(void) { return (0); } static int iflib_module_event_handler(module_t mod, int what, void *arg) { int err; switch (what) { case MOD_LOAD: if ((err = iflib_module_init()) != 0) return (err); break; case MOD_UNLOAD: return (EBUSY); default: return (EOPNOTSUPP); } return (0); } /********************************************************************* * * PUBLIC FUNCTION DEFINITIONS * ordered as in iflib.h * **********************************************************************/ static void _iflib_assert(if_shared_ctx_t sctx) { int i; MPASS(sctx->isc_tx_maxsize); MPASS(sctx->isc_tx_maxsegsize); MPASS(sctx->isc_rx_maxsize); MPASS(sctx->isc_rx_nsegments); MPASS(sctx->isc_rx_maxsegsize); MPASS(sctx->isc_nrxqs >= 1 && sctx->isc_nrxqs <= 8); for (i = 0; i < sctx->isc_nrxqs; i++) { MPASS(sctx->isc_nrxd_min[i]); MPASS(powerof2(sctx->isc_nrxd_min[i])); MPASS(sctx->isc_nrxd_max[i]); MPASS(powerof2(sctx->isc_nrxd_max[i])); MPASS(sctx->isc_nrxd_default[i]); MPASS(powerof2(sctx->isc_nrxd_default[i])); } MPASS(sctx->isc_ntxqs >= 1 && sctx->isc_ntxqs <= 8); for (i = 0; i < sctx->isc_ntxqs; i++) { MPASS(sctx->isc_ntxd_min[i]); MPASS(powerof2(sctx->isc_ntxd_min[i])); MPASS(sctx->isc_ntxd_max[i]); MPASS(powerof2(sctx->isc_ntxd_max[i])); MPASS(sctx->isc_ntxd_default[i]); MPASS(powerof2(sctx->isc_ntxd_default[i])); } } static void _iflib_pre_assert(if_softc_ctx_t scctx) { MPASS(scctx->isc_txrx->ift_txd_encap); MPASS(scctx->isc_txrx->ift_txd_flush); MPASS(scctx->isc_txrx->ift_txd_credits_update); MPASS(scctx->isc_txrx->ift_rxd_available); MPASS(scctx->isc_txrx->ift_rxd_pkt_get); MPASS(scctx->isc_txrx->ift_rxd_refill); MPASS(scctx->isc_txrx->ift_rxd_flush); } static int iflib_register(if_ctx_t ctx) { if_shared_ctx_t sctx = ctx->ifc_sctx; driver_t *driver = sctx->isc_driver; device_t dev = ctx->ifc_dev; if_t ifp; u_char type; int iflags; if ((sctx->isc_flags & IFLIB_PSEUDO) == 0) _iflib_assert(sctx); CTX_LOCK_INIT(ctx); STATE_LOCK_INIT(ctx, device_get_nameunit(ctx->ifc_dev)); if (sctx->isc_flags & IFLIB_PSEUDO) { if (sctx->isc_flags & IFLIB_PSEUDO_ETHER) type = IFT_ETHER; else type = IFT_PPP; } else type = IFT_ETHER; ifp = ctx->ifc_ifp = if_alloc(type); if (ifp == NULL) { device_printf(dev, "can not allocate ifnet structure\n"); return (ENOMEM); } /* * Initialize our context's device specific methods */ kobj_init((kobj_t) ctx, (kobj_class_t) driver); kobj_class_compile((kobj_class_t) driver); if_initname(ifp, device_get_name(dev), device_get_unit(dev)); if_setsoftc(ifp, ctx); if_setdev(ifp, dev); if_setinitfn(ifp, iflib_if_init); if_setioctlfn(ifp, iflib_if_ioctl); #ifdef ALTQ if_setstartfn(ifp, iflib_altq_if_start); if_settransmitfn(ifp, iflib_altq_if_transmit); if_setsendqready(ifp); #else if_settransmitfn(ifp, iflib_if_transmit); #endif if_setqflushfn(ifp, iflib_if_qflush); iflags = IFF_MULTICAST | IFF_KNOWSEPOCH; if ((sctx->isc_flags & IFLIB_PSEUDO) && (sctx->isc_flags & IFLIB_PSEUDO_ETHER) == 0) iflags |= IFF_POINTOPOINT; else iflags |= IFF_BROADCAST | IFF_SIMPLEX; if_setflags(ifp, iflags); ctx->ifc_vlan_attach_event = EVENTHANDLER_REGISTER(vlan_config, iflib_vlan_register, ctx, EVENTHANDLER_PRI_FIRST); ctx->ifc_vlan_detach_event = EVENTHANDLER_REGISTER(vlan_unconfig, iflib_vlan_unregister, ctx, EVENTHANDLER_PRI_FIRST); if ((sctx->isc_flags & IFLIB_DRIVER_MEDIA) == 0) { ctx->ifc_mediap = &ctx->ifc_media; ifmedia_init(ctx->ifc_mediap, IFM_IMASK, iflib_media_change, iflib_media_status); } return (0); } static void iflib_unregister_vlan_handlers(if_ctx_t ctx) { /* Unregister VLAN events */ if (ctx->ifc_vlan_attach_event != NULL) { EVENTHANDLER_DEREGISTER(vlan_config, ctx->ifc_vlan_attach_event); ctx->ifc_vlan_attach_event = NULL; } if (ctx->ifc_vlan_detach_event != NULL) { EVENTHANDLER_DEREGISTER(vlan_unconfig, ctx->ifc_vlan_detach_event); ctx->ifc_vlan_detach_event = NULL; } } static void iflib_deregister(if_ctx_t ctx) { if_t ifp = ctx->ifc_ifp; /* Remove all media */ ifmedia_removeall(&ctx->ifc_media); /* Ensure that VLAN event handlers are unregistered */ iflib_unregister_vlan_handlers(ctx); /* Release kobject reference */ kobj_delete((kobj_t) ctx, NULL); /* Free the ifnet structure */ if_free(ifp); STATE_LOCK_DESTROY(ctx); /* ether_ifdetach calls if_qflush - lock must be destroy afterwards*/ CTX_LOCK_DESTROY(ctx); } static int iflib_queues_alloc(if_ctx_t ctx) { if_shared_ctx_t sctx = ctx->ifc_sctx; if_softc_ctx_t scctx = &ctx->ifc_softc_ctx; device_t dev = ctx->ifc_dev; int nrxqsets = scctx->isc_nrxqsets; int ntxqsets = scctx->isc_ntxqsets; iflib_txq_t txq; iflib_rxq_t rxq; iflib_fl_t fl = NULL; int i, j, cpu, err, txconf, rxconf; iflib_dma_info_t ifdip; uint32_t *rxqsizes = scctx->isc_rxqsizes; uint32_t *txqsizes = scctx->isc_txqsizes; uint8_t nrxqs = sctx->isc_nrxqs; uint8_t ntxqs = sctx->isc_ntxqs; int nfree_lists = sctx->isc_nfl ? sctx->isc_nfl : 1; caddr_t *vaddrs; uint64_t *paddrs; KASSERT(ntxqs > 0, ("number of queues per qset must be at least 1")); KASSERT(nrxqs > 0, ("number of queues per qset must be at least 1")); /* Allocate the TX ring struct memory */ if (!(ctx->ifc_txqs = (iflib_txq_t) malloc(sizeof(struct iflib_txq) * ntxqsets, M_IFLIB, M_NOWAIT | M_ZERO))) { device_printf(dev, "Unable to allocate TX ring memory\n"); err = ENOMEM; goto fail; } /* Now allocate the RX */ if (!(ctx->ifc_rxqs = (iflib_rxq_t) malloc(sizeof(struct iflib_rxq) * nrxqsets, M_IFLIB, M_NOWAIT | M_ZERO))) { device_printf(dev, "Unable to allocate RX ring memory\n"); err = ENOMEM; goto rx_fail; } txq = ctx->ifc_txqs; rxq = ctx->ifc_rxqs; /* * XXX handle allocation failure */ for (txconf = i = 0, cpu = CPU_FIRST(); i < ntxqsets; i++, txconf++, txq++, cpu = CPU_NEXT(cpu)) { /* Set up some basics */ if ((ifdip = malloc(sizeof(struct iflib_dma_info) * ntxqs, M_IFLIB, M_NOWAIT | M_ZERO)) == NULL) { device_printf(dev, "Unable to allocate TX DMA info memory\n"); err = ENOMEM; goto err_tx_desc; } txq->ift_ifdi = ifdip; for (j = 0; j < ntxqs; j++, ifdip++) { if (iflib_dma_alloc(ctx, txqsizes[j], ifdip, 0)) { device_printf(dev, "Unable to allocate TX descriptors\n"); err = ENOMEM; goto err_tx_desc; } txq->ift_txd_size[j] = scctx->isc_txd_size[j]; bzero((void *)ifdip->idi_vaddr, txqsizes[j]); } txq->ift_ctx = ctx; txq->ift_id = i; if (sctx->isc_flags & IFLIB_HAS_TXCQ) { txq->ift_br_offset = 1; } else { txq->ift_br_offset = 0; } /* XXX fix this */ txq->ift_timer.c_cpu = cpu; if (iflib_txsd_alloc(txq)) { device_printf(dev, "Critical Failure setting up TX buffers\n"); err = ENOMEM; goto err_tx_desc; } /* Initialize the TX lock */ snprintf(txq->ift_mtx_name, MTX_NAME_LEN, "%s:TX(%d):callout", device_get_nameunit(dev), txq->ift_id); mtx_init(&txq->ift_mtx, txq->ift_mtx_name, NULL, MTX_DEF); callout_init_mtx(&txq->ift_timer, &txq->ift_mtx, 0); err = ifmp_ring_alloc(&txq->ift_br, 2048, txq, iflib_txq_drain, iflib_txq_can_drain, M_IFLIB, M_WAITOK); if (err) { /* XXX free any allocated rings */ device_printf(dev, "Unable to allocate buf_ring\n"); goto err_tx_desc; } } for (rxconf = i = 0; i < nrxqsets; i++, rxconf++, rxq++) { /* Set up some basics */ callout_init(&rxq->ifr_watchdog, 1); if ((ifdip = malloc(sizeof(struct iflib_dma_info) * nrxqs, M_IFLIB, M_NOWAIT | M_ZERO)) == NULL) { device_printf(dev, "Unable to allocate RX DMA info memory\n"); err = ENOMEM; goto err_tx_desc; } rxq->ifr_ifdi = ifdip; /* XXX this needs to be changed if #rx queues != #tx queues */ rxq->ifr_ntxqirq = 1; rxq->ifr_txqid[0] = i; for (j = 0; j < nrxqs; j++, ifdip++) { if (iflib_dma_alloc(ctx, rxqsizes[j], ifdip, 0)) { device_printf(dev, "Unable to allocate RX descriptors\n"); err = ENOMEM; goto err_tx_desc; } bzero((void *)ifdip->idi_vaddr, rxqsizes[j]); } rxq->ifr_ctx = ctx; rxq->ifr_id = i; if (sctx->isc_flags & IFLIB_HAS_RXCQ) { rxq->ifr_fl_offset = 1; } else { rxq->ifr_fl_offset = 0; } rxq->ifr_nfl = nfree_lists; if (!(fl = (iflib_fl_t) malloc(sizeof(struct iflib_fl) * nfree_lists, M_IFLIB, M_NOWAIT | M_ZERO))) { device_printf(dev, "Unable to allocate free list memory\n"); err = ENOMEM; goto err_tx_desc; } rxq->ifr_fl = fl; for (j = 0; j < nfree_lists; j++) { fl[j].ifl_rxq = rxq; fl[j].ifl_id = j; fl[j].ifl_ifdi = &rxq->ifr_ifdi[j + rxq->ifr_fl_offset]; fl[j].ifl_rxd_size = scctx->isc_rxd_size[j]; } /* Allocate receive buffers for the ring */ if (iflib_rxsd_alloc(rxq)) { device_printf(dev, "Critical Failure setting up receive buffers\n"); err = ENOMEM; goto err_rx_desc; } for (j = 0, fl = rxq->ifr_fl; j < rxq->ifr_nfl; j++, fl++) fl->ifl_rx_bitmap = bit_alloc(fl->ifl_size, M_IFLIB, M_WAITOK); } /* TXQs */ vaddrs = malloc(sizeof(caddr_t)*ntxqsets*ntxqs, M_IFLIB, M_WAITOK); paddrs = malloc(sizeof(uint64_t)*ntxqsets*ntxqs, M_IFLIB, M_WAITOK); for (i = 0; i < ntxqsets; i++) { iflib_dma_info_t di = ctx->ifc_txqs[i].ift_ifdi; for (j = 0; j < ntxqs; j++, di++) { vaddrs[i*ntxqs + j] = di->idi_vaddr; paddrs[i*ntxqs + j] = di->idi_paddr; } } if ((err = IFDI_TX_QUEUES_ALLOC(ctx, vaddrs, paddrs, ntxqs, ntxqsets)) != 0) { device_printf(ctx->ifc_dev, "Unable to allocate device TX queue\n"); iflib_tx_structures_free(ctx); free(vaddrs, M_IFLIB); free(paddrs, M_IFLIB); goto err_rx_desc; } free(vaddrs, M_IFLIB); free(paddrs, M_IFLIB); /* RXQs */ vaddrs = malloc(sizeof(caddr_t)*nrxqsets*nrxqs, M_IFLIB, M_WAITOK); paddrs = malloc(sizeof(uint64_t)*nrxqsets*nrxqs, M_IFLIB, M_WAITOK); for (i = 0; i < nrxqsets; i++) { iflib_dma_info_t di = ctx->ifc_rxqs[i].ifr_ifdi; for (j = 0; j < nrxqs; j++, di++) { vaddrs[i*nrxqs + j] = di->idi_vaddr; paddrs[i*nrxqs + j] = di->idi_paddr; } } if ((err = IFDI_RX_QUEUES_ALLOC(ctx, vaddrs, paddrs, nrxqs, nrxqsets)) != 0) { device_printf(ctx->ifc_dev, "Unable to allocate device RX queue\n"); iflib_tx_structures_free(ctx); free(vaddrs, M_IFLIB); free(paddrs, M_IFLIB); goto err_rx_desc; } free(vaddrs, M_IFLIB); free(paddrs, M_IFLIB); return (0); /* XXX handle allocation failure changes */ err_rx_desc: err_tx_desc: rx_fail: if (ctx->ifc_rxqs != NULL) free(ctx->ifc_rxqs, M_IFLIB); ctx->ifc_rxqs = NULL; if (ctx->ifc_txqs != NULL) free(ctx->ifc_txqs, M_IFLIB); ctx->ifc_txqs = NULL; fail: return (err); } static int iflib_tx_structures_setup(if_ctx_t ctx) { iflib_txq_t txq = ctx->ifc_txqs; int i; for (i = 0; i < NTXQSETS(ctx); i++, txq++) iflib_txq_setup(txq); return (0); } static void iflib_tx_structures_free(if_ctx_t ctx) { iflib_txq_t txq = ctx->ifc_txqs; if_shared_ctx_t sctx = ctx->ifc_sctx; int i, j; for (i = 0; i < NTXQSETS(ctx); i++, txq++) { for (j = 0; j < sctx->isc_ntxqs; j++) iflib_dma_free(&txq->ift_ifdi[j]); iflib_txq_destroy(txq); } free(ctx->ifc_txqs, M_IFLIB); ctx->ifc_txqs = NULL; IFDI_QUEUES_FREE(ctx); } /********************************************************************* * * Initialize all receive rings. * **********************************************************************/ static int iflib_rx_structures_setup(if_ctx_t ctx) { iflib_rxq_t rxq = ctx->ifc_rxqs; int q; #if defined(INET6) || defined(INET) int err, i; #endif for (q = 0; q < ctx->ifc_softc_ctx.isc_nrxqsets; q++, rxq++) { #if defined(INET6) || defined(INET) if (if_getcapabilities(ctx->ifc_ifp) & IFCAP_LRO) { err = tcp_lro_init_args(&rxq->ifr_lc, ctx->ifc_ifp, TCP_LRO_ENTRIES, min(1024, ctx->ifc_softc_ctx.isc_nrxd[rxq->ifr_fl_offset])); if (err != 0) { device_printf(ctx->ifc_dev, "LRO Initialization failed!\n"); goto fail; } } #endif IFDI_RXQ_SETUP(ctx, rxq->ifr_id); } return (0); #if defined(INET6) || defined(INET) fail: /* * Free LRO resources allocated so far, we will only handle * the rings that completed, the failing case will have * cleaned up for itself. 'q' failed, so its the terminus. */ rxq = ctx->ifc_rxqs; for (i = 0; i < q; ++i, rxq++) { if (if_getcapabilities(ctx->ifc_ifp) & IFCAP_LRO) tcp_lro_free(&rxq->ifr_lc); } return (err); #endif } /********************************************************************* * * Free all receive rings. * **********************************************************************/ static void iflib_rx_structures_free(if_ctx_t ctx) { iflib_rxq_t rxq = ctx->ifc_rxqs; if_shared_ctx_t sctx = ctx->ifc_sctx; int i, j; for (i = 0; i < ctx->ifc_softc_ctx.isc_nrxqsets; i++, rxq++) { for (j = 0; j < sctx->isc_nrxqs; j++) iflib_dma_free(&rxq->ifr_ifdi[j]); iflib_rx_sds_free(rxq); #if defined(INET6) || defined(INET) if (if_getcapabilities(ctx->ifc_ifp) & IFCAP_LRO) tcp_lro_free(&rxq->ifr_lc); #endif } free(ctx->ifc_rxqs, M_IFLIB); ctx->ifc_rxqs = NULL; } static int iflib_qset_structures_setup(if_ctx_t ctx) { int err; /* * It is expected that the caller takes care of freeing queues if this * fails. */ if ((err = iflib_tx_structures_setup(ctx)) != 0) { device_printf(ctx->ifc_dev, "iflib_tx_structures_setup failed: %d\n", err); return (err); } if ((err = iflib_rx_structures_setup(ctx)) != 0) device_printf(ctx->ifc_dev, "iflib_rx_structures_setup failed: %d\n", err); return (err); } int iflib_irq_alloc(if_ctx_t ctx, if_irq_t irq, int rid, driver_filter_t filter, void *filter_arg, driver_intr_t handler, void *arg, const char *name) { return (_iflib_irq_alloc(ctx, irq, rid, filter, handler, arg, name)); } #ifdef SMP static int find_nth(if_ctx_t ctx, int qid) { cpuset_t cpus; int i, cpuid, eqid, count; CPU_COPY(&ctx->ifc_cpus, &cpus); count = CPU_COUNT(&cpus); eqid = qid % count; /* clear up to the qid'th bit */ for (i = 0; i < eqid; i++) { cpuid = CPU_FFS(&cpus); MPASS(cpuid != 0); CPU_CLR(cpuid-1, &cpus); } cpuid = CPU_FFS(&cpus); MPASS(cpuid != 0); return (cpuid-1); } #ifdef SCHED_ULE extern struct cpu_group *cpu_top; /* CPU topology */ static int find_child_with_core(int cpu, struct cpu_group *grp) { int i; if (grp->cg_children == 0) return -1; MPASS(grp->cg_child); for (i = 0; i < grp->cg_children; i++) { if (CPU_ISSET(cpu, &grp->cg_child[i].cg_mask)) return i; } return -1; } /* * Find the nth "close" core to the specified core * "close" is defined as the deepest level that shares * at least an L2 cache. With threads, this will be * threads on the same core. If the shared cache is L3 * or higher, simply returns the same core. */ static int find_close_core(int cpu, int core_offset) { struct cpu_group *grp; int i; int fcpu; cpuset_t cs; grp = cpu_top; if (grp == NULL) return cpu; i = 0; while ((i = find_child_with_core(cpu, grp)) != -1) { /* If the child only has one cpu, don't descend */ if (grp->cg_child[i].cg_count <= 1) break; grp = &grp->cg_child[i]; } /* If they don't share at least an L2 cache, use the same CPU */ if (grp->cg_level > CG_SHARE_L2 || grp->cg_level == CG_SHARE_NONE) return cpu; /* Now pick one */ CPU_COPY(&grp->cg_mask, &cs); /* Add the selected CPU offset to core offset. */ for (i = 0; (fcpu = CPU_FFS(&cs)) != 0; i++) { if (fcpu - 1 == cpu) break; CPU_CLR(fcpu - 1, &cs); } MPASS(fcpu); core_offset += i; CPU_COPY(&grp->cg_mask, &cs); for (i = core_offset % grp->cg_count; i > 0; i--) { MPASS(CPU_FFS(&cs)); CPU_CLR(CPU_FFS(&cs) - 1, &cs); } MPASS(CPU_FFS(&cs)); return CPU_FFS(&cs) - 1; } #else static int find_close_core(int cpu, int core_offset __unused) { return cpu; } #endif static int get_core_offset(if_ctx_t ctx, iflib_intr_type_t type, int qid) { switch (type) { case IFLIB_INTR_TX: /* TX queues get cores which share at least an L2 cache with the corresponding RX queue */ /* XXX handle multiple RX threads per core and more than two core per L2 group */ return qid / CPU_COUNT(&ctx->ifc_cpus) + 1; case IFLIB_INTR_RX: case IFLIB_INTR_RXTX: /* RX queues get the specified core */ return qid / CPU_COUNT(&ctx->ifc_cpus); default: return -1; } } #else #define get_core_offset(ctx, type, qid) CPU_FIRST() #define find_close_core(cpuid, tid) CPU_FIRST() #define find_nth(ctx, gid) CPU_FIRST() #endif /* Just to avoid copy/paste */ static inline int iflib_irq_set_affinity(if_ctx_t ctx, if_irq_t irq, iflib_intr_type_t type, int qid, struct grouptask *gtask, struct taskqgroup *tqg, void *uniq, const char *name) { device_t dev; int co, cpuid, err, tid; dev = ctx->ifc_dev; co = ctx->ifc_sysctl_core_offset; if (ctx->ifc_sysctl_separate_txrx && type == IFLIB_INTR_TX) co += ctx->ifc_softc_ctx.isc_nrxqsets; cpuid = find_nth(ctx, qid + co); tid = get_core_offset(ctx, type, qid); if (tid < 0) { device_printf(dev, "get_core_offset failed\n"); return (EOPNOTSUPP); } cpuid = find_close_core(cpuid, tid); err = taskqgroup_attach_cpu(tqg, gtask, uniq, cpuid, dev, irq->ii_res, name); if (err) { device_printf(dev, "taskqgroup_attach_cpu failed %d\n", err); return (err); } #ifdef notyet if (cpuid > ctx->ifc_cpuid_highest) ctx->ifc_cpuid_highest = cpuid; #endif return (0); } int iflib_irq_alloc_generic(if_ctx_t ctx, if_irq_t irq, int rid, iflib_intr_type_t type, driver_filter_t *filter, void *filter_arg, int qid, const char *name) { device_t dev; struct grouptask *gtask; struct taskqgroup *tqg; iflib_filter_info_t info; gtask_fn_t *fn; int tqrid, err; driver_filter_t *intr_fast; void *q; info = &ctx->ifc_filter_info; tqrid = rid; switch (type) { /* XXX merge tx/rx for netmap? */ case IFLIB_INTR_TX: q = &ctx->ifc_txqs[qid]; info = &ctx->ifc_txqs[qid].ift_filter_info; gtask = &ctx->ifc_txqs[qid].ift_task; tqg = qgroup_if_io_tqg; fn = _task_fn_tx; intr_fast = iflib_fast_intr; GROUPTASK_INIT(gtask, 0, fn, q); ctx->ifc_flags |= IFC_NETMAP_TX_IRQ; break; case IFLIB_INTR_RX: q = &ctx->ifc_rxqs[qid]; info = &ctx->ifc_rxqs[qid].ifr_filter_info; gtask = &ctx->ifc_rxqs[qid].ifr_task; tqg = qgroup_if_io_tqg; fn = _task_fn_rx; intr_fast = iflib_fast_intr; NET_GROUPTASK_INIT(gtask, 0, fn, q); break; case IFLIB_INTR_RXTX: q = &ctx->ifc_rxqs[qid]; info = &ctx->ifc_rxqs[qid].ifr_filter_info; gtask = &ctx->ifc_rxqs[qid].ifr_task; tqg = qgroup_if_io_tqg; fn = _task_fn_rx; intr_fast = iflib_fast_intr_rxtx; NET_GROUPTASK_INIT(gtask, 0, fn, q); break; case IFLIB_INTR_ADMIN: q = ctx; tqrid = -1; info = &ctx->ifc_filter_info; gtask = &ctx->ifc_admin_task; tqg = qgroup_if_config_tqg; fn = _task_fn_admin; intr_fast = iflib_fast_intr_ctx; break; default: device_printf(ctx->ifc_dev, "%s: unknown net intr type\n", __func__); return (EINVAL); } info->ifi_filter = filter; info->ifi_filter_arg = filter_arg; info->ifi_task = gtask; info->ifi_ctx = q; dev = ctx->ifc_dev; err = _iflib_irq_alloc(ctx, irq, rid, intr_fast, NULL, info, name); if (err != 0) { device_printf(dev, "_iflib_irq_alloc failed %d\n", err); return (err); } if (type == IFLIB_INTR_ADMIN) return (0); if (tqrid != -1) { err = iflib_irq_set_affinity(ctx, irq, type, qid, gtask, tqg, q, name); if (err) return (err); } else { taskqgroup_attach(tqg, gtask, q, dev, irq->ii_res, name); } return (0); } void iflib_softirq_alloc_generic(if_ctx_t ctx, if_irq_t irq, iflib_intr_type_t type, void *arg, int qid, const char *name) { struct grouptask *gtask; struct taskqgroup *tqg; gtask_fn_t *fn; void *q; int err; switch (type) { case IFLIB_INTR_TX: q = &ctx->ifc_txqs[qid]; gtask = &ctx->ifc_txqs[qid].ift_task; tqg = qgroup_if_io_tqg; fn = _task_fn_tx; GROUPTASK_INIT(gtask, 0, fn, q); break; case IFLIB_INTR_RX: q = &ctx->ifc_rxqs[qid]; gtask = &ctx->ifc_rxqs[qid].ifr_task; tqg = qgroup_if_io_tqg; fn = _task_fn_rx; NET_GROUPTASK_INIT(gtask, 0, fn, q); break; case IFLIB_INTR_IOV: q = ctx; gtask = &ctx->ifc_vflr_task; tqg = qgroup_if_config_tqg; fn = _task_fn_iov; GROUPTASK_INIT(gtask, 0, fn, q); break; default: panic("unknown net intr type"); } if (irq != NULL) { err = iflib_irq_set_affinity(ctx, irq, type, qid, gtask, tqg, q, name); if (err) taskqgroup_attach(tqg, gtask, q, ctx->ifc_dev, irq->ii_res, name); } else { taskqgroup_attach(tqg, gtask, q, NULL, NULL, name); } } void iflib_irq_free(if_ctx_t ctx, if_irq_t irq) { if (irq->ii_tag) bus_teardown_intr(ctx->ifc_dev, irq->ii_res, irq->ii_tag); if (irq->ii_res) bus_release_resource(ctx->ifc_dev, SYS_RES_IRQ, rman_get_rid(irq->ii_res), irq->ii_res); } static int iflib_legacy_setup(if_ctx_t ctx, driver_filter_t filter, void *filter_arg, int *rid, const char *name) { iflib_txq_t txq = ctx->ifc_txqs; iflib_rxq_t rxq = ctx->ifc_rxqs; if_irq_t irq = &ctx->ifc_legacy_irq; iflib_filter_info_t info; device_t dev; struct grouptask *gtask; struct resource *res; struct taskqgroup *tqg; void *q; int err, tqrid; bool rx_only; q = &ctx->ifc_rxqs[0]; info = &rxq[0].ifr_filter_info; gtask = &rxq[0].ifr_task; tqg = qgroup_if_io_tqg; tqrid = *rid; rx_only = (ctx->ifc_sctx->isc_flags & IFLIB_SINGLE_IRQ_RX_ONLY) != 0; ctx->ifc_flags |= IFC_LEGACY; info->ifi_filter = filter; info->ifi_filter_arg = filter_arg; info->ifi_task = gtask; info->ifi_ctx = rx_only ? ctx : q; dev = ctx->ifc_dev; /* We allocate a single interrupt resource */ err = _iflib_irq_alloc(ctx, irq, tqrid, rx_only ? iflib_fast_intr_ctx : iflib_fast_intr_rxtx, NULL, info, name); if (err != 0) return (err); NET_GROUPTASK_INIT(gtask, 0, _task_fn_rx, q); res = irq->ii_res; taskqgroup_attach(tqg, gtask, q, dev, res, name); GROUPTASK_INIT(&txq->ift_task, 0, _task_fn_tx, txq); taskqgroup_attach(qgroup_if_io_tqg, &txq->ift_task, txq, dev, res, "tx"); return (0); } void iflib_led_create(if_ctx_t ctx) { ctx->ifc_led_dev = led_create(iflib_led_func, ctx, device_get_nameunit(ctx->ifc_dev)); } void iflib_tx_intr_deferred(if_ctx_t ctx, int txqid) { GROUPTASK_ENQUEUE(&ctx->ifc_txqs[txqid].ift_task); } void iflib_rx_intr_deferred(if_ctx_t ctx, int rxqid) { GROUPTASK_ENQUEUE(&ctx->ifc_rxqs[rxqid].ifr_task); } void iflib_admin_intr_deferred(if_ctx_t ctx) { MPASS(ctx->ifc_admin_task.gt_taskqueue != NULL); GROUPTASK_ENQUEUE(&ctx->ifc_admin_task); } void iflib_iov_intr_deferred(if_ctx_t ctx) { GROUPTASK_ENQUEUE(&ctx->ifc_vflr_task); } void iflib_io_tqg_attach(struct grouptask *gt, void *uniq, int cpu, const char *name) { taskqgroup_attach_cpu(qgroup_if_io_tqg, gt, uniq, cpu, NULL, NULL, name); } void iflib_config_gtask_init(void *ctx, struct grouptask *gtask, gtask_fn_t *fn, const char *name) { GROUPTASK_INIT(gtask, 0, fn, ctx); taskqgroup_attach(qgroup_if_config_tqg, gtask, gtask, NULL, NULL, name); } void iflib_config_gtask_deinit(struct grouptask *gtask) { taskqgroup_detach(qgroup_if_config_tqg, gtask); } void iflib_link_state_change(if_ctx_t ctx, int link_state, uint64_t baudrate) { if_t ifp = ctx->ifc_ifp; iflib_txq_t txq = ctx->ifc_txqs; if_setbaudrate(ifp, baudrate); if (baudrate >= IF_Gbps(10)) { STATE_LOCK(ctx); ctx->ifc_flags |= IFC_PREFETCH; STATE_UNLOCK(ctx); } /* If link down, disable watchdog */ if ((ctx->ifc_link_state == LINK_STATE_UP) && (link_state == LINK_STATE_DOWN)) { for (int i = 0; i < ctx->ifc_softc_ctx.isc_ntxqsets; i++, txq++) txq->ift_qstatus = IFLIB_QUEUE_IDLE; } ctx->ifc_link_state = link_state; if_link_state_change(ifp, link_state); } static int iflib_tx_credits_update(if_ctx_t ctx, iflib_txq_t txq) { int credits; #ifdef INVARIANTS int credits_pre = txq->ift_cidx_processed; #endif bus_dmamap_sync(txq->ift_ifdi->idi_tag, txq->ift_ifdi->idi_map, BUS_DMASYNC_POSTREAD); if ((credits = ctx->isc_txd_credits_update(ctx->ifc_softc, txq->ift_id, true)) == 0) return (0); txq->ift_processed += credits; txq->ift_cidx_processed += credits; MPASS(credits_pre + credits == txq->ift_cidx_processed); if (txq->ift_cidx_processed >= txq->ift_size) txq->ift_cidx_processed -= txq->ift_size; return (credits); } static int iflib_rxd_avail(if_ctx_t ctx, iflib_rxq_t rxq, qidx_t cidx, qidx_t budget) { iflib_fl_t fl; u_int i; for (i = 0, fl = &rxq->ifr_fl[0]; i < rxq->ifr_nfl; i++, fl++) bus_dmamap_sync(fl->ifl_ifdi->idi_tag, fl->ifl_ifdi->idi_map, BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); return (ctx->isc_rxd_available(ctx->ifc_softc, rxq->ifr_id, cidx, budget)); } void iflib_add_int_delay_sysctl(if_ctx_t ctx, const char *name, const char *description, if_int_delay_info_t info, int offset, int value) { info->iidi_ctx = ctx; info->iidi_offset = offset; info->iidi_value = value; SYSCTL_ADD_PROC(device_get_sysctl_ctx(ctx->ifc_dev), SYSCTL_CHILDREN(device_get_sysctl_tree(ctx->ifc_dev)), OID_AUTO, name, CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE, info, 0, iflib_sysctl_int_delay, "I", description); } struct sx * iflib_ctx_lock_get(if_ctx_t ctx) { return (&ctx->ifc_ctx_sx); } static int iflib_msix_init(if_ctx_t ctx) { device_t dev = ctx->ifc_dev; if_shared_ctx_t sctx = ctx->ifc_sctx; if_softc_ctx_t scctx = &ctx->ifc_softc_ctx; int admincnt, bar, err, iflib_num_rx_queues, iflib_num_tx_queues; int msgs, queuemsgs, queues, rx_queues, tx_queues, vectors; iflib_num_tx_queues = ctx->ifc_sysctl_ntxqs; iflib_num_rx_queues = ctx->ifc_sysctl_nrxqs; if (bootverbose) device_printf(dev, "msix_init qsets capped at %d\n", imax(scctx->isc_ntxqsets, scctx->isc_nrxqsets)); /* Override by tuneable */ if (scctx->isc_disable_msix) goto msi; /* First try MSI-X */ if ((msgs = pci_msix_count(dev)) == 0) { if (bootverbose) device_printf(dev, "MSI-X not supported or disabled\n"); goto msi; } bar = ctx->ifc_softc_ctx.isc_msix_bar; /* * bar == -1 => "trust me I know what I'm doing" * Some drivers are for hardware that is so shoddily * documented that no one knows which bars are which * so the developer has to map all bars. This hack * allows shoddy garbage to use MSI-X in this framework. */ if (bar != -1) { ctx->ifc_msix_mem = bus_alloc_resource_any(dev, SYS_RES_MEMORY, &bar, RF_ACTIVE); if (ctx->ifc_msix_mem == NULL) { device_printf(dev, "Unable to map MSI-X table\n"); goto msi; } } admincnt = sctx->isc_admin_intrcnt; #if IFLIB_DEBUG /* use only 1 qset in debug mode */ queuemsgs = min(msgs - admincnt, 1); #else queuemsgs = msgs - admincnt; #endif #ifdef RSS queues = imin(queuemsgs, rss_getnumbuckets()); #else queues = queuemsgs; #endif queues = imin(CPU_COUNT(&ctx->ifc_cpus), queues); if (bootverbose) device_printf(dev, "intr CPUs: %d queue msgs: %d admincnt: %d\n", CPU_COUNT(&ctx->ifc_cpus), queuemsgs, admincnt); #ifdef RSS /* If we're doing RSS, clamp at the number of RSS buckets */ if (queues > rss_getnumbuckets()) queues = rss_getnumbuckets(); #endif if (iflib_num_rx_queues > 0 && iflib_num_rx_queues < queuemsgs - admincnt) rx_queues = iflib_num_rx_queues; else rx_queues = queues; if (rx_queues > scctx->isc_nrxqsets) rx_queues = scctx->isc_nrxqsets; /* * We want this to be all logical CPUs by default */ if (iflib_num_tx_queues > 0 && iflib_num_tx_queues < queues) tx_queues = iflib_num_tx_queues; else tx_queues = mp_ncpus; if (tx_queues > scctx->isc_ntxqsets) tx_queues = scctx->isc_ntxqsets; if (ctx->ifc_sysctl_qs_eq_override == 0) { #ifdef INVARIANTS if (tx_queues != rx_queues) device_printf(dev, "queue equality override not set, capping rx_queues at %d and tx_queues at %d\n", min(rx_queues, tx_queues), min(rx_queues, tx_queues)); #endif tx_queues = min(rx_queues, tx_queues); rx_queues = min(rx_queues, tx_queues); } vectors = rx_queues + admincnt; if (msgs < vectors) { device_printf(dev, "insufficient number of MSI-X vectors " "(supported %d, need %d)\n", msgs, vectors); goto msi; } device_printf(dev, "Using %d RX queues %d TX queues\n", rx_queues, tx_queues); msgs = vectors; if ((err = pci_alloc_msix(dev, &vectors)) == 0) { if (vectors != msgs) { device_printf(dev, "Unable to allocate sufficient MSI-X vectors " "(got %d, need %d)\n", vectors, msgs); pci_release_msi(dev); if (bar != -1) { bus_release_resource(dev, SYS_RES_MEMORY, bar, ctx->ifc_msix_mem); ctx->ifc_msix_mem = NULL; } goto msi; } device_printf(dev, "Using MSI-X interrupts with %d vectors\n", vectors); scctx->isc_vectors = vectors; scctx->isc_nrxqsets = rx_queues; scctx->isc_ntxqsets = tx_queues; scctx->isc_intr = IFLIB_INTR_MSIX; return (vectors); } else { device_printf(dev, "failed to allocate %d MSI-X vectors, err: %d\n", vectors, err); if (bar != -1) { bus_release_resource(dev, SYS_RES_MEMORY, bar, ctx->ifc_msix_mem); ctx->ifc_msix_mem = NULL; } } msi: vectors = pci_msi_count(dev); scctx->isc_nrxqsets = 1; scctx->isc_ntxqsets = 1; scctx->isc_vectors = vectors; if (vectors == 1 && pci_alloc_msi(dev, &vectors) == 0) { device_printf(dev,"Using an MSI interrupt\n"); scctx->isc_intr = IFLIB_INTR_MSI; } else { scctx->isc_vectors = 1; device_printf(dev,"Using a Legacy interrupt\n"); scctx->isc_intr = IFLIB_INTR_LEGACY; } return (vectors); } static const char *ring_states[] = { "IDLE", "BUSY", "STALLED", "ABDICATED" }; static int mp_ring_state_handler(SYSCTL_HANDLER_ARGS) { int rc; uint16_t *state = ((uint16_t *)oidp->oid_arg1); struct sbuf *sb; const char *ring_state = "UNKNOWN"; /* XXX needed ? */ rc = sysctl_wire_old_buffer(req, 0); MPASS(rc == 0); if (rc != 0) return (rc); sb = sbuf_new_for_sysctl(NULL, NULL, 80, req); MPASS(sb != NULL); if (sb == NULL) return (ENOMEM); if (state[3] <= 3) ring_state = ring_states[state[3]]; sbuf_printf(sb, "pidx_head: %04hd pidx_tail: %04hd cidx: %04hd state: %s", state[0], state[1], state[2], ring_state); rc = sbuf_finish(sb); sbuf_delete(sb); return(rc); } enum iflib_ndesc_handler { IFLIB_NTXD_HANDLER, IFLIB_NRXD_HANDLER, }; static int mp_ndesc_handler(SYSCTL_HANDLER_ARGS) { if_ctx_t ctx = (void *)arg1; enum iflib_ndesc_handler type = arg2; char buf[256] = {0}; qidx_t *ndesc; char *p, *next; int nqs, rc, i; nqs = 8; switch(type) { case IFLIB_NTXD_HANDLER: ndesc = ctx->ifc_sysctl_ntxds; if (ctx->ifc_sctx) nqs = ctx->ifc_sctx->isc_ntxqs; break; case IFLIB_NRXD_HANDLER: ndesc = ctx->ifc_sysctl_nrxds; if (ctx->ifc_sctx) nqs = ctx->ifc_sctx->isc_nrxqs; break; default: printf("%s: unhandled type\n", __func__); return (EINVAL); } if (nqs == 0) nqs = 8; for (i=0; i<8; i++) { if (i >= nqs) break; if (i) strcat(buf, ","); sprintf(strchr(buf, 0), "%d", ndesc[i]); } rc = sysctl_handle_string(oidp, buf, sizeof(buf), req); if (rc || req->newptr == NULL) return rc; for (i = 0, next = buf, p = strsep(&next, " ,"); i < 8 && p; i++, p = strsep(&next, " ,")) { ndesc[i] = strtoul(p, NULL, 10); } return(rc); } #define NAME_BUFLEN 32 static void iflib_add_device_sysctl_pre(if_ctx_t ctx) { device_t dev = iflib_get_dev(ctx); struct sysctl_oid_list *child, *oid_list; struct sysctl_ctx_list *ctx_list; struct sysctl_oid *node; ctx_list = device_get_sysctl_ctx(dev); child = SYSCTL_CHILDREN(device_get_sysctl_tree(dev)); ctx->ifc_sysctl_node = node = SYSCTL_ADD_NODE(ctx_list, child, OID_AUTO, "iflib", CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, "IFLIB fields"); oid_list = SYSCTL_CHILDREN(node); SYSCTL_ADD_CONST_STRING(ctx_list, oid_list, OID_AUTO, "driver_version", CTLFLAG_RD, ctx->ifc_sctx->isc_driver_version, "driver version"); SYSCTL_ADD_U16(ctx_list, oid_list, OID_AUTO, "override_ntxqs", CTLFLAG_RWTUN, &ctx->ifc_sysctl_ntxqs, 0, "# of txqs to use, 0 => use default #"); SYSCTL_ADD_U16(ctx_list, oid_list, OID_AUTO, "override_nrxqs", CTLFLAG_RWTUN, &ctx->ifc_sysctl_nrxqs, 0, "# of rxqs to use, 0 => use default #"); SYSCTL_ADD_U16(ctx_list, oid_list, OID_AUTO, "override_qs_enable", CTLFLAG_RWTUN, &ctx->ifc_sysctl_qs_eq_override, 0, "permit #txq != #rxq"); SYSCTL_ADD_INT(ctx_list, oid_list, OID_AUTO, "disable_msix", CTLFLAG_RWTUN, &ctx->ifc_softc_ctx.isc_disable_msix, 0, "disable MSI-X (default 0)"); SYSCTL_ADD_U16(ctx_list, oid_list, OID_AUTO, "rx_budget", CTLFLAG_RWTUN, &ctx->ifc_sysctl_rx_budget, 0, "set the RX budget"); SYSCTL_ADD_U16(ctx_list, oid_list, OID_AUTO, "tx_abdicate", CTLFLAG_RWTUN, &ctx->ifc_sysctl_tx_abdicate, 0, "cause TX to abdicate instead of running to completion"); ctx->ifc_sysctl_core_offset = CORE_OFFSET_UNSPECIFIED; SYSCTL_ADD_U16(ctx_list, oid_list, OID_AUTO, "core_offset", CTLFLAG_RDTUN, &ctx->ifc_sysctl_core_offset, 0, "offset to start using cores at"); SYSCTL_ADD_U8(ctx_list, oid_list, OID_AUTO, "separate_txrx", CTLFLAG_RDTUN, &ctx->ifc_sysctl_separate_txrx, 0, "use separate cores for TX and RX"); /* XXX change for per-queue sizes */ SYSCTL_ADD_PROC(ctx_list, oid_list, OID_AUTO, "override_ntxds", CTLTYPE_STRING | CTLFLAG_RWTUN | CTLFLAG_NEEDGIANT, ctx, IFLIB_NTXD_HANDLER, mp_ndesc_handler, "A", "list of # of TX descriptors to use, 0 = use default #"); SYSCTL_ADD_PROC(ctx_list, oid_list, OID_AUTO, "override_nrxds", CTLTYPE_STRING | CTLFLAG_RWTUN | CTLFLAG_NEEDGIANT, ctx, IFLIB_NRXD_HANDLER, mp_ndesc_handler, "A", "list of # of RX descriptors to use, 0 = use default #"); } static void iflib_add_device_sysctl_post(if_ctx_t ctx) { if_shared_ctx_t sctx = ctx->ifc_sctx; if_softc_ctx_t scctx = &ctx->ifc_softc_ctx; device_t dev = iflib_get_dev(ctx); struct sysctl_oid_list *child; struct sysctl_ctx_list *ctx_list; iflib_fl_t fl; iflib_txq_t txq; iflib_rxq_t rxq; int i, j; char namebuf[NAME_BUFLEN]; char *qfmt; struct sysctl_oid *queue_node, *fl_node, *node; struct sysctl_oid_list *queue_list, *fl_list; ctx_list = device_get_sysctl_ctx(dev); node = ctx->ifc_sysctl_node; child = SYSCTL_CHILDREN(node); if (scctx->isc_ntxqsets > 100) qfmt = "txq%03d"; else if (scctx->isc_ntxqsets > 10) qfmt = "txq%02d"; else qfmt = "txq%d"; for (i = 0, txq = ctx->ifc_txqs; i < scctx->isc_ntxqsets; i++, txq++) { snprintf(namebuf, NAME_BUFLEN, qfmt, i); queue_node = SYSCTL_ADD_NODE(ctx_list, child, OID_AUTO, namebuf, CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, "Queue Name"); queue_list = SYSCTL_CHILDREN(queue_node); #if MEMORY_LOGGING SYSCTL_ADD_QUAD(ctx_list, queue_list, OID_AUTO, "txq_dequeued", CTLFLAG_RD, &txq->ift_dequeued, "total mbufs freed"); SYSCTL_ADD_QUAD(ctx_list, queue_list, OID_AUTO, "txq_enqueued", CTLFLAG_RD, &txq->ift_enqueued, "total mbufs enqueued"); #endif SYSCTL_ADD_QUAD(ctx_list, queue_list, OID_AUTO, "mbuf_defrag", CTLFLAG_RD, &txq->ift_mbuf_defrag, "# of times m_defrag was called"); SYSCTL_ADD_QUAD(ctx_list, queue_list, OID_AUTO, "m_pullups", CTLFLAG_RD, &txq->ift_pullups, "# of times m_pullup was called"); SYSCTL_ADD_QUAD(ctx_list, queue_list, OID_AUTO, "mbuf_defrag_failed", CTLFLAG_RD, &txq->ift_mbuf_defrag_failed, "# of times m_defrag failed"); SYSCTL_ADD_QUAD(ctx_list, queue_list, OID_AUTO, "no_desc_avail", CTLFLAG_RD, &txq->ift_no_desc_avail, "# of times no descriptors were available"); SYSCTL_ADD_QUAD(ctx_list, queue_list, OID_AUTO, "tx_map_failed", CTLFLAG_RD, &txq->ift_map_failed, "# of times DMA map failed"); SYSCTL_ADD_QUAD(ctx_list, queue_list, OID_AUTO, "txd_encap_efbig", CTLFLAG_RD, &txq->ift_txd_encap_efbig, "# of times txd_encap returned EFBIG"); SYSCTL_ADD_QUAD(ctx_list, queue_list, OID_AUTO, "no_tx_dma_setup", CTLFLAG_RD, &txq->ift_no_tx_dma_setup, "# of times map failed for other than EFBIG"); SYSCTL_ADD_U16(ctx_list, queue_list, OID_AUTO, "txq_pidx", CTLFLAG_RD, &txq->ift_pidx, 1, "Producer Index"); SYSCTL_ADD_U16(ctx_list, queue_list, OID_AUTO, "txq_cidx", CTLFLAG_RD, &txq->ift_cidx, 1, "Consumer Index"); SYSCTL_ADD_U16(ctx_list, queue_list, OID_AUTO, "txq_cidx_processed", CTLFLAG_RD, &txq->ift_cidx_processed, 1, "Consumer Index seen by credit update"); SYSCTL_ADD_U16(ctx_list, queue_list, OID_AUTO, "txq_in_use", CTLFLAG_RD, &txq->ift_in_use, 1, "descriptors in use"); SYSCTL_ADD_QUAD(ctx_list, queue_list, OID_AUTO, "txq_processed", CTLFLAG_RD, &txq->ift_processed, "descriptors procesed for clean"); SYSCTL_ADD_QUAD(ctx_list, queue_list, OID_AUTO, "txq_cleaned", CTLFLAG_RD, &txq->ift_cleaned, "total cleaned"); SYSCTL_ADD_PROC(ctx_list, queue_list, OID_AUTO, "ring_state", CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_NEEDGIANT, __DEVOLATILE(uint64_t *, &txq->ift_br->state), 0, mp_ring_state_handler, "A", "soft ring state"); SYSCTL_ADD_COUNTER_U64(ctx_list, queue_list, OID_AUTO, "r_enqueues", CTLFLAG_RD, &txq->ift_br->enqueues, "# of enqueues to the mp_ring for this queue"); SYSCTL_ADD_COUNTER_U64(ctx_list, queue_list, OID_AUTO, "r_drops", CTLFLAG_RD, &txq->ift_br->drops, "# of drops in the mp_ring for this queue"); SYSCTL_ADD_COUNTER_U64(ctx_list, queue_list, OID_AUTO, "r_starts", CTLFLAG_RD, &txq->ift_br->starts, "# of normal consumer starts in the mp_ring for this queue"); SYSCTL_ADD_COUNTER_U64(ctx_list, queue_list, OID_AUTO, "r_stalls", CTLFLAG_RD, &txq->ift_br->stalls, "# of consumer stalls in the mp_ring for this queue"); SYSCTL_ADD_COUNTER_U64(ctx_list, queue_list, OID_AUTO, "r_restarts", CTLFLAG_RD, &txq->ift_br->restarts, "# of consumer restarts in the mp_ring for this queue"); SYSCTL_ADD_COUNTER_U64(ctx_list, queue_list, OID_AUTO, "r_abdications", CTLFLAG_RD, &txq->ift_br->abdications, "# of consumer abdications in the mp_ring for this queue"); } if (scctx->isc_nrxqsets > 100) qfmt = "rxq%03d"; else if (scctx->isc_nrxqsets > 10) qfmt = "rxq%02d"; else qfmt = "rxq%d"; for (i = 0, rxq = ctx->ifc_rxqs; i < scctx->isc_nrxqsets; i++, rxq++) { snprintf(namebuf, NAME_BUFLEN, qfmt, i); queue_node = SYSCTL_ADD_NODE(ctx_list, child, OID_AUTO, namebuf, CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, "Queue Name"); queue_list = SYSCTL_CHILDREN(queue_node); if (sctx->isc_flags & IFLIB_HAS_RXCQ) { SYSCTL_ADD_U16(ctx_list, queue_list, OID_AUTO, "rxq_cq_cidx", CTLFLAG_RD, &rxq->ifr_cq_cidx, 1, "Consumer Index"); } for (j = 0, fl = rxq->ifr_fl; j < rxq->ifr_nfl; j++, fl++) { snprintf(namebuf, NAME_BUFLEN, "rxq_fl%d", j); fl_node = SYSCTL_ADD_NODE(ctx_list, queue_list, OID_AUTO, namebuf, CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, "freelist Name"); fl_list = SYSCTL_CHILDREN(fl_node); SYSCTL_ADD_U16(ctx_list, fl_list, OID_AUTO, "pidx", CTLFLAG_RD, &fl->ifl_pidx, 1, "Producer Index"); SYSCTL_ADD_U16(ctx_list, fl_list, OID_AUTO, "cidx", CTLFLAG_RD, &fl->ifl_cidx, 1, "Consumer Index"); SYSCTL_ADD_U16(ctx_list, fl_list, OID_AUTO, "credits", CTLFLAG_RD, &fl->ifl_credits, 1, "credits available"); SYSCTL_ADD_U16(ctx_list, fl_list, OID_AUTO, "buf_size", CTLFLAG_RD, &fl->ifl_buf_size, 1, "buffer size"); #if MEMORY_LOGGING SYSCTL_ADD_QUAD(ctx_list, fl_list, OID_AUTO, "fl_m_enqueued", CTLFLAG_RD, &fl->ifl_m_enqueued, "mbufs allocated"); SYSCTL_ADD_QUAD(ctx_list, fl_list, OID_AUTO, "fl_m_dequeued", CTLFLAG_RD, &fl->ifl_m_dequeued, "mbufs freed"); SYSCTL_ADD_QUAD(ctx_list, fl_list, OID_AUTO, "fl_cl_enqueued", CTLFLAG_RD, &fl->ifl_cl_enqueued, "clusters allocated"); SYSCTL_ADD_QUAD(ctx_list, fl_list, OID_AUTO, "fl_cl_dequeued", CTLFLAG_RD, &fl->ifl_cl_dequeued, "clusters freed"); #endif } } } void iflib_request_reset(if_ctx_t ctx) { STATE_LOCK(ctx); ctx->ifc_flags |= IFC_DO_RESET; STATE_UNLOCK(ctx); } #ifndef __NO_STRICT_ALIGNMENT static struct mbuf * iflib_fixup_rx(struct mbuf *m) { struct mbuf *n; if (m->m_len <= (MCLBYTES - ETHER_HDR_LEN)) { bcopy(m->m_data, m->m_data + ETHER_HDR_LEN, m->m_len); m->m_data += ETHER_HDR_LEN; n = m; } else { MGETHDR(n, M_NOWAIT, MT_DATA); if (n == NULL) { m_freem(m); return (NULL); } bcopy(m->m_data, n->m_data, ETHER_HDR_LEN); m->m_data += ETHER_HDR_LEN; m->m_len -= ETHER_HDR_LEN; n->m_len = ETHER_HDR_LEN; M_MOVE_PKTHDR(n, m); n->m_next = m; } return (n); } #endif #ifdef DEBUGNET static void iflib_debugnet_init(if_t ifp, int *nrxr, int *ncl, int *clsize) { if_ctx_t ctx; ctx = if_getsoftc(ifp); CTX_LOCK(ctx); *nrxr = NRXQSETS(ctx); *ncl = ctx->ifc_rxqs[0].ifr_fl->ifl_size; *clsize = ctx->ifc_rxqs[0].ifr_fl->ifl_buf_size; CTX_UNLOCK(ctx); } static void iflib_debugnet_event(if_t ifp, enum debugnet_ev event) { if_ctx_t ctx; if_softc_ctx_t scctx; iflib_fl_t fl; iflib_rxq_t rxq; int i, j; ctx = if_getsoftc(ifp); scctx = &ctx->ifc_softc_ctx; switch (event) { case DEBUGNET_START: for (i = 0; i < scctx->isc_nrxqsets; i++) { rxq = &ctx->ifc_rxqs[i]; for (j = 0; j < rxq->ifr_nfl; j++) { fl = rxq->ifr_fl; fl->ifl_zone = m_getzone(fl->ifl_buf_size); } } iflib_no_tx_batch = 1; break; default: break; } } static int iflib_debugnet_transmit(if_t ifp, struct mbuf *m) { if_ctx_t ctx; iflib_txq_t txq; int error; ctx = if_getsoftc(ifp); if ((if_getdrvflags(ifp) & (IFF_DRV_RUNNING | IFF_DRV_OACTIVE)) != IFF_DRV_RUNNING) return (EBUSY); txq = &ctx->ifc_txqs[0]; error = iflib_encap(txq, &m); if (error == 0) (void)iflib_txd_db_check(ctx, txq, true, txq->ift_in_use); return (error); } static int iflib_debugnet_poll(if_t ifp, int count) { struct epoch_tracker et; if_ctx_t ctx; if_softc_ctx_t scctx; iflib_txq_t txq; int i; ctx = if_getsoftc(ifp); scctx = &ctx->ifc_softc_ctx; if ((if_getdrvflags(ifp) & (IFF_DRV_RUNNING | IFF_DRV_OACTIVE)) != IFF_DRV_RUNNING) return (EBUSY); txq = &ctx->ifc_txqs[0]; (void)iflib_completed_tx_reclaim(txq, RECLAIM_THRESH(ctx)); NET_EPOCH_ENTER(et); for (i = 0; i < scctx->isc_nrxqsets; i++) (void)iflib_rxeof(&ctx->ifc_rxqs[i], 16 /* XXX */); NET_EPOCH_EXIT(et); return (0); } #endif /* DEBUGNET */ Index: projects/clang1100-import/sys/net80211/ieee80211_ioctl.c =================================================================== --- projects/clang1100-import/sys/net80211/ieee80211_ioctl.c (revision 364034) +++ projects/clang1100-import/sys/net80211/ieee80211_ioctl.c (revision 364035) @@ -1,3709 +1,3716 @@ /*- * SPDX-License-Identifier: BSD-2-Clause-FreeBSD * * Copyright (c) 2001 Atsushi Onoe * Copyright (c) 2002-2009 Sam Leffler, Errno Consulting * 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. */ #include __FBSDID("$FreeBSD$"); /* * IEEE 802.11 ioctl support (FreeBSD-specific) */ #include "opt_inet.h" #include "opt_wlan.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef INET #include #include #endif #include #include #include #include #define IS_UP_AUTO(_vap) \ (IFNET_IS_UP_RUNNING((_vap)->iv_ifp) && \ (_vap)->iv_roaming == IEEE80211_ROAMING_AUTO) static const uint8_t zerobssid[IEEE80211_ADDR_LEN]; static struct ieee80211_channel *findchannel(struct ieee80211com *, int ieee, int mode); static int ieee80211_scanreq(struct ieee80211vap *, struct ieee80211_scan_req *); static int ieee80211_ioctl_getkey(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211com *ic = vap->iv_ic; struct ieee80211_node *ni; struct ieee80211req_key ik; struct ieee80211_key *wk; const struct ieee80211_cipher *cip; u_int kid; int error; if (ireq->i_len != sizeof(ik)) return EINVAL; error = copyin(ireq->i_data, &ik, sizeof(ik)); if (error) return error; kid = ik.ik_keyix; if (kid == IEEE80211_KEYIX_NONE) { ni = ieee80211_find_vap_node(&ic->ic_sta, vap, ik.ik_macaddr); if (ni == NULL) return ENOENT; wk = &ni->ni_ucastkey; } else { if (kid >= IEEE80211_WEP_NKID) return EINVAL; wk = &vap->iv_nw_keys[kid]; IEEE80211_ADDR_COPY(&ik.ik_macaddr, vap->iv_bss->ni_macaddr); ni = NULL; } cip = wk->wk_cipher; ik.ik_type = cip->ic_cipher; ik.ik_keylen = wk->wk_keylen; ik.ik_flags = wk->wk_flags & (IEEE80211_KEY_XMIT | IEEE80211_KEY_RECV); if (wk->wk_keyix == vap->iv_def_txkey) ik.ik_flags |= IEEE80211_KEY_DEFAULT; /* XXX TODO: move priv check to ieee80211_freebsd.c */ if (priv_check(curthread, PRIV_NET80211_VAP_GETKEY) == 0) { /* NB: only root can read key data */ ik.ik_keyrsc = wk->wk_keyrsc[IEEE80211_NONQOS_TID]; ik.ik_keytsc = wk->wk_keytsc; memcpy(ik.ik_keydata, wk->wk_key, wk->wk_keylen); if (cip->ic_cipher == IEEE80211_CIPHER_TKIP) { memcpy(ik.ik_keydata+wk->wk_keylen, wk->wk_key + IEEE80211_KEYBUF_SIZE, IEEE80211_MICBUF_SIZE); ik.ik_keylen += IEEE80211_MICBUF_SIZE; } } else { ik.ik_keyrsc = 0; ik.ik_keytsc = 0; memset(ik.ik_keydata, 0, sizeof(ik.ik_keydata)); } if (ni != NULL) ieee80211_free_node(ni); return copyout(&ik, ireq->i_data, sizeof(ik)); } static int ieee80211_ioctl_getchanlist(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211com *ic = vap->iv_ic; if (sizeof(ic->ic_chan_active) < ireq->i_len) ireq->i_len = sizeof(ic->ic_chan_active); return copyout(&ic->ic_chan_active, ireq->i_data, ireq->i_len); } static int ieee80211_ioctl_getchaninfo(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211com *ic = vap->iv_ic; uint32_t space; space = __offsetof(struct ieee80211req_chaninfo, ic_chans[ic->ic_nchans]); if (space > ireq->i_len) space = ireq->i_len; /* XXX assumes compatible layout */ return copyout(&ic->ic_nchans, ireq->i_data, space); } static int ieee80211_ioctl_getwpaie(struct ieee80211vap *vap, struct ieee80211req *ireq, int req) { struct ieee80211_node *ni; struct ieee80211req_wpaie2 *wpaie; int error; if (ireq->i_len < IEEE80211_ADDR_LEN) return EINVAL; wpaie = IEEE80211_MALLOC(sizeof(*wpaie), M_TEMP, IEEE80211_M_NOWAIT | IEEE80211_M_ZERO); if (wpaie == NULL) return ENOMEM; error = copyin(ireq->i_data, wpaie->wpa_macaddr, IEEE80211_ADDR_LEN); if (error != 0) goto bad; ni = ieee80211_find_vap_node(&vap->iv_ic->ic_sta, vap, wpaie->wpa_macaddr); if (ni == NULL) { error = ENOENT; goto bad; } if (ni->ni_ies.wpa_ie != NULL) { int ielen = ni->ni_ies.wpa_ie[1] + 2; if (ielen > sizeof(wpaie->wpa_ie)) ielen = sizeof(wpaie->wpa_ie); memcpy(wpaie->wpa_ie, ni->ni_ies.wpa_ie, ielen); } if (req == IEEE80211_IOC_WPAIE2) { if (ni->ni_ies.rsn_ie != NULL) { int ielen = ni->ni_ies.rsn_ie[1] + 2; if (ielen > sizeof(wpaie->rsn_ie)) ielen = sizeof(wpaie->rsn_ie); memcpy(wpaie->rsn_ie, ni->ni_ies.rsn_ie, ielen); } if (ireq->i_len > sizeof(struct ieee80211req_wpaie2)) ireq->i_len = sizeof(struct ieee80211req_wpaie2); } else { /* compatibility op, may overwrite wpa ie */ /* XXX check ic_flags? */ if (ni->ni_ies.rsn_ie != NULL) { int ielen = ni->ni_ies.rsn_ie[1] + 2; if (ielen > sizeof(wpaie->wpa_ie)) ielen = sizeof(wpaie->wpa_ie); memcpy(wpaie->wpa_ie, ni->ni_ies.rsn_ie, ielen); } if (ireq->i_len > sizeof(struct ieee80211req_wpaie)) ireq->i_len = sizeof(struct ieee80211req_wpaie); } ieee80211_free_node(ni); error = copyout(wpaie, ireq->i_data, ireq->i_len); bad: IEEE80211_FREE(wpaie, M_TEMP); return error; } static int ieee80211_ioctl_getstastats(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211_node *ni; uint8_t macaddr[IEEE80211_ADDR_LEN]; const size_t off = __offsetof(struct ieee80211req_sta_stats, is_stats); int error; if (ireq->i_len < off) return EINVAL; error = copyin(ireq->i_data, macaddr, IEEE80211_ADDR_LEN); if (error != 0) return error; ni = ieee80211_find_vap_node(&vap->iv_ic->ic_sta, vap, macaddr); if (ni == NULL) return ENOENT; if (ireq->i_len > sizeof(struct ieee80211req_sta_stats)) ireq->i_len = sizeof(struct ieee80211req_sta_stats); /* NB: copy out only the statistics */ error = copyout(&ni->ni_stats, (uint8_t *) ireq->i_data + off, ireq->i_len - off); ieee80211_free_node(ni); return error; } struct scanreq { struct ieee80211req_scan_result *sr; size_t space; }; static size_t scan_space(const struct ieee80211_scan_entry *se, int *ielen) { size_t len; *ielen = se->se_ies.len; /* * NB: ie's can be no more than 255 bytes and the max 802.11 * packet is <3Kbytes so we are sure this doesn't overflow * 16-bits; if this is a concern we can drop the ie's. */ len = sizeof(struct ieee80211req_scan_result) + se->se_ssid[1] + se->se_meshid[1] + *ielen; return roundup(len, sizeof(uint32_t)); } static void get_scan_space(void *arg, const struct ieee80211_scan_entry *se) { struct scanreq *req = arg; int ielen; req->space += scan_space(se, &ielen); } static void get_scan_result(void *arg, const struct ieee80211_scan_entry *se) { struct scanreq *req = arg; struct ieee80211req_scan_result *sr; int ielen, len, nr, nxr; uint8_t *cp; len = scan_space(se, &ielen); if (len > req->space) return; sr = req->sr; KASSERT(len <= 65535 && ielen <= 65535, ("len %u ssid %u ie %u", len, se->se_ssid[1], ielen)); sr->isr_len = len; sr->isr_ie_off = sizeof(struct ieee80211req_scan_result); sr->isr_ie_len = ielen; sr->isr_freq = se->se_chan->ic_freq; sr->isr_flags = se->se_chan->ic_flags; sr->isr_rssi = se->se_rssi; sr->isr_noise = se->se_noise; sr->isr_intval = se->se_intval; sr->isr_capinfo = se->se_capinfo; sr->isr_erp = se->se_erp; IEEE80211_ADDR_COPY(sr->isr_bssid, se->se_bssid); nr = min(se->se_rates[1], IEEE80211_RATE_MAXSIZE); memcpy(sr->isr_rates, se->se_rates+2, nr); nxr = min(se->se_xrates[1], IEEE80211_RATE_MAXSIZE - nr); memcpy(sr->isr_rates+nr, se->se_xrates+2, nxr); sr->isr_nrates = nr + nxr; /* copy SSID */ sr->isr_ssid_len = se->se_ssid[1]; cp = ((uint8_t *)sr) + sr->isr_ie_off; memcpy(cp, se->se_ssid+2, sr->isr_ssid_len); /* copy mesh id */ cp += sr->isr_ssid_len; sr->isr_meshid_len = se->se_meshid[1]; memcpy(cp, se->se_meshid+2, sr->isr_meshid_len); cp += sr->isr_meshid_len; if (ielen) memcpy(cp, se->se_ies.data, ielen); req->space -= len; req->sr = (struct ieee80211req_scan_result *)(((uint8_t *)sr) + len); } static int ieee80211_ioctl_getscanresults(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct scanreq req; int error; if (ireq->i_len < sizeof(struct scanreq)) return EFAULT; error = 0; req.space = 0; ieee80211_scan_iterate(vap, get_scan_space, &req); if (req.space > ireq->i_len) req.space = ireq->i_len; if (req.space > 0) { uint32_t space; void *p; space = req.space; /* XXX M_WAITOK after driver lock released */ p = IEEE80211_MALLOC(space, M_TEMP, IEEE80211_M_NOWAIT | IEEE80211_M_ZERO); if (p == NULL) return ENOMEM; req.sr = p; ieee80211_scan_iterate(vap, get_scan_result, &req); ireq->i_len = space - req.space; error = copyout(p, ireq->i_data, ireq->i_len); IEEE80211_FREE(p, M_TEMP); } else ireq->i_len = 0; return error; } struct stainforeq { struct ieee80211req_sta_info *si; size_t space; }; static size_t sta_space(const struct ieee80211_node *ni, size_t *ielen) { *ielen = ni->ni_ies.len; return roundup(sizeof(struct ieee80211req_sta_info) + *ielen, sizeof(uint32_t)); } static void get_sta_space(void *arg, struct ieee80211_node *ni) { struct stainforeq *req = arg; size_t ielen; if (ni->ni_vap->iv_opmode == IEEE80211_M_HOSTAP && ni->ni_associd == 0) /* only associated stations */ return; req->space += sta_space(ni, &ielen); } static void get_sta_info(void *arg, struct ieee80211_node *ni) { struct stainforeq *req = arg; struct ieee80211vap *vap = ni->ni_vap; struct ieee80211req_sta_info *si; size_t ielen, len; uint8_t *cp; if (vap->iv_opmode == IEEE80211_M_HOSTAP && ni->ni_associd == 0) /* only associated stations */ return; if (ni->ni_chan == IEEE80211_CHAN_ANYC) /* XXX bogus entry */ return; len = sta_space(ni, &ielen); if (len > req->space) return; si = req->si; si->isi_len = len; si->isi_ie_off = sizeof(struct ieee80211req_sta_info); si->isi_ie_len = ielen; si->isi_freq = ni->ni_chan->ic_freq; si->isi_flags = ni->ni_chan->ic_flags; si->isi_state = ni->ni_flags; si->isi_authmode = ni->ni_authmode; vap->iv_ic->ic_node_getsignal(ni, &si->isi_rssi, &si->isi_noise); vap->iv_ic->ic_node_getmimoinfo(ni, &si->isi_mimo); si->isi_capinfo = ni->ni_capinfo; si->isi_erp = ni->ni_erp; IEEE80211_ADDR_COPY(si->isi_macaddr, ni->ni_macaddr); si->isi_nrates = ni->ni_rates.rs_nrates; if (si->isi_nrates > 15) si->isi_nrates = 15; memcpy(si->isi_rates, ni->ni_rates.rs_rates, si->isi_nrates); si->isi_txrate = ni->ni_txrate; if (si->isi_txrate & IEEE80211_RATE_MCS) { const struct ieee80211_mcs_rates *mcs = &ieee80211_htrates[ni->ni_txrate &~ IEEE80211_RATE_MCS]; if (IEEE80211_IS_CHAN_HT40(ni->ni_chan)) { if (ni->ni_flags & IEEE80211_NODE_SGI40) si->isi_txmbps = mcs->ht40_rate_800ns; else si->isi_txmbps = mcs->ht40_rate_400ns; } else { if (ni->ni_flags & IEEE80211_NODE_SGI20) si->isi_txmbps = mcs->ht20_rate_800ns; else si->isi_txmbps = mcs->ht20_rate_400ns; } } else si->isi_txmbps = si->isi_txrate; si->isi_associd = ni->ni_associd; si->isi_txpower = ni->ni_txpower; si->isi_vlan = ni->ni_vlan; if (ni->ni_flags & IEEE80211_NODE_QOS) { memcpy(si->isi_txseqs, ni->ni_txseqs, sizeof(ni->ni_txseqs)); memcpy(si->isi_rxseqs, ni->ni_rxseqs, sizeof(ni->ni_rxseqs)); } else { si->isi_txseqs[0] = ni->ni_txseqs[IEEE80211_NONQOS_TID]; si->isi_rxseqs[0] = ni->ni_rxseqs[IEEE80211_NONQOS_TID]; } /* NB: leave all cases in case we relax ni_associd == 0 check */ if (ieee80211_node_is_authorized(ni)) si->isi_inact = vap->iv_inact_run; else if (ni->ni_associd != 0 || (vap->iv_opmode == IEEE80211_M_WDS && (vap->iv_flags_ext & IEEE80211_FEXT_WDSLEGACY))) si->isi_inact = vap->iv_inact_auth; else si->isi_inact = vap->iv_inact_init; si->isi_inact = (si->isi_inact - ni->ni_inact) * IEEE80211_INACT_WAIT; si->isi_localid = ni->ni_mllid; si->isi_peerid = ni->ni_mlpid; si->isi_peerstate = ni->ni_mlstate; if (ielen) { cp = ((uint8_t *)si) + si->isi_ie_off; memcpy(cp, ni->ni_ies.data, ielen); } req->si = (struct ieee80211req_sta_info *)(((uint8_t *)si) + len); req->space -= len; } static int getstainfo_common(struct ieee80211vap *vap, struct ieee80211req *ireq, struct ieee80211_node *ni, size_t off) { struct ieee80211com *ic = vap->iv_ic; struct stainforeq req; size_t space; void *p; int error; error = 0; req.space = 0; if (ni == NULL) { ieee80211_iterate_nodes_vap(&ic->ic_sta, vap, get_sta_space, &req); } else get_sta_space(&req, ni); if (req.space > ireq->i_len) req.space = ireq->i_len; if (req.space > 0) { space = req.space; /* XXX M_WAITOK after driver lock released */ p = IEEE80211_MALLOC(space, M_TEMP, IEEE80211_M_NOWAIT | IEEE80211_M_ZERO); if (p == NULL) { error = ENOMEM; goto bad; } req.si = p; if (ni == NULL) { ieee80211_iterate_nodes_vap(&ic->ic_sta, vap, get_sta_info, &req); } else get_sta_info(&req, ni); ireq->i_len = space - req.space; error = copyout(p, (uint8_t *) ireq->i_data+off, ireq->i_len); IEEE80211_FREE(p, M_TEMP); } else ireq->i_len = 0; bad: if (ni != NULL) ieee80211_free_node(ni); return error; } static int ieee80211_ioctl_getstainfo(struct ieee80211vap *vap, struct ieee80211req *ireq) { uint8_t macaddr[IEEE80211_ADDR_LEN]; const size_t off = __offsetof(struct ieee80211req_sta_req, info); struct ieee80211_node *ni; int error; if (ireq->i_len < sizeof(struct ieee80211req_sta_req)) return EFAULT; error = copyin(ireq->i_data, macaddr, IEEE80211_ADDR_LEN); if (error != 0) return error; if (IEEE80211_ADDR_EQ(macaddr, vap->iv_ifp->if_broadcastaddr)) { ni = NULL; } else { ni = ieee80211_find_vap_node(&vap->iv_ic->ic_sta, vap, macaddr); if (ni == NULL) return ENOENT; } return getstainfo_common(vap, ireq, ni, off); } static int ieee80211_ioctl_getstatxpow(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211_node *ni; struct ieee80211req_sta_txpow txpow; int error; if (ireq->i_len != sizeof(txpow)) return EINVAL; error = copyin(ireq->i_data, &txpow, sizeof(txpow)); if (error != 0) return error; ni = ieee80211_find_vap_node(&vap->iv_ic->ic_sta, vap, txpow.it_macaddr); if (ni == NULL) return ENOENT; txpow.it_txpow = ni->ni_txpower; error = copyout(&txpow, ireq->i_data, sizeof(txpow)); ieee80211_free_node(ni); return error; } static int ieee80211_ioctl_getwmeparam(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211com *ic = vap->iv_ic; struct ieee80211_wme_state *wme = &ic->ic_wme; struct wmeParams *wmep; int ac; if ((ic->ic_caps & IEEE80211_C_WME) == 0) return EINVAL; ac = (ireq->i_len & IEEE80211_WMEPARAM_VAL); if (ac >= WME_NUM_AC) ac = WME_AC_BE; if (ireq->i_len & IEEE80211_WMEPARAM_BSS) wmep = &wme->wme_wmeBssChanParams.cap_wmeParams[ac]; else wmep = &wme->wme_wmeChanParams.cap_wmeParams[ac]; switch (ireq->i_type) { case IEEE80211_IOC_WME_CWMIN: /* WME: CWmin */ ireq->i_val = wmep->wmep_logcwmin; break; case IEEE80211_IOC_WME_CWMAX: /* WME: CWmax */ ireq->i_val = wmep->wmep_logcwmax; break; case IEEE80211_IOC_WME_AIFS: /* WME: AIFS */ ireq->i_val = wmep->wmep_aifsn; break; case IEEE80211_IOC_WME_TXOPLIMIT: /* WME: txops limit */ ireq->i_val = wmep->wmep_txopLimit; break; case IEEE80211_IOC_WME_ACM: /* WME: ACM (bss only) */ wmep = &wme->wme_wmeBssChanParams.cap_wmeParams[ac]; ireq->i_val = wmep->wmep_acm; break; case IEEE80211_IOC_WME_ACKPOLICY: /* WME: ACK policy (!bss only)*/ wmep = &wme->wme_wmeChanParams.cap_wmeParams[ac]; ireq->i_val = !wmep->wmep_noackPolicy; break; } return 0; } static int ieee80211_ioctl_getmaccmd(struct ieee80211vap *vap, struct ieee80211req *ireq) { const struct ieee80211_aclator *acl = vap->iv_acl; return (acl == NULL ? EINVAL : acl->iac_getioctl(vap, ireq)); } static int ieee80211_ioctl_getcurchan(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211com *ic = vap->iv_ic; struct ieee80211_channel *c; if (ireq->i_len != sizeof(struct ieee80211_channel)) return EINVAL; /* * vap's may have different operating channels when HT is * in use. When in RUN state report the vap-specific channel. * Otherwise return curchan. */ if (vap->iv_state == IEEE80211_S_RUN || vap->iv_state == IEEE80211_S_SLEEP) c = vap->iv_bss->ni_chan; else c = ic->ic_curchan; return copyout(c, ireq->i_data, sizeof(*c)); } static int getappie(const struct ieee80211_appie *aie, struct ieee80211req *ireq) { if (aie == NULL) return EINVAL; /* NB: truncate, caller can check length */ if (ireq->i_len > aie->ie_len) ireq->i_len = aie->ie_len; return copyout(aie->ie_data, ireq->i_data, ireq->i_len); } static int ieee80211_ioctl_getappie(struct ieee80211vap *vap, struct ieee80211req *ireq) { uint8_t fc0; fc0 = ireq->i_val & 0xff; if ((fc0 & IEEE80211_FC0_TYPE_MASK) != IEEE80211_FC0_TYPE_MGT) return EINVAL; /* NB: could check iv_opmode and reject but hardly worth the effort */ switch (fc0 & IEEE80211_FC0_SUBTYPE_MASK) { case IEEE80211_FC0_SUBTYPE_BEACON: return getappie(vap->iv_appie_beacon, ireq); case IEEE80211_FC0_SUBTYPE_PROBE_RESP: return getappie(vap->iv_appie_proberesp, ireq); case IEEE80211_FC0_SUBTYPE_ASSOC_RESP: return getappie(vap->iv_appie_assocresp, ireq); case IEEE80211_FC0_SUBTYPE_PROBE_REQ: return getappie(vap->iv_appie_probereq, ireq); case IEEE80211_FC0_SUBTYPE_ASSOC_REQ: return getappie(vap->iv_appie_assocreq, ireq); case IEEE80211_FC0_SUBTYPE_BEACON|IEEE80211_FC0_SUBTYPE_PROBE_RESP: return getappie(vap->iv_appie_wpa, ireq); } return EINVAL; } static int ieee80211_ioctl_getregdomain(struct ieee80211vap *vap, const struct ieee80211req *ireq) { struct ieee80211com *ic = vap->iv_ic; if (ireq->i_len != sizeof(ic->ic_regdomain)) return EINVAL; return copyout(&ic->ic_regdomain, ireq->i_data, sizeof(ic->ic_regdomain)); } static int ieee80211_ioctl_getroam(struct ieee80211vap *vap, const struct ieee80211req *ireq) { size_t len = ireq->i_len; /* NB: accept short requests for backwards compat */ if (len > sizeof(vap->iv_roamparms)) len = sizeof(vap->iv_roamparms); return copyout(vap->iv_roamparms, ireq->i_data, len); } static int ieee80211_ioctl_gettxparams(struct ieee80211vap *vap, const struct ieee80211req *ireq) { size_t len = ireq->i_len; /* NB: accept short requests for backwards compat */ if (len > sizeof(vap->iv_txparms)) len = sizeof(vap->iv_txparms); return copyout(vap->iv_txparms, ireq->i_data, len); } static int ieee80211_ioctl_getdevcaps(struct ieee80211com *ic, const struct ieee80211req *ireq) { struct ieee80211_devcaps_req *dc; struct ieee80211req_chaninfo *ci; int maxchans, error; maxchans = 1 + ((ireq->i_len - sizeof(struct ieee80211_devcaps_req)) / sizeof(struct ieee80211_channel)); /* NB: require 1 so we know ic_nchans is accessible */ if (maxchans < 1) return EINVAL; /* constrain max request size, 2K channels is ~24Kbytes */ if (maxchans > 2048) maxchans = 2048; dc = (struct ieee80211_devcaps_req *) IEEE80211_MALLOC(IEEE80211_DEVCAPS_SIZE(maxchans), M_TEMP, IEEE80211_M_NOWAIT | IEEE80211_M_ZERO); if (dc == NULL) return ENOMEM; dc->dc_drivercaps = ic->ic_caps; dc->dc_cryptocaps = ic->ic_cryptocaps; dc->dc_htcaps = ic->ic_htcaps; dc->dc_vhtcaps = ic->ic_vhtcaps; ci = &dc->dc_chaninfo; ic->ic_getradiocaps(ic, maxchans, &ci->ic_nchans, ci->ic_chans); KASSERT(ci->ic_nchans <= maxchans, ("nchans %d maxchans %d", ci->ic_nchans, maxchans)); ieee80211_sort_channels(ci->ic_chans, ci->ic_nchans); error = copyout(dc, ireq->i_data, IEEE80211_DEVCAPS_SPACE(dc)); IEEE80211_FREE(dc, M_TEMP); return error; } static int ieee80211_ioctl_getstavlan(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211_node *ni; struct ieee80211req_sta_vlan vlan; int error; if (ireq->i_len != sizeof(vlan)) return EINVAL; error = copyin(ireq->i_data, &vlan, sizeof(vlan)); if (error != 0) return error; if (!IEEE80211_ADDR_EQ(vlan.sv_macaddr, zerobssid)) { ni = ieee80211_find_vap_node(&vap->iv_ic->ic_sta, vap, vlan.sv_macaddr); if (ni == NULL) return ENOENT; } else ni = ieee80211_ref_node(vap->iv_bss); vlan.sv_vlan = ni->ni_vlan; error = copyout(&vlan, ireq->i_data, sizeof(vlan)); ieee80211_free_node(ni); return error; } /* * Dummy ioctl get handler so the linker set is defined. */ static int dummy_ioctl_get(struct ieee80211vap *vap, struct ieee80211req *ireq) { return ENOSYS; } IEEE80211_IOCTL_GET(dummy, dummy_ioctl_get); static int ieee80211_ioctl_getdefault(struct ieee80211vap *vap, struct ieee80211req *ireq) { ieee80211_ioctl_getfunc * const *get; int error; SET_FOREACH(get, ieee80211_ioctl_getset) { error = (*get)(vap, ireq); if (error != ENOSYS) return error; } return EINVAL; } static int ieee80211_ioctl_get80211(struct ieee80211vap *vap, u_long cmd, struct ieee80211req *ireq) { #define MS(_v, _f) (((_v) & _f) >> _f##_S) struct ieee80211com *ic = vap->iv_ic; u_int kid, len; uint8_t tmpkey[IEEE80211_KEYBUF_SIZE]; char tmpssid[IEEE80211_NWID_LEN]; int error = 0; switch (ireq->i_type) { + case IEEE80211_IOC_IC_NAME: + len = strlen(ic->ic_name) + 1; + if (len > ireq->i_len) + return (EINVAL); + ireq->i_len = len; + error = copyout(ic->ic_name, ireq->i_data, ireq->i_len); + break; case IEEE80211_IOC_SSID: switch (vap->iv_state) { case IEEE80211_S_INIT: case IEEE80211_S_SCAN: ireq->i_len = vap->iv_des_ssid[0].len; memcpy(tmpssid, vap->iv_des_ssid[0].ssid, ireq->i_len); break; default: ireq->i_len = vap->iv_bss->ni_esslen; memcpy(tmpssid, vap->iv_bss->ni_essid, ireq->i_len); break; } error = copyout(tmpssid, ireq->i_data, ireq->i_len); break; case IEEE80211_IOC_NUMSSIDS: ireq->i_val = 1; break; case IEEE80211_IOC_WEP: if ((vap->iv_flags & IEEE80211_F_PRIVACY) == 0) ireq->i_val = IEEE80211_WEP_OFF; else if (vap->iv_flags & IEEE80211_F_DROPUNENC) ireq->i_val = IEEE80211_WEP_ON; else ireq->i_val = IEEE80211_WEP_MIXED; break; case IEEE80211_IOC_WEPKEY: kid = (u_int) ireq->i_val; if (kid >= IEEE80211_WEP_NKID) return EINVAL; len = (u_int) vap->iv_nw_keys[kid].wk_keylen; /* NB: only root can read WEP keys */ /* XXX TODO: move priv check to ieee80211_freebsd.c */ if (priv_check(curthread, PRIV_NET80211_VAP_GETKEY) == 0) { bcopy(vap->iv_nw_keys[kid].wk_key, tmpkey, len); } else { bzero(tmpkey, len); } ireq->i_len = len; error = copyout(tmpkey, ireq->i_data, len); break; case IEEE80211_IOC_NUMWEPKEYS: ireq->i_val = IEEE80211_WEP_NKID; break; case IEEE80211_IOC_WEPTXKEY: ireq->i_val = vap->iv_def_txkey; break; case IEEE80211_IOC_AUTHMODE: if (vap->iv_flags & IEEE80211_F_WPA) ireq->i_val = IEEE80211_AUTH_WPA; else ireq->i_val = vap->iv_bss->ni_authmode; break; case IEEE80211_IOC_CHANNEL: ireq->i_val = ieee80211_chan2ieee(ic, ic->ic_curchan); break; case IEEE80211_IOC_POWERSAVE: if (vap->iv_flags & IEEE80211_F_PMGTON) ireq->i_val = IEEE80211_POWERSAVE_ON; else ireq->i_val = IEEE80211_POWERSAVE_OFF; break; case IEEE80211_IOC_POWERSAVESLEEP: ireq->i_val = ic->ic_lintval; break; case IEEE80211_IOC_RTSTHRESHOLD: ireq->i_val = vap->iv_rtsthreshold; break; case IEEE80211_IOC_PROTMODE: ireq->i_val = vap->iv_protmode; break; case IEEE80211_IOC_TXPOWER: /* * Tx power limit is the min of max regulatory * power, any user-set limit, and the max the * radio can do. * * TODO: methodize this */ ireq->i_val = 2*ic->ic_curchan->ic_maxregpower; if (ireq->i_val > ic->ic_txpowlimit) ireq->i_val = ic->ic_txpowlimit; if (ireq->i_val > ic->ic_curchan->ic_maxpower) ireq->i_val = ic->ic_curchan->ic_maxpower; break; case IEEE80211_IOC_WPA: switch (vap->iv_flags & IEEE80211_F_WPA) { case IEEE80211_F_WPA1: ireq->i_val = 1; break; case IEEE80211_F_WPA2: ireq->i_val = 2; break; case IEEE80211_F_WPA1 | IEEE80211_F_WPA2: ireq->i_val = 3; break; default: ireq->i_val = 0; break; } break; case IEEE80211_IOC_CHANLIST: error = ieee80211_ioctl_getchanlist(vap, ireq); break; case IEEE80211_IOC_ROAMING: ireq->i_val = vap->iv_roaming; break; case IEEE80211_IOC_PRIVACY: ireq->i_val = (vap->iv_flags & IEEE80211_F_PRIVACY) != 0; break; case IEEE80211_IOC_DROPUNENCRYPTED: ireq->i_val = (vap->iv_flags & IEEE80211_F_DROPUNENC) != 0; break; case IEEE80211_IOC_COUNTERMEASURES: ireq->i_val = (vap->iv_flags & IEEE80211_F_COUNTERM) != 0; break; case IEEE80211_IOC_WME: ireq->i_val = (vap->iv_flags & IEEE80211_F_WME) != 0; break; case IEEE80211_IOC_HIDESSID: ireq->i_val = (vap->iv_flags & IEEE80211_F_HIDESSID) != 0; break; case IEEE80211_IOC_APBRIDGE: ireq->i_val = (vap->iv_flags & IEEE80211_F_NOBRIDGE) == 0; break; case IEEE80211_IOC_WPAKEY: error = ieee80211_ioctl_getkey(vap, ireq); break; case IEEE80211_IOC_CHANINFO: error = ieee80211_ioctl_getchaninfo(vap, ireq); break; case IEEE80211_IOC_BSSID: if (ireq->i_len != IEEE80211_ADDR_LEN) return EINVAL; if (vap->iv_state == IEEE80211_S_RUN || vap->iv_state == IEEE80211_S_SLEEP) { error = copyout(vap->iv_opmode == IEEE80211_M_WDS ? vap->iv_bss->ni_macaddr : vap->iv_bss->ni_bssid, ireq->i_data, ireq->i_len); } else error = copyout(vap->iv_des_bssid, ireq->i_data, ireq->i_len); break; case IEEE80211_IOC_WPAIE: case IEEE80211_IOC_WPAIE2: error = ieee80211_ioctl_getwpaie(vap, ireq, ireq->i_type); break; case IEEE80211_IOC_SCAN_RESULTS: error = ieee80211_ioctl_getscanresults(vap, ireq); break; case IEEE80211_IOC_STA_STATS: error = ieee80211_ioctl_getstastats(vap, ireq); break; case IEEE80211_IOC_TXPOWMAX: ireq->i_val = vap->iv_bss->ni_txpower; break; case IEEE80211_IOC_STA_TXPOW: error = ieee80211_ioctl_getstatxpow(vap, ireq); break; case IEEE80211_IOC_STA_INFO: error = ieee80211_ioctl_getstainfo(vap, ireq); break; case IEEE80211_IOC_WME_CWMIN: /* WME: CWmin */ case IEEE80211_IOC_WME_CWMAX: /* WME: CWmax */ case IEEE80211_IOC_WME_AIFS: /* WME: AIFS */ case IEEE80211_IOC_WME_TXOPLIMIT: /* WME: txops limit */ case IEEE80211_IOC_WME_ACM: /* WME: ACM (bss only) */ case IEEE80211_IOC_WME_ACKPOLICY: /* WME: ACK policy (!bss only) */ error = ieee80211_ioctl_getwmeparam(vap, ireq); break; case IEEE80211_IOC_DTIM_PERIOD: ireq->i_val = vap->iv_dtim_period; break; case IEEE80211_IOC_BEACON_INTERVAL: /* NB: get from ic_bss for station mode */ ireq->i_val = vap->iv_bss->ni_intval; break; case IEEE80211_IOC_PUREG: ireq->i_val = (vap->iv_flags & IEEE80211_F_PUREG) != 0; break; case IEEE80211_IOC_QUIET: ireq->i_val = vap->iv_quiet; break; case IEEE80211_IOC_QUIET_COUNT: ireq->i_val = vap->iv_quiet_count; break; case IEEE80211_IOC_QUIET_PERIOD: ireq->i_val = vap->iv_quiet_period; break; case IEEE80211_IOC_QUIET_DUR: ireq->i_val = vap->iv_quiet_duration; break; case IEEE80211_IOC_QUIET_OFFSET: ireq->i_val = vap->iv_quiet_offset; break; case IEEE80211_IOC_BGSCAN: ireq->i_val = (vap->iv_flags & IEEE80211_F_BGSCAN) != 0; break; case IEEE80211_IOC_BGSCAN_IDLE: ireq->i_val = vap->iv_bgscanidle*hz/1000; /* ms */ break; case IEEE80211_IOC_BGSCAN_INTERVAL: ireq->i_val = vap->iv_bgscanintvl/hz; /* seconds */ break; case IEEE80211_IOC_SCANVALID: ireq->i_val = vap->iv_scanvalid/hz; /* seconds */ break; case IEEE80211_IOC_FRAGTHRESHOLD: ireq->i_val = vap->iv_fragthreshold; break; case IEEE80211_IOC_MACCMD: error = ieee80211_ioctl_getmaccmd(vap, ireq); break; case IEEE80211_IOC_BURST: ireq->i_val = (vap->iv_flags & IEEE80211_F_BURST) != 0; break; case IEEE80211_IOC_BMISSTHRESHOLD: ireq->i_val = vap->iv_bmissthreshold; break; case IEEE80211_IOC_CURCHAN: error = ieee80211_ioctl_getcurchan(vap, ireq); break; case IEEE80211_IOC_SHORTGI: ireq->i_val = 0; if (vap->iv_flags_ht & IEEE80211_FHT_SHORTGI20) ireq->i_val |= IEEE80211_HTCAP_SHORTGI20; if (vap->iv_flags_ht & IEEE80211_FHT_SHORTGI40) ireq->i_val |= IEEE80211_HTCAP_SHORTGI40; break; case IEEE80211_IOC_AMPDU: ireq->i_val = 0; if (vap->iv_flags_ht & IEEE80211_FHT_AMPDU_TX) ireq->i_val |= 1; if (vap->iv_flags_ht & IEEE80211_FHT_AMPDU_RX) ireq->i_val |= 2; break; case IEEE80211_IOC_AMPDU_LIMIT: /* XXX TODO: make this a per-node thing; and leave this as global */ if (vap->iv_opmode == IEEE80211_M_HOSTAP) ireq->i_val = vap->iv_ampdu_rxmax; else if (vap->iv_state == IEEE80211_S_RUN || vap->iv_state == IEEE80211_S_SLEEP) /* * XXX TODO: this isn't completely correct, as we've * negotiated the higher of the two. */ ireq->i_val = MS(vap->iv_bss->ni_htparam, IEEE80211_HTCAP_MAXRXAMPDU); else ireq->i_val = vap->iv_ampdu_limit; break; case IEEE80211_IOC_AMPDU_DENSITY: /* XXX TODO: make this a per-node thing; and leave this as global */ if (vap->iv_opmode == IEEE80211_M_STA && (vap->iv_state == IEEE80211_S_RUN || vap->iv_state == IEEE80211_S_SLEEP)) /* * XXX TODO: this isn't completely correct, as we've * negotiated the higher of the two. */ ireq->i_val = MS(vap->iv_bss->ni_htparam, IEEE80211_HTCAP_MPDUDENSITY); else ireq->i_val = vap->iv_ampdu_density; break; case IEEE80211_IOC_AMSDU: ireq->i_val = 0; if (vap->iv_flags_ht & IEEE80211_FHT_AMSDU_TX) ireq->i_val |= 1; if (vap->iv_flags_ht & IEEE80211_FHT_AMSDU_RX) ireq->i_val |= 2; break; case IEEE80211_IOC_AMSDU_LIMIT: ireq->i_val = vap->iv_amsdu_limit; /* XXX truncation? */ break; case IEEE80211_IOC_PUREN: ireq->i_val = (vap->iv_flags_ht & IEEE80211_FHT_PUREN) != 0; break; case IEEE80211_IOC_DOTH: ireq->i_val = (vap->iv_flags & IEEE80211_F_DOTH) != 0; break; case IEEE80211_IOC_REGDOMAIN: error = ieee80211_ioctl_getregdomain(vap, ireq); break; case IEEE80211_IOC_ROAM: error = ieee80211_ioctl_getroam(vap, ireq); break; case IEEE80211_IOC_TXPARAMS: error = ieee80211_ioctl_gettxparams(vap, ireq); break; case IEEE80211_IOC_HTCOMPAT: ireq->i_val = (vap->iv_flags_ht & IEEE80211_FHT_HTCOMPAT) != 0; break; case IEEE80211_IOC_DWDS: ireq->i_val = (vap->iv_flags & IEEE80211_F_DWDS) != 0; break; case IEEE80211_IOC_INACTIVITY: ireq->i_val = (vap->iv_flags_ext & IEEE80211_FEXT_INACT) != 0; break; case IEEE80211_IOC_APPIE: error = ieee80211_ioctl_getappie(vap, ireq); break; case IEEE80211_IOC_WPS: ireq->i_val = (vap->iv_flags_ext & IEEE80211_FEXT_WPS) != 0; break; case IEEE80211_IOC_TSN: ireq->i_val = (vap->iv_flags_ext & IEEE80211_FEXT_TSN) != 0; break; case IEEE80211_IOC_DFS: ireq->i_val = (vap->iv_flags_ext & IEEE80211_FEXT_DFS) != 0; break; case IEEE80211_IOC_DOTD: ireq->i_val = (vap->iv_flags_ext & IEEE80211_FEXT_DOTD) != 0; break; case IEEE80211_IOC_DEVCAPS: error = ieee80211_ioctl_getdevcaps(ic, ireq); break; case IEEE80211_IOC_HTPROTMODE: ireq->i_val = vap->iv_htprotmode; break; case IEEE80211_IOC_HTCONF: if (vap->iv_flags_ht & IEEE80211_FHT_HT) { ireq->i_val = 1; if (vap->iv_flags_ht & IEEE80211_FHT_USEHT40) ireq->i_val |= 2; } else ireq->i_val = 0; break; case IEEE80211_IOC_STA_VLAN: error = ieee80211_ioctl_getstavlan(vap, ireq); break; case IEEE80211_IOC_SMPS: if (vap->iv_opmode == IEEE80211_M_STA && (vap->iv_state == IEEE80211_S_RUN || vap->iv_state == IEEE80211_S_SLEEP)) { if (vap->iv_bss->ni_flags & IEEE80211_NODE_MIMO_RTS) ireq->i_val = IEEE80211_HTCAP_SMPS_DYNAMIC; else if (vap->iv_bss->ni_flags & IEEE80211_NODE_MIMO_PS) ireq->i_val = IEEE80211_HTCAP_SMPS_ENA; else ireq->i_val = IEEE80211_HTCAP_SMPS_OFF; } else ireq->i_val = vap->iv_htcaps & IEEE80211_HTCAP_SMPS; break; case IEEE80211_IOC_RIFS: if (vap->iv_opmode == IEEE80211_M_STA && (vap->iv_state == IEEE80211_S_RUN || vap->iv_state == IEEE80211_S_SLEEP)) ireq->i_val = (vap->iv_bss->ni_flags & IEEE80211_NODE_RIFS) != 0; else ireq->i_val = (vap->iv_flags_ht & IEEE80211_FHT_RIFS) != 0; break; case IEEE80211_IOC_STBC: ireq->i_val = 0; if (vap->iv_flags_ht & IEEE80211_FHT_STBC_TX) ireq->i_val |= 1; if (vap->iv_flags_ht & IEEE80211_FHT_STBC_RX) ireq->i_val |= 2; break; case IEEE80211_IOC_LDPC: ireq->i_val = 0; if (vap->iv_flags_ht & IEEE80211_FHT_LDPC_TX) ireq->i_val |= 1; if (vap->iv_flags_ht & IEEE80211_FHT_LDPC_RX) ireq->i_val |= 2; break; case IEEE80211_IOC_UAPSD: ireq->i_val = 0; if (vap->iv_flags_ext & IEEE80211_FEXT_UAPSD) ireq->i_val = 1; break; /* VHT */ case IEEE80211_IOC_VHTCONF: ireq->i_val = 0; if (vap->iv_flags_vht & IEEE80211_FVHT_VHT) ireq->i_val |= 1; if (vap->iv_flags_vht & IEEE80211_FVHT_USEVHT40) ireq->i_val |= 2; if (vap->iv_flags_vht & IEEE80211_FVHT_USEVHT80) ireq->i_val |= 4; if (vap->iv_flags_vht & IEEE80211_FVHT_USEVHT80P80) ireq->i_val |= 8; if (vap->iv_flags_vht & IEEE80211_FVHT_USEVHT160) ireq->i_val |= 16; break; default: error = ieee80211_ioctl_getdefault(vap, ireq); break; } return error; #undef MS } static int ieee80211_ioctl_setkey(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211req_key ik; struct ieee80211_node *ni; struct ieee80211_key *wk; uint16_t kid; int error, i; if (ireq->i_len != sizeof(ik)) return EINVAL; error = copyin(ireq->i_data, &ik, sizeof(ik)); if (error) return error; /* NB: cipher support is verified by ieee80211_crypt_newkey */ /* NB: this also checks ik->ik_keylen > sizeof(wk->wk_key) */ if (ik.ik_keylen > sizeof(ik.ik_keydata)) return E2BIG; kid = ik.ik_keyix; if (kid == IEEE80211_KEYIX_NONE) { /* XXX unicast keys currently must be tx/rx */ if (ik.ik_flags != (IEEE80211_KEY_XMIT | IEEE80211_KEY_RECV)) return EINVAL; if (vap->iv_opmode == IEEE80211_M_STA) { ni = ieee80211_ref_node(vap->iv_bss); if (!IEEE80211_ADDR_EQ(ik.ik_macaddr, ni->ni_bssid)) { ieee80211_free_node(ni); return EADDRNOTAVAIL; } } else { ni = ieee80211_find_vap_node(&vap->iv_ic->ic_sta, vap, ik.ik_macaddr); if (ni == NULL) return ENOENT; } wk = &ni->ni_ucastkey; } else { if (kid >= IEEE80211_WEP_NKID) return EINVAL; wk = &vap->iv_nw_keys[kid]; /* * Global slots start off w/o any assigned key index. * Force one here for consistency with IEEE80211_IOC_WEPKEY. */ if (wk->wk_keyix == IEEE80211_KEYIX_NONE) wk->wk_keyix = kid; ni = NULL; } error = 0; ieee80211_key_update_begin(vap); if (ieee80211_crypto_newkey(vap, ik.ik_type, ik.ik_flags, wk)) { wk->wk_keylen = ik.ik_keylen; /* NB: MIC presence is implied by cipher type */ if (wk->wk_keylen > IEEE80211_KEYBUF_SIZE) wk->wk_keylen = IEEE80211_KEYBUF_SIZE; for (i = 0; i < IEEE80211_TID_SIZE; i++) wk->wk_keyrsc[i] = ik.ik_keyrsc; wk->wk_keytsc = 0; /* new key, reset */ memset(wk->wk_key, 0, sizeof(wk->wk_key)); memcpy(wk->wk_key, ik.ik_keydata, ik.ik_keylen); IEEE80211_ADDR_COPY(wk->wk_macaddr, ni != NULL ? ni->ni_macaddr : ik.ik_macaddr); if (!ieee80211_crypto_setkey(vap, wk)) error = EIO; else if ((ik.ik_flags & IEEE80211_KEY_DEFAULT)) /* * Inform the driver that this is the default * transmit key. Now, ideally we'd just set * a flag in the key update that would * say "yes, we're the default key", but * that currently isn't the way the ioctl -> * key interface works. */ ieee80211_crypto_set_deftxkey(vap, kid); } else error = ENXIO; ieee80211_key_update_end(vap); if (ni != NULL) ieee80211_free_node(ni); return error; } static int ieee80211_ioctl_delkey(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211req_del_key dk; int kid, error; if (ireq->i_len != sizeof(dk)) return EINVAL; error = copyin(ireq->i_data, &dk, sizeof(dk)); if (error) return error; kid = dk.idk_keyix; /* XXX uint8_t -> uint16_t */ if (dk.idk_keyix == (uint8_t) IEEE80211_KEYIX_NONE) { struct ieee80211_node *ni; if (vap->iv_opmode == IEEE80211_M_STA) { ni = ieee80211_ref_node(vap->iv_bss); if (!IEEE80211_ADDR_EQ(dk.idk_macaddr, ni->ni_bssid)) { ieee80211_free_node(ni); return EADDRNOTAVAIL; } } else { ni = ieee80211_find_vap_node(&vap->iv_ic->ic_sta, vap, dk.idk_macaddr); if (ni == NULL) return ENOENT; } /* XXX error return */ ieee80211_node_delucastkey(ni); ieee80211_free_node(ni); } else { if (kid >= IEEE80211_WEP_NKID) return EINVAL; /* XXX error return */ ieee80211_crypto_delkey(vap, &vap->iv_nw_keys[kid]); } return 0; } struct mlmeop { struct ieee80211vap *vap; int op; int reason; }; static void mlmedebug(struct ieee80211vap *vap, const uint8_t mac[IEEE80211_ADDR_LEN], int op, int reason) { #ifdef IEEE80211_DEBUG static const struct { int mask; const char *opstr; } ops[] = { { 0, "op#0" }, { IEEE80211_MSG_IOCTL | IEEE80211_MSG_STATE | IEEE80211_MSG_ASSOC, "assoc" }, { IEEE80211_MSG_IOCTL | IEEE80211_MSG_STATE | IEEE80211_MSG_ASSOC, "disassoc" }, { IEEE80211_MSG_IOCTL | IEEE80211_MSG_STATE | IEEE80211_MSG_AUTH, "deauth" }, { IEEE80211_MSG_IOCTL | IEEE80211_MSG_STATE | IEEE80211_MSG_AUTH, "authorize" }, { IEEE80211_MSG_IOCTL | IEEE80211_MSG_STATE | IEEE80211_MSG_AUTH, "unauthorize" }, }; if (op == IEEE80211_MLME_AUTH) { IEEE80211_NOTE_MAC(vap, IEEE80211_MSG_IOCTL | IEEE80211_MSG_STATE | IEEE80211_MSG_AUTH, mac, "station authenticate %s via MLME (reason: %d (%s))", reason == IEEE80211_STATUS_SUCCESS ? "ACCEPT" : "REJECT", reason, ieee80211_reason_to_string(reason)); } else if (!(IEEE80211_MLME_ASSOC <= op && op <= IEEE80211_MLME_AUTH)) { IEEE80211_NOTE_MAC(vap, IEEE80211_MSG_ANY, mac, "unknown MLME request %d (reason: %d (%s))", op, reason, ieee80211_reason_to_string(reason)); } else if (reason == IEEE80211_STATUS_SUCCESS) { IEEE80211_NOTE_MAC(vap, ops[op].mask, mac, "station %s via MLME", ops[op].opstr); } else { IEEE80211_NOTE_MAC(vap, ops[op].mask, mac, "station %s via MLME (reason: %d (%s))", ops[op].opstr, reason, ieee80211_reason_to_string(reason)); } #endif /* IEEE80211_DEBUG */ } static void domlme(void *arg, struct ieee80211_node *ni) { struct mlmeop *mop = arg; struct ieee80211vap *vap = ni->ni_vap; if (vap != mop->vap) return; /* * NB: if ni_associd is zero then the node is already cleaned * up and we don't need to do this (we're safely holding a * reference but should otherwise not modify it's state). */ if (ni->ni_associd == 0) return; mlmedebug(vap, ni->ni_macaddr, mop->op, mop->reason); if (mop->op == IEEE80211_MLME_DEAUTH) { IEEE80211_SEND_MGMT(ni, IEEE80211_FC0_SUBTYPE_DEAUTH, mop->reason); } else { IEEE80211_SEND_MGMT(ni, IEEE80211_FC0_SUBTYPE_DISASSOC, mop->reason); } ieee80211_node_leave(ni); } static int setmlme_dropsta(struct ieee80211vap *vap, const uint8_t mac[IEEE80211_ADDR_LEN], struct mlmeop *mlmeop) { struct ieee80211_node_table *nt = &vap->iv_ic->ic_sta; struct ieee80211_node *ni; int error = 0; /* NB: the broadcast address means do 'em all */ if (!IEEE80211_ADDR_EQ(mac, vap->iv_ifp->if_broadcastaddr)) { IEEE80211_NODE_LOCK(nt); ni = ieee80211_find_node_locked(nt, mac); IEEE80211_NODE_UNLOCK(nt); /* * Don't do the node update inside the node * table lock. This unfortunately causes LORs * with drivers and their TX paths. */ if (ni != NULL) { domlme(mlmeop, ni); ieee80211_free_node(ni); } else error = ENOENT; } else { ieee80211_iterate_nodes(nt, domlme, mlmeop); } return error; } static int setmlme_common(struct ieee80211vap *vap, int op, const uint8_t mac[IEEE80211_ADDR_LEN], int reason) { struct ieee80211com *ic = vap->iv_ic; struct ieee80211_node_table *nt = &ic->ic_sta; struct ieee80211_node *ni; struct mlmeop mlmeop; int error; error = 0; switch (op) { case IEEE80211_MLME_DISASSOC: case IEEE80211_MLME_DEAUTH: switch (vap->iv_opmode) { case IEEE80211_M_STA: mlmedebug(vap, vap->iv_bss->ni_macaddr, op, reason); /* XXX not quite right */ ieee80211_new_state(vap, IEEE80211_S_INIT, reason); break; case IEEE80211_M_HOSTAP: mlmeop.vap = vap; mlmeop.op = op; mlmeop.reason = reason; error = setmlme_dropsta(vap, mac, &mlmeop); break; case IEEE80211_M_WDS: /* XXX user app should send raw frame? */ if (op != IEEE80211_MLME_DEAUTH) { error = EINVAL; break; } #if 0 /* XXX accept any address, simplifies user code */ if (!IEEE80211_ADDR_EQ(mac, vap->iv_bss->ni_macaddr)) { error = EINVAL; break; } #endif mlmedebug(vap, vap->iv_bss->ni_macaddr, op, reason); ni = ieee80211_ref_node(vap->iv_bss); IEEE80211_SEND_MGMT(ni, IEEE80211_FC0_SUBTYPE_DEAUTH, reason); ieee80211_free_node(ni); break; case IEEE80211_M_MBSS: IEEE80211_NODE_LOCK(nt); ni = ieee80211_find_node_locked(nt, mac); /* * Don't do the node update inside the node * table lock. This unfortunately causes LORs * with drivers and their TX paths. */ IEEE80211_NODE_UNLOCK(nt); if (ni != NULL) { ieee80211_node_leave(ni); ieee80211_free_node(ni); } else { error = ENOENT; } break; default: error = EINVAL; break; } break; case IEEE80211_MLME_AUTHORIZE: case IEEE80211_MLME_UNAUTHORIZE: if (vap->iv_opmode != IEEE80211_M_HOSTAP && vap->iv_opmode != IEEE80211_M_WDS) { error = EINVAL; break; } IEEE80211_NODE_LOCK(nt); ni = ieee80211_find_vap_node_locked(nt, vap, mac); /* * Don't do the node update inside the node * table lock. This unfortunately causes LORs * with drivers and their TX paths. */ IEEE80211_NODE_UNLOCK(nt); if (ni != NULL) { mlmedebug(vap, mac, op, reason); if (op == IEEE80211_MLME_AUTHORIZE) ieee80211_node_authorize(ni); else ieee80211_node_unauthorize(ni); ieee80211_free_node(ni); } else error = ENOENT; break; case IEEE80211_MLME_AUTH: if (vap->iv_opmode != IEEE80211_M_HOSTAP) { error = EINVAL; break; } IEEE80211_NODE_LOCK(nt); ni = ieee80211_find_vap_node_locked(nt, vap, mac); /* * Don't do the node update inside the node * table lock. This unfortunately causes LORs * with drivers and their TX paths. */ IEEE80211_NODE_UNLOCK(nt); if (ni != NULL) { mlmedebug(vap, mac, op, reason); if (reason == IEEE80211_STATUS_SUCCESS) { IEEE80211_SEND_MGMT(ni, IEEE80211_FC0_SUBTYPE_AUTH, 2); /* * For shared key auth, just continue the * exchange. Otherwise when 802.1x is not in * use mark the port authorized at this point * so traffic can flow. */ if (ni->ni_authmode != IEEE80211_AUTH_8021X && ni->ni_challenge == NULL) ieee80211_node_authorize(ni); } else { vap->iv_stats.is_rx_acl++; ieee80211_send_error(ni, ni->ni_macaddr, IEEE80211_FC0_SUBTYPE_AUTH, 2|(reason<<16)); ieee80211_node_leave(ni); } ieee80211_free_node(ni); } else error = ENOENT; break; default: error = EINVAL; break; } return error; } struct scanlookup { const uint8_t *mac; int esslen; const uint8_t *essid; const struct ieee80211_scan_entry *se; }; /* * Match mac address and any ssid. */ static void mlmelookup(void *arg, const struct ieee80211_scan_entry *se) { struct scanlookup *look = arg; if (!IEEE80211_ADDR_EQ(look->mac, se->se_macaddr)) return; if (look->esslen != 0) { if (se->se_ssid[1] != look->esslen) return; if (memcmp(look->essid, se->se_ssid+2, look->esslen)) return; } look->se = se; } static int setmlme_assoc_sta(struct ieee80211vap *vap, const uint8_t mac[IEEE80211_ADDR_LEN], int ssid_len, const uint8_t ssid[IEEE80211_NWID_LEN]) { struct scanlookup lookup; KASSERT(vap->iv_opmode == IEEE80211_M_STA, ("expected opmode STA not %s", ieee80211_opmode_name[vap->iv_opmode])); /* NB: this is racey if roaming is !manual */ lookup.se = NULL; lookup.mac = mac; lookup.esslen = ssid_len; lookup.essid = ssid; ieee80211_scan_iterate(vap, mlmelookup, &lookup); if (lookup.se == NULL) return ENOENT; mlmedebug(vap, mac, IEEE80211_MLME_ASSOC, 0); if (!ieee80211_sta_join(vap, lookup.se->se_chan, lookup.se)) return EIO; /* XXX unique but could be better */ return 0; } static int setmlme_assoc_adhoc(struct ieee80211vap *vap, const uint8_t mac[IEEE80211_ADDR_LEN], int ssid_len, const uint8_t ssid[IEEE80211_NWID_LEN]) { struct ieee80211_scan_req *sr; int error; KASSERT(vap->iv_opmode == IEEE80211_M_IBSS || vap->iv_opmode == IEEE80211_M_AHDEMO, ("expected opmode IBSS or AHDEMO not %s", ieee80211_opmode_name[vap->iv_opmode])); if (ssid_len == 0) return EINVAL; sr = IEEE80211_MALLOC(sizeof(*sr), M_TEMP, IEEE80211_M_NOWAIT | IEEE80211_M_ZERO); if (sr == NULL) return ENOMEM; /* NB: IEEE80211_IOC_SSID call missing for ap_scan=2. */ memset(vap->iv_des_ssid[0].ssid, 0, IEEE80211_NWID_LEN); vap->iv_des_ssid[0].len = ssid_len; memcpy(vap->iv_des_ssid[0].ssid, ssid, ssid_len); vap->iv_des_nssid = 1; sr->sr_flags = IEEE80211_IOC_SCAN_ACTIVE | IEEE80211_IOC_SCAN_ONCE; sr->sr_duration = IEEE80211_IOC_SCAN_FOREVER; memcpy(sr->sr_ssid[0].ssid, ssid, ssid_len); sr->sr_ssid[0].len = ssid_len; sr->sr_nssid = 1; error = ieee80211_scanreq(vap, sr); IEEE80211_FREE(sr, M_TEMP); return error; } static int ieee80211_ioctl_setmlme(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211req_mlme mlme; int error; if (ireq->i_len != sizeof(mlme)) return EINVAL; error = copyin(ireq->i_data, &mlme, sizeof(mlme)); if (error) return error; if (vap->iv_opmode == IEEE80211_M_STA && mlme.im_op == IEEE80211_MLME_ASSOC) return setmlme_assoc_sta(vap, mlme.im_macaddr, vap->iv_des_ssid[0].len, vap->iv_des_ssid[0].ssid); else if ((vap->iv_opmode == IEEE80211_M_IBSS || vap->iv_opmode == IEEE80211_M_AHDEMO) && mlme.im_op == IEEE80211_MLME_ASSOC) return setmlme_assoc_adhoc(vap, mlme.im_macaddr, mlme.im_ssid_len, mlme.im_ssid); else return setmlme_common(vap, mlme.im_op, mlme.im_macaddr, mlme.im_reason); } static int ieee80211_ioctl_macmac(struct ieee80211vap *vap, struct ieee80211req *ireq) { uint8_t mac[IEEE80211_ADDR_LEN]; const struct ieee80211_aclator *acl = vap->iv_acl; int error; if (ireq->i_len != sizeof(mac)) return EINVAL; error = copyin(ireq->i_data, mac, ireq->i_len); if (error) return error; if (acl == NULL) { acl = ieee80211_aclator_get("mac"); if (acl == NULL || !acl->iac_attach(vap)) return EINVAL; vap->iv_acl = acl; } if (ireq->i_type == IEEE80211_IOC_ADDMAC) acl->iac_add(vap, mac); else acl->iac_remove(vap, mac); return 0; } static int ieee80211_ioctl_setmaccmd(struct ieee80211vap *vap, struct ieee80211req *ireq) { const struct ieee80211_aclator *acl = vap->iv_acl; switch (ireq->i_val) { case IEEE80211_MACCMD_POLICY_OPEN: case IEEE80211_MACCMD_POLICY_ALLOW: case IEEE80211_MACCMD_POLICY_DENY: case IEEE80211_MACCMD_POLICY_RADIUS: if (acl == NULL) { acl = ieee80211_aclator_get("mac"); if (acl == NULL || !acl->iac_attach(vap)) return EINVAL; vap->iv_acl = acl; } acl->iac_setpolicy(vap, ireq->i_val); break; case IEEE80211_MACCMD_FLUSH: if (acl != NULL) acl->iac_flush(vap); /* NB: silently ignore when not in use */ break; case IEEE80211_MACCMD_DETACH: if (acl != NULL) { vap->iv_acl = NULL; acl->iac_detach(vap); } break; default: if (acl == NULL) return EINVAL; else return acl->iac_setioctl(vap, ireq); } return 0; } static int ieee80211_ioctl_setchanlist(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211com *ic = vap->iv_ic; uint8_t *chanlist, *list; int i, nchan, maxchan, error; if (ireq->i_len > sizeof(ic->ic_chan_active)) ireq->i_len = sizeof(ic->ic_chan_active); list = IEEE80211_MALLOC(ireq->i_len + IEEE80211_CHAN_BYTES, M_TEMP, IEEE80211_M_NOWAIT | IEEE80211_M_ZERO); if (list == NULL) return ENOMEM; error = copyin(ireq->i_data, list, ireq->i_len); if (error) { IEEE80211_FREE(list, M_TEMP); return error; } nchan = 0; chanlist = list + ireq->i_len; /* NB: zero'd already */ maxchan = ireq->i_len * NBBY; for (i = 0; i < ic->ic_nchans; i++) { const struct ieee80211_channel *c = &ic->ic_channels[i]; /* * Calculate the intersection of the user list and the * available channels so users can do things like specify * 1-255 to get all available channels. */ if (c->ic_ieee < maxchan && isset(list, c->ic_ieee)) { setbit(chanlist, c->ic_ieee); nchan++; } } if (nchan == 0) { IEEE80211_FREE(list, M_TEMP); return EINVAL; } if (ic->ic_bsschan != IEEE80211_CHAN_ANYC && /* XXX */ isclr(chanlist, ic->ic_bsschan->ic_ieee)) ic->ic_bsschan = IEEE80211_CHAN_ANYC; memcpy(ic->ic_chan_active, chanlist, IEEE80211_CHAN_BYTES); ieee80211_scan_flush(vap); IEEE80211_FREE(list, M_TEMP); return ENETRESET; } static int ieee80211_ioctl_setstastats(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211_node *ni; uint8_t macaddr[IEEE80211_ADDR_LEN]; int error; /* * NB: we could copyin ieee80211req_sta_stats so apps * could make selective changes but that's overkill; * just clear all stats for now. */ if (ireq->i_len < IEEE80211_ADDR_LEN) return EINVAL; error = copyin(ireq->i_data, macaddr, IEEE80211_ADDR_LEN); if (error != 0) return error; ni = ieee80211_find_vap_node(&vap->iv_ic->ic_sta, vap, macaddr); if (ni == NULL) return ENOENT; /* XXX require ni_vap == vap? */ memset(&ni->ni_stats, 0, sizeof(ni->ni_stats)); ieee80211_free_node(ni); return 0; } static int ieee80211_ioctl_setstatxpow(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211_node *ni; struct ieee80211req_sta_txpow txpow; int error; if (ireq->i_len != sizeof(txpow)) return EINVAL; error = copyin(ireq->i_data, &txpow, sizeof(txpow)); if (error != 0) return error; ni = ieee80211_find_vap_node(&vap->iv_ic->ic_sta, vap, txpow.it_macaddr); if (ni == NULL) return ENOENT; ni->ni_txpower = txpow.it_txpow; ieee80211_free_node(ni); return error; } static int ieee80211_ioctl_setwmeparam(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211com *ic = vap->iv_ic; struct ieee80211_wme_state *wme = &ic->ic_wme; struct wmeParams *wmep, *chanp; int isbss, ac, aggrmode; if ((ic->ic_caps & IEEE80211_C_WME) == 0) return EOPNOTSUPP; isbss = (ireq->i_len & IEEE80211_WMEPARAM_BSS); ac = (ireq->i_len & IEEE80211_WMEPARAM_VAL); aggrmode = (wme->wme_flags & WME_F_AGGRMODE); if (ac >= WME_NUM_AC) ac = WME_AC_BE; if (isbss) { chanp = &wme->wme_bssChanParams.cap_wmeParams[ac]; wmep = &wme->wme_wmeBssChanParams.cap_wmeParams[ac]; } else { chanp = &wme->wme_chanParams.cap_wmeParams[ac]; wmep = &wme->wme_wmeChanParams.cap_wmeParams[ac]; } switch (ireq->i_type) { case IEEE80211_IOC_WME_CWMIN: /* WME: CWmin */ wmep->wmep_logcwmin = ireq->i_val; if (!isbss || !aggrmode) chanp->wmep_logcwmin = ireq->i_val; break; case IEEE80211_IOC_WME_CWMAX: /* WME: CWmax */ wmep->wmep_logcwmax = ireq->i_val; if (!isbss || !aggrmode) chanp->wmep_logcwmax = ireq->i_val; break; case IEEE80211_IOC_WME_AIFS: /* WME: AIFS */ wmep->wmep_aifsn = ireq->i_val; if (!isbss || !aggrmode) chanp->wmep_aifsn = ireq->i_val; break; case IEEE80211_IOC_WME_TXOPLIMIT: /* WME: txops limit */ wmep->wmep_txopLimit = ireq->i_val; if (!isbss || !aggrmode) chanp->wmep_txopLimit = ireq->i_val; break; case IEEE80211_IOC_WME_ACM: /* WME: ACM (bss only) */ wmep->wmep_acm = ireq->i_val; if (!aggrmode) chanp->wmep_acm = ireq->i_val; break; case IEEE80211_IOC_WME_ACKPOLICY: /* WME: ACK policy (!bss only)*/ wmep->wmep_noackPolicy = chanp->wmep_noackPolicy = (ireq->i_val) == 0; break; } ieee80211_wme_updateparams(vap); return 0; } static int find11gchannel(struct ieee80211com *ic, int start, int freq) { const struct ieee80211_channel *c; int i; for (i = start+1; i < ic->ic_nchans; i++) { c = &ic->ic_channels[i]; if (c->ic_freq == freq && IEEE80211_IS_CHAN_ANYG(c)) return 1; } /* NB: should not be needed but in case things are mis-sorted */ for (i = 0; i < start; i++) { c = &ic->ic_channels[i]; if (c->ic_freq == freq && IEEE80211_IS_CHAN_ANYG(c)) return 1; } return 0; } static struct ieee80211_channel * findchannel(struct ieee80211com *ic, int ieee, int mode) { static const u_int chanflags[IEEE80211_MODE_MAX] = { [IEEE80211_MODE_AUTO] = 0, [IEEE80211_MODE_11A] = IEEE80211_CHAN_A, [IEEE80211_MODE_11B] = IEEE80211_CHAN_B, [IEEE80211_MODE_11G] = IEEE80211_CHAN_G, [IEEE80211_MODE_FH] = IEEE80211_CHAN_FHSS, [IEEE80211_MODE_TURBO_A] = IEEE80211_CHAN_108A, [IEEE80211_MODE_TURBO_G] = IEEE80211_CHAN_108G, [IEEE80211_MODE_STURBO_A] = IEEE80211_CHAN_STURBO, [IEEE80211_MODE_HALF] = IEEE80211_CHAN_HALF, [IEEE80211_MODE_QUARTER] = IEEE80211_CHAN_QUARTER, /* NB: handled specially below */ [IEEE80211_MODE_11NA] = IEEE80211_CHAN_A, [IEEE80211_MODE_11NG] = IEEE80211_CHAN_G, [IEEE80211_MODE_VHT_5GHZ] = IEEE80211_CHAN_A, [IEEE80211_MODE_VHT_2GHZ] = IEEE80211_CHAN_G, }; u_int modeflags; int i; modeflags = chanflags[mode]; for (i = 0; i < ic->ic_nchans; i++) { struct ieee80211_channel *c = &ic->ic_channels[i]; if (c->ic_ieee != ieee) continue; if (mode == IEEE80211_MODE_AUTO) { /* ignore turbo channels for autoselect */ if (IEEE80211_IS_CHAN_TURBO(c)) continue; /* * XXX special-case 11b/g channels so we * always select the g channel if both * are present. * XXX prefer HT to non-HT? */ if (!IEEE80211_IS_CHAN_B(c) || !find11gchannel(ic, i, c->ic_freq)) return c; } else { /* must check VHT specifically */ if ((mode == IEEE80211_MODE_VHT_5GHZ || mode == IEEE80211_MODE_VHT_2GHZ) && !IEEE80211_IS_CHAN_VHT(c)) continue; /* * Must check HT specially - only match on HT, * not HT+VHT channels */ if ((mode == IEEE80211_MODE_11NA || mode == IEEE80211_MODE_11NG) && !IEEE80211_IS_CHAN_HT(c)) continue; if ((mode == IEEE80211_MODE_11NA || mode == IEEE80211_MODE_11NG) && IEEE80211_IS_CHAN_VHT(c)) continue; /* Check that the modeflags above match */ if ((c->ic_flags & modeflags) == modeflags) return c; } } return NULL; } /* * Check the specified against any desired mode (aka netband). * This is only used (presently) when operating in hostap mode * to enforce consistency. */ static int check_mode_consistency(const struct ieee80211_channel *c, int mode) { KASSERT(c != IEEE80211_CHAN_ANYC, ("oops, no channel")); switch (mode) { case IEEE80211_MODE_11B: return (IEEE80211_IS_CHAN_B(c)); case IEEE80211_MODE_11G: return (IEEE80211_IS_CHAN_ANYG(c) && !IEEE80211_IS_CHAN_HT(c)); case IEEE80211_MODE_11A: return (IEEE80211_IS_CHAN_A(c) && !IEEE80211_IS_CHAN_HT(c)); case IEEE80211_MODE_STURBO_A: return (IEEE80211_IS_CHAN_STURBO(c)); case IEEE80211_MODE_11NA: return (IEEE80211_IS_CHAN_HTA(c)); case IEEE80211_MODE_11NG: return (IEEE80211_IS_CHAN_HTG(c)); } return 1; } /* * Common code to set the current channel. If the device * is up and running this may result in an immediate channel * change or a kick of the state machine. */ static int setcurchan(struct ieee80211vap *vap, struct ieee80211_channel *c) { struct ieee80211com *ic = vap->iv_ic; int error; if (c != IEEE80211_CHAN_ANYC) { if (IEEE80211_IS_CHAN_RADAR(c)) return EBUSY; /* XXX better code? */ if (vap->iv_opmode == IEEE80211_M_HOSTAP) { if (IEEE80211_IS_CHAN_NOHOSTAP(c)) return EINVAL; if (!check_mode_consistency(c, vap->iv_des_mode)) return EINVAL; } else if (vap->iv_opmode == IEEE80211_M_IBSS) { if (IEEE80211_IS_CHAN_NOADHOC(c)) return EINVAL; } if ((vap->iv_state == IEEE80211_S_RUN || vap->iv_state == IEEE80211_S_SLEEP) && vap->iv_bss->ni_chan == c) return 0; /* NB: nothing to do */ } vap->iv_des_chan = c; error = 0; if (vap->iv_opmode == IEEE80211_M_MONITOR && vap->iv_des_chan != IEEE80211_CHAN_ANYC) { /* * Monitor mode can switch directly. */ if (IFNET_IS_UP_RUNNING(vap->iv_ifp)) { /* XXX need state machine for other vap's to follow */ ieee80211_setcurchan(ic, vap->iv_des_chan); vap->iv_bss->ni_chan = ic->ic_curchan; } else { ic->ic_curchan = vap->iv_des_chan; ic->ic_rt = ieee80211_get_ratetable(ic->ic_curchan); } } else { /* * Need to go through the state machine in case we * need to reassociate or the like. The state machine * will pickup the desired channel and avoid scanning. */ if (IS_UP_AUTO(vap)) ieee80211_new_state(vap, IEEE80211_S_SCAN, 0); else if (vap->iv_des_chan != IEEE80211_CHAN_ANYC) { /* * When not up+running and a real channel has * been specified fix the current channel so * there is immediate feedback; e.g. via ifconfig. */ ic->ic_curchan = vap->iv_des_chan; ic->ic_rt = ieee80211_get_ratetable(ic->ic_curchan); } } return error; } /* * Old api for setting the current channel; this is * deprecated because channel numbers are ambiguous. */ static int ieee80211_ioctl_setchannel(struct ieee80211vap *vap, const struct ieee80211req *ireq) { struct ieee80211com *ic = vap->iv_ic; struct ieee80211_channel *c; /* XXX 0xffff overflows 16-bit signed */ if (ireq->i_val == 0 || ireq->i_val == (int16_t) IEEE80211_CHAN_ANY) { c = IEEE80211_CHAN_ANYC; } else { struct ieee80211_channel *c2; c = findchannel(ic, ireq->i_val, vap->iv_des_mode); if (c == NULL) { c = findchannel(ic, ireq->i_val, IEEE80211_MODE_AUTO); if (c == NULL) return EINVAL; } /* * Fine tune channel selection based on desired mode: * if 11b is requested, find the 11b version of any * 11g channel returned, * if static turbo, find the turbo version of any * 11a channel return, * if 11na is requested, find the ht version of any * 11a channel returned, * if 11ng is requested, find the ht version of any * 11g channel returned, * if 11ac is requested, find the 11ac version * of any 11a/11na channel returned, * (TBD) 11acg (2GHz VHT) * otherwise we should be ok with what we've got. */ switch (vap->iv_des_mode) { case IEEE80211_MODE_11B: if (IEEE80211_IS_CHAN_ANYG(c)) { c2 = findchannel(ic, ireq->i_val, IEEE80211_MODE_11B); /* NB: should not happen, =>'s 11g w/o 11b */ if (c2 != NULL) c = c2; } break; case IEEE80211_MODE_TURBO_A: if (IEEE80211_IS_CHAN_A(c)) { c2 = findchannel(ic, ireq->i_val, IEEE80211_MODE_TURBO_A); if (c2 != NULL) c = c2; } break; case IEEE80211_MODE_11NA: if (IEEE80211_IS_CHAN_A(c)) { c2 = findchannel(ic, ireq->i_val, IEEE80211_MODE_11NA); if (c2 != NULL) c = c2; } break; case IEEE80211_MODE_11NG: if (IEEE80211_IS_CHAN_ANYG(c)) { c2 = findchannel(ic, ireq->i_val, IEEE80211_MODE_11NG); if (c2 != NULL) c = c2; } break; case IEEE80211_MODE_VHT_2GHZ: printf("%s: TBD\n", __func__); break; case IEEE80211_MODE_VHT_5GHZ: if (IEEE80211_IS_CHAN_A(c)) { c2 = findchannel(ic, ireq->i_val, IEEE80211_MODE_VHT_5GHZ); if (c2 != NULL) c = c2; } break; default: /* NB: no static turboG */ break; } } return setcurchan(vap, c); } /* * New/current api for setting the current channel; a complete * channel description is provide so there is no ambiguity in * identifying the channel. */ static int ieee80211_ioctl_setcurchan(struct ieee80211vap *vap, const struct ieee80211req *ireq) { struct ieee80211com *ic = vap->iv_ic; struct ieee80211_channel chan, *c; int error; if (ireq->i_len != sizeof(chan)) return EINVAL; error = copyin(ireq->i_data, &chan, sizeof(chan)); if (error != 0) return error; /* XXX 0xffff overflows 16-bit signed */ if (chan.ic_freq == 0 || chan.ic_freq == IEEE80211_CHAN_ANY) { c = IEEE80211_CHAN_ANYC; } else { c = ieee80211_find_channel(ic, chan.ic_freq, chan.ic_flags); if (c == NULL) return EINVAL; } return setcurchan(vap, c); } static int ieee80211_ioctl_setregdomain(struct ieee80211vap *vap, const struct ieee80211req *ireq) { struct ieee80211_regdomain_req *reg; int nchans, error; nchans = 1 + ((ireq->i_len - sizeof(struct ieee80211_regdomain_req)) / sizeof(struct ieee80211_channel)); if (!(1 <= nchans && nchans <= IEEE80211_CHAN_MAX)) { IEEE80211_DPRINTF(vap, IEEE80211_MSG_IOCTL, "%s: bad # chans, i_len %d nchans %d\n", __func__, ireq->i_len, nchans); return EINVAL; } reg = (struct ieee80211_regdomain_req *) IEEE80211_MALLOC(IEEE80211_REGDOMAIN_SIZE(nchans), M_TEMP, IEEE80211_M_NOWAIT | IEEE80211_M_ZERO); if (reg == NULL) { IEEE80211_DPRINTF(vap, IEEE80211_MSG_IOCTL, "%s: no memory, nchans %d\n", __func__, nchans); return ENOMEM; } error = copyin(ireq->i_data, reg, IEEE80211_REGDOMAIN_SIZE(nchans)); if (error == 0) { /* NB: validate inline channel count against storage size */ if (reg->chaninfo.ic_nchans != nchans) { IEEE80211_DPRINTF(vap, IEEE80211_MSG_IOCTL, "%s: chan cnt mismatch, %d != %d\n", __func__, reg->chaninfo.ic_nchans, nchans); error = EINVAL; } else error = ieee80211_setregdomain(vap, reg); } IEEE80211_FREE(reg, M_TEMP); return (error == 0 ? ENETRESET : error); } static int checkrate(const struct ieee80211_rateset *rs, int rate) { int i; if (rate == IEEE80211_FIXED_RATE_NONE) return 1; for (i = 0; i < rs->rs_nrates; i++) if ((rs->rs_rates[i] & IEEE80211_RATE_VAL) == rate) return 1; return 0; } static int checkmcs(const struct ieee80211_htrateset *rs, int mcs) { int rate_val = IEEE80211_RV(mcs); int i; if (mcs == IEEE80211_FIXED_RATE_NONE) return 1; if ((mcs & IEEE80211_RATE_MCS) == 0) /* MCS always have 0x80 set */ return 0; for (i = 0; i < rs->rs_nrates; i++) if (IEEE80211_RV(rs->rs_rates[i]) == rate_val) return 1; return 0; } static int ieee80211_ioctl_setroam(struct ieee80211vap *vap, const struct ieee80211req *ireq) { struct ieee80211com *ic = vap->iv_ic; struct ieee80211_roamparams_req *parms; struct ieee80211_roamparam *src, *dst; const struct ieee80211_htrateset *rs_ht; const struct ieee80211_rateset *rs; int changed, error, mode, is11n, nmodes; if (ireq->i_len != sizeof(vap->iv_roamparms)) return EINVAL; parms = IEEE80211_MALLOC(sizeof(*parms), M_TEMP, IEEE80211_M_NOWAIT | IEEE80211_M_ZERO); if (parms == NULL) return ENOMEM; error = copyin(ireq->i_data, parms, ireq->i_len); if (error != 0) goto fail; changed = 0; nmodes = IEEE80211_MODE_MAX; /* validate parameters and check if anything changed */ for (mode = IEEE80211_MODE_11A; mode < nmodes; mode++) { if (isclr(ic->ic_modecaps, mode)) continue; src = &parms->params[mode]; dst = &vap->iv_roamparms[mode]; rs = &ic->ic_sup_rates[mode]; /* NB: 11n maps to legacy */ rs_ht = &ic->ic_sup_htrates; is11n = (mode == IEEE80211_MODE_11NA || mode == IEEE80211_MODE_11NG); /* XXX TODO: 11ac */ if (src->rate != dst->rate) { if (!checkrate(rs, src->rate) && (!is11n || !checkmcs(rs_ht, src->rate))) { error = EINVAL; goto fail; } changed++; } if (src->rssi != dst->rssi) changed++; } if (changed) { /* * Copy new parameters in place and notify the * driver so it can push state to the device. */ /* XXX locking? */ for (mode = IEEE80211_MODE_11A; mode < nmodes; mode++) { if (isset(ic->ic_modecaps, mode)) vap->iv_roamparms[mode] = parms->params[mode]; } if (vap->iv_roaming == IEEE80211_ROAMING_DEVICE) error = ERESTART; } fail: IEEE80211_FREE(parms, M_TEMP); return error; } static int ieee80211_ioctl_settxparams(struct ieee80211vap *vap, const struct ieee80211req *ireq) { struct ieee80211com *ic = vap->iv_ic; struct ieee80211_txparams_req parms; /* XXX stack use? */ struct ieee80211_txparam *src, *dst; const struct ieee80211_htrateset *rs_ht; const struct ieee80211_rateset *rs; int error, mode, changed, is11n, nmodes; /* NB: accept short requests for backwards compat */ if (ireq->i_len > sizeof(parms)) return EINVAL; error = copyin(ireq->i_data, &parms, ireq->i_len); if (error != 0) return error; nmodes = ireq->i_len / sizeof(struct ieee80211_txparam); changed = 0; /* validate parameters and check if anything changed */ for (mode = IEEE80211_MODE_11A; mode < nmodes; mode++) { if (isclr(ic->ic_modecaps, mode)) continue; src = &parms.params[mode]; dst = &vap->iv_txparms[mode]; rs = &ic->ic_sup_rates[mode]; /* NB: 11n maps to legacy */ rs_ht = &ic->ic_sup_htrates; is11n = (mode == IEEE80211_MODE_11NA || mode == IEEE80211_MODE_11NG); if (src->ucastrate != dst->ucastrate) { if (!checkrate(rs, src->ucastrate) && (!is11n || !checkmcs(rs_ht, src->ucastrate))) return EINVAL; changed++; } if (src->mcastrate != dst->mcastrate) { if (!checkrate(rs, src->mcastrate) && (!is11n || !checkmcs(rs_ht, src->mcastrate))) return EINVAL; changed++; } if (src->mgmtrate != dst->mgmtrate) { if (!checkrate(rs, src->mgmtrate) && (!is11n || !checkmcs(rs_ht, src->mgmtrate))) return EINVAL; changed++; } if (src->maxretry != dst->maxretry) /* NB: no bounds */ changed++; } if (changed) { /* * Copy new parameters in place and notify the * driver so it can push state to the device. */ for (mode = IEEE80211_MODE_11A; mode < nmodes; mode++) { if (isset(ic->ic_modecaps, mode)) vap->iv_txparms[mode] = parms.params[mode]; } /* XXX could be more intelligent, e.g. don't reset if setting not being used */ return ENETRESET; } return 0; } /* * Application Information Element support. */ static int setappie(struct ieee80211_appie **aie, const struct ieee80211req *ireq) { struct ieee80211_appie *app = *aie; struct ieee80211_appie *napp; int error; if (ireq->i_len == 0) { /* delete any existing ie */ if (app != NULL) { *aie = NULL; /* XXX racey */ IEEE80211_FREE(app, M_80211_NODE_IE); } return 0; } if (!(2 <= ireq->i_len && ireq->i_len <= IEEE80211_MAX_APPIE)) return EINVAL; /* * Allocate a new appie structure and copy in the user data. * When done swap in the new structure. Note that we do not * guard against users holding a ref to the old structure; * this must be handled outside this code. * * XXX bad bad bad */ napp = (struct ieee80211_appie *) IEEE80211_MALLOC( sizeof(struct ieee80211_appie) + ireq->i_len, M_80211_NODE_IE, IEEE80211_M_NOWAIT); if (napp == NULL) return ENOMEM; /* XXX holding ic lock */ error = copyin(ireq->i_data, napp->ie_data, ireq->i_len); if (error) { IEEE80211_FREE(napp, M_80211_NODE_IE); return error; } napp->ie_len = ireq->i_len; *aie = napp; if (app != NULL) IEEE80211_FREE(app, M_80211_NODE_IE); return 0; } static void setwparsnie(struct ieee80211vap *vap, uint8_t *ie, int space) { /* validate data is present as best we can */ if (space == 0 || 2+ie[1] > space) return; if (ie[0] == IEEE80211_ELEMID_VENDOR) vap->iv_wpa_ie = ie; else if (ie[0] == IEEE80211_ELEMID_RSN) vap->iv_rsn_ie = ie; } static int ieee80211_ioctl_setappie_locked(struct ieee80211vap *vap, const struct ieee80211req *ireq, int fc0) { int error; IEEE80211_LOCK_ASSERT(vap->iv_ic); switch (fc0 & IEEE80211_FC0_SUBTYPE_MASK) { case IEEE80211_FC0_SUBTYPE_BEACON: if (vap->iv_opmode != IEEE80211_M_HOSTAP && vap->iv_opmode != IEEE80211_M_IBSS) { error = EINVAL; break; } error = setappie(&vap->iv_appie_beacon, ireq); if (error == 0) ieee80211_beacon_notify(vap, IEEE80211_BEACON_APPIE); break; case IEEE80211_FC0_SUBTYPE_PROBE_RESP: error = setappie(&vap->iv_appie_proberesp, ireq); break; case IEEE80211_FC0_SUBTYPE_ASSOC_RESP: if (vap->iv_opmode == IEEE80211_M_HOSTAP) error = setappie(&vap->iv_appie_assocresp, ireq); else error = EINVAL; break; case IEEE80211_FC0_SUBTYPE_PROBE_REQ: error = setappie(&vap->iv_appie_probereq, ireq); break; case IEEE80211_FC0_SUBTYPE_ASSOC_REQ: if (vap->iv_opmode == IEEE80211_M_STA) error = setappie(&vap->iv_appie_assocreq, ireq); else error = EINVAL; break; case (IEEE80211_APPIE_WPA & IEEE80211_FC0_SUBTYPE_MASK): error = setappie(&vap->iv_appie_wpa, ireq); if (error == 0) { /* * Must split single blob of data into separate * WPA and RSN ie's because they go in different * locations in the mgt frames. * XXX use IEEE80211_IOC_WPA2 so user code does split */ vap->iv_wpa_ie = NULL; vap->iv_rsn_ie = NULL; if (vap->iv_appie_wpa != NULL) { struct ieee80211_appie *appie = vap->iv_appie_wpa; uint8_t *data = appie->ie_data; /* XXX ie length validate is painful, cheat */ setwparsnie(vap, data, appie->ie_len); setwparsnie(vap, data + 2 + data[1], appie->ie_len - (2 + data[1])); } if (vap->iv_opmode == IEEE80211_M_HOSTAP || vap->iv_opmode == IEEE80211_M_IBSS) { /* * Must rebuild beacon frame as the update * mechanism doesn't handle WPA/RSN ie's. * Could extend it but it doesn't normally * change; this is just to deal with hostapd * plumbing the ie after the interface is up. */ error = ENETRESET; } } break; default: error = EINVAL; break; } return error; } static int ieee80211_ioctl_setappie(struct ieee80211vap *vap, const struct ieee80211req *ireq) { struct ieee80211com *ic = vap->iv_ic; int error; uint8_t fc0; fc0 = ireq->i_val & 0xff; if ((fc0 & IEEE80211_FC0_TYPE_MASK) != IEEE80211_FC0_TYPE_MGT) return EINVAL; /* NB: could check iv_opmode and reject but hardly worth the effort */ IEEE80211_LOCK(ic); error = ieee80211_ioctl_setappie_locked(vap, ireq, fc0); IEEE80211_UNLOCK(ic); return error; } static int ieee80211_ioctl_chanswitch(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211com *ic = vap->iv_ic; struct ieee80211_chanswitch_req csr; struct ieee80211_channel *c; int error; if (ireq->i_len != sizeof(csr)) return EINVAL; error = copyin(ireq->i_data, &csr, sizeof(csr)); if (error != 0) return error; /* XXX adhoc mode not supported */ if (vap->iv_opmode != IEEE80211_M_HOSTAP || (vap->iv_flags & IEEE80211_F_DOTH) == 0) return EOPNOTSUPP; c = ieee80211_find_channel(ic, csr.csa_chan.ic_freq, csr.csa_chan.ic_flags); if (c == NULL) return ENOENT; IEEE80211_LOCK(ic); if ((ic->ic_flags & IEEE80211_F_CSAPENDING) == 0) ieee80211_csa_startswitch(ic, c, csr.csa_mode, csr.csa_count); else if (csr.csa_count == 0) ieee80211_csa_cancelswitch(ic); else error = EBUSY; IEEE80211_UNLOCK(ic); return error; } static int ieee80211_scanreq(struct ieee80211vap *vap, struct ieee80211_scan_req *sr) { #define IEEE80211_IOC_SCAN_FLAGS \ (IEEE80211_IOC_SCAN_NOPICK | IEEE80211_IOC_SCAN_ACTIVE | \ IEEE80211_IOC_SCAN_PICK1ST | IEEE80211_IOC_SCAN_BGSCAN | \ IEEE80211_IOC_SCAN_ONCE | IEEE80211_IOC_SCAN_NOBCAST | \ IEEE80211_IOC_SCAN_NOJOIN | IEEE80211_IOC_SCAN_FLUSH | \ IEEE80211_IOC_SCAN_CHECK) struct ieee80211com *ic = vap->iv_ic; int error, i; /* convert duration */ if (sr->sr_duration == IEEE80211_IOC_SCAN_FOREVER) sr->sr_duration = IEEE80211_SCAN_FOREVER; else { if (sr->sr_duration < IEEE80211_IOC_SCAN_DURATION_MIN || sr->sr_duration > IEEE80211_IOC_SCAN_DURATION_MAX) return EINVAL; sr->sr_duration = msecs_to_ticks(sr->sr_duration); } /* convert min/max channel dwell */ if (sr->sr_mindwell != 0) sr->sr_mindwell = msecs_to_ticks(sr->sr_mindwell); if (sr->sr_maxdwell != 0) sr->sr_maxdwell = msecs_to_ticks(sr->sr_maxdwell); /* NB: silently reduce ssid count to what is supported */ if (sr->sr_nssid > IEEE80211_SCAN_MAX_SSID) sr->sr_nssid = IEEE80211_SCAN_MAX_SSID; for (i = 0; i < sr->sr_nssid; i++) if (sr->sr_ssid[i].len > IEEE80211_NWID_LEN) return EINVAL; /* cleanse flags just in case, could reject if invalid flags */ sr->sr_flags &= IEEE80211_IOC_SCAN_FLAGS; /* * Add an implicit NOPICK if the vap is not marked UP. This * allows applications to scan without joining a bss (or picking * a channel and setting up a bss) and without forcing manual * roaming mode--you just need to mark the parent device UP. */ if ((vap->iv_ifp->if_flags & IFF_UP) == 0) sr->sr_flags |= IEEE80211_IOC_SCAN_NOPICK; IEEE80211_DPRINTF(vap, IEEE80211_MSG_SCAN, "%s: flags 0x%x%s duration 0x%x mindwell %u maxdwell %u nssid %d\n", __func__, sr->sr_flags, (vap->iv_ifp->if_flags & IFF_UP) == 0 ? " (!IFF_UP)" : "", sr->sr_duration, sr->sr_mindwell, sr->sr_maxdwell, sr->sr_nssid); /* * If we are in INIT state then the driver has never had a chance * to setup hardware state to do a scan; we must use the state * machine to get us up to the SCAN state but once we reach SCAN * state we then want to use the supplied params. Stash the * parameters in the vap and mark IEEE80211_FEXT_SCANREQ; the * state machines will recognize this and use the stashed params * to issue the scan request. * * Otherwise just invoke the scan machinery directly. */ IEEE80211_LOCK(ic); if (ic->ic_nrunning == 0) { IEEE80211_UNLOCK(ic); return ENXIO; } if (vap->iv_state == IEEE80211_S_INIT) { /* NB: clobbers previous settings */ vap->iv_scanreq_flags = sr->sr_flags; vap->iv_scanreq_duration = sr->sr_duration; vap->iv_scanreq_nssid = sr->sr_nssid; for (i = 0; i < sr->sr_nssid; i++) { vap->iv_scanreq_ssid[i].len = sr->sr_ssid[i].len; memcpy(vap->iv_scanreq_ssid[i].ssid, sr->sr_ssid[i].ssid, sr->sr_ssid[i].len); } vap->iv_flags_ext |= IEEE80211_FEXT_SCANREQ; IEEE80211_UNLOCK(ic); ieee80211_new_state(vap, IEEE80211_S_SCAN, 0); } else { vap->iv_flags_ext &= ~IEEE80211_FEXT_SCANREQ; IEEE80211_UNLOCK(ic); if (sr->sr_flags & IEEE80211_IOC_SCAN_CHECK) { error = ieee80211_check_scan(vap, sr->sr_flags, sr->sr_duration, sr->sr_mindwell, sr->sr_maxdwell, sr->sr_nssid, /* NB: cheat, we assume structures are compatible */ (const struct ieee80211_scan_ssid *) &sr->sr_ssid[0]); } else { error = ieee80211_start_scan(vap, sr->sr_flags, sr->sr_duration, sr->sr_mindwell, sr->sr_maxdwell, sr->sr_nssid, /* NB: cheat, we assume structures are compatible */ (const struct ieee80211_scan_ssid *) &sr->sr_ssid[0]); } if (error == 0) return EINPROGRESS; } return 0; #undef IEEE80211_IOC_SCAN_FLAGS } static int ieee80211_ioctl_scanreq(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211_scan_req *sr; int error; if (ireq->i_len != sizeof(*sr)) return EINVAL; sr = IEEE80211_MALLOC(sizeof(*sr), M_TEMP, IEEE80211_M_NOWAIT | IEEE80211_M_ZERO); if (sr == NULL) return ENOMEM; error = copyin(ireq->i_data, sr, sizeof(*sr)); if (error != 0) goto bad; error = ieee80211_scanreq(vap, sr); bad: IEEE80211_FREE(sr, M_TEMP); return error; } static int ieee80211_ioctl_setstavlan(struct ieee80211vap *vap, struct ieee80211req *ireq) { struct ieee80211_node *ni; struct ieee80211req_sta_vlan vlan; int error; if (ireq->i_len != sizeof(vlan)) return EINVAL; error = copyin(ireq->i_data, &vlan, sizeof(vlan)); if (error != 0) return error; if (!IEEE80211_ADDR_EQ(vlan.sv_macaddr, zerobssid)) { ni = ieee80211_find_vap_node(&vap->iv_ic->ic_sta, vap, vlan.sv_macaddr); if (ni == NULL) return ENOENT; } else ni = ieee80211_ref_node(vap->iv_bss); ni->ni_vlan = vlan.sv_vlan; ieee80211_free_node(ni); return error; } static int isvap11g(const struct ieee80211vap *vap) { const struct ieee80211_node *bss = vap->iv_bss; return bss->ni_chan != IEEE80211_CHAN_ANYC && IEEE80211_IS_CHAN_ANYG(bss->ni_chan); } static int isvapht(const struct ieee80211vap *vap) { const struct ieee80211_node *bss = vap->iv_bss; return bss->ni_chan != IEEE80211_CHAN_ANYC && IEEE80211_IS_CHAN_HT(bss->ni_chan); } /* * Dummy ioctl set handler so the linker set is defined. */ static int dummy_ioctl_set(struct ieee80211vap *vap, struct ieee80211req *ireq) { return ENOSYS; } IEEE80211_IOCTL_SET(dummy, dummy_ioctl_set); static int ieee80211_ioctl_setdefault(struct ieee80211vap *vap, struct ieee80211req *ireq) { ieee80211_ioctl_setfunc * const *set; int error; SET_FOREACH(set, ieee80211_ioctl_setset) { error = (*set)(vap, ireq); if (error != ENOSYS) return error; } return EINVAL; } static int ieee80211_ioctl_set80211(struct ieee80211vap *vap, u_long cmd, struct ieee80211req *ireq) { struct ieee80211com *ic = vap->iv_ic; int error; const struct ieee80211_authenticator *auth; uint8_t tmpkey[IEEE80211_KEYBUF_SIZE]; char tmpssid[IEEE80211_NWID_LEN]; uint8_t tmpbssid[IEEE80211_ADDR_LEN]; struct ieee80211_key *k; u_int kid; uint32_t flags; error = 0; switch (ireq->i_type) { case IEEE80211_IOC_SSID: if (ireq->i_val != 0 || ireq->i_len > IEEE80211_NWID_LEN) return EINVAL; error = copyin(ireq->i_data, tmpssid, ireq->i_len); if (error) break; memset(vap->iv_des_ssid[0].ssid, 0, IEEE80211_NWID_LEN); vap->iv_des_ssid[0].len = ireq->i_len; memcpy(vap->iv_des_ssid[0].ssid, tmpssid, ireq->i_len); vap->iv_des_nssid = (ireq->i_len > 0); error = ENETRESET; break; case IEEE80211_IOC_WEP: switch (ireq->i_val) { case IEEE80211_WEP_OFF: vap->iv_flags &= ~IEEE80211_F_PRIVACY; vap->iv_flags &= ~IEEE80211_F_DROPUNENC; break; case IEEE80211_WEP_ON: vap->iv_flags |= IEEE80211_F_PRIVACY; vap->iv_flags |= IEEE80211_F_DROPUNENC; break; case IEEE80211_WEP_MIXED: vap->iv_flags |= IEEE80211_F_PRIVACY; vap->iv_flags &= ~IEEE80211_F_DROPUNENC; break; } error = ENETRESET; break; case IEEE80211_IOC_WEPKEY: kid = (u_int) ireq->i_val; if (kid >= IEEE80211_WEP_NKID) return EINVAL; k = &vap->iv_nw_keys[kid]; if (ireq->i_len == 0) { /* zero-len =>'s delete any existing key */ (void) ieee80211_crypto_delkey(vap, k); break; } if (ireq->i_len > sizeof(tmpkey)) return EINVAL; memset(tmpkey, 0, sizeof(tmpkey)); error = copyin(ireq->i_data, tmpkey, ireq->i_len); if (error) break; ieee80211_key_update_begin(vap); k->wk_keyix = kid; /* NB: force fixed key id */ if (ieee80211_crypto_newkey(vap, IEEE80211_CIPHER_WEP, IEEE80211_KEY_XMIT | IEEE80211_KEY_RECV, k)) { k->wk_keylen = ireq->i_len; memcpy(k->wk_key, tmpkey, sizeof(tmpkey)); IEEE80211_ADDR_COPY(k->wk_macaddr, vap->iv_myaddr); if (!ieee80211_crypto_setkey(vap, k)) error = EINVAL; } else error = EINVAL; ieee80211_key_update_end(vap); break; case IEEE80211_IOC_WEPTXKEY: kid = (u_int) ireq->i_val; if (kid >= IEEE80211_WEP_NKID && (uint16_t) kid != IEEE80211_KEYIX_NONE) return EINVAL; /* * Firmware devices may need to be told about an explicit * key index here, versus just inferring it from the * key set / change. Since we may also need to pause * things like transmit before the key is updated, * give the driver a chance to flush things by tying * into key update begin/end. */ ieee80211_key_update_begin(vap); ieee80211_crypto_set_deftxkey(vap, kid); ieee80211_key_update_end(vap); break; case IEEE80211_IOC_AUTHMODE: switch (ireq->i_val) { case IEEE80211_AUTH_WPA: case IEEE80211_AUTH_8021X: /* 802.1x */ case IEEE80211_AUTH_OPEN: /* open */ case IEEE80211_AUTH_SHARED: /* shared-key */ case IEEE80211_AUTH_AUTO: /* auto */ auth = ieee80211_authenticator_get(ireq->i_val); if (auth == NULL) return EINVAL; break; default: return EINVAL; } switch (ireq->i_val) { case IEEE80211_AUTH_WPA: /* WPA w/ 802.1x */ vap->iv_flags |= IEEE80211_F_PRIVACY; ireq->i_val = IEEE80211_AUTH_8021X; break; case IEEE80211_AUTH_OPEN: /* open */ vap->iv_flags &= ~(IEEE80211_F_WPA|IEEE80211_F_PRIVACY); break; case IEEE80211_AUTH_SHARED: /* shared-key */ case IEEE80211_AUTH_8021X: /* 802.1x */ vap->iv_flags &= ~IEEE80211_F_WPA; /* both require a key so mark the PRIVACY capability */ vap->iv_flags |= IEEE80211_F_PRIVACY; break; case IEEE80211_AUTH_AUTO: /* auto */ vap->iv_flags &= ~IEEE80211_F_WPA; /* XXX PRIVACY handling? */ /* XXX what's the right way to do this? */ break; } /* NB: authenticator attach/detach happens on state change */ vap->iv_bss->ni_authmode = ireq->i_val; /* XXX mixed/mode/usage? */ vap->iv_auth = auth; error = ENETRESET; break; case IEEE80211_IOC_CHANNEL: error = ieee80211_ioctl_setchannel(vap, ireq); break; case IEEE80211_IOC_POWERSAVE: switch (ireq->i_val) { case IEEE80211_POWERSAVE_OFF: if (vap->iv_flags & IEEE80211_F_PMGTON) { ieee80211_syncflag(vap, -IEEE80211_F_PMGTON); error = ERESTART; } break; case IEEE80211_POWERSAVE_ON: if ((vap->iv_caps & IEEE80211_C_PMGT) == 0) error = EOPNOTSUPP; else if ((vap->iv_flags & IEEE80211_F_PMGTON) == 0) { ieee80211_syncflag(vap, IEEE80211_F_PMGTON); error = ERESTART; } break; default: error = EINVAL; break; } break; case IEEE80211_IOC_POWERSAVESLEEP: if (ireq->i_val < 0) return EINVAL; ic->ic_lintval = ireq->i_val; error = ERESTART; break; case IEEE80211_IOC_RTSTHRESHOLD: if (!(IEEE80211_RTS_MIN <= ireq->i_val && ireq->i_val <= IEEE80211_RTS_MAX)) return EINVAL; vap->iv_rtsthreshold = ireq->i_val; error = ERESTART; break; case IEEE80211_IOC_PROTMODE: if (ireq->i_val > IEEE80211_PROT_RTSCTS) return EINVAL; vap->iv_protmode = (enum ieee80211_protmode)ireq->i_val; /* NB: if not operating in 11g this can wait */ if (ic->ic_bsschan != IEEE80211_CHAN_ANYC && IEEE80211_IS_CHAN_ANYG(ic->ic_bsschan)) error = ERESTART; /* driver callback for protection mode update */ ieee80211_vap_update_erp_protmode(vap); break; case IEEE80211_IOC_TXPOWER: if ((ic->ic_caps & IEEE80211_C_TXPMGT) == 0) return EOPNOTSUPP; if (!(IEEE80211_TXPOWER_MIN <= ireq->i_val && ireq->i_val <= IEEE80211_TXPOWER_MAX)) return EINVAL; ic->ic_txpowlimit = ireq->i_val; error = ERESTART; break; case IEEE80211_IOC_ROAMING: if (!(IEEE80211_ROAMING_DEVICE <= ireq->i_val && ireq->i_val <= IEEE80211_ROAMING_MANUAL)) return EINVAL; vap->iv_roaming = (enum ieee80211_roamingmode)ireq->i_val; /* XXXX reset? */ break; case IEEE80211_IOC_PRIVACY: if (ireq->i_val) { /* XXX check for key state? */ vap->iv_flags |= IEEE80211_F_PRIVACY; } else vap->iv_flags &= ~IEEE80211_F_PRIVACY; /* XXX ERESTART? */ break; case IEEE80211_IOC_DROPUNENCRYPTED: if (ireq->i_val) vap->iv_flags |= IEEE80211_F_DROPUNENC; else vap->iv_flags &= ~IEEE80211_F_DROPUNENC; /* XXX ERESTART? */ break; case IEEE80211_IOC_WPAKEY: error = ieee80211_ioctl_setkey(vap, ireq); break; case IEEE80211_IOC_DELKEY: error = ieee80211_ioctl_delkey(vap, ireq); break; case IEEE80211_IOC_MLME: error = ieee80211_ioctl_setmlme(vap, ireq); break; case IEEE80211_IOC_COUNTERMEASURES: if (ireq->i_val) { if ((vap->iv_flags & IEEE80211_F_WPA) == 0) return EOPNOTSUPP; vap->iv_flags |= IEEE80211_F_COUNTERM; } else vap->iv_flags &= ~IEEE80211_F_COUNTERM; /* XXX ERESTART? */ break; case IEEE80211_IOC_WPA: if (ireq->i_val > 3) return EINVAL; /* XXX verify ciphers available */ flags = vap->iv_flags & ~IEEE80211_F_WPA; switch (ireq->i_val) { case 0: /* wpa_supplicant calls this to clear the WPA config */ break; case 1: if (!(vap->iv_caps & IEEE80211_C_WPA1)) return EOPNOTSUPP; flags |= IEEE80211_F_WPA1; break; case 2: if (!(vap->iv_caps & IEEE80211_C_WPA2)) return EOPNOTSUPP; flags |= IEEE80211_F_WPA2; break; case 3: if ((vap->iv_caps & IEEE80211_C_WPA) != IEEE80211_C_WPA) return EOPNOTSUPP; flags |= IEEE80211_F_WPA1 | IEEE80211_F_WPA2; break; default: /* Can't set any -> error */ return EOPNOTSUPP; } vap->iv_flags = flags; error = ERESTART; /* NB: can change beacon frame */ break; case IEEE80211_IOC_WME: if (ireq->i_val) { if ((vap->iv_caps & IEEE80211_C_WME) == 0) return EOPNOTSUPP; ieee80211_syncflag(vap, IEEE80211_F_WME); } else ieee80211_syncflag(vap, -IEEE80211_F_WME); error = ERESTART; /* NB: can change beacon frame */ break; case IEEE80211_IOC_HIDESSID: if (ireq->i_val) vap->iv_flags |= IEEE80211_F_HIDESSID; else vap->iv_flags &= ~IEEE80211_F_HIDESSID; error = ERESTART; /* XXX ENETRESET? */ break; case IEEE80211_IOC_APBRIDGE: if (ireq->i_val == 0) vap->iv_flags |= IEEE80211_F_NOBRIDGE; else vap->iv_flags &= ~IEEE80211_F_NOBRIDGE; break; case IEEE80211_IOC_BSSID: if (ireq->i_len != sizeof(tmpbssid)) return EINVAL; error = copyin(ireq->i_data, tmpbssid, ireq->i_len); if (error) break; IEEE80211_ADDR_COPY(vap->iv_des_bssid, tmpbssid); if (IEEE80211_ADDR_EQ(vap->iv_des_bssid, zerobssid)) vap->iv_flags &= ~IEEE80211_F_DESBSSID; else vap->iv_flags |= IEEE80211_F_DESBSSID; error = ENETRESET; break; case IEEE80211_IOC_CHANLIST: error = ieee80211_ioctl_setchanlist(vap, ireq); break; #define OLD_IEEE80211_IOC_SCAN_REQ 23 #ifdef OLD_IEEE80211_IOC_SCAN_REQ case OLD_IEEE80211_IOC_SCAN_REQ: IEEE80211_DPRINTF(vap, IEEE80211_MSG_SCAN, "%s: active scan request\n", __func__); /* * If we are in INIT state then the driver has never * had a chance to setup hardware state to do a scan; * use the state machine to get us up the SCAN state. * Otherwise just invoke the scan machinery to start * a one-time scan. */ if (vap->iv_state == IEEE80211_S_INIT) ieee80211_new_state(vap, IEEE80211_S_SCAN, 0); else (void) ieee80211_start_scan(vap, IEEE80211_SCAN_ACTIVE | IEEE80211_SCAN_NOPICK | IEEE80211_SCAN_ONCE, IEEE80211_SCAN_FOREVER, 0, 0, /* XXX use ioctl params */ vap->iv_des_nssid, vap->iv_des_ssid); break; #endif /* OLD_IEEE80211_IOC_SCAN_REQ */ case IEEE80211_IOC_SCAN_REQ: error = ieee80211_ioctl_scanreq(vap, ireq); break; case IEEE80211_IOC_SCAN_CANCEL: IEEE80211_DPRINTF(vap, IEEE80211_MSG_SCAN, "%s: cancel scan\n", __func__); ieee80211_cancel_scan(vap); break; case IEEE80211_IOC_HTCONF: if (ireq->i_val & 1) ieee80211_syncflag_ht(vap, IEEE80211_FHT_HT); else ieee80211_syncflag_ht(vap, -IEEE80211_FHT_HT); if (ireq->i_val & 2) ieee80211_syncflag_ht(vap, IEEE80211_FHT_USEHT40); else ieee80211_syncflag_ht(vap, -IEEE80211_FHT_USEHT40); error = ENETRESET; break; case IEEE80211_IOC_ADDMAC: case IEEE80211_IOC_DELMAC: error = ieee80211_ioctl_macmac(vap, ireq); break; case IEEE80211_IOC_MACCMD: error = ieee80211_ioctl_setmaccmd(vap, ireq); break; case IEEE80211_IOC_STA_STATS: error = ieee80211_ioctl_setstastats(vap, ireq); break; case IEEE80211_IOC_STA_TXPOW: error = ieee80211_ioctl_setstatxpow(vap, ireq); break; case IEEE80211_IOC_WME_CWMIN: /* WME: CWmin */ case IEEE80211_IOC_WME_CWMAX: /* WME: CWmax */ case IEEE80211_IOC_WME_AIFS: /* WME: AIFS */ case IEEE80211_IOC_WME_TXOPLIMIT: /* WME: txops limit */ case IEEE80211_IOC_WME_ACM: /* WME: ACM (bss only) */ case IEEE80211_IOC_WME_ACKPOLICY: /* WME: ACK policy (!bss only) */ error = ieee80211_ioctl_setwmeparam(vap, ireq); break; case IEEE80211_IOC_DTIM_PERIOD: if (vap->iv_opmode != IEEE80211_M_HOSTAP && vap->iv_opmode != IEEE80211_M_MBSS && vap->iv_opmode != IEEE80211_M_IBSS) return EINVAL; if (IEEE80211_DTIM_MIN <= ireq->i_val && ireq->i_val <= IEEE80211_DTIM_MAX) { vap->iv_dtim_period = ireq->i_val; error = ENETRESET; /* requires restart */ } else error = EINVAL; break; case IEEE80211_IOC_BEACON_INTERVAL: if (vap->iv_opmode != IEEE80211_M_HOSTAP && vap->iv_opmode != IEEE80211_M_MBSS && vap->iv_opmode != IEEE80211_M_IBSS) return EINVAL; if (IEEE80211_BINTVAL_MIN <= ireq->i_val && ireq->i_val <= IEEE80211_BINTVAL_MAX) { ic->ic_bintval = ireq->i_val; error = ENETRESET; /* requires restart */ } else error = EINVAL; break; case IEEE80211_IOC_PUREG: if (ireq->i_val) vap->iv_flags |= IEEE80211_F_PUREG; else vap->iv_flags &= ~IEEE80211_F_PUREG; /* NB: reset only if we're operating on an 11g channel */ if (isvap11g(vap)) error = ENETRESET; break; case IEEE80211_IOC_QUIET: vap->iv_quiet= ireq->i_val; break; case IEEE80211_IOC_QUIET_COUNT: vap->iv_quiet_count=ireq->i_val; break; case IEEE80211_IOC_QUIET_PERIOD: vap->iv_quiet_period=ireq->i_val; break; case IEEE80211_IOC_QUIET_OFFSET: vap->iv_quiet_offset=ireq->i_val; break; case IEEE80211_IOC_QUIET_DUR: if(ireq->i_val < vap->iv_bss->ni_intval) vap->iv_quiet_duration = ireq->i_val; else error = EINVAL; break; case IEEE80211_IOC_BGSCAN: if (ireq->i_val) { if ((vap->iv_caps & IEEE80211_C_BGSCAN) == 0) return EOPNOTSUPP; vap->iv_flags |= IEEE80211_F_BGSCAN; } else vap->iv_flags &= ~IEEE80211_F_BGSCAN; break; case IEEE80211_IOC_BGSCAN_IDLE: if (ireq->i_val >= IEEE80211_BGSCAN_IDLE_MIN) vap->iv_bgscanidle = ireq->i_val*hz/1000; else error = EINVAL; break; case IEEE80211_IOC_BGSCAN_INTERVAL: if (ireq->i_val >= IEEE80211_BGSCAN_INTVAL_MIN) vap->iv_bgscanintvl = ireq->i_val*hz; else error = EINVAL; break; case IEEE80211_IOC_SCANVALID: if (ireq->i_val >= IEEE80211_SCAN_VALID_MIN) vap->iv_scanvalid = ireq->i_val*hz; else error = EINVAL; break; case IEEE80211_IOC_FRAGTHRESHOLD: if ((vap->iv_caps & IEEE80211_C_TXFRAG) == 0 && ireq->i_val != IEEE80211_FRAG_MAX) return EOPNOTSUPP; if (!(IEEE80211_FRAG_MIN <= ireq->i_val && ireq->i_val <= IEEE80211_FRAG_MAX)) return EINVAL; vap->iv_fragthreshold = ireq->i_val; error = ERESTART; break; case IEEE80211_IOC_BURST: if (ireq->i_val) { if ((vap->iv_caps & IEEE80211_C_BURST) == 0) return EOPNOTSUPP; ieee80211_syncflag(vap, IEEE80211_F_BURST); } else ieee80211_syncflag(vap, -IEEE80211_F_BURST); error = ERESTART; break; case IEEE80211_IOC_BMISSTHRESHOLD: if (!(IEEE80211_HWBMISS_MIN <= ireq->i_val && ireq->i_val <= IEEE80211_HWBMISS_MAX)) return EINVAL; vap->iv_bmissthreshold = ireq->i_val; error = ERESTART; break; case IEEE80211_IOC_CURCHAN: error = ieee80211_ioctl_setcurchan(vap, ireq); break; case IEEE80211_IOC_SHORTGI: if (ireq->i_val) { #define IEEE80211_HTCAP_SHORTGI \ (IEEE80211_HTCAP_SHORTGI20 | IEEE80211_HTCAP_SHORTGI40) if (((ireq->i_val ^ vap->iv_htcaps) & IEEE80211_HTCAP_SHORTGI) != 0) return EINVAL; if (ireq->i_val & IEEE80211_HTCAP_SHORTGI20) vap->iv_flags_ht |= IEEE80211_FHT_SHORTGI20; if (ireq->i_val & IEEE80211_HTCAP_SHORTGI40) vap->iv_flags_ht |= IEEE80211_FHT_SHORTGI40; #undef IEEE80211_HTCAP_SHORTGI } else vap->iv_flags_ht &= ~(IEEE80211_FHT_SHORTGI20 | IEEE80211_FHT_SHORTGI40); error = ERESTART; break; case IEEE80211_IOC_AMPDU: if (ireq->i_val && (vap->iv_htcaps & IEEE80211_HTC_AMPDU) == 0) return EINVAL; if (ireq->i_val & 1) vap->iv_flags_ht |= IEEE80211_FHT_AMPDU_TX; else vap->iv_flags_ht &= ~IEEE80211_FHT_AMPDU_TX; if (ireq->i_val & 2) vap->iv_flags_ht |= IEEE80211_FHT_AMPDU_RX; else vap->iv_flags_ht &= ~IEEE80211_FHT_AMPDU_RX; /* NB: reset only if we're operating on an 11n channel */ if (isvapht(vap)) error = ERESTART; break; case IEEE80211_IOC_AMPDU_LIMIT: /* XXX TODO: figure out ampdu_limit versus ampdu_rxmax */ if (!(IEEE80211_HTCAP_MAXRXAMPDU_8K <= ireq->i_val && ireq->i_val <= IEEE80211_HTCAP_MAXRXAMPDU_64K)) return EINVAL; if (vap->iv_opmode == IEEE80211_M_HOSTAP) vap->iv_ampdu_rxmax = ireq->i_val; else vap->iv_ampdu_limit = ireq->i_val; error = ERESTART; break; case IEEE80211_IOC_AMPDU_DENSITY: if (!(IEEE80211_HTCAP_MPDUDENSITY_NA <= ireq->i_val && ireq->i_val <= IEEE80211_HTCAP_MPDUDENSITY_16)) return EINVAL; vap->iv_ampdu_density = ireq->i_val; error = ERESTART; break; case IEEE80211_IOC_AMSDU: if (ireq->i_val && (vap->iv_htcaps & IEEE80211_HTC_AMSDU) == 0) return EINVAL; if (ireq->i_val & 1) vap->iv_flags_ht |= IEEE80211_FHT_AMSDU_TX; else vap->iv_flags_ht &= ~IEEE80211_FHT_AMSDU_TX; if (ireq->i_val & 2) vap->iv_flags_ht |= IEEE80211_FHT_AMSDU_RX; else vap->iv_flags_ht &= ~IEEE80211_FHT_AMSDU_RX; /* NB: reset only if we're operating on an 11n channel */ if (isvapht(vap)) error = ERESTART; break; case IEEE80211_IOC_AMSDU_LIMIT: /* XXX validate */ vap->iv_amsdu_limit = ireq->i_val; /* XXX truncation? */ break; case IEEE80211_IOC_PUREN: if (ireq->i_val) { if ((vap->iv_flags_ht & IEEE80211_FHT_HT) == 0) return EINVAL; vap->iv_flags_ht |= IEEE80211_FHT_PUREN; } else vap->iv_flags_ht &= ~IEEE80211_FHT_PUREN; /* NB: reset only if we're operating on an 11n channel */ if (isvapht(vap)) error = ERESTART; break; case IEEE80211_IOC_DOTH: if (ireq->i_val) { #if 0 /* XXX no capability */ if ((vap->iv_caps & IEEE80211_C_DOTH) == 0) return EOPNOTSUPP; #endif vap->iv_flags |= IEEE80211_F_DOTH; } else vap->iv_flags &= ~IEEE80211_F_DOTH; error = ENETRESET; break; case IEEE80211_IOC_REGDOMAIN: error = ieee80211_ioctl_setregdomain(vap, ireq); break; case IEEE80211_IOC_ROAM: error = ieee80211_ioctl_setroam(vap, ireq); break; case IEEE80211_IOC_TXPARAMS: error = ieee80211_ioctl_settxparams(vap, ireq); break; case IEEE80211_IOC_HTCOMPAT: if (ireq->i_val) { if ((vap->iv_flags_ht & IEEE80211_FHT_HT) == 0) return EOPNOTSUPP; vap->iv_flags_ht |= IEEE80211_FHT_HTCOMPAT; } else vap->iv_flags_ht &= ~IEEE80211_FHT_HTCOMPAT; /* NB: reset only if we're operating on an 11n channel */ if (isvapht(vap)) error = ERESTART; break; case IEEE80211_IOC_DWDS: if (ireq->i_val) { /* NB: DWDS only makes sense for WDS-capable devices */ if ((ic->ic_caps & IEEE80211_C_WDS) == 0) return EOPNOTSUPP; /* NB: DWDS is used only with ap+sta vaps */ if (vap->iv_opmode != IEEE80211_M_HOSTAP && vap->iv_opmode != IEEE80211_M_STA) return EINVAL; vap->iv_flags |= IEEE80211_F_DWDS; if (vap->iv_opmode == IEEE80211_M_STA) vap->iv_flags_ext |= IEEE80211_FEXT_4ADDR; } else { vap->iv_flags &= ~IEEE80211_F_DWDS; if (vap->iv_opmode == IEEE80211_M_STA) vap->iv_flags_ext &= ~IEEE80211_FEXT_4ADDR; } break; case IEEE80211_IOC_INACTIVITY: if (ireq->i_val) vap->iv_flags_ext |= IEEE80211_FEXT_INACT; else vap->iv_flags_ext &= ~IEEE80211_FEXT_INACT; break; case IEEE80211_IOC_APPIE: error = ieee80211_ioctl_setappie(vap, ireq); break; case IEEE80211_IOC_WPS: if (ireq->i_val) { if ((vap->iv_caps & IEEE80211_C_WPA) == 0) return EOPNOTSUPP; vap->iv_flags_ext |= IEEE80211_FEXT_WPS; } else vap->iv_flags_ext &= ~IEEE80211_FEXT_WPS; break; case IEEE80211_IOC_TSN: if (ireq->i_val) { if ((vap->iv_caps & IEEE80211_C_WPA) == 0) return EOPNOTSUPP; vap->iv_flags_ext |= IEEE80211_FEXT_TSN; } else vap->iv_flags_ext &= ~IEEE80211_FEXT_TSN; break; case IEEE80211_IOC_CHANSWITCH: error = ieee80211_ioctl_chanswitch(vap, ireq); break; case IEEE80211_IOC_DFS: if (ireq->i_val) { if ((vap->iv_caps & IEEE80211_C_DFS) == 0) return EOPNOTSUPP; /* NB: DFS requires 11h support */ if ((vap->iv_flags & IEEE80211_F_DOTH) == 0) return EINVAL; vap->iv_flags_ext |= IEEE80211_FEXT_DFS; } else vap->iv_flags_ext &= ~IEEE80211_FEXT_DFS; break; case IEEE80211_IOC_DOTD: if (ireq->i_val) vap->iv_flags_ext |= IEEE80211_FEXT_DOTD; else vap->iv_flags_ext &= ~IEEE80211_FEXT_DOTD; if (vap->iv_opmode == IEEE80211_M_STA) error = ENETRESET; break; case IEEE80211_IOC_HTPROTMODE: if (ireq->i_val > IEEE80211_PROT_RTSCTS) return EINVAL; vap->iv_htprotmode = ireq->i_val ? IEEE80211_PROT_RTSCTS : IEEE80211_PROT_NONE; /* NB: if not operating in 11n this can wait */ if (isvapht(vap)) error = ERESTART; /* Notify driver layer of HT protmode changes */ ieee80211_vap_update_ht_protmode(vap); break; case IEEE80211_IOC_STA_VLAN: error = ieee80211_ioctl_setstavlan(vap, ireq); break; case IEEE80211_IOC_SMPS: if ((ireq->i_val &~ IEEE80211_HTCAP_SMPS) != 0 || ireq->i_val == 0x0008) /* value of 2 is reserved */ return EINVAL; if (ireq->i_val != IEEE80211_HTCAP_SMPS_OFF && (vap->iv_htcaps & IEEE80211_HTC_SMPS) == 0) return EOPNOTSUPP; vap->iv_htcaps = (vap->iv_htcaps &~ IEEE80211_HTCAP_SMPS) | ireq->i_val; /* NB: if not operating in 11n this can wait */ if (isvapht(vap)) error = ERESTART; break; case IEEE80211_IOC_RIFS: if (ireq->i_val != 0) { if ((vap->iv_htcaps & IEEE80211_HTC_RIFS) == 0) return EOPNOTSUPP; vap->iv_flags_ht |= IEEE80211_FHT_RIFS; } else vap->iv_flags_ht &= ~IEEE80211_FHT_RIFS; /* NB: if not operating in 11n this can wait */ if (isvapht(vap)) error = ERESTART; break; case IEEE80211_IOC_STBC: /* Check if we can do STBC TX/RX before changing the setting */ if ((ireq->i_val & 1) && ((vap->iv_htcaps & IEEE80211_HTCAP_TXSTBC) == 0)) return EOPNOTSUPP; if ((ireq->i_val & 2) && ((vap->iv_htcaps & IEEE80211_HTCAP_RXSTBC) == 0)) return EOPNOTSUPP; /* TX */ if (ireq->i_val & 1) vap->iv_flags_ht |= IEEE80211_FHT_STBC_TX; else vap->iv_flags_ht &= ~IEEE80211_FHT_STBC_TX; /* RX */ if (ireq->i_val & 2) vap->iv_flags_ht |= IEEE80211_FHT_STBC_RX; else vap->iv_flags_ht &= ~IEEE80211_FHT_STBC_RX; /* NB: reset only if we're operating on an 11n channel */ if (isvapht(vap)) error = ERESTART; break; case IEEE80211_IOC_LDPC: /* Check if we can do LDPC TX/RX before changing the setting */ if ((ireq->i_val & 1) && (vap->iv_htcaps & IEEE80211_HTC_TXLDPC) == 0) return EOPNOTSUPP; if ((ireq->i_val & 2) && (vap->iv_htcaps & IEEE80211_HTCAP_LDPC) == 0) return EOPNOTSUPP; /* TX */ if (ireq->i_val & 1) vap->iv_flags_ht |= IEEE80211_FHT_LDPC_TX; else vap->iv_flags_ht &= ~IEEE80211_FHT_LDPC_TX; /* RX */ if (ireq->i_val & 2) vap->iv_flags_ht |= IEEE80211_FHT_LDPC_RX; else vap->iv_flags_ht &= ~IEEE80211_FHT_LDPC_RX; /* NB: reset only if we're operating on an 11n channel */ if (isvapht(vap)) error = ERESTART; break; case IEEE80211_IOC_UAPSD: if ((vap->iv_caps & IEEE80211_C_UAPSD) == 0) return EOPNOTSUPP; if (ireq->i_val == 0) vap->iv_flags_ext &= ~IEEE80211_FEXT_UAPSD; else if (ireq->i_val == 1) vap->iv_flags_ext |= IEEE80211_FEXT_UAPSD; else return EINVAL; break; /* VHT */ case IEEE80211_IOC_VHTCONF: if (ireq->i_val & 1) ieee80211_syncflag_vht(vap, IEEE80211_FVHT_VHT); else ieee80211_syncflag_vht(vap, -IEEE80211_FVHT_VHT); if (ireq->i_val & 2) ieee80211_syncflag_vht(vap, IEEE80211_FVHT_USEVHT40); else ieee80211_syncflag_vht(vap, -IEEE80211_FVHT_USEVHT40); if (ireq->i_val & 4) ieee80211_syncflag_vht(vap, IEEE80211_FVHT_USEVHT80); else ieee80211_syncflag_vht(vap, -IEEE80211_FVHT_USEVHT80); if (ireq->i_val & 8) ieee80211_syncflag_vht(vap, IEEE80211_FVHT_USEVHT80P80); else ieee80211_syncflag_vht(vap, -IEEE80211_FVHT_USEVHT80P80); if (ireq->i_val & 16) ieee80211_syncflag_vht(vap, IEEE80211_FVHT_USEVHT160); else ieee80211_syncflag_vht(vap, -IEEE80211_FVHT_USEVHT160); error = ENETRESET; break; default: error = ieee80211_ioctl_setdefault(vap, ireq); break; } /* * The convention is that ENETRESET means an operation * requires a complete re-initialization of the device (e.g. * changing something that affects the association state). * ERESTART means the request may be handled with only a * reload of the hardware state. We hand ERESTART requests * to the iv_reset callback so the driver can decide. If * a device does not fillin iv_reset then it defaults to one * that returns ENETRESET. Otherwise a driver may return * ENETRESET (in which case a full reset will be done) or * 0 to mean there's no need to do anything (e.g. when the * change has no effect on the driver/device). */ if (error == ERESTART) error = IFNET_IS_UP_RUNNING(vap->iv_ifp) ? vap->iv_reset(vap, ireq->i_type) : 0; if (error == ENETRESET) { /* XXX need to re-think AUTO handling */ if (IS_UP_AUTO(vap)) ieee80211_init(vap); error = 0; } return error; } int ieee80211_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data) { struct ieee80211vap *vap = ifp->if_softc; struct ieee80211com *ic = vap->iv_ic; int error = 0, wait = 0, ic_used; struct ifreq *ifr; struct ifaddr *ifa; /* XXX */ ic_used = (cmd != SIOCSIFMTU && cmd != SIOCG80211STATS); if (ic_used && (error = ieee80211_com_vincref(vap)) != 0) return (error); switch (cmd) { case SIOCSIFFLAGS: IEEE80211_LOCK(ic); if ((ifp->if_flags ^ vap->iv_ifflags) & IFF_PROMISC) { /* * Enable promiscuous mode when: * 1. Interface is not a member of bridge, or * 2. Requested by user, or * 3. In monitor (or adhoc-demo) mode. */ if (ifp->if_bridge == NULL || (ifp->if_flags & IFF_PPROMISC) != 0 || vap->iv_opmode == IEEE80211_M_MONITOR || (vap->iv_opmode == IEEE80211_M_AHDEMO && (vap->iv_caps & IEEE80211_C_TDMA) == 0)) { ieee80211_promisc(vap, ifp->if_flags & IFF_PROMISC); vap->iv_ifflags ^= IFF_PROMISC; } } if ((ifp->if_flags ^ vap->iv_ifflags) & IFF_ALLMULTI) { ieee80211_allmulti(vap, ifp->if_flags & IFF_ALLMULTI); vap->iv_ifflags ^= IFF_ALLMULTI; } if (ifp->if_flags & IFF_UP) { /* * Bring ourself up unless we're already operational. * If we're the first vap and the parent is not up * then it will automatically be brought up as a * side-effect of bringing ourself up. */ if (vap->iv_state == IEEE80211_S_INIT) { if (ic->ic_nrunning == 0) wait = 1; ieee80211_start_locked(vap); } } else if (ifp->if_drv_flags & IFF_DRV_RUNNING) { /* * Stop ourself. If we are the last vap to be * marked down the parent will also be taken down. */ if (ic->ic_nrunning == 1) wait = 1; ieee80211_stop_locked(vap); } IEEE80211_UNLOCK(ic); /* Wait for parent ioctl handler if it was queued */ if (wait) { struct epoch_tracker et; ieee80211_waitfor_parent(ic); /* * Check if the MAC address was changed * via SIOCSIFLLADDR ioctl. * * NB: device may be detached during initialization; * use if_ioctl for existence check. */ NET_EPOCH_ENTER(et); if (ifp->if_ioctl == ieee80211_ioctl && (ifp->if_flags & IFF_UP) == 0 && !IEEE80211_ADDR_EQ(vap->iv_myaddr, IF_LLADDR(ifp))) IEEE80211_ADDR_COPY(vap->iv_myaddr, IF_LLADDR(ifp)); NET_EPOCH_EXIT(et); } break; case SIOCADDMULTI: case SIOCDELMULTI: ieee80211_runtask(ic, &ic->ic_mcast_task); break; case SIOCSIFMEDIA: case SIOCGIFMEDIA: ifr = (struct ifreq *)data; error = ifmedia_ioctl(ifp, ifr, &vap->iv_media, cmd); break; case SIOCG80211: error = ieee80211_ioctl_get80211(vap, cmd, (struct ieee80211req *) data); break; case SIOCS80211: /* XXX TODO: move priv check to ieee80211_freebsd.c */ error = priv_check(curthread, PRIV_NET80211_VAP_MANAGE); if (error == 0) error = ieee80211_ioctl_set80211(vap, cmd, (struct ieee80211req *) data); break; case SIOCG80211STATS: ifr = (struct ifreq *)data; copyout(&vap->iv_stats, ifr_data_get_ptr(ifr), sizeof (vap->iv_stats)); break; case SIOCSIFMTU: ifr = (struct ifreq *)data; if (!(IEEE80211_MTU_MIN <= ifr->ifr_mtu && ifr->ifr_mtu <= IEEE80211_MTU_MAX)) error = EINVAL; else ifp->if_mtu = ifr->ifr_mtu; break; case SIOCSIFADDR: /* * XXX Handle this directly so we can suppress if_init calls. * XXX This should be done in ether_ioctl but for the moment * XXX there are too many other parts of the system that * XXX set IFF_UP and so suppress if_init being called when * XXX it should be. */ ifa = (struct ifaddr *) data; switch (ifa->ifa_addr->sa_family) { #ifdef INET case AF_INET: if ((ifp->if_flags & IFF_UP) == 0) { ifp->if_flags |= IFF_UP; ifp->if_init(ifp->if_softc); } arp_ifinit(ifp, ifa); break; #endif default: if ((ifp->if_flags & IFF_UP) == 0) { ifp->if_flags |= IFF_UP; ifp->if_init(ifp->if_softc); } break; } break; case SIOCSIFLLADDR: /* XXX TODO: move priv check to ieee80211_freebsd.c */ error = priv_check(curthread, PRIV_NET80211_VAP_SETMAC); if (error == 0) break; /* Fallthrough */ default: /* * Pass unknown ioctls first to the driver, and if it * returns ENOTTY, then to the generic Ethernet handler. */ if (ic->ic_ioctl != NULL && (error = ic->ic_ioctl(ic, cmd, data)) != ENOTTY) break; error = ether_ioctl(ifp, cmd, data); break; } if (ic_used) ieee80211_com_vdecref(vap); return (error); } Index: projects/clang1100-import/sys/net80211/ieee80211_ioctl.h =================================================================== --- projects/clang1100-import/sys/net80211/ieee80211_ioctl.h (revision 364034) +++ projects/clang1100-import/sys/net80211/ieee80211_ioctl.h (revision 364035) @@ -1,877 +1,880 @@ /*- * SPDX-License-Identifier: BSD-2-Clause-FreeBSD * * Copyright (c) 2001 Atsushi Onoe * Copyright (c) 2002-2009 Sam Leffler, Errno Consulting * 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. * * $FreeBSD$ */ #ifndef _NET80211_IEEE80211_IOCTL_H_ #define _NET80211_IEEE80211_IOCTL_H_ /* * IEEE 802.11 ioctls. */ #include #include #include /* * Per/node (station) statistics. */ struct ieee80211_nodestats { uint32_t ns_rx_data; /* rx data frames */ uint32_t ns_rx_mgmt; /* rx management frames */ uint32_t ns_rx_ctrl; /* rx control frames */ uint32_t ns_rx_ucast; /* rx unicast frames */ uint32_t ns_rx_mcast; /* rx multi/broadcast frames */ uint64_t ns_rx_bytes; /* rx data count (bytes) */ uint64_t ns_rx_beacons; /* rx beacon frames */ uint32_t ns_rx_proberesp; /* rx probe response frames */ uint32_t ns_rx_dup; /* rx discard 'cuz dup */ uint32_t ns_rx_noprivacy; /* rx w/ wep but privacy off */ uint32_t ns_rx_wepfail; /* rx wep processing failed */ uint32_t ns_rx_demicfail; /* rx demic failed */ uint32_t ns_rx_decap; /* rx decapsulation failed */ uint32_t ns_rx_defrag; /* rx defragmentation failed */ uint32_t ns_rx_disassoc; /* rx disassociation */ uint32_t ns_rx_deauth; /* rx deauthentication */ uint32_t ns_rx_action; /* rx action */ uint32_t ns_rx_decryptcrc; /* rx decrypt failed on crc */ uint32_t ns_rx_unauth; /* rx on unauthorized port */ uint32_t ns_rx_unencrypted; /* rx unecrypted w/ privacy */ uint32_t ns_rx_drop; /* rx discard other reason */ uint32_t ns_tx_data; /* tx data frames */ uint32_t ns_tx_mgmt; /* tx management frames */ uint32_t ns_tx_ctrl; /* tx control frames */ uint32_t ns_tx_ucast; /* tx unicast frames */ uint32_t ns_tx_mcast; /* tx multi/broadcast frames */ uint64_t ns_tx_bytes; /* tx data count (bytes) */ uint32_t ns_tx_probereq; /* tx probe request frames */ uint32_t ns_tx_novlantag; /* tx discard 'cuz no tag */ uint32_t ns_tx_vlanmismatch; /* tx discard 'cuz bad tag */ uint32_t ns_ps_discard; /* ps discard 'cuz of age */ /* MIB-related state */ uint32_t ns_tx_assoc; /* [re]associations */ uint32_t ns_tx_assoc_fail; /* [re]association failures */ uint32_t ns_tx_auth; /* [re]authentications */ uint32_t ns_tx_auth_fail; /* [re]authentication failures*/ uint32_t ns_tx_deauth; /* deauthentications */ uint32_t ns_tx_deauth_code; /* last deauth reason */ uint32_t ns_tx_disassoc; /* disassociations */ uint32_t ns_tx_disassoc_code; /* last disassociation reason */ /* Hardware A-MSDU decode */ uint32_t ns_rx_amsdu_more; /* RX decap A-MSDU, more coming from A-MSDU */ uint32_t ns_rx_amsdu_more_end; /* RX decap A-MSDU (or any other frame), no more coming */ uint32_t ns_spare[6]; }; /* * Summary statistics. */ struct ieee80211_stats { uint32_t is_rx_badversion; /* rx frame with bad version */ uint32_t is_rx_tooshort; /* rx frame too short */ uint32_t is_rx_wrongbss; /* rx from wrong bssid */ uint32_t is_rx_dup; /* rx discard 'cuz dup */ uint32_t is_rx_wrongdir; /* rx w/ wrong direction */ uint32_t is_rx_mcastecho; /* rx discard 'cuz mcast echo */ uint32_t is_rx_notassoc; /* rx discard 'cuz sta !assoc */ uint32_t is_rx_noprivacy; /* rx w/ wep but privacy off */ uint32_t is_rx_unencrypted; /* rx w/o wep and privacy on */ uint32_t is_rx_wepfail; /* rx wep processing failed */ uint32_t is_rx_decap; /* rx decapsulation failed */ uint32_t is_rx_mgtdiscard; /* rx discard mgt frames */ uint32_t is_rx_ctl; /* rx ctrl frames */ uint32_t is_rx_beacon; /* rx beacon frames */ uint32_t is_rx_rstoobig; /* rx rate set truncated */ uint32_t is_rx_elem_missing; /* rx required element missing*/ uint32_t is_rx_elem_toobig; /* rx element too big */ uint32_t is_rx_elem_toosmall; /* rx element too small */ uint32_t is_rx_elem_unknown; /* rx element unknown */ uint32_t is_rx_badchan; /* rx frame w/ invalid chan */ uint32_t is_rx_chanmismatch; /* rx frame chan mismatch */ uint32_t is_rx_nodealloc; /* rx frame dropped */ uint32_t is_rx_ssidmismatch; /* rx frame ssid mismatch */ uint32_t is_rx_auth_unsupported; /* rx w/ unsupported auth alg */ uint32_t is_rx_auth_fail; /* rx sta auth failure */ uint32_t is_rx_auth_countermeasures;/* rx auth discard 'cuz CM */ uint32_t is_rx_assoc_bss; /* rx assoc from wrong bssid */ uint32_t is_rx_assoc_notauth; /* rx assoc w/o auth */ uint32_t is_rx_assoc_capmismatch;/* rx assoc w/ cap mismatch */ uint32_t is_rx_assoc_norate; /* rx assoc w/ no rate match */ uint32_t is_rx_assoc_badwpaie; /* rx assoc w/ bad WPA IE */ uint32_t is_rx_deauth; /* rx deauthentication */ uint32_t is_rx_disassoc; /* rx disassociation */ uint32_t is_rx_badsubtype; /* rx frame w/ unknown subtype*/ uint32_t is_rx_nobuf; /* rx failed for lack of buf */ uint32_t is_rx_decryptcrc; /* rx decrypt failed on crc */ uint32_t is_rx_ahdemo_mgt; /* rx discard ahdemo mgt frame*/ uint32_t is_rx_bad_auth; /* rx bad auth request */ uint32_t is_rx_unauth; /* rx on unauthorized port */ uint32_t is_rx_badkeyid; /* rx w/ incorrect keyid */ uint32_t is_rx_ccmpreplay; /* rx seq# violation (CCMP) */ uint32_t is_rx_ccmpformat; /* rx format bad (CCMP) */ uint32_t is_rx_ccmpmic; /* rx MIC check failed (CCMP) */ uint32_t is_rx_tkipreplay; /* rx seq# violation (TKIP) */ uint32_t is_rx_tkipformat; /* rx format bad (TKIP) */ uint32_t is_rx_tkipmic; /* rx MIC check failed (TKIP) */ uint32_t is_rx_tkipicv; /* rx ICV check failed (TKIP) */ uint32_t is_rx_badcipher; /* rx failed 'cuz key type */ uint32_t is_rx_nocipherctx; /* rx failed 'cuz key !setup */ uint32_t is_rx_acl; /* rx discard 'cuz acl policy */ uint32_t is_tx_nobuf; /* tx failed for lack of buf */ uint32_t is_tx_nonode; /* tx failed for no node */ uint32_t is_tx_unknownmgt; /* tx of unknown mgt frame */ uint32_t is_tx_badcipher; /* tx failed 'cuz key type */ uint32_t is_tx_nodefkey; /* tx failed 'cuz no defkey */ uint32_t is_tx_noheadroom; /* tx failed 'cuz no space */ uint32_t is_tx_fragframes; /* tx frames fragmented */ uint32_t is_tx_frags; /* tx fragments created */ uint32_t is_scan_active; /* active scans started */ uint32_t is_scan_passive; /* passive scans started */ uint32_t is_node_timeout; /* nodes timed out inactivity */ uint32_t is_crypto_nomem; /* no memory for crypto ctx */ uint32_t is_crypto_tkip; /* tkip crypto done in s/w */ uint32_t is_crypto_tkipenmic; /* tkip en-MIC done in s/w */ uint32_t is_crypto_tkipdemic; /* tkip de-MIC done in s/w */ uint32_t is_crypto_tkipcm; /* tkip counter measures */ uint32_t is_crypto_ccmp; /* ccmp crypto done in s/w */ uint32_t is_crypto_wep; /* wep crypto done in s/w */ uint32_t is_crypto_setkey_cipher;/* cipher rejected key */ uint32_t is_crypto_setkey_nokey; /* no key index for setkey */ uint32_t is_crypto_delkey; /* driver key delete failed */ uint32_t is_crypto_badcipher; /* unknown cipher */ uint32_t is_crypto_nocipher; /* cipher not available */ uint32_t is_crypto_attachfail; /* cipher attach failed */ uint32_t is_crypto_swfallback; /* cipher fallback to s/w */ uint32_t is_crypto_keyfail; /* driver key alloc failed */ uint32_t is_crypto_enmicfail; /* en-MIC failed */ uint32_t is_ibss_capmismatch; /* merge failed-cap mismatch */ uint32_t is_ibss_norate; /* merge failed-rate mismatch */ uint32_t is_ps_unassoc; /* ps-poll for unassoc. sta */ uint32_t is_ps_badaid; /* ps-poll w/ incorrect aid */ uint32_t is_ps_qempty; /* ps-poll w/ nothing to send */ uint32_t is_ff_badhdr; /* fast frame rx'd w/ bad hdr */ uint32_t is_ff_tooshort; /* fast frame rx decap error */ uint32_t is_ff_split; /* fast frame rx split error */ uint32_t is_ff_decap; /* fast frames decap'd */ uint32_t is_ff_encap; /* fast frames encap'd for tx */ uint32_t is_rx_badbintval; /* rx frame w/ bogus bintval */ uint32_t is_rx_demicfail; /* rx demic failed */ uint32_t is_rx_defrag; /* rx defragmentation failed */ uint32_t is_rx_mgmt; /* rx management frames */ uint32_t is_rx_action; /* rx action mgt frames */ uint32_t is_amsdu_tooshort; /* A-MSDU rx decap error */ uint32_t is_amsdu_split; /* A-MSDU rx split error */ uint32_t is_amsdu_decap; /* A-MSDU decap'd */ uint32_t is_amsdu_encap; /* A-MSDU encap'd for tx */ uint32_t is_ampdu_bar_bad; /* A-MPDU BAR out of window */ uint32_t is_ampdu_bar_oow; /* A-MPDU BAR before ADDBA */ uint32_t is_ampdu_bar_move; /* A-MPDU BAR moved window */ uint32_t is_ampdu_bar_rx; /* A-MPDU BAR frames handled */ uint32_t is_ampdu_rx_flush; /* A-MPDU frames flushed */ uint32_t is_ampdu_rx_oor; /* A-MPDU frames out-of-order */ uint32_t is_ampdu_rx_copy; /* A-MPDU frames copied down */ uint32_t is_ampdu_rx_drop; /* A-MPDU frames dropped */ uint32_t is_tx_badstate; /* tx discard state != RUN */ uint32_t is_tx_notassoc; /* tx failed, sta not assoc */ uint32_t is_tx_classify; /* tx classification failed */ uint32_t is_dwds_mcast; /* discard mcast over dwds */ uint32_t is_dwds_qdrop; /* dwds pending frame q full */ uint32_t is_ht_assoc_nohtcap; /* non-HT sta rejected */ uint32_t is_ht_assoc_downgrade; /* HT sta forced to legacy */ uint32_t is_ht_assoc_norate; /* HT assoc w/ rate mismatch */ uint32_t is_ampdu_rx_age; /* A-MPDU sent up 'cuz of age */ uint32_t is_ampdu_rx_move; /* A-MPDU MSDU moved window */ uint32_t is_addba_reject; /* ADDBA reject 'cuz disabled */ uint32_t is_addba_norequest; /* ADDBA response w/o ADDBA */ uint32_t is_addba_badtoken; /* ADDBA response w/ wrong dialogtoken */ uint32_t is_addba_badpolicy; /* ADDBA resp w/ wrong policy */ uint32_t is_ampdu_stop; /* A-MPDU stream stopped */ uint32_t is_ampdu_stop_failed; /* A-MPDU stream not running */ uint32_t is_ampdu_rx_reorder; /* A-MPDU held for rx reorder */ uint32_t is_scan_bg; /* background scans started */ uint8_t is_rx_deauth_code; /* last rx'd deauth reason */ uint8_t is_rx_disassoc_code; /* last rx'd disassoc reason */ uint8_t is_rx_authfail_code; /* last rx'd auth fail reason */ uint32_t is_beacon_miss; /* beacon miss notification */ uint32_t is_rx_badstate; /* rx discard state != RUN */ uint32_t is_ff_flush; /* ff's flush'd from stageq */ uint32_t is_tx_ctl; /* tx ctrl frames */ uint32_t is_ampdu_rexmt; /* A-MPDU frames rexmt ok */ uint32_t is_ampdu_rexmt_fail; /* A-MPDU frames rexmt fail */ uint32_t is_mesh_wrongmesh; /* dropped 'cuz not mesh sta*/ uint32_t is_mesh_nolink; /* dropped 'cuz link not estab*/ uint32_t is_mesh_fwd_ttl; /* mesh not fwd'd 'cuz ttl 0 */ uint32_t is_mesh_fwd_nobuf; /* mesh not fwd'd 'cuz no mbuf*/ uint32_t is_mesh_fwd_tooshort; /* mesh not fwd'd 'cuz no hdr */ uint32_t is_mesh_fwd_disabled; /* mesh not fwd'd 'cuz disabled */ uint32_t is_mesh_fwd_nopath; /* mesh not fwd'd 'cuz path unknown */ uint32_t is_hwmp_wrongseq; /* wrong hwmp seq no. */ uint32_t is_hwmp_rootreqs; /* root PREQs sent */ uint32_t is_hwmp_rootrann; /* root RANNs sent */ uint32_t is_mesh_badae; /* dropped 'cuz invalid AE */ uint32_t is_mesh_rtaddfailed; /* route add failed */ uint32_t is_mesh_notproxy; /* dropped 'cuz not proxying */ uint32_t is_rx_badalign; /* dropped 'cuz misaligned */ uint32_t is_hwmp_proxy; /* PREP for proxy route */ uint32_t is_beacon_bad; /* Number of bad beacons */ uint32_t is_ampdu_bar_tx; /* A-MPDU BAR frames TXed */ uint32_t is_ampdu_bar_tx_retry; /* A-MPDU BAR frames TX rtry */ uint32_t is_ampdu_bar_tx_fail; /* A-MPDU BAR frames TX fail */ uint32_t is_ff_encapfail; /* failed FF encap */ uint32_t is_amsdu_encapfail; /* failed A-MSDU encap */ uint32_t is_spare[5]; }; /* * Max size of optional information elements. We artificially * constrain this; it's limited only by the max frame size (and * the max parameter size of the wireless extensions). */ #define IEEE80211_MAX_OPT_IE 256 /* * WPA/RSN get/set key request. Specify the key/cipher * type and whether the key is to be used for sending and/or * receiving. The key index should be set only when working * with global keys (use IEEE80211_KEYIX_NONE for ``no index''). * Otherwise a unicast/pairwise key is specified by the bssid * (on a station) or mac address (on an ap). They key length * must include any MIC key data; otherwise it should be no * more than IEEE80211_KEYBUF_SIZE. */ struct ieee80211req_key { uint8_t ik_type; /* key/cipher type */ uint8_t ik_pad; uint16_t ik_keyix; /* key index */ uint8_t ik_keylen; /* key length in bytes */ uint8_t ik_flags; /* NB: IEEE80211_KEY_XMIT and IEEE80211_KEY_RECV defined elsewhere */ #define IEEE80211_KEY_DEFAULT 0x80 /* default xmit key */ uint8_t ik_macaddr[IEEE80211_ADDR_LEN]; uint64_t ik_keyrsc; /* key receive sequence counter */ uint64_t ik_keytsc; /* key transmit sequence counter */ uint8_t ik_keydata[IEEE80211_KEYBUF_SIZE+IEEE80211_MICBUF_SIZE]; }; /* * Delete a key either by index or address. Set the index * to IEEE80211_KEYIX_NONE when deleting a unicast key. */ struct ieee80211req_del_key { uint8_t idk_keyix; /* key index */ uint8_t idk_macaddr[IEEE80211_ADDR_LEN]; }; /* * MLME state manipulation request. IEEE80211_MLME_ASSOC * only makes sense when operating as a station. The other * requests can be used when operating as a station or an * ap (to effect a station). */ struct ieee80211req_mlme { uint8_t im_op; /* operation to perform */ #define IEEE80211_MLME_ASSOC 1 /* associate station */ #define IEEE80211_MLME_DISASSOC 2 /* disassociate station */ #define IEEE80211_MLME_DEAUTH 3 /* deauthenticate station */ #define IEEE80211_MLME_AUTHORIZE 4 /* authorize station */ #define IEEE80211_MLME_UNAUTHORIZE 5 /* unauthorize station */ #define IEEE80211_MLME_AUTH 6 /* authenticate station */ uint8_t im_ssid_len; /* length of optional ssid */ uint16_t im_reason; /* 802.11 reason code */ uint8_t im_macaddr[IEEE80211_ADDR_LEN]; uint8_t im_ssid[IEEE80211_NWID_LEN]; }; /* * MAC ACL operations. */ enum { IEEE80211_MACCMD_POLICY_OPEN = 0, /* set policy: no ACL's */ IEEE80211_MACCMD_POLICY_ALLOW = 1, /* set policy: allow traffic */ IEEE80211_MACCMD_POLICY_DENY = 2, /* set policy: deny traffic */ IEEE80211_MACCMD_FLUSH = 3, /* flush ACL database */ IEEE80211_MACCMD_DETACH = 4, /* detach ACL policy */ IEEE80211_MACCMD_POLICY = 5, /* get ACL policy */ IEEE80211_MACCMD_LIST = 6, /* get ACL database */ IEEE80211_MACCMD_POLICY_RADIUS = 7, /* set policy: RADIUS managed */ }; struct ieee80211req_maclist { uint8_t ml_macaddr[IEEE80211_ADDR_LEN]; } __packed; /* * Mesh Routing Table Operations. */ enum { IEEE80211_MESH_RTCMD_LIST = 0, /* list HWMP routing table */ IEEE80211_MESH_RTCMD_FLUSH = 1, /* flush HWMP routing table */ IEEE80211_MESH_RTCMD_ADD = 2, /* add entry to the table */ IEEE80211_MESH_RTCMD_DELETE = 3, /* delete an entry from the table */ }; struct ieee80211req_mesh_route { uint8_t imr_flags; #define IEEE80211_MESHRT_FLAGS_DISCOVER 0x01 #define IEEE80211_MESHRT_FLAGS_VALID 0x02 #define IEEE80211_MESHRT_FLAGS_PROXY 0x04 #define IEEE80211_MESHRT_FLAGS_GATE 0x08 uint8_t imr_dest[IEEE80211_ADDR_LEN]; uint8_t imr_nexthop[IEEE80211_ADDR_LEN]; uint16_t imr_nhops; uint8_t imr_pad; uint32_t imr_metric; uint32_t imr_lifetime; uint32_t imr_lastmseq; }; /* * HWMP root modes */ enum { IEEE80211_HWMP_ROOTMODE_DISABLED = 0, /* disabled */ IEEE80211_HWMP_ROOTMODE_NORMAL = 1, /* normal PREPs */ IEEE80211_HWMP_ROOTMODE_PROACTIVE = 2, /* proactive PREPS */ IEEE80211_HWMP_ROOTMODE_RANN = 3, /* use RANN elemid */ }; /* * Set the active channel list by IEEE channel #: each channel * to be marked active is set in a bit vector. Note this list is * intersected with the available channel list in calculating * the set of channels actually used in scanning. */ struct ieee80211req_chanlist { uint8_t ic_channels[32]; /* NB: can be variable length */ }; /* * Get the active channel list info. */ struct ieee80211req_chaninfo { u_int ic_nchans; struct ieee80211_channel ic_chans[1]; /* NB: variable length */ }; #define IEEE80211_CHANINFO_SIZE(_nchan) \ (sizeof(struct ieee80211req_chaninfo) + \ (((_nchan)-1) * sizeof(struct ieee80211_channel))) #define IEEE80211_CHANINFO_SPACE(_ci) \ IEEE80211_CHANINFO_SIZE((_ci)->ic_nchans) /* * Retrieve the WPA/RSN information element for an associated station. */ struct ieee80211req_wpaie { /* old version w/ only one ie */ uint8_t wpa_macaddr[IEEE80211_ADDR_LEN]; uint8_t wpa_ie[IEEE80211_MAX_OPT_IE]; }; struct ieee80211req_wpaie2 { uint8_t wpa_macaddr[IEEE80211_ADDR_LEN]; uint8_t wpa_ie[IEEE80211_MAX_OPT_IE]; uint8_t rsn_ie[IEEE80211_MAX_OPT_IE]; }; /* * Retrieve per-node statistics. */ struct ieee80211req_sta_stats { union { /* NB: explicitly force 64-bit alignment */ uint8_t macaddr[IEEE80211_ADDR_LEN]; uint64_t pad; } is_u; struct ieee80211_nodestats is_stats; }; /* * Station information block; the mac address is used * to retrieve other data like stats, unicast key, etc. */ struct ieee80211req_sta_info { uint16_t isi_len; /* total length (mult of 4) */ uint16_t isi_ie_off; /* offset to IE data */ uint16_t isi_ie_len; /* IE length */ uint16_t isi_freq; /* MHz */ uint32_t isi_flags; /* channel flags */ uint32_t isi_state; /* state flags */ uint8_t isi_authmode; /* authentication algorithm */ int8_t isi_rssi; /* receive signal strength */ int8_t isi_noise; /* noise floor */ uint8_t isi_capinfo; /* capabilities */ uint8_t isi_erp; /* ERP element */ uint8_t isi_macaddr[IEEE80211_ADDR_LEN]; uint8_t isi_nrates; /* negotiated rates */ uint8_t isi_rates[IEEE80211_RATE_MAXSIZE]; uint8_t isi_txrate; /* legacy/IEEE rate or MCS */ uint16_t isi_associd; /* assoc response */ uint16_t isi_txpower; /* current tx power */ uint16_t isi_vlan; /* vlan tag */ /* NB: [IEEE80211_NONQOS_TID] holds seq#'s for non-QoS stations */ uint16_t isi_txseqs[IEEE80211_TID_SIZE];/* tx seq #/TID */ uint16_t isi_rxseqs[IEEE80211_TID_SIZE];/* rx seq#/TID */ uint16_t isi_inact; /* inactivity timer */ uint16_t isi_txmbps; /* current tx rate in .5 Mb/s */ uint16_t isi_pad; uint32_t isi_jointime; /* time of assoc/join */ struct ieee80211_mimo_info isi_mimo; /* MIMO info for 11n sta's */ /* 11s info */ uint16_t isi_peerid; uint16_t isi_localid; uint8_t isi_peerstate; /* XXX frag state? */ /* variable length IE data */ }; /* * Retrieve per-station information; to retrieve all * specify a mac address of ff:ff:ff:ff:ff:ff. */ struct ieee80211req_sta_req { union { /* NB: explicitly force 64-bit alignment */ uint8_t macaddr[IEEE80211_ADDR_LEN]; uint64_t pad; } is_u; struct ieee80211req_sta_info info[1]; /* variable length */ }; /* * Get/set per-station tx power cap. */ struct ieee80211req_sta_txpow { uint8_t it_macaddr[IEEE80211_ADDR_LEN]; uint8_t it_txpow; }; /* * WME parameters manipulated with IEEE80211_IOC_WME_CWMIN * through IEEE80211_IOC_WME_ACKPOLICY are set and return * using i_val and i_len. i_val holds the value itself. * i_len specifies the AC and, as appropriate, then high bit * specifies whether the operation is to be applied to the * BSS or ourself. */ #define IEEE80211_WMEPARAM_SELF 0x0000 /* parameter applies to self */ #define IEEE80211_WMEPARAM_BSS 0x8000 /* parameter applies to BSS */ #define IEEE80211_WMEPARAM_VAL 0x7fff /* parameter value */ /* * Application Information Elements can be appended to a variety * of frames with the IEE80211_IOC_APPIE request. This request * piggybacks on a normal ieee80211req; the frame type is passed * in i_val as the 802.11 FC0 bytes and the length of the IE data * is passed in i_len. The data is referenced in i_data. If i_len * is zero then any previously configured IE data is removed. At * most IEEE80211_MAX_APPIE data be appened. Note that multiple * IE's can be supplied; the data is treated opaquely. */ #define IEEE80211_MAX_APPIE 1024 /* max app IE data */ /* * Hack: the WPA authenticator uses this mechanism to specify WPA * ie's that are used instead of the ones normally constructed using * the cipher state setup with separate ioctls. This avoids issues * like the authenticator ordering ie data differently than the * net80211 layer and needing to keep separate state for WPA and RSN. */ #define IEEE80211_APPIE_WPA \ (IEEE80211_FC0_TYPE_MGT | IEEE80211_FC0_SUBTYPE_BEACON | \ IEEE80211_FC0_SUBTYPE_PROBE_RESP) /* * Station mode roaming parameters. These are maintained * per band/mode and control the roaming algorithm. */ struct ieee80211_roamparams_req { struct ieee80211_roamparam params[IEEE80211_MODE_MAX]; }; /* * Transmit parameters. These can be used to set fixed transmit * rate for each operating mode when operating as client or on a * per-client basis according to the capabilities of the client * (e.g. an 11b client associated to an 11g ap) when operating as * an ap. * * MCS are distinguished from legacy rates by or'ing in 0x80. */ struct ieee80211_txparams_req { struct ieee80211_txparam params[IEEE80211_MODE_MAX]; }; /* * Set regulatory domain state with IEEE80211_IOC_REGDOMAIN. * Note this is both the regulatory description and the channel * list. The get request for IEEE80211_IOC_REGDOMAIN returns * only the regdomain info; the channel list is obtained * separately with IEEE80211_IOC_CHANINFO. */ struct ieee80211_regdomain_req { struct ieee80211_regdomain rd; struct ieee80211req_chaninfo chaninfo; }; #define IEEE80211_REGDOMAIN_SIZE(_nchan) \ (sizeof(struct ieee80211_regdomain_req) + \ (((_nchan)-1) * sizeof(struct ieee80211_channel))) #define IEEE80211_REGDOMAIN_SPACE(_req) \ IEEE80211_REGDOMAIN_SIZE((_req)->chaninfo.ic_nchans) /* * Get driver capabilities. Driver, hardware crypto, and * HT/802.11n capabilities, and a table that describes what * the radio can do. */ struct ieee80211_devcaps_req { uint32_t dc_drivercaps; /* general driver caps */ uint32_t dc_cryptocaps; /* hardware crypto support */ uint32_t dc_htcaps; /* HT/802.11n support */ uint32_t dc_vhtcaps; /* VHT/802.11ac capabilities */ struct ieee80211req_chaninfo dc_chaninfo; }; #define IEEE80211_DEVCAPS_SIZE(_nchan) \ (sizeof(struct ieee80211_devcaps_req) + \ (((_nchan)-1) * sizeof(struct ieee80211_channel))) #define IEEE80211_DEVCAPS_SPACE(_dc) \ IEEE80211_DEVCAPS_SIZE((_dc)->dc_chaninfo.ic_nchans) struct ieee80211_chanswitch_req { struct ieee80211_channel csa_chan; /* new channel */ int csa_mode; /* CSA mode */ int csa_count; /* beacon count to switch */ }; /* * Get/set per-station vlan tag. */ struct ieee80211req_sta_vlan { uint8_t sv_macaddr[IEEE80211_ADDR_LEN]; uint16_t sv_vlan; }; #ifdef __FreeBSD__ /* * FreeBSD-style ioctls. */ /* the first member must be matched with struct ifreq */ struct ieee80211req { char i_name[IFNAMSIZ]; /* if_name, e.g. "wi0" */ uint16_t i_type; /* req type */ int16_t i_val; /* Index or simple value */ uint16_t i_len; /* Index or simple value */ void *i_data; /* Extra data */ }; #define SIOCS80211 _IOW('i', 234, struct ieee80211req) #define SIOCG80211 _IOWR('i', 235, struct ieee80211req) #define SIOCG80211STATS _IOWR('i', 236, struct ifreq) #define IEEE80211_IOC_SSID 1 #define IEEE80211_IOC_NUMSSIDS 2 #define IEEE80211_IOC_WEP 3 #define IEEE80211_WEP_NOSUP -1 #define IEEE80211_WEP_OFF 0 #define IEEE80211_WEP_ON 1 #define IEEE80211_WEP_MIXED 2 #define IEEE80211_IOC_WEPKEY 4 #define IEEE80211_IOC_NUMWEPKEYS 5 #define IEEE80211_IOC_WEPTXKEY 6 #define IEEE80211_IOC_AUTHMODE 7 #define IEEE80211_IOC_STATIONNAME 8 #define IEEE80211_IOC_CHANNEL 9 #define IEEE80211_IOC_POWERSAVE 10 #define IEEE80211_POWERSAVE_NOSUP -1 #define IEEE80211_POWERSAVE_OFF 0 #define IEEE80211_POWERSAVE_CAM 1 #define IEEE80211_POWERSAVE_PSP 2 #define IEEE80211_POWERSAVE_PSP_CAM 3 #define IEEE80211_POWERSAVE_ON IEEE80211_POWERSAVE_CAM #define IEEE80211_IOC_POWERSAVESLEEP 11 #define IEEE80211_IOC_RTSTHRESHOLD 12 #define IEEE80211_IOC_PROTMODE 13 #define IEEE80211_PROTMODE_OFF 0 #define IEEE80211_PROTMODE_CTS 1 #define IEEE80211_PROTMODE_RTSCTS 2 #define IEEE80211_IOC_TXPOWER 14 /* global tx power limit */ #define IEEE80211_IOC_BSSID 15 #define IEEE80211_IOC_ROAMING 16 /* roaming mode */ #define IEEE80211_IOC_PRIVACY 17 /* privacy invoked */ #define IEEE80211_IOC_DROPUNENCRYPTED 18 /* discard unencrypted frames */ #define IEEE80211_IOC_WPAKEY 19 #define IEEE80211_IOC_DELKEY 20 #define IEEE80211_IOC_MLME 21 /* 22 was IEEE80211_IOC_OPTIE, replaced by IEEE80211_IOC_APPIE */ /* 23 was IEEE80211_IOC_SCAN_REQ */ /* 24 was IEEE80211_IOC_SCAN_RESULTS */ #define IEEE80211_IOC_COUNTERMEASURES 25 /* WPA/TKIP countermeasures */ #define IEEE80211_IOC_WPA 26 /* WPA mode (0,1,2) */ #define IEEE80211_IOC_CHANLIST 27 /* channel list */ #define IEEE80211_IOC_WME 28 /* WME mode (on, off) */ #define IEEE80211_IOC_HIDESSID 29 /* hide SSID mode (on, off) */ #define IEEE80211_IOC_APBRIDGE 30 /* AP inter-sta bridging */ /* 31-35,37-38 were for WPA authenticator settings */ /* 36 was IEEE80211_IOC_DRIVER_CAPS */ #define IEEE80211_IOC_WPAIE 39 /* WPA information element */ #define IEEE80211_IOC_STA_STATS 40 /* per-station statistics */ #define IEEE80211_IOC_MACCMD 41 /* MAC ACL operation */ #define IEEE80211_IOC_CHANINFO 42 /* channel info list */ #define IEEE80211_IOC_TXPOWMAX 43 /* max tx power for channel */ #define IEEE80211_IOC_STA_TXPOW 44 /* per-station tx power limit */ /* 45 was IEEE80211_IOC_STA_INFO */ #define IEEE80211_IOC_WME_CWMIN 46 /* WME: ECWmin */ #define IEEE80211_IOC_WME_CWMAX 47 /* WME: ECWmax */ #define IEEE80211_IOC_WME_AIFS 48 /* WME: AIFSN */ #define IEEE80211_IOC_WME_TXOPLIMIT 49 /* WME: txops limit */ #define IEEE80211_IOC_WME_ACM 50 /* WME: ACM (bss only) */ #define IEEE80211_IOC_WME_ACKPOLICY 51 /* WME: ACK policy (!bss only)*/ #define IEEE80211_IOC_DTIM_PERIOD 52 /* DTIM period (beacons) */ #define IEEE80211_IOC_BEACON_INTERVAL 53 /* beacon interval (ms) */ #define IEEE80211_IOC_ADDMAC 54 /* add sta to MAC ACL table */ #define IEEE80211_IOC_DELMAC 55 /* del sta from MAC ACL table */ #define IEEE80211_IOC_PUREG 56 /* pure 11g (no 11b stations) */ #define IEEE80211_IOC_FF 57 /* ATH fast frames (on, off) */ #define IEEE80211_IOC_TURBOP 58 /* ATH turbo' (on, off) */ #define IEEE80211_IOC_BGSCAN 59 /* bg scanning (on, off) */ #define IEEE80211_IOC_BGSCAN_IDLE 60 /* bg scan idle threshold */ #define IEEE80211_IOC_BGSCAN_INTERVAL 61 /* bg scan interval */ #define IEEE80211_IOC_SCANVALID 65 /* scan cache valid threshold */ /* 66-72 were IEEE80211_IOC_ROAM_* and IEEE80211_IOC_MCAST_RATE */ #define IEEE80211_IOC_FRAGTHRESHOLD 73 /* tx fragmentation threshold */ #define IEEE80211_IOC_BURST 75 /* packet bursting */ #define IEEE80211_IOC_SCAN_RESULTS 76 /* get scan results */ #define IEEE80211_IOC_BMISSTHRESHOLD 77 /* beacon miss threshold */ #define IEEE80211_IOC_STA_INFO 78 /* station/neighbor info */ #define IEEE80211_IOC_WPAIE2 79 /* WPA+RSN info elements */ #define IEEE80211_IOC_CURCHAN 80 /* current channel */ #define IEEE80211_IOC_SHORTGI 81 /* 802.11n half GI */ #define IEEE80211_IOC_AMPDU 82 /* 802.11n A-MPDU (on, off) */ #define IEEE80211_IOC_AMPDU_LIMIT 83 /* A-MPDU length limit */ #define IEEE80211_IOC_AMPDU_DENSITY 84 /* A-MPDU density */ #define IEEE80211_IOC_AMSDU 85 /* 802.11n A-MSDU (on, off) */ #define IEEE80211_IOC_AMSDU_LIMIT 86 /* A-MSDU length limit */ #define IEEE80211_IOC_PUREN 87 /* pure 11n (no legacy sta's) */ #define IEEE80211_IOC_DOTH 88 /* 802.11h (on, off) */ /* 89-91 were regulatory items */ #define IEEE80211_IOC_HTCOMPAT 92 /* support pre-D1.10 HT ie's */ #define IEEE80211_IOC_DWDS 93 /* DWDS/4-address handling */ #define IEEE80211_IOC_INACTIVITY 94 /* sta inactivity handling */ #define IEEE80211_IOC_APPIE 95 /* application IE's */ #define IEEE80211_IOC_WPS 96 /* WPS operation */ #define IEEE80211_IOC_TSN 97 /* TSN operation */ #define IEEE80211_IOC_DEVCAPS 98 /* driver+device capabilities */ #define IEEE80211_IOC_CHANSWITCH 99 /* start 11h channel switch */ #define IEEE80211_IOC_DFS 100 /* DFS (on, off) */ #define IEEE80211_IOC_DOTD 101 /* 802.11d (on, off) */ #define IEEE80211_IOC_HTPROTMODE 102 /* HT protection (off, rts) */ #define IEEE80211_IOC_SCAN_REQ 103 /* scan w/ specified params */ #define IEEE80211_IOC_SCAN_CANCEL 104 /* cancel ongoing scan */ #define IEEE80211_IOC_HTCONF 105 /* HT config (off, HT20, HT40)*/ #define IEEE80211_IOC_REGDOMAIN 106 /* regulatory domain info */ #define IEEE80211_IOC_ROAM 107 /* roaming params en masse */ #define IEEE80211_IOC_TXPARAMS 108 /* tx parameters */ #define IEEE80211_IOC_STA_VLAN 109 /* per-station vlan tag */ #define IEEE80211_IOC_SMPS 110 /* MIMO power save */ #define IEEE80211_IOC_RIFS 111 /* RIFS config (on, off) */ #define IEEE80211_IOC_GREENFIELD 112 /* Greenfield (on, off) */ #define IEEE80211_IOC_STBC 113 /* STBC Tx/RX (on, off) */ #define IEEE80211_IOC_LDPC 114 /* LDPC Tx/RX (on, off) */ #define IEEE80211_IOC_UAPSD 115 /* UAPSD (on, off) */ #define IEEE80211_IOC_UAPSD_INFO 116 /* UAPSD (SP, per-AC enable) */ /* VHT */ #define IEEE80211_IOC_VHTCONF 130 /* VHT config (off, on; widths) */ #define IEEE80211_IOC_MESH_ID 170 /* mesh identifier */ #define IEEE80211_IOC_MESH_AP 171 /* accepting peerings */ #define IEEE80211_IOC_MESH_FWRD 172 /* forward frames */ #define IEEE80211_IOC_MESH_PROTO 173 /* mesh protocols */ #define IEEE80211_IOC_MESH_TTL 174 /* mesh TTL */ #define IEEE80211_IOC_MESH_RTCMD 175 /* mesh routing table commands*/ #define IEEE80211_IOC_MESH_PR_METRIC 176 /* mesh metric protocol */ #define IEEE80211_IOC_MESH_PR_PATH 177 /* mesh path protocol */ #define IEEE80211_IOC_MESH_PR_SIG 178 /* mesh sig protocol */ #define IEEE80211_IOC_MESH_PR_CC 179 /* mesh congestion protocol */ #define IEEE80211_IOC_MESH_PR_AUTH 180 /* mesh auth protocol */ #define IEEE80211_IOC_MESH_GATE 181 /* mesh gate XXX: 173? */ #define IEEE80211_IOC_HWMP_ROOTMODE 190 /* HWMP root mode */ #define IEEE80211_IOC_HWMP_MAXHOPS 191 /* number of hops before drop */ #define IEEE80211_IOC_HWMP_TTL 192 /* HWMP TTL */ #define IEEE80211_IOC_TDMA_SLOT 201 /* TDMA: assigned slot */ #define IEEE80211_IOC_TDMA_SLOTCNT 202 /* TDMA: slots in bss */ #define IEEE80211_IOC_TDMA_SLOTLEN 203 /* TDMA: slot length (usecs) */ #define IEEE80211_IOC_TDMA_BINTERVAL 204 /* TDMA: beacon intvl (slots) */ #define IEEE80211_IOC_QUIET 205 /* Quiet Enable/Disable */ #define IEEE80211_IOC_QUIET_PERIOD 206 /* Quiet Period */ #define IEEE80211_IOC_QUIET_OFFSET 207 /* Quiet Offset */ #define IEEE80211_IOC_QUIET_DUR 208 /* Quiet Duration */ #define IEEE80211_IOC_QUIET_COUNT 209 /* Quiet Count */ + +#define IEEE80211_IOC_IC_NAME 210 /* HW device name. */ + /* * Parameters for controlling a scan requested with * IEEE80211_IOC_SCAN_REQ. * * Active scans cause ProbeRequest frames to be issued for each * specified ssid and, by default, a broadcast ProbeRequest frame. * The set of ssid's is specified in the request. * * By default the scan will cause a BSS to be joined (in station/adhoc * mode) or a channel to be selected for operation (hostap mode). * To disable that specify IEEE80211_IOC_SCAN_NOPICK and if the * * If the station is currently associated to an AP then a scan request * will cause the station to leave the current channel and potentially * miss frames from the AP. Alternatively the station may notify the * AP that it is going into power save mode before it leaves the channel. * This ensures frames for the station are buffered by the AP. This is * termed a ``bg scan'' and is requested with the IEEE80211_IOC_SCAN_BGSCAN * flag. Background scans may take longer than foreground scans and may * be preempted by traffic. If a station is not associated to an AP * then a request for a background scan is automatically done in the * foreground. * * The results of the scan request are cached by the system. This * information is aged out and/or invalidated based on events like not * being able to associated to an AP. To flush the current cache * contents before doing a scan the IEEE80211_IOC_SCAN_FLUSH flag may * be specified. * * By default the scan will be done until a suitable AP is located * or a channel is found for use. A scan can also be constrained * to be done once (IEEE80211_IOC_SCAN_ONCE) or to last for no more * than a specified duration. */ struct ieee80211_scan_req { int sr_flags; #define IEEE80211_IOC_SCAN_NOPICK 0x00001 /* scan only, no selection */ #define IEEE80211_IOC_SCAN_ACTIVE 0x00002 /* active scan (probe req) */ #define IEEE80211_IOC_SCAN_PICK1ST 0x00004 /* ``hey sailor'' mode */ #define IEEE80211_IOC_SCAN_BGSCAN 0x00008 /* bg scan, exit ps at end */ #define IEEE80211_IOC_SCAN_ONCE 0x00010 /* do one complete pass */ #define IEEE80211_IOC_SCAN_NOBCAST 0x00020 /* don't send bcast probe req */ #define IEEE80211_IOC_SCAN_NOJOIN 0x00040 /* no auto-sequencing */ #define IEEE80211_IOC_SCAN_FLUSH 0x10000 /* flush scan cache first */ #define IEEE80211_IOC_SCAN_CHECK 0x20000 /* check scan cache first */ u_int sr_duration; /* duration (ms) */ #define IEEE80211_IOC_SCAN_DURATION_MIN 1 #define IEEE80211_IOC_SCAN_DURATION_MAX 0x7fffffff #define IEEE80211_IOC_SCAN_FOREVER IEEE80211_IOC_SCAN_DURATION_MAX u_int sr_mindwell; /* min channel dwelltime (ms) */ u_int sr_maxdwell; /* max channel dwelltime (ms) */ int sr_nssid; #define IEEE80211_IOC_SCAN_MAX_SSID 3 struct { int len; /* length in bytes */ uint8_t ssid[IEEE80211_NWID_LEN]; /* ssid contents */ } sr_ssid[IEEE80211_IOC_SCAN_MAX_SSID]; }; /* * Scan result data returned for IEEE80211_IOC_SCAN_RESULTS. * Each result is a fixed size structure followed by a variable * length SSID and one or more variable length information elements. * The size of each variable length item is found in the fixed * size structure and the entire length of the record is specified * in isr_len. Result records are rounded to a multiple of 4 bytes. */ struct ieee80211req_scan_result { uint16_t isr_len; /* total length (mult of 4) */ uint16_t isr_ie_off; /* offset to SSID+IE data */ uint16_t isr_ie_len; /* IE length */ uint16_t isr_freq; /* MHz */ uint16_t isr_flags; /* channel flags */ int8_t isr_noise; int8_t isr_rssi; uint16_t isr_intval; /* beacon interval */ uint8_t isr_capinfo; /* capabilities */ uint8_t isr_erp; /* ERP element */ uint8_t isr_bssid[IEEE80211_ADDR_LEN]; uint8_t isr_nrates; uint8_t isr_rates[IEEE80211_RATE_MAXSIZE]; uint8_t isr_ssid_len; /* SSID length */ uint8_t isr_meshid_len; /* MESH ID length */ /* variable length SSID, followed by variable length MESH ID, followed by IE data */ }; /* * Virtual AP cloning parameters. The parent device must * be a vap-capable device. All parameters specified with * the clone request are fixed for the lifetime of the vap. * * There are two flavors of WDS vaps: legacy and dynamic. * Legacy WDS operation implements a static binding between * two stations encapsulating traffic in 4-address frames. * Dynamic WDS vaps are created when a station associates to * an AP and sends a 4-address frame. If the AP vap is * configured to support WDS then this will generate an * event to user programs listening on the routing socket * and a Dynamic WDS vap will be created to handle traffic * to/from that station. In both cases the bssid of the * peer must be specified when creating the vap. * * By default a vap will inherit the mac address/bssid of * the underlying device. To request a unique address the * IEEE80211_CLONE_BSSID flag should be supplied. This is * meaningless for WDS vaps as they share the bssid of an * AP vap that must otherwise exist. Note that some devices * may not be able to support multiple addresses. * * Station mode vap's normally depend on the device to notice * when the AP stops sending beacon frames. If IEEE80211_CLONE_NOBEACONS * is specified the net80211 layer will do this in s/w. This * is mostly useful when setting up a WDS repeater/extender where * an AP vap is combined with a sta vap and the device isn't able * to track beacon frames in hardware. */ struct ieee80211_clone_params { char icp_parent[IFNAMSIZ]; /* parent device */ uint16_t icp_opmode; /* operating mode */ uint16_t icp_flags; /* see below */ uint8_t icp_bssid[IEEE80211_ADDR_LEN]; /* for WDS links */ uint8_t icp_macaddr[IEEE80211_ADDR_LEN];/* local address */ }; #define IEEE80211_CLONE_BSSID 0x0001 /* allocate unique mac/bssid */ #define IEEE80211_CLONE_NOBEACONS 0x0002 /* don't setup beacon timers */ #define IEEE80211_CLONE_WDSLEGACY 0x0004 /* legacy WDS processing */ #define IEEE80211_CLONE_MACADDR 0x0008 /* use specified mac addr */ #define IEEE80211_CLONE_TDMA 0x0010 /* operate in TDMA mode */ #endif /* __FreeBSD__ */ #endif /* _NET80211_IEEE80211_IOCTL_H_ */ Index: projects/clang1100-import/sys/netinet/udp_usrreq.c =================================================================== --- projects/clang1100-import/sys/netinet/udp_usrreq.c (revision 364034) +++ projects/clang1100-import/sys/netinet/udp_usrreq.c (revision 364035) @@ -1,1734 +1,1795 @@ /*- * SPDX-License-Identifier: BSD-3-Clause * * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995 * The Regents of the University of California. * Copyright (c) 2008 Robert N. M. Watson * Copyright (c) 2010-2011 Juniper Networks, Inc. * Copyright (c) 2014 Kevin Lo * All rights reserved. * * Portions of this software were developed by Robert N. M. Watson under * contract to Juniper Networks, Inc. * * 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. * * @(#)udp_usrreq.c 8.6 (Berkeley) 5/23/95 */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" #include "opt_rss.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 #include #include #include #include #include #include #include #ifdef INET6 #include #endif #include #include #include #include #ifdef INET6 #include #endif #include #include #include #include #include #include #include /* * UDP and UDP-Lite protocols implementation. * Per RFC 768, August, 1980. * Per RFC 3828, July, 2004. */ /* * BSD 4.2 defaulted the udp checksum to be off. Turning off udp checksums * removes the only data integrity mechanism for packets and malformed * packets that would otherwise be discarded due to bad checksums, and may * cause problems (especially for NFS data blocks). */ VNET_DEFINE(int, udp_cksum) = 1; SYSCTL_INT(_net_inet_udp, UDPCTL_CHECKSUM, checksum, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(udp_cksum), 0, "compute udp checksum"); VNET_DEFINE(int, udp_log_in_vain) = 0; SYSCTL_INT(_net_inet_udp, OID_AUTO, log_in_vain, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(udp_log_in_vain), 0, "Log all incoming UDP packets"); VNET_DEFINE(int, udp_blackhole) = 0; SYSCTL_INT(_net_inet_udp, OID_AUTO, blackhole, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(udp_blackhole), 0, "Do not send port unreachables for refused connects"); u_long udp_sendspace = 9216; /* really max datagram size */ SYSCTL_ULONG(_net_inet_udp, UDPCTL_MAXDGRAM, maxdgram, CTLFLAG_RW, &udp_sendspace, 0, "Maximum outgoing UDP datagram size"); u_long udp_recvspace = 40 * (1024 + #ifdef INET6 sizeof(struct sockaddr_in6) #else sizeof(struct sockaddr_in) #endif ); /* 40 1K datagrams */ SYSCTL_ULONG(_net_inet_udp, UDPCTL_RECVSPACE, recvspace, CTLFLAG_RW, &udp_recvspace, 0, "Maximum space for incoming UDP datagrams"); VNET_DEFINE(struct inpcbhead, udb); /* from udp_var.h */ VNET_DEFINE(struct inpcbinfo, udbinfo); VNET_DEFINE(struct inpcbhead, ulitecb); VNET_DEFINE(struct inpcbinfo, ulitecbinfo); VNET_DEFINE_STATIC(uma_zone_t, udpcb_zone); #define V_udpcb_zone VNET(udpcb_zone) #ifndef UDBHASHSIZE #define UDBHASHSIZE 128 #endif VNET_PCPUSTAT_DEFINE(struct udpstat, udpstat); /* from udp_var.h */ VNET_PCPUSTAT_SYSINIT(udpstat); SYSCTL_VNET_PCPUSTAT(_net_inet_udp, UDPCTL_STATS, stats, struct udpstat, udpstat, "UDP statistics (struct udpstat, netinet/udp_var.h)"); #ifdef VIMAGE VNET_PCPUSTAT_SYSUNINIT(udpstat); #endif /* VIMAGE */ #ifdef INET static void udp_detach(struct socket *so); static int udp_output(struct inpcb *, struct mbuf *, struct sockaddr *, - struct mbuf *, struct thread *); + struct mbuf *, struct thread *, int); #endif static void udp_zone_change(void *tag) { uma_zone_set_max(V_udbinfo.ipi_zone, maxsockets); uma_zone_set_max(V_udpcb_zone, maxsockets); } static int udp_inpcb_init(void *mem, int size, int flags) { struct inpcb *inp; inp = mem; INP_LOCK_INIT(inp, "inp", "udpinp"); return (0); } static int udplite_inpcb_init(void *mem, int size, int flags) { struct inpcb *inp; inp = mem; INP_LOCK_INIT(inp, "inp", "udpliteinp"); return (0); } void udp_init(void) { /* * For now default to 2-tuple UDP hashing - until the fragment * reassembly code can also update the flowid. * * Once we can calculate the flowid that way and re-establish * a 4-tuple, flip this to 4-tuple. */ in_pcbinfo_init(&V_udbinfo, "udp", &V_udb, UDBHASHSIZE, UDBHASHSIZE, "udp_inpcb", udp_inpcb_init, IPI_HASHFIELDS_2TUPLE); V_udpcb_zone = uma_zcreate("udpcb", sizeof(struct udpcb), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0); uma_zone_set_max(V_udpcb_zone, maxsockets); uma_zone_set_warning(V_udpcb_zone, "kern.ipc.maxsockets limit reached"); EVENTHANDLER_REGISTER(maxsockets_change, udp_zone_change, NULL, EVENTHANDLER_PRI_ANY); } void udplite_init(void) { in_pcbinfo_init(&V_ulitecbinfo, "udplite", &V_ulitecb, UDBHASHSIZE, UDBHASHSIZE, "udplite_inpcb", udplite_inpcb_init, IPI_HASHFIELDS_2TUPLE); } /* * Kernel module interface for updating udpstat. The argument is an index * into udpstat treated as an array of u_long. While this encodes the * general layout of udpstat into the caller, it doesn't encode its location, * so that future changes to add, for example, per-CPU stats support won't * cause binary compatibility problems for kernel modules. */ void kmod_udpstat_inc(int statnum) { counter_u64_add(VNET(udpstat)[statnum], 1); } int udp_newudpcb(struct inpcb *inp) { struct udpcb *up; up = uma_zalloc(V_udpcb_zone, M_NOWAIT | M_ZERO); if (up == NULL) return (ENOBUFS); inp->inp_ppcb = up; return (0); } void udp_discardcb(struct udpcb *up) { uma_zfree(V_udpcb_zone, up); } #ifdef VIMAGE static void udp_destroy(void *unused __unused) { in_pcbinfo_destroy(&V_udbinfo); uma_zdestroy(V_udpcb_zone); } VNET_SYSUNINIT(udp, SI_SUB_PROTO_DOMAIN, SI_ORDER_FOURTH, udp_destroy, NULL); static void udplite_destroy(void *unused __unused) { in_pcbinfo_destroy(&V_ulitecbinfo); } VNET_SYSUNINIT(udplite, SI_SUB_PROTO_DOMAIN, SI_ORDER_FOURTH, udplite_destroy, NULL); #endif #ifdef INET /* * Subroutine of udp_input(), which appends the provided mbuf chain to the * passed pcb/socket. The caller must provide a sockaddr_in via udp_in that * contains the source address. If the socket ends up being an IPv6 socket, * udp_append() will convert to a sockaddr_in6 before passing the address * into the socket code. * * In the normal case udp_append() will return 0, indicating that you * must unlock the inp. However if a tunneling protocol is in place we increment * the inpcb refcnt and unlock the inp, on return from the tunneling protocol we * then decrement the reference count. If the inp_rele returns 1, indicating the * inp is gone, we return that to the caller to tell them *not* to unlock * the inp. In the case of multi-cast this will cause the distribution * to stop (though most tunneling protocols known currently do *not* use * multicast). */ static int udp_append(struct inpcb *inp, struct ip *ip, struct mbuf *n, int off, struct sockaddr_in *udp_in) { struct sockaddr *append_sa; struct socket *so; struct mbuf *tmpopts, *opts = NULL; #ifdef INET6 struct sockaddr_in6 udp_in6; #endif struct udpcb *up; INP_LOCK_ASSERT(inp); /* * Engage the tunneling protocol. */ up = intoudpcb(inp); if (up->u_tun_func != NULL) { in_pcbref(inp); INP_RUNLOCK(inp); (*up->u_tun_func)(n, off, inp, (struct sockaddr *)&udp_in[0], up->u_tun_ctx); INP_RLOCK(inp); return (in_pcbrele_rlocked(inp)); } off += sizeof(struct udphdr); #if defined(IPSEC) || defined(IPSEC_SUPPORT) /* Check AH/ESP integrity. */ if (IPSEC_ENABLED(ipv4) && IPSEC_CHECK_POLICY(ipv4, n, inp) != 0) { m_freem(n); return (0); } if (up->u_flags & UF_ESPINUDP) {/* IPSec UDP encaps. */ if (IPSEC_ENABLED(ipv4) && UDPENCAP_INPUT(n, off, AF_INET) != 0) return (0); /* Consumed. */ } #endif /* IPSEC */ #ifdef MAC if (mac_inpcb_check_deliver(inp, n) != 0) { m_freem(n); return (0); } #endif /* MAC */ if (inp->inp_flags & INP_CONTROLOPTS || inp->inp_socket->so_options & (SO_TIMESTAMP | SO_BINTIME)) { #ifdef INET6 if (inp->inp_vflag & INP_IPV6) (void)ip6_savecontrol_v4(inp, n, &opts, NULL); else #endif /* INET6 */ ip_savecontrol(inp, &opts, ip, n); } if ((inp->inp_vflag & INP_IPV4) && (inp->inp_flags2 & INP_ORIGDSTADDR)) { tmpopts = sbcreatecontrol((caddr_t)&udp_in[1], sizeof(struct sockaddr_in), IP_ORIGDSTADDR, IPPROTO_IP); if (tmpopts) { if (opts) { tmpopts->m_next = opts; opts = tmpopts; } else opts = tmpopts; } } #ifdef INET6 if (inp->inp_vflag & INP_IPV6) { bzero(&udp_in6, sizeof(udp_in6)); udp_in6.sin6_len = sizeof(udp_in6); udp_in6.sin6_family = AF_INET6; in6_sin_2_v4mapsin6(&udp_in[0], &udp_in6); append_sa = (struct sockaddr *)&udp_in6; } else #endif /* INET6 */ append_sa = (struct sockaddr *)&udp_in[0]; m_adj(n, off); so = inp->inp_socket; SOCKBUF_LOCK(&so->so_rcv); if (sbappendaddr_locked(&so->so_rcv, append_sa, n, opts) == 0) { SOCKBUF_UNLOCK(&so->so_rcv); m_freem(n); if (opts) m_freem(opts); UDPSTAT_INC(udps_fullsock); } else sorwakeup_locked(so); return (0); } int udp_input(struct mbuf **mp, int *offp, int proto) { struct ip *ip; struct udphdr *uh; struct ifnet *ifp; struct inpcb *inp; uint16_t len, ip_len; struct inpcbinfo *pcbinfo; struct ip save_ip; struct sockaddr_in udp_in[2]; struct mbuf *m; struct m_tag *fwd_tag; int cscov_partial, iphlen; m = *mp; iphlen = *offp; ifp = m->m_pkthdr.rcvif; *mp = NULL; UDPSTAT_INC(udps_ipackets); /* * Strip IP options, if any; should skip this, make available to * user, and use on returned packets, but we don't yet have a way to * check the checksum with options still present. */ if (iphlen > sizeof (struct ip)) { ip_stripoptions(m); iphlen = sizeof(struct ip); } /* * Get IP and UDP header together in first mbuf. */ if (m->m_len < iphlen + sizeof(struct udphdr)) { if ((m = m_pullup(m, iphlen + sizeof(struct udphdr))) == NULL) { UDPSTAT_INC(udps_hdrops); return (IPPROTO_DONE); } } ip = mtod(m, struct ip *); uh = (struct udphdr *)((caddr_t)ip + iphlen); cscov_partial = (proto == IPPROTO_UDPLITE) ? 1 : 0; /* * Destination port of 0 is illegal, based on RFC768. */ if (uh->uh_dport == 0) goto badunlocked; /* * Construct sockaddr format source address. Stuff source address * and datagram in user buffer. */ bzero(&udp_in[0], sizeof(struct sockaddr_in) * 2); udp_in[0].sin_len = sizeof(struct sockaddr_in); udp_in[0].sin_family = AF_INET; udp_in[0].sin_port = uh->uh_sport; udp_in[0].sin_addr = ip->ip_src; udp_in[1].sin_len = sizeof(struct sockaddr_in); udp_in[1].sin_family = AF_INET; udp_in[1].sin_port = uh->uh_dport; udp_in[1].sin_addr = ip->ip_dst; /* * Make mbuf data length reflect UDP length. If not enough data to * reflect UDP length, drop. */ len = ntohs((u_short)uh->uh_ulen); ip_len = ntohs(ip->ip_len) - iphlen; if (proto == IPPROTO_UDPLITE && (len == 0 || len == ip_len)) { /* Zero means checksum over the complete packet. */ if (len == 0) len = ip_len; cscov_partial = 0; } if (ip_len != len) { if (len > ip_len || len < sizeof(struct udphdr)) { UDPSTAT_INC(udps_badlen); goto badunlocked; } if (proto == IPPROTO_UDP) m_adj(m, len - ip_len); } /* * Save a copy of the IP header in case we want restore it for * sending an ICMP error message in response. */ if (!V_udp_blackhole) save_ip = *ip; else memset(&save_ip, 0, sizeof(save_ip)); /* * Checksum extended UDP header and data. */ if (uh->uh_sum) { u_short uh_sum; if ((m->m_pkthdr.csum_flags & CSUM_DATA_VALID) && !cscov_partial) { if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR) uh_sum = m->m_pkthdr.csum_data; else uh_sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, htonl((u_short)len + m->m_pkthdr.csum_data + proto)); uh_sum ^= 0xffff; } else { char b[9]; bcopy(((struct ipovly *)ip)->ih_x1, b, 9); bzero(((struct ipovly *)ip)->ih_x1, 9); ((struct ipovly *)ip)->ih_len = (proto == IPPROTO_UDP) ? uh->uh_ulen : htons(ip_len); uh_sum = in_cksum(m, len + sizeof (struct ip)); bcopy(b, ((struct ipovly *)ip)->ih_x1, 9); } if (uh_sum) { UDPSTAT_INC(udps_badsum); m_freem(m); return (IPPROTO_DONE); } } else { if (proto == IPPROTO_UDP) { UDPSTAT_INC(udps_nosum); } else { /* UDPLite requires a checksum */ /* XXX: What is the right UDPLite MIB counter here? */ m_freem(m); return (IPPROTO_DONE); } } pcbinfo = udp_get_inpcbinfo(proto); if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) || in_broadcast(ip->ip_dst, ifp)) { struct inpcb *last; struct inpcbhead *pcblist; NET_EPOCH_ASSERT(); pcblist = udp_get_pcblist(proto); last = NULL; CK_LIST_FOREACH(inp, pcblist, inp_list) { if (inp->inp_lport != uh->uh_dport) continue; #ifdef INET6 if ((inp->inp_vflag & INP_IPV4) == 0) continue; #endif if (inp->inp_laddr.s_addr != INADDR_ANY && inp->inp_laddr.s_addr != ip->ip_dst.s_addr) continue; if (inp->inp_faddr.s_addr != INADDR_ANY && inp->inp_faddr.s_addr != ip->ip_src.s_addr) continue; if (inp->inp_fport != 0 && inp->inp_fport != uh->uh_sport) continue; INP_RLOCK(inp); if (__predict_false(inp->inp_flags2 & INP_FREED)) { INP_RUNLOCK(inp); continue; } /* * XXXRW: Because we weren't holding either the inpcb * or the hash lock when we checked for a match * before, we should probably recheck now that the * inpcb lock is held. */ /* * Handle socket delivery policy for any-source * and source-specific multicast. [RFC3678] */ if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { struct ip_moptions *imo; struct sockaddr_in group; int blocked; imo = inp->inp_moptions; if (imo == NULL) { INP_RUNLOCK(inp); continue; } bzero(&group, sizeof(struct sockaddr_in)); group.sin_len = sizeof(struct sockaddr_in); group.sin_family = AF_INET; group.sin_addr = ip->ip_dst; blocked = imo_multi_filter(imo, ifp, (struct sockaddr *)&group, (struct sockaddr *)&udp_in[0]); if (blocked != MCAST_PASS) { if (blocked == MCAST_NOTGMEMBER) IPSTAT_INC(ips_notmember); if (blocked == MCAST_NOTSMEMBER || blocked == MCAST_MUTED) UDPSTAT_INC(udps_filtermcast); INP_RUNLOCK(inp); continue; } } if (last != NULL) { struct mbuf *n; if ((n = m_copym(m, 0, M_COPYALL, M_NOWAIT)) != NULL) { if (proto == IPPROTO_UDPLITE) UDPLITE_PROBE(receive, NULL, last, ip, last, uh); else UDP_PROBE(receive, NULL, last, ip, last, uh); if (udp_append(last, ip, n, iphlen, udp_in)) { goto inp_lost; } } INP_RUNLOCK(last); } last = inp; /* * Don't look for additional matches if this one does * not have either the SO_REUSEPORT or SO_REUSEADDR * socket options set. This heuristic avoids * searching through all pcbs in the common case of a * non-shared port. It assumes that an application * will never clear these options after setting them. */ if ((last->inp_socket->so_options & (SO_REUSEPORT|SO_REUSEPORT_LB|SO_REUSEADDR)) == 0) break; } if (last == NULL) { /* * No matching pcb found; discard datagram. (No need * to send an ICMP Port Unreachable for a broadcast * or multicast datgram.) */ UDPSTAT_INC(udps_noportbcast); if (inp) INP_RUNLOCK(inp); goto badunlocked; } if (proto == IPPROTO_UDPLITE) UDPLITE_PROBE(receive, NULL, last, ip, last, uh); else UDP_PROBE(receive, NULL, last, ip, last, uh); if (udp_append(last, ip, m, iphlen, udp_in) == 0) INP_RUNLOCK(last); inp_lost: return (IPPROTO_DONE); } /* * Locate pcb for datagram. */ /* * Grab info from PACKET_TAG_IPFORWARD tag prepended to the chain. */ if ((m->m_flags & M_IP_NEXTHOP) && (fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL) { struct sockaddr_in *next_hop; next_hop = (struct sockaddr_in *)(fwd_tag + 1); /* * Transparently forwarded. Pretend to be the destination. * Already got one like this? */ inp = in_pcblookup_mbuf(pcbinfo, ip->ip_src, uh->uh_sport, ip->ip_dst, uh->uh_dport, INPLOOKUP_RLOCKPCB, ifp, m); if (!inp) { /* * It's new. Try to find the ambushing socket. * Because we've rewritten the destination address, * any hardware-generated hash is ignored. */ inp = in_pcblookup(pcbinfo, ip->ip_src, uh->uh_sport, next_hop->sin_addr, next_hop->sin_port ? htons(next_hop->sin_port) : uh->uh_dport, INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB, ifp); } /* Remove the tag from the packet. We don't need it anymore. */ m_tag_delete(m, fwd_tag); m->m_flags &= ~M_IP_NEXTHOP; } else inp = in_pcblookup_mbuf(pcbinfo, ip->ip_src, uh->uh_sport, ip->ip_dst, uh->uh_dport, INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB, ifp, m); if (inp == NULL) { if (V_udp_log_in_vain) { char src[INET_ADDRSTRLEN]; char dst[INET_ADDRSTRLEN]; log(LOG_INFO, "Connection attempt to UDP %s:%d from %s:%d\n", inet_ntoa_r(ip->ip_dst, dst), ntohs(uh->uh_dport), inet_ntoa_r(ip->ip_src, src), ntohs(uh->uh_sport)); } if (proto == IPPROTO_UDPLITE) UDPLITE_PROBE(receive, NULL, NULL, ip, NULL, uh); else UDP_PROBE(receive, NULL, NULL, ip, NULL, uh); UDPSTAT_INC(udps_noport); if (m->m_flags & (M_BCAST | M_MCAST)) { UDPSTAT_INC(udps_noportbcast); goto badunlocked; } if (V_udp_blackhole) goto badunlocked; if (badport_bandlim(BANDLIM_ICMP_UNREACH) < 0) goto badunlocked; *ip = save_ip; icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0); return (IPPROTO_DONE); } /* * Check the minimum TTL for socket. */ INP_RLOCK_ASSERT(inp); if (inp->inp_ip_minttl && inp->inp_ip_minttl > ip->ip_ttl) { if (proto == IPPROTO_UDPLITE) UDPLITE_PROBE(receive, NULL, inp, ip, inp, uh); else UDP_PROBE(receive, NULL, inp, ip, inp, uh); INP_RUNLOCK(inp); m_freem(m); return (IPPROTO_DONE); } if (cscov_partial) { struct udpcb *up; up = intoudpcb(inp); if (up->u_rxcslen == 0 || up->u_rxcslen > len) { INP_RUNLOCK(inp); m_freem(m); return (IPPROTO_DONE); } } if (proto == IPPROTO_UDPLITE) UDPLITE_PROBE(receive, NULL, inp, ip, inp, uh); else UDP_PROBE(receive, NULL, inp, ip, inp, uh); if (udp_append(inp, ip, m, iphlen, udp_in) == 0) INP_RUNLOCK(inp); return (IPPROTO_DONE); badunlocked: m_freem(m); return (IPPROTO_DONE); } #endif /* INET */ /* * Notify a udp user of an asynchronous error; just wake up so that they can * collect error status. */ struct inpcb * udp_notify(struct inpcb *inp, int errno) { INP_WLOCK_ASSERT(inp); if ((errno == EHOSTUNREACH || errno == ENETUNREACH || errno == EHOSTDOWN) && inp->inp_route.ro_nh) { NH_FREE(inp->inp_route.ro_nh); inp->inp_route.ro_nh = (struct nhop_object *)NULL; } inp->inp_socket->so_error = errno; sorwakeup(inp->inp_socket); sowwakeup(inp->inp_socket); return (inp); } #ifdef INET static void udp_common_ctlinput(int cmd, struct sockaddr *sa, void *vip, struct inpcbinfo *pcbinfo) { struct ip *ip = vip; struct udphdr *uh; struct in_addr faddr; struct inpcb *inp; faddr = ((struct sockaddr_in *)sa)->sin_addr; if (sa->sa_family != AF_INET || faddr.s_addr == INADDR_ANY) return; if (PRC_IS_REDIRECT(cmd)) { /* signal EHOSTDOWN, as it flushes the cached route */ in_pcbnotifyall(&V_udbinfo, faddr, EHOSTDOWN, udp_notify); return; } /* * Hostdead is ugly because it goes linearly through all PCBs. * * XXX: We never get this from ICMP, otherwise it makes an excellent * DoS attack on machines with many connections. */ if (cmd == PRC_HOSTDEAD) ip = NULL; else if ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0) return; if (ip != NULL) { uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2)); inp = in_pcblookup(pcbinfo, faddr, uh->uh_dport, ip->ip_src, uh->uh_sport, INPLOOKUP_WLOCKPCB, NULL); if (inp != NULL) { INP_WLOCK_ASSERT(inp); if (inp->inp_socket != NULL) { udp_notify(inp, inetctlerrmap[cmd]); } INP_WUNLOCK(inp); } else { inp = in_pcblookup(pcbinfo, faddr, uh->uh_dport, ip->ip_src, uh->uh_sport, INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB, NULL); if (inp != NULL) { struct udpcb *up; void *ctx; udp_tun_icmp_t func; up = intoudpcb(inp); ctx = up->u_tun_ctx; func = up->u_icmp_func; INP_RUNLOCK(inp); if (func != NULL) (*func)(cmd, sa, vip, ctx); } } } else in_pcbnotifyall(pcbinfo, faddr, inetctlerrmap[cmd], udp_notify); } void udp_ctlinput(int cmd, struct sockaddr *sa, void *vip) { return (udp_common_ctlinput(cmd, sa, vip, &V_udbinfo)); } void udplite_ctlinput(int cmd, struct sockaddr *sa, void *vip) { return (udp_common_ctlinput(cmd, sa, vip, &V_ulitecbinfo)); } #endif /* INET */ static int udp_pcblist(SYSCTL_HANDLER_ARGS) { struct xinpgen xig; struct epoch_tracker et; struct inpcb *inp; int error; if (req->newptr != 0) return (EPERM); if (req->oldptr == 0) { int n; n = V_udbinfo.ipi_count; n += imax(n / 8, 10); req->oldidx = 2 * (sizeof xig) + n * sizeof(struct xinpcb); return (0); } if ((error = sysctl_wire_old_buffer(req, 0)) != 0) return (error); bzero(&xig, sizeof(xig)); xig.xig_len = sizeof xig; xig.xig_count = V_udbinfo.ipi_count; xig.xig_gen = V_udbinfo.ipi_gencnt; xig.xig_sogen = so_gencnt; error = SYSCTL_OUT(req, &xig, sizeof xig); if (error) return (error); NET_EPOCH_ENTER(et); for (inp = CK_LIST_FIRST(V_udbinfo.ipi_listhead); inp != NULL; inp = CK_LIST_NEXT(inp, inp_list)) { INP_RLOCK(inp); if (inp->inp_gencnt <= xig.xig_gen && cr_canseeinpcb(req->td->td_ucred, inp) == 0) { struct xinpcb xi; in_pcbtoxinpcb(inp, &xi); INP_RUNLOCK(inp); error = SYSCTL_OUT(req, &xi, sizeof xi); if (error) break; } else INP_RUNLOCK(inp); } NET_EPOCH_EXIT(et); if (!error) { /* * Give the user an updated idea of our state. If the * generation differs from what we told her before, she knows * that something happened while we were processing this * request, and it might be necessary to retry. */ xig.xig_gen = V_udbinfo.ipi_gencnt; xig.xig_sogen = so_gencnt; xig.xig_count = V_udbinfo.ipi_count; error = SYSCTL_OUT(req, &xig, sizeof xig); } return (error); } SYSCTL_PROC(_net_inet_udp, UDPCTL_PCBLIST, pcblist, CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, 0, udp_pcblist, "S,xinpcb", "List of active UDP sockets"); #ifdef INET static int udp_getcred(SYSCTL_HANDLER_ARGS) { struct xucred xuc; struct sockaddr_in addrs[2]; struct epoch_tracker et; struct inpcb *inp; int error; error = priv_check(req->td, PRIV_NETINET_GETCRED); if (error) return (error); error = SYSCTL_IN(req, addrs, sizeof(addrs)); if (error) return (error); NET_EPOCH_ENTER(et); inp = in_pcblookup(&V_udbinfo, addrs[1].sin_addr, addrs[1].sin_port, addrs[0].sin_addr, addrs[0].sin_port, INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB, NULL); NET_EPOCH_EXIT(et); if (inp != NULL) { INP_RLOCK_ASSERT(inp); if (inp->inp_socket == NULL) error = ENOENT; if (error == 0) error = cr_canseeinpcb(req->td->td_ucred, inp); if (error == 0) cru2x(inp->inp_cred, &xuc); INP_RUNLOCK(inp); } else error = ENOENT; if (error == 0) error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred)); return (error); } SYSCTL_PROC(_net_inet_udp, OID_AUTO, getcred, CTLTYPE_OPAQUE | CTLFLAG_RW | CTLFLAG_PRISON | CTLFLAG_MPSAFE, 0, 0, udp_getcred, "S,xucred", "Get the xucred of a UDP connection"); #endif /* INET */ int udp_ctloutput(struct socket *so, struct sockopt *sopt) { struct inpcb *inp; struct udpcb *up; int isudplite, error, optval; error = 0; isudplite = (so->so_proto->pr_protocol == IPPROTO_UDPLITE) ? 1 : 0; inp = sotoinpcb(so); KASSERT(inp != NULL, ("%s: inp == NULL", __func__)); INP_WLOCK(inp); if (sopt->sopt_level != so->so_proto->pr_protocol) { #ifdef INET6 if (INP_CHECK_SOCKAF(so, AF_INET6)) { INP_WUNLOCK(inp); error = ip6_ctloutput(so, sopt); } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { INP_WUNLOCK(inp); error = ip_ctloutput(so, sopt); } #endif return (error); } switch (sopt->sopt_dir) { case SOPT_SET: switch (sopt->sopt_name) { #if defined(IPSEC) || defined(IPSEC_SUPPORT) #ifdef INET case UDP_ENCAP: if (!IPSEC_ENABLED(ipv4)) { INP_WUNLOCK(inp); return (ENOPROTOOPT); } error = UDPENCAP_PCBCTL(inp, sopt); break; #endif /* INET */ #endif /* IPSEC */ case UDPLITE_SEND_CSCOV: case UDPLITE_RECV_CSCOV: if (!isudplite) { INP_WUNLOCK(inp); error = ENOPROTOOPT; break; } INP_WUNLOCK(inp); error = sooptcopyin(sopt, &optval, sizeof(optval), sizeof(optval)); if (error != 0) break; inp = sotoinpcb(so); KASSERT(inp != NULL, ("%s: inp == NULL", __func__)); INP_WLOCK(inp); up = intoudpcb(inp); KASSERT(up != NULL, ("%s: up == NULL", __func__)); if ((optval != 0 && optval < 8) || (optval > 65535)) { INP_WUNLOCK(inp); error = EINVAL; break; } if (sopt->sopt_name == UDPLITE_SEND_CSCOV) up->u_txcslen = optval; else up->u_rxcslen = optval; INP_WUNLOCK(inp); break; default: INP_WUNLOCK(inp); error = ENOPROTOOPT; break; } break; case SOPT_GET: switch (sopt->sopt_name) { #if defined(IPSEC) || defined(IPSEC_SUPPORT) #ifdef INET case UDP_ENCAP: if (!IPSEC_ENABLED(ipv4)) { INP_WUNLOCK(inp); return (ENOPROTOOPT); } error = UDPENCAP_PCBCTL(inp, sopt); break; #endif /* INET */ #endif /* IPSEC */ case UDPLITE_SEND_CSCOV: case UDPLITE_RECV_CSCOV: if (!isudplite) { INP_WUNLOCK(inp); error = ENOPROTOOPT; break; } up = intoudpcb(inp); KASSERT(up != NULL, ("%s: up == NULL", __func__)); if (sopt->sopt_name == UDPLITE_SEND_CSCOV) optval = up->u_txcslen; else optval = up->u_rxcslen; INP_WUNLOCK(inp); error = sooptcopyout(sopt, &optval, sizeof(optval)); break; default: INP_WUNLOCK(inp); error = ENOPROTOOPT; break; } break; } return (error); } #ifdef INET +#ifdef INET6 +/* The logic here is derived from ip6_setpktopt(). See comments there. */ static int +udp_v4mapped_pktinfo(struct cmsghdr *cm, struct sockaddr_in * src, + struct inpcb *inp, int flags) +{ + struct ifnet *ifp; + struct in6_pktinfo *pktinfo; + struct in_addr ia; + + if ((flags & PRUS_IPV6) == 0) + return (0); + + if (cm->cmsg_level != IPPROTO_IPV6) + return (0); + + if (cm->cmsg_type != IPV6_2292PKTINFO && + cm->cmsg_type != IPV6_PKTINFO) + return (0); + + if (cm->cmsg_len != + CMSG_LEN(sizeof(struct in6_pktinfo))) + return (EINVAL); + + pktinfo = (struct in6_pktinfo *)CMSG_DATA(cm); + if (!IN6_IS_ADDR_V4MAPPED(&pktinfo->ipi6_addr) && + !IN6_IS_ADDR_UNSPECIFIED(&pktinfo->ipi6_addr)) + return (EINVAL); + + /* Validate the interface index if specified. */ + if (pktinfo->ipi6_ifindex > V_if_index) + return (ENXIO); + + ifp = NULL; + if (pktinfo->ipi6_ifindex) { + ifp = ifnet_byindex(pktinfo->ipi6_ifindex); + if (ifp == NULL) + return (ENXIO); + } + if (ifp != NULL && !IN6_IS_ADDR_UNSPECIFIED(&pktinfo->ipi6_addr)) { + + ia.s_addr = pktinfo->ipi6_addr.s6_addr32[3]; + if (in_ifhasaddr(ifp, ia) == 0) + return (EADDRNOTAVAIL); + } + + bzero(src, sizeof(*src)); + src->sin_family = AF_INET; + src->sin_len = sizeof(*src); + src->sin_port = inp->inp_lport; + src->sin_addr.s_addr = pktinfo->ipi6_addr.s6_addr32[3]; + + return (0); +} +#endif + +static int udp_output(struct inpcb *inp, struct mbuf *m, struct sockaddr *addr, - struct mbuf *control, struct thread *td) + struct mbuf *control, struct thread *td, int flags) { struct udpiphdr *ui; int len = m->m_pkthdr.len; struct in_addr faddr, laddr; struct cmsghdr *cm; struct inpcbinfo *pcbinfo; struct sockaddr_in *sin, src; struct epoch_tracker et; int cscov_partial = 0; int error = 0; int ipflags; u_short fport, lport; u_char tos; uint8_t pr; uint16_t cscov = 0; uint32_t flowid = 0; uint8_t flowtype = M_HASHTYPE_NONE; if (len + sizeof(struct udpiphdr) > IP_MAXPACKET) { if (control) m_freem(control); m_freem(m); return (EMSGSIZE); } src.sin_family = 0; sin = (struct sockaddr_in *)addr; /* * udp_output() may need to temporarily bind or connect the current * inpcb. As such, we don't know up front whether we will need the * pcbinfo lock or not. Do any work to decide what is needed up * front before acquiring any locks. * * We will need network epoch in either case, to safely lookup into * pcb hash. */ if (sin == NULL || (inp->inp_laddr.s_addr == INADDR_ANY && inp->inp_lport == 0)) INP_WLOCK(inp); else INP_RLOCK(inp); NET_EPOCH_ENTER(et); tos = inp->inp_ip_tos; if (control != NULL) { /* * XXX: Currently, we assume all the optional information is * stored in a single mbuf. */ if (control->m_next) { m_freem(control); error = EINVAL; goto release; } for (; control->m_len > 0; control->m_data += CMSG_ALIGN(cm->cmsg_len), control->m_len -= CMSG_ALIGN(cm->cmsg_len)) { cm = mtod(control, struct cmsghdr *); if (control->m_len < sizeof(*cm) || cm->cmsg_len == 0 || cm->cmsg_len > control->m_len) { error = EINVAL; break; } +#ifdef INET6 + error = udp_v4mapped_pktinfo(cm, &src, inp, flags); + if (error != 0) + break; +#endif if (cm->cmsg_level != IPPROTO_IP) continue; switch (cm->cmsg_type) { case IP_SENDSRCADDR: if (cm->cmsg_len != CMSG_LEN(sizeof(struct in_addr))) { error = EINVAL; break; } bzero(&src, sizeof(src)); src.sin_family = AF_INET; src.sin_len = sizeof(src); src.sin_port = inp->inp_lport; src.sin_addr = *(struct in_addr *)CMSG_DATA(cm); break; case IP_TOS: if (cm->cmsg_len != CMSG_LEN(sizeof(u_char))) { error = EINVAL; break; } tos = *(u_char *)CMSG_DATA(cm); break; case IP_FLOWID: if (cm->cmsg_len != CMSG_LEN(sizeof(uint32_t))) { error = EINVAL; break; } flowid = *(uint32_t *) CMSG_DATA(cm); break; case IP_FLOWTYPE: if (cm->cmsg_len != CMSG_LEN(sizeof(uint32_t))) { error = EINVAL; break; } flowtype = *(uint32_t *) CMSG_DATA(cm); break; #ifdef RSS case IP_RSSBUCKETID: if (cm->cmsg_len != CMSG_LEN(sizeof(uint32_t))) { error = EINVAL; break; } /* This is just a placeholder for now */ break; #endif /* RSS */ default: error = ENOPROTOOPT; break; } if (error) break; } m_freem(control); } if (error) goto release; pr = inp->inp_socket->so_proto->pr_protocol; pcbinfo = udp_get_inpcbinfo(pr); /* * If the IP_SENDSRCADDR control message was specified, override the * source address for this datagram. Its use is invalidated if the * address thus specified is incomplete or clobbers other inpcbs. */ laddr = inp->inp_laddr; lport = inp->inp_lport; if (src.sin_family == AF_INET) { INP_HASH_LOCK_ASSERT(pcbinfo); if ((lport == 0) || (laddr.s_addr == INADDR_ANY && src.sin_addr.s_addr == INADDR_ANY)) { error = EINVAL; goto release; } error = in_pcbbind_setup(inp, (struct sockaddr *)&src, &laddr.s_addr, &lport, td->td_ucred); if (error) goto release; } /* * If a UDP socket has been connected, then a local address/port will * have been selected and bound. * * If a UDP socket has not been connected to, then an explicit * destination address must be used, in which case a local * address/port may not have been selected and bound. */ if (sin != NULL) { INP_LOCK_ASSERT(inp); if (inp->inp_faddr.s_addr != INADDR_ANY) { error = EISCONN; goto release; } /* * Jail may rewrite the destination address, so let it do * that before we use it. */ error = prison_remote_ip4(td->td_ucred, &sin->sin_addr); if (error) goto release; /* * If a local address or port hasn't yet been selected, or if * the destination address needs to be rewritten due to using * a special INADDR_ constant, invoke in_pcbconnect_setup() * to do the heavy lifting. Once a port is selected, we * commit the binding back to the socket; we also commit the * binding of the address if in jail. * * If we already have a valid binding and we're not * requesting a destination address rewrite, use a fast path. */ if (inp->inp_laddr.s_addr == INADDR_ANY || inp->inp_lport == 0 || sin->sin_addr.s_addr == INADDR_ANY || sin->sin_addr.s_addr == INADDR_BROADCAST) { INP_HASH_LOCK_ASSERT(pcbinfo); error = in_pcbconnect_setup(inp, addr, &laddr.s_addr, &lport, &faddr.s_addr, &fport, NULL, td->td_ucred); if (error) goto release; /* * XXXRW: Why not commit the port if the address is * !INADDR_ANY? */ /* Commit the local port if newly assigned. */ if (inp->inp_laddr.s_addr == INADDR_ANY && inp->inp_lport == 0) { INP_WLOCK_ASSERT(inp); /* * Remember addr if jailed, to prevent * rebinding. */ if (prison_flag(td->td_ucred, PR_IP4)) inp->inp_laddr = laddr; inp->inp_lport = lport; INP_HASH_WLOCK(pcbinfo); error = in_pcbinshash(inp); INP_HASH_WUNLOCK(pcbinfo); if (error != 0) { inp->inp_lport = 0; error = EAGAIN; goto release; } inp->inp_flags |= INP_ANONPORT; } } else { faddr = sin->sin_addr; fport = sin->sin_port; } } else { INP_LOCK_ASSERT(inp); faddr = inp->inp_faddr; fport = inp->inp_fport; if (faddr.s_addr == INADDR_ANY) { error = ENOTCONN; goto release; } } /* * Calculate data length and get a mbuf for UDP, IP, and possible * link-layer headers. Immediate slide the data pointer back forward * since we won't use that space at this layer. */ M_PREPEND(m, sizeof(struct udpiphdr) + max_linkhdr, M_NOWAIT); if (m == NULL) { error = ENOBUFS; goto release; } m->m_data += max_linkhdr; m->m_len -= max_linkhdr; m->m_pkthdr.len -= max_linkhdr; /* * Fill in mbuf with extended UDP header and addresses and length put * into network format. */ ui = mtod(m, struct udpiphdr *); bzero(ui->ui_x1, sizeof(ui->ui_x1)); /* XXX still needed? */ ui->ui_v = IPVERSION << 4; ui->ui_pr = pr; ui->ui_src = laddr; ui->ui_dst = faddr; ui->ui_sport = lport; ui->ui_dport = fport; ui->ui_ulen = htons((u_short)len + sizeof(struct udphdr)); if (pr == IPPROTO_UDPLITE) { struct udpcb *up; uint16_t plen; up = intoudpcb(inp); cscov = up->u_txcslen; plen = (u_short)len + sizeof(struct udphdr); if (cscov >= plen) cscov = 0; ui->ui_len = htons(plen); ui->ui_ulen = htons(cscov); /* * For UDP-Lite, checksum coverage length of zero means * the entire UDPLite packet is covered by the checksum. */ cscov_partial = (cscov == 0) ? 0 : 1; } /* * Set the Don't Fragment bit in the IP header. */ if (inp->inp_flags & INP_DONTFRAG) { struct ip *ip; ip = (struct ip *)&ui->ui_i; ip->ip_off |= htons(IP_DF); } ipflags = 0; if (inp->inp_socket->so_options & SO_DONTROUTE) ipflags |= IP_ROUTETOIF; if (inp->inp_socket->so_options & SO_BROADCAST) ipflags |= IP_ALLOWBROADCAST; if (inp->inp_flags & INP_ONESBCAST) ipflags |= IP_SENDONES; #ifdef MAC mac_inpcb_create_mbuf(inp, m); #endif /* * Set up checksum and output datagram. */ ui->ui_sum = 0; if (pr == IPPROTO_UDPLITE) { if (inp->inp_flags & INP_ONESBCAST) faddr.s_addr = INADDR_BROADCAST; if (cscov_partial) { if ((ui->ui_sum = in_cksum(m, sizeof(struct ip) + cscov)) == 0) ui->ui_sum = 0xffff; } else { if ((ui->ui_sum = in_cksum(m, sizeof(struct udpiphdr) + len)) == 0) ui->ui_sum = 0xffff; } } else if (V_udp_cksum) { if (inp->inp_flags & INP_ONESBCAST) faddr.s_addr = INADDR_BROADCAST; ui->ui_sum = in_pseudo(ui->ui_src.s_addr, faddr.s_addr, htons((u_short)len + sizeof(struct udphdr) + pr)); m->m_pkthdr.csum_flags = CSUM_UDP; m->m_pkthdr.csum_data = offsetof(struct udphdr, uh_sum); } ((struct ip *)ui)->ip_len = htons(sizeof(struct udpiphdr) + len); ((struct ip *)ui)->ip_ttl = inp->inp_ip_ttl; /* XXX */ ((struct ip *)ui)->ip_tos = tos; /* XXX */ UDPSTAT_INC(udps_opackets); /* * Setup flowid / RSS information for outbound socket. * * Once the UDP code decides to set a flowid some other way, * this allows the flowid to be overridden by userland. */ if (flowtype != M_HASHTYPE_NONE) { m->m_pkthdr.flowid = flowid; M_HASHTYPE_SET(m, flowtype); } #ifdef RSS else { uint32_t hash_val, hash_type; /* * Calculate an appropriate RSS hash for UDP and * UDP Lite. * * The called function will take care of figuring out * whether a 2-tuple or 4-tuple hash is required based * on the currently configured scheme. * * Later later on connected socket values should be * cached in the inpcb and reused, rather than constantly * re-calculating it. * * UDP Lite is a different protocol number and will * likely end up being hashed as a 2-tuple until * RSS / NICs grow UDP Lite protocol awareness. */ if (rss_proto_software_hash_v4(faddr, laddr, fport, lport, pr, &hash_val, &hash_type) == 0) { m->m_pkthdr.flowid = hash_val; M_HASHTYPE_SET(m, hash_type); } } /* * Don't override with the inp cached flowid value. * * Depending upon the kind of send being done, the inp * flowid/flowtype values may actually not be appropriate * for this particular socket send. * * We should either leave the flowid at zero (which is what is * currently done) or set it to some software generated * hash value based on the packet contents. */ ipflags |= IP_NODEFAULTFLOWID; #endif /* RSS */ if (pr == IPPROTO_UDPLITE) UDPLITE_PROBE(send, NULL, inp, &ui->ui_i, inp, &ui->ui_u); else UDP_PROBE(send, NULL, inp, &ui->ui_i, inp, &ui->ui_u); error = ip_output(m, inp->inp_options, INP_WLOCKED(inp) ? &inp->inp_route : NULL, ipflags, inp->inp_moptions, inp); INP_UNLOCK(inp); NET_EPOCH_EXIT(et); return (error); release: INP_UNLOCK(inp); NET_EPOCH_EXIT(et); m_freem(m); return (error); } static void udp_abort(struct socket *so) { struct inpcb *inp; struct inpcbinfo *pcbinfo; pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol); inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_abort: inp == NULL")); INP_WLOCK(inp); if (inp->inp_faddr.s_addr != INADDR_ANY) { INP_HASH_WLOCK(pcbinfo); in_pcbdisconnect(inp); inp->inp_laddr.s_addr = INADDR_ANY; INP_HASH_WUNLOCK(pcbinfo); soisdisconnected(so); } INP_WUNLOCK(inp); } static int udp_attach(struct socket *so, int proto, struct thread *td) { static uint32_t udp_flowid; struct inpcb *inp; struct inpcbinfo *pcbinfo; int error; pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol); inp = sotoinpcb(so); KASSERT(inp == NULL, ("udp_attach: inp != NULL")); error = soreserve(so, udp_sendspace, udp_recvspace); if (error) return (error); INP_INFO_WLOCK(pcbinfo); error = in_pcballoc(so, pcbinfo); if (error) { INP_INFO_WUNLOCK(pcbinfo); return (error); } inp = sotoinpcb(so); inp->inp_vflag |= INP_IPV4; inp->inp_ip_ttl = V_ip_defttl; inp->inp_flowid = atomic_fetchadd_int(&udp_flowid, 1); inp->inp_flowtype = M_HASHTYPE_OPAQUE; error = udp_newudpcb(inp); if (error) { in_pcbdetach(inp); in_pcbfree(inp); INP_INFO_WUNLOCK(pcbinfo); return (error); } INP_WUNLOCK(inp); INP_INFO_WUNLOCK(pcbinfo); return (0); } #endif /* INET */ int udp_set_kernel_tunneling(struct socket *so, udp_tun_func_t f, udp_tun_icmp_t i, void *ctx) { struct inpcb *inp; struct udpcb *up; KASSERT(so->so_type == SOCK_DGRAM, ("udp_set_kernel_tunneling: !dgram")); inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_set_kernel_tunneling: inp == NULL")); INP_WLOCK(inp); up = intoudpcb(inp); if ((up->u_tun_func != NULL) || (up->u_icmp_func != NULL)) { INP_WUNLOCK(inp); return (EBUSY); } up->u_tun_func = f; up->u_icmp_func = i; up->u_tun_ctx = ctx; INP_WUNLOCK(inp); return (0); } #ifdef INET static int udp_bind(struct socket *so, struct sockaddr *nam, struct thread *td) { struct inpcb *inp; struct inpcbinfo *pcbinfo; int error; pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol); inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_bind: inp == NULL")); INP_WLOCK(inp); INP_HASH_WLOCK(pcbinfo); error = in_pcbbind(inp, nam, td->td_ucred); INP_HASH_WUNLOCK(pcbinfo); INP_WUNLOCK(inp); return (error); } static void udp_close(struct socket *so) { struct inpcb *inp; struct inpcbinfo *pcbinfo; pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol); inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_close: inp == NULL")); INP_WLOCK(inp); if (inp->inp_faddr.s_addr != INADDR_ANY) { INP_HASH_WLOCK(pcbinfo); in_pcbdisconnect(inp); inp->inp_laddr.s_addr = INADDR_ANY; INP_HASH_WUNLOCK(pcbinfo); soisdisconnected(so); } INP_WUNLOCK(inp); } static int udp_connect(struct socket *so, struct sockaddr *nam, struct thread *td) { struct epoch_tracker et; struct inpcb *inp; struct inpcbinfo *pcbinfo; struct sockaddr_in *sin; int error; pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol); inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_connect: inp == NULL")); INP_WLOCK(inp); if (inp->inp_faddr.s_addr != INADDR_ANY) { INP_WUNLOCK(inp); return (EISCONN); } sin = (struct sockaddr_in *)nam; error = prison_remote_ip4(td->td_ucred, &sin->sin_addr); if (error != 0) { INP_WUNLOCK(inp); return (error); } NET_EPOCH_ENTER(et); INP_HASH_WLOCK(pcbinfo); error = in_pcbconnect(inp, nam, td->td_ucred); INP_HASH_WUNLOCK(pcbinfo); NET_EPOCH_EXIT(et); if (error == 0) soisconnected(so); INP_WUNLOCK(inp); return (error); } static void udp_detach(struct socket *so) { struct inpcb *inp; struct inpcbinfo *pcbinfo; struct udpcb *up; pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol); inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_detach: inp == NULL")); KASSERT(inp->inp_faddr.s_addr == INADDR_ANY, ("udp_detach: not disconnected")); INP_INFO_WLOCK(pcbinfo); INP_WLOCK(inp); up = intoudpcb(inp); KASSERT(up != NULL, ("%s: up == NULL", __func__)); inp->inp_ppcb = NULL; in_pcbdetach(inp); in_pcbfree(inp); INP_INFO_WUNLOCK(pcbinfo); udp_discardcb(up); } static int udp_disconnect(struct socket *so) { struct inpcb *inp; struct inpcbinfo *pcbinfo; pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol); inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_disconnect: inp == NULL")); INP_WLOCK(inp); if (inp->inp_faddr.s_addr == INADDR_ANY) { INP_WUNLOCK(inp); return (ENOTCONN); } INP_HASH_WLOCK(pcbinfo); in_pcbdisconnect(inp); inp->inp_laddr.s_addr = INADDR_ANY; INP_HASH_WUNLOCK(pcbinfo); SOCK_LOCK(so); so->so_state &= ~SS_ISCONNECTED; /* XXX */ SOCK_UNLOCK(so); INP_WUNLOCK(inp); return (0); } static int udp_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr, struct mbuf *control, struct thread *td) { struct inpcb *inp; inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_send: inp == NULL")); - return (udp_output(inp, m, addr, control, td)); + return (udp_output(inp, m, addr, control, td, flags)); } #endif /* INET */ int udp_shutdown(struct socket *so) { struct inpcb *inp; inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp_shutdown: inp == NULL")); INP_WLOCK(inp); socantsendmore(so); INP_WUNLOCK(inp); return (0); } #ifdef INET struct pr_usrreqs udp_usrreqs = { .pru_abort = udp_abort, .pru_attach = udp_attach, .pru_bind = udp_bind, .pru_connect = udp_connect, .pru_control = in_control, .pru_detach = udp_detach, .pru_disconnect = udp_disconnect, .pru_peeraddr = in_getpeeraddr, .pru_send = udp_send, .pru_soreceive = soreceive_dgram, .pru_sosend = sosend_dgram, .pru_shutdown = udp_shutdown, .pru_sockaddr = in_getsockaddr, .pru_sosetlabel = in_pcbsosetlabel, .pru_close = udp_close, }; #endif /* INET */ Index: projects/clang1100-import/sys/netinet6/udp6_usrreq.c =================================================================== --- projects/clang1100-import/sys/netinet6/udp6_usrreq.c (revision 364034) +++ projects/clang1100-import/sys/netinet6/udp6_usrreq.c (revision 364035) @@ -1,1379 +1,1379 @@ /*- * SPDX-License-Identifier: BSD-3-Clause * * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * Copyright (c) 2010-2011 Juniper Networks, Inc. * Copyright (c) 2014 Kevin Lo * All rights reserved. * * Portions of this software were developed by Robert N. M. Watson under * contract to Juniper Networks, Inc. * * 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 project 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 PROJECT 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 PROJECT 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: udp6_usrreq.c,v 1.27 2001/05/21 05:45:10 jinmei Exp $ * $KAME: udp6_output.c,v 1.31 2001/05/21 16:39:15 jinmei Exp $ */ /*- * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995 * 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. * 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. * * @(#)udp_usrreq.c 8.6 (Berkeley) 5/23/95 */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" #include "opt_rss.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 #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * UDP protocol implementation. * Per RFC 768, August, 1980. */ extern struct protosw inetsw[]; static void udp6_detach(struct socket *so); static int udp6_append(struct inpcb *inp, struct mbuf *n, int off, struct sockaddr_in6 *fromsa) { struct socket *so; struct mbuf *opts = NULL, *tmp_opts; struct udpcb *up; INP_LOCK_ASSERT(inp); /* * Engage the tunneling protocol. */ up = intoudpcb(inp); if (up->u_tun_func != NULL) { in_pcbref(inp); INP_RUNLOCK(inp); (*up->u_tun_func)(n, off, inp, (struct sockaddr *)&fromsa[0], up->u_tun_ctx); INP_RLOCK(inp); return (in_pcbrele_rlocked(inp)); } #if defined(IPSEC) || defined(IPSEC_SUPPORT) /* Check AH/ESP integrity. */ if (IPSEC_ENABLED(ipv6)) { if (IPSEC_CHECK_POLICY(ipv6, n, inp) != 0) { m_freem(n); return (0); } } #endif /* IPSEC */ #ifdef MAC if (mac_inpcb_check_deliver(inp, n) != 0) { m_freem(n); return (0); } #endif opts = NULL; if (inp->inp_flags & INP_CONTROLOPTS || inp->inp_socket->so_options & SO_TIMESTAMP) ip6_savecontrol(inp, n, &opts); if ((inp->inp_vflag & INP_IPV6) && (inp->inp_flags2 & INP_ORIGDSTADDR)) { tmp_opts = sbcreatecontrol((caddr_t)&fromsa[1], sizeof(struct sockaddr_in6), IPV6_ORIGDSTADDR, IPPROTO_IPV6); if (tmp_opts) { if (opts) { tmp_opts->m_next = opts; opts = tmp_opts; } else opts = tmp_opts; } } m_adj(n, off + sizeof(struct udphdr)); so = inp->inp_socket; SOCKBUF_LOCK(&so->so_rcv); if (sbappendaddr_locked(&so->so_rcv, (struct sockaddr *)&fromsa[0], n, opts) == 0) { SOCKBUF_UNLOCK(&so->so_rcv); m_freem(n); if (opts) m_freem(opts); UDPSTAT_INC(udps_fullsock); } else sorwakeup_locked(so); return (0); } int udp6_input(struct mbuf **mp, int *offp, int proto) { struct mbuf *m = *mp; struct ifnet *ifp; struct ip6_hdr *ip6; struct udphdr *uh; struct inpcb *inp; struct inpcbinfo *pcbinfo; struct udpcb *up; int off = *offp; int cscov_partial; int plen, ulen; struct sockaddr_in6 fromsa[2]; struct m_tag *fwd_tag; uint16_t uh_sum; uint8_t nxt; NET_EPOCH_ASSERT(); ifp = m->m_pkthdr.rcvif; if (m->m_len < off + sizeof(struct udphdr)) { m = m_pullup(m, off + sizeof(struct udphdr)); if (m == NULL) { IP6STAT_INC(ip6s_exthdrtoolong); *mp = NULL; return (IPPROTO_DONE); } } ip6 = mtod(m, struct ip6_hdr *); uh = (struct udphdr *)((caddr_t)ip6 + off); UDPSTAT_INC(udps_ipackets); /* * Destination port of 0 is illegal, based on RFC768. */ if (uh->uh_dport == 0) goto badunlocked; plen = ntohs(ip6->ip6_plen) - off + sizeof(*ip6); ulen = ntohs((u_short)uh->uh_ulen); nxt = proto; cscov_partial = (nxt == IPPROTO_UDPLITE) ? 1 : 0; if (nxt == IPPROTO_UDPLITE) { /* Zero means checksum over the complete packet. */ if (ulen == 0) ulen = plen; if (ulen == plen) cscov_partial = 0; if ((ulen < sizeof(struct udphdr)) || (ulen > plen)) { /* XXX: What is the right UDPLite MIB counter? */ goto badunlocked; } if (uh->uh_sum == 0) { /* XXX: What is the right UDPLite MIB counter? */ goto badunlocked; } } else { if ((ulen < sizeof(struct udphdr)) || (plen != ulen)) { UDPSTAT_INC(udps_badlen); goto badunlocked; } if (uh->uh_sum == 0) { UDPSTAT_INC(udps_nosum); goto badunlocked; } } if ((m->m_pkthdr.csum_flags & CSUM_DATA_VALID_IPV6) && !cscov_partial) { if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR) uh_sum = m->m_pkthdr.csum_data; else uh_sum = in6_cksum_pseudo(ip6, ulen, nxt, m->m_pkthdr.csum_data); uh_sum ^= 0xffff; } else uh_sum = in6_cksum_partial(m, nxt, off, plen, ulen); if (uh_sum != 0) { UDPSTAT_INC(udps_badsum); goto badunlocked; } /* * Construct sockaddr format source address. */ init_sin6(&fromsa[0], m, 0); fromsa[0].sin6_port = uh->uh_sport; init_sin6(&fromsa[1], m, 1); fromsa[1].sin6_port = uh->uh_dport; pcbinfo = udp_get_inpcbinfo(nxt); if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { struct inpcb *last; struct inpcbhead *pcblist; struct ip6_moptions *imo; /* * In the event that laddr should be set to the link-local * address (this happens in RIPng), the multicast address * specified in the received packet will not match laddr. To * handle this situation, matching is relaxed if the * receiving interface is the same as one specified in the * socket and if the destination multicast address matches * one of the multicast groups specified in the socket. */ /* * KAME note: traditionally we dropped udpiphdr from mbuf * here. We need udphdr for IPsec processing so we do that * later. */ pcblist = udp_get_pcblist(nxt); last = NULL; CK_LIST_FOREACH(inp, pcblist, inp_list) { if ((inp->inp_vflag & INP_IPV6) == 0) continue; if (inp->inp_lport != uh->uh_dport) continue; if (inp->inp_fport != 0 && inp->inp_fport != uh->uh_sport) continue; if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) { if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &ip6->ip6_dst)) continue; } if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) { if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &ip6->ip6_src) || inp->inp_fport != uh->uh_sport) continue; } /* * XXXRW: Because we weren't holding either the inpcb * or the hash lock when we checked for a match * before, we should probably recheck now that the * inpcb lock is (supposed to be) held. */ /* * Handle socket delivery policy for any-source * and source-specific multicast. [RFC3678] */ imo = inp->in6p_moptions; if (imo && IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { struct sockaddr_in6 mcaddr; int blocked; INP_RLOCK(inp); if (__predict_false(inp->inp_flags2 & INP_FREED)) { INP_RUNLOCK(inp); continue; } bzero(&mcaddr, sizeof(struct sockaddr_in6)); mcaddr.sin6_len = sizeof(struct sockaddr_in6); mcaddr.sin6_family = AF_INET6; mcaddr.sin6_addr = ip6->ip6_dst; blocked = im6o_mc_filter(imo, ifp, (struct sockaddr *)&mcaddr, (struct sockaddr *)&fromsa[0]); if (blocked != MCAST_PASS) { if (blocked == MCAST_NOTGMEMBER) IP6STAT_INC(ip6s_notmember); if (blocked == MCAST_NOTSMEMBER || blocked == MCAST_MUTED) UDPSTAT_INC(udps_filtermcast); INP_RUNLOCK(inp); /* XXX */ continue; } INP_RUNLOCK(inp); } if (last != NULL) { struct mbuf *n; if ((n = m_copym(m, 0, M_COPYALL, M_NOWAIT)) != NULL) { INP_RLOCK(last); if (__predict_true(last->inp_flags2 & INP_FREED) == 0) { if (nxt == IPPROTO_UDPLITE) UDPLITE_PROBE(receive, NULL, last, ip6, last, uh); else UDP_PROBE(receive, NULL, last, ip6, last, uh); if (udp6_append(last, n, off, fromsa)) { /* XXX-BZ do we leak m here? */ *mp = NULL; return (IPPROTO_DONE); } } INP_RUNLOCK(last); } } last = inp; /* * Don't look for additional matches if this one does * not have either the SO_REUSEPORT or SO_REUSEADDR * socket options set. This heuristic avoids * searching through all pcbs in the common case of a * non-shared port. It assumes that an application * will never clear these options after setting them. */ if ((last->inp_socket->so_options & (SO_REUSEPORT|SO_REUSEPORT_LB|SO_REUSEADDR)) == 0) break; } if (last == NULL) { /* * No matching pcb found; discard datagram. (No need * to send an ICMP Port Unreachable for a broadcast * or multicast datgram.) */ UDPSTAT_INC(udps_noport); UDPSTAT_INC(udps_noportmcast); goto badunlocked; } INP_RLOCK(last); if (__predict_true(last->inp_flags2 & INP_FREED) == 0) { if (nxt == IPPROTO_UDPLITE) UDPLITE_PROBE(receive, NULL, last, ip6, last, uh); else UDP_PROBE(receive, NULL, last, ip6, last, uh); if (udp6_append(last, m, off, fromsa) == 0) INP_RUNLOCK(last); } else INP_RUNLOCK(last); *mp = NULL; return (IPPROTO_DONE); } /* * Locate pcb for datagram. */ /* * Grab info from PACKET_TAG_IPFORWARD tag prepended to the chain. */ if ((m->m_flags & M_IP6_NEXTHOP) && (fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL) { struct sockaddr_in6 *next_hop6; next_hop6 = (struct sockaddr_in6 *)(fwd_tag + 1); /* * Transparently forwarded. Pretend to be the destination. * Already got one like this? */ inp = in6_pcblookup_mbuf(pcbinfo, &ip6->ip6_src, uh->uh_sport, &ip6->ip6_dst, uh->uh_dport, INPLOOKUP_RLOCKPCB, m->m_pkthdr.rcvif, m); if (!inp) { /* * It's new. Try to find the ambushing socket. * Because we've rewritten the destination address, * any hardware-generated hash is ignored. */ inp = in6_pcblookup(pcbinfo, &ip6->ip6_src, uh->uh_sport, &next_hop6->sin6_addr, next_hop6->sin6_port ? htons(next_hop6->sin6_port) : uh->uh_dport, INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB, m->m_pkthdr.rcvif); } /* Remove the tag from the packet. We don't need it anymore. */ m_tag_delete(m, fwd_tag); m->m_flags &= ~M_IP6_NEXTHOP; } else inp = in6_pcblookup_mbuf(pcbinfo, &ip6->ip6_src, uh->uh_sport, &ip6->ip6_dst, uh->uh_dport, INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB, m->m_pkthdr.rcvif, m); if (inp == NULL) { if (V_udp_log_in_vain) { char ip6bufs[INET6_ADDRSTRLEN]; char ip6bufd[INET6_ADDRSTRLEN]; log(LOG_INFO, "Connection attempt to UDP [%s]:%d from [%s]:%d\n", ip6_sprintf(ip6bufd, &ip6->ip6_dst), ntohs(uh->uh_dport), ip6_sprintf(ip6bufs, &ip6->ip6_src), ntohs(uh->uh_sport)); } if (nxt == IPPROTO_UDPLITE) UDPLITE_PROBE(receive, NULL, NULL, ip6, NULL, uh); else UDP_PROBE(receive, NULL, NULL, ip6, NULL, uh); UDPSTAT_INC(udps_noport); if (m->m_flags & M_MCAST) { printf("UDP6: M_MCAST is set in a unicast packet.\n"); UDPSTAT_INC(udps_noportmcast); goto badunlocked; } if (V_udp_blackhole) goto badunlocked; icmp6_error(m, ICMP6_DST_UNREACH, ICMP6_DST_UNREACH_NOPORT, 0); *mp = NULL; return (IPPROTO_DONE); } INP_RLOCK_ASSERT(inp); up = intoudpcb(inp); if (cscov_partial) { if (up->u_rxcslen == 0 || up->u_rxcslen > ulen) { INP_RUNLOCK(inp); m_freem(m); *mp = NULL; return (IPPROTO_DONE); } } if (nxt == IPPROTO_UDPLITE) UDPLITE_PROBE(receive, NULL, inp, ip6, inp, uh); else UDP_PROBE(receive, NULL, inp, ip6, inp, uh); if (udp6_append(inp, m, off, fromsa) == 0) INP_RUNLOCK(inp); *mp = NULL; return (IPPROTO_DONE); badunlocked: m_freem(m); *mp = NULL; return (IPPROTO_DONE); } static void udp6_common_ctlinput(int cmd, struct sockaddr *sa, void *d, struct inpcbinfo *pcbinfo) { struct udphdr uh; struct ip6_hdr *ip6; struct mbuf *m; int off = 0; struct ip6ctlparam *ip6cp = NULL; const struct sockaddr_in6 *sa6_src = NULL; void *cmdarg; struct inpcb *(*notify)(struct inpcb *, int) = udp_notify; struct udp_portonly { u_int16_t uh_sport; u_int16_t uh_dport; } *uhp; if (sa->sa_family != AF_INET6 || sa->sa_len != sizeof(struct sockaddr_in6)) return; if ((unsigned)cmd >= PRC_NCMDS) return; if (PRC_IS_REDIRECT(cmd)) notify = in6_rtchange, d = NULL; else if (cmd == PRC_HOSTDEAD) d = NULL; else if (inet6ctlerrmap[cmd] == 0) return; /* if the parameter is from icmp6, decode it. */ if (d != NULL) { ip6cp = (struct ip6ctlparam *)d; m = ip6cp->ip6c_m; ip6 = ip6cp->ip6c_ip6; off = ip6cp->ip6c_off; cmdarg = ip6cp->ip6c_cmdarg; sa6_src = ip6cp->ip6c_src; } else { m = NULL; ip6 = NULL; cmdarg = NULL; sa6_src = &sa6_any; } if (ip6) { /* * XXX: We assume that when IPV6 is non NULL, * M and OFF are valid. */ /* Check if we can safely examine src and dst ports. */ if (m->m_pkthdr.len < off + sizeof(*uhp)) return; bzero(&uh, sizeof(uh)); m_copydata(m, off, sizeof(*uhp), (caddr_t)&uh); if (!PRC_IS_REDIRECT(cmd)) { /* Check to see if its tunneled */ struct inpcb *inp; inp = in6_pcblookup_mbuf(pcbinfo, &ip6->ip6_dst, uh.uh_dport, &ip6->ip6_src, uh.uh_sport, INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB, m->m_pkthdr.rcvif, m); if (inp != NULL) { struct udpcb *up; up = intoudpcb(inp); if (up->u_icmp_func) { /* Yes it is. */ INP_RUNLOCK(inp); (*up->u_icmp_func)(cmd, (struct sockaddr *)ip6cp->ip6c_src, d, up->u_tun_ctx); return; } else { /* Can't find it. */ INP_RUNLOCK(inp); } } } (void)in6_pcbnotify(pcbinfo, sa, uh.uh_dport, (struct sockaddr *)ip6cp->ip6c_src, uh.uh_sport, cmd, cmdarg, notify); } else (void)in6_pcbnotify(pcbinfo, sa, 0, (const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify); } void udp6_ctlinput(int cmd, struct sockaddr *sa, void *d) { return (udp6_common_ctlinput(cmd, sa, d, &V_udbinfo)); } void udplite6_ctlinput(int cmd, struct sockaddr *sa, void *d) { return (udp6_common_ctlinput(cmd, sa, d, &V_ulitecbinfo)); } static int udp6_getcred(SYSCTL_HANDLER_ARGS) { struct xucred xuc; struct sockaddr_in6 addrs[2]; struct epoch_tracker et; struct inpcb *inp; int error; error = priv_check(req->td, PRIV_NETINET_GETCRED); if (error) return (error); if (req->newlen != sizeof(addrs)) return (EINVAL); if (req->oldlen != sizeof(struct xucred)) return (EINVAL); error = SYSCTL_IN(req, addrs, sizeof(addrs)); if (error) return (error); if ((error = sa6_embedscope(&addrs[0], V_ip6_use_defzone)) != 0 || (error = sa6_embedscope(&addrs[1], V_ip6_use_defzone)) != 0) { return (error); } NET_EPOCH_ENTER(et); inp = in6_pcblookup(&V_udbinfo, &addrs[1].sin6_addr, addrs[1].sin6_port, &addrs[0].sin6_addr, addrs[0].sin6_port, INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB, NULL); NET_EPOCH_EXIT(et); if (inp != NULL) { INP_RLOCK_ASSERT(inp); if (inp->inp_socket == NULL) error = ENOENT; if (error == 0) error = cr_canseesocket(req->td->td_ucred, inp->inp_socket); if (error == 0) cru2x(inp->inp_cred, &xuc); INP_RUNLOCK(inp); } else error = ENOENT; if (error == 0) error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred)); return (error); } SYSCTL_PROC(_net_inet6_udp6, OID_AUTO, getcred, CTLTYPE_OPAQUE | CTLFLAG_RW | CTLFLAG_MPSAFE, 0, 0, udp6_getcred, "S,xucred", "Get the xucred of a UDP6 connection"); static int udp6_output(struct socket *so, int flags_arg, struct mbuf *m, struct sockaddr *addr6, struct mbuf *control, struct thread *td) { struct inpcb *inp; struct ip6_hdr *ip6; struct udphdr *udp6; struct in6_addr *laddr, *faddr, in6a; struct ip6_pktopts *optp, opt; struct sockaddr_in6 *sin6, tmp; struct epoch_tracker et; int cscov_partial, error, flags, hlen, scope_ambiguous; u_int32_t ulen, plen; uint16_t cscov; u_short fport; uint8_t nxt; /* addr6 has been validated in udp6_send(). */ sin6 = (struct sockaddr_in6 *)addr6; /* * In contrast to to IPv4 we do not validate the max. packet length * here due to IPv6 Jumbograms (RFC2675). */ scope_ambiguous = 0; if (sin6) { /* Protect *addr6 from overwrites. */ tmp = *sin6; sin6 = &tmp; /* * Application should provide a proper zone ID or the use of * default zone IDs should be enabled. Unfortunately, some * applications do not behave as it should, so we need a * workaround. Even if an appropriate ID is not determined, * we'll see if we can determine the outgoing interface. If we * can, determine the zone ID based on the interface below. */ if (sin6->sin6_scope_id == 0 && !V_ip6_use_defzone) scope_ambiguous = 1; if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0) { if (control) m_freem(control); m_freem(m); return (error); } } inp = sotoinpcb(so); KASSERT(inp != NULL, ("%s: inp == NULL", __func__)); /* * In the following cases we want a write lock on the inp for either * local operations or for possible route cache updates in the IPv6 * output path: * - on connected sockets (sin6 is NULL) for route cache updates, * - when we are not bound to an address and source port (it is * in6_pcbsetport() which will require the write lock). * * We check the inp fields before actually locking the inp, so * here exists a race, and we may WLOCK the inp and end with already * bound one by other thread. This is fine. */ if (sin6 == NULL || (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) && inp->inp_lport == 0)) INP_WLOCK(inp); else INP_RLOCK(inp); nxt = (inp->inp_socket->so_proto->pr_protocol == IPPROTO_UDP) ? IPPROTO_UDP : IPPROTO_UDPLITE; #ifdef INET if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0) { int hasv4addr; if (sin6 == NULL) hasv4addr = (inp->inp_vflag & INP_IPV4); else hasv4addr = IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr) ? 1 : 0; if (hasv4addr) { struct pr_usrreqs *pru; /* * XXXRW: We release UDP-layer locks before calling * udp_send() in order to avoid recursion. However, * this does mean there is a short window where inp's * fields are unstable. Could this lead to a * potential race in which the factors causing us to * select the UDPv4 output routine are invalidated? */ INP_UNLOCK(inp); if (sin6) in6_sin6_2_sin_in_sock((struct sockaddr *)sin6); pru = inetsw[ip_protox[nxt]].pr_usrreqs; /* addr will just be freed in sendit(). */ - return ((*pru->pru_send)(so, flags_arg, m, + return ((*pru->pru_send)(so, flags_arg | PRUS_IPV6, m, (struct sockaddr *)sin6, control, td)); } } else #endif if (sin6 && IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { /* * Given this is either an IPv6-only socket or no INET is * supported we will fail the send if the given destination * address is a v4mapped address. * * XXXGL: do we leak m and control? */ INP_UNLOCK(inp); return (EINVAL); } if (control) { if ((error = ip6_setpktopts(control, &opt, inp->in6p_outputopts, td->td_ucred, nxt)) != 0) { INP_UNLOCK(inp); ip6_clearpktopts(&opt, -1); if (control) m_freem(control); m_freem(m); return (error); } optp = &opt; } else optp = inp->in6p_outputopts; NET_EPOCH_ENTER(et); if (sin6) { /* * Since we saw no essential reason for calling in_pcbconnect, * we get rid of such kind of logic, and call in6_selectsrc * and in6_pcbsetport in order to fill in the local address * and the local port. */ if (sin6->sin6_port == 0) { error = EADDRNOTAVAIL; goto release; } if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) { /* how about ::ffff:0.0.0.0 case? */ error = EISCONN; goto release; } /* * Given we handle the v4mapped case in the INET block above * assert here that it must not happen anymore. */ KASSERT(!IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr), ("%s: sin6(%p)->sin6_addr is v4mapped which we " "should have handled.", __func__, sin6)); /* This only requires read-locking. */ error = in6_selectsrc_socket(sin6, optp, inp, td->td_ucred, scope_ambiguous, &in6a, NULL); if (error) goto release; laddr = &in6a; if (inp->inp_lport == 0) { struct inpcbinfo *pcbinfo; INP_WLOCK_ASSERT(inp); pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol); INP_HASH_WLOCK(pcbinfo); error = in6_pcbsetport(laddr, inp, td->td_ucred); INP_HASH_WUNLOCK(pcbinfo); if (error != 0) { /* Undo an address bind that may have occurred. */ inp->in6p_laddr = in6addr_any; goto release; } } faddr = &sin6->sin6_addr; fport = sin6->sin6_port; /* allow 0 port */ } else { if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) { error = ENOTCONN; goto release; } laddr = &inp->in6p_laddr; faddr = &inp->in6p_faddr; fport = inp->inp_fport; } ulen = m->m_pkthdr.len; plen = sizeof(struct udphdr) + ulen; hlen = sizeof(struct ip6_hdr); /* * Calculate data length and get a mbuf * for UDP and IP6 headers. */ M_PREPEND(m, hlen + sizeof(struct udphdr), M_NOWAIT); if (m == NULL) { error = ENOBUFS; goto release; } /* * Stuff checksum and output datagram. */ cscov = cscov_partial = 0; udp6 = (struct udphdr *)(mtod(m, caddr_t) + hlen); udp6->uh_sport = inp->inp_lport; /* lport is always set in the PCB */ udp6->uh_dport = fport; if (nxt == IPPROTO_UDPLITE) { struct udpcb *up; up = intoudpcb(inp); cscov = up->u_txcslen; if (cscov >= plen) cscov = 0; udp6->uh_ulen = htons(cscov); /* * For UDP-Lite, checksum coverage length of zero means * the entire UDPLite packet is covered by the checksum. */ cscov_partial = (cscov == 0) ? 0 : 1; } else if (plen <= 0xffff) udp6->uh_ulen = htons((u_short)plen); else udp6->uh_ulen = 0; udp6->uh_sum = 0; ip6 = mtod(m, struct ip6_hdr *); ip6->ip6_flow = inp->inp_flow & IPV6_FLOWINFO_MASK; ip6->ip6_vfc &= ~IPV6_VERSION_MASK; ip6->ip6_vfc |= IPV6_VERSION; ip6->ip6_plen = htons((u_short)plen); ip6->ip6_nxt = nxt; ip6->ip6_hlim = in6_selecthlim(inp, NULL); ip6->ip6_src = *laddr; ip6->ip6_dst = *faddr; #ifdef MAC mac_inpcb_create_mbuf(inp, m); #endif if (cscov_partial) { if ((udp6->uh_sum = in6_cksum_partial(m, nxt, sizeof(struct ip6_hdr), plen, cscov)) == 0) udp6->uh_sum = 0xffff; } else { udp6->uh_sum = in6_cksum_pseudo(ip6, plen, nxt, 0); m->m_pkthdr.csum_flags = CSUM_UDP_IPV6; m->m_pkthdr.csum_data = offsetof(struct udphdr, uh_sum); } flags = 0; #ifdef RSS { uint32_t hash_val, hash_type; uint8_t pr; pr = inp->inp_socket->so_proto->pr_protocol; /* * Calculate an appropriate RSS hash for UDP and * UDP Lite. * * The called function will take care of figuring out * whether a 2-tuple or 4-tuple hash is required based * on the currently configured scheme. * * Later later on connected socket values should be * cached in the inpcb and reused, rather than constantly * re-calculating it. * * UDP Lite is a different protocol number and will * likely end up being hashed as a 2-tuple until * RSS / NICs grow UDP Lite protocol awareness. */ if (rss_proto_software_hash_v6(faddr, laddr, fport, inp->inp_lport, pr, &hash_val, &hash_type) == 0) { m->m_pkthdr.flowid = hash_val; M_HASHTYPE_SET(m, hash_type); } /* * Don't override with the inp cached flowid. * * Until the whole UDP path is vetted, it may actually * be incorrect. */ flags |= IP_NODEFAULTFLOWID; } #endif UDPSTAT_INC(udps_opackets); if (nxt == IPPROTO_UDPLITE) UDPLITE_PROBE(send, NULL, inp, ip6, inp, udp6); else UDP_PROBE(send, NULL, inp, ip6, inp, udp6); error = ip6_output(m, optp, INP_WLOCKED(inp) ? &inp->inp_route6 : NULL, flags, inp->in6p_moptions, NULL, inp); INP_UNLOCK(inp); NET_EPOCH_EXIT(et); if (control) { ip6_clearpktopts(&opt, -1); m_freem(control); } return (error); release: INP_UNLOCK(inp); NET_EPOCH_EXIT(et); if (control) { ip6_clearpktopts(&opt, -1); m_freem(control); } m_freem(m); return (error); } static void udp6_abort(struct socket *so) { struct inpcb *inp; struct inpcbinfo *pcbinfo; pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol); inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp6_abort: inp == NULL")); INP_WLOCK(inp); #ifdef INET if (inp->inp_vflag & INP_IPV4) { struct pr_usrreqs *pru; uint8_t nxt; nxt = (inp->inp_socket->so_proto->pr_protocol == IPPROTO_UDP) ? IPPROTO_UDP : IPPROTO_UDPLITE; INP_WUNLOCK(inp); pru = inetsw[ip_protox[nxt]].pr_usrreqs; (*pru->pru_abort)(so); return; } #endif if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) { INP_HASH_WLOCK(pcbinfo); in6_pcbdisconnect(inp); inp->in6p_laddr = in6addr_any; INP_HASH_WUNLOCK(pcbinfo); soisdisconnected(so); } INP_WUNLOCK(inp); } static int udp6_attach(struct socket *so, int proto, struct thread *td) { struct inpcb *inp; struct inpcbinfo *pcbinfo; int error; pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol); inp = sotoinpcb(so); KASSERT(inp == NULL, ("udp6_attach: inp != NULL")); if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { error = soreserve(so, udp_sendspace, udp_recvspace); if (error) return (error); } INP_INFO_WLOCK(pcbinfo); error = in_pcballoc(so, pcbinfo); if (error) { INP_INFO_WUNLOCK(pcbinfo); return (error); } inp = (struct inpcb *)so->so_pcb; inp->inp_vflag |= INP_IPV6; if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0) inp->inp_vflag |= INP_IPV4; inp->in6p_hops = -1; /* use kernel default */ inp->in6p_cksum = -1; /* just to be sure */ /* * XXX: ugly!! * IPv4 TTL initialization is necessary for an IPv6 socket as well, * because the socket may be bound to an IPv6 wildcard address, * which may match an IPv4-mapped IPv6 address. */ inp->inp_ip_ttl = V_ip_defttl; error = udp_newudpcb(inp); if (error) { in_pcbdetach(inp); in_pcbfree(inp); INP_INFO_WUNLOCK(pcbinfo); return (error); } INP_WUNLOCK(inp); INP_INFO_WUNLOCK(pcbinfo); return (0); } static int udp6_bind(struct socket *so, struct sockaddr *nam, struct thread *td) { struct inpcb *inp; struct inpcbinfo *pcbinfo; int error; u_char vflagsav; pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol); inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp6_bind: inp == NULL")); INP_WLOCK(inp); INP_HASH_WLOCK(pcbinfo); vflagsav = inp->inp_vflag; inp->inp_vflag &= ~INP_IPV4; inp->inp_vflag |= INP_IPV6; if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0) { struct sockaddr_in6 *sin6_p; sin6_p = (struct sockaddr_in6 *)nam; if (IN6_IS_ADDR_UNSPECIFIED(&sin6_p->sin6_addr)) inp->inp_vflag |= INP_IPV4; #ifdef INET else if (IN6_IS_ADDR_V4MAPPED(&sin6_p->sin6_addr)) { struct sockaddr_in sin; in6_sin6_2_sin(&sin, sin6_p); inp->inp_vflag |= INP_IPV4; inp->inp_vflag &= ~INP_IPV6; error = in_pcbbind(inp, (struct sockaddr *)&sin, td->td_ucred); goto out; } #endif } error = in6_pcbbind(inp, nam, td->td_ucred); #ifdef INET out: #endif if (error != 0) inp->inp_vflag = vflagsav; INP_HASH_WUNLOCK(pcbinfo); INP_WUNLOCK(inp); return (error); } static void udp6_close(struct socket *so) { struct inpcb *inp; struct inpcbinfo *pcbinfo; pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol); inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp6_close: inp == NULL")); INP_WLOCK(inp); #ifdef INET if (inp->inp_vflag & INP_IPV4) { struct pr_usrreqs *pru; uint8_t nxt; nxt = (inp->inp_socket->so_proto->pr_protocol == IPPROTO_UDP) ? IPPROTO_UDP : IPPROTO_UDPLITE; INP_WUNLOCK(inp); pru = inetsw[ip_protox[nxt]].pr_usrreqs; (*pru->pru_disconnect)(so); return; } #endif if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) { INP_HASH_WLOCK(pcbinfo); in6_pcbdisconnect(inp); inp->in6p_laddr = in6addr_any; INP_HASH_WUNLOCK(pcbinfo); soisdisconnected(so); } INP_WUNLOCK(inp); } static int udp6_connect(struct socket *so, struct sockaddr *nam, struct thread *td) { #ifdef INET struct epoch_tracker et; #endif struct inpcb *inp; struct inpcbinfo *pcbinfo; struct sockaddr_in6 *sin6; int error; u_char vflagsav; pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol); inp = sotoinpcb(so); sin6 = (struct sockaddr_in6 *)nam; KASSERT(inp != NULL, ("udp6_connect: inp == NULL")); /* * XXXRW: Need to clarify locking of v4/v6 flags. */ INP_WLOCK(inp); #ifdef INET if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { struct sockaddr_in sin; if ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0) { error = EINVAL; goto out; } if ((inp->inp_vflag & INP_IPV4) == 0) { error = EAFNOSUPPORT; goto out; } if (inp->inp_faddr.s_addr != INADDR_ANY) { error = EISCONN; goto out; } in6_sin6_2_sin(&sin, sin6); error = prison_remote_ip4(td->td_ucred, &sin.sin_addr); if (error != 0) goto out; vflagsav = inp->inp_vflag; inp->inp_vflag |= INP_IPV4; inp->inp_vflag &= ~INP_IPV6; NET_EPOCH_ENTER(et); INP_HASH_WLOCK(pcbinfo); error = in_pcbconnect(inp, (struct sockaddr *)&sin, td->td_ucred); INP_HASH_WUNLOCK(pcbinfo); NET_EPOCH_EXIT(et); /* * If connect succeeds, mark socket as connected. If * connect fails and socket is unbound, reset inp_vflag * field. */ if (error == 0) soisconnected(so); else if (inp->inp_laddr.s_addr == INADDR_ANY && inp->inp_lport == 0) inp->inp_vflag = vflagsav; goto out; } else { if ((inp->inp_vflag & INP_IPV6) == 0) { error = EAFNOSUPPORT; goto out; } } #endif if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) { error = EISCONN; goto out; } error = prison_remote_ip6(td->td_ucred, &sin6->sin6_addr); if (error != 0) goto out; vflagsav = inp->inp_vflag; inp->inp_vflag &= ~INP_IPV4; inp->inp_vflag |= INP_IPV6; INP_HASH_WLOCK(pcbinfo); error = in6_pcbconnect(inp, nam, td->td_ucred); INP_HASH_WUNLOCK(pcbinfo); /* * If connect succeeds, mark socket as connected. If * connect fails and socket is unbound, reset inp_vflag * field. */ if (error == 0) soisconnected(so); else if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) && inp->inp_lport == 0) inp->inp_vflag = vflagsav; out: INP_WUNLOCK(inp); return (error); } static void udp6_detach(struct socket *so) { struct inpcb *inp; struct inpcbinfo *pcbinfo; struct udpcb *up; pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol); inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp6_detach: inp == NULL")); INP_INFO_WLOCK(pcbinfo); INP_WLOCK(inp); up = intoudpcb(inp); KASSERT(up != NULL, ("%s: up == NULL", __func__)); in_pcbdetach(inp); in_pcbfree(inp); INP_INFO_WUNLOCK(pcbinfo); udp_discardcb(up); } static int udp6_disconnect(struct socket *so) { struct inpcb *inp; struct inpcbinfo *pcbinfo; pcbinfo = udp_get_inpcbinfo(so->so_proto->pr_protocol); inp = sotoinpcb(so); KASSERT(inp != NULL, ("udp6_disconnect: inp == NULL")); INP_WLOCK(inp); #ifdef INET if (inp->inp_vflag & INP_IPV4) { struct pr_usrreqs *pru; uint8_t nxt; nxt = (inp->inp_socket->so_proto->pr_protocol == IPPROTO_UDP) ? IPPROTO_UDP : IPPROTO_UDPLITE; INP_WUNLOCK(inp); pru = inetsw[ip_protox[nxt]].pr_usrreqs; (void)(*pru->pru_disconnect)(so); return (0); } #endif if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) { INP_WUNLOCK(inp); return (ENOTCONN); } INP_HASH_WLOCK(pcbinfo); in6_pcbdisconnect(inp); inp->in6p_laddr = in6addr_any; INP_HASH_WUNLOCK(pcbinfo); SOCK_LOCK(so); so->so_state &= ~SS_ISCONNECTED; /* XXX */ SOCK_UNLOCK(so); INP_WUNLOCK(inp); return (0); } static int udp6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr, struct mbuf *control, struct thread *td) { int error; if (addr) { if (addr->sa_len != sizeof(struct sockaddr_in6)) { error = EINVAL; goto bad; } if (addr->sa_family != AF_INET6) { error = EAFNOSUPPORT; goto bad; } } return (udp6_output(so, flags, m, addr, control, td)); bad: if (control) m_freem(control); m_freem(m); return (error); } struct pr_usrreqs udp6_usrreqs = { .pru_abort = udp6_abort, .pru_attach = udp6_attach, .pru_bind = udp6_bind, .pru_connect = udp6_connect, .pru_control = in6_control, .pru_detach = udp6_detach, .pru_disconnect = udp6_disconnect, .pru_peeraddr = in6_mapped_peeraddr, .pru_send = udp6_send, .pru_shutdown = udp_shutdown, .pru_sockaddr = in6_mapped_sockaddr, .pru_soreceive = soreceive_dgram, .pru_sosend = sosend_dgram, .pru_sosetlabel = in_pcbsosetlabel, .pru_close = udp6_close }; Index: projects/clang1100-import/sys/sys/protosw.h =================================================================== --- projects/clang1100-import/sys/sys/protosw.h (revision 364034) +++ projects/clang1100-import/sys/sys/protosw.h (revision 364035) @@ -1,352 +1,353 @@ /*- * SPDX-License-Identifier: BSD-3-Clause * * Copyright (c) 1982, 1986, 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. * 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. * * @(#)protosw.h 8.1 (Berkeley) 6/2/93 * $FreeBSD$ */ #ifndef _SYS_PROTOSW_H_ #define _SYS_PROTOSW_H_ /* Forward declare these structures referenced from prototypes below. */ struct kaiocb; struct mbuf; struct thread; struct sockaddr; struct socket; struct sockopt; /*#ifdef _KERNEL*/ /* * Protocol switch table. * * Each protocol has a handle initializing one of these structures, * which is used for protocol-protocol and system-protocol communication. * * A protocol is called through the pr_init entry before any other. * Thereafter it is called every 200ms through the pr_fasttimo entry and * every 500ms through the pr_slowtimo for timer based actions. * The system will call the pr_drain entry if it is low on space and * this should throw away any non-critical data. * * Protocols pass data between themselves as chains of mbufs using * the pr_input and pr_output hooks. Pr_input passes data up (towards * the users) and pr_output passes it down (towards the interfaces); control * information passes up and down on pr_ctlinput and pr_ctloutput. * The protocol is responsible for the space occupied by any the * arguments to these entries and must dispose it. * * In retrospect, it would be a lot nicer to use an interface * similar to the vnode VOP interface. */ /* USE THESE FOR YOUR PROTOTYPES ! */ typedef int pr_input_t (struct mbuf **, int*, int); typedef int pr_output_t (struct mbuf *, struct socket *, ...); typedef void pr_ctlinput_t (int, struct sockaddr *, void *); typedef int pr_ctloutput_t (struct socket *, struct sockopt *); typedef void pr_init_t (void); typedef void pr_fasttimo_t (void); typedef void pr_slowtimo_t (void); typedef void pr_drain_t (void); struct protosw { short pr_type; /* socket type used for */ struct domain *pr_domain; /* domain protocol a member of */ short pr_protocol; /* protocol number */ short pr_flags; /* see below */ /* protocol-protocol hooks */ pr_input_t *pr_input; /* input to protocol (from below) */ pr_output_t *pr_output; /* output to protocol (from above) */ pr_ctlinput_t *pr_ctlinput; /* control input (from below) */ pr_ctloutput_t *pr_ctloutput; /* control output (from above) */ /* utility hooks */ pr_init_t *pr_init; pr_fasttimo_t *pr_fasttimo; /* fast timeout (200ms) */ pr_slowtimo_t *pr_slowtimo; /* slow timeout (500ms) */ pr_drain_t *pr_drain; /* flush any excess space possible */ struct pr_usrreqs *pr_usrreqs; /* user-protocol hook */ }; /*#endif*/ #define PR_SLOWHZ 2 /* 2 slow timeouts per second */ #define PR_FASTHZ 5 /* 5 fast timeouts per second */ /* * This number should be defined again within each protocol family to avoid * confusion. */ #define PROTO_SPACER 32767 /* spacer for loadable protocols */ /* * Values for pr_flags. * PR_ADDR requires PR_ATOMIC; * PR_ADDR and PR_CONNREQUIRED are mutually exclusive. * PR_IMPLOPCL means that the protocol allows sendto without prior connect, * and the protocol understands the MSG_EOF flag. The first property is * is only relevant if PR_CONNREQUIRED is set (otherwise sendto is allowed * anyhow). */ #define PR_ATOMIC 0x01 /* exchange atomic messages only */ #define PR_ADDR 0x02 /* addresses given with messages */ #define PR_CONNREQUIRED 0x04 /* connection required by protocol */ #define PR_WANTRCVD 0x08 /* want PRU_RCVD calls */ #define PR_RIGHTS 0x10 /* passes capabilities */ #define PR_IMPLOPCL 0x20 /* implied open/close */ #define PR_LASTHDR 0x40 /* enforce ipsec policy; last header */ /* * In earlier BSD network stacks, a single pr_usrreq() function pointer was * invoked with an operation number indicating what operation was desired. * We now provide individual function pointers which protocols can implement, * which offers a number of benefits (such as type checking for arguments). * These older constants are still present in order to support TCP debugging. */ #define PRU_ATTACH 0 /* attach protocol to up */ #define PRU_DETACH 1 /* detach protocol from up */ #define PRU_BIND 2 /* bind socket to address */ #define PRU_LISTEN 3 /* listen for connection */ #define PRU_CONNECT 4 /* establish connection to peer */ #define PRU_ACCEPT 5 /* accept connection from peer */ #define PRU_DISCONNECT 6 /* disconnect from peer */ #define PRU_SHUTDOWN 7 /* won't send any more data */ #define PRU_RCVD 8 /* have taken data; more room now */ #define PRU_SEND 9 /* send this data */ #define PRU_ABORT 10 /* abort (fast DISCONNECT, DETATCH) */ #define PRU_CONTROL 11 /* control operations on protocol */ #define PRU_SENSE 12 /* return status into m */ #define PRU_RCVOOB 13 /* retrieve out of band data */ #define PRU_SENDOOB 14 /* send out of band data */ #define PRU_SOCKADDR 15 /* fetch socket's address */ #define PRU_PEERADDR 16 /* fetch peer's address */ #define PRU_CONNECT2 17 /* connect two sockets */ /* begin for protocols internal use */ #define PRU_FASTTIMO 18 /* 200ms timeout */ #define PRU_SLOWTIMO 19 /* 500ms timeout */ #define PRU_PROTORCV 20 /* receive from below */ #define PRU_PROTOSEND 21 /* send to below */ /* end for protocol's internal use */ #define PRU_SEND_EOF 22 /* send and close */ #define PRU_SOSETLABEL 23 /* MAC label change */ #define PRU_CLOSE 24 /* socket close */ #define PRU_FLUSH 25 /* flush the socket */ #define PRU_NREQ 25 #ifdef PRUREQUESTS const char *prurequests[] = { "ATTACH", "DETACH", "BIND", "LISTEN", "CONNECT", "ACCEPT", "DISCONNECT", "SHUTDOWN", "RCVD", "SEND", "ABORT", "CONTROL", "SENSE", "RCVOOB", "SENDOOB", "SOCKADDR", "PEERADDR", "CONNECT2", "FASTTIMO", "SLOWTIMO", "PROTORCV", "PROTOSEND", "SEND_EOF", "SOSETLABEL", "CLOSE", "FLUSH", }; #endif #ifdef _KERNEL /* users shouldn't see this decl */ struct ifnet; struct stat; struct ucred; struct uio; /* * If the ordering here looks odd, that's because it's alphabetical. These * should eventually be merged back into struct protosw. * * Some fields initialized to defaults if they are NULL. * See uipc_domain.c:net_init_domain() */ struct pr_usrreqs { void (*pru_abort)(struct socket *so); int (*pru_accept)(struct socket *so, struct sockaddr **nam); int (*pru_attach)(struct socket *so, int proto, struct thread *td); int (*pru_bind)(struct socket *so, struct sockaddr *nam, struct thread *td); int (*pru_connect)(struct socket *so, struct sockaddr *nam, struct thread *td); int (*pru_connect2)(struct socket *so1, struct socket *so2); int (*pru_control)(struct socket *so, u_long cmd, caddr_t data, struct ifnet *ifp, struct thread *td); void (*pru_detach)(struct socket *so); int (*pru_disconnect)(struct socket *so); int (*pru_listen)(struct socket *so, int backlog, struct thread *td); int (*pru_peeraddr)(struct socket *so, struct sockaddr **nam); int (*pru_rcvd)(struct socket *so, int flags); int (*pru_rcvoob)(struct socket *so, struct mbuf *m, int flags); int (*pru_send)(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr, struct mbuf *control, struct thread *td); #define PRUS_OOB 0x1 #define PRUS_EOF 0x2 #define PRUS_MORETOCOME 0x4 #define PRUS_NOTREADY 0x8 +#define PRUS_IPV6 0x10 int (*pru_ready)(struct socket *so, struct mbuf *m, int count); int (*pru_sense)(struct socket *so, struct stat *sb); int (*pru_shutdown)(struct socket *so); int (*pru_flush)(struct socket *so, int direction); int (*pru_sockaddr)(struct socket *so, struct sockaddr **nam); int (*pru_sosend)(struct socket *so, struct sockaddr *addr, struct uio *uio, struct mbuf *top, struct mbuf *control, int flags, struct thread *td); int (*pru_soreceive)(struct socket *so, struct sockaddr **paddr, struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, int *flagsp); int (*pru_sopoll)(struct socket *so, int events, struct ucred *cred, struct thread *td); void (*pru_sosetlabel)(struct socket *so); void (*pru_close)(struct socket *so); int (*pru_bindat)(int fd, struct socket *so, struct sockaddr *nam, struct thread *td); int (*pru_connectat)(int fd, struct socket *so, struct sockaddr *nam, struct thread *td); int (*pru_aio_queue)(struct socket *so, struct kaiocb *job); }; /* * All nonvoid pru_*() functions below return EOPNOTSUPP. */ int pru_accept_notsupp(struct socket *so, struct sockaddr **nam); int pru_aio_queue_notsupp(struct socket *so, struct kaiocb *job); int pru_attach_notsupp(struct socket *so, int proto, struct thread *td); int pru_bind_notsupp(struct socket *so, struct sockaddr *nam, struct thread *td); int pru_bindat_notsupp(int fd, struct socket *so, struct sockaddr *nam, struct thread *td); int pru_connect_notsupp(struct socket *so, struct sockaddr *nam, struct thread *td); int pru_connectat_notsupp(int fd, struct socket *so, struct sockaddr *nam, struct thread *td); int pru_connect2_notsupp(struct socket *so1, struct socket *so2); int pru_control_notsupp(struct socket *so, u_long cmd, caddr_t data, struct ifnet *ifp, struct thread *td); int pru_disconnect_notsupp(struct socket *so); int pru_listen_notsupp(struct socket *so, int backlog, struct thread *td); int pru_peeraddr_notsupp(struct socket *so, struct sockaddr **nam); int pru_rcvd_notsupp(struct socket *so, int flags); int pru_rcvoob_notsupp(struct socket *so, struct mbuf *m, int flags); int pru_send_notsupp(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr, struct mbuf *control, struct thread *td); int pru_ready_notsupp(struct socket *so, struct mbuf *m, int count); int pru_sense_null(struct socket *so, struct stat *sb); int pru_shutdown_notsupp(struct socket *so); int pru_sockaddr_notsupp(struct socket *so, struct sockaddr **nam); int pru_sosend_notsupp(struct socket *so, struct sockaddr *addr, struct uio *uio, struct mbuf *top, struct mbuf *control, int flags, struct thread *td); int pru_soreceive_notsupp(struct socket *so, struct sockaddr **paddr, struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, int *flagsp); int pru_sopoll_notsupp(struct socket *so, int events, struct ucred *cred, struct thread *td); #endif /* _KERNEL */ /* * The arguments to the ctlinput routine are * (*protosw[].pr_ctlinput)(cmd, sa, arg); * where cmd is one of the commands below, sa is a pointer to a sockaddr, * and arg is a `void *' argument used within a protocol family. */ #define PRC_IFDOWN 0 /* interface transition */ #define PRC_ROUTEDEAD 1 /* select new route if possible ??? */ #define PRC_IFUP 2 /* interface has come back up */ /* was PRC_QUENCH2 3 DEC congestion bit says slow down */ /* was PRC_QUENCH 4 Deprecated by RFC 6633 */ #define PRC_MSGSIZE 5 /* message size forced drop */ #define PRC_HOSTDEAD 6 /* host appears to be down */ #define PRC_HOSTUNREACH 7 /* deprecated (use PRC_UNREACH_HOST) */ #define PRC_UNREACH_NET 8 /* no route to network */ #define PRC_UNREACH_HOST 9 /* no route to host */ #define PRC_UNREACH_PROTOCOL 10 /* dst says bad protocol */ #define PRC_UNREACH_PORT 11 /* bad port # */ /* was PRC_UNREACH_NEEDFRAG 12 (use PRC_MSGSIZE) */ #define PRC_UNREACH_SRCFAIL 13 /* source route failed */ #define PRC_REDIRECT_NET 14 /* net routing redirect */ #define PRC_REDIRECT_HOST 15 /* host routing redirect */ #define PRC_REDIRECT_TOSNET 16 /* redirect for type of service & net */ #define PRC_REDIRECT_TOSHOST 17 /* redirect for tos & host */ #define PRC_TIMXCEED_INTRANS 18 /* packet lifetime expired in transit */ #define PRC_TIMXCEED_REASS 19 /* lifetime expired on reass q */ #define PRC_PARAMPROB 20 /* header incorrect */ #define PRC_UNREACH_ADMIN_PROHIB 21 /* packet administrativly prohibited */ #define PRC_NCMDS 22 #define PRC_IS_REDIRECT(cmd) \ ((cmd) >= PRC_REDIRECT_NET && (cmd) <= PRC_REDIRECT_TOSHOST) #ifdef PRCREQUESTS char *prcrequests[] = { "IFDOWN", "ROUTEDEAD", "IFUP", "DEC-BIT-QUENCH2", "QUENCH", "MSGSIZE", "HOSTDEAD", "#7", "NET-UNREACH", "HOST-UNREACH", "PROTO-UNREACH", "PORT-UNREACH", "#12", "SRCFAIL-UNREACH", "NET-REDIRECT", "HOST-REDIRECT", "TOSNET-REDIRECT", "TOSHOST-REDIRECT", "TX-INTRANS", "TX-REASS", "PARAMPROB", "ADMIN-UNREACH" }; #endif /* * The arguments to ctloutput are: * (*protosw[].pr_ctloutput)(req, so, level, optname, optval, p); * req is one of the actions listed below, so is a (struct socket *), * level is an indication of which protocol layer the option is intended. * optname is a protocol dependent socket option request, * optval is a pointer to a mbuf-chain pointer, for value-return results. * The protocol is responsible for disposal of the mbuf chain *optval * if supplied, * the caller is responsible for any space held by *optval, when returned. * A non-zero return from ctloutput gives an * UNIX error number which should be passed to higher level software. */ #define PRCO_GETOPT 0 #define PRCO_SETOPT 1 #define PRCO_NCMDS 2 #ifdef PRCOREQUESTS char *prcorequests[] = { "GETOPT", "SETOPT", }; #endif #ifdef _KERNEL void pfctlinput(int, struct sockaddr *); struct domain *pffinddomain(int family); struct protosw *pffindproto(int family, int protocol, int type); struct protosw *pffindtype(int family, int type); int pf_proto_register(int family, struct protosw *npr); int pf_proto_unregister(int family, int protocol, int type); #endif #endif Index: projects/clang1100-import/tools/build/Makefile =================================================================== --- projects/clang1100-import/tools/build/Makefile (revision 364034) +++ projects/clang1100-import/tools/build/Makefile (revision 364035) @@ -1,180 +1,179 @@ # $FreeBSD$ .PATH: ${.CURDIR}/../../include LIB= egacy SRC= INCSGROUPS= INCS SYSINCS CASPERINC UFSINCS FFSINCS MSDOSFSINCS DISKINCS INCS= SYSINCSDIR= ${INCLUDEDIR}/sys CASPERINCDIR= ${INCLUDEDIR}/casper # Also add ufs/ffs/msdosfs/disk headers to allow building makefs as a bootstrap tool UFSINCSDIR= ${INCLUDEDIR}/ufs/ufs FFSINCSDIR= ${INCLUDEDIR}/ufs/ffs MSDOSFSINCSDIR= ${INCLUDEDIR}/fs/msdosfs DISKINCSDIR= ${INCLUDEDIR}/sys/disk BOOTSTRAPPING?= 0 _WITH_PWCACHEDB!= grep -c pwcache_groupdb /usr/include/grp.h || true .if ${_WITH_PWCACHEDB} == 0 .PATH: ${.CURDIR}/../../contrib/libc-pwcache CFLAGS+= -I${.CURDIR}/../../contrib/libc-pwcache \ -I${.CURDIR}/../../lib/libc/include SRCS+= pwcache.c .endif _WITH_STRSVIS!= grep -c strsvis /usr/include/vis.h || true .if ${_WITH_STRSVIS} == 0 .PATH: ${.CURDIR}/../../contrib/libc-vis SRCS+= vis.c CFLAGS+= -I${.CURDIR}/../../contrib/libc-vis \ -I${.CURDIR}/../../lib/libc/include .endif _WITH_REALLOCARRAY!= grep -c reallocarray /usr/include/stdlib.h || true .if ${_WITH_REALLOCARRAY} == 0 .PATH: ${.CURDIR}/../../lib/libc/stdlib INCS+= stdlib.h SRCS+= reallocarray.c CFLAGS+= -I${.CURDIR}/../../lib/libc/include .endif _WITH_UTIMENS!= grep -c utimensat /usr/include/sys/stat.h || true .if ${_WITH_UTIMENS} == 0 SYSINCS+= stat.h SRCS+= futimens.c utimensat.c .endif _WITH_EXPLICIT_BZERO!= grep -c explicit_bzero /usr/include/strings.h || true .if ${_WITH_EXPLICIT_BZERO} == 0 .PATH: ${SRCTOP}/sys/libkern INCS+= strings.h SRCS+= explicit_bzero.c .endif .if exists(/usr/include/capsicum_helpers.h) _WITH_CAPH_ENTER!= grep -c caph_enter /usr/include/capsicum_helpers.h || true _WITH_CAPH_RIGHTS_LIMIT!= grep -c caph_rights_limit /usr/include/capsicum_helpers.h || true .endif .if !defined(_WITH_CAPH_ENTER) || ${_WITH_CAPH_ENTER} == 0 || ${_WITH_CAPH_RIGHTS_LIMIT} == 0 .PATH: ${SRCTOP}/lib/libcapsicum INCS+= capsicum_helpers.h .PATH: ${SRCTOP}/lib/libcasper/libcasper INCS+= libcasper.h .endif CASPERINC+= ${SRCTOP}/lib/libcasper/services/cap_fileargs/cap_fileargs.h .if empty(SRCS) SRCS= dummy.c .endif .if defined(CROSS_BUILD_TESTING) SUBDIR= cross-build .endif # To allow bootstrapping makefs on FreeBSD 11 or non-FreeBSD systems: UFSINCS+= ${SRCTOP}/sys/ufs/ufs/dinode.h UFSINCS+= ${SRCTOP}/sys/ufs/ufs/dir.h FFSINCS+= ${SRCTOP}/sys/ufs/ffs/fs.h MSDOSFSINCS+= ${SRCTOP}/sys/fs/msdosfs/bootsect.h MSDOSFSINCS+= ${SRCTOP}/sys/fs/msdosfs/bpb.h MSDOSFSINCS+= ${SRCTOP}/sys/fs/msdosfs/denode.h MSDOSFSINCS+= ${SRCTOP}/sys/fs/msdosfs/direntry.h MSDOSFSINCS+= ${SRCTOP}/sys/fs/msdosfs/fat.h MSDOSFSINCS+= ${SRCTOP}/sys/fs/msdosfs/msdosfsmount.h DISKINCS+= ${SRCTOP}/sys/sys/disk/bsd.h # Needed to build config (since it uses libnv) SYSINCS+= ${SRCTOP}/sys/sys/nv.h ${SRCTOP}/sys/sys/cnv.h \ ${SRCTOP}/sys/sys/dnv.h # vtfontcvt is using sys/font.h SYSINCS+= ${SRCTOP}/sys/sys/font.h # We want to run the build with only ${WORLDTMP} in $PATH to ensure we don't # accidentally run tools that are incompatible but happen to be in $PATH. # This is especially important when building on Linux/MacOS where many of the # programs used during the build accept different flags or generate different # output. On those platforms we only symlink the tools known to be compatible # (e.g. basic utilities such as mkdir) into ${WORLDTMP} and build all others # from the FreeBSD sources during the bootstrap-tools stage. # basic commands: It is fine to use the host version for all of these even on # Linux/MacOS since we only use flags that are supported by all of them. _host_tools_to_symlink= basename bzip2 bunzip2 chmod chown cmp comm cp date dd \ dirname echo env false find fmt gzip gunzip head hostname id ln ls \ mkdir mv nice patch rm realpath sh sleep stat tee touch tr true uname \ uniq wc which # We also need a symlink to the absolute path to the make binary used for # the toplevel makefile. This is not necessarily the same as `which make` # since e.g. on Linux and MacOS that will be GNU make. _make_abs!= which "${MAKE}" _host_abs_tools_to_symlink= ${_make_abs}:make ${_make_abs}:bmake host-symlinks: @echo "Linking host tools into ${DESTDIR}/bin" .for _tool in ${_host_tools_to_symlink} - @if [ ! -e "${DESTDIR}/bin/${_tool}" ]; then \ - source_path=`which ${_tool}`; \ - if [ ! -e "$${source_path}" ] ; then \ - echo "Cannot find host tool '${_tool}'"; false; \ - fi; \ - ln -sfnv "$${source_path}" "${DESTDIR}/bin/${_tool}"; \ - fi + @source_path=`which ${_tool}`; \ + if [ ! -e "$${source_path}" ] ; then \ + echo "Cannot find host tool '${_tool}'"; false; \ + fi; \ + rm -f "${DESTDIR}/bin/${_tool}"; \ + cp -f "$${source_path}" "${DESTDIR}/bin/${_tool}" .endfor .for _tool in ${_host_abs_tools_to_symlink} @source_path="${_tool:S/:/ /:[1]}"; \ target_path="${DESTDIR}/bin/${_tool:S/:/ /:[2]}"; \ - if [ ! -e "$${target_path}" ] ; then \ - if [ ! -e "$${source_path}" ] ; then \ - echo "Host tool '${src_path}' is missing"; false; \ - fi; \ - ln -sfnv "$${source_path}" "$${target_path}"; \ - fi + if [ ! -e "$${source_path}" ] ; then \ + echo "Host tool '${src_path}' is missing"; false; \ + fi; \ + rm -f "$${target_path}"; \ + cp -f "$${source_path}" "$${target_path}" .endfor .if exists(/usr/libexec/flua) - ln -sf /usr/libexec/flua ${DESTDIR}/usr/libexec/flua + rm -f ${DESTDIR}/usr/libexec/flua + cp -f /usr/libexec/flua ${DESTDIR}/usr/libexec/flua .endif # Create all the directories that are needed during the legacy, bootstrap-tools # and cross-tools stages. We do this here using mkdir since mtree may not exist # yet (this happens if we are crossbuilding from Linux/Mac). INSTALLDIR_LIST= \ bin \ lib/casper \ lib/geom \ usr/include/casper \ usr/include/private/ucl \ usr/include/private/zstd \ usr/lib \ usr/libexec installdirs: mkdir -p ${INSTALLDIR_LIST:S,^,${DESTDIR}/,} # Link usr/bin, sbin, and usr/sbin to bin so that it doesn't matter whether a # bootstrap tool was added to WORLTMP with a symlink or by building it in the # bootstrap-tools phase. We could also overrride BINDIR when building bootstrap # tools but adding the symlinks is easier and means all tools are also # in the directory that they are installed to normally. .for _dir in sbin usr/sbin usr/bin # delete existing directories from before r340157 @if [ ! -L ${DESTDIR}/${_dir} ]; then \ echo "removing old non-symlink ${DESTDIR}/${_dir}"; \ rm -rf "${DESTDIR}/${_dir}"; \ fi .endfor ln -sfn bin ${DESTDIR}/sbin ln -sfn ../bin ${DESTDIR}/usr/bin ln -sfn ../bin ${DESTDIR}/usr/sbin .for _group in ${INCSGROUPS:NINCS} mkdir -p "${DESTDIR}/${${_group}DIR}" .endfor .include Index: projects/clang1100-import/usr.bin/grep/Makefile =================================================================== --- projects/clang1100-import/usr.bin/grep/Makefile (revision 364034) +++ projects/clang1100-import/usr.bin/grep/Makefile (revision 364035) @@ -1,71 +1,71 @@ # $NetBSD: Makefile,v 1.4 2011/02/16 01:31:33 joerg Exp $ # $FreeBSD$ # $OpenBSD: Makefile,v 1.6 2003/06/25 15:00:04 millert Exp $ .include -.if ${MK_BSD_GREP} == "yes" +.if ${MK_BSD_GREP} == "yes" || defined(BOOTSTRAPPING) PROG= grep MAN1= grep.1 zgrep.1 .else PROG= bsdgrep CLEANFILES+= bsdgrep.1 MAN1= bsdgrep.1 zgrep.1 bsdgrep.1: grep.1 ${CP} ${.ALLSRC} ${.TARGET} .endif SRCS= file.c grep.c queue.c util.c SCRIPTS= zgrep.sh LINKS= ${BINDIR}/zgrep ${BINDIR}/zfgrep \ ${BINDIR}/zgrep ${BINDIR}/zegrep \ ${BINDIR}/zgrep ${BINDIR}/bzgrep \ ${BINDIR}/zgrep ${BINDIR}/bzegrep \ ${BINDIR}/zgrep ${BINDIR}/bzfgrep \ ${BINDIR}/zgrep ${BINDIR}/lzgrep \ ${BINDIR}/zgrep ${BINDIR}/lzegrep \ ${BINDIR}/zgrep ${BINDIR}/lzfgrep \ ${BINDIR}/zgrep ${BINDIR}/xzgrep \ ${BINDIR}/zgrep ${BINDIR}/xzegrep \ ${BINDIR}/zgrep ${BINDIR}/xzfgrep \ ${BINDIR}/zgrep ${BINDIR}/zstdgrep \ ${BINDIR}/zgrep ${BINDIR}/zstdegrep \ ${BINDIR}/zgrep ${BINDIR}/zstdegrep MLINKS= zgrep.1 zfgrep.1 \ zgrep.1 zegrep.1 \ zgrep.1 bzgrep.1 \ zgrep.1 bzegrep.1 \ zgrep.1 bzfgrep.1 \ zgrep.1 lzgrep.1 \ zgrep.1 lzegrep.1 \ zgrep.1 lzfgrep.1 \ zgrep.1 xzgrep.1 \ zgrep.1 xzegrep.1 \ zgrep.1 xzfgrep.1 \ zgrep.1 zstdgrep.1 \ zgrep.1 zstdegrep.1 \ zgrep.1 zstdfgrep.1 CFLAGS.gcc+= --param max-inline-insns-single=500 -.if ${MK_BSD_GREP} == "yes" +.if ${MK_BSD_GREP} == "yes" || defined(BOOTSTRAPPING) LINKS+= ${BINDIR}/grep ${BINDIR}/egrep \ ${BINDIR}/grep ${BINDIR}/fgrep \ ${BINDIR}/grep ${BINDIR}/rgrep \ MLINKS+= grep.1 egrep.1 \ grep.1 fgrep.1 \ grep.1 rgrep.1 .endif .if ${MK_GNU_GREP_COMPAT} != "no" CFLAGS+= -DWITH_GNU_COMPAT LIBADD+= regex .endif HAS_TESTS= SUBDIR.${MK_TESTS}+= tests .include Index: projects/clang1100-import/usr.bin/grep/grep.1 =================================================================== --- projects/clang1100-import/usr.bin/grep/grep.1 (revision 364034) +++ projects/clang1100-import/usr.bin/grep/grep.1 (revision 364035) @@ -1,530 +1,530 @@ .\" $NetBSD: grep.1,v 1.2 2011/02/16 01:31:33 joerg Exp $ .\" $FreeBSD$ .\" $OpenBSD: grep.1,v 1.38 2010/04/05 06:30:59 jmc Exp $ .\" Copyright (c) 1980, 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. .\" 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. .\" .\" @(#)grep.1 8.3 (Berkeley) 4/18/94 .\" -.Dd August 21, 2018 +.Dd August 7, 2020 .Dt GREP 1 .Os .Sh NAME .Nm grep , .Nm egrep , .Nm fgrep , .Nm rgrep .Nd file pattern searcher .Sh SYNOPSIS .Nm grep .Bk -words .Op Fl abcdDEFGHhIiLlmnOopqRSsUVvwxz .Op Fl A Ar num .Op Fl B Ar num .Op Fl C Ns Op Ar num .Op Fl e Ar pattern .Op Fl f Ar file .Op Fl Fl binary-files= Ns Ar value .Op Fl Fl color Ns Op Cm = Ns Ar when .Op Fl Fl colour Ns Op Cm = Ns Ar when .Op Fl Fl context Ns Op Cm = Ns Ar num .Op Fl Fl label .Op Fl Fl line-buffered .Op Fl Fl null .Op Ar pattern .Op Ar .Ek .Sh DESCRIPTION The .Nm grep utility searches any given input files, selecting lines that match one or more patterns. By default, a pattern matches an input line if the regular expression (RE) in the pattern matches the input line without its trailing newline. An empty expression matches every line. Each input line that matches at least one of the patterns is written to the standard output. .Pp .Nm grep is used for simple patterns and basic regular expressions .Pq BREs ; .Nm egrep can handle extended regular expressions .Pq EREs . See .Xr re_format 7 for more information on regular expressions. .Nm fgrep is quicker than both .Nm grep and .Nm egrep , but can only handle fixed patterns (i.e., it does not interpret regular expressions). Patterns may consist of one or more lines, allowing any of the pattern lines to match a portion of the input. .Pp The following options are available: .Bl -tag -width indent .It Fl A Ar num , Fl Fl after-context= Ns Ar num Print .Ar num lines of trailing context after each match. See also the .Fl B and .Fl C options. .It Fl a , Fl Fl text Treat all files as ASCII text. Normally .Nm will simply print .Dq Binary file ... matches if files contain binary characters. Use of this option forces .Nm to output lines matching the specified pattern. .It Fl B Ar num , Fl Fl before-context= Ns Ar num Print .Ar num lines of leading context before each match. See also the .Fl A and .Fl C options. .It Fl b , Fl Fl byte-offset The offset in bytes of a matched pattern is displayed in front of the respective matched line. .It Fl C Ns Oo Ar num Oc , Fl Fl context Ns Oo = Ns Ar num Oc Print .Ar num lines of leading and trailing context surrounding each match. The default value of .Ar num is .Dq 2 and is equivalent to .Dq Fl A Ar 2 Fl B Ar 2 . Note: no whitespace may be given between the option and its argument. .It Fl c , Fl Fl count Only a count of selected lines is written to standard output. .It Fl Fl colour= Ns Oo Ar when Oc , Fl Fl color= Ns Oo Ar when Oc Mark up the matching text with the expression stored in the .Ev GREP_COLOR environment variable. The possible values of .Ar when are .Dq Cm never , .Dq Cm always and .Dq Cm auto . .It Fl D Ar action , Fl Fl devices= Ns Ar action Specify the demanded .Ar action for devices, FIFOs and sockets. The default .Ar action is .Dq Cm read , which means, that they are read as if they were normal files. If the .Ar action is set to .Dq Cm skip , devices are silently skipped. .It Fl d Ar action , Fl Fl directories= Ns Ar action Specify the demanded .Ar action for directories. It is .Dq Cm read by default, which means that the directories are read in the same manner as normal files. Other possible values are .Dq Cm skip to silently ignore the directories, and .Dq Cm recurse to read them recursively, which has the same effect as the .Fl R and .Fl r option. .It Fl E , Fl Fl extended-regexp Interpret .Ar pattern as an extended regular expression (i.e., force .Nm grep to behave as .Nm egrep ) . .It Fl e Ar pattern , Fl Fl regexp= Ns Ar pattern Specify a .Ar pattern used during the search of the input: an input line is selected if it matches any of the specified patterns. This option is most useful when multiple .Fl e options are used to specify multiple patterns, or when a .Ar pattern begins with a dash .Pq Sq - . .It Fl Fl exclude Ar pattern If specified, it excludes files matching the given filename .Ar pattern from the search. Note that .Fl Fl exclude and .Fl Fl include patterns are processed in the order given. -If a name patches multiple patterns, the latest matching rule wins. +If a name matches multiple patterns, the latest matching rule wins. If no .Fl Fl include pattern is specified, all files are searched that are not excluded. Patterns are matched to the full path specified, not only to the filename component. .It Fl Fl exclude-dir Ar pattern If .Fl R is specified, it excludes directories matching the given filename .Ar pattern from the search. Note that .Fl Fl exclude-dir and .Fl Fl include-dir patterns are processed in the order given. -If a name patches multiple patterns, the latest matching rule wins. +If a name matches multiple patterns, the latest matching rule wins. If no .Fl Fl include-dir pattern is specified, all directories are searched that are not excluded. .It Fl F , Fl Fl fixed-strings Interpret .Ar pattern as a set of fixed strings (i.e., force .Nm grep to behave as .Nm fgrep ) . .It Fl f Ar file , Fl Fl file= Ns Ar file Read one or more newline separated patterns from .Ar file . Empty pattern lines match every input line. Newlines are not considered part of a pattern. If .Ar file is empty, nothing is matched. .It Fl G , Fl Fl basic-regexp Interpret .Ar pattern as a basic regular expression (i.e., force .Nm grep to behave as traditional .Nm grep ) . .It Fl H Always print filename headers with output lines. .It Fl h , Fl Fl no-filename Never print filename headers .Pq i.e., filenames with output lines. .It Fl Fl help Print a brief help message. .It Fl I Ignore binary files. This option is equivalent to the .Dq Fl Fl binary-file= Ns Cm without-match option. .It Fl i , Fl Fl ignore-case Perform case insensitive matching. By default, .Nm grep is case sensitive. .It Fl Fl include Ar pattern If specified, only files matching the given filename .Ar pattern are searched. Note that .Fl Fl include and .Fl Fl exclude patterns are processed in the order given. -If a name patches multiple patterns, the latest matching rule wins. +If a name matches multiple patterns, the latest matching rule wins. Patterns are matched to the full path specified, not only to the filename component. .It Fl Fl include-dir Ar pattern If .Fl R is specified, only directories matching the given filename .Ar pattern are searched. Note that .Fl Fl include-dir and .Fl Fl exclude-dir patterns are processed in the order given. -If a name patches multiple patterns, the latest matching rule wins. +If a name matches multiple patterns, the latest matching rule wins. .It Fl L , Fl Fl files-without-match Only the names of files not containing selected lines are written to standard output. Pathnames are listed once per file searched. If the standard input is searched, the string .Dq (standard input) is written unless a .Fl Fl label is specified. .It Fl l , Fl Fl files-with-matches Only the names of files containing selected lines are written to standard output. .Nm grep will only search a file until a match has been found, making searches potentially less expensive. Pathnames are listed once per file searched. If the standard input is searched, the string .Dq (standard input) is written unless a .Fl Fl label is specified. .It Fl Fl label Label to use in place of .Dq (standard input) for a file name where a file name would normally be printed. This option applies to .Fl H , .Fl L , and .Fl l . .It Fl Fl mmap Use .Xr mmap 2 instead of .Xr read 2 to read input, which can result in better performance under some circumstances but can cause undefined behaviour. .It Fl m Ar num , Fl Fl max-count= Ns Ar num Stop reading the file after .Ar num matches. .It Fl n , Fl Fl line-number Each output line is preceded by its relative line number in the file, starting at line 1. The line number counter is reset for each file processed. This option is ignored if .Fl c , .Fl L , .Fl l , or .Fl q is specified. .It Fl Fl null Prints a zero-byte after the file name. .It Fl O If .Fl R is specified, follow symbolic links only if they were explicitly listed on the command line. The default is not to follow symbolic links. .It Fl o , Fl Fl only-matching Prints only the matching part of the lines. .It Fl p If .Fl R is specified, no symbolic links are followed. This is the default. .It Fl q , Fl Fl quiet , Fl Fl silent Quiet mode: suppress normal output. .Nm grep will only search a file until a match has been found, making searches potentially less expensive. .It Fl R , Fl r , Fl Fl recursive Recursively search subdirectories listed. (i.e., force .Nm grep to behave as .Nm rgrep ) . .It Fl S If .Fl R is specified, all symbolic links are followed. The default is not to follow symbolic links. .It Fl s , Fl Fl no-messages Silent mode. Nonexistent and unreadable files are ignored (i.e., their error messages are suppressed). .It Fl U , Fl Fl binary Search binary files, but do not attempt to print them. .It Fl u This option has no effect and is provided only for compatibility with GNU grep. .It Fl V , Fl Fl version Display version information and exit. .It Fl v , Fl Fl invert-match Selected lines are those .Em not matching any of the specified patterns. .It Fl w , Fl Fl word-regexp The expression is searched for as a word (as if surrounded by .Sq [[:<:]] and .Sq [[:>:]] ; see .Xr re_format 7 ) . .It Fl x , Fl Fl line-regexp Only input lines selected against an entire fixed string or regular expression are considered to be matching lines. .It Fl y Equivalent to .Fl i . Obsoleted. .It Fl z , Fl Fl null-data Treat input and output data as sequences of lines terminated by a zero-byte instead of a newline. .It Fl Fl binary-files= Ns Ar value Controls searching and printing of binary files. Options are: .Bl -tag -compact -width "binary (default)" .It Cm binary No (default) Search binary files but do not print them. .It Cm without-match Do not search binary files. .It Cm text Treat all files as text. .El .It Fl Fl line-buffered Force output to be line buffered. By default, output is line buffered when standard output is a terminal and block buffered otherwise. .El .Pp If no file arguments are specified, the standard input is used. Additionally, .Dq Cm - may be used in place of a file name, anywhere that a file name is accepted, to read from standard input. This includes both .Fl f and file arguments. .Sh EXIT STATUS The .Nm grep utility exits with one of the following values: .Pp .Bl -tag -width flag -compact .It Li 0 One or more lines were selected. .It Li 1 No lines were selected. .It Li \*(Gt1 An error occurred. .El .Sh EXAMPLES .Bl -dash .It To find all occurrences of the word .Sq patricia in a file: .Pp .Dl $ grep 'patricia' myfile .It To find all occurrences of the pattern .Ql .Pp at the beginning of a line: .Pp .Dl $ grep '^\e.Pp' myfile .Pp The apostrophes ensure the entire expression is evaluated by .Nm grep instead of by the user's shell. The caret .Ql ^ matches the null string at the beginning of a line, and the .Ql \e escapes the .Ql \&. , which would otherwise match any character. .It To find all lines in a file which do not contain the words .Sq foo or .Sq bar : .Pp .Dl $ grep -v -e 'foo' -e 'bar' myfile .It A simple example of an extended regular expression: .Pp .Dl $ egrep '19|20|25' calendar .Pp Peruses the file .Sq calendar looking for either 19, 20, or 25. .El .Sh SEE ALSO .Xr ed 1 , .Xr ex 1 , .Xr sed 1 , .Xr zgrep 1 , .Xr re_format 7 .Sh STANDARDS The .Nm utility is compliant with the .St -p1003.1-2008 specification. .Pp The flags .Op Fl AaBbCDdGHhILmoPRSUVw are extensions to that specification, and the behaviour of the .Fl f flag when used with an empty pattern file is left undefined. .Pp All long options are provided for compatibility with GNU versions of this utility. .Pp Historic versions of the .Nm grep utility also supported the flags .Op Fl ruy . This implementation supports those options; however, their use is strongly discouraged. .Sh HISTORY The .Nm grep command first appeared in .At v6 . Index: projects/clang1100-import/usr.sbin/makefs/msdos/msdosfs_vnops.c =================================================================== --- projects/clang1100-import/usr.sbin/makefs/msdos/msdosfs_vnops.c (revision 364034) +++ projects/clang1100-import/usr.sbin/makefs/msdos/msdosfs_vnops.c (revision 364035) @@ -1,645 +1,644 @@ /* $NetBSD: msdosfs_vnops.c,v 1.19 2017/04/13 17:10:12 christos Exp $ */ /*- * SPDX-License-Identifier: BSD-4-Clause * * Copyright (C) 1994, 1995, 1997 Wolfgang Solfrank. * Copyright (C) 1994, 1995, 1997 TooLs GmbH. * All rights reserved. * Original code by Paul Popelka (paulp@uts.amdahl.com) (see below). * * 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 TooLs GmbH. * 4. The name of TooLs GmbH may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``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 TOOLS GMBH 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. */ /*- * Written by Paul Popelka (paulp@uts.amdahl.com) * * You can do anything you want with this software, just don't say you wrote * it, and don't remove this notice. * * This software is provided "as is". * * The author supplies this software to be publicly redistributed on the * understanding that the author is not responsible for the correct * functioning of this software in any circumstances and is not liable for * any damages caused by this software. * * October 1992 */ #include __FBSDID("$FreeBSD$"); #include -#include #include #include #include #include #include #include #include #include #include #include "ffs/buf.h" #include #include "msdos/direntry.h" #include #include #include #include "makefs.h" #include "msdos.h" /* * Some general notes: * * In the ufs filesystem the inodes, superblocks, and indirect blocks are * read/written using the vnode for the filesystem. Blocks that represent * the contents of a file are read/written using the vnode for the file * (including directories when they are read/written as files). This * presents problems for the dos filesystem because data that should be in * an inode (if dos had them) resides in the directory itself. Since we * must update directory entries without the benefit of having the vnode * for the directory we must use the vnode for the filesystem. This means * that when a directory is actually read/written (via read, write, or * readdir, or seek) we must use the vnode for the filesystem instead of * the vnode for the directory as would happen in ufs. This is to insure we * retrieve the correct block from the buffer cache since the hash value is * based upon the vnode address and the desired block number. */ static int msdosfs_wfile(const char *, struct denode *, fsnode *); static void unix2fattime(const struct timespec *tsp, uint16_t *ddp, uint16_t *dtp); static void msdosfs_times(struct denode *dep, const struct stat *st) { if (stampst.st_ino) st = &stampst; #ifdef HAVE_STRUCT_STAT_BIRTHTIME unix2fattime(&st->st_birthtim, &dep->de_CDate, &dep->de_CTime); #else unix2fattime(&st->st_ctim, &dep->de_CDate, &dep->de_CTime); #endif unix2fattime(&st->st_atim, &dep->de_ADate, NULL); unix2fattime(&st->st_mtim, &dep->de_MDate, &dep->de_MTime); } static void unix2fattime(const struct timespec *tsp, uint16_t *ddp, uint16_t *dtp) { time_t t1; struct tm lt = {0}; t1 = tsp->tv_sec; localtime_r(&t1, <); unsigned long fat_time = ((lt.tm_year - 80) << 25) | ((lt.tm_mon + 1) << 21) | (lt.tm_mday << 16) | (lt.tm_hour << 11) | (lt.tm_min << 5) | (lt.tm_sec >> 1); if (ddp != NULL) *ddp = (uint16_t)(fat_time >> 16); if (dtp != NULL) *dtp = (uint16_t)fat_time; } /* * When we search a directory the blocks containing directory entries are * read and examined. The directory entries contain information that would * normally be in the inode of a unix filesystem. This means that some of * a directory's contents may also be in memory resident denodes (sort of * an inode). This can cause problems if we are searching while some other * process is modifying a directory. To prevent one process from accessing * incompletely modified directory information we depend upon being the * sole owner of a directory block. bread/brelse provide this service. * This being the case, when a process modifies a directory it must first * acquire the disk block that contains the directory entry to be modified. * Then update the disk block and the denode, and then write the disk block * out to disk. This way disk blocks containing directory entries and in * memory denode's will be in synch. */ static int msdosfs_findslot(struct denode *dp, struct componentname *cnp) { daddr_t bn; int error; int slotcount; int slotoffset = 0; int frcn; u_long cluster; int blkoff; u_int diroff; int blsize; struct msdosfsmount *pmp; struct buf *bp = 0; struct direntry *dep; u_char dosfilename[12]; int wincnt = 1; int chksum = -1, chksum_ok; int olddos = 1; pmp = dp->de_pmp; switch (unix2dosfn((const u_char *)cnp->cn_nameptr, dosfilename, cnp->cn_namelen, 0)) { case 0: return (EINVAL); case 1: break; case 2: wincnt = winSlotCnt((const u_char *)cnp->cn_nameptr, cnp->cn_namelen) + 1; break; case 3: olddos = 0; wincnt = winSlotCnt((const u_char *)cnp->cn_nameptr, cnp->cn_namelen) + 1; break; } if (pmp->pm_flags & MSDOSFSMNT_SHORTNAME) wincnt = 1; /* * Suppress search for slots unless creating * file and at end of pathname, in which case * we watch for a place to put the new file in * case it doesn't already exist. */ slotcount = 0; MSDOSFS_DPRINTF(("%s(): dos filename: %s\n", __func__, dosfilename)); /* * Search the directory pointed at by vdp for the name pointed at * by cnp->cn_nameptr. */ /* * The outer loop ranges over the clusters that make up the * directory. Note that the root directory is different from all * other directories. It has a fixed number of blocks that are not * part of the pool of allocatable clusters. So, we treat it a * little differently. The root directory starts at "cluster" 0. */ diroff = 0; for (frcn = 0; diroff < dp->de_FileSize; frcn++) { if ((error = pcbmap(dp, frcn, &bn, &cluster, &blsize)) != 0) { if (error == E2BIG) break; return (error); } error = bread(pmp->pm_devvp, bn, blsize, 0, &bp); if (error) { return (error); } for (blkoff = 0; blkoff < blsize; blkoff += sizeof(struct direntry), diroff += sizeof(struct direntry)) { dep = (struct direntry *)(bp->b_data + blkoff); /* * If the slot is empty and we are still looking * for an empty then remember this one. If the * slot is not empty then check to see if it * matches what we are looking for. If the slot * has never been filled with anything, then the * remainder of the directory has never been used, * so there is no point in searching it. */ if (dep->deName[0] == SLOT_EMPTY || dep->deName[0] == SLOT_DELETED) { /* * Drop memory of previous long matches */ chksum = -1; if (slotcount < wincnt) { slotcount++; slotoffset = diroff; } if (dep->deName[0] == SLOT_EMPTY) { brelse(bp); goto notfound; } } else { /* * If there wasn't enough space for our * winentries, forget about the empty space */ if (slotcount < wincnt) slotcount = 0; /* * Check for Win95 long filename entry */ if (dep->deAttributes == ATTR_WIN95) { if (pmp->pm_flags & MSDOSFSMNT_SHORTNAME) continue; chksum = winChkName( (const u_char *)cnp->cn_nameptr, cnp->cn_namelen, (struct winentry *)dep, chksum); continue; } /* * Ignore volume labels (anywhere, not just * the root directory). */ if (dep->deAttributes & ATTR_VOLUME) { chksum = -1; continue; } /* * Check for a checksum or name match */ chksum_ok = (chksum == winChksum(dep->deName)); if (!chksum_ok && (!olddos || memcmp(dosfilename, dep->deName, 11))) { chksum = -1; continue; } MSDOSFS_DPRINTF(("%s(): match blkoff %d, diroff %u\n", __func__, blkoff, diroff)); /* * Remember where this directory * entry came from for whoever did * this lookup. */ dp->de_fndoffset = diroff; dp->de_fndcnt = 0; return EEXIST; } } /* for (blkoff = 0; .... */ /* * Release the buffer holding the directory cluster just * searched. */ brelse(bp); } /* for (frcn = 0; ; frcn++) */ notfound: /* * We hold no disk buffers at this point. */ /* * If we get here we didn't find the entry we were looking for. But * that's ok if we are creating or renaming and are at the end of * the pathname and the directory hasn't been removed. */ MSDOSFS_DPRINTF(("%s(): refcnt %ld, slotcount %d, slotoffset %d\n", __func__, dp->de_refcnt, slotcount, slotoffset)); /* * Fixup the slot description to point to the place where * we might put the new DOS direntry (putting the Win95 * long name entries before that) */ if (!slotcount) { slotcount = 1; slotoffset = diroff; } if (wincnt > slotcount) { slotoffset += sizeof(struct direntry) * (wincnt - slotcount); } /* * Return an indication of where the new directory * entry should be put. */ dp->de_fndoffset = slotoffset; dp->de_fndcnt = wincnt - 1; /* * We return with the directory locked, so that * the parameters we set up above will still be * valid if we actually decide to do a direnter(). * We return ni_vp == NULL to indicate that the entry * does not currently exist; we leave a pointer to * the (locked) directory inode in ndp->ni_dvp. * * NB - if the directory is unlocked, then this * information cannot be used. */ return 0; } /* * Create a regular file. On entry the directory to contain the file being * created is locked. We must release before we return. */ struct denode * msdosfs_mkfile(const char *path, struct denode *pdep, fsnode *node) { struct componentname cn; struct denode ndirent; struct denode *dep; int error; struct stat *st = &node->inode->st; cn.cn_nameptr = node->name; cn.cn_namelen = strlen(node->name); MSDOSFS_DPRINTF(("%s(name %s, mode 0%o size %zu)\n", __func__, node->name, st->st_mode, (size_t)st->st_size)); /* * If this is the root directory and there is no space left we * can't do anything. This is because the root directory can not * change size. */ if (pdep->de_StartCluster == MSDOSFSROOT && pdep->de_fndoffset >= pdep->de_FileSize) { error = ENOSPC; goto bad; } /* * Create a directory entry for the file, then call createde() to * have it installed. NOTE: DOS files are always executable. We * use the absence of the owner write bit to make the file * readonly. */ memset(&ndirent, 0, sizeof(ndirent)); if ((error = uniqdosname(pdep, &cn, ndirent.de_Name)) != 0) goto bad; ndirent.de_Attributes = (st->st_mode & S_IWUSR) ? ATTR_ARCHIVE : ATTR_ARCHIVE | ATTR_READONLY; ndirent.de_StartCluster = 0; ndirent.de_FileSize = 0; ndirent.de_pmp = pdep->de_pmp; ndirent.de_flag = DE_ACCESS | DE_CREATE | DE_UPDATE; msdosfs_times(&ndirent, &node->inode->st); if ((error = msdosfs_findslot(pdep, &cn)) != 0) goto bad; if ((error = createde(&ndirent, pdep, &dep, &cn)) != 0) goto bad; if ((error = msdosfs_wfile(path, dep, node)) != 0) goto bad; return dep; bad: errno = error; return NULL; } static int msdosfs_updatede(struct denode *dep) { struct buf *bp; struct direntry *dirp; int error; dep->de_flag &= ~DE_MODIFIED; error = readde(dep, &bp, &dirp); if (error) return error; DE_EXTERNALIZE(dirp, dep); error = bwrite(bp); return error; } /* * Write data to a file or directory. */ static int msdosfs_wfile(const char *path, struct denode *dep, fsnode *node) { int error, fd; size_t osize = dep->de_FileSize; struct stat *st = &node->inode->st; size_t nsize, offs; struct msdosfsmount *pmp = dep->de_pmp; struct buf *bp; char *dat; u_long cn = 0; error = 0; /* XXX: gcc/vax */ MSDOSFS_DPRINTF(("%s(diroff %lu, dirclust %lu, startcluster %lu)\n", __func__, dep->de_diroffset, dep->de_dirclust, dep->de_StartCluster)); if (st->st_size == 0) return 0; /* Don't bother to try to write files larger than the fs limit */ if (st->st_size > MSDOSFS_FILESIZE_MAX) return EFBIG; nsize = st->st_size; MSDOSFS_DPRINTF(("%s(nsize=%zu, osize=%zu)\n", __func__, nsize, osize)); if (nsize > osize) { if ((error = deextend(dep, nsize, NULL)) != 0) return error; if ((error = msdosfs_updatede(dep)) != 0) return error; } if ((fd = open(path, O_RDONLY)) == -1) { error = errno; MSDOSFS_DPRINTF(("open %s: %s", path, strerror(error))); return error; } if ((dat = mmap(0, nsize, PROT_READ, MAP_FILE | MAP_PRIVATE, fd, 0)) == MAP_FAILED) { error = errno; MSDOSFS_DPRINTF(("%s: mmap %s: %s", __func__, node->name, strerror(error))); close(fd); goto out; } close(fd); for (offs = 0; offs < nsize;) { int blsize, cpsize; daddr_t bn; u_long on = offs & pmp->pm_crbomask; if ((error = pcbmap(dep, cn++, &bn, NULL, &blsize)) != 0) { MSDOSFS_DPRINTF(("%s: pcbmap %lu", __func__, (unsigned long)bn)); goto out; } MSDOSFS_DPRINTF(("%s(cn=%lu, bn=%llu, blsize=%d)\n", __func__, cn, (unsigned long long)bn, blsize)); if ((error = bread(pmp->pm_devvp, bn, blsize, 0, &bp)) != 0) { MSDOSFS_DPRINTF(("bread %d\n", error)); goto out; } cpsize = MIN((nsize - offs), blsize - on); memcpy(bp->b_data + on, dat + offs, cpsize); bwrite(bp); offs += cpsize; } munmap(dat, nsize); return 0; out: munmap(dat, nsize); return error; } static const struct { struct direntry dot; struct direntry dotdot; } dosdirtemplate = { { ". ", /* the . entry */ ATTR_DIRECTORY, /* file attribute */ 0, /* reserved */ 0, { 0, 0 }, { 0, 0 }, /* create time & date */ { 0, 0 }, /* access date */ { 0, 0 }, /* high bits of start cluster */ { 210, 4 }, { 210, 4 }, /* modify time & date */ { 0, 0 }, /* startcluster */ { 0, 0, 0, 0 } /* filesize */ }, { ".. ", /* the .. entry */ ATTR_DIRECTORY, /* file attribute */ 0, /* reserved */ 0, { 0, 0 }, { 0, 0 }, /* create time & date */ { 0, 0 }, /* access date */ { 0, 0 }, /* high bits of start cluster */ { 210, 4 }, { 210, 4 }, /* modify time & date */ { 0, 0 }, /* startcluster */ { 0, 0, 0, 0 } /* filesize */ } }; struct denode * msdosfs_mkdire(const char *path, struct denode *pdep, fsnode *node) { struct denode ndirent; struct denode *dep; struct componentname cn; struct msdosfsmount *pmp = pdep->de_pmp; int error; u_long newcluster, pcl, bn; struct direntry *denp; struct buf *bp; cn.cn_nameptr = node->name; cn.cn_namelen = strlen(node->name); /* * If this is the root directory and there is no space left we * can't do anything. This is because the root directory can not * change size. */ if (pdep->de_StartCluster == MSDOSFSROOT && pdep->de_fndoffset >= pdep->de_FileSize) { error = ENOSPC; goto bad2; } /* * Allocate a cluster to hold the about to be created directory. */ error = clusteralloc(pmp, 0, 1, CLUST_EOFE, &newcluster, NULL); if (error) goto bad2; memset(&ndirent, 0, sizeof(ndirent)); ndirent.de_pmp = pmp; ndirent.de_flag = DE_ACCESS | DE_CREATE | DE_UPDATE; msdosfs_times(&ndirent, &node->inode->st); /* * Now fill the cluster with the "." and ".." entries. And write * the cluster to disk. This way it is there for the parent * directory to be pointing at if there were a crash. */ bn = cntobn(pmp, newcluster); MSDOSFS_DPRINTF(("%s(newcluster %lu, bn=%lu)\n", __func__, newcluster, bn)); /* always succeeds */ bp = getblk(pmp->pm_devvp, bn, pmp->pm_bpcluster, 0, 0, 0); memset(bp->b_data, 0, pmp->pm_bpcluster); memcpy(bp->b_data, &dosdirtemplate, sizeof dosdirtemplate); denp = (struct direntry *)bp->b_data; putushort(denp[0].deStartCluster, newcluster); putushort(denp[0].deCDate, ndirent.de_CDate); putushort(denp[0].deCTime, ndirent.de_CTime); denp[0].deCHundredth = ndirent.de_CHun; putushort(denp[0].deADate, ndirent.de_ADate); putushort(denp[0].deMDate, ndirent.de_MDate); putushort(denp[0].deMTime, ndirent.de_MTime); pcl = pdep->de_StartCluster; MSDOSFS_DPRINTF(("%s(pcl %lu, rootdirblk=%lu)\n", __func__, pcl, pmp->pm_rootdirblk)); if (FAT32(pmp) && pcl == pmp->pm_rootdirblk) pcl = 0; putushort(denp[1].deStartCluster, pcl); putushort(denp[1].deCDate, ndirent.de_CDate); putushort(denp[1].deCTime, ndirent.de_CTime); denp[1].deCHundredth = ndirent.de_CHun; putushort(denp[1].deADate, ndirent.de_ADate); putushort(denp[1].deMDate, ndirent.de_MDate); putushort(denp[1].deMTime, ndirent.de_MTime); if (FAT32(pmp)) { putushort(denp[0].deHighClust, newcluster >> 16); putushort(denp[1].deHighClust, pdep->de_StartCluster >> 16); } else { putushort(denp[0].deHighClust, 0); putushort(denp[1].deHighClust, 0); } if ((error = bwrite(bp)) != 0) goto bad; /* * Now build up a directory entry pointing to the newly allocated * cluster. This will be written to an empty slot in the parent * directory. */ if ((error = uniqdosname(pdep, &cn, ndirent.de_Name)) != 0) goto bad; ndirent.de_Attributes = ATTR_DIRECTORY; ndirent.de_StartCluster = newcluster; ndirent.de_FileSize = 0; ndirent.de_pmp = pdep->de_pmp; if ((error = msdosfs_findslot(pdep, &cn)) != 0) goto bad; if ((error = createde(&ndirent, pdep, &dep, &cn)) != 0) goto bad; if ((error = msdosfs_updatede(dep)) != 0) goto bad; return dep; bad: clusterfree(pmp, newcluster, NULL); bad2: errno = error; return NULL; } Index: projects/clang1100-import/usr.sbin/pwd_mkdb/Makefile =================================================================== --- projects/clang1100-import/usr.sbin/pwd_mkdb/Makefile (revision 364034) +++ projects/clang1100-import/usr.sbin/pwd_mkdb/Makefile (revision 364035) @@ -1,13 +1,16 @@ # @(#)Makefile 8.1 (Berkeley) 6/6/93 # $FreeBSD$ .PATH: ${SRCTOP}/lib/libc/gen # for pw_scan.c PACKAGE= runtime PROG= pwd_mkdb MAN= pwd_mkdb.8 SRCS= pw_scan.c pwd_mkdb.c CFLAGS+= -I${SRCTOP}/lib/libc/gen # for pw_scan.h +.if defined(BOOTSTRAPPING) +CFLAGS+=-I${.CURDIR} +.endif .include Index: projects/clang1100-import/usr.sbin/pwd_mkdb/pwd.h =================================================================== --- projects/clang1100-import/usr.sbin/pwd_mkdb/pwd.h (nonexistent) +++ projects/clang1100-import/usr.sbin/pwd_mkdb/pwd.h (revision 364035) @@ -0,0 +1,66 @@ +/*- + * SPDX-License-Identifier: BSD-2-Clause + * + * Copyright 2018-2020 Alex Richardson + * + * This software was developed by SRI International and the University of + * Cambridge Computer Laboratory (Department of Computer Science and + * Technology) under DARPA contract HR0011-18-C-0016 ("ECATS"), as part of the + * DARPA SSITH research programme. + * + * This software was developed by SRI International and the University of + * Cambridge Computer Laboratory under DARPA/AFRL contract (FA8750-10-C-0237) + * ("CTSRD"), as part of the DARPA CRASH research programme. + * + * 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. + * + * $FreeBSD$ + */ + +/* + * When building pwd_mkdb we need to use target systems definition of + * struct passwd. This protects against future changes to struct passwd and + * is essential to allow cross-building from Linux/macOS hosts since the + * structure is not compatible there. + */ +#include +#include +/* + * Note: pwd_mkdb always stores uint32_t for all integer fields (including + * time_t!) so these definitions do not need to match sys/sys/_types.h + */ +typedef uint32_t _bootstrap_gid_t; +typedef uint32_t _bootstrap_uid_t; +typedef uint64_t _bootstrap_time_t; +#define _GID_T_DECLARED +#define _TIME_T_DECLARED +#define _UID_T_DECLARED +#define _SIZE_T_DECLARED + +#define gid_t _bootstrap_gid_t +#define uid_t _bootstrap_uid_t +#define time_t _bootstrap_time_t +#define passwd _bootstrap_passwd +#include "../../include/pwd.h" +#undef gid_t +#undef uid_t +#undef time_t Property changes on: projects/clang1100-import/usr.sbin/pwd_mkdb/pwd.h ___________________________________________________________________ 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/clang1100-import/usr.sbin/tzsetup/Makefile =================================================================== --- projects/clang1100-import/usr.sbin/tzsetup/Makefile (revision 364034) +++ projects/clang1100-import/usr.sbin/tzsetup/Makefile (revision 364035) @@ -1,16 +1,16 @@ # $FreeBSD$ .include PROG= tzsetup MAN= tzsetup.8 CFLAGS+= -I. -.if ${MK_DIALOG} != no +.if ${MK_DIALOG} != no && !defined(BOOTSTRAPPING) WARNS?= 3 CFLAGS+= -I${SRCTOP}/contrib/dialog -DHAVE_DIALOG LIBADD= dialog ncursesw .endif .include Index: projects/clang1100-import =================================================================== --- projects/clang1100-import (revision 364034) +++ projects/clang1100-import (revision 364035) Property changes on: projects/clang1100-import ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head:r363989-364034