Index: projects/vnet/UPDATING =================================================================== --- projects/vnet/UPDATING (revision 302157) +++ projects/vnet/UPDATING (revision 302158) @@ -1,1614 +1,1622 @@ 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: http://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: FreeBSD has switched from gcc to clang. If you have trouble bootstrapping from older versions of FreeBSD, try WITHOUT_CLANG and WITH_GCC to bootstrap to the tip of head, and then rebuild without this option. The bootstrap process from older version of current across the gcc/clang cutover is a bit fragile. NOTE TO PEOPLE WHO THINK THAT FreeBSD 11.x IS SLOW: FreeBSD 11.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".) +20160622: + The libc stub for the pipe(2) system call has been replaced with + a wrapper which calls the pipe2(2) system call and the pipe(2) is now + only implemented by the kernels which 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 effect 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 kernel 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 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 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. 20131010: The rc.d/jail script has been updated to support jail(8) configuration file. The "jail__*" rc.conf(5) variables for per-jail configuration are automatically converted to /var/run/jail..conf before the jail(8) utility is invoked. This is transparently backward compatible. See below about some incompatibilities and rc.conf(5) manual page for more details. These variables are now deprecated in favor of jail(8) configuration file. One can use "rc.d/jail config " command to generate a jail(8) configuration file in /var/run/jail..conf without running the jail(8) utility. The default pathname of the configuration file is /etc/jail.conf and can be specified by using $jail_conf or $jail__conf variables. Please note that jail_devfs_ruleset accepts an integer at this moment. Please consider to rewrite the ruleset name with an integer. 20130930: BIND has been removed from the base system. If all you need is a local resolver, simply enable and start the local_unbound service instead. Otherwise, several versions of BIND are available in the ports tree. The dns/bind99 port is one example. With this change, nslookup(1) and dig(1) are no longer in the base system. Users should instead use host(1) and drill(1) which are in the base system. Alternatively, nslookup and dig can be obtained by installing the dns/bind-tools port. 20130916: With the addition of unbound(8), a new unbound user is now required during installworld. "mergemaster -p" can be used to add the user prior to installworld, as documented in the handbook. 20130911: OpenSSH is now built with DNSSEC support, and will by default silently trust signed SSHFP records. This can be controlled with the VerifyHostKeyDNS client configuration setting. DNSSEC support can be disabled entirely with the WITHOUT_LDNS option in src.conf. 20130906: The GNU Compiler Collection and C++ standard library (libstdc++) are no longer built by default on platforms where clang is the system compiler. You can enable them with the WITH_GCC and WITH_GNUCXX options in src.conf. 20130905: The PROCDESC kernel option is now part of the GENERIC kernel configuration and is required for the rwhod(8) to work. If you are using custom kernel configuration, you should include 'options PROCDESC'. 20130905: The API and ABI related to the Capsicum framework was modified in backward incompatible way. The userland libraries and programs have to be recompiled to work with the new kernel. This includes the following libraries and programs, but the whole buildworld is advised: libc, libprocstat, dhclient, tcpdump, hastd, hastctl, kdump, procstat, rwho, rwhod, uniq. 20130903: AES-NI intrinsic support has been added to gcc. The AES-NI module has been updated to use this support. A new gcc is required to build the aesni module on both i386 and amd64. 20130821: The PADLOCK_RNG and RDRAND_RNG kernel options are now devices. Thus "device padlock_rng" and "device rdrand_rng" should be used instead of "options PADLOCK_RNG" & "options RDRAND_RNG". 20130813: WITH_ICONV has been split into two feature sets. WITH_ICONV now enables just the iconv* functionality and is now on by default. WITH_LIBICONV_COMPAT enables the libiconv api and link time compatibility. Set WITHOUT_ICONV to build the old way. If you have been using WITH_ICONV before, you will very likely need to turn on WITH_LIBICONV_COMPAT. 20130806: INVARIANTS option now enables DEBUG for code with OpenSolaris and Illumos origin, including ZFS. If you have INVARIANTS in your kernel configuration, then there is no need to set DEBUG or ZFS_DEBUG explicitly. DEBUG used to enable witness(9) tracking of OpenSolaris (mostly ZFS) locks if WITNESS option was set. Because that generated a lot of witness(9) reports and all of them were believed to be false positives, this is no longer done. New option OPENSOLARIS_WITNESS can be used to achieve the previous behavior. 20130806: Timer values in IPv6 data structures now use time_uptime instead of time_second. Although this is not a user-visible functional change, userland utilities which directly use them---ndp(8), rtadvd(8), and rtsold(8) in the base system---need to be updated to r253970 or later. 20130802: find -delete can now delete the pathnames given as arguments, instead of only files found below them or if the pathname did not contain any slashes. Formerly, the following error message would result: find: -delete: : relative path potentially not safe Deleting the pathnames given as arguments can be prevented without error messages using -mindepth 1 or by changing directory and passing "." as argument to find. This works in the old as well as the new version of find. 20130726: Behavior of devfs rules path matching has been changed. Pattern is now always matched against fully qualified devfs path and slash characters must be explicitly matched by slashes in pattern (FNM_PATHNAME). Rulesets involving devfs subdirectories must be reviewed. 20130716: The default ARM ABI has changed to the ARM EABI. The old ABI is incompatible with the ARM EABI and all programs and modules will need to be rebuilt to work with a new kernel. To keep using the old ABI ensure the WITHOUT_ARM_EABI knob is set. NOTE: Support for the old ABI will be removed in the future and users are advised to upgrade. 20130709: pkg_install has been disconnected from the build if you really need it you should add WITH_PKGTOOLS in your src.conf(5). 20130709: Most of network statistics structures were changed to be able keep 64-bits counters. Thus all tools, that work with networking statistics, must be rebuilt (netstat(1), bsnmpd(1), etc.) 20130618: Fix a bug that allowed a tracing process (e.g. gdb) to write to a memory-mapped file in the traced process's address space even if neither the traced process nor the tracing process had write access to that file. 20130615: CVS has been removed from the base system. An exact copy of the code is available from the devel/cvs port. 20130613: Some people report the following error after the switch to bmake: make: illegal option -- J usage: make [-BPSXeiknpqrstv] [-C directory] [-D variable] ... *** [buildworld] Error code 2 this likely due to an old instance of make in ${MAKEPATH} (${MAKEOBJDIRPREFIX}${.CURDIR}/make.${MACHINE}) which src/Makefile will use that blindly, if it exists, so if you see the above error: rm -rf `make -V MAKEPATH` should resolve it. 20130516: Use bmake by default. Whereas before one could choose to build with bmake via -DWITH_BMAKE one must now use -DWITHOUT_BMAKE to use the old make. The goal is to remove these knobs for 10-RELEASE. It is worth noting that bmake (like gmake) treats the command line as the unit of failure, rather than statements within the command line. Thus '(cd some/where && dosomething)' is safer than 'cd some/where; dosomething'. The '()' allows consistent behavior in parallel build. 20130429: Fix a bug that allows NFS clients to issue READDIR on files. 20130426: The WITHOUT_IDEA option has been removed because the IDEA patent expired. 20130426: The sysctl which controls TRIM support under ZFS has been renamed from vfs.zfs.trim_disable -> vfs.zfs.trim.enabled and has been enabled by default. 20130425: The mergemaster command now uses the default MAKEOBJDIRPREFIX rather than creating it's own in the temporary directory in order allow access to bootstrapped versions of tools such as install and mtree. When upgrading from version of FreeBSD where the install command does not support -l, you will need to install a new mergemaster command if mergemaster -p is required. This can be accomplished with the command (cd src/usr.sbin/mergemaster && make install). 20130404: Legacy ATA stack, disabled and replaced by new CAM-based one since FreeBSD 9.0, completely removed from the sources. Kernel modules atadisk and atapi*, user-level tools atacontrol and burncd are removed. Kernel option `options ATA_CAM` is now permanently enabled and removed. 20130319: SOCK_CLOEXEC and SOCK_NONBLOCK flags have been added to socket(2) and socketpair(2). Software, in particular Kerberos, may automatically detect and use these during building. The resulting binaries will not work on older kernels. 20130308: CTL_DISABLE has also been added to the sparc64 GENERIC (for further information, see the respective 20130304 entry). 20130304: Recent commits to callout(9) changed the size of struct callout, so the KBI is probably heavily disturbed. Also, some functions in callout(9)/sleep(9)/sleepqueue(9)/condvar(9) KPIs were replaced by macros. Every kernel module using it won't load, so rebuild is requested. The ctl device has been re-enabled in GENERIC for i386 and amd64, but does not initialize by default (because of the new CTL_DISABLE option) to save memory. To re-enable it, remove the CTL_DISABLE option from the kernel config file or set kern.cam.ctl.disable=0 in /boot/loader.conf. 20130301: The ctl device has been disabled in GENERIC for i386 and amd64. This was done due to the extra memory being allocated at system initialisation time by the ctl driver which was only used if a CAM target device was created. This makes a FreeBSD system unusable on 128MB or less of RAM. 20130208: A new compression method (lz4) has been merged to -HEAD. Please refer to zpool-features(7) for more information. Please refer to the "ZFS notes" section of this file for information on upgrading boot ZFS pools. 20130129: A BSD-licensed patch(1) variant has been added and is installed as bsdpatch, being the GNU version the default patch. To inverse the logic and use the BSD-licensed one as default, while having the GNU version installed as gnupatch, rebuild and install world with the WITH_BSD_PATCH knob set. 20130121: Due to the use of the new -l option to install(1) during build and install, you must take care not to directly set the INSTALL make variable in your /etc/make.conf, /etc/src.conf, or on the command line. If you wish to use the -C flag for all installs you may be able to add INSTALL+=-C to /etc/make.conf or /etc/src.conf. 20130118: The install(1) option -M has changed meaning and now takes an argument that is a file or path to append logs to. In the unlikely event that -M was the last option on the command line and the command line contained at least two files and a target directory the first file will have logs appended to it. The -M option served little practical purpose in the last decade so its use is expected to be extremely rare. 20121223: After switching to Clang as the default compiler some users of ZFS on i386 systems started to experience stack overflow kernel panics. Please consider using 'options KSTACK_PAGES=4' in such configurations. 20121222: GEOM_LABEL now mangles label names read from file system metadata. Mangling affect labels containing spaces, non-printable characters, '%' or '"'. Device names in /etc/fstab and other places may need to be updated. 20121217: By default, only the 10 most recent kernel dumps will be saved. To restore the previous behaviour (no limit on the number of kernel dumps stored in the dump directory) add the following line to /etc/rc.conf: savecore_flags="" 20121201: With the addition of auditdistd(8), a new auditdistd user is now required during installworld. "mergemaster -p" can be used to add the user prior to installworld, as documented in the handbook. 20121117: The sin6_scope_id member variable in struct sockaddr_in6 is now filled by the kernel before passing the structure to the userland via sysctl or routing socket. This means the KAME-specific embedded scope id in sin6_addr.s6_addr[2] is always cleared in userland application. This behavior can be controlled by net.inet6.ip6.deembed_scopeid. __FreeBSD_version is bumped to 1000025. 20121105: On i386 and amd64 systems WITH_CLANG_IS_CC is now the default. This means that the world and kernel will be compiled with clang and that clang will be installed as /usr/bin/cc, /usr/bin/c++, and /usr/bin/cpp. To disable this behavior and revert to building with gcc, compile with WITHOUT_CLANG_IS_CC. Really old versions of current may need to bootstrap WITHOUT_CLANG first if the clang build fails (its compatibility window doesn't extend to the 9 stable branch point). 20121102: The IPFIREWALL_FORWARD kernel option has been removed. Its functionality now turned on by default. 20121023: The ZERO_COPY_SOCKET kernel option has been removed and split into SOCKET_SEND_COW and SOCKET_RECV_PFLIP. NB: SOCKET_SEND_COW uses the VM page based copy-on-write mechanism which is not safe and may result in kernel crashes. NB: The SOCKET_RECV_PFLIP mechanism is useless as no current driver supports disposeable external page sized mbuf storage. Proper replacements for both zero-copy mechanisms are under consideration and will eventually lead to complete removal of the two kernel options. 20121023: The IPv4 network stack has been converted to network byte order. The following modules need to be recompiled together with kernel: carp(4), divert(4), gif(4), siftr(4), gre(4), pf(4), ipfw(4), ng_ipfw(4), stf(4). 20121022: Support for non-MPSAFE filesystems was removed from VFS. The VFS_VERSION was bumped, all filesystem modules shall be recompiled. 20121018: All the non-MPSAFE filesystems have been disconnected from the build. The full list includes: codafs, hpfs, ntfs, nwfs, portalfs, smbfs, xfs. 20121016: The interface cloning API and ABI has changed. The following modules need to be recompiled together with kernel: ipfw(4), pfsync(4), pflog(4), usb(4), wlan(4), stf(4), vlan(4), disc(4), edsc(4), if_bridge(4), gif(4), tap(4), faith(4), epair(4), enc(4), tun(4), if_lagg(4), gre(4). 20121015: The sdhci driver was split in two parts: sdhci (generic SD Host Controller logic) and sdhci_pci (actual hardware driver). No kernel config modifications are required, but if you load sdhc as a module you must switch to sdhci_pci instead. 20121014: Import the FUSE kernel and userland support into base system. 20121013: The GNU sort(1) program has been removed since the BSD-licensed sort(1) has been the default for quite some time and no serious problems have been reported. The corresponding WITH_GNU_SORT knob has also gone. 20121006: The pfil(9) API/ABI for AF_INET family has been changed. Packet filtering modules: pf(4), ipfw(4), ipfilter(4) need to be recompiled with new kernel. 20121001: The net80211(4) ABI has been changed to allow for improved driver PS-POLL and power-save support. All wireless drivers need to be recompiled to work with the new kernel. 20120913: The random(4) support for the VIA hardware random number generator (`PADLOCK') is no longer enabled unconditionally. Add the padlock_rng device in the custom kernel config if needed. The GENERIC kernels on i386 and amd64 do include the device, so the change only affects the custom kernel configurations. 20120908: The pf(4) packet filter ABI has been changed. pfctl(8) and snmp_pf module need to be recompiled to work with new kernel. 20120828: A new ZFS feature flag "com.delphix:empty_bpobj" has been merged to -HEAD. Pools that have empty_bpobj in active state can not be imported read-write with ZFS implementations that do not support this feature. For more information read the zpool-features(5) manual page. 20120727: The sparc64 ZFS loader has been changed to no longer try to auto- detect ZFS providers based on diskN aliases but now requires these to be explicitly listed in the OFW boot-device environment variable. 20120712: The OpenSSL has been upgraded to 1.0.1c. Any binaries requiring libcrypto.so.6 or libssl.so.6 must be recompiled. Also, there are configuration changes. Make sure to merge /etc/ssl/openssl.cnf. 20120712: The following sysctls and tunables have been renamed for consistency with other variables: kern.cam.da.da_send_ordered -> kern.cam.da.send_ordered kern.cam.ada.ada_send_ordered -> kern.cam.ada.send_ordered 20120628: The sort utility has been replaced with BSD sort. For now, GNU sort is also available as "gnusort" or the default can be set back to GNU sort by setting WITH_GNU_SORT. In this case, BSD sort will be installed as "bsdsort". 20120611: A new version of ZFS (pool version 5000) has been merged to -HEAD. Starting with this version the old system of ZFS pool versioning is superseded by "feature flags". This concept enables forward compatibility against certain future changes in functionality of ZFS pools. The first read-only compatible "feature flag" for ZFS pools is named "com.delphix:async_destroy". For more information read the new zpool-features(5) manual page. Please refer to the "ZFS notes" section of this file for information on upgrading boot ZFS pools. 20120417: The malloc(3) implementation embedded in libc now uses sources imported as contrib/jemalloc. The most disruptive API change is to /etc/malloc.conf. If your system has an old-style /etc/malloc.conf, delete it prior to installworld, and optionally re-create it using the new format after rebooting. See malloc.conf(5) for details (specifically the TUNING section and the "opt.*" entries in the MALLCTL NAMESPACE section). 20120328: Big-endian MIPS TARGET_ARCH values no longer end in "eb". mips64eb is now spelled mips64. mipsn32eb is now spelled mipsn32. mipseb is now spelled mips. This is to aid compatibility with third-party software that expects this naming scheme in uname(3). Little-endian settings are unchanged. If you are updating a big-endian mips64 machine from before this change, you may need to set MACHINE_ARCH=mips64 in your environment before the new build system will recognize your machine. 20120306: Disable by default the option VFS_ALLOW_NONMPSAFE for all supported platforms. 20120229: Now unix domain sockets behave "as expected" on nullfs(5). Previously nullfs(5) did not pass through all behaviours to the underlying layer, as a result if we bound to a socket on the lower layer we could connect only to the lower path; if we bound to the upper layer we could connect only to the upper path. The new behavior is one can connect to both the lower and the upper paths regardless what layer path one binds to. 20120211: The getifaddrs upgrade path broken with 20111215 has been restored. If you have upgraded in between 20111215 and 20120209 you need to recompile libc again with your kernel. You still need to recompile world to be able to configure CARP but this restriction already comes from 20111215. 20120114: The set_rcvar() function has been removed from /etc/rc.subr. All base and ports rc.d scripts have been updated, so if you have a port installed with a script in /usr/local/etc/rc.d you can either hand-edit the rcvar= line, or reinstall the port. An easy way to handle the mass-update of /etc/rc.d: rm /etc/rc.d/* && mergemaster -i 20120109: panic(9) now stops other CPUs in the SMP systems, disables interrupts on the current CPU and prevents other threads from running. This behavior can be reverted using the kern.stop_scheduler_on_panic tunable/sysctl. The new behavior can be incompatible with kern.sync_on_panic. 20111215: The carp(4) facility has been changed significantly. Configuration of the CARP protocol via ifconfig(8) has changed, as well as format of CARP events submitted to devd(8) has changed. See manual pages for more information. The arpbalance feature of carp(4) is currently not supported anymore. Size of struct in_aliasreq, struct in6_aliasreq has changed. User utilities using SIOCAIFADDR, SIOCAIFADDR_IN6, e.g. ifconfig(8), need to be recompiled. 20111122: The acpi_wmi(4) status device /dev/wmistat has been renamed to /dev/wmistat0. 20111108: The option VFS_ALLOW_NONMPSAFE option has been added in order to explicitely support non-MPSAFE filesystems. It is on by default for all supported platform at this present time. 20111101: The broken amd(4) driver has been replaced with esp(4) in the amd64, i386 and pc98 GENERIC kernel configuration files. 20110930: sysinstall has been removed 20110923: The stable/9 branch created in subversion. This corresponds to the RELENG_9 branch in CVS. COMMON ITEMS: General Notes ------------- Avoid using make -j when upgrading. While generally safe, there are sometimes problems using -j to upgrade. If your upgrade fails with -j, please try again without -j. From time to time in the past there have been problems using -j with buildworld and/or installworld. This is especially true when upgrading between "distant" versions (eg one that cross a major release boundary or several minor releases, or when several months have passed on the -current branch). 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. 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 before reporting problems with a major version upgrade. 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 just build a kernel when you know that it won't mess you up -------------------------------------------------------------- This assumes you are already running a CURRENT system. Replace ${arch} with the architecture of your machine (e.g. "i386", "arm", "amd64", "ia64", "pc98", "sparc64", "powerpc", "mips", etc). cd src/sys/${arch}/conf config KERNEL_NAME_HERE cd ../compile/KERNEL_NAME_HERE make depend make make install If this fails, go to the "To build a kernel" section. 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 kernel 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 kernel KERNCONF=YOUR_KERNEL_HERE [8] [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 cd src 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 noop. 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] In order to have a kernel that can run the 4.x binaries needed to do an installworld, you must include the COMPAT_FREEBSD4 option in your kernel. Failure to do so may leave you with a system that is hard to boot to recover. A similar kernel option COMPAT_FREEBSD5 is required to run the 5.x binaries on more recent kernels. And so on for COMPAT_FREEBSD6 and COMPAT_FREEBSD7. Make sure that you merge any new devices from GENERIC since the last time you updated your kernel config file. [9] When checking out sources, you must include the -P flag to have cvs prune empty directories. 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. All Rights Reserved. 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/vnet/cddl/lib/libavl/Makefile =================================================================== --- projects/vnet/cddl/lib/libavl/Makefile (revision 302157) +++ projects/vnet/cddl/lib/libavl/Makefile (revision 302158) @@ -1,11 +1,11 @@ # $FreeBSD$ .PATH: ${.CURDIR}/../../../sys/cddl/contrib/opensolaris/common/avl LIB= avl SRCS= avl.c -WARNS?= 0 +WARNS?= 3 CFLAGS+= -I${.CURDIR}/../../../sys/cddl/compat/opensolaris CFLAGS+= -I${.CURDIR}/../../../sys/cddl/contrib/opensolaris/uts/common .include Index: projects/vnet/cddl/lib/libctf/Makefile =================================================================== --- projects/vnet/cddl/lib/libctf/Makefile (revision 302157) +++ projects/vnet/cddl/lib/libctf/Makefile (revision 302158) @@ -1,33 +1,33 @@ # $FreeBSD$ .PATH: ${.CURDIR}/../../../cddl/contrib/opensolaris/common/ctf .PATH: ${.CURDIR}/../../../cddl/contrib/opensolaris/lib/libctf/common .PATH: ${.CURDIR}/../../../sys/cddl/contrib/opensolaris/common/ctf LIB= ctf SRCS= ctf_create.c \ ctf_decl.c \ ctf_error.c \ ctf_hash.c \ ctf_labels.c \ ctf_lib.c \ ctf_lookup.c \ ctf_open.c \ ctf_subr.c \ ctf_types.c \ ctf_util.c MAN= ctf.5 -WARNS?= 0 +WARNS?= 2 CFLAGS+= -DCTF_OLD_VERSIONS CFLAGS+= -I${.CURDIR}/../../../sys/cddl/compat/opensolaris \ -I${.CURDIR}/../../../cddl/compat/opensolaris/include \ -I${OPENSOLARIS_USR_DISTDIR}/head \ -I${OPENSOLARIS_USR_DISTDIR}/common/ctf \ -I${OPENSOLARIS_USR_DISTDIR}/lib/libctf/common \ -I${OPENSOLARIS_SYS_DISTDIR}/uts/common LIBADD+= z .include Index: projects/vnet/cddl/lib/libnvpair/Makefile =================================================================== --- projects/vnet/cddl/lib/libnvpair/Makefile (revision 302157) +++ projects/vnet/cddl/lib/libnvpair/Makefile (revision 302158) @@ -1,34 +1,34 @@ # $FreeBSD$ .PATH: ${.CURDIR}/../../../cddl/contrib/opensolaris/lib/libnvpair .PATH: ${.CURDIR}/../../../sys/cddl/contrib/opensolaris/common/nvpair LIB= nvpair SRCS= libnvpair.c \ nvpair_alloc_system.c \ nvpair_json.c \ opensolaris_fnvpair.c \ opensolaris_nvpair.c \ opensolaris_nvpair_alloc_fixed.c -WARNS?= 0 +WARNS?= 1 CFLAGS+= -I${.CURDIR}/../../../cddl/compat/opensolaris/include CFLAGS+= -I${.CURDIR}/../../../cddl/contrib/opensolaris/lib/libzpool/common CFLAGS+= -I${.CURDIR}/../../../sys/cddl/compat/opensolaris CFLAGS+= -I${.CURDIR}/../../../sys/cddl/contrib/opensolaris/uts/common CFLAGS+= -I${.CURDIR}/../../../sys/cddl/contrib/opensolaris/uts/common/fs/zfs CFLAGS+= -I${.CURDIR}/../../../sys CFLAGS+= -I${.CURDIR}/../../../cddl/contrib/opensolaris/head CFLAGS+= -I${.CURDIR}/../../../cddl/compat/opensolaris/lib/libumem # This library uses macros to define fprintf behavior for several object types # The compiler will see the non-string literal arguments to the fprintf calls and # omit warnings for them. Quiesce these warnings in contrib code: # # cddl/contrib/opensolaris/lib/libnvpair/libnvpair.c:743:12: warning: format # string is not a string literal (potentially insecure) [-Wformat-security] # ARENDER(pctl, nvlist_array, nvl, name, val, nelem); # CFLAGS+= -Wno-format-security .include Index: projects/vnet/cddl/lib/libumem/Makefile =================================================================== --- projects/vnet/cddl/lib/libumem/Makefile (revision 302157) +++ projects/vnet/cddl/lib/libumem/Makefile (revision 302158) @@ -1,10 +1,10 @@ # $FreeBSD$ .PATH: ${.CURDIR}/../../../cddl/compat/opensolaris/lib/libumem LIB= umem SRCS= umem.c -WARNS?= 0 +WARNS?= 3 CFLAGS+= -I${.CURDIR}/../../../cddl/compat/opensolaris/lib/libumem .include Index: projects/vnet/cddl/lib/libuutil/Makefile =================================================================== --- projects/vnet/cddl/lib/libuutil/Makefile (revision 302157) +++ projects/vnet/cddl/lib/libuutil/Makefile (revision 302158) @@ -1,26 +1,26 @@ # $FreeBSD$ .PATH: ${.CURDIR}/../../../cddl/contrib/opensolaris/lib/libuutil/common .PATH: ${.CURDIR}/../../../sys/cddl/contrib/opensolaris/common/avl LIB= uutil SRCS= avl.c \ uu_alloc.c \ uu_avl.c \ uu_dprintf.c \ uu_ident.c \ uu_list.c \ uu_misc.c \ uu_open.c \ uu_pname.c \ uu_strtoint.c -WARNS?= 0 +WARNS?= 1 CFLAGS+= -DNATIVE_BUILD CFLAGS+= -I${.CURDIR}/../../../cddl/contrib/opensolaris/lib/libuutil/common CFLAGS+= -I${.CURDIR}/../../../sys/cddl/compat/opensolaris CFLAGS+= -I${.CURDIR}/../../../sys/cddl/contrib/opensolaris/uts/common CFLAGS+= -I${.CURDIR}/../../../cddl/compat/opensolaris/include CFLAGS+= -I${.CURDIR}/../../../cddl/contrib/opensolaris/head .include Index: projects/vnet/cddl =================================================================== --- projects/vnet/cddl (revision 302157) +++ projects/vnet/cddl (revision 302158) Property changes on: projects/vnet/cddl ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/cddl:r301116-302157 Index: projects/vnet/etc/defaults/rc.conf =================================================================== --- projects/vnet/etc/defaults/rc.conf (revision 302157) +++ projects/vnet/etc/defaults/rc.conf (revision 302158) @@ -1,743 +1,749 @@ #!/bin/sh # This is rc.conf - a file full of useful variables that you can set # to change the default startup behavior of your system. You should # not edit this file! Put any overrides into one of the ${rc_conf_files} # instead and you will be able to update these defaults later without # spamming your local configuration information. # # The ${rc_conf_files} files should only contain values which override # values set in this file. This eases the upgrade path when defaults # are changed and new features are added. # # All arguments must be in double or single quotes. # # For a more detailed explanation of all the rc.conf variables, please # refer to the rc.conf(5) manual page. # # $FreeBSD$ ############################################################## ### Important initial Boot-time options #################### ############################################################## #rc_debug="NO" # Set to YES to enable debugging output from rc.d rc_info="NO" # Enables display of informational messages at boot. rc_startmsgs="YES" # Show "Starting foo:" messages at boot rcshutdown_timeout="90" # Seconds to wait before terminating rc.shutdown early_late_divider="FILESYSTEMS" # Script that separates early/late # stages of the boot process. Make sure you know # the ramifications if you change this. # See rc.conf(5) for more details. always_force_depends="NO" # Set to check that indicated dependencies are # running during boot (can increase boot time). apm_enable="NO" # Set to YES to enable APM BIOS functions (or NO). apmd_enable="NO" # Run apmd to handle APM event from userland. apmd_flags="" # Flags to apmd (if enabled). ddb_enable="NO" # Set to YES to load ddb scripts at boot. ddb_config="/etc/ddb.conf" # ddb(8) config file. devd_enable="YES" # Run devd, to trigger programs on device tree changes. devd_flags="" # Additional flags for devd(8). #kld_list="" # Kernel modules to load after local disks are mounted kldxref_enable="NO" # Build linker.hints files with kldxref(8). kldxref_clobber="NO" # Overwrite old linker.hints at boot. kldxref_module_path="" # Override kern.module_path. A ';'-delimited list. powerd_enable="NO" # Run powerd to lower our power usage. powerd_flags="" # Flags to powerd (if enabled). tmpmfs="AUTO" # Set to YES to always create an mfs /tmp, NO to never tmpsize="20m" # Size of mfs /tmp if created tmpmfs_flags="-S" # Extra mdmfs options for the mfs /tmp varmfs="AUTO" # Set to YES to always create an mfs /var, NO to never varsize="32m" # Size of mfs /var if created varmfs_flags="-S" # Extra mount options for the mfs /var populate_var="AUTO" # Set to YES to always (re)populate /var, NO to never cleanvar_enable="YES" # Clean the /var directory local_startup="/usr/local/etc/rc.d" # startup script dirs. script_name_sep=" " # Change if your startup scripts' names contain spaces rc_conf_files="/etc/rc.conf /etc/rc.conf.local" # ZFS support zfs_enable="NO" # Set to YES to automatically mount ZFS file systems # ZFSD support zfsd_enable="NO" # Set to YES to automatically start the ZFS fault # management daemon. gptboot_enable="YES" # GPT boot success/failure reporting. # Experimental - test before enabling gbde_autoattach_all="NO" # YES automatically mounts gbde devices from fstab gbde_devices="NO" # Devices to automatically attach (list, or AUTO) gbde_attach_attempts="3" # Number of times to attempt attaching gbde devices gbde_lockdir="/etc" # Where to look for gbde lockfiles # GELI disk encryption configuration. geli_devices="" # List of devices to automatically attach in addition to # GELI devices listed in /etc/fstab. geli_tries="" # Number of times to attempt attaching geli device. # If empty, kern.geom.eli.tries will be used. geli_default_flags="" # Default flags for geli(8). geli_autodetach="YES" # Automatically detach on last close. # Providers are marked as such when all file systems are # mounted. # Example use. #geli_devices="da1 mirror/home" #geli_da1_flags="-p -k /etc/geli/da1.keys" #geli_da1_autodetach="NO" #geli_mirror_home_flags="-k /etc/geli/home.keys" root_rw_mount="YES" # Set to NO to inhibit remounting root read-write. root_hold_delay="30" # Time to wait for root mount hold release. fsck_y_enable="NO" # Set to YES to do fsck -y if the initial preen fails. fsck_y_flags="" # Additional flags for fsck -y background_fsck="YES" # Attempt to run fsck in the background where possible. background_fsck_delay="60" # Time to wait (seconds) before starting the fsck. netfs_types="nfs:NFS smbfs:SMB" # Net filesystems. extra_netfs_types="NO" # List of network extra filesystem types for delayed # mount at startup (or NO). ############################################################## ### Network configuration sub-section ###################### ############################################################## ### Basic network and firewall/security options: ### hostname="" # Set this! hostid_enable="YES" # Set host UUID. hostid_file="/etc/hostid" # File with hostuuid. nisdomainname="NO" # Set to NIS domain if using NIS (or NO). dhclient_program="/sbin/dhclient" # Path to dhcp client program. dhclient_flags="" # Extra flags to pass to dhcp client. #dhclient_flags_fxp0="" # Extra dhclient flags for fxp0 only background_dhclient="NO" # Start dhcp client in the background. #background_dhclient_fxp0="YES" # Start dhcp client on fxp0 in the background. synchronous_dhclient="NO" # Start dhclient directly on configured # interfaces during startup. defaultroute_delay="30" # Time to wait for a default route on a DHCP interface. defaultroute_carrier_delay="5" # Time to wait for carrier while waiting for a default route. netif_enable="YES" # Set to YES to initialize network interfaces netif_ipexpand_max="2048" # Maximum number of IP addrs in a range spec. wpa_supplicant_program="/usr/sbin/wpa_supplicant" wpa_supplicant_flags="-s" # Extra flags to pass to wpa_supplicant wpa_supplicant_conf_file="/etc/wpa_supplicant.conf" # firewall_enable="NO" # Set to YES to enable firewall functionality firewall_script="/etc/rc.firewall" # Which script to run to set up the firewall firewall_type="UNKNOWN" # Firewall type (see /etc/rc.firewall) firewall_quiet="NO" # Set to YES to suppress rule display firewall_logging="NO" # Set to YES to enable events logging firewall_logif="NO" # Set to YES to create logging-pseudo interface firewall_flags="" # Flags passed to ipfw when type is a file firewall_coscripts="" # List of executables/scripts to run after # firewall starts/stops firewall_client_net="192.0.2.0/24" # IPv4 Network address for "client" # firewall. #firewall_client_net_ipv6="2001:db8:2:1::/64" # IPv6 network prefix for # "client" firewall. firewall_simple_iif="ed1" # Inside network interface for "simple" # firewall. firewall_simple_inet="192.0.2.16/28" # Inside network address for "simple" # firewall. firewall_simple_oif="ed0" # Outside network interface for "simple" # firewall. firewall_simple_onet="192.0.2.0/28" # Outside network address for "simple" # firewall. #firewall_simple_iif_ipv6="ed1" # Inside IPv6 network interface for "simple" # firewall. #firewall_simple_inet_ipv6="2001:db8:2:800::/56" # Inside IPv6 network prefix # for "simple" firewall. #firewall_simple_oif_ipv6="ed0" # Outside IPv6 network interface for "simple" # firewall. #firewall_simple_onet_ipv6="2001:db8:2:0::/56" # Outside IPv6 network prefix # for "simple" firewall. firewall_myservices="" # List of TCP ports on which this host # offers services for "workstation" firewall. firewall_allowservices="" # List of IPs which have access to # $firewall_myservices for "workstation" # firewall. firewall_trusted="" # List of IPs which have full access to this # host for "workstation" firewall. firewall_logdeny="NO" # Set to YES to log default denied incoming # packets for "workstation" firewall. firewall_nologports="135-139,445 1026,1027 1433,1434" # List of TCP/UDP ports # for which denied incoming packets are not # logged for "workstation" firewall. firewall_nat_enable="NO" # Enable kernel NAT (if firewall_enable == YES) firewall_nat_interface="" # Public interface or IPaddress to use firewall_nat_flags="" # Additional configuration parameters dummynet_enable="NO" # Load the dummynet(4) module ip_portrange_first="NO" # Set first dynamically allocated port ip_portrange_last="NO" # Set last dynamically allocated port ike_enable="NO" # Enable IKE daemon (usually racoon or isakmpd) ike_program="/usr/local/sbin/isakmpd" # Path to IKE daemon ike_flags="" # Additional flags for IKE daemon ipsec_enable="NO" # Set to YES to run setkey on ipsec_file ipsec_file="/etc/ipsec.conf" # Name of config file for setkey natd_program="/sbin/natd" # path to natd, if you want a different one. natd_enable="NO" # Enable natd (if firewall_enable == YES). natd_interface="" # Public interface or IPaddress to use. natd_flags="" # Additional flags for natd. ipfilter_enable="NO" # Set to YES to enable ipfilter functionality ipfilter_program="/sbin/ipf" # where the ipfilter program lives ipfilter_rules="/etc/ipf.rules" # rules definition file for ipfilter, see # /usr/src/contrib/ipfilter/rules for examples ipfilter_flags="" # additional flags for ipfilter ipnat_enable="NO" # Set to YES to enable ipnat functionality ipnat_program="/sbin/ipnat" # where the ipnat program lives ipnat_rules="/etc/ipnat.rules" # rules definition file for ipnat ipnat_flags="" # additional flags for ipnat ipmon_enable="NO" # Set to YES for ipmon; needs ipfilter or ipnat ipmon_program="/sbin/ipmon" # where the ipfilter monitor program lives ipmon_flags="-Ds" # typically "-Ds" or "-D /var/log/ipflog" ipfs_enable="NO" # Set to YES to enable saving and restoring # of state tables at shutdown and boot ipfs_program="/sbin/ipfs" # where the ipfs program lives ipfs_flags="" # additional flags for ipfs pf_enable="NO" # Set to YES to enable packet filter (pf) pf_rules="/etc/pf.conf" # rules definition file for pf pf_program="/sbin/pfctl" # where the pfctl program lives pf_flags="" # additional flags for pfctl pflog_enable="NO" # Set to YES to enable packet filter logging pflog_logfile="/var/log/pflog" # where pflogd should store the logfile pflog_program="/sbin/pflogd" # where the pflogd program lives pflog_flags="" # additional flags for pflogd ftpproxy_enable="NO" # Set to YES to enable ftp-proxy(8) for pf ftpproxy_flags="" # additional flags for ftp-proxy(8) pfsync_enable="NO" # Expose pf state to other hosts for syncing pfsync_syncdev="" # Interface for pfsync to work through pfsync_syncpeer="" # IP address of pfsync peer host pfsync_ifconfig="" # Additional options to ifconfig(8) for pfsync tcp_extensions="YES" # Set to NO to turn off RFC1323 extensions. log_in_vain="0" # >=1 to log connects to ports w/o listeners. tcp_keepalive="YES" # Enable stale TCP connection timeout (or NO). tcp_drop_synfin="NO" # Set to YES to drop TCP packets with SYN+FIN # NOTE: this violates the TCP specification icmp_drop_redirect="NO" # Set to YES to ignore ICMP REDIRECT packets icmp_log_redirect="NO" # Set to YES to log ICMP REDIRECT packets network_interfaces="auto" # List of network interfaces (or "auto"). cloned_interfaces="" # List of cloned network interfaces to create. #cloned_interfaces="gif0 gif1 gif2 gif3" # Pre-cloning GENERIC config. #ifconfig_lo0="inet 127.0.0.1" # default loopback device configuration. #ifconfig_lo0_alias0="inet 127.0.0.254 netmask 0xffffffff" # Sample alias entry. #ifconfig_ed0_ipv6="inet6 2001:db8:1::1 prefixlen 64" # Sample IPv6 addr entry #ifconfig_ed0_alias0="inet6 2001:db8:2::1 prefixlen 64" # Sample IPv6 alias #ifconfig_fxp0_name="net0" # Change interface name from fxp0 to net0. #vlans_fxp0="101 vlan0" # vlan(4) interfaces for fxp0 device #create_args_vlan0="vlan 102" # vlan tag for vlan0 device #wlans_ath0="wlan0" # wlan(4) interfaces for ath0 device #wlandebug_wlan0="scan+auth+assoc" # Set debug flags with wlandebug(8) #ipv4_addrs_fxp0="192.168.0.1/24 192.168.1.1-5/28" # example IPv4 address entry. # #autobridge_interfaces="bridge0" # List of bridges to check #autobridge_bridge0="tap* vlan0" # Interface glob to automatically add to the bridge # # If you have any sppp(4) interfaces above, you might also want to set # the following parameters. Refer to spppcontrol(8) for their meaning. sppp_interfaces="" # List of sppp interfaces. #sppp_interfaces="...0" # example: sppp over ... #spppconfig_...0="authproto=chap myauthname=foo myauthsecret='top secret' hisauthname=some-gw hisauthsecret='another secret'" # User ppp configuration. ppp_enable="NO" # Start user-ppp (or NO). ppp_program="/usr/sbin/ppp" # Path to user-ppp program. ppp_mode="auto" # Choice of "auto", "ddial", "direct" or "dedicated". # For details see man page for ppp(8). Default is auto. ppp_nat="YES" # Use PPP's internal network address translation or NO. ppp_profile="papchap" # Which profile to use from /etc/ppp/ppp.conf. ppp_user="root" # Which user to run ppp as # Start multiple instances of ppp at boot time #ppp_profile="profile1 profile2 profile3" # Which profiles to use #ppp_profile1_mode="ddial" # Override ppp mode for profile1 #ppp_profile2_nat="NO" # Override nat mode for profile2 # profile3 uses default ppp_mode and ppp_nat ### Network daemon (miscellaneous) ### hostapd_enable="NO" # Run hostap daemon. syslogd_enable="YES" # Run syslog daemon (or NO). syslogd_program="/usr/sbin/syslogd" # path to syslogd, if you want a different one. syslogd_flags="-s" # Flags to syslogd (if enabled). syslogd_oomprotect="YES" # Don't kill syslogd when swap space is exhausted. altlog_proglist="" # List of chrooted applicatioins in /var inetd_enable="NO" # Run the network daemon dispatcher (YES/NO). inetd_program="/usr/sbin/inetd" # path to inetd, if you want a different one. inetd_flags="-wW -C 60" # Optional flags to inetd iscsid_enable="NO" # iSCSI initiator daemon. iscsictl_enable="NO" # iSCSI initiator autostart. iscsictl_flags="-Aa" # Optional flags to iscsictl. hastd_enable="NO" # Run the HAST daemon (YES/NO). hastd_program="/sbin/hastd" # path to hastd, if you want a different one. hastd_flags="" # Optional flags to hastd. ctld_enable="NO" # CAM Target Layer / iSCSI target daemon. local_unbound_enable="NO" # local caching resolver blacklistd_enable="NO" # Run blacklistd daemon (YES/NO). blacklistd_flags="" # Optional flags for blacklistd(8). # # kerberos. Do not run the admin daemons on slave servers # kdc_enable="NO" # Run a kerberos 5 KDC (or NO). kdc_program="/usr/libexec/kdc" # path to kerberos 5 KDC kdc_flags="" # Additional flags to the kerberos 5 KDC kadmind_enable="NO" # Run kadmind (or NO) kadmind_program="/usr/libexec/kadmind" # path to kadmind kpasswdd_enable="NO" # Run kpasswdd (or NO) kpasswdd_program="/usr/libexec/kpasswdd" # path to kpasswdd kfd_enable="NO" # Run kfd (or NO) kfd_program="/usr/libexec/kfd" # path to kerberos 5 kfd daemon kfd_flags="" ipropd_master_enable="NO" # Run Heimdal incremental propagation daemon # (master daemon). ipropd_master_program="/usr/libexec/ipropd-master" ipropd_master_flags="" # Flags to ipropd-master. ipropd_master_keytab="/etc/krb5.keytab" # keytab for ipropd-master. ipropd_master_slaves="" # slave node names used for /var/heimdal/slaves. ipropd_slave_enable="NO" # Run Heimdal incremental propagation daemon # (slave daemon). ipropd_slave_program="/usr/libexec/ipropd-slave" ipropd_slave_flags="" # Flags to ipropd-slave. ipropd_slave_keytab="/etc/krb5.keytab" # keytab for ipropd-slave. ipropd_slave_master="" # master node name. gssd_enable="NO" # Run the gssd daemon (or NO). gssd_program="/usr/sbin/gssd" # Path to gssd. gssd_flags="" # Flags for gssd. rwhod_enable="NO" # Run the rwho daemon (or NO). rwhod_flags="" # Flags for rwhod rarpd_enable="NO" # Run rarpd (or NO). rarpd_flags="-a" # Flags to rarpd. bootparamd_enable="NO" # Run bootparamd (or NO). bootparamd_flags="" # Flags to bootparamd pppoed_enable="NO" # Run the PPP over Ethernet daemon. pppoed_provider="*" # Provider and ppp(8) config file entry. pppoed_flags="-P /var/run/pppoed.pid" # Flags to pppoed (if enabled). pppoed_interface="fxp0" # The interface that pppoed runs on. sshd_enable="NO" # Enable sshd sshd_program="/usr/sbin/sshd" # path to sshd, if you want a different one. sshd_flags="" # Additional flags for sshd. ftpd_enable="NO" # Enable stand-alone ftpd. ftpd_program="/usr/libexec/ftpd" # Path to ftpd, if you want a different one. ftpd_flags="" # Additional flags to stand-alone ftpd. ### Network daemon (NFS): All need rpcbind_enable="YES" ### amd_enable="NO" # Run amd service with $amd_flags (or NO). amd_program="/usr/sbin/amd" # path to amd, if you want a different one. amd_flags="-a /.amd_mnt -l syslog /host /etc/amd.map /net /etc/amd.map" amd_map_program="NO" # Can be set to "ypcat -k amd.master" autofs_enable="NO" # Run autofs daemons. automount_flags="" # Flags to automount(8) (if autofs enabled). automountd_flags="" # Flags to automountd(8) (if autofs enabled). autounmountd_flags="" # Flags to autounmountd(8) (if autofs enabled). nfs_client_enable="NO" # This host is an NFS client (or NO). nfs_access_cache="60" # Client cache timeout in seconds nfs_server_enable="NO" # This host is an NFS server (or NO). nfs_server_flags="-u -t" # Flags to nfsd (if enabled). nfs_server_managegids="NO" # The NFS server maps gids for AUTH_SYS (or NO). mountd_enable="NO" # Run mountd (or NO). mountd_flags="-r -S" # Flags to mountd (if NFS server enabled). weak_mountd_authentication="NO" # Allow non-root mount requests to be served. nfs_reserved_port_only="NO" # Provide NFS only on secure port (or NO). nfs_bufpackets="" # bufspace (in packets) for client rpc_lockd_enable="NO" # Run NFS rpc.lockd needed for client/server. rpc_lockd_flags="" # Flags to rpc.lockd (if enabled). rpc_statd_enable="NO" # Run NFS rpc.statd needed for client/server. rpc_statd_flags="" # Flags to rpc.statd (if enabled). rpcbind_enable="NO" # Run the portmapper service (YES/NO). rpcbind_program="/usr/sbin/rpcbind" # path to rpcbind, if you want a different one. rpcbind_flags="" # Flags to rpcbind (if enabled). rpc_ypupdated_enable="NO" # Run if NIS master and SecureRPC (or NO). keyserv_enable="NO" # Run the SecureRPC keyserver (or NO). keyserv_flags="" # Flags to keyserv (if enabled). nfsv4_server_enable="NO" # Enable support for NFSv4 nfscbd_enable="NO" # NFSv4 client side callback daemon nfscbd_flags="" # Flags for nfscbd nfsuserd_enable="NO" # NFSv4 user/group name mapping daemon nfsuserd_flags="" # Flags for nfsuserd ### Network Time Services options: ### timed_enable="NO" # Run the time daemon (or NO). timed_flags="" # Flags to timed (if enabled). ntpdate_enable="NO" # Run ntpdate to sync time on boot (or NO). ntpdate_program="/usr/sbin/ntpdate" # path to ntpdate, if you want a different one. ntpdate_flags="-b" # Flags to ntpdate (if enabled). ntpdate_config="/etc/ntp.conf" # ntpdate(8) configuration file ntpdate_hosts="" # Whitespace-separated list of ntpdate(8) servers. ntpd_enable="NO" # Run ntpd Network Time Protocol (or NO). ntpd_program="/usr/sbin/ntpd" # path to ntpd, if you want a different one. ntpd_config="/etc/ntp.conf" # ntpd(8) configuration file ntpd_sync_on_start="NO" # Sync time on ntpd startup, even if offset is high ntpd_flags="-p /var/run/ntpd.pid -f /var/db/ntpd.drift" # Flags to ntpd (if enabled). ntp_src_leapfile="/etc/ntp/leap-seconds" # Initial source for ntpd leapfile ntp_db_leapfile="/var/db/ntpd.leap-seconds.list" # Working copy (updated weekly) leapfile ntp_leapfile_sources="https://www.ietf.org/timezones/data/leap-seconds.list" # Source from which to fetch leapfile ntp_leapfile_fetch_opts="-mq" # Options to use for ntp leapfile fetch, # e.g. --no-verify-peer ntp_leapfile_expiry_days=30 # Check for new leapfile 30 days prior to # expiry. ntp_leapfile_fetch_verbose="NO" # Be verbose during NTP leapfile fetch # Network Information Services (NIS) options: All need rpcbind_enable="YES" ### nis_client_enable="NO" # We're an NIS client (or NO). nis_client_flags="" # Flags to ypbind (if enabled). nis_ypset_enable="NO" # Run ypset at boot time (or NO). nis_ypset_flags="" # Flags to ypset (if enabled). nis_server_enable="NO" # We're an NIS server (or NO). nis_server_flags="" # Flags to ypserv (if enabled). nis_ypxfrd_enable="NO" # Run rpc.ypxfrd at boot time (or NO). nis_ypxfrd_flags="" # Flags to rpc.ypxfrd (if enabled). nis_yppasswdd_enable="NO" # Run rpc.yppasswdd at boot time (or NO). nis_yppasswdd_flags="" # Flags to rpc.yppasswdd (if enabled). nis_ypldap_enable="NO" # Run ypldap at boot time (or NO). nis_ypldap_flags="" # Flags to ypldap (if enabled). ### SNMP daemon ### # Be sure to understand the security implications of running SNMP v1/v2 # in your network. bsnmpd_enable="NO" # Run the SNMP daemon (or NO). bsnmpd_flags="" # Flags for bsnmpd. ### Network routing options: ### defaultrouter="NO" # Set to default gateway (or NO). static_arp_pairs="" # Set to static ARP list (or leave empty). static_ndp_pairs="" # Set to static NDP list (or leave empty). static_routes="" # Set to static route list (or leave empty). natm_static_routes="" # Set to static route list for NATM (or leave empty). gateway_enable="NO" # Set to YES if this host will be a gateway. routed_enable="NO" # Set to YES to enable a routing daemon. routed_program="/sbin/routed" # Name of routing daemon to use if enabled. routed_flags="-q" # Flags for routing daemon. arpproxy_all="NO" # replaces obsolete kernel option ARP_PROXYALL. forward_sourceroute="NO" # do source routing (only if gateway_enable is set to "YES") accept_sourceroute="NO" # accept source routed packets to us ### ATM interface options: ### atm_enable="NO" # Configure ATM interfaces (or NO). #atm_netif_hea0="atm 1" # Network interfaces for physical interface. #atm_sigmgr_hea0="uni31" # Signalling manager for physical interface. #atm_prefix_hea0="ILMI" # NSAP prefix (UNI interfaces only) (or ILMI). #atm_macaddr_hea0="NO" # Override physical MAC address (or NO). #atm_arpserver_atm0="0x47.0005.80.999999.9999.9999.9999.999999999999.00" # ATMARP server address (or local). #atm_scsparp_atm0="NO" # Run SCSP/ATMARP on network interface (or NO). atm_pvcs="" # Set to PVC list (or leave empty). atm_arps="" # Set to permanent ARP list (or leave empty). ### Bluetooth ### hcsecd_enable="NO" # Enable hcsecd(8) (or NO) hcsecd_config="/etc/bluetooth/hcsecd.conf" # hcsecd(8) configuration file sdpd_enable="NO" # Enable sdpd(8) (or NO) sdpd_control="/var/run/sdp" # sdpd(8) control socket sdpd_groupname="nobody" # set spdp(8) user/group to run as after sdpd_username="nobody" # it initializes bthidd_enable="NO" # Enable bthidd(8) (or NO) bthidd_config="/etc/bluetooth/bthidd.conf" # bthidd(8) configuration file bthidd_hids="/var/db/bthidd.hids" # bthidd(8) known HID devices file rfcomm_pppd_server_enable="NO" # Enable rfcomm_pppd(8) in server mode (or NO) rfcomm_pppd_server_profile="one two" # Profile to use from /etc/ppp/ppp.conf # #rfcomm_pppd_server_one_bdaddr="" # Override local bdaddr for 'one' rfcomm_pppd_server_one_channel="1" # Override local channel for 'one' #rfcomm_pppd_server_one_register_sp="NO" # Override SP and DUN register #rfcomm_pppd_server_one_register_dun="NO" # for 'one' # #rfcomm_pppd_server_two_bdaddr="" # Override local bdaddr for 'two' rfcomm_pppd_server_two_channel="3" # Override local channel for 'two' #rfcomm_pppd_server_two_register_sp="NO" # Override SP and DUN register #rfcomm_pppd_server_two_register_dun="NO" # for 'two' ubthidhci_enable="NO" # Switch an USB BT controller present on #ubthidhci_busnum="3" # bus 3 and addr 2 from HID mode to HCI mode. #ubthidhci_addr="2" # Check usbconfig list to find the correct # numbers for your system. ### Network link/usability verification options netwait_enable="NO" # Enable rc.d/netwait (or NO) #netwait_ip="" # Wait for ping response from any IP in this list. netwait_timeout="60" # Total number of seconds to perform pings. #netwait_if="" # Wait for active link on each intf in this list. netwait_if_timeout="30" # Total number of seconds to monitor link state. ### Miscellaneous network options: ### icmp_bmcastecho="NO" # respond to broadcast ping packets ### IPv6 options: ### ipv6_network_interfaces="auto" # List of IPv6 network interfaces # (or "auto" or "none"). ipv6_activate_all_interfaces="NO" # If NO, interfaces which have no # corresponding $ifconfig_IF_ipv6 is # marked as IFDISABLED for security # reason. ipv6_defaultrouter="NO" # Set to IPv6 default gateway (or NO). #ipv6_defaultrouter="2002:c058:6301::" # Use this for 6to4 (RFC 3068) ipv6_static_routes="" # Set to static route list (or leave empty). #ipv6_static_routes="xxx" # An example to set fec0:0000:0000:0006::/64 # route toward loopback interface. #ipv6_route_xxx="fec0:0000:0000:0006:: -prefixlen 64 ::1" ipv6_gateway_enable="NO" # Set to YES if this host will be a gateway. ipv6_cpe_wanif="NO" # Set to the upstream interface name if this # node will work as a router to forward IPv6 # packets not explicitly addressed to itself. ipv6_privacy="NO" # Use privacy address on RA-receiving IFs # (RFC 4941) route6d_enable="NO" # Set to YES to enable an IPv6 routing daemon. route6d_program="/usr/sbin/route6d" # Name of IPv6 routing daemon. route6d_flags="" # Flags to IPv6 routing daemon. #route6d_flags="-l" # Example for route6d with only IPv6 site local # addrs. #route6d_flags="-q" # If you want to run a routing daemon on an end # node, you should stop advertisement. #ipv6_network_interfaces="ed0 ep0" # Examples for router # or static configuration for end node. # Choose correct prefix value. #ipv6_prefix_ed0="fec0:0000:0000:0001 fec0:0000:0000:0002" # Examples for rtr. #ipv6_prefix_ep0="fec0:0000:0000:0003 fec0:0000:0000:0004" # Examples for rtr. ipv6_default_interface="NO" # Default output interface for scoped addrs. # This works only with # ipv6_gateway_enable="NO". rtsol_flags="" # Flags to IPv6 router solicitation. rtsold_enable="NO" # Set to YES to enable an IPv6 router # solicitation daemon. rtsold_flags="-a" # Flags to an IPv6 router solicitation # daemon. rtadvd_enable="NO" # Set to YES to enable an IPv6 router # advertisement daemon. If set to YES, # this router becomes a possible candidate # IPv6 default router for local subnets. rtadvd_interfaces="" # Interfaces rtadvd sends RA packets. mroute6d_enable="NO" # Do IPv6 multicast routing. mroute6d_program="/usr/local/sbin/pim6dd" # Name of IPv6 multicast # routing daemon. You need to # install it from package or # port. mroute6d_flags="" # Flags to IPv6 multicast routing daemon. stf_interface_ipv4addr="" # Local IPv4 addr for 6to4 IPv6 over IPv4 # tunneling interface. Specify this entry # to enable 6to4 interface. stf_interface_ipv4plen="0" # Prefix length for 6to4 IPv4 addr, # to limit peer addr range. Effective value # is 0-31. stf_interface_ipv6_ifid="0:0:0:1" # IPv6 interface id for stf0. # If you like, you can set "AUTO" for this. stf_interface_ipv6_slaid="0000" # IPv6 Site Level Aggregator for stf0 ipv6_ipv4mapping="NO" # Set to "YES" to enable IPv4 mapped IPv6 addr # communication. (like ::ffff:a.b.c.d) ipv6_ipfilter_rules="/etc/ipf6.rules" # rules definition file for ipfilter, # see /usr/src/contrib/ipfilter/rules # for examples ip6addrctl_enable="YES" # Set to YES to enable default address selection ip6addrctl_verbose="NO" # Set to YES to enable verbose configuration messages ip6addrctl_policy="AUTO" # A pre-defined address selection policy # (ipv4_prefer, ipv6_prefer, or AUTO) ############################################################## ### System console options ################################# ############################################################## keyboard="" # keyboard device to use (default /dev/kbd0). keymap="NO" # keymap in /usr/share/{syscons,vt}/keymaps/* (or NO). keyrate="NO" # keyboard rate to: slow, normal, fast (or NO). keybell="NO" # See kbdcontrol(1) for options. Use "off" to disable. keychange="NO" # function keys default values (or NO). cursor="NO" # cursor type {normal|blink|destructive} (or NO). scrnmap="NO" # screen map in /usr/share/syscons/scrnmaps/* (or NO). font8x16="NO" # font 8x16 from /usr/share/{syscons,vt}/fonts/* (or NO). font8x14="NO" # font 8x14 from /usr/share/{syscons,vt}/fonts/* (or NO). font8x8="NO" # font 8x8 from /usr/share/{syscons,vt}/fonts/* (or NO). blanktime="300" # blank time (in seconds) or "NO" to turn it off. saver="NO" # screen saver: Uses /boot/kernel/${saver}_saver.ko moused_nondefault_enable="YES" # Treat non-default mice as enabled unless # specifically overriden in rc.conf(5). moused_enable="NO" # Run the mouse daemon. moused_type="auto" # See man page for rc.conf(5) for available settings. moused_port="/dev/psm0" # Set to your mouse port. moused_flags="" # Any additional flags to moused. mousechar_start="NO" # if 0xd0-0xd3 default range is occupied in your # language code table, specify alternative range # start like mousechar_start=3, see vidcontrol(1) allscreens_flags="" # Set this vidcontrol mode for all virtual screens allscreens_kbdflags="" # Set this kbdcontrol mode for all virtual screens ############################################################## ### Mail Transfer Agent (MTA) options ###################### ############################################################## mta_start_script="/etc/rc.sendmail" # Script to start your chosen MTA, called by /etc/rc. # Settings for /etc/rc.sendmail and /etc/rc.d/sendmail: sendmail_enable="NO" # Run the sendmail inbound daemon (YES/NO). sendmail_pidfile="/var/run/sendmail.pid" # sendmail pid file sendmail_procname="/usr/sbin/sendmail" # sendmail process name sendmail_flags="-L sm-mta -bd -q30m" # Flags to sendmail (as a server) sendmail_cert_create="YES" # Create a server certificate if none (YES/NO) #sendmail_cert_cn="CN" # CN of the generate certificate sendmail_submit_enable="YES" # Start a localhost-only MTA for mail submission sendmail_submit_flags="-L sm-mta -bd -q30m -ODaemonPortOptions=Addr=localhost" # Flags for localhost-only MTA sendmail_outbound_enable="YES" # Dequeue stuck mail (YES/NO). sendmail_outbound_flags="-L sm-queue -q30m" # Flags to sendmail (outbound only) sendmail_msp_queue_enable="YES" # Dequeue stuck clientmqueue mail (YES/NO). sendmail_msp_queue_flags="-L sm-msp-queue -Ac -q30m" # Flags for sendmail_msp_queue daemon. sendmail_rebuild_aliases="NO" # Run newaliases if necessary (YES/NO). ############################################################## ### Miscellaneous administrative options ################### ############################################################## auditd_enable="NO" # Run the audit daemon. auditd_program="/usr/sbin/auditd" # Path to the audit daemon. auditd_flags="" # Which options to pass to the audit daemon. auditdistd_enable="NO" # Run the audit daemon. auditdistd_program="/usr/sbin/auditdistd" # Path to the auditdistd daemon. auditdistd_flags="" # Which options to pass to the auditdistd daemon. cron_enable="YES" # Run the periodic job daemon. cron_program="/usr/sbin/cron" # Which cron executable to run (if enabled). cron_dst="YES" # Handle DST transitions intelligently (YES/NO) cron_flags="" # Which options to pass to the cron daemon. lpd_enable="NO" # Run the line printer daemon. lpd_program="/usr/sbin/lpd" # path to lpd, if you want a different one. lpd_flags="" # Flags to lpd (if enabled). nscd_enable="NO" # Run the nsswitch caching daemon. chkprintcap_enable="NO" # Run chkprintcap(8) before running lpd. chkprintcap_flags="-d" # Create missing directories by default. dumpdev="AUTO" # Device to crashdump to (device name, AUTO, or NO). dumpdir="/var/crash" # Directory where crash dumps are to be stored savecore_enable="YES" # Extract core from dump devices if any savecore_flags="-m 10" # Used if dumpdev is enabled above, and present. # By default, only the 10 most recent kernel dumps # are saved. crashinfo_enable="YES" # Automatically generate crash dump summary. crashinfo_program="/usr/sbin/crashinfo" # Script to generate crash dump summary. quota_enable="NO" # turn on quotas on startup (or NO). check_quotas="YES" # Check quotas on startup (or NO). quotaon_flags="-a" # Turn quotas on for all file systems (if enabled) quotaoff_flags="-a" # Turn quotas off for all file systems at shutdown quotacheck_flags="-a" # Check all file system quotas (if enabled) accounting_enable="NO" # Turn on process accounting (or NO). ibcs2_enable="NO" # Ibcs2 (SCO) emulation loaded at startup (or NO). ibcs2_loaders="coff" # List of additional Ibcs2 loaders (or NO). firstboot_sentinel="/firstboot" # Scripts with "firstboot" keyword are run if # this file exists. Should be on a R/W filesystem so # the file can be deleted after the boot completes. # Emulation/compatibility services provided by /etc/rc.d/abi sysvipc_enable="NO" # Load System V IPC primitives at startup (or NO). linux_enable="NO" # Linux binary compatibility loaded at startup (or NO). svr4_enable="NO" # SysVR4 emulation loaded at startup (or NO). clear_tmp_enable="NO" # Clear /tmp at startup. clear_tmp_X="YES" # Clear and recreate X11-related directories in /tmp ldconfig_insecure="NO" # Set to YES to disable ldconfig security checks ldconfig_paths="/usr/lib/compat /usr/local/lib /usr/local/lib/compat/pkg" # shared library search paths ldconfig32_paths="/usr/lib32 /usr/lib32/compat" # 32-bit compatibility shared library search paths ldconfigsoft_paths="/usr/libsoft /usr/libsoft/compat /usr/local/libsoft" # soft float compatibility shared library search paths # Note: temporarily with extra stuff for transition ldconfig_paths_aout="/usr/lib/compat/aout /usr/local/lib/aout" # a.out shared library search paths ldconfig_local_dirs="/usr/local/libdata/ldconfig" # Local directories with ldconfig configuration files. ldconfig_local32_dirs="/usr/local/libdata/ldconfig32" # Local directories with 32-bit compatibility ldconfig # configuration files. ldconfig_localsoft_dirs="/usr/local/libdata/ldconfigsoft" # Local directories with soft float compatibility ldconfig # configuration files. kern_securelevel_enable="NO" # kernel security level (see security(7)) kern_securelevel="-1" # range: -1..3 ; `-1' is the most insecure # Note that setting securelevel to 0 will result # in the system booting with securelevel set to 1, as # init(8) will raise the level when rc(8) completes. update_motd="YES" # update version info in /etc/motd (or NO) entropy_boot_file="/boot/entropy" # Set to NO to disable very early # (used at early boot time) entropy caching through reboots. entropy_file="/entropy" # Set to NO to disable late (used when going multi-user) # entropy through reboots. # /var/db/entropy-file is preferred if / is not avail. entropy_dir="/var/db/entropy" # Set to NO to disable caching entropy via cron. entropy_save_sz="4096" # Size of the entropy cache files. entropy_save_num="8" # Number of entropy cache files to save. harvest_mask="511" # Entropy device harvests all but the very invasive sources. # (See 'sysctl kern.random.harvest' and random(4)) dmesg_enable="YES" # Save dmesg(8) to /var/run/dmesg.boot watchdogd_enable="NO" # Start the software watchdog daemon watchdogd_flags="" # Flags to watchdogd (if enabled) devfs_rulesets="/etc/defaults/devfs.rules /etc/devfs.rules" # Files containing # devfs(8) rules. devfs_system_ruleset="" # The name (NOT number) of a ruleset to apply to /dev devfs_set_rulesets="" # A list of /mount/dev=ruleset_name settings to # apply (must be mounted already, i.e. fstab(5)) devfs_load_rulesets="YES" # Enable to always load the default rulesets performance_cx_lowest="C2" # Online CPU idle state performance_cpu_freq="NONE" # Online CPU frequency economy_cx_lowest="Cmax" # Offline CPU idle state economy_cpu_freq="NONE" # Offline CPU frequency virecover_enable="YES" # Perform housekeeping for the vi(1) editor ugidfw_enable="NO" # Load mac_bsdextended(4) rules on boot bsdextended_script="/etc/rc.bsdextended" # Default mac_bsdextended(4) # ruleset file. newsyslog_enable="YES" # Run newsyslog at startup. newsyslog_flags="-CN" # Newsyslog flags to create marked files mixer_enable="YES" # Run the sound mixer. opensm_enable="NO" # Opensm(8) for infiniband devices defaults to off # rctl(8) requires kernel options RACCT and RCTL rctl_enable="YES" # Load rctl(8) rules on boot rctl_rules="/etc/rctl.conf" # rctl(8) ruleset. See rctl.conf(5). iovctl_files="" # Config files for iovctl(8) ############################################################## ### Jail Configuration (see rc.conf(5) manual page) ########## ############################################################## jail_enable="NO" # Set to NO to disable starting of any jails jail_parallel_start="NO" # Start jails in the background jail_list="" # Space separated list of names of jails jail_reverse_stop="NO" # Stop jails in reverse order ############################################################## ### Define source_rc_confs, the mechanism used by /etc/rc.* ## ### scripts to source rc_conf_files overrides safely. ## ############################################################## if [ -z "${source_rc_confs_defined}" ]; then source_rc_confs_defined=yes source_rc_confs() { local i sourced_files for i in ${rc_conf_files}; do case ${sourced_files} in *:$i:*) ;; *) sourced_files="${sourced_files}:$i:" if [ -r $i ]; then . $i fi ;; esac done # Re-do process to pick up [possibly] redefined $rc_conf_files for i in ${rc_conf_files}; do case ${sourced_files} in *:$i:*) ;; *) sourced_files="${sourced_files}:$i:" if [ -r $i ]; then . $i fi ;; esac done } fi + +# Allow vendors to override FreeBSD defaults in /etc/default/rc.conf +# without the need to carefully manage /etc/rc.conf. +if [ -r /etc/defaults/vendor.conf ]; then + . /etc/defaults/vendor.conf +fi Index: projects/vnet/lib/libusb/Makefile =================================================================== --- projects/vnet/lib/libusb/Makefile (revision 302157) +++ projects/vnet/lib/libusb/Makefile (revision 302158) @@ -1,269 +1,273 @@ # # $FreeBSD$ # # Makefile for the FreeBSD specific LibUSB 2.0 # PACKAGE=lib${LIB} LIB= usb SHLIB_MAJOR= 3 SHLIB_MINOR= 0 SRCS= libusb20.c SRCS+= libusb20_desc.c SRCS+= libusb20_ugen20.c INCS+= libusb20.h INCS+= libusb20_desc.h MAN= libusb.3 libusb20.3 MKLINT= no NOGCCERROR= PTHREAD_LIBS?= -lpthread WARNS?= 2 LIBADD= pthread MLINKS+= libusb.3 usb.3 # libusb 0.1 compat INCS+= usb.h SRCS+= libusb01.c # libusb 1.0 compat INCS+= libusb.h SRCS+= libusb10.c SRCS+= libusb10_desc.c SRCS+= libusb10_hotplug.c SRCS+= libusb10_io.c .if defined(COMPAT_32BIT) CFLAGS+= -DCOMPAT_32BIT .endif FILES= libusb-0.1.pc libusb-1.0.pc libusb-2.0.pc FILESDIR= ${LIBDATADIR}/pkgconfig # # Cross platform support # # Examples: # make LIBUSB_GLOBAL_INCLUDE_FILE=libusb_global_linux.h # make COMPAT_32BIT=YES \ # LIBUSB_GLOBAL_INCLUDE_FILE=libusb_global_linux.h \ # DEBUG_FLAGS="-g" # # From Ubuntu 10.04: # freebsd-make LIBUSB_GLOBAL_INCLUDE_FILE=libusb_global_linux.h \ # PTHREAD_LIBS="-lpthread -lrt" # freebsd-make COMPAT32_BIT=YES \ # LIBUSB_GLOBAL_INCLUDE_FILE=libusb_global_linux.h \ # PTHREAD_LIBS="-lpthread -lrt" # # .if defined(LIBUSB_GLOBAL_INCLUDE_FILE) CFLAGS+= -DLIBUSB_GLOBAL_INCLUDE_FILE=\"${LIBUSB_GLOBAL_INCLUDE_FILE}\" CFLAGS+= -DUSB_GLOBAL_INCLUDE_FILE=\"${LIBUSB_GLOBAL_INCLUDE_FILE}\" CFLAGS+= -I ../../sys .endif # LibUSB v1.0 MLINKS += libusb.3 libusb_get_version.3 MLINKS += libusb.3 libusb_init.3 MLINKS += libusb.3 libusb_exit.3 MLINKS += libusb.3 libusb_strerror.3 MLINKS += libusb.3 libusb_error_name.3 MLINKS += libusb.3 libusb_set_debug.3 MLINKS += libusb.3 libusb_get_device_list.3 MLINKS += libusb.3 libusb_free_device_list.3 MLINKS += libusb.3 libusb_get_bus_number.3 MLINKS += libusb.3 libusb_get_port_number.3 MLINKS += libusb.3 libusb_get_device_address.3 MLINKS += libusb.3 libusb_get_device_speed.3 MLINKS += libusb.3 libusb_get_max_packet_size.3 MLINKS += libusb.3 libusb_get_max_iso_packet_size.3 MLINKS += libusb.3 libusb_ref_device.3 MLINKS += libusb.3 libusb_unref_device.3 MLINKS += libusb.3 libusb_open.3 MLINKS += libusb.3 libusb_open_device_with_vid_pid.3 MLINKS += libusb.3 libusb_close.3 MLINKS += libusb.3 libusb_get_device.3 MLINKS += libusb.3 libusb_get_configuration.3 MLINKS += libusb.3 libusb_set_configuration.3 MLINKS += libusb.3 libusb_claim_interface.3 MLINKS += libusb.3 libusb_release_interface.3 MLINKS += libusb.3 libusb_set_interface_alt_setting.3 MLINKS += libusb.3 libusb_clear_halt.3 MLINKS += libusb.3 libusb_reset_device.3 MLINKS += libusb.3 libusb_check_connected.3 MLINKS += libusb.3 libusb_kernel_driver_active.3 MLINKS += libusb.3 libusb_get_driver.3 MLINKS += libusb.3 libusb_get_driver_np.3 MLINKS += libusb.3 libusb_detach_kernel_driver.3 MLINKS += libusb.3 libusb_detach_kernel_driver_np.3 MLINKS += libusb.3 libusb_attach_kernel_driver.3 MLINKS += libusb.3 libusb_set_auto_detach_kernel_driver.3 MLINKS += libusb.3 libusb_get_device_descriptor.3 MLINKS += libusb.3 libusb_get_active_config_descriptor.3 MLINKS += libusb.3 libusb_get_config_descriptor.3 MLINKS += libusb.3 libusb_get_config_descriptor_by_value.3 MLINKS += libusb.3 libusb_free_config_descriptor.3 MLINKS += libusb.3 libusb_get_string_descriptor.3 MLINKS += libusb.3 libusb_get_string_descriptor_ascii.3 MLINKS += libusb.3 libusb_parse_ss_endpoint_comp.3 MLINKS += libusb.3 libusb_free_ss_endpoint_comp.3 MLINKS += libusb.3 libusb_get_ss_endpoint_companion_descriptor.3 MLINKS += libusb.3 libusb_free_ss_endpoint_companion_descriptor.3 MLINKS += libusb.3 libusb_parse_bos_descriptor.3 MLINKS += libusb.3 libusb_free_bos_descriptor.3 MLINKS += libusb.3 libusb_get_usb_2_0_extension_descriptor.3 MLINKS += libusb.3 libusb_free_usb_2_0_extension_descriptor.3 MLINKS += libusb.3 libusb_get_ss_usb_device_capability_descriptor.3 MLINKS += libusb.3 libusb_free_ss_usb_device_capability_descriptor.3 MLINKS += libusb.3 libusb_get_container_id_descriptor.3 MLINKS += libusb.3 libusb_free_container_id_descriptor.3 +MLINKS += libusb.3 libusb_alloc_streams.3 +MLINKS += libusb.3 libusb_free_streams.3 MLINKS += libusb.3 libusb_alloc_transfer.3 MLINKS += libusb.3 libusb_free_transfer.3 +MLINKS += libusb.3 libusb_transfer_set_stream_id.3 +MLINKS += libusb.3 libusb_transfer_get_stream_id.3 MLINKS += libusb.3 libusb_submit_transfer.3 MLINKS += libusb.3 libusb_cancel_transfer.3 MLINKS += libusb.3 libusb_control_transfer.3 MLINKS += libusb.3 libusb_bulk_transfer.3 MLINKS += libusb.3 libusb_interrupt_transfer.3 MLINKS += libusb.3 libusb_try_lock_events.3 MLINKS += libusb.3 libusb_lock_events.3 MLINKS += libusb.3 libusb_unlock_events.3 MLINKS += libusb.3 libusb_event_handling_ok.3 MLINKS += libusb.3 libusb_event_handler_active.3 MLINKS += libusb.3 libusb_lock_event_waiters.3 MLINKS += libusb.3 libusb_unlock_event_waiters.3 MLINKS += libusb.3 libusb_wait_for_event.3 MLINKS += libusb.3 libusb_handle_events_timeout_completed.3 MLINKS += libusb.3 libusb_handle_events_completed.3 MLINKS += libusb.3 libusb_handle_events_timeout.3 MLINKS += libusb.3 libusb_handle_events.3 MLINKS += libusb.3 libusb_handle_events_locked.3 MLINKS += libusb.3 libusb_get_next_timeout.3 MLINKS += libusb.3 libusb_set_pollfd_notifiers.3 MLINKS += libusb.3 libusb_get_pollfds.3 MLINKS += libusb.3 libusb_hotplug_register_callback.3 MLINKS += libusb.3 libusb_hotplug_deregister_callback.3 # LibUSB v0.1 MLINKS += libusb.3 usb_open.3 MLINKS += libusb.3 usb_close.3 MLINKS += libusb.3 usb_get_string.3 MLINKS += libusb.3 usb_get_string_simple.3 MLINKS += libusb.3 usb_get_descriptor_by_endpoint.3 MLINKS += libusb.3 usb_get_descriptor.3 MLINKS += libusb.3 usb_parse_descriptor.3 MLINKS += libusb.3 usb_parse_configuration.3 MLINKS += libusb.3 usb_destroy_configuration.3 MLINKS += libusb.3 usb_fetch_and_parse_descriptors.3 MLINKS += libusb.3 usb_bulk_write.3 MLINKS += libusb.3 usb_bulk_read.3 MLINKS += libusb.3 usb_interrupt_write.3 MLINKS += libusb.3 usb_interrupt_read.3 MLINKS += libusb.3 usb_control_msg.3 MLINKS += libusb.3 usb_set_configuration.3 MLINKS += libusb.3 usb_claim_interface.3 MLINKS += libusb.3 usb_release_interface.3 MLINKS += libusb.3 usb_set_altinterface.3 MLINKS += libusb.3 usb_resetep.3 MLINKS += libusb.3 usb_clear_halt.3 MLINKS += libusb.3 usb_reset.3 MLINKS += libusb.3 usb_strerror.3 MLINKS += libusb.3 usb_init.3 MLINKS += libusb.3 usb_set_debug.3 MLINKS += libusb.3 usb_find_busses.3 MLINKS += libusb.3 usb_find_devices.3 MLINKS += libusb.3 usb_device.3 MLINKS += libusb.3 usb_get_busses.3 MLINKS += libusb.3 usb_check_connected.3 # LibUSB v2.0 MLINKS += libusb20.3 libusb20_tr_close.3 MLINKS += libusb20.3 libusb20_tr_open.3 MLINKS += libusb20.3 libusb20_tr_get_pointer.3 MLINKS += libusb20.3 libusb20_tr_get_time_complete.3 MLINKS += libusb20.3 libusb20_tr_get_actual_frames.3 MLINKS += libusb20.3 libusb20_tr_get_actual_length.3 MLINKS += libusb20.3 libusb20_tr_get_max_frames.3 MLINKS += libusb20.3 libusb20_tr_get_max_packet_length.3 MLINKS += libusb20.3 libusb20_tr_get_max_total_length.3 MLINKS += libusb20.3 libusb20_tr_get_status.3 MLINKS += libusb20.3 libusb20_tr_pending.3 MLINKS += libusb20.3 libusb20_tr_callback_wrapper.3 MLINKS += libusb20.3 libusb20_tr_clear_stall_sync.3 MLINKS += libusb20.3 libusb20_tr_drain.3 MLINKS += libusb20.3 libusb20_tr_set_buffer.3 MLINKS += libusb20.3 libusb20_tr_set_callback.3 MLINKS += libusb20.3 libusb20_tr_set_flags.3 MLINKS += libusb20.3 libusb20_tr_get_length.3 MLINKS += libusb20.3 libusb20_tr_set_length.3 MLINKS += libusb20.3 libusb20_tr_set_priv_sc0.3 MLINKS += libusb20.3 libusb20_tr_set_priv_sc1.3 MLINKS += libusb20.3 libusb20_tr_set_timeout.3 MLINKS += libusb20.3 libusb20_tr_set_total_frames.3 MLINKS += libusb20.3 libusb20_tr_setup_bulk.3 MLINKS += libusb20.3 libusb20_tr_setup_control.3 MLINKS += libusb20.3 libusb20_tr_setup_intr.3 MLINKS += libusb20.3 libusb20_tr_setup_isoc.3 MLINKS += libusb20.3 libusb20_tr_bulk_intr_sync.3 MLINKS += libusb20.3 libusb20_tr_start.3 MLINKS += libusb20.3 libusb20_tr_stop.3 MLINKS += libusb20.3 libusb20_tr_submit.3 MLINKS += libusb20.3 libusb20_tr_get_priv_sc0.3 MLINKS += libusb20.3 libusb20_tr_get_priv_sc1.3 MLINKS += libusb20.3 libusb20_dev_get_backend_name.3 MLINKS += libusb20.3 libusb20_dev_get_info.3 MLINKS += libusb20.3 libusb20_dev_get_iface_desc.3 MLINKS += libusb20.3 libusb20_dev_get_desc.3 MLINKS += libusb20.3 libusb20_dev_close.3 MLINKS += libusb20.3 libusb20_dev_detach_kernel_driver.3 MLINKS += libusb20.3 libusb20_dev_set_config_index.3 MLINKS += libusb20.3 libusb20_dev_get_debug.3 MLINKS += libusb20.3 libusb20_dev_get_fd.3 MLINKS += libusb20.3 libusb20_dev_kernel_driver_active.3 MLINKS += libusb20.3 libusb20_dev_open.3 MLINKS += libusb20.3 libusb20_dev_process.3 MLINKS += libusb20.3 libusb20_dev_request_sync.3 MLINKS += libusb20.3 libusb20_dev_req_string_sync.3 MLINKS += libusb20.3 libusb20_dev_req_string_simple_sync.3 MLINKS += libusb20.3 libusb20_dev_reset.3 MLINKS += libusb20.3 libusb20_dev_check_connected.3 MLINKS += libusb20.3 libusb20_dev_set_power_mode.3 MLINKS += libusb20.3 libusb20_dev_get_power_mode.3 MLINKS += libusb20.3 libusb20_dev_get_port_path.3 MLINKS += libusb20.3 libusb20_dev_get_power_usage.3 MLINKS += libusb20.3 libusb20_dev_set_alt_index.3 MLINKS += libusb20.3 libusb20_dev_get_device_desc.3 MLINKS += libusb20.3 libusb20_dev_alloc_config.3 MLINKS += libusb20.3 libusb20_dev_alloc.3 MLINKS += libusb20.3 libusb20_dev_get_address.3 MLINKS += libusb20.3 libusb20_dev_get_parent_address.3 MLINKS += libusb20.3 libusb20_dev_get_parent_port.3 MLINKS += libusb20.3 libusb20_dev_get_bus_number.3 MLINKS += libusb20.3 libusb20_dev_get_mode.3 MLINKS += libusb20.3 libusb20_dev_get_speed.3 MLINKS += libusb20.3 libusb20_dev_get_config_index.3 MLINKS += libusb20.3 libusb20_dev_free.3 MLINKS += libusb20.3 libusb20_dev_set_debug.3 MLINKS += libusb20.3 libusb20_dev_wait_process.3 MLINKS += libusb20.3 libusb20_be_get_template.3 MLINKS += libusb20.3 libusb20_be_set_template.3 MLINKS += libusb20.3 libusb20_be_get_dev_quirk.3 MLINKS += libusb20.3 libusb20_be_get_quirk_name.3 MLINKS += libusb20.3 libusb20_be_add_dev_quirk.3 MLINKS += libusb20.3 libusb20_be_remove_dev_quirk.3 MLINKS += libusb20.3 libusb20_be_alloc_default.3 MLINKS += libusb20.3 libusb20_be_device_foreach.3 MLINKS += libusb20.3 libusb20_be_dequeue_device.3 MLINKS += libusb20.3 libusb20_be_enqueue_device.3 MLINKS += libusb20.3 libusb20_be_free.3 MLINKS += libusb20.3 libusb20_me_get_1.3 MLINKS += libusb20.3 libusb20_me_get_2.3 MLINKS += libusb20.3 libusb20_me_encode.3 MLINKS += libusb20.3 libusb20_me_decode.3 MLINKS += libusb20.3 libusb20_desc_foreach.3 MLINKS += libusb20.3 libusb20_strerror.3 MLINKS += libusb20.3 libusb20_error_name.3 .include Index: projects/vnet/lib/libusb/libusb.3 =================================================================== --- projects/vnet/lib/libusb/libusb.3 (revision 302157) +++ projects/vnet/lib/libusb/libusb.3 (revision 302158) @@ -1,731 +1,754 @@ .\" .\" Copyright (c) 2009 Sylvestre Gallon .\" .\" 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 June 22, 2016 +.Dd June 23, 2016 .Dt LIBUSB 3 .Os .Sh NAME .Nm libusb .Nd "USB access library" .Sh LIBRARY USB access library .Pq libusb, -lusb .Sh SYNOPSIS .In libusb.h .Sh DESCRIPTION The .Nm library contains interfaces for directly managing a usb device. The current implementation supports v1.0 of the libusb API. .Sh LIBRARY INITIALISATION AND DEINITIALISATION .Ft "const struct libusb_version *" .Fn libusb_get_version "void" This function returns version information about LibUSB. .Pp .Ft int .Fn libusb_init "libusb_context **ctx" This function initialises libusb. It must be called at the beginning of the program, before other libusb routines are used. This function returns 0 on success or LIBUSB_ERROR on failure. .Pp .Ft void .Fn libusb_exit "libusb_context *ctx" Deinitialise libusb. Must be called at the end of the application. Other libusb routines may not be called after this function. .Pp .Ft const char * .Fn libusb_strerror "int code" Get the ASCII representation of the error given by the .Fa code argument. This function does not return NULL. .Pp .Ft const char * .Fn libusb_error_name "int code" Get the ASCII representation of the error enum given by the .Fa code argument. This function does not return NULL. .Pp .Ft void .Fn libusb_set_debug "libusb_context *ctx" "int level" Set the debug level to .Fa level . .Pp .Ft ssize_t .Fn libusb_get_device_list "libusb_context *ctx" "libusb_device ***list" Populate .Fa list with the list of usb devices available, adding a reference to each device in the list. All the list entries created by this function must have their reference counter decremented when you are done with them, and the list itself must be freed. This function returns the number of devices in the list or a LIBUSB_ERROR code. .Pp .Ft void .Fn libusb_free_device_list "libusb_device **list" "int unref_devices" Free the list of devices discovered by libusb_get_device_list. If .Fa unref_device is set to 1 all devices in the list have their reference counter decremented once. .Pp .Ft uint8_t .Fn libusb_get_bus_number "libusb_device *dev" Returns the number of the bus contained by the device .Fa dev . .Pp .Ft uint8_t .Fn libusb_get_port_number "libusb_device *dev" Returns the port number which the device given by .Fa dev is attached to. .Pp .Ft int .Fn libusb_get_port_numbers "libusb_device *dev" "uint8_t *buf" "uint8_t bufsize" Stores, in the buffer .Fa buf of size .Fa bufsize , the list of all port numbers from root for the device .Fa dev . .Pp .Ft int .Fn libusb_get_port_path "libusb_context *ctx" "libusb_device *dev" "uint8_t *buf" "uint8_t bufsize" Deprecated function equivalent to libusb_get_port_numbers. .Pp .Ft uint8_t .Fn libusb_get_device_address "libusb_device *dev" Returns the device_address contained by the device .Fa dev . .Pp .Ft enum libusb_speed .Fn libusb_get_device_speed "libusb_device *dev" Returns the wire speed at which the device is connected. See the LIBUSB_SPEED_XXX enums for more information. LIBUSB_SPEED_UNKNOWN is returned in case of unknown wire speed. .Pp .Ft int .Fn libusb_get_max_packet_size "libusb_device *dev" "unsigned char endpoint" Returns the wMaxPacketSize value on success, LIBUSB_ERROR_NOT_FOUND if the endpoint does not exist and LIBUSB_ERROR_OTHERS on other failure. .Pp .Ft int .Fn libusb_get_max_iso_packet_size "libusb_device *dev" "unsigned char endpoint" Returns the packet size multiplied by the packet multiplier on success, LIBUSB_ERROR_NOT_FOUND if the endpoint does not exist and LIBUSB_ERROR_OTHERS on other failure. .Pp .Ft libusb_device * .Fn libusb_ref_device "libusb_device *dev" Increment the reference counter of the device .Fa dev . .Pp .Ft void .Fn libusb_unref_device "libusb_device *dev" Decrement the reference counter of the device .Fa dev . .Pp .Ft int .Fn libusb_open "libusb_device *dev" "libusb_device_handle **devh" Open a device and obtain a device_handle. Returns 0 on success, LIBUSB_ERROR_NO_MEM on memory allocation problems, LIBUSB_ERROR_ACCESS on permissions problems, LIBUSB_ERROR_NO_DEVICE if the device has been disconnected and a LIBUSB_ERROR code on other errors. .Pp .Ft libusb_device_handle * .Fn libusb_open_device_with_vid_pid "libusb_context *ctx" "uint16_t vid" "uint16_t pid" A convenience function to open a device by vendor and product IDs .Fa vid and .Fa pid . Returns NULL on error. .Pp .Ft void .Fn libusb_close "libusb_device_handle *devh" Close a device handle. .Pp .Ft libusb_device * .Fn libusb_get_device "libusb_device_handle *devh" Get the device contained by devh. Returns NULL on error. .Pp .Ft int .Fn libusb_get_configuration "libusb_device_handle *devh" "int *config" Returns the value of the current configuration. Returns 0 on success, LIBUSB_ERROR_NO_DEVICE if the device has been disconnected and a LIBUSB_ERROR code on error. .Pp .Ft int .Fn libusb_set_configuration "libusb_device_handle *devh" "int config" Set the active configuration to .Fa config for the device contained by .Fa devh . This function returns 0 on success, LIBUSB_ERROR_NOT_FOUND if the requested configuration does not exist, LIBUSB_ERROR_BUSY if the interfaces are currently claimed, LIBUSB_ERROR_NO_DEVICE if the device has been disconnected and a LIBUSB_ERROR code on failure. .Pp .Ft int .Fn libusb_claim_interface "libusb_device_handle *devh" "int interface_number" Claim an interface in a given libusb_handle .Fa devh . This is a non-blocking function. It returns 0 on success, LIBUSB_ERROR_NOT_FOUND if the requested interface does not exist, LIBUSB_ERROR_BUSY if a program or driver has claimed the interface, LIBUSB_ERROR_NO_DEVICE if the device has been disconnected and a LIBUSB_ERROR code on failure. .Pp .Ft int .Fn libusb_release_interface "libusb_device_handle *devh" "int interface_number" This function releases an interface. All the claimed interfaces on a device must be released before closing the device. Returns 0 on success, LIBUSB_ERROR_NOT_FOUND if the interface was not claimed, LIBUSB_ERROR_NO_DEVICE if the device has been disconnected and LIBUSB_ERROR on failure. .Pp .Ft int .Fn libusb_set_interface_alt_setting "libusb_device_handle *dev" "int interface_number" "int alternate_setting" Activate an alternate setting for an interface. Returns 0 on success, LIBUSB_ERROR_NOT_FOUND if the interface was not claimed or the requested setting does not exist, LIBUSB_ERROR_NO_DEVICE if the device has been disconnected and a LIBUSB_ERROR code on failure. .Pp .Ft int .Fn libusb_clear_halt "libusb_device_handle *devh" "unsigned char endpoint" Clear an halt/stall for a endpoint. Returns 0 on success, LIBUSB_ERROR_NOT_FOUND if the endpoint does not exist, LIBUSB_ERROR_NO_DEVICE if the device has been disconnected and a LIBUSB_ERROR code on failure. .Pp .Ft int .Fn libusb_reset_device "libusb_device_handle *devh" Perform an USB port reset for an usb device. Returns 0 on success, LIBUSB_ERROR_NOT_FOUND if re-enumeration is required or if the device has been disconnected and a LIBUSB_ERROR code on failure. .Pp .Ft int .Fn libusb_check_connected "libusb_device_handle *devh" Test if the USB device is still connected. Returns 0 on success, LIBUSB_ERROR_NO_DEVICE if it has been disconnected and a LIBUSB_ERROR code on failure. .Pp .Ft int .Fn libusb_kernel_driver_active "libusb_device_handle *devh" "int interface" Determine if a driver is active on a interface. Returns 0 if no kernel driver is active and 1 if a kernel driver is active, LIBUSB_ERROR_NO_DEVICE if the device has been disconnected and a LIBUSB_ERROR code on failure. .Pp .Ft int .Fn libusb_get_driver "libusb_device_handle *devh" "int interface" "char *name" "int namelen" or .Ft int .Fn libusb_get_driver_np "libusb_device_handle *devh" "int interface" "char *name" "int namelen" Copy the name of the driver attached to the given .Fa device and .Fa interface into the buffer .Fa name of length .Fa namelen . Returns 0 on success, LIBUSB_ERROR_NOT_FOUND if no kernel driver is attached to the given interface and LIBUSB_ERROR_INVALID_PARAM if the interface does not exist. This function is non-portable. The buffer pointed to by .Fa name is only zero terminated on success. .Pp .Ft int .Fn libusb_detach_kernel_driver "libusb_device_handle *devh" "int interface" or .Ft int .Fn libusb_detach_kernel_driver_np "libusb_device_handle *devh" "int interface" Detach a kernel driver from an interface. This is needed to claim an interface already claimed by a kernel driver. Returns 0 on success, LIBUSB_ERROR_NOT_FOUND if no kernel driver was active, LIBUSB_ERROR_INVALID_PARAM if the interface does not exist, LIBUSB_ERROR_NO_DEVICE if the device has been disconnected and a LIBUSB_ERROR code on failure. This function is non-portable. .Pp .Ft int .Fn libusb_attach_kernel_driver "libusb_device_handle *devh" "int interface" Re-attach an interface kernel driver that was previously detached. Returns 0 on success, LIBUSB_ERROR_INVALID_PARAM if the interface does not exist, LIBUSB_ERROR_NO_DEVICE if the device has been disconnected, LIBUSB_ERROR_BUSY if the driver cannot be attached because the interface is claimed by a program or driver and a LIBUSB_ERROR code on failure. .Pp .Ft int .Fn libusb_set_auto_detach_kernel_driver "libusb_device_handle *devh" "int enable" This function enables automatic kernel interface driver detach when an interface is claimed. When the interface is restored the kernel driver is allowed to be re-attached. If the .Fa enable argument is non-zero the feature is enabled. Else disabled. Returns 0 on success and a LIBUSB_ERROR code on failure. .Sh USB DESCRIPTORS .Ft int .Fn libusb_get_device_descriptor "libusb_device *dev" "libusb_device_descriptor *desc" Get the USB device descriptor for the device .Fa dev . This is a non-blocking function. Returns 0 on success and a LIBUSB_ERROR code on failure. .Pp .Ft int .Fn libusb_get_active_config_descriptor "libusb_device *dev" "struct libusb_config_descriptor **config" Get the USB configuration descriptor for the active configuration. Returns 0 on success, LIBUSB_ERROR_NOT_FOUND if the device is in an unconfigured state and a LIBUSB_ERROR code on error. .Pp .Ft int .Fn libusb_get_config_descriptor "libusb_device *dev" "uint8_t config_index" "libusb_config_descriptor **config" Get a USB configuration descriptor based on its index .Fa idx. Returns 0 on success, LIBUSB_ERROR_NOT_FOUND if the configuration does not exist and a LIBUSB_ERROR code on error. .Pp .Ft int .Fn libusb_get_config_descriptor_by_value "libusb_device *dev" "uint8 bConfigurationValue" "libusb_config_descriptor **config" Get a USB configuration descriptor with a specific bConfigurationValue. This is a non-blocking function which does not send a request through the device. Returns 0 on success, LIBUSB_ERROR_NOT_FOUND if the configuration does not exist and a LIBUSB_ERROR code on failure. .Pp .Ft void .Fn libusb_free_config_descriptor "libusb_config_descriptor *config" Free a configuration descriptor. .Pp .Ft int .Fn libusb_get_string_descriptor "libusb_device_handle *devh" "uint8_t desc_idx" "uint16_t langid" "unsigned char *data" "int length" Retrieve a string descriptor in raw format. Returns the number of bytes actually transferred on success or a negative LIBUSB_ERROR code on failure. .Pp .Ft int .Fn libusb_get_string_descriptor_ascii "libusb_device_handle *devh" "uint8_t desc_idx" "unsigned char *data" "int length" Retrieve a string descriptor in C style ASCII. Returns the positive number of bytes in the resulting ASCII string on success and a LIBUSB_ERROR code on failure. .Pp .Ft int .Fn libusb_parse_ss_endpoint_comp "const void *buf" "int len" "libusb_ss_endpoint_companion_descriptor **ep_comp" This function parses the USB 3.0 endpoint companion descriptor in host endian format pointed to by .Fa buf and having a length of .Fa len . Typically these arguments are the extra and extra_length fields of the endpoint descriptor. On success the pointer to resulting descriptor is stored at the location given by .Fa ep_comp . Returns zero on success and a LIBUSB_ERROR code on failure. On success the parsed USB 3.0 endpoint companion descriptor must be freed using the libusb_free_ss_endpoint_comp function. .Pp .Ft void .Fn libusb_free_ss_endpoint_comp "libusb_ss_endpoint_companion_descriptor *ep_comp" This function is NULL safe and frees a parsed USB 3.0 endpoint companion descriptor given by .Fa ep_comp . .Pp .Ft int .Fn libusb_get_ss_endpoint_companion_descriptor "struct libusb_context *ctx" "const struct libusb_endpoint_descriptor *endpoint" "struct libusb_ss_endpoint_companion_descriptor **ep_comp" This function finds and parses the USB 3.0 endpoint companion descriptor given by .Fa endpoint . Returns zero on success and a LIBUSB_ERROR code on failure. On success the parsed USB 3.0 endpoint companion descriptor must be freed using the libusb_free_ss_endpoint_companion_descriptor function. .Pp .Ft void .Fn libusb_free_ss_endpoint_companion_descriptor "struct libusb_ss_endpoint_companion_descriptor *ep_comp" This function is NULL safe and frees a parsed USB 3.0 endpoint companion descriptor given by .Fa ep_comp . .Pp .Ft int .Fn libusb_get_bos_descriptor "libusb_device_handle *handle" "struct libusb_bos_descriptor **bos" This function queries the USB device given by .Fa handle and stores a pointer to a parsed BOS descriptor into .Fa bos . Returns zero on success and a LIBUSB_ERROR code on failure. On success the parsed BOS descriptor must be freed using the libusb_free_bos_descriptor function. .Pp .Ft int .Fn libusb_parse_bos_descriptor "const void *buf" "int len" "libusb_bos_descriptor **bos" This function parses a Binary Object Store, BOS, descriptor into host endian format pointed to by .Fa buf and having a length of .Fa len . On success the pointer to resulting descriptor is stored at the location given by .Fa bos . Returns zero on success and a LIBUSB_ERROR code on failure. On success the parsed BOS descriptor must be freed using the libusb_free_bos_descriptor function. .Pp .Ft void .Fn libusb_free_bos_descriptor "libusb_bos_descriptor *bos" This function is NULL safe and frees a parsed BOS descriptor given by .Fa bos . .Pp .Ft int .Fn libusb_get_usb_2_0_extension_descriptor "struct libusb_context *ctx" "struct libusb_bos_dev_capability_descriptor *dev_cap" "struct libusb_usb_2_0_extension_descriptor **usb_2_0_extension" This function parses the USB 2.0 extension descriptor from the descriptor given by .Fa dev_cap and stores a pointer to the parsed descriptor into .Fa usb_2_0_extension . Returns zero on success and a LIBUSB_ERROR code on failure. On success the parsed USB 2.0 extension descriptor must be freed using the libusb_free_usb_2_0_extension_descriptor function. .Pp .Ft void .Fn libusb_free_usb_2_0_extension_descriptor "struct libusb_usb_2_0_extension_descriptor *usb_2_0_extension" This function is NULL safe and frees a parsed USB 2.0 extension descriptor given by .Fa usb_2_0_extension . .Pp .Ft int .Fn libusb_get_ss_usb_device_capability_descriptor "struct libusb_context *ctx" "struct libusb_bos_dev_capability_descriptor *dev_cap" "struct libusb_ss_usb_device_capability_descriptor **ss_usb_device_capability" This function parses the SuperSpeed device capability descriptor from the descriptor given by .Fa dev_cap and stores a pointer to the parsed descriptor into .Fa ss_usb_device_capability . Returns zero on success and a LIBUSB_ERROR code on failure. On success the parsed SuperSpeed device capability descriptor must be freed using the libusb_free_ss_usb_device_capability_descriptor function. .Pp .Ft void .Fn libusb_free_ss_usb_device_capability_descriptor "struct libusb_ss_usb_device_capability_descriptor *ss_usb_device_capability" This function is NULL safe and frees a parsed SuperSpeed device capability descriptor given by .Fa ss_usb_device_capability . .Pp .Ft int .Fn libusb_get_container_id_descriptor "struct libusb_context *ctx" "struct libusb_bos_dev_capability_descriptor *dev_cap" "struct libusb_container_id_descriptor **container_id" This function parses the container ID descriptor from the descriptor given by .Fa dev_cap and stores a pointer to the parsed descriptor into .Fa container_id . Returns zero on success and a LIBUSB_ERROR code on failure. On success the parsed container ID descriptor must be freed using the libusb_free_container_id_descriptor function. .Pp .Ft void .Fn libusb_free_container_id_descriptor "struct libusb_container_id_descriptor *container_id" This function is NULL safe and frees a parsed container ID descriptor given by .Fa container_id . .Sh USB ASYNCHRONOUS I/O .Ft struct libusb_transfer * .Fn libusb_alloc_transfer "int iso_packets" Allocate a transfer with the number of isochronous packet descriptors specified by .Fa iso_packets . Returns NULL on error. .Pp .Ft void .Fn libusb_free_transfer "struct libusb_transfer *tr" Free a transfer. .Pp .Ft int .Fn libusb_submit_transfer "struct libusb_transfer *tr" This function will submit a transfer and returns immediately. Returns 0 on success, LIBUSB_ERROR_NO_DEVICE if the device has been disconnected and a LIBUSB_ERROR code on other failure. .Pp .Ft int .Fn libusb_cancel_transfer "struct libusb_transfer *tr" This function asynchronously cancels a transfer. Returns 0 on success and a LIBUSB_ERROR code on failure. .Sh USB SYNCHRONOUS I/O .Ft int .Fn libusb_control_transfer "libusb_device_handle *devh" "uint8_t bmRequestType" "uint8_t bRequest" "uint16_t wValue" "uint16_t wIndex" "unsigned char *data" "uint16_t wLength" "unsigned int timeout" Perform a USB control transfer. Returns the actual number of bytes transferred on success, in the range from and including zero up to and including .Fa wLength . On error a LIBUSB_ERROR code is returned, for example LIBUSB_ERROR_TIMEOUT if the transfer timed out, LIBUSB_ERROR_PIPE if the control request was not supported, LIBUSB_ERROR_NO_DEVICE if the device has been disconnected and another LIBUSB_ERROR code on other failures. The LIBUSB_ERROR codes are all negative. .Pp .Ft int .Fn libusb_bulk_transfer "struct libusb_device_handle *devh" "unsigned char endpoint" "unsigned char *data" "int length" "int *transferred" "unsigned int timeout" Perform an USB bulk transfer. A timeout value of zero means no timeout. The timeout value is given in milliseconds. Returns 0 on success, LIBUSB_ERROR_TIMEOUT if the transfer timed out, LIBUSB_ERROR_PIPE if the control request was not supported, LIBUSB_ERROR_OVERFLOW if the device offered more data, LIBUSB_ERROR_NO_DEVICE if the device has been disconnected and a LIBUSB_ERROR code on other failure. .Pp .Ft int .Fn libusb_interrupt_transfer "struct libusb_device_handle *devh" "unsigned char endpoint" "unsigned char *data" "int length" "int *transferred" "unsigned int timeout" Perform an USB Interrupt transfer. A timeout value of zero means no timeout. The timeout value is given in milliseconds. Returns 0 on success, LIBUSB_ERROR_TIMEOUT if the transfer timed out, LIBUSB_ERROR_PIPE if the control request was not supported, LIBUSB_ERROR_OVERFLOW if the device offered more data, LIBUSB_ERROR_NO_DEVICE if the device has been disconnected and a LIBUSB_ERROR code on other failure. +.Sh USB STREAMS SUPPORT +.Ft int +.Fn libusb_alloc_streams "libusb_device_handle *dev" "uint32_t num_streams" "unsigned char *endpoints" "int num_endpoints" +This function verifies that the given number of streams using the +given number of endpoints is allowed and allocates the resources +needed to use so-called USB streams. +Currently only a single stream per endpoint is supported to simplify +the internals of LibUSB. +This function returns 0 on success or a LIBUSB_ERROR code on failure. +.Pp +.Ft int +.Fn libusb_free_streams "libusb_device_handle *dev" "unsigned char *endpoints" "int num_endpoints" +This function release resources needed for streams usage. +Returns 0 on success or a LIBUSB_ERROR code on failure. +.Pp +.Ft void +.Fn libusb_transfer_set_stream_id "struct libusb_transfer *transfer" "uint32_t stream_id" +This function sets the stream ID for the given USB transfer. +.Pp +.Ft uint32_t +.Fn libusb_transfer_get_stream_id "struct libusb_transfer *transfer" +This function returns the stream ID for the given USB transfer. +If no stream ID is used a value of zero is returned. .Sh USB EVENTS .Ft int .Fn libusb_try_lock_events "libusb_context *ctx" Try to acquire the event handling lock. Returns 0 if the lock was obtained and 1 if not. .Pp .Ft void .Fn libusb_lock_events "libusb_context *ctx" Acquire the event handling lock. This function is blocking. .Pp .Ft void .Fn libusb_unlock_events "libusb_context *ctx" Release the event handling lock. This will wake up any thread blocked on .Fn libusb_wait_for_event . .Pp .Ft int .Fn libusb_event_handling_ok "libusb_context *ctx" Determine if it still OK for this thread to be doing event handling. Returns 1 if event handling can start or continue. Returns 0 if this thread must give up the events lock. .Pp .Ft int .Fn libusb_event_handler_active "libusb_context *ctx" Determine if an active thread is handling events. Returns 1 if there is a thread handling events and 0 if there are no threads currently handling events. .Pp .Ft void .Fn libusb_lock_event_waiters "libusb_context *ctx" Acquire the event_waiters lock. This lock is designed to be obtained in the situation where you want to be aware when events are completed, but some other thread is event handling so calling libusb_handle_events() is not allowed. .Pp .Ft void .Fn libusb_unlock_event_waiters "libusb_context *ctx" Release the event_waiters lock. .Pp .Ft int .Fn libusb_wait_for_event "libusb_context *ctx" "struct timeval *tv" Wait for another thread to signal completion of an event. Must be called with the event waiters lock held, see libusb_lock_event_waiters(). This will block until the timeout expires or a transfer completes or a thread releases the event handling lock through libusb_unlock_events(). Returns 0 after a transfer completes or another thread stops event handling, and 1 if the timeout expired. .Pp .Ft int .Fn libusb_handle_events_timeout_completed "libusb_context *ctx" "struct timeval *tv" "int *completed" Handle any pending events by checking if timeouts have expired and by checking the set of file descriptors for activity. If the .Fa completed argument is not equal to NULL, this function will loop until a transfer completion callback sets the variable pointed to by the .Fa completed argument to non-zero. If the .Fa tv argument is not equal to NULL, this function will return LIBUSB_ERROR_TIMEOUT after the given timeout. Returns 0 on success, or a LIBUSB_ERROR code on failure or timeout. .Pp .Ft int .Fn libusb_handle_events_completed "libusb_context *ctx" "int *completed" Handle any pending events by checking the set of file descriptors for activity. If the .Fa completed argument is not equal to NULL, this function will loop until a transfer completion callback sets the variable pointed to by the .Fa completed argument to non-zero. Returns 0 on success, or a LIBUSB_ERROR code on failure. .Pp .Ft int .Fn libusb_handle_events_timeout "libusb_context *ctx" "struct timeval *tv" Handle any pending events by checking if timeouts have expired and by checking the set of file descriptors for activity. Returns 0 on success, or a LIBUSB_ERROR code on failure or timeout. .Pp .Ft int .Fn libusb_handle_events "libusb_context *ctx" Handle any pending events in blocking mode with a sensible timeout. Returns 0 on success and a LIBUSB_ERROR code on failure. .Pp .Ft int .Fn libusb_handle_events_locked "libusb_context *ctx" "struct timeval *tv" Handle any pending events by polling file descriptors, without checking if another thread is already doing so. Must be called with the event lock held. .Pp .Ft int .Fn libusb_get_next_timeout "libusb_context *ctx" "struct timeval *tv" Determine the next internal timeout that libusb needs to handle. Returns 0 if there are no pending timeouts, 1 if a timeout was returned, or a LIBUSB_ERROR code on failure or timeout. .Pp .Ft void .Fn libusb_set_pollfd_notifiers "libusb_context *ctx" "libusb_pollfd_added_cb added_cb" "libusb_pollfd_removed_cb remove_cb" "void *user_data" Register notification functions for file descriptor additions/removals. These functions will be invoked for every new or removed file descriptor that libusb uses as an event source. .Pp .Ft const struct libusb_pollfd ** .Fn libusb_get_pollfds "libusb_context *ctx" Retrive a list of file descriptors that should be polled by your main loop as libusb event sources. Returns a NULL-terminated list on success or NULL on failure. .Pp .Ft int .Fn libusb_hotplug_register_callback "libusb_context *ctx" "libusb_hotplug_event events" "libusb_hotplug_flag flags" "int vendor_id" "int product_id" "int dev_class" "libusb_hotplug_callback_fn cb_fn" "void *user_data" "libusb_hotplug_callback_handle *handle" This function registers a hotplug filter. The .Fa events argument select which events makes the hotplug filter trigger. Available event values are LIBUSB_HOTPLUG_EVENT_DEVICE_ARRIVED and LIBUSB_HOTPLUG_EVENT_DEVICE_LEFT. One or more events must be specified. The .Fa vendor_id , .Fa product_id and .Fa dev_class arguments can be set to LIBUSB_HOTPLUG_MATCH_ANY to match any value in the USB device descriptor. Else the specified value is used for matching. If the .Fa flags argument is set to LIBUSB_HOTPLUG_ENUMERATE, all currently attached and matching USB devices will be passed to the hotplug filter, given by the .Fa cb_fn argument. Else the .Fa flags argument should be set to LIBUSB_HOTPLUG_NO_FLAGS. This function returns 0 upon success or a LIBUSB_ERROR code on failure. .Pp .Ft int .Fn libusb_hotplug_callback_fn "libusb_context *ctx" "libusb_device *device" "libusb_hotplug_event event" "void *user_data" The hotplug filter function. If this function returns non-zero, the filter is removed. Else the filter is kept and can receive more events. The .Fa user_data argument is the same as given when the filter was registered. The .Fa event argument can be either of LIBUSB_HOTPLUG_EVENT_DEVICE_ARRIVED or LIBUSB_HOTPLUG_EVENT_DEVICE_LEFT. .Pp .Ft void .Fn libusb_hotplug_deregister_callback "libusb_context *ctx" "libusb_hotplug_callback_handle handle" This function unregisters a hotplug filter. .Sh LIBUSB VERSION 0.1 COMPATIBILITY The library is also compliant with LibUSB version 0.1.12. .Pp .Fn usb_open .Fn usb_close .Fn usb_get_string .Fn usb_get_string_simple .Fn usb_get_descriptor_by_endpoint .Fn usb_get_descriptor .Fn usb_parse_descriptor .Fn usb_parse_configuration .Fn usb_destroy_configuration .Fn usb_fetch_and_parse_descriptors .Fn usb_bulk_write .Fn usb_bulk_read .Fn usb_interrupt_write .Fn usb_interrupt_read .Fn usb_control_msg .Fn usb_set_configuration .Fn usb_claim_interface .Fn usb_release_interface .Fn usb_set_altinterface .Fn usb_resetep .Fn usb_clear_halt .Fn usb_reset .Fn usb_strerror .Fn usb_init .Fn usb_set_debug .Fn usb_find_busses .Fn usb_find_devices .Fn usb_device .Fn usb_get_busses .Fn usb_check_connected .Fn usb_get_driver_np .Fn usb_detach_kernel_driver_np .Sh SEE ALSO .Xr libusb20 3 , .Xr usb 4 , .Xr usbconfig 8 , .Xr usbdump 8 .Pp .Pa http://libusb.sourceforge.net/ .Sh HISTORY .Nm support first appeared in .Fx 8.0 . Index: projects/vnet/lib/libusb/libusb.h =================================================================== --- projects/vnet/lib/libusb/libusb.h (revision 302157) +++ projects/vnet/lib/libusb/libusb.h (revision 302158) @@ -1,572 +1,579 @@ /* $FreeBSD$ */ /*- * Copyright (c) 2009 Sylvestre Gallon. 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. */ #ifndef __LIBUSB_H__ #define __LIBUSB_H__ #ifndef LIBUSB_GLOBAL_INCLUDE_FILE #include #include #include #endif #define LIBUSB_API_VERSION 0x01000102 #define LIBUSB_CALL #ifdef __cplusplus extern "C" { #endif #if 0 } /* indent fix */ #endif /* libusb enums */ enum libusb_class_code { LIBUSB_CLASS_PER_INTERFACE = 0, LIBUSB_CLASS_AUDIO = 1, LIBUSB_CLASS_COMM = 2, LIBUSB_CLASS_HID = 3, LIBUSB_CLASS_PTP = 6, LIBUSB_CLASS_IMAGE = 6, LIBUSB_CLASS_PRINTER = 7, LIBUSB_CLASS_MASS_STORAGE = 8, LIBUSB_CLASS_HUB = 9, LIBUSB_CLASS_DATA = 10, LIBUSB_CLASS_SMART_CARD = 11, LIBUSB_CLASS_CONTENT_SECURITY = 13, LIBUSB_CLASS_VIDEO = 14, LIBUSB_CLASS_PERSONAL_HEALTHCARE = 15, LIBUSB_CLASS_DIAGNOSTIC_DEVICE = 0xdc, LIBUSB_CLASS_WIRELESS = 0xe0, LIBUSB_CLASS_APPLICATION = 0xfe, LIBUSB_CLASS_VENDOR_SPEC = 0xff, }; enum libusb_descriptor_type { LIBUSB_DT_DEVICE = 0x01, LIBUSB_DT_CONFIG = 0x02, LIBUSB_DT_STRING = 0x03, LIBUSB_DT_INTERFACE = 0x04, LIBUSB_DT_ENDPOINT = 0x05, LIBUSB_DT_HID = 0x21, LIBUSB_DT_REPORT = 0x22, LIBUSB_DT_PHYSICAL = 0x23, LIBUSB_DT_HUB = 0x29, LIBUSB_DT_BOS = 0x0f, LIBUSB_DT_DEVICE_CAPABILITY = 0x10, LIBUSB_DT_SS_ENDPOINT_COMPANION = 0x30, }; enum libusb_device_capability_type { LIBUSB_WIRELESS_USB_DEVICE_CAPABILITY = 0x1, LIBUSB_USB_2_0_EXTENSION_DEVICE_CAPABILITY = 0x2, LIBUSB_SS_USB_DEVICE_CAPABILITY = 0x3, LIBUSB_CONTAINER_ID_DEVICE_CAPABILITY = 0x4, }; #define LIBUSB_DT_DEVICE_SIZE 18 #define LIBUSB_DT_CONFIG_SIZE 9 #define LIBUSB_DT_INTERFACE_SIZE 9 #define LIBUSB_DT_ENDPOINT_SIZE 7 #define LIBUSB_DT_ENDPOINT_AUDIO_SIZE 9 #define LIBUSB_DT_HUB_NONVAR_SIZE 7 #define LIBUSB_DT_SS_ENDPOINT_COMPANION_SIZE 6 #define LIBUSB_DT_BOS_SIZE 5 #define LIBUSB_USB_2_0_EXTENSION_DEVICE_CAPABILITY_SIZE 7 #define LIBUSB_SS_USB_DEVICE_CAPABILITY_SIZE 10 #define LIBUSB_BT_USB_2_0_EXTENSION_SIZE 7 #define LIBUSB_BT_SS_USB_DEVICE_CAPABILITY_SIZE 10 #define LIBUSB_BT_CONTAINER_ID_SIZE 20 #define LIBUSB_ENDPOINT_ADDRESS_MASK 0x0f #define LIBUSB_ENDPOINT_DIR_MASK 0x80 enum libusb_endpoint_direction { LIBUSB_ENDPOINT_IN = 0x80, LIBUSB_ENDPOINT_OUT = 0x00, }; #define LIBUSB_TRANSFER_TYPE_MASK 0x03 enum libusb_transfer_type { LIBUSB_TRANSFER_TYPE_CONTROL = 0, LIBUSB_TRANSFER_TYPE_ISOCHRONOUS = 1, LIBUSB_TRANSFER_TYPE_BULK = 2, LIBUSB_TRANSFER_TYPE_INTERRUPT = 3, }; enum libusb_standard_request { LIBUSB_REQUEST_GET_STATUS = 0x00, LIBUSB_REQUEST_CLEAR_FEATURE = 0x01, LIBUSB_REQUEST_SET_FEATURE = 0x03, LIBUSB_REQUEST_SET_ADDRESS = 0x05, LIBUSB_REQUEST_GET_DESCRIPTOR = 0x06, LIBUSB_REQUEST_SET_DESCRIPTOR = 0x07, LIBUSB_REQUEST_GET_CONFIGURATION = 0x08, LIBUSB_REQUEST_SET_CONFIGURATION = 0x09, LIBUSB_REQUEST_GET_INTERFACE = 0x0A, LIBUSB_REQUEST_SET_INTERFACE = 0x0B, LIBUSB_REQUEST_SYNCH_FRAME = 0x0C, LIBUSB_REQUEST_SET_SEL = 0x30, LIBUSB_REQUEST_SET_ISOCH_DELAY = 0x31, }; enum libusb_request_type { LIBUSB_REQUEST_TYPE_STANDARD = (0x00 << 5), LIBUSB_REQUEST_TYPE_CLASS = (0x01 << 5), LIBUSB_REQUEST_TYPE_VENDOR = (0x02 << 5), LIBUSB_REQUEST_TYPE_RESERVED = (0x03 << 5), }; enum libusb_request_recipient { LIBUSB_RECIPIENT_DEVICE = 0x00, LIBUSB_RECIPIENT_INTERFACE = 0x01, LIBUSB_RECIPIENT_ENDPOINT = 0x02, LIBUSB_RECIPIENT_OTHER = 0x03, }; #define LIBUSB_ISO_SYNC_TYPE_MASK 0x0C enum libusb_iso_sync_type { LIBUSB_ISO_SYNC_TYPE_NONE = 0, LIBUSB_ISO_SYNC_TYPE_ASYNC = 1, LIBUSB_ISO_SYNC_TYPE_ADAPTIVE = 2, LIBUSB_ISO_SYNC_TYPE_SYNC = 3, }; #define LIBUSB_ISO_USAGE_TYPE_MASK 0x30 enum libusb_iso_usage_type { LIBUSB_ISO_USAGE_TYPE_DATA = 0, LIBUSB_ISO_USAGE_TYPE_FEEDBACK = 1, LIBUSB_ISO_USAGE_TYPE_IMPLICIT = 2, }; enum libusb_bos_type { LIBUSB_BT_WIRELESS_USB_DEVICE_CAPABILITY = 1, LIBUSB_BT_USB_2_0_EXTENSION = 2, LIBUSB_BT_SS_USB_DEVICE_CAPABILITY = 3, LIBUSB_BT_CONTAINER_ID = 4, }; enum libusb_error { LIBUSB_SUCCESS = 0, LIBUSB_ERROR_IO = -1, LIBUSB_ERROR_INVALID_PARAM = -2, LIBUSB_ERROR_ACCESS = -3, LIBUSB_ERROR_NO_DEVICE = -4, LIBUSB_ERROR_NOT_FOUND = -5, LIBUSB_ERROR_BUSY = -6, LIBUSB_ERROR_TIMEOUT = -7, LIBUSB_ERROR_OVERFLOW = -8, LIBUSB_ERROR_PIPE = -9, LIBUSB_ERROR_INTERRUPTED = -10, LIBUSB_ERROR_NO_MEM = -11, LIBUSB_ERROR_NOT_SUPPORTED = -12, LIBUSB_ERROR_OTHER = -99, }; enum libusb_speed { LIBUSB_SPEED_UNKNOWN = 0, LIBUSB_SPEED_LOW = 1, LIBUSB_SPEED_FULL = 2, LIBUSB_SPEED_HIGH = 3, LIBUSB_SPEED_SUPER = 4, }; enum libusb_transfer_status { LIBUSB_TRANSFER_COMPLETED, LIBUSB_TRANSFER_ERROR, LIBUSB_TRANSFER_TIMED_OUT, LIBUSB_TRANSFER_CANCELLED, LIBUSB_TRANSFER_STALL, LIBUSB_TRANSFER_NO_DEVICE, LIBUSB_TRANSFER_OVERFLOW, }; enum libusb_transfer_flags { LIBUSB_TRANSFER_SHORT_NOT_OK = 1 << 0, LIBUSB_TRANSFER_FREE_BUFFER = 1 << 1, LIBUSB_TRANSFER_FREE_TRANSFER = 1 << 2, }; enum libusb_log_level { LIBUSB_LOG_LEVEL_NONE = 0, LIBUSB_LOG_LEVEL_ERROR, LIBUSB_LOG_LEVEL_WARNING, LIBUSB_LOG_LEVEL_INFO, LIBUSB_LOG_LEVEL_DEBUG }; /* XXX */ /* libusb_set_debug should take parameters from libusb_log_level * above according to * http://libusb.sourceforge.net/api-1.0/group__lib.html */ enum libusb_debug_level { LIBUSB_DEBUG_NO=0, LIBUSB_DEBUG_FUNCTION=1, LIBUSB_DEBUG_TRANSFER=2, }; #define LIBUSB_HOTPLUG_MATCH_ANY -1 typedef enum { LIBUSB_HOTPLUG_NO_FLAGS = 0, LIBUSB_HOTPLUG_ENUMERATE = 1 << 0, } libusb_hotplug_flag; typedef enum { LIBUSB_HOTPLUG_EVENT_DEVICE_ARRIVED = 1, LIBUSB_HOTPLUG_EVENT_DEVICE_LEFT = 2, } libusb_hotplug_event; /* libusb structures */ struct libusb_context; struct libusb_device; struct libusb_transfer; struct libusb_device_handle; struct libusb_hotplug_callback_handle; struct libusb_pollfd { int fd; short events; }; struct libusb_version { const uint16_t major; const uint16_t minor; const uint16_t micro; const uint16_t nano; const char *rc; const char *describe; }; typedef struct libusb_context libusb_context; typedef struct libusb_device libusb_device; typedef struct libusb_device_handle libusb_device_handle; typedef struct libusb_pollfd libusb_pollfd; typedef void (*libusb_pollfd_added_cb) (int fd, short events, void *user_data); typedef void (*libusb_pollfd_removed_cb) (int fd, void *user_data); typedef struct libusb_hotplug_callback_handle *libusb_hotplug_callback_handle; typedef struct libusb_device_descriptor { uint8_t bLength; uint8_t bDescriptorType; uint16_t bcdUSB; uint8_t bDeviceClass; uint8_t bDeviceSubClass; uint8_t bDeviceProtocol; uint8_t bMaxPacketSize0; uint16_t idVendor; uint16_t idProduct; uint16_t bcdDevice; uint8_t iManufacturer; uint8_t iProduct; uint8_t iSerialNumber; uint8_t bNumConfigurations; } libusb_device_descriptor; typedef struct libusb_endpoint_descriptor { uint8_t bLength; uint8_t bDescriptorType; uint8_t bEndpointAddress; uint8_t bmAttributes; uint16_t wMaxPacketSize; uint8_t bInterval; uint8_t bRefresh; uint8_t bSynchAddress; uint8_t *extra; int extra_length; } libusb_endpoint_descriptor __aligned(sizeof(void *)); typedef struct libusb_ss_endpoint_companion_descriptor { uint8_t bLength; uint8_t bDescriptorType; uint8_t bMaxBurst; uint8_t bmAttributes; uint16_t wBytesPerInterval; } libusb_ss_endpoint_companion_descriptor __aligned(sizeof(void *)); typedef struct libusb_interface_descriptor { uint8_t bLength; uint8_t bDescriptorType; uint8_t bInterfaceNumber; uint8_t bAlternateSetting; uint8_t bNumEndpoints; uint8_t bInterfaceClass; uint8_t bInterfaceSubClass; uint8_t bInterfaceProtocol; uint8_t iInterface; struct libusb_endpoint_descriptor *endpoint; uint8_t *extra; int extra_length; } libusb_interface_descriptor __aligned(sizeof(void *)); typedef struct libusb_interface { struct libusb_interface_descriptor *altsetting; int num_altsetting; } libusb_interface __aligned(sizeof(void *)); typedef struct libusb_config_descriptor { uint8_t bLength; uint8_t bDescriptorType; uint16_t wTotalLength; uint8_t bNumInterfaces; uint8_t bConfigurationValue; uint8_t iConfiguration; uint8_t bmAttributes; uint8_t MaxPower; struct libusb_interface *interface; uint8_t *extra; int extra_length; } libusb_config_descriptor __aligned(sizeof(void *)); typedef struct libusb_usb_2_0_device_capability_descriptor { uint8_t bLength; uint8_t bDescriptorType; uint8_t bDevCapabilityType; uint32_t bmAttributes; #define LIBUSB_USB_2_0_CAPABILITY_LPM_SUPPORT (1 << 1) } libusb_usb_2_0_device_capability_descriptor __aligned(sizeof(void *)); typedef struct libusb_ss_usb_device_capability_descriptor { uint8_t bLength; uint8_t bDescriptorType; uint8_t bDevCapabilityType; uint8_t bmAttributes; #define LIBUSB_SS_USB_CAPABILITY_LPM_SUPPORT (1 << 1) uint16_t wSpeedSupported; #define LIBUSB_CAPABILITY_LOW_SPEED_OPERATION (1) #define LIBUSB_CAPABILITY_FULL_SPEED_OPERATION (1 << 1) #define LIBUSB_CAPABILITY_HIGH_SPEED_OPERATION (1 << 2) #define LIBUSB_CAPABILITY_5GBPS_OPERATION (1 << 3) uint8_t bFunctionalitySupport; uint8_t bU1DevExitLat; uint16_t wU2DevExitLat; } libusb_ss_usb_device_capability_descriptor __aligned(sizeof(void *)); typedef struct libusb_bos_dev_capability_descriptor { uint8_t bLength; uint8_t bDescriptorType; uint8_t bDevCapabilityType; uint8_t dev_capability_data[0]; } libusb_bos_dev_capability_descriptor __aligned(sizeof(void *)); typedef struct libusb_bos_descriptor { uint8_t bLength; uint8_t bDescriptorType; uint16_t wTotalLength; uint8_t bNumDeviceCapabilities; struct libusb_usb_2_0_device_capability_descriptor *usb_2_0_ext_cap; struct libusb_ss_usb_device_capability_descriptor *ss_usb_cap; } libusb_bos_descriptor __aligned(sizeof(void *)); typedef struct libusb_usb_2_0_extension_descriptor { uint8_t bLength; uint8_t bDescriptorType; uint8_t bDevCapabilityType; uint32_t bmAttributes; } libusb_usb_2_0_extension_descriptor __aligned(sizeof(void *)); typedef struct libusb_container_id_descriptor { uint8_t bLength; uint8_t bDescriptorType; uint8_t bDevCapabilityType; uint8_t bReserved; uint8_t ContainerID[16]; } libusb_container_id_descriptor __aligned(sizeof(void *)); typedef struct libusb_control_setup { uint8_t bmRequestType; uint8_t bRequest; uint16_t wValue; uint16_t wIndex; uint16_t wLength; } libusb_control_setup; #define LIBUSB_CONTROL_SETUP_SIZE 8 /* bytes */ typedef struct libusb_iso_packet_descriptor { uint32_t length; uint32_t actual_length; enum libusb_transfer_status status; } libusb_iso_packet_descriptor __aligned(sizeof(void *)); typedef void (*libusb_transfer_cb_fn) (struct libusb_transfer *transfer); typedef struct libusb_transfer { libusb_device_handle *dev_handle; uint8_t flags; uint8_t endpoint; uint8_t type; uint32_t timeout; enum libusb_transfer_status status; int length; int actual_length; libusb_transfer_cb_fn callback; void *user_data; uint8_t *buffer; int num_iso_packets; struct libusb_iso_packet_descriptor iso_packet_desc[0]; } libusb_transfer __aligned(sizeof(void *)); /* Library initialisation */ void libusb_set_debug(libusb_context * ctx, int level); const struct libusb_version *libusb_get_version(void); const char *libusb_strerror(int code); const char *libusb_error_name(int code); int libusb_init(libusb_context ** context); void libusb_exit(struct libusb_context *ctx); /* Device handling and enumeration */ ssize_t libusb_get_device_list(libusb_context * ctx, libusb_device *** list); void libusb_free_device_list(libusb_device ** list, int unref_devices); uint8_t libusb_get_bus_number(libusb_device * dev); uint8_t libusb_get_port_number(libusb_device * dev); int libusb_get_port_numbers(libusb_device *dev, uint8_t *buf, uint8_t bufsize); int libusb_get_port_path(libusb_context *ctx, libusb_device *dev, uint8_t *buf, uint8_t bufsize); uint8_t libusb_get_device_address(libusb_device * dev); enum libusb_speed libusb_get_device_speed(libusb_device * dev); int libusb_clear_halt(libusb_device_handle *devh, uint8_t endpoint); int libusb_get_max_packet_size(libusb_device * dev, uint8_t endpoint); int libusb_get_max_iso_packet_size(libusb_device * dev, uint8_t endpoint); libusb_device *libusb_ref_device(libusb_device * dev); void libusb_unref_device(libusb_device * dev); int libusb_open(libusb_device * dev, libusb_device_handle ** devh); libusb_device_handle *libusb_open_device_with_vid_pid(libusb_context * ctx, uint16_t vendor_id, uint16_t product_id); void libusb_close(libusb_device_handle * devh); libusb_device *libusb_get_device(libusb_device_handle * devh); int libusb_get_configuration(libusb_device_handle * devh, int *config); int libusb_set_configuration(libusb_device_handle * devh, int configuration); int libusb_claim_interface(libusb_device_handle * devh, int interface_number); int libusb_release_interface(libusb_device_handle * devh, int interface_number); int libusb_reset_device(libusb_device_handle * devh); int libusb_check_connected(libusb_device_handle * devh); int libusb_kernel_driver_active(libusb_device_handle * devh, int interface); int libusb_get_driver_np(libusb_device_handle * devh, int interface, char *name, int namelen); int libusb_get_driver(libusb_device_handle * devh, int interface, char *name, int namelen); int libusb_detach_kernel_driver_np(libusb_device_handle * devh, int interface); int libusb_detach_kernel_driver(libusb_device_handle * devh, int interface); int libusb_attach_kernel_driver(libusb_device_handle * devh, int interface); int libusb_set_auto_detach_kernel_driver(libusb_device_handle *dev, int enable); int libusb_set_interface_alt_setting(libusb_device_handle * devh, int interface_number, int alternate_setting); /* USB Descriptors */ int libusb_get_device_descriptor(libusb_device * dev, struct libusb_device_descriptor *desc); int libusb_get_active_config_descriptor(libusb_device * dev, struct libusb_config_descriptor **config); int libusb_get_config_descriptor(libusb_device * dev, uint8_t config_index, struct libusb_config_descriptor **config); int libusb_get_config_descriptor_by_value(libusb_device * dev, uint8_t bConfigurationValue, struct libusb_config_descriptor **config); void libusb_free_config_descriptor(struct libusb_config_descriptor *config); int libusb_get_ss_endpoint_companion_descriptor(struct libusb_context *ctx, const struct libusb_endpoint_descriptor *endpoint, struct libusb_ss_endpoint_companion_descriptor **ep_comp); void libusb_free_ss_endpoint_companion_descriptor(struct libusb_ss_endpoint_companion_descriptor *ep_comp); int libusb_get_string_descriptor(libusb_device_handle * devh, uint8_t desc_index, uint16_t langid, unsigned char *data, int length); int libusb_get_string_descriptor_ascii(libusb_device_handle * devh, uint8_t desc_index, uint8_t *data, int length); int libusb_get_descriptor(libusb_device_handle * devh, uint8_t desc_type, uint8_t desc_index, uint8_t *data, int length); int libusb_parse_ss_endpoint_comp(const void *buf, int len, struct libusb_ss_endpoint_companion_descriptor **ep_comp); void libusb_free_ss_endpoint_comp(struct libusb_ss_endpoint_companion_descriptor *ep_comp); int libusb_parse_bos_descriptor(const void *buf, int len, struct libusb_bos_descriptor **bos); void libusb_free_bos_descriptor(struct libusb_bos_descriptor *bos); int libusb_get_bos_descriptor(libusb_device_handle *handle, struct libusb_bos_descriptor **bos); int libusb_get_usb_2_0_extension_descriptor(struct libusb_context *ctx, struct libusb_bos_dev_capability_descriptor *dev_cap, struct libusb_usb_2_0_extension_descriptor **usb_2_0_extension); void libusb_free_usb_2_0_extension_descriptor(struct libusb_usb_2_0_extension_descriptor *usb_2_0_extension); int libusb_get_ss_usb_device_capability_descriptor(struct libusb_context *ctx, struct libusb_bos_dev_capability_descriptor *dev_cap, struct libusb_ss_usb_device_capability_descriptor **ss_usb_device_capability); void libusb_free_ss_usb_device_capability_descriptor(struct libusb_ss_usb_device_capability_descriptor *ss_usb_device_capability); int libusb_get_container_id_descriptor(struct libusb_context *ctx, struct libusb_bos_dev_capability_descriptor *dev_cap, struct libusb_container_id_descriptor **container_id); void libusb_free_container_id_descriptor(struct libusb_container_id_descriptor *container_id); /* Asynchronous device I/O */ struct libusb_transfer *libusb_alloc_transfer(int iso_packets); void libusb_free_transfer(struct libusb_transfer *transfer); int libusb_submit_transfer(struct libusb_transfer *transfer); int libusb_cancel_transfer(struct libusb_transfer *transfer); uint8_t *libusb_get_iso_packet_buffer(struct libusb_transfer *transfer, uint32_t index); uint8_t *libusb_get_iso_packet_buffer_simple(struct libusb_transfer *transfer, uint32_t index); void libusb_set_iso_packet_lengths(struct libusb_transfer *transfer, uint32_t length); uint8_t *libusb_control_transfer_get_data(struct libusb_transfer *transfer); struct libusb_control_setup *libusb_control_transfer_get_setup(struct libusb_transfer *transfer); void libusb_fill_control_setup(uint8_t *buf, uint8_t bmRequestType, uint8_t bRequest, uint16_t wValue, uint16_t wIndex, uint16_t wLength); void libusb_fill_control_transfer(struct libusb_transfer *transfer, libusb_device_handle *devh, uint8_t *buf, libusb_transfer_cb_fn callback, void *user_data, uint32_t timeout); void libusb_fill_bulk_transfer(struct libusb_transfer *transfer, libusb_device_handle *devh, uint8_t endpoint, uint8_t *buf, int length, libusb_transfer_cb_fn callback, void *user_data, uint32_t timeout); void libusb_fill_interrupt_transfer(struct libusb_transfer *transfer, libusb_device_handle *devh, uint8_t endpoint, uint8_t *buf, int length, libusb_transfer_cb_fn callback, void *user_data, uint32_t timeout); void libusb_fill_iso_transfer(struct libusb_transfer *transfer, libusb_device_handle *devh, uint8_t endpoint, uint8_t *buf, int length, int npacket, libusb_transfer_cb_fn callback, void *user_data, uint32_t timeout); /* Polling and timing */ int libusb_try_lock_events(libusb_context * ctx); void libusb_lock_events(libusb_context * ctx); void libusb_unlock_events(libusb_context * ctx); int libusb_event_handling_ok(libusb_context * ctx); int libusb_event_handler_active(libusb_context * ctx); void libusb_lock_event_waiters(libusb_context * ctx); void libusb_unlock_event_waiters(libusb_context * ctx); int libusb_wait_for_event(libusb_context * ctx, struct timeval *tv); int libusb_handle_events_timeout_completed(libusb_context * ctx, struct timeval *tv, int *completed); int libusb_handle_events_completed(libusb_context * ctx, int *completed); int libusb_handle_events_timeout(libusb_context * ctx, struct timeval *tv); int libusb_handle_events(libusb_context * ctx); int libusb_handle_events_locked(libusb_context * ctx, struct timeval *tv); int libusb_get_next_timeout(libusb_context * ctx, struct timeval *tv); void libusb_set_pollfd_notifiers(libusb_context * ctx, libusb_pollfd_added_cb added_cb, libusb_pollfd_removed_cb removed_cb, void *user_data); const struct libusb_pollfd **libusb_get_pollfds(libusb_context * ctx); /* Synchronous device I/O */ int libusb_control_transfer(libusb_device_handle * devh, uint8_t bmRequestType, uint8_t bRequest, uint16_t wValue, uint16_t wIndex, uint8_t *data, uint16_t wLength, uint32_t timeout); int libusb_bulk_transfer(libusb_device_handle * devh, uint8_t endpoint, uint8_t *data, int length, int *transferred, uint32_t timeout); int libusb_interrupt_transfer(libusb_device_handle * devh, uint8_t endpoint, uint8_t *data, int length, int *transferred, uint32_t timeout); /* Byte-order */ uint16_t libusb_cpu_to_le16(uint16_t x); uint16_t libusb_le16_to_cpu(uint16_t x); /* Hotplug support */ typedef int (*libusb_hotplug_callback_fn)(libusb_context *ctx, libusb_device *device, libusb_hotplug_event event, void *user_data); int libusb_hotplug_register_callback(libusb_context *ctx, libusb_hotplug_event events, libusb_hotplug_flag flags, int vendor_id, int product_id, int dev_class, libusb_hotplug_callback_fn cb_fn, void *user_data, libusb_hotplug_callback_handle *handle); void libusb_hotplug_deregister_callback(libusb_context *ctx, libusb_hotplug_callback_handle handle); +/* Streams support */ + +int libusb_alloc_streams(libusb_device_handle *dev, uint32_t num_streams, unsigned char *endpoints, int num_endpoints); +int libusb_free_streams(libusb_device_handle *dev, unsigned char *endpoints, int num_endpoints); +void libusb_transfer_set_stream_id(struct libusb_transfer *transfer, uint32_t stream_id); +uint32_t libusb_transfer_get_stream_id(struct libusb_transfer *transfer); + #if 0 { /* indent fix */ #endif #ifdef __cplusplus } #endif #endif /* __LIBUSB_H__ */ Index: projects/vnet/lib/libusb/libusb10.c =================================================================== --- projects/vnet/lib/libusb/libusb10.c (revision 302157) +++ projects/vnet/lib/libusb/libusb10.c (revision 302158) @@ -1,1655 +1,1656 @@ /* $FreeBSD$ */ /*- * Copyright (c) 2009 Sylvestre Gallon. All rights reserved. * Copyright (c) 2009 Hans Petter Selasky. 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. */ #ifdef LIBUSB_GLOBAL_INCLUDE_FILE #include LIBUSB_GLOBAL_INCLUDE_FILE #else #include #include #include #include #include #include #include #include #include #include #include #include #include #endif #define libusb_device_handle libusb20_device #include "libusb20.h" #include "libusb20_desc.h" #include "libusb20_int.h" #include "libusb.h" #include "libusb10.h" #define LIBUSB_NUM_SW_ENDPOINTS (16 * 4) static pthread_mutex_t default_context_lock = PTHREAD_MUTEX_INITIALIZER; struct libusb_context *usbi_default_context = NULL; /* Prototypes */ static struct libusb20_transfer *libusb10_get_transfer(struct libusb20_device *, uint8_t, uint8_t); static int libusb10_get_buffsize(struct libusb20_device *, libusb_transfer *); static int libusb10_convert_error(uint8_t status); static void libusb10_complete_transfer(struct libusb20_transfer *, struct libusb_super_transfer *, int); static void libusb10_isoc_proxy(struct libusb20_transfer *); static void libusb10_bulk_intr_proxy(struct libusb20_transfer *); static void libusb10_ctrl_proxy(struct libusb20_transfer *); static void libusb10_submit_transfer_sub(struct libusb20_device *, uint8_t); /* Library initialisation / deinitialisation */ static const struct libusb_version libusb_version = { .major = 1, .minor = 0, .micro = 0, .nano = 2016, .rc = "", .describe = "http://www.freebsd.org" }; const struct libusb_version * libusb_get_version(void) { return (&libusb_version); } void libusb_set_debug(libusb_context *ctx, int level) { ctx = GET_CONTEXT(ctx); if (ctx) ctx->debug = level; } static void libusb_set_nonblocking(int f) { int flags; /* * We ignore any failures in this function, hence the * non-blocking flag is not critical to the operation of * libUSB. We use F_GETFL and F_SETFL to be compatible with * Linux. */ flags = fcntl(f, F_GETFL, NULL); if (flags == -1) return; flags |= O_NONBLOCK; fcntl(f, F_SETFL, flags); } int libusb_init(libusb_context **context) { struct libusb_context *ctx; pthread_condattr_t attr; char *debug; int ret; ctx = malloc(sizeof(*ctx)); if (!ctx) return (LIBUSB_ERROR_INVALID_PARAM); memset(ctx, 0, sizeof(*ctx)); debug = getenv("LIBUSB_DEBUG"); if (debug != NULL) { ctx->debug = atoi(debug); if (ctx->debug != 0) ctx->debug_fixed = 1; } TAILQ_INIT(&ctx->pollfds); TAILQ_INIT(&ctx->tr_done); TAILQ_INIT(&ctx->hotplug_cbh); TAILQ_INIT(&ctx->hotplug_devs); if (pthread_mutex_init(&ctx->ctx_lock, NULL) != 0) { free(ctx); return (LIBUSB_ERROR_NO_MEM); } if (pthread_mutex_init(&ctx->hotplug_lock, NULL) != 0) { pthread_mutex_destroy(&ctx->ctx_lock); free(ctx); return (LIBUSB_ERROR_NO_MEM); } if (pthread_condattr_init(&attr) != 0) { pthread_mutex_destroy(&ctx->ctx_lock); pthread_mutex_destroy(&ctx->hotplug_lock); free(ctx); return (LIBUSB_ERROR_NO_MEM); } if (pthread_condattr_setclock(&attr, CLOCK_MONOTONIC) != 0) { pthread_mutex_destroy(&ctx->ctx_lock); pthread_mutex_destroy(&ctx->hotplug_lock); pthread_condattr_destroy(&attr); free(ctx); return (LIBUSB_ERROR_OTHER); } if (pthread_cond_init(&ctx->ctx_cond, &attr) != 0) { pthread_mutex_destroy(&ctx->ctx_lock); pthread_mutex_destroy(&ctx->hotplug_lock); pthread_condattr_destroy(&attr); free(ctx); return (LIBUSB_ERROR_NO_MEM); } pthread_condattr_destroy(&attr); ctx->ctx_handler = NO_THREAD; ctx->hotplug_handler = NO_THREAD; ret = pipe(ctx->ctrl_pipe); if (ret < 0) { pthread_mutex_destroy(&ctx->ctx_lock); pthread_mutex_destroy(&ctx->hotplug_lock); pthread_cond_destroy(&ctx->ctx_cond); free(ctx); return (LIBUSB_ERROR_OTHER); } /* set non-blocking mode on the control pipe to avoid deadlock */ libusb_set_nonblocking(ctx->ctrl_pipe[0]); libusb_set_nonblocking(ctx->ctrl_pipe[1]); libusb10_add_pollfd(ctx, &ctx->ctx_poll, NULL, ctx->ctrl_pipe[0], POLLIN); pthread_mutex_lock(&default_context_lock); if (usbi_default_context == NULL) { usbi_default_context = ctx; } pthread_mutex_unlock(&default_context_lock); if (context) *context = ctx; DPRINTF(ctx, LIBUSB_DEBUG_FUNCTION, "libusb_init complete"); return (0); } void libusb_exit(libusb_context *ctx) { ctx = GET_CONTEXT(ctx); if (ctx == NULL) return; /* stop hotplug thread, if any */ if (ctx->hotplug_handler != NO_THREAD) { pthread_t td; void *ptr; HOTPLUG_LOCK(ctx); td = ctx->hotplug_handler; ctx->hotplug_handler = NO_THREAD; HOTPLUG_UNLOCK(ctx); pthread_join(td, &ptr); } /* XXX cleanup devices */ libusb10_remove_pollfd(ctx, &ctx->ctx_poll); close(ctx->ctrl_pipe[0]); close(ctx->ctrl_pipe[1]); pthread_mutex_destroy(&ctx->ctx_lock); pthread_mutex_destroy(&ctx->hotplug_lock); pthread_cond_destroy(&ctx->ctx_cond); pthread_mutex_lock(&default_context_lock); if (ctx == usbi_default_context) { usbi_default_context = NULL; } pthread_mutex_unlock(&default_context_lock); free(ctx); } /* Device handling and initialisation. */ ssize_t libusb_get_device_list(libusb_context *ctx, libusb_device ***list) { struct libusb20_backend *usb_backend; struct libusb20_device *pdev; struct libusb_device *dev; int i; ctx = GET_CONTEXT(ctx); if (ctx == NULL) return (LIBUSB_ERROR_INVALID_PARAM); if (list == NULL) return (LIBUSB_ERROR_INVALID_PARAM); usb_backend = libusb20_be_alloc_default(); if (usb_backend == NULL) return (LIBUSB_ERROR_NO_MEM); /* figure out how many USB devices are present */ pdev = NULL; i = 0; while ((pdev = libusb20_be_device_foreach(usb_backend, pdev))) i++; /* allocate device pointer list */ *list = malloc((i + 1) * sizeof(void *)); if (*list == NULL) { libusb20_be_free(usb_backend); return (LIBUSB_ERROR_NO_MEM); } /* create libusb v1.0 compliant devices */ i = 0; while ((pdev = libusb20_be_device_foreach(usb_backend, NULL))) { dev = malloc(sizeof(*dev)); if (dev == NULL) { while (i != 0) { libusb_unref_device((*list)[i - 1]); i--; } free(*list); *list = NULL; libusb20_be_free(usb_backend); return (LIBUSB_ERROR_NO_MEM); } /* get device into libUSB v1.0 list */ libusb20_be_dequeue_device(usb_backend, pdev); memset(dev, 0, sizeof(*dev)); /* init transfer queues */ TAILQ_INIT(&dev->tr_head); /* set context we belong to */ dev->ctx = ctx; /* link together the two structures */ dev->os_priv = pdev; pdev->privLuData = dev; (*list)[i] = libusb_ref_device(dev); i++; } (*list)[i] = NULL; libusb20_be_free(usb_backend); return (i); } void libusb_free_device_list(libusb_device **list, int unref_devices) { int i; if (list == NULL) return; /* be NULL safe */ if (unref_devices) { for (i = 0; list[i] != NULL; i++) libusb_unref_device(list[i]); } free(list); } uint8_t libusb_get_bus_number(libusb_device *dev) { if (dev == NULL) return (0); /* should not happen */ return (libusb20_dev_get_bus_number(dev->os_priv)); } uint8_t libusb_get_port_number(libusb_device *dev) { if (dev == NULL) return (0); /* should not happen */ return (libusb20_dev_get_parent_port(dev->os_priv)); } int libusb_get_port_numbers(libusb_device *dev, uint8_t *buf, uint8_t bufsize) { return (libusb20_dev_get_port_path(dev->os_priv, buf, bufsize)); } int libusb_get_port_path(libusb_context *ctx, libusb_device *dev, uint8_t *buf, uint8_t bufsize) { return (libusb20_dev_get_port_path(dev->os_priv, buf, bufsize)); } uint8_t libusb_get_device_address(libusb_device *dev) { if (dev == NULL) return (0); /* should not happen */ return (libusb20_dev_get_address(dev->os_priv)); } enum libusb_speed libusb_get_device_speed(libusb_device *dev) { if (dev == NULL) return (LIBUSB_SPEED_UNKNOWN); /* should not happen */ switch (libusb20_dev_get_speed(dev->os_priv)) { case LIBUSB20_SPEED_LOW: return (LIBUSB_SPEED_LOW); case LIBUSB20_SPEED_FULL: return (LIBUSB_SPEED_FULL); case LIBUSB20_SPEED_HIGH: return (LIBUSB_SPEED_HIGH); case LIBUSB20_SPEED_SUPER: return (LIBUSB_SPEED_SUPER); default: break; } return (LIBUSB_SPEED_UNKNOWN); } int libusb_get_max_packet_size(libusb_device *dev, uint8_t endpoint) { struct libusb_config_descriptor *pdconf; struct libusb_interface *pinf; struct libusb_interface_descriptor *pdinf; struct libusb_endpoint_descriptor *pdend; int i; int j; int k; int ret; if (dev == NULL) return (LIBUSB_ERROR_NO_DEVICE); ret = libusb_get_active_config_descriptor(dev, &pdconf); if (ret < 0) return (ret); ret = LIBUSB_ERROR_NOT_FOUND; for (i = 0; i < pdconf->bNumInterfaces; i++) { pinf = &pdconf->interface[i]; for (j = 0; j < pinf->num_altsetting; j++) { pdinf = &pinf->altsetting[j]; for (k = 0; k < pdinf->bNumEndpoints; k++) { pdend = &pdinf->endpoint[k]; if (pdend->bEndpointAddress == endpoint) { ret = pdend->wMaxPacketSize; goto out; } } } } out: libusb_free_config_descriptor(pdconf); return (ret); } int libusb_get_max_iso_packet_size(libusb_device *dev, uint8_t endpoint) { int multiplier; int ret; ret = libusb_get_max_packet_size(dev, endpoint); switch (libusb20_dev_get_speed(dev->os_priv)) { case LIBUSB20_SPEED_LOW: case LIBUSB20_SPEED_FULL: break; default: if (ret > -1) { multiplier = (1 + ((ret >> 11) & 3)); if (multiplier > 3) multiplier = 3; ret = (ret & 0x7FF) * multiplier; } break; } return (ret); } libusb_device * libusb_ref_device(libusb_device *dev) { if (dev == NULL) return (NULL); /* be NULL safe */ CTX_LOCK(dev->ctx); dev->refcnt++; CTX_UNLOCK(dev->ctx); return (dev); } void libusb_unref_device(libusb_device *dev) { if (dev == NULL) return; /* be NULL safe */ CTX_LOCK(dev->ctx); dev->refcnt--; CTX_UNLOCK(dev->ctx); if (dev->refcnt == 0) { libusb20_dev_free(dev->os_priv); free(dev); } } int libusb_open(libusb_device *dev, libusb_device_handle **devh) { libusb_context *ctx = dev->ctx; struct libusb20_device *pdev = dev->os_priv; uint8_t dummy; int err; if (devh == NULL) return (LIBUSB_ERROR_INVALID_PARAM); /* set default device handle value */ *devh = NULL; dev = libusb_ref_device(dev); if (dev == NULL) return (LIBUSB_ERROR_INVALID_PARAM); err = libusb20_dev_open(pdev, LIBUSB_NUM_SW_ENDPOINTS); if (err) { libusb_unref_device(dev); return (LIBUSB_ERROR_NO_MEM); } libusb10_add_pollfd(ctx, &dev->dev_poll, pdev, libusb20_dev_get_fd(pdev), POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM); /* make sure our event loop detects the new device */ dummy = 0; err = write(ctx->ctrl_pipe[1], &dummy, sizeof(dummy)); if (err < (int)sizeof(dummy)) { /* ignore error, if any */ DPRINTF(ctx, LIBUSB_DEBUG_FUNCTION, "libusb_open write failed!"); } *devh = pdev; return (0); } libusb_device_handle * libusb_open_device_with_vid_pid(libusb_context *ctx, uint16_t vendor_id, uint16_t product_id) { struct libusb_device **devs; struct libusb20_device *pdev; struct LIBUSB20_DEVICE_DESC_DECODED *pdesc; int i; int j; ctx = GET_CONTEXT(ctx); if (ctx == NULL) return (NULL); /* be NULL safe */ DPRINTF(ctx, LIBUSB_DEBUG_FUNCTION, "libusb_open_device_width_vid_pid enter"); if ((i = libusb_get_device_list(ctx, &devs)) < 0) return (NULL); pdev = NULL; for (j = 0; j < i; j++) { struct libusb20_device *tdev; tdev = devs[j]->os_priv; pdesc = libusb20_dev_get_device_desc(tdev); /* * NOTE: The USB library will automatically swap the * fields in the device descriptor to be of host * endian type! */ if (pdesc->idVendor == vendor_id && pdesc->idProduct == product_id) { libusb_open(devs[j], &pdev); break; } } libusb_free_device_list(devs, 1); DPRINTF(ctx, LIBUSB_DEBUG_FUNCTION, "libusb_open_device_width_vid_pid leave"); return (pdev); } void libusb_close(struct libusb20_device *pdev) { libusb_context *ctx; struct libusb_device *dev; uint8_t dummy; int err; if (pdev == NULL) return; /* be NULL safe */ dev = libusb_get_device(pdev); ctx = dev->ctx; libusb10_remove_pollfd(ctx, &dev->dev_poll); libusb20_dev_close(pdev); /* unref will free the "pdev" when the refcount reaches zero */ libusb_unref_device(dev); /* make sure our event loop detects the closed device */ dummy = 0; err = write(ctx->ctrl_pipe[1], &dummy, sizeof(dummy)); if (err < (int)sizeof(dummy)) { /* ignore error, if any */ DPRINTF(ctx, LIBUSB_DEBUG_FUNCTION, "libusb_close write failed!"); } } libusb_device * libusb_get_device(struct libusb20_device *pdev) { if (pdev == NULL) return (NULL); return ((libusb_device *)pdev->privLuData); } int libusb_get_configuration(struct libusb20_device *pdev, int *config) { struct libusb20_config *pconf; if (pdev == NULL || config == NULL) return (LIBUSB_ERROR_INVALID_PARAM); pconf = libusb20_dev_alloc_config(pdev, libusb20_dev_get_config_index(pdev)); if (pconf == NULL) return (LIBUSB_ERROR_NO_MEM); *config = pconf->desc.bConfigurationValue; free(pconf); return (0); } int libusb_set_configuration(struct libusb20_device *pdev, int configuration) { struct libusb20_config *pconf; struct libusb_device *dev; int err; uint8_t i; dev = libusb_get_device(pdev); if (dev == NULL) return (LIBUSB_ERROR_INVALID_PARAM); if (configuration < 1) { /* unconfigure */ i = 255; } else { for (i = 0; i != 255; i++) { uint8_t found; pconf = libusb20_dev_alloc_config(pdev, i); if (pconf == NULL) return (LIBUSB_ERROR_INVALID_PARAM); found = (pconf->desc.bConfigurationValue == configuration); free(pconf); if (found) goto set_config; } return (LIBUSB_ERROR_INVALID_PARAM); } set_config: libusb10_cancel_all_transfer(dev); libusb10_remove_pollfd(dev->ctx, &dev->dev_poll); err = libusb20_dev_set_config_index(pdev, i); libusb10_add_pollfd(dev->ctx, &dev->dev_poll, pdev, libusb20_dev_get_fd(pdev), POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM); return (err ? LIBUSB_ERROR_INVALID_PARAM : 0); } int libusb_claim_interface(struct libusb20_device *pdev, int interface_number) { libusb_device *dev; int err = 0; dev = libusb_get_device(pdev); if (dev == NULL) return (LIBUSB_ERROR_INVALID_PARAM); if (interface_number < 0 || interface_number > 31) return (LIBUSB_ERROR_INVALID_PARAM); if (pdev->auto_detach != 0) { err = libusb_detach_kernel_driver(pdev, interface_number); if (err != 0) goto done; } CTX_LOCK(dev->ctx); dev->claimed_interfaces |= (1 << interface_number); CTX_UNLOCK(dev->ctx); done: return (err); } int libusb_release_interface(struct libusb20_device *pdev, int interface_number) { libusb_device *dev; int err = 0; dev = libusb_get_device(pdev); if (dev == NULL) return (LIBUSB_ERROR_INVALID_PARAM); if (interface_number < 0 || interface_number > 31) return (LIBUSB_ERROR_INVALID_PARAM); if (pdev->auto_detach != 0) { err = libusb_attach_kernel_driver(pdev, interface_number); if (err != 0) goto done; } CTX_LOCK(dev->ctx); if (!(dev->claimed_interfaces & (1 << interface_number))) err = LIBUSB_ERROR_NOT_FOUND; else dev->claimed_interfaces &= ~(1 << interface_number); CTX_UNLOCK(dev->ctx); done: return (err); } int libusb_set_interface_alt_setting(struct libusb20_device *pdev, int interface_number, int alternate_setting) { libusb_device *dev; int err = 0; dev = libusb_get_device(pdev); if (dev == NULL) return (LIBUSB_ERROR_INVALID_PARAM); if (interface_number < 0 || interface_number > 31) return (LIBUSB_ERROR_INVALID_PARAM); CTX_LOCK(dev->ctx); if (!(dev->claimed_interfaces & (1 << interface_number))) err = LIBUSB_ERROR_NOT_FOUND; CTX_UNLOCK(dev->ctx); if (err) return (err); libusb10_cancel_all_transfer(dev); libusb10_remove_pollfd(dev->ctx, &dev->dev_poll); err = libusb20_dev_set_alt_index(pdev, interface_number, alternate_setting); libusb10_add_pollfd(dev->ctx, &dev->dev_poll, pdev, libusb20_dev_get_fd(pdev), POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM); return (err ? LIBUSB_ERROR_OTHER : 0); } static struct libusb20_transfer * libusb10_get_transfer(struct libusb20_device *pdev, uint8_t endpoint, uint8_t xfer_index) { xfer_index &= 1; /* double buffering */ xfer_index |= (endpoint & LIBUSB20_ENDPOINT_ADDRESS_MASK) * 4; if (endpoint & LIBUSB20_ENDPOINT_DIR_MASK) { /* this is an IN endpoint */ xfer_index |= 2; } return (libusb20_tr_get_pointer(pdev, xfer_index)); } int libusb_clear_halt(struct libusb20_device *pdev, uint8_t endpoint) { struct libusb20_transfer *xfer; struct libusb_device *dev; int err; xfer = libusb10_get_transfer(pdev, endpoint, 0); if (xfer == NULL) return (LIBUSB_ERROR_INVALID_PARAM); dev = libusb_get_device(pdev); if (dev == NULL) return (LIBUSB_ERROR_INVALID_PARAM); CTX_LOCK(dev->ctx); err = libusb20_tr_open(xfer, 0, 1, endpoint); CTX_UNLOCK(dev->ctx); if (err != 0 && err != LIBUSB20_ERROR_BUSY) return (LIBUSB_ERROR_OTHER); libusb20_tr_clear_stall_sync(xfer); /* check if we opened the transfer */ if (err == 0) { CTX_LOCK(dev->ctx); libusb20_tr_close(xfer); CTX_UNLOCK(dev->ctx); } return (0); /* success */ } int libusb_reset_device(struct libusb20_device *pdev) { libusb_device *dev; int err; dev = libusb_get_device(pdev); if (dev == NULL) return (LIBUSB_ERROR_INVALID_PARAM); libusb10_cancel_all_transfer(dev); libusb10_remove_pollfd(dev->ctx, &dev->dev_poll); err = libusb20_dev_reset(pdev); libusb10_add_pollfd(dev->ctx, &dev->dev_poll, pdev, libusb20_dev_get_fd(pdev), POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM); return (err ? LIBUSB_ERROR_OTHER : 0); } int libusb_check_connected(struct libusb20_device *pdev) { libusb_device *dev; int err; dev = libusb_get_device(pdev); if (dev == NULL) return (LIBUSB_ERROR_INVALID_PARAM); err = libusb20_dev_check_connected(pdev); return (err ? LIBUSB_ERROR_NO_DEVICE : 0); } int libusb_kernel_driver_active(struct libusb20_device *pdev, int interface) { if (pdev == NULL) return (LIBUSB_ERROR_INVALID_PARAM); if (libusb20_dev_kernel_driver_active(pdev, interface)) return (0); /* no kernel driver is active */ else return (1); /* kernel driver is active */ } int libusb_get_driver_np(struct libusb20_device *pdev, int interface, char *name, int namelen) { return (libusb_get_driver(pdev, interface, name, namelen)); } int libusb_get_driver(struct libusb20_device *pdev, int interface, char *name, int namelen) { char *ptr; int err; if (pdev == NULL) return (LIBUSB_ERROR_INVALID_PARAM); if (namelen < 1) return (LIBUSB_ERROR_INVALID_PARAM); if (namelen > 255) namelen = 255; err = libusb20_dev_get_iface_desc( pdev, interface, name, namelen); if (err != 0) return (LIBUSB_ERROR_OTHER); /* we only want the driver name */ ptr = strstr(name, ":"); if (ptr != NULL) *ptr = 0; return (0); } int libusb_detach_kernel_driver_np(struct libusb20_device *pdev, int interface) { return (libusb_detach_kernel_driver(pdev, interface)); } int libusb_detach_kernel_driver(struct libusb20_device *pdev, int interface) { int err; if (pdev == NULL) return (LIBUSB_ERROR_INVALID_PARAM); err = libusb20_dev_detach_kernel_driver( pdev, interface); return (err ? LIBUSB_ERROR_OTHER : 0); } int libusb_attach_kernel_driver(struct libusb20_device *pdev, int interface) { if (pdev == NULL) return (LIBUSB_ERROR_INVALID_PARAM); /* stub - currently not supported by libusb20 */ return (0); } int libusb_set_auto_detach_kernel_driver(libusb_device_handle *dev, int enable) { dev->auto_detach = (enable ? 1 : 0); return (0); } /* Asynchronous device I/O */ struct libusb_transfer * libusb_alloc_transfer(int iso_packets) { struct libusb_transfer *uxfer; struct libusb_super_transfer *sxfer; int len; len = sizeof(struct libusb_transfer) + sizeof(struct libusb_super_transfer) + (iso_packets * sizeof(libusb_iso_packet_descriptor)); sxfer = malloc(len); if (sxfer == NULL) return (NULL); memset(sxfer, 0, len); uxfer = (struct libusb_transfer *)( ((uint8_t *)sxfer) + sizeof(*sxfer)); /* set default value */ uxfer->num_iso_packets = iso_packets; return (uxfer); } void libusb_free_transfer(struct libusb_transfer *uxfer) { struct libusb_super_transfer *sxfer; if (uxfer == NULL) return; /* be NULL safe */ /* check if we should free the transfer buffer */ if (uxfer->flags & LIBUSB_TRANSFER_FREE_BUFFER) free(uxfer->buffer); sxfer = (struct libusb_super_transfer *)( (uint8_t *)uxfer - sizeof(*sxfer)); free(sxfer); } static uint32_t libusb10_get_maxframe(struct libusb20_device *pdev, libusb_transfer *xfer) { uint32_t ret; switch (xfer->type) { case LIBUSB_TRANSFER_TYPE_ISOCHRONOUS: ret = 60 | LIBUSB20_MAX_FRAME_PRE_SCALE; /* 60ms */ break; case LIBUSB_TRANSFER_TYPE_CONTROL: ret = 2; break; default: ret = 1; break; } return (ret); } static int libusb10_get_buffsize(struct libusb20_device *pdev, libusb_transfer *xfer) { int ret; int usb_speed; usb_speed = libusb20_dev_get_speed(pdev); switch (xfer->type) { case LIBUSB_TRANSFER_TYPE_ISOCHRONOUS: ret = 0; /* kernel will auto-select */ break; case LIBUSB_TRANSFER_TYPE_CONTROL: ret = 1024; break; default: switch (usb_speed) { case LIBUSB20_SPEED_LOW: ret = 256; break; case LIBUSB20_SPEED_FULL: ret = 4096; break; case LIBUSB20_SPEED_SUPER: ret = 65536; break; default: ret = 16384; break; } break; } return (ret); } static int libusb10_convert_error(uint8_t status) { ; /* indent fix */ switch (status) { case LIBUSB20_TRANSFER_START: case LIBUSB20_TRANSFER_COMPLETED: return (LIBUSB_TRANSFER_COMPLETED); case LIBUSB20_TRANSFER_OVERFLOW: return (LIBUSB_TRANSFER_OVERFLOW); case LIBUSB20_TRANSFER_NO_DEVICE: return (LIBUSB_TRANSFER_NO_DEVICE); case LIBUSB20_TRANSFER_STALL: return (LIBUSB_TRANSFER_STALL); case LIBUSB20_TRANSFER_CANCELLED: return (LIBUSB_TRANSFER_CANCELLED); case LIBUSB20_TRANSFER_TIMED_OUT: return (LIBUSB_TRANSFER_TIMED_OUT); default: return (LIBUSB_TRANSFER_ERROR); } } /* This function must be called locked */ static void libusb10_complete_transfer(struct libusb20_transfer *pxfer, struct libusb_super_transfer *sxfer, int status) { struct libusb_transfer *uxfer; struct libusb_device *dev; uxfer = (struct libusb_transfer *)( ((uint8_t *)sxfer) + sizeof(*sxfer)); if (pxfer != NULL) libusb20_tr_set_priv_sc1(pxfer, NULL); /* set transfer status */ uxfer->status = status; /* update super transfer state */ sxfer->state = LIBUSB_SUPER_XFER_ST_NONE; dev = libusb_get_device(uxfer->dev_handle); TAILQ_INSERT_TAIL(&dev->ctx->tr_done, sxfer, entry); } /* This function must be called locked */ static void libusb10_isoc_proxy(struct libusb20_transfer *pxfer) { struct libusb_super_transfer *sxfer; struct libusb_transfer *uxfer; uint32_t actlen; uint16_t iso_packets; uint16_t i; uint8_t status; status = libusb20_tr_get_status(pxfer); sxfer = libusb20_tr_get_priv_sc1(pxfer); actlen = libusb20_tr_get_actual_length(pxfer); iso_packets = libusb20_tr_get_max_frames(pxfer); if (sxfer == NULL) return; /* cancelled - nothing to do */ uxfer = (struct libusb_transfer *)( ((uint8_t *)sxfer) + sizeof(*sxfer)); if (iso_packets > uxfer->num_iso_packets) iso_packets = uxfer->num_iso_packets; if (iso_packets == 0) return; /* nothing to do */ /* make sure that the number of ISOCHRONOUS packets is valid */ uxfer->num_iso_packets = iso_packets; switch (status) { case LIBUSB20_TRANSFER_COMPLETED: /* update actual length */ uxfer->actual_length = actlen; for (i = 0; i != iso_packets; i++) { uxfer->iso_packet_desc[i].actual_length = libusb20_tr_get_length(pxfer, i); } libusb10_complete_transfer(pxfer, sxfer, LIBUSB_TRANSFER_COMPLETED); break; case LIBUSB20_TRANSFER_START: /* setup length(s) */ actlen = 0; for (i = 0; i != iso_packets; i++) { libusb20_tr_setup_isoc(pxfer, &uxfer->buffer[actlen], uxfer->iso_packet_desc[i].length, i); actlen += uxfer->iso_packet_desc[i].length; } /* no remainder */ sxfer->rem_len = 0; libusb20_tr_set_total_frames(pxfer, iso_packets); libusb20_tr_submit(pxfer); /* fork another USB transfer, if any */ libusb10_submit_transfer_sub(libusb20_tr_get_priv_sc0(pxfer), uxfer->endpoint); break; default: libusb10_complete_transfer(pxfer, sxfer, libusb10_convert_error(status)); break; } } /* This function must be called locked */ static void libusb10_bulk_intr_proxy(struct libusb20_transfer *pxfer) { struct libusb_super_transfer *sxfer; struct libusb_transfer *uxfer; uint32_t max_bulk; uint32_t actlen; uint8_t status; uint8_t flags; status = libusb20_tr_get_status(pxfer); sxfer = libusb20_tr_get_priv_sc1(pxfer); max_bulk = libusb20_tr_get_max_total_length(pxfer); actlen = libusb20_tr_get_actual_length(pxfer); if (sxfer == NULL) return; /* cancelled - nothing to do */ uxfer = (struct libusb_transfer *)( ((uint8_t *)sxfer) + sizeof(*sxfer)); flags = uxfer->flags; switch (status) { case LIBUSB20_TRANSFER_COMPLETED: uxfer->actual_length += actlen; /* check for short packet */ if (sxfer->last_len != actlen) { if (flags & LIBUSB_TRANSFER_SHORT_NOT_OK) { libusb10_complete_transfer(pxfer, sxfer, LIBUSB_TRANSFER_ERROR); } else { libusb10_complete_transfer(pxfer, sxfer, LIBUSB_TRANSFER_COMPLETED); } break; } /* check for end of data */ if (sxfer->rem_len == 0) { libusb10_complete_transfer(pxfer, sxfer, LIBUSB_TRANSFER_COMPLETED); break; } /* FALLTHROUGH */ case LIBUSB20_TRANSFER_START: if (max_bulk > sxfer->rem_len) { max_bulk = sxfer->rem_len; } /* setup new BULK or INTERRUPT transaction */ libusb20_tr_setup_bulk(pxfer, sxfer->curr_data, max_bulk, uxfer->timeout); /* update counters */ sxfer->last_len = max_bulk; sxfer->curr_data += max_bulk; sxfer->rem_len -= max_bulk; libusb20_tr_submit(pxfer); /* check if we can fork another USB transfer */ if (sxfer->rem_len == 0) libusb10_submit_transfer_sub(libusb20_tr_get_priv_sc0(pxfer), uxfer->endpoint); break; default: libusb10_complete_transfer(pxfer, sxfer, libusb10_convert_error(status)); break; } } /* This function must be called locked */ static void libusb10_ctrl_proxy(struct libusb20_transfer *pxfer) { struct libusb_super_transfer *sxfer; struct libusb_transfer *uxfer; uint32_t max_bulk; uint32_t actlen; uint8_t status; uint8_t flags; status = libusb20_tr_get_status(pxfer); sxfer = libusb20_tr_get_priv_sc1(pxfer); max_bulk = libusb20_tr_get_max_total_length(pxfer); actlen = libusb20_tr_get_actual_length(pxfer); if (sxfer == NULL) return; /* cancelled - nothing to do */ uxfer = (struct libusb_transfer *)( ((uint8_t *)sxfer) + sizeof(*sxfer)); flags = uxfer->flags; switch (status) { case LIBUSB20_TRANSFER_COMPLETED: uxfer->actual_length += actlen; /* subtract length of SETUP packet, if any */ actlen -= libusb20_tr_get_length(pxfer, 0); /* check for short packet */ if (sxfer->last_len != actlen) { if (flags & LIBUSB_TRANSFER_SHORT_NOT_OK) { libusb10_complete_transfer(pxfer, sxfer, LIBUSB_TRANSFER_ERROR); } else { libusb10_complete_transfer(pxfer, sxfer, LIBUSB_TRANSFER_COMPLETED); } break; } /* check for end of data */ if (sxfer->rem_len == 0) { libusb10_complete_transfer(pxfer, sxfer, LIBUSB_TRANSFER_COMPLETED); break; } /* FALLTHROUGH */ case LIBUSB20_TRANSFER_START: if (max_bulk > sxfer->rem_len) { max_bulk = sxfer->rem_len; } /* setup new CONTROL transaction */ if (status == LIBUSB20_TRANSFER_COMPLETED) { /* next fragment - don't send SETUP packet */ libusb20_tr_set_length(pxfer, 0, 0); } else { /* first fragment - send SETUP packet */ libusb20_tr_set_length(pxfer, 8, 0); libusb20_tr_set_buffer(pxfer, uxfer->buffer, 0); } if (max_bulk != 0) { libusb20_tr_set_length(pxfer, max_bulk, 1); libusb20_tr_set_buffer(pxfer, sxfer->curr_data, 1); libusb20_tr_set_total_frames(pxfer, 2); } else { libusb20_tr_set_total_frames(pxfer, 1); } /* update counters */ sxfer->last_len = max_bulk; sxfer->curr_data += max_bulk; sxfer->rem_len -= max_bulk; libusb20_tr_submit(pxfer); /* check if we can fork another USB transfer */ if (sxfer->rem_len == 0) libusb10_submit_transfer_sub(libusb20_tr_get_priv_sc0(pxfer), uxfer->endpoint); break; default: libusb10_complete_transfer(pxfer, sxfer, libusb10_convert_error(status)); break; } } /* The following function must be called locked */ static void libusb10_submit_transfer_sub(struct libusb20_device *pdev, uint8_t endpoint) { struct libusb20_transfer *pxfer0; struct libusb20_transfer *pxfer1; struct libusb_super_transfer *sxfer; struct libusb_transfer *uxfer; struct libusb_device *dev; int err; int buffsize; int maxframe; int temp; uint8_t dummy; dev = libusb_get_device(pdev); pxfer0 = libusb10_get_transfer(pdev, endpoint, 0); pxfer1 = libusb10_get_transfer(pdev, endpoint, 1); if (pxfer0 == NULL || pxfer1 == NULL) return; /* shouldn't happen */ temp = 0; if (libusb20_tr_pending(pxfer0)) temp |= 1; if (libusb20_tr_pending(pxfer1)) temp |= 2; switch (temp) { case 3: /* wait till one of the transfers complete */ return; case 2: sxfer = libusb20_tr_get_priv_sc1(pxfer1); if (sxfer == NULL) return; /* cancelling */ if (sxfer->rem_len) return; /* cannot queue another one */ /* swap transfers */ pxfer1 = pxfer0; break; case 1: sxfer = libusb20_tr_get_priv_sc1(pxfer0); if (sxfer == NULL) return; /* cancelling */ if (sxfer->rem_len) return; /* cannot queue another one */ /* swap transfers */ pxfer0 = pxfer1; break; default: break; } /* find next transfer on same endpoint */ TAILQ_FOREACH(sxfer, &dev->tr_head, entry) { uxfer = (struct libusb_transfer *)( ((uint8_t *)sxfer) + sizeof(*sxfer)); if (uxfer->endpoint == endpoint) { TAILQ_REMOVE(&dev->tr_head, sxfer, entry); sxfer->entry.tqe_prev = NULL; goto found; } } return; /* success */ found: libusb20_tr_set_priv_sc0(pxfer0, pdev); libusb20_tr_set_priv_sc1(pxfer0, sxfer); /* reset super transfer state */ sxfer->rem_len = uxfer->length; sxfer->curr_data = uxfer->buffer; uxfer->actual_length = 0; switch (uxfer->type) { case LIBUSB_TRANSFER_TYPE_ISOCHRONOUS: libusb20_tr_set_callback(pxfer0, libusb10_isoc_proxy); break; case LIBUSB_TRANSFER_TYPE_BULK: case LIBUSB_TRANSFER_TYPE_INTERRUPT: libusb20_tr_set_callback(pxfer0, libusb10_bulk_intr_proxy); break; case LIBUSB_TRANSFER_TYPE_CONTROL: libusb20_tr_set_callback(pxfer0, libusb10_ctrl_proxy); if (sxfer->rem_len < 8) goto failure; /* remove SETUP packet from data */ sxfer->rem_len -= 8; sxfer->curr_data += 8; break; default: goto failure; } buffsize = libusb10_get_buffsize(pdev, uxfer); maxframe = libusb10_get_maxframe(pdev, uxfer); /* make sure the transfer is opened */ - err = libusb20_tr_open(pxfer0, buffsize, maxframe, endpoint); + err = libusb20_tr_open_stream(pxfer0, buffsize, maxframe, + endpoint, sxfer->stream_id); if (err && (err != LIBUSB20_ERROR_BUSY)) { goto failure; } libusb20_tr_start(pxfer0); return; failure: libusb10_complete_transfer(pxfer0, sxfer, LIBUSB_TRANSFER_ERROR); /* make sure our event loop spins the done handler */ dummy = 0; err = write(dev->ctx->ctrl_pipe[1], &dummy, sizeof(dummy)); } /* The following function must be called unlocked */ int libusb_submit_transfer(struct libusb_transfer *uxfer) { struct libusb20_transfer *pxfer0; struct libusb20_transfer *pxfer1; struct libusb_super_transfer *sxfer; struct libusb_device *dev; uint8_t endpoint; int err; if (uxfer == NULL) return (LIBUSB_ERROR_INVALID_PARAM); if (uxfer->dev_handle == NULL) return (LIBUSB_ERROR_INVALID_PARAM); endpoint = uxfer->endpoint; dev = libusb_get_device(uxfer->dev_handle); DPRINTF(dev->ctx, LIBUSB_DEBUG_FUNCTION, "libusb_submit_transfer enter"); sxfer = (struct libusb_super_transfer *)( (uint8_t *)uxfer - sizeof(*sxfer)); CTX_LOCK(dev->ctx); pxfer0 = libusb10_get_transfer(uxfer->dev_handle, endpoint, 0); pxfer1 = libusb10_get_transfer(uxfer->dev_handle, endpoint, 1); if (pxfer0 == NULL || pxfer1 == NULL) { err = LIBUSB_ERROR_OTHER; } else if ((sxfer->entry.tqe_prev != NULL) || (libusb20_tr_get_priv_sc1(pxfer0) == sxfer) || (libusb20_tr_get_priv_sc1(pxfer1) == sxfer)) { err = LIBUSB_ERROR_BUSY; } else { /* set pending state */ sxfer->state = LIBUSB_SUPER_XFER_ST_PEND; /* insert transfer into transfer head list */ TAILQ_INSERT_TAIL(&dev->tr_head, sxfer, entry); /* start work transfers */ libusb10_submit_transfer_sub( uxfer->dev_handle, endpoint); err = 0; /* success */ } CTX_UNLOCK(dev->ctx); DPRINTF(dev->ctx, LIBUSB_DEBUG_FUNCTION, "libusb_submit_transfer leave %d", err); return (err); } /* Asynchronous transfer cancel */ int libusb_cancel_transfer(struct libusb_transfer *uxfer) { struct libusb20_transfer *pxfer0; struct libusb20_transfer *pxfer1; struct libusb_super_transfer *sxfer; struct libusb_device *dev; uint8_t endpoint; int retval; if (uxfer == NULL) return (LIBUSB_ERROR_INVALID_PARAM); /* check if not initialised */ if (uxfer->dev_handle == NULL) return (LIBUSB_ERROR_NOT_FOUND); endpoint = uxfer->endpoint; dev = libusb_get_device(uxfer->dev_handle); DPRINTF(dev->ctx, LIBUSB_DEBUG_FUNCTION, "libusb_cancel_transfer enter"); sxfer = (struct libusb_super_transfer *)( (uint8_t *)uxfer - sizeof(*sxfer)); retval = 0; CTX_LOCK(dev->ctx); pxfer0 = libusb10_get_transfer(uxfer->dev_handle, endpoint, 0); pxfer1 = libusb10_get_transfer(uxfer->dev_handle, endpoint, 1); if (sxfer->state != LIBUSB_SUPER_XFER_ST_PEND) { /* only update the transfer status */ uxfer->status = LIBUSB_TRANSFER_CANCELLED; retval = LIBUSB_ERROR_NOT_FOUND; } else if (sxfer->entry.tqe_prev != NULL) { /* we are lucky - transfer is on a queue */ TAILQ_REMOVE(&dev->tr_head, sxfer, entry); sxfer->entry.tqe_prev = NULL; libusb10_complete_transfer(NULL, sxfer, LIBUSB_TRANSFER_CANCELLED); } else if (pxfer0 == NULL || pxfer1 == NULL) { /* not started */ retval = LIBUSB_ERROR_NOT_FOUND; } else if (libusb20_tr_get_priv_sc1(pxfer0) == sxfer) { libusb10_complete_transfer(pxfer0, sxfer, LIBUSB_TRANSFER_CANCELLED); libusb20_tr_stop(pxfer0); /* make sure the queue doesn't stall */ libusb10_submit_transfer_sub( uxfer->dev_handle, endpoint); } else if (libusb20_tr_get_priv_sc1(pxfer1) == sxfer) { libusb10_complete_transfer(pxfer1, sxfer, LIBUSB_TRANSFER_CANCELLED); libusb20_tr_stop(pxfer1); /* make sure the queue doesn't stall */ libusb10_submit_transfer_sub( uxfer->dev_handle, endpoint); } else { /* not started */ retval = LIBUSB_ERROR_NOT_FOUND; } CTX_UNLOCK(dev->ctx); DPRINTF(dev->ctx, LIBUSB_DEBUG_FUNCTION, "libusb_cancel_transfer leave"); return (retval); } UNEXPORTED void libusb10_cancel_all_transfer(libusb_device *dev) { struct libusb20_device *pdev = dev->os_priv; unsigned x; for (x = 0; x != LIBUSB_NUM_SW_ENDPOINTS; x++) { struct libusb20_transfer *xfer; xfer = libusb20_tr_get_pointer(pdev, x); if (xfer == NULL) continue; libusb20_tr_close(xfer); } } uint16_t libusb_cpu_to_le16(uint16_t x) { return (htole16(x)); } uint16_t libusb_le16_to_cpu(uint16_t x) { return (le16toh(x)); } const char * libusb_strerror(int code) { switch (code) { case LIBUSB_SUCCESS: return ("Success"); case LIBUSB_ERROR_IO: return ("I/O error"); case LIBUSB_ERROR_INVALID_PARAM: return ("Invalid parameter"); case LIBUSB_ERROR_ACCESS: return ("Permissions error"); case LIBUSB_ERROR_NO_DEVICE: return ("No device"); case LIBUSB_ERROR_NOT_FOUND: return ("Not found"); case LIBUSB_ERROR_BUSY: return ("Device busy"); case LIBUSB_ERROR_TIMEOUT: return ("Timeout"); case LIBUSB_ERROR_OVERFLOW: return ("Overflow"); case LIBUSB_ERROR_PIPE: return ("Pipe error"); case LIBUSB_ERROR_INTERRUPTED: return ("Interrupted"); case LIBUSB_ERROR_NO_MEM: return ("Out of memory"); case LIBUSB_ERROR_NOT_SUPPORTED: return ("Not supported"); case LIBUSB_ERROR_OTHER: return ("Other error"); default: return ("Unknown error"); } } const char * libusb_error_name(int code) { switch (code) { case LIBUSB_SUCCESS: return ("LIBUSB_SUCCESS"); case LIBUSB_ERROR_IO: return ("LIBUSB_ERROR_IO"); case LIBUSB_ERROR_INVALID_PARAM: return ("LIBUSB_ERROR_INVALID_PARAM"); case LIBUSB_ERROR_ACCESS: return ("LIBUSB_ERROR_ACCESS"); case LIBUSB_ERROR_NO_DEVICE: return ("LIBUSB_ERROR_NO_DEVICE"); case LIBUSB_ERROR_NOT_FOUND: return ("LIBUSB_ERROR_NOT_FOUND"); case LIBUSB_ERROR_BUSY: return ("LIBUSB_ERROR_BUSY"); case LIBUSB_ERROR_TIMEOUT: return ("LIBUSB_ERROR_TIMEOUT"); case LIBUSB_ERROR_OVERFLOW: return ("LIBUSB_ERROR_OVERFLOW"); case LIBUSB_ERROR_PIPE: return ("LIBUSB_ERROR_PIPE"); case LIBUSB_ERROR_INTERRUPTED: return ("LIBUSB_ERROR_INTERRUPTED"); case LIBUSB_ERROR_NO_MEM: return ("LIBUSB_ERROR_NO_MEM"); case LIBUSB_ERROR_NOT_SUPPORTED: return ("LIBUSB_ERROR_NOT_SUPPORTED"); case LIBUSB_ERROR_OTHER: return ("LIBUSB_ERROR_OTHER"); default: return ("LIBUSB_ERROR_UNKNOWN"); } } Index: projects/vnet/lib/libusb/libusb10.h =================================================================== --- projects/vnet/lib/libusb/libusb10.h (revision 302157) +++ projects/vnet/lib/libusb/libusb10.h (revision 302158) @@ -1,135 +1,136 @@ /* $FreeBSD$ */ /*- * Copyright (c) 2009 Sylvestre Gallon. 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. */ #ifndef __LIBUSB10_H__ #define __LIBUSB10_H__ #ifndef LIBUSB_GLOBAL_INCLUDE_FILE #include #endif #define GET_CONTEXT(ctx) (((ctx) == NULL) ? usbi_default_context : (ctx)) #define UNEXPORTED __attribute__((__visibility__("hidden"))) #define CTX_LOCK(ctx) pthread_mutex_lock(&(ctx)->ctx_lock) #define CTX_TRYLOCK(ctx) pthread_mutex_trylock(&(ctx)->ctx_lock) #define CTX_UNLOCK(ctx) pthread_mutex_unlock(&(ctx)->ctx_lock) #define HOTPLUG_LOCK(ctx) pthread_mutex_lock(&(ctx)->hotplug_lock) #define HOTPLUG_UNLOCK(ctx) pthread_mutex_unlock(&(ctx)->hotplug_lock) #define DPRINTF(ctx, dbg, format, args...) do { \ if ((ctx)->debug == dbg) { \ switch (dbg) { \ case LIBUSB_DEBUG_FUNCTION: \ printf("LIBUSB_FUNCTION: " \ format "\n", ## args); \ break; \ case LIBUSB_DEBUG_TRANSFER: \ printf("LIBUSB_TRANSFER: " \ format "\n", ## args); \ break; \ default: \ break; \ } \ } \ } while(0) /* internal structures */ struct libusb_super_pollfd { TAILQ_ENTRY(libusb_super_pollfd) entry; struct libusb20_device *pdev; struct libusb_pollfd pollfd; }; struct libusb_super_transfer { TAILQ_ENTRY(libusb_super_transfer) entry; uint8_t *curr_data; uint32_t rem_len; uint32_t last_len; + uint32_t stream_id; uint8_t state; #define LIBUSB_SUPER_XFER_ST_NONE 0 #define LIBUSB_SUPER_XFER_ST_PEND 1 }; struct libusb_hotplug_callback_handle { TAILQ_ENTRY(libusb_hotplug_callback_handle) entry; int events; int vendor; int product; int devclass; libusb_hotplug_callback_fn fn; void *user_data; }; struct libusb_context { int debug; int debug_fixed; int ctrl_pipe[2]; int tr_done_ref; int tr_done_gen; pthread_mutex_t ctx_lock; pthread_mutex_t hotplug_lock; pthread_cond_t ctx_cond; pthread_t hotplug_handler; pthread_t ctx_handler; #define NO_THREAD ((pthread_t)-1) TAILQ_HEAD(, libusb_super_pollfd) pollfds; TAILQ_HEAD(, libusb_super_transfer) tr_done; TAILQ_HEAD(, libusb_hotplug_callback_handle) hotplug_cbh; TAILQ_HEAD(, libusb_device) hotplug_devs; struct libusb_super_pollfd ctx_poll; libusb_pollfd_added_cb fd_added_cb; libusb_pollfd_removed_cb fd_removed_cb; void *fd_cb_user_data; }; struct libusb_device { int refcnt; uint32_t claimed_interfaces; struct libusb_super_pollfd dev_poll; struct libusb_context *ctx; TAILQ_ENTRY(libusb_device) hotplug_entry; TAILQ_HEAD(, libusb_super_transfer) tr_head; struct libusb20_device *os_priv; }; extern struct libusb_context *usbi_default_context; void libusb10_add_pollfd(libusb_context *ctx, struct libusb_super_pollfd *pollfd, struct libusb20_device *pdev, int fd, short events); void libusb10_remove_pollfd(libusb_context *ctx, struct libusb_super_pollfd *pollfd); void libusb10_cancel_all_transfer(libusb_device *dev); #endif /* __LIBUSB10_H__ */ Index: projects/vnet/lib/libusb/libusb10_io.c =================================================================== --- projects/vnet/lib/libusb/libusb10_io.c (revision 302157) +++ projects/vnet/lib/libusb/libusb10_io.c (revision 302158) @@ -1,769 +1,814 @@ /* $FreeBSD$ */ /*- * Copyright (c) 2009 Sylvestre Gallon. 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. */ #ifdef LIBUSB_GLOBAL_INCLUDE_FILE #include LIBUSB_GLOBAL_INCLUDE_FILE #else #include #include #include #include #include #include #include #include #include #include #endif #define libusb_device_handle libusb20_device #include "libusb20.h" #include "libusb20_desc.h" #include "libusb20_int.h" #include "libusb.h" #include "libusb10.h" UNEXPORTED void libusb10_add_pollfd(libusb_context *ctx, struct libusb_super_pollfd *pollfd, struct libusb20_device *pdev, int fd, short events) { if (ctx == NULL) return; /* invalid */ if (pollfd->entry.tqe_prev != NULL) return; /* already queued */ if (fd < 0) return; /* invalid */ pollfd->pdev = pdev; pollfd->pollfd.fd = fd; pollfd->pollfd.events = events; CTX_LOCK(ctx); TAILQ_INSERT_TAIL(&ctx->pollfds, pollfd, entry); CTX_UNLOCK(ctx); if (ctx->fd_added_cb) ctx->fd_added_cb(fd, events, ctx->fd_cb_user_data); } UNEXPORTED void libusb10_remove_pollfd(libusb_context *ctx, struct libusb_super_pollfd *pollfd) { if (ctx == NULL) return; /* invalid */ if (pollfd->entry.tqe_prev == NULL) return; /* already dequeued */ CTX_LOCK(ctx); TAILQ_REMOVE(&ctx->pollfds, pollfd, entry); pollfd->entry.tqe_prev = NULL; CTX_UNLOCK(ctx); if (ctx->fd_removed_cb) ctx->fd_removed_cb(pollfd->pollfd.fd, ctx->fd_cb_user_data); } /* This function must be called locked */ static int libusb10_handle_events_sub(struct libusb_context *ctx, struct timeval *tv) { struct libusb_device *dev; struct libusb20_device **ppdev; struct libusb_super_pollfd *pfd; struct pollfd *fds; struct libusb_super_transfer *sxfer; struct libusb_transfer *uxfer; nfds_t nfds; int timeout; int i; int err; DPRINTF(ctx, LIBUSB_DEBUG_FUNCTION, "libusb10_handle_events_sub enter"); nfds = 0; i = 0; TAILQ_FOREACH(pfd, &ctx->pollfds, entry) nfds++; fds = alloca(sizeof(*fds) * nfds); if (fds == NULL) return (LIBUSB_ERROR_NO_MEM); ppdev = alloca(sizeof(*ppdev) * nfds); if (ppdev == NULL) return (LIBUSB_ERROR_NO_MEM); TAILQ_FOREACH(pfd, &ctx->pollfds, entry) { fds[i].fd = pfd->pollfd.fd; fds[i].events = pfd->pollfd.events; fds[i].revents = 0; ppdev[i] = pfd->pdev; if (pfd->pdev != NULL) libusb_get_device(pfd->pdev)->refcnt++; i++; } if (tv == NULL) timeout = -1; else timeout = (tv->tv_sec * 1000) + ((tv->tv_usec + 999) / 1000); CTX_UNLOCK(ctx); err = poll(fds, nfds, timeout); CTX_LOCK(ctx); if ((err == -1) && (errno == EINTR)) err = LIBUSB_ERROR_INTERRUPTED; else if (err < 0) err = LIBUSB_ERROR_IO; if (err < 1) { for (i = 0; i != (int)nfds; i++) { if (ppdev[i] != NULL) { CTX_UNLOCK(ctx); libusb_unref_device(libusb_get_device(ppdev[i])); CTX_LOCK(ctx); } } goto do_done; } for (i = 0; i != (int)nfds; i++) { if (ppdev[i] != NULL) { dev = libusb_get_device(ppdev[i]); if (fds[i].revents == 0) err = 0; /* nothing to do */ else err = libusb20_dev_process(ppdev[i]); if (err) { /* cancel all transfers - device is gone */ libusb10_cancel_all_transfer(dev); /* remove USB device from polling loop */ libusb10_remove_pollfd(dev->ctx, &dev->dev_poll); } CTX_UNLOCK(ctx); libusb_unref_device(dev); CTX_LOCK(ctx); } else { uint8_t dummy; while (1) { if (read(fds[i].fd, &dummy, 1) != 1) break; } } } err = 0; do_done: /* Do all done callbacks */ while ((sxfer = TAILQ_FIRST(&ctx->tr_done))) { uint8_t flags; TAILQ_REMOVE(&ctx->tr_done, sxfer, entry); sxfer->entry.tqe_prev = NULL; ctx->tr_done_ref++; CTX_UNLOCK(ctx); uxfer = (struct libusb_transfer *)( ((uint8_t *)sxfer) + sizeof(*sxfer)); /* Allow the callback to free the transfer itself. */ flags = uxfer->flags; if (uxfer->callback != NULL) (uxfer->callback) (uxfer); /* Check if the USB transfer should be automatically freed. */ if (flags & LIBUSB_TRANSFER_FREE_TRANSFER) libusb_free_transfer(uxfer); CTX_LOCK(ctx); ctx->tr_done_ref--; ctx->tr_done_gen++; } /* Wakeup other waiters */ pthread_cond_broadcast(&ctx->ctx_cond); return (err); } /* Polling and timing */ int libusb_try_lock_events(libusb_context *ctx) { int err; ctx = GET_CONTEXT(ctx); if (ctx == NULL) return (1); err = CTX_TRYLOCK(ctx); if (err) return (1); err = (ctx->ctx_handler != NO_THREAD); if (err) CTX_UNLOCK(ctx); else ctx->ctx_handler = pthread_self(); return (err); } void libusb_lock_events(libusb_context *ctx) { ctx = GET_CONTEXT(ctx); CTX_LOCK(ctx); if (ctx->ctx_handler == NO_THREAD) ctx->ctx_handler = pthread_self(); } void libusb_unlock_events(libusb_context *ctx) { ctx = GET_CONTEXT(ctx); if (ctx->ctx_handler == pthread_self()) { ctx->ctx_handler = NO_THREAD; pthread_cond_broadcast(&ctx->ctx_cond); } CTX_UNLOCK(ctx); } int libusb_event_handling_ok(libusb_context *ctx) { ctx = GET_CONTEXT(ctx); return (ctx->ctx_handler == pthread_self()); } int libusb_event_handler_active(libusb_context *ctx) { ctx = GET_CONTEXT(ctx); return (ctx->ctx_handler != NO_THREAD); } void libusb_lock_event_waiters(libusb_context *ctx) { ctx = GET_CONTEXT(ctx); CTX_LOCK(ctx); } void libusb_unlock_event_waiters(libusb_context *ctx) { ctx = GET_CONTEXT(ctx); CTX_UNLOCK(ctx); } int libusb_wait_for_event(libusb_context *ctx, struct timeval *tv) { struct timespec ts; int err; ctx = GET_CONTEXT(ctx); DPRINTF(ctx, LIBUSB_DEBUG_FUNCTION, "libusb_wait_for_event enter"); if (tv == NULL) { pthread_cond_wait(&ctx->ctx_cond, &ctx->ctx_lock); return (0); } err = clock_gettime(CLOCK_MONOTONIC, &ts); if (err < 0) return (LIBUSB_ERROR_OTHER); /* * The "tv" arguments points to a relative time structure and * not an absolute time structure. */ ts.tv_sec += tv->tv_sec; ts.tv_nsec += tv->tv_usec * 1000; if (ts.tv_nsec >= 1000000000) { ts.tv_nsec -= 1000000000; ts.tv_sec++; } err = pthread_cond_timedwait(&ctx->ctx_cond, &ctx->ctx_lock, &ts); if (err == ETIMEDOUT) return (1); return (0); } int libusb_handle_events_timeout_completed(libusb_context *ctx, struct timeval *tv, int *completed) { int err = 0; ctx = GET_CONTEXT(ctx); DPRINTF(ctx, LIBUSB_DEBUG_FUNCTION, "libusb_handle_events_timeout_completed enter"); libusb_lock_events(ctx); while (1) { if (completed != NULL) { if (*completed != 0 || err != 0) break; } err = libusb_handle_events_locked(ctx, tv); if (completed == NULL) break; } libusb_unlock_events(ctx); DPRINTF(ctx, LIBUSB_DEBUG_FUNCTION, "libusb_handle_events_timeout_completed exit"); return (err); } int libusb_handle_events_completed(libusb_context *ctx, int *completed) { return (libusb_handle_events_timeout_completed(ctx, NULL, completed)); } int libusb_handle_events_timeout(libusb_context *ctx, struct timeval *tv) { return (libusb_handle_events_timeout_completed(ctx, tv, NULL)); } int libusb_handle_events(libusb_context *ctx) { return (libusb_handle_events_timeout_completed(ctx, NULL, NULL)); } int libusb_handle_events_locked(libusb_context *ctx, struct timeval *tv) { int err; ctx = GET_CONTEXT(ctx); if (libusb_event_handling_ok(ctx)) { err = libusb10_handle_events_sub(ctx, tv); } else { err = libusb_wait_for_event(ctx, tv); if (err != 0) err = LIBUSB_ERROR_TIMEOUT; } return (err); } int libusb_get_next_timeout(libusb_context *ctx, struct timeval *tv) { /* all timeouts are currently being done by the kernel */ timerclear(tv); return (0); } void libusb_set_pollfd_notifiers(libusb_context *ctx, libusb_pollfd_added_cb added_cb, libusb_pollfd_removed_cb removed_cb, void *user_data) { ctx = GET_CONTEXT(ctx); ctx->fd_added_cb = added_cb; ctx->fd_removed_cb = removed_cb; ctx->fd_cb_user_data = user_data; } const struct libusb_pollfd ** libusb_get_pollfds(libusb_context *ctx) { struct libusb_super_pollfd *pollfd; libusb_pollfd **ret; int i; ctx = GET_CONTEXT(ctx); CTX_LOCK(ctx); i = 0; TAILQ_FOREACH(pollfd, &ctx->pollfds, entry) i++; ret = calloc(i + 1, sizeof(struct libusb_pollfd *)); if (ret == NULL) goto done; i = 0; TAILQ_FOREACH(pollfd, &ctx->pollfds, entry) ret[i++] = &pollfd->pollfd; ret[i] = NULL; done: CTX_UNLOCK(ctx); return ((const struct libusb_pollfd **)ret); } /* Synchronous device I/O */ int libusb_control_transfer(libusb_device_handle *devh, uint8_t bmRequestType, uint8_t bRequest, uint16_t wValue, uint16_t wIndex, uint8_t *data, uint16_t wLength, unsigned int timeout) { struct LIBUSB20_CONTROL_SETUP_DECODED req; int err; uint16_t actlen; if (devh == NULL) return (LIBUSB_ERROR_INVALID_PARAM); if ((wLength != 0) && (data == NULL)) return (LIBUSB_ERROR_INVALID_PARAM); LIBUSB20_INIT(LIBUSB20_CONTROL_SETUP, &req); req.bmRequestType = bmRequestType; req.bRequest = bRequest; req.wValue = wValue; req.wIndex = wIndex; req.wLength = wLength; err = libusb20_dev_request_sync(devh, &req, data, &actlen, timeout, 0); if (err == LIBUSB20_ERROR_PIPE) return (LIBUSB_ERROR_PIPE); else if (err == LIBUSB20_ERROR_TIMEOUT) return (LIBUSB_ERROR_TIMEOUT); else if (err) return (LIBUSB_ERROR_NO_DEVICE); return (actlen); } static void libusb10_do_transfer_cb(struct libusb_transfer *transfer) { libusb_context *ctx; int *pdone; ctx = GET_CONTEXT(NULL); DPRINTF(ctx, LIBUSB_DEBUG_TRANSFER, "sync I/O done"); pdone = transfer->user_data; *pdone = 1; } /* * TODO: Replace the following function. Allocating and freeing on a * per-transfer basis is slow. --HPS */ static int libusb10_do_transfer(libusb_device_handle *devh, uint8_t endpoint, uint8_t *data, int length, int *transferred, unsigned int timeout, int type) { libusb_context *ctx; struct libusb_transfer *xfer; int done; int ret; if (devh == NULL) return (LIBUSB_ERROR_INVALID_PARAM); if ((length != 0) && (data == NULL)) return (LIBUSB_ERROR_INVALID_PARAM); xfer = libusb_alloc_transfer(0); if (xfer == NULL) return (LIBUSB_ERROR_NO_MEM); ctx = libusb_get_device(devh)->ctx; xfer->dev_handle = devh; xfer->endpoint = endpoint; xfer->type = type; xfer->timeout = timeout; xfer->buffer = data; xfer->length = length; xfer->user_data = (void *)&done; xfer->callback = libusb10_do_transfer_cb; done = 0; if ((ret = libusb_submit_transfer(xfer)) < 0) { libusb_free_transfer(xfer); return (ret); } while (done == 0) { if ((ret = libusb_handle_events(ctx)) < 0) { libusb_cancel_transfer(xfer); usleep(1000); /* nice it */ } } *transferred = xfer->actual_length; switch (xfer->status) { case LIBUSB_TRANSFER_COMPLETED: ret = 0; break; case LIBUSB_TRANSFER_TIMED_OUT: ret = LIBUSB_ERROR_TIMEOUT; break; case LIBUSB_TRANSFER_OVERFLOW: ret = LIBUSB_ERROR_OVERFLOW; break; case LIBUSB_TRANSFER_STALL: ret = LIBUSB_ERROR_PIPE; break; case LIBUSB_TRANSFER_NO_DEVICE: ret = LIBUSB_ERROR_NO_DEVICE; break; default: ret = LIBUSB_ERROR_OTHER; break; } libusb_free_transfer(xfer); return (ret); } int libusb_bulk_transfer(libusb_device_handle *devh, uint8_t endpoint, uint8_t *data, int length, int *transferred, unsigned int timeout) { libusb_context *ctx; int ret; ctx = GET_CONTEXT(NULL); DPRINTF(ctx, LIBUSB_DEBUG_FUNCTION, "libusb_bulk_transfer enter"); ret = libusb10_do_transfer(devh, endpoint, data, length, transferred, timeout, LIBUSB_TRANSFER_TYPE_BULK); DPRINTF(ctx, LIBUSB_DEBUG_FUNCTION, "libusb_bulk_transfer leave"); return (ret); } int libusb_interrupt_transfer(libusb_device_handle *devh, uint8_t endpoint, uint8_t *data, int length, int *transferred, unsigned int timeout) { libusb_context *ctx; int ret; ctx = GET_CONTEXT(NULL); DPRINTF(ctx, LIBUSB_DEBUG_FUNCTION, "libusb_interrupt_transfer enter"); ret = libusb10_do_transfer(devh, endpoint, data, length, transferred, timeout, LIBUSB_TRANSFER_TYPE_INTERRUPT); DPRINTF(ctx, LIBUSB_DEBUG_FUNCTION, "libusb_interrupt_transfer leave"); return (ret); } uint8_t * libusb_get_iso_packet_buffer(struct libusb_transfer *transfer, uint32_t off) { uint8_t *ptr; uint32_t n; if (transfer->num_iso_packets < 0) return (NULL); if (off >= (uint32_t)transfer->num_iso_packets) return (NULL); ptr = transfer->buffer; if (ptr == NULL) return (NULL); for (n = 0; n != off; n++) { ptr += transfer->iso_packet_desc[n].length; } return (ptr); } uint8_t * libusb_get_iso_packet_buffer_simple(struct libusb_transfer *transfer, uint32_t off) { uint8_t *ptr; if (transfer->num_iso_packets < 0) return (NULL); if (off >= (uint32_t)transfer->num_iso_packets) return (NULL); ptr = transfer->buffer; if (ptr == NULL) return (NULL); ptr += transfer->iso_packet_desc[0].length * off; return (ptr); } void libusb_set_iso_packet_lengths(struct libusb_transfer *transfer, uint32_t length) { int n; if (transfer->num_iso_packets < 0) return; for (n = 0; n != transfer->num_iso_packets; n++) transfer->iso_packet_desc[n].length = length; } uint8_t * libusb_control_transfer_get_data(struct libusb_transfer *transfer) { if (transfer->buffer == NULL) return (NULL); return (transfer->buffer + LIBUSB_CONTROL_SETUP_SIZE); } struct libusb_control_setup * libusb_control_transfer_get_setup(struct libusb_transfer *transfer) { return ((struct libusb_control_setup *)transfer->buffer); } void libusb_fill_control_setup(uint8_t *buf, uint8_t bmRequestType, uint8_t bRequest, uint16_t wValue, uint16_t wIndex, uint16_t wLength) { struct libusb_control_setup *req = (struct libusb_control_setup *)buf; /* The alignment is OK for all fields below. */ req->bmRequestType = bmRequestType; req->bRequest = bRequest; req->wValue = htole16(wValue); req->wIndex = htole16(wIndex); req->wLength = htole16(wLength); } void libusb_fill_control_transfer(struct libusb_transfer *transfer, libusb_device_handle *devh, uint8_t *buf, libusb_transfer_cb_fn callback, void *user_data, uint32_t timeout) { struct libusb_control_setup *setup = (struct libusb_control_setup *)buf; transfer->dev_handle = devh; transfer->endpoint = 0; transfer->type = LIBUSB_TRANSFER_TYPE_CONTROL; transfer->timeout = timeout; transfer->buffer = buf; if (setup != NULL) transfer->length = LIBUSB_CONTROL_SETUP_SIZE + le16toh(setup->wLength); else transfer->length = 0; transfer->user_data = user_data; transfer->callback = callback; } void libusb_fill_bulk_transfer(struct libusb_transfer *transfer, libusb_device_handle *devh, uint8_t endpoint, uint8_t *buf, int length, libusb_transfer_cb_fn callback, void *user_data, uint32_t timeout) { transfer->dev_handle = devh; transfer->endpoint = endpoint; transfer->type = LIBUSB_TRANSFER_TYPE_BULK; transfer->timeout = timeout; transfer->buffer = buf; transfer->length = length; transfer->user_data = user_data; transfer->callback = callback; } void libusb_fill_interrupt_transfer(struct libusb_transfer *transfer, libusb_device_handle *devh, uint8_t endpoint, uint8_t *buf, int length, libusb_transfer_cb_fn callback, void *user_data, uint32_t timeout) { transfer->dev_handle = devh; transfer->endpoint = endpoint; transfer->type = LIBUSB_TRANSFER_TYPE_INTERRUPT; transfer->timeout = timeout; transfer->buffer = buf; transfer->length = length; transfer->user_data = user_data; transfer->callback = callback; } void libusb_fill_iso_transfer(struct libusb_transfer *transfer, libusb_device_handle *devh, uint8_t endpoint, uint8_t *buf, int length, int npacket, libusb_transfer_cb_fn callback, void *user_data, uint32_t timeout) { transfer->dev_handle = devh; transfer->endpoint = endpoint; transfer->type = LIBUSB_TRANSFER_TYPE_ISOCHRONOUS; transfer->timeout = timeout; transfer->buffer = buf; transfer->length = length; transfer->num_iso_packets = npacket; transfer->user_data = user_data; transfer->callback = callback; } +int +libusb_alloc_streams(libusb_device_handle *dev, uint32_t num_streams, + unsigned char *endpoints, int num_endpoints) +{ + if (num_streams > 1) + return (LIBUSB_ERROR_INVALID_PARAM); + return (0); +} + +int +libusb_free_streams(libusb_device_handle *dev, unsigned char *endpoints, int num_endpoints) +{ + + return (0); +} + +void +libusb_transfer_set_stream_id(struct libusb_transfer *transfer, uint32_t stream_id) +{ + struct libusb_super_transfer *sxfer; + + if (transfer == NULL) + return; + + sxfer = (struct libusb_super_transfer *)( + ((uint8_t *)transfer) - sizeof(*sxfer)); + + /* set stream ID */ + sxfer->stream_id = stream_id; +} + +uint32_t +libusb_transfer_get_stream_id(struct libusb_transfer *transfer) +{ + struct libusb_super_transfer *sxfer; + + if (transfer == NULL) + return (0); + + sxfer = (struct libusb_super_transfer *)( + ((uint8_t *)transfer) - sizeof(*sxfer)); + + /* get stream ID */ + return (sxfer->stream_id); +} Index: projects/vnet/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c =================================================================== --- projects/vnet/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c (revision 302157) +++ projects/vnet/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c (revision 302158) @@ -1,6800 +1,6803 @@ /* * 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) 2011-2012 Pawel Jakub Dawidek . * All rights reserved. * Copyright 2013 Martin Matuska . All rights reserved. * Copyright 2014 Xin Li . All rights reserved. * Copyright 2015, OmniTI Computer Consulting, Inc. All rights reserved. * Copyright 2015 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2014, Joyent, Inc. All rights reserved. * Copyright (c) 2011, 2015 by Delphix. All rights reserved. * Copyright (c) 2013 by Saso Kiselkov. All rights reserved. * Copyright (c) 2013 Steven Hartland. All rights reserved. * Copyright (c) 2014 Integros [integros.com] */ /* * ZFS ioctls. * * This file handles the ioctls to /dev/zfs, used for configuring ZFS storage * pools and filesystems, e.g. with /sbin/zfs and /sbin/zpool. * * There are two ways that we handle ioctls: the legacy way where almost * all of the logic is in the ioctl callback, and the new way where most * of the marshalling is handled in the common entry point, zfsdev_ioctl(). * * Non-legacy ioctls should be registered by calling * zfs_ioctl_register() from zfs_ioctl_init(). The ioctl is invoked * from userland by lzc_ioctl(). * * The registration arguments are as follows: * * const char *name * The name of the ioctl. This is used for history logging. If the * ioctl returns successfully (the callback returns 0), and allow_log * is true, then a history log entry will be recorded with the input & * output nvlists. The log entry can be printed with "zpool history -i". * * zfs_ioc_t ioc * The ioctl request number, which userland will pass to ioctl(2). * The ioctl numbers can change from release to release, because * the caller (libzfs) must be matched to the kernel. * * zfs_secpolicy_func_t *secpolicy * This function will be called before the zfs_ioc_func_t, to * determine if this operation is permitted. It should return EPERM * on failure, and 0 on success. Checks include determining if the * dataset is visible in this zone, and if the user has either all * zfs privileges in the zone (SYS_MOUNT), or has been granted permission * to do this operation on this dataset with "zfs allow". * * zfs_ioc_namecheck_t namecheck * This specifies what to expect in the zfs_cmd_t:zc_name -- a pool * name, a dataset name, or nothing. If the name is not well-formed, * the ioctl will fail and the callback will not be called. * Therefore, the callback can assume that the name is well-formed * (e.g. is null-terminated, doesn't have more than one '@' character, * doesn't have invalid characters). * * zfs_ioc_poolcheck_t pool_check * This specifies requirements on the pool state. If the pool does * not meet them (is suspended or is readonly), the ioctl will fail * and the callback will not be called. If any checks are specified * (i.e. it is not POOL_CHECK_NONE), namecheck must not be NO_NAME. * Multiple checks can be or-ed together (e.g. POOL_CHECK_SUSPENDED | * POOL_CHECK_READONLY). * * boolean_t smush_outnvlist * If smush_outnvlist is true, then the output is presumed to be a * list of errors, and it will be "smushed" down to fit into the * caller's buffer, by removing some entries and replacing them with a * single "N_MORE_ERRORS" entry indicating how many were removed. See * nvlist_smush() for details. If smush_outnvlist is false, and the * outnvlist does not fit into the userland-provided buffer, then the * ioctl will fail with ENOMEM. * * zfs_ioc_func_t *func * The callback function that will perform the operation. * * The callback should return 0 on success, or an error number on * failure. If the function fails, the userland ioctl will return -1, * and errno will be set to the callback's return value. The callback * will be called with the following arguments: * * const char *name * The name of the pool or dataset to operate on, from * zfs_cmd_t:zc_name. The 'namecheck' argument specifies the * expected type (pool, dataset, or none). * * nvlist_t *innvl * The input nvlist, deserialized from zfs_cmd_t:zc_nvlist_src. Or * NULL if no input nvlist was provided. Changes to this nvlist are * ignored. If the input nvlist could not be deserialized, the * ioctl will fail and the callback will not be called. * * nvlist_t *outnvl * The output nvlist, initially empty. The callback can fill it in, * and it will be returned to userland by serializing it into * zfs_cmd_t:zc_nvlist_dst. If it is non-empty, and serialization * fails (e.g. because the caller didn't supply a large enough * buffer), then the overall ioctl will fail. See the * 'smush_nvlist' argument above for additional behaviors. * * There are two typical uses of the output nvlist: * - To return state, e.g. property values. In this case, * smush_outnvlist should be false. If the buffer was not large * enough, the caller will reallocate a larger buffer and try * the ioctl again. * * - To return multiple errors from an ioctl which makes on-disk * changes. In this case, smush_outnvlist should be true. * Ioctls which make on-disk modifications should generally not * use the outnvl if they succeed, because the caller can not * distinguish between the operation failing, and * deserialization failing. */ #ifdef __FreeBSD__ #include "opt_kstack_pages.h" #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "zfs_namecheck.h" #include "zfs_prop.h" #include "zfs_deleg.h" #include "zfs_comutil.h" #include "zfs_ioctl_compat.h" CTASSERT(sizeof(zfs_cmd_t) < IOCPARM_MAX); static struct cdev *zfsdev; extern void zfs_init(void); extern void zfs_fini(void); uint_t zfs_fsyncer_key; extern uint_t rrw_tsd_key; static uint_t zfs_allow_log_key; extern uint_t zfs_geom_probe_vdev_key; typedef int zfs_ioc_legacy_func_t(zfs_cmd_t *); typedef int zfs_ioc_func_t(const char *, nvlist_t *, nvlist_t *); typedef int zfs_secpolicy_func_t(zfs_cmd_t *, nvlist_t *, cred_t *); typedef enum { NO_NAME, POOL_NAME, DATASET_NAME } zfs_ioc_namecheck_t; typedef enum { POOL_CHECK_NONE = 1 << 0, POOL_CHECK_SUSPENDED = 1 << 1, POOL_CHECK_READONLY = 1 << 2, } zfs_ioc_poolcheck_t; typedef struct zfs_ioc_vec { zfs_ioc_legacy_func_t *zvec_legacy_func; zfs_ioc_func_t *zvec_func; zfs_secpolicy_func_t *zvec_secpolicy; zfs_ioc_namecheck_t zvec_namecheck; boolean_t zvec_allow_log; zfs_ioc_poolcheck_t zvec_pool_check; boolean_t zvec_smush_outnvlist; const char *zvec_name; } zfs_ioc_vec_t; /* This array is indexed by zfs_userquota_prop_t */ static const char *userquota_perms[] = { ZFS_DELEG_PERM_USERUSED, ZFS_DELEG_PERM_USERQUOTA, ZFS_DELEG_PERM_GROUPUSED, ZFS_DELEG_PERM_GROUPQUOTA, }; static int zfs_ioc_userspace_upgrade(zfs_cmd_t *zc); static int zfs_check_settable(const char *name, nvpair_t *property, cred_t *cr); static int zfs_check_clearable(char *dataset, nvlist_t *props, nvlist_t **errors); static int zfs_fill_zplprops_root(uint64_t, nvlist_t *, nvlist_t *, boolean_t *); int zfs_set_prop_nvlist(const char *, zprop_source_t, nvlist_t *, nvlist_t *); static int get_nvlist(uint64_t nvl, uint64_t size, int iflag, nvlist_t **nvp); static void zfsdev_close(void *data); static int zfs_prop_activate_feature(spa_t *spa, spa_feature_t feature); /* _NOTE(PRINTFLIKE(4)) - this is printf-like, but lint is too whiney */ void __dprintf(const char *file, const char *func, int line, const char *fmt, ...) { const char *newfile; char buf[512]; va_list adx; /* * Get rid of annoying "../common/" prefix to filename. */ newfile = strrchr(file, '/'); if (newfile != NULL) { newfile = newfile + 1; /* Get rid of leading / */ } else { newfile = file; } va_start(adx, fmt); (void) vsnprintf(buf, sizeof (buf), fmt, adx); va_end(adx); /* * To get this data, use the zfs-dprintf probe as so: * dtrace -q -n 'zfs-dprintf \ * /stringof(arg0) == "dbuf.c"/ \ * {printf("%s: %s", stringof(arg1), stringof(arg3))}' * arg0 = file name * arg1 = function name * arg2 = line number * arg3 = message */ DTRACE_PROBE4(zfs__dprintf, char *, newfile, char *, func, int, line, char *, buf); } static void history_str_free(char *buf) { kmem_free(buf, HIS_MAX_RECORD_LEN); } static char * history_str_get(zfs_cmd_t *zc) { char *buf; if (zc->zc_history == 0) return (NULL); buf = kmem_alloc(HIS_MAX_RECORD_LEN, KM_SLEEP); if (copyinstr((void *)(uintptr_t)zc->zc_history, buf, HIS_MAX_RECORD_LEN, NULL) != 0) { history_str_free(buf); return (NULL); } buf[HIS_MAX_RECORD_LEN -1] = '\0'; return (buf); } /* * Check to see if the named dataset is currently defined as bootable */ static boolean_t zfs_is_bootfs(const char *name) { objset_t *os; if (dmu_objset_hold(name, FTAG, &os) == 0) { boolean_t ret; ret = (dmu_objset_id(os) == spa_bootfs(dmu_objset_spa(os))); dmu_objset_rele(os, FTAG); return (ret); } return (B_FALSE); } /* * Return non-zero if the spa version is less than requested version. */ static int zfs_earlier_version(const char *name, int version) { spa_t *spa; if (spa_open(name, &spa, FTAG) == 0) { if (spa_version(spa) < version) { spa_close(spa, FTAG); return (1); } spa_close(spa, FTAG); } return (0); } /* * Return TRUE if the ZPL version is less than requested version. */ static boolean_t zpl_earlier_version(const char *name, int version) { objset_t *os; boolean_t rc = B_TRUE; if (dmu_objset_hold(name, FTAG, &os) == 0) { uint64_t zplversion; if (dmu_objset_type(os) != DMU_OST_ZFS) { dmu_objset_rele(os, FTAG); return (B_TRUE); } /* XXX reading from non-owned objset */ if (zfs_get_zplprop(os, ZFS_PROP_VERSION, &zplversion) == 0) rc = zplversion < version; dmu_objset_rele(os, FTAG); } return (rc); } static void zfs_log_history(zfs_cmd_t *zc) { spa_t *spa; char *buf; if ((buf = history_str_get(zc)) == NULL) return; if (spa_open(zc->zc_name, &spa, FTAG) == 0) { if (spa_version(spa) >= SPA_VERSION_ZPOOL_HISTORY) (void) spa_history_log(spa, buf); spa_close(spa, FTAG); } history_str_free(buf); } /* * Policy for top-level read operations (list pools). Requires no privileges, * and can be used in the local zone, as there is no associated dataset. */ /* ARGSUSED */ static int zfs_secpolicy_none(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (0); } /* * Policy for dataset read operations (list children, get statistics). Requires * no privileges, but must be visible in the local zone. */ /* ARGSUSED */ static int zfs_secpolicy_read(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { if (INGLOBALZONE(curthread) || zone_dataset_visible(zc->zc_name, NULL)) return (0); return (SET_ERROR(ENOENT)); } static int zfs_dozonecheck_impl(const char *dataset, uint64_t zoned, cred_t *cr) { int writable = 1; /* * The dataset must be visible by this zone -- check this first * so they don't see EPERM on something they shouldn't know about. */ if (!INGLOBALZONE(curthread) && !zone_dataset_visible(dataset, &writable)) return (SET_ERROR(ENOENT)); if (INGLOBALZONE(curthread)) { /* * If the fs is zoned, only root can access it from the * global zone. */ if (secpolicy_zfs(cr) && zoned) return (SET_ERROR(EPERM)); } else { /* * If we are in a local zone, the 'zoned' property must be set. */ if (!zoned) return (SET_ERROR(EPERM)); /* must be writable by this zone */ if (!writable) return (SET_ERROR(EPERM)); } return (0); } static int zfs_dozonecheck(const char *dataset, cred_t *cr) { uint64_t zoned; if (dsl_prop_get_integer(dataset, "jailed", &zoned, NULL)) return (SET_ERROR(ENOENT)); return (zfs_dozonecheck_impl(dataset, zoned, cr)); } static int zfs_dozonecheck_ds(const char *dataset, dsl_dataset_t *ds, cred_t *cr) { uint64_t zoned; if (dsl_prop_get_int_ds(ds, "jailed", &zoned)) return (SET_ERROR(ENOENT)); return (zfs_dozonecheck_impl(dataset, zoned, cr)); } static int zfs_secpolicy_write_perms_ds(const char *name, dsl_dataset_t *ds, const char *perm, cred_t *cr) { int error; error = zfs_dozonecheck_ds(name, ds, cr); if (error == 0) { error = secpolicy_zfs(cr); if (error != 0) error = dsl_deleg_access_impl(ds, perm, cr); } return (error); } static int zfs_secpolicy_write_perms(const char *name, const char *perm, cred_t *cr) { int error; dsl_dataset_t *ds; dsl_pool_t *dp; error = dsl_pool_hold(name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, name, FTAG, &ds); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } error = zfs_secpolicy_write_perms_ds(name, ds, perm, cr); dsl_dataset_rele(ds, FTAG); dsl_pool_rele(dp, FTAG); return (error); } #ifdef SECLABEL /* * Policy for setting the security label property. * * Returns 0 for success, non-zero for access and other errors. */ static int zfs_set_slabel_policy(const char *name, char *strval, cred_t *cr) { char ds_hexsl[MAXNAMELEN]; bslabel_t ds_sl, new_sl; boolean_t new_default = FALSE; uint64_t zoned; int needed_priv = -1; int error; /* First get the existing dataset label. */ error = dsl_prop_get(name, zfs_prop_to_name(ZFS_PROP_MLSLABEL), 1, sizeof (ds_hexsl), &ds_hexsl, NULL); if (error != 0) return (SET_ERROR(EPERM)); if (strcasecmp(strval, ZFS_MLSLABEL_DEFAULT) == 0) new_default = TRUE; /* The label must be translatable */ if (!new_default && (hexstr_to_label(strval, &new_sl) != 0)) return (SET_ERROR(EINVAL)); /* * In a non-global zone, disallow attempts to set a label that * doesn't match that of the zone; otherwise no other checks * are needed. */ if (!INGLOBALZONE(curproc)) { if (new_default || !blequal(&new_sl, CR_SL(CRED()))) return (SET_ERROR(EPERM)); return (0); } /* * For global-zone datasets (i.e., those whose zoned property is * "off", verify that the specified new label is valid for the * global zone. */ if (dsl_prop_get_integer(name, zfs_prop_to_name(ZFS_PROP_ZONED), &zoned, NULL)) return (SET_ERROR(EPERM)); if (!zoned) { if (zfs_check_global_label(name, strval) != 0) return (SET_ERROR(EPERM)); } /* * If the existing dataset label is nondefault, check if the * dataset is mounted (label cannot be changed while mounted). * Get the zfsvfs; if there isn't one, then the dataset isn't * mounted (or isn't a dataset, doesn't exist, ...). */ if (strcasecmp(ds_hexsl, ZFS_MLSLABEL_DEFAULT) != 0) { objset_t *os; static char *setsl_tag = "setsl_tag"; /* * Try to own the dataset; abort if there is any error, * (e.g., already mounted, in use, or other error). */ error = dmu_objset_own(name, DMU_OST_ZFS, B_TRUE, setsl_tag, &os); if (error != 0) return (SET_ERROR(EPERM)); dmu_objset_disown(os, setsl_tag); if (new_default) { needed_priv = PRIV_FILE_DOWNGRADE_SL; goto out_check; } if (hexstr_to_label(strval, &new_sl) != 0) return (SET_ERROR(EPERM)); if (blstrictdom(&ds_sl, &new_sl)) needed_priv = PRIV_FILE_DOWNGRADE_SL; else if (blstrictdom(&new_sl, &ds_sl)) needed_priv = PRIV_FILE_UPGRADE_SL; } else { /* dataset currently has a default label */ if (!new_default) needed_priv = PRIV_FILE_UPGRADE_SL; } out_check: if (needed_priv != -1) return (PRIV_POLICY(cr, needed_priv, B_FALSE, EPERM, NULL)); return (0); } #endif /* SECLABEL */ static int zfs_secpolicy_setprop(const char *dsname, zfs_prop_t prop, nvpair_t *propval, cred_t *cr) { char *strval; /* * Check permissions for special properties. */ switch (prop) { case ZFS_PROP_ZONED: /* * Disallow setting of 'zoned' from within a local zone. */ if (!INGLOBALZONE(curthread)) return (SET_ERROR(EPERM)); break; case ZFS_PROP_QUOTA: case ZFS_PROP_FILESYSTEM_LIMIT: case ZFS_PROP_SNAPSHOT_LIMIT: if (!INGLOBALZONE(curthread)) { uint64_t zoned; char setpoint[MAXNAMELEN]; /* * Unprivileged users are allowed to modify the * limit on things *under* (ie. contained by) * the thing they own. */ if (dsl_prop_get_integer(dsname, "jailed", &zoned, setpoint)) return (SET_ERROR(EPERM)); if (!zoned || strlen(dsname) <= strlen(setpoint)) return (SET_ERROR(EPERM)); } break; case ZFS_PROP_MLSLABEL: #ifdef SECLABEL if (!is_system_labeled()) return (SET_ERROR(EPERM)); if (nvpair_value_string(propval, &strval) == 0) { int err; err = zfs_set_slabel_policy(dsname, strval, CRED()); if (err != 0) return (err); } #else return (EOPNOTSUPP); #endif break; } return (zfs_secpolicy_write_perms(dsname, zfs_prop_to_name(prop), cr)); } /* ARGSUSED */ static int zfs_secpolicy_set_fsacl(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int error; error = zfs_dozonecheck(zc->zc_name, cr); if (error != 0) return (error); /* * permission to set permissions will be evaluated later in * dsl_deleg_can_allow() */ return (0); } /* ARGSUSED */ static int zfs_secpolicy_rollback(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_ROLLBACK, cr)); } /* ARGSUSED */ static int zfs_secpolicy_send(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { dsl_pool_t *dp; dsl_dataset_t *ds; char *cp; int error; /* * Generate the current snapshot name from the given objsetid, then * use that name for the secpolicy/zone checks. */ cp = strchr(zc->zc_name, '@'); if (cp == NULL) return (SET_ERROR(EINVAL)); error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold_obj(dp, zc->zc_sendobj, FTAG, &ds); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } dsl_dataset_name(ds, zc->zc_name); error = zfs_secpolicy_write_perms_ds(zc->zc_name, ds, ZFS_DELEG_PERM_SEND, cr); dsl_dataset_rele(ds, FTAG); dsl_pool_rele(dp, FTAG); return (error); } /* ARGSUSED */ static int zfs_secpolicy_send_new(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_SEND, cr)); } /* ARGSUSED */ static int zfs_secpolicy_deleg_share(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { vnode_t *vp; int error; if ((error = lookupname(zc->zc_value, UIO_SYSSPACE, NO_FOLLOW, NULL, &vp)) != 0) return (error); /* Now make sure mntpnt and dataset are ZFS */ if (strcmp(vp->v_vfsp->mnt_stat.f_fstypename, "zfs") != 0 || (strcmp((char *)refstr_value(vp->v_vfsp->vfs_resource), zc->zc_name) != 0)) { VN_RELE(vp); return (SET_ERROR(EPERM)); } VN_RELE(vp); return (dsl_deleg_access(zc->zc_name, ZFS_DELEG_PERM_SHARE, cr)); } int zfs_secpolicy_share(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { if (!INGLOBALZONE(curthread)) return (SET_ERROR(EPERM)); if (secpolicy_nfs(cr) == 0) { return (0); } else { return (zfs_secpolicy_deleg_share(zc, innvl, cr)); } } int zfs_secpolicy_smb_acl(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { if (!INGLOBALZONE(curthread)) return (SET_ERROR(EPERM)); if (secpolicy_smb(cr) == 0) { return (0); } else { return (zfs_secpolicy_deleg_share(zc, innvl, cr)); } } static int zfs_get_parent(const char *datasetname, char *parent, int parentsize) { char *cp; /* * Remove the @bla or /bla from the end of the name to get the parent. */ (void) strncpy(parent, datasetname, parentsize); cp = strrchr(parent, '@'); if (cp != NULL) { cp[0] = '\0'; } else { cp = strrchr(parent, '/'); if (cp == NULL) return (SET_ERROR(ENOENT)); cp[0] = '\0'; } return (0); } int zfs_secpolicy_destroy_perms(const char *name, cred_t *cr) { int error; if ((error = zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_MOUNT, cr)) != 0) return (error); return (zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_DESTROY, cr)); } /* ARGSUSED */ static int zfs_secpolicy_destroy(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_destroy_perms(zc->zc_name, cr)); } /* * Destroying snapshots with delegated permissions requires * descendant mount and destroy permissions. */ /* ARGSUSED */ static int zfs_secpolicy_destroy_snaps(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { nvlist_t *snaps; nvpair_t *pair, *nextpair; int error = 0; if (nvlist_lookup_nvlist(innvl, "snaps", &snaps) != 0) return (SET_ERROR(EINVAL)); for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; pair = nextpair) { nextpair = nvlist_next_nvpair(snaps, pair); error = zfs_secpolicy_destroy_perms(nvpair_name(pair), cr); if (error == ENOENT) { /* * Ignore any snapshots that don't exist (we consider * them "already destroyed"). Remove the name from the * nvl here in case the snapshot is created between * now and when we try to destroy it (in which case * we don't want to destroy it since we haven't * checked for permission). */ fnvlist_remove_nvpair(snaps, pair); error = 0; } if (error != 0) break; } return (error); } int zfs_secpolicy_rename_perms(const char *from, const char *to, cred_t *cr) { char parentname[MAXNAMELEN]; int error; if ((error = zfs_secpolicy_write_perms(from, ZFS_DELEG_PERM_RENAME, cr)) != 0) return (error); if ((error = zfs_secpolicy_write_perms(from, ZFS_DELEG_PERM_MOUNT, cr)) != 0) return (error); if ((error = zfs_get_parent(to, parentname, sizeof (parentname))) != 0) return (error); if ((error = zfs_secpolicy_write_perms(parentname, ZFS_DELEG_PERM_CREATE, cr)) != 0) return (error); if ((error = zfs_secpolicy_write_perms(parentname, ZFS_DELEG_PERM_MOUNT, cr)) != 0) return (error); return (error); } /* ARGSUSED */ static int zfs_secpolicy_rename(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { char *at = NULL; int error; if ((zc->zc_cookie & 1) != 0) { /* * This is recursive rename, so the starting snapshot might * not exist. Check file system or volume permission instead. */ at = strchr(zc->zc_name, '@'); if (at == NULL) return (EINVAL); *at = '\0'; } error = zfs_secpolicy_rename_perms(zc->zc_name, zc->zc_value, cr); if (at != NULL) *at = '@'; return (error); } /* ARGSUSED */ static int zfs_secpolicy_promote(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { dsl_pool_t *dp; dsl_dataset_t *clone; int error; error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_PROMOTE, cr); if (error != 0) return (error); error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, zc->zc_name, FTAG, &clone); if (error == 0) { char parentname[MAXNAMELEN]; dsl_dataset_t *origin = NULL; dsl_dir_t *dd; dd = clone->ds_dir; error = dsl_dataset_hold_obj(dd->dd_pool, dsl_dir_phys(dd)->dd_origin_obj, FTAG, &origin); if (error != 0) { dsl_dataset_rele(clone, FTAG); dsl_pool_rele(dp, FTAG); return (error); } error = zfs_secpolicy_write_perms_ds(zc->zc_name, clone, ZFS_DELEG_PERM_MOUNT, cr); dsl_dataset_name(origin, parentname); if (error == 0) { error = zfs_secpolicy_write_perms_ds(parentname, origin, ZFS_DELEG_PERM_PROMOTE, cr); } dsl_dataset_rele(clone, FTAG); dsl_dataset_rele(origin, FTAG); } dsl_pool_rele(dp, FTAG); return (error); } /* ARGSUSED */ static int zfs_secpolicy_recv(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int error; if ((error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_RECEIVE, cr)) != 0) return (error); if ((error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_MOUNT, cr)) != 0) return (error); return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_CREATE, cr)); } int zfs_secpolicy_snapshot_perms(const char *name, cred_t *cr) { return (zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_SNAPSHOT, cr)); } /* * Check for permission to create each snapshot in the nvlist. */ /* ARGSUSED */ static int zfs_secpolicy_snapshot(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { nvlist_t *snaps; int error; nvpair_t *pair; if (nvlist_lookup_nvlist(innvl, "snaps", &snaps) != 0) return (SET_ERROR(EINVAL)); for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; pair = nvlist_next_nvpair(snaps, pair)) { char *name = nvpair_name(pair); char *atp = strchr(name, '@'); if (atp == NULL) { error = SET_ERROR(EINVAL); break; } *atp = '\0'; error = zfs_secpolicy_snapshot_perms(name, cr); *atp = '@'; if (error != 0) break; } return (error); } /* * Check for permission to create each snapshot in the nvlist. */ /* ARGSUSED */ static int zfs_secpolicy_bookmark(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int error = 0; for (nvpair_t *pair = nvlist_next_nvpair(innvl, NULL); pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) { char *name = nvpair_name(pair); char *hashp = strchr(name, '#'); if (hashp == NULL) { error = SET_ERROR(EINVAL); break; } *hashp = '\0'; error = zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_BOOKMARK, cr); *hashp = '#'; if (error != 0) break; } return (error); } /* ARGSUSED */ static int zfs_secpolicy_destroy_bookmarks(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { nvpair_t *pair, *nextpair; int error = 0; for (pair = nvlist_next_nvpair(innvl, NULL); pair != NULL; pair = nextpair) { char *name = nvpair_name(pair); char *hashp = strchr(name, '#'); nextpair = nvlist_next_nvpair(innvl, pair); if (hashp == NULL) { error = SET_ERROR(EINVAL); break; } *hashp = '\0'; error = zfs_secpolicy_write_perms(name, ZFS_DELEG_PERM_DESTROY, cr); *hashp = '#'; if (error == ENOENT) { /* * Ignore any filesystems that don't exist (we consider * their bookmarks "already destroyed"). Remove * the name from the nvl here in case the filesystem * is created between now and when we try to destroy * the bookmark (in which case we don't want to * destroy it since we haven't checked for permission). */ fnvlist_remove_nvpair(innvl, pair); error = 0; } if (error != 0) break; } return (error); } /* ARGSUSED */ static int zfs_secpolicy_log_history(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { /* * Even root must have a proper TSD so that we know what pool * to log to. */ if (tsd_get(zfs_allow_log_key) == NULL) return (SET_ERROR(EPERM)); return (0); } static int zfs_secpolicy_create_clone(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { char parentname[MAXNAMELEN]; int error; char *origin; if ((error = zfs_get_parent(zc->zc_name, parentname, sizeof (parentname))) != 0) return (error); if (nvlist_lookup_string(innvl, "origin", &origin) == 0 && (error = zfs_secpolicy_write_perms(origin, ZFS_DELEG_PERM_CLONE, cr)) != 0) return (error); if ((error = zfs_secpolicy_write_perms(parentname, ZFS_DELEG_PERM_CREATE, cr)) != 0) return (error); return (zfs_secpolicy_write_perms(parentname, ZFS_DELEG_PERM_MOUNT, cr)); } /* * Policy for pool operations - create/destroy pools, add vdevs, etc. Requires * SYS_CONFIG privilege, which is not available in a local zone. */ /* ARGSUSED */ static int zfs_secpolicy_config(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { if (secpolicy_sys_config(cr, B_FALSE) != 0) return (SET_ERROR(EPERM)); return (0); } /* * Policy for object to name lookups. */ /* ARGSUSED */ static int zfs_secpolicy_diff(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int error; if ((error = secpolicy_sys_config(cr, B_FALSE)) == 0) return (0); error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_DIFF, cr); return (error); } /* * Policy for fault injection. Requires all privileges. */ /* ARGSUSED */ static int zfs_secpolicy_inject(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (secpolicy_zinject(cr)); } /* ARGSUSED */ static int zfs_secpolicy_inherit_prop(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { zfs_prop_t prop = zfs_name_to_prop(zc->zc_value); if (prop == ZPROP_INVAL) { if (!zfs_prop_user(zc->zc_value)) return (SET_ERROR(EINVAL)); return (zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_USERPROP, cr)); } else { return (zfs_secpolicy_setprop(zc->zc_name, prop, NULL, cr)); } } static int zfs_secpolicy_userspace_one(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int err = zfs_secpolicy_read(zc, innvl, cr); if (err) return (err); if (zc->zc_objset_type >= ZFS_NUM_USERQUOTA_PROPS) return (SET_ERROR(EINVAL)); if (zc->zc_value[0] == 0) { /* * They are asking about a posix uid/gid. If it's * themself, allow it. */ if (zc->zc_objset_type == ZFS_PROP_USERUSED || zc->zc_objset_type == ZFS_PROP_USERQUOTA) { if (zc->zc_guid == crgetuid(cr)) return (0); } else { if (groupmember(zc->zc_guid, cr)) return (0); } } return (zfs_secpolicy_write_perms(zc->zc_name, userquota_perms[zc->zc_objset_type], cr)); } static int zfs_secpolicy_userspace_many(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { int err = zfs_secpolicy_read(zc, innvl, cr); if (err) return (err); if (zc->zc_objset_type >= ZFS_NUM_USERQUOTA_PROPS) return (SET_ERROR(EINVAL)); return (zfs_secpolicy_write_perms(zc->zc_name, userquota_perms[zc->zc_objset_type], cr)); } /* ARGSUSED */ static int zfs_secpolicy_userspace_upgrade(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { return (zfs_secpolicy_setprop(zc->zc_name, ZFS_PROP_VERSION, NULL, cr)); } /* ARGSUSED */ static int zfs_secpolicy_hold(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { nvpair_t *pair; nvlist_t *holds; int error; error = nvlist_lookup_nvlist(innvl, "holds", &holds); if (error != 0) return (SET_ERROR(EINVAL)); for (pair = nvlist_next_nvpair(holds, NULL); pair != NULL; pair = nvlist_next_nvpair(holds, pair)) { char fsname[MAXNAMELEN]; error = dmu_fsname(nvpair_name(pair), fsname); if (error != 0) return (error); error = zfs_secpolicy_write_perms(fsname, ZFS_DELEG_PERM_HOLD, cr); if (error != 0) return (error); } return (0); } /* ARGSUSED */ static int zfs_secpolicy_release(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { nvpair_t *pair; int error; for (pair = nvlist_next_nvpair(innvl, NULL); pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) { char fsname[MAXNAMELEN]; error = dmu_fsname(nvpair_name(pair), fsname); if (error != 0) return (error); error = zfs_secpolicy_write_perms(fsname, ZFS_DELEG_PERM_RELEASE, cr); if (error != 0) return (error); } return (0); } /* * Policy for allowing temporary snapshots to be taken or released */ static int zfs_secpolicy_tmp_snapshot(zfs_cmd_t *zc, nvlist_t *innvl, cred_t *cr) { /* * A temporary snapshot is the same as a snapshot, * hold, destroy and release all rolled into one. * Delegated diff alone is sufficient that we allow this. */ int error; if ((error = zfs_secpolicy_write_perms(zc->zc_name, ZFS_DELEG_PERM_DIFF, cr)) == 0) return (0); error = zfs_secpolicy_snapshot_perms(zc->zc_name, cr); if (error == 0) error = zfs_secpolicy_hold(zc, innvl, cr); if (error == 0) error = zfs_secpolicy_release(zc, innvl, cr); if (error == 0) error = zfs_secpolicy_destroy(zc, innvl, cr); return (error); } /* * Returns the nvlist as specified by the user in the zfs_cmd_t. */ static int get_nvlist(uint64_t nvl, uint64_t size, int iflag, nvlist_t **nvp) { char *packed; int error; nvlist_t *list = NULL; /* * Read in and unpack the user-supplied nvlist. */ if (size == 0) return (SET_ERROR(EINVAL)); packed = kmem_alloc(size, KM_SLEEP); if ((error = ddi_copyin((void *)(uintptr_t)nvl, packed, size, iflag)) != 0) { kmem_free(packed, size); return (SET_ERROR(EFAULT)); } if ((error = nvlist_unpack(packed, size, &list, 0)) != 0) { kmem_free(packed, size); return (error); } kmem_free(packed, size); *nvp = list; return (0); } /* * Reduce the size of this nvlist until it can be serialized in 'max' bytes. * Entries will be removed from the end of the nvlist, and one int32 entry * named "N_MORE_ERRORS" will be added indicating how many entries were * removed. */ static int nvlist_smush(nvlist_t *errors, size_t max) { size_t size; size = fnvlist_size(errors); if (size > max) { nvpair_t *more_errors; int n = 0; if (max < 1024) return (SET_ERROR(ENOMEM)); fnvlist_add_int32(errors, ZPROP_N_MORE_ERRORS, 0); more_errors = nvlist_prev_nvpair(errors, NULL); do { nvpair_t *pair = nvlist_prev_nvpair(errors, more_errors); fnvlist_remove_nvpair(errors, pair); n++; size = fnvlist_size(errors); } while (size > max); fnvlist_remove_nvpair(errors, more_errors); fnvlist_add_int32(errors, ZPROP_N_MORE_ERRORS, n); ASSERT3U(fnvlist_size(errors), <=, max); } return (0); } static int put_nvlist(zfs_cmd_t *zc, nvlist_t *nvl) { char *packed = NULL; int error = 0; size_t size; size = fnvlist_size(nvl); if (size > zc->zc_nvlist_dst_size) { /* * Solaris returns ENOMEM here, because even if an error is * returned from an ioctl(2), new zc_nvlist_dst_size will be * passed to the userland. This is not the case for FreeBSD. * We need to return 0, so the kernel will copy the * zc_nvlist_dst_size back and the userland can discover that a * bigger buffer is needed. */ error = 0; } else { packed = fnvlist_pack(nvl, &size); if (ddi_copyout(packed, (void *)(uintptr_t)zc->zc_nvlist_dst, size, zc->zc_iflags) != 0) error = SET_ERROR(EFAULT); fnvlist_pack_free(packed, size); } zc->zc_nvlist_dst_size = size; zc->zc_nvlist_dst_filled = B_TRUE; return (error); } static int getzfsvfs(const char *dsname, zfsvfs_t **zfvp) { objset_t *os; + vfs_t *vfsp; int error; error = dmu_objset_hold(dsname, FTAG, &os); if (error != 0) return (error); if (dmu_objset_type(os) != DMU_OST_ZFS) { dmu_objset_rele(os, FTAG); return (SET_ERROR(EINVAL)); } mutex_enter(&os->os_user_ptr_lock); *zfvp = dmu_objset_get_user(os); if (*zfvp) { -#ifdef illumos - VFS_HOLD((*zfvp)->z_vfs); -#else - if (vfs_busy((*zfvp)->z_vfs, 0) != 0) { - *zfvp = NULL; - error = SET_ERROR(ESRCH); - } -#endif + vfsp = (*zfvp)->z_vfs; + vfs_ref(vfsp); } else { error = SET_ERROR(ESRCH); } mutex_exit(&os->os_user_ptr_lock); dmu_objset_rele(os, FTAG); + if (error == 0) { + error = vfs_busy(vfsp, 0); + vfs_rel(vfsp); + if (error != 0) { + *zfvp = NULL; + error = SET_ERROR(ESRCH); + } + } return (error); } /* * Find a zfsvfs_t for a mounted filesystem, or create our own, in which * case its z_vfs will be NULL, and it will be opened as the owner. * If 'writer' is set, the z_teardown_lock will be held for RW_WRITER, * which prevents all vnode ops from running. */ static int zfsvfs_hold(const char *name, void *tag, zfsvfs_t **zfvp, boolean_t writer) { int error = 0; if (getzfsvfs(name, zfvp) != 0) error = zfsvfs_create(name, zfvp); if (error == 0) { rrm_enter(&(*zfvp)->z_teardown_lock, (writer) ? RW_WRITER : RW_READER, tag); if ((*zfvp)->z_unmounted) { /* * XXX we could probably try again, since the unmounting * thread should be just about to disassociate the * objset from the zfsvfs. */ rrm_exit(&(*zfvp)->z_teardown_lock, tag); return (SET_ERROR(EBUSY)); } } return (error); } static void zfsvfs_rele(zfsvfs_t *zfsvfs, void *tag) { rrm_exit(&zfsvfs->z_teardown_lock, tag); if (zfsvfs->z_vfs) { #ifdef illumos VFS_RELE(zfsvfs->z_vfs); #else vfs_unbusy(zfsvfs->z_vfs); #endif } else { dmu_objset_disown(zfsvfs->z_os, zfsvfs); zfsvfs_free(zfsvfs); } } static int zfs_ioc_pool_create(zfs_cmd_t *zc) { int error; nvlist_t *config, *props = NULL; nvlist_t *rootprops = NULL; nvlist_t *zplprops = NULL; if (error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config)) return (error); if (zc->zc_nvlist_src_size != 0 && (error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &props))) { nvlist_free(config); return (error); } if (props) { nvlist_t *nvl = NULL; uint64_t version = SPA_VERSION; (void) nvlist_lookup_uint64(props, zpool_prop_to_name(ZPOOL_PROP_VERSION), &version); if (!SPA_VERSION_IS_SUPPORTED(version)) { error = SET_ERROR(EINVAL); goto pool_props_bad; } (void) nvlist_lookup_nvlist(props, ZPOOL_ROOTFS_PROPS, &nvl); if (nvl) { error = nvlist_dup(nvl, &rootprops, KM_SLEEP); if (error != 0) { nvlist_free(config); nvlist_free(props); return (error); } (void) nvlist_remove_all(props, ZPOOL_ROOTFS_PROPS); } VERIFY(nvlist_alloc(&zplprops, NV_UNIQUE_NAME, KM_SLEEP) == 0); error = zfs_fill_zplprops_root(version, rootprops, zplprops, NULL); if (error != 0) goto pool_props_bad; } error = spa_create(zc->zc_name, config, props, zplprops); /* * Set the remaining root properties */ if (!error && (error = zfs_set_prop_nvlist(zc->zc_name, ZPROP_SRC_LOCAL, rootprops, NULL)) != 0) (void) spa_destroy(zc->zc_name); pool_props_bad: nvlist_free(rootprops); nvlist_free(zplprops); nvlist_free(config); nvlist_free(props); return (error); } static int zfs_ioc_pool_destroy(zfs_cmd_t *zc) { int error; zfs_log_history(zc); error = spa_destroy(zc->zc_name); if (error == 0) zvol_remove_minors(zc->zc_name); return (error); } static int zfs_ioc_pool_import(zfs_cmd_t *zc) { nvlist_t *config, *props = NULL; uint64_t guid; int error; if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config)) != 0) return (error); if (zc->zc_nvlist_src_size != 0 && (error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &props))) { nvlist_free(config); return (error); } if (nvlist_lookup_uint64(config, ZPOOL_CONFIG_POOL_GUID, &guid) != 0 || guid != zc->zc_guid) error = SET_ERROR(EINVAL); else error = spa_import(zc->zc_name, config, props, zc->zc_cookie); if (zc->zc_nvlist_dst != 0) { int err; if ((err = put_nvlist(zc, config)) != 0) error = err; } nvlist_free(config); nvlist_free(props); return (error); } static int zfs_ioc_pool_export(zfs_cmd_t *zc) { int error; boolean_t force = (boolean_t)zc->zc_cookie; boolean_t hardforce = (boolean_t)zc->zc_guid; zfs_log_history(zc); error = spa_export(zc->zc_name, NULL, force, hardforce); if (error == 0) zvol_remove_minors(zc->zc_name); return (error); } static int zfs_ioc_pool_configs(zfs_cmd_t *zc) { nvlist_t *configs; int error; if ((configs = spa_all_configs(&zc->zc_cookie)) == NULL) return (SET_ERROR(EEXIST)); error = put_nvlist(zc, configs); nvlist_free(configs); return (error); } /* * inputs: * zc_name name of the pool * * outputs: * zc_cookie real errno * zc_nvlist_dst config nvlist * zc_nvlist_dst_size size of config nvlist */ static int zfs_ioc_pool_stats(zfs_cmd_t *zc) { nvlist_t *config; int error; int ret = 0; error = spa_get_stats(zc->zc_name, &config, zc->zc_value, sizeof (zc->zc_value)); if (config != NULL) { ret = put_nvlist(zc, config); nvlist_free(config); /* * The config may be present even if 'error' is non-zero. * In this case we return success, and preserve the real errno * in 'zc_cookie'. */ zc->zc_cookie = error; } else { ret = error; } return (ret); } /* * Try to import the given pool, returning pool stats as appropriate so that * user land knows which devices are available and overall pool health. */ static int zfs_ioc_pool_tryimport(zfs_cmd_t *zc) { nvlist_t *tryconfig, *config; int error; if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &tryconfig)) != 0) return (error); config = spa_tryimport(tryconfig); nvlist_free(tryconfig); if (config == NULL) return (SET_ERROR(EINVAL)); error = put_nvlist(zc, config); nvlist_free(config); return (error); } /* * inputs: * zc_name name of the pool * zc_cookie scan func (pool_scan_func_t) */ static int zfs_ioc_pool_scan(zfs_cmd_t *zc) { spa_t *spa; int error; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); if (zc->zc_cookie == POOL_SCAN_NONE) error = spa_scan_stop(spa); else error = spa_scan(spa, zc->zc_cookie); spa_close(spa, FTAG); return (error); } static int zfs_ioc_pool_freeze(zfs_cmd_t *zc) { spa_t *spa; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error == 0) { spa_freeze(spa); spa_close(spa, FTAG); } return (error); } static int zfs_ioc_pool_upgrade(zfs_cmd_t *zc) { spa_t *spa; int error; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); if (zc->zc_cookie < spa_version(spa) || !SPA_VERSION_IS_SUPPORTED(zc->zc_cookie)) { spa_close(spa, FTAG); return (SET_ERROR(EINVAL)); } spa_upgrade(spa, zc->zc_cookie); spa_close(spa, FTAG); return (error); } static int zfs_ioc_pool_get_history(zfs_cmd_t *zc) { spa_t *spa; char *hist_buf; uint64_t size; int error; if ((size = zc->zc_history_len) == 0) return (SET_ERROR(EINVAL)); if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); if (spa_version(spa) < SPA_VERSION_ZPOOL_HISTORY) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } hist_buf = kmem_alloc(size, KM_SLEEP); if ((error = spa_history_get(spa, &zc->zc_history_offset, &zc->zc_history_len, hist_buf)) == 0) { error = ddi_copyout(hist_buf, (void *)(uintptr_t)zc->zc_history, zc->zc_history_len, zc->zc_iflags); } spa_close(spa, FTAG); kmem_free(hist_buf, size); return (error); } static int zfs_ioc_pool_reguid(zfs_cmd_t *zc) { spa_t *spa; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error == 0) { error = spa_change_guid(spa); spa_close(spa, FTAG); } return (error); } static int zfs_ioc_dsobj_to_dsname(zfs_cmd_t *zc) { return (dsl_dsobj_to_dsname(zc->zc_name, zc->zc_obj, zc->zc_value)); } /* * inputs: * zc_name name of filesystem * zc_obj object to find * * outputs: * zc_value name of object */ static int zfs_ioc_obj_to_path(zfs_cmd_t *zc) { objset_t *os; int error; /* XXX reading from objset not owned */ if ((error = dmu_objset_hold(zc->zc_name, FTAG, &os)) != 0) return (error); if (dmu_objset_type(os) != DMU_OST_ZFS) { dmu_objset_rele(os, FTAG); return (SET_ERROR(EINVAL)); } error = zfs_obj_to_path(os, zc->zc_obj, zc->zc_value, sizeof (zc->zc_value)); dmu_objset_rele(os, FTAG); return (error); } /* * inputs: * zc_name name of filesystem * zc_obj object to find * * outputs: * zc_stat stats on object * zc_value path to object */ static int zfs_ioc_obj_to_stats(zfs_cmd_t *zc) { objset_t *os; int error; /* XXX reading from objset not owned */ if ((error = dmu_objset_hold(zc->zc_name, FTAG, &os)) != 0) return (error); if (dmu_objset_type(os) != DMU_OST_ZFS) { dmu_objset_rele(os, FTAG); return (SET_ERROR(EINVAL)); } error = zfs_obj_to_stats(os, zc->zc_obj, &zc->zc_stat, zc->zc_value, sizeof (zc->zc_value)); dmu_objset_rele(os, FTAG); return (error); } static int zfs_ioc_vdev_add(zfs_cmd_t *zc) { spa_t *spa; int error; nvlist_t *config, **l2cache, **spares; uint_t nl2cache = 0, nspares = 0; error = spa_open(zc->zc_name, &spa, FTAG); if (error != 0) return (error); error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config); (void) nvlist_lookup_nvlist_array(config, ZPOOL_CONFIG_L2CACHE, &l2cache, &nl2cache); (void) nvlist_lookup_nvlist_array(config, ZPOOL_CONFIG_SPARES, &spares, &nspares); #ifdef illumos /* * A root pool with concatenated devices is not supported. * Thus, can not add a device to a root pool. * * Intent log device can not be added to a rootpool because * during mountroot, zil is replayed, a seperated log device * can not be accessed during the mountroot time. * * l2cache and spare devices are ok to be added to a rootpool. */ if (spa_bootfs(spa) != 0 && nl2cache == 0 && nspares == 0) { nvlist_free(config); spa_close(spa, FTAG); return (SET_ERROR(EDOM)); } #endif /* illumos */ if (error == 0) { error = spa_vdev_add(spa, config); nvlist_free(config); } spa_close(spa, FTAG); return (error); } /* * inputs: * zc_name name of the pool * zc_nvlist_conf nvlist of devices to remove * zc_cookie to stop the remove? */ static int zfs_ioc_vdev_remove(zfs_cmd_t *zc) { spa_t *spa; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error != 0) return (error); error = spa_vdev_remove(spa, zc->zc_guid, B_FALSE); spa_close(spa, FTAG); return (error); } static int zfs_ioc_vdev_set_state(zfs_cmd_t *zc) { spa_t *spa; int error; vdev_state_t newstate = VDEV_STATE_UNKNOWN; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); switch (zc->zc_cookie) { case VDEV_STATE_ONLINE: error = vdev_online(spa, zc->zc_guid, zc->zc_obj, &newstate); break; case VDEV_STATE_OFFLINE: error = vdev_offline(spa, zc->zc_guid, zc->zc_obj); break; case VDEV_STATE_FAULTED: if (zc->zc_obj != VDEV_AUX_ERR_EXCEEDED && zc->zc_obj != VDEV_AUX_EXTERNAL) zc->zc_obj = VDEV_AUX_ERR_EXCEEDED; error = vdev_fault(spa, zc->zc_guid, zc->zc_obj); break; case VDEV_STATE_DEGRADED: if (zc->zc_obj != VDEV_AUX_ERR_EXCEEDED && zc->zc_obj != VDEV_AUX_EXTERNAL) zc->zc_obj = VDEV_AUX_ERR_EXCEEDED; error = vdev_degrade(spa, zc->zc_guid, zc->zc_obj); break; default: error = SET_ERROR(EINVAL); } zc->zc_cookie = newstate; spa_close(spa, FTAG); return (error); } static int zfs_ioc_vdev_attach(zfs_cmd_t *zc) { spa_t *spa; int replacing = zc->zc_cookie; nvlist_t *config; int error; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); if ((error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config)) == 0) { error = spa_vdev_attach(spa, zc->zc_guid, config, replacing); nvlist_free(config); } spa_close(spa, FTAG); return (error); } static int zfs_ioc_vdev_detach(zfs_cmd_t *zc) { spa_t *spa; int error; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); error = spa_vdev_detach(spa, zc->zc_guid, 0, B_FALSE); spa_close(spa, FTAG); return (error); } static int zfs_ioc_vdev_split(zfs_cmd_t *zc) { spa_t *spa; nvlist_t *config, *props = NULL; int error; boolean_t exp = !!(zc->zc_cookie & ZPOOL_EXPORT_AFTER_SPLIT); if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); if (error = get_nvlist(zc->zc_nvlist_conf, zc->zc_nvlist_conf_size, zc->zc_iflags, &config)) { spa_close(spa, FTAG); return (error); } if (zc->zc_nvlist_src_size != 0 && (error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &props))) { spa_close(spa, FTAG); nvlist_free(config); return (error); } error = spa_vdev_split_mirror(spa, zc->zc_string, config, props, exp); spa_close(spa, FTAG); nvlist_free(config); nvlist_free(props); return (error); } static int zfs_ioc_vdev_setpath(zfs_cmd_t *zc) { spa_t *spa; char *path = zc->zc_value; uint64_t guid = zc->zc_guid; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error != 0) return (error); error = spa_vdev_setpath(spa, guid, path); spa_close(spa, FTAG); return (error); } static int zfs_ioc_vdev_setfru(zfs_cmd_t *zc) { spa_t *spa; char *fru = zc->zc_value; uint64_t guid = zc->zc_guid; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error != 0) return (error); error = spa_vdev_setfru(spa, guid, fru); spa_close(spa, FTAG); return (error); } static int zfs_ioc_objset_stats_impl(zfs_cmd_t *zc, objset_t *os) { int error = 0; nvlist_t *nv; dmu_objset_fast_stat(os, &zc->zc_objset_stats); if (zc->zc_nvlist_dst != 0 && (error = dsl_prop_get_all(os, &nv)) == 0) { dmu_objset_stats(os, nv); /* * NB: zvol_get_stats() will read the objset contents, * which we aren't supposed to do with a * DS_MODE_USER hold, because it could be * inconsistent. So this is a bit of a workaround... * XXX reading with out owning */ if (!zc->zc_objset_stats.dds_inconsistent && dmu_objset_type(os) == DMU_OST_ZVOL) { error = zvol_get_stats(os, nv); if (error == EIO) return (error); VERIFY0(error); } error = put_nvlist(zc, nv); nvlist_free(nv); } return (error); } /* * inputs: * zc_name name of filesystem * zc_nvlist_dst_size size of buffer for property nvlist * * outputs: * zc_objset_stats stats * zc_nvlist_dst property nvlist * zc_nvlist_dst_size size of property nvlist */ static int zfs_ioc_objset_stats(zfs_cmd_t *zc) { objset_t *os; int error; error = dmu_objset_hold(zc->zc_name, FTAG, &os); if (error == 0) { error = zfs_ioc_objset_stats_impl(zc, os); dmu_objset_rele(os, FTAG); } if (error == ENOMEM) error = 0; return (error); } /* * inputs: * zc_name name of filesystem * zc_nvlist_dst_size size of buffer for property nvlist * * outputs: * zc_nvlist_dst received property nvlist * zc_nvlist_dst_size size of received property nvlist * * Gets received properties (distinct from local properties on or after * SPA_VERSION_RECVD_PROPS) for callers who want to differentiate received from * local property values. */ static int zfs_ioc_objset_recvd_props(zfs_cmd_t *zc) { int error = 0; nvlist_t *nv; /* * Without this check, we would return local property values if the * caller has not already received properties on or after * SPA_VERSION_RECVD_PROPS. */ if (!dsl_prop_get_hasrecvd(zc->zc_name)) return (SET_ERROR(ENOTSUP)); if (zc->zc_nvlist_dst != 0 && (error = dsl_prop_get_received(zc->zc_name, &nv)) == 0) { error = put_nvlist(zc, nv); nvlist_free(nv); } return (error); } static int nvl_add_zplprop(objset_t *os, nvlist_t *props, zfs_prop_t prop) { uint64_t value; int error; /* * zfs_get_zplprop() will either find a value or give us * the default value (if there is one). */ if ((error = zfs_get_zplprop(os, prop, &value)) != 0) return (error); VERIFY(nvlist_add_uint64(props, zfs_prop_to_name(prop), value) == 0); return (0); } /* * inputs: * zc_name name of filesystem * zc_nvlist_dst_size size of buffer for zpl property nvlist * * outputs: * zc_nvlist_dst zpl property nvlist * zc_nvlist_dst_size size of zpl property nvlist */ static int zfs_ioc_objset_zplprops(zfs_cmd_t *zc) { objset_t *os; int err; /* XXX reading without owning */ if (err = dmu_objset_hold(zc->zc_name, FTAG, &os)) return (err); dmu_objset_fast_stat(os, &zc->zc_objset_stats); /* * NB: nvl_add_zplprop() will read the objset contents, * which we aren't supposed to do with a DS_MODE_USER * hold, because it could be inconsistent. */ if (zc->zc_nvlist_dst != 0 && !zc->zc_objset_stats.dds_inconsistent && dmu_objset_type(os) == DMU_OST_ZFS) { nvlist_t *nv; VERIFY(nvlist_alloc(&nv, NV_UNIQUE_NAME, KM_SLEEP) == 0); if ((err = nvl_add_zplprop(os, nv, ZFS_PROP_VERSION)) == 0 && (err = nvl_add_zplprop(os, nv, ZFS_PROP_NORMALIZE)) == 0 && (err = nvl_add_zplprop(os, nv, ZFS_PROP_UTF8ONLY)) == 0 && (err = nvl_add_zplprop(os, nv, ZFS_PROP_CASE)) == 0) err = put_nvlist(zc, nv); nvlist_free(nv); } else { err = SET_ERROR(ENOENT); } dmu_objset_rele(os, FTAG); return (err); } boolean_t dataset_name_hidden(const char *name) { /* * Skip over datasets that are not visible in this zone, * internal datasets (which have a $ in their name), and * temporary datasets (which have a % in their name). */ if (strchr(name, '$') != NULL) return (B_TRUE); if (strchr(name, '%') != NULL) return (B_TRUE); if (!INGLOBALZONE(curthread) && !zone_dataset_visible(name, NULL)) return (B_TRUE); return (B_FALSE); } /* * inputs: * zc_name name of filesystem * zc_cookie zap cursor * zc_nvlist_dst_size size of buffer for property nvlist * * outputs: * zc_name name of next filesystem * zc_cookie zap cursor * zc_objset_stats stats * zc_nvlist_dst property nvlist * zc_nvlist_dst_size size of property nvlist */ static int zfs_ioc_dataset_list_next(zfs_cmd_t *zc) { objset_t *os; int error; char *p; size_t orig_len = strlen(zc->zc_name); top: if (error = dmu_objset_hold(zc->zc_name, FTAG, &os)) { if (error == ENOENT) error = SET_ERROR(ESRCH); return (error); } p = strrchr(zc->zc_name, '/'); if (p == NULL || p[1] != '\0') (void) strlcat(zc->zc_name, "/", sizeof (zc->zc_name)); p = zc->zc_name + strlen(zc->zc_name); do { error = dmu_dir_list_next(os, sizeof (zc->zc_name) - (p - zc->zc_name), p, NULL, &zc->zc_cookie); if (error == ENOENT) error = SET_ERROR(ESRCH); } while (error == 0 && dataset_name_hidden(zc->zc_name)); dmu_objset_rele(os, FTAG); /* * If it's an internal dataset (ie. with a '$' in its name), * don't try to get stats for it, otherwise we'll return ENOENT. */ if (error == 0 && strchr(zc->zc_name, '$') == NULL) { error = zfs_ioc_objset_stats(zc); /* fill in the stats */ if (error == ENOENT) { /* We lost a race with destroy, get the next one. */ zc->zc_name[orig_len] = '\0'; goto top; } } return (error); } /* * inputs: * zc_name name of filesystem * zc_cookie zap cursor * zc_nvlist_dst_size size of buffer for property nvlist * zc_simple when set, only name is requested * * outputs: * zc_name name of next snapshot * zc_objset_stats stats * zc_nvlist_dst property nvlist * zc_nvlist_dst_size size of property nvlist */ static int zfs_ioc_snapshot_list_next(zfs_cmd_t *zc) { objset_t *os; int error; error = dmu_objset_hold(zc->zc_name, FTAG, &os); if (error != 0) { return (error == ENOENT ? ESRCH : error); } /* * A dataset name of maximum length cannot have any snapshots, * so exit immediately. */ if (strlcat(zc->zc_name, "@", sizeof (zc->zc_name)) >= MAXNAMELEN) { dmu_objset_rele(os, FTAG); return (SET_ERROR(ESRCH)); } error = dmu_snapshot_list_next(os, sizeof (zc->zc_name) - strlen(zc->zc_name), zc->zc_name + strlen(zc->zc_name), &zc->zc_obj, &zc->zc_cookie, NULL); if (error == 0 && !zc->zc_simple) { dsl_dataset_t *ds; dsl_pool_t *dp = os->os_dsl_dataset->ds_dir->dd_pool; error = dsl_dataset_hold_obj(dp, zc->zc_obj, FTAG, &ds); if (error == 0) { objset_t *ossnap; error = dmu_objset_from_ds(ds, &ossnap); if (error == 0) error = zfs_ioc_objset_stats_impl(zc, ossnap); dsl_dataset_rele(ds, FTAG); } } else if (error == ENOENT) { error = SET_ERROR(ESRCH); } dmu_objset_rele(os, FTAG); /* if we failed, undo the @ that we tacked on to zc_name */ if (error != 0) *strchr(zc->zc_name, '@') = '\0'; return (error); } static int zfs_prop_set_userquota(const char *dsname, nvpair_t *pair) { const char *propname = nvpair_name(pair); uint64_t *valary; unsigned int vallen; const char *domain; char *dash; zfs_userquota_prop_t type; uint64_t rid; uint64_t quota; zfsvfs_t *zfsvfs; int err; if (nvpair_type(pair) == DATA_TYPE_NVLIST) { nvlist_t *attrs; VERIFY(nvpair_value_nvlist(pair, &attrs) == 0); if (nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &pair) != 0) return (SET_ERROR(EINVAL)); } /* * A correctly constructed propname is encoded as * userquota@-. */ if ((dash = strchr(propname, '-')) == NULL || nvpair_value_uint64_array(pair, &valary, &vallen) != 0 || vallen != 3) return (SET_ERROR(EINVAL)); domain = dash + 1; type = valary[0]; rid = valary[1]; quota = valary[2]; err = zfsvfs_hold(dsname, FTAG, &zfsvfs, B_FALSE); if (err == 0) { err = zfs_set_userquota(zfsvfs, type, domain, rid, quota); zfsvfs_rele(zfsvfs, FTAG); } return (err); } /* * If the named property is one that has a special function to set its value, * return 0 on success and a positive error code on failure; otherwise if it is * not one of the special properties handled by this function, return -1. * * XXX: It would be better for callers of the property interface if we handled * these special cases in dsl_prop.c (in the dsl layer). */ static int zfs_prop_set_special(const char *dsname, zprop_source_t source, nvpair_t *pair) { const char *propname = nvpair_name(pair); zfs_prop_t prop = zfs_name_to_prop(propname); uint64_t intval; int err = -1; if (prop == ZPROP_INVAL) { if (zfs_prop_userquota(propname)) return (zfs_prop_set_userquota(dsname, pair)); return (-1); } if (nvpair_type(pair) == DATA_TYPE_NVLIST) { nvlist_t *attrs; VERIFY(nvpair_value_nvlist(pair, &attrs) == 0); VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &pair) == 0); } if (zfs_prop_get_type(prop) == PROP_TYPE_STRING) return (-1); VERIFY(0 == nvpair_value_uint64(pair, &intval)); switch (prop) { case ZFS_PROP_QUOTA: err = dsl_dir_set_quota(dsname, source, intval); break; case ZFS_PROP_REFQUOTA: err = dsl_dataset_set_refquota(dsname, source, intval); break; case ZFS_PROP_FILESYSTEM_LIMIT: case ZFS_PROP_SNAPSHOT_LIMIT: if (intval == UINT64_MAX) { /* clearing the limit, just do it */ err = 0; } else { err = dsl_dir_activate_fs_ss_limit(dsname); } /* * Set err to -1 to force the zfs_set_prop_nvlist code down the * default path to set the value in the nvlist. */ if (err == 0) err = -1; break; case ZFS_PROP_RESERVATION: err = dsl_dir_set_reservation(dsname, source, intval); break; case ZFS_PROP_REFRESERVATION: err = dsl_dataset_set_refreservation(dsname, source, intval); break; case ZFS_PROP_VOLSIZE: err = zvol_set_volsize(dsname, intval); break; case ZFS_PROP_VERSION: { zfsvfs_t *zfsvfs; if ((err = zfsvfs_hold(dsname, FTAG, &zfsvfs, B_TRUE)) != 0) break; err = zfs_set_version(zfsvfs, intval); zfsvfs_rele(zfsvfs, FTAG); if (err == 0 && intval >= ZPL_VERSION_USERSPACE) { zfs_cmd_t *zc; zc = kmem_zalloc(sizeof (zfs_cmd_t), KM_SLEEP); (void) strcpy(zc->zc_name, dsname); (void) zfs_ioc_userspace_upgrade(zc); kmem_free(zc, sizeof (zfs_cmd_t)); } break; } default: err = -1; } return (err); } /* * This function is best effort. If it fails to set any of the given properties, * it continues to set as many as it can and returns the last error * encountered. If the caller provides a non-NULL errlist, it will be filled in * with the list of names of all the properties that failed along with the * corresponding error numbers. * * If every property is set successfully, zero is returned and errlist is not * modified. */ int zfs_set_prop_nvlist(const char *dsname, zprop_source_t source, nvlist_t *nvl, nvlist_t *errlist) { nvpair_t *pair; nvpair_t *propval; int rv = 0; uint64_t intval; char *strval; nvlist_t *genericnvl = fnvlist_alloc(); nvlist_t *retrynvl = fnvlist_alloc(); retry: pair = NULL; while ((pair = nvlist_next_nvpair(nvl, pair)) != NULL) { const char *propname = nvpair_name(pair); zfs_prop_t prop = zfs_name_to_prop(propname); int err = 0; /* decode the property value */ propval = pair; if (nvpair_type(pair) == DATA_TYPE_NVLIST) { nvlist_t *attrs; attrs = fnvpair_value_nvlist(pair); if (nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &propval) != 0) err = SET_ERROR(EINVAL); } /* Validate value type */ if (err == 0 && prop == ZPROP_INVAL) { if (zfs_prop_user(propname)) { if (nvpair_type(propval) != DATA_TYPE_STRING) err = SET_ERROR(EINVAL); } else if (zfs_prop_userquota(propname)) { if (nvpair_type(propval) != DATA_TYPE_UINT64_ARRAY) err = SET_ERROR(EINVAL); } else { err = SET_ERROR(EINVAL); } } else if (err == 0) { if (nvpair_type(propval) == DATA_TYPE_STRING) { if (zfs_prop_get_type(prop) != PROP_TYPE_STRING) err = SET_ERROR(EINVAL); } else if (nvpair_type(propval) == DATA_TYPE_UINT64) { const char *unused; intval = fnvpair_value_uint64(propval); switch (zfs_prop_get_type(prop)) { case PROP_TYPE_NUMBER: break; case PROP_TYPE_STRING: err = SET_ERROR(EINVAL); break; case PROP_TYPE_INDEX: if (zfs_prop_index_to_string(prop, intval, &unused) != 0) err = SET_ERROR(EINVAL); break; default: cmn_err(CE_PANIC, "unknown property type"); } } else { err = SET_ERROR(EINVAL); } } /* Validate permissions */ if (err == 0) err = zfs_check_settable(dsname, pair, CRED()); if (err == 0) { err = zfs_prop_set_special(dsname, source, pair); if (err == -1) { /* * For better performance we build up a list of * properties to set in a single transaction. */ err = nvlist_add_nvpair(genericnvl, pair); } else if (err != 0 && nvl != retrynvl) { /* * This may be a spurious error caused by * receiving quota and reservation out of order. * Try again in a second pass. */ err = nvlist_add_nvpair(retrynvl, pair); } } if (err != 0) { if (errlist != NULL) fnvlist_add_int32(errlist, propname, err); rv = err; } } if (nvl != retrynvl && !nvlist_empty(retrynvl)) { nvl = retrynvl; goto retry; } if (!nvlist_empty(genericnvl) && dsl_props_set(dsname, source, genericnvl) != 0) { /* * If this fails, we still want to set as many properties as we * can, so try setting them individually. */ pair = NULL; while ((pair = nvlist_next_nvpair(genericnvl, pair)) != NULL) { const char *propname = nvpair_name(pair); int err = 0; propval = pair; if (nvpair_type(pair) == DATA_TYPE_NVLIST) { nvlist_t *attrs; attrs = fnvpair_value_nvlist(pair); propval = fnvlist_lookup_nvpair(attrs, ZPROP_VALUE); } if (nvpair_type(propval) == DATA_TYPE_STRING) { strval = fnvpair_value_string(propval); err = dsl_prop_set_string(dsname, propname, source, strval); } else { intval = fnvpair_value_uint64(propval); err = dsl_prop_set_int(dsname, propname, source, intval); } if (err != 0) { if (errlist != NULL) { fnvlist_add_int32(errlist, propname, err); } rv = err; } } } nvlist_free(genericnvl); nvlist_free(retrynvl); return (rv); } /* * Check that all the properties are valid user properties. */ static int zfs_check_userprops(const char *fsname, nvlist_t *nvl) { nvpair_t *pair = NULL; int error = 0; while ((pair = nvlist_next_nvpair(nvl, pair)) != NULL) { const char *propname = nvpair_name(pair); if (!zfs_prop_user(propname) || nvpair_type(pair) != DATA_TYPE_STRING) return (SET_ERROR(EINVAL)); if (error = zfs_secpolicy_write_perms(fsname, ZFS_DELEG_PERM_USERPROP, CRED())) return (error); if (strlen(propname) >= ZAP_MAXNAMELEN) return (SET_ERROR(ENAMETOOLONG)); if (strlen(fnvpair_value_string(pair)) >= ZAP_MAXVALUELEN) return (E2BIG); } return (0); } static void props_skip(nvlist_t *props, nvlist_t *skipped, nvlist_t **newprops) { nvpair_t *pair; VERIFY(nvlist_alloc(newprops, NV_UNIQUE_NAME, KM_SLEEP) == 0); pair = NULL; while ((pair = nvlist_next_nvpair(props, pair)) != NULL) { if (nvlist_exists(skipped, nvpair_name(pair))) continue; VERIFY(nvlist_add_nvpair(*newprops, pair) == 0); } } static int clear_received_props(const char *dsname, nvlist_t *props, nvlist_t *skipped) { int err = 0; nvlist_t *cleared_props = NULL; props_skip(props, skipped, &cleared_props); if (!nvlist_empty(cleared_props)) { /* * Acts on local properties until the dataset has received * properties at least once on or after SPA_VERSION_RECVD_PROPS. */ zprop_source_t flags = (ZPROP_SRC_NONE | (dsl_prop_get_hasrecvd(dsname) ? ZPROP_SRC_RECEIVED : 0)); err = zfs_set_prop_nvlist(dsname, flags, cleared_props, NULL); } nvlist_free(cleared_props); return (err); } /* * inputs: * zc_name name of filesystem * zc_value name of property to set * zc_nvlist_src{_size} nvlist of properties to apply * zc_cookie received properties flag * * outputs: * zc_nvlist_dst{_size} error for each unapplied received property */ static int zfs_ioc_set_prop(zfs_cmd_t *zc) { nvlist_t *nvl; boolean_t received = zc->zc_cookie; zprop_source_t source = (received ? ZPROP_SRC_RECEIVED : ZPROP_SRC_LOCAL); nvlist_t *errors; int error; if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &nvl)) != 0) return (error); if (received) { nvlist_t *origprops; if (dsl_prop_get_received(zc->zc_name, &origprops) == 0) { (void) clear_received_props(zc->zc_name, origprops, nvl); nvlist_free(origprops); } error = dsl_prop_set_hasrecvd(zc->zc_name); } errors = fnvlist_alloc(); if (error == 0) error = zfs_set_prop_nvlist(zc->zc_name, source, nvl, errors); if (zc->zc_nvlist_dst != 0 && errors != NULL) { (void) put_nvlist(zc, errors); } nvlist_free(errors); nvlist_free(nvl); return (error); } /* * inputs: * zc_name name of filesystem * zc_value name of property to inherit * zc_cookie revert to received value if TRUE * * outputs: none */ static int zfs_ioc_inherit_prop(zfs_cmd_t *zc) { const char *propname = zc->zc_value; zfs_prop_t prop = zfs_name_to_prop(propname); boolean_t received = zc->zc_cookie; zprop_source_t source = (received ? ZPROP_SRC_NONE /* revert to received value, if any */ : ZPROP_SRC_INHERITED); /* explicitly inherit */ if (received) { nvlist_t *dummy; nvpair_t *pair; zprop_type_t type; int err; /* * zfs_prop_set_special() expects properties in the form of an * nvpair with type info. */ if (prop == ZPROP_INVAL) { if (!zfs_prop_user(propname)) return (SET_ERROR(EINVAL)); type = PROP_TYPE_STRING; } else if (prop == ZFS_PROP_VOLSIZE || prop == ZFS_PROP_VERSION) { return (SET_ERROR(EINVAL)); } else { type = zfs_prop_get_type(prop); } VERIFY(nvlist_alloc(&dummy, NV_UNIQUE_NAME, KM_SLEEP) == 0); switch (type) { case PROP_TYPE_STRING: VERIFY(0 == nvlist_add_string(dummy, propname, "")); break; case PROP_TYPE_NUMBER: case PROP_TYPE_INDEX: VERIFY(0 == nvlist_add_uint64(dummy, propname, 0)); break; default: nvlist_free(dummy); return (SET_ERROR(EINVAL)); } pair = nvlist_next_nvpair(dummy, NULL); err = zfs_prop_set_special(zc->zc_name, source, pair); nvlist_free(dummy); if (err != -1) return (err); /* special property already handled */ } else { /* * Only check this in the non-received case. We want to allow * 'inherit -S' to revert non-inheritable properties like quota * and reservation to the received or default values even though * they are not considered inheritable. */ if (prop != ZPROP_INVAL && !zfs_prop_inheritable(prop)) return (SET_ERROR(EINVAL)); } /* property name has been validated by zfs_secpolicy_inherit_prop() */ return (dsl_prop_inherit(zc->zc_name, zc->zc_value, source)); } static int zfs_ioc_pool_set_props(zfs_cmd_t *zc) { nvlist_t *props; spa_t *spa; int error; nvpair_t *pair; if (error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &props)) return (error); /* * If the only property is the configfile, then just do a spa_lookup() * to handle the faulted case. */ pair = nvlist_next_nvpair(props, NULL); if (pair != NULL && strcmp(nvpair_name(pair), zpool_prop_to_name(ZPOOL_PROP_CACHEFILE)) == 0 && nvlist_next_nvpair(props, pair) == NULL) { mutex_enter(&spa_namespace_lock); if ((spa = spa_lookup(zc->zc_name)) != NULL) { spa_configfile_set(spa, props, B_FALSE); spa_config_sync(spa, B_FALSE, B_TRUE); } mutex_exit(&spa_namespace_lock); if (spa != NULL) { nvlist_free(props); return (0); } } if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) { nvlist_free(props); return (error); } error = spa_prop_set(spa, props); nvlist_free(props); spa_close(spa, FTAG); return (error); } static int zfs_ioc_pool_get_props(zfs_cmd_t *zc) { spa_t *spa; int error; nvlist_t *nvp = NULL; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) { /* * If the pool is faulted, there may be properties we can still * get (such as altroot and cachefile), so attempt to get them * anyway. */ mutex_enter(&spa_namespace_lock); if ((spa = spa_lookup(zc->zc_name)) != NULL) error = spa_prop_get(spa, &nvp); mutex_exit(&spa_namespace_lock); } else { error = spa_prop_get(spa, &nvp); spa_close(spa, FTAG); } if (error == 0 && zc->zc_nvlist_dst != 0) error = put_nvlist(zc, nvp); else error = SET_ERROR(EFAULT); nvlist_free(nvp); return (error); } /* * inputs: * zc_name name of filesystem * zc_nvlist_src{_size} nvlist of delegated permissions * zc_perm_action allow/unallow flag * * outputs: none */ static int zfs_ioc_set_fsacl(zfs_cmd_t *zc) { int error; nvlist_t *fsaclnv = NULL; if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &fsaclnv)) != 0) return (error); /* * Verify nvlist is constructed correctly */ if ((error = zfs_deleg_verify_nvlist(fsaclnv)) != 0) { nvlist_free(fsaclnv); return (SET_ERROR(EINVAL)); } /* * If we don't have PRIV_SYS_MOUNT, then validate * that user is allowed to hand out each permission in * the nvlist(s) */ error = secpolicy_zfs(CRED()); if (error != 0) { if (zc->zc_perm_action == B_FALSE) { error = dsl_deleg_can_allow(zc->zc_name, fsaclnv, CRED()); } else { error = dsl_deleg_can_unallow(zc->zc_name, fsaclnv, CRED()); } } if (error == 0) error = dsl_deleg_set(zc->zc_name, fsaclnv, zc->zc_perm_action); nvlist_free(fsaclnv); return (error); } /* * inputs: * zc_name name of filesystem * * outputs: * zc_nvlist_src{_size} nvlist of delegated permissions */ static int zfs_ioc_get_fsacl(zfs_cmd_t *zc) { nvlist_t *nvp; int error; if ((error = dsl_deleg_get(zc->zc_name, &nvp)) == 0) { error = put_nvlist(zc, nvp); nvlist_free(nvp); } return (error); } /* * Search the vfs list for a specified resource. Returns a pointer to it * or NULL if no suitable entry is found. The caller of this routine * is responsible for releasing the returned vfs pointer. */ static vfs_t * zfs_get_vfs(const char *resource) { vfs_t *vfsp; mtx_lock(&mountlist_mtx); TAILQ_FOREACH(vfsp, &mountlist, mnt_list) { if (strcmp(refstr_value(vfsp->vfs_resource), resource) == 0) { if (vfs_busy(vfsp, MBF_MNTLSTLOCK) != 0) vfsp = NULL; break; } } if (vfsp == NULL) mtx_unlock(&mountlist_mtx); return (vfsp); } /* ARGSUSED */ static void zfs_create_cb(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx) { zfs_creat_t *zct = arg; zfs_create_fs(os, cr, zct->zct_zplprops, tx); } #define ZFS_PROP_UNDEFINED ((uint64_t)-1) /* * inputs: * os parent objset pointer (NULL if root fs) * fuids_ok fuids allowed in this version of the spa? * sa_ok SAs allowed in this version of the spa? * createprops list of properties requested by creator * * outputs: * zplprops values for the zplprops we attach to the master node object * is_ci true if requested file system will be purely case-insensitive * * Determine the settings for utf8only, normalization and * casesensitivity. Specific values may have been requested by the * creator and/or we can inherit values from the parent dataset. If * the file system is of too early a vintage, a creator can not * request settings for these properties, even if the requested * setting is the default value. We don't actually want to create dsl * properties for these, so remove them from the source nvlist after * processing. */ static int zfs_fill_zplprops_impl(objset_t *os, uint64_t zplver, boolean_t fuids_ok, boolean_t sa_ok, nvlist_t *createprops, nvlist_t *zplprops, boolean_t *is_ci) { uint64_t sense = ZFS_PROP_UNDEFINED; uint64_t norm = ZFS_PROP_UNDEFINED; uint64_t u8 = ZFS_PROP_UNDEFINED; ASSERT(zplprops != NULL); /* * Pull out creator prop choices, if any. */ if (createprops) { (void) nvlist_lookup_uint64(createprops, zfs_prop_to_name(ZFS_PROP_VERSION), &zplver); (void) nvlist_lookup_uint64(createprops, zfs_prop_to_name(ZFS_PROP_NORMALIZE), &norm); (void) nvlist_remove_all(createprops, zfs_prop_to_name(ZFS_PROP_NORMALIZE)); (void) nvlist_lookup_uint64(createprops, zfs_prop_to_name(ZFS_PROP_UTF8ONLY), &u8); (void) nvlist_remove_all(createprops, zfs_prop_to_name(ZFS_PROP_UTF8ONLY)); (void) nvlist_lookup_uint64(createprops, zfs_prop_to_name(ZFS_PROP_CASE), &sense); (void) nvlist_remove_all(createprops, zfs_prop_to_name(ZFS_PROP_CASE)); } /* * If the zpl version requested is whacky or the file system * or pool is version is too "young" to support normalization * and the creator tried to set a value for one of the props, * error out. */ if ((zplver < ZPL_VERSION_INITIAL || zplver > ZPL_VERSION) || (zplver >= ZPL_VERSION_FUID && !fuids_ok) || (zplver >= ZPL_VERSION_SA && !sa_ok) || (zplver < ZPL_VERSION_NORMALIZATION && (norm != ZFS_PROP_UNDEFINED || u8 != ZFS_PROP_UNDEFINED || sense != ZFS_PROP_UNDEFINED))) return (SET_ERROR(ENOTSUP)); /* * Put the version in the zplprops */ VERIFY(nvlist_add_uint64(zplprops, zfs_prop_to_name(ZFS_PROP_VERSION), zplver) == 0); if (norm == ZFS_PROP_UNDEFINED) VERIFY(zfs_get_zplprop(os, ZFS_PROP_NORMALIZE, &norm) == 0); VERIFY(nvlist_add_uint64(zplprops, zfs_prop_to_name(ZFS_PROP_NORMALIZE), norm) == 0); /* * If we're normalizing, names must always be valid UTF-8 strings. */ if (norm) u8 = 1; if (u8 == ZFS_PROP_UNDEFINED) VERIFY(zfs_get_zplprop(os, ZFS_PROP_UTF8ONLY, &u8) == 0); VERIFY(nvlist_add_uint64(zplprops, zfs_prop_to_name(ZFS_PROP_UTF8ONLY), u8) == 0); if (sense == ZFS_PROP_UNDEFINED) VERIFY(zfs_get_zplprop(os, ZFS_PROP_CASE, &sense) == 0); VERIFY(nvlist_add_uint64(zplprops, zfs_prop_to_name(ZFS_PROP_CASE), sense) == 0); if (is_ci) *is_ci = (sense == ZFS_CASE_INSENSITIVE); return (0); } static int zfs_fill_zplprops(const char *dataset, nvlist_t *createprops, nvlist_t *zplprops, boolean_t *is_ci) { boolean_t fuids_ok, sa_ok; uint64_t zplver = ZPL_VERSION; objset_t *os = NULL; char parentname[MAXNAMELEN]; char *cp; spa_t *spa; uint64_t spa_vers; int error; (void) strlcpy(parentname, dataset, sizeof (parentname)); cp = strrchr(parentname, '/'); ASSERT(cp != NULL); cp[0] = '\0'; if ((error = spa_open(dataset, &spa, FTAG)) != 0) return (error); spa_vers = spa_version(spa); spa_close(spa, FTAG); zplver = zfs_zpl_version_map(spa_vers); fuids_ok = (zplver >= ZPL_VERSION_FUID); sa_ok = (zplver >= ZPL_VERSION_SA); /* * Open parent object set so we can inherit zplprop values. */ if ((error = dmu_objset_hold(parentname, FTAG, &os)) != 0) return (error); error = zfs_fill_zplprops_impl(os, zplver, fuids_ok, sa_ok, createprops, zplprops, is_ci); dmu_objset_rele(os, FTAG); return (error); } static int zfs_fill_zplprops_root(uint64_t spa_vers, nvlist_t *createprops, nvlist_t *zplprops, boolean_t *is_ci) { boolean_t fuids_ok; boolean_t sa_ok; uint64_t zplver = ZPL_VERSION; int error; zplver = zfs_zpl_version_map(spa_vers); fuids_ok = (zplver >= ZPL_VERSION_FUID); sa_ok = (zplver >= ZPL_VERSION_SA); error = zfs_fill_zplprops_impl(NULL, zplver, fuids_ok, sa_ok, createprops, zplprops, is_ci); return (error); } /* * innvl: { * "type" -> dmu_objset_type_t (int32) * (optional) "props" -> { prop -> value } * } * * outnvl: propname -> error code (int32) */ static int zfs_ioc_create(const char *fsname, nvlist_t *innvl, nvlist_t *outnvl) { int error = 0; zfs_creat_t zct = { 0 }; nvlist_t *nvprops = NULL; void (*cbfunc)(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx); int32_t type32; dmu_objset_type_t type; boolean_t is_insensitive = B_FALSE; if (nvlist_lookup_int32(innvl, "type", &type32) != 0) return (SET_ERROR(EINVAL)); type = type32; (void) nvlist_lookup_nvlist(innvl, "props", &nvprops); switch (type) { case DMU_OST_ZFS: cbfunc = zfs_create_cb; break; case DMU_OST_ZVOL: cbfunc = zvol_create_cb; break; default: cbfunc = NULL; break; } if (strchr(fsname, '@') || strchr(fsname, '%')) return (SET_ERROR(EINVAL)); zct.zct_props = nvprops; if (cbfunc == NULL) return (SET_ERROR(EINVAL)); if (type == DMU_OST_ZVOL) { uint64_t volsize, volblocksize; if (nvprops == NULL) return (SET_ERROR(EINVAL)); if (nvlist_lookup_uint64(nvprops, zfs_prop_to_name(ZFS_PROP_VOLSIZE), &volsize) != 0) return (SET_ERROR(EINVAL)); if ((error = nvlist_lookup_uint64(nvprops, zfs_prop_to_name(ZFS_PROP_VOLBLOCKSIZE), &volblocksize)) != 0 && error != ENOENT) return (SET_ERROR(EINVAL)); if (error != 0) volblocksize = zfs_prop_default_numeric( ZFS_PROP_VOLBLOCKSIZE); if ((error = zvol_check_volblocksize( volblocksize)) != 0 || (error = zvol_check_volsize(volsize, volblocksize)) != 0) return (error); } else if (type == DMU_OST_ZFS) { int error; /* * We have to have normalization and * case-folding flags correct when we do the * file system creation, so go figure them out * now. */ VERIFY(nvlist_alloc(&zct.zct_zplprops, NV_UNIQUE_NAME, KM_SLEEP) == 0); error = zfs_fill_zplprops(fsname, nvprops, zct.zct_zplprops, &is_insensitive); if (error != 0) { nvlist_free(zct.zct_zplprops); return (error); } } error = dmu_objset_create(fsname, type, is_insensitive ? DS_FLAG_CI_DATASET : 0, cbfunc, &zct); nvlist_free(zct.zct_zplprops); /* * It would be nice to do this atomically. */ if (error == 0) { error = zfs_set_prop_nvlist(fsname, ZPROP_SRC_LOCAL, nvprops, outnvl); if (error != 0) (void) dsl_destroy_head(fsname); } #ifdef __FreeBSD__ if (error == 0 && type == DMU_OST_ZVOL) zvol_create_minors(fsname); #endif return (error); } /* * innvl: { * "origin" -> name of origin snapshot * (optional) "props" -> { prop -> value } * } * * outnvl: propname -> error code (int32) */ static int zfs_ioc_clone(const char *fsname, nvlist_t *innvl, nvlist_t *outnvl) { int error = 0; nvlist_t *nvprops = NULL; char *origin_name; if (nvlist_lookup_string(innvl, "origin", &origin_name) != 0) return (SET_ERROR(EINVAL)); (void) nvlist_lookup_nvlist(innvl, "props", &nvprops); if (strchr(fsname, '@') || strchr(fsname, '%')) return (SET_ERROR(EINVAL)); if (dataset_namecheck(origin_name, NULL, NULL) != 0) return (SET_ERROR(EINVAL)); error = dmu_objset_clone(fsname, origin_name); if (error != 0) return (error); /* * It would be nice to do this atomically. */ if (error == 0) { error = zfs_set_prop_nvlist(fsname, ZPROP_SRC_LOCAL, nvprops, outnvl); if (error != 0) (void) dsl_destroy_head(fsname); } #ifdef __FreeBSD__ if (error == 0) zvol_create_minors(fsname); #endif return (error); } /* * innvl: { * "snaps" -> { snapshot1, snapshot2 } * (optional) "props" -> { prop -> value (string) } * } * * outnvl: snapshot -> error code (int32) */ static int zfs_ioc_snapshot(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { nvlist_t *snaps; nvlist_t *props = NULL; int error, poollen; nvpair_t *pair; (void) nvlist_lookup_nvlist(innvl, "props", &props); if ((error = zfs_check_userprops(poolname, props)) != 0) return (error); if (!nvlist_empty(props) && zfs_earlier_version(poolname, SPA_VERSION_SNAP_PROPS)) return (SET_ERROR(ENOTSUP)); if (nvlist_lookup_nvlist(innvl, "snaps", &snaps) != 0) return (SET_ERROR(EINVAL)); poollen = strlen(poolname); for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; pair = nvlist_next_nvpair(snaps, pair)) { const char *name = nvpair_name(pair); const char *cp = strchr(name, '@'); /* * The snap name must contain an @, and the part after it must * contain only valid characters. */ if (cp == NULL || zfs_component_namecheck(cp + 1, NULL, NULL) != 0) return (SET_ERROR(EINVAL)); /* * The snap must be in the specified pool. */ if (strncmp(name, poolname, poollen) != 0 || (name[poollen] != '/' && name[poollen] != '@')) return (SET_ERROR(EXDEV)); /* This must be the only snap of this fs. */ for (nvpair_t *pair2 = nvlist_next_nvpair(snaps, pair); pair2 != NULL; pair2 = nvlist_next_nvpair(snaps, pair2)) { if (strncmp(name, nvpair_name(pair2), cp - name + 1) == 0) { return (SET_ERROR(EXDEV)); } } } error = dsl_dataset_snapshot(snaps, props, outnvl); return (error); } /* * innvl: "message" -> string */ /* ARGSUSED */ static int zfs_ioc_log_history(const char *unused, nvlist_t *innvl, nvlist_t *outnvl) { char *message; spa_t *spa; int error; char *poolname; /* * The poolname in the ioctl is not set, we get it from the TSD, * which was set at the end of the last successful ioctl that allows * logging. The secpolicy func already checked that it is set. * Only one log ioctl is allowed after each successful ioctl, so * we clear the TSD here. */ poolname = tsd_get(zfs_allow_log_key); (void) tsd_set(zfs_allow_log_key, NULL); error = spa_open(poolname, &spa, FTAG); strfree(poolname); if (error != 0) return (error); if (nvlist_lookup_string(innvl, "message", &message) != 0) { spa_close(spa, FTAG); return (SET_ERROR(EINVAL)); } if (spa_version(spa) < SPA_VERSION_ZPOOL_HISTORY) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } error = spa_history_log(spa, message); spa_close(spa, FTAG); return (error); } /* * The dp_config_rwlock must not be held when calling this, because the * unmount may need to write out data. * * This function is best-effort. Callers must deal gracefully if it * remains mounted (or is remounted after this call). * * Returns 0 if the argument is not a snapshot, or it is not currently a * filesystem, or we were able to unmount it. Returns error code otherwise. */ int zfs_unmount_snap(const char *snapname) { vfs_t *vfsp; zfsvfs_t *zfsvfs; int err; if (strchr(snapname, '@') == NULL) return (0); vfsp = zfs_get_vfs(snapname); if (vfsp == NULL) return (0); zfsvfs = vfsp->vfs_data; ASSERT(!dsl_pool_config_held(dmu_objset_pool(zfsvfs->z_os))); err = vn_vfswlock(vfsp->vfs_vnodecovered); #ifdef illumos VFS_RELE(vfsp); #else vfs_unbusy(vfsp); #endif if (err != 0) return (SET_ERROR(err)); /* * Always force the unmount for snapshots. */ #ifdef illumos (void) dounmount(vfsp, MS_FORCE, kcred); #else vfs_ref(vfsp); (void) dounmount(vfsp, MS_FORCE, curthread); #endif return (0); } /* ARGSUSED */ static int zfs_unmount_snap_cb(const char *snapname, void *arg) { return (zfs_unmount_snap(snapname)); } /* * When a clone is destroyed, its origin may also need to be destroyed, * in which case it must be unmounted. This routine will do that unmount * if necessary. */ void zfs_destroy_unmount_origin(const char *fsname) { int error; objset_t *os; dsl_dataset_t *ds; error = dmu_objset_hold(fsname, FTAG, &os); if (error != 0) return; ds = dmu_objset_ds(os); if (dsl_dir_is_clone(ds->ds_dir) && DS_IS_DEFER_DESTROY(ds->ds_prev)) { char originname[MAXNAMELEN]; dsl_dataset_name(ds->ds_prev, originname); dmu_objset_rele(os, FTAG); (void) zfs_unmount_snap(originname); } else { dmu_objset_rele(os, FTAG); } } /* * innvl: { * "snaps" -> { snapshot1, snapshot2 } * (optional boolean) "defer" * } * * outnvl: snapshot -> error code (int32) * */ /* ARGSUSED */ static int zfs_ioc_destroy_snaps(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { int error, poollen; nvlist_t *snaps; nvpair_t *pair; boolean_t defer; if (nvlist_lookup_nvlist(innvl, "snaps", &snaps) != 0) return (SET_ERROR(EINVAL)); defer = nvlist_exists(innvl, "defer"); poollen = strlen(poolname); for (pair = nvlist_next_nvpair(snaps, NULL); pair != NULL; pair = nvlist_next_nvpair(snaps, pair)) { const char *name = nvpair_name(pair); /* * The snap must be in the specified pool to prevent the * invalid removal of zvol minors below. */ if (strncmp(name, poolname, poollen) != 0 || (name[poollen] != '/' && name[poollen] != '@')) return (SET_ERROR(EXDEV)); error = zfs_unmount_snap(name); if (error != 0) return (error); #if defined(__FreeBSD__) zvol_remove_minors(name); #endif } return (dsl_destroy_snapshots_nvl(snaps, defer, outnvl)); } /* * Create bookmarks. Bookmark names are of the form #. * All bookmarks must be in the same pool. * * innvl: { * bookmark1 -> snapshot1, bookmark2 -> snapshot2 * } * * outnvl: bookmark -> error code (int32) * */ /* ARGSUSED */ static int zfs_ioc_bookmark(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { for (nvpair_t *pair = nvlist_next_nvpair(innvl, NULL); pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) { char *snap_name; /* * Verify the snapshot argument. */ if (nvpair_value_string(pair, &snap_name) != 0) return (SET_ERROR(EINVAL)); /* Verify that the keys (bookmarks) are unique */ for (nvpair_t *pair2 = nvlist_next_nvpair(innvl, pair); pair2 != NULL; pair2 = nvlist_next_nvpair(innvl, pair2)) { if (strcmp(nvpair_name(pair), nvpair_name(pair2)) == 0) return (SET_ERROR(EINVAL)); } } return (dsl_bookmark_create(innvl, outnvl)); } /* * innvl: { * property 1, property 2, ... * } * * outnvl: { * bookmark name 1 -> { property 1, property 2, ... }, * bookmark name 2 -> { property 1, property 2, ... } * } * */ static int zfs_ioc_get_bookmarks(const char *fsname, nvlist_t *innvl, nvlist_t *outnvl) { return (dsl_get_bookmarks(fsname, innvl, outnvl)); } /* * innvl: { * bookmark name 1, bookmark name 2 * } * * outnvl: bookmark -> error code (int32) * */ static int zfs_ioc_destroy_bookmarks(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) { int error, poollen; poollen = strlen(poolname); for (nvpair_t *pair = nvlist_next_nvpair(innvl, NULL); pair != NULL; pair = nvlist_next_nvpair(innvl, pair)) { const char *name = nvpair_name(pair); const char *cp = strchr(name, '#'); /* * The bookmark name must contain an #, and the part after it * must contain only valid characters. */ if (cp == NULL || zfs_component_namecheck(cp + 1, NULL, NULL) != 0) return (SET_ERROR(EINVAL)); /* * The bookmark must be in the specified pool. */ if (strncmp(name, poolname, poollen) != 0 || (name[poollen] != '/' && name[poollen] != '#')) return (SET_ERROR(EXDEV)); } error = dsl_bookmark_destroy(innvl, outnvl); return (error); } /* * inputs: * zc_name name of dataset to destroy * zc_objset_type type of objset * zc_defer_destroy mark for deferred destroy * * outputs: none */ static int zfs_ioc_destroy(zfs_cmd_t *zc) { int err; if (zc->zc_objset_type == DMU_OST_ZFS) { err = zfs_unmount_snap(zc->zc_name); if (err != 0) return (err); } if (strchr(zc->zc_name, '@')) err = dsl_destroy_snapshot(zc->zc_name, zc->zc_defer_destroy); else err = dsl_destroy_head(zc->zc_name); if (zc->zc_objset_type == DMU_OST_ZVOL && err == 0) #ifdef __FreeBSD__ zvol_remove_minors(zc->zc_name); #else (void) zvol_remove_minor(zc->zc_name); #endif return (err); } /* * fsname is name of dataset to rollback (to most recent snapshot) * * innvl is not used. * * outnvl: "target" -> name of most recent snapshot * } */ /* ARGSUSED */ static int zfs_ioc_rollback(const char *fsname, nvlist_t *args, nvlist_t *outnvl) { zfsvfs_t *zfsvfs; int error; if (getzfsvfs(fsname, &zfsvfs) == 0) { error = zfs_suspend_fs(zfsvfs); if (error == 0) { int resume_err; error = dsl_dataset_rollback(fsname, zfsvfs, outnvl); resume_err = zfs_resume_fs(zfsvfs, fsname); error = error ? error : resume_err; } #ifdef illumos VFS_RELE(zfsvfs->z_vfs); #else vfs_unbusy(zfsvfs->z_vfs); #endif } else { error = dsl_dataset_rollback(fsname, NULL, outnvl); } return (error); } static int recursive_unmount(const char *fsname, void *arg) { const char *snapname = arg; char fullname[MAXNAMELEN]; (void) snprintf(fullname, sizeof (fullname), "%s@%s", fsname, snapname); return (zfs_unmount_snap(fullname)); } /* * inputs: * zc_name old name of dataset * zc_value new name of dataset * zc_cookie recursive flag (only valid for snapshots) * * outputs: none */ static int zfs_ioc_rename(zfs_cmd_t *zc) { boolean_t recursive = zc->zc_cookie & 1; char *at; boolean_t allow_mounted = B_TRUE; #ifdef __FreeBSD__ allow_mounted = (zc->zc_cookie & 2) != 0; #endif zc->zc_value[sizeof (zc->zc_value) - 1] = '\0'; if (dataset_namecheck(zc->zc_value, NULL, NULL) != 0 || strchr(zc->zc_value, '%')) return (SET_ERROR(EINVAL)); at = strchr(zc->zc_name, '@'); if (at != NULL) { /* snaps must be in same fs */ int error; if (strncmp(zc->zc_name, zc->zc_value, at - zc->zc_name + 1)) return (SET_ERROR(EXDEV)); *at = '\0'; if (zc->zc_objset_type == DMU_OST_ZFS && !allow_mounted) { error = dmu_objset_find(zc->zc_name, recursive_unmount, at + 1, recursive ? DS_FIND_CHILDREN : 0); if (error != 0) { *at = '@'; return (error); } } error = dsl_dataset_rename_snapshot(zc->zc_name, at + 1, strchr(zc->zc_value, '@') + 1, recursive); *at = '@'; return (error); } else { #ifdef illumos if (zc->zc_objset_type == DMU_OST_ZVOL) (void) zvol_remove_minor(zc->zc_name); #endif return (dsl_dir_rename(zc->zc_name, zc->zc_value)); } } static int zfs_check_settable(const char *dsname, nvpair_t *pair, cred_t *cr) { const char *propname = nvpair_name(pair); boolean_t issnap = (strchr(dsname, '@') != NULL); zfs_prop_t prop = zfs_name_to_prop(propname); uint64_t intval; int err; if (prop == ZPROP_INVAL) { if (zfs_prop_user(propname)) { if (err = zfs_secpolicy_write_perms(dsname, ZFS_DELEG_PERM_USERPROP, cr)) return (err); return (0); } if (!issnap && zfs_prop_userquota(propname)) { const char *perm = NULL; const char *uq_prefix = zfs_userquota_prop_prefixes[ZFS_PROP_USERQUOTA]; const char *gq_prefix = zfs_userquota_prop_prefixes[ZFS_PROP_GROUPQUOTA]; if (strncmp(propname, uq_prefix, strlen(uq_prefix)) == 0) { perm = ZFS_DELEG_PERM_USERQUOTA; } else if (strncmp(propname, gq_prefix, strlen(gq_prefix)) == 0) { perm = ZFS_DELEG_PERM_GROUPQUOTA; } else { /* USERUSED and GROUPUSED are read-only */ return (SET_ERROR(EINVAL)); } if (err = zfs_secpolicy_write_perms(dsname, perm, cr)) return (err); return (0); } return (SET_ERROR(EINVAL)); } if (issnap) return (SET_ERROR(EINVAL)); if (nvpair_type(pair) == DATA_TYPE_NVLIST) { /* * dsl_prop_get_all_impl() returns properties in this * format. */ nvlist_t *attrs; VERIFY(nvpair_value_nvlist(pair, &attrs) == 0); VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &pair) == 0); } /* * Check that this value is valid for this pool version */ switch (prop) { case ZFS_PROP_COMPRESSION: /* * If the user specified gzip compression, make sure * the SPA supports it. We ignore any errors here since * we'll catch them later. */ if (nvpair_value_uint64(pair, &intval) == 0) { if (intval >= ZIO_COMPRESS_GZIP_1 && intval <= ZIO_COMPRESS_GZIP_9 && zfs_earlier_version(dsname, SPA_VERSION_GZIP_COMPRESSION)) { return (SET_ERROR(ENOTSUP)); } if (intval == ZIO_COMPRESS_ZLE && zfs_earlier_version(dsname, SPA_VERSION_ZLE_COMPRESSION)) return (SET_ERROR(ENOTSUP)); if (intval == ZIO_COMPRESS_LZ4) { spa_t *spa; if ((err = spa_open(dsname, &spa, FTAG)) != 0) return (err); if (!spa_feature_is_enabled(spa, SPA_FEATURE_LZ4_COMPRESS)) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } spa_close(spa, FTAG); } /* * If this is a bootable dataset then * verify that the compression algorithm * is supported for booting. We must return * something other than ENOTSUP since it * implies a downrev pool version. */ if (zfs_is_bootfs(dsname) && !BOOTFS_COMPRESS_VALID(intval)) { return (SET_ERROR(ERANGE)); } } break; case ZFS_PROP_COPIES: if (zfs_earlier_version(dsname, SPA_VERSION_DITTO_BLOCKS)) return (SET_ERROR(ENOTSUP)); break; case ZFS_PROP_RECORDSIZE: /* Record sizes above 128k need the feature to be enabled */ if (nvpair_value_uint64(pair, &intval) == 0 && intval > SPA_OLD_MAXBLOCKSIZE) { spa_t *spa; /* * If this is a bootable dataset then * the we don't allow large (>128K) blocks, * because GRUB doesn't support them. */ if (zfs_is_bootfs(dsname) && intval > SPA_OLD_MAXBLOCKSIZE) { return (SET_ERROR(ERANGE)); } /* * We don't allow setting the property above 1MB, * unless the tunable has been changed. */ if (intval > zfs_max_recordsize || intval > SPA_MAXBLOCKSIZE) return (SET_ERROR(ERANGE)); if ((err = spa_open(dsname, &spa, FTAG)) != 0) return (err); if (!spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_BLOCKS)) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } spa_close(spa, FTAG); } break; case ZFS_PROP_SHARESMB: if (zpl_earlier_version(dsname, ZPL_VERSION_FUID)) return (SET_ERROR(ENOTSUP)); break; case ZFS_PROP_ACLINHERIT: if (nvpair_type(pair) == DATA_TYPE_UINT64 && nvpair_value_uint64(pair, &intval) == 0) { if (intval == ZFS_ACL_PASSTHROUGH_X && zfs_earlier_version(dsname, SPA_VERSION_PASSTHROUGH_X)) return (SET_ERROR(ENOTSUP)); } break; case ZFS_PROP_CHECKSUM: case ZFS_PROP_DEDUP: { spa_feature_t feature; spa_t *spa; /* dedup feature version checks */ if (prop == ZFS_PROP_DEDUP && zfs_earlier_version(dsname, SPA_VERSION_DEDUP)) return (SET_ERROR(ENOTSUP)); if (nvpair_value_uint64(pair, &intval) != 0) return (SET_ERROR(EINVAL)); /* check prop value is enabled in features */ feature = zio_checksum_to_feature(intval & ZIO_CHECKSUM_MASK); if (feature == SPA_FEATURE_NONE) break; if ((err = spa_open(dsname, &spa, FTAG)) != 0) return (err); /* * Salted checksums are not supported on root pools. */ if (spa_bootfs(spa) != 0 && intval < ZIO_CHECKSUM_FUNCTIONS && (zio_checksum_table[intval].ci_flags & ZCHECKSUM_FLAG_SALTED)) { spa_close(spa, FTAG); return (SET_ERROR(ERANGE)); } if (!spa_feature_is_enabled(spa, feature)) { spa_close(spa, FTAG); return (SET_ERROR(ENOTSUP)); } spa_close(spa, FTAG); break; } } return (zfs_secpolicy_setprop(dsname, prop, pair, CRED())); } /* * Checks for a race condition to make sure we don't increment a feature flag * multiple times. */ static int zfs_prop_activate_feature_check(void *arg, dmu_tx_t *tx) { spa_t *spa = dmu_tx_pool(tx)->dp_spa; spa_feature_t *featurep = arg; if (!spa_feature_is_active(spa, *featurep)) return (0); else return (SET_ERROR(EBUSY)); } /* * The callback invoked on feature activation in the sync task caused by * zfs_prop_activate_feature. */ static void zfs_prop_activate_feature_sync(void *arg, dmu_tx_t *tx) { spa_t *spa = dmu_tx_pool(tx)->dp_spa; spa_feature_t *featurep = arg; spa_feature_incr(spa, *featurep, tx); } /* * Activates a feature on a pool in response to a property setting. This * creates a new sync task which modifies the pool to reflect the feature * as being active. */ static int zfs_prop_activate_feature(spa_t *spa, spa_feature_t feature) { int err; /* EBUSY here indicates that the feature is already active */ err = dsl_sync_task(spa_name(spa), zfs_prop_activate_feature_check, zfs_prop_activate_feature_sync, &feature, 2, ZFS_SPACE_CHECK_RESERVED); if (err != 0 && err != EBUSY) return (err); else return (0); } /* * Removes properties from the given props list that fail permission checks * needed to clear them and to restore them in case of a receive error. For each * property, make sure we have both set and inherit permissions. * * Returns the first error encountered if any permission checks fail. If the * caller provides a non-NULL errlist, it also gives the complete list of names * of all the properties that failed a permission check along with the * corresponding error numbers. The caller is responsible for freeing the * returned errlist. * * If every property checks out successfully, zero is returned and the list * pointed at by errlist is NULL. */ static int zfs_check_clearable(char *dataset, nvlist_t *props, nvlist_t **errlist) { zfs_cmd_t *zc; nvpair_t *pair, *next_pair; nvlist_t *errors; int err, rv = 0; if (props == NULL) return (0); VERIFY(nvlist_alloc(&errors, NV_UNIQUE_NAME, KM_SLEEP) == 0); zc = kmem_alloc(sizeof (zfs_cmd_t), KM_SLEEP); (void) strcpy(zc->zc_name, dataset); pair = nvlist_next_nvpair(props, NULL); while (pair != NULL) { next_pair = nvlist_next_nvpair(props, pair); (void) strcpy(zc->zc_value, nvpair_name(pair)); if ((err = zfs_check_settable(dataset, pair, CRED())) != 0 || (err = zfs_secpolicy_inherit_prop(zc, NULL, CRED())) != 0) { VERIFY(nvlist_remove_nvpair(props, pair) == 0); VERIFY(nvlist_add_int32(errors, zc->zc_value, err) == 0); } pair = next_pair; } kmem_free(zc, sizeof (zfs_cmd_t)); if ((pair = nvlist_next_nvpair(errors, NULL)) == NULL) { nvlist_free(errors); errors = NULL; } else { VERIFY(nvpair_value_int32(pair, &rv) == 0); } if (errlist == NULL) nvlist_free(errors); else *errlist = errors; return (rv); } static boolean_t propval_equals(nvpair_t *p1, nvpair_t *p2) { if (nvpair_type(p1) == DATA_TYPE_NVLIST) { /* dsl_prop_get_all_impl() format */ nvlist_t *attrs; VERIFY(nvpair_value_nvlist(p1, &attrs) == 0); VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &p1) == 0); } if (nvpair_type(p2) == DATA_TYPE_NVLIST) { nvlist_t *attrs; VERIFY(nvpair_value_nvlist(p2, &attrs) == 0); VERIFY(nvlist_lookup_nvpair(attrs, ZPROP_VALUE, &p2) == 0); } if (nvpair_type(p1) != nvpair_type(p2)) return (B_FALSE); if (nvpair_type(p1) == DATA_TYPE_STRING) { char *valstr1, *valstr2; VERIFY(nvpair_value_string(p1, (char **)&valstr1) == 0); VERIFY(nvpair_value_string(p2, (char **)&valstr2) == 0); return (strcmp(valstr1, valstr2) == 0); } else { uint64_t intval1, intval2; VERIFY(nvpair_value_uint64(p1, &intval1) == 0); VERIFY(nvpair_value_uint64(p2, &intval2) == 0); return (intval1 == intval2); } } /* * Remove properties from props if they are not going to change (as determined * by comparison with origprops). Remove them from origprops as well, since we * do not need to clear or restore properties that won't change. */ static void props_reduce(nvlist_t *props, nvlist_t *origprops) { nvpair_t *pair, *next_pair; if (origprops == NULL) return; /* all props need to be received */ pair = nvlist_next_nvpair(props, NULL); while (pair != NULL) { const char *propname = nvpair_name(pair); nvpair_t *match; next_pair = nvlist_next_nvpair(props, pair); if ((nvlist_lookup_nvpair(origprops, propname, &match) != 0) || !propval_equals(pair, match)) goto next; /* need to set received value */ /* don't clear the existing received value */ (void) nvlist_remove_nvpair(origprops, match); /* don't bother receiving the property */ (void) nvlist_remove_nvpair(props, pair); next: pair = next_pair; } } /* * Extract properties that cannot be set PRIOR to the receipt of a dataset. * For example, refquota cannot be set until after the receipt of a dataset, * because in replication streams, an older/earlier snapshot may exceed the * refquota. We want to receive the older/earlier snapshot, but setting * refquota pre-receipt will set the dsl's ACTUAL quota, which will prevent * the older/earlier snapshot from being received (with EDQUOT). * * The ZFS test "zfs_receive_011_pos" demonstrates such a scenario. * * libzfs will need to be judicious handling errors encountered by props * extracted by this function. */ static nvlist_t * extract_delay_props(nvlist_t *props) { nvlist_t *delayprops; nvpair_t *nvp, *tmp; static const zfs_prop_t delayable[] = { ZFS_PROP_REFQUOTA, 0 }; int i; VERIFY(nvlist_alloc(&delayprops, NV_UNIQUE_NAME, KM_SLEEP) == 0); for (nvp = nvlist_next_nvpair(props, NULL); nvp != NULL; nvp = nvlist_next_nvpair(props, nvp)) { /* * strcmp() is safe because zfs_prop_to_name() always returns * a bounded string. */ for (i = 0; delayable[i] != 0; i++) { if (strcmp(zfs_prop_to_name(delayable[i]), nvpair_name(nvp)) == 0) { break; } } if (delayable[i] != 0) { tmp = nvlist_prev_nvpair(props, nvp); VERIFY(nvlist_add_nvpair(delayprops, nvp) == 0); VERIFY(nvlist_remove_nvpair(props, nvp) == 0); nvp = tmp; } } if (nvlist_empty(delayprops)) { nvlist_free(delayprops); delayprops = NULL; } return (delayprops); } #ifdef DEBUG static boolean_t zfs_ioc_recv_inject_err; #endif /* * inputs: * zc_name name of containing filesystem * zc_nvlist_src{_size} nvlist of properties to apply * zc_value name of snapshot to create * zc_string name of clone origin (if DRR_FLAG_CLONE) * zc_cookie file descriptor to recv from * zc_begin_record the BEGIN record of the stream (not byteswapped) * zc_guid force flag * zc_cleanup_fd cleanup-on-exit file descriptor * zc_action_handle handle for this guid/ds mapping (or zero on first call) * zc_resumable if data is incomplete assume sender will resume * * outputs: * zc_cookie number of bytes read * zc_nvlist_dst{_size} error for each unapplied received property * zc_obj zprop_errflags_t * zc_action_handle handle for this guid/ds mapping */ static int zfs_ioc_recv(zfs_cmd_t *zc) { file_t *fp; dmu_recv_cookie_t drc; boolean_t force = (boolean_t)zc->zc_guid; int fd; int error = 0; int props_error = 0; nvlist_t *errors; offset_t off; nvlist_t *props = NULL; /* sent properties */ nvlist_t *origprops = NULL; /* existing properties */ nvlist_t *delayprops = NULL; /* sent properties applied post-receive */ char *origin = NULL; char *tosnap; char tofs[ZFS_MAXNAMELEN]; cap_rights_t rights; boolean_t first_recvd_props = B_FALSE; if (dataset_namecheck(zc->zc_value, NULL, NULL) != 0 || strchr(zc->zc_value, '@') == NULL || strchr(zc->zc_value, '%')) return (SET_ERROR(EINVAL)); (void) strcpy(tofs, zc->zc_value); tosnap = strchr(tofs, '@'); *tosnap++ = '\0'; if (zc->zc_nvlist_src != 0 && (error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &props)) != 0) return (error); fd = zc->zc_cookie; #ifdef illumos fp = getf(fd); #else fget_read(curthread, fd, cap_rights_init(&rights, CAP_PREAD), &fp); #endif if (fp == NULL) { nvlist_free(props); return (SET_ERROR(EBADF)); } errors = fnvlist_alloc(); if (zc->zc_string[0]) origin = zc->zc_string; error = dmu_recv_begin(tofs, tosnap, &zc->zc_begin_record, force, zc->zc_resumable, origin, &drc); if (error != 0) goto out; /* * Set properties before we receive the stream so that they are applied * to the new data. Note that we must call dmu_recv_stream() if * dmu_recv_begin() succeeds. */ if (props != NULL && !drc.drc_newfs) { if (spa_version(dsl_dataset_get_spa(drc.drc_ds)) >= SPA_VERSION_RECVD_PROPS && !dsl_prop_get_hasrecvd(tofs)) first_recvd_props = B_TRUE; /* * If new received properties are supplied, they are to * completely replace the existing received properties, so stash * away the existing ones. */ if (dsl_prop_get_received(tofs, &origprops) == 0) { nvlist_t *errlist = NULL; /* * Don't bother writing a property if its value won't * change (and avoid the unnecessary security checks). * * The first receive after SPA_VERSION_RECVD_PROPS is a * special case where we blow away all local properties * regardless. */ if (!first_recvd_props) props_reduce(props, origprops); if (zfs_check_clearable(tofs, origprops, &errlist) != 0) (void) nvlist_merge(errors, errlist, 0); nvlist_free(errlist); if (clear_received_props(tofs, origprops, first_recvd_props ? NULL : props) != 0) zc->zc_obj |= ZPROP_ERR_NOCLEAR; } else { zc->zc_obj |= ZPROP_ERR_NOCLEAR; } } if (props != NULL) { props_error = dsl_prop_set_hasrecvd(tofs); if (props_error == 0) { delayprops = extract_delay_props(props); (void) zfs_set_prop_nvlist(tofs, ZPROP_SRC_RECEIVED, props, errors); } } off = fp->f_offset; error = dmu_recv_stream(&drc, fp, &off, zc->zc_cleanup_fd, &zc->zc_action_handle); if (error == 0) { zfsvfs_t *zfsvfs = NULL; if (getzfsvfs(tofs, &zfsvfs) == 0) { /* online recv */ int end_err; error = zfs_suspend_fs(zfsvfs); /* * If the suspend fails, then the recv_end will * likely also fail, and clean up after itself. */ end_err = dmu_recv_end(&drc, zfsvfs); if (error == 0) error = zfs_resume_fs(zfsvfs, tofs); error = error ? error : end_err; #ifdef illumos VFS_RELE(zfsvfs->z_vfs); #else vfs_unbusy(zfsvfs->z_vfs); #endif } else { error = dmu_recv_end(&drc, NULL); } /* Set delayed properties now, after we're done receiving. */ if (delayprops != NULL && error == 0) { (void) zfs_set_prop_nvlist(tofs, ZPROP_SRC_RECEIVED, delayprops, errors); } } if (delayprops != NULL) { /* * Merge delayed props back in with initial props, in case * we're DEBUG and zfs_ioc_recv_inject_err is set (which means * we have to make sure clear_received_props() includes * the delayed properties). * * Since zfs_ioc_recv_inject_err is only in DEBUG kernels, * using ASSERT() will be just like a VERIFY. */ ASSERT(nvlist_merge(props, delayprops, 0) == 0); nvlist_free(delayprops); } /* * Now that all props, initial and delayed, are set, report the prop * errors to the caller. */ if (zc->zc_nvlist_dst_size != 0 && (nvlist_smush(errors, zc->zc_nvlist_dst_size) != 0 || put_nvlist(zc, errors) != 0)) { /* * Caller made zc->zc_nvlist_dst less than the minimum expected * size or supplied an invalid address. */ props_error = SET_ERROR(EINVAL); } zc->zc_cookie = off - fp->f_offset; if (off >= 0 && off <= MAXOFFSET_T) fp->f_offset = off; #ifdef DEBUG if (zfs_ioc_recv_inject_err) { zfs_ioc_recv_inject_err = B_FALSE; error = 1; } #endif #ifdef __FreeBSD__ if (error == 0) zvol_create_minors(tofs); #endif /* * On error, restore the original props. */ if (error != 0 && props != NULL && !drc.drc_newfs) { if (clear_received_props(tofs, props, NULL) != 0) { /* * We failed to clear the received properties. * Since we may have left a $recvd value on the * system, we can't clear the $hasrecvd flag. */ zc->zc_obj |= ZPROP_ERR_NORESTORE; } else if (first_recvd_props) { dsl_prop_unset_hasrecvd(tofs); } if (origprops == NULL && !drc.drc_newfs) { /* We failed to stash the original properties. */ zc->zc_obj |= ZPROP_ERR_NORESTORE; } /* * dsl_props_set() will not convert RECEIVED to LOCAL on or * after SPA_VERSION_RECVD_PROPS, so we need to specify LOCAL * explictly if we're restoring local properties cleared in the * first new-style receive. */ if (origprops != NULL && zfs_set_prop_nvlist(tofs, (first_recvd_props ? ZPROP_SRC_LOCAL : ZPROP_SRC_RECEIVED), origprops, NULL) != 0) { /* * We stashed the original properties but failed to * restore them. */ zc->zc_obj |= ZPROP_ERR_NORESTORE; } } out: nvlist_free(props); nvlist_free(origprops); nvlist_free(errors); releasef(fd); if (error == 0) error = props_error; return (error); } /* * inputs: * zc_name name of snapshot to send * zc_cookie file descriptor to send stream to * zc_obj fromorigin flag (mutually exclusive with zc_fromobj) * zc_sendobj objsetid of snapshot to send * zc_fromobj objsetid of incremental fromsnap (may be zero) * zc_guid if set, estimate size of stream only. zc_cookie is ignored. * output size in zc_objset_type. * zc_flags lzc_send_flags * * outputs: * zc_objset_type estimated size, if zc_guid is set */ static int zfs_ioc_send(zfs_cmd_t *zc) { int error; offset_t off; boolean_t estimate = (zc->zc_guid != 0); boolean_t embedok = (zc->zc_flags & 0x1); boolean_t large_block_ok = (zc->zc_flags & 0x2); if (zc->zc_obj != 0) { dsl_pool_t *dp; dsl_dataset_t *tosnap; error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold_obj(dp, zc->zc_sendobj, FTAG, &tosnap); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } if (dsl_dir_is_clone(tosnap->ds_dir)) zc->zc_fromobj = dsl_dir_phys(tosnap->ds_dir)->dd_origin_obj; dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); } if (estimate) { dsl_pool_t *dp; dsl_dataset_t *tosnap; dsl_dataset_t *fromsnap = NULL; error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold_obj(dp, zc->zc_sendobj, FTAG, &tosnap); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } if (zc->zc_fromobj != 0) { error = dsl_dataset_hold_obj(dp, zc->zc_fromobj, FTAG, &fromsnap); if (error != 0) { dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); return (error); } } error = dmu_send_estimate(tosnap, fromsnap, &zc->zc_objset_type); if (fromsnap != NULL) dsl_dataset_rele(fromsnap, FTAG); dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); } else { file_t *fp; cap_rights_t rights; #ifdef illumos fp = getf(zc->zc_cookie); #else fget_write(curthread, zc->zc_cookie, cap_rights_init(&rights, CAP_WRITE), &fp); #endif if (fp == NULL) return (SET_ERROR(EBADF)); off = fp->f_offset; error = dmu_send_obj(zc->zc_name, zc->zc_sendobj, zc->zc_fromobj, embedok, large_block_ok, #ifdef illumos zc->zc_cookie, fp->f_vnode, &off); #else zc->zc_cookie, fp, &off); #endif if (off >= 0 && off <= MAXOFFSET_T) fp->f_offset = off; releasef(zc->zc_cookie); } return (error); } /* * inputs: * zc_name name of snapshot on which to report progress * zc_cookie file descriptor of send stream * * outputs: * zc_cookie number of bytes written in send stream thus far */ static int zfs_ioc_send_progress(zfs_cmd_t *zc) { dsl_pool_t *dp; dsl_dataset_t *ds; dmu_sendarg_t *dsp = NULL; int error; error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, zc->zc_name, FTAG, &ds); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } mutex_enter(&ds->ds_sendstream_lock); /* * Iterate over all the send streams currently active on this dataset. * If there's one which matches the specified file descriptor _and_ the * stream was started by the current process, return the progress of * that stream. */ for (dsp = list_head(&ds->ds_sendstreams); dsp != NULL; dsp = list_next(&ds->ds_sendstreams, dsp)) { if (dsp->dsa_outfd == zc->zc_cookie && dsp->dsa_proc == curproc) break; } if (dsp != NULL) zc->zc_cookie = *(dsp->dsa_off); else error = SET_ERROR(ENOENT); mutex_exit(&ds->ds_sendstream_lock); dsl_dataset_rele(ds, FTAG); dsl_pool_rele(dp, FTAG); return (error); } static int zfs_ioc_inject_fault(zfs_cmd_t *zc) { int id, error; error = zio_inject_fault(zc->zc_name, (int)zc->zc_guid, &id, &zc->zc_inject_record); if (error == 0) zc->zc_guid = (uint64_t)id; return (error); } static int zfs_ioc_clear_fault(zfs_cmd_t *zc) { return (zio_clear_fault((int)zc->zc_guid)); } static int zfs_ioc_inject_list_next(zfs_cmd_t *zc) { int id = (int)zc->zc_guid; int error; error = zio_inject_list_next(&id, zc->zc_name, sizeof (zc->zc_name), &zc->zc_inject_record); zc->zc_guid = id; return (error); } static int zfs_ioc_error_log(zfs_cmd_t *zc) { spa_t *spa; int error; size_t count = (size_t)zc->zc_nvlist_dst_size; if ((error = spa_open(zc->zc_name, &spa, FTAG)) != 0) return (error); error = spa_get_errlog(spa, (void *)(uintptr_t)zc->zc_nvlist_dst, &count); if (error == 0) zc->zc_nvlist_dst_size = count; else zc->zc_nvlist_dst_size = spa_get_errlog_size(spa); spa_close(spa, FTAG); return (error); } static int zfs_ioc_clear(zfs_cmd_t *zc) { spa_t *spa; vdev_t *vd; int error; /* * On zpool clear we also fix up missing slogs */ mutex_enter(&spa_namespace_lock); spa = spa_lookup(zc->zc_name); if (spa == NULL) { mutex_exit(&spa_namespace_lock); return (SET_ERROR(EIO)); } if (spa_get_log_state(spa) == SPA_LOG_MISSING) { /* we need to let spa_open/spa_load clear the chains */ spa_set_log_state(spa, SPA_LOG_CLEAR); } spa->spa_last_open_failed = 0; mutex_exit(&spa_namespace_lock); if (zc->zc_cookie & ZPOOL_NO_REWIND) { error = spa_open(zc->zc_name, &spa, FTAG); } else { nvlist_t *policy; nvlist_t *config = NULL; if (zc->zc_nvlist_src == 0) return (SET_ERROR(EINVAL)); if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &policy)) == 0) { error = spa_open_rewind(zc->zc_name, &spa, FTAG, policy, &config); if (config != NULL) { int err; if ((err = put_nvlist(zc, config)) != 0) error = err; nvlist_free(config); } nvlist_free(policy); } } if (error != 0) return (error); spa_vdev_state_enter(spa, SCL_NONE); if (zc->zc_guid == 0) { vd = NULL; } else { vd = spa_lookup_by_guid(spa, zc->zc_guid, B_TRUE); if (vd == NULL) { (void) spa_vdev_state_exit(spa, NULL, ENODEV); spa_close(spa, FTAG); return (SET_ERROR(ENODEV)); } } vdev_clear(spa, vd); (void) spa_vdev_state_exit(spa, NULL, 0); /* * Resume any suspended I/Os. */ if (zio_resume(spa) != 0) error = SET_ERROR(EIO); spa_close(spa, FTAG); return (error); } static int zfs_ioc_pool_reopen(zfs_cmd_t *zc) { spa_t *spa; int error; error = spa_open(zc->zc_name, &spa, FTAG); if (error != 0) return (error); spa_vdev_state_enter(spa, SCL_NONE); /* * If a resilver is already in progress then set the * spa_scrub_reopen flag to B_TRUE so that we don't restart * the scan as a side effect of the reopen. Otherwise, let * vdev_open() decided if a resilver is required. */ spa->spa_scrub_reopen = dsl_scan_resilvering(spa->spa_dsl_pool); vdev_reopen(spa->spa_root_vdev); spa->spa_scrub_reopen = B_FALSE; (void) spa_vdev_state_exit(spa, NULL, 0); spa_close(spa, FTAG); return (0); } /* * inputs: * zc_name name of filesystem * zc_value name of origin snapshot * * outputs: * zc_string name of conflicting snapshot, if there is one */ static int zfs_ioc_promote(zfs_cmd_t *zc) { char *cp; /* * We don't need to unmount *all* the origin fs's snapshots, but * it's easier. */ cp = strchr(zc->zc_value, '@'); if (cp) *cp = '\0'; (void) dmu_objset_find(zc->zc_value, zfs_unmount_snap_cb, NULL, DS_FIND_SNAPSHOTS); return (dsl_dataset_promote(zc->zc_name, zc->zc_string)); } /* * Retrieve a single {user|group}{used|quota}@... property. * * inputs: * zc_name name of filesystem * zc_objset_type zfs_userquota_prop_t * zc_value domain name (eg. "S-1-234-567-89") * zc_guid RID/UID/GID * * outputs: * zc_cookie property value */ static int zfs_ioc_userspace_one(zfs_cmd_t *zc) { zfsvfs_t *zfsvfs; int error; if (zc->zc_objset_type >= ZFS_NUM_USERQUOTA_PROPS) return (SET_ERROR(EINVAL)); error = zfsvfs_hold(zc->zc_name, FTAG, &zfsvfs, B_FALSE); if (error != 0) return (error); error = zfs_userspace_one(zfsvfs, zc->zc_objset_type, zc->zc_value, zc->zc_guid, &zc->zc_cookie); zfsvfs_rele(zfsvfs, FTAG); return (error); } /* * inputs: * zc_name name of filesystem * zc_cookie zap cursor * zc_objset_type zfs_userquota_prop_t * zc_nvlist_dst[_size] buffer to fill (not really an nvlist) * * outputs: * zc_nvlist_dst[_size] data buffer (array of zfs_useracct_t) * zc_cookie zap cursor */ static int zfs_ioc_userspace_many(zfs_cmd_t *zc) { zfsvfs_t *zfsvfs; int bufsize = zc->zc_nvlist_dst_size; if (bufsize <= 0) return (SET_ERROR(ENOMEM)); int error = zfsvfs_hold(zc->zc_name, FTAG, &zfsvfs, B_FALSE); if (error != 0) return (error); void *buf = kmem_alloc(bufsize, KM_SLEEP); error = zfs_userspace_many(zfsvfs, zc->zc_objset_type, &zc->zc_cookie, buf, &zc->zc_nvlist_dst_size); if (error == 0) { error = ddi_copyout(buf, (void *)(uintptr_t)zc->zc_nvlist_dst, zc->zc_nvlist_dst_size, zc->zc_iflags); } kmem_free(buf, bufsize); zfsvfs_rele(zfsvfs, FTAG); return (error); } /* * inputs: * zc_name name of filesystem * * outputs: * none */ static int zfs_ioc_userspace_upgrade(zfs_cmd_t *zc) { objset_t *os; int error = 0; zfsvfs_t *zfsvfs; if (getzfsvfs(zc->zc_name, &zfsvfs) == 0) { if (!dmu_objset_userused_enabled(zfsvfs->z_os)) { /* * If userused is not enabled, it may be because the * objset needs to be closed & reopened (to grow the * objset_phys_t). Suspend/resume the fs will do that. */ error = zfs_suspend_fs(zfsvfs); if (error == 0) { dmu_objset_refresh_ownership(zfsvfs->z_os, zfsvfs); error = zfs_resume_fs(zfsvfs, zc->zc_name); } } if (error == 0) error = dmu_objset_userspace_upgrade(zfsvfs->z_os); #ifdef illumos VFS_RELE(zfsvfs->z_vfs); #else vfs_unbusy(zfsvfs->z_vfs); #endif } else { /* XXX kind of reading contents without owning */ error = dmu_objset_hold(zc->zc_name, FTAG, &os); if (error != 0) return (error); error = dmu_objset_userspace_upgrade(os); dmu_objset_rele(os, FTAG); } return (error); } #ifdef illumos /* * We don't want to have a hard dependency * against some special symbols in sharefs * nfs, and smbsrv. Determine them if needed when * the first file system is shared. * Neither sharefs, nfs or smbsrv are unloadable modules. */ int (*znfsexport_fs)(void *arg); int (*zshare_fs)(enum sharefs_sys_op, share_t *, uint32_t); int (*zsmbexport_fs)(void *arg, boolean_t add_share); int zfs_nfsshare_inited; int zfs_smbshare_inited; ddi_modhandle_t nfs_mod; ddi_modhandle_t sharefs_mod; ddi_modhandle_t smbsrv_mod; #endif /* illumos */ kmutex_t zfs_share_lock; #ifdef illumos static int zfs_init_sharefs() { int error; ASSERT(MUTEX_HELD(&zfs_share_lock)); /* Both NFS and SMB shares also require sharetab support. */ if (sharefs_mod == NULL && ((sharefs_mod = ddi_modopen("fs/sharefs", KRTLD_MODE_FIRST, &error)) == NULL)) { return (SET_ERROR(ENOSYS)); } if (zshare_fs == NULL && ((zshare_fs = (int (*)(enum sharefs_sys_op, share_t *, uint32_t)) ddi_modsym(sharefs_mod, "sharefs_impl", &error)) == NULL)) { return (SET_ERROR(ENOSYS)); } return (0); } #endif /* illumos */ static int zfs_ioc_share(zfs_cmd_t *zc) { #ifdef illumos int error; int opcode; switch (zc->zc_share.z_sharetype) { case ZFS_SHARE_NFS: case ZFS_UNSHARE_NFS: if (zfs_nfsshare_inited == 0) { mutex_enter(&zfs_share_lock); if (nfs_mod == NULL && ((nfs_mod = ddi_modopen("fs/nfs", KRTLD_MODE_FIRST, &error)) == NULL)) { mutex_exit(&zfs_share_lock); return (SET_ERROR(ENOSYS)); } if (znfsexport_fs == NULL && ((znfsexport_fs = (int (*)(void *)) ddi_modsym(nfs_mod, "nfs_export", &error)) == NULL)) { mutex_exit(&zfs_share_lock); return (SET_ERROR(ENOSYS)); } error = zfs_init_sharefs(); if (error != 0) { mutex_exit(&zfs_share_lock); return (SET_ERROR(ENOSYS)); } zfs_nfsshare_inited = 1; mutex_exit(&zfs_share_lock); } break; case ZFS_SHARE_SMB: case ZFS_UNSHARE_SMB: if (zfs_smbshare_inited == 0) { mutex_enter(&zfs_share_lock); if (smbsrv_mod == NULL && ((smbsrv_mod = ddi_modopen("drv/smbsrv", KRTLD_MODE_FIRST, &error)) == NULL)) { mutex_exit(&zfs_share_lock); return (SET_ERROR(ENOSYS)); } if (zsmbexport_fs == NULL && ((zsmbexport_fs = (int (*)(void *, boolean_t))ddi_modsym(smbsrv_mod, "smb_server_share", &error)) == NULL)) { mutex_exit(&zfs_share_lock); return (SET_ERROR(ENOSYS)); } error = zfs_init_sharefs(); if (error != 0) { mutex_exit(&zfs_share_lock); return (SET_ERROR(ENOSYS)); } zfs_smbshare_inited = 1; mutex_exit(&zfs_share_lock); } break; default: return (SET_ERROR(EINVAL)); } switch (zc->zc_share.z_sharetype) { case ZFS_SHARE_NFS: case ZFS_UNSHARE_NFS: if (error = znfsexport_fs((void *) (uintptr_t)zc->zc_share.z_exportdata)) return (error); break; case ZFS_SHARE_SMB: case ZFS_UNSHARE_SMB: if (error = zsmbexport_fs((void *) (uintptr_t)zc->zc_share.z_exportdata, zc->zc_share.z_sharetype == ZFS_SHARE_SMB ? B_TRUE: B_FALSE)) { return (error); } break; } opcode = (zc->zc_share.z_sharetype == ZFS_SHARE_NFS || zc->zc_share.z_sharetype == ZFS_SHARE_SMB) ? SHAREFS_ADD : SHAREFS_REMOVE; /* * Add or remove share from sharetab */ error = zshare_fs(opcode, (void *)(uintptr_t)zc->zc_share.z_sharedata, zc->zc_share.z_sharemax); return (error); #else /* !illumos */ return (ENOSYS); #endif /* illumos */ } ace_t full_access[] = { {(uid_t)-1, ACE_ALL_PERMS, ACE_EVERYONE, 0} }; /* * inputs: * zc_name name of containing filesystem * zc_obj object # beyond which we want next in-use object # * * outputs: * zc_obj next in-use object # */ static int zfs_ioc_next_obj(zfs_cmd_t *zc) { objset_t *os = NULL; int error; error = dmu_objset_hold(zc->zc_name, FTAG, &os); if (error != 0) return (error); error = dmu_object_next(os, &zc->zc_obj, B_FALSE, dsl_dataset_phys(os->os_dsl_dataset)->ds_prev_snap_txg); dmu_objset_rele(os, FTAG); return (error); } /* * inputs: * zc_name name of filesystem * zc_value prefix name for snapshot * zc_cleanup_fd cleanup-on-exit file descriptor for calling process * * outputs: * zc_value short name of new snapshot */ static int zfs_ioc_tmp_snapshot(zfs_cmd_t *zc) { char *snap_name; char *hold_name; int error; minor_t minor; error = zfs_onexit_fd_hold(zc->zc_cleanup_fd, &minor); if (error != 0) return (error); snap_name = kmem_asprintf("%s-%016llx", zc->zc_value, (u_longlong_t)ddi_get_lbolt64()); hold_name = kmem_asprintf("%%%s", zc->zc_value); error = dsl_dataset_snapshot_tmp(zc->zc_name, snap_name, minor, hold_name); if (error == 0) (void) strcpy(zc->zc_value, snap_name); strfree(snap_name); strfree(hold_name); zfs_onexit_fd_rele(zc->zc_cleanup_fd); return (error); } /* * inputs: * zc_name name of "to" snapshot * zc_value name of "from" snapshot * zc_cookie file descriptor to write diff data on * * outputs: * dmu_diff_record_t's to the file descriptor */ static int zfs_ioc_diff(zfs_cmd_t *zc) { file_t *fp; cap_rights_t rights; offset_t off; int error; #ifdef illumos fp = getf(zc->zc_cookie); #else fget_write(curthread, zc->zc_cookie, cap_rights_init(&rights, CAP_WRITE), &fp); #endif if (fp == NULL) return (SET_ERROR(EBADF)); off = fp->f_offset; #ifdef illumos error = dmu_diff(zc->zc_name, zc->zc_value, fp->f_vnode, &off); #else error = dmu_diff(zc->zc_name, zc->zc_value, fp, &off); #endif if (off >= 0 && off <= MAXOFFSET_T) fp->f_offset = off; releasef(zc->zc_cookie); return (error); } #ifdef illumos /* * Remove all ACL files in shares dir */ static int zfs_smb_acl_purge(znode_t *dzp) { zap_cursor_t zc; zap_attribute_t zap; zfsvfs_t *zfsvfs = dzp->z_zfsvfs; int error; for (zap_cursor_init(&zc, zfsvfs->z_os, dzp->z_id); (error = zap_cursor_retrieve(&zc, &zap)) == 0; zap_cursor_advance(&zc)) { if ((error = VOP_REMOVE(ZTOV(dzp), zap.za_name, kcred, NULL, 0)) != 0) break; } zap_cursor_fini(&zc); return (error); } #endif /* illumos */ static int zfs_ioc_smb_acl(zfs_cmd_t *zc) { #ifdef illumos vnode_t *vp; znode_t *dzp; vnode_t *resourcevp = NULL; znode_t *sharedir; zfsvfs_t *zfsvfs; nvlist_t *nvlist; char *src, *target; vattr_t vattr; vsecattr_t vsec; int error = 0; if ((error = lookupname(zc->zc_value, UIO_SYSSPACE, NO_FOLLOW, NULL, &vp)) != 0) return (error); /* Now make sure mntpnt and dataset are ZFS */ if (strcmp(vp->v_vfsp->mnt_stat.f_fstypename, "zfs") != 0 || (strcmp((char *)refstr_value(vp->v_vfsp->vfs_resource), zc->zc_name) != 0)) { VN_RELE(vp); return (SET_ERROR(EINVAL)); } dzp = VTOZ(vp); zfsvfs = dzp->z_zfsvfs; ZFS_ENTER(zfsvfs); /* * Create share dir if its missing. */ mutex_enter(&zfsvfs->z_lock); if (zfsvfs->z_shares_dir == 0) { dmu_tx_t *tx; tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, TRUE, ZFS_SHARES_DIR); dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL); error = dmu_tx_assign(tx, TXG_WAIT); if (error != 0) { dmu_tx_abort(tx); } else { error = zfs_create_share_dir(zfsvfs, tx); dmu_tx_commit(tx); } if (error != 0) { mutex_exit(&zfsvfs->z_lock); VN_RELE(vp); ZFS_EXIT(zfsvfs); return (error); } } mutex_exit(&zfsvfs->z_lock); ASSERT(zfsvfs->z_shares_dir); if ((error = zfs_zget(zfsvfs, zfsvfs->z_shares_dir, &sharedir)) != 0) { VN_RELE(vp); ZFS_EXIT(zfsvfs); return (error); } switch (zc->zc_cookie) { case ZFS_SMB_ACL_ADD: vattr.va_mask = AT_MODE|AT_UID|AT_GID|AT_TYPE; vattr.va_type = VREG; vattr.va_mode = S_IFREG|0777; vattr.va_uid = 0; vattr.va_gid = 0; vsec.vsa_mask = VSA_ACE; vsec.vsa_aclentp = &full_access; vsec.vsa_aclentsz = sizeof (full_access); vsec.vsa_aclcnt = 1; error = VOP_CREATE(ZTOV(sharedir), zc->zc_string, &vattr, EXCL, 0, &resourcevp, kcred, 0, NULL, &vsec); if (resourcevp) VN_RELE(resourcevp); break; case ZFS_SMB_ACL_REMOVE: error = VOP_REMOVE(ZTOV(sharedir), zc->zc_string, kcred, NULL, 0); break; case ZFS_SMB_ACL_RENAME: if ((error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &nvlist)) != 0) { VN_RELE(vp); VN_RELE(ZTOV(sharedir)); ZFS_EXIT(zfsvfs); return (error); } if (nvlist_lookup_string(nvlist, ZFS_SMB_ACL_SRC, &src) || nvlist_lookup_string(nvlist, ZFS_SMB_ACL_TARGET, &target)) { VN_RELE(vp); VN_RELE(ZTOV(sharedir)); ZFS_EXIT(zfsvfs); nvlist_free(nvlist); return (error); } error = VOP_RENAME(ZTOV(sharedir), src, ZTOV(sharedir), target, kcred, NULL, 0); nvlist_free(nvlist); break; case ZFS_SMB_ACL_PURGE: error = zfs_smb_acl_purge(sharedir); break; default: error = SET_ERROR(EINVAL); break; } VN_RELE(vp); VN_RELE(ZTOV(sharedir)); ZFS_EXIT(zfsvfs); return (error); #else /* !illumos */ return (EOPNOTSUPP); #endif /* illumos */ } /* * innvl: { * "holds" -> { snapname -> holdname (string), ... } * (optional) "cleanup_fd" -> fd (int32) * } * * outnvl: { * snapname -> error value (int32) * ... * } */ /* ARGSUSED */ static int zfs_ioc_hold(const char *pool, nvlist_t *args, nvlist_t *errlist) { nvpair_t *pair; nvlist_t *holds; int cleanup_fd = -1; int error; minor_t minor = 0; error = nvlist_lookup_nvlist(args, "holds", &holds); if (error != 0) return (SET_ERROR(EINVAL)); /* make sure the user didn't pass us any invalid (empty) tags */ for (pair = nvlist_next_nvpair(holds, NULL); pair != NULL; pair = nvlist_next_nvpair(holds, pair)) { char *htag; error = nvpair_value_string(pair, &htag); if (error != 0) return (SET_ERROR(error)); if (strlen(htag) == 0) return (SET_ERROR(EINVAL)); } if (nvlist_lookup_int32(args, "cleanup_fd", &cleanup_fd) == 0) { error = zfs_onexit_fd_hold(cleanup_fd, &minor); if (error != 0) return (error); } error = dsl_dataset_user_hold(holds, minor, errlist); if (minor != 0) zfs_onexit_fd_rele(cleanup_fd); return (error); } /* * innvl is not used. * * outnvl: { * holdname -> time added (uint64 seconds since epoch) * ... * } */ /* ARGSUSED */ static int zfs_ioc_get_holds(const char *snapname, nvlist_t *args, nvlist_t *outnvl) { return (dsl_dataset_get_holds(snapname, outnvl)); } /* * innvl: { * snapname -> { holdname, ... } * ... * } * * outnvl: { * snapname -> error value (int32) * ... * } */ /* ARGSUSED */ static int zfs_ioc_release(const char *pool, nvlist_t *holds, nvlist_t *errlist) { return (dsl_dataset_user_release(holds, errlist)); } /* * inputs: * zc_name name of new filesystem or snapshot * zc_value full name of old snapshot * * outputs: * zc_cookie space in bytes * zc_objset_type compressed space in bytes * zc_perm_action uncompressed space in bytes */ static int zfs_ioc_space_written(zfs_cmd_t *zc) { int error; dsl_pool_t *dp; dsl_dataset_t *new, *old; error = dsl_pool_hold(zc->zc_name, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, zc->zc_name, FTAG, &new); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } error = dsl_dataset_hold(dp, zc->zc_value, FTAG, &old); if (error != 0) { dsl_dataset_rele(new, FTAG); dsl_pool_rele(dp, FTAG); return (error); } error = dsl_dataset_space_written(old, new, &zc->zc_cookie, &zc->zc_objset_type, &zc->zc_perm_action); dsl_dataset_rele(old, FTAG); dsl_dataset_rele(new, FTAG); dsl_pool_rele(dp, FTAG); return (error); } /* * innvl: { * "firstsnap" -> snapshot name * } * * outnvl: { * "used" -> space in bytes * "compressed" -> compressed space in bytes * "uncompressed" -> uncompressed space in bytes * } */ static int zfs_ioc_space_snaps(const char *lastsnap, nvlist_t *innvl, nvlist_t *outnvl) { int error; dsl_pool_t *dp; dsl_dataset_t *new, *old; char *firstsnap; uint64_t used, comp, uncomp; if (nvlist_lookup_string(innvl, "firstsnap", &firstsnap) != 0) return (SET_ERROR(EINVAL)); error = dsl_pool_hold(lastsnap, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, lastsnap, FTAG, &new); if (error == 0 && !new->ds_is_snapshot) { dsl_dataset_rele(new, FTAG); error = SET_ERROR(EINVAL); } if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } error = dsl_dataset_hold(dp, firstsnap, FTAG, &old); if (error == 0 && !old->ds_is_snapshot) { dsl_dataset_rele(old, FTAG); error = SET_ERROR(EINVAL); } if (error != 0) { dsl_dataset_rele(new, FTAG); dsl_pool_rele(dp, FTAG); return (error); } error = dsl_dataset_space_wouldfree(old, new, &used, &comp, &uncomp); dsl_dataset_rele(old, FTAG); dsl_dataset_rele(new, FTAG); dsl_pool_rele(dp, FTAG); fnvlist_add_uint64(outnvl, "used", used); fnvlist_add_uint64(outnvl, "compressed", comp); fnvlist_add_uint64(outnvl, "uncompressed", uncomp); return (error); } static int zfs_ioc_jail(zfs_cmd_t *zc) { return (zone_dataset_attach(curthread->td_ucred, zc->zc_name, (int)zc->zc_jailid)); } static int zfs_ioc_unjail(zfs_cmd_t *zc) { return (zone_dataset_detach(curthread->td_ucred, zc->zc_name, (int)zc->zc_jailid)); } /* * innvl: { * "fd" -> file descriptor to write stream to (int32) * (optional) "fromsnap" -> full snap name to send an incremental from * (optional) "largeblockok" -> (value ignored) * indicates that blocks > 128KB are permitted * (optional) "embedok" -> (value ignored) * presence indicates DRR_WRITE_EMBEDDED records are permitted * (optional) "resume_object" and "resume_offset" -> (uint64) * if present, resume send stream from specified object and offset. * } * * outnvl is unused */ /* ARGSUSED */ static int zfs_ioc_send_new(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl) { cap_rights_t rights; file_t *fp; int error; offset_t off; char *fromname = NULL; int fd; boolean_t largeblockok; boolean_t embedok; uint64_t resumeobj = 0; uint64_t resumeoff = 0; error = nvlist_lookup_int32(innvl, "fd", &fd); if (error != 0) return (SET_ERROR(EINVAL)); (void) nvlist_lookup_string(innvl, "fromsnap", &fromname); largeblockok = nvlist_exists(innvl, "largeblockok"); embedok = nvlist_exists(innvl, "embedok"); (void) nvlist_lookup_uint64(innvl, "resume_object", &resumeobj); (void) nvlist_lookup_uint64(innvl, "resume_offset", &resumeoff); #ifdef illumos file_t *fp = getf(fd); #else fget_write(curthread, fd, cap_rights_init(&rights, CAP_WRITE), &fp); #endif if (fp == NULL) return (SET_ERROR(EBADF)); off = fp->f_offset; error = dmu_send(snapname, fromname, embedok, largeblockok, fd, #ifdef illumos resumeobj, resumeoff, fp->f_vnode, &off); #else resumeobj, resumeoff, fp, &off); #endif #ifdef illumos if (VOP_SEEK(fp->f_vnode, fp->f_offset, &off, NULL) == 0) fp->f_offset = off; #else fp->f_offset = off; #endif releasef(fd); return (error); } /* * Determine approximately how large a zfs send stream will be -- the number * of bytes that will be written to the fd supplied to zfs_ioc_send_new(). * * innvl: { * (optional) "from" -> full snap or bookmark name to send an incremental * from * } * * outnvl: { * "space" -> bytes of space (uint64) * } */ static int zfs_ioc_send_space(const char *snapname, nvlist_t *innvl, nvlist_t *outnvl) { dsl_pool_t *dp; dsl_dataset_t *tosnap; int error; char *fromname; uint64_t space; error = dsl_pool_hold(snapname, FTAG, &dp); if (error != 0) return (error); error = dsl_dataset_hold(dp, snapname, FTAG, &tosnap); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } error = nvlist_lookup_string(innvl, "from", &fromname); if (error == 0) { if (strchr(fromname, '@') != NULL) { /* * If from is a snapshot, hold it and use the more * efficient dmu_send_estimate to estimate send space * size using deadlists. */ dsl_dataset_t *fromsnap; error = dsl_dataset_hold(dp, fromname, FTAG, &fromsnap); if (error != 0) goto out; error = dmu_send_estimate(tosnap, fromsnap, &space); dsl_dataset_rele(fromsnap, FTAG); } else if (strchr(fromname, '#') != NULL) { /* * If from is a bookmark, fetch the creation TXG of the * snapshot it was created from and use that to find * blocks that were born after it. */ zfs_bookmark_phys_t frombm; error = dsl_bookmark_lookup(dp, fromname, tosnap, &frombm); if (error != 0) goto out; error = dmu_send_estimate_from_txg(tosnap, frombm.zbm_creation_txg, &space); } else { /* * from is not properly formatted as a snapshot or * bookmark */ error = SET_ERROR(EINVAL); goto out; } } else { // If estimating the size of a full send, use dmu_send_estimate error = dmu_send_estimate(tosnap, NULL, &space); } fnvlist_add_uint64(outnvl, "space", space); out: dsl_dataset_rele(tosnap, FTAG); dsl_pool_rele(dp, FTAG); return (error); } static zfs_ioc_vec_t zfs_ioc_vec[ZFS_IOC_LAST - ZFS_IOC_FIRST]; static void zfs_ioctl_register_legacy(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy, zfs_ioc_namecheck_t namecheck, boolean_t log_history, zfs_ioc_poolcheck_t pool_check) { zfs_ioc_vec_t *vec = &zfs_ioc_vec[ioc - ZFS_IOC_FIRST]; ASSERT3U(ioc, >=, ZFS_IOC_FIRST); ASSERT3U(ioc, <, ZFS_IOC_LAST); ASSERT3P(vec->zvec_legacy_func, ==, NULL); ASSERT3P(vec->zvec_func, ==, NULL); vec->zvec_legacy_func = func; vec->zvec_secpolicy = secpolicy; vec->zvec_namecheck = namecheck; vec->zvec_allow_log = log_history; vec->zvec_pool_check = pool_check; } /* * See the block comment at the beginning of this file for details on * each argument to this function. */ static void zfs_ioctl_register(const char *name, zfs_ioc_t ioc, zfs_ioc_func_t *func, zfs_secpolicy_func_t *secpolicy, zfs_ioc_namecheck_t namecheck, zfs_ioc_poolcheck_t pool_check, boolean_t smush_outnvlist, boolean_t allow_log) { zfs_ioc_vec_t *vec = &zfs_ioc_vec[ioc - ZFS_IOC_FIRST]; ASSERT3U(ioc, >=, ZFS_IOC_FIRST); ASSERT3U(ioc, <, ZFS_IOC_LAST); ASSERT3P(vec->zvec_legacy_func, ==, NULL); ASSERT3P(vec->zvec_func, ==, NULL); /* if we are logging, the name must be valid */ ASSERT(!allow_log || namecheck != NO_NAME); vec->zvec_name = name; vec->zvec_func = func; vec->zvec_secpolicy = secpolicy; vec->zvec_namecheck = namecheck; vec->zvec_pool_check = pool_check; vec->zvec_smush_outnvlist = smush_outnvlist; vec->zvec_allow_log = allow_log; } static void zfs_ioctl_register_pool(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy, boolean_t log_history, zfs_ioc_poolcheck_t pool_check) { zfs_ioctl_register_legacy(ioc, func, secpolicy, POOL_NAME, log_history, pool_check); } static void zfs_ioctl_register_dataset_nolog(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy, zfs_ioc_poolcheck_t pool_check) { zfs_ioctl_register_legacy(ioc, func, secpolicy, DATASET_NAME, B_FALSE, pool_check); } static void zfs_ioctl_register_pool_modify(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func) { zfs_ioctl_register_legacy(ioc, func, zfs_secpolicy_config, POOL_NAME, B_TRUE, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); } static void zfs_ioctl_register_pool_meta(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy) { zfs_ioctl_register_legacy(ioc, func, secpolicy, NO_NAME, B_FALSE, POOL_CHECK_NONE); } static void zfs_ioctl_register_dataset_read_secpolicy(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy) { zfs_ioctl_register_legacy(ioc, func, secpolicy, DATASET_NAME, B_FALSE, POOL_CHECK_SUSPENDED); } static void zfs_ioctl_register_dataset_read(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func) { zfs_ioctl_register_dataset_read_secpolicy(ioc, func, zfs_secpolicy_read); } static void zfs_ioctl_register_dataset_modify(zfs_ioc_t ioc, zfs_ioc_legacy_func_t *func, zfs_secpolicy_func_t *secpolicy) { zfs_ioctl_register_legacy(ioc, func, secpolicy, DATASET_NAME, B_TRUE, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); } static void zfs_ioctl_init(void) { zfs_ioctl_register("snapshot", ZFS_IOC_SNAPSHOT, zfs_ioc_snapshot, zfs_secpolicy_snapshot, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); zfs_ioctl_register("log_history", ZFS_IOC_LOG_HISTORY, zfs_ioc_log_history, zfs_secpolicy_log_history, NO_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_FALSE); zfs_ioctl_register("space_snaps", ZFS_IOC_SPACE_SNAPS, zfs_ioc_space_snaps, zfs_secpolicy_read, DATASET_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE); zfs_ioctl_register("send", ZFS_IOC_SEND_NEW, zfs_ioc_send_new, zfs_secpolicy_send_new, DATASET_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE); zfs_ioctl_register("send_space", ZFS_IOC_SEND_SPACE, zfs_ioc_send_space, zfs_secpolicy_read, DATASET_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE); zfs_ioctl_register("create", ZFS_IOC_CREATE, zfs_ioc_create, zfs_secpolicy_create_clone, DATASET_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); zfs_ioctl_register("clone", ZFS_IOC_CLONE, zfs_ioc_clone, zfs_secpolicy_create_clone, DATASET_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); zfs_ioctl_register("destroy_snaps", ZFS_IOC_DESTROY_SNAPS, zfs_ioc_destroy_snaps, zfs_secpolicy_destroy_snaps, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); zfs_ioctl_register("hold", ZFS_IOC_HOLD, zfs_ioc_hold, zfs_secpolicy_hold, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); zfs_ioctl_register("release", ZFS_IOC_RELEASE, zfs_ioc_release, zfs_secpolicy_release, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); zfs_ioctl_register("get_holds", ZFS_IOC_GET_HOLDS, zfs_ioc_get_holds, zfs_secpolicy_read, DATASET_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE); zfs_ioctl_register("rollback", ZFS_IOC_ROLLBACK, zfs_ioc_rollback, zfs_secpolicy_rollback, DATASET_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_FALSE, B_TRUE); zfs_ioctl_register("bookmark", ZFS_IOC_BOOKMARK, zfs_ioc_bookmark, zfs_secpolicy_bookmark, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); zfs_ioctl_register("get_bookmarks", ZFS_IOC_GET_BOOKMARKS, zfs_ioc_get_bookmarks, zfs_secpolicy_read, DATASET_NAME, POOL_CHECK_SUSPENDED, B_FALSE, B_FALSE); zfs_ioctl_register("destroy_bookmarks", ZFS_IOC_DESTROY_BOOKMARKS, zfs_ioc_destroy_bookmarks, zfs_secpolicy_destroy_bookmarks, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); /* IOCTLS that use the legacy function signature */ zfs_ioctl_register_legacy(ZFS_IOC_POOL_FREEZE, zfs_ioc_pool_freeze, zfs_secpolicy_config, NO_NAME, B_FALSE, POOL_CHECK_READONLY); zfs_ioctl_register_pool(ZFS_IOC_POOL_CREATE, zfs_ioc_pool_create, zfs_secpolicy_config, B_TRUE, POOL_CHECK_NONE); zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_SCAN, zfs_ioc_pool_scan); zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_UPGRADE, zfs_ioc_pool_upgrade); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_ADD, zfs_ioc_vdev_add); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_REMOVE, zfs_ioc_vdev_remove); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SET_STATE, zfs_ioc_vdev_set_state); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_ATTACH, zfs_ioc_vdev_attach); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_DETACH, zfs_ioc_vdev_detach); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SETPATH, zfs_ioc_vdev_setpath); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SETFRU, zfs_ioc_vdev_setfru); zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_SET_PROPS, zfs_ioc_pool_set_props); zfs_ioctl_register_pool_modify(ZFS_IOC_VDEV_SPLIT, zfs_ioc_vdev_split); zfs_ioctl_register_pool_modify(ZFS_IOC_POOL_REGUID, zfs_ioc_pool_reguid); zfs_ioctl_register_pool_meta(ZFS_IOC_POOL_CONFIGS, zfs_ioc_pool_configs, zfs_secpolicy_none); zfs_ioctl_register_pool_meta(ZFS_IOC_POOL_TRYIMPORT, zfs_ioc_pool_tryimport, zfs_secpolicy_config); zfs_ioctl_register_pool_meta(ZFS_IOC_INJECT_FAULT, zfs_ioc_inject_fault, zfs_secpolicy_inject); zfs_ioctl_register_pool_meta(ZFS_IOC_CLEAR_FAULT, zfs_ioc_clear_fault, zfs_secpolicy_inject); zfs_ioctl_register_pool_meta(ZFS_IOC_INJECT_LIST_NEXT, zfs_ioc_inject_list_next, zfs_secpolicy_inject); /* * pool destroy, and export don't log the history as part of * zfsdev_ioctl, but rather zfs_ioc_pool_export * does the logging of those commands. */ zfs_ioctl_register_pool(ZFS_IOC_POOL_DESTROY, zfs_ioc_pool_destroy, zfs_secpolicy_config, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_POOL_EXPORT, zfs_ioc_pool_export, zfs_secpolicy_config, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_POOL_STATS, zfs_ioc_pool_stats, zfs_secpolicy_read, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_POOL_GET_PROPS, zfs_ioc_pool_get_props, zfs_secpolicy_read, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_ERROR_LOG, zfs_ioc_error_log, zfs_secpolicy_inject, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_DSOBJ_TO_DSNAME, zfs_ioc_dsobj_to_dsname, zfs_secpolicy_diff, B_FALSE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_POOL_GET_HISTORY, zfs_ioc_pool_get_history, zfs_secpolicy_config, B_FALSE, POOL_CHECK_SUSPENDED); zfs_ioctl_register_pool(ZFS_IOC_POOL_IMPORT, zfs_ioc_pool_import, zfs_secpolicy_config, B_TRUE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_CLEAR, zfs_ioc_clear, zfs_secpolicy_config, B_TRUE, POOL_CHECK_NONE); zfs_ioctl_register_pool(ZFS_IOC_POOL_REOPEN, zfs_ioc_pool_reopen, zfs_secpolicy_config, B_TRUE, POOL_CHECK_SUSPENDED); zfs_ioctl_register_dataset_read(ZFS_IOC_SPACE_WRITTEN, zfs_ioc_space_written); zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_RECVD_PROPS, zfs_ioc_objset_recvd_props); zfs_ioctl_register_dataset_read(ZFS_IOC_NEXT_OBJ, zfs_ioc_next_obj); zfs_ioctl_register_dataset_read(ZFS_IOC_GET_FSACL, zfs_ioc_get_fsacl); zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_STATS, zfs_ioc_objset_stats); zfs_ioctl_register_dataset_read(ZFS_IOC_OBJSET_ZPLPROPS, zfs_ioc_objset_zplprops); zfs_ioctl_register_dataset_read(ZFS_IOC_DATASET_LIST_NEXT, zfs_ioc_dataset_list_next); zfs_ioctl_register_dataset_read(ZFS_IOC_SNAPSHOT_LIST_NEXT, zfs_ioc_snapshot_list_next); zfs_ioctl_register_dataset_read(ZFS_IOC_SEND_PROGRESS, zfs_ioc_send_progress); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_DIFF, zfs_ioc_diff, zfs_secpolicy_diff); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_OBJ_TO_STATS, zfs_ioc_obj_to_stats, zfs_secpolicy_diff); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_OBJ_TO_PATH, zfs_ioc_obj_to_path, zfs_secpolicy_diff); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_USERSPACE_ONE, zfs_ioc_userspace_one, zfs_secpolicy_userspace_one); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_USERSPACE_MANY, zfs_ioc_userspace_many, zfs_secpolicy_userspace_many); zfs_ioctl_register_dataset_read_secpolicy(ZFS_IOC_SEND, zfs_ioc_send, zfs_secpolicy_send); zfs_ioctl_register_dataset_modify(ZFS_IOC_SET_PROP, zfs_ioc_set_prop, zfs_secpolicy_none); zfs_ioctl_register_dataset_modify(ZFS_IOC_DESTROY, zfs_ioc_destroy, zfs_secpolicy_destroy); zfs_ioctl_register_dataset_modify(ZFS_IOC_RENAME, zfs_ioc_rename, zfs_secpolicy_rename); zfs_ioctl_register_dataset_modify(ZFS_IOC_RECV, zfs_ioc_recv, zfs_secpolicy_recv); zfs_ioctl_register_dataset_modify(ZFS_IOC_PROMOTE, zfs_ioc_promote, zfs_secpolicy_promote); zfs_ioctl_register_dataset_modify(ZFS_IOC_INHERIT_PROP, zfs_ioc_inherit_prop, zfs_secpolicy_inherit_prop); zfs_ioctl_register_dataset_modify(ZFS_IOC_SET_FSACL, zfs_ioc_set_fsacl, zfs_secpolicy_set_fsacl); zfs_ioctl_register_dataset_nolog(ZFS_IOC_SHARE, zfs_ioc_share, zfs_secpolicy_share, POOL_CHECK_NONE); zfs_ioctl_register_dataset_nolog(ZFS_IOC_SMB_ACL, zfs_ioc_smb_acl, zfs_secpolicy_smb_acl, POOL_CHECK_NONE); zfs_ioctl_register_dataset_nolog(ZFS_IOC_USERSPACE_UPGRADE, zfs_ioc_userspace_upgrade, zfs_secpolicy_userspace_upgrade, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); zfs_ioctl_register_dataset_nolog(ZFS_IOC_TMP_SNAPSHOT, zfs_ioc_tmp_snapshot, zfs_secpolicy_tmp_snapshot, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY); #ifdef __FreeBSD__ zfs_ioctl_register_dataset_nolog(ZFS_IOC_JAIL, zfs_ioc_jail, zfs_secpolicy_config, POOL_CHECK_NONE); zfs_ioctl_register_dataset_nolog(ZFS_IOC_UNJAIL, zfs_ioc_unjail, zfs_secpolicy_config, POOL_CHECK_NONE); #endif } int pool_status_check(const char *name, zfs_ioc_namecheck_t type, zfs_ioc_poolcheck_t check) { spa_t *spa; int error; ASSERT(type == POOL_NAME || type == DATASET_NAME); if (check & POOL_CHECK_NONE) return (0); error = spa_open(name, &spa, FTAG); if (error == 0) { if ((check & POOL_CHECK_SUSPENDED) && spa_suspended(spa)) error = SET_ERROR(EAGAIN); else if ((check & POOL_CHECK_READONLY) && !spa_writeable(spa)) error = SET_ERROR(EROFS); spa_close(spa, FTAG); } return (error); } /* * Find a free minor number. */ minor_t zfsdev_minor_alloc(void) { static minor_t last_minor; minor_t m; ASSERT(MUTEX_HELD(&spa_namespace_lock)); for (m = last_minor + 1; m != last_minor; m++) { if (m > ZFSDEV_MAX_MINOR) m = 1; if (ddi_get_soft_state(zfsdev_state, m) == NULL) { last_minor = m; return (m); } } return (0); } static int zfs_ctldev_init(struct cdev *devp) { minor_t minor; zfs_soft_state_t *zs; ASSERT(MUTEX_HELD(&spa_namespace_lock)); minor = zfsdev_minor_alloc(); if (minor == 0) return (SET_ERROR(ENXIO)); if (ddi_soft_state_zalloc(zfsdev_state, minor) != DDI_SUCCESS) return (SET_ERROR(EAGAIN)); devfs_set_cdevpriv((void *)(uintptr_t)minor, zfsdev_close); zs = ddi_get_soft_state(zfsdev_state, minor); zs->zss_type = ZSST_CTLDEV; zfs_onexit_init((zfs_onexit_t **)&zs->zss_data); return (0); } static void zfs_ctldev_destroy(zfs_onexit_t *zo, minor_t minor) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); zfs_onexit_destroy(zo); ddi_soft_state_free(zfsdev_state, minor); } void * zfsdev_get_soft_state(minor_t minor, enum zfs_soft_state_type which) { zfs_soft_state_t *zp; zp = ddi_get_soft_state(zfsdev_state, minor); if (zp == NULL || zp->zss_type != which) return (NULL); return (zp->zss_data); } static int zfsdev_open(struct cdev *devp, int flag, int mode, struct thread *td) { int error = 0; #ifdef illumos if (getminor(*devp) != 0) return (zvol_open(devp, flag, otyp, cr)); #endif /* This is the control device. Allocate a new minor if requested. */ if (flag & FEXCL) { mutex_enter(&spa_namespace_lock); error = zfs_ctldev_init(devp); mutex_exit(&spa_namespace_lock); } return (error); } static void zfsdev_close(void *data) { zfs_onexit_t *zo; minor_t minor = (minor_t)(uintptr_t)data; if (minor == 0) return; mutex_enter(&spa_namespace_lock); zo = zfsdev_get_soft_state(minor, ZSST_CTLDEV); if (zo == NULL) { mutex_exit(&spa_namespace_lock); return; } zfs_ctldev_destroy(zo, minor); mutex_exit(&spa_namespace_lock); } static int zfsdev_ioctl(struct cdev *dev, u_long zcmd, caddr_t arg, int flag, struct thread *td) { zfs_cmd_t *zc; uint_t vecnum; int error, rc, len; #ifdef illumos minor_t minor = getminor(dev); #else zfs_iocparm_t *zc_iocparm; int cflag, cmd, oldvecnum; boolean_t newioc, compat; void *compat_zc = NULL; cred_t *cr = td->td_ucred; #endif const zfs_ioc_vec_t *vec; char *saved_poolname = NULL; nvlist_t *innvl = NULL; cflag = ZFS_CMD_COMPAT_NONE; compat = B_FALSE; newioc = B_TRUE; /* "new" style (zfs_iocparm_t) ioctl */ len = IOCPARM_LEN(zcmd); vecnum = cmd = zcmd & 0xff; /* * Check if we are talking to supported older binaries * and translate zfs_cmd if necessary */ if (len != sizeof(zfs_iocparm_t)) { newioc = B_FALSE; compat = B_TRUE; vecnum = cmd; switch (len) { case sizeof(zfs_cmd_zcmd_t): cflag = ZFS_CMD_COMPAT_LZC; break; case sizeof(zfs_cmd_deadman_t): cflag = ZFS_CMD_COMPAT_DEADMAN; break; case sizeof(zfs_cmd_v28_t): cflag = ZFS_CMD_COMPAT_V28; break; case sizeof(zfs_cmd_v15_t): cflag = ZFS_CMD_COMPAT_V15; vecnum = zfs_ioctl_v15_to_v28[cmd]; /* * Return without further handling * if the command is blacklisted. */ if (vecnum == ZFS_IOC_COMPAT_PASS) return (0); else if (vecnum == ZFS_IOC_COMPAT_FAIL) return (ENOTSUP); break; default: return (EINVAL); } } #ifdef illumos vecnum = cmd - ZFS_IOC_FIRST; ASSERT3U(getmajor(dev), ==, ddi_driver_major(zfs_dip)); #endif if (vecnum >= sizeof (zfs_ioc_vec) / sizeof (zfs_ioc_vec[0])) return (SET_ERROR(EINVAL)); vec = &zfs_ioc_vec[vecnum]; zc = kmem_zalloc(sizeof(zfs_cmd_t), KM_SLEEP); #ifdef illumos error = ddi_copyin((void *)arg, zc, sizeof (zfs_cmd_t), flag); if (error != 0) { error = SET_ERROR(EFAULT); goto out; } #else /* !illumos */ bzero(zc, sizeof(zfs_cmd_t)); if (newioc) { zc_iocparm = (void *)arg; switch (zc_iocparm->zfs_ioctl_version) { case ZFS_IOCVER_CURRENT: if (zc_iocparm->zfs_cmd_size != sizeof(zfs_cmd_t)) { error = SET_ERROR(EINVAL); goto out; } break; case ZFS_IOCVER_RESUME: if (zc_iocparm->zfs_cmd_size != sizeof(zfs_cmd_resume_t)) { error = SET_ERROR(EFAULT); goto out; } compat = B_TRUE; cflag = ZFS_CMD_COMPAT_RESUME; break; case ZFS_IOCVER_EDBP: if (zc_iocparm->zfs_cmd_size != sizeof(zfs_cmd_edbp_t)) { error = SET_ERROR(EFAULT); goto out; } compat = B_TRUE; cflag = ZFS_CMD_COMPAT_EDBP; break; case ZFS_IOCVER_ZCMD: if (zc_iocparm->zfs_cmd_size > sizeof(zfs_cmd_t) || zc_iocparm->zfs_cmd_size < sizeof(zfs_cmd_zcmd_t)) { error = SET_ERROR(EFAULT); goto out; } compat = B_TRUE; cflag = ZFS_CMD_COMPAT_ZCMD; break; default: error = SET_ERROR(EINVAL); goto out; /* NOTREACHED */ } if (compat) { ASSERT(sizeof(zfs_cmd_t) >= zc_iocparm->zfs_cmd_size); compat_zc = kmem_zalloc(sizeof(zfs_cmd_t), KM_SLEEP); bzero(compat_zc, sizeof(zfs_cmd_t)); error = ddi_copyin((void *)(uintptr_t)zc_iocparm->zfs_cmd, compat_zc, zc_iocparm->zfs_cmd_size, flag); if (error != 0) { error = SET_ERROR(EFAULT); goto out; } } else { error = ddi_copyin((void *)(uintptr_t)zc_iocparm->zfs_cmd, zc, zc_iocparm->zfs_cmd_size, flag); if (error != 0) { error = SET_ERROR(EFAULT); goto out; } } } if (compat) { if (newioc) { ASSERT(compat_zc != NULL); zfs_cmd_compat_get(zc, compat_zc, cflag); } else { ASSERT(compat_zc == NULL); zfs_cmd_compat_get(zc, arg, cflag); } oldvecnum = vecnum; error = zfs_ioctl_compat_pre(zc, &vecnum, cflag); if (error != 0) goto out; if (oldvecnum != vecnum) vec = &zfs_ioc_vec[vecnum]; } #endif /* !illumos */ zc->zc_iflags = flag & FKIOCTL; if (zc->zc_nvlist_src_size != 0) { error = get_nvlist(zc->zc_nvlist_src, zc->zc_nvlist_src_size, zc->zc_iflags, &innvl); if (error != 0) goto out; } /* rewrite innvl for backwards compatibility */ if (compat) innvl = zfs_ioctl_compat_innvl(zc, innvl, vecnum, cflag); /* * Ensure that all pool/dataset names are valid before we pass down to * the lower layers. */ zc->zc_name[sizeof (zc->zc_name) - 1] = '\0'; switch (vec->zvec_namecheck) { case POOL_NAME: if (pool_namecheck(zc->zc_name, NULL, NULL) != 0) error = SET_ERROR(EINVAL); else error = pool_status_check(zc->zc_name, vec->zvec_namecheck, vec->zvec_pool_check); break; case DATASET_NAME: if (dataset_namecheck(zc->zc_name, NULL, NULL) != 0) error = SET_ERROR(EINVAL); else error = pool_status_check(zc->zc_name, vec->zvec_namecheck, vec->zvec_pool_check); break; case NO_NAME: break; } if (error == 0 && !(flag & FKIOCTL)) error = vec->zvec_secpolicy(zc, innvl, cr); if (error != 0) goto out; /* legacy ioctls can modify zc_name */ len = strcspn(zc->zc_name, "/@#") + 1; saved_poolname = kmem_alloc(len, KM_SLEEP); (void) strlcpy(saved_poolname, zc->zc_name, len); if (vec->zvec_func != NULL) { nvlist_t *outnvl; int puterror = 0; spa_t *spa; nvlist_t *lognv = NULL; ASSERT(vec->zvec_legacy_func == NULL); /* * Add the innvl to the lognv before calling the func, * in case the func changes the innvl. */ if (vec->zvec_allow_log) { lognv = fnvlist_alloc(); fnvlist_add_string(lognv, ZPOOL_HIST_IOCTL, vec->zvec_name); if (!nvlist_empty(innvl)) { fnvlist_add_nvlist(lognv, ZPOOL_HIST_INPUT_NVL, innvl); } } outnvl = fnvlist_alloc(); error = vec->zvec_func(zc->zc_name, innvl, outnvl); if (error == 0 && vec->zvec_allow_log && spa_open(zc->zc_name, &spa, FTAG) == 0) { if (!nvlist_empty(outnvl)) { fnvlist_add_nvlist(lognv, ZPOOL_HIST_OUTPUT_NVL, outnvl); } (void) spa_history_log_nvl(spa, lognv); spa_close(spa, FTAG); } fnvlist_free(lognv); /* rewrite outnvl for backwards compatibility */ if (compat) outnvl = zfs_ioctl_compat_outnvl(zc, outnvl, vecnum, cflag); if (!nvlist_empty(outnvl) || zc->zc_nvlist_dst_size != 0) { int smusherror = 0; if (vec->zvec_smush_outnvlist) { smusherror = nvlist_smush(outnvl, zc->zc_nvlist_dst_size); } if (smusherror == 0) puterror = put_nvlist(zc, outnvl); } if (puterror != 0) error = puterror; nvlist_free(outnvl); } else { error = vec->zvec_legacy_func(zc); } out: nvlist_free(innvl); #ifdef illumos rc = ddi_copyout(zc, (void *)arg, sizeof (zfs_cmd_t), flag); if (error == 0 && rc != 0) error = SET_ERROR(EFAULT); #else if (compat) { zfs_ioctl_compat_post(zc, cmd, cflag); if (newioc) { ASSERT(compat_zc != NULL); ASSERT(sizeof(zfs_cmd_t) >= zc_iocparm->zfs_cmd_size); zfs_cmd_compat_put(zc, compat_zc, vecnum, cflag); rc = ddi_copyout(compat_zc, (void *)(uintptr_t)zc_iocparm->zfs_cmd, zc_iocparm->zfs_cmd_size, flag); if (error == 0 && rc != 0) error = SET_ERROR(EFAULT); kmem_free(compat_zc, sizeof (zfs_cmd_t)); } else { zfs_cmd_compat_put(zc, arg, vecnum, cflag); } } else { ASSERT(newioc); rc = ddi_copyout(zc, (void *)(uintptr_t)zc_iocparm->zfs_cmd, sizeof (zfs_cmd_t), flag); if (error == 0 && rc != 0) error = SET_ERROR(EFAULT); } #endif if (error == 0 && vec->zvec_allow_log) { char *s = tsd_get(zfs_allow_log_key); if (s != NULL) strfree(s); (void) tsd_set(zfs_allow_log_key, saved_poolname); } else { if (saved_poolname != NULL) strfree(saved_poolname); } kmem_free(zc, sizeof (zfs_cmd_t)); return (error); } #ifdef illumos static int zfs_attach(dev_info_t *dip, ddi_attach_cmd_t cmd) { if (cmd != DDI_ATTACH) return (DDI_FAILURE); if (ddi_create_minor_node(dip, "zfs", S_IFCHR, 0, DDI_PSEUDO, 0) == DDI_FAILURE) return (DDI_FAILURE); zfs_dip = dip; ddi_report_dev(dip); return (DDI_SUCCESS); } static int zfs_detach(dev_info_t *dip, ddi_detach_cmd_t cmd) { if (spa_busy() || zfs_busy() || zvol_busy()) return (DDI_FAILURE); if (cmd != DDI_DETACH) return (DDI_FAILURE); zfs_dip = NULL; ddi_prop_remove_all(dip); ddi_remove_minor_node(dip, NULL); return (DDI_SUCCESS); } /*ARGSUSED*/ static int zfs_info(dev_info_t *dip, ddi_info_cmd_t infocmd, void *arg, void **result) { switch (infocmd) { case DDI_INFO_DEVT2DEVINFO: *result = zfs_dip; return (DDI_SUCCESS); case DDI_INFO_DEVT2INSTANCE: *result = (void *)0; return (DDI_SUCCESS); } return (DDI_FAILURE); } #endif /* illumos */ /* * OK, so this is a little weird. * * /dev/zfs is the control node, i.e. minor 0. * /dev/zvol/[r]dsk/pool/dataset are the zvols, minor > 0. * * /dev/zfs has basically nothing to do except serve up ioctls, * so most of the standard driver entry points are in zvol.c. */ #ifdef illumos static struct cb_ops zfs_cb_ops = { zfsdev_open, /* open */ zfsdev_close, /* close */ zvol_strategy, /* strategy */ nodev, /* print */ zvol_dump, /* dump */ zvol_read, /* read */ zvol_write, /* write */ zfsdev_ioctl, /* ioctl */ nodev, /* devmap */ nodev, /* mmap */ nodev, /* segmap */ nochpoll, /* poll */ ddi_prop_op, /* prop_op */ NULL, /* streamtab */ D_NEW | D_MP | D_64BIT, /* Driver compatibility flag */ CB_REV, /* version */ nodev, /* async read */ nodev, /* async write */ }; static struct dev_ops zfs_dev_ops = { DEVO_REV, /* version */ 0, /* refcnt */ zfs_info, /* info */ nulldev, /* identify */ nulldev, /* probe */ zfs_attach, /* attach */ zfs_detach, /* detach */ nodev, /* reset */ &zfs_cb_ops, /* driver operations */ NULL, /* no bus operations */ NULL, /* power */ ddi_quiesce_not_needed, /* quiesce */ }; static struct modldrv zfs_modldrv = { &mod_driverops, "ZFS storage pool", &zfs_dev_ops }; static struct modlinkage modlinkage = { MODREV_1, (void *)&zfs_modlfs, (void *)&zfs_modldrv, NULL }; #endif /* illumos */ static struct cdevsw zfs_cdevsw = { .d_version = D_VERSION, .d_open = zfsdev_open, .d_ioctl = zfsdev_ioctl, .d_name = ZFS_DEV_NAME }; static void zfs_allow_log_destroy(void *arg) { char *poolname = arg; strfree(poolname); } static void zfsdev_init(void) { zfsdev = make_dev(&zfs_cdevsw, 0x0, UID_ROOT, GID_OPERATOR, 0666, ZFS_DEV_NAME); } static void zfsdev_fini(void) { if (zfsdev != NULL) destroy_dev(zfsdev); } static struct root_hold_token *zfs_root_token; struct proc *zfsproc; #ifdef illumos int _init(void) { int error; spa_init(FREAD | FWRITE); zfs_init(); zvol_init(); zfs_ioctl_init(); if ((error = mod_install(&modlinkage)) != 0) { zvol_fini(); zfs_fini(); spa_fini(); return (error); } tsd_create(&zfs_fsyncer_key, NULL); tsd_create(&rrw_tsd_key, rrw_tsd_destroy); tsd_create(&zfs_allow_log_key, zfs_allow_log_destroy); error = ldi_ident_from_mod(&modlinkage, &zfs_li); ASSERT(error == 0); mutex_init(&zfs_share_lock, NULL, MUTEX_DEFAULT, NULL); return (0); } int _fini(void) { int error; if (spa_busy() || zfs_busy() || zvol_busy() || zio_injection_enabled) return (SET_ERROR(EBUSY)); if ((error = mod_remove(&modlinkage)) != 0) return (error); zvol_fini(); zfs_fini(); spa_fini(); if (zfs_nfsshare_inited) (void) ddi_modclose(nfs_mod); if (zfs_smbshare_inited) (void) ddi_modclose(smbsrv_mod); if (zfs_nfsshare_inited || zfs_smbshare_inited) (void) ddi_modclose(sharefs_mod); tsd_destroy(&zfs_fsyncer_key); ldi_ident_release(zfs_li); zfs_li = NULL; mutex_destroy(&zfs_share_lock); return (error); } int _info(struct modinfo *modinfop) { return (mod_info(&modlinkage, modinfop)); } #endif /* illumos */ static int zfs__init(void); static int zfs__fini(void); static void zfs_shutdown(void *, int); static eventhandler_tag zfs_shutdown_event_tag; #ifdef __FreeBSD__ #define ZFS_MIN_KSTACK_PAGES 4 #endif int zfs__init(void) { #ifdef __FreeBSD__ #if KSTACK_PAGES < ZFS_MIN_KSTACK_PAGES printf("ZFS NOTICE: KSTACK_PAGES is %d which could result in stack " "overflow panic!\nPlease consider adding " "'options KSTACK_PAGES=%d' to your kernel config\n", KSTACK_PAGES, ZFS_MIN_KSTACK_PAGES); #endif #endif zfs_root_token = root_mount_hold("ZFS"); mutex_init(&zfs_share_lock, NULL, MUTEX_DEFAULT, NULL); spa_init(FREAD | FWRITE); zfs_init(); zvol_init(); zfs_ioctl_init(); tsd_create(&zfs_fsyncer_key, NULL); tsd_create(&rrw_tsd_key, rrw_tsd_destroy); tsd_create(&zfs_allow_log_key, zfs_allow_log_destroy); tsd_create(&zfs_geom_probe_vdev_key, NULL); printf("ZFS storage pool version: features support (" SPA_VERSION_STRING ")\n"); root_mount_rel(zfs_root_token); zfsdev_init(); return (0); } int zfs__fini(void) { if (spa_busy() || zfs_busy() || zvol_busy() || zio_injection_enabled) { return (EBUSY); } zfsdev_fini(); zvol_fini(); zfs_fini(); spa_fini(); tsd_destroy(&zfs_fsyncer_key); tsd_destroy(&rrw_tsd_key); tsd_destroy(&zfs_allow_log_key); mutex_destroy(&zfs_share_lock); return (0); } static void zfs_shutdown(void *arg __unused, int howto __unused) { /* * ZFS fini routines can not properly work in a panic-ed system. */ if (panicstr == NULL) (void)zfs__fini(); } static int zfs_modevent(module_t mod, int type, void *unused __unused) { int err; switch (type) { case MOD_LOAD: err = zfs__init(); if (err == 0) zfs_shutdown_event_tag = EVENTHANDLER_REGISTER( shutdown_post_sync, zfs_shutdown, NULL, SHUTDOWN_PRI_FIRST); return (err); case MOD_UNLOAD: err = zfs__fini(); if (err == 0 && zfs_shutdown_event_tag != NULL) EVENTHANDLER_DEREGISTER(shutdown_post_sync, zfs_shutdown_event_tag); return (err); case MOD_SHUTDOWN: return (0); default: break; } return (EOPNOTSUPP); } static moduledata_t zfs_mod = { "zfsctrl", zfs_modevent, 0 }; DECLARE_MODULE(zfsctrl, zfs_mod, SI_SUB_VFS, SI_ORDER_ANY); MODULE_VERSION(zfsctrl, 1); MODULE_DEPEND(zfsctrl, opensolaris, 1, 1, 1); MODULE_DEPEND(zfsctrl, krpc, 1, 1, 1); MODULE_DEPEND(zfsctrl, acl_nfs4, 1, 1, 1); Index: projects/vnet/sys/cddl/contrib/opensolaris =================================================================== --- projects/vnet/sys/cddl/contrib/opensolaris (revision 302157) +++ projects/vnet/sys/cddl/contrib/opensolaris (revision 302158) Property changes on: projects/vnet/sys/cddl/contrib/opensolaris ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/sys/cddl/contrib/opensolaris:r302084-302157 Index: projects/vnet/sys/dev/mlx5/mlx5_core/mlx5_vport.c =================================================================== --- projects/vnet/sys/dev/mlx5/mlx5_core/mlx5_vport.c (revision 302157) +++ projects/vnet/sys/dev/mlx5/mlx5_core/mlx5_vport.c (revision 302158) @@ -1,1280 +1,1280 @@ /*- * Copyright (c) 2013-2015, 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, 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. * * $FreeBSD$ */ #include #include #include #include "mlx5_core.h" u8 mlx5_query_vport_state(struct mlx5_core_dev *mdev, u8 opmod) { u32 in[MLX5_ST_SZ_DW(query_vport_state_in)]; u32 out[MLX5_ST_SZ_DW(query_vport_state_out)]; int err; memset(in, 0, sizeof(in)); MLX5_SET(query_vport_state_in, in, opcode, MLX5_CMD_OP_QUERY_VPORT_STATE); MLX5_SET(query_vport_state_in, in, op_mod, opmod); err = mlx5_cmd_exec_check_status(mdev, in, sizeof(in), out, sizeof(out)); if (err) mlx5_core_warn(mdev, "MLX5_CMD_OP_QUERY_VPORT_STATE failed\n"); return MLX5_GET(query_vport_state_out, out, state); } EXPORT_SYMBOL_GPL(mlx5_query_vport_state); static int mlx5_query_nic_vport_context(struct mlx5_core_dev *mdev, u32 vport, u32 *out, int outlen) { u32 in[MLX5_ST_SZ_DW(query_nic_vport_context_in)]; memset(in, 0, sizeof(in)); MLX5_SET(query_nic_vport_context_in, in, opcode, MLX5_CMD_OP_QUERY_NIC_VPORT_CONTEXT); MLX5_SET(query_nic_vport_context_in, in, vport_number, vport); if (vport) MLX5_SET(query_nic_vport_context_in, in, other_vport, 1); return mlx5_cmd_exec_check_status(mdev, in, sizeof(in), out, outlen); } int mlx5_vport_alloc_q_counter(struct mlx5_core_dev *mdev, int *counter_set_id) { u32 in[MLX5_ST_SZ_DW(alloc_q_counter_in)]; - u32 out[MLX5_ST_SZ_DW(alloc_q_counter_in)]; + u32 out[MLX5_ST_SZ_DW(alloc_q_counter_out)]; int err; memset(in, 0, sizeof(in)); memset(out, 0, sizeof(out)); MLX5_SET(alloc_q_counter_in, in, opcode, MLX5_CMD_OP_ALLOC_Q_COUNTER); err = mlx5_cmd_exec_check_status(mdev, in, sizeof(in), out, sizeof(out)); if (err) return err; *counter_set_id = MLX5_GET(alloc_q_counter_out, out, counter_set_id); return err; } int mlx5_vport_dealloc_q_counter(struct mlx5_core_dev *mdev, int counter_set_id) { u32 in[MLX5_ST_SZ_DW(dealloc_q_counter_in)]; u32 out[MLX5_ST_SZ_DW(dealloc_q_counter_out)]; memset(in, 0, sizeof(in)); memset(out, 0, sizeof(out)); MLX5_SET(dealloc_q_counter_in, in, opcode, MLX5_CMD_OP_DEALLOC_Q_COUNTER); MLX5_SET(dealloc_q_counter_in, in, counter_set_id, counter_set_id); return mlx5_cmd_exec_check_status(mdev, in, sizeof(in), out, sizeof(out)); } static int mlx5_vport_query_q_counter(struct mlx5_core_dev *mdev, int counter_set_id, int reset, void *out, int out_size) { u32 in[MLX5_ST_SZ_DW(query_q_counter_in)]; memset(in, 0, sizeof(in)); MLX5_SET(query_q_counter_in, in, opcode, MLX5_CMD_OP_QUERY_Q_COUNTER); MLX5_SET(query_q_counter_in, in, clear, reset); MLX5_SET(query_q_counter_in, in, counter_set_id, counter_set_id); return mlx5_cmd_exec_check_status(mdev, in, sizeof(in), out, out_size); } int mlx5_vport_query_out_of_rx_buffer(struct mlx5_core_dev *mdev, int counter_set_id, u32 *out_of_rx_buffer) { u32 out[MLX5_ST_SZ_DW(query_q_counter_out)]; int err; memset(out, 0, sizeof(out)); err = mlx5_vport_query_q_counter(mdev, counter_set_id, 0, out, sizeof(out)); if (err) return err; *out_of_rx_buffer = MLX5_GET(query_q_counter_out, out, out_of_buffer); return err; } int mlx5_query_nic_vport_mac_address(struct mlx5_core_dev *mdev, u32 vport, u8 *addr) { u32 *out; int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out); u8 *out_addr; int err; out = mlx5_vzalloc(outlen); if (!out) return -ENOMEM; out_addr = MLX5_ADDR_OF(query_nic_vport_context_out, out, nic_vport_context.permanent_address); err = mlx5_query_nic_vport_context(mdev, vport, out, outlen); if (err) goto out; ether_addr_copy(addr, &out_addr[2]); out: kvfree(out); return err; } EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_mac_address); int mlx5_query_nic_vport_system_image_guid(struct mlx5_core_dev *mdev, u64 *system_image_guid) { u32 *out; int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out); int err; out = mlx5_vzalloc(outlen); if (!out) return -ENOMEM; err = mlx5_query_nic_vport_context(mdev, 0, out, outlen); if (err) goto out; *system_image_guid = MLX5_GET64(query_nic_vport_context_out, out, nic_vport_context.system_image_guid); out: kvfree(out); return err; } EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_system_image_guid); int mlx5_query_nic_vport_node_guid(struct mlx5_core_dev *mdev, u64 *node_guid) { u32 *out; int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out); int err; out = mlx5_vzalloc(outlen); if (!out) return -ENOMEM; err = mlx5_query_nic_vport_context(mdev, 0, out, outlen); if (err) goto out; *node_guid = MLX5_GET64(query_nic_vport_context_out, out, nic_vport_context.node_guid); out: kvfree(out); return err; } EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_node_guid); int mlx5_query_nic_vport_port_guid(struct mlx5_core_dev *mdev, u64 *port_guid) { u32 *out; int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out); int err; out = mlx5_vzalloc(outlen); if (!out) return -ENOMEM; err = mlx5_query_nic_vport_context(mdev, 0, out, outlen); if (err) goto out; *port_guid = MLX5_GET64(query_nic_vport_context_out, out, nic_vport_context.port_guid); out: kvfree(out); return err; } EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_port_guid); int mlx5_query_nic_vport_qkey_viol_cntr(struct mlx5_core_dev *mdev, u16 *qkey_viol_cntr) { u32 *out; int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out); int err; out = mlx5_vzalloc(outlen); if (!out) return -ENOMEM; err = mlx5_query_nic_vport_context(mdev, 0, out, outlen); if (err) goto out; *qkey_viol_cntr = MLX5_GET(query_nic_vport_context_out, out, nic_vport_context.qkey_violation_counter); out: kvfree(out); return err; } EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_qkey_viol_cntr); static int mlx5_modify_nic_vport_context(struct mlx5_core_dev *mdev, void *in, int inlen) { u32 out[MLX5_ST_SZ_DW(modify_nic_vport_context_out)]; MLX5_SET(modify_nic_vport_context_in, in, opcode, MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT); memset(out, 0, sizeof(out)); return mlx5_cmd_exec_check_status(mdev, in, inlen, out, sizeof(out)); } static int mlx5_nic_vport_enable_disable_roce(struct mlx5_core_dev *mdev, int enable_disable) { void *in; int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in); int err; in = mlx5_vzalloc(inlen); if (!in) { mlx5_core_warn(mdev, "failed to allocate inbox\n"); return -ENOMEM; } MLX5_SET(modify_nic_vport_context_in, in, field_select.roce_en, 1); MLX5_SET(modify_nic_vport_context_in, in, nic_vport_context.roce_en, enable_disable); err = mlx5_modify_nic_vport_context(mdev, in, inlen); kvfree(in); return err; } int mlx5_set_nic_vport_current_mac(struct mlx5_core_dev *mdev, int vport, bool other_vport, u8 *addr) { void *in; int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in) + MLX5_ST_SZ_BYTES(mac_address_layout); u8 *mac_layout; u8 *mac_ptr; int err; in = mlx5_vzalloc(inlen); if (!in) { mlx5_core_warn(mdev, "failed to allocate inbox\n"); return -ENOMEM; } MLX5_SET(modify_nic_vport_context_in, in, opcode, MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT); MLX5_SET(modify_nic_vport_context_in, in, vport_number, vport); MLX5_SET(modify_nic_vport_context_in, in, other_vport, other_vport); MLX5_SET(modify_nic_vport_context_in, in, field_select.addresses_list, 1); MLX5_SET(modify_nic_vport_context_in, in, nic_vport_context.allowed_list_type, MLX5_NIC_VPORT_LIST_TYPE_UC); MLX5_SET(modify_nic_vport_context_in, in, nic_vport_context.allowed_list_size, 1); mac_layout = (u8 *)MLX5_ADDR_OF(modify_nic_vport_context_in, in, nic_vport_context.current_uc_mac_address); mac_ptr = (u8 *)MLX5_ADDR_OF(mac_address_layout, mac_layout, mac_addr_47_32); ether_addr_copy(mac_ptr, addr); err = mlx5_modify_nic_vport_context(mdev, in, inlen); kvfree(in); return err; } EXPORT_SYMBOL_GPL(mlx5_set_nic_vport_current_mac); int mlx5_set_nic_vport_vlan_list(struct mlx5_core_dev *dev, u32 vport, u16 *vlan_list, int list_len) { void *in, *ctx; int i, err; int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in) + MLX5_ST_SZ_BYTES(vlan_layout) * (int)list_len; int max_list_size = 1 << MLX5_CAP_GEN_MAX(dev, log_max_vlan_list); if (list_len > max_list_size) { mlx5_core_warn(dev, "Requested list size (%d) > (%d) max_list_size\n", list_len, max_list_size); return -ENOSPC; } in = mlx5_vzalloc(inlen); if (!in) { mlx5_core_warn(dev, "failed to allocate inbox\n"); return -ENOMEM; } MLX5_SET(modify_nic_vport_context_in, in, vport_number, vport); if (vport) MLX5_SET(modify_nic_vport_context_in, in, other_vport, 1); MLX5_SET(modify_nic_vport_context_in, in, field_select.addresses_list, 1); ctx = MLX5_ADDR_OF(modify_nic_vport_context_in, in, nic_vport_context); MLX5_SET(nic_vport_context, ctx, allowed_list_type, MLX5_NIC_VPORT_LIST_TYPE_VLAN); MLX5_SET(nic_vport_context, ctx, allowed_list_size, list_len); for (i = 0; i < list_len; i++) { u8 *vlan_lout = MLX5_ADDR_OF(nic_vport_context, ctx, current_uc_mac_address[i]); MLX5_SET(vlan_layout, vlan_lout, vlan, vlan_list[i]); } err = mlx5_modify_nic_vport_context(dev, in, inlen); kvfree(in); return err; } EXPORT_SYMBOL_GPL(mlx5_set_nic_vport_vlan_list); int mlx5_set_nic_vport_mc_list(struct mlx5_core_dev *mdev, int vport, u64 *addr_list, size_t addr_list_len) { void *in, *ctx; int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in) + MLX5_ST_SZ_BYTES(mac_address_layout) * (int)addr_list_len; int err; size_t i; int max_list_sz = 1 << MLX5_CAP_GEN_MAX(mdev, log_max_current_mc_list); if ((int)addr_list_len > max_list_sz) { mlx5_core_warn(mdev, "Requested list size (%d) > (%d) max_list_size\n", (int)addr_list_len, max_list_sz); return -ENOSPC; } in = mlx5_vzalloc(inlen); if (!in) { mlx5_core_warn(mdev, "failed to allocate inbox\n"); return -ENOMEM; } MLX5_SET(modify_nic_vport_context_in, in, vport_number, vport); if (vport) MLX5_SET(modify_nic_vport_context_in, in, other_vport, 1); MLX5_SET(modify_nic_vport_context_in, in, field_select.addresses_list, 1); ctx = MLX5_ADDR_OF(modify_nic_vport_context_in, in, nic_vport_context); MLX5_SET(nic_vport_context, ctx, allowed_list_type, MLX5_NIC_VPORT_LIST_TYPE_MC); MLX5_SET(nic_vport_context, ctx, allowed_list_size, addr_list_len); for (i = 0; i < addr_list_len; i++) { u8 *mac_lout = (u8 *)MLX5_ADDR_OF(nic_vport_context, ctx, current_uc_mac_address[i]); u8 *mac_ptr = (u8 *)MLX5_ADDR_OF(mac_address_layout, mac_lout, mac_addr_47_32); ether_addr_copy(mac_ptr, (u8 *)&addr_list[i]); } err = mlx5_modify_nic_vport_context(mdev, in, inlen); kvfree(in); return err; } EXPORT_SYMBOL_GPL(mlx5_set_nic_vport_mc_list); int mlx5_set_nic_vport_promisc(struct mlx5_core_dev *mdev, int vport, bool promisc_mc, bool promisc_uc, bool promisc_all) { u8 in[MLX5_ST_SZ_BYTES(modify_nic_vport_context_in)]; u8 *ctx = MLX5_ADDR_OF(modify_nic_vport_context_in, in, nic_vport_context); memset(in, 0, MLX5_ST_SZ_BYTES(modify_nic_vport_context_in)); MLX5_SET(modify_nic_vport_context_in, in, vport_number, vport); if (vport) MLX5_SET(modify_nic_vport_context_in, in, other_vport, 1); MLX5_SET(modify_nic_vport_context_in, in, field_select.promisc, 1); if (promisc_mc) MLX5_SET(nic_vport_context, ctx, promisc_mc, 1); if (promisc_uc) MLX5_SET(nic_vport_context, ctx, promisc_uc, 1); if (promisc_all) MLX5_SET(nic_vport_context, ctx, promisc_all, 1); return mlx5_modify_nic_vport_context(mdev, in, sizeof(in)); } EXPORT_SYMBOL_GPL(mlx5_set_nic_vport_promisc); int mlx5_query_nic_vport_mac_list(struct mlx5_core_dev *dev, u32 vport, enum mlx5_list_type list_type, u8 addr_list[][ETH_ALEN], int *list_size) { u32 in[MLX5_ST_SZ_DW(query_nic_vport_context_in)]; void *nic_vport_ctx; int max_list_size; int req_list_size; u8 *mac_addr; int out_sz; void *out; int err; int i; req_list_size = *list_size; max_list_size = (list_type == MLX5_NIC_VPORT_LIST_TYPE_UC) ? 1 << MLX5_CAP_GEN_MAX(dev, log_max_current_uc_list) : 1 << MLX5_CAP_GEN_MAX(dev, log_max_current_mc_list); if (req_list_size > max_list_size) { mlx5_core_warn(dev, "Requested list size (%d) > (%d) max_list_size\n", req_list_size, max_list_size); req_list_size = max_list_size; } out_sz = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in) + req_list_size * MLX5_ST_SZ_BYTES(mac_address_layout); memset(in, 0, sizeof(in)); out = kzalloc(out_sz, GFP_KERNEL); if (!out) return -ENOMEM; MLX5_SET(query_nic_vport_context_in, in, opcode, MLX5_CMD_OP_QUERY_NIC_VPORT_CONTEXT); MLX5_SET(query_nic_vport_context_in, in, allowed_list_type, list_type); MLX5_SET(query_nic_vport_context_in, in, vport_number, vport); if (vport) MLX5_SET(query_nic_vport_context_in, in, other_vport, 1); err = mlx5_cmd_exec_check_status(dev, in, sizeof(in), out, out_sz); if (err) goto out; nic_vport_ctx = MLX5_ADDR_OF(query_nic_vport_context_out, out, nic_vport_context); req_list_size = MLX5_GET(nic_vport_context, nic_vport_ctx, allowed_list_size); *list_size = req_list_size; for (i = 0; i < req_list_size; i++) { mac_addr = MLX5_ADDR_OF(nic_vport_context, nic_vport_ctx, current_uc_mac_address[i]) + 2; ether_addr_copy(addr_list[i], mac_addr); } out: kfree(out); return err; } EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_mac_list); int mlx5_modify_nic_vport_mac_list(struct mlx5_core_dev *dev, enum mlx5_list_type list_type, u8 addr_list[][ETH_ALEN], int list_size) { u32 out[MLX5_ST_SZ_DW(modify_nic_vport_context_out)]; void *nic_vport_ctx; int max_list_size; int in_sz; void *in; int err; int i; max_list_size = list_type == MLX5_NIC_VPORT_LIST_TYPE_UC ? 1 << MLX5_CAP_GEN(dev, log_max_current_uc_list) : 1 << MLX5_CAP_GEN(dev, log_max_current_mc_list); if (list_size > max_list_size) return -ENOSPC; in_sz = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in) + list_size * MLX5_ST_SZ_BYTES(mac_address_layout); memset(out, 0, sizeof(out)); in = kzalloc(in_sz, GFP_KERNEL); if (!in) return -ENOMEM; MLX5_SET(modify_nic_vport_context_in, in, opcode, MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT); MLX5_SET(modify_nic_vport_context_in, in, field_select.addresses_list, 1); nic_vport_ctx = MLX5_ADDR_OF(modify_nic_vport_context_in, in, nic_vport_context); MLX5_SET(nic_vport_context, nic_vport_ctx, allowed_list_type, list_type); MLX5_SET(nic_vport_context, nic_vport_ctx, allowed_list_size, list_size); for (i = 0; i < list_size; i++) { u8 *curr_mac = MLX5_ADDR_OF(nic_vport_context, nic_vport_ctx, current_uc_mac_address[i]) + 2; ether_addr_copy(curr_mac, addr_list[i]); } err = mlx5_cmd_exec_check_status(dev, in, in_sz, out, sizeof(out)); kfree(in); return err; } EXPORT_SYMBOL_GPL(mlx5_modify_nic_vport_mac_list); int mlx5_query_nic_vport_vlan_list(struct mlx5_core_dev *dev, u32 vport, u16 *vlan_list, int *list_size) { u32 in[MLX5_ST_SZ_DW(query_nic_vport_context_in)]; void *nic_vport_ctx; int max_list_size; int req_list_size; int out_sz; void *out; void *vlan_addr; int err; int i; req_list_size = *list_size; max_list_size = 1 << MLX5_CAP_GEN_MAX(dev, log_max_vlan_list); if (req_list_size > max_list_size) { mlx5_core_warn(dev, "Requested list size (%d) > (%d) max_list_size\n", req_list_size, max_list_size); req_list_size = max_list_size; } out_sz = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in) + req_list_size * MLX5_ST_SZ_BYTES(vlan_layout); memset(in, 0, sizeof(in)); out = kzalloc(out_sz, GFP_KERNEL); if (!out) return -ENOMEM; MLX5_SET(query_nic_vport_context_in, in, opcode, MLX5_CMD_OP_QUERY_NIC_VPORT_CONTEXT); MLX5_SET(query_nic_vport_context_in, in, allowed_list_type, MLX5_NIC_VPORT_CONTEXT_ALLOWED_LIST_TYPE_VLAN_LIST); MLX5_SET(query_nic_vport_context_in, in, vport_number, vport); if (vport) MLX5_SET(query_nic_vport_context_in, in, other_vport, 1); err = mlx5_cmd_exec_check_status(dev, in, sizeof(in), out, out_sz); if (err) goto out; nic_vport_ctx = MLX5_ADDR_OF(query_nic_vport_context_out, out, nic_vport_context); req_list_size = MLX5_GET(nic_vport_context, nic_vport_ctx, allowed_list_size); *list_size = req_list_size; for (i = 0; i < req_list_size; i++) { vlan_addr = MLX5_ADDR_OF(nic_vport_context, nic_vport_ctx, current_uc_mac_address[i]); vlan_list[i] = MLX5_GET(vlan_layout, vlan_addr, vlan); } out: kfree(out); return err; } EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_vlan_list); int mlx5_modify_nic_vport_vlans(struct mlx5_core_dev *dev, u16 vlans[], int list_size) { u32 out[MLX5_ST_SZ_DW(modify_nic_vport_context_out)]; void *nic_vport_ctx; int max_list_size; int in_sz; void *in; int err; int i; max_list_size = 1 << MLX5_CAP_GEN(dev, log_max_vlan_list); if (list_size > max_list_size) return -ENOSPC; in_sz = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in) + list_size * MLX5_ST_SZ_BYTES(vlan_layout); memset(out, 0, sizeof(out)); in = kzalloc(in_sz, GFP_KERNEL); if (!in) return -ENOMEM; MLX5_SET(modify_nic_vport_context_in, in, opcode, MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT); MLX5_SET(modify_nic_vport_context_in, in, field_select.addresses_list, 1); nic_vport_ctx = MLX5_ADDR_OF(modify_nic_vport_context_in, in, nic_vport_context); MLX5_SET(nic_vport_context, nic_vport_ctx, allowed_list_type, MLX5_NIC_VPORT_LIST_TYPE_VLAN); MLX5_SET(nic_vport_context, nic_vport_ctx, allowed_list_size, list_size); for (i = 0; i < list_size; i++) { void *vlan_addr = MLX5_ADDR_OF(nic_vport_context, nic_vport_ctx, current_uc_mac_address[i]); MLX5_SET(vlan_layout, vlan_addr, vlan, vlans[i]); } err = mlx5_cmd_exec_check_status(dev, in, in_sz, out, sizeof(out)); kfree(in); return err; } EXPORT_SYMBOL_GPL(mlx5_modify_nic_vport_vlans); int mlx5_set_nic_vport_permanent_mac(struct mlx5_core_dev *mdev, int vport, u8 *addr) { void *in; int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in); u8 *mac_ptr; int err; in = mlx5_vzalloc(inlen); if (!in) { mlx5_core_warn(mdev, "failed to allocate inbox\n"); return -ENOMEM; } MLX5_SET(modify_nic_vport_context_in, in, opcode, MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT); MLX5_SET(modify_nic_vport_context_in, in, vport_number, vport); MLX5_SET(modify_nic_vport_context_in, in, other_vport, 1); MLX5_SET(modify_nic_vport_context_in, in, field_select.permanent_address, 1); mac_ptr = (u8 *)MLX5_ADDR_OF(modify_nic_vport_context_in, in, nic_vport_context.permanent_address.mac_addr_47_32); ether_addr_copy(mac_ptr, addr); err = mlx5_modify_nic_vport_context(mdev, in, inlen); kvfree(in); return err; } EXPORT_SYMBOL_GPL(mlx5_set_nic_vport_permanent_mac); int mlx5_nic_vport_enable_roce(struct mlx5_core_dev *mdev) { return mlx5_nic_vport_enable_disable_roce(mdev, 1); } EXPORT_SYMBOL_GPL(mlx5_nic_vport_enable_roce); int mlx5_nic_vport_disable_roce(struct mlx5_core_dev *mdev) { return mlx5_nic_vport_enable_disable_roce(mdev, 0); } EXPORT_SYMBOL_GPL(mlx5_nic_vport_disable_roce); int mlx5_query_hca_vport_context(struct mlx5_core_dev *mdev, u8 port_num, u8 vport_num, u32 *out, int outlen) { u32 in[MLX5_ST_SZ_DW(query_hca_vport_context_in)]; int is_group_manager; is_group_manager = MLX5_CAP_GEN(mdev, vport_group_manager); memset(in, 0, sizeof(in)); MLX5_SET(query_hca_vport_context_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_VPORT_CONTEXT); if (vport_num) { if (is_group_manager) { MLX5_SET(query_hca_vport_context_in, in, other_vport, 1); MLX5_SET(query_hca_vport_context_in, in, vport_number, vport_num); } else { return -EPERM; } } if (MLX5_CAP_GEN(mdev, num_ports) == 2) MLX5_SET(query_hca_vport_context_in, in, port_num, port_num); return mlx5_cmd_exec_check_status(mdev, in, sizeof(in), out, outlen); } int mlx5_query_hca_vport_system_image_guid(struct mlx5_core_dev *mdev, u64 *system_image_guid) { u32 *out; int outlen = MLX5_ST_SZ_BYTES(query_hca_vport_context_out); int err; out = mlx5_vzalloc(outlen); if (!out) return -ENOMEM; err = mlx5_query_hca_vport_context(mdev, 1, 0, out, outlen); if (err) goto out; *system_image_guid = MLX5_GET64(query_hca_vport_context_out, out, hca_vport_context.system_image_guid); out: kvfree(out); return err; } EXPORT_SYMBOL_GPL(mlx5_query_hca_vport_system_image_guid); int mlx5_query_hca_vport_node_guid(struct mlx5_core_dev *mdev, u64 *node_guid) { u32 *out; int outlen = MLX5_ST_SZ_BYTES(query_hca_vport_context_out); int err; out = mlx5_vzalloc(outlen); if (!out) return -ENOMEM; err = mlx5_query_hca_vport_context(mdev, 1, 0, out, outlen); if (err) goto out; *node_guid = MLX5_GET64(query_hca_vport_context_out, out, hca_vport_context.node_guid); out: kvfree(out); return err; } EXPORT_SYMBOL_GPL(mlx5_query_hca_vport_node_guid); int mlx5_query_hca_vport_gid(struct mlx5_core_dev *dev, u8 port_num, u16 vport_num, u16 gid_index, union ib_gid *gid) { int in_sz = MLX5_ST_SZ_BYTES(query_hca_vport_gid_in); int out_sz = MLX5_ST_SZ_BYTES(query_hca_vport_gid_out); int is_group_manager; void *out = NULL; void *in = NULL; union ib_gid *tmp; int tbsz; int nout; int err; is_group_manager = MLX5_CAP_GEN(dev, vport_group_manager); tbsz = mlx5_get_gid_table_len(MLX5_CAP_GEN(dev, gid_table_size)); if (gid_index > tbsz && gid_index != 0xffff) return -EINVAL; if (gid_index == 0xffff) nout = tbsz; else nout = 1; out_sz += nout * sizeof(*gid); in = mlx5_vzalloc(in_sz); out = mlx5_vzalloc(out_sz); if (!in || !out) { err = -ENOMEM; goto out; } MLX5_SET(query_hca_vport_gid_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_VPORT_GID); if (vport_num) { if (is_group_manager) { MLX5_SET(query_hca_vport_gid_in, in, vport_number, vport_num); MLX5_SET(query_hca_vport_gid_in, in, other_vport, 1); } else { err = -EPERM; goto out; } } MLX5_SET(query_hca_vport_gid_in, in, gid_index, gid_index); if (MLX5_CAP_GEN(dev, num_ports) == 2) MLX5_SET(query_hca_vport_gid_in, in, port_num, port_num); err = mlx5_cmd_exec(dev, in, in_sz, out, out_sz); if (err) goto out; err = mlx5_cmd_status_to_err_v2(out); if (err) goto out; tmp = (union ib_gid *)MLX5_ADDR_OF(query_hca_vport_gid_out, out, gid); gid->global.subnet_prefix = tmp->global.subnet_prefix; gid->global.interface_id = tmp->global.interface_id; out: kvfree(in); kvfree(out); return err; } EXPORT_SYMBOL_GPL(mlx5_query_hca_vport_gid); int mlx5_query_hca_vport_pkey(struct mlx5_core_dev *dev, u8 other_vport, u8 port_num, u16 vf_num, u16 pkey_index, u16 *pkey) { int in_sz = MLX5_ST_SZ_BYTES(query_hca_vport_pkey_in); int out_sz = MLX5_ST_SZ_BYTES(query_hca_vport_pkey_out); int is_group_manager; void *out = NULL; void *in = NULL; void *pkarr; int nout; int tbsz; int err; int i; is_group_manager = MLX5_CAP_GEN(dev, vport_group_manager); tbsz = mlx5_to_sw_pkey_sz(MLX5_CAP_GEN(dev, pkey_table_size)); if (pkey_index > tbsz && pkey_index != 0xffff) return -EINVAL; if (pkey_index == 0xffff) nout = tbsz; else nout = 1; out_sz += nout * MLX5_ST_SZ_BYTES(pkey); in = kzalloc(in_sz, GFP_KERNEL); out = kzalloc(out_sz, GFP_KERNEL); MLX5_SET(query_hca_vport_pkey_in, in, opcode, MLX5_CMD_OP_QUERY_HCA_VPORT_PKEY); if (other_vport) { if (is_group_manager) { MLX5_SET(query_hca_vport_pkey_in, in, vport_number, vf_num); MLX5_SET(query_hca_vport_pkey_in, in, other_vport, 1); } else { err = -EPERM; goto out; } } MLX5_SET(query_hca_vport_pkey_in, in, pkey_index, pkey_index); if (MLX5_CAP_GEN(dev, num_ports) == 2) MLX5_SET(query_hca_vport_pkey_in, in, port_num, port_num); err = mlx5_cmd_exec(dev, in, in_sz, out, out_sz); if (err) goto out; err = mlx5_cmd_status_to_err_v2(out); if (err) goto out; pkarr = MLX5_ADDR_OF(query_hca_vport_pkey_out, out, pkey); for (i = 0; i < nout; i++, pkey++, pkarr += MLX5_ST_SZ_BYTES(pkey)) *pkey = MLX5_GET_PR(pkey, pkarr, pkey); out: kfree(in); kfree(out); return err; } EXPORT_SYMBOL_GPL(mlx5_query_hca_vport_pkey); static int mlx5_modify_eswitch_vport_context(struct mlx5_core_dev *mdev, u16 vport, void *in, int inlen) { u32 out[MLX5_ST_SZ_DW(modify_esw_vport_context_out)]; int err; memset(out, 0, sizeof(out)); MLX5_SET(modify_esw_vport_context_in, in, vport_number, vport); if (vport) MLX5_SET(modify_esw_vport_context_in, in, other_vport, 1); MLX5_SET(modify_esw_vport_context_in, in, opcode, MLX5_CMD_OP_MODIFY_ESW_VPORT_CONTEXT); err = mlx5_cmd_exec_check_status(mdev, in, inlen, out, sizeof(out)); if (err) mlx5_core_warn(mdev, "MLX5_CMD_OP_MODIFY_ESW_VPORT_CONTEXT failed\n"); return err; } int mlx5_set_eswitch_cvlan_info(struct mlx5_core_dev *mdev, u8 vport, u8 insert_mode, u8 strip_mode, u16 vlan, u8 cfi, u8 pcp) { u32 in[MLX5_ST_SZ_DW(modify_esw_vport_context_in)]; memset(in, 0, sizeof(in)); if (insert_mode != MLX5_MODIFY_ESW_VPORT_CONTEXT_CVLAN_INSERT_NONE) { MLX5_SET(modify_esw_vport_context_in, in, esw_vport_context.cvlan_cfi, cfi); MLX5_SET(modify_esw_vport_context_in, in, esw_vport_context.cvlan_pcp, pcp); MLX5_SET(modify_esw_vport_context_in, in, esw_vport_context.cvlan_id, vlan); } MLX5_SET(modify_esw_vport_context_in, in, esw_vport_context.vport_cvlan_insert, insert_mode); MLX5_SET(modify_esw_vport_context_in, in, esw_vport_context.vport_cvlan_strip, strip_mode); MLX5_SET(modify_esw_vport_context_in, in, field_select, MLX5_MODIFY_ESW_VPORT_CONTEXT_FIELD_SELECT_CVLAN_STRIP | MLX5_MODIFY_ESW_VPORT_CONTEXT_FIELD_SELECT_CVLAN_INSERT); return mlx5_modify_eswitch_vport_context(mdev, vport, in, sizeof(in)); } EXPORT_SYMBOL_GPL(mlx5_set_eswitch_cvlan_info); int mlx5_arm_vport_context_events(struct mlx5_core_dev *mdev, u8 vport, u32 events_mask) { u32 *in; u32 inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in); void *nic_vport_ctx; int err; in = mlx5_vzalloc(inlen); if (!in) return -ENOMEM; MLX5_SET(modify_nic_vport_context_in, in, opcode, MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT); MLX5_SET(modify_nic_vport_context_in, in, field_select.change_event, 1); MLX5_SET(modify_nic_vport_context_in, in, vport_number, vport); if (vport) MLX5_SET(modify_nic_vport_context_in, in, other_vport, 1); nic_vport_ctx = MLX5_ADDR_OF(modify_nic_vport_context_in, in, nic_vport_context); MLX5_SET(nic_vport_context, nic_vport_ctx, arm_change_event, 1); if (events_mask & MLX5_UC_ADDR_CHANGE) MLX5_SET(nic_vport_context, nic_vport_ctx, event_on_uc_address_change, 1); if (events_mask & MLX5_MC_ADDR_CHANGE) MLX5_SET(nic_vport_context, nic_vport_ctx, event_on_mc_address_change, 1); if (events_mask & MLX5_VLAN_CHANGE) MLX5_SET(nic_vport_context, nic_vport_ctx, event_on_vlan_change, 1); if (events_mask & MLX5_PROMISC_CHANGE) MLX5_SET(nic_vport_context, nic_vport_ctx, event_on_promisc_change, 1); if (events_mask & MLX5_MTU_CHANGE) MLX5_SET(nic_vport_context, nic_vport_ctx, event_on_mtu, 1); err = mlx5_modify_nic_vport_context(mdev, in, inlen); kvfree(in); return err; } EXPORT_SYMBOL_GPL(mlx5_arm_vport_context_events); int mlx5_query_vport_promisc(struct mlx5_core_dev *mdev, u32 vport, u8 *promisc_uc, u8 *promisc_mc, u8 *promisc_all) { u32 *out; int outlen = MLX5_ST_SZ_BYTES(query_nic_vport_context_out); int err; out = kzalloc(outlen, GFP_KERNEL); if (!out) return -ENOMEM; err = mlx5_query_nic_vport_context(mdev, vport, out, outlen); if (err) goto out; *promisc_uc = MLX5_GET(query_nic_vport_context_out, out, nic_vport_context.promisc_uc); *promisc_mc = MLX5_GET(query_nic_vport_context_out, out, nic_vport_context.promisc_mc); *promisc_all = MLX5_GET(query_nic_vport_context_out, out, nic_vport_context.promisc_all); out: kfree(out); return err; } EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_promisc); int mlx5_modify_nic_vport_promisc(struct mlx5_core_dev *mdev, int promisc_uc, int promisc_mc, int promisc_all) { void *in; int inlen = MLX5_ST_SZ_BYTES(modify_nic_vport_context_in); int err; in = mlx5_vzalloc(inlen); if (!in) { mlx5_core_err(mdev, "failed to allocate inbox\n"); return -ENOMEM; } MLX5_SET(modify_nic_vport_context_in, in, field_select.promisc, 1); MLX5_SET(modify_nic_vport_context_in, in, nic_vport_context.promisc_uc, promisc_uc); MLX5_SET(modify_nic_vport_context_in, in, nic_vport_context.promisc_mc, promisc_mc); MLX5_SET(modify_nic_vport_context_in, in, nic_vport_context.promisc_all, promisc_all); err = mlx5_modify_nic_vport_context(mdev, in, inlen); kvfree(in); return err; } EXPORT_SYMBOL_GPL(mlx5_modify_nic_vport_promisc); int mlx5_query_vport_counter(struct mlx5_core_dev *dev, u8 port_num, u16 vport_num, void *out, int out_size) { int in_sz = MLX5_ST_SZ_BYTES(query_vport_counter_in); int is_group_manager; void *in; int err; is_group_manager = MLX5_CAP_GEN(dev, vport_group_manager); in = mlx5_vzalloc(in_sz); if (!in) return -ENOMEM; MLX5_SET(query_vport_counter_in, in, opcode, MLX5_CMD_OP_QUERY_VPORT_COUNTER); if (vport_num) { if (is_group_manager) { MLX5_SET(query_vport_counter_in, in, other_vport, 1); MLX5_SET(query_vport_counter_in, in, vport_number, vport_num); } else { err = -EPERM; goto ex; } } if (MLX5_CAP_GEN(dev, num_ports) == 2) MLX5_SET(query_vport_counter_in, in, port_num, port_num); err = mlx5_cmd_exec(dev, in, in_sz, out, out_size); if (err) goto ex; err = mlx5_cmd_status_to_err_v2(out); if (err) goto ex; ex: kvfree(in); return err; } EXPORT_SYMBOL_GPL(mlx5_query_vport_counter); int mlx5_get_vport_counters(struct mlx5_core_dev *dev, u8 port_num, struct mlx5_vport_counters *vc) { int out_sz = MLX5_ST_SZ_BYTES(query_vport_counter_out); void *out; int err; out = mlx5_vzalloc(out_sz); if (!out) return -ENOMEM; err = mlx5_query_vport_counter(dev, port_num, 0, out, out_sz); if (err) goto ex; vc->received_errors.packets = MLX5_GET64(query_vport_counter_out, out, received_errors.packets); vc->received_errors.octets = MLX5_GET64(query_vport_counter_out, out, received_errors.octets); vc->transmit_errors.packets = MLX5_GET64(query_vport_counter_out, out, transmit_errors.packets); vc->transmit_errors.octets = MLX5_GET64(query_vport_counter_out, out, transmit_errors.octets); vc->received_ib_unicast.packets = MLX5_GET64(query_vport_counter_out, out, received_ib_unicast.packets); vc->received_ib_unicast.octets = MLX5_GET64(query_vport_counter_out, out, received_ib_unicast.octets); vc->transmitted_ib_unicast.packets = MLX5_GET64(query_vport_counter_out, out, transmitted_ib_unicast.packets); vc->transmitted_ib_unicast.octets = MLX5_GET64(query_vport_counter_out, out, transmitted_ib_unicast.octets); vc->received_ib_multicast.packets = MLX5_GET64(query_vport_counter_out, out, received_ib_multicast.packets); vc->received_ib_multicast.octets = MLX5_GET64(query_vport_counter_out, out, received_ib_multicast.octets); vc->transmitted_ib_multicast.packets = MLX5_GET64(query_vport_counter_out, out, transmitted_ib_multicast.packets); vc->transmitted_ib_multicast.octets = MLX5_GET64(query_vport_counter_out, out, transmitted_ib_multicast.octets); vc->received_eth_broadcast.packets = MLX5_GET64(query_vport_counter_out, out, received_eth_broadcast.packets); vc->received_eth_broadcast.octets = MLX5_GET64(query_vport_counter_out, out, received_eth_broadcast.octets); vc->transmitted_eth_broadcast.packets = MLX5_GET64(query_vport_counter_out, out, transmitted_eth_broadcast.packets); vc->transmitted_eth_broadcast.octets = MLX5_GET64(query_vport_counter_out, out, transmitted_eth_broadcast.octets); vc->received_eth_unicast.octets = MLX5_GET64(query_vport_counter_out, out, received_eth_unicast.octets); vc->received_eth_unicast.packets = MLX5_GET64(query_vport_counter_out, out, received_eth_unicast.packets); vc->transmitted_eth_unicast.octets = MLX5_GET64(query_vport_counter_out, out, transmitted_eth_unicast.octets); vc->transmitted_eth_unicast.packets = MLX5_GET64(query_vport_counter_out, out, transmitted_eth_unicast.packets); vc->received_eth_multicast.octets = MLX5_GET64(query_vport_counter_out, out, received_eth_multicast.octets); vc->received_eth_multicast.packets = MLX5_GET64(query_vport_counter_out, out, received_eth_multicast.packets); vc->transmitted_eth_multicast.octets = MLX5_GET64(query_vport_counter_out, out, transmitted_eth_multicast.octets); vc->transmitted_eth_multicast.packets = MLX5_GET64(query_vport_counter_out, out, transmitted_eth_multicast.packets); ex: kvfree(out); return err; } Index: projects/vnet/sys/geom/geom_disk.c =================================================================== --- projects/vnet/sys/geom/geom_disk.c (revision 302157) +++ projects/vnet/sys/geom/geom_disk.c (revision 302158) @@ -1,1008 +1,1003 @@ /*- * Copyright (c) 2002 Poul-Henning Kamp * Copyright (c) 2002 Networks Associates Technology, Inc. * All rights reserved. * * This software was developed for the FreeBSD Project by Poul-Henning Kamp * and NAI Labs, the Security Research Division of Network Associates, Inc. * under DARPA/SPAWAR contract N66001-01-C-8035 ("CBOSS"), as part of the * DARPA CHATS research program. * * 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 names of the authors may not be used to endorse or promote * products derived from this software without specific prior written * permission. * * 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_geom.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include struct g_disk_softc { struct mtx done_mtx; struct disk *dp; struct sysctl_ctx_list sysctl_ctx; struct sysctl_oid *sysctl_tree; char led[64]; uint32_t state; struct mtx start_mtx; }; static g_access_t g_disk_access; static g_start_t g_disk_start; static g_ioctl_t g_disk_ioctl; static g_dumpconf_t g_disk_dumpconf; static g_provgone_t g_disk_providergone; static struct g_class g_disk_class = { .name = G_DISK_CLASS_NAME, .version = G_VERSION, .start = g_disk_start, .access = g_disk_access, .ioctl = g_disk_ioctl, .providergone = g_disk_providergone, .dumpconf = g_disk_dumpconf, }; SYSCTL_DECL(_kern_geom); static SYSCTL_NODE(_kern_geom, OID_AUTO, disk, CTLFLAG_RW, 0, "GEOM_DISK stuff"); DECLARE_GEOM_CLASS(g_disk_class, g_disk); static int g_disk_access(struct g_provider *pp, int r, int w, int e) { struct disk *dp; struct g_disk_softc *sc; int error; g_trace(G_T_ACCESS, "g_disk_access(%s, %d, %d, %d)", pp->name, r, w, e); g_topology_assert(); sc = pp->private; if (sc == NULL || (dp = sc->dp) == NULL || dp->d_destroyed) { /* * Allow decreasing access count even if disk is not * available anymore. */ if (r <= 0 && w <= 0 && e <= 0) return (0); return (ENXIO); } r += pp->acr; w += pp->acw; e += pp->ace; error = 0; if ((pp->acr + pp->acw + pp->ace) == 0 && (r + w + e) > 0) { if (dp->d_open != NULL) { error = dp->d_open(dp); if (bootverbose && error != 0) printf("Opened disk %s -> %d\n", pp->name, error); if (error != 0) return (error); } pp->mediasize = dp->d_mediasize; pp->sectorsize = dp->d_sectorsize; if (dp->d_maxsize == 0) { printf("WARNING: Disk drive %s%d has no d_maxsize\n", dp->d_name, dp->d_unit); dp->d_maxsize = DFLTPHYS; } if (dp->d_delmaxsize == 0) { if (bootverbose && dp->d_flags & DISKFLAG_CANDELETE) { printf("WARNING: Disk drive %s%d has no " "d_delmaxsize\n", dp->d_name, dp->d_unit); } dp->d_delmaxsize = dp->d_maxsize; } pp->stripeoffset = dp->d_stripeoffset; pp->stripesize = dp->d_stripesize; dp->d_flags |= DISKFLAG_OPEN; } else if ((pp->acr + pp->acw + pp->ace) > 0 && (r + w + e) == 0) { if (dp->d_close != NULL) { error = dp->d_close(dp); if (error != 0) printf("Closed disk %s -> %d\n", pp->name, error); } sc->state = G_STATE_ACTIVE; if (sc->led[0] != 0) led_set(sc->led, "0"); dp->d_flags &= ~DISKFLAG_OPEN; } return (error); } static void g_disk_kerneldump(struct bio *bp, struct disk *dp) { struct g_kerneldump *gkd; struct g_geom *gp; gkd = (struct g_kerneldump*)bp->bio_data; gp = bp->bio_to->geom; g_trace(G_T_TOPOLOGY, "g_disk_kerneldump(%s, %jd, %jd)", gp->name, (intmax_t)gkd->offset, (intmax_t)gkd->length); if (dp->d_dump == NULL) { g_io_deliver(bp, ENODEV); return; } gkd->di.dumper = dp->d_dump; gkd->di.priv = dp; gkd->di.blocksize = dp->d_sectorsize; gkd->di.maxiosize = dp->d_maxsize; gkd->di.mediaoffset = gkd->offset; if ((gkd->offset + gkd->length) > dp->d_mediasize) gkd->length = dp->d_mediasize - gkd->offset; gkd->di.mediasize = gkd->length; g_io_deliver(bp, 0); } static void g_disk_setstate(struct bio *bp, struct g_disk_softc *sc) { const char *cmd; memcpy(&sc->state, bp->bio_data, sizeof(sc->state)); if (sc->led[0] != 0) { switch (sc->state) { case G_STATE_FAILED: cmd = "1"; break; case G_STATE_REBUILD: cmd = "f5"; break; case G_STATE_RESYNC: cmd = "f1"; break; default: cmd = "0"; break; } led_set(sc->led, cmd); } g_io_deliver(bp, 0); } static void g_disk_done(struct bio *bp) { struct bintime now; struct bio *bp2; struct g_disk_softc *sc; /* See "notes" for why we need a mutex here */ /* XXX: will witness accept a mix of Giant/unGiant drivers here ? */ bp2 = bp->bio_parent; sc = bp2->bio_to->private; bp->bio_completed = bp->bio_length - bp->bio_resid; binuptime(&now); mtx_lock(&sc->done_mtx); if (bp2->bio_error == 0) bp2->bio_error = bp->bio_error; bp2->bio_completed += bp->bio_completed; switch (bp->bio_cmd) { case BIO_ZONE: bcopy(&bp->bio_zone, &bp2->bio_zone, sizeof(bp->bio_zone)); /*FALLTHROUGH*/ case BIO_READ: case BIO_WRITE: case BIO_DELETE: case BIO_FLUSH: devstat_end_transaction_bio_bt(sc->dp->d_devstat, bp, &now); break; default: break; } bp2->bio_inbed++; if (bp2->bio_children == bp2->bio_inbed) { mtx_unlock(&sc->done_mtx); bp2->bio_resid = bp2->bio_bcount - bp2->bio_completed; g_io_deliver(bp2, bp2->bio_error); } else mtx_unlock(&sc->done_mtx); g_destroy_bio(bp); } static int g_disk_ioctl(struct g_provider *pp, u_long cmd, void * data, int fflag, struct thread *td) { struct disk *dp; struct g_disk_softc *sc; int error; sc = pp->private; dp = sc->dp; if (dp->d_ioctl == NULL) return (ENOIOCTL); error = dp->d_ioctl(dp, cmd, data, fflag, td); return (error); } static off_t g_disk_maxsize(struct disk *dp, struct bio *bp) { if (bp->bio_cmd == BIO_DELETE) return (dp->d_delmaxsize); return (dp->d_maxsize); } static int g_disk_maxsegs(struct disk *dp, struct bio *bp) { return ((g_disk_maxsize(dp, bp) / PAGE_SIZE) + 1); } static void g_disk_advance(struct disk *dp, struct bio *bp, off_t off) { bp->bio_offset += off; bp->bio_length -= off; if ((bp->bio_flags & BIO_VLIST) != 0) { bus_dma_segment_t *seg, *end; seg = (bus_dma_segment_t *)bp->bio_data; end = (bus_dma_segment_t *)bp->bio_data + bp->bio_ma_n; off += bp->bio_ma_offset; while (off >= seg->ds_len) { KASSERT((seg != end), ("vlist request runs off the end")); off -= seg->ds_len; seg++; } bp->bio_ma_offset = off; bp->bio_ma_n = end - seg; bp->bio_data = (void *)seg; } else if ((bp->bio_flags & BIO_UNMAPPED) != 0) { bp->bio_ma += off / PAGE_SIZE; bp->bio_ma_offset += off; bp->bio_ma_offset %= PAGE_SIZE; bp->bio_ma_n -= off / PAGE_SIZE; } else { bp->bio_data += off; } } static void g_disk_seg_limit(bus_dma_segment_t *seg, off_t *poffset, off_t *plength, int *ppages) { uintptr_t seg_page_base; uintptr_t seg_page_end; off_t offset; off_t length; int seg_pages; offset = *poffset; length = *plength; if (length > seg->ds_len - offset) length = seg->ds_len - offset; seg_page_base = trunc_page(seg->ds_addr + offset); seg_page_end = round_page(seg->ds_addr + offset + length); seg_pages = (seg_page_end - seg_page_base) >> PAGE_SHIFT; if (seg_pages > *ppages) { seg_pages = *ppages; length = (seg_page_base + (seg_pages << PAGE_SHIFT)) - (seg->ds_addr + offset); } *poffset = 0; *plength -= length; *ppages -= seg_pages; } static off_t g_disk_vlist_limit(struct disk *dp, struct bio *bp, bus_dma_segment_t **pendseg) { bus_dma_segment_t *seg, *end; off_t residual; off_t offset; int pages; seg = (bus_dma_segment_t *)bp->bio_data; end = (bus_dma_segment_t *)bp->bio_data + bp->bio_ma_n; residual = bp->bio_length; offset = bp->bio_ma_offset; pages = g_disk_maxsegs(dp, bp); while (residual != 0 && pages != 0) { KASSERT((seg != end), ("vlist limit runs off the end")); g_disk_seg_limit(seg, &offset, &residual, &pages); seg++; } if (pendseg != NULL) *pendseg = seg; return (residual); } static bool g_disk_limit(struct disk *dp, struct bio *bp) { bool limited = false; off_t maxsz; maxsz = g_disk_maxsize(dp, bp); /* * XXX: If we have a stripesize we should really use it here. * Care should be taken in the delete case if this is done * as deletes can be very sensitive to size given how they * are processed. */ if (bp->bio_length > maxsz) { bp->bio_length = maxsz; limited = true; } if ((bp->bio_flags & BIO_VLIST) != 0) { bus_dma_segment_t *firstseg, *endseg; off_t residual; firstseg = (bus_dma_segment_t*)bp->bio_data; residual = g_disk_vlist_limit(dp, bp, &endseg); if (residual != 0) { bp->bio_ma_n = endseg - firstseg; bp->bio_length -= residual; limited = true; } } else if ((bp->bio_flags & BIO_UNMAPPED) != 0) { bp->bio_ma_n = howmany(bp->bio_ma_offset + bp->bio_length, PAGE_SIZE); } return (limited); } static void g_disk_start(struct bio *bp) { struct bio *bp2, *bp3; struct disk *dp; struct g_disk_softc *sc; int error; off_t off; sc = bp->bio_to->private; if (sc == NULL || (dp = sc->dp) == NULL || dp->d_destroyed) { g_io_deliver(bp, ENXIO); return; } error = EJUSTRETURN; switch(bp->bio_cmd) { case BIO_DELETE: if (!(dp->d_flags & DISKFLAG_CANDELETE)) { error = EOPNOTSUPP; break; } /* fall-through */ case BIO_READ: case BIO_WRITE: KASSERT((dp->d_flags & DISKFLAG_UNMAPPED_BIO) != 0 || (bp->bio_flags & BIO_UNMAPPED) == 0, ("unmapped bio not supported by disk %s", dp->d_name)); off = 0; bp3 = NULL; bp2 = g_clone_bio(bp); if (bp2 == NULL) { error = ENOMEM; break; } for (;;) { if (g_disk_limit(dp, bp2)) { off += bp2->bio_length; /* * To avoid a race, we need to grab the next bio * before we schedule this one. See "notes". */ bp3 = g_clone_bio(bp); if (bp3 == NULL) bp->bio_error = ENOMEM; } bp2->bio_done = g_disk_done; bp2->bio_pblkno = bp2->bio_offset / dp->d_sectorsize; bp2->bio_bcount = bp2->bio_length; bp2->bio_disk = dp; mtx_lock(&sc->start_mtx); devstat_start_transaction_bio(dp->d_devstat, bp2); mtx_unlock(&sc->start_mtx); dp->d_strategy(bp2); if (bp3 == NULL) break; bp2 = bp3; bp3 = NULL; g_disk_advance(dp, bp2, off); } break; case BIO_GETATTR: /* Give the driver a chance to override */ if (dp->d_getattr != NULL) { if (bp->bio_disk == NULL) bp->bio_disk = dp; error = dp->d_getattr(bp); if (error != -1) break; error = EJUSTRETURN; } if (g_handleattr_int(bp, "GEOM::candelete", (dp->d_flags & DISKFLAG_CANDELETE) != 0)) break; else if (g_handleattr_int(bp, "GEOM::fwsectors", dp->d_fwsectors)) break; else if (g_handleattr_int(bp, "GEOM::fwheads", dp->d_fwheads)) break; else if (g_handleattr_off_t(bp, "GEOM::frontstuff", 0)) break; else if (g_handleattr_str(bp, "GEOM::ident", dp->d_ident)) break; else if (g_handleattr_uint16_t(bp, "GEOM::hba_vendor", dp->d_hba_vendor)) break; else if (g_handleattr_uint16_t(bp, "GEOM::hba_device", dp->d_hba_device)) break; else if (g_handleattr_uint16_t(bp, "GEOM::hba_subvendor", dp->d_hba_subvendor)) break; else if (g_handleattr_uint16_t(bp, "GEOM::hba_subdevice", dp->d_hba_subdevice)) break; else if (!strcmp(bp->bio_attribute, "GEOM::kerneldump")) g_disk_kerneldump(bp, dp); else if (!strcmp(bp->bio_attribute, "GEOM::setstate")) g_disk_setstate(bp, sc); else if (g_handleattr_uint16_t(bp, "GEOM::rotation_rate", dp->d_rotation_rate)) break; else error = ENOIOCTL; break; case BIO_FLUSH: g_trace(G_T_BIO, "g_disk_flushcache(%s)", bp->bio_to->name); if (!(dp->d_flags & DISKFLAG_CANFLUSHCACHE)) { error = EOPNOTSUPP; break; } /*FALLTHROUGH*/ case BIO_ZONE: if (bp->bio_cmd == BIO_ZONE) { if (!(dp->d_flags & DISKFLAG_CANZONE)) { error = EOPNOTSUPP; break; } g_trace(G_T_BIO, "g_disk_zone(%s)", bp->bio_to->name); } bp2 = g_clone_bio(bp); if (bp2 == NULL) { g_io_deliver(bp, ENOMEM); return; } bp2->bio_done = g_disk_done; bp2->bio_disk = dp; mtx_lock(&sc->start_mtx); devstat_start_transaction_bio(dp->d_devstat, bp2); mtx_unlock(&sc->start_mtx); dp->d_strategy(bp2); break; default: error = EOPNOTSUPP; break; } if (error != EJUSTRETURN) g_io_deliver(bp, error); return; } static void g_disk_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp) { struct bio *bp; struct disk *dp; struct g_disk_softc *sc; char *buf; int res = 0; sc = gp->softc; if (sc == NULL || (dp = sc->dp) == NULL) return; if (indent == NULL) { sbuf_printf(sb, " hd %u", dp->d_fwheads); sbuf_printf(sb, " sc %u", dp->d_fwsectors); return; } if (pp != NULL) { sbuf_printf(sb, "%s%u\n", indent, dp->d_fwheads); sbuf_printf(sb, "%s%u\n", indent, dp->d_fwsectors); /* * "rotationrate" is a little complicated, because the value * returned by the drive might not be the RPM; 0 and 1 are * special cases, and there's also a valid range. */ sbuf_printf(sb, "%s", indent); if (dp->d_rotation_rate == 0) /* Old drives don't */ sbuf_printf(sb, "unknown"); /* report RPM. */ else if (dp->d_rotation_rate == 1) /* Since 0 is used */ sbuf_printf(sb, "0"); /* above, SSDs use 1. */ else if ((dp->d_rotation_rate >= 0x041) && (dp->d_rotation_rate <= 0xfffe)) sbuf_printf(sb, "%u", dp->d_rotation_rate); else sbuf_printf(sb, "invalid"); sbuf_printf(sb, "\n"); if (dp->d_getattr != NULL) { buf = g_malloc(DISK_IDENT_SIZE, M_WAITOK); bp = g_alloc_bio(); bp->bio_disk = dp; bp->bio_attribute = "GEOM::ident"; bp->bio_length = DISK_IDENT_SIZE; bp->bio_data = buf; res = dp->d_getattr(bp); sbuf_printf(sb, "%s", indent); g_conf_printf_escaped(sb, "%s", res == 0 ? buf: dp->d_ident); sbuf_printf(sb, "\n"); bp->bio_attribute = "GEOM::lunid"; bp->bio_length = DISK_IDENT_SIZE; bp->bio_data = buf; if (dp->d_getattr(bp) == 0) { sbuf_printf(sb, "%s", indent); g_conf_printf_escaped(sb, "%s", buf); sbuf_printf(sb, "\n"); } bp->bio_attribute = "GEOM::lunname"; bp->bio_length = DISK_IDENT_SIZE; bp->bio_data = buf; if (dp->d_getattr(bp) == 0) { sbuf_printf(sb, "%s", indent); g_conf_printf_escaped(sb, "%s", buf); sbuf_printf(sb, "\n"); } g_destroy_bio(bp); g_free(buf); } else { sbuf_printf(sb, "%s", indent); g_conf_printf_escaped(sb, "%s", dp->d_ident); sbuf_printf(sb, "\n"); } sbuf_printf(sb, "%s", indent); g_conf_printf_escaped(sb, "%s", dp->d_descr); sbuf_printf(sb, "\n"); } } static void g_disk_resize(void *ptr, int flag) { struct disk *dp; struct g_geom *gp; struct g_provider *pp; if (flag == EV_CANCEL) return; g_topology_assert(); dp = ptr; gp = dp->d_geom; if (dp->d_destroyed || gp == NULL) return; LIST_FOREACH(pp, &gp->provider, provider) { if (pp->sectorsize != 0 && pp->sectorsize != dp->d_sectorsize) g_wither_provider(pp, ENXIO); else g_resize_provider(pp, dp->d_mediasize); } } static void g_disk_create(void *arg, int flag) { struct g_geom *gp; struct g_provider *pp; struct disk *dp; struct g_disk_softc *sc; char tmpstr[80]; if (flag == EV_CANCEL) return; g_topology_assert(); dp = arg; - mtx_lock(&dp->d_mtx); + mtx_pool_lock(mtxpool_sleep, dp); dp->d_init_level = DISK_INIT_START; /* * If the disk has already gone away, we can just stop here and * call the user's callback to tell him we've cleaned things up. */ if (dp->d_goneflag != 0) { - mtx_unlock(&dp->d_mtx); + mtx_pool_unlock(mtxpool_sleep, dp); if (dp->d_gone != NULL) dp->d_gone(dp); return; } - mtx_unlock(&dp->d_mtx); + mtx_pool_unlock(mtxpool_sleep, dp); sc = g_malloc(sizeof(*sc), M_WAITOK | M_ZERO); mtx_init(&sc->start_mtx, "g_disk_start", NULL, MTX_DEF); mtx_init(&sc->done_mtx, "g_disk_done", NULL, MTX_DEF); sc->dp = dp; gp = g_new_geomf(&g_disk_class, "%s%d", dp->d_name, dp->d_unit); gp->softc = sc; pp = g_new_providerf(gp, "%s", gp->name); devstat_remove_entry(pp->stat); pp->stat = NULL; dp->d_devstat->id = pp; pp->mediasize = dp->d_mediasize; pp->sectorsize = dp->d_sectorsize; pp->stripeoffset = dp->d_stripeoffset; pp->stripesize = dp->d_stripesize; if ((dp->d_flags & DISKFLAG_UNMAPPED_BIO) != 0) pp->flags |= G_PF_ACCEPT_UNMAPPED; if ((dp->d_flags & DISKFLAG_DIRECT_COMPLETION) != 0) pp->flags |= G_PF_DIRECT_SEND; pp->flags |= G_PF_DIRECT_RECEIVE; if (bootverbose) printf("GEOM: new disk %s\n", gp->name); sysctl_ctx_init(&sc->sysctl_ctx); snprintf(tmpstr, sizeof(tmpstr), "GEOM disk %s", gp->name); sc->sysctl_tree = SYSCTL_ADD_NODE(&sc->sysctl_ctx, SYSCTL_STATIC_CHILDREN(_kern_geom_disk), OID_AUTO, gp->name, CTLFLAG_RD, 0, tmpstr); if (sc->sysctl_tree != NULL) { SYSCTL_ADD_STRING(&sc->sysctl_ctx, SYSCTL_CHILDREN(sc->sysctl_tree), OID_AUTO, "led", CTLFLAG_RWTUN, sc->led, sizeof(sc->led), "LED name"); } pp->private = sc; dp->d_geom = gp; g_error_provider(pp, 0); - mtx_lock(&dp->d_mtx); + mtx_pool_lock(mtxpool_sleep, dp); dp->d_init_level = DISK_INIT_DONE; /* * If the disk has gone away at this stage, start the withering * process for it. */ if (dp->d_goneflag != 0) { - mtx_unlock(&dp->d_mtx); + mtx_pool_unlock(mtxpool_sleep, dp); g_wither_provider(pp, ENXIO); return; } - mtx_unlock(&dp->d_mtx); + mtx_pool_unlock(mtxpool_sleep, dp); } /* * We get this callback after all of the consumers have gone away, and just * before the provider is freed. If the disk driver provided a d_gone * callback, let them know that it is okay to free resources -- they won't * be getting any more accesses from GEOM. */ static void g_disk_providergone(struct g_provider *pp) { struct disk *dp; struct g_disk_softc *sc; sc = (struct g_disk_softc *)pp->private; dp = sc->dp; if (dp != NULL && dp->d_gone != NULL) dp->d_gone(dp); if (sc->sysctl_tree != NULL) { sysctl_ctx_free(&sc->sysctl_ctx); sc->sysctl_tree = NULL; } if (sc->led[0] != 0) { led_set(sc->led, "0"); sc->led[0] = 0; } pp->private = NULL; pp->geom->softc = NULL; mtx_destroy(&sc->done_mtx); mtx_destroy(&sc->start_mtx); g_free(sc); } static void g_disk_destroy(void *ptr, int flag) { struct disk *dp; struct g_geom *gp; struct g_disk_softc *sc; g_topology_assert(); dp = ptr; gp = dp->d_geom; if (gp != NULL) { sc = gp->softc; if (sc != NULL) sc->dp = NULL; dp->d_geom = NULL; g_wither_geom(gp, ENXIO); } - mtx_destroy(&dp->d_mtx); - g_free(dp); } /* * We only allow printable characters in disk ident, * the rest is converted to 'x'. */ static void g_disk_ident_adjust(char *ident, size_t size) { char *p, tmp[4], newid[DISK_IDENT_SIZE]; newid[0] = '\0'; for (p = ident; *p != '\0'; p++) { if (isprint(*p)) { tmp[0] = *p; tmp[1] = '\0'; } else { snprintf(tmp, sizeof(tmp), "x%02hhx", *(unsigned char *)p); } if (strlcat(newid, tmp, sizeof(newid)) >= sizeof(newid)) break; } bzero(ident, size); strlcpy(ident, newid, size); } struct disk * disk_alloc(void) { return (g_malloc(sizeof(struct disk), M_WAITOK | M_ZERO)); } void disk_create(struct disk *dp, int version) { if (version != DISK_VERSION) { printf("WARNING: Attempt to add disk %s%d %s", dp->d_name, dp->d_unit, " using incompatible ABI version of disk(9)\n"); printf("WARNING: Ignoring disk %s%d\n", dp->d_name, dp->d_unit); return; } if (dp->d_flags & DISKFLAG_RESERVED) { printf("WARNING: Attempt to add non-MPSAFE disk %s%d\n", dp->d_name, dp->d_unit); printf("WARNING: Ignoring disk %s%d\n", dp->d_name, dp->d_unit); return; } KASSERT(dp->d_strategy != NULL, ("disk_create need d_strategy")); KASSERT(dp->d_name != NULL, ("disk_create need d_name")); KASSERT(*dp->d_name != 0, ("disk_create need d_name")); KASSERT(strlen(dp->d_name) < SPECNAMELEN - 4, ("disk name too long")); if (dp->d_devstat == NULL) dp->d_devstat = devstat_new_entry(dp->d_name, dp->d_unit, dp->d_sectorsize, DEVSTAT_ALL_SUPPORTED, DEVSTAT_TYPE_DIRECT, DEVSTAT_PRIORITY_MAX); dp->d_geom = NULL; - snprintf(dp->d_mtx_name, sizeof(dp->d_mtx_name), "%s%ddlk", - dp->d_name, dp->d_unit); - mtx_init(&dp->d_mtx, dp->d_mtx_name, NULL, MTX_DEF); dp->d_init_level = DISK_INIT_NONE; g_disk_ident_adjust(dp->d_ident, sizeof(dp->d_ident)); g_post_event(g_disk_create, dp, M_WAITOK, dp, NULL); } void disk_destroy(struct disk *dp) { g_cancel_event(dp); dp->d_destroyed = 1; if (dp->d_devstat != NULL) devstat_remove_entry(dp->d_devstat); g_post_event(g_disk_destroy, dp, M_WAITOK, NULL); } void disk_gone(struct disk *dp) { struct g_geom *gp; struct g_provider *pp; - mtx_lock(&dp->d_mtx); + mtx_pool_lock(mtxpool_sleep, dp); dp->d_goneflag = 1; /* * If we're still in the process of creating this disk (the * g_disk_create() function is still queued, or is in * progress), the init level will not yet be DISK_INIT_DONE. * * If that is the case, g_disk_create() will see d_goneflag * and take care of cleaning things up. * * If the disk has already been created, we default to * withering the provider as usual below. * * If the caller has not set a d_gone() callback, he will * not be any worse off by returning here, because the geom * has not been fully setup in any case. */ if (dp->d_init_level < DISK_INIT_DONE) { - mtx_unlock(&dp->d_mtx); + mtx_pool_unlock(mtxpool_sleep, dp); return; } - mtx_unlock(&dp->d_mtx); + mtx_pool_unlock(mtxpool_sleep, dp); gp = dp->d_geom; if (gp != NULL) { pp = LIST_FIRST(&gp->provider); if (pp != NULL) { KASSERT(LIST_NEXT(pp, provider) == NULL, ("geom %p has more than one provider", gp)); g_wither_provider(pp, ENXIO); } } } void disk_attr_changed(struct disk *dp, const char *attr, int flag) { struct g_geom *gp; struct g_provider *pp; char devnamebuf[128]; gp = dp->d_geom; if (gp != NULL) LIST_FOREACH(pp, &gp->provider, provider) (void)g_attr_changed(pp, attr, flag); snprintf(devnamebuf, sizeof(devnamebuf), "devname=%s%d", dp->d_name, dp->d_unit); devctl_notify("GEOM", "disk", attr, devnamebuf); } void disk_media_changed(struct disk *dp, int flag) { struct g_geom *gp; struct g_provider *pp; gp = dp->d_geom; if (gp != NULL) { pp = LIST_FIRST(&gp->provider); if (pp != NULL) { KASSERT(LIST_NEXT(pp, provider) == NULL, ("geom %p has more than one provider", gp)); g_media_changed(pp, flag); } } } void disk_media_gone(struct disk *dp, int flag) { struct g_geom *gp; struct g_provider *pp; gp = dp->d_geom; if (gp != NULL) { pp = LIST_FIRST(&gp->provider); if (pp != NULL) { KASSERT(LIST_NEXT(pp, provider) == NULL, ("geom %p has more than one provider", gp)); g_media_gone(pp, flag); } } } int disk_resize(struct disk *dp, int flag) { if (dp->d_destroyed || dp->d_geom == NULL) return (0); return (g_post_event(g_disk_resize, dp, flag, NULL)); } static void g_kern_disks(void *p, int flag __unused) { struct sbuf *sb; struct g_geom *gp; char *sp; sb = p; sp = ""; g_topology_assert(); LIST_FOREACH(gp, &g_disk_class.geom, geom) { sbuf_printf(sb, "%s%s", sp, gp->name); sp = " "; } sbuf_finish(sb); } static int sysctl_disks(SYSCTL_HANDLER_ARGS) { int error; struct sbuf *sb; sb = sbuf_new_auto(); g_waitfor_event(g_kern_disks, sb, M_WAITOK, NULL); error = SYSCTL_OUT(req, sbuf_data(sb), sbuf_len(sb) + 1); sbuf_delete(sb); return error; } SYSCTL_PROC(_kern, OID_AUTO, disks, CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, 0, sysctl_disks, "A", "names of available disks"); Index: projects/vnet/sys/geom/geom_disk.h =================================================================== --- projects/vnet/sys/geom/geom_disk.h (revision 302157) +++ projects/vnet/sys/geom/geom_disk.h (revision 302158) @@ -1,142 +1,140 @@ /*- * Copyright (c) 2003 Poul-Henning Kamp * All rights reserved. * * This software was developed for the FreeBSD Project by Poul-Henning Kamp * and NAI Labs, the Security Research Division of Network Associates, Inc. * under DARPA/SPAWAR contract N66001-01-C-8035 ("CBOSS"), as part of the * DARPA CHATS research program. * * 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 names of the authors may not be used to endorse or promote * products derived from this software without specific prior written * permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #ifndef _GEOM_GEOM_DISK_H_ #define _GEOM_GEOM_DISK_H_ #ifdef _KERNEL #include #include #include #include #define G_DISK_CLASS_NAME "DISK" struct disk; typedef int disk_open_t(struct disk *); typedef int disk_close_t(struct disk *); typedef void disk_strategy_t(struct bio *bp); typedef int disk_getattr_t(struct bio *bp); typedef void disk_gone_t(struct disk *); typedef int disk_ioctl_t(struct disk *, u_long cmd, void *data, int fflag, struct thread *td); /* NB: disk_ioctl_t SHALL be cast'able to d_ioctl_t */ struct g_geom; struct devstat; typedef enum { DISK_INIT_NONE, DISK_INIT_START, DISK_INIT_DONE } disk_init_level; struct disk { /* Fields which are private to geom_disk */ struct g_geom *d_geom; struct devstat *d_devstat; int d_goneflag; int d_destroyed; - struct mtx d_mtx; - char d_mtx_name[24]; disk_init_level d_init_level; /* Shared fields */ u_int d_flags; const char *d_name; u_int d_unit; struct bio_queue_head *d_queue; struct mtx *d_lock; /* Disk methods */ disk_open_t *d_open; disk_close_t *d_close; disk_strategy_t *d_strategy; disk_ioctl_t *d_ioctl; dumper_t *d_dump; disk_getattr_t *d_getattr; disk_gone_t *d_gone; /* Info fields from driver to geom_disk.c. Valid when open */ u_int d_sectorsize; off_t d_mediasize; u_int d_fwsectors; u_int d_fwheads; u_int d_maxsize; off_t d_delmaxsize; u_int d_stripeoffset; u_int d_stripesize; char d_ident[DISK_IDENT_SIZE]; char d_descr[DISK_IDENT_SIZE]; uint16_t d_hba_vendor; uint16_t d_hba_device; uint16_t d_hba_subvendor; uint16_t d_hba_subdevice; uint16_t d_rotation_rate; /* Fields private to the driver */ void *d_drv1; }; #define DISKFLAG_RESERVED 0x1 /* Was NEEDSGIANT */ #define DISKFLAG_OPEN 0x2 #define DISKFLAG_CANDELETE 0x4 #define DISKFLAG_CANFLUSHCACHE 0x8 #define DISKFLAG_UNMAPPED_BIO 0x10 #define DISKFLAG_DIRECT_COMPLETION 0x20 #define DISKFLAG_CANZONE 0x80 struct disk *disk_alloc(void); void disk_create(struct disk *disk, int version); void disk_destroy(struct disk *disk); void disk_gone(struct disk *disk); void disk_attr_changed(struct disk *dp, const char *attr, int flag); void disk_media_changed(struct disk *dp, int flag); void disk_media_gone(struct disk *dp, int flag); int disk_resize(struct disk *dp, int flag); #define DISK_VERSION_00 0x58561059 #define DISK_VERSION_01 0x5856105a #define DISK_VERSION_02 0x5856105b #define DISK_VERSION_03 0x5856105c #define DISK_VERSION_04 0x5856105d #define DISK_VERSION_05 0x5856105e #define DISK_VERSION DISK_VERSION_05 #endif /* _KERNEL */ #endif /* _GEOM_GEOM_DISK_H_ */ Index: projects/vnet/sys/kern/uipc_shm.c =================================================================== --- projects/vnet/sys/kern/uipc_shm.c (revision 302157) +++ projects/vnet/sys/kern/uipc_shm.c (revision 302158) @@ -1,1106 +1,1106 @@ /*- * Copyright (c) 2006, 2011 Robert N. M. Watson * 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. */ /* * Support for shared swap-backed anonymous memory objects via * shm_open(2) and shm_unlink(2). While most of the implementation is * here, vm_mmap.c contains mapping logic changes. * * TODO: * * (1) Need to export data to a userland tool via a sysctl. Should ipcs(1) * and ipcrm(1) be expanded or should new tools to manage both POSIX * kernel semaphores and POSIX shared memory be written? * * (2) Add support for this file type to fstat(1). * * (3) Resource limits? Does this need its own resource limits or are the * existing limits in mmap(2) sufficient? */ #include __FBSDID("$FreeBSD$"); #include "opt_capsicum.h" #include "opt_ktrace.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 struct shm_mapping { char *sm_path; Fnv32_t sm_fnv; struct shmfd *sm_shmfd; LIST_ENTRY(shm_mapping) sm_link; }; static MALLOC_DEFINE(M_SHMFD, "shmfd", "shared memory file descriptor"); static LIST_HEAD(, shm_mapping) *shm_dictionary; static struct sx shm_dict_lock; static struct mtx shm_timestamp_lock; static u_long shm_hash; static struct unrhdr *shm_ino_unr; static dev_t shm_dev_ino; #define SHM_HASH(fnv) (&shm_dictionary[(fnv) & shm_hash]) static void shm_init(void *arg); static void shm_insert(char *path, Fnv32_t fnv, struct shmfd *shmfd); static struct shmfd *shm_lookup(char *path, Fnv32_t fnv); static int shm_remove(char *path, Fnv32_t fnv, struct ucred *ucred); static fo_rdwr_t shm_read; static fo_rdwr_t shm_write; static fo_truncate_t shm_truncate; static fo_stat_t shm_stat; static fo_close_t shm_close; static fo_chmod_t shm_chmod; static fo_chown_t shm_chown; static fo_seek_t shm_seek; static fo_fill_kinfo_t shm_fill_kinfo; static fo_mmap_t shm_mmap; /* File descriptor operations. */ struct fileops shm_ops = { .fo_read = shm_read, .fo_write = shm_write, .fo_truncate = shm_truncate, .fo_ioctl = invfo_ioctl, .fo_poll = invfo_poll, .fo_kqfilter = invfo_kqfilter, .fo_stat = shm_stat, .fo_close = shm_close, .fo_chmod = shm_chmod, .fo_chown = shm_chown, .fo_sendfile = vn_sendfile, .fo_seek = shm_seek, .fo_fill_kinfo = shm_fill_kinfo, .fo_mmap = shm_mmap, .fo_flags = DFLAG_PASSABLE | DFLAG_SEEKABLE }; FEATURE(posix_shm, "POSIX shared memory"); static int uiomove_object_page(vm_object_t obj, size_t len, struct uio *uio) { vm_page_t m; vm_pindex_t idx; size_t tlen; int error, offset, rv; idx = OFF_TO_IDX(uio->uio_offset); offset = uio->uio_offset & PAGE_MASK; tlen = MIN(PAGE_SIZE - offset, len); VM_OBJECT_WLOCK(obj); /* * Read I/O without either a corresponding resident page or swap * page: use zero_region. This is intended to avoid instantiating * pages on read from a sparse region. */ if (uio->uio_rw == UIO_READ && vm_page_lookup(obj, idx) == NULL && !vm_pager_has_page(obj, idx, NULL, NULL)) { VM_OBJECT_WUNLOCK(obj); return (uiomove(__DECONST(void *, zero_region), tlen, uio)); } /* * Parallel reads of the page content from disk are prevented * by exclusive busy. * * Although the tmpfs vnode lock is held here, it is * nonetheless safe to sleep waiting for a free page. The * pageout daemon does not need to acquire the tmpfs vnode * lock to page out tobj's pages because tobj is a OBJT_SWAP * type object. */ m = vm_page_grab(obj, idx, VM_ALLOC_NORMAL); if (m->valid != VM_PAGE_BITS_ALL) { if (vm_pager_has_page(obj, idx, NULL, NULL)) { rv = vm_pager_get_pages(obj, &m, 1, NULL, NULL); if (rv != VM_PAGER_OK) { printf( "uiomove_object: vm_obj %p idx %jd valid %x pager error %d\n", obj, idx, m->valid, rv); vm_page_lock(m); vm_page_free(m); vm_page_unlock(m); VM_OBJECT_WUNLOCK(obj); return (EIO); } } else vm_page_zero_invalid(m, TRUE); } vm_page_xunbusy(m); vm_page_lock(m); vm_page_hold(m); if (m->queue == PQ_NONE) { vm_page_deactivate(m); } else { /* Requeue to maintain LRU ordering. */ vm_page_requeue(m); } vm_page_unlock(m); VM_OBJECT_WUNLOCK(obj); error = uiomove_fromphys(&m, offset, tlen, uio); if (uio->uio_rw == UIO_WRITE && error == 0) { VM_OBJECT_WLOCK(obj); vm_page_dirty(m); vm_pager_page_unswapped(m); VM_OBJECT_WUNLOCK(obj); } vm_page_lock(m); vm_page_unhold(m); vm_page_unlock(m); return (error); } int uiomove_object(vm_object_t obj, off_t obj_size, struct uio *uio) { ssize_t resid; size_t len; int error; error = 0; while ((resid = uio->uio_resid) > 0) { if (obj_size <= uio->uio_offset) break; len = MIN(obj_size - uio->uio_offset, resid); if (len == 0) break; error = uiomove_object_page(obj, len, uio); if (error != 0 || resid == uio->uio_resid) break; } return (error); } static int shm_seek(struct file *fp, off_t offset, int whence, struct thread *td) { struct shmfd *shmfd; off_t foffset; int error; shmfd = fp->f_data; foffset = foffset_lock(fp, 0); error = 0; switch (whence) { case L_INCR: if (foffset < 0 || (offset > 0 && foffset > OFF_MAX - offset)) { error = EOVERFLOW; break; } offset += foffset; break; case L_XTND: if (offset > 0 && shmfd->shm_size > OFF_MAX - offset) { error = EOVERFLOW; break; } offset += shmfd->shm_size; break; case L_SET: break; default: error = EINVAL; } if (error == 0) { if (offset < 0 || offset > shmfd->shm_size) error = EINVAL; else td->td_uretoff.tdu_off = offset; } foffset_unlock(fp, offset, error != 0 ? FOF_NOUPDATE : 0); return (error); } static int shm_read(struct file *fp, struct uio *uio, struct ucred *active_cred, int flags, struct thread *td) { struct shmfd *shmfd; void *rl_cookie; int error; shmfd = fp->f_data; - foffset_lock_uio(fp, uio, flags); - rl_cookie = rangelock_rlock(&shmfd->shm_rl, uio->uio_offset, - uio->uio_offset + uio->uio_resid, &shmfd->shm_mtx); #ifdef MAC error = mac_posixshm_check_read(active_cred, fp->f_cred, shmfd); if (error) return (error); #endif + foffset_lock_uio(fp, uio, flags); + rl_cookie = rangelock_rlock(&shmfd->shm_rl, uio->uio_offset, + uio->uio_offset + uio->uio_resid, &shmfd->shm_mtx); error = uiomove_object(shmfd->shm_object, shmfd->shm_size, uio); rangelock_unlock(&shmfd->shm_rl, rl_cookie, &shmfd->shm_mtx); foffset_unlock_uio(fp, uio, flags); return (error); } static int shm_write(struct file *fp, struct uio *uio, struct ucred *active_cred, int flags, struct thread *td) { struct shmfd *shmfd; void *rl_cookie; int error; shmfd = fp->f_data; #ifdef MAC error = mac_posixshm_check_write(active_cred, fp->f_cred, shmfd); if (error) return (error); #endif foffset_lock_uio(fp, uio, flags); if ((flags & FOF_OFFSET) == 0) { rl_cookie = rangelock_wlock(&shmfd->shm_rl, 0, OFF_MAX, &shmfd->shm_mtx); } else { rl_cookie = rangelock_wlock(&shmfd->shm_rl, uio->uio_offset, uio->uio_offset + uio->uio_resid, &shmfd->shm_mtx); } error = uiomove_object(shmfd->shm_object, shmfd->shm_size, uio); rangelock_unlock(&shmfd->shm_rl, rl_cookie, &shmfd->shm_mtx); foffset_unlock_uio(fp, uio, flags); return (error); } static int shm_truncate(struct file *fp, off_t length, struct ucred *active_cred, struct thread *td) { struct shmfd *shmfd; #ifdef MAC int error; #endif shmfd = fp->f_data; #ifdef MAC error = mac_posixshm_check_truncate(active_cred, fp->f_cred, shmfd); if (error) return (error); #endif return (shm_dotruncate(shmfd, length)); } static int shm_stat(struct file *fp, struct stat *sb, struct ucred *active_cred, struct thread *td) { struct shmfd *shmfd; #ifdef MAC int error; #endif shmfd = fp->f_data; #ifdef MAC error = mac_posixshm_check_stat(active_cred, fp->f_cred, shmfd); if (error) return (error); #endif /* * Attempt to return sanish values for fstat() on a memory file * descriptor. */ bzero(sb, sizeof(*sb)); sb->st_blksize = PAGE_SIZE; sb->st_size = shmfd->shm_size; sb->st_blocks = howmany(sb->st_size, sb->st_blksize); mtx_lock(&shm_timestamp_lock); sb->st_atim = shmfd->shm_atime; sb->st_ctim = shmfd->shm_ctime; sb->st_mtim = shmfd->shm_mtime; sb->st_birthtim = shmfd->shm_birthtime; sb->st_mode = S_IFREG | shmfd->shm_mode; /* XXX */ sb->st_uid = shmfd->shm_uid; sb->st_gid = shmfd->shm_gid; mtx_unlock(&shm_timestamp_lock); sb->st_dev = shm_dev_ino; sb->st_ino = shmfd->shm_ino; return (0); } static int shm_close(struct file *fp, struct thread *td) { struct shmfd *shmfd; shmfd = fp->f_data; fp->f_data = NULL; shm_drop(shmfd); return (0); } int shm_dotruncate(struct shmfd *shmfd, off_t length) { vm_object_t object; vm_page_t m; vm_pindex_t idx, nobjsize; vm_ooffset_t delta; int base, rv; object = shmfd->shm_object; VM_OBJECT_WLOCK(object); if (length == shmfd->shm_size) { VM_OBJECT_WUNLOCK(object); return (0); } nobjsize = OFF_TO_IDX(length + PAGE_MASK); /* Are we shrinking? If so, trim the end. */ if (length < shmfd->shm_size) { /* * Disallow any requests to shrink the size if this * object is mapped into the kernel. */ if (shmfd->shm_kmappings > 0) { VM_OBJECT_WUNLOCK(object); return (EBUSY); } /* * Zero the truncated part of the last page. */ base = length & PAGE_MASK; if (base != 0) { idx = OFF_TO_IDX(length); retry: m = vm_page_lookup(object, idx); if (m != NULL) { if (vm_page_sleep_if_busy(m, "shmtrc")) goto retry; } else if (vm_pager_has_page(object, idx, NULL, NULL)) { m = vm_page_alloc(object, idx, VM_ALLOC_NORMAL); if (m == NULL) { VM_OBJECT_WUNLOCK(object); VM_WAIT; VM_OBJECT_WLOCK(object); goto retry; } else if (m->valid != VM_PAGE_BITS_ALL) rv = vm_pager_get_pages(object, &m, 1, NULL, NULL); else /* A cached page was reactivated. */ rv = VM_PAGER_OK; vm_page_lock(m); if (rv == VM_PAGER_OK) { vm_page_deactivate(m); vm_page_unlock(m); vm_page_xunbusy(m); } else { vm_page_free(m); vm_page_unlock(m); VM_OBJECT_WUNLOCK(object); return (EIO); } } if (m != NULL) { pmap_zero_page_area(m, base, PAGE_SIZE - base); KASSERT(m->valid == VM_PAGE_BITS_ALL, ("shm_dotruncate: page %p is invalid", m)); vm_page_dirty(m); vm_pager_page_unswapped(m); } } delta = ptoa(object->size - nobjsize); /* Toss in memory pages. */ if (nobjsize < object->size) vm_object_page_remove(object, nobjsize, object->size, 0); /* Toss pages from swap. */ if (object->type == OBJT_SWAP) swap_pager_freespace(object, nobjsize, delta); /* Free the swap accounted for shm */ swap_release_by_cred(delta, object->cred); object->charge -= delta; } else { /* Attempt to reserve the swap */ delta = ptoa(nobjsize - object->size); if (!swap_reserve_by_cred(delta, object->cred)) { VM_OBJECT_WUNLOCK(object); return (ENOMEM); } object->charge += delta; } shmfd->shm_size = length; mtx_lock(&shm_timestamp_lock); vfs_timestamp(&shmfd->shm_ctime); shmfd->shm_mtime = shmfd->shm_ctime; mtx_unlock(&shm_timestamp_lock); object->size = nobjsize; VM_OBJECT_WUNLOCK(object); return (0); } /* * shmfd object management including creation and reference counting * routines. */ struct shmfd * shm_alloc(struct ucred *ucred, mode_t mode) { struct shmfd *shmfd; int ino; shmfd = malloc(sizeof(*shmfd), M_SHMFD, M_WAITOK | M_ZERO); shmfd->shm_size = 0; shmfd->shm_uid = ucred->cr_uid; shmfd->shm_gid = ucred->cr_gid; shmfd->shm_mode = mode; shmfd->shm_object = vm_pager_allocate(OBJT_DEFAULT, NULL, shmfd->shm_size, VM_PROT_DEFAULT, 0, ucred); KASSERT(shmfd->shm_object != NULL, ("shm_create: vm_pager_allocate")); shmfd->shm_object->pg_color = 0; VM_OBJECT_WLOCK(shmfd->shm_object); vm_object_clear_flag(shmfd->shm_object, OBJ_ONEMAPPING); vm_object_set_flag(shmfd->shm_object, OBJ_COLORED | OBJ_NOSPLIT); VM_OBJECT_WUNLOCK(shmfd->shm_object); vfs_timestamp(&shmfd->shm_birthtime); shmfd->shm_atime = shmfd->shm_mtime = shmfd->shm_ctime = shmfd->shm_birthtime; ino = alloc_unr(shm_ino_unr); if (ino == -1) shmfd->shm_ino = 0; else shmfd->shm_ino = ino; refcount_init(&shmfd->shm_refs, 1); mtx_init(&shmfd->shm_mtx, "shmrl", NULL, MTX_DEF); rangelock_init(&shmfd->shm_rl); #ifdef MAC mac_posixshm_init(shmfd); mac_posixshm_create(ucred, shmfd); #endif return (shmfd); } struct shmfd * shm_hold(struct shmfd *shmfd) { refcount_acquire(&shmfd->shm_refs); return (shmfd); } void shm_drop(struct shmfd *shmfd) { if (refcount_release(&shmfd->shm_refs)) { #ifdef MAC mac_posixshm_destroy(shmfd); #endif rangelock_destroy(&shmfd->shm_rl); mtx_destroy(&shmfd->shm_mtx); vm_object_deallocate(shmfd->shm_object); if (shmfd->shm_ino != 0) free_unr(shm_ino_unr, shmfd->shm_ino); free(shmfd, M_SHMFD); } } /* * Determine if the credentials have sufficient permissions for a * specified combination of FREAD and FWRITE. */ int shm_access(struct shmfd *shmfd, struct ucred *ucred, int flags) { accmode_t accmode; int error; accmode = 0; if (flags & FREAD) accmode |= VREAD; if (flags & FWRITE) accmode |= VWRITE; mtx_lock(&shm_timestamp_lock); error = vaccess(VREG, shmfd->shm_mode, shmfd->shm_uid, shmfd->shm_gid, accmode, ucred, NULL); mtx_unlock(&shm_timestamp_lock); return (error); } /* * Dictionary management. We maintain an in-kernel dictionary to map * paths to shmfd objects. We use the FNV hash on the path to store * the mappings in a hash table. */ static void shm_init(void *arg) { mtx_init(&shm_timestamp_lock, "shm timestamps", NULL, MTX_DEF); sx_init(&shm_dict_lock, "shm dictionary"); shm_dictionary = hashinit(1024, M_SHMFD, &shm_hash); shm_ino_unr = new_unrhdr(1, INT32_MAX, NULL); KASSERT(shm_ino_unr != NULL, ("shm fake inodes not initialized")); shm_dev_ino = devfs_alloc_cdp_inode(); KASSERT(shm_dev_ino > 0, ("shm dev inode not initialized")); } SYSINIT(shm_init, SI_SUB_SYSV_SHM, SI_ORDER_ANY, shm_init, NULL); static struct shmfd * shm_lookup(char *path, Fnv32_t fnv) { struct shm_mapping *map; LIST_FOREACH(map, SHM_HASH(fnv), sm_link) { if (map->sm_fnv != fnv) continue; if (strcmp(map->sm_path, path) == 0) return (map->sm_shmfd); } return (NULL); } static void shm_insert(char *path, Fnv32_t fnv, struct shmfd *shmfd) { struct shm_mapping *map; map = malloc(sizeof(struct shm_mapping), M_SHMFD, M_WAITOK); map->sm_path = path; map->sm_fnv = fnv; map->sm_shmfd = shm_hold(shmfd); shmfd->shm_path = path; LIST_INSERT_HEAD(SHM_HASH(fnv), map, sm_link); } static int shm_remove(char *path, Fnv32_t fnv, struct ucred *ucred) { struct shm_mapping *map; int error; LIST_FOREACH(map, SHM_HASH(fnv), sm_link) { if (map->sm_fnv != fnv) continue; if (strcmp(map->sm_path, path) == 0) { #ifdef MAC error = mac_posixshm_check_unlink(ucred, map->sm_shmfd); if (error) return (error); #endif error = shm_access(map->sm_shmfd, ucred, FREAD | FWRITE); if (error) return (error); map->sm_shmfd->shm_path = NULL; LIST_REMOVE(map, sm_link); shm_drop(map->sm_shmfd); free(map->sm_path, M_SHMFD); free(map, M_SHMFD); return (0); } } return (ENOENT); } int kern_shm_open(struct thread *td, const char *userpath, int flags, mode_t mode, struct filecaps *fcaps) { struct filedesc *fdp; struct shmfd *shmfd; struct file *fp; char *path; const char *pr_path; size_t pr_pathlen; Fnv32_t fnv; mode_t cmode; int fd, error; #ifdef CAPABILITY_MODE /* * shm_open(2) is only allowed for anonymous objects. */ if (IN_CAPABILITY_MODE(td) && (userpath != SHM_ANON)) return (ECAPMODE); #endif if ((flags & O_ACCMODE) != O_RDONLY && (flags & O_ACCMODE) != O_RDWR) return (EINVAL); if ((flags & ~(O_ACCMODE | O_CREAT | O_EXCL | O_TRUNC | O_CLOEXEC)) != 0) return (EINVAL); fdp = td->td_proc->p_fd; cmode = (mode & ~fdp->fd_cmask) & ACCESSPERMS; error = falloc_caps(td, &fp, &fd, O_CLOEXEC, fcaps); if (error) return (error); /* A SHM_ANON path pointer creates an anonymous object. */ if (userpath == SHM_ANON) { /* A read-only anonymous object is pointless. */ if ((flags & O_ACCMODE) == O_RDONLY) { fdclose(td, fp, fd); fdrop(fp, td); return (EINVAL); } shmfd = shm_alloc(td->td_ucred, cmode); } else { path = malloc(MAXPATHLEN, M_SHMFD, M_WAITOK); pr_path = td->td_ucred->cr_prison->pr_path; /* Construct a full pathname for jailed callers. */ pr_pathlen = strcmp(pr_path, "/") == 0 ? 0 : strlcpy(path, pr_path, MAXPATHLEN); error = copyinstr(userpath, path + pr_pathlen, MAXPATHLEN - pr_pathlen, NULL); #ifdef KTRACE if (error == 0 && KTRPOINT(curthread, KTR_NAMEI)) ktrnamei(path); #endif /* Require paths to start with a '/' character. */ if (error == 0 && path[pr_pathlen] != '/') error = EINVAL; if (error) { fdclose(td, fp, fd); fdrop(fp, td); free(path, M_SHMFD); return (error); } fnv = fnv_32_str(path, FNV1_32_INIT); sx_xlock(&shm_dict_lock); shmfd = shm_lookup(path, fnv); if (shmfd == NULL) { /* Object does not yet exist, create it if requested. */ if (flags & O_CREAT) { #ifdef MAC error = mac_posixshm_check_create(td->td_ucred, path); if (error == 0) { #endif shmfd = shm_alloc(td->td_ucred, cmode); shm_insert(path, fnv, shmfd); #ifdef MAC } #endif } else { free(path, M_SHMFD); error = ENOENT; } } else { /* * Object already exists, obtain a new * reference if requested and permitted. */ free(path, M_SHMFD); if ((flags & (O_CREAT | O_EXCL)) == (O_CREAT | O_EXCL)) error = EEXIST; else { #ifdef MAC error = mac_posixshm_check_open(td->td_ucred, shmfd, FFLAGS(flags & O_ACCMODE)); if (error == 0) #endif error = shm_access(shmfd, td->td_ucred, FFLAGS(flags & O_ACCMODE)); } /* * Truncate the file back to zero length if * O_TRUNC was specified and the object was * opened with read/write. */ if (error == 0 && (flags & (O_ACCMODE | O_TRUNC)) == (O_RDWR | O_TRUNC)) { #ifdef MAC error = mac_posixshm_check_truncate( td->td_ucred, fp->f_cred, shmfd); if (error == 0) #endif shm_dotruncate(shmfd, 0); } if (error == 0) shm_hold(shmfd); } sx_xunlock(&shm_dict_lock); if (error) { fdclose(td, fp, fd); fdrop(fp, td); return (error); } } finit(fp, FFLAGS(flags & O_ACCMODE), DTYPE_SHM, shmfd, &shm_ops); td->td_retval[0] = fd; fdrop(fp, td); return (0); } /* System calls. */ int sys_shm_open(struct thread *td, struct shm_open_args *uap) { return (kern_shm_open(td, uap->path, uap->flags, uap->mode, NULL)); } int sys_shm_unlink(struct thread *td, struct shm_unlink_args *uap) { char *path; const char *pr_path; size_t pr_pathlen; Fnv32_t fnv; int error; path = malloc(MAXPATHLEN, M_TEMP, M_WAITOK); pr_path = td->td_ucred->cr_prison->pr_path; pr_pathlen = strcmp(pr_path, "/") == 0 ? 0 : strlcpy(path, pr_path, MAXPATHLEN); error = copyinstr(uap->path, path + pr_pathlen, MAXPATHLEN - pr_pathlen, NULL); if (error) { free(path, M_TEMP); return (error); } #ifdef KTRACE if (KTRPOINT(curthread, KTR_NAMEI)) ktrnamei(path); #endif fnv = fnv_32_str(path, FNV1_32_INIT); sx_xlock(&shm_dict_lock); error = shm_remove(path, fnv, td->td_ucred); sx_xunlock(&shm_dict_lock); free(path, M_TEMP); return (error); } int shm_mmap(struct file *fp, vm_map_t map, vm_offset_t *addr, vm_size_t objsize, vm_prot_t prot, vm_prot_t cap_maxprot, int flags, vm_ooffset_t foff, struct thread *td) { struct shmfd *shmfd; vm_prot_t maxprot; int error; shmfd = fp->f_data; maxprot = VM_PROT_NONE; /* FREAD should always be set. */ if ((fp->f_flag & FREAD) != 0) maxprot |= VM_PROT_EXECUTE | VM_PROT_READ; if ((fp->f_flag & FWRITE) != 0) maxprot |= VM_PROT_WRITE; /* Don't permit shared writable mappings on read-only descriptors. */ if ((flags & MAP_SHARED) != 0 && (maxprot & VM_PROT_WRITE) == 0 && (prot & VM_PROT_WRITE) != 0) return (EACCES); maxprot &= cap_maxprot; #ifdef MAC error = mac_posixshm_check_mmap(td->td_ucred, shmfd, prot, flags); if (error != 0) return (error); #endif /* * XXXRW: This validation is probably insufficient, and subject to * sign errors. It should be fixed. */ if (foff >= shmfd->shm_size || foff + objsize > round_page(shmfd->shm_size)) return (EINVAL); mtx_lock(&shm_timestamp_lock); vfs_timestamp(&shmfd->shm_atime); mtx_unlock(&shm_timestamp_lock); vm_object_reference(shmfd->shm_object); error = vm_mmap_object(map, addr, objsize, prot, maxprot, flags, shmfd->shm_object, foff, FALSE, td); if (error != 0) vm_object_deallocate(shmfd->shm_object); return (0); } static int shm_chmod(struct file *fp, mode_t mode, struct ucred *active_cred, struct thread *td) { struct shmfd *shmfd; int error; error = 0; shmfd = fp->f_data; mtx_lock(&shm_timestamp_lock); /* * SUSv4 says that x bits of permission need not be affected. * Be consistent with our shm_open there. */ #ifdef MAC error = mac_posixshm_check_setmode(active_cred, shmfd, mode); if (error != 0) goto out; #endif error = vaccess(VREG, shmfd->shm_mode, shmfd->shm_uid, shmfd->shm_gid, VADMIN, active_cred, NULL); if (error != 0) goto out; shmfd->shm_mode = mode & ACCESSPERMS; out: mtx_unlock(&shm_timestamp_lock); return (error); } static int shm_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred, struct thread *td) { struct shmfd *shmfd; int error; error = 0; shmfd = fp->f_data; mtx_lock(&shm_timestamp_lock); #ifdef MAC error = mac_posixshm_check_setowner(active_cred, shmfd, uid, gid); if (error != 0) goto out; #endif if (uid == (uid_t)-1) uid = shmfd->shm_uid; if (gid == (gid_t)-1) gid = shmfd->shm_gid; if (((uid != shmfd->shm_uid && uid != active_cred->cr_uid) || (gid != shmfd->shm_gid && !groupmember(gid, active_cred))) && (error = priv_check_cred(active_cred, PRIV_VFS_CHOWN, 0))) goto out; shmfd->shm_uid = uid; shmfd->shm_gid = gid; out: mtx_unlock(&shm_timestamp_lock); return (error); } /* * Helper routines to allow the backing object of a shared memory file * descriptor to be mapped in the kernel. */ int shm_map(struct file *fp, size_t size, off_t offset, void **memp) { struct shmfd *shmfd; vm_offset_t kva, ofs; vm_object_t obj; int rv; if (fp->f_type != DTYPE_SHM) return (EINVAL); shmfd = fp->f_data; obj = shmfd->shm_object; VM_OBJECT_WLOCK(obj); /* * XXXRW: This validation is probably insufficient, and subject to * sign errors. It should be fixed. */ if (offset >= shmfd->shm_size || offset + size > round_page(shmfd->shm_size)) { VM_OBJECT_WUNLOCK(obj); return (EINVAL); } shmfd->shm_kmappings++; vm_object_reference_locked(obj); VM_OBJECT_WUNLOCK(obj); /* Map the object into the kernel_map and wire it. */ kva = vm_map_min(kernel_map); ofs = offset & PAGE_MASK; offset = trunc_page(offset); size = round_page(size + ofs); rv = vm_map_find(kernel_map, obj, offset, &kva, size, 0, VMFS_OPTIMAL_SPACE, VM_PROT_READ | VM_PROT_WRITE, VM_PROT_READ | VM_PROT_WRITE, 0); if (rv == KERN_SUCCESS) { rv = vm_map_wire(kernel_map, kva, kva + size, VM_MAP_WIRE_SYSTEM | VM_MAP_WIRE_NOHOLES); if (rv == KERN_SUCCESS) { *memp = (void *)(kva + ofs); return (0); } vm_map_remove(kernel_map, kva, kva + size); } else vm_object_deallocate(obj); /* On failure, drop our mapping reference. */ VM_OBJECT_WLOCK(obj); shmfd->shm_kmappings--; VM_OBJECT_WUNLOCK(obj); return (vm_mmap_to_errno(rv)); } /* * We require the caller to unmap the entire entry. This allows us to * safely decrement shm_kmappings when a mapping is removed. */ int shm_unmap(struct file *fp, void *mem, size_t size) { struct shmfd *shmfd; vm_map_entry_t entry; vm_offset_t kva, ofs; vm_object_t obj; vm_pindex_t pindex; vm_prot_t prot; boolean_t wired; vm_map_t map; int rv; if (fp->f_type != DTYPE_SHM) return (EINVAL); shmfd = fp->f_data; kva = (vm_offset_t)mem; ofs = kva & PAGE_MASK; kva = trunc_page(kva); size = round_page(size + ofs); map = kernel_map; rv = vm_map_lookup(&map, kva, VM_PROT_READ | VM_PROT_WRITE, &entry, &obj, &pindex, &prot, &wired); if (rv != KERN_SUCCESS) return (EINVAL); if (entry->start != kva || entry->end != kva + size) { vm_map_lookup_done(map, entry); return (EINVAL); } vm_map_lookup_done(map, entry); if (obj != shmfd->shm_object) return (EINVAL); vm_map_remove(map, kva, kva + size); VM_OBJECT_WLOCK(obj); KASSERT(shmfd->shm_kmappings > 0, ("shm_unmap: object not mapped")); shmfd->shm_kmappings--; VM_OBJECT_WUNLOCK(obj); return (0); } static int shm_fill_kinfo(struct file *fp, struct kinfo_file *kif, struct filedesc *fdp) { const char *path, *pr_path; struct shmfd *shmfd; size_t pr_pathlen; kif->kf_type = KF_TYPE_SHM; shmfd = fp->f_data; mtx_lock(&shm_timestamp_lock); kif->kf_un.kf_file.kf_file_mode = S_IFREG | shmfd->shm_mode; /* XXX */ mtx_unlock(&shm_timestamp_lock); kif->kf_un.kf_file.kf_file_size = shmfd->shm_size; if (shmfd->shm_path != NULL) { sx_slock(&shm_dict_lock); if (shmfd->shm_path != NULL) { path = shmfd->shm_path; pr_path = curthread->td_ucred->cr_prison->pr_path; if (strcmp(pr_path, "/") != 0) { /* Return the jail-rooted pathname. */ pr_pathlen = strlen(pr_path); if (strncmp(path, pr_path, pr_pathlen) == 0 && path[pr_pathlen] == '/') path += pr_pathlen; } strlcpy(kif->kf_path, path, sizeof(kif->kf_path)); } sx_sunlock(&shm_dict_lock); } return (0); } Index: projects/vnet/sys/net/if_var.h =================================================================== --- projects/vnet/sys/net/if_var.h (revision 302157) +++ projects/vnet/sys/net/if_var.h (revision 302158) @@ -1,664 +1,666 @@ /*- * Copyright (c) 1982, 1986, 1989, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * From: @(#)if.h 8.1 (Berkeley) 6/10/93 * $FreeBSD$ */ #ifndef _NET_IF_VAR_H_ #define _NET_IF_VAR_H_ /* * Structures defining a network interface, providing a packet * transport mechanism (ala level 0 of the PUP protocols). * * Each interface accepts output datagrams of a specified maximum * length, and provides higher level routines with input datagrams * received from its medium. * * Output occurs when the routine if_output is called, with three parameters: * (*ifp->if_output)(ifp, m, dst, rt) * Here m is the mbuf chain to be sent and dst is the destination address. * The output routine encapsulates the supplied datagram if necessary, * and then transmits it on its medium. * * On input, each interface unwraps the data received by it, and either * places it on the input queue of an internetwork datagram routine * and posts the associated software interrupt, or passes the datagram to a raw * packet input routine. * * Routines exist for locating interfaces by their addresses * or for locating an interface on a certain network, as well as more general * routing and gateway routines maintaining information used to locate * interfaces. These routines live in the files if.c and route.c */ struct rtentry; /* ifa_rtrequest */ struct rt_addrinfo; /* ifa_rtrequest */ struct socket; struct carp_if; struct carp_softc; struct ifvlantrunk; struct route; /* if_output */ struct vnet; struct ifmedia; struct netmap_adapter; #ifdef _KERNEL #include /* ifqueue only? */ #include #include #endif /* _KERNEL */ #include #include /* XXX */ #include /* struct ifqueue */ #include /* XXX */ #include /* XXX */ #include /* if_link_task */ #define IF_DUNIT_NONE -1 #include TAILQ_HEAD(ifnethead, ifnet); /* we use TAILQs so that the order of */ TAILQ_HEAD(ifaddrhead, ifaddr); /* instantiation is preserved in the list */ TAILQ_HEAD(ifmultihead, ifmultiaddr); TAILQ_HEAD(ifgrouphead, ifg_group); #ifdef _KERNEL VNET_DECLARE(struct pfil_head, link_pfil_hook); /* packet filter hooks */ #define V_link_pfil_hook VNET(link_pfil_hook) #define HHOOK_IPSEC_INET 0 #define HHOOK_IPSEC_INET6 1 #define HHOOK_IPSEC_COUNT 2 VNET_DECLARE(struct hhook_head *, ipsec_hhh_in[HHOOK_IPSEC_COUNT]); VNET_DECLARE(struct hhook_head *, ipsec_hhh_out[HHOOK_IPSEC_COUNT]); #define V_ipsec_hhh_in VNET(ipsec_hhh_in) #define V_ipsec_hhh_out VNET(ipsec_hhh_out) #endif /* _KERNEL */ typedef enum { IFCOUNTER_IPACKETS = 0, IFCOUNTER_IERRORS, IFCOUNTER_OPACKETS, IFCOUNTER_OERRORS, IFCOUNTER_COLLISIONS, IFCOUNTER_IBYTES, IFCOUNTER_OBYTES, IFCOUNTER_IMCASTS, IFCOUNTER_OMCASTS, IFCOUNTER_IQDROPS, IFCOUNTER_OQDROPS, IFCOUNTER_NOPROTO, IFCOUNTERS /* Array size. */ } ift_counter; typedef struct ifnet * if_t; typedef void (*if_start_fn_t)(if_t); typedef int (*if_ioctl_fn_t)(if_t, u_long, caddr_t); typedef void (*if_init_fn_t)(void *); typedef void (*if_qflush_fn_t)(if_t); typedef int (*if_transmit_fn_t)(if_t, struct mbuf *); typedef uint64_t (*if_get_counter_t)(if_t, ift_counter); struct ifnet_hw_tsomax { u_int tsomaxbytes; /* TSO total burst length limit in bytes */ u_int tsomaxsegcount; /* TSO maximum segment count */ u_int tsomaxsegsize; /* TSO maximum segment size in bytes */ }; /* Interface encap request types */ typedef enum { IFENCAP_LL = 1 /* pre-calculate link-layer header */ } ife_type; /* * The structure below allows to request various pre-calculated L2/L3 headers * for different media. Requests varies by type (rtype field). * * IFENCAP_LL type: pre-calculates link header based on address family * and destination lladdr. * * Input data fields: * buf: pointer to destination buffer * bufsize: buffer size * flags: IFENCAP_FLAG_BROADCAST if destination is broadcast * family: address family defined by AF_ constant. * lladdr: pointer to link-layer address * lladdr_len: length of link-layer address * hdata: pointer to L3 header (optional, used for ARP requests). * Output data fields: * buf: encap data is stored here * bufsize: resulting encap length is stored here * lladdr_off: offset of link-layer address from encap hdr start * hdata: L3 header may be altered if necessary */ struct if_encap_req { u_char *buf; /* Destination buffer (w) */ size_t bufsize; /* size of provided buffer (r) */ ife_type rtype; /* request type (r) */ uint32_t flags; /* Request flags (r) */ int family; /* Address family AF_* (r) */ int lladdr_off; /* offset from header start (w) */ int lladdr_len; /* lladdr length (r) */ char *lladdr; /* link-level address pointer (r) */ char *hdata; /* Upper layer header data (rw) */ }; #define IFENCAP_FLAG_BROADCAST 0x02 /* Destination is broadcast */ /* * Structure defining a network interface. * * Size ILP32: 592 (approx) * LP64: 1048 (approx) */ struct ifnet { /* General book keeping of interface lists. */ TAILQ_ENTRY(ifnet) if_link; /* all struct ifnets are chained */ LIST_ENTRY(ifnet) if_clones; /* interfaces of a cloner */ TAILQ_HEAD(, ifg_list) if_groups; /* linked list of groups per if */ /* protected by if_addr_lock */ u_char if_alloctype; /* if_type at time of allocation */ /* Driver and protocol specific information that remains stable. */ void *if_softc; /* pointer to driver state */ void *if_llsoftc; /* link layer softc */ void *if_l2com; /* pointer to protocol bits */ const char *if_dname; /* driver name */ int if_dunit; /* unit or IF_DUNIT_NONE */ u_short if_index; /* numeric abbreviation for this if */ short if_index_reserved; /* spare space to grow if_index */ char if_xname[IFNAMSIZ]; /* external name (name + unit) */ char *if_description; /* interface description */ /* Variable fields that are touched by the stack and drivers. */ int if_flags; /* up/down, broadcast, etc. */ int if_drv_flags; /* driver-managed status flags */ int if_capabilities; /* interface features & capabilities */ int if_capenable; /* enabled features & capabilities */ void *if_linkmib; /* link-type-specific MIB data */ size_t if_linkmiblen; /* length of above data */ u_int if_refcount; /* reference count */ /* These fields are shared with struct if_data. */ uint8_t if_type; /* ethernet, tokenring, etc */ uint8_t if_addrlen; /* media address length */ uint8_t if_hdrlen; /* media header length */ uint8_t if_link_state; /* current link state */ uint32_t if_mtu; /* maximum transmission unit */ uint32_t if_metric; /* routing metric (external only) */ uint64_t if_baudrate; /* linespeed */ uint64_t if_hwassist; /* HW offload capabilities, see IFCAP */ time_t if_epoch; /* uptime at attach or stat reset */ struct timeval if_lastchange; /* time of last administrative change */ struct ifaltq if_snd; /* output queue (includes altq) */ struct task if_linktask; /* task for link change events */ /* Addresses of different protocol families assigned to this if. */ struct rwlock if_addr_lock; /* lock to protect address lists */ /* * if_addrhead is the list of all addresses associated to * an interface. * Some code in the kernel assumes that first element * of the list has type AF_LINK, and contains sockaddr_dl * addresses which store the link-level address and the name * of the interface. * However, access to the AF_LINK address through this * field is deprecated. Use if_addr or ifaddr_byindex() instead. */ struct ifaddrhead if_addrhead; /* linked list of addresses per if */ struct ifmultihead if_multiaddrs; /* multicast addresses configured */ int if_amcount; /* number of all-multicast requests */ struct ifaddr *if_addr; /* pointer to link-level address */ const u_int8_t *if_broadcastaddr; /* linklevel broadcast bytestring */ struct rwlock if_afdata_lock; void *if_afdata[AF_MAX]; int if_afdata_initialized; /* Additional features hung off the interface. */ u_int if_fib; /* interface FIB */ struct vnet *if_vnet; /* pointer to network stack instance */ struct vnet *if_home_vnet; /* where this ifnet originates from */ struct ifvlantrunk *if_vlantrunk; /* pointer to 802.1q data */ struct bpf_if *if_bpf; /* packet filter structure */ int if_pcount; /* number of promiscuous listeners */ void *if_bridge; /* bridge glue */ void *if_lagg; /* lagg glue */ void *if_pf_kif; /* pf glue */ struct carp_if *if_carp; /* carp interface structure */ struct label *if_label; /* interface MAC label */ struct netmap_adapter *if_netmap; /* netmap(4) softc */ /* Various procedures of the layer2 encapsulation and drivers. */ int (*if_output) /* output routine (enqueue) */ (struct ifnet *, struct mbuf *, const struct sockaddr *, struct route *); void (*if_input) /* input routine (from h/w driver) */ (struct ifnet *, struct mbuf *); if_start_fn_t if_start; /* initiate output routine */ if_ioctl_fn_t if_ioctl; /* ioctl routine */ if_init_fn_t if_init; /* Init routine */ int (*if_resolvemulti) /* validate/resolve multicast */ (struct ifnet *, struct sockaddr **, struct sockaddr *); if_qflush_fn_t if_qflush; /* flush any queue */ if_transmit_fn_t if_transmit; /* initiate output routine */ void (*if_reassign) /* reassign to vnet routine */ (struct ifnet *, struct vnet *, char *); if_get_counter_t if_get_counter; /* get counter values */ int (*if_requestencap) /* make link header from request */ (struct ifnet *, struct if_encap_req *); /* Statistics. */ counter_u64_t if_counters[IFCOUNTERS]; /* Stuff that's only temporary and doesn't belong here. */ /* * Network adapter TSO limits: * =========================== * * If the "if_hw_tsomax" field is zero the maximum segment * length limit does not apply. If the "if_hw_tsomaxsegcount" * or the "if_hw_tsomaxsegsize" field is zero the TSO segment * count limit does not apply. If all three fields are zero, * there is no TSO limit. * * NOTE: The TSO limits should reflect the values used in the * BUSDMA tag a network adapter is using to load a mbuf chain * for transmission. The TCP/IP network stack will subtract * space for all linklevel and protocol level headers and * ensure that the full mbuf chain passed to the network * adapter fits within the given limits. */ u_int if_hw_tsomax; /* TSO maximum size in bytes */ u_int if_hw_tsomaxsegcount; /* TSO maximum segment count */ u_int if_hw_tsomaxsegsize; /* TSO maximum segment size in bytes */ /* * Spare fields to be added before branching a stable branch, so * that structure can be enhanced without changing the kernel * binary interface. */ + void *if_pspare[4]; /* packet pacing / general use */ + int if_ispare[4]; /* packet pacing / general use */ }; /* for compatibility with other BSDs */ #define if_addrlist if_addrhead #define if_list if_link #define if_name(ifp) ((ifp)->if_xname) /* * Locks for address lists on the network interface. */ #define IF_ADDR_LOCK_INIT(if) rw_init(&(if)->if_addr_lock, "if_addr_lock") #define IF_ADDR_LOCK_DESTROY(if) rw_destroy(&(if)->if_addr_lock) #define IF_ADDR_WLOCK(if) rw_wlock(&(if)->if_addr_lock) #define IF_ADDR_WUNLOCK(if) rw_wunlock(&(if)->if_addr_lock) #define IF_ADDR_RLOCK(if) rw_rlock(&(if)->if_addr_lock) #define IF_ADDR_RUNLOCK(if) rw_runlock(&(if)->if_addr_lock) #define IF_ADDR_LOCK_ASSERT(if) rw_assert(&(if)->if_addr_lock, RA_LOCKED) #define IF_ADDR_WLOCK_ASSERT(if) rw_assert(&(if)->if_addr_lock, RA_WLOCKED) /* * Function variations on locking macros intended to be used by loadable * kernel modules in order to divorce them from the internals of address list * locking. */ void if_addr_rlock(struct ifnet *ifp); /* if_addrhead */ void if_addr_runlock(struct ifnet *ifp); /* if_addrhead */ void if_maddr_rlock(if_t ifp); /* if_multiaddrs */ void if_maddr_runlock(if_t ifp); /* if_multiaddrs */ #ifdef _KERNEL #ifdef _SYS_EVENTHANDLER_H_ /* interface link layer address change event */ typedef void (*iflladdr_event_handler_t)(void *, struct ifnet *); EVENTHANDLER_DECLARE(iflladdr_event, iflladdr_event_handler_t); /* interface address change event */ typedef void (*ifaddr_event_handler_t)(void *, struct ifnet *); EVENTHANDLER_DECLARE(ifaddr_event, ifaddr_event_handler_t); /* new interface arrival event */ typedef void (*ifnet_arrival_event_handler_t)(void *, struct ifnet *); EVENTHANDLER_DECLARE(ifnet_arrival_event, ifnet_arrival_event_handler_t); /* interface departure event */ typedef void (*ifnet_departure_event_handler_t)(void *, struct ifnet *); EVENTHANDLER_DECLARE(ifnet_departure_event, ifnet_departure_event_handler_t); /* Interface link state change event */ typedef void (*ifnet_link_event_handler_t)(void *, struct ifnet *, int); EVENTHANDLER_DECLARE(ifnet_link_event, ifnet_link_event_handler_t); #endif /* _SYS_EVENTHANDLER_H_ */ /* * interface groups */ struct ifg_group { char ifg_group[IFNAMSIZ]; u_int ifg_refcnt; void *ifg_pf_kif; TAILQ_HEAD(, ifg_member) ifg_members; TAILQ_ENTRY(ifg_group) ifg_next; }; struct ifg_member { TAILQ_ENTRY(ifg_member) ifgm_next; struct ifnet *ifgm_ifp; }; struct ifg_list { struct ifg_group *ifgl_group; TAILQ_ENTRY(ifg_list) ifgl_next; }; #ifdef _SYS_EVENTHANDLER_H_ /* group attach event */ typedef void (*group_attach_event_handler_t)(void *, struct ifg_group *); EVENTHANDLER_DECLARE(group_attach_event, group_attach_event_handler_t); /* group detach event */ typedef void (*group_detach_event_handler_t)(void *, struct ifg_group *); EVENTHANDLER_DECLARE(group_detach_event, group_detach_event_handler_t); /* group change event */ typedef void (*group_change_event_handler_t)(void *, const char *); EVENTHANDLER_DECLARE(group_change_event, group_change_event_handler_t); #endif /* _SYS_EVENTHANDLER_H_ */ #define IF_AFDATA_LOCK_INIT(ifp) \ rw_init(&(ifp)->if_afdata_lock, "if_afdata") #define IF_AFDATA_WLOCK(ifp) rw_wlock(&(ifp)->if_afdata_lock) #define IF_AFDATA_RLOCK(ifp) rw_rlock(&(ifp)->if_afdata_lock) #define IF_AFDATA_WUNLOCK(ifp) rw_wunlock(&(ifp)->if_afdata_lock) #define IF_AFDATA_RUNLOCK(ifp) rw_runlock(&(ifp)->if_afdata_lock) #define IF_AFDATA_LOCK(ifp) IF_AFDATA_WLOCK(ifp) #define IF_AFDATA_UNLOCK(ifp) IF_AFDATA_WUNLOCK(ifp) #define IF_AFDATA_TRYLOCK(ifp) rw_try_wlock(&(ifp)->if_afdata_lock) #define IF_AFDATA_DESTROY(ifp) rw_destroy(&(ifp)->if_afdata_lock) #define IF_AFDATA_LOCK_ASSERT(ifp) rw_assert(&(ifp)->if_afdata_lock, RA_LOCKED) #define IF_AFDATA_RLOCK_ASSERT(ifp) rw_assert(&(ifp)->if_afdata_lock, RA_RLOCKED) #define IF_AFDATA_WLOCK_ASSERT(ifp) rw_assert(&(ifp)->if_afdata_lock, RA_WLOCKED) #define IF_AFDATA_UNLOCK_ASSERT(ifp) rw_assert(&(ifp)->if_afdata_lock, RA_UNLOCKED) /* * 72 was chosen below because it is the size of a TCP/IP * header (40) + the minimum mss (32). */ #define IF_MINMTU 72 #define IF_MAXMTU 65535 #define TOEDEV(ifp) ((ifp)->if_llsoftc) /* * The ifaddr structure contains information about one address * of an interface. They are maintained by the different address families, * are allocated and attached when an address is set, and are linked * together so all addresses for an interface can be located. * * NOTE: a 'struct ifaddr' is always at the beginning of a larger * chunk of malloc'ed memory, where we store the three addresses * (ifa_addr, ifa_dstaddr and ifa_netmask) referenced here. */ struct ifaddr { struct sockaddr *ifa_addr; /* address of interface */ struct sockaddr *ifa_dstaddr; /* other end of p-to-p link */ #define ifa_broadaddr ifa_dstaddr /* broadcast address interface */ struct sockaddr *ifa_netmask; /* used to determine subnet */ struct ifnet *ifa_ifp; /* back-pointer to interface */ struct carp_softc *ifa_carp; /* pointer to CARP data */ TAILQ_ENTRY(ifaddr) ifa_link; /* queue macro glue */ void (*ifa_rtrequest) /* check or clean routes (+ or -)'d */ (int, struct rtentry *, struct rt_addrinfo *); u_short ifa_flags; /* mostly rt_flags for cloning */ #define IFA_ROUTE RTF_UP /* route installed */ #define IFA_RTSELF RTF_HOST /* loopback route to self installed */ u_int ifa_refcnt; /* references to this structure */ counter_u64_t ifa_ipackets; counter_u64_t ifa_opackets; counter_u64_t ifa_ibytes; counter_u64_t ifa_obytes; }; /* For compatibility with other BSDs. SCTP uses it. */ #define ifa_list ifa_link struct ifaddr * ifa_alloc(size_t size, int flags); void ifa_free(struct ifaddr *ifa); void ifa_ref(struct ifaddr *ifa); /* * Multicast address structure. This is analogous to the ifaddr * structure except that it keeps track of multicast addresses. */ struct ifmultiaddr { TAILQ_ENTRY(ifmultiaddr) ifma_link; /* queue macro glue */ struct sockaddr *ifma_addr; /* address this membership is for */ struct sockaddr *ifma_lladdr; /* link-layer translation, if any */ struct ifnet *ifma_ifp; /* back-pointer to interface */ u_int ifma_refcount; /* reference count */ void *ifma_protospec; /* protocol-specific state, if any */ struct ifmultiaddr *ifma_llifma; /* pointer to ifma for ifma_lladdr */ }; extern struct rwlock ifnet_rwlock; extern struct sx ifnet_sxlock; #define IFNET_WLOCK() do { \ sx_xlock(&ifnet_sxlock); \ rw_wlock(&ifnet_rwlock); \ } while (0) #define IFNET_WUNLOCK() do { \ rw_wunlock(&ifnet_rwlock); \ sx_xunlock(&ifnet_sxlock); \ } while (0) /* * To assert the ifnet lock, you must know not only whether it's for read or * write, but also whether it was acquired with sleep support or not. */ #define IFNET_RLOCK_ASSERT() sx_assert(&ifnet_sxlock, SA_SLOCKED) #define IFNET_RLOCK_NOSLEEP_ASSERT() rw_assert(&ifnet_rwlock, RA_RLOCKED) #define IFNET_WLOCK_ASSERT() do { \ sx_assert(&ifnet_sxlock, SA_XLOCKED); \ rw_assert(&ifnet_rwlock, RA_WLOCKED); \ } while (0) #define IFNET_RLOCK() sx_slock(&ifnet_sxlock) #define IFNET_RLOCK_NOSLEEP() rw_rlock(&ifnet_rwlock) #define IFNET_RUNLOCK() sx_sunlock(&ifnet_sxlock) #define IFNET_RUNLOCK_NOSLEEP() rw_runlock(&ifnet_rwlock) /* * Look up an ifnet given its index; the _ref variant also acquires a * reference that must be freed using if_rele(). It is almost always a bug * to call ifnet_byindex() instead if ifnet_byindex_ref(). */ struct ifnet *ifnet_byindex(u_short idx); struct ifnet *ifnet_byindex_locked(u_short idx); struct ifnet *ifnet_byindex_ref(u_short idx); /* * Given the index, ifaddr_byindex() returns the one and only * link-level ifaddr for the interface. You are not supposed to use * it to traverse the list of addresses associated to the interface. */ struct ifaddr *ifaddr_byindex(u_short idx); VNET_DECLARE(struct ifnethead, ifnet); VNET_DECLARE(struct ifgrouphead, ifg_head); VNET_DECLARE(int, if_index); VNET_DECLARE(struct ifnet *, loif); /* first loopback interface */ #define V_ifnet VNET(ifnet) #define V_ifg_head VNET(ifg_head) #define V_if_index VNET(if_index) #define V_loif VNET(loif) int if_addgroup(struct ifnet *, const char *); int if_delgroup(struct ifnet *, const char *); int if_addmulti(struct ifnet *, struct sockaddr *, struct ifmultiaddr **); int if_allmulti(struct ifnet *, int); struct ifnet* if_alloc(u_char); void if_attach(struct ifnet *); void if_dead(struct ifnet *); int if_delmulti(struct ifnet *, struct sockaddr *); void if_delmulti_ifma(struct ifmultiaddr *); void if_detach(struct ifnet *); void if_purgeaddrs(struct ifnet *); void if_delallmulti(struct ifnet *); void if_down(struct ifnet *); struct ifmultiaddr * if_findmulti(struct ifnet *, const struct sockaddr *); void if_free(struct ifnet *); void if_initname(struct ifnet *, const char *, int); void if_link_state_change(struct ifnet *, int); int if_printf(struct ifnet *, const char *, ...) __printflike(2, 3); void if_ref(struct ifnet *); void if_rele(struct ifnet *); int if_setlladdr(struct ifnet *, const u_char *, int); void if_up(struct ifnet *); int ifioctl(struct socket *, u_long, caddr_t, struct thread *); int ifpromisc(struct ifnet *, int); struct ifnet *ifunit(const char *); struct ifnet *ifunit_ref(const char *); int ifa_add_loopback_route(struct ifaddr *, struct sockaddr *); int ifa_del_loopback_route(struct ifaddr *, struct sockaddr *); int ifa_switch_loopback_route(struct ifaddr *, struct sockaddr *); struct ifaddr *ifa_ifwithaddr(const struct sockaddr *); int ifa_ifwithaddr_check(const struct sockaddr *); struct ifaddr *ifa_ifwithbroadaddr(const struct sockaddr *, int); struct ifaddr *ifa_ifwithdstaddr(const struct sockaddr *, int); struct ifaddr *ifa_ifwithnet(const struct sockaddr *, int, int); struct ifaddr *ifa_ifwithroute(int, const struct sockaddr *, struct sockaddr *, u_int); struct ifaddr *ifaof_ifpforaddr(const struct sockaddr *, struct ifnet *); int ifa_preferred(struct ifaddr *, struct ifaddr *); int if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen); typedef void *if_com_alloc_t(u_char type, struct ifnet *ifp); typedef void if_com_free_t(void *com, u_char type); void if_register_com_alloc(u_char type, if_com_alloc_t *a, if_com_free_t *f); void if_deregister_com_alloc(u_char type); void if_data_copy(struct ifnet *, struct if_data *); uint64_t if_get_counter_default(struct ifnet *, ift_counter); void if_inc_counter(struct ifnet *, ift_counter, int64_t); #define IF_LLADDR(ifp) \ LLADDR((struct sockaddr_dl *)((ifp)->if_addr->ifa_addr)) uint64_t if_setbaudrate(if_t ifp, uint64_t baudrate); uint64_t if_getbaudrate(if_t ifp); int if_setcapabilities(if_t ifp, int capabilities); int if_setcapabilitiesbit(if_t ifp, int setbit, int clearbit); int if_getcapabilities(if_t ifp); int if_togglecapenable(if_t ifp, int togglecap); int if_setcapenable(if_t ifp, int capenable); int if_setcapenablebit(if_t ifp, int setcap, int clearcap); int if_getcapenable(if_t ifp); const char *if_getdname(if_t ifp); int if_setdev(if_t ifp, void *dev); int if_setdrvflagbits(if_t ifp, int if_setflags, int clear_flags); int if_getdrvflags(if_t ifp); int if_setdrvflags(if_t ifp, int flags); int if_clearhwassist(if_t ifp); int if_sethwassistbits(if_t ifp, int toset, int toclear); int if_sethwassist(if_t ifp, int hwassist_bit); int if_gethwassist(if_t ifp); int if_setsoftc(if_t ifp, void *softc); void *if_getsoftc(if_t ifp); int if_setflags(if_t ifp, int flags); int if_setmtu(if_t ifp, int mtu); int if_getmtu(if_t ifp); int if_getmtu_family(if_t ifp, int family); int if_setflagbits(if_t ifp, int set, int clear); int if_getflags(if_t ifp); int if_sendq_empty(if_t ifp); int if_setsendqready(if_t ifp); int if_setsendqlen(if_t ifp, int tx_desc_count); int if_input(if_t ifp, struct mbuf* sendmp); int if_sendq_prepend(if_t ifp, struct mbuf *m); struct mbuf *if_dequeue(if_t ifp); int if_setifheaderlen(if_t ifp, int len); void if_setrcvif(struct mbuf *m, if_t ifp); void if_setvtag(struct mbuf *m, u_int16_t tag); u_int16_t if_getvtag(struct mbuf *m); int if_vlantrunkinuse(if_t ifp); caddr_t if_getlladdr(if_t ifp); void *if_gethandle(u_char); void if_bpfmtap(if_t ifp, struct mbuf *m); void if_etherbpfmtap(if_t ifp, struct mbuf *m); void if_vlancap(if_t ifp); int if_setupmultiaddr(if_t ifp, void *mta, int *cnt, int max); int if_multiaddr_array(if_t ifp, void *mta, int *cnt, int max); int if_multiaddr_count(if_t ifp, int max); int if_multi_apply(struct ifnet *ifp, int (*filter)(void *, struct ifmultiaddr *, int), void *arg); int if_getamcount(if_t ifp); struct ifaddr * if_getifaddr(if_t ifp); /* Functions */ void if_setinitfn(if_t ifp, void (*)(void *)); void if_setioctlfn(if_t ifp, int (*)(if_t, u_long, caddr_t)); void if_setstartfn(if_t ifp, void (*)(if_t)); void if_settransmitfn(if_t ifp, if_transmit_fn_t); void if_setqflushfn(if_t ifp, if_qflush_fn_t); void if_setgetcounterfn(if_t ifp, if_get_counter_t); /* Revisit the below. These are inline functions originally */ int drbr_inuse_drv(if_t ifp, struct buf_ring *br); struct mbuf* drbr_dequeue_drv(if_t ifp, struct buf_ring *br); int drbr_needs_enqueue_drv(if_t ifp, struct buf_ring *br); int drbr_enqueue_drv(if_t ifp, struct buf_ring *br, struct mbuf *m); /* TSO */ void if_hw_tsomax_common(if_t ifp, struct ifnet_hw_tsomax *); int if_hw_tsomax_update(if_t ifp, struct ifnet_hw_tsomax *); #ifdef DEVICE_POLLING enum poll_cmd { POLL_ONLY, POLL_AND_CHECK_STATUS }; typedef int poll_handler_t(if_t ifp, enum poll_cmd cmd, int count); int ether_poll_register(poll_handler_t *h, if_t ifp); int ether_poll_deregister(if_t ifp); #endif /* DEVICE_POLLING */ #endif /* _KERNEL */ #include /* XXXAO: temporary unconditional include */ #endif /* !_NET_IF_VAR_H_ */ Index: projects/vnet/sys/net/pfvar.h =================================================================== --- projects/vnet/sys/net/pfvar.h (revision 302157) +++ projects/vnet/sys/net/pfvar.h (revision 302158) @@ -1,1758 +1,1757 @@ /* * Copyright (c) 2001 Daniel Hartmeier * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * * - Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * - 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 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 HOLDERS 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. * * $OpenBSD: pfvar.h,v 1.282 2009/01/29 15:12:28 pyr Exp $ * $FreeBSD$ */ #ifndef _NET_PFVAR_H_ #define _NET_PFVAR_H_ #include #include #include #include #include #include #include #include #include #include #include #include struct pf_addr { union { struct in_addr v4; struct in6_addr v6; u_int8_t addr8[16]; u_int16_t addr16[8]; u_int32_t addr32[4]; } pfa; /* 128-bit address */ #define v4 pfa.v4 #define v6 pfa.v6 #define addr8 pfa.addr8 #define addr16 pfa.addr16 #define addr32 pfa.addr32 }; #define PFI_AFLAG_NETWORK 0x01 #define PFI_AFLAG_BROADCAST 0x02 #define PFI_AFLAG_PEER 0x04 #define PFI_AFLAG_MODEMASK 0x07 #define PFI_AFLAG_NOALIAS 0x08 struct pf_addr_wrap { union { struct { struct pf_addr addr; struct pf_addr mask; } a; char ifname[IFNAMSIZ]; char tblname[PF_TABLE_NAME_SIZE]; } v; union { struct pfi_dynaddr *dyn; struct pfr_ktable *tbl; int dyncnt; int tblcnt; } p; u_int8_t type; /* PF_ADDR_* */ u_int8_t iflags; /* PFI_AFLAG_* */ }; #ifdef _KERNEL struct pfi_dynaddr { TAILQ_ENTRY(pfi_dynaddr) entry; struct pf_addr pfid_addr4; struct pf_addr pfid_mask4; struct pf_addr pfid_addr6; struct pf_addr pfid_mask6; struct pfr_ktable *pfid_kt; struct pfi_kif *pfid_kif; int pfid_net; /* mask or 128 */ int pfid_acnt4; /* address count IPv4 */ int pfid_acnt6; /* address count IPv6 */ sa_family_t pfid_af; /* rule af */ u_int8_t pfid_iflags; /* PFI_AFLAG_* */ }; /* * Address manipulation macros */ #define HTONL(x) (x) = htonl((__uint32_t)(x)) #define HTONS(x) (x) = htons((__uint16_t)(x)) #define NTOHL(x) (x) = ntohl((__uint32_t)(x)) #define NTOHS(x) (x) = ntohs((__uint16_t)(x)) #define PF_NAME "pf" #define PF_HASHROW_ASSERT(h) mtx_assert(&(h)->lock, MA_OWNED) #define PF_HASHROW_LOCK(h) mtx_lock(&(h)->lock) #define PF_HASHROW_UNLOCK(h) mtx_unlock(&(h)->lock) #define PF_STATE_LOCK(s) \ do { \ struct pf_idhash *_ih = &V_pf_idhash[PF_IDHASH(s)]; \ PF_HASHROW_LOCK(_ih); \ } while (0) #define PF_STATE_UNLOCK(s) \ do { \ struct pf_idhash *_ih = &V_pf_idhash[PF_IDHASH((s))]; \ PF_HASHROW_UNLOCK(_ih); \ } while (0) #ifdef INVARIANTS #define PF_STATE_LOCK_ASSERT(s) \ do { \ struct pf_idhash *_ih = &V_pf_idhash[PF_IDHASH(s)]; \ PF_HASHROW_ASSERT(_ih); \ } while (0) #else /* !INVARIANTS */ #define PF_STATE_LOCK_ASSERT(s) do {} while (0) #endif /* INVARIANTS */ extern struct mtx pf_unlnkdrules_mtx; #define PF_UNLNKDRULES_LOCK() mtx_lock(&pf_unlnkdrules_mtx) #define PF_UNLNKDRULES_UNLOCK() mtx_unlock(&pf_unlnkdrules_mtx) extern struct rwlock pf_rules_lock; #define PF_RULES_RLOCK() rw_rlock(&pf_rules_lock) #define PF_RULES_RUNLOCK() rw_runlock(&pf_rules_lock) #define PF_RULES_WLOCK() rw_wlock(&pf_rules_lock) #define PF_RULES_WUNLOCK() rw_wunlock(&pf_rules_lock) #define PF_RULES_ASSERT() rw_assert(&pf_rules_lock, RA_LOCKED) #define PF_RULES_RASSERT() rw_assert(&pf_rules_lock, RA_RLOCKED) #define PF_RULES_WASSERT() rw_assert(&pf_rules_lock, RA_WLOCKED) #define PF_MODVER 1 #define PFLOG_MODVER 1 #define PFSYNC_MODVER 1 #define PFLOG_MINVER 1 #define PFLOG_PREFVER PFLOG_MODVER #define PFLOG_MAXVER 1 #define PFSYNC_MINVER 1 #define PFSYNC_PREFVER PFSYNC_MODVER #define PFSYNC_MAXVER 1 #ifdef INET #ifndef INET6 #define PF_INET_ONLY #endif /* ! INET6 */ #endif /* INET */ #ifdef INET6 #ifndef INET #define PF_INET6_ONLY #endif /* ! INET */ #endif /* INET6 */ #ifdef INET #ifdef INET6 #define PF_INET_INET6 #endif /* INET6 */ #endif /* INET */ #else #define PF_INET_INET6 #endif /* _KERNEL */ /* Both IPv4 and IPv6 */ #ifdef PF_INET_INET6 #define PF_AEQ(a, b, c) \ ((c == AF_INET && (a)->addr32[0] == (b)->addr32[0]) || \ (c == AF_INET6 && (a)->addr32[3] == (b)->addr32[3] && \ (a)->addr32[2] == (b)->addr32[2] && \ (a)->addr32[1] == (b)->addr32[1] && \ (a)->addr32[0] == (b)->addr32[0])) \ #define PF_ANEQ(a, b, c) \ ((c == AF_INET && (a)->addr32[0] != (b)->addr32[0]) || \ (c == AF_INET6 && ((a)->addr32[0] != (b)->addr32[0] || \ (a)->addr32[1] != (b)->addr32[1] || \ (a)->addr32[2] != (b)->addr32[2] || \ (a)->addr32[3] != (b)->addr32[3]))) \ #define PF_AZERO(a, c) \ ((c == AF_INET && !(a)->addr32[0]) || \ (c == AF_INET6 && !(a)->addr32[0] && !(a)->addr32[1] && \ !(a)->addr32[2] && !(a)->addr32[3] )) \ #define PF_MATCHA(n, a, m, b, f) \ pf_match_addr(n, a, m, b, f) #define PF_ACPY(a, b, f) \ pf_addrcpy(a, b, f) #define PF_AINC(a, f) \ pf_addr_inc(a, f) #define PF_POOLMASK(a, b, c, d, f) \ pf_poolmask(a, b, c, d, f) #else /* Just IPv6 */ #ifdef PF_INET6_ONLY #define PF_AEQ(a, b, c) \ ((a)->addr32[3] == (b)->addr32[3] && \ (a)->addr32[2] == (b)->addr32[2] && \ (a)->addr32[1] == (b)->addr32[1] && \ (a)->addr32[0] == (b)->addr32[0]) \ #define PF_ANEQ(a, b, c) \ ((a)->addr32[3] != (b)->addr32[3] || \ (a)->addr32[2] != (b)->addr32[2] || \ (a)->addr32[1] != (b)->addr32[1] || \ (a)->addr32[0] != (b)->addr32[0]) \ #define PF_AZERO(a, c) \ (!(a)->addr32[0] && \ !(a)->addr32[1] && \ !(a)->addr32[2] && \ !(a)->addr32[3] ) \ #define PF_MATCHA(n, a, m, b, f) \ pf_match_addr(n, a, m, b, f) #define PF_ACPY(a, b, f) \ pf_addrcpy(a, b, f) #define PF_AINC(a, f) \ pf_addr_inc(a, f) #define PF_POOLMASK(a, b, c, d, f) \ pf_poolmask(a, b, c, d, f) #else /* Just IPv4 */ #ifdef PF_INET_ONLY #define PF_AEQ(a, b, c) \ ((a)->addr32[0] == (b)->addr32[0]) #define PF_ANEQ(a, b, c) \ ((a)->addr32[0] != (b)->addr32[0]) #define PF_AZERO(a, c) \ (!(a)->addr32[0]) #define PF_MATCHA(n, a, m, b, f) \ pf_match_addr(n, a, m, b, f) #define PF_ACPY(a, b, f) \ (a)->v4.s_addr = (b)->v4.s_addr #define PF_AINC(a, f) \ do { \ (a)->addr32[0] = htonl(ntohl((a)->addr32[0]) + 1); \ } while (0) #define PF_POOLMASK(a, b, c, d, f) \ do { \ (a)->addr32[0] = ((b)->addr32[0] & (c)->addr32[0]) | \ (((c)->addr32[0] ^ 0xffffffff ) & (d)->addr32[0]); \ } while (0) #endif /* PF_INET_ONLY */ #endif /* PF_INET6_ONLY */ #endif /* PF_INET_INET6 */ /* * XXX callers not FIB-aware in our version of pf yet. * OpenBSD fixed it later it seems, 2010/05/07 13:33:16 claudio. */ #define PF_MISMATCHAW(aw, x, af, neg, ifp, rtid) \ ( \ (((aw)->type == PF_ADDR_NOROUTE && \ pf_routable((x), (af), NULL, (rtid))) || \ (((aw)->type == PF_ADDR_URPFFAILED && (ifp) != NULL && \ pf_routable((x), (af), (ifp), (rtid))) || \ ((aw)->type == PF_ADDR_TABLE && \ !pfr_match_addr((aw)->p.tbl, (x), (af))) || \ ((aw)->type == PF_ADDR_DYNIFTL && \ !pfi_match_addr((aw)->p.dyn, (x), (af))) || \ ((aw)->type == PF_ADDR_RANGE && \ !pf_match_addr_range(&(aw)->v.a.addr, \ &(aw)->v.a.mask, (x), (af))) || \ ((aw)->type == PF_ADDR_ADDRMASK && \ !PF_AZERO(&(aw)->v.a.mask, (af)) && \ !PF_MATCHA(0, &(aw)->v.a.addr, \ &(aw)->v.a.mask, (x), (af))))) != \ (neg) \ ) struct pf_rule_uid { uid_t uid[2]; u_int8_t op; }; struct pf_rule_gid { uid_t gid[2]; u_int8_t op; }; struct pf_rule_addr { struct pf_addr_wrap addr; u_int16_t port[2]; u_int8_t neg; u_int8_t port_op; }; struct pf_pooladdr { struct pf_addr_wrap addr; TAILQ_ENTRY(pf_pooladdr) entries; char ifname[IFNAMSIZ]; struct pfi_kif *kif; }; TAILQ_HEAD(pf_palist, pf_pooladdr); struct pf_poolhashkey { union { u_int8_t key8[16]; u_int16_t key16[8]; u_int32_t key32[4]; } pfk; /* 128-bit hash key */ #define key8 pfk.key8 #define key16 pfk.key16 #define key32 pfk.key32 }; struct pf_pool { struct pf_palist list; struct pf_pooladdr *cur; struct pf_poolhashkey key; struct pf_addr counter; int tblidx; u_int16_t proxy_port[2]; u_int8_t opts; }; /* A packed Operating System description for fingerprinting */ typedef u_int32_t pf_osfp_t; #define PF_OSFP_ANY ((pf_osfp_t)0) #define PF_OSFP_UNKNOWN ((pf_osfp_t)-1) #define PF_OSFP_NOMATCH ((pf_osfp_t)-2) struct pf_osfp_entry { SLIST_ENTRY(pf_osfp_entry) fp_entry; pf_osfp_t fp_os; int fp_enflags; #define PF_OSFP_EXPANDED 0x001 /* expanded entry */ #define PF_OSFP_GENERIC 0x002 /* generic signature */ #define PF_OSFP_NODETAIL 0x004 /* no p0f details */ #define PF_OSFP_LEN 32 char fp_class_nm[PF_OSFP_LEN]; char fp_version_nm[PF_OSFP_LEN]; char fp_subtype_nm[PF_OSFP_LEN]; }; #define PF_OSFP_ENTRY_EQ(a, b) \ ((a)->fp_os == (b)->fp_os && \ memcmp((a)->fp_class_nm, (b)->fp_class_nm, PF_OSFP_LEN) == 0 && \ memcmp((a)->fp_version_nm, (b)->fp_version_nm, PF_OSFP_LEN) == 0 && \ memcmp((a)->fp_subtype_nm, (b)->fp_subtype_nm, PF_OSFP_LEN) == 0) /* handle pf_osfp_t packing */ #define _FP_RESERVED_BIT 1 /* For the special negative #defines */ #define _FP_UNUSED_BITS 1 #define _FP_CLASS_BITS 10 /* OS Class (Windows, Linux) */ #define _FP_VERSION_BITS 10 /* OS version (95, 98, NT, 2.4.54, 3.2) */ #define _FP_SUBTYPE_BITS 10 /* patch level (NT SP4, SP3, ECN patch) */ #define PF_OSFP_UNPACK(osfp, class, version, subtype) do { \ (class) = ((osfp) >> (_FP_VERSION_BITS+_FP_SUBTYPE_BITS)) & \ ((1 << _FP_CLASS_BITS) - 1); \ (version) = ((osfp) >> _FP_SUBTYPE_BITS) & \ ((1 << _FP_VERSION_BITS) - 1);\ (subtype) = (osfp) & ((1 << _FP_SUBTYPE_BITS) - 1); \ } while(0) #define PF_OSFP_PACK(osfp, class, version, subtype) do { \ (osfp) = ((class) & ((1 << _FP_CLASS_BITS) - 1)) << (_FP_VERSION_BITS \ + _FP_SUBTYPE_BITS); \ (osfp) |= ((version) & ((1 << _FP_VERSION_BITS) - 1)) << \ _FP_SUBTYPE_BITS; \ (osfp) |= (subtype) & ((1 << _FP_SUBTYPE_BITS) - 1); \ } while(0) /* the fingerprint of an OSes TCP SYN packet */ typedef u_int64_t pf_tcpopts_t; struct pf_os_fingerprint { SLIST_HEAD(pf_osfp_enlist, pf_osfp_entry) fp_oses; /* list of matches */ pf_tcpopts_t fp_tcpopts; /* packed TCP options */ u_int16_t fp_wsize; /* TCP window size */ u_int16_t fp_psize; /* ip->ip_len */ u_int16_t fp_mss; /* TCP MSS */ u_int16_t fp_flags; #define PF_OSFP_WSIZE_MOD 0x0001 /* Window modulus */ #define PF_OSFP_WSIZE_DC 0x0002 /* Window don't care */ #define PF_OSFP_WSIZE_MSS 0x0004 /* Window multiple of MSS */ #define PF_OSFP_WSIZE_MTU 0x0008 /* Window multiple of MTU */ #define PF_OSFP_PSIZE_MOD 0x0010 /* packet size modulus */ #define PF_OSFP_PSIZE_DC 0x0020 /* packet size don't care */ #define PF_OSFP_WSCALE 0x0040 /* TCP window scaling */ #define PF_OSFP_WSCALE_MOD 0x0080 /* TCP window scale modulus */ #define PF_OSFP_WSCALE_DC 0x0100 /* TCP window scale dont-care */ #define PF_OSFP_MSS 0x0200 /* TCP MSS */ #define PF_OSFP_MSS_MOD 0x0400 /* TCP MSS modulus */ #define PF_OSFP_MSS_DC 0x0800 /* TCP MSS dont-care */ #define PF_OSFP_DF 0x1000 /* IPv4 don't fragment bit */ #define PF_OSFP_TS0 0x2000 /* Zero timestamp */ #define PF_OSFP_INET6 0x4000 /* IPv6 */ u_int8_t fp_optcnt; /* TCP option count */ u_int8_t fp_wscale; /* TCP window scaling */ u_int8_t fp_ttl; /* IPv4 TTL */ #define PF_OSFP_MAXTTL_OFFSET 40 /* TCP options packing */ #define PF_OSFP_TCPOPT_NOP 0x0 /* TCP NOP option */ #define PF_OSFP_TCPOPT_WSCALE 0x1 /* TCP window scaling option */ #define PF_OSFP_TCPOPT_MSS 0x2 /* TCP max segment size opt */ #define PF_OSFP_TCPOPT_SACK 0x3 /* TCP SACK OK option */ #define PF_OSFP_TCPOPT_TS 0x4 /* TCP timestamp option */ #define PF_OSFP_TCPOPT_BITS 3 /* bits used by each option */ #define PF_OSFP_MAX_OPTS \ (sizeof(((struct pf_os_fingerprint *)0)->fp_tcpopts) * 8) \ / PF_OSFP_TCPOPT_BITS SLIST_ENTRY(pf_os_fingerprint) fp_next; }; struct pf_osfp_ioctl { struct pf_osfp_entry fp_os; pf_tcpopts_t fp_tcpopts; /* packed TCP options */ u_int16_t fp_wsize; /* TCP window size */ u_int16_t fp_psize; /* ip->ip_len */ u_int16_t fp_mss; /* TCP MSS */ u_int16_t fp_flags; u_int8_t fp_optcnt; /* TCP option count */ u_int8_t fp_wscale; /* TCP window scaling */ u_int8_t fp_ttl; /* IPv4 TTL */ int fp_getnum; /* DIOCOSFPGET number */ }; union pf_rule_ptr { struct pf_rule *ptr; u_int32_t nr; }; #define PF_ANCHOR_NAME_SIZE 64 struct pf_rule { struct pf_rule_addr src; struct pf_rule_addr dst; #define PF_SKIP_IFP 0 #define PF_SKIP_DIR 1 #define PF_SKIP_AF 2 #define PF_SKIP_PROTO 3 #define PF_SKIP_SRC_ADDR 4 #define PF_SKIP_SRC_PORT 5 #define PF_SKIP_DST_ADDR 6 #define PF_SKIP_DST_PORT 7 #define PF_SKIP_COUNT 8 union pf_rule_ptr skip[PF_SKIP_COUNT]; #define PF_RULE_LABEL_SIZE 64 char label[PF_RULE_LABEL_SIZE]; char ifname[IFNAMSIZ]; char qname[PF_QNAME_SIZE]; char pqname[PF_QNAME_SIZE]; #define PF_TAG_NAME_SIZE 64 char tagname[PF_TAG_NAME_SIZE]; char match_tagname[PF_TAG_NAME_SIZE]; char overload_tblname[PF_TABLE_NAME_SIZE]; TAILQ_ENTRY(pf_rule) entries; struct pf_pool rpool; u_int64_t evaluations; u_int64_t packets[2]; u_int64_t bytes[2]; struct pfi_kif *kif; struct pf_anchor *anchor; struct pfr_ktable *overload_tbl; pf_osfp_t os_fingerprint; int rtableid; u_int32_t timeout[PFTM_MAX]; u_int32_t max_states; u_int32_t max_src_nodes; u_int32_t max_src_states; u_int32_t max_src_conn; struct { u_int32_t limit; u_int32_t seconds; } max_src_conn_rate; u_int32_t qid; u_int32_t pqid; u_int32_t rt_listid; u_int32_t nr; u_int32_t prob; uid_t cuid; pid_t cpid; counter_u64_t states_cur; counter_u64_t states_tot; counter_u64_t src_nodes; u_int16_t return_icmp; u_int16_t return_icmp6; u_int16_t max_mss; u_int16_t tag; u_int16_t match_tag; u_int16_t scrub_flags; struct pf_rule_uid uid; struct pf_rule_gid gid; u_int32_t rule_flag; u_int8_t action; u_int8_t direction; u_int8_t log; u_int8_t logif; u_int8_t quick; u_int8_t ifnot; u_int8_t match_tag_not; u_int8_t natpass; #define PF_STATE_NORMAL 0x1 #define PF_STATE_MODULATE 0x2 #define PF_STATE_SYNPROXY 0x3 u_int8_t keep_state; sa_family_t af; u_int8_t proto; u_int8_t type; u_int8_t code; u_int8_t flags; u_int8_t flagset; u_int8_t min_ttl; u_int8_t allow_opts; u_int8_t rt; u_int8_t return_ttl; u_int8_t tos; u_int8_t set_tos; u_int8_t anchor_relative; u_int8_t anchor_wildcard; #define PF_FLUSH 0x01 #define PF_FLUSH_GLOBAL 0x02 u_int8_t flush; #define PF_PRIO_ZERO 0xff /* match "prio 0" packets */ #define PF_PRIO_MAX 7 u_int8_t prio; u_int8_t set_prio[2]; struct { struct pf_addr addr; u_int16_t port; } divert; uint64_t u_states_cur; uint64_t u_states_tot; uint64_t u_src_nodes; }; /* rule flags */ #define PFRULE_DROP 0x0000 #define PFRULE_RETURNRST 0x0001 #define PFRULE_FRAGMENT 0x0002 #define PFRULE_RETURNICMP 0x0004 #define PFRULE_RETURN 0x0008 #define PFRULE_NOSYNC 0x0010 #define PFRULE_SRCTRACK 0x0020 /* track source states */ #define PFRULE_RULESRCTRACK 0x0040 /* per rule */ #define PFRULE_REFS 0x0080 /* rule has references */ /* scrub flags */ #define PFRULE_NODF 0x0100 #define PFRULE_RANDOMID 0x0800 #define PFRULE_REASSEMBLE_TCP 0x1000 #define PFRULE_SET_TOS 0x2000 /* rule flags again */ #define PFRULE_IFBOUND 0x00010000 /* if-bound */ #define PFRULE_STATESLOPPY 0x00020000 /* sloppy state tracking */ #define PFSTATE_HIWAT 10000 /* default state table size */ #define PFSTATE_ADAPT_START 6000 /* default adaptive timeout start */ #define PFSTATE_ADAPT_END 12000 /* default adaptive timeout end */ struct pf_threshold { u_int32_t limit; #define PF_THRESHOLD_MULT 1000 #define PF_THRESHOLD_MAX 0xffffffff / PF_THRESHOLD_MULT u_int32_t seconds; u_int32_t count; u_int32_t last; }; struct pf_src_node { LIST_ENTRY(pf_src_node) entry; struct pf_addr addr; struct pf_addr raddr; union pf_rule_ptr rule; struct pfi_kif *kif; u_int64_t bytes[2]; u_int64_t packets[2]; u_int32_t states; u_int32_t conn; struct pf_threshold conn_rate; u_int32_t creation; u_int32_t expire; sa_family_t af; u_int8_t ruletype; }; #define PFSNODE_HIWAT 10000 /* default source node table size */ struct pf_state_scrub { struct timeval pfss_last; /* time received last packet */ u_int32_t pfss_tsecr; /* last echoed timestamp */ u_int32_t pfss_tsval; /* largest timestamp */ u_int32_t pfss_tsval0; /* original timestamp */ u_int16_t pfss_flags; #define PFSS_TIMESTAMP 0x0001 /* modulate timestamp */ #define PFSS_PAWS 0x0010 /* stricter PAWS checks */ #define PFSS_PAWS_IDLED 0x0020 /* was idle too long. no PAWS */ #define PFSS_DATA_TS 0x0040 /* timestamp on data packets */ #define PFSS_DATA_NOTS 0x0080 /* no timestamp on data packets */ u_int8_t pfss_ttl; /* stashed TTL */ u_int8_t pad; u_int32_t pfss_ts_mod; /* timestamp modulation */ }; struct pf_state_host { struct pf_addr addr; u_int16_t port; u_int16_t pad; }; struct pf_state_peer { struct pf_state_scrub *scrub; /* state is scrubbed */ u_int32_t seqlo; /* Max sequence number sent */ u_int32_t seqhi; /* Max the other end ACKd + win */ u_int32_t seqdiff; /* Sequence number modulator */ u_int16_t max_win; /* largest window (pre scaling) */ u_int16_t mss; /* Maximum segment size option */ u_int8_t state; /* active state level */ u_int8_t wscale; /* window scaling factor */ u_int8_t tcp_est; /* Did we reach TCPS_ESTABLISHED */ u_int8_t pad[1]; }; /* Keep synced with struct pf_state_key. */ struct pf_state_key_cmp { struct pf_addr addr[2]; u_int16_t port[2]; sa_family_t af; u_int8_t proto; u_int8_t pad[2]; }; struct pf_state_key { struct pf_addr addr[2]; u_int16_t port[2]; sa_family_t af; u_int8_t proto; u_int8_t pad[2]; LIST_ENTRY(pf_state_key) entry; TAILQ_HEAD(, pf_state) states[2]; }; /* Keep synced with struct pf_state. */ struct pf_state_cmp { u_int64_t id; u_int32_t creatorid; u_int8_t direction; u_int8_t pad[3]; }; struct pf_state { u_int64_t id; u_int32_t creatorid; u_int8_t direction; u_int8_t pad[3]; u_int refs; TAILQ_ENTRY(pf_state) sync_list; TAILQ_ENTRY(pf_state) key_list[2]; LIST_ENTRY(pf_state) entry; struct pf_state_peer src; struct pf_state_peer dst; union pf_rule_ptr rule; union pf_rule_ptr anchor; union pf_rule_ptr nat_rule; struct pf_addr rt_addr; struct pf_state_key *key[2]; /* addresses stack and wire */ struct pfi_kif *kif; struct pfi_kif *rt_kif; struct pf_src_node *src_node; struct pf_src_node *nat_src_node; u_int64_t packets[2]; u_int64_t bytes[2]; u_int32_t creation; u_int32_t expire; u_int32_t pfsync_time; u_int16_t tag; u_int8_t log; u_int8_t state_flags; #define PFSTATE_ALLOWOPTS 0x01 #define PFSTATE_SLOPPY 0x02 /* was PFSTATE_PFLOW 0x04 */ #define PFSTATE_NOSYNC 0x08 #define PFSTATE_ACK 0x10 #define PFSTATE_SETPRIO 0x0200 #define PFSTATE_SETMASK (PFSTATE_SETPRIO) u_int8_t timeout; u_int8_t sync_state; /* PFSYNC_S_x */ /* XXX */ u_int8_t sync_updates; u_int8_t _tail[3]; }; /* * Unified state structures for pulling states out of the kernel * used by pfsync(4) and the pf(4) ioctl. */ struct pfsync_state_scrub { u_int16_t pfss_flags; u_int8_t pfss_ttl; /* stashed TTL */ #define PFSYNC_SCRUB_FLAG_VALID 0x01 u_int8_t scrub_flag; u_int32_t pfss_ts_mod; /* timestamp modulation */ } __packed; struct pfsync_state_peer { struct pfsync_state_scrub scrub; /* state is scrubbed */ u_int32_t seqlo; /* Max sequence number sent */ u_int32_t seqhi; /* Max the other end ACKd + win */ u_int32_t seqdiff; /* Sequence number modulator */ u_int16_t max_win; /* largest window (pre scaling) */ u_int16_t mss; /* Maximum segment size option */ u_int8_t state; /* active state level */ u_int8_t wscale; /* window scaling factor */ u_int8_t pad[6]; } __packed; struct pfsync_state_key { struct pf_addr addr[2]; u_int16_t port[2]; }; struct pfsync_state { u_int64_t id; char ifname[IFNAMSIZ]; struct pfsync_state_key key[2]; struct pfsync_state_peer src; struct pfsync_state_peer dst; struct pf_addr rt_addr; u_int32_t rule; u_int32_t anchor; u_int32_t nat_rule; u_int32_t creation; u_int32_t expire; u_int32_t packets[2][2]; u_int32_t bytes[2][2]; u_int32_t creatorid; sa_family_t af; u_int8_t proto; u_int8_t direction; u_int8_t __spare[2]; u_int8_t log; u_int8_t state_flags; u_int8_t timeout; u_int8_t sync_flags; u_int8_t updates; } __packed; #ifdef _KERNEL /* pfsync */ typedef int pfsync_state_import_t(struct pfsync_state *, u_int8_t); typedef void pfsync_insert_state_t(struct pf_state *); typedef void pfsync_update_state_t(struct pf_state *); typedef void pfsync_delete_state_t(struct pf_state *); typedef void pfsync_clear_states_t(u_int32_t, const char *); typedef int pfsync_defer_t(struct pf_state *, struct mbuf *); extern pfsync_state_import_t *pfsync_state_import_ptr; extern pfsync_insert_state_t *pfsync_insert_state_ptr; extern pfsync_update_state_t *pfsync_update_state_ptr; extern pfsync_delete_state_t *pfsync_delete_state_ptr; extern pfsync_clear_states_t *pfsync_clear_states_ptr; extern pfsync_defer_t *pfsync_defer_ptr; void pfsync_state_export(struct pfsync_state *, struct pf_state *); /* pflog */ struct pf_ruleset; struct pf_pdesc; typedef int pflog_packet_t(struct pfi_kif *, struct mbuf *, sa_family_t, u_int8_t, u_int8_t, struct pf_rule *, struct pf_rule *, struct pf_ruleset *, struct pf_pdesc *, int); extern pflog_packet_t *pflog_packet_ptr; -#define V_pf_end_threads VNET(pf_end_threads) #endif /* _KERNEL */ #define PFSYNC_FLAG_SRCNODE 0x04 #define PFSYNC_FLAG_NATSRCNODE 0x08 /* for copies to/from network byte order */ /* ioctl interface also uses network byte order */ #define pf_state_peer_hton(s,d) do { \ (d)->seqlo = htonl((s)->seqlo); \ (d)->seqhi = htonl((s)->seqhi); \ (d)->seqdiff = htonl((s)->seqdiff); \ (d)->max_win = htons((s)->max_win); \ (d)->mss = htons((s)->mss); \ (d)->state = (s)->state; \ (d)->wscale = (s)->wscale; \ if ((s)->scrub) { \ (d)->scrub.pfss_flags = \ htons((s)->scrub->pfss_flags & PFSS_TIMESTAMP); \ (d)->scrub.pfss_ttl = (s)->scrub->pfss_ttl; \ (d)->scrub.pfss_ts_mod = htonl((s)->scrub->pfss_ts_mod);\ (d)->scrub.scrub_flag = PFSYNC_SCRUB_FLAG_VALID; \ } \ } while (0) #define pf_state_peer_ntoh(s,d) do { \ (d)->seqlo = ntohl((s)->seqlo); \ (d)->seqhi = ntohl((s)->seqhi); \ (d)->seqdiff = ntohl((s)->seqdiff); \ (d)->max_win = ntohs((s)->max_win); \ (d)->mss = ntohs((s)->mss); \ (d)->state = (s)->state; \ (d)->wscale = (s)->wscale; \ if ((s)->scrub.scrub_flag == PFSYNC_SCRUB_FLAG_VALID && \ (d)->scrub != NULL) { \ (d)->scrub->pfss_flags = \ ntohs((s)->scrub.pfss_flags) & PFSS_TIMESTAMP; \ (d)->scrub->pfss_ttl = (s)->scrub.pfss_ttl; \ (d)->scrub->pfss_ts_mod = ntohl((s)->scrub.pfss_ts_mod);\ } \ } while (0) #define pf_state_counter_hton(s,d) do { \ d[0] = htonl((s>>32)&0xffffffff); \ d[1] = htonl(s&0xffffffff); \ } while (0) #define pf_state_counter_from_pfsync(s) \ (((u_int64_t)(s[0])<<32) | (u_int64_t)(s[1])) #define pf_state_counter_ntoh(s,d) do { \ d = ntohl(s[0]); \ d = d<<32; \ d += ntohl(s[1]); \ } while (0) TAILQ_HEAD(pf_rulequeue, pf_rule); struct pf_anchor; struct pf_ruleset { struct { struct pf_rulequeue queues[2]; struct { struct pf_rulequeue *ptr; struct pf_rule **ptr_array; u_int32_t rcount; u_int32_t ticket; int open; } active, inactive; } rules[PF_RULESET_MAX]; struct pf_anchor *anchor; u_int32_t tticket; int tables; int topen; }; RB_HEAD(pf_anchor_global, pf_anchor); RB_HEAD(pf_anchor_node, pf_anchor); struct pf_anchor { RB_ENTRY(pf_anchor) entry_global; RB_ENTRY(pf_anchor) entry_node; struct pf_anchor *parent; struct pf_anchor_node children; char name[PF_ANCHOR_NAME_SIZE]; char path[MAXPATHLEN]; struct pf_ruleset ruleset; int refcnt; /* anchor rules */ int match; /* XXX: used for pfctl black magic */ }; RB_PROTOTYPE(pf_anchor_global, pf_anchor, entry_global, pf_anchor_compare); RB_PROTOTYPE(pf_anchor_node, pf_anchor, entry_node, pf_anchor_compare); #define PF_RESERVED_ANCHOR "_pf" #define PFR_TFLAG_PERSIST 0x00000001 #define PFR_TFLAG_CONST 0x00000002 #define PFR_TFLAG_ACTIVE 0x00000004 #define PFR_TFLAG_INACTIVE 0x00000008 #define PFR_TFLAG_REFERENCED 0x00000010 #define PFR_TFLAG_REFDANCHOR 0x00000020 #define PFR_TFLAG_COUNTERS 0x00000040 /* Adjust masks below when adding flags. */ #define PFR_TFLAG_USRMASK (PFR_TFLAG_PERSIST | \ PFR_TFLAG_CONST | \ PFR_TFLAG_COUNTERS) #define PFR_TFLAG_SETMASK (PFR_TFLAG_ACTIVE | \ PFR_TFLAG_INACTIVE | \ PFR_TFLAG_REFERENCED | \ PFR_TFLAG_REFDANCHOR) #define PFR_TFLAG_ALLMASK (PFR_TFLAG_PERSIST | \ PFR_TFLAG_CONST | \ PFR_TFLAG_ACTIVE | \ PFR_TFLAG_INACTIVE | \ PFR_TFLAG_REFERENCED | \ PFR_TFLAG_REFDANCHOR | \ PFR_TFLAG_COUNTERS) struct pf_anchor_stackframe; struct pfr_table { char pfrt_anchor[MAXPATHLEN]; char pfrt_name[PF_TABLE_NAME_SIZE]; u_int32_t pfrt_flags; u_int8_t pfrt_fback; }; enum { PFR_FB_NONE, PFR_FB_MATCH, PFR_FB_ADDED, PFR_FB_DELETED, PFR_FB_CHANGED, PFR_FB_CLEARED, PFR_FB_DUPLICATE, PFR_FB_NOTMATCH, PFR_FB_CONFLICT, PFR_FB_NOCOUNT, PFR_FB_MAX }; struct pfr_addr { union { struct in_addr _pfra_ip4addr; struct in6_addr _pfra_ip6addr; } pfra_u; u_int8_t pfra_af; u_int8_t pfra_net; u_int8_t pfra_not; u_int8_t pfra_fback; }; #define pfra_ip4addr pfra_u._pfra_ip4addr #define pfra_ip6addr pfra_u._pfra_ip6addr enum { PFR_DIR_IN, PFR_DIR_OUT, PFR_DIR_MAX }; enum { PFR_OP_BLOCK, PFR_OP_PASS, PFR_OP_ADDR_MAX, PFR_OP_TABLE_MAX }; #define PFR_OP_XPASS PFR_OP_ADDR_MAX struct pfr_astats { struct pfr_addr pfras_a; u_int64_t pfras_packets[PFR_DIR_MAX][PFR_OP_ADDR_MAX]; u_int64_t pfras_bytes[PFR_DIR_MAX][PFR_OP_ADDR_MAX]; long pfras_tzero; }; enum { PFR_REFCNT_RULE, PFR_REFCNT_ANCHOR, PFR_REFCNT_MAX }; struct pfr_tstats { struct pfr_table pfrts_t; u_int64_t pfrts_packets[PFR_DIR_MAX][PFR_OP_TABLE_MAX]; u_int64_t pfrts_bytes[PFR_DIR_MAX][PFR_OP_TABLE_MAX]; u_int64_t pfrts_match; u_int64_t pfrts_nomatch; long pfrts_tzero; int pfrts_cnt; int pfrts_refcnt[PFR_REFCNT_MAX]; }; #define pfrts_name pfrts_t.pfrt_name #define pfrts_flags pfrts_t.pfrt_flags #ifndef _SOCKADDR_UNION_DEFINED #define _SOCKADDR_UNION_DEFINED union sockaddr_union { struct sockaddr sa; struct sockaddr_in sin; struct sockaddr_in6 sin6; }; #endif /* _SOCKADDR_UNION_DEFINED */ struct pfr_kcounters { u_int64_t pfrkc_packets[PFR_DIR_MAX][PFR_OP_ADDR_MAX]; u_int64_t pfrkc_bytes[PFR_DIR_MAX][PFR_OP_ADDR_MAX]; }; SLIST_HEAD(pfr_kentryworkq, pfr_kentry); struct pfr_kentry { struct radix_node pfrke_node[2]; union sockaddr_union pfrke_sa; SLIST_ENTRY(pfr_kentry) pfrke_workq; struct pfr_kcounters *pfrke_counters; long pfrke_tzero; u_int8_t pfrke_af; u_int8_t pfrke_net; u_int8_t pfrke_not; u_int8_t pfrke_mark; }; SLIST_HEAD(pfr_ktableworkq, pfr_ktable); RB_HEAD(pfr_ktablehead, pfr_ktable); struct pfr_ktable { struct pfr_tstats pfrkt_ts; RB_ENTRY(pfr_ktable) pfrkt_tree; SLIST_ENTRY(pfr_ktable) pfrkt_workq; struct radix_node_head *pfrkt_ip4; struct radix_node_head *pfrkt_ip6; struct pfr_ktable *pfrkt_shadow; struct pfr_ktable *pfrkt_root; struct pf_ruleset *pfrkt_rs; long pfrkt_larg; int pfrkt_nflags; }; #define pfrkt_t pfrkt_ts.pfrts_t #define pfrkt_name pfrkt_t.pfrt_name #define pfrkt_anchor pfrkt_t.pfrt_anchor #define pfrkt_ruleset pfrkt_t.pfrt_ruleset #define pfrkt_flags pfrkt_t.pfrt_flags #define pfrkt_cnt pfrkt_ts.pfrts_cnt #define pfrkt_refcnt pfrkt_ts.pfrts_refcnt #define pfrkt_packets pfrkt_ts.pfrts_packets #define pfrkt_bytes pfrkt_ts.pfrts_bytes #define pfrkt_match pfrkt_ts.pfrts_match #define pfrkt_nomatch pfrkt_ts.pfrts_nomatch #define pfrkt_tzero pfrkt_ts.pfrts_tzero /* keep synced with pfi_kif, used in RB_FIND */ struct pfi_kif_cmp { char pfik_name[IFNAMSIZ]; }; struct pfi_kif { char pfik_name[IFNAMSIZ]; union { RB_ENTRY(pfi_kif) _pfik_tree; LIST_ENTRY(pfi_kif) _pfik_list; } _pfik_glue; #define pfik_tree _pfik_glue._pfik_tree #define pfik_list _pfik_glue._pfik_list u_int64_t pfik_packets[2][2][2]; u_int64_t pfik_bytes[2][2][2]; u_int32_t pfik_tzero; u_int pfik_flags; struct ifnet *pfik_ifp; struct ifg_group *pfik_group; u_int pfik_rulerefs; TAILQ_HEAD(, pfi_dynaddr) pfik_dynaddrs; }; #define PFI_IFLAG_REFS 0x0001 /* has state references */ #define PFI_IFLAG_SKIP 0x0100 /* skip filtering on interface */ struct pf_pdesc { struct { int done; uid_t uid; gid_t gid; } lookup; u_int64_t tot_len; /* Make Mickey money */ union { struct tcphdr *tcp; struct udphdr *udp; struct icmp *icmp; #ifdef INET6 struct icmp6_hdr *icmp6; #endif /* INET6 */ void *any; } hdr; struct pf_rule *nat_rule; /* nat/rdr rule applied to packet */ struct pf_addr *src; /* src address */ struct pf_addr *dst; /* dst address */ u_int16_t *sport; u_int16_t *dport; struct pf_mtag *pf_mtag; u_int32_t p_len; /* total length of payload */ u_int16_t *ip_sum; u_int16_t *proto_sum; u_int16_t flags; /* Let SCRUB trigger behavior in * state code. Easier than tags */ #define PFDESC_TCP_NORM 0x0001 /* TCP shall be statefully scrubbed */ #define PFDESC_IP_REAS 0x0002 /* IP frags would've been reassembled */ sa_family_t af; u_int8_t proto; u_int8_t tos; u_int8_t dir; /* direction */ u_int8_t sidx; /* key index for source */ u_int8_t didx; /* key index for destination */ }; /* flags for RDR options */ #define PF_DPORT_RANGE 0x01 /* Dest port uses range */ #define PF_RPORT_RANGE 0x02 /* RDR'ed port uses range */ /* UDP state enumeration */ #define PFUDPS_NO_TRAFFIC 0 #define PFUDPS_SINGLE 1 #define PFUDPS_MULTIPLE 2 #define PFUDPS_NSTATES 3 /* number of state levels */ #define PFUDPS_NAMES { \ "NO_TRAFFIC", \ "SINGLE", \ "MULTIPLE", \ NULL \ } /* Other protocol state enumeration */ #define PFOTHERS_NO_TRAFFIC 0 #define PFOTHERS_SINGLE 1 #define PFOTHERS_MULTIPLE 2 #define PFOTHERS_NSTATES 3 /* number of state levels */ #define PFOTHERS_NAMES { \ "NO_TRAFFIC", \ "SINGLE", \ "MULTIPLE", \ NULL \ } #define ACTION_SET(a, x) \ do { \ if ((a) != NULL) \ *(a) = (x); \ } while (0) #define REASON_SET(a, x) \ do { \ if ((a) != NULL) \ *(a) = (x); \ if (x < PFRES_MAX) \ counter_u64_add(V_pf_status.counters[x], 1); \ } while (0) struct pf_kstatus { counter_u64_t counters[PFRES_MAX]; /* reason for passing/dropping */ counter_u64_t lcounters[LCNT_MAX]; /* limit counters */ counter_u64_t fcounters[FCNT_MAX]; /* state operation counters */ counter_u64_t scounters[SCNT_MAX]; /* src_node operation counters */ uint32_t states; uint32_t src_nodes; uint32_t running; uint32_t since; uint32_t debug; uint32_t hostid; char ifname[IFNAMSIZ]; uint8_t pf_chksum[PF_MD5_DIGEST_LENGTH]; }; struct pf_divert { union { struct in_addr ipv4; struct in6_addr ipv6; } addr; u_int16_t port; }; #define PFFRAG_FRENT_HIWAT 5000 /* Number of fragment entries */ #define PFR_KENTRY_HIWAT 200000 /* Number of table entries */ /* * ioctl parameter structures */ struct pfioc_pooladdr { u_int32_t action; u_int32_t ticket; u_int32_t nr; u_int32_t r_num; u_int8_t r_action; u_int8_t r_last; u_int8_t af; char anchor[MAXPATHLEN]; struct pf_pooladdr addr; }; struct pfioc_rule { u_int32_t action; u_int32_t ticket; u_int32_t pool_ticket; u_int32_t nr; char anchor[MAXPATHLEN]; char anchor_call[MAXPATHLEN]; struct pf_rule rule; }; struct pfioc_natlook { struct pf_addr saddr; struct pf_addr daddr; struct pf_addr rsaddr; struct pf_addr rdaddr; u_int16_t sport; u_int16_t dport; u_int16_t rsport; u_int16_t rdport; sa_family_t af; u_int8_t proto; u_int8_t direction; }; struct pfioc_state { struct pfsync_state state; }; struct pfioc_src_node_kill { sa_family_t psnk_af; struct pf_rule_addr psnk_src; struct pf_rule_addr psnk_dst; u_int psnk_killed; }; struct pfioc_state_kill { struct pf_state_cmp psk_pfcmp; sa_family_t psk_af; int psk_proto; struct pf_rule_addr psk_src; struct pf_rule_addr psk_dst; char psk_ifname[IFNAMSIZ]; char psk_label[PF_RULE_LABEL_SIZE]; u_int psk_killed; }; struct pfioc_states { int ps_len; union { caddr_t psu_buf; struct pfsync_state *psu_states; } ps_u; #define ps_buf ps_u.psu_buf #define ps_states ps_u.psu_states }; struct pfioc_src_nodes { int psn_len; union { caddr_t psu_buf; struct pf_src_node *psu_src_nodes; } psn_u; #define psn_buf psn_u.psu_buf #define psn_src_nodes psn_u.psu_src_nodes }; struct pfioc_if { char ifname[IFNAMSIZ]; }; struct pfioc_tm { int timeout; int seconds; }; struct pfioc_limit { int index; unsigned limit; }; struct pfioc_altq { u_int32_t action; u_int32_t ticket; u_int32_t nr; struct pf_altq altq; }; struct pfioc_qstats { u_int32_t ticket; u_int32_t nr; void *buf; int nbytes; u_int8_t scheduler; }; struct pfioc_ruleset { u_int32_t nr; char path[MAXPATHLEN]; char name[PF_ANCHOR_NAME_SIZE]; }; #define PF_RULESET_ALTQ (PF_RULESET_MAX) #define PF_RULESET_TABLE (PF_RULESET_MAX+1) struct pfioc_trans { int size; /* number of elements */ int esize; /* size of each element in bytes */ struct pfioc_trans_e { int rs_num; char anchor[MAXPATHLEN]; u_int32_t ticket; } *array; }; #define PFR_FLAG_ATOMIC 0x00000001 /* unused */ #define PFR_FLAG_DUMMY 0x00000002 #define PFR_FLAG_FEEDBACK 0x00000004 #define PFR_FLAG_CLSTATS 0x00000008 #define PFR_FLAG_ADDRSTOO 0x00000010 #define PFR_FLAG_REPLACE 0x00000020 #define PFR_FLAG_ALLRSETS 0x00000040 #define PFR_FLAG_ALLMASK 0x0000007F #ifdef _KERNEL #define PFR_FLAG_USERIOCTL 0x10000000 #endif struct pfioc_table { struct pfr_table pfrio_table; void *pfrio_buffer; int pfrio_esize; int pfrio_size; int pfrio_size2; int pfrio_nadd; int pfrio_ndel; int pfrio_nchange; int pfrio_flags; u_int32_t pfrio_ticket; }; #define pfrio_exists pfrio_nadd #define pfrio_nzero pfrio_nadd #define pfrio_nmatch pfrio_nadd #define pfrio_naddr pfrio_size2 #define pfrio_setflag pfrio_size2 #define pfrio_clrflag pfrio_nadd struct pfioc_iface { char pfiio_name[IFNAMSIZ]; void *pfiio_buffer; int pfiio_esize; int pfiio_size; int pfiio_nzero; int pfiio_flags; }; /* * ioctl operations */ #define DIOCSTART _IO ('D', 1) #define DIOCSTOP _IO ('D', 2) #define DIOCADDRULE _IOWR('D', 4, struct pfioc_rule) #define DIOCGETRULES _IOWR('D', 6, struct pfioc_rule) #define DIOCGETRULE _IOWR('D', 7, struct pfioc_rule) /* XXX cut 8 - 17 */ #define DIOCCLRSTATES _IOWR('D', 18, struct pfioc_state_kill) #define DIOCGETSTATE _IOWR('D', 19, struct pfioc_state) #define DIOCSETSTATUSIF _IOWR('D', 20, struct pfioc_if) #define DIOCGETSTATUS _IOWR('D', 21, struct pf_status) #define DIOCCLRSTATUS _IO ('D', 22) #define DIOCNATLOOK _IOWR('D', 23, struct pfioc_natlook) #define DIOCSETDEBUG _IOWR('D', 24, u_int32_t) #define DIOCGETSTATES _IOWR('D', 25, struct pfioc_states) #define DIOCCHANGERULE _IOWR('D', 26, struct pfioc_rule) /* XXX cut 26 - 28 */ #define DIOCSETTIMEOUT _IOWR('D', 29, struct pfioc_tm) #define DIOCGETTIMEOUT _IOWR('D', 30, struct pfioc_tm) #define DIOCADDSTATE _IOWR('D', 37, struct pfioc_state) #define DIOCCLRRULECTRS _IO ('D', 38) #define DIOCGETLIMIT _IOWR('D', 39, struct pfioc_limit) #define DIOCSETLIMIT _IOWR('D', 40, struct pfioc_limit) #define DIOCKILLSTATES _IOWR('D', 41, struct pfioc_state_kill) #define DIOCSTARTALTQ _IO ('D', 42) #define DIOCSTOPALTQ _IO ('D', 43) #define DIOCADDALTQ _IOWR('D', 45, struct pfioc_altq) #define DIOCGETALTQS _IOWR('D', 47, struct pfioc_altq) #define DIOCGETALTQ _IOWR('D', 48, struct pfioc_altq) #define DIOCCHANGEALTQ _IOWR('D', 49, struct pfioc_altq) #define DIOCGETQSTATS _IOWR('D', 50, struct pfioc_qstats) #define DIOCBEGINADDRS _IOWR('D', 51, struct pfioc_pooladdr) #define DIOCADDADDR _IOWR('D', 52, struct pfioc_pooladdr) #define DIOCGETADDRS _IOWR('D', 53, struct pfioc_pooladdr) #define DIOCGETADDR _IOWR('D', 54, struct pfioc_pooladdr) #define DIOCCHANGEADDR _IOWR('D', 55, struct pfioc_pooladdr) /* XXX cut 55 - 57 */ #define DIOCGETRULESETS _IOWR('D', 58, struct pfioc_ruleset) #define DIOCGETRULESET _IOWR('D', 59, struct pfioc_ruleset) #define DIOCRCLRTABLES _IOWR('D', 60, struct pfioc_table) #define DIOCRADDTABLES _IOWR('D', 61, struct pfioc_table) #define DIOCRDELTABLES _IOWR('D', 62, struct pfioc_table) #define DIOCRGETTABLES _IOWR('D', 63, struct pfioc_table) #define DIOCRGETTSTATS _IOWR('D', 64, struct pfioc_table) #define DIOCRCLRTSTATS _IOWR('D', 65, struct pfioc_table) #define DIOCRCLRADDRS _IOWR('D', 66, struct pfioc_table) #define DIOCRADDADDRS _IOWR('D', 67, struct pfioc_table) #define DIOCRDELADDRS _IOWR('D', 68, struct pfioc_table) #define DIOCRSETADDRS _IOWR('D', 69, struct pfioc_table) #define DIOCRGETADDRS _IOWR('D', 70, struct pfioc_table) #define DIOCRGETASTATS _IOWR('D', 71, struct pfioc_table) #define DIOCRCLRASTATS _IOWR('D', 72, struct pfioc_table) #define DIOCRTSTADDRS _IOWR('D', 73, struct pfioc_table) #define DIOCRSETTFLAGS _IOWR('D', 74, struct pfioc_table) #define DIOCRINADEFINE _IOWR('D', 77, struct pfioc_table) #define DIOCOSFPFLUSH _IO('D', 78) #define DIOCOSFPADD _IOWR('D', 79, struct pf_osfp_ioctl) #define DIOCOSFPGET _IOWR('D', 80, struct pf_osfp_ioctl) #define DIOCXBEGIN _IOWR('D', 81, struct pfioc_trans) #define DIOCXCOMMIT _IOWR('D', 82, struct pfioc_trans) #define DIOCXROLLBACK _IOWR('D', 83, struct pfioc_trans) #define DIOCGETSRCNODES _IOWR('D', 84, struct pfioc_src_nodes) #define DIOCCLRSRCNODES _IO('D', 85) #define DIOCSETHOSTID _IOWR('D', 86, u_int32_t) #define DIOCIGETIFACES _IOWR('D', 87, struct pfioc_iface) #define DIOCSETIFFLAG _IOWR('D', 89, struct pfioc_iface) #define DIOCCLRIFFLAG _IOWR('D', 90, struct pfioc_iface) #define DIOCKILLSRCNODES _IOWR('D', 91, struct pfioc_src_node_kill) struct pf_ifspeed { char ifname[IFNAMSIZ]; u_int32_t baudrate; }; #define DIOCGIFSPEED _IOWR('D', 92, struct pf_ifspeed) #ifdef _KERNEL LIST_HEAD(pf_src_node_list, pf_src_node); struct pf_srchash { struct pf_src_node_list nodes; struct mtx lock; }; struct pf_keyhash { LIST_HEAD(, pf_state_key) keys; struct mtx lock; }; struct pf_idhash { LIST_HEAD(, pf_state) states; struct mtx lock; }; extern u_long pf_hashmask; extern u_long pf_srchashmask; #define PF_HASHSIZ (32768) VNET_DECLARE(struct pf_keyhash *, pf_keyhash); VNET_DECLARE(struct pf_idhash *, pf_idhash); #define V_pf_keyhash VNET(pf_keyhash) #define V_pf_idhash VNET(pf_idhash) VNET_DECLARE(struct pf_srchash *, pf_srchash); #define V_pf_srchash VNET(pf_srchash) #define PF_IDHASH(s) (be64toh((s)->id) % (pf_hashmask + 1)) VNET_DECLARE(void *, pf_swi_cookie); #define V_pf_swi_cookie VNET(pf_swi_cookie) VNET_DECLARE(uint64_t, pf_stateid[MAXCPU]); #define V_pf_stateid VNET(pf_stateid) TAILQ_HEAD(pf_altqqueue, pf_altq); VNET_DECLARE(struct pf_altqqueue, pf_altqs[2]); #define V_pf_altqs VNET(pf_altqs) VNET_DECLARE(struct pf_palist, pf_pabuf); #define V_pf_pabuf VNET(pf_pabuf) VNET_DECLARE(u_int32_t, ticket_altqs_active); #define V_ticket_altqs_active VNET(ticket_altqs_active) VNET_DECLARE(u_int32_t, ticket_altqs_inactive); #define V_ticket_altqs_inactive VNET(ticket_altqs_inactive) VNET_DECLARE(int, altqs_inactive_open); #define V_altqs_inactive_open VNET(altqs_inactive_open) VNET_DECLARE(u_int32_t, ticket_pabuf); #define V_ticket_pabuf VNET(ticket_pabuf) VNET_DECLARE(struct pf_altqqueue *, pf_altqs_active); #define V_pf_altqs_active VNET(pf_altqs_active) VNET_DECLARE(struct pf_altqqueue *, pf_altqs_inactive); #define V_pf_altqs_inactive VNET(pf_altqs_inactive) VNET_DECLARE(struct pf_rulequeue, pf_unlinked_rules); #define V_pf_unlinked_rules VNET(pf_unlinked_rules) void pf_initialize(void); void pf_mtag_initialize(void); void pf_mtag_cleanup(void); void pf_cleanup(void); struct pf_mtag *pf_get_mtag(struct mbuf *); extern void pf_calc_skip_steps(struct pf_rulequeue *); #ifdef ALTQ extern void pf_altq_ifnet_event(struct ifnet *, int); #endif VNET_DECLARE(uma_zone_t, pf_state_z); #define V_pf_state_z VNET(pf_state_z) VNET_DECLARE(uma_zone_t, pf_state_key_z); #define V_pf_state_key_z VNET(pf_state_key_z) VNET_DECLARE(uma_zone_t, pf_state_scrub_z); #define V_pf_state_scrub_z VNET(pf_state_scrub_z) extern void pf_purge_thread(void *); extern void pf_unload_vnet_purge(void); extern void pf_intr(void *); extern void pf_purge_expired_src_nodes(void); extern int pf_unlink_state(struct pf_state *, u_int); #define PF_ENTER_LOCKED 0x00000001 #define PF_RETURN_LOCKED 0x00000002 extern int pf_state_insert(struct pfi_kif *, struct pf_state_key *, struct pf_state_key *, struct pf_state *); extern void pf_free_state(struct pf_state *); static __inline void pf_ref_state(struct pf_state *s) { refcount_acquire(&s->refs); } static __inline int pf_release_state(struct pf_state *s) { if (refcount_release(&s->refs)) { pf_free_state(s); return (1); } else return (0); } extern struct pf_state *pf_find_state_byid(uint64_t, uint32_t); extern struct pf_state *pf_find_state_all(struct pf_state_key_cmp *, u_int, int *); extern struct pf_src_node *pf_find_src_node(struct pf_addr *, struct pf_rule *, sa_family_t, int); extern void pf_unlink_src_node(struct pf_src_node *); extern u_int pf_free_src_nodes(struct pf_src_node_list *); extern void pf_print_state(struct pf_state *); extern void pf_print_flags(u_int8_t); extern u_int16_t pf_cksum_fixup(u_int16_t, u_int16_t, u_int16_t, u_int8_t); extern u_int16_t pf_proto_cksum_fixup(struct mbuf *, u_int16_t, u_int16_t, u_int16_t, u_int8_t); VNET_DECLARE(struct ifnet *, sync_ifp); #define V_sync_ifp VNET(sync_ifp); VNET_DECLARE(struct pf_rule, pf_default_rule); #define V_pf_default_rule VNET(pf_default_rule) extern void pf_addrcpy(struct pf_addr *, struct pf_addr *, u_int8_t); void pf_free_rule(struct pf_rule *); #ifdef INET int pf_test(int, struct ifnet *, struct mbuf **, struct inpcb *); int pf_normalize_ip(struct mbuf **, int, struct pfi_kif *, u_short *, struct pf_pdesc *); #endif /* INET */ #ifdef INET6 int pf_test6(int, struct ifnet *, struct mbuf **, struct inpcb *); int pf_normalize_ip6(struct mbuf **, int, struct pfi_kif *, u_short *, struct pf_pdesc *); void pf_poolmask(struct pf_addr *, struct pf_addr*, struct pf_addr *, struct pf_addr *, u_int8_t); void pf_addr_inc(struct pf_addr *, sa_family_t); int pf_refragment6(struct ifnet *, struct mbuf **, struct m_tag *); #endif /* INET6 */ u_int32_t pf_new_isn(struct pf_state *); void *pf_pull_hdr(struct mbuf *, int, void *, int, u_short *, u_short *, sa_family_t); void pf_change_a(void *, u_int16_t *, u_int32_t, u_int8_t); void pf_change_proto_a(struct mbuf *, void *, u_int16_t *, u_int32_t, u_int8_t); void pf_change_tcp_a(struct mbuf *, void *, u_int16_t *, u_int32_t); void pf_send_deferred_syn(struct pf_state *); int pf_match_addr(u_int8_t, struct pf_addr *, struct pf_addr *, struct pf_addr *, sa_family_t); int pf_match_addr_range(struct pf_addr *, struct pf_addr *, struct pf_addr *, sa_family_t); int pf_match_port(u_int8_t, u_int16_t, u_int16_t, u_int16_t); void pf_normalize_init(void); void pf_normalize_cleanup(void); int pf_normalize_tcp(int, struct pfi_kif *, struct mbuf *, int, int, void *, struct pf_pdesc *); void pf_normalize_tcp_cleanup(struct pf_state *); int pf_normalize_tcp_init(struct mbuf *, int, struct pf_pdesc *, struct tcphdr *, struct pf_state_peer *, struct pf_state_peer *); int pf_normalize_tcp_stateful(struct mbuf *, int, struct pf_pdesc *, u_short *, struct tcphdr *, struct pf_state *, struct pf_state_peer *, struct pf_state_peer *, int *); u_int32_t pf_state_expires(const struct pf_state *); void pf_purge_expired_fragments(void); int pf_routable(struct pf_addr *addr, sa_family_t af, struct pfi_kif *, int); int pf_socket_lookup(int, struct pf_pdesc *, struct mbuf *); struct pf_state_key *pf_alloc_state_key(int); void pfr_initialize(void); void pfr_cleanup(void); int pfr_match_addr(struct pfr_ktable *, struct pf_addr *, sa_family_t); void pfr_update_stats(struct pfr_ktable *, struct pf_addr *, sa_family_t, u_int64_t, int, int, int); int pfr_pool_get(struct pfr_ktable *, int *, struct pf_addr *, sa_family_t); void pfr_dynaddr_update(struct pfr_ktable *, struct pfi_dynaddr *); struct pfr_ktable * pfr_attach_table(struct pf_ruleset *, char *); void pfr_detach_table(struct pfr_ktable *); int pfr_clr_tables(struct pfr_table *, int *, int); int pfr_add_tables(struct pfr_table *, int, int *, int); int pfr_del_tables(struct pfr_table *, int, int *, int); int pfr_get_tables(struct pfr_table *, struct pfr_table *, int *, int); int pfr_get_tstats(struct pfr_table *, struct pfr_tstats *, int *, int); int pfr_clr_tstats(struct pfr_table *, int, int *, int); int pfr_set_tflags(struct pfr_table *, int, int, int, int *, int *, int); int pfr_clr_addrs(struct pfr_table *, int *, int); int pfr_insert_kentry(struct pfr_ktable *, struct pfr_addr *, long); int pfr_add_addrs(struct pfr_table *, struct pfr_addr *, int, int *, int); int pfr_del_addrs(struct pfr_table *, struct pfr_addr *, int, int *, int); int pfr_set_addrs(struct pfr_table *, struct pfr_addr *, int, int *, int *, int *, int *, int, u_int32_t); int pfr_get_addrs(struct pfr_table *, struct pfr_addr *, int *, int); int pfr_get_astats(struct pfr_table *, struct pfr_astats *, int *, int); int pfr_clr_astats(struct pfr_table *, struct pfr_addr *, int, int *, int); int pfr_tst_addrs(struct pfr_table *, struct pfr_addr *, int, int *, int); int pfr_ina_begin(struct pfr_table *, u_int32_t *, int *, int); int pfr_ina_rollback(struct pfr_table *, u_int32_t, int *, int); int pfr_ina_commit(struct pfr_table *, u_int32_t, int *, int *, int); int pfr_ina_define(struct pfr_table *, struct pfr_addr *, int, int *, int *, u_int32_t, int); MALLOC_DECLARE(PFI_MTYPE); VNET_DECLARE(struct pfi_kif *, pfi_all); #define V_pfi_all VNET(pfi_all) void pfi_initialize(void); void pfi_initialize_vnet(void); void pfi_cleanup(void); void pfi_cleanup_vnet(void); void pfi_kif_ref(struct pfi_kif *); void pfi_kif_unref(struct pfi_kif *); struct pfi_kif *pfi_kif_find(const char *); struct pfi_kif *pfi_kif_attach(struct pfi_kif *, const char *); int pfi_kif_match(struct pfi_kif *, struct pfi_kif *); void pfi_kif_purge(void); int pfi_match_addr(struct pfi_dynaddr *, struct pf_addr *, sa_family_t); int pfi_dynaddr_setup(struct pf_addr_wrap *, sa_family_t); void pfi_dynaddr_remove(struct pfi_dynaddr *); void pfi_dynaddr_copyout(struct pf_addr_wrap *); void pfi_update_status(const char *, struct pf_status *); void pfi_get_ifaces(const char *, struct pfi_kif *, int *); int pfi_set_flags(const char *, int); int pfi_clear_flags(const char *, int); int pf_match_tag(struct mbuf *, struct pf_rule *, int *, int); int pf_tag_packet(struct mbuf *, struct pf_pdesc *, int); int pf_addr_cmp(struct pf_addr *, struct pf_addr *, sa_family_t); void pf_qid2qname(u_int32_t, char *); VNET_DECLARE(struct pf_kstatus, pf_status); #define V_pf_status VNET(pf_status) struct pf_limit { uma_zone_t zone; u_int limit; }; VNET_DECLARE(struct pf_limit, pf_limits[PF_LIMIT_MAX]); #define V_pf_limits VNET(pf_limits) #endif /* _KERNEL */ #ifdef _KERNEL VNET_DECLARE(struct pf_anchor_global, pf_anchors); #define V_pf_anchors VNET(pf_anchors) VNET_DECLARE(struct pf_anchor, pf_main_anchor); #define V_pf_main_anchor VNET(pf_main_anchor) #define pf_main_ruleset V_pf_main_anchor.ruleset #endif /* these ruleset functions can be linked into userland programs (pfctl) */ int pf_get_ruleset_number(u_int8_t); void pf_init_ruleset(struct pf_ruleset *); int pf_anchor_setup(struct pf_rule *, const struct pf_ruleset *, const char *); int pf_anchor_copyout(const struct pf_ruleset *, const struct pf_rule *, struct pfioc_rule *); void pf_anchor_remove(struct pf_rule *); void pf_remove_if_empty_ruleset(struct pf_ruleset *); struct pf_ruleset *pf_find_ruleset(const char *); struct pf_ruleset *pf_find_or_create_ruleset(const char *); void pf_rs_initialize(void); /* The fingerprint functions can be linked into userland programs (tcpdump) */ int pf_osfp_add(struct pf_osfp_ioctl *); #ifdef _KERNEL struct pf_osfp_enlist * pf_osfp_fingerprint(struct pf_pdesc *, struct mbuf *, int, const struct tcphdr *); #endif /* _KERNEL */ void pf_osfp_flush(void); int pf_osfp_get(struct pf_osfp_ioctl *); int pf_osfp_match(struct pf_osfp_enlist *, pf_osfp_t); #ifdef _KERNEL void pf_print_host(struct pf_addr *, u_int16_t, u_int8_t); void pf_step_into_anchor(struct pf_anchor_stackframe *, int *, struct pf_ruleset **, int, struct pf_rule **, struct pf_rule **, int *); int pf_step_out_of_anchor(struct pf_anchor_stackframe *, int *, struct pf_ruleset **, int, struct pf_rule **, struct pf_rule **, int *); int pf_map_addr(u_int8_t, struct pf_rule *, struct pf_addr *, struct pf_addr *, struct pf_addr *, struct pf_src_node **); struct pf_rule *pf_get_translation(struct pf_pdesc *, struct mbuf *, int, int, struct pfi_kif *, struct pf_src_node **, struct pf_state_key **, struct pf_state_key **, struct pf_addr *, struct pf_addr *, uint16_t, uint16_t, struct pf_anchor_stackframe *); struct pf_state_key *pf_state_key_setup(struct pf_pdesc *, struct pf_addr *, struct pf_addr *, u_int16_t, u_int16_t); struct pf_state_key *pf_state_key_clone(struct pf_state_key *); #endif /* _KERNEL */ #endif /* _NET_PFVAR_H_ */ Index: projects/vnet/sys/netinet/in_pcb.h =================================================================== --- projects/vnet/sys/netinet/in_pcb.h (revision 302157) +++ projects/vnet/sys/netinet/in_pcb.h (revision 302158) @@ -1,741 +1,741 @@ /*- * Copyright (c) 1982, 1986, 1990, 1993 * The Regents of the University of California. * Copyright (c) 2010-2011 Juniper Networks, Inc. * 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. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)in_pcb.h 8.1 (Berkeley) 6/10/93 * $FreeBSD$ */ #ifndef _NETINET_IN_PCB_H_ #define _NETINET_IN_PCB_H_ #include #include #include #include #include #ifdef _KERNEL #include #include #include #include #endif #define in6pcb inpcb /* for KAME src sync over BSD*'s */ #define in6p_sp inp_sp /* for KAME src sync over BSD*'s */ struct inpcbpolicy; /* * struct inpcb is the common protocol control block structure used in most * IP transport protocols. * * Pointers to local and foreign host table entries, local and foreign socket * numbers, and pointers up (to a socket structure) and down (to a * protocol-specific control block) are stored here. */ LIST_HEAD(inpcbhead, inpcb); LIST_HEAD(inpcbporthead, inpcbport); typedef u_quad_t inp_gen_t; /* * PCB with AF_INET6 null bind'ed laddr can receive AF_INET input packet. * So, AF_INET6 null laddr is also used as AF_INET null laddr, by utilizing * the following structure. */ struct in_addr_4in6 { u_int32_t ia46_pad32[3]; struct in_addr ia46_addr4; }; /* * NOTE: ipv6 addrs should be 64-bit aligned, per RFC 2553. in_conninfo has * some extra padding to accomplish this. * NOTE 2: tcp_syncache.c uses first 5 32-bit words, which identify fport, * lport, faddr to generate hash, so these fields shouldn't be moved. */ struct in_endpoints { u_int16_t ie_fport; /* foreign port */ u_int16_t ie_lport; /* local port */ /* protocol dependent part, local and foreign addr */ union { /* foreign host table entry */ struct in_addr_4in6 ie46_foreign; struct in6_addr ie6_foreign; } ie_dependfaddr; union { /* local host table entry */ struct in_addr_4in6 ie46_local; struct in6_addr ie6_local; } ie_dependladdr; u_int32_t ie6_zoneid; /* scope zone id */ }; #define ie_faddr ie_dependfaddr.ie46_foreign.ia46_addr4 #define ie_laddr ie_dependladdr.ie46_local.ia46_addr4 #define ie6_faddr ie_dependfaddr.ie6_foreign #define ie6_laddr ie_dependladdr.ie6_local /* * XXX The defines for inc_* are hacks and should be changed to direct * references. */ struct in_conninfo { u_int8_t inc_flags; u_int8_t inc_len; u_int16_t inc_fibnum; /* XXX was pad, 16 bits is plenty */ /* protocol dependent part */ struct in_endpoints inc_ie; }; /* * Flags for inc_flags. */ #define INC_ISIPV6 0x01 #define inc_isipv6 inc_flags /* temp compatibility */ #define inc_fport inc_ie.ie_fport #define inc_lport inc_ie.ie_lport #define inc_faddr inc_ie.ie_faddr #define inc_laddr inc_ie.ie_laddr #define inc6_faddr inc_ie.ie6_faddr #define inc6_laddr inc_ie.ie6_laddr #define inc6_zoneid inc_ie.ie6_zoneid struct icmp6_filter; /*- * struct inpcb captures the network layer state for TCP, UDP, and raw IPv4 and * IPv6 sockets. In the case of TCP and UDP, further per-connection state is * hung off of inp_ppcb most of the time. Almost all fields of struct inpcb * are static after creation or protected by a per-inpcb rwlock, inp_lock. A * few fields are protected by multiple locks as indicated in the locking notes * below. For these fields, all of the listed locks must be write-locked for * any modifications. However, these fields can be safely read while any one of * the listed locks are read-locked. This model can permit greater concurrency * for read operations. For example, connections can be looked up while only * holding a read lock on the global pcblist lock. This is important for * performance when attempting to find the connection for a packet given its IP * and port tuple. * * One noteworthy exception is that the global pcbinfo lock follows a different * set of rules in relation to the inp_list field. Rather than being * write-locked for modifications and read-locked for list iterations, it must * be read-locked during modifications and write-locked during list iterations. * This ensures that the relatively rare global list iterations safely walk a * stable snapshot of connections while allowing more common list modifications * to safely grab the pcblist lock just while adding or removing a connection * from the global list. * * Key: * (c) - Constant after initialization * (g) - Protected by the pcbgroup lock * (i) - Protected by the inpcb lock * (p) - Protected by the pcbinfo lock for the inpcb * (l) - Protected by the pcblist lock for the inpcb * (h) - Protected by the pcbhash lock for the inpcb * (s) - Protected by another subsystem's locks * (x) - Undefined locking * * A few other notes: * * When a read lock is held, stability of the field is guaranteed; to write * to a field, a write lock must generally be held. * * netinet/netinet6-layer code should not assume that the inp_socket pointer * is safe to dereference without inp_lock being held, even for protocols * other than TCP (where the inpcb persists during TIMEWAIT even after the * socket has been freed), or there may be close(2)-related races. * * The inp_vflag field is overloaded, and would otherwise ideally be (c). * * TODO: Currently only the TCP stack is leveraging the global pcbinfo lock * read-lock usage during modification, this model can be applied to other * protocols (especially SCTP). */ struct inpcb { LIST_ENTRY(inpcb) inp_hash; /* (h/i) hash list */ LIST_ENTRY(inpcb) inp_pcbgrouphash; /* (g/i) hash list */ LIST_ENTRY(inpcb) inp_list; /* (p/l) list for all PCBs for proto */ /* (p[w]) for list iteration */ /* (p[r]/l) for addition/removal */ void *inp_ppcb; /* (i) pointer to per-protocol pcb */ struct inpcbinfo *inp_pcbinfo; /* (c) PCB list info */ struct inpcbgroup *inp_pcbgroup; /* (g/i) PCB group list */ LIST_ENTRY(inpcb) inp_pcbgroup_wild; /* (g/i/h) group wildcard entry */ struct socket *inp_socket; /* (i) back pointer to socket */ struct ucred *inp_cred; /* (c) cache of socket cred */ u_int32_t inp_flow; /* (i) IPv6 flow information */ int inp_flags; /* (i) generic IP/datagram flags */ int inp_flags2; /* (i) generic IP/datagram flags #2*/ u_char inp_vflag; /* (i) IP version flag (v4/v6) */ u_char inp_ip_ttl; /* (i) time to live proto */ u_char inp_ip_p; /* (c) protocol proto */ u_char inp_ip_minttl; /* (i) minimum TTL or drop */ uint32_t inp_flowid; /* (x) flow id / queue id */ u_int inp_refcount; /* (i) refcount */ - void *inp_pspare[5]; /* (x) route caching / general use */ + void *inp_pspare[5]; /* (x) packet pacing / general use */ uint32_t inp_flowtype; /* (x) M_HASHTYPE value */ uint32_t inp_rss_listen_bucket; /* (x) overridden RSS listen bucket */ - u_int inp_ispare[4]; /* (x) route caching / user cookie / + u_int inp_ispare[4]; /* (x) packet pacing / user cookie / * general use */ /* Local and foreign ports, local and foreign addr. */ struct in_conninfo inp_inc; /* (i) list for PCB's local port */ /* MAC and IPSEC policy information. */ struct label *inp_label; /* (i) MAC label */ struct inpcbpolicy *inp_sp; /* (s) for IPSEC */ /* Protocol-dependent part; options. */ struct { u_char inp4_ip_tos; /* (i) type of service proto */ struct mbuf *inp4_options; /* (i) IP options */ struct ip_moptions *inp4_moptions; /* (i) IP mcast options */ } inp_depend4; struct { /* (i) IP options */ struct mbuf *inp6_options; /* (i) IP6 options for outgoing packets */ struct ip6_pktopts *inp6_outputopts; /* (i) IP multicast options */ struct ip6_moptions *inp6_moptions; /* (i) ICMPv6 code type filter */ struct icmp6_filter *inp6_icmp6filt; /* (i) IPV6_CHECKSUM setsockopt */ int inp6_cksum; short inp6_hops; } inp_depend6; LIST_ENTRY(inpcb) inp_portlist; /* (i/h) */ struct inpcbport *inp_phd; /* (i/h) head of this list */ #define inp_zero_size offsetof(struct inpcb, inp_gencnt) inp_gen_t inp_gencnt; /* (c) generation count */ struct llentry *inp_lle; /* cached L2 information */ struct rwlock inp_lock; rt_gen_t inp_rt_cookie; /* generation for route entry */ union { /* cached L3 information */ struct route inpu_route; struct route_in6 inpu_route6; } inp_rtu; #define inp_route inp_rtu.inpu_route #define inp_route6 inp_rtu.inpu_route6 }; #define inp_fport inp_inc.inc_fport #define inp_lport inp_inc.inc_lport #define inp_faddr inp_inc.inc_faddr #define inp_laddr inp_inc.inc_laddr #define inp_ip_tos inp_depend4.inp4_ip_tos #define inp_options inp_depend4.inp4_options #define inp_moptions inp_depend4.inp4_moptions #define in6p_faddr inp_inc.inc6_faddr #define in6p_laddr inp_inc.inc6_laddr #define in6p_zoneid inp_inc.inc6_zoneid #define in6p_hops inp_depend6.inp6_hops /* default hop limit */ #define in6p_flowinfo inp_flow #define in6p_options inp_depend6.inp6_options #define in6p_outputopts inp_depend6.inp6_outputopts #define in6p_moptions inp_depend6.inp6_moptions #define in6p_icmp6filt inp_depend6.inp6_icmp6filt #define in6p_cksum inp_depend6.inp6_cksum #define inp_vnet inp_pcbinfo->ipi_vnet /* * The range of the generation count, as used in this implementation, is 9e19. * We would have to create 300 billion connections per second for this number * to roll over in a year. This seems sufficiently unlikely that we simply * don't concern ourselves with that possibility. */ /* * Interface exported to userland by various protocols which use inpcbs. Hack * alert -- only define if struct xsocket is in scope. */ #ifdef _SYS_SOCKETVAR_H_ struct xinpcb { size_t xi_len; /* length of this structure */ struct inpcb xi_inp; struct xsocket xi_socket; u_quad_t xi_alignment_hack; }; struct xinpgen { size_t xig_len; /* length of this structure */ u_int xig_count; /* number of PCBs at this time */ inp_gen_t xig_gen; /* generation count at this time */ so_gen_t xig_sogen; /* socket generation count at this time */ }; #endif /* _SYS_SOCKETVAR_H_ */ struct inpcbport { LIST_ENTRY(inpcbport) phd_hash; struct inpcbhead phd_pcblist; u_short phd_port; }; /*- * Global data structure for each high-level protocol (UDP, TCP, ...) in both * IPv4 and IPv6. Holds inpcb lists and information for managing them. * * Each pcbinfo is protected by three locks: ipi_lock, ipi_hash_lock and * ipi_list_lock: * - ipi_lock covering the global pcb list stability during loop iteration, * - ipi_hash_lock covering the hashed lookup tables, * - ipi_list_lock covering mutable global fields (such as the global * pcb list) * * The lock order is: * * ipi_lock (before) * inpcb locks (before) * ipi_list locks (before) * {ipi_hash_lock, pcbgroup locks} * * Locking key: * * (c) Constant or nearly constant after initialisation * (g) Locked by ipi_lock * (l) Locked by ipi_list_lock * (h) Read using either ipi_hash_lock or inpcb lock; write requires both * (p) Protected by one or more pcbgroup locks * (x) Synchronisation properties poorly defined */ struct inpcbinfo { /* * Global lock protecting full inpcb list traversal */ struct rwlock ipi_lock; /* * Global list of inpcbs on the protocol. */ struct inpcbhead *ipi_listhead; /* (g/l) */ u_int ipi_count; /* (l) */ /* * Generation count -- incremented each time a connection is allocated * or freed. */ u_quad_t ipi_gencnt; /* (l) */ /* * Fields associated with port lookup and allocation. */ u_short ipi_lastport; /* (x) */ u_short ipi_lastlow; /* (x) */ u_short ipi_lasthi; /* (x) */ /* * UMA zone from which inpcbs are allocated for this protocol. */ struct uma_zone *ipi_zone; /* (c) */ /* * Connection groups associated with this protocol. These fields are * constant, but pcbgroup structures themselves are protected by * per-pcbgroup locks. */ struct inpcbgroup *ipi_pcbgroups; /* (c) */ u_int ipi_npcbgroups; /* (c) */ u_int ipi_hashfields; /* (c) */ /* * Global lock protecting non-pcbgroup hash lookup tables. */ struct rwlock ipi_hash_lock; /* * Global hash of inpcbs, hashed by local and foreign addresses and * port numbers. */ struct inpcbhead *ipi_hashbase; /* (h) */ u_long ipi_hashmask; /* (h) */ /* * Global hash of inpcbs, hashed by only local port number. */ struct inpcbporthead *ipi_porthashbase; /* (h) */ u_long ipi_porthashmask; /* (h) */ /* * List of wildcard inpcbs for use with pcbgroups. In the past, was * per-pcbgroup but is now global. All pcbgroup locks must be held * to modify the list, so any is sufficient to read it. */ struct inpcbhead *ipi_wildbase; /* (p) */ u_long ipi_wildmask; /* (p) */ /* * Pointer to network stack instance */ struct vnet *ipi_vnet; /* (c) */ /* * general use 2 */ void *ipi_pspare[2]; /* * Global lock protecting global inpcb list, inpcb count, etc. */ struct rwlock ipi_list_lock; }; #ifdef _KERNEL /* * Connection groups hold sets of connections that have similar CPU/thread * affinity. Each connection belongs to exactly one connection group. */ struct inpcbgroup { /* * Per-connection group hash of inpcbs, hashed by local and foreign * addresses and port numbers. */ struct inpcbhead *ipg_hashbase; /* (c) */ u_long ipg_hashmask; /* (c) */ /* * Notional affinity of this pcbgroup. */ u_int ipg_cpu; /* (p) */ /* * Per-connection group lock, not to be confused with ipi_lock. * Protects the hash table hung off the group, but also the global * wildcard list in inpcbinfo. */ struct mtx ipg_lock; } __aligned(CACHE_LINE_SIZE); #define INP_LOCK_INIT(inp, d, t) \ rw_init_flags(&(inp)->inp_lock, (t), RW_RECURSE | RW_DUPOK) #define INP_LOCK_DESTROY(inp) rw_destroy(&(inp)->inp_lock) #define INP_RLOCK(inp) rw_rlock(&(inp)->inp_lock) #define INP_WLOCK(inp) rw_wlock(&(inp)->inp_lock) #define INP_TRY_RLOCK(inp) rw_try_rlock(&(inp)->inp_lock) #define INP_TRY_WLOCK(inp) rw_try_wlock(&(inp)->inp_lock) #define INP_RUNLOCK(inp) rw_runlock(&(inp)->inp_lock) #define INP_WUNLOCK(inp) rw_wunlock(&(inp)->inp_lock) #define INP_TRY_UPGRADE(inp) rw_try_upgrade(&(inp)->inp_lock) #define INP_DOWNGRADE(inp) rw_downgrade(&(inp)->inp_lock) #define INP_WLOCKED(inp) rw_wowned(&(inp)->inp_lock) #define INP_LOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_LOCKED) #define INP_RLOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_RLOCKED) #define INP_WLOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_WLOCKED) #define INP_UNLOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_UNLOCKED) /* * These locking functions are for inpcb consumers outside of sys/netinet, * more specifically, they were added for the benefit of TOE drivers. The * macros are reserved for use by the stack. */ void inp_wlock(struct inpcb *); void inp_wunlock(struct inpcb *); void inp_rlock(struct inpcb *); void inp_runlock(struct inpcb *); #ifdef INVARIANTS void inp_lock_assert(struct inpcb *); void inp_unlock_assert(struct inpcb *); #else static __inline void inp_lock_assert(struct inpcb *inp __unused) { } static __inline void inp_unlock_assert(struct inpcb *inp __unused) { } #endif void inp_apply_all(void (*func)(struct inpcb *, void *), void *arg); int inp_ip_tos_get(const struct inpcb *inp); void inp_ip_tos_set(struct inpcb *inp, int val); struct socket * inp_inpcbtosocket(struct inpcb *inp); struct tcpcb * inp_inpcbtotcpcb(struct inpcb *inp); void inp_4tuple_get(struct inpcb *inp, uint32_t *laddr, uint16_t *lp, uint32_t *faddr, uint16_t *fp); short inp_so_options(const struct inpcb *inp); #endif /* _KERNEL */ #define INP_INFO_LOCK_INIT(ipi, d) \ rw_init_flags(&(ipi)->ipi_lock, (d), RW_RECURSE) #define INP_INFO_LOCK_DESTROY(ipi) rw_destroy(&(ipi)->ipi_lock) #define INP_INFO_RLOCK(ipi) rw_rlock(&(ipi)->ipi_lock) #define INP_INFO_WLOCK(ipi) rw_wlock(&(ipi)->ipi_lock) #define INP_INFO_TRY_RLOCK(ipi) rw_try_rlock(&(ipi)->ipi_lock) #define INP_INFO_TRY_WLOCK(ipi) rw_try_wlock(&(ipi)->ipi_lock) #define INP_INFO_TRY_UPGRADE(ipi) rw_try_upgrade(&(ipi)->ipi_lock) #define INP_INFO_WLOCKED(ipi) rw_wowned(&(ipi)->ipi_lock) #define INP_INFO_RUNLOCK(ipi) rw_runlock(&(ipi)->ipi_lock) #define INP_INFO_WUNLOCK(ipi) rw_wunlock(&(ipi)->ipi_lock) #define INP_INFO_LOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_LOCKED) #define INP_INFO_RLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_RLOCKED) #define INP_INFO_WLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_WLOCKED) #define INP_INFO_UNLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_UNLOCKED) #define INP_LIST_LOCK_INIT(ipi, d) \ rw_init_flags(&(ipi)->ipi_list_lock, (d), 0) #define INP_LIST_LOCK_DESTROY(ipi) rw_destroy(&(ipi)->ipi_list_lock) #define INP_LIST_RLOCK(ipi) rw_rlock(&(ipi)->ipi_list_lock) #define INP_LIST_WLOCK(ipi) rw_wlock(&(ipi)->ipi_list_lock) #define INP_LIST_TRY_RLOCK(ipi) rw_try_rlock(&(ipi)->ipi_list_lock) #define INP_LIST_TRY_WLOCK(ipi) rw_try_wlock(&(ipi)->ipi_list_lock) #define INP_LIST_TRY_UPGRADE(ipi) rw_try_upgrade(&(ipi)->ipi_list_lock) #define INP_LIST_RUNLOCK(ipi) rw_runlock(&(ipi)->ipi_list_lock) #define INP_LIST_WUNLOCK(ipi) rw_wunlock(&(ipi)->ipi_list_lock) #define INP_LIST_LOCK_ASSERT(ipi) \ rw_assert(&(ipi)->ipi_list_lock, RA_LOCKED) #define INP_LIST_RLOCK_ASSERT(ipi) \ rw_assert(&(ipi)->ipi_list_lock, RA_RLOCKED) #define INP_LIST_WLOCK_ASSERT(ipi) \ rw_assert(&(ipi)->ipi_list_lock, RA_WLOCKED) #define INP_LIST_UNLOCK_ASSERT(ipi) \ rw_assert(&(ipi)->ipi_list_lock, RA_UNLOCKED) #define INP_HASH_LOCK_INIT(ipi, d) \ rw_init_flags(&(ipi)->ipi_hash_lock, (d), 0) #define INP_HASH_LOCK_DESTROY(ipi) rw_destroy(&(ipi)->ipi_hash_lock) #define INP_HASH_RLOCK(ipi) rw_rlock(&(ipi)->ipi_hash_lock) #define INP_HASH_WLOCK(ipi) rw_wlock(&(ipi)->ipi_hash_lock) #define INP_HASH_RUNLOCK(ipi) rw_runlock(&(ipi)->ipi_hash_lock) #define INP_HASH_WUNLOCK(ipi) rw_wunlock(&(ipi)->ipi_hash_lock) #define INP_HASH_LOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_hash_lock, \ RA_LOCKED) #define INP_HASH_WLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_hash_lock, \ RA_WLOCKED) #define INP_GROUP_LOCK_INIT(ipg, d) mtx_init(&(ipg)->ipg_lock, (d), NULL, \ MTX_DEF | MTX_DUPOK) #define INP_GROUP_LOCK_DESTROY(ipg) mtx_destroy(&(ipg)->ipg_lock) #define INP_GROUP_LOCK(ipg) mtx_lock(&(ipg)->ipg_lock) #define INP_GROUP_LOCK_ASSERT(ipg) mtx_assert(&(ipg)->ipg_lock, MA_OWNED) #define INP_GROUP_UNLOCK(ipg) mtx_unlock(&(ipg)->ipg_lock) #define INP_PCBHASH(faddr, lport, fport, mask) \ (((faddr) ^ ((faddr) >> 16) ^ ntohs((lport) ^ (fport))) & (mask)) #define INP_PCBPORTHASH(lport, mask) \ (ntohs((lport)) & (mask)) #define INP6_PCBHASHKEY(faddr) ((faddr)->s6_addr32[3]) /* * Flags for inp_vflags -- historically version flags only */ #define INP_IPV4 0x1 #define INP_IPV6 0x2 #define INP_IPV6PROTO 0x4 /* opened under IPv6 protocol */ /* * Flags for inp_flags. */ #define INP_RECVOPTS 0x00000001 /* receive incoming IP options */ #define INP_RECVRETOPTS 0x00000002 /* receive IP options for reply */ #define INP_RECVDSTADDR 0x00000004 /* receive IP dst address */ #define INP_HDRINCL 0x00000008 /* user supplies entire IP header */ #define INP_HIGHPORT 0x00000010 /* user wants "high" port binding */ #define INP_LOWPORT 0x00000020 /* user wants "low" port binding */ #define INP_ANONPORT 0x00000040 /* port chosen for user */ #define INP_RECVIF 0x00000080 /* receive incoming interface */ #define INP_MTUDISC 0x00000100 /* user can do MTU discovery */ /* 0x000200 unused: was INP_FAITH */ #define INP_RECVTTL 0x00000400 /* receive incoming IP TTL */ #define INP_DONTFRAG 0x00000800 /* don't fragment packet */ #define INP_BINDANY 0x00001000 /* allow bind to any address */ #define INP_INHASHLIST 0x00002000 /* in_pcbinshash() has been called */ #define INP_RECVTOS 0x00004000 /* receive incoming IP TOS */ #define IN6P_IPV6_V6ONLY 0x00008000 /* restrict AF_INET6 socket for v6 */ #define IN6P_PKTINFO 0x00010000 /* receive IP6 dst and I/F */ #define IN6P_HOPLIMIT 0x00020000 /* receive hoplimit */ #define IN6P_HOPOPTS 0x00040000 /* receive hop-by-hop options */ #define IN6P_DSTOPTS 0x00080000 /* receive dst options after rthdr */ #define IN6P_RTHDR 0x00100000 /* receive routing header */ #define IN6P_RTHDRDSTOPTS 0x00200000 /* receive dstoptions before rthdr */ #define IN6P_TCLASS 0x00400000 /* receive traffic class value */ #define IN6P_AUTOFLOWLABEL 0x00800000 /* attach flowlabel automatically */ #define INP_TIMEWAIT 0x01000000 /* in TIMEWAIT, ppcb is tcptw */ #define INP_ONESBCAST 0x02000000 /* send all-ones broadcast */ #define INP_DROPPED 0x04000000 /* protocol drop flag */ #define INP_SOCKREF 0x08000000 /* strong socket reference */ #define INP_RESERVED_0 0x10000000 /* reserved field */ #define INP_RESERVED_1 0x20000000 /* reserved field */ #define IN6P_RFC2292 0x40000000 /* used RFC2292 API on the socket */ #define IN6P_MTU 0x80000000 /* receive path MTU */ #define INP_CONTROLOPTS (INP_RECVOPTS|INP_RECVRETOPTS|INP_RECVDSTADDR|\ INP_RECVIF|INP_RECVTTL|INP_RECVTOS|\ IN6P_PKTINFO|IN6P_HOPLIMIT|IN6P_HOPOPTS|\ IN6P_DSTOPTS|IN6P_RTHDR|IN6P_RTHDRDSTOPTS|\ IN6P_TCLASS|IN6P_AUTOFLOWLABEL|IN6P_RFC2292|\ IN6P_MTU) /* * Flags for inp_flags2. */ #define INP_LLE_VALID 0x00000001 /* cached lle is valid */ #define INP_RT_VALID 0x00000002 /* cached rtentry is valid */ #define INP_PCBGROUPWILD 0x00000004 /* in pcbgroup wildcard list */ #define INP_REUSEPORT 0x00000008 /* SO_REUSEPORT option is set */ #define INP_FREED 0x00000010 /* inp itself is not valid */ #define INP_REUSEADDR 0x00000020 /* SO_REUSEADDR option is set */ #define INP_BINDMULTI 0x00000040 /* IP_BINDMULTI option is set */ #define INP_RSS_BUCKET_SET 0x00000080 /* IP_RSS_LISTEN_BUCKET is set */ #define INP_RECVFLOWID 0x00000100 /* populate recv datagram with flow info */ #define INP_RECVRSSBUCKETID 0x00000200 /* populate recv datagram with bucket id */ /* * Flags passed to in_pcblookup*() functions. */ #define INPLOOKUP_WILDCARD 0x00000001 /* Allow wildcard sockets. */ #define INPLOOKUP_RLOCKPCB 0x00000002 /* Return inpcb read-locked. */ #define INPLOOKUP_WLOCKPCB 0x00000004 /* Return inpcb write-locked. */ #define INPLOOKUP_MASK (INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB | \ INPLOOKUP_WLOCKPCB) #define sotoinpcb(so) ((struct inpcb *)(so)->so_pcb) #define sotoin6pcb(so) sotoinpcb(so) /* for KAME src sync over BSD*'s */ #define INP_SOCKAF(so) so->so_proto->pr_domain->dom_family #define INP_CHECK_SOCKAF(so, af) (INP_SOCKAF(so) == af) /* * Constants for pcbinfo.ipi_hashfields. */ #define IPI_HASHFIELDS_NONE 0 #define IPI_HASHFIELDS_2TUPLE 1 #define IPI_HASHFIELDS_4TUPLE 2 #ifdef _KERNEL VNET_DECLARE(int, ipport_reservedhigh); VNET_DECLARE(int, ipport_reservedlow); VNET_DECLARE(int, ipport_lowfirstauto); VNET_DECLARE(int, ipport_lowlastauto); VNET_DECLARE(int, ipport_firstauto); VNET_DECLARE(int, ipport_lastauto); VNET_DECLARE(int, ipport_hifirstauto); VNET_DECLARE(int, ipport_hilastauto); VNET_DECLARE(int, ipport_randomized); VNET_DECLARE(int, ipport_randomcps); VNET_DECLARE(int, ipport_randomtime); VNET_DECLARE(int, ipport_stoprandom); VNET_DECLARE(int, ipport_tcpallocs); #define V_ipport_reservedhigh VNET(ipport_reservedhigh) #define V_ipport_reservedlow VNET(ipport_reservedlow) #define V_ipport_lowfirstauto VNET(ipport_lowfirstauto) #define V_ipport_lowlastauto VNET(ipport_lowlastauto) #define V_ipport_firstauto VNET(ipport_firstauto) #define V_ipport_lastauto VNET(ipport_lastauto) #define V_ipport_hifirstauto VNET(ipport_hifirstauto) #define V_ipport_hilastauto VNET(ipport_hilastauto) #define V_ipport_randomized VNET(ipport_randomized) #define V_ipport_randomcps VNET(ipport_randomcps) #define V_ipport_randomtime VNET(ipport_randomtime) #define V_ipport_stoprandom VNET(ipport_stoprandom) #define V_ipport_tcpallocs VNET(ipport_tcpallocs) void in_pcbinfo_destroy(struct inpcbinfo *); void in_pcbinfo_init(struct inpcbinfo *, const char *, struct inpcbhead *, int, int, char *, uma_init, uma_fini, uint32_t, u_int); int in_pcbbind_check_bindmulti(const struct inpcb *ni, const struct inpcb *oi); struct inpcbgroup * in_pcbgroup_byhash(struct inpcbinfo *, u_int, uint32_t); struct inpcbgroup * in_pcbgroup_byinpcb(struct inpcb *); struct inpcbgroup * in_pcbgroup_bytuple(struct inpcbinfo *, struct in_addr, u_short, struct in_addr, u_short); void in_pcbgroup_destroy(struct inpcbinfo *); int in_pcbgroup_enabled(struct inpcbinfo *); void in_pcbgroup_init(struct inpcbinfo *, u_int, int); void in_pcbgroup_remove(struct inpcb *); void in_pcbgroup_update(struct inpcb *); void in_pcbgroup_update_mbuf(struct inpcb *, struct mbuf *); void in_pcbpurgeif0(struct inpcbinfo *, struct ifnet *); int in_pcballoc(struct socket *, struct inpcbinfo *); int in_pcbbind(struct inpcb *, struct sockaddr *, struct ucred *); int in_pcb_lport(struct inpcb *, struct in_addr *, u_short *, struct ucred *, int); int in_pcbbind_setup(struct inpcb *, struct sockaddr *, in_addr_t *, u_short *, struct ucred *); int in_pcbconnect(struct inpcb *, struct sockaddr *, struct ucred *); int in_pcbconnect_mbuf(struct inpcb *, struct sockaddr *, struct ucred *, struct mbuf *); int in_pcbconnect_setup(struct inpcb *, struct sockaddr *, in_addr_t *, u_short *, in_addr_t *, u_short *, struct inpcb **, struct ucred *); void in_pcbdetach(struct inpcb *); void in_pcbdisconnect(struct inpcb *); void in_pcbdrop(struct inpcb *); void in_pcbfree(struct inpcb *); int in_pcbinshash(struct inpcb *); int in_pcbinshash_nopcbgroup(struct inpcb *); int in_pcbladdr(struct inpcb *, struct in_addr *, struct in_addr *, struct ucred *); struct inpcb * in_pcblookup_local(struct inpcbinfo *, struct in_addr, u_short, int, struct ucred *); struct inpcb * in_pcblookup(struct inpcbinfo *, struct in_addr, u_int, struct in_addr, u_int, int, struct ifnet *); struct inpcb * in_pcblookup_mbuf(struct inpcbinfo *, struct in_addr, u_int, struct in_addr, u_int, int, struct ifnet *, struct mbuf *); void in_pcbnotifyall(struct inpcbinfo *pcbinfo, struct in_addr, int, struct inpcb *(*)(struct inpcb *, int)); void in_pcbref(struct inpcb *); void in_pcbrehash(struct inpcb *); void in_pcbrehash_mbuf(struct inpcb *, struct mbuf *); int in_pcbrele(struct inpcb *); int in_pcbrele_rlocked(struct inpcb *); int in_pcbrele_wlocked(struct inpcb *); void in_losing(struct inpcb *); void in_pcbsetsolabel(struct socket *so); int in_getpeeraddr(struct socket *so, struct sockaddr **nam); int in_getsockaddr(struct socket *so, struct sockaddr **nam); struct sockaddr * in_sockaddr(in_port_t port, struct in_addr *addr); void in_pcbsosetlabel(struct socket *so); #endif /* _KERNEL */ #endif /* !_NETINET_IN_PCB_H_ */ Index: projects/vnet/sys/netinet/sctp_usrreq.c =================================================================== --- projects/vnet/sys/netinet/sctp_usrreq.c (revision 302157) +++ projects/vnet/sys/netinet/sctp_usrreq.c (revision 302158) @@ -1,7500 +1,7499 @@ /*- * Copyright (c) 2001-2008, by Cisco Systems, Inc. All rights reserved. * Copyright (c) 2008-2012, by Randall Stewart. All rights reserved. * Copyright (c) 2008-2012, by Michael Tuexen. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * a) Redistributions of source code must retain the above copyright notice, * this list of conditions and the following disclaimer. * * b) Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in * the documentation and/or other materials provided with the distribution. * * c) Neither the name of Cisco Systems, Inc. nor the names of its * contributors may be used to endorse or promote products derived * from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF * THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #ifdef INET6 #include #endif #include #include #include #include #include #include #include #include #include #include extern const struct sctp_cc_functions sctp_cc_functions[]; extern const struct sctp_ss_functions sctp_ss_functions[]; void sctp_init(void) { u_long sb_max_adj; /* Initialize and modify the sysctled variables */ sctp_init_sysctls(); if ((nmbclusters / 8) > SCTP_ASOC_MAX_CHUNKS_ON_QUEUE) SCTP_BASE_SYSCTL(sctp_max_chunks_on_queue) = (nmbclusters / 8); /* * Allow a user to take no more than 1/2 the number of clusters or * the SB_MAX whichever is smaller for the send window. */ sb_max_adj = (u_long)((u_quad_t) (SB_MAX) * MCLBYTES / (MSIZE + MCLBYTES)); SCTP_BASE_SYSCTL(sctp_sendspace) = min(sb_max_adj, (((uint32_t) nmbclusters / 2) * SCTP_DEFAULT_MAXSEGMENT)); /* * Now for the recv window, should we take the same amount? or * should I do 1/2 the SB_MAX instead in the SB_MAX min above. For * now I will just copy. */ SCTP_BASE_SYSCTL(sctp_recvspace) = SCTP_BASE_SYSCTL(sctp_sendspace); SCTP_BASE_VAR(first_time) = 0; SCTP_BASE_VAR(sctp_pcb_initialized) = 0; sctp_pcb_init(); #if defined(SCTP_PACKET_LOGGING) SCTP_BASE_VAR(packet_log_writers) = 0; SCTP_BASE_VAR(packet_log_end) = 0; bzero(&SCTP_BASE_VAR(packet_log_buffer), SCTP_PACKET_LOG_SIZE); #endif } #ifdef VIMAGE static void sctp_finish(void *unused __unused) { sctp_pcb_finish(); } VNET_SYSUNINIT(sctp, SI_SUB_PROTO_DOMAIN, SI_ORDER_FOURTH, sctp_finish, NULL); #endif void sctp_pathmtu_adjustment(struct sctp_tcb *stcb, uint16_t nxtsz) { struct sctp_tmit_chunk *chk; uint16_t overhead; /* Adjust that too */ stcb->asoc.smallest_mtu = nxtsz; /* now off to subtract IP_DF flag if needed */ overhead = IP_HDR_SIZE; if (sctp_auth_is_required_chunk(SCTP_DATA, stcb->asoc.peer_auth_chunks)) { overhead += sctp_get_auth_chunk_len(stcb->asoc.peer_hmac_id); } TAILQ_FOREACH(chk, &stcb->asoc.send_queue, sctp_next) { if ((chk->send_size + overhead) > nxtsz) { chk->flags |= CHUNK_FLAGS_FRAGMENT_OK; } } TAILQ_FOREACH(chk, &stcb->asoc.sent_queue, sctp_next) { if ((chk->send_size + overhead) > nxtsz) { /* * For this guy we also mark for immediate resend * since we sent to big of chunk */ chk->flags |= CHUNK_FLAGS_FRAGMENT_OK; if (chk->sent < SCTP_DATAGRAM_RESEND) { sctp_flight_size_decrease(chk); sctp_total_flight_decrease(stcb, chk); chk->sent = SCTP_DATAGRAM_RESEND; sctp_ucount_incr(stcb->asoc.sent_queue_retran_cnt); chk->rec.data.doing_fast_retransmit = 0; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_FLIGHT_LOGGING_ENABLE) { sctp_misc_ints(SCTP_FLIGHT_LOG_DOWN_PMTU, chk->whoTo->flight_size, chk->book_size, (uint32_t) (uintptr_t) chk->whoTo, chk->rec.data.TSN_seq); } /* Clear any time so NO RTT is being done */ chk->do_rtt = 0; } } } } #ifdef INET void sctp_notify(struct sctp_inpcb *inp, struct sctp_tcb *stcb, struct sctp_nets *net, uint8_t icmp_type, uint8_t icmp_code, uint16_t ip_len, uint16_t next_mtu) { #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) struct socket *so; #endif int timer_stopped; if (icmp_type != ICMP_UNREACH) { /* We only care about unreachable */ SCTP_TCB_UNLOCK(stcb); return; } if ((icmp_code == ICMP_UNREACH_NET) || (icmp_code == ICMP_UNREACH_HOST) || (icmp_code == ICMP_UNREACH_NET_UNKNOWN) || (icmp_code == ICMP_UNREACH_HOST_UNKNOWN) || (icmp_code == ICMP_UNREACH_ISOLATED) || (icmp_code == ICMP_UNREACH_NET_PROHIB) || (icmp_code == ICMP_UNREACH_HOST_PROHIB) || (icmp_code == ICMP_UNREACH_FILTER_PROHIB)) { /* Mark the net unreachable. */ if (net->dest_state & SCTP_ADDR_REACHABLE) { /* OK, that destination is NOT reachable. */ net->dest_state &= ~SCTP_ADDR_REACHABLE; net->dest_state &= ~SCTP_ADDR_PF; sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_DOWN, stcb, 0, (void *)net, SCTP_SO_NOT_LOCKED); } SCTP_TCB_UNLOCK(stcb); } else if ((icmp_code == ICMP_UNREACH_PROTOCOL) || (icmp_code == ICMP_UNREACH_PORT)) { /* Treat it like an ABORT. */ sctp_abort_notification(stcb, 1, 0, NULL, SCTP_SO_NOT_LOCKED); #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) so = SCTP_INP_SO(inp); atomic_add_int(&stcb->asoc.refcnt, 1); SCTP_TCB_UNLOCK(stcb); SCTP_SOCKET_LOCK(so, 1); SCTP_TCB_LOCK(stcb); atomic_subtract_int(&stcb->asoc.refcnt, 1); #endif (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_2); #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) SCTP_SOCKET_UNLOCK(so, 1); /* SCTP_TCB_UNLOCK(stcb); MT: I think this is not needed. */ #endif /* no need to unlock here, since the TCB is gone */ } else if (icmp_code == ICMP_UNREACH_NEEDFRAG) { /* Find the next (smaller) MTU */ if (next_mtu == 0) { /* * Old type router that does not tell us what the * next MTU is. Rats we will have to guess (in a * educated fashion of course). */ next_mtu = sctp_get_prev_mtu(ip_len); } /* Stop the PMTU timer. */ if (SCTP_OS_TIMER_PENDING(&net->pmtu_timer.timer)) { timer_stopped = 1; sctp_timer_stop(SCTP_TIMER_TYPE_PATHMTURAISE, inp, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_1); } else { timer_stopped = 0; } /* Update the path MTU. */ if (net->mtu > next_mtu) { net->mtu = next_mtu; if (net->port) { net->mtu -= sizeof(struct udphdr); } } /* Update the association MTU */ if (stcb->asoc.smallest_mtu > next_mtu) { sctp_pathmtu_adjustment(stcb, next_mtu); } /* Finally, start the PMTU timer if it was running before. */ if (timer_stopped) { sctp_timer_start(SCTP_TIMER_TYPE_PATHMTURAISE, inp, stcb, net); } SCTP_TCB_UNLOCK(stcb); } else { SCTP_TCB_UNLOCK(stcb); } } void sctp_ctlinput(int cmd, struct sockaddr *sa, void *vip) { struct ip *outer_ip; struct ip *inner_ip; struct sctphdr *sh; struct icmp *icmp; struct sctp_inpcb *inp; struct sctp_tcb *stcb; struct sctp_nets *net; struct sctp_init_chunk *ch; struct sockaddr_in src, dst; if (sa->sa_family != AF_INET || ((struct sockaddr_in *)sa)->sin_addr.s_addr == INADDR_ANY) { return; } if (PRC_IS_REDIRECT(cmd)) { vip = NULL; } else if ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0) { return; } if (vip != NULL) { inner_ip = (struct ip *)vip; icmp = (struct icmp *)((caddr_t)inner_ip - (sizeof(struct icmp) - sizeof(struct ip))); outer_ip = (struct ip *)((caddr_t)icmp - sizeof(struct ip)); sh = (struct sctphdr *)((caddr_t)inner_ip + (inner_ip->ip_hl << 2)); memset(&src, 0, sizeof(struct sockaddr_in)); src.sin_family = AF_INET; src.sin_len = sizeof(struct sockaddr_in); src.sin_port = sh->src_port; src.sin_addr = inner_ip->ip_src; memset(&dst, 0, sizeof(struct sockaddr_in)); dst.sin_family = AF_INET; dst.sin_len = sizeof(struct sockaddr_in); dst.sin_port = sh->dest_port; dst.sin_addr = inner_ip->ip_dst; /* * 'dst' holds the dest of the packet that failed to be * sent. 'src' holds our local endpoint address. Thus we * reverse the dst and the src in the lookup. */ inp = NULL; net = NULL; stcb = sctp_findassociation_addr_sa((struct sockaddr *)&dst, (struct sockaddr *)&src, &inp, &net, 1, SCTP_DEFAULT_VRFID); if ((stcb != NULL) && (net != NULL) && - (inp != NULL) && - (inp->sctp_socket != NULL)) { + (inp != NULL)) { /* Check the verification tag */ if (ntohl(sh->v_tag) != 0) { /* * This must be the verification tag used * for sending out packets. We don't * consider packets reflecting the * verification tag. */ if (ntohl(sh->v_tag) != stcb->asoc.peer_vtag) { SCTP_TCB_UNLOCK(stcb); return; } } else { if (ntohs(outer_ip->ip_len) >= sizeof(struct ip) + 8 + (inner_ip->ip_hl << 2) + 20) { /* * In this case we can check if we * got an INIT chunk and if the * initiate tag matches. */ ch = (struct sctp_init_chunk *)(sh + 1); if ((ch->ch.chunk_type != SCTP_INITIATION) || (ntohl(ch->init.initiate_tag) != stcb->asoc.my_vtag)) { SCTP_TCB_UNLOCK(stcb); return; } } else { SCTP_TCB_UNLOCK(stcb); return; } } sctp_notify(inp, stcb, net, icmp->icmp_type, icmp->icmp_code, ntohs(inner_ip->ip_len), ntohs(icmp->icmp_nextmtu)); } else { if ((stcb == NULL) && (inp != NULL)) { /* reduce ref-count */ SCTP_INP_WLOCK(inp); SCTP_INP_DECR_REF(inp); SCTP_INP_WUNLOCK(inp); } if (stcb) { SCTP_TCB_UNLOCK(stcb); } } } return; } #endif static int sctp_getcred(SYSCTL_HANDLER_ARGS) { struct xucred xuc; struct sockaddr_in addrs[2]; struct sctp_inpcb *inp; struct sctp_nets *net; struct sctp_tcb *stcb; int error; uint32_t vrf_id; /* FIX, for non-bsd is this right? */ vrf_id = SCTP_DEFAULT_VRFID; error = priv_check(req->td, PRIV_NETINET_GETCRED); if (error) return (error); error = SYSCTL_IN(req, addrs, sizeof(addrs)); if (error) return (error); stcb = sctp_findassociation_addr_sa(sintosa(&addrs[1]), sintosa(&addrs[0]), &inp, &net, 1, vrf_id); if (stcb == NULL || inp == NULL || inp->sctp_socket == NULL) { if ((inp != NULL) && (stcb == NULL)) { /* reduce ref-count */ SCTP_INP_WLOCK(inp); SCTP_INP_DECR_REF(inp); goto cred_can_cont; } SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); error = ENOENT; goto out; } SCTP_TCB_UNLOCK(stcb); /* * We use the write lock here, only since in the error leg we need * it. If we used RLOCK, then we would have to * wlock/decr/unlock/rlock. Which in theory could create a hole. * Better to use higher wlock. */ SCTP_INP_WLOCK(inp); cred_can_cont: error = cr_canseesocket(req->td->td_ucred, inp->sctp_socket); if (error) { SCTP_INP_WUNLOCK(inp); goto out; } cru2x(inp->sctp_socket->so_cred, &xuc); SCTP_INP_WUNLOCK(inp); error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred)); out: return (error); } SYSCTL_PROC(_net_inet_sctp, OID_AUTO, getcred, CTLTYPE_OPAQUE | CTLFLAG_RW, 0, 0, sctp_getcred, "S,ucred", "Get the ucred of a SCTP connection"); #ifdef INET static void sctp_abort(struct socket *so) { struct sctp_inpcb *inp; uint32_t flags; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { return; } sctp_must_try_again: flags = inp->sctp_flags; #ifdef SCTP_LOG_CLOSING sctp_log_closing(inp, NULL, 17); #endif if (((flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) && (atomic_cmpset_int(&inp->sctp_flags, flags, (flags | SCTP_PCB_FLAGS_SOCKET_GONE | SCTP_PCB_FLAGS_CLOSE_IP)))) { #ifdef SCTP_LOG_CLOSING sctp_log_closing(inp, NULL, 16); #endif sctp_inpcb_free(inp, SCTP_FREE_SHOULD_USE_ABORT, SCTP_CALLED_AFTER_CMPSET_OFCLOSE); SOCK_LOCK(so); SCTP_SB_CLEAR(so->so_snd); /* * same for the rcv ones, they are only here for the * accounting/select. */ SCTP_SB_CLEAR(so->so_rcv); /* Now null out the reference, we are completely detached. */ so->so_pcb = NULL; SOCK_UNLOCK(so); } else { flags = inp->sctp_flags; if ((flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) { goto sctp_must_try_again; } } return; } static int sctp_attach(struct socket *so, int proto SCTP_UNUSED, struct thread *p SCTP_UNUSED) { struct sctp_inpcb *inp; struct inpcb *ip_inp; int error; uint32_t vrf_id = SCTP_DEFAULT_VRFID; inp = (struct sctp_inpcb *)so->so_pcb; if (inp != NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { error = SCTP_SORESERVE(so, SCTP_BASE_SYSCTL(sctp_sendspace), SCTP_BASE_SYSCTL(sctp_recvspace)); if (error) { return (error); } } error = sctp_inpcb_alloc(so, vrf_id); if (error) { return (error); } inp = (struct sctp_inpcb *)so->so_pcb; SCTP_INP_WLOCK(inp); inp->sctp_flags &= ~SCTP_PCB_FLAGS_BOUND_V6; /* I'm not v6! */ ip_inp = &inp->ip_inp.inp; ip_inp->inp_vflag |= INP_IPV4; ip_inp->inp_ip_ttl = MODULE_GLOBAL(ip_defttl); SCTP_INP_WUNLOCK(inp); return (0); } static int sctp_bind(struct socket *so, struct sockaddr *addr, struct thread *p) { struct sctp_inpcb *inp; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } if (addr != NULL) { if ((addr->sa_family != AF_INET) || (addr->sa_len != sizeof(struct sockaddr_in))) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } } return (sctp_inpcb_bind(so, addr, NULL, p)); } #endif void sctp_close(struct socket *so) { struct sctp_inpcb *inp; uint32_t flags; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) return; /* * Inform all the lower layer assoc that we are done. */ sctp_must_try_again: flags = inp->sctp_flags; #ifdef SCTP_LOG_CLOSING sctp_log_closing(inp, NULL, 17); #endif if (((flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) && (atomic_cmpset_int(&inp->sctp_flags, flags, (flags | SCTP_PCB_FLAGS_SOCKET_GONE | SCTP_PCB_FLAGS_CLOSE_IP)))) { if (((so->so_options & SO_LINGER) && (so->so_linger == 0)) || (so->so_rcv.sb_cc > 0)) { #ifdef SCTP_LOG_CLOSING sctp_log_closing(inp, NULL, 13); #endif sctp_inpcb_free(inp, SCTP_FREE_SHOULD_USE_ABORT, SCTP_CALLED_AFTER_CMPSET_OFCLOSE); } else { #ifdef SCTP_LOG_CLOSING sctp_log_closing(inp, NULL, 14); #endif sctp_inpcb_free(inp, SCTP_FREE_SHOULD_USE_GRACEFUL_CLOSE, SCTP_CALLED_AFTER_CMPSET_OFCLOSE); } /* * The socket is now detached, no matter what the state of * the SCTP association. */ SOCK_LOCK(so); SCTP_SB_CLEAR(so->so_snd); /* * same for the rcv ones, they are only here for the * accounting/select. */ SCTP_SB_CLEAR(so->so_rcv); /* Now null out the reference, we are completely detached. */ so->so_pcb = NULL; SOCK_UNLOCK(so); } else { flags = inp->sctp_flags; if ((flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) { goto sctp_must_try_again; } } return; } int sctp_sendm(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr, struct mbuf *control, struct thread *p); int sctp_sendm(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr, struct mbuf *control, struct thread *p) { struct sctp_inpcb *inp; int error; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { if (control) { sctp_m_freem(control); control = NULL; } SCTP_LTRACE_ERR_RET_PKT(m, inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); sctp_m_freem(m); return (EINVAL); } /* Got to have an to address if we are NOT a connected socket */ if ((addr == NULL) && ((inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) || (inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE))) { goto connected_type; } else if (addr == NULL) { SCTP_LTRACE_ERR_RET_PKT(m, inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EDESTADDRREQ); error = EDESTADDRREQ; sctp_m_freem(m); if (control) { sctp_m_freem(control); control = NULL; } return (error); } #ifdef INET6 if (addr->sa_family != AF_INET) { /* must be a v4 address! */ SCTP_LTRACE_ERR_RET_PKT(m, inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EDESTADDRREQ); sctp_m_freem(m); if (control) { sctp_m_freem(control); control = NULL; } error = EDESTADDRREQ; return (error); } #endif /* INET6 */ connected_type: /* now what about control */ if (control) { if (inp->control) { SCTP_PRINTF("huh? control set?\n"); sctp_m_freem(inp->control); inp->control = NULL; } inp->control = control; } /* Place the data */ if (inp->pkt) { SCTP_BUF_NEXT(inp->pkt_last) = m; inp->pkt_last = m; } else { inp->pkt_last = inp->pkt = m; } if ( /* FreeBSD uses a flag passed */ ((flags & PRUS_MORETOCOME) == 0) ) { /* * note with the current version this code will only be used * by OpenBSD-- NetBSD, FreeBSD, and MacOS have methods for * re-defining sosend to use the sctp_sosend. One can * optionally switch back to this code (by changing back the * definitions) but this is not advisable. This code is used * by FreeBSD when sending a file with sendfile() though. */ int ret; ret = sctp_output(inp, inp->pkt, addr, inp->control, p, flags); inp->pkt = NULL; inp->control = NULL; return (ret); } else { return (0); } } int sctp_disconnect(struct socket *so) { struct sctp_inpcb *inp; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOTCONN); return (ENOTCONN); } SCTP_INP_RLOCK(inp); if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) { if (LIST_EMPTY(&inp->sctp_asoc_list)) { /* No connection */ SCTP_INP_RUNLOCK(inp); return (0); } else { struct sctp_association *asoc; struct sctp_tcb *stcb; stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb == NULL) { SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } SCTP_TCB_LOCK(stcb); asoc = &stcb->asoc; if (stcb->asoc.state & SCTP_STATE_ABOUT_TO_BE_FREED) { /* We are about to be freed, out of here */ SCTP_TCB_UNLOCK(stcb); SCTP_INP_RUNLOCK(inp); return (0); } if (((so->so_options & SO_LINGER) && (so->so_linger == 0)) || (so->so_rcv.sb_cc > 0)) { if (SCTP_GET_STATE(asoc) != SCTP_STATE_COOKIE_WAIT) { /* Left with Data unread */ struct mbuf *err; err = sctp_get_mbuf_for_msg(sizeof(struct sctp_paramhdr), 0, M_NOWAIT, 1, MT_DATA); if (err) { /* * Fill in the user * initiated abort */ struct sctp_paramhdr *ph; ph = mtod(err, struct sctp_paramhdr *); SCTP_BUF_LEN(err) = sizeof(struct sctp_paramhdr); ph->param_type = htons(SCTP_CAUSE_USER_INITIATED_ABT); ph->param_length = htons(SCTP_BUF_LEN(err)); } sctp_send_abort_tcb(stcb, err, SCTP_SO_LOCKED); SCTP_STAT_INCR_COUNTER32(sctps_aborted); } SCTP_INP_RUNLOCK(inp); if ((SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_OPEN) || (SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_SHUTDOWN_RECEIVED)) { SCTP_STAT_DECR_GAUGE32(sctps_currestab); } (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_3); /* No unlock tcb assoc is gone */ return (0); } if (TAILQ_EMPTY(&asoc->send_queue) && TAILQ_EMPTY(&asoc->sent_queue) && (asoc->stream_queue_cnt == 0)) { /* there is nothing queued to send, so done */ if (asoc->locked_on_sending) { goto abort_anyway; } if ((SCTP_GET_STATE(asoc) != SCTP_STATE_SHUTDOWN_SENT) && (SCTP_GET_STATE(asoc) != SCTP_STATE_SHUTDOWN_ACK_SENT)) { /* only send SHUTDOWN 1st time thru */ struct sctp_nets *netp; if ((SCTP_GET_STATE(asoc) == SCTP_STATE_OPEN) || (SCTP_GET_STATE(asoc) == SCTP_STATE_SHUTDOWN_RECEIVED)) { SCTP_STAT_DECR_GAUGE32(sctps_currestab); } SCTP_SET_STATE(asoc, SCTP_STATE_SHUTDOWN_SENT); SCTP_CLEAR_SUBSTATE(asoc, SCTP_STATE_SHUTDOWN_PENDING); sctp_stop_timers_for_shutdown(stcb); if (stcb->asoc.alternate) { netp = stcb->asoc.alternate; } else { netp = stcb->asoc.primary_destination; } sctp_send_shutdown(stcb, netp); sctp_timer_start(SCTP_TIMER_TYPE_SHUTDOWN, stcb->sctp_ep, stcb, netp); sctp_timer_start(SCTP_TIMER_TYPE_SHUTDOWNGUARD, stcb->sctp_ep, stcb, netp); sctp_chunk_output(stcb->sctp_ep, stcb, SCTP_OUTPUT_FROM_T3, SCTP_SO_LOCKED); } } else { /* * we still got (or just got) data to send, * so set SHUTDOWN_PENDING */ /* * XXX sockets draft says that SCTP_EOF * should be sent with no data. currently, * we will allow user data to be sent first * and move to SHUTDOWN-PENDING */ struct sctp_nets *netp; if (stcb->asoc.alternate) { netp = stcb->asoc.alternate; } else { netp = stcb->asoc.primary_destination; } asoc->state |= SCTP_STATE_SHUTDOWN_PENDING; sctp_timer_start(SCTP_TIMER_TYPE_SHUTDOWNGUARD, stcb->sctp_ep, stcb, netp); if (asoc->locked_on_sending) { /* Locked to send out the data */ struct sctp_stream_queue_pending *sp; sp = TAILQ_LAST(&asoc->locked_on_sending->outqueue, sctp_streamhead); if (sp == NULL) { SCTP_PRINTF("Error, sp is NULL, locked on sending is non-null strm:%d\n", asoc->locked_on_sending->stream_no); } else { if ((sp->length == 0) && (sp->msg_is_complete == 0)) asoc->state |= SCTP_STATE_PARTIAL_MSG_LEFT; } } if (TAILQ_EMPTY(&asoc->send_queue) && TAILQ_EMPTY(&asoc->sent_queue) && (asoc->state & SCTP_STATE_PARTIAL_MSG_LEFT)) { struct mbuf *op_err; abort_anyway: op_err = sctp_generate_cause(SCTP_CAUSE_USER_INITIATED_ABT, ""); stcb->sctp_ep->last_abort_code = SCTP_FROM_SCTP_USRREQ + SCTP_LOC_4; sctp_send_abort_tcb(stcb, op_err, SCTP_SO_LOCKED); SCTP_STAT_INCR_COUNTER32(sctps_aborted); if ((SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_OPEN) || (SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_SHUTDOWN_RECEIVED)) { SCTP_STAT_DECR_GAUGE32(sctps_currestab); } SCTP_INP_RUNLOCK(inp); (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_5); return (0); } else { sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_CLOSING, SCTP_SO_LOCKED); } } soisdisconnecting(so); SCTP_TCB_UNLOCK(stcb); SCTP_INP_RUNLOCK(inp); return (0); } /* not reached */ } else { /* UDP model does not support this */ SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); return (EOPNOTSUPP); } } int sctp_flush(struct socket *so, int how) { /* * We will just clear out the values and let subsequent close clear * out the data, if any. Note if the user did a shutdown(SHUT_RD) * they will not be able to read the data, the socket will block * that from happening. */ struct sctp_inpcb *inp; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } SCTP_INP_RLOCK(inp); /* For the 1 to many model this does nothing */ if (inp->sctp_flags & SCTP_PCB_FLAGS_UDPTYPE) { SCTP_INP_RUNLOCK(inp); return (0); } SCTP_INP_RUNLOCK(inp); if ((how == PRU_FLUSH_RD) || (how == PRU_FLUSH_RDWR)) { /* * First make sure the sb will be happy, we don't use these * except maybe the count */ SCTP_INP_WLOCK(inp); SCTP_INP_READ_LOCK(inp); inp->sctp_flags |= SCTP_PCB_FLAGS_SOCKET_CANT_READ; SCTP_INP_READ_UNLOCK(inp); SCTP_INP_WUNLOCK(inp); so->so_rcv.sb_cc = 0; so->so_rcv.sb_mbcnt = 0; so->so_rcv.sb_mb = NULL; } if ((how == PRU_FLUSH_WR) || (how == PRU_FLUSH_RDWR)) { /* * First make sure the sb will be happy, we don't use these * except maybe the count */ so->so_snd.sb_cc = 0; so->so_snd.sb_mbcnt = 0; so->so_snd.sb_mb = NULL; } return (0); } int sctp_shutdown(struct socket *so) { struct sctp_inpcb *inp; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } SCTP_INP_RLOCK(inp); /* For UDP model this is a invalid call */ if (!((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL))) { /* Restore the flags that the soshutdown took away. */ SOCKBUF_LOCK(&so->so_rcv); so->so_rcv.sb_state &= ~SBS_CANTRCVMORE; SOCKBUF_UNLOCK(&so->so_rcv); /* This proc will wakeup for read and do nothing (I hope) */ SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); return (EOPNOTSUPP); } else { /* * Ok, if we reach here its the TCP model and it is either a * SHUT_WR or SHUT_RDWR. This means we put the shutdown flag * against it. */ struct sctp_tcb *stcb; struct sctp_association *asoc; struct sctp_nets *netp; if ((so->so_state & (SS_ISCONNECTED | SS_ISCONNECTING | SS_ISDISCONNECTING)) == 0) { SCTP_INP_RUNLOCK(inp); return (ENOTCONN); } socantsendmore(so); stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb == NULL) { /* * Ok, we hit the case that the shutdown call was * made after an abort or something. Nothing to do * now. */ SCTP_INP_RUNLOCK(inp); return (0); } SCTP_TCB_LOCK(stcb); asoc = &stcb->asoc; if (asoc->state & SCTP_STATE_ABOUT_TO_BE_FREED) { SCTP_TCB_UNLOCK(stcb); SCTP_INP_RUNLOCK(inp); return (0); } if ((SCTP_GET_STATE(asoc) != SCTP_STATE_COOKIE_WAIT) && (SCTP_GET_STATE(asoc) != SCTP_STATE_COOKIE_ECHOED) && (SCTP_GET_STATE(asoc) != SCTP_STATE_OPEN)) { /* * If we are not in or before ESTABLISHED, there is * no protocol action required. */ SCTP_TCB_UNLOCK(stcb); SCTP_INP_RUNLOCK(inp); return (0); } if (stcb->asoc.alternate) { netp = stcb->asoc.alternate; } else { netp = stcb->asoc.primary_destination; } if ((SCTP_GET_STATE(asoc) == SCTP_STATE_OPEN) && TAILQ_EMPTY(&asoc->send_queue) && TAILQ_EMPTY(&asoc->sent_queue) && (asoc->stream_queue_cnt == 0)) { if (asoc->locked_on_sending) { goto abort_anyway; } /* there is nothing queued to send, so I'm done... */ SCTP_STAT_DECR_GAUGE32(sctps_currestab); SCTP_SET_STATE(asoc, SCTP_STATE_SHUTDOWN_SENT); SCTP_CLEAR_SUBSTATE(asoc, SCTP_STATE_SHUTDOWN_PENDING); sctp_stop_timers_for_shutdown(stcb); sctp_send_shutdown(stcb, netp); sctp_timer_start(SCTP_TIMER_TYPE_SHUTDOWN, stcb->sctp_ep, stcb, netp); } else { /* * We still got (or just got) data to send, so set * SHUTDOWN_PENDING. */ SCTP_ADD_SUBSTATE(asoc, SCTP_STATE_SHUTDOWN_PENDING); if (asoc->locked_on_sending) { /* Locked to send out the data */ struct sctp_stream_queue_pending *sp; sp = TAILQ_LAST(&asoc->locked_on_sending->outqueue, sctp_streamhead); if (sp == NULL) { SCTP_PRINTF("Error, sp is NULL, locked on sending is non-null strm:%d\n", asoc->locked_on_sending->stream_no); } else { if ((sp->length == 0) && (sp->msg_is_complete == 0)) { SCTP_ADD_SUBSTATE(asoc, SCTP_STATE_PARTIAL_MSG_LEFT); } } } if (TAILQ_EMPTY(&asoc->send_queue) && TAILQ_EMPTY(&asoc->sent_queue) && (asoc->state & SCTP_STATE_PARTIAL_MSG_LEFT)) { struct mbuf *op_err; abort_anyway: op_err = sctp_generate_cause(SCTP_CAUSE_USER_INITIATED_ABT, ""); stcb->sctp_ep->last_abort_code = SCTP_FROM_SCTP_USRREQ + SCTP_LOC_6; sctp_abort_an_association(stcb->sctp_ep, stcb, op_err, SCTP_SO_LOCKED); SCTP_INP_RUNLOCK(inp); return (0); } } sctp_timer_start(SCTP_TIMER_TYPE_SHUTDOWNGUARD, stcb->sctp_ep, stcb, netp); /* * XXX: Why do this in the case where we have still data * queued? */ sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_CLOSING, SCTP_SO_LOCKED); SCTP_TCB_UNLOCK(stcb); SCTP_INP_RUNLOCK(inp); return (0); } } /* * copies a "user" presentable address and removes embedded scope, etc. * returns 0 on success, 1 on error */ static uint32_t sctp_fill_user_address(struct sockaddr_storage *ss, struct sockaddr *sa) { #ifdef INET6 struct sockaddr_in6 lsa6; sa = (struct sockaddr *)sctp_recover_scope((struct sockaddr_in6 *)sa, &lsa6); #endif memcpy(ss, sa, sa->sa_len); return (0); } /* * NOTE: assumes addr lock is held */ static size_t sctp_fill_up_addresses_vrf(struct sctp_inpcb *inp, struct sctp_tcb *stcb, size_t limit, struct sockaddr_storage *sas, uint32_t vrf_id) { struct sctp_ifn *sctp_ifn; struct sctp_ifa *sctp_ifa; size_t actual; int loopback_scope; #if defined(INET) int ipv4_local_scope, ipv4_addr_legal; #endif #if defined(INET6) int local_scope, site_scope, ipv6_addr_legal; #endif struct sctp_vrf *vrf; actual = 0; if (limit <= 0) return (actual); if (stcb) { /* Turn on all the appropriate scope */ loopback_scope = stcb->asoc.scope.loopback_scope; #if defined(INET) ipv4_local_scope = stcb->asoc.scope.ipv4_local_scope; ipv4_addr_legal = stcb->asoc.scope.ipv4_addr_legal; #endif #if defined(INET6) local_scope = stcb->asoc.scope.local_scope; site_scope = stcb->asoc.scope.site_scope; ipv6_addr_legal = stcb->asoc.scope.ipv6_addr_legal; #endif } else { /* Use generic values for endpoints. */ loopback_scope = 1; #if defined(INET) ipv4_local_scope = 1; #endif #if defined(INET6) local_scope = 1; site_scope = 1; #endif if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { #if defined(INET6) ipv6_addr_legal = 1; #endif #if defined(INET) if (SCTP_IPV6_V6ONLY(inp)) { ipv4_addr_legal = 0; } else { ipv4_addr_legal = 1; } #endif } else { #if defined(INET6) ipv6_addr_legal = 0; #endif #if defined(INET) ipv4_addr_legal = 1; #endif } } vrf = sctp_find_vrf(vrf_id); if (vrf == NULL) { return (0); } if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) { LIST_FOREACH(sctp_ifn, &vrf->ifnlist, next_ifn) { if ((loopback_scope == 0) && SCTP_IFN_IS_IFT_LOOP(sctp_ifn)) { /* Skip loopback if loopback_scope not set */ continue; } LIST_FOREACH(sctp_ifa, &sctp_ifn->ifalist, next_ifa) { if (stcb) { /* * For the BOUND-ALL case, the list * associated with a TCB is Always * considered a reverse list.. i.e. * it lists addresses that are NOT * part of the association. If this * is one of those we must skip it. */ if (sctp_is_addr_restricted(stcb, sctp_ifa)) { continue; } } switch (sctp_ifa->address.sa.sa_family) { #ifdef INET case AF_INET: if (ipv4_addr_legal) { struct sockaddr_in *sin; sin = &sctp_ifa->address.sin; if (sin->sin_addr.s_addr == 0) { /* * we skip * unspecifed * addresses */ continue; } if (prison_check_ip4(inp->ip_inp.inp.inp_cred, &sin->sin_addr) != 0) { continue; } if ((ipv4_local_scope == 0) && (IN4_ISPRIVATE_ADDRESS(&sin->sin_addr))) { continue; } #ifdef INET6 if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4)) { in6_sin_2_v4mapsin6(sin, (struct sockaddr_in6 *)sas); ((struct sockaddr_in6 *)sas)->sin6_port = inp->sctp_lport; sas = (struct sockaddr_storage *)((caddr_t)sas + sizeof(struct sockaddr_in6)); actual += sizeof(struct sockaddr_in6); } else { #endif memcpy(sas, sin, sizeof(*sin)); ((struct sockaddr_in *)sas)->sin_port = inp->sctp_lport; sas = (struct sockaddr_storage *)((caddr_t)sas + sizeof(*sin)); actual += sizeof(*sin); #ifdef INET6 } #endif if (actual >= limit) { return (actual); } } else { continue; } break; #endif #ifdef INET6 case AF_INET6: if (ipv6_addr_legal) { struct sockaddr_in6 *sin6; sin6 = &sctp_ifa->address.sin6; if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { /* * we skip * unspecifed * addresses */ continue; } if (prison_check_ip6(inp->ip_inp.inp.inp_cred, &sin6->sin6_addr) != 0) { continue; } if (IN6_IS_ADDR_LINKLOCAL(&sin6->sin6_addr)) { if (local_scope == 0) continue; if (sin6->sin6_scope_id == 0) { if (sa6_recoverscope(sin6) != 0) /* * * bad * * li * nk * * loc * al * * add * re * ss * */ continue; } } if ((site_scope == 0) && (IN6_IS_ADDR_SITELOCAL(&sin6->sin6_addr))) { continue; } memcpy(sas, sin6, sizeof(*sin6)); ((struct sockaddr_in6 *)sas)->sin6_port = inp->sctp_lport; sas = (struct sockaddr_storage *)((caddr_t)sas + sizeof(*sin6)); actual += sizeof(*sin6); if (actual >= limit) { return (actual); } } else { continue; } break; #endif default: /* TSNH */ break; } } } } else { struct sctp_laddr *laddr; LIST_FOREACH(laddr, &inp->sctp_addr_list, sctp_nxt_addr) { if (stcb) { if (sctp_is_addr_restricted(stcb, laddr->ifa)) { continue; } } if (sctp_fill_user_address(sas, &laddr->ifa->address.sa)) continue; switch (laddr->ifa->address.sa.sa_family) { #ifdef INET case AF_INET: ((struct sockaddr_in *)sas)->sin_port = inp->sctp_lport; break; #endif #ifdef INET6 case AF_INET6: ((struct sockaddr_in6 *)sas)->sin6_port = inp->sctp_lport; break; #endif default: /* TSNH */ break; } sas = (struct sockaddr_storage *)((caddr_t)sas + laddr->ifa->address.sa.sa_len); actual += laddr->ifa->address.sa.sa_len; if (actual >= limit) { return (actual); } } } return (actual); } static size_t sctp_fill_up_addresses(struct sctp_inpcb *inp, struct sctp_tcb *stcb, size_t limit, struct sockaddr_storage *sas) { size_t size = 0; SCTP_IPI_ADDR_RLOCK(); /* fill up addresses for the endpoint's default vrf */ size = sctp_fill_up_addresses_vrf(inp, stcb, limit, sas, inp->def_vrf_id); SCTP_IPI_ADDR_RUNLOCK(); return (size); } /* * NOTE: assumes addr lock is held */ static int sctp_count_max_addresses_vrf(struct sctp_inpcb *inp, uint32_t vrf_id) { int cnt = 0; struct sctp_vrf *vrf = NULL; /* * In both sub-set bound an bound_all cases we return the MAXIMUM * number of addresses that you COULD get. In reality the sub-set * bound may have an exclusion list for a given TCB OR in the * bound-all case a TCB may NOT include the loopback or other * addresses as well. */ vrf = sctp_find_vrf(vrf_id); if (vrf == NULL) { return (0); } if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) { struct sctp_ifn *sctp_ifn; struct sctp_ifa *sctp_ifa; LIST_FOREACH(sctp_ifn, &vrf->ifnlist, next_ifn) { LIST_FOREACH(sctp_ifa, &sctp_ifn->ifalist, next_ifa) { /* Count them if they are the right type */ switch (sctp_ifa->address.sa.sa_family) { #ifdef INET case AF_INET: #ifdef INET6 if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4)) cnt += sizeof(struct sockaddr_in6); else cnt += sizeof(struct sockaddr_in); #else cnt += sizeof(struct sockaddr_in); #endif break; #endif #ifdef INET6 case AF_INET6: cnt += sizeof(struct sockaddr_in6); break; #endif default: break; } } } } else { struct sctp_laddr *laddr; LIST_FOREACH(laddr, &inp->sctp_addr_list, sctp_nxt_addr) { switch (laddr->ifa->address.sa.sa_family) { #ifdef INET case AF_INET: #ifdef INET6 if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4)) cnt += sizeof(struct sockaddr_in6); else cnt += sizeof(struct sockaddr_in); #else cnt += sizeof(struct sockaddr_in); #endif break; #endif #ifdef INET6 case AF_INET6: cnt += sizeof(struct sockaddr_in6); break; #endif default: break; } } } return (cnt); } static int sctp_count_max_addresses(struct sctp_inpcb *inp) { int cnt = 0; SCTP_IPI_ADDR_RLOCK(); /* count addresses for the endpoint's default VRF */ cnt = sctp_count_max_addresses_vrf(inp, inp->def_vrf_id); SCTP_IPI_ADDR_RUNLOCK(); return (cnt); } static int sctp_do_connect_x(struct socket *so, struct sctp_inpcb *inp, void *optval, size_t optsize, void *p, int delay) { int error = 0; int creat_lock_on = 0; struct sctp_tcb *stcb = NULL; struct sockaddr *sa; unsigned int num_v6 = 0, num_v4 = 0, *totaddrp, totaddr; uint32_t vrf_id; int bad_addresses = 0; sctp_assoc_t *a_id; SCTPDBG(SCTP_DEBUG_PCB1, "Connectx called\n"); if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) && (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED)) { /* We are already connected AND the TCP model */ SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, EADDRINUSE); return (EADDRINUSE); } if ((inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) && (sctp_is_feature_off(inp, SCTP_PCB_FLAGS_PORTREUSE))) { SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } if (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) { SCTP_INP_RLOCK(inp); stcb = LIST_FIRST(&inp->sctp_asoc_list); SCTP_INP_RUNLOCK(inp); } if (stcb) { SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, EALREADY); return (EALREADY); } SCTP_INP_INCR_REF(inp); SCTP_ASOC_CREATE_LOCK(inp); creat_lock_on = 1; if ((inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE) || (inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE)) { SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, EFAULT); error = EFAULT; goto out_now; } totaddrp = (unsigned int *)optval; totaddr = *totaddrp; sa = (struct sockaddr *)(totaddrp + 1); stcb = sctp_connectx_helper_find(inp, sa, &totaddr, &num_v4, &num_v6, &error, (unsigned int)(optsize - sizeof(int)), &bad_addresses); if ((stcb != NULL) || bad_addresses) { /* Already have or am bring up an association */ SCTP_ASOC_CREATE_UNLOCK(inp); creat_lock_on = 0; if (stcb) SCTP_TCB_UNLOCK(stcb); if (bad_addresses == 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EALREADY); error = EALREADY; } goto out_now; } #ifdef INET6 if (((inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) == 0) && (num_v6 > 0)) { error = EINVAL; goto out_now; } if ((inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) && (num_v4 > 0)) { struct in6pcb *inp6; inp6 = (struct in6pcb *)inp; if (SCTP_IPV6_V6ONLY(inp6)) { /* * if IPV6_V6ONLY flag, ignore connections destined * to a v4 addr or v4-mapped addr */ SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto out_now; } } #endif /* INET6 */ if ((inp->sctp_flags & SCTP_PCB_FLAGS_UNBOUND) == SCTP_PCB_FLAGS_UNBOUND) { /* Bind a ephemeral port */ error = sctp_inpcb_bind(so, NULL, NULL, p); if (error) { goto out_now; } } /* FIX ME: do we want to pass in a vrf on the connect call? */ vrf_id = inp->def_vrf_id; /* We are GOOD to go */ stcb = sctp_aloc_assoc(inp, sa, &error, 0, vrf_id, inp->sctp_ep.pre_open_stream_count, inp->sctp_ep.port, (struct thread *)p ); if (stcb == NULL) { /* Gak! no memory */ goto out_now; } if (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) { stcb->sctp_ep->sctp_flags |= SCTP_PCB_FLAGS_CONNECTED; /* Set the connected flag so we can queue data */ soisconnecting(so); } SCTP_SET_STATE(&stcb->asoc, SCTP_STATE_COOKIE_WAIT); /* move to second address */ switch (sa->sa_family) { #ifdef INET case AF_INET: sa = (struct sockaddr *)((caddr_t)sa + sizeof(struct sockaddr_in)); break; #endif #ifdef INET6 case AF_INET6: sa = (struct sockaddr *)((caddr_t)sa + sizeof(struct sockaddr_in6)); break; #endif default: break; } error = 0; sctp_connectx_helper_add(stcb, sa, (totaddr - 1), &error); /* Fill in the return id */ if (error) { (void)sctp_free_assoc(inp, stcb, SCTP_PCBFREE_FORCE, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_7); goto out_now; } a_id = (sctp_assoc_t *) optval; *a_id = sctp_get_associd(stcb); /* initialize authentication parameters for the assoc */ sctp_initialize_auth_params(inp, stcb); if (delay) { /* doing delayed connection */ stcb->asoc.delayed_connection = 1; sctp_timer_start(SCTP_TIMER_TYPE_INIT, inp, stcb, stcb->asoc.primary_destination); } else { (void)SCTP_GETTIME_TIMEVAL(&stcb->asoc.time_entered); sctp_send_initiate(inp, stcb, SCTP_SO_LOCKED); } SCTP_TCB_UNLOCK(stcb); if (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) { stcb->sctp_ep->sctp_flags |= SCTP_PCB_FLAGS_CONNECTED; /* Set the connected flag so we can queue data */ soisconnecting(so); } out_now: if (creat_lock_on) { SCTP_ASOC_CREATE_UNLOCK(inp); } SCTP_INP_DECR_REF(inp); return (error); } #define SCTP_FIND_STCB(inp, stcb, assoc_id) { \ if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) ||\ (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) { \ SCTP_INP_RLOCK(inp); \ stcb = LIST_FIRST(&inp->sctp_asoc_list); \ if (stcb) { \ SCTP_TCB_LOCK(stcb); \ } \ SCTP_INP_RUNLOCK(inp); \ } else if (assoc_id > SCTP_ALL_ASSOC) { \ stcb = sctp_findassociation_ep_asocid(inp, assoc_id, 1); \ if (stcb == NULL) { \ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); \ error = ENOENT; \ break; \ } \ } else { \ stcb = NULL; \ } \ } #define SCTP_CHECK_AND_CAST(destp, srcp, type, size) {\ if (size < sizeof(type)) { \ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); \ error = EINVAL; \ break; \ } else { \ destp = (type *)srcp; \ } \ } static int sctp_getopt(struct socket *so, int optname, void *optval, size_t *optsize, void *p) { struct sctp_inpcb *inp = NULL; int error, val = 0; struct sctp_tcb *stcb = NULL; if (optval == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return EINVAL; } error = 0; switch (optname) { case SCTP_NODELAY: case SCTP_AUTOCLOSE: case SCTP_EXPLICIT_EOR: case SCTP_AUTO_ASCONF: case SCTP_DISABLE_FRAGMENTS: case SCTP_I_WANT_MAPPED_V4_ADDR: case SCTP_USE_EXT_RCVINFO: SCTP_INP_RLOCK(inp); switch (optname) { case SCTP_DISABLE_FRAGMENTS: val = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NO_FRAGMENT); break; case SCTP_I_WANT_MAPPED_V4_ADDR: val = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4); break; case SCTP_AUTO_ASCONF: if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) { /* only valid for bound all sockets */ val = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_AUTO_ASCONF); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto flags_out; } break; case SCTP_EXPLICIT_EOR: val = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_EXPLICIT_EOR); break; case SCTP_NODELAY: val = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NODELAY); break; case SCTP_USE_EXT_RCVINFO: val = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_EXT_RCVINFO); break; case SCTP_AUTOCLOSE: if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_AUTOCLOSE)) val = TICKS_TO_SEC(inp->sctp_ep.auto_close_time); else val = 0; break; default: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOPROTOOPT); error = ENOPROTOOPT; } /* end switch (sopt->sopt_name) */ if (*optsize < sizeof(val)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } flags_out: SCTP_INP_RUNLOCK(inp); if (error == 0) { /* return the option value */ *(int *)optval = val; *optsize = sizeof(val); } break; case SCTP_GET_PACKET_LOG: { #ifdef SCTP_PACKET_LOGGING uint8_t *target; int ret; SCTP_CHECK_AND_CAST(target, optval, uint8_t, *optsize); ret = sctp_copy_out_packet_log(target, (int)*optsize); *optsize = ret; #else SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); error = EOPNOTSUPP; #endif break; } case SCTP_REUSE_PORT: { uint32_t *value; if ((inp->sctp_flags & SCTP_PCB_FLAGS_UDPTYPE)) { /* Can't do this for a 1-m socket */ error = EINVAL; break; } SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); *value = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_PORTREUSE); *optsize = sizeof(uint32_t); break; } case SCTP_PARTIAL_DELIVERY_POINT: { uint32_t *value; SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); *value = inp->partial_delivery_point; *optsize = sizeof(uint32_t); break; } case SCTP_FRAGMENT_INTERLEAVE: { uint32_t *value; SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_FRAG_INTERLEAVE)) { if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_INTERLEAVE_STRMS)) { *value = SCTP_FRAG_LEVEL_2; } else { *value = SCTP_FRAG_LEVEL_1; } } else { *value = SCTP_FRAG_LEVEL_0; } *optsize = sizeof(uint32_t); break; } case SCTP_INTERLEAVING_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.idata_supported; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); if (inp->idata_supported) { av->assoc_value = 1; } else { av->assoc_value = 0; } SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_CMT_ON_OFF: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.sctp_cmt_on_off; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); av->assoc_value = inp->sctp_cmt_on_off; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_PLUGGABLE_CC: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.congestion_control_module; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); av->assoc_value = inp->sctp_ep.sctp_default_cc_module; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_CC_OPTION: { struct sctp_cc_option *cc_opt; SCTP_CHECK_AND_CAST(cc_opt, optval, struct sctp_cc_option, *optsize); SCTP_FIND_STCB(inp, stcb, cc_opt->aid_value.assoc_id); if (stcb == NULL) { error = EINVAL; } else { if (stcb->asoc.cc_functions.sctp_cwnd_socket_option == NULL) { error = ENOTSUP; } else { error = (*stcb->asoc.cc_functions.sctp_cwnd_socket_option) (stcb, 0, cc_opt); *optsize = sizeof(struct sctp_cc_option); } SCTP_TCB_UNLOCK(stcb); } break; } case SCTP_PLUGGABLE_SS: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.stream_scheduling_module; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); av->assoc_value = inp->sctp_ep.sctp_default_ss_module; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_SS_VALUE: { struct sctp_stream_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_stream_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { if ((av->stream_id >= stcb->asoc.streamoutcnt) || (stcb->asoc.ss_functions.sctp_ss_get_value(stcb, &stcb->asoc, &stcb->asoc.strmout[av->stream_id], &av->stream_value) < 0)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } else { *optsize = sizeof(struct sctp_stream_value); } SCTP_TCB_UNLOCK(stcb); } else { /* * Can't get stream value without * association */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } break; } case SCTP_GET_ADDR_LEN: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); error = EINVAL; #ifdef INET if (av->assoc_value == AF_INET) { av->assoc_value = sizeof(struct sockaddr_in); error = 0; } #endif #ifdef INET6 if (av->assoc_value == AF_INET6) { av->assoc_value = sizeof(struct sockaddr_in6); error = 0; } #endif if (error) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); } else { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_GET_ASSOC_NUMBER: { uint32_t *value, cnt; SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); SCTP_INP_RLOCK(inp); if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) { /* Can't do this for a 1-1 socket */ error = EINVAL; SCTP_INP_RUNLOCK(inp); break; } cnt = 0; LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { cnt++; } SCTP_INP_RUNLOCK(inp); *value = cnt; *optsize = sizeof(uint32_t); break; } case SCTP_GET_ASSOC_ID_LIST: { struct sctp_assoc_ids *ids; uint32_t at; size_t limit; SCTP_CHECK_AND_CAST(ids, optval, struct sctp_assoc_ids, *optsize); SCTP_INP_RLOCK(inp); if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) { /* Can't do this for a 1-1 socket */ error = EINVAL; SCTP_INP_RUNLOCK(inp); break; } at = 0; limit = (*optsize - sizeof(uint32_t)) / sizeof(sctp_assoc_t); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { if (at < limit) { ids->gaids_assoc_id[at++] = sctp_get_associd(stcb); if (at == 0) { error = EINVAL; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else { error = EINVAL; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } SCTP_INP_RUNLOCK(inp); if (error == 0) { ids->gaids_number_of_ids = at; *optsize = ((at * sizeof(sctp_assoc_t)) + sizeof(uint32_t)); } break; } case SCTP_CONTEXT: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.context; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); av->assoc_value = inp->sctp_context; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_VRF_ID: { uint32_t *default_vrfid; SCTP_CHECK_AND_CAST(default_vrfid, optval, uint32_t, *optsize); *default_vrfid = inp->def_vrf_id; *optsize = sizeof(uint32_t); break; } case SCTP_GET_ASOC_VRF: { struct sctp_assoc_value *id; SCTP_CHECK_AND_CAST(id, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, id->assoc_id); if (stcb == NULL) { error = EINVAL; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); } else { id->assoc_value = stcb->asoc.vrf_id; *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_GET_VRF_IDS: { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); error = EOPNOTSUPP; break; } case SCTP_GET_NONCE_VALUES: { struct sctp_get_nonce_values *gnv; SCTP_CHECK_AND_CAST(gnv, optval, struct sctp_get_nonce_values, *optsize); SCTP_FIND_STCB(inp, stcb, gnv->gn_assoc_id); if (stcb) { gnv->gn_peers_tag = stcb->asoc.peer_vtag; gnv->gn_local_tag = stcb->asoc.my_vtag; SCTP_TCB_UNLOCK(stcb); *optsize = sizeof(struct sctp_get_nonce_values); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOTCONN); error = ENOTCONN; } break; } case SCTP_DELAYED_SACK: { struct sctp_sack_info *sack; SCTP_CHECK_AND_CAST(sack, optval, struct sctp_sack_info, *optsize); SCTP_FIND_STCB(inp, stcb, sack->sack_assoc_id); if (stcb) { sack->sack_delay = stcb->asoc.delayed_ack; sack->sack_freq = stcb->asoc.sack_freq; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (sack->sack_assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); sack->sack_delay = TICKS_TO_MSEC(inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_RECV]); sack->sack_freq = inp->sctp_ep.sctp_sack_freq; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_sack_info); } break; } case SCTP_GET_SNDBUF_USE: { struct sctp_sockstat *ss; SCTP_CHECK_AND_CAST(ss, optval, struct sctp_sockstat, *optsize); SCTP_FIND_STCB(inp, stcb, ss->ss_assoc_id); if (stcb) { ss->ss_total_sndbuf = stcb->asoc.total_output_queue_size; ss->ss_total_recv_buf = (stcb->asoc.size_on_reasm_queue + stcb->asoc.size_on_all_streams); SCTP_TCB_UNLOCK(stcb); *optsize = sizeof(struct sctp_sockstat); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOTCONN); error = ENOTCONN; } break; } case SCTP_MAX_BURST: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.max_burst; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); av->assoc_value = inp->sctp_ep.max_burst; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_MAXSEG: { struct sctp_assoc_value *av; int ovh; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = sctp_get_frag_point(stcb, &stcb->asoc); SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { ovh = SCTP_MED_OVERHEAD; } else { ovh = SCTP_MED_V4_OVERHEAD; } if (inp->sctp_frag_point >= SCTP_DEFAULT_MAXSEGMENT) av->assoc_value = 0; else av->assoc_value = inp->sctp_frag_point - ovh; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_GET_STAT_LOG: error = sctp_fill_stat_log(optval, optsize); break; case SCTP_EVENTS: { struct sctp_event_subscribe *events; SCTP_CHECK_AND_CAST(events, optval, struct sctp_event_subscribe, *optsize); memset(events, 0, sizeof(struct sctp_event_subscribe)); SCTP_INP_RLOCK(inp); if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVDATAIOEVNT)) events->sctp_data_io_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVASSOCEVNT)) events->sctp_association_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVPADDREVNT)) events->sctp_address_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVSENDFAILEVNT)) events->sctp_send_failure_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVPEERERR)) events->sctp_peer_error_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVSHUTDOWNEVNT)) events->sctp_shutdown_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_PDAPIEVNT)) events->sctp_partial_delivery_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_ADAPTATIONEVNT)) events->sctp_adaptation_layer_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_AUTHEVNT)) events->sctp_authentication_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_DRYEVNT)) events->sctp_sender_dry_event = 1; if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_STREAM_RESETEVNT)) events->sctp_stream_reset_event = 1; SCTP_INP_RUNLOCK(inp); *optsize = sizeof(struct sctp_event_subscribe); break; } case SCTP_ADAPTATION_LAYER: { uint32_t *value; SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); SCTP_INP_RLOCK(inp); *value = inp->sctp_ep.adaptation_layer_indicator; SCTP_INP_RUNLOCK(inp); *optsize = sizeof(uint32_t); break; } case SCTP_SET_INITIAL_DBG_SEQ: { uint32_t *value; SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); SCTP_INP_RLOCK(inp); *value = inp->sctp_ep.initial_sequence_debug; SCTP_INP_RUNLOCK(inp); *optsize = sizeof(uint32_t); break; } case SCTP_GET_LOCAL_ADDR_SIZE: { uint32_t *value; SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); SCTP_INP_RLOCK(inp); *value = sctp_count_max_addresses(inp); SCTP_INP_RUNLOCK(inp); *optsize = sizeof(uint32_t); break; } case SCTP_GET_REMOTE_ADDR_SIZE: { uint32_t *value; size_t size; struct sctp_nets *net; SCTP_CHECK_AND_CAST(value, optval, uint32_t, *optsize); /* FIXME MT: change to sctp_assoc_value? */ SCTP_FIND_STCB(inp, stcb, (sctp_assoc_t) * value); if (stcb) { size = 0; /* Count the sizes */ TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { switch (net->ro._l_addr.sa.sa_family) { #ifdef INET case AF_INET: #ifdef INET6 if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4)) { size += sizeof(struct sockaddr_in6); } else { size += sizeof(struct sockaddr_in); } #else size += sizeof(struct sockaddr_in); #endif break; #endif #ifdef INET6 case AF_INET6: size += sizeof(struct sockaddr_in6); break; #endif default: break; } } SCTP_TCB_UNLOCK(stcb); *value = (uint32_t) size; *optsize = sizeof(uint32_t); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOTCONN); error = ENOTCONN; } break; } case SCTP_GET_PEER_ADDRESSES: /* * Get the address information, an array is passed in to * fill up we pack it. */ { size_t cpsz, left; struct sockaddr_storage *sas; struct sctp_nets *net; struct sctp_getaddresses *saddr; SCTP_CHECK_AND_CAST(saddr, optval, struct sctp_getaddresses, *optsize); SCTP_FIND_STCB(inp, stcb, saddr->sget_assoc_id); if (stcb) { left = (*optsize) - sizeof(struct sctp_getaddresses); *optsize = sizeof(struct sctp_getaddresses); sas = (struct sockaddr_storage *)&saddr->addr[0]; TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { switch (net->ro._l_addr.sa.sa_family) { #ifdef INET case AF_INET: #ifdef INET6 if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4)) { cpsz = sizeof(struct sockaddr_in6); } else { cpsz = sizeof(struct sockaddr_in); } #else cpsz = sizeof(struct sockaddr_in); #endif break; #endif #ifdef INET6 case AF_INET6: cpsz = sizeof(struct sockaddr_in6); break; #endif default: cpsz = 0; break; } if (cpsz == 0) { break; } if (left < cpsz) { /* not enough room. */ break; } #if defined(INET) && defined(INET6) if ((sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4)) && (net->ro._l_addr.sa.sa_family == AF_INET)) { /* Must map the address */ in6_sin_2_v4mapsin6(&net->ro._l_addr.sin, (struct sockaddr_in6 *)sas); } else { memcpy(sas, &net->ro._l_addr, cpsz); } #else memcpy(sas, &net->ro._l_addr, cpsz); #endif ((struct sockaddr_in *)sas)->sin_port = stcb->rport; sas = (struct sockaddr_storage *)((caddr_t)sas + cpsz); left -= cpsz; *optsize += cpsz; } SCTP_TCB_UNLOCK(stcb); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); error = ENOENT; } break; } case SCTP_GET_LOCAL_ADDRESSES: { size_t limit, actual; struct sockaddr_storage *sas; struct sctp_getaddresses *saddr; SCTP_CHECK_AND_CAST(saddr, optval, struct sctp_getaddresses, *optsize); SCTP_FIND_STCB(inp, stcb, saddr->sget_assoc_id); sas = (struct sockaddr_storage *)&saddr->addr[0]; limit = *optsize - sizeof(sctp_assoc_t); actual = sctp_fill_up_addresses(inp, stcb, limit, sas); if (stcb) { SCTP_TCB_UNLOCK(stcb); } *optsize = sizeof(struct sockaddr_storage) + actual; break; } case SCTP_PEER_ADDR_PARAMS: { struct sctp_paddrparams *paddrp; struct sctp_nets *net; struct sockaddr *addr; #if defined(INET) && defined(INET6) struct sockaddr_in sin_store; #endif SCTP_CHECK_AND_CAST(paddrp, optval, struct sctp_paddrparams, *optsize); SCTP_FIND_STCB(inp, stcb, paddrp->spp_assoc_id); #if defined(INET) && defined(INET6) if (paddrp->spp_address.ss_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)&paddrp->spp_address; if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { in6_sin6_2_sin(&sin_store, sin6); addr = (struct sockaddr *)&sin_store; } else { addr = (struct sockaddr *)&paddrp->spp_address; } } else { addr = (struct sockaddr *)&paddrp->spp_address; } #else addr = (struct sockaddr *)&paddrp->spp_address; #endif if (stcb != NULL) { net = sctp_findnet(stcb, addr); } else { /* * We increment here since * sctp_findassociation_ep_addr() wil do a * decrement if it finds the stcb as long as * the locked tcb (last argument) is NOT a * TCB.. aka NULL. */ net = NULL; SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, addr, &net, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } } if ((stcb != NULL) && (net == NULL)) { #ifdef INET if (addr->sa_family == AF_INET) { struct sockaddr_in *sin; sin = (struct sockaddr_in *)addr; if (sin->sin_addr.s_addr != INADDR_ANY) { error = EINVAL; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif #ifdef INET6 if (addr->sa_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)addr; if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { error = EINVAL; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif { error = EAFNOSUPPORT; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } if (stcb != NULL) { /* Applies to the specific association */ paddrp->spp_flags = 0; if (net != NULL) { paddrp->spp_hbinterval = net->heart_beat_delay; paddrp->spp_pathmaxrxt = net->failure_threshold; paddrp->spp_pathmtu = net->mtu; switch (net->ro._l_addr.sa.sa_family) { #ifdef INET case AF_INET: paddrp->spp_pathmtu -= SCTP_MIN_V4_OVERHEAD; break; #endif #ifdef INET6 case AF_INET6: paddrp->spp_pathmtu -= SCTP_MIN_V4_OVERHEAD; break; #endif default: break; } /* get flags for HB */ if (net->dest_state & SCTP_ADDR_NOHB) { paddrp->spp_flags |= SPP_HB_DISABLE; } else { paddrp->spp_flags |= SPP_HB_ENABLE; } /* get flags for PMTU */ if (net->dest_state & SCTP_ADDR_NO_PMTUD) { paddrp->spp_flags |= SPP_PMTUD_DISABLE; } else { paddrp->spp_flags |= SPP_PMTUD_ENABLE; } if (net->dscp & 0x01) { paddrp->spp_dscp = net->dscp & 0xfc; paddrp->spp_flags |= SPP_DSCP; } #ifdef INET6 if ((net->ro._l_addr.sa.sa_family == AF_INET6) && (net->flowlabel & 0x80000000)) { paddrp->spp_ipv6_flowlabel = net->flowlabel & 0x000fffff; paddrp->spp_flags |= SPP_IPV6_FLOWLABEL; } #endif } else { /* * No destination so return default * value */ paddrp->spp_pathmaxrxt = stcb->asoc.def_net_failure; paddrp->spp_pathmtu = 0; if (stcb->asoc.default_dscp & 0x01) { paddrp->spp_dscp = stcb->asoc.default_dscp & 0xfc; paddrp->spp_flags |= SPP_DSCP; } #ifdef INET6 if (stcb->asoc.default_flowlabel & 0x80000000) { paddrp->spp_ipv6_flowlabel = stcb->asoc.default_flowlabel & 0x000fffff; paddrp->spp_flags |= SPP_IPV6_FLOWLABEL; } #endif /* default settings should be these */ if (sctp_stcb_is_feature_on(inp, stcb, SCTP_PCB_FLAGS_DONOT_HEARTBEAT)) { paddrp->spp_flags |= SPP_HB_DISABLE; } else { paddrp->spp_flags |= SPP_HB_ENABLE; } if (sctp_stcb_is_feature_on(inp, stcb, SCTP_PCB_FLAGS_DO_NOT_PMTUD)) { paddrp->spp_flags |= SPP_PMTUD_DISABLE; } else { paddrp->spp_flags |= SPP_PMTUD_ENABLE; } paddrp->spp_hbinterval = stcb->asoc.heart_beat_delay; } paddrp->spp_assoc_id = sctp_get_associd(stcb); SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (paddrp->spp_assoc_id == SCTP_FUTURE_ASSOC)) { /* Use endpoint defaults */ SCTP_INP_RLOCK(inp); paddrp->spp_pathmaxrxt = inp->sctp_ep.def_net_failure; paddrp->spp_hbinterval = TICKS_TO_MSEC(inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_HEARTBEAT]); paddrp->spp_assoc_id = SCTP_FUTURE_ASSOC; /* get inp's default */ if (inp->sctp_ep.default_dscp & 0x01) { paddrp->spp_dscp = inp->sctp_ep.default_dscp & 0xfc; paddrp->spp_flags |= SPP_DSCP; } #ifdef INET6 if ((inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) && (inp->sctp_ep.default_flowlabel & 0x80000000)) { paddrp->spp_ipv6_flowlabel = inp->sctp_ep.default_flowlabel & 0x000fffff; paddrp->spp_flags |= SPP_IPV6_FLOWLABEL; } #endif /* can't return this */ paddrp->spp_pathmtu = 0; if (sctp_is_feature_off(inp, SCTP_PCB_FLAGS_DONOT_HEARTBEAT)) { paddrp->spp_flags |= SPP_HB_ENABLE; } else { paddrp->spp_flags |= SPP_HB_DISABLE; } if (sctp_is_feature_off(inp, SCTP_PCB_FLAGS_DO_NOT_PMTUD)) { paddrp->spp_flags |= SPP_PMTUD_ENABLE; } else { paddrp->spp_flags |= SPP_PMTUD_DISABLE; } SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_paddrparams); } break; } case SCTP_GET_PEER_ADDR_INFO: { struct sctp_paddrinfo *paddri; struct sctp_nets *net; struct sockaddr *addr; #if defined(INET) && defined(INET6) struct sockaddr_in sin_store; #endif SCTP_CHECK_AND_CAST(paddri, optval, struct sctp_paddrinfo, *optsize); SCTP_FIND_STCB(inp, stcb, paddri->spinfo_assoc_id); #if defined(INET) && defined(INET6) if (paddri->spinfo_address.ss_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)&paddri->spinfo_address; if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { in6_sin6_2_sin(&sin_store, sin6); addr = (struct sockaddr *)&sin_store; } else { addr = (struct sockaddr *)&paddri->spinfo_address; } } else { addr = (struct sockaddr *)&paddri->spinfo_address; } #else addr = (struct sockaddr *)&paddri->spinfo_address; #endif if (stcb != NULL) { net = sctp_findnet(stcb, addr); } else { /* * We increment here since * sctp_findassociation_ep_addr() wil do a * decrement if it finds the stcb as long as * the locked tcb (last argument) is NOT a * TCB.. aka NULL. */ net = NULL; SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, addr, &net, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } } if ((stcb != NULL) && (net != NULL)) { if (net->dest_state & SCTP_ADDR_UNCONFIRMED) { /* It's unconfirmed */ paddri->spinfo_state = SCTP_UNCONFIRMED; } else if (net->dest_state & SCTP_ADDR_REACHABLE) { /* It's active */ paddri->spinfo_state = SCTP_ACTIVE; } else { /* It's inactive */ paddri->spinfo_state = SCTP_INACTIVE; } paddri->spinfo_cwnd = net->cwnd; paddri->spinfo_srtt = net->lastsa >> SCTP_RTT_SHIFT; paddri->spinfo_rto = net->RTO; paddri->spinfo_assoc_id = sctp_get_associd(stcb); paddri->spinfo_mtu = net->mtu; switch (addr->sa_family) { #if defined(INET) case AF_INET: paddri->spinfo_mtu -= SCTP_MIN_V4_OVERHEAD; break; #endif #if defined(INET6) case AF_INET6: paddri->spinfo_mtu -= SCTP_MIN_OVERHEAD; break; #endif default: break; } SCTP_TCB_UNLOCK(stcb); *optsize = sizeof(struct sctp_paddrinfo); } else { if (stcb != NULL) { SCTP_TCB_UNLOCK(stcb); } SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); error = ENOENT; } break; } case SCTP_PCB_STATUS: { struct sctp_pcbinfo *spcb; SCTP_CHECK_AND_CAST(spcb, optval, struct sctp_pcbinfo, *optsize); sctp_fill_pcbinfo(spcb); *optsize = sizeof(struct sctp_pcbinfo); break; } case SCTP_STATUS: { struct sctp_nets *net; struct sctp_status *sstat; SCTP_CHECK_AND_CAST(sstat, optval, struct sctp_status, *optsize); SCTP_FIND_STCB(inp, stcb, sstat->sstat_assoc_id); if (stcb == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } sstat->sstat_state = sctp_map_assoc_state(stcb->asoc.state); sstat->sstat_assoc_id = sctp_get_associd(stcb); sstat->sstat_rwnd = stcb->asoc.peers_rwnd; sstat->sstat_unackdata = stcb->asoc.sent_queue_cnt; /* * We can't include chunks that have been passed to * the socket layer. Only things in queue. */ sstat->sstat_penddata = (stcb->asoc.cnt_on_reasm_queue + stcb->asoc.cnt_on_all_streams); sstat->sstat_instrms = stcb->asoc.streamincnt; sstat->sstat_outstrms = stcb->asoc.streamoutcnt; sstat->sstat_fragmentation_point = sctp_get_frag_point(stcb, &stcb->asoc); memcpy(&sstat->sstat_primary.spinfo_address, &stcb->asoc.primary_destination->ro._l_addr, ((struct sockaddr *)(&stcb->asoc.primary_destination->ro._l_addr))->sa_len); net = stcb->asoc.primary_destination; ((struct sockaddr_in *)&sstat->sstat_primary.spinfo_address)->sin_port = stcb->rport; /* * Again the user can get info from sctp_constants.h * for what the state of the network is. */ if (net->dest_state & SCTP_ADDR_UNCONFIRMED) { /* It's unconfirmed */ sstat->sstat_primary.spinfo_state = SCTP_UNCONFIRMED; } else if (net->dest_state & SCTP_ADDR_REACHABLE) { /* It's active */ sstat->sstat_primary.spinfo_state = SCTP_ACTIVE; } else { /* It's inactive */ sstat->sstat_primary.spinfo_state = SCTP_INACTIVE; } sstat->sstat_primary.spinfo_cwnd = net->cwnd; sstat->sstat_primary.spinfo_srtt = net->lastsa >> SCTP_RTT_SHIFT; sstat->sstat_primary.spinfo_rto = net->RTO; sstat->sstat_primary.spinfo_mtu = net->mtu; switch (stcb->asoc.primary_destination->ro._l_addr.sa.sa_family) { #if defined(INET) case AF_INET: sstat->sstat_primary.spinfo_mtu -= SCTP_MIN_V4_OVERHEAD; break; #endif #if defined(INET6) case AF_INET6: sstat->sstat_primary.spinfo_mtu -= SCTP_MIN_OVERHEAD; break; #endif default: break; } sstat->sstat_primary.spinfo_assoc_id = sctp_get_associd(stcb); SCTP_TCB_UNLOCK(stcb); *optsize = sizeof(struct sctp_status); break; } case SCTP_RTOINFO: { struct sctp_rtoinfo *srto; SCTP_CHECK_AND_CAST(srto, optval, struct sctp_rtoinfo, *optsize); SCTP_FIND_STCB(inp, stcb, srto->srto_assoc_id); if (stcb) { srto->srto_initial = stcb->asoc.initial_rto; srto->srto_max = stcb->asoc.maxrto; srto->srto_min = stcb->asoc.minrto; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (srto->srto_assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); srto->srto_initial = inp->sctp_ep.initial_rto; srto->srto_max = inp->sctp_ep.sctp_maxrto; srto->srto_min = inp->sctp_ep.sctp_minrto; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_rtoinfo); } break; } case SCTP_TIMEOUTS: { struct sctp_timeouts *stimo; SCTP_CHECK_AND_CAST(stimo, optval, struct sctp_timeouts, *optsize); SCTP_FIND_STCB(inp, stcb, stimo->stimo_assoc_id); if (stcb) { stimo->stimo_init = stcb->asoc.timoinit; stimo->stimo_data = stcb->asoc.timodata; stimo->stimo_sack = stcb->asoc.timosack; stimo->stimo_shutdown = stcb->asoc.timoshutdown; stimo->stimo_heartbeat = stcb->asoc.timoheartbeat; stimo->stimo_cookie = stcb->asoc.timocookie; stimo->stimo_shutdownack = stcb->asoc.timoshutdownack; SCTP_TCB_UNLOCK(stcb); *optsize = sizeof(struct sctp_timeouts); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } break; } case SCTP_ASSOCINFO: { struct sctp_assocparams *sasoc; SCTP_CHECK_AND_CAST(sasoc, optval, struct sctp_assocparams, *optsize); SCTP_FIND_STCB(inp, stcb, sasoc->sasoc_assoc_id); if (stcb) { sasoc->sasoc_cookie_life = TICKS_TO_MSEC(stcb->asoc.cookie_life); sasoc->sasoc_asocmaxrxt = stcb->asoc.max_send_times; sasoc->sasoc_number_peer_destinations = stcb->asoc.numnets; sasoc->sasoc_peer_rwnd = stcb->asoc.peers_rwnd; sasoc->sasoc_local_rwnd = stcb->asoc.my_rwnd; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (sasoc->sasoc_assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); sasoc->sasoc_cookie_life = TICKS_TO_MSEC(inp->sctp_ep.def_cookie_life); sasoc->sasoc_asocmaxrxt = inp->sctp_ep.max_send_times; sasoc->sasoc_number_peer_destinations = 0; sasoc->sasoc_peer_rwnd = 0; sasoc->sasoc_local_rwnd = sbspace(&inp->sctp_socket->so_rcv); SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assocparams); } break; } case SCTP_DEFAULT_SEND_PARAM: { struct sctp_sndrcvinfo *s_info; SCTP_CHECK_AND_CAST(s_info, optval, struct sctp_sndrcvinfo, *optsize); SCTP_FIND_STCB(inp, stcb, s_info->sinfo_assoc_id); if (stcb) { memcpy(s_info, &stcb->asoc.def_send, sizeof(stcb->asoc.def_send)); SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (s_info->sinfo_assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); memcpy(s_info, &inp->def_send, sizeof(inp->def_send)); SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_sndrcvinfo); } break; } case SCTP_INITMSG: { struct sctp_initmsg *sinit; SCTP_CHECK_AND_CAST(sinit, optval, struct sctp_initmsg, *optsize); SCTP_INP_RLOCK(inp); sinit->sinit_num_ostreams = inp->sctp_ep.pre_open_stream_count; sinit->sinit_max_instreams = inp->sctp_ep.max_open_streams_intome; sinit->sinit_max_attempts = inp->sctp_ep.max_init_times; sinit->sinit_max_init_timeo = inp->sctp_ep.initial_init_rto_max; SCTP_INP_RUNLOCK(inp); *optsize = sizeof(struct sctp_initmsg); break; } case SCTP_PRIMARY_ADDR: /* we allow a "get" operation on this */ { struct sctp_setprim *ssp; SCTP_CHECK_AND_CAST(ssp, optval, struct sctp_setprim, *optsize); SCTP_FIND_STCB(inp, stcb, ssp->ssp_assoc_id); if (stcb) { union sctp_sockstore *addr; addr = &stcb->asoc.primary_destination->ro._l_addr; switch (addr->sa.sa_family) { #ifdef INET case AF_INET: #ifdef INET6 if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4)) { in6_sin_2_v4mapsin6(&addr->sin, (struct sockaddr_in6 *)&ssp->ssp_addr); } else { memcpy(&ssp->ssp_addr, &addr->sin, sizeof(struct sockaddr_in)); } #else memcpy(&ssp->ssp_addr, &addr->sin, sizeof(struct sockaddr_in)); #endif break; #endif #ifdef INET6 case AF_INET6: memcpy(&ssp->ssp_addr, &addr->sin6, sizeof(struct sockaddr_in6)); break; #endif default: break; } SCTP_TCB_UNLOCK(stcb); *optsize = sizeof(struct sctp_setprim); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } break; } case SCTP_HMAC_IDENT: { struct sctp_hmacalgo *shmac; sctp_hmaclist_t *hmaclist; uint32_t size; int i; SCTP_CHECK_AND_CAST(shmac, optval, struct sctp_hmacalgo, *optsize); SCTP_INP_RLOCK(inp); hmaclist = inp->sctp_ep.local_hmacs; if (hmaclist == NULL) { /* no HMACs to return */ *optsize = sizeof(*shmac); SCTP_INP_RUNLOCK(inp); break; } /* is there room for all of the hmac ids? */ size = sizeof(*shmac) + (hmaclist->num_algo * sizeof(shmac->shmac_idents[0])); if ((size_t)(*optsize) < size) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; SCTP_INP_RUNLOCK(inp); break; } /* copy in the list */ shmac->shmac_number_of_idents = hmaclist->num_algo; for (i = 0; i < hmaclist->num_algo; i++) { shmac->shmac_idents[i] = hmaclist->hmac[i]; } SCTP_INP_RUNLOCK(inp); *optsize = size; break; } case SCTP_AUTH_ACTIVE_KEY: { struct sctp_authkeyid *scact; SCTP_CHECK_AND_CAST(scact, optval, struct sctp_authkeyid, *optsize); SCTP_FIND_STCB(inp, stcb, scact->scact_assoc_id); if (stcb) { /* get the active key on the assoc */ scact->scact_keynumber = stcb->asoc.authinfo.active_keyid; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (scact->scact_assoc_id == SCTP_FUTURE_ASSOC)) { /* get the endpoint active key */ SCTP_INP_RLOCK(inp); scact->scact_keynumber = inp->sctp_ep.default_keyid; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_authkeyid); } break; } case SCTP_LOCAL_AUTH_CHUNKS: { struct sctp_authchunks *sac; sctp_auth_chklist_t *chklist = NULL; size_t size = 0; SCTP_CHECK_AND_CAST(sac, optval, struct sctp_authchunks, *optsize); SCTP_FIND_STCB(inp, stcb, sac->gauth_assoc_id); if (stcb) { /* get off the assoc */ chklist = stcb->asoc.local_auth_chunks; /* is there enough space? */ size = sctp_auth_get_chklist_size(chklist); if (*optsize < (sizeof(struct sctp_authchunks) + size)) { error = EINVAL; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); } else { /* copy in the chunks */ (void)sctp_serialize_auth_chunks(chklist, sac->gauth_chunks); sac->gauth_number_of_chunks = (uint32_t) size; *optsize = sizeof(struct sctp_authchunks) + size; } SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (sac->gauth_assoc_id == SCTP_FUTURE_ASSOC)) { /* get off the endpoint */ SCTP_INP_RLOCK(inp); chklist = inp->sctp_ep.local_auth_chunks; /* is there enough space? */ size = sctp_auth_get_chklist_size(chklist); if (*optsize < (sizeof(struct sctp_authchunks) + size)) { error = EINVAL; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); } else { /* copy in the chunks */ (void)sctp_serialize_auth_chunks(chklist, sac->gauth_chunks); sac->gauth_number_of_chunks = (uint32_t) size; *optsize = sizeof(struct sctp_authchunks) + size; } SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_PEER_AUTH_CHUNKS: { struct sctp_authchunks *sac; sctp_auth_chklist_t *chklist = NULL; size_t size = 0; SCTP_CHECK_AND_CAST(sac, optval, struct sctp_authchunks, *optsize); SCTP_FIND_STCB(inp, stcb, sac->gauth_assoc_id); if (stcb) { /* get off the assoc */ chklist = stcb->asoc.peer_auth_chunks; /* is there enough space? */ size = sctp_auth_get_chklist_size(chklist); if (*optsize < (sizeof(struct sctp_authchunks) + size)) { error = EINVAL; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); } else { /* copy in the chunks */ (void)sctp_serialize_auth_chunks(chklist, sac->gauth_chunks); sac->gauth_number_of_chunks = (uint32_t) size; *optsize = sizeof(struct sctp_authchunks) + size; } SCTP_TCB_UNLOCK(stcb); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); error = ENOENT; } break; } case SCTP_EVENT: { struct sctp_event *event; uint32_t event_type; SCTP_CHECK_AND_CAST(event, optval, struct sctp_event, *optsize); SCTP_FIND_STCB(inp, stcb, event->se_assoc_id); switch (event->se_type) { case SCTP_ASSOC_CHANGE: event_type = SCTP_PCB_FLAGS_RECVASSOCEVNT; break; case SCTP_PEER_ADDR_CHANGE: event_type = SCTP_PCB_FLAGS_RECVPADDREVNT; break; case SCTP_REMOTE_ERROR: event_type = SCTP_PCB_FLAGS_RECVPEERERR; break; case SCTP_SEND_FAILED: event_type = SCTP_PCB_FLAGS_RECVSENDFAILEVNT; break; case SCTP_SHUTDOWN_EVENT: event_type = SCTP_PCB_FLAGS_RECVSHUTDOWNEVNT; break; case SCTP_ADAPTATION_INDICATION: event_type = SCTP_PCB_FLAGS_ADAPTATIONEVNT; break; case SCTP_PARTIAL_DELIVERY_EVENT: event_type = SCTP_PCB_FLAGS_PDAPIEVNT; break; case SCTP_AUTHENTICATION_EVENT: event_type = SCTP_PCB_FLAGS_AUTHEVNT; break; case SCTP_STREAM_RESET_EVENT: event_type = SCTP_PCB_FLAGS_STREAM_RESETEVNT; break; case SCTP_SENDER_DRY_EVENT: event_type = SCTP_PCB_FLAGS_DRYEVNT; break; case SCTP_NOTIFICATIONS_STOPPED_EVENT: event_type = 0; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOTSUP); error = ENOTSUP; break; case SCTP_ASSOC_RESET_EVENT: event_type = SCTP_PCB_FLAGS_ASSOC_RESETEVNT; break; case SCTP_STREAM_CHANGE_EVENT: event_type = SCTP_PCB_FLAGS_STREAM_CHANGEEVNT; break; case SCTP_SEND_FAILED_EVENT: event_type = SCTP_PCB_FLAGS_RECVNSENDFAILEVNT; break; default: event_type = 0; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } if (event_type > 0) { if (stcb) { event->se_on = sctp_stcb_is_feature_on(inp, stcb, event_type); SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (event->se_assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); event->se_on = sctp_is_feature_on(inp, event_type); SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } } if (error == 0) { *optsize = sizeof(struct sctp_event); } break; } case SCTP_RECVRCVINFO: { int onoff; if (*optsize < sizeof(int)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } else { SCTP_INP_RLOCK(inp); onoff = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVRCVINFO); SCTP_INP_RUNLOCK(inp); } if (error == 0) { /* return the option value */ *(int *)optval = onoff; *optsize = sizeof(int); } break; } case SCTP_RECVNXTINFO: { int onoff; if (*optsize < sizeof(int)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } else { SCTP_INP_RLOCK(inp); onoff = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVNXTINFO); SCTP_INP_RUNLOCK(inp); } if (error == 0) { /* return the option value */ *(int *)optval = onoff; *optsize = sizeof(int); } break; } case SCTP_DEFAULT_SNDINFO: { struct sctp_sndinfo *info; SCTP_CHECK_AND_CAST(info, optval, struct sctp_sndinfo, *optsize); SCTP_FIND_STCB(inp, stcb, info->snd_assoc_id); if (stcb) { info->snd_sid = stcb->asoc.def_send.sinfo_stream; info->snd_flags = stcb->asoc.def_send.sinfo_flags; info->snd_flags &= 0xfff0; info->snd_ppid = stcb->asoc.def_send.sinfo_ppid; info->snd_context = stcb->asoc.def_send.sinfo_context; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (info->snd_assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); info->snd_sid = inp->def_send.sinfo_stream; info->snd_flags = inp->def_send.sinfo_flags; info->snd_flags &= 0xfff0; info->snd_ppid = inp->def_send.sinfo_ppid; info->snd_context = inp->def_send.sinfo_context; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_sndinfo); } break; } case SCTP_DEFAULT_PRINFO: { struct sctp_default_prinfo *info; SCTP_CHECK_AND_CAST(info, optval, struct sctp_default_prinfo, *optsize); SCTP_FIND_STCB(inp, stcb, info->pr_assoc_id); if (stcb) { info->pr_policy = PR_SCTP_POLICY(stcb->asoc.def_send.sinfo_flags); info->pr_value = stcb->asoc.def_send.sinfo_timetolive; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (info->pr_assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); info->pr_policy = PR_SCTP_POLICY(inp->def_send.sinfo_flags); info->pr_value = inp->def_send.sinfo_timetolive; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_default_prinfo); } break; } case SCTP_PEER_ADDR_THLDS: { struct sctp_paddrthlds *thlds; struct sctp_nets *net; struct sockaddr *addr; #if defined(INET) && defined(INET6) struct sockaddr_in sin_store; #endif SCTP_CHECK_AND_CAST(thlds, optval, struct sctp_paddrthlds, *optsize); SCTP_FIND_STCB(inp, stcb, thlds->spt_assoc_id); #if defined(INET) && defined(INET6) if (thlds->spt_address.ss_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)&thlds->spt_address; if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { in6_sin6_2_sin(&sin_store, sin6); addr = (struct sockaddr *)&sin_store; } else { addr = (struct sockaddr *)&thlds->spt_address; } } else { addr = (struct sockaddr *)&thlds->spt_address; } #else addr = (struct sockaddr *)&thlds->spt_address; #endif if (stcb != NULL) { net = sctp_findnet(stcb, addr); } else { /* * We increment here since * sctp_findassociation_ep_addr() wil do a * decrement if it finds the stcb as long as * the locked tcb (last argument) is NOT a * TCB.. aka NULL. */ net = NULL; SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, addr, &net, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } } if ((stcb != NULL) && (net == NULL)) { #ifdef INET if (addr->sa_family == AF_INET) { struct sockaddr_in *sin; sin = (struct sockaddr_in *)addr; if (sin->sin_addr.s_addr != INADDR_ANY) { error = EINVAL; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif #ifdef INET6 if (addr->sa_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)addr; if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { error = EINVAL; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif { error = EAFNOSUPPORT; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } if (stcb != NULL) { if (net != NULL) { thlds->spt_pathmaxrxt = net->failure_threshold; thlds->spt_pathpfthld = net->pf_threshold; thlds->spt_pathcpthld = 0xffff; } else { thlds->spt_pathmaxrxt = stcb->asoc.def_net_failure; thlds->spt_pathpfthld = stcb->asoc.def_net_pf_threshold; thlds->spt_pathcpthld = 0xffff; } thlds->spt_assoc_id = sctp_get_associd(stcb); SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (thlds->spt_assoc_id == SCTP_FUTURE_ASSOC)) { /* Use endpoint defaults */ SCTP_INP_RLOCK(inp); thlds->spt_pathmaxrxt = inp->sctp_ep.def_net_failure; thlds->spt_pathpfthld = inp->sctp_ep.def_net_pf_threshold; thlds->spt_pathcpthld = 0xffff; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_paddrthlds); } break; } case SCTP_REMOTE_UDP_ENCAPS_PORT: { struct sctp_udpencaps *encaps; struct sctp_nets *net; struct sockaddr *addr; #if defined(INET) && defined(INET6) struct sockaddr_in sin_store; #endif SCTP_CHECK_AND_CAST(encaps, optval, struct sctp_udpencaps, *optsize); SCTP_FIND_STCB(inp, stcb, encaps->sue_assoc_id); #if defined(INET) && defined(INET6) if (encaps->sue_address.ss_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)&encaps->sue_address; if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { in6_sin6_2_sin(&sin_store, sin6); addr = (struct sockaddr *)&sin_store; } else { addr = (struct sockaddr *)&encaps->sue_address; } } else { addr = (struct sockaddr *)&encaps->sue_address; } #else addr = (struct sockaddr *)&encaps->sue_address; #endif if (stcb) { net = sctp_findnet(stcb, addr); } else { /* * We increment here since * sctp_findassociation_ep_addr() wil do a * decrement if it finds the stcb as long as * the locked tcb (last argument) is NOT a * TCB.. aka NULL. */ net = NULL; SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, addr, &net, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } } if ((stcb != NULL) && (net == NULL)) { #ifdef INET if (addr->sa_family == AF_INET) { struct sockaddr_in *sin; sin = (struct sockaddr_in *)addr; if (sin->sin_addr.s_addr != INADDR_ANY) { error = EINVAL; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif #ifdef INET6 if (addr->sa_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)addr; if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { error = EINVAL; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif { error = EAFNOSUPPORT; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } if (stcb != NULL) { if (net) { encaps->sue_port = net->port; } else { encaps->sue_port = stcb->asoc.port; } SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (encaps->sue_assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); encaps->sue_port = inp->sctp_ep.port; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_udpencaps); } break; } case SCTP_ECN_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.ecn_supported; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); av->assoc_value = inp->ecn_supported; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_PR_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.prsctp_supported; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); av->assoc_value = inp->prsctp_supported; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_AUTH_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.auth_supported; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); av->assoc_value = inp->auth_supported; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_ASCONF_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.asconf_supported; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); av->assoc_value = inp->asconf_supported; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_RECONFIG_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.reconfig_supported; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); av->assoc_value = inp->reconfig_supported; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_NRSACK_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.nrsack_supported; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); av->assoc_value = inp->nrsack_supported; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_PKTDROP_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.pktdrop_supported; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); av->assoc_value = inp->pktdrop_supported; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_ENABLE_STREAM_RESET: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = (uint32_t) stcb->asoc.local_strreset_support; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); av->assoc_value = (uint32_t) inp->local_strreset_support; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } case SCTP_PR_STREAM_STATUS: { struct sctp_prstatus *sprstat; uint16_t sid; uint16_t policy; SCTP_CHECK_AND_CAST(sprstat, optval, struct sctp_prstatus, *optsize); SCTP_FIND_STCB(inp, stcb, sprstat->sprstat_assoc_id); sid = sprstat->sprstat_sid; policy = sprstat->sprstat_policy; #if defined(SCTP_DETAILED_STR_STATS) if ((stcb != NULL) && (sid < stcb->asoc.streamoutcnt) && (policy != SCTP_PR_SCTP_NONE) && ((policy <= SCTP_PR_SCTP_MAX) || (policy == SCTP_PR_SCTP_ALL))) { if (policy == SCTP_PR_SCTP_ALL) { sprstat->sprstat_abandoned_unsent = stcb->asoc.strmout[sid].abandoned_unsent[0]; sprstat->sprstat_abandoned_sent = stcb->asoc.strmout[sid].abandoned_sent[0]; } else { sprstat->sprstat_abandoned_unsent = stcb->asoc.strmout[sid].abandoned_unsent[policy]; sprstat->sprstat_abandoned_sent = stcb->asoc.strmout[sid].abandoned_sent[policy]; } #else if ((stcb != NULL) && (sid < stcb->asoc.streamoutcnt) && (policy == SCTP_PR_SCTP_ALL)) { sprstat->sprstat_abandoned_unsent = stcb->asoc.strmout[sid].abandoned_unsent[0]; sprstat->sprstat_abandoned_sent = stcb->asoc.strmout[sid].abandoned_sent[0]; #endif SCTP_TCB_UNLOCK(stcb); *optsize = sizeof(struct sctp_prstatus); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } break; } case SCTP_PR_ASSOC_STATUS: { struct sctp_prstatus *sprstat; uint16_t policy; SCTP_CHECK_AND_CAST(sprstat, optval, struct sctp_prstatus, *optsize); SCTP_FIND_STCB(inp, stcb, sprstat->sprstat_assoc_id); policy = sprstat->sprstat_policy; if ((stcb != NULL) && (policy != SCTP_PR_SCTP_NONE) && ((policy <= SCTP_PR_SCTP_MAX) || (policy == SCTP_PR_SCTP_ALL))) { if (policy == SCTP_PR_SCTP_ALL) { sprstat->sprstat_abandoned_unsent = stcb->asoc.abandoned_unsent[0]; sprstat->sprstat_abandoned_sent = stcb->asoc.abandoned_sent[0]; } else { sprstat->sprstat_abandoned_unsent = stcb->asoc.abandoned_unsent[policy]; sprstat->sprstat_abandoned_sent = stcb->asoc.abandoned_sent[policy]; } SCTP_TCB_UNLOCK(stcb); *optsize = sizeof(struct sctp_prstatus); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } break; } case SCTP_MAX_CWND: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, *optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { av->assoc_value = stcb->asoc.max_cwnd; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_RLOCK(inp); av->assoc_value = inp->max_cwnd; SCTP_INP_RUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } if (error == 0) { *optsize = sizeof(struct sctp_assoc_value); } break; } default: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOPROTOOPT); error = ENOPROTOOPT; break; } /* end switch (sopt->sopt_name) */ if (error) { *optsize = 0; } return (error); } static int sctp_setopt(struct socket *so, int optname, void *optval, size_t optsize, void *p) { int error, set_opt; uint32_t *mopt; struct sctp_tcb *stcb = NULL; struct sctp_inpcb *inp = NULL; uint32_t vrf_id; if (optval == NULL) { SCTP_PRINTF("optval is NULL\n"); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { SCTP_PRINTF("inp is NULL?\n"); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } vrf_id = inp->def_vrf_id; error = 0; switch (optname) { case SCTP_NODELAY: case SCTP_AUTOCLOSE: case SCTP_AUTO_ASCONF: case SCTP_EXPLICIT_EOR: case SCTP_DISABLE_FRAGMENTS: case SCTP_USE_EXT_RCVINFO: case SCTP_I_WANT_MAPPED_V4_ADDR: /* copy in the option value */ SCTP_CHECK_AND_CAST(mopt, optval, uint32_t, optsize); set_opt = 0; if (error) break; switch (optname) { case SCTP_DISABLE_FRAGMENTS: set_opt = SCTP_PCB_FLAGS_NO_FRAGMENT; break; case SCTP_AUTO_ASCONF: /* * NOTE: we don't really support this flag */ if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) { /* only valid for bound all sockets */ if ((SCTP_BASE_SYSCTL(sctp_auto_asconf) == 0) && (*mopt != 0)) { /* forbidden by admin */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EPERM); return (EPERM); } set_opt = SCTP_PCB_FLAGS_AUTO_ASCONF; } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } break; case SCTP_EXPLICIT_EOR: set_opt = SCTP_PCB_FLAGS_EXPLICIT_EOR; break; case SCTP_USE_EXT_RCVINFO: set_opt = SCTP_PCB_FLAGS_EXT_RCVINFO; break; case SCTP_I_WANT_MAPPED_V4_ADDR: if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { set_opt = SCTP_PCB_FLAGS_NEEDS_MAPPED_V4; } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } break; case SCTP_NODELAY: set_opt = SCTP_PCB_FLAGS_NODELAY; break; case SCTP_AUTOCLOSE: if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } set_opt = SCTP_PCB_FLAGS_AUTOCLOSE; /* * The value is in ticks. Note this does not effect * old associations, only new ones. */ inp->sctp_ep.auto_close_time = SEC_TO_TICKS(*mopt); break; } SCTP_INP_WLOCK(inp); if (*mopt != 0) { sctp_feature_on(inp, set_opt); } else { sctp_feature_off(inp, set_opt); } SCTP_INP_WUNLOCK(inp); break; case SCTP_REUSE_PORT: { SCTP_CHECK_AND_CAST(mopt, optval, uint32_t, optsize); if ((inp->sctp_flags & SCTP_PCB_FLAGS_UNBOUND) == 0) { /* Can't set it after we are bound */ error = EINVAL; break; } if ((inp->sctp_flags & SCTP_PCB_FLAGS_UDPTYPE)) { /* Can't do this for a 1-m socket */ error = EINVAL; break; } if (optval) sctp_feature_on(inp, SCTP_PCB_FLAGS_PORTREUSE); else sctp_feature_off(inp, SCTP_PCB_FLAGS_PORTREUSE); break; } case SCTP_PARTIAL_DELIVERY_POINT: { uint32_t *value; SCTP_CHECK_AND_CAST(value, optval, uint32_t, optsize); if (*value > SCTP_SB_LIMIT_RCV(so)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } inp->partial_delivery_point = *value; break; } case SCTP_FRAGMENT_INTERLEAVE: /* not yet until we re-write sctp_recvmsg() */ { uint32_t *level; SCTP_CHECK_AND_CAST(level, optval, uint32_t, optsize); if (*level == SCTP_FRAG_LEVEL_2) { sctp_feature_on(inp, SCTP_PCB_FLAGS_FRAG_INTERLEAVE); sctp_feature_on(inp, SCTP_PCB_FLAGS_INTERLEAVE_STRMS); } else if (*level == SCTP_FRAG_LEVEL_1) { sctp_feature_on(inp, SCTP_PCB_FLAGS_FRAG_INTERLEAVE); sctp_feature_off(inp, SCTP_PCB_FLAGS_INTERLEAVE_STRMS); } else if (*level == SCTP_FRAG_LEVEL_0) { sctp_feature_off(inp, SCTP_PCB_FLAGS_FRAG_INTERLEAVE); sctp_feature_off(inp, SCTP_PCB_FLAGS_INTERLEAVE_STRMS); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } break; } case SCTP_INTERLEAVING_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_WLOCK(inp); if (av->assoc_value == 0) { inp->idata_supported = 0; } else { if ((sctp_is_feature_on(inp, SCTP_PCB_FLAGS_FRAG_INTERLEAVE)) && (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_INTERLEAVE_STRMS))) { inp->idata_supported = 1; } else { /* * Must have Frag * interleave and * stream interleave * on */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } SCTP_INP_WUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_CMT_ON_OFF: if (SCTP_BASE_SYSCTL(sctp_cmt_on_off)) { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); if (av->assoc_value > SCTP_CMT_MAX) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { stcb->asoc.sctp_cmt_on_off = av->assoc_value; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC) || (av->assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_WLOCK(inp); inp->sctp_cmt_on_off = av->assoc_value; SCTP_INP_WUNLOCK(inp); } if ((av->assoc_id == SCTP_CURRENT_ASSOC) || (av->assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); stcb->asoc.sctp_cmt_on_off = av->assoc_value; SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } } } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOPROTOOPT); error = ENOPROTOOPT; } break; case SCTP_PLUGGABLE_CC: { struct sctp_assoc_value *av; struct sctp_nets *net; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); if ((av->assoc_value != SCTP_CC_RFC2581) && (av->assoc_value != SCTP_CC_HSTCP) && (av->assoc_value != SCTP_CC_HTCP) && (av->assoc_value != SCTP_CC_RTCC)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { stcb->asoc.cc_functions = sctp_cc_functions[av->assoc_value]; stcb->asoc.congestion_control_module = av->assoc_value; if (stcb->asoc.cc_functions.sctp_set_initial_cc_param != NULL) { TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { stcb->asoc.cc_functions.sctp_set_initial_cc_param(stcb, net); } } SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC) || (av->assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_WLOCK(inp); inp->sctp_ep.sctp_default_cc_module = av->assoc_value; SCTP_INP_WUNLOCK(inp); } if ((av->assoc_id == SCTP_CURRENT_ASSOC) || (av->assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); stcb->asoc.cc_functions = sctp_cc_functions[av->assoc_value]; stcb->asoc.congestion_control_module = av->assoc_value; if (stcb->asoc.cc_functions.sctp_set_initial_cc_param != NULL) { TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { stcb->asoc.cc_functions.sctp_set_initial_cc_param(stcb, net); } } SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } } break; } case SCTP_CC_OPTION: { struct sctp_cc_option *cc_opt; SCTP_CHECK_AND_CAST(cc_opt, optval, struct sctp_cc_option, optsize); SCTP_FIND_STCB(inp, stcb, cc_opt->aid_value.assoc_id); if (stcb == NULL) { if (cc_opt->aid_value.assoc_id == SCTP_CURRENT_ASSOC) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); if (stcb->asoc.cc_functions.sctp_cwnd_socket_option) { (*stcb->asoc.cc_functions.sctp_cwnd_socket_option) (stcb, 1, cc_opt); } SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } else { error = EINVAL; } } else { if (stcb->asoc.cc_functions.sctp_cwnd_socket_option == NULL) { error = ENOTSUP; } else { error = (*stcb->asoc.cc_functions.sctp_cwnd_socket_option) (stcb, 1, cc_opt); } SCTP_TCB_UNLOCK(stcb); } break; } case SCTP_PLUGGABLE_SS: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); if ((av->assoc_value != SCTP_SS_DEFAULT) && (av->assoc_value != SCTP_SS_ROUND_ROBIN) && (av->assoc_value != SCTP_SS_ROUND_ROBIN_PACKET) && (av->assoc_value != SCTP_SS_PRIORITY) && (av->assoc_value != SCTP_SS_FAIR_BANDWITH) && (av->assoc_value != SCTP_SS_FIRST_COME)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { stcb->asoc.ss_functions.sctp_ss_clear(stcb, &stcb->asoc, 1, 1); stcb->asoc.ss_functions = sctp_ss_functions[av->assoc_value]; stcb->asoc.stream_scheduling_module = av->assoc_value; stcb->asoc.ss_functions.sctp_ss_init(stcb, &stcb->asoc, 1); SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC) || (av->assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_WLOCK(inp); inp->sctp_ep.sctp_default_ss_module = av->assoc_value; SCTP_INP_WUNLOCK(inp); } if ((av->assoc_id == SCTP_CURRENT_ASSOC) || (av->assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); stcb->asoc.ss_functions.sctp_ss_clear(stcb, &stcb->asoc, 1, 1); stcb->asoc.ss_functions = sctp_ss_functions[av->assoc_value]; stcb->asoc.stream_scheduling_module = av->assoc_value; stcb->asoc.ss_functions.sctp_ss_init(stcb, &stcb->asoc, 1); SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } } break; } case SCTP_SS_VALUE: { struct sctp_stream_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_stream_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { if ((av->stream_id >= stcb->asoc.streamoutcnt) || (stcb->asoc.ss_functions.sctp_ss_set_value(stcb, &stcb->asoc, &stcb->asoc.strmout[av->stream_id], av->stream_value) < 0)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_TCB_UNLOCK(stcb); } else { if (av->assoc_id == SCTP_CURRENT_ASSOC) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); if (av->stream_id < stcb->asoc.streamoutcnt) { stcb->asoc.ss_functions.sctp_ss_set_value(stcb, &stcb->asoc, &stcb->asoc.strmout[av->stream_id], av->stream_value); } SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } else { /* * Can't set stream value without * association */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_CLR_STAT_LOG: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); error = EOPNOTSUPP; break; case SCTP_CONTEXT: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { stcb->asoc.context = av->assoc_value; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC) || (av->assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_WLOCK(inp); inp->sctp_context = av->assoc_value; SCTP_INP_WUNLOCK(inp); } if ((av->assoc_id == SCTP_CURRENT_ASSOC) || (av->assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); stcb->asoc.context = av->assoc_value; SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } } break; } case SCTP_VRF_ID: { uint32_t *default_vrfid; SCTP_CHECK_AND_CAST(default_vrfid, optval, uint32_t, optsize); if (*default_vrfid > SCTP_MAX_VRF_ID) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } inp->def_vrf_id = *default_vrfid; break; } case SCTP_DEL_VRF_ID: { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); error = EOPNOTSUPP; break; } case SCTP_ADD_VRF_ID: { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); error = EOPNOTSUPP; break; } case SCTP_DELAYED_SACK: { struct sctp_sack_info *sack; SCTP_CHECK_AND_CAST(sack, optval, struct sctp_sack_info, optsize); SCTP_FIND_STCB(inp, stcb, sack->sack_assoc_id); if (sack->sack_delay) { if (sack->sack_delay > SCTP_MAX_SACK_DELAY) sack->sack_delay = SCTP_MAX_SACK_DELAY; if (MSEC_TO_TICKS(sack->sack_delay) < 1) { sack->sack_delay = TICKS_TO_MSEC(1); } } if (stcb) { if (sack->sack_delay) { stcb->asoc.delayed_ack = sack->sack_delay; } if (sack->sack_freq) { stcb->asoc.sack_freq = sack->sack_freq; } SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (sack->sack_assoc_id == SCTP_FUTURE_ASSOC) || (sack->sack_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_WLOCK(inp); if (sack->sack_delay) { inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_RECV] = MSEC_TO_TICKS(sack->sack_delay); } if (sack->sack_freq) { inp->sctp_ep.sctp_sack_freq = sack->sack_freq; } SCTP_INP_WUNLOCK(inp); } if ((sack->sack_assoc_id == SCTP_CURRENT_ASSOC) || (sack->sack_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); if (sack->sack_delay) { stcb->asoc.delayed_ack = sack->sack_delay; } if (sack->sack_freq) { stcb->asoc.sack_freq = sack->sack_freq; } SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } } break; } case SCTP_AUTH_CHUNK: { struct sctp_authchunk *sauth; SCTP_CHECK_AND_CAST(sauth, optval, struct sctp_authchunk, optsize); SCTP_INP_WLOCK(inp); if (sctp_auth_add_chunk(sauth->sauth_chunk, inp->sctp_ep.local_auth_chunks)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_INP_WUNLOCK(inp); break; } case SCTP_AUTH_KEY: { struct sctp_authkey *sca; struct sctp_keyhead *shared_keys; sctp_sharedkey_t *shared_key; sctp_key_t *key = NULL; size_t size; SCTP_CHECK_AND_CAST(sca, optval, struct sctp_authkey, optsize); if (sca->sca_keylength == 0) { size = optsize - sizeof(struct sctp_authkey); } else { if (sca->sca_keylength + sizeof(struct sctp_authkey) <= optsize) { size = sca->sca_keylength; } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } } SCTP_FIND_STCB(inp, stcb, sca->sca_assoc_id); if (stcb) { shared_keys = &stcb->asoc.shared_keys; /* clear the cached keys for this key id */ sctp_clear_cachedkeys(stcb, sca->sca_keynumber); /* * create the new shared key and * insert/replace it */ if (size > 0) { key = sctp_set_key(sca->sca_key, (uint32_t) size); if (key == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOMEM); error = ENOMEM; SCTP_TCB_UNLOCK(stcb); break; } } shared_key = sctp_alloc_sharedkey(); if (shared_key == NULL) { sctp_free_key(key); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOMEM); error = ENOMEM; SCTP_TCB_UNLOCK(stcb); break; } shared_key->key = key; shared_key->keyid = sca->sca_keynumber; error = sctp_insert_sharedkey(shared_keys, shared_key); SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (sca->sca_assoc_id == SCTP_FUTURE_ASSOC) || (sca->sca_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_WLOCK(inp); shared_keys = &inp->sctp_ep.shared_keys; /* * clear the cached keys on all * assocs for this key id */ sctp_clear_cachedkeys_ep(inp, sca->sca_keynumber); /* * create the new shared key and * insert/replace it */ if (size > 0) { key = sctp_set_key(sca->sca_key, (uint32_t) size); if (key == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOMEM); error = ENOMEM; SCTP_INP_WUNLOCK(inp); break; } } shared_key = sctp_alloc_sharedkey(); if (shared_key == NULL) { sctp_free_key(key); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOMEM); error = ENOMEM; SCTP_INP_WUNLOCK(inp); break; } shared_key->key = key; shared_key->keyid = sca->sca_keynumber; error = sctp_insert_sharedkey(shared_keys, shared_key); SCTP_INP_WUNLOCK(inp); } if ((sca->sca_assoc_id == SCTP_CURRENT_ASSOC) || (sca->sca_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); shared_keys = &stcb->asoc.shared_keys; /* * clear the cached keys for * this key id */ sctp_clear_cachedkeys(stcb, sca->sca_keynumber); /* * create the new shared key * and insert/replace it */ if (size > 0) { key = sctp_set_key(sca->sca_key, (uint32_t) size); if (key == NULL) { SCTP_TCB_UNLOCK(stcb); continue; } } shared_key = sctp_alloc_sharedkey(); if (shared_key == NULL) { sctp_free_key(key); SCTP_TCB_UNLOCK(stcb); continue; } shared_key->key = key; shared_key->keyid = sca->sca_keynumber; error = sctp_insert_sharedkey(shared_keys, shared_key); SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } } break; } case SCTP_HMAC_IDENT: { struct sctp_hmacalgo *shmac; sctp_hmaclist_t *hmaclist; uint16_t hmacid; uint32_t i; SCTP_CHECK_AND_CAST(shmac, optval, struct sctp_hmacalgo, optsize); if ((optsize < sizeof(struct sctp_hmacalgo) + shmac->shmac_number_of_idents * sizeof(uint16_t)) || (shmac->shmac_number_of_idents > 0xffff)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } hmaclist = sctp_alloc_hmaclist((uint16_t) shmac->shmac_number_of_idents); if (hmaclist == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOMEM); error = ENOMEM; break; } for (i = 0; i < shmac->shmac_number_of_idents; i++) { hmacid = shmac->shmac_idents[i]; if (sctp_auth_add_hmacid(hmaclist, hmacid)) { /* invalid HMACs were found */ ; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; sctp_free_hmaclist(hmaclist); goto sctp_set_hmac_done; } } for (i = 0; i < hmaclist->num_algo; i++) { if (hmaclist->hmac[i] == SCTP_AUTH_HMAC_ID_SHA1) { /* already in list */ break; } } if (i == hmaclist->num_algo) { /* not found in list */ sctp_free_hmaclist(hmaclist); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } /* set it on the endpoint */ SCTP_INP_WLOCK(inp); if (inp->sctp_ep.local_hmacs) sctp_free_hmaclist(inp->sctp_ep.local_hmacs); inp->sctp_ep.local_hmacs = hmaclist; SCTP_INP_WUNLOCK(inp); sctp_set_hmac_done: break; } case SCTP_AUTH_ACTIVE_KEY: { struct sctp_authkeyid *scact; SCTP_CHECK_AND_CAST(scact, optval, struct sctp_authkeyid, optsize); SCTP_FIND_STCB(inp, stcb, scact->scact_assoc_id); /* set the active key on the right place */ if (stcb) { /* set the active key on the assoc */ if (sctp_auth_setactivekey(stcb, scact->scact_keynumber)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (scact->scact_assoc_id == SCTP_FUTURE_ASSOC) || (scact->scact_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_WLOCK(inp); if (sctp_auth_setactivekey_ep(inp, scact->scact_keynumber)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_INP_WUNLOCK(inp); } if ((scact->scact_assoc_id == SCTP_CURRENT_ASSOC) || (scact->scact_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); sctp_auth_setactivekey(stcb, scact->scact_keynumber); SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } } break; } case SCTP_AUTH_DELETE_KEY: { struct sctp_authkeyid *scdel; SCTP_CHECK_AND_CAST(scdel, optval, struct sctp_authkeyid, optsize); SCTP_FIND_STCB(inp, stcb, scdel->scact_assoc_id); /* delete the key from the right place */ if (stcb) { if (sctp_delete_sharedkey(stcb, scdel->scact_keynumber)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (scdel->scact_assoc_id == SCTP_FUTURE_ASSOC) || (scdel->scact_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_WLOCK(inp); if (sctp_delete_sharedkey_ep(inp, scdel->scact_keynumber)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_INP_WUNLOCK(inp); } if ((scdel->scact_assoc_id == SCTP_CURRENT_ASSOC) || (scdel->scact_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); sctp_delete_sharedkey(stcb, scdel->scact_keynumber); SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } } break; } case SCTP_AUTH_DEACTIVATE_KEY: { struct sctp_authkeyid *keyid; SCTP_CHECK_AND_CAST(keyid, optval, struct sctp_authkeyid, optsize); SCTP_FIND_STCB(inp, stcb, keyid->scact_assoc_id); /* deactivate the key from the right place */ if (stcb) { if (sctp_deact_sharedkey(stcb, keyid->scact_keynumber)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (keyid->scact_assoc_id == SCTP_FUTURE_ASSOC) || (keyid->scact_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_WLOCK(inp); if (sctp_deact_sharedkey_ep(inp, keyid->scact_keynumber)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_INP_WUNLOCK(inp); } if ((keyid->scact_assoc_id == SCTP_CURRENT_ASSOC) || (keyid->scact_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); sctp_deact_sharedkey(stcb, keyid->scact_keynumber); SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } } break; } case SCTP_ENABLE_STREAM_RESET: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); if (av->assoc_value & (~SCTP_ENABLE_VALUE_MASK)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { stcb->asoc.local_strreset_support = (uint8_t) av->assoc_value; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC) || (av->assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_WLOCK(inp); inp->local_strreset_support = (uint8_t) av->assoc_value; SCTP_INP_WUNLOCK(inp); } if ((av->assoc_id == SCTP_CURRENT_ASSOC) || (av->assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); stcb->asoc.local_strreset_support = (uint8_t) av->assoc_value; SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } } break; } case SCTP_RESET_STREAMS: { struct sctp_reset_streams *strrst; int i, send_out = 0; int send_in = 0; SCTP_CHECK_AND_CAST(strrst, optval, struct sctp_reset_streams, optsize); SCTP_FIND_STCB(inp, stcb, strrst->srs_assoc_id); if (stcb == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); error = ENOENT; break; } if (stcb->asoc.reconfig_supported == 0) { /* * Peer does not support the chunk type. */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); error = EOPNOTSUPP; SCTP_TCB_UNLOCK(stcb); break; } if (sizeof(struct sctp_reset_streams) + strrst->srs_number_streams * sizeof(uint16_t) > optsize) { error = EINVAL; SCTP_TCB_UNLOCK(stcb); break; } if (strrst->srs_flags & SCTP_STREAM_RESET_INCOMING) { send_in = 1; if (stcb->asoc.stream_reset_outstanding) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EALREADY); error = EALREADY; SCTP_TCB_UNLOCK(stcb); break; } } if (strrst->srs_flags & SCTP_STREAM_RESET_OUTGOING) { send_out = 1; } if ((strrst->srs_number_streams > SCTP_MAX_STREAMS_AT_ONCE_RESET) && send_in) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOMEM); error = ENOMEM; SCTP_TCB_UNLOCK(stcb); break; } if ((send_in == 0) && (send_out == 0)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; SCTP_TCB_UNLOCK(stcb); break; } for (i = 0; i < strrst->srs_number_streams; i++) { if ((send_in) && (strrst->srs_stream_list[i] > stcb->asoc.streamincnt)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } if ((send_out) && (strrst->srs_stream_list[i] > stcb->asoc.streamoutcnt)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } } if (error) { SCTP_TCB_UNLOCK(stcb); break; } if (send_out) { int cnt; uint16_t strm; if (strrst->srs_number_streams) { for (i = 0, cnt = 0; i < strrst->srs_number_streams; i++) { strm = strrst->srs_stream_list[i]; if (stcb->asoc.strmout[strm].state == SCTP_STREAM_OPEN) { stcb->asoc.strmout[strm].state = SCTP_STREAM_RESET_PENDING; cnt++; } } } else { /* Its all */ for (i = 0, cnt = 0; i < stcb->asoc.streamoutcnt; i++) { if (stcb->asoc.strmout[i].state == SCTP_STREAM_OPEN) { stcb->asoc.strmout[i].state = SCTP_STREAM_RESET_PENDING; cnt++; } } } } if (send_in) { error = sctp_send_str_reset_req(stcb, strrst->srs_number_streams, strrst->srs_stream_list, send_in, 0, 0, 0, 0, 0); } else { error = sctp_send_stream_reset_out_if_possible(stcb, SCTP_SO_LOCKED); } if (error == 0) { sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_STRRST_REQ, SCTP_SO_LOCKED); } else { /* * For outgoing streams don't report any * problems in sending the request to the * application. XXX: Double check resetting * incoming streams. */ error = 0; } SCTP_TCB_UNLOCK(stcb); break; } case SCTP_ADD_STREAMS: { struct sctp_add_streams *stradd; uint8_t addstream = 0; uint16_t add_o_strmcnt = 0; uint16_t add_i_strmcnt = 0; SCTP_CHECK_AND_CAST(stradd, optval, struct sctp_add_streams, optsize); SCTP_FIND_STCB(inp, stcb, stradd->sas_assoc_id); if (stcb == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); error = ENOENT; break; } if (stcb->asoc.reconfig_supported == 0) { /* * Peer does not support the chunk type. */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); error = EOPNOTSUPP; SCTP_TCB_UNLOCK(stcb); break; } if (stcb->asoc.stream_reset_outstanding) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EALREADY); error = EALREADY; SCTP_TCB_UNLOCK(stcb); break; } if ((stradd->sas_outstrms == 0) && (stradd->sas_instrms == 0)) { error = EINVAL; goto skip_stuff; } if (stradd->sas_outstrms) { addstream = 1; /* We allocate here */ add_o_strmcnt = stradd->sas_outstrms; if ((((int)add_o_strmcnt) + ((int)stcb->asoc.streamoutcnt)) > 0x0000ffff) { /* You can't have more than 64k */ error = EINVAL; goto skip_stuff; } } if (stradd->sas_instrms) { int cnt; addstream |= 2; /* * We allocate inside * sctp_send_str_reset_req() */ add_i_strmcnt = stradd->sas_instrms; cnt = add_i_strmcnt; cnt += stcb->asoc.streamincnt; if (cnt > 0x0000ffff) { /* You can't have more than 64k */ error = EINVAL; goto skip_stuff; } if (cnt > (int)stcb->asoc.max_inbound_streams) { /* More than you are allowed */ error = EINVAL; goto skip_stuff; } } error = sctp_send_str_reset_req(stcb, 0, NULL, 0, 0, addstream, add_o_strmcnt, add_i_strmcnt, 0); sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_STRRST_REQ, SCTP_SO_LOCKED); skip_stuff: SCTP_TCB_UNLOCK(stcb); break; } case SCTP_RESET_ASSOC: { int i; uint32_t *value; SCTP_CHECK_AND_CAST(value, optval, uint32_t, optsize); SCTP_FIND_STCB(inp, stcb, (sctp_assoc_t) * value); if (stcb == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); error = ENOENT; break; } if (stcb->asoc.reconfig_supported == 0) { /* * Peer does not support the chunk type. */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); error = EOPNOTSUPP; SCTP_TCB_UNLOCK(stcb); break; } if (stcb->asoc.stream_reset_outstanding) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EALREADY); error = EALREADY; SCTP_TCB_UNLOCK(stcb); break; } /* * Is there any data pending in the send or sent * queues? */ if (!TAILQ_EMPTY(&stcb->asoc.send_queue) || !TAILQ_EMPTY(&stcb->asoc.sent_queue)) { busy_out: error = EBUSY; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); SCTP_TCB_UNLOCK(stcb); break; } /* Do any streams have data queued? */ for (i = 0; i < stcb->asoc.streamoutcnt; i++) { if (!TAILQ_EMPTY(&stcb->asoc.strmout[i].outqueue)) { goto busy_out; } } error = sctp_send_str_reset_req(stcb, 0, NULL, 0, 1, 0, 0, 0, 0); sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_STRRST_REQ, SCTP_SO_LOCKED); SCTP_TCB_UNLOCK(stcb); break; } case SCTP_CONNECT_X: if (optsize < (sizeof(int) + sizeof(struct sockaddr_in))) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } error = sctp_do_connect_x(so, inp, optval, optsize, p, 0); break; case SCTP_CONNECT_X_DELAYED: if (optsize < (sizeof(int) + sizeof(struct sockaddr_in))) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } error = sctp_do_connect_x(so, inp, optval, optsize, p, 1); break; case SCTP_CONNECT_X_COMPLETE: { struct sockaddr *sa; /* FIXME MT: check correct? */ SCTP_CHECK_AND_CAST(sa, optval, struct sockaddr, optsize); /* find tcb */ if (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) { SCTP_INP_RLOCK(inp); stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb) { SCTP_TCB_LOCK(stcb); } SCTP_INP_RUNLOCK(inp); } else { /* * We increment here since * sctp_findassociation_ep_addr() wil do a * decrement if it finds the stcb as long as * the locked tcb (last argument) is NOT a * TCB.. aka NULL. */ SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, sa, NULL, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } } if (stcb == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); error = ENOENT; break; } if (stcb->asoc.delayed_connection == 1) { stcb->asoc.delayed_connection = 0; (void)SCTP_GETTIME_TIMEVAL(&stcb->asoc.time_entered); sctp_timer_stop(SCTP_TIMER_TYPE_INIT, inp, stcb, stcb->asoc.primary_destination, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_8); sctp_send_initiate(inp, stcb, SCTP_SO_LOCKED); } else { /* * already expired or did not use delayed * connectx */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EALREADY); error = EALREADY; } SCTP_TCB_UNLOCK(stcb); break; } case SCTP_MAX_BURST: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { stcb->asoc.max_burst = av->assoc_value; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC) || (av->assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_WLOCK(inp); inp->sctp_ep.max_burst = av->assoc_value; SCTP_INP_WUNLOCK(inp); } if ((av->assoc_id == SCTP_CURRENT_ASSOC) || (av->assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); stcb->asoc.max_burst = av->assoc_value; SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } } break; } case SCTP_MAXSEG: { struct sctp_assoc_value *av; int ovh; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { ovh = SCTP_MED_OVERHEAD; } else { ovh = SCTP_MED_V4_OVERHEAD; } if (stcb) { if (av->assoc_value) { stcb->asoc.sctp_frag_point = (av->assoc_value + ovh); } else { stcb->asoc.sctp_frag_point = SCTP_DEFAULT_MAXSEGMENT; } SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_WLOCK(inp); /* * FIXME MT: I think this is not in * tune with the API ID */ if (av->assoc_value) { inp->sctp_frag_point = (av->assoc_value + ovh); } else { inp->sctp_frag_point = SCTP_DEFAULT_MAXSEGMENT; } SCTP_INP_WUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_EVENTS: { struct sctp_event_subscribe *events; SCTP_CHECK_AND_CAST(events, optval, struct sctp_event_subscribe, optsize); SCTP_INP_WLOCK(inp); if (events->sctp_data_io_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_RECVDATAIOEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_RECVDATAIOEVNT); } if (events->sctp_association_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_RECVASSOCEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_RECVASSOCEVNT); } if (events->sctp_address_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_RECVPADDREVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_RECVPADDREVNT); } if (events->sctp_send_failure_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_RECVSENDFAILEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_RECVSENDFAILEVNT); } if (events->sctp_peer_error_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_RECVPEERERR); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_RECVPEERERR); } if (events->sctp_shutdown_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_RECVSHUTDOWNEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_RECVSHUTDOWNEVNT); } if (events->sctp_partial_delivery_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_PDAPIEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_PDAPIEVNT); } if (events->sctp_adaptation_layer_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_ADAPTATIONEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_ADAPTATIONEVNT); } if (events->sctp_authentication_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_AUTHEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_AUTHEVNT); } if (events->sctp_sender_dry_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_DRYEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_DRYEVNT); } if (events->sctp_stream_reset_event) { sctp_feature_on(inp, SCTP_PCB_FLAGS_STREAM_RESETEVNT); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_STREAM_RESETEVNT); } SCTP_INP_WUNLOCK(inp); SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); if (events->sctp_association_event) { sctp_stcb_feature_on(inp, stcb, SCTP_PCB_FLAGS_RECVASSOCEVNT); } else { sctp_stcb_feature_off(inp, stcb, SCTP_PCB_FLAGS_RECVASSOCEVNT); } if (events->sctp_address_event) { sctp_stcb_feature_on(inp, stcb, SCTP_PCB_FLAGS_RECVPADDREVNT); } else { sctp_stcb_feature_off(inp, stcb, SCTP_PCB_FLAGS_RECVPADDREVNT); } if (events->sctp_send_failure_event) { sctp_stcb_feature_on(inp, stcb, SCTP_PCB_FLAGS_RECVSENDFAILEVNT); } else { sctp_stcb_feature_off(inp, stcb, SCTP_PCB_FLAGS_RECVSENDFAILEVNT); } if (events->sctp_peer_error_event) { sctp_stcb_feature_on(inp, stcb, SCTP_PCB_FLAGS_RECVPEERERR); } else { sctp_stcb_feature_off(inp, stcb, SCTP_PCB_FLAGS_RECVPEERERR); } if (events->sctp_shutdown_event) { sctp_stcb_feature_on(inp, stcb, SCTP_PCB_FLAGS_RECVSHUTDOWNEVNT); } else { sctp_stcb_feature_off(inp, stcb, SCTP_PCB_FLAGS_RECVSHUTDOWNEVNT); } if (events->sctp_partial_delivery_event) { sctp_stcb_feature_on(inp, stcb, SCTP_PCB_FLAGS_PDAPIEVNT); } else { sctp_stcb_feature_off(inp, stcb, SCTP_PCB_FLAGS_PDAPIEVNT); } if (events->sctp_adaptation_layer_event) { sctp_stcb_feature_on(inp, stcb, SCTP_PCB_FLAGS_ADAPTATIONEVNT); } else { sctp_stcb_feature_off(inp, stcb, SCTP_PCB_FLAGS_ADAPTATIONEVNT); } if (events->sctp_authentication_event) { sctp_stcb_feature_on(inp, stcb, SCTP_PCB_FLAGS_AUTHEVNT); } else { sctp_stcb_feature_off(inp, stcb, SCTP_PCB_FLAGS_AUTHEVNT); } if (events->sctp_sender_dry_event) { sctp_stcb_feature_on(inp, stcb, SCTP_PCB_FLAGS_DRYEVNT); } else { sctp_stcb_feature_off(inp, stcb, SCTP_PCB_FLAGS_DRYEVNT); } if (events->sctp_stream_reset_event) { sctp_stcb_feature_on(inp, stcb, SCTP_PCB_FLAGS_STREAM_RESETEVNT); } else { sctp_stcb_feature_off(inp, stcb, SCTP_PCB_FLAGS_STREAM_RESETEVNT); } SCTP_TCB_UNLOCK(stcb); } /* * Send up the sender dry event only for 1-to-1 * style sockets. */ if (events->sctp_sender_dry_event) { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) { stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb) { SCTP_TCB_LOCK(stcb); if (TAILQ_EMPTY(&stcb->asoc.send_queue) && TAILQ_EMPTY(&stcb->asoc.sent_queue) && (stcb->asoc.stream_queue_cnt == 0)) { sctp_ulp_notify(SCTP_NOTIFY_SENDER_DRY, stcb, 0, NULL, SCTP_SO_LOCKED); } SCTP_TCB_UNLOCK(stcb); } } } SCTP_INP_RUNLOCK(inp); break; } case SCTP_ADAPTATION_LAYER: { struct sctp_setadaptation *adap_bits; SCTP_CHECK_AND_CAST(adap_bits, optval, struct sctp_setadaptation, optsize); SCTP_INP_WLOCK(inp); inp->sctp_ep.adaptation_layer_indicator = adap_bits->ssb_adaptation_ind; inp->sctp_ep.adaptation_layer_indicator_provided = 1; SCTP_INP_WUNLOCK(inp); break; } #ifdef SCTP_DEBUG case SCTP_SET_INITIAL_DBG_SEQ: { uint32_t *vvv; SCTP_CHECK_AND_CAST(vvv, optval, uint32_t, optsize); SCTP_INP_WLOCK(inp); inp->sctp_ep.initial_sequence_debug = *vvv; SCTP_INP_WUNLOCK(inp); break; } #endif case SCTP_DEFAULT_SEND_PARAM: { struct sctp_sndrcvinfo *s_info; SCTP_CHECK_AND_CAST(s_info, optval, struct sctp_sndrcvinfo, optsize); SCTP_FIND_STCB(inp, stcb, s_info->sinfo_assoc_id); if (stcb) { if (s_info->sinfo_stream < stcb->asoc.streamoutcnt) { memcpy(&stcb->asoc.def_send, s_info, min(optsize, sizeof(stcb->asoc.def_send))); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (s_info->sinfo_assoc_id == SCTP_FUTURE_ASSOC) || (s_info->sinfo_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_WLOCK(inp); memcpy(&inp->def_send, s_info, min(optsize, sizeof(inp->def_send))); SCTP_INP_WUNLOCK(inp); } if ((s_info->sinfo_assoc_id == SCTP_CURRENT_ASSOC) || (s_info->sinfo_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); if (s_info->sinfo_stream < stcb->asoc.streamoutcnt) { memcpy(&stcb->asoc.def_send, s_info, min(optsize, sizeof(stcb->asoc.def_send))); } SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } } break; } case SCTP_PEER_ADDR_PARAMS: { struct sctp_paddrparams *paddrp; struct sctp_nets *net; struct sockaddr *addr; #if defined(INET) && defined(INET6) struct sockaddr_in sin_store; #endif SCTP_CHECK_AND_CAST(paddrp, optval, struct sctp_paddrparams, optsize); SCTP_FIND_STCB(inp, stcb, paddrp->spp_assoc_id); #if defined(INET) && defined(INET6) if (paddrp->spp_address.ss_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)&paddrp->spp_address; if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { in6_sin6_2_sin(&sin_store, sin6); addr = (struct sockaddr *)&sin_store; } else { addr = (struct sockaddr *)&paddrp->spp_address; } } else { addr = (struct sockaddr *)&paddrp->spp_address; } #else addr = (struct sockaddr *)&paddrp->spp_address; #endif if (stcb != NULL) { net = sctp_findnet(stcb, addr); } else { /* * We increment here since * sctp_findassociation_ep_addr() wil do a * decrement if it finds the stcb as long as * the locked tcb (last argument) is NOT a * TCB.. aka NULL. */ net = NULL; SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, addr, &net, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } } if ((stcb != NULL) && (net == NULL)) { #ifdef INET if (addr->sa_family == AF_INET) { struct sockaddr_in *sin; sin = (struct sockaddr_in *)addr; if (sin->sin_addr.s_addr != INADDR_ANY) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); SCTP_TCB_UNLOCK(stcb); error = EINVAL; break; } } else #endif #ifdef INET6 if (addr->sa_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)addr; if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); SCTP_TCB_UNLOCK(stcb); error = EINVAL; break; } } else #endif { error = EAFNOSUPPORT; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } /* sanity checks */ if ((paddrp->spp_flags & SPP_HB_ENABLE) && (paddrp->spp_flags & SPP_HB_DISABLE)) { if (stcb) SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } if ((paddrp->spp_flags & SPP_PMTUD_ENABLE) && (paddrp->spp_flags & SPP_PMTUD_DISABLE)) { if (stcb) SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } if (stcb != NULL) { /************************TCB SPECIFIC SET ******************/ if (net != NULL) { /************************NET SPECIFIC SET ******************/ if (paddrp->spp_flags & SPP_HB_DISABLE) { if (!(net->dest_state & SCTP_ADDR_UNCONFIRMED) && !(net->dest_state & SCTP_ADDR_NOHB)) { sctp_timer_stop(SCTP_TIMER_TYPE_HEARTBEAT, inp, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_9); } net->dest_state |= SCTP_ADDR_NOHB; } if (paddrp->spp_flags & SPP_HB_ENABLE) { if (paddrp->spp_hbinterval) { net->heart_beat_delay = paddrp->spp_hbinterval; } else if (paddrp->spp_flags & SPP_HB_TIME_IS_ZERO) { net->heart_beat_delay = 0; } sctp_timer_stop(SCTP_TIMER_TYPE_HEARTBEAT, inp, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_10); sctp_timer_start(SCTP_TIMER_TYPE_HEARTBEAT, inp, stcb, net); net->dest_state &= ~SCTP_ADDR_NOHB; } if (paddrp->spp_flags & SPP_HB_DEMAND) { /* on demand HB */ sctp_send_hb(stcb, net, SCTP_SO_LOCKED); sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_SOCKOPT, SCTP_SO_LOCKED); sctp_timer_start(SCTP_TIMER_TYPE_HEARTBEAT, inp, stcb, net); } if ((paddrp->spp_flags & SPP_PMTUD_DISABLE) && (paddrp->spp_pathmtu >= SCTP_SMALLEST_PMTU)) { if (SCTP_OS_TIMER_PENDING(&net->pmtu_timer.timer)) { sctp_timer_stop(SCTP_TIMER_TYPE_PATHMTURAISE, inp, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_11); } net->dest_state |= SCTP_ADDR_NO_PMTUD; net->mtu = paddrp->spp_pathmtu; switch (net->ro._l_addr.sa.sa_family) { #ifdef INET case AF_INET: net->mtu += SCTP_MIN_V4_OVERHEAD; break; #endif #ifdef INET6 case AF_INET6: net->mtu += SCTP_MIN_OVERHEAD; break; #endif default: break; } if (net->mtu < stcb->asoc.smallest_mtu) { sctp_pathmtu_adjustment(stcb, net->mtu); } } if (paddrp->spp_flags & SPP_PMTUD_ENABLE) { if (!SCTP_OS_TIMER_PENDING(&net->pmtu_timer.timer)) { sctp_timer_start(SCTP_TIMER_TYPE_PATHMTURAISE, inp, stcb, net); } net->dest_state &= ~SCTP_ADDR_NO_PMTUD; } if (paddrp->spp_pathmaxrxt) { if (net->dest_state & SCTP_ADDR_PF) { if (net->error_count > paddrp->spp_pathmaxrxt) { net->dest_state &= ~SCTP_ADDR_PF; } } else { if ((net->error_count <= paddrp->spp_pathmaxrxt) && (net->error_count > net->pf_threshold)) { net->dest_state |= SCTP_ADDR_PF; sctp_send_hb(stcb, net, SCTP_SO_LOCKED); sctp_timer_stop(SCTP_TIMER_TYPE_HEARTBEAT, stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_12); sctp_timer_start(SCTP_TIMER_TYPE_HEARTBEAT, stcb->sctp_ep, stcb, net); } } if (net->dest_state & SCTP_ADDR_REACHABLE) { if (net->error_count > paddrp->spp_pathmaxrxt) { net->dest_state &= ~SCTP_ADDR_REACHABLE; sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_DOWN, stcb, 0, net, SCTP_SO_LOCKED); } } else { if (net->error_count <= paddrp->spp_pathmaxrxt) { net->dest_state |= SCTP_ADDR_REACHABLE; sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_UP, stcb, 0, net, SCTP_SO_LOCKED); } } net->failure_threshold = paddrp->spp_pathmaxrxt; } if (paddrp->spp_flags & SPP_DSCP) { net->dscp = paddrp->spp_dscp & 0xfc; net->dscp |= 0x01; } #ifdef INET6 if (paddrp->spp_flags & SPP_IPV6_FLOWLABEL) { if (net->ro._l_addr.sa.sa_family == AF_INET6) { net->flowlabel = paddrp->spp_ipv6_flowlabel & 0x000fffff; net->flowlabel |= 0x80000000; } } #endif } else { /************************ASSOC ONLY -- NO NET SPECIFIC SET ******************/ if (paddrp->spp_pathmaxrxt != 0) { stcb->asoc.def_net_failure = paddrp->spp_pathmaxrxt; TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { if (net->dest_state & SCTP_ADDR_PF) { if (net->error_count > paddrp->spp_pathmaxrxt) { net->dest_state &= ~SCTP_ADDR_PF; } } else { if ((net->error_count <= paddrp->spp_pathmaxrxt) && (net->error_count > net->pf_threshold)) { net->dest_state |= SCTP_ADDR_PF; sctp_send_hb(stcb, net, SCTP_SO_LOCKED); sctp_timer_stop(SCTP_TIMER_TYPE_HEARTBEAT, stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_13); sctp_timer_start(SCTP_TIMER_TYPE_HEARTBEAT, stcb->sctp_ep, stcb, net); } } if (net->dest_state & SCTP_ADDR_REACHABLE) { if (net->error_count > paddrp->spp_pathmaxrxt) { net->dest_state &= ~SCTP_ADDR_REACHABLE; sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_DOWN, stcb, 0, net, SCTP_SO_LOCKED); } } else { if (net->error_count <= paddrp->spp_pathmaxrxt) { net->dest_state |= SCTP_ADDR_REACHABLE; sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_UP, stcb, 0, net, SCTP_SO_LOCKED); } } net->failure_threshold = paddrp->spp_pathmaxrxt; } } if (paddrp->spp_flags & SPP_HB_ENABLE) { if (paddrp->spp_hbinterval != 0) { stcb->asoc.heart_beat_delay = paddrp->spp_hbinterval; } else if (paddrp->spp_flags & SPP_HB_TIME_IS_ZERO) { stcb->asoc.heart_beat_delay = 0; } /* Turn back on the timer */ TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { if (paddrp->spp_hbinterval != 0) { net->heart_beat_delay = paddrp->spp_hbinterval; } else if (paddrp->spp_flags & SPP_HB_TIME_IS_ZERO) { net->heart_beat_delay = 0; } if (net->dest_state & SCTP_ADDR_NOHB) { net->dest_state &= ~SCTP_ADDR_NOHB; } sctp_timer_stop(SCTP_TIMER_TYPE_HEARTBEAT, inp, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_14); sctp_timer_start(SCTP_TIMER_TYPE_HEARTBEAT, inp, stcb, net); } sctp_stcb_feature_off(inp, stcb, SCTP_PCB_FLAGS_DONOT_HEARTBEAT); } if (paddrp->spp_flags & SPP_HB_DISABLE) { TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { if (!(net->dest_state & SCTP_ADDR_NOHB)) { net->dest_state |= SCTP_ADDR_NOHB; if (!(net->dest_state & SCTP_ADDR_UNCONFIRMED)) { sctp_timer_stop(SCTP_TIMER_TYPE_HEARTBEAT, inp, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_15); } } } sctp_stcb_feature_on(inp, stcb, SCTP_PCB_FLAGS_DONOT_HEARTBEAT); } if ((paddrp->spp_flags & SPP_PMTUD_DISABLE) && (paddrp->spp_pathmtu >= SCTP_SMALLEST_PMTU)) { TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { if (SCTP_OS_TIMER_PENDING(&net->pmtu_timer.timer)) { sctp_timer_stop(SCTP_TIMER_TYPE_PATHMTURAISE, inp, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_16); } net->dest_state |= SCTP_ADDR_NO_PMTUD; net->mtu = paddrp->spp_pathmtu; switch (net->ro._l_addr.sa.sa_family) { #ifdef INET case AF_INET: net->mtu += SCTP_MIN_V4_OVERHEAD; break; #endif #ifdef INET6 case AF_INET6: net->mtu += SCTP_MIN_OVERHEAD; break; #endif default: break; } if (net->mtu < stcb->asoc.smallest_mtu) { sctp_pathmtu_adjustment(stcb, net->mtu); } } sctp_stcb_feature_on(inp, stcb, SCTP_PCB_FLAGS_DO_NOT_PMTUD); } if (paddrp->spp_flags & SPP_PMTUD_ENABLE) { TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { if (!SCTP_OS_TIMER_PENDING(&net->pmtu_timer.timer)) { sctp_timer_start(SCTP_TIMER_TYPE_PATHMTURAISE, inp, stcb, net); } net->dest_state &= ~SCTP_ADDR_NO_PMTUD; } sctp_stcb_feature_off(inp, stcb, SCTP_PCB_FLAGS_DO_NOT_PMTUD); } if (paddrp->spp_flags & SPP_DSCP) { TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { net->dscp = paddrp->spp_dscp & 0xfc; net->dscp |= 0x01; } stcb->asoc.default_dscp = paddrp->spp_dscp & 0xfc; stcb->asoc.default_dscp |= 0x01; } #ifdef INET6 if (paddrp->spp_flags & SPP_IPV6_FLOWLABEL) { TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { if (net->ro._l_addr.sa.sa_family == AF_INET6) { net->flowlabel = paddrp->spp_ipv6_flowlabel & 0x000fffff; net->flowlabel |= 0x80000000; } } stcb->asoc.default_flowlabel = paddrp->spp_ipv6_flowlabel & 0x000fffff; stcb->asoc.default_flowlabel |= 0x80000000; } #endif } SCTP_TCB_UNLOCK(stcb); } else { /************************NO TCB, SET TO default stuff ******************/ if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (paddrp->spp_assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_WLOCK(inp); /* * For the TOS/FLOWLABEL stuff you * set it with the options on the * socket */ if (paddrp->spp_pathmaxrxt != 0) { inp->sctp_ep.def_net_failure = paddrp->spp_pathmaxrxt; } if (paddrp->spp_flags & SPP_HB_TIME_IS_ZERO) inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_HEARTBEAT] = 0; else if (paddrp->spp_hbinterval != 0) { if (paddrp->spp_hbinterval > SCTP_MAX_HB_INTERVAL) paddrp->spp_hbinterval = SCTP_MAX_HB_INTERVAL; inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_HEARTBEAT] = MSEC_TO_TICKS(paddrp->spp_hbinterval); } if (paddrp->spp_flags & SPP_HB_ENABLE) { if (paddrp->spp_flags & SPP_HB_TIME_IS_ZERO) { inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_HEARTBEAT] = 0; } else if (paddrp->spp_hbinterval) { inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_HEARTBEAT] = MSEC_TO_TICKS(paddrp->spp_hbinterval); } sctp_feature_off(inp, SCTP_PCB_FLAGS_DONOT_HEARTBEAT); } else if (paddrp->spp_flags & SPP_HB_DISABLE) { sctp_feature_on(inp, SCTP_PCB_FLAGS_DONOT_HEARTBEAT); } if (paddrp->spp_flags & SPP_PMTUD_ENABLE) { sctp_feature_off(inp, SCTP_PCB_FLAGS_DO_NOT_PMTUD); } else if (paddrp->spp_flags & SPP_PMTUD_DISABLE) { sctp_feature_on(inp, SCTP_PCB_FLAGS_DO_NOT_PMTUD); } if (paddrp->spp_flags & SPP_DSCP) { inp->sctp_ep.default_dscp = paddrp->spp_dscp & 0xfc; inp->sctp_ep.default_dscp |= 0x01; } #ifdef INET6 if (paddrp->spp_flags & SPP_IPV6_FLOWLABEL) { if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { inp->sctp_ep.default_flowlabel = paddrp->spp_ipv6_flowlabel & 0x000fffff; inp->sctp_ep.default_flowlabel |= 0x80000000; } } #endif SCTP_INP_WUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_RTOINFO: { struct sctp_rtoinfo *srto; uint32_t new_init, new_min, new_max; SCTP_CHECK_AND_CAST(srto, optval, struct sctp_rtoinfo, optsize); SCTP_FIND_STCB(inp, stcb, srto->srto_assoc_id); if (stcb) { if (srto->srto_initial) new_init = srto->srto_initial; else new_init = stcb->asoc.initial_rto; if (srto->srto_max) new_max = srto->srto_max; else new_max = stcb->asoc.maxrto; if (srto->srto_min) new_min = srto->srto_min; else new_min = stcb->asoc.minrto; if ((new_min <= new_init) && (new_init <= new_max)) { stcb->asoc.initial_rto = new_init; stcb->asoc.maxrto = new_max; stcb->asoc.minrto = new_min; } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (srto->srto_assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_WLOCK(inp); if (srto->srto_initial) new_init = srto->srto_initial; else new_init = inp->sctp_ep.initial_rto; if (srto->srto_max) new_max = srto->srto_max; else new_max = inp->sctp_ep.sctp_maxrto; if (srto->srto_min) new_min = srto->srto_min; else new_min = inp->sctp_ep.sctp_minrto; if ((new_min <= new_init) && (new_init <= new_max)) { inp->sctp_ep.initial_rto = new_init; inp->sctp_ep.sctp_maxrto = new_max; inp->sctp_ep.sctp_minrto = new_min; } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_INP_WUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_ASSOCINFO: { struct sctp_assocparams *sasoc; SCTP_CHECK_AND_CAST(sasoc, optval, struct sctp_assocparams, optsize); SCTP_FIND_STCB(inp, stcb, sasoc->sasoc_assoc_id); if (sasoc->sasoc_cookie_life) { /* boundary check the cookie life */ if (sasoc->sasoc_cookie_life < 1000) sasoc->sasoc_cookie_life = 1000; if (sasoc->sasoc_cookie_life > SCTP_MAX_COOKIE_LIFE) { sasoc->sasoc_cookie_life = SCTP_MAX_COOKIE_LIFE; } } if (stcb) { if (sasoc->sasoc_asocmaxrxt) stcb->asoc.max_send_times = sasoc->sasoc_asocmaxrxt; if (sasoc->sasoc_cookie_life) { stcb->asoc.cookie_life = MSEC_TO_TICKS(sasoc->sasoc_cookie_life); } SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (sasoc->sasoc_assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_WLOCK(inp); if (sasoc->sasoc_asocmaxrxt) inp->sctp_ep.max_send_times = sasoc->sasoc_asocmaxrxt; if (sasoc->sasoc_cookie_life) { inp->sctp_ep.def_cookie_life = MSEC_TO_TICKS(sasoc->sasoc_cookie_life); } SCTP_INP_WUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_INITMSG: { struct sctp_initmsg *sinit; SCTP_CHECK_AND_CAST(sinit, optval, struct sctp_initmsg, optsize); SCTP_INP_WLOCK(inp); if (sinit->sinit_num_ostreams) inp->sctp_ep.pre_open_stream_count = sinit->sinit_num_ostreams; if (sinit->sinit_max_instreams) inp->sctp_ep.max_open_streams_intome = sinit->sinit_max_instreams; if (sinit->sinit_max_attempts) inp->sctp_ep.max_init_times = sinit->sinit_max_attempts; if (sinit->sinit_max_init_timeo) inp->sctp_ep.initial_init_rto_max = sinit->sinit_max_init_timeo; SCTP_INP_WUNLOCK(inp); break; } case SCTP_PRIMARY_ADDR: { struct sctp_setprim *spa; struct sctp_nets *net; struct sockaddr *addr; #if defined(INET) && defined(INET6) struct sockaddr_in sin_store; #endif SCTP_CHECK_AND_CAST(spa, optval, struct sctp_setprim, optsize); SCTP_FIND_STCB(inp, stcb, spa->ssp_assoc_id); #if defined(INET) && defined(INET6) if (spa->ssp_addr.ss_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)&spa->ssp_addr; if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { in6_sin6_2_sin(&sin_store, sin6); addr = (struct sockaddr *)&sin_store; } else { addr = (struct sockaddr *)&spa->ssp_addr; } } else { addr = (struct sockaddr *)&spa->ssp_addr; } #else addr = (struct sockaddr *)&spa->ssp_addr; #endif if (stcb != NULL) { net = sctp_findnet(stcb, addr); } else { /* * We increment here since * sctp_findassociation_ep_addr() wil do a * decrement if it finds the stcb as long as * the locked tcb (last argument) is NOT a * TCB.. aka NULL. */ net = NULL; SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, addr, &net, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } } if ((stcb != NULL) && (net != NULL)) { if (net != stcb->asoc.primary_destination) { if (!(net->dest_state & SCTP_ADDR_UNCONFIRMED)) { /* Ok we need to set it */ if (sctp_set_primary_addr(stcb, (struct sockaddr *)NULL, net) == 0) { if ((stcb->asoc.alternate) && (!(net->dest_state & SCTP_ADDR_PF)) && (net->dest_state & SCTP_ADDR_REACHABLE)) { sctp_free_remote_addr(stcb->asoc.alternate); stcb->asoc.alternate = NULL; } } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } if (stcb != NULL) { SCTP_TCB_UNLOCK(stcb); } break; } case SCTP_SET_DYNAMIC_PRIMARY: { union sctp_sockstore *ss; error = priv_check(curthread, PRIV_NETINET_RESERVEDPORT); if (error) break; SCTP_CHECK_AND_CAST(ss, optval, union sctp_sockstore, optsize); /* SUPER USER CHECK? */ error = sctp_dynamic_set_primary(&ss->sa, vrf_id); break; } case SCTP_SET_PEER_PRIMARY_ADDR: { struct sctp_setpeerprim *sspp; struct sockaddr *addr; #if defined(INET) && defined(INET6) struct sockaddr_in sin_store; #endif SCTP_CHECK_AND_CAST(sspp, optval, struct sctp_setpeerprim, optsize); SCTP_FIND_STCB(inp, stcb, sspp->sspp_assoc_id); if (stcb != NULL) { struct sctp_ifa *ifa; #if defined(INET) && defined(INET6) if (sspp->sspp_addr.ss_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)&sspp->sspp_addr; if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { in6_sin6_2_sin(&sin_store, sin6); addr = (struct sockaddr *)&sin_store; } else { addr = (struct sockaddr *)&sspp->sspp_addr; } } else { addr = (struct sockaddr *)&sspp->sspp_addr; } #else addr = (struct sockaddr *)&sspp->sspp_addr; #endif ifa = sctp_find_ifa_by_addr(addr, stcb->asoc.vrf_id, SCTP_ADDR_NOT_LOCKED); if (ifa == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto out_of_it; } if ((inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) == 0) { /* * Must validate the ifa found is in * our ep */ struct sctp_laddr *laddr; int found = 0; LIST_FOREACH(laddr, &inp->sctp_addr_list, sctp_nxt_addr) { if (laddr->ifa == NULL) { SCTPDBG(SCTP_DEBUG_OUTPUT1, "%s: NULL ifa\n", __func__); continue; } if ((sctp_is_addr_restricted(stcb, laddr->ifa)) && (!sctp_is_addr_pending(stcb, laddr->ifa))) { continue; } if (laddr->ifa == ifa) { found = 1; break; } } if (!found) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto out_of_it; } } else { switch (addr->sa_family) { #ifdef INET case AF_INET: { struct sockaddr_in *sin; sin = (struct sockaddr_in *)addr; if (prison_check_ip4(inp->ip_inp.inp.inp_cred, &sin->sin_addr) != 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto out_of_it; } break; } #endif #ifdef INET6 case AF_INET6: { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)addr; if (prison_check_ip6(inp->ip_inp.inp.inp_cred, &sin6->sin6_addr) != 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto out_of_it; } break; } #endif default: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto out_of_it; } } if (sctp_set_primary_ip_address_sa(stcb, addr) != 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_SOCKOPT, SCTP_SO_LOCKED); out_of_it: SCTP_TCB_UNLOCK(stcb); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } break; } case SCTP_BINDX_ADD_ADDR: { struct sctp_getaddresses *addrs; struct thread *td; td = (struct thread *)p; SCTP_CHECK_AND_CAST(addrs, optval, struct sctp_getaddresses, optsize); #ifdef INET if (addrs->addr->sa_family == AF_INET) { if (optsize < sizeof(struct sctp_getaddresses) - sizeof(struct sockaddr) + sizeof(struct sockaddr_in)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } if (td != NULL && (error = prison_local_ip4(td->td_ucred, &(((struct sockaddr_in *)(addrs->addr))->sin_addr)))) { SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif #ifdef INET6 if (addrs->addr->sa_family == AF_INET6) { if (optsize < sizeof(struct sctp_getaddresses) - sizeof(struct sockaddr) + sizeof(struct sockaddr_in6)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } if (td != NULL && (error = prison_local_ip6(td->td_ucred, &(((struct sockaddr_in6 *)(addrs->addr))->sin6_addr), (SCTP_IPV6_V6ONLY(inp) != 0))) != 0) { SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif { error = EAFNOSUPPORT; break; } sctp_bindx_add_address(so, inp, addrs->addr, addrs->sget_assoc_id, vrf_id, &error, p); break; } case SCTP_BINDX_REM_ADDR: { struct sctp_getaddresses *addrs; struct thread *td; td = (struct thread *)p; SCTP_CHECK_AND_CAST(addrs, optval, struct sctp_getaddresses, optsize); #ifdef INET if (addrs->addr->sa_family == AF_INET) { if (optsize < sizeof(struct sctp_getaddresses) - sizeof(struct sockaddr) + sizeof(struct sockaddr_in)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } if (td != NULL && (error = prison_local_ip4(td->td_ucred, &(((struct sockaddr_in *)(addrs->addr))->sin_addr)))) { SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif #ifdef INET6 if (addrs->addr->sa_family == AF_INET6) { if (optsize < sizeof(struct sctp_getaddresses) - sizeof(struct sockaddr) + sizeof(struct sockaddr_in6)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } if (td != NULL && (error = prison_local_ip6(td->td_ucred, &(((struct sockaddr_in6 *)(addrs->addr))->sin6_addr), (SCTP_IPV6_V6ONLY(inp) != 0))) != 0) { SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } else #endif { error = EAFNOSUPPORT; break; } sctp_bindx_delete_address(inp, addrs->addr, addrs->sget_assoc_id, vrf_id, &error); break; } case SCTP_EVENT: { struct sctp_event *event; uint32_t event_type; SCTP_CHECK_AND_CAST(event, optval, struct sctp_event, optsize); SCTP_FIND_STCB(inp, stcb, event->se_assoc_id); switch (event->se_type) { case SCTP_ASSOC_CHANGE: event_type = SCTP_PCB_FLAGS_RECVASSOCEVNT; break; case SCTP_PEER_ADDR_CHANGE: event_type = SCTP_PCB_FLAGS_RECVPADDREVNT; break; case SCTP_REMOTE_ERROR: event_type = SCTP_PCB_FLAGS_RECVPEERERR; break; case SCTP_SEND_FAILED: event_type = SCTP_PCB_FLAGS_RECVSENDFAILEVNT; break; case SCTP_SHUTDOWN_EVENT: event_type = SCTP_PCB_FLAGS_RECVSHUTDOWNEVNT; break; case SCTP_ADAPTATION_INDICATION: event_type = SCTP_PCB_FLAGS_ADAPTATIONEVNT; break; case SCTP_PARTIAL_DELIVERY_EVENT: event_type = SCTP_PCB_FLAGS_PDAPIEVNT; break; case SCTP_AUTHENTICATION_EVENT: event_type = SCTP_PCB_FLAGS_AUTHEVNT; break; case SCTP_STREAM_RESET_EVENT: event_type = SCTP_PCB_FLAGS_STREAM_RESETEVNT; break; case SCTP_SENDER_DRY_EVENT: event_type = SCTP_PCB_FLAGS_DRYEVNT; break; case SCTP_NOTIFICATIONS_STOPPED_EVENT: event_type = 0; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOTSUP); error = ENOTSUP; break; case SCTP_ASSOC_RESET_EVENT: event_type = SCTP_PCB_FLAGS_ASSOC_RESETEVNT; break; case SCTP_STREAM_CHANGE_EVENT: event_type = SCTP_PCB_FLAGS_STREAM_CHANGEEVNT; break; case SCTP_SEND_FAILED_EVENT: event_type = SCTP_PCB_FLAGS_RECVNSENDFAILEVNT; break; default: event_type = 0; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } if (event_type > 0) { if (stcb) { if (event->se_on) { sctp_stcb_feature_on(inp, stcb, event_type); if (event_type == SCTP_PCB_FLAGS_DRYEVNT) { if (TAILQ_EMPTY(&stcb->asoc.send_queue) && TAILQ_EMPTY(&stcb->asoc.sent_queue) && (stcb->asoc.stream_queue_cnt == 0)) { sctp_ulp_notify(SCTP_NOTIFY_SENDER_DRY, stcb, 0, NULL, SCTP_SO_LOCKED); } } } else { sctp_stcb_feature_off(inp, stcb, event_type); } SCTP_TCB_UNLOCK(stcb); } else { /* * We don't want to send up a storm * of events, so return an error for * sender dry events */ if ((event_type == SCTP_PCB_FLAGS_DRYEVNT) && ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) == 0) && ((inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) == 0) && ((event->se_assoc_id == SCTP_ALL_ASSOC) || (event->se_assoc_id == SCTP_CURRENT_ASSOC))) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOTSUP); error = ENOTSUP; break; } if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (event->se_assoc_id == SCTP_FUTURE_ASSOC) || (event->se_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_WLOCK(inp); if (event->se_on) { sctp_feature_on(inp, event_type); } else { sctp_feature_off(inp, event_type); } SCTP_INP_WUNLOCK(inp); } if ((event->se_assoc_id == SCTP_CURRENT_ASSOC) || (event->se_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); if (event->se_on) { sctp_stcb_feature_on(inp, stcb, event_type); } else { sctp_stcb_feature_off(inp, stcb, event_type); } SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } } } break; } case SCTP_RECVRCVINFO: { int *onoff; SCTP_CHECK_AND_CAST(onoff, optval, int, optsize); SCTP_INP_WLOCK(inp); if (*onoff != 0) { sctp_feature_on(inp, SCTP_PCB_FLAGS_RECVRCVINFO); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_RECVRCVINFO); } SCTP_INP_WUNLOCK(inp); break; } case SCTP_RECVNXTINFO: { int *onoff; SCTP_CHECK_AND_CAST(onoff, optval, int, optsize); SCTP_INP_WLOCK(inp); if (*onoff != 0) { sctp_feature_on(inp, SCTP_PCB_FLAGS_RECVNXTINFO); } else { sctp_feature_off(inp, SCTP_PCB_FLAGS_RECVNXTINFO); } SCTP_INP_WUNLOCK(inp); break; } case SCTP_DEFAULT_SNDINFO: { struct sctp_sndinfo *info; uint16_t policy; SCTP_CHECK_AND_CAST(info, optval, struct sctp_sndinfo, optsize); SCTP_FIND_STCB(inp, stcb, info->snd_assoc_id); if (stcb) { if (info->snd_sid < stcb->asoc.streamoutcnt) { stcb->asoc.def_send.sinfo_stream = info->snd_sid; policy = PR_SCTP_POLICY(stcb->asoc.def_send.sinfo_flags); stcb->asoc.def_send.sinfo_flags = info->snd_flags; stcb->asoc.def_send.sinfo_flags |= policy; stcb->asoc.def_send.sinfo_ppid = info->snd_ppid; stcb->asoc.def_send.sinfo_context = info->snd_context; } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (info->snd_assoc_id == SCTP_FUTURE_ASSOC) || (info->snd_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_WLOCK(inp); inp->def_send.sinfo_stream = info->snd_sid; policy = PR_SCTP_POLICY(inp->def_send.sinfo_flags); inp->def_send.sinfo_flags = info->snd_flags; inp->def_send.sinfo_flags |= policy; inp->def_send.sinfo_ppid = info->snd_ppid; inp->def_send.sinfo_context = info->snd_context; SCTP_INP_WUNLOCK(inp); } if ((info->snd_assoc_id == SCTP_CURRENT_ASSOC) || (info->snd_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); if (info->snd_sid < stcb->asoc.streamoutcnt) { stcb->asoc.def_send.sinfo_stream = info->snd_sid; policy = PR_SCTP_POLICY(stcb->asoc.def_send.sinfo_flags); stcb->asoc.def_send.sinfo_flags = info->snd_flags; stcb->asoc.def_send.sinfo_flags |= policy; stcb->asoc.def_send.sinfo_ppid = info->snd_ppid; stcb->asoc.def_send.sinfo_context = info->snd_context; } SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } } break; } case SCTP_DEFAULT_PRINFO: { struct sctp_default_prinfo *info; SCTP_CHECK_AND_CAST(info, optval, struct sctp_default_prinfo, optsize); SCTP_FIND_STCB(inp, stcb, info->pr_assoc_id); if (info->pr_policy > SCTP_PR_SCTP_MAX) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; } if (stcb) { stcb->asoc.def_send.sinfo_flags &= 0xfff0; stcb->asoc.def_send.sinfo_flags |= info->pr_policy; stcb->asoc.def_send.sinfo_timetolive = info->pr_value; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (info->pr_assoc_id == SCTP_FUTURE_ASSOC) || (info->pr_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_WLOCK(inp); inp->def_send.sinfo_flags &= 0xfff0; inp->def_send.sinfo_flags |= info->pr_policy; inp->def_send.sinfo_timetolive = info->pr_value; SCTP_INP_WUNLOCK(inp); } if ((info->pr_assoc_id == SCTP_CURRENT_ASSOC) || (info->pr_assoc_id == SCTP_ALL_ASSOC)) { SCTP_INP_RLOCK(inp); LIST_FOREACH(stcb, &inp->sctp_asoc_list, sctp_tcblist) { SCTP_TCB_LOCK(stcb); stcb->asoc.def_send.sinfo_flags &= 0xfff0; stcb->asoc.def_send.sinfo_flags |= info->pr_policy; stcb->asoc.def_send.sinfo_timetolive = info->pr_value; SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } } break; } case SCTP_PEER_ADDR_THLDS: /* Applies to the specific association */ { struct sctp_paddrthlds *thlds; struct sctp_nets *net; struct sockaddr *addr; #if defined(INET) && defined(INET6) struct sockaddr_in sin_store; #endif SCTP_CHECK_AND_CAST(thlds, optval, struct sctp_paddrthlds, optsize); SCTP_FIND_STCB(inp, stcb, thlds->spt_assoc_id); #if defined(INET) && defined(INET6) if (thlds->spt_address.ss_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)&thlds->spt_address; if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { in6_sin6_2_sin(&sin_store, sin6); addr = (struct sockaddr *)&sin_store; } else { addr = (struct sockaddr *)&thlds->spt_address; } } else { addr = (struct sockaddr *)&thlds->spt_address; } #else addr = (struct sockaddr *)&thlds->spt_address; #endif if (stcb != NULL) { net = sctp_findnet(stcb, addr); } else { /* * We increment here since * sctp_findassociation_ep_addr() wil do a * decrement if it finds the stcb as long as * the locked tcb (last argument) is NOT a * TCB.. aka NULL. */ net = NULL; SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, addr, &net, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } } if ((stcb != NULL) && (net == NULL)) { #ifdef INET if (addr->sa_family == AF_INET) { struct sockaddr_in *sin; sin = (struct sockaddr_in *)addr; if (sin->sin_addr.s_addr != INADDR_ANY) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); SCTP_TCB_UNLOCK(stcb); error = EINVAL; break; } } else #endif #ifdef INET6 if (addr->sa_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)addr; if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); SCTP_TCB_UNLOCK(stcb); error = EINVAL; break; } } else #endif { error = EAFNOSUPPORT; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } if (thlds->spt_pathcpthld != 0xffff) { error = EINVAL; SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } if (stcb != NULL) { if (net != NULL) { net->failure_threshold = thlds->spt_pathmaxrxt; net->pf_threshold = thlds->spt_pathpfthld; if (net->dest_state & SCTP_ADDR_PF) { if ((net->error_count > net->failure_threshold) || (net->error_count <= net->pf_threshold)) { net->dest_state &= ~SCTP_ADDR_PF; } } else { if ((net->error_count > net->pf_threshold) && (net->error_count <= net->failure_threshold)) { net->dest_state |= SCTP_ADDR_PF; sctp_send_hb(stcb, net, SCTP_SO_LOCKED); sctp_timer_stop(SCTP_TIMER_TYPE_HEARTBEAT, stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_17); sctp_timer_start(SCTP_TIMER_TYPE_HEARTBEAT, stcb->sctp_ep, stcb, net); } } if (net->dest_state & SCTP_ADDR_REACHABLE) { if (net->error_count > net->failure_threshold) { net->dest_state &= ~SCTP_ADDR_REACHABLE; sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_DOWN, stcb, 0, net, SCTP_SO_LOCKED); } } else { if (net->error_count <= net->failure_threshold) { net->dest_state |= SCTP_ADDR_REACHABLE; sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_UP, stcb, 0, net, SCTP_SO_LOCKED); } } } else { TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { net->failure_threshold = thlds->spt_pathmaxrxt; net->pf_threshold = thlds->spt_pathpfthld; if (net->dest_state & SCTP_ADDR_PF) { if ((net->error_count > net->failure_threshold) || (net->error_count <= net->pf_threshold)) { net->dest_state &= ~SCTP_ADDR_PF; } } else { if ((net->error_count > net->pf_threshold) && (net->error_count <= net->failure_threshold)) { net->dest_state |= SCTP_ADDR_PF; sctp_send_hb(stcb, net, SCTP_SO_LOCKED); sctp_timer_stop(SCTP_TIMER_TYPE_HEARTBEAT, stcb->sctp_ep, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_18); sctp_timer_start(SCTP_TIMER_TYPE_HEARTBEAT, stcb->sctp_ep, stcb, net); } } if (net->dest_state & SCTP_ADDR_REACHABLE) { if (net->error_count > net->failure_threshold) { net->dest_state &= ~SCTP_ADDR_REACHABLE; sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_DOWN, stcb, 0, net, SCTP_SO_LOCKED); } } else { if (net->error_count <= net->failure_threshold) { net->dest_state |= SCTP_ADDR_REACHABLE; sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_UP, stcb, 0, net, SCTP_SO_LOCKED); } } } stcb->asoc.def_net_failure = thlds->spt_pathmaxrxt; stcb->asoc.def_net_pf_threshold = thlds->spt_pathpfthld; } SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (thlds->spt_assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_WLOCK(inp); inp->sctp_ep.def_net_failure = thlds->spt_pathmaxrxt; inp->sctp_ep.def_net_pf_threshold = thlds->spt_pathpfthld; SCTP_INP_WUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_REMOTE_UDP_ENCAPS_PORT: { struct sctp_udpencaps *encaps; struct sctp_nets *net; struct sockaddr *addr; #if defined(INET) && defined(INET6) struct sockaddr_in sin_store; #endif SCTP_CHECK_AND_CAST(encaps, optval, struct sctp_udpencaps, optsize); SCTP_FIND_STCB(inp, stcb, encaps->sue_assoc_id); #if defined(INET) && defined(INET6) if (encaps->sue_address.ss_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)&encaps->sue_address; if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { in6_sin6_2_sin(&sin_store, sin6); addr = (struct sockaddr *)&sin_store; } else { addr = (struct sockaddr *)&encaps->sue_address; } } else { addr = (struct sockaddr *)&encaps->sue_address; } #else addr = (struct sockaddr *)&encaps->sue_address; #endif if (stcb != NULL) { net = sctp_findnet(stcb, addr); } else { /* * We increment here since * sctp_findassociation_ep_addr() wil do a * decrement if it finds the stcb as long as * the locked tcb (last argument) is NOT a * TCB.. aka NULL. */ net = NULL; SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, addr, &net, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } } if ((stcb != NULL) && (net == NULL)) { #ifdef INET if (addr->sa_family == AF_INET) { struct sockaddr_in *sin; sin = (struct sockaddr_in *)addr; if (sin->sin_addr.s_addr != INADDR_ANY) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); SCTP_TCB_UNLOCK(stcb); error = EINVAL; break; } } else #endif #ifdef INET6 if (addr->sa_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)addr; if (!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); SCTP_TCB_UNLOCK(stcb); error = EINVAL; break; } } else #endif { error = EAFNOSUPPORT; SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); break; } } if (stcb != NULL) { if (net != NULL) { net->port = encaps->sue_port; } else { stcb->asoc.port = encaps->sue_port; } SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (encaps->sue_assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_WLOCK(inp); inp->sctp_ep.port = encaps->sue_port; SCTP_INP_WUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_ECN_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_WLOCK(inp); if (av->assoc_value == 0) { inp->ecn_supported = 0; } else { inp->ecn_supported = 1; } SCTP_INP_WUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_PR_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_WLOCK(inp); if (av->assoc_value == 0) { inp->prsctp_supported = 0; } else { inp->prsctp_supported = 1; } SCTP_INP_WUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_AUTH_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { if ((av->assoc_value == 0) && (inp->asconf_supported == 1)) { /* * AUTH is required for * ASCONF */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } else { SCTP_INP_WLOCK(inp); if (av->assoc_value == 0) { inp->auth_supported = 0; } else { inp->auth_supported = 1; } SCTP_INP_WUNLOCK(inp); } } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_ASCONF_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { if ((av->assoc_value != 0) && (inp->auth_supported == 0)) { /* * AUTH is required for * ASCONF */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } else { SCTP_INP_WLOCK(inp); if (av->assoc_value == 0) { inp->asconf_supported = 0; sctp_auth_delete_chunk(SCTP_ASCONF, inp->sctp_ep.local_auth_chunks); sctp_auth_delete_chunk(SCTP_ASCONF_ACK, inp->sctp_ep.local_auth_chunks); } else { inp->asconf_supported = 1; sctp_auth_add_chunk(SCTP_ASCONF, inp->sctp_ep.local_auth_chunks); sctp_auth_add_chunk(SCTP_ASCONF_ACK, inp->sctp_ep.local_auth_chunks); } SCTP_INP_WUNLOCK(inp); } } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_RECONFIG_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_WLOCK(inp); if (av->assoc_value == 0) { inp->reconfig_supported = 0; } else { inp->reconfig_supported = 1; } SCTP_INP_WUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_NRSACK_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_WLOCK(inp); if (av->assoc_value == 0) { inp->nrsack_supported = 0; } else { inp->nrsack_supported = 1; } SCTP_INP_WUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_PKTDROP_SUPPORTED: { struct sctp_assoc_value *av; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_WLOCK(inp); if (av->assoc_value == 0) { inp->pktdrop_supported = 0; } else { inp->pktdrop_supported = 1; } SCTP_INP_WUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } case SCTP_MAX_CWND: { struct sctp_assoc_value *av; struct sctp_nets *net; SCTP_CHECK_AND_CAST(av, optval, struct sctp_assoc_value, optsize); SCTP_FIND_STCB(inp, stcb, av->assoc_id); if (stcb) { stcb->asoc.max_cwnd = av->assoc_value; if (stcb->asoc.max_cwnd > 0) { TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { if ((net->cwnd > stcb->asoc.max_cwnd) && (net->cwnd > (net->mtu - sizeof(struct sctphdr)))) { net->cwnd = stcb->asoc.max_cwnd; if (net->cwnd < (net->mtu - sizeof(struct sctphdr))) { net->cwnd = net->mtu - sizeof(struct sctphdr); } } } } SCTP_TCB_UNLOCK(stcb); } else { if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || (av->assoc_id == SCTP_FUTURE_ASSOC)) { SCTP_INP_WLOCK(inp); inp->max_cwnd = av->assoc_value; SCTP_INP_WUNLOCK(inp); } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } } break; } default: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOPROTOOPT); error = ENOPROTOOPT; break; } /* end switch (opt) */ return (error); } int sctp_ctloutput(struct socket *so, struct sockopt *sopt) { void *optval = NULL; size_t optsize = 0; void *p; int error = 0; struct sctp_inpcb *inp; if ((sopt->sopt_level == SOL_SOCKET) && (sopt->sopt_name == SO_SETFIB)) { inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { SCTP_LTRACE_ERR_RET(so->so_pcb, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOBUFS); return (EINVAL); } SCTP_INP_WLOCK(inp); inp->fibnum = so->so_fibnum; SCTP_INP_WUNLOCK(inp); return (0); } if (sopt->sopt_level != IPPROTO_SCTP) { /* wrong proto level... send back up to IP */ #ifdef INET6 if (INP_CHECK_SOCKAF(so, AF_INET6)) error = ip6_ctloutput(so, sopt); #endif /* INET6 */ #if defined(INET) && defined(INET6) else #endif #ifdef INET error = ip_ctloutput(so, sopt); #endif return (error); } optsize = sopt->sopt_valsize; if (optsize) { SCTP_MALLOC(optval, void *, optsize, SCTP_M_SOCKOPT); if (optval == NULL) { SCTP_LTRACE_ERR_RET(so->so_pcb, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOBUFS); return (ENOBUFS); } error = sooptcopyin(sopt, optval, optsize, optsize); if (error) { SCTP_FREE(optval, SCTP_M_SOCKOPT); goto out; } } p = (void *)sopt->sopt_td; if (sopt->sopt_dir == SOPT_SET) { error = sctp_setopt(so, sopt->sopt_name, optval, optsize, p); } else if (sopt->sopt_dir == SOPT_GET) { error = sctp_getopt(so, sopt->sopt_name, optval, &optsize, p); } else { SCTP_LTRACE_ERR_RET(so->so_pcb, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; } if ((error == 0) && (optval != NULL)) { error = sooptcopyout(sopt, optval, optsize); SCTP_FREE(optval, SCTP_M_SOCKOPT); } else if (optval != NULL) { SCTP_FREE(optval, SCTP_M_SOCKOPT); } out: return (error); } #ifdef INET static int sctp_connect(struct socket *so, struct sockaddr *addr, struct thread *p) { int error = 0; int create_lock_on = 0; uint32_t vrf_id; struct sctp_inpcb *inp; struct sctp_tcb *stcb = NULL; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { /* I made the same as TCP since we are not setup? */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (ECONNRESET); } if (addr == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return EINVAL; } switch (addr->sa_family) { #ifdef INET6 case AF_INET6: { struct sockaddr_in6 *sin6p; if (addr->sa_len != sizeof(struct sockaddr_in6)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } sin6p = (struct sockaddr_in6 *)addr; if (p != NULL && (error = prison_remote_ip6(p->td_ucred, &sin6p->sin6_addr)) != 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); return (error); } break; } #endif #ifdef INET case AF_INET: { struct sockaddr_in *sinp; if (addr->sa_len != sizeof(struct sockaddr_in)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (EINVAL); } sinp = (struct sockaddr_in *)addr; if (p != NULL && (error = prison_remote_ip4(p->td_ucred, &sinp->sin_addr)) != 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, error); return (error); } break; } #endif default: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EAFNOSUPPORT); return (EAFNOSUPPORT); } SCTP_INP_INCR_REF(inp); SCTP_ASOC_CREATE_LOCK(inp); create_lock_on = 1; if ((inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE) || (inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE)) { /* Should I really unlock ? */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EFAULT); error = EFAULT; goto out_now; } #ifdef INET6 if (((inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) == 0) && (addr->sa_family == AF_INET6)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto out_now; } #endif if ((inp->sctp_flags & SCTP_PCB_FLAGS_UNBOUND) == SCTP_PCB_FLAGS_UNBOUND) { /* Bind a ephemeral port */ error = sctp_inpcb_bind(so, NULL, NULL, p); if (error) { goto out_now; } } /* Now do we connect? */ if ((inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) && (sctp_is_feature_off(inp, SCTP_PCB_FLAGS_PORTREUSE))) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; goto out_now; } if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) && (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED)) { /* We are already connected AND the TCP model */ SCTP_LTRACE_ERR_RET(inp, stcb, NULL, SCTP_FROM_SCTP_USRREQ, EADDRINUSE); error = EADDRINUSE; goto out_now; } if (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) { SCTP_INP_RLOCK(inp); stcb = LIST_FIRST(&inp->sctp_asoc_list); SCTP_INP_RUNLOCK(inp); } else { /* * We increment here since sctp_findassociation_ep_addr() * will do a decrement if it finds the stcb as long as the * locked tcb (last argument) is NOT a TCB.. aka NULL. */ SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, addr, NULL, NULL, NULL); if (stcb == NULL) { SCTP_INP_DECR_REF(inp); } else { SCTP_TCB_UNLOCK(stcb); } } if (stcb != NULL) { /* Already have or am bring up an association */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EALREADY); error = EALREADY; goto out_now; } vrf_id = inp->def_vrf_id; /* We are GOOD to go */ stcb = sctp_aloc_assoc(inp, addr, &error, 0, vrf_id, inp->sctp_ep.pre_open_stream_count, inp->sctp_ep.port, p); if (stcb == NULL) { /* Gak! no memory */ goto out_now; } if (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) { stcb->sctp_ep->sctp_flags |= SCTP_PCB_FLAGS_CONNECTED; /* Set the connected flag so we can queue data */ soisconnecting(so); } SCTP_SET_STATE(&stcb->asoc, SCTP_STATE_COOKIE_WAIT); (void)SCTP_GETTIME_TIMEVAL(&stcb->asoc.time_entered); /* initialize authentication parameters for the assoc */ sctp_initialize_auth_params(inp, stcb); sctp_send_initiate(inp, stcb, SCTP_SO_LOCKED); SCTP_TCB_UNLOCK(stcb); out_now: if (create_lock_on) { SCTP_ASOC_CREATE_UNLOCK(inp); } SCTP_INP_DECR_REF(inp); return (error); } #endif int sctp_listen(struct socket *so, int backlog, struct thread *p) { /* * Note this module depends on the protocol processing being called * AFTER any socket level flags and backlog are applied to the * socket. The traditional way that the socket flags are applied is * AFTER protocol processing. We have made a change to the * sys/kern/uipc_socket.c module to reverse this but this MUST be in * place if the socket API for SCTP is to work properly. */ int error = 0; struct sctp_inpcb *inp; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { /* I made the same as TCP since we are not setup? */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (ECONNRESET); } if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_PORTREUSE)) { /* See if we have a listener */ struct sctp_inpcb *tinp; union sctp_sockstore store; if ((inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) == 0) { /* not bound all */ struct sctp_laddr *laddr; LIST_FOREACH(laddr, &inp->sctp_addr_list, sctp_nxt_addr) { memcpy(&store, &laddr->ifa->address, sizeof(store)); switch (store.sa.sa_family) { #ifdef INET case AF_INET: store.sin.sin_port = inp->sctp_lport; break; #endif #ifdef INET6 case AF_INET6: store.sin6.sin6_port = inp->sctp_lport; break; #endif default: break; } tinp = sctp_pcb_findep(&store.sa, 0, 0, inp->def_vrf_id); if (tinp && (tinp != inp) && ((tinp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE) == 0) && ((tinp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) && (tinp->sctp_socket->so_qlimit)) { /* * we have a listener already and * its not this inp. */ SCTP_INP_DECR_REF(tinp); return (EADDRINUSE); } else if (tinp) { SCTP_INP_DECR_REF(tinp); } } } else { /* Setup a local addr bound all */ memset(&store, 0, sizeof(store)); #ifdef INET6 if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { store.sa.sa_family = AF_INET6; store.sa.sa_len = sizeof(struct sockaddr_in6); } #endif #ifdef INET if ((inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) == 0) { store.sa.sa_family = AF_INET; store.sa.sa_len = sizeof(struct sockaddr_in); } #endif switch (store.sa.sa_family) { #ifdef INET case AF_INET: store.sin.sin_port = inp->sctp_lport; break; #endif #ifdef INET6 case AF_INET6: store.sin6.sin6_port = inp->sctp_lport; break; #endif default: break; } tinp = sctp_pcb_findep(&store.sa, 0, 0, inp->def_vrf_id); if (tinp && (tinp != inp) && ((tinp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE) == 0) && ((tinp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) && (tinp->sctp_socket->so_qlimit)) { /* * we have a listener already and its not * this inp. */ SCTP_INP_DECR_REF(tinp); return (EADDRINUSE); } else if (tinp) { SCTP_INP_DECR_REF(tinp); } } } SCTP_INP_RLOCK(inp); #ifdef SCTP_LOCK_LOGGING if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_LOCK_LOGGING_ENABLE) { sctp_log_lock(inp, (struct sctp_tcb *)NULL, SCTP_LOG_LOCK_SOCK); } #endif SOCK_LOCK(so); error = solisten_proto_check(so); SOCK_UNLOCK(so); if (error) { SCTP_INP_RUNLOCK(inp); return (error); } if ((sctp_is_feature_on(inp, SCTP_PCB_FLAGS_PORTREUSE)) && (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) { /* * The unlucky case - We are in the tcp pool with this guy. * - Someone else is in the main inp slot. - We must move * this guy (the listener) to the main slot - We must then * move the guy that was listener to the TCP Pool. */ if (sctp_swap_inpcb_for_listen(inp)) { SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EADDRINUSE); return (EADDRINUSE); } } if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) && (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED)) { /* We are already connected AND the TCP model */ SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EADDRINUSE); return (EADDRINUSE); } SCTP_INP_RUNLOCK(inp); if (inp->sctp_flags & SCTP_PCB_FLAGS_UNBOUND) { /* We must do a bind. */ if ((error = sctp_inpcb_bind(so, NULL, NULL, p))) { /* bind error, probably perm */ return (error); } } SOCK_LOCK(so); /* It appears for 7.0 and on, we must always call this. */ solisten_proto(so, backlog); if (inp->sctp_flags & SCTP_PCB_FLAGS_UDPTYPE) { /* remove the ACCEPTCONN flag for one-to-many sockets */ so->so_options &= ~SO_ACCEPTCONN; } if (backlog == 0) { /* turning off listen */ so->so_options &= ~SO_ACCEPTCONN; } SOCK_UNLOCK(so); return (error); } static int sctp_defered_wakeup_cnt = 0; int sctp_accept(struct socket *so, struct sockaddr **addr) { struct sctp_tcb *stcb; struct sctp_inpcb *inp; union sctp_sockstore store; #ifdef INET6 int error; #endif inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (ECONNRESET); } SCTP_INP_RLOCK(inp); if (inp->sctp_flags & SCTP_PCB_FLAGS_UDPTYPE) { SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EOPNOTSUPP); return (EOPNOTSUPP); } if (so->so_state & SS_ISDISCONNECTED) { SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ECONNABORTED); return (ECONNABORTED); } stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb == NULL) { SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (ECONNRESET); } SCTP_TCB_LOCK(stcb); SCTP_INP_RUNLOCK(inp); store = stcb->asoc.primary_destination->ro._l_addr; stcb->asoc.state &= ~SCTP_STATE_IN_ACCEPT_QUEUE; SCTP_TCB_UNLOCK(stcb); switch (store.sa.sa_family) { #ifdef INET case AF_INET: { struct sockaddr_in *sin; SCTP_MALLOC_SONAME(sin, struct sockaddr_in *, sizeof *sin); if (sin == NULL) return (ENOMEM); sin->sin_family = AF_INET; sin->sin_len = sizeof(*sin); sin->sin_port = store.sin.sin_port; sin->sin_addr = store.sin.sin_addr; *addr = (struct sockaddr *)sin; break; } #endif #ifdef INET6 case AF_INET6: { struct sockaddr_in6 *sin6; SCTP_MALLOC_SONAME(sin6, struct sockaddr_in6 *, sizeof *sin6); if (sin6 == NULL) return (ENOMEM); sin6->sin6_family = AF_INET6; sin6->sin6_len = sizeof(*sin6); sin6->sin6_port = store.sin6.sin6_port; sin6->sin6_addr = store.sin6.sin6_addr; if ((error = sa6_recoverscope(sin6)) != 0) { SCTP_FREE_SONAME(sin6); return (error); } *addr = (struct sockaddr *)sin6; break; } #endif default: /* TSNH */ break; } /* Wake any delayed sleep action */ if (inp->sctp_flags & SCTP_PCB_FLAGS_DONT_WAKE) { SCTP_INP_WLOCK(inp); inp->sctp_flags &= ~SCTP_PCB_FLAGS_DONT_WAKE; if (inp->sctp_flags & SCTP_PCB_FLAGS_WAKEOUTPUT) { inp->sctp_flags &= ~SCTP_PCB_FLAGS_WAKEOUTPUT; SCTP_INP_WUNLOCK(inp); SOCKBUF_LOCK(&inp->sctp_socket->so_snd); if (sowriteable(inp->sctp_socket)) { sowwakeup_locked(inp->sctp_socket); } else { SOCKBUF_UNLOCK(&inp->sctp_socket->so_snd); } SCTP_INP_WLOCK(inp); } if (inp->sctp_flags & SCTP_PCB_FLAGS_WAKEINPUT) { inp->sctp_flags &= ~SCTP_PCB_FLAGS_WAKEINPUT; SCTP_INP_WUNLOCK(inp); SOCKBUF_LOCK(&inp->sctp_socket->so_rcv); if (soreadable(inp->sctp_socket)) { sctp_defered_wakeup_cnt++; sorwakeup_locked(inp->sctp_socket); } else { SOCKBUF_UNLOCK(&inp->sctp_socket->so_rcv); } SCTP_INP_WLOCK(inp); } SCTP_INP_WUNLOCK(inp); } if (stcb->asoc.state & SCTP_STATE_ABOUT_TO_BE_FREED) { SCTP_TCB_LOCK(stcb); sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_19); } return (0); } #ifdef INET int sctp_ingetaddr(struct socket *so, struct sockaddr **addr) { struct sockaddr_in *sin; uint32_t vrf_id; struct sctp_inpcb *inp; struct sctp_ifa *sctp_ifa; /* * Do the malloc first in case it blocks. */ SCTP_MALLOC_SONAME(sin, struct sockaddr_in *, sizeof *sin); if (sin == NULL) return (ENOMEM); sin->sin_family = AF_INET; sin->sin_len = sizeof(*sin); inp = (struct sctp_inpcb *)so->so_pcb; if (!inp) { SCTP_FREE_SONAME(sin); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (ECONNRESET); } SCTP_INP_RLOCK(inp); sin->sin_port = inp->sctp_lport; if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) { if (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) { struct sctp_tcb *stcb; struct sockaddr_in *sin_a; struct sctp_nets *net; int fnd; stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb == NULL) { goto notConn; } fnd = 0; sin_a = NULL; SCTP_TCB_LOCK(stcb); TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { sin_a = (struct sockaddr_in *)&net->ro._l_addr; if (sin_a == NULL) /* this will make coverity happy */ continue; if (sin_a->sin_family == AF_INET) { fnd = 1; break; } } if ((!fnd) || (sin_a == NULL)) { /* punt */ SCTP_TCB_UNLOCK(stcb); goto notConn; } vrf_id = inp->def_vrf_id; sctp_ifa = sctp_source_address_selection(inp, stcb, (sctp_route_t *) & net->ro, net, 0, vrf_id); if (sctp_ifa) { sin->sin_addr = sctp_ifa->address.sin.sin_addr; sctp_free_ifa(sctp_ifa); } SCTP_TCB_UNLOCK(stcb); } else { /* For the bound all case you get back 0 */ notConn: sin->sin_addr.s_addr = 0; } } else { /* Take the first IPv4 address in the list */ struct sctp_laddr *laddr; int fnd = 0; LIST_FOREACH(laddr, &inp->sctp_addr_list, sctp_nxt_addr) { if (laddr->ifa->address.sa.sa_family == AF_INET) { struct sockaddr_in *sin_a; sin_a = &laddr->ifa->address.sin; sin->sin_addr = sin_a->sin_addr; fnd = 1; break; } } if (!fnd) { SCTP_FREE_SONAME(sin); SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); return (ENOENT); } } SCTP_INP_RUNLOCK(inp); (*addr) = (struct sockaddr *)sin; return (0); } int sctp_peeraddr(struct socket *so, struct sockaddr **addr) { struct sockaddr_in *sin; int fnd; struct sockaddr_in *sin_a; struct sctp_inpcb *inp; struct sctp_tcb *stcb; struct sctp_nets *net; /* Do the malloc first in case it blocks. */ SCTP_MALLOC_SONAME(sin, struct sockaddr_in *, sizeof *sin); if (sin == NULL) return (ENOMEM); sin->sin_family = AF_INET; sin->sin_len = sizeof(*sin); inp = (struct sctp_inpcb *)so->so_pcb; if ((inp == NULL) || ((inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) == 0)) { /* UDP type and listeners will drop out here */ SCTP_FREE_SONAME(sin); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOTCONN); return (ENOTCONN); } SCTP_INP_RLOCK(inp); stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb) { SCTP_TCB_LOCK(stcb); } SCTP_INP_RUNLOCK(inp); if (stcb == NULL) { SCTP_FREE_SONAME(sin); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); return (ECONNRESET); } fnd = 0; TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { sin_a = (struct sockaddr_in *)&net->ro._l_addr; if (sin_a->sin_family == AF_INET) { fnd = 1; sin->sin_port = stcb->rport; sin->sin_addr = sin_a->sin_addr; break; } } SCTP_TCB_UNLOCK(stcb); if (!fnd) { /* No IPv4 address */ SCTP_FREE_SONAME(sin); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, ENOENT); return (ENOENT); } (*addr) = (struct sockaddr *)sin; return (0); } struct pr_usrreqs sctp_usrreqs = { .pru_abort = sctp_abort, .pru_accept = sctp_accept, .pru_attach = sctp_attach, .pru_bind = sctp_bind, .pru_connect = sctp_connect, .pru_control = in_control, .pru_close = sctp_close, .pru_detach = sctp_close, .pru_sopoll = sopoll_generic, .pru_flush = sctp_flush, .pru_disconnect = sctp_disconnect, .pru_listen = sctp_listen, .pru_peeraddr = sctp_peeraddr, .pru_send = sctp_sendm, .pru_shutdown = sctp_shutdown, .pru_sockaddr = sctp_ingetaddr, .pru_sosend = sctp_sosend, .pru_soreceive = sctp_soreceive }; #endif Index: projects/vnet/sys/netinet/sctputil.c =================================================================== --- projects/vnet/sys/netinet/sctputil.c (revision 302157) +++ projects/vnet/sys/netinet/sctputil.c (revision 302158) @@ -1,7284 +1,7255 @@ /*- * Copyright (c) 2001-2008, by Cisco Systems, Inc. All rights reserved. * Copyright (c) 2008-2012, by Randall Stewart. All rights reserved. * Copyright (c) 2008-2012, by Michael Tuexen. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * a) Redistributions of source code must retain the above copyright notice, * this list of conditions and the following disclaimer. * * b) Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in * the documentation and/or other materials provided with the distribution. * * c) Neither the name of Cisco Systems, Inc. nor the names of its * contributors may be used to endorse or promote products derived * from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF * THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #ifdef INET6 #include #endif #include #include #include #include #include /* for sctp_deliver_data() */ #include #include #include #include #include #include #ifdef INET6 #include #endif #ifndef KTR_SCTP #define KTR_SCTP KTR_SUBSYS #endif extern const struct sctp_cc_functions sctp_cc_functions[]; extern const struct sctp_ss_functions sctp_ss_functions[]; void sctp_sblog(struct sockbuf *sb, struct sctp_tcb *stcb, int from, int incr) { struct sctp_cwnd_log sctp_clog; sctp_clog.x.sb.stcb = stcb; sctp_clog.x.sb.so_sbcc = sb->sb_cc; if (stcb) sctp_clog.x.sb.stcb_sbcc = stcb->asoc.sb_cc; else sctp_clog.x.sb.stcb_sbcc = 0; sctp_clog.x.sb.incr = incr; SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_SB, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } void sctp_log_closing(struct sctp_inpcb *inp, struct sctp_tcb *stcb, int16_t loc) { struct sctp_cwnd_log sctp_clog; sctp_clog.x.close.inp = (void *)inp; sctp_clog.x.close.sctp_flags = inp->sctp_flags; if (stcb) { sctp_clog.x.close.stcb = (void *)stcb; sctp_clog.x.close.state = (uint16_t) stcb->asoc.state; } else { sctp_clog.x.close.stcb = 0; sctp_clog.x.close.state = 0; } sctp_clog.x.close.loc = loc; SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_CLOSE, 0, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } void rto_logging(struct sctp_nets *net, int from) { struct sctp_cwnd_log sctp_clog; memset(&sctp_clog, 0, sizeof(sctp_clog)); sctp_clog.x.rto.net = (void *)net; sctp_clog.x.rto.rtt = net->rtt / 1000; SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_RTT, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } void sctp_log_strm_del_alt(struct sctp_tcb *stcb, uint32_t tsn, uint16_t sseq, uint16_t stream, int from) { struct sctp_cwnd_log sctp_clog; sctp_clog.x.strlog.stcb = stcb; sctp_clog.x.strlog.n_tsn = tsn; sctp_clog.x.strlog.n_sseq = sseq; sctp_clog.x.strlog.e_tsn = 0; sctp_clog.x.strlog.e_sseq = 0; sctp_clog.x.strlog.strm = stream; SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_STRM, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } void sctp_log_nagle_event(struct sctp_tcb *stcb, int action) { struct sctp_cwnd_log sctp_clog; sctp_clog.x.nagle.stcb = (void *)stcb; sctp_clog.x.nagle.total_flight = stcb->asoc.total_flight; sctp_clog.x.nagle.total_in_queue = stcb->asoc.total_output_queue_size; sctp_clog.x.nagle.count_in_queue = stcb->asoc.chunks_on_out_queue; sctp_clog.x.nagle.count_in_flight = stcb->asoc.total_flight_count; SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_NAGLE, action, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } void sctp_log_sack(uint32_t old_cumack, uint32_t cumack, uint32_t tsn, uint16_t gaps, uint16_t dups, int from) { struct sctp_cwnd_log sctp_clog; sctp_clog.x.sack.cumack = cumack; sctp_clog.x.sack.oldcumack = old_cumack; sctp_clog.x.sack.tsn = tsn; sctp_clog.x.sack.numGaps = gaps; sctp_clog.x.sack.numDups = dups; SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_SACK, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } void sctp_log_map(uint32_t map, uint32_t cum, uint32_t high, int from) { struct sctp_cwnd_log sctp_clog; memset(&sctp_clog, 0, sizeof(sctp_clog)); sctp_clog.x.map.base = map; sctp_clog.x.map.cum = cum; sctp_clog.x.map.high = high; SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_MAP, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } void sctp_log_fr(uint32_t biggest_tsn, uint32_t biggest_new_tsn, uint32_t tsn, int from) { struct sctp_cwnd_log sctp_clog; memset(&sctp_clog, 0, sizeof(sctp_clog)); sctp_clog.x.fr.largest_tsn = biggest_tsn; sctp_clog.x.fr.largest_new_tsn = biggest_new_tsn; sctp_clog.x.fr.tsn = tsn; SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_FR, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } #ifdef SCTP_MBUF_LOGGING void sctp_log_mb(struct mbuf *m, int from) { struct sctp_cwnd_log sctp_clog; sctp_clog.x.mb.mp = m; sctp_clog.x.mb.mbuf_flags = (uint8_t) (SCTP_BUF_GET_FLAGS(m)); sctp_clog.x.mb.size = (uint16_t) (SCTP_BUF_LEN(m)); sctp_clog.x.mb.data = SCTP_BUF_AT(m, 0); if (SCTP_BUF_IS_EXTENDED(m)) { sctp_clog.x.mb.ext = SCTP_BUF_EXTEND_BASE(m); sctp_clog.x.mb.refcnt = (uint8_t) (SCTP_BUF_EXTEND_REFCNT(m)); } else { sctp_clog.x.mb.ext = 0; sctp_clog.x.mb.refcnt = 0; } SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_MBUF, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } void sctp_log_mbc(struct mbuf *m, int from) { struct mbuf *mat; for (mat = m; mat; mat = SCTP_BUF_NEXT(mat)) { sctp_log_mb(mat, from); } } #endif void sctp_log_strm_del(struct sctp_queued_to_read *control, struct sctp_queued_to_read *poschk, int from) { struct sctp_cwnd_log sctp_clog; if (control == NULL) { SCTP_PRINTF("Gak log of NULL?\n"); return; } sctp_clog.x.strlog.stcb = control->stcb; sctp_clog.x.strlog.n_tsn = control->sinfo_tsn; sctp_clog.x.strlog.n_sseq = control->sinfo_ssn; sctp_clog.x.strlog.strm = control->sinfo_stream; if (poschk != NULL) { sctp_clog.x.strlog.e_tsn = poschk->sinfo_tsn; sctp_clog.x.strlog.e_sseq = poschk->sinfo_ssn; } else { sctp_clog.x.strlog.e_tsn = 0; sctp_clog.x.strlog.e_sseq = 0; } SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_STRM, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } void sctp_log_cwnd(struct sctp_tcb *stcb, struct sctp_nets *net, int augment, uint8_t from) { struct sctp_cwnd_log sctp_clog; sctp_clog.x.cwnd.net = net; if (stcb->asoc.send_queue_cnt > 255) sctp_clog.x.cwnd.cnt_in_send = 255; else sctp_clog.x.cwnd.cnt_in_send = stcb->asoc.send_queue_cnt; if (stcb->asoc.stream_queue_cnt > 255) sctp_clog.x.cwnd.cnt_in_str = 255; else sctp_clog.x.cwnd.cnt_in_str = stcb->asoc.stream_queue_cnt; if (net) { sctp_clog.x.cwnd.cwnd_new_value = net->cwnd; sctp_clog.x.cwnd.inflight = net->flight_size; sctp_clog.x.cwnd.pseudo_cumack = net->pseudo_cumack; sctp_clog.x.cwnd.meets_pseudo_cumack = net->new_pseudo_cumack; sctp_clog.x.cwnd.need_new_pseudo_cumack = net->find_pseudo_cumack; } if (SCTP_CWNDLOG_PRESEND == from) { sctp_clog.x.cwnd.meets_pseudo_cumack = stcb->asoc.peers_rwnd; } sctp_clog.x.cwnd.cwnd_augment = augment; SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_CWND, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } void sctp_log_lock(struct sctp_inpcb *inp, struct sctp_tcb *stcb, uint8_t from) { struct sctp_cwnd_log sctp_clog; memset(&sctp_clog, 0, sizeof(sctp_clog)); if (inp) { sctp_clog.x.lock.sock = (void *)inp->sctp_socket; } else { sctp_clog.x.lock.sock = (void *)NULL; } sctp_clog.x.lock.inp = (void *)inp; if (stcb) { sctp_clog.x.lock.tcb_lock = mtx_owned(&stcb->tcb_mtx); } else { sctp_clog.x.lock.tcb_lock = SCTP_LOCK_UNKNOWN; } if (inp) { sctp_clog.x.lock.inp_lock = mtx_owned(&inp->inp_mtx); sctp_clog.x.lock.create_lock = mtx_owned(&inp->inp_create_mtx); } else { sctp_clog.x.lock.inp_lock = SCTP_LOCK_UNKNOWN; sctp_clog.x.lock.create_lock = SCTP_LOCK_UNKNOWN; } sctp_clog.x.lock.info_lock = rw_wowned(&SCTP_BASE_INFO(ipi_ep_mtx)); if (inp && (inp->sctp_socket)) { sctp_clog.x.lock.sock_lock = mtx_owned(&(inp->sctp_socket->so_rcv.sb_mtx)); sctp_clog.x.lock.sockrcvbuf_lock = mtx_owned(&(inp->sctp_socket->so_rcv.sb_mtx)); sctp_clog.x.lock.socksndbuf_lock = mtx_owned(&(inp->sctp_socket->so_snd.sb_mtx)); } else { sctp_clog.x.lock.sock_lock = SCTP_LOCK_UNKNOWN; sctp_clog.x.lock.sockrcvbuf_lock = SCTP_LOCK_UNKNOWN; sctp_clog.x.lock.socksndbuf_lock = SCTP_LOCK_UNKNOWN; } SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_LOCK_EVENT, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } void sctp_log_maxburst(struct sctp_tcb *stcb, struct sctp_nets *net, int error, int burst, uint8_t from) { struct sctp_cwnd_log sctp_clog; memset(&sctp_clog, 0, sizeof(sctp_clog)); sctp_clog.x.cwnd.net = net; sctp_clog.x.cwnd.cwnd_new_value = error; sctp_clog.x.cwnd.inflight = net->flight_size; sctp_clog.x.cwnd.cwnd_augment = burst; if (stcb->asoc.send_queue_cnt > 255) sctp_clog.x.cwnd.cnt_in_send = 255; else sctp_clog.x.cwnd.cnt_in_send = stcb->asoc.send_queue_cnt; if (stcb->asoc.stream_queue_cnt > 255) sctp_clog.x.cwnd.cnt_in_str = 255; else sctp_clog.x.cwnd.cnt_in_str = stcb->asoc.stream_queue_cnt; SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_MAXBURST, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } void sctp_log_rwnd(uint8_t from, uint32_t peers_rwnd, uint32_t snd_size, uint32_t overhead) { struct sctp_cwnd_log sctp_clog; sctp_clog.x.rwnd.rwnd = peers_rwnd; sctp_clog.x.rwnd.send_size = snd_size; sctp_clog.x.rwnd.overhead = overhead; sctp_clog.x.rwnd.new_rwnd = 0; SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_RWND, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } void sctp_log_rwnd_set(uint8_t from, uint32_t peers_rwnd, uint32_t flight_size, uint32_t overhead, uint32_t a_rwndval) { struct sctp_cwnd_log sctp_clog; sctp_clog.x.rwnd.rwnd = peers_rwnd; sctp_clog.x.rwnd.send_size = flight_size; sctp_clog.x.rwnd.overhead = overhead; sctp_clog.x.rwnd.new_rwnd = a_rwndval; SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_RWND, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } #ifdef SCTP_MBCNT_LOGGING static void sctp_log_mbcnt(uint8_t from, uint32_t total_oq, uint32_t book, uint32_t total_mbcnt_q, uint32_t mbcnt) { struct sctp_cwnd_log sctp_clog; sctp_clog.x.mbcnt.total_queue_size = total_oq; sctp_clog.x.mbcnt.size_change = book; sctp_clog.x.mbcnt.total_queue_mb_size = total_mbcnt_q; sctp_clog.x.mbcnt.mbcnt_change = mbcnt; SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_MBCNT, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } #endif void sctp_misc_ints(uint8_t from, uint32_t a, uint32_t b, uint32_t c, uint32_t d) { SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_MISC_EVENT, from, a, b, c, d); } void sctp_wakeup_log(struct sctp_tcb *stcb, uint32_t wake_cnt, int from) { struct sctp_cwnd_log sctp_clog; sctp_clog.x.wake.stcb = (void *)stcb; sctp_clog.x.wake.wake_cnt = wake_cnt; sctp_clog.x.wake.flight = stcb->asoc.total_flight_count; sctp_clog.x.wake.send_q = stcb->asoc.send_queue_cnt; sctp_clog.x.wake.sent_q = stcb->asoc.sent_queue_cnt; if (stcb->asoc.stream_queue_cnt < 0xff) sctp_clog.x.wake.stream_qcnt = (uint8_t) stcb->asoc.stream_queue_cnt; else sctp_clog.x.wake.stream_qcnt = 0xff; if (stcb->asoc.chunks_on_out_queue < 0xff) sctp_clog.x.wake.chunks_on_oque = (uint8_t) stcb->asoc.chunks_on_out_queue; else sctp_clog.x.wake.chunks_on_oque = 0xff; sctp_clog.x.wake.sctpflags = 0; /* set in the defered mode stuff */ if (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_DONT_WAKE) sctp_clog.x.wake.sctpflags |= 1; if (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_WAKEOUTPUT) sctp_clog.x.wake.sctpflags |= 2; if (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_WAKEINPUT) sctp_clog.x.wake.sctpflags |= 4; /* what about the sb */ if (stcb->sctp_socket) { struct socket *so = stcb->sctp_socket; sctp_clog.x.wake.sbflags = (uint8_t) ((so->so_snd.sb_flags & 0x00ff)); } else { sctp_clog.x.wake.sbflags = 0xff; } SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_WAKE, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } void sctp_log_block(uint8_t from, struct sctp_association *asoc, size_t sendlen) { struct sctp_cwnd_log sctp_clog; sctp_clog.x.blk.onsb = asoc->total_output_queue_size; sctp_clog.x.blk.send_sent_qcnt = (uint16_t) (asoc->send_queue_cnt + asoc->sent_queue_cnt); sctp_clog.x.blk.peer_rwnd = asoc->peers_rwnd; sctp_clog.x.blk.stream_qcnt = (uint16_t) asoc->stream_queue_cnt; sctp_clog.x.blk.chunks_on_oque = (uint16_t) asoc->chunks_on_out_queue; sctp_clog.x.blk.flight_size = (uint16_t) (asoc->total_flight / 1024); sctp_clog.x.blk.sndlen = (uint32_t) sendlen; SCTP_CTR6(KTR_SCTP, "SCTP:%d[%d]:%x-%x-%x-%x", SCTP_LOG_EVENT_BLOCK, from, sctp_clog.x.misc.log1, sctp_clog.x.misc.log2, sctp_clog.x.misc.log3, sctp_clog.x.misc.log4); } int sctp_fill_stat_log(void *optval SCTP_UNUSED, size_t *optsize SCTP_UNUSED) { /* May need to fix this if ktrdump does not work */ return (0); } #ifdef SCTP_AUDITING_ENABLED uint8_t sctp_audit_data[SCTP_AUDIT_SIZE][2]; static int sctp_audit_indx = 0; static void sctp_print_audit_report(void) { int i; int cnt; cnt = 0; for (i = sctp_audit_indx; i < SCTP_AUDIT_SIZE; i++) { if ((sctp_audit_data[i][0] == 0xe0) && (sctp_audit_data[i][1] == 0x01)) { cnt = 0; SCTP_PRINTF("\n"); } else if (sctp_audit_data[i][0] == 0xf0) { cnt = 0; SCTP_PRINTF("\n"); } else if ((sctp_audit_data[i][0] == 0xc0) && (sctp_audit_data[i][1] == 0x01)) { SCTP_PRINTF("\n"); cnt = 0; } SCTP_PRINTF("%2.2x%2.2x ", (uint32_t) sctp_audit_data[i][0], (uint32_t) sctp_audit_data[i][1]); cnt++; if ((cnt % 14) == 0) SCTP_PRINTF("\n"); } for (i = 0; i < sctp_audit_indx; i++) { if ((sctp_audit_data[i][0] == 0xe0) && (sctp_audit_data[i][1] == 0x01)) { cnt = 0; SCTP_PRINTF("\n"); } else if (sctp_audit_data[i][0] == 0xf0) { cnt = 0; SCTP_PRINTF("\n"); } else if ((sctp_audit_data[i][0] == 0xc0) && (sctp_audit_data[i][1] == 0x01)) { SCTP_PRINTF("\n"); cnt = 0; } SCTP_PRINTF("%2.2x%2.2x ", (uint32_t) sctp_audit_data[i][0], (uint32_t) sctp_audit_data[i][1]); cnt++; if ((cnt % 14) == 0) SCTP_PRINTF("\n"); } SCTP_PRINTF("\n"); } void sctp_auditing(int from, struct sctp_inpcb *inp, struct sctp_tcb *stcb, struct sctp_nets *net) { int resend_cnt, tot_out, rep, tot_book_cnt; struct sctp_nets *lnet; struct sctp_tmit_chunk *chk; sctp_audit_data[sctp_audit_indx][0] = 0xAA; sctp_audit_data[sctp_audit_indx][1] = 0x000000ff & from; sctp_audit_indx++; if (sctp_audit_indx >= SCTP_AUDIT_SIZE) { sctp_audit_indx = 0; } if (inp == NULL) { sctp_audit_data[sctp_audit_indx][0] = 0xAF; sctp_audit_data[sctp_audit_indx][1] = 0x01; sctp_audit_indx++; if (sctp_audit_indx >= SCTP_AUDIT_SIZE) { sctp_audit_indx = 0; } return; } if (stcb == NULL) { sctp_audit_data[sctp_audit_indx][0] = 0xAF; sctp_audit_data[sctp_audit_indx][1] = 0x02; sctp_audit_indx++; if (sctp_audit_indx >= SCTP_AUDIT_SIZE) { sctp_audit_indx = 0; } return; } sctp_audit_data[sctp_audit_indx][0] = 0xA1; sctp_audit_data[sctp_audit_indx][1] = (0x000000ff & stcb->asoc.sent_queue_retran_cnt); sctp_audit_indx++; if (sctp_audit_indx >= SCTP_AUDIT_SIZE) { sctp_audit_indx = 0; } rep = 0; tot_book_cnt = 0; resend_cnt = tot_out = 0; TAILQ_FOREACH(chk, &stcb->asoc.sent_queue, sctp_next) { if (chk->sent == SCTP_DATAGRAM_RESEND) { resend_cnt++; } else if (chk->sent < SCTP_DATAGRAM_RESEND) { tot_out += chk->book_size; tot_book_cnt++; } } if (resend_cnt != stcb->asoc.sent_queue_retran_cnt) { sctp_audit_data[sctp_audit_indx][0] = 0xAF; sctp_audit_data[sctp_audit_indx][1] = 0xA1; sctp_audit_indx++; if (sctp_audit_indx >= SCTP_AUDIT_SIZE) { sctp_audit_indx = 0; } SCTP_PRINTF("resend_cnt:%d asoc-tot:%d\n", resend_cnt, stcb->asoc.sent_queue_retran_cnt); rep = 1; stcb->asoc.sent_queue_retran_cnt = resend_cnt; sctp_audit_data[sctp_audit_indx][0] = 0xA2; sctp_audit_data[sctp_audit_indx][1] = (0x000000ff & stcb->asoc.sent_queue_retran_cnt); sctp_audit_indx++; if (sctp_audit_indx >= SCTP_AUDIT_SIZE) { sctp_audit_indx = 0; } } if (tot_out != stcb->asoc.total_flight) { sctp_audit_data[sctp_audit_indx][0] = 0xAF; sctp_audit_data[sctp_audit_indx][1] = 0xA2; sctp_audit_indx++; if (sctp_audit_indx >= SCTP_AUDIT_SIZE) { sctp_audit_indx = 0; } rep = 1; SCTP_PRINTF("tot_flt:%d asoc_tot:%d\n", tot_out, (int)stcb->asoc.total_flight); stcb->asoc.total_flight = tot_out; } if (tot_book_cnt != stcb->asoc.total_flight_count) { sctp_audit_data[sctp_audit_indx][0] = 0xAF; sctp_audit_data[sctp_audit_indx][1] = 0xA5; sctp_audit_indx++; if (sctp_audit_indx >= SCTP_AUDIT_SIZE) { sctp_audit_indx = 0; } rep = 1; SCTP_PRINTF("tot_flt_book:%d\n", tot_book_cnt); stcb->asoc.total_flight_count = tot_book_cnt; } tot_out = 0; TAILQ_FOREACH(lnet, &stcb->asoc.nets, sctp_next) { tot_out += lnet->flight_size; } if (tot_out != stcb->asoc.total_flight) { sctp_audit_data[sctp_audit_indx][0] = 0xAF; sctp_audit_data[sctp_audit_indx][1] = 0xA3; sctp_audit_indx++; if (sctp_audit_indx >= SCTP_AUDIT_SIZE) { sctp_audit_indx = 0; } rep = 1; SCTP_PRINTF("real flight:%d net total was %d\n", stcb->asoc.total_flight, tot_out); /* now corrective action */ TAILQ_FOREACH(lnet, &stcb->asoc.nets, sctp_next) { tot_out = 0; TAILQ_FOREACH(chk, &stcb->asoc.sent_queue, sctp_next) { if ((chk->whoTo == lnet) && (chk->sent < SCTP_DATAGRAM_RESEND)) { tot_out += chk->book_size; } } if (lnet->flight_size != tot_out) { SCTP_PRINTF("net:%p flight was %d corrected to %d\n", (void *)lnet, lnet->flight_size, tot_out); lnet->flight_size = tot_out; } } } if (rep) { sctp_print_audit_report(); } } void sctp_audit_log(uint8_t ev, uint8_t fd) { sctp_audit_data[sctp_audit_indx][0] = ev; sctp_audit_data[sctp_audit_indx][1] = fd; sctp_audit_indx++; if (sctp_audit_indx >= SCTP_AUDIT_SIZE) { sctp_audit_indx = 0; } } #endif /* * sctp_stop_timers_for_shutdown() should be called * when entering the SHUTDOWN_SENT or SHUTDOWN_ACK_SENT * state to make sure that all timers are stopped. */ void sctp_stop_timers_for_shutdown(struct sctp_tcb *stcb) { struct sctp_association *asoc; struct sctp_nets *net; asoc = &stcb->asoc; (void)SCTP_OS_TIMER_STOP(&asoc->dack_timer.timer); (void)SCTP_OS_TIMER_STOP(&asoc->strreset_timer.timer); (void)SCTP_OS_TIMER_STOP(&asoc->asconf_timer.timer); (void)SCTP_OS_TIMER_STOP(&asoc->autoclose_timer.timer); (void)SCTP_OS_TIMER_STOP(&asoc->delayed_event_timer.timer); TAILQ_FOREACH(net, &asoc->nets, sctp_next) { (void)SCTP_OS_TIMER_STOP(&net->pmtu_timer.timer); (void)SCTP_OS_TIMER_STOP(&net->hb_timer.timer); } } /* * a list of sizes based on typical mtu's, used only if next hop size not * returned. */ static uint32_t sctp_mtu_sizes[] = { 68, 296, 508, 512, 544, 576, 1006, 1492, 1500, 1536, 2002, 2048, 4352, 4464, 8166, 17914, 32000, 65535 }; /* * Return the largest MTU smaller than val. If there is no * entry, just return val. */ uint32_t sctp_get_prev_mtu(uint32_t val) { uint32_t i; if (val <= sctp_mtu_sizes[0]) { return (val); } for (i = 1; i < (sizeof(sctp_mtu_sizes) / sizeof(uint32_t)); i++) { if (val <= sctp_mtu_sizes[i]) { break; } } return (sctp_mtu_sizes[i - 1]); } /* * Return the smallest MTU larger than val. If there is no * entry, just return val. */ uint32_t sctp_get_next_mtu(uint32_t val) { /* select another MTU that is just bigger than this one */ uint32_t i; for (i = 0; i < (sizeof(sctp_mtu_sizes) / sizeof(uint32_t)); i++) { if (val < sctp_mtu_sizes[i]) { return (sctp_mtu_sizes[i]); } } return (val); } void sctp_fill_random_store(struct sctp_pcb *m) { /* * Here we use the MD5/SHA-1 to hash with our good randomNumbers and * our counter. The result becomes our good random numbers and we * then setup to give these out. Note that we do no locking to * protect this. This is ok, since if competing folks call this we * will get more gobbled gook in the random store which is what we * want. There is a danger that two guys will use the same random * numbers, but thats ok too since that is random as well :-> */ m->store_at = 0; (void)sctp_hmac(SCTP_HMAC, (uint8_t *) m->random_numbers, sizeof(m->random_numbers), (uint8_t *) & m->random_counter, sizeof(m->random_counter), (uint8_t *) m->random_store); m->random_counter++; } uint32_t sctp_select_initial_TSN(struct sctp_pcb *inp) { /* * A true implementation should use random selection process to get * the initial stream sequence number, using RFC1750 as a good * guideline */ uint32_t x, *xp; uint8_t *p; int store_at, new_store; if (inp->initial_sequence_debug != 0) { uint32_t ret; ret = inp->initial_sequence_debug; inp->initial_sequence_debug++; return (ret); } retry: store_at = inp->store_at; new_store = store_at + sizeof(uint32_t); if (new_store >= (SCTP_SIGNATURE_SIZE - 3)) { new_store = 0; } if (!atomic_cmpset_int(&inp->store_at, store_at, new_store)) { goto retry; } if (new_store == 0) { /* Refill the random store */ sctp_fill_random_store(inp); } p = &inp->random_store[store_at]; xp = (uint32_t *) p; x = *xp; return (x); } uint32_t sctp_select_a_tag(struct sctp_inpcb *inp, uint16_t lport, uint16_t rport, int check) { uint32_t x; struct timeval now; if (check) { (void)SCTP_GETTIME_TIMEVAL(&now); } for (;;) { x = sctp_select_initial_TSN(&inp->sctp_ep); if (x == 0) { /* we never use 0 */ continue; } if (!check || sctp_is_vtag_good(x, lport, rport, &now)) { break; } } return (x); } int32_t sctp_map_assoc_state(int kernel_state) { int32_t user_state; if (kernel_state & SCTP_STATE_WAS_ABORTED) { user_state = SCTP_CLOSED; } else if (kernel_state & SCTP_STATE_SHUTDOWN_PENDING) { user_state = SCTP_SHUTDOWN_PENDING; } else { switch (kernel_state & SCTP_STATE_MASK) { case SCTP_STATE_EMPTY: user_state = SCTP_CLOSED; break; case SCTP_STATE_INUSE: user_state = SCTP_CLOSED; break; case SCTP_STATE_COOKIE_WAIT: user_state = SCTP_COOKIE_WAIT; break; case SCTP_STATE_COOKIE_ECHOED: user_state = SCTP_COOKIE_ECHOED; break; case SCTP_STATE_OPEN: user_state = SCTP_ESTABLISHED; break; case SCTP_STATE_SHUTDOWN_SENT: user_state = SCTP_SHUTDOWN_SENT; break; case SCTP_STATE_SHUTDOWN_RECEIVED: user_state = SCTP_SHUTDOWN_RECEIVED; break; case SCTP_STATE_SHUTDOWN_ACK_SENT: user_state = SCTP_SHUTDOWN_ACK_SENT; break; default: user_state = SCTP_CLOSED; break; } } return (user_state); } int sctp_init_asoc(struct sctp_inpcb *inp, struct sctp_tcb *stcb, uint32_t override_tag, uint32_t vrf_id, uint16_t o_strms) { struct sctp_association *asoc; /* * Anything set to zero is taken care of by the allocation routine's * bzero */ /* * Up front select what scoping to apply on addresses I tell my peer * Not sure what to do with these right now, we will need to come up * with a way to set them. We may need to pass them through from the * caller in the sctp_aloc_assoc() function. */ int i; #if defined(SCTP_DETAILED_STR_STATS) int j; #endif asoc = &stcb->asoc; /* init all variables to a known value. */ SCTP_SET_STATE(&stcb->asoc, SCTP_STATE_INUSE); asoc->max_burst = inp->sctp_ep.max_burst; asoc->fr_max_burst = inp->sctp_ep.fr_max_burst; asoc->heart_beat_delay = TICKS_TO_MSEC(inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_HEARTBEAT]); asoc->cookie_life = inp->sctp_ep.def_cookie_life; asoc->sctp_cmt_on_off = inp->sctp_cmt_on_off; asoc->ecn_supported = inp->ecn_supported; asoc->prsctp_supported = inp->prsctp_supported; asoc->idata_supported = inp->idata_supported; asoc->auth_supported = inp->auth_supported; asoc->asconf_supported = inp->asconf_supported; asoc->reconfig_supported = inp->reconfig_supported; asoc->nrsack_supported = inp->nrsack_supported; asoc->pktdrop_supported = inp->pktdrop_supported; asoc->idata_supported = inp->idata_supported; asoc->sctp_cmt_pf = (uint8_t) 0; asoc->sctp_frag_point = inp->sctp_frag_point; asoc->sctp_features = inp->sctp_features; asoc->default_dscp = inp->sctp_ep.default_dscp; asoc->max_cwnd = inp->max_cwnd; #ifdef INET6 if (inp->sctp_ep.default_flowlabel) { asoc->default_flowlabel = inp->sctp_ep.default_flowlabel; } else { if (inp->ip_inp.inp.inp_flags & IN6P_AUTOFLOWLABEL) { asoc->default_flowlabel = sctp_select_initial_TSN(&inp->sctp_ep); asoc->default_flowlabel &= 0x000fffff; asoc->default_flowlabel |= 0x80000000; } else { asoc->default_flowlabel = 0; } } #endif asoc->sb_send_resv = 0; if (override_tag) { asoc->my_vtag = override_tag; } else { asoc->my_vtag = sctp_select_a_tag(inp, stcb->sctp_ep->sctp_lport, stcb->rport, 1); } /* Get the nonce tags */ asoc->my_vtag_nonce = sctp_select_a_tag(inp, stcb->sctp_ep->sctp_lport, stcb->rport, 0); asoc->peer_vtag_nonce = sctp_select_a_tag(inp, stcb->sctp_ep->sctp_lport, stcb->rport, 0); asoc->vrf_id = vrf_id; #ifdef SCTP_ASOCLOG_OF_TSNS asoc->tsn_in_at = 0; asoc->tsn_out_at = 0; asoc->tsn_in_wrapped = 0; asoc->tsn_out_wrapped = 0; asoc->cumack_log_at = 0; asoc->cumack_log_atsnt = 0; #endif #ifdef SCTP_FS_SPEC_LOG asoc->fs_index = 0; #endif asoc->refcnt = 0; asoc->assoc_up_sent = 0; asoc->asconf_seq_out = asoc->str_reset_seq_out = asoc->init_seq_number = asoc->sending_seq = sctp_select_initial_TSN(&inp->sctp_ep); asoc->asconf_seq_out_acked = asoc->asconf_seq_out - 1; /* we are optimisitic here */ asoc->peer_supports_nat = 0; asoc->sent_queue_retran_cnt = 0; /* for CMT */ asoc->last_net_cmt_send_started = NULL; /* This will need to be adjusted */ asoc->last_acked_seq = asoc->init_seq_number - 1; asoc->advanced_peer_ack_point = asoc->last_acked_seq; asoc->asconf_seq_in = asoc->last_acked_seq; /* here we are different, we hold the next one we expect */ asoc->str_reset_seq_in = asoc->last_acked_seq + 1; asoc->initial_init_rto_max = inp->sctp_ep.initial_init_rto_max; asoc->initial_rto = inp->sctp_ep.initial_rto; asoc->max_init_times = inp->sctp_ep.max_init_times; asoc->max_send_times = inp->sctp_ep.max_send_times; asoc->def_net_failure = inp->sctp_ep.def_net_failure; asoc->def_net_pf_threshold = inp->sctp_ep.def_net_pf_threshold; asoc->free_chunk_cnt = 0; asoc->iam_blocking = 0; asoc->context = inp->sctp_context; asoc->local_strreset_support = inp->local_strreset_support; asoc->def_send = inp->def_send; asoc->delayed_ack = TICKS_TO_MSEC(inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_RECV]); asoc->sack_freq = inp->sctp_ep.sctp_sack_freq; asoc->pr_sctp_cnt = 0; asoc->total_output_queue_size = 0; if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { asoc->scope.ipv6_addr_legal = 1; if (SCTP_IPV6_V6ONLY(inp) == 0) { asoc->scope.ipv4_addr_legal = 1; } else { asoc->scope.ipv4_addr_legal = 0; } } else { asoc->scope.ipv6_addr_legal = 0; asoc->scope.ipv4_addr_legal = 1; } asoc->my_rwnd = max(SCTP_SB_LIMIT_RCV(inp->sctp_socket), SCTP_MINIMAL_RWND); asoc->peers_rwnd = SCTP_SB_LIMIT_RCV(inp->sctp_socket); asoc->smallest_mtu = inp->sctp_frag_point; asoc->minrto = inp->sctp_ep.sctp_minrto; asoc->maxrto = inp->sctp_ep.sctp_maxrto; asoc->locked_on_sending = NULL; asoc->stream_locked_on = 0; asoc->ecn_echo_cnt_onq = 0; asoc->stream_locked = 0; asoc->send_sack = 1; LIST_INIT(&asoc->sctp_restricted_addrs); TAILQ_INIT(&asoc->nets); TAILQ_INIT(&asoc->pending_reply_queue); TAILQ_INIT(&asoc->asconf_ack_sent); /* Setup to fill the hb random cache at first HB */ asoc->hb_random_idx = 4; asoc->sctp_autoclose_ticks = inp->sctp_ep.auto_close_time; stcb->asoc.congestion_control_module = inp->sctp_ep.sctp_default_cc_module; stcb->asoc.cc_functions = sctp_cc_functions[inp->sctp_ep.sctp_default_cc_module]; stcb->asoc.stream_scheduling_module = inp->sctp_ep.sctp_default_ss_module; stcb->asoc.ss_functions = sctp_ss_functions[inp->sctp_ep.sctp_default_ss_module]; /* * Now the stream parameters, here we allocate space for all streams * that we request by default. */ asoc->strm_realoutsize = asoc->streamoutcnt = asoc->pre_open_streams = o_strms; SCTP_MALLOC(asoc->strmout, struct sctp_stream_out *, asoc->streamoutcnt * sizeof(struct sctp_stream_out), SCTP_M_STRMO); if (asoc->strmout == NULL) { /* big trouble no memory */ SCTP_LTRACE_ERR_RET(NULL, stcb, NULL, SCTP_FROM_SCTPUTIL, ENOMEM); return (ENOMEM); } for (i = 0; i < asoc->streamoutcnt; i++) { /* * inbound side must be set to 0xffff, also NOTE when we get * the INIT-ACK back (for INIT sender) we MUST reduce the * count (streamoutcnt) but first check if we sent to any of * the upper streams that were dropped (if some were). Those * that were dropped must be notified to the upper layer as * failed to send. */ asoc->strmout[i].next_mid_ordered = 0; asoc->strmout[i].next_mid_unordered = 0; TAILQ_INIT(&asoc->strmout[i].outqueue); asoc->strmout[i].chunks_on_queues = 0; #if defined(SCTP_DETAILED_STR_STATS) for (j = 0; j < SCTP_PR_SCTP_MAX + 1; j++) { asoc->strmout[i].abandoned_sent[j] = 0; asoc->strmout[i].abandoned_unsent[j] = 0; } #else asoc->strmout[i].abandoned_sent[0] = 0; asoc->strmout[i].abandoned_unsent[0] = 0; #endif asoc->strmout[i].stream_no = i; asoc->strmout[i].last_msg_incomplete = 0; asoc->strmout[i].state = SCTP_STREAM_OPENING; asoc->ss_functions.sctp_ss_init_stream(&asoc->strmout[i], NULL); } asoc->ss_functions.sctp_ss_init(stcb, asoc, 0); /* Now the mapping array */ asoc->mapping_array_size = SCTP_INITIAL_MAPPING_ARRAY; SCTP_MALLOC(asoc->mapping_array, uint8_t *, asoc->mapping_array_size, SCTP_M_MAP); if (asoc->mapping_array == NULL) { SCTP_FREE(asoc->strmout, SCTP_M_STRMO); SCTP_LTRACE_ERR_RET(NULL, stcb, NULL, SCTP_FROM_SCTPUTIL, ENOMEM); return (ENOMEM); } memset(asoc->mapping_array, 0, asoc->mapping_array_size); SCTP_MALLOC(asoc->nr_mapping_array, uint8_t *, asoc->mapping_array_size, SCTP_M_MAP); if (asoc->nr_mapping_array == NULL) { SCTP_FREE(asoc->strmout, SCTP_M_STRMO); SCTP_FREE(asoc->mapping_array, SCTP_M_MAP); SCTP_LTRACE_ERR_RET(NULL, stcb, NULL, SCTP_FROM_SCTPUTIL, ENOMEM); return (ENOMEM); } memset(asoc->nr_mapping_array, 0, asoc->mapping_array_size); /* Now the init of the other outqueues */ TAILQ_INIT(&asoc->free_chunks); TAILQ_INIT(&asoc->control_send_queue); TAILQ_INIT(&asoc->asconf_send_queue); TAILQ_INIT(&asoc->send_queue); TAILQ_INIT(&asoc->sent_queue); TAILQ_INIT(&asoc->resetHead); asoc->max_inbound_streams = inp->sctp_ep.max_open_streams_intome; TAILQ_INIT(&asoc->asconf_queue); /* authentication fields */ asoc->authinfo.random = NULL; asoc->authinfo.active_keyid = 0; asoc->authinfo.assoc_key = NULL; asoc->authinfo.assoc_keyid = 0; asoc->authinfo.recv_key = NULL; asoc->authinfo.recv_keyid = 0; LIST_INIT(&asoc->shared_keys); asoc->marked_retrans = 0; asoc->port = inp->sctp_ep.port; asoc->timoinit = 0; asoc->timodata = 0; asoc->timosack = 0; asoc->timoshutdown = 0; asoc->timoheartbeat = 0; asoc->timocookie = 0; asoc->timoshutdownack = 0; (void)SCTP_GETTIME_TIMEVAL(&asoc->start_time); asoc->discontinuity_time = asoc->start_time; for (i = 0; i < SCTP_PR_SCTP_MAX + 1; i++) { asoc->abandoned_unsent[i] = 0; asoc->abandoned_sent[i] = 0; } /* * sa_ignore MEMLEAK {memory is put in the assoc mapping array and * freed later when the association is freed. */ return (0); } void sctp_print_mapping_array(struct sctp_association *asoc) { unsigned int i, limit; SCTP_PRINTF("Mapping array size: %d, baseTSN: %8.8x, cumAck: %8.8x, highestTSN: (%8.8x, %8.8x).\n", asoc->mapping_array_size, asoc->mapping_array_base_tsn, asoc->cumulative_tsn, asoc->highest_tsn_inside_map, asoc->highest_tsn_inside_nr_map); for (limit = asoc->mapping_array_size; limit > 1; limit--) { if (asoc->mapping_array[limit - 1] != 0) { break; } } SCTP_PRINTF("Renegable mapping array (last %d entries are zero):\n", asoc->mapping_array_size - limit); for (i = 0; i < limit; i++) { SCTP_PRINTF("%2.2x%c", asoc->mapping_array[i], ((i + 1) % 16) ? ' ' : '\n'); } if (limit % 16) SCTP_PRINTF("\n"); for (limit = asoc->mapping_array_size; limit > 1; limit--) { if (asoc->nr_mapping_array[limit - 1]) { break; } } SCTP_PRINTF("Non renegable mapping array (last %d entries are zero):\n", asoc->mapping_array_size - limit); for (i = 0; i < limit; i++) { SCTP_PRINTF("%2.2x%c", asoc->nr_mapping_array[i], ((i + 1) % 16) ? ' ' : '\n'); } if (limit % 16) SCTP_PRINTF("\n"); } int sctp_expand_mapping_array(struct sctp_association *asoc, uint32_t needed) { /* mapping array needs to grow */ uint8_t *new_array1, *new_array2; uint32_t new_size; new_size = asoc->mapping_array_size + ((needed + 7) / 8 + SCTP_MAPPING_ARRAY_INCR); SCTP_MALLOC(new_array1, uint8_t *, new_size, SCTP_M_MAP); SCTP_MALLOC(new_array2, uint8_t *, new_size, SCTP_M_MAP); if ((new_array1 == NULL) || (new_array2 == NULL)) { /* can't get more, forget it */ SCTP_PRINTF("No memory for expansion of SCTP mapping array %d\n", new_size); if (new_array1) { SCTP_FREE(new_array1, SCTP_M_MAP); } if (new_array2) { SCTP_FREE(new_array2, SCTP_M_MAP); } return (-1); } memset(new_array1, 0, new_size); memset(new_array2, 0, new_size); memcpy(new_array1, asoc->mapping_array, asoc->mapping_array_size); memcpy(new_array2, asoc->nr_mapping_array, asoc->mapping_array_size); SCTP_FREE(asoc->mapping_array, SCTP_M_MAP); SCTP_FREE(asoc->nr_mapping_array, SCTP_M_MAP); asoc->mapping_array = new_array1; asoc->nr_mapping_array = new_array2; asoc->mapping_array_size = new_size; return (0); } static void sctp_iterator_work(struct sctp_iterator *it) { int iteration_count = 0; int inp_skip = 0; int first_in = 1; struct sctp_inpcb *tinp; SCTP_INP_INFO_RLOCK(); SCTP_ITERATOR_LOCK(); if (it->inp) { SCTP_INP_RLOCK(it->inp); SCTP_INP_DECR_REF(it->inp); } if (it->inp == NULL) { /* iterator is complete */ done_with_iterator: SCTP_ITERATOR_UNLOCK(); SCTP_INP_INFO_RUNLOCK(); if (it->function_atend != NULL) { (*it->function_atend) (it->pointer, it->val); } SCTP_FREE(it, SCTP_M_ITER); return; } select_a_new_ep: if (first_in) { first_in = 0; } else { SCTP_INP_RLOCK(it->inp); } while (((it->pcb_flags) && ((it->inp->sctp_flags & it->pcb_flags) != it->pcb_flags)) || ((it->pcb_features) && ((it->inp->sctp_features & it->pcb_features) != it->pcb_features))) { /* endpoint flags or features don't match, so keep looking */ if (it->iterator_flags & SCTP_ITERATOR_DO_SINGLE_INP) { SCTP_INP_RUNLOCK(it->inp); goto done_with_iterator; } tinp = it->inp; it->inp = LIST_NEXT(it->inp, sctp_list); SCTP_INP_RUNLOCK(tinp); if (it->inp == NULL) { goto done_with_iterator; } SCTP_INP_RLOCK(it->inp); } /* now go through each assoc which is in the desired state */ if (it->done_current_ep == 0) { if (it->function_inp != NULL) inp_skip = (*it->function_inp) (it->inp, it->pointer, it->val); it->done_current_ep = 1; } if (it->stcb == NULL) { /* run the per instance function */ it->stcb = LIST_FIRST(&it->inp->sctp_asoc_list); } if ((inp_skip) || it->stcb == NULL) { if (it->function_inp_end != NULL) { inp_skip = (*it->function_inp_end) (it->inp, it->pointer, it->val); } SCTP_INP_RUNLOCK(it->inp); goto no_stcb; } while (it->stcb) { SCTP_TCB_LOCK(it->stcb); if (it->asoc_state && ((it->stcb->asoc.state & it->asoc_state) != it->asoc_state)) { /* not in the right state... keep looking */ SCTP_TCB_UNLOCK(it->stcb); goto next_assoc; } /* see if we have limited out the iterator loop */ iteration_count++; if (iteration_count > SCTP_ITERATOR_MAX_AT_ONCE) { /* Pause to let others grab the lock */ atomic_add_int(&it->stcb->asoc.refcnt, 1); SCTP_TCB_UNLOCK(it->stcb); SCTP_INP_INCR_REF(it->inp); SCTP_INP_RUNLOCK(it->inp); SCTP_ITERATOR_UNLOCK(); SCTP_INP_INFO_RUNLOCK(); SCTP_INP_INFO_RLOCK(); SCTP_ITERATOR_LOCK(); if (sctp_it_ctl.iterator_flags) { /* We won't be staying here */ SCTP_INP_DECR_REF(it->inp); atomic_add_int(&it->stcb->asoc.refcnt, -1); if (sctp_it_ctl.iterator_flags & SCTP_ITERATOR_STOP_CUR_IT) { sctp_it_ctl.iterator_flags &= ~SCTP_ITERATOR_STOP_CUR_IT; goto done_with_iterator; } if (sctp_it_ctl.iterator_flags & SCTP_ITERATOR_STOP_CUR_INP) { sctp_it_ctl.iterator_flags &= ~SCTP_ITERATOR_STOP_CUR_INP; goto no_stcb; } /* If we reach here huh? */ SCTP_PRINTF("Unknown it ctl flag %x\n", sctp_it_ctl.iterator_flags); sctp_it_ctl.iterator_flags = 0; } SCTP_INP_RLOCK(it->inp); SCTP_INP_DECR_REF(it->inp); SCTP_TCB_LOCK(it->stcb); atomic_add_int(&it->stcb->asoc.refcnt, -1); iteration_count = 0; } /* run function on this one */ (*it->function_assoc) (it->inp, it->stcb, it->pointer, it->val); /* * we lie here, it really needs to have its own type but * first I must verify that this won't effect things :-0 */ if (it->no_chunk_output == 0) sctp_chunk_output(it->inp, it->stcb, SCTP_OUTPUT_FROM_T3, SCTP_SO_NOT_LOCKED); SCTP_TCB_UNLOCK(it->stcb); next_assoc: it->stcb = LIST_NEXT(it->stcb, sctp_tcblist); if (it->stcb == NULL) { /* Run last function */ if (it->function_inp_end != NULL) { inp_skip = (*it->function_inp_end) (it->inp, it->pointer, it->val); } } } SCTP_INP_RUNLOCK(it->inp); no_stcb: /* done with all assocs on this endpoint, move on to next endpoint */ it->done_current_ep = 0; if (it->iterator_flags & SCTP_ITERATOR_DO_SINGLE_INP) { it->inp = NULL; } else { it->inp = LIST_NEXT(it->inp, sctp_list); } if (it->inp == NULL) { goto done_with_iterator; } goto select_a_new_ep; } void sctp_iterator_worker(void) { struct sctp_iterator *it, *nit; /* This function is called with the WQ lock in place */ sctp_it_ctl.iterator_running = 1; TAILQ_FOREACH_SAFE(it, &sctp_it_ctl.iteratorhead, sctp_nxt_itr, nit) { sctp_it_ctl.cur_it = it; /* now lets work on this one */ TAILQ_REMOVE(&sctp_it_ctl.iteratorhead, it, sctp_nxt_itr); SCTP_IPI_ITERATOR_WQ_UNLOCK(); CURVNET_SET(it->vn); sctp_iterator_work(it); sctp_it_ctl.cur_it = NULL; CURVNET_RESTORE(); SCTP_IPI_ITERATOR_WQ_LOCK(); /* sa_ignore FREED_MEMORY */ } sctp_it_ctl.iterator_running = 0; return; } static void sctp_handle_addr_wq(void) { /* deal with the ADDR wq from the rtsock calls */ struct sctp_laddr *wi, *nwi; struct sctp_asconf_iterator *asc; SCTP_MALLOC(asc, struct sctp_asconf_iterator *, sizeof(struct sctp_asconf_iterator), SCTP_M_ASC_IT); if (asc == NULL) { /* Try later, no memory */ sctp_timer_start(SCTP_TIMER_TYPE_ADDR_WQ, (struct sctp_inpcb *)NULL, (struct sctp_tcb *)NULL, (struct sctp_nets *)NULL); return; } LIST_INIT(&asc->list_of_work); asc->cnt = 0; SCTP_WQ_ADDR_LOCK(); LIST_FOREACH_SAFE(wi, &SCTP_BASE_INFO(addr_wq), sctp_nxt_addr, nwi) { LIST_REMOVE(wi, sctp_nxt_addr); LIST_INSERT_HEAD(&asc->list_of_work, wi, sctp_nxt_addr); asc->cnt++; } SCTP_WQ_ADDR_UNLOCK(); if (asc->cnt == 0) { SCTP_FREE(asc, SCTP_M_ASC_IT); } else { int ret; ret = sctp_initiate_iterator(sctp_asconf_iterator_ep, sctp_asconf_iterator_stcb, NULL, /* No ep end for boundall */ SCTP_PCB_FLAGS_BOUNDALL, SCTP_PCB_ANY_FEATURES, SCTP_ASOC_ANY_STATE, (void *)asc, 0, sctp_asconf_iterator_end, NULL, 0); if (ret) { SCTP_PRINTF("Failed to initiate iterator for handle_addr_wq\n"); /* * Freeing if we are stopping or put back on the * addr_wq. */ if (SCTP_BASE_VAR(sctp_pcb_initialized) == 0) { sctp_asconf_iterator_end(asc, 0); } else { SCTP_WQ_ADDR_LOCK(); LIST_FOREACH(wi, &asc->list_of_work, sctp_nxt_addr) { LIST_INSERT_HEAD(&SCTP_BASE_INFO(addr_wq), wi, sctp_nxt_addr); } SCTP_WQ_ADDR_UNLOCK(); SCTP_FREE(asc, SCTP_M_ASC_IT); } } } } void sctp_timeout_handler(void *t) { struct sctp_inpcb *inp; struct sctp_tcb *stcb; struct sctp_nets *net; struct sctp_timer *tmr; struct mbuf *op_err; #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) struct socket *so; #endif int did_output; int type; tmr = (struct sctp_timer *)t; inp = (struct sctp_inpcb *)tmr->ep; stcb = (struct sctp_tcb *)tmr->tcb; net = (struct sctp_nets *)tmr->net; CURVNET_SET((struct vnet *)tmr->vnet); did_output = 1; #ifdef SCTP_AUDITING_ENABLED sctp_audit_log(0xF0, (uint8_t) tmr->type); sctp_auditing(3, inp, stcb, net); #endif /* sanity checks... */ if (tmr->self != (void *)tmr) { /* * SCTP_PRINTF("Stale SCTP timer fired (%p), ignoring...\n", * (void *)tmr); */ CURVNET_RESTORE(); return; } tmr->stopped_from = 0xa001; if (!SCTP_IS_TIMER_TYPE_VALID(tmr->type)) { /* * SCTP_PRINTF("SCTP timer fired with invalid type: 0x%x\n", * tmr->type); */ CURVNET_RESTORE(); return; } tmr->stopped_from = 0xa002; if ((tmr->type != SCTP_TIMER_TYPE_ADDR_WQ) && (inp == NULL)) { CURVNET_RESTORE(); return; } /* if this is an iterator timeout, get the struct and clear inp */ tmr->stopped_from = 0xa003; if (inp) { SCTP_INP_INCR_REF(inp); if ((inp->sctp_socket == NULL) && ((tmr->type != SCTP_TIMER_TYPE_INPKILL) && (tmr->type != SCTP_TIMER_TYPE_INIT) && (tmr->type != SCTP_TIMER_TYPE_SEND) && (tmr->type != SCTP_TIMER_TYPE_RECV) && (tmr->type != SCTP_TIMER_TYPE_HEARTBEAT) && (tmr->type != SCTP_TIMER_TYPE_SHUTDOWN) && (tmr->type != SCTP_TIMER_TYPE_SHUTDOWNACK) && (tmr->type != SCTP_TIMER_TYPE_SHUTDOWNGUARD) && (tmr->type != SCTP_TIMER_TYPE_ASOCKILL)) ) { SCTP_INP_DECR_REF(inp); CURVNET_RESTORE(); return; } } tmr->stopped_from = 0xa004; if (stcb) { atomic_add_int(&stcb->asoc.refcnt, 1); if (stcb->asoc.state == 0) { atomic_add_int(&stcb->asoc.refcnt, -1); if (inp) { SCTP_INP_DECR_REF(inp); } CURVNET_RESTORE(); return; } } type = tmr->type; tmr->stopped_from = 0xa005; SCTPDBG(SCTP_DEBUG_TIMER1, "Timer type %d goes off\n", type); if (!SCTP_OS_TIMER_ACTIVE(&tmr->timer)) { if (inp) { SCTP_INP_DECR_REF(inp); } if (stcb) { atomic_add_int(&stcb->asoc.refcnt, -1); } CURVNET_RESTORE(); return; } tmr->stopped_from = 0xa006; if (stcb) { SCTP_TCB_LOCK(stcb); atomic_add_int(&stcb->asoc.refcnt, -1); if ((type != SCTP_TIMER_TYPE_ASOCKILL) && ((stcb->asoc.state == 0) || (stcb->asoc.state & SCTP_STATE_ABOUT_TO_BE_FREED))) { SCTP_TCB_UNLOCK(stcb); if (inp) { SCTP_INP_DECR_REF(inp); } CURVNET_RESTORE(); return; } } /* record in stopped what t-o occurred */ tmr->stopped_from = type; /* mark as being serviced now */ if (SCTP_OS_TIMER_PENDING(&tmr->timer)) { /* * Callout has been rescheduled. */ goto get_out; } if (!SCTP_OS_TIMER_ACTIVE(&tmr->timer)) { /* * Not active, so no action. */ goto get_out; } SCTP_OS_TIMER_DEACTIVATE(&tmr->timer); /* call the handler for the appropriate timer type */ switch (type) { case SCTP_TIMER_TYPE_ZERO_COPY: if (inp == NULL) { break; } if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_ZERO_COPY_ACTIVE)) { SCTP_ZERO_COPY_EVENT(inp, inp->sctp_socket); } break; case SCTP_TIMER_TYPE_ZCOPY_SENDQ: if (inp == NULL) { break; } if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_ZERO_COPY_ACTIVE)) { SCTP_ZERO_COPY_SENDQ_EVENT(inp, inp->sctp_socket); } break; case SCTP_TIMER_TYPE_ADDR_WQ: sctp_handle_addr_wq(); break; case SCTP_TIMER_TYPE_SEND: if ((stcb == NULL) || (inp == NULL)) { break; } SCTP_STAT_INCR(sctps_timodata); stcb->asoc.timodata++; stcb->asoc.num_send_timers_up--; if (stcb->asoc.num_send_timers_up < 0) { stcb->asoc.num_send_timers_up = 0; } SCTP_TCB_LOCK_ASSERT(stcb); if (sctp_t3rxt_timer(inp, stcb, net)) { /* no need to unlock on tcb its gone */ goto out_decr; } SCTP_TCB_LOCK_ASSERT(stcb); #ifdef SCTP_AUDITING_ENABLED sctp_auditing(4, inp, stcb, net); #endif sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_T3, SCTP_SO_NOT_LOCKED); if ((stcb->asoc.num_send_timers_up == 0) && (stcb->asoc.sent_queue_cnt > 0)) { struct sctp_tmit_chunk *chk; /* * safeguard. If there on some on the sent queue * somewhere but no timers running something is * wrong... so we start a timer on the first chunk * on the send queue on whatever net it is sent to. */ chk = TAILQ_FIRST(&stcb->asoc.sent_queue); sctp_timer_start(SCTP_TIMER_TYPE_SEND, inp, stcb, chk->whoTo); } break; case SCTP_TIMER_TYPE_INIT: if ((stcb == NULL) || (inp == NULL)) { break; } SCTP_STAT_INCR(sctps_timoinit); stcb->asoc.timoinit++; if (sctp_t1init_timer(inp, stcb, net)) { /* no need to unlock on tcb its gone */ goto out_decr; } /* We do output but not here */ did_output = 0; break; case SCTP_TIMER_TYPE_RECV: if ((stcb == NULL) || (inp == NULL)) { break; } SCTP_STAT_INCR(sctps_timosack); stcb->asoc.timosack++; sctp_send_sack(stcb, SCTP_SO_NOT_LOCKED); #ifdef SCTP_AUDITING_ENABLED sctp_auditing(4, inp, stcb, net); #endif sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_SACK_TMR, SCTP_SO_NOT_LOCKED); break; case SCTP_TIMER_TYPE_SHUTDOWN: if ((stcb == NULL) || (inp == NULL)) { break; } if (sctp_shutdown_timer(inp, stcb, net)) { /* no need to unlock on tcb its gone */ goto out_decr; } SCTP_STAT_INCR(sctps_timoshutdown); stcb->asoc.timoshutdown++; #ifdef SCTP_AUDITING_ENABLED sctp_auditing(4, inp, stcb, net); #endif sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_SHUT_TMR, SCTP_SO_NOT_LOCKED); break; case SCTP_TIMER_TYPE_HEARTBEAT: if ((stcb == NULL) || (inp == NULL) || (net == NULL)) { break; } SCTP_STAT_INCR(sctps_timoheartbeat); stcb->asoc.timoheartbeat++; if (sctp_heartbeat_timer(inp, stcb, net)) { /* no need to unlock on tcb its gone */ goto out_decr; } #ifdef SCTP_AUDITING_ENABLED sctp_auditing(4, inp, stcb, net); #endif if (!(net->dest_state & SCTP_ADDR_NOHB)) { sctp_timer_start(SCTP_TIMER_TYPE_HEARTBEAT, inp, stcb, net); sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_HB_TMR, SCTP_SO_NOT_LOCKED); } break; case SCTP_TIMER_TYPE_COOKIE: if ((stcb == NULL) || (inp == NULL)) { break; } if (sctp_cookie_timer(inp, stcb, net)) { /* no need to unlock on tcb its gone */ goto out_decr; } SCTP_STAT_INCR(sctps_timocookie); stcb->asoc.timocookie++; #ifdef SCTP_AUDITING_ENABLED sctp_auditing(4, inp, stcb, net); #endif /* * We consider T3 and Cookie timer pretty much the same with * respect to where from in chunk_output. */ sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_T3, SCTP_SO_NOT_LOCKED); break; case SCTP_TIMER_TYPE_NEWCOOKIE: { struct timeval tv; int i, secret; if (inp == NULL) { break; } SCTP_STAT_INCR(sctps_timosecret); (void)SCTP_GETTIME_TIMEVAL(&tv); SCTP_INP_WLOCK(inp); inp->sctp_ep.time_of_secret_change = tv.tv_sec; inp->sctp_ep.last_secret_number = inp->sctp_ep.current_secret_number; inp->sctp_ep.current_secret_number++; if (inp->sctp_ep.current_secret_number >= SCTP_HOW_MANY_SECRETS) { inp->sctp_ep.current_secret_number = 0; } secret = (int)inp->sctp_ep.current_secret_number; for (i = 0; i < SCTP_NUMBER_OF_SECRETS; i++) { inp->sctp_ep.secret_key[secret][i] = sctp_select_initial_TSN(&inp->sctp_ep); } SCTP_INP_WUNLOCK(inp); sctp_timer_start(SCTP_TIMER_TYPE_NEWCOOKIE, inp, stcb, net); } did_output = 0; break; case SCTP_TIMER_TYPE_PATHMTURAISE: if ((stcb == NULL) || (inp == NULL)) { break; } SCTP_STAT_INCR(sctps_timopathmtu); sctp_pathmtu_timer(inp, stcb, net); did_output = 0; break; case SCTP_TIMER_TYPE_SHUTDOWNACK: if ((stcb == NULL) || (inp == NULL)) { break; } if (sctp_shutdownack_timer(inp, stcb, net)) { /* no need to unlock on tcb its gone */ goto out_decr; } SCTP_STAT_INCR(sctps_timoshutdownack); stcb->asoc.timoshutdownack++; #ifdef SCTP_AUDITING_ENABLED sctp_auditing(4, inp, stcb, net); #endif sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_SHUT_ACK_TMR, SCTP_SO_NOT_LOCKED); break; case SCTP_TIMER_TYPE_SHUTDOWNGUARD: if ((stcb == NULL) || (inp == NULL)) { break; } SCTP_STAT_INCR(sctps_timoshutdownguard); op_err = sctp_generate_cause(SCTP_BASE_SYSCTL(sctp_diag_info_code), "Shutdown guard timer expired"); sctp_abort_an_association(inp, stcb, op_err, SCTP_SO_NOT_LOCKED); /* no need to unlock on tcb its gone */ goto out_decr; case SCTP_TIMER_TYPE_STRRESET: if ((stcb == NULL) || (inp == NULL)) { break; } if (sctp_strreset_timer(inp, stcb, net)) { /* no need to unlock on tcb its gone */ goto out_decr; } SCTP_STAT_INCR(sctps_timostrmrst); sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_STRRST_TMR, SCTP_SO_NOT_LOCKED); break; case SCTP_TIMER_TYPE_ASCONF: if ((stcb == NULL) || (inp == NULL)) { break; } if (sctp_asconf_timer(inp, stcb, net)) { /* no need to unlock on tcb its gone */ goto out_decr; } SCTP_STAT_INCR(sctps_timoasconf); #ifdef SCTP_AUDITING_ENABLED sctp_auditing(4, inp, stcb, net); #endif sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_ASCONF_TMR, SCTP_SO_NOT_LOCKED); break; case SCTP_TIMER_TYPE_PRIM_DELETED: if ((stcb == NULL) || (inp == NULL)) { break; } sctp_delete_prim_timer(inp, stcb, net); SCTP_STAT_INCR(sctps_timodelprim); break; case SCTP_TIMER_TYPE_AUTOCLOSE: if ((stcb == NULL) || (inp == NULL)) { break; } SCTP_STAT_INCR(sctps_timoautoclose); sctp_autoclose_timer(inp, stcb, net); sctp_chunk_output(inp, stcb, SCTP_OUTPUT_FROM_AUTOCLOSE_TMR, SCTP_SO_NOT_LOCKED); did_output = 0; break; case SCTP_TIMER_TYPE_ASOCKILL: if ((stcb == NULL) || (inp == NULL)) { break; } SCTP_STAT_INCR(sctps_timoassockill); /* Can we free it yet? */ SCTP_INP_DECR_REF(inp); sctp_timer_stop(SCTP_TIMER_TYPE_ASOCKILL, inp, stcb, NULL, SCTP_FROM_SCTPUTIL + SCTP_LOC_1); #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) so = SCTP_INP_SO(inp); atomic_add_int(&stcb->asoc.refcnt, 1); SCTP_TCB_UNLOCK(stcb); SCTP_SOCKET_LOCK(so, 1); SCTP_TCB_LOCK(stcb); atomic_subtract_int(&stcb->asoc.refcnt, 1); #endif (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTPUTIL + SCTP_LOC_2); #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) SCTP_SOCKET_UNLOCK(so, 1); #endif /* * free asoc, always unlocks (or destroy's) so prevent * duplicate unlock or unlock of a free mtx :-0 */ stcb = NULL; goto out_no_decr; case SCTP_TIMER_TYPE_INPKILL: SCTP_STAT_INCR(sctps_timoinpkill); if (inp == NULL) { break; } /* * special case, take away our increment since WE are the * killer */ SCTP_INP_DECR_REF(inp); sctp_timer_stop(SCTP_TIMER_TYPE_INPKILL, inp, NULL, NULL, SCTP_FROM_SCTPUTIL + SCTP_LOC_3); sctp_inpcb_free(inp, SCTP_FREE_SHOULD_USE_ABORT, SCTP_CALLED_FROM_INPKILL_TIMER); inp = NULL; goto out_no_decr; default: SCTPDBG(SCTP_DEBUG_TIMER1, "sctp_timeout_handler:unknown timer %d\n", type); break; } #ifdef SCTP_AUDITING_ENABLED sctp_audit_log(0xF1, (uint8_t) type); if (inp) sctp_auditing(5, inp, stcb, net); #endif if ((did_output) && stcb) { /* * Now we need to clean up the control chunk chain if an * ECNE is on it. It must be marked as UNSENT again so next * call will continue to send it until such time that we get * a CWR, to remove it. It is, however, less likely that we * will find a ecn echo on the chain though. */ sctp_fix_ecn_echo(&stcb->asoc); } get_out: if (stcb) { SCTP_TCB_UNLOCK(stcb); } out_decr: if (inp) { SCTP_INP_DECR_REF(inp); } out_no_decr: SCTPDBG(SCTP_DEBUG_TIMER1, "Timer now complete (type = %d)\n", type); CURVNET_RESTORE(); } void sctp_timer_start(int t_type, struct sctp_inpcb *inp, struct sctp_tcb *stcb, struct sctp_nets *net) { uint32_t to_ticks; struct sctp_timer *tmr; if ((t_type != SCTP_TIMER_TYPE_ADDR_WQ) && (inp == NULL)) return; tmr = NULL; if (stcb) { SCTP_TCB_LOCK_ASSERT(stcb); } switch (t_type) { case SCTP_TIMER_TYPE_ZERO_COPY: tmr = &inp->sctp_ep.zero_copy_timer; to_ticks = SCTP_ZERO_COPY_TICK_DELAY; break; case SCTP_TIMER_TYPE_ZCOPY_SENDQ: tmr = &inp->sctp_ep.zero_copy_sendq_timer; to_ticks = SCTP_ZERO_COPY_SENDQ_TICK_DELAY; break; case SCTP_TIMER_TYPE_ADDR_WQ: /* Only 1 tick away :-) */ tmr = &SCTP_BASE_INFO(addr_wq_timer); to_ticks = SCTP_ADDRESS_TICK_DELAY; break; case SCTP_TIMER_TYPE_SEND: /* Here we use the RTO timer */ { int rto_val; if ((stcb == NULL) || (net == NULL)) { return; } tmr = &net->rxt_timer; if (net->RTO == 0) { rto_val = stcb->asoc.initial_rto; } else { rto_val = net->RTO; } to_ticks = MSEC_TO_TICKS(rto_val); } break; case SCTP_TIMER_TYPE_INIT: /* * Here we use the INIT timer default usually about 1 * minute. */ if ((stcb == NULL) || (net == NULL)) { return; } tmr = &net->rxt_timer; if (net->RTO == 0) { to_ticks = MSEC_TO_TICKS(stcb->asoc.initial_rto); } else { to_ticks = MSEC_TO_TICKS(net->RTO); } break; case SCTP_TIMER_TYPE_RECV: /* * Here we use the Delayed-Ack timer value from the inp * ususually about 200ms. */ if (stcb == NULL) { return; } tmr = &stcb->asoc.dack_timer; to_ticks = MSEC_TO_TICKS(stcb->asoc.delayed_ack); break; case SCTP_TIMER_TYPE_SHUTDOWN: /* Here we use the RTO of the destination. */ if ((stcb == NULL) || (net == NULL)) { return; } if (net->RTO == 0) { to_ticks = MSEC_TO_TICKS(stcb->asoc.initial_rto); } else { to_ticks = MSEC_TO_TICKS(net->RTO); } tmr = &net->rxt_timer; break; case SCTP_TIMER_TYPE_HEARTBEAT: /* * the net is used here so that we can add in the RTO. Even * though we use a different timer. We also add the HB timer * PLUS a random jitter. */ if ((stcb == NULL) || (net == NULL)) { return; } else { uint32_t rndval; uint32_t jitter; if ((net->dest_state & SCTP_ADDR_NOHB) && !(net->dest_state & SCTP_ADDR_UNCONFIRMED)) { return; } if (net->RTO == 0) { to_ticks = stcb->asoc.initial_rto; } else { to_ticks = net->RTO; } rndval = sctp_select_initial_TSN(&inp->sctp_ep); jitter = rndval % to_ticks; if (jitter >= (to_ticks >> 1)) { to_ticks = to_ticks + (jitter - (to_ticks >> 1)); } else { to_ticks = to_ticks - jitter; } if (!(net->dest_state & SCTP_ADDR_UNCONFIRMED) && !(net->dest_state & SCTP_ADDR_PF)) { to_ticks += net->heart_beat_delay; } /* * Now we must convert the to_ticks that are now in * ms to ticks. */ to_ticks = MSEC_TO_TICKS(to_ticks); tmr = &net->hb_timer; } break; case SCTP_TIMER_TYPE_COOKIE: /* * Here we can use the RTO timer from the network since one * RTT was compelete. If a retran happened then we will be * using the RTO initial value. */ if ((stcb == NULL) || (net == NULL)) { return; } if (net->RTO == 0) { to_ticks = MSEC_TO_TICKS(stcb->asoc.initial_rto); } else { to_ticks = MSEC_TO_TICKS(net->RTO); } tmr = &net->rxt_timer; break; case SCTP_TIMER_TYPE_NEWCOOKIE: /* * nothing needed but the endpoint here ususually about 60 * minutes. */ tmr = &inp->sctp_ep.signature_change; to_ticks = inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_SIGNATURE]; break; case SCTP_TIMER_TYPE_ASOCKILL: if (stcb == NULL) { return; } tmr = &stcb->asoc.strreset_timer; to_ticks = MSEC_TO_TICKS(SCTP_ASOC_KILL_TIMEOUT); break; case SCTP_TIMER_TYPE_INPKILL: /* * The inp is setup to die. We re-use the signature_chage * timer since that has stopped and we are in the GONE * state. */ tmr = &inp->sctp_ep.signature_change; to_ticks = MSEC_TO_TICKS(SCTP_INP_KILL_TIMEOUT); break; case SCTP_TIMER_TYPE_PATHMTURAISE: /* * Here we use the value found in the EP for PMTU ususually * about 10 minutes. */ if ((stcb == NULL) || (net == NULL)) { return; } if (net->dest_state & SCTP_ADDR_NO_PMTUD) { return; } to_ticks = inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_PMTU]; tmr = &net->pmtu_timer; break; case SCTP_TIMER_TYPE_SHUTDOWNACK: /* Here we use the RTO of the destination */ if ((stcb == NULL) || (net == NULL)) { return; } if (net->RTO == 0) { to_ticks = MSEC_TO_TICKS(stcb->asoc.initial_rto); } else { to_ticks = MSEC_TO_TICKS(net->RTO); } tmr = &net->rxt_timer; break; case SCTP_TIMER_TYPE_SHUTDOWNGUARD: /* * Here we use the endpoints shutdown guard timer usually * about 3 minutes. */ if (stcb == NULL) { return; } if (inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_MAXSHUTDOWN] == 0) { to_ticks = 5 * MSEC_TO_TICKS(stcb->asoc.maxrto); } else { to_ticks = inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_MAXSHUTDOWN]; } tmr = &stcb->asoc.shut_guard_timer; break; case SCTP_TIMER_TYPE_STRRESET: /* * Here the timer comes from the stcb but its value is from * the net's RTO. */ if ((stcb == NULL) || (net == NULL)) { return; } if (net->RTO == 0) { to_ticks = MSEC_TO_TICKS(stcb->asoc.initial_rto); } else { to_ticks = MSEC_TO_TICKS(net->RTO); } tmr = &stcb->asoc.strreset_timer; break; case SCTP_TIMER_TYPE_ASCONF: /* * Here the timer comes from the stcb but its value is from * the net's RTO. */ if ((stcb == NULL) || (net == NULL)) { return; } if (net->RTO == 0) { to_ticks = MSEC_TO_TICKS(stcb->asoc.initial_rto); } else { to_ticks = MSEC_TO_TICKS(net->RTO); } tmr = &stcb->asoc.asconf_timer; break; case SCTP_TIMER_TYPE_PRIM_DELETED: if ((stcb == NULL) || (net != NULL)) { return; } to_ticks = MSEC_TO_TICKS(stcb->asoc.initial_rto); tmr = &stcb->asoc.delete_prim_timer; break; case SCTP_TIMER_TYPE_AUTOCLOSE: if (stcb == NULL) { return; } if (stcb->asoc.sctp_autoclose_ticks == 0) { /* * Really an error since stcb is NOT set to * autoclose */ return; } to_ticks = stcb->asoc.sctp_autoclose_ticks; tmr = &stcb->asoc.autoclose_timer; break; default: SCTPDBG(SCTP_DEBUG_TIMER1, "%s: Unknown timer type %d\n", __func__, t_type); return; break; } if ((to_ticks <= 0) || (tmr == NULL)) { SCTPDBG(SCTP_DEBUG_TIMER1, "%s: %d:software error to_ticks:%d tmr:%p not set ??\n", __func__, t_type, to_ticks, (void *)tmr); return; } if (SCTP_OS_TIMER_PENDING(&tmr->timer)) { /* * we do NOT allow you to have it already running. if it is * we leave the current one up unchanged */ return; } /* At this point we can proceed */ if (t_type == SCTP_TIMER_TYPE_SEND) { stcb->asoc.num_send_timers_up++; } tmr->stopped_from = 0; tmr->type = t_type; tmr->ep = (void *)inp; tmr->tcb = (void *)stcb; tmr->net = (void *)net; tmr->self = (void *)tmr; tmr->vnet = (void *)curvnet; tmr->ticks = sctp_get_tick_count(); (void)SCTP_OS_TIMER_START(&tmr->timer, to_ticks, sctp_timeout_handler, tmr); return; } void sctp_timer_stop(int t_type, struct sctp_inpcb *inp, struct sctp_tcb *stcb, struct sctp_nets *net, uint32_t from) { struct sctp_timer *tmr; if ((t_type != SCTP_TIMER_TYPE_ADDR_WQ) && (inp == NULL)) return; tmr = NULL; if (stcb) { SCTP_TCB_LOCK_ASSERT(stcb); } switch (t_type) { case SCTP_TIMER_TYPE_ZERO_COPY: tmr = &inp->sctp_ep.zero_copy_timer; break; case SCTP_TIMER_TYPE_ZCOPY_SENDQ: tmr = &inp->sctp_ep.zero_copy_sendq_timer; break; case SCTP_TIMER_TYPE_ADDR_WQ: tmr = &SCTP_BASE_INFO(addr_wq_timer); break; case SCTP_TIMER_TYPE_SEND: if ((stcb == NULL) || (net == NULL)) { return; } tmr = &net->rxt_timer; break; case SCTP_TIMER_TYPE_INIT: if ((stcb == NULL) || (net == NULL)) { return; } tmr = &net->rxt_timer; break; case SCTP_TIMER_TYPE_RECV: if (stcb == NULL) { return; } tmr = &stcb->asoc.dack_timer; break; case SCTP_TIMER_TYPE_SHUTDOWN: if ((stcb == NULL) || (net == NULL)) { return; } tmr = &net->rxt_timer; break; case SCTP_TIMER_TYPE_HEARTBEAT: if ((stcb == NULL) || (net == NULL)) { return; } tmr = &net->hb_timer; break; case SCTP_TIMER_TYPE_COOKIE: if ((stcb == NULL) || (net == NULL)) { return; } tmr = &net->rxt_timer; break; case SCTP_TIMER_TYPE_NEWCOOKIE: /* nothing needed but the endpoint here */ tmr = &inp->sctp_ep.signature_change; /* * We re-use the newcookie timer for the INP kill timer. We * must assure that we do not kill it by accident. */ break; case SCTP_TIMER_TYPE_ASOCKILL: /* * Stop the asoc kill timer. */ if (stcb == NULL) { return; } tmr = &stcb->asoc.strreset_timer; break; case SCTP_TIMER_TYPE_INPKILL: /* * The inp is setup to die. We re-use the signature_chage * timer since that has stopped and we are in the GONE * state. */ tmr = &inp->sctp_ep.signature_change; break; case SCTP_TIMER_TYPE_PATHMTURAISE: if ((stcb == NULL) || (net == NULL)) { return; } tmr = &net->pmtu_timer; break; case SCTP_TIMER_TYPE_SHUTDOWNACK: if ((stcb == NULL) || (net == NULL)) { return; } tmr = &net->rxt_timer; break; case SCTP_TIMER_TYPE_SHUTDOWNGUARD: if (stcb == NULL) { return; } tmr = &stcb->asoc.shut_guard_timer; break; case SCTP_TIMER_TYPE_STRRESET: if (stcb == NULL) { return; } tmr = &stcb->asoc.strreset_timer; break; case SCTP_TIMER_TYPE_ASCONF: if (stcb == NULL) { return; } tmr = &stcb->asoc.asconf_timer; break; case SCTP_TIMER_TYPE_PRIM_DELETED: if (stcb == NULL) { return; } tmr = &stcb->asoc.delete_prim_timer; break; case SCTP_TIMER_TYPE_AUTOCLOSE: if (stcb == NULL) { return; } tmr = &stcb->asoc.autoclose_timer; break; default: SCTPDBG(SCTP_DEBUG_TIMER1, "%s: Unknown timer type %d\n", __func__, t_type); break; } if (tmr == NULL) { return; } if ((tmr->type != t_type) && tmr->type) { /* * Ok we have a timer that is under joint use. Cookie timer * per chance with the SEND timer. We therefore are NOT * running the timer that the caller wants stopped. So just * return. */ return; } if ((t_type == SCTP_TIMER_TYPE_SEND) && (stcb != NULL)) { stcb->asoc.num_send_timers_up--; if (stcb->asoc.num_send_timers_up < 0) { stcb->asoc.num_send_timers_up = 0; } } tmr->self = NULL; tmr->stopped_from = from; (void)SCTP_OS_TIMER_STOP(&tmr->timer); return; } uint32_t sctp_calculate_len(struct mbuf *m) { uint32_t tlen = 0; struct mbuf *at; at = m; while (at) { tlen += SCTP_BUF_LEN(at); at = SCTP_BUF_NEXT(at); } return (tlen); } void sctp_mtu_size_reset(struct sctp_inpcb *inp, struct sctp_association *asoc, uint32_t mtu) { /* * Reset the P-MTU size on this association, this involves changing * the asoc MTU, going through ANY chunk+overhead larger than mtu to * allow the DF flag to be cleared. */ struct sctp_tmit_chunk *chk; unsigned int eff_mtu, ovh; asoc->smallest_mtu = mtu; if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) { ovh = SCTP_MIN_OVERHEAD; } else { ovh = SCTP_MIN_V4_OVERHEAD; } eff_mtu = mtu - ovh; TAILQ_FOREACH(chk, &asoc->send_queue, sctp_next) { if (chk->send_size > eff_mtu) { chk->flags |= CHUNK_FLAGS_FRAGMENT_OK; } } TAILQ_FOREACH(chk, &asoc->sent_queue, sctp_next) { if (chk->send_size > eff_mtu) { chk->flags |= CHUNK_FLAGS_FRAGMENT_OK; } } } /* * given an association and starting time of the current RTT period return * RTO in number of msecs net should point to the current network */ uint32_t sctp_calculate_rto(struct sctp_tcb *stcb, struct sctp_association *asoc, struct sctp_nets *net, struct timeval *told, int safe, int rtt_from_sack) { /*- * given an association and the starting time of the current RTT * period (in value1/value2) return RTO in number of msecs. */ int32_t rtt; /* RTT in ms */ uint32_t new_rto; int first_measure = 0; struct timeval now, then, *old; /* Copy it out for sparc64 */ if (safe == sctp_align_unsafe_makecopy) { old = &then; memcpy(&then, told, sizeof(struct timeval)); } else if (safe == sctp_align_safe_nocopy) { old = told; } else { /* error */ SCTP_PRINTF("Huh, bad rto calc call\n"); return (0); } /************************/ /* 1. calculate new RTT */ /************************/ /* get the current time */ if (stcb->asoc.use_precise_time) { (void)SCTP_GETPTIME_TIMEVAL(&now); } else { (void)SCTP_GETTIME_TIMEVAL(&now); } timevalsub(&now, old); /* store the current RTT in us */ net->rtt = (uint64_t) 1000000 *(uint64_t) now.tv_sec + (uint64_t) now.tv_usec; /* compute rtt in ms */ rtt = (int32_t) (net->rtt / 1000); if ((asoc->cc_functions.sctp_rtt_calculated) && (rtt_from_sack == SCTP_RTT_FROM_DATA)) { /* * Tell the CC module that a new update has just occurred * from a sack */ (*asoc->cc_functions.sctp_rtt_calculated) (stcb, net, &now); } /* * Do we need to determine the lan? We do this only on sacks i.e. * RTT being determined from data not non-data (HB/INIT->INITACK). */ if ((rtt_from_sack == SCTP_RTT_FROM_DATA) && (net->lan_type == SCTP_LAN_UNKNOWN)) { if (net->rtt > SCTP_LOCAL_LAN_RTT) { net->lan_type = SCTP_LAN_INTERNET; } else { net->lan_type = SCTP_LAN_LOCAL; } } /***************************/ /* 2. update RTTVAR & SRTT */ /***************************/ /*- * Compute the scaled average lastsa and the * scaled variance lastsv as described in van Jacobson * Paper "Congestion Avoidance and Control", Annex A. * * (net->lastsa >> SCTP_RTT_SHIFT) is the srtt * (net->lastsa >> SCTP_RTT_VAR_SHIFT) is the rttvar */ if (net->RTO_measured) { rtt -= (net->lastsa >> SCTP_RTT_SHIFT); net->lastsa += rtt; if (rtt < 0) { rtt = -rtt; } rtt -= (net->lastsv >> SCTP_RTT_VAR_SHIFT); net->lastsv += rtt; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_RTTVAR_LOGGING_ENABLE) { rto_logging(net, SCTP_LOG_RTTVAR); } } else { /* First RTO measurment */ net->RTO_measured = 1; first_measure = 1; net->lastsa = rtt << SCTP_RTT_SHIFT; net->lastsv = (rtt / 2) << SCTP_RTT_VAR_SHIFT; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_RTTVAR_LOGGING_ENABLE) { rto_logging(net, SCTP_LOG_INITIAL_RTT); } } if (net->lastsv == 0) { net->lastsv = SCTP_CLOCK_GRANULARITY; } new_rto = (net->lastsa >> SCTP_RTT_SHIFT) + net->lastsv; if ((new_rto > SCTP_SAT_NETWORK_MIN) && (stcb->asoc.sat_network_lockout == 0)) { stcb->asoc.sat_network = 1; } else if ((!first_measure) && stcb->asoc.sat_network) { stcb->asoc.sat_network = 0; stcb->asoc.sat_network_lockout = 1; } /* bound it, per C6/C7 in Section 5.3.1 */ if (new_rto < stcb->asoc.minrto) { new_rto = stcb->asoc.minrto; } if (new_rto > stcb->asoc.maxrto) { new_rto = stcb->asoc.maxrto; } /* we are now returning the RTO */ return (new_rto); } /* * return a pointer to a contiguous piece of data from the given mbuf chain * starting at 'off' for 'len' bytes. If the desired piece spans more than * one mbuf, a copy is made at 'ptr'. caller must ensure that the buffer size * is >= 'len' returns NULL if there there isn't 'len' bytes in the chain. */ caddr_t sctp_m_getptr(struct mbuf *m, int off, int len, uint8_t * in_ptr) { uint32_t count; uint8_t *ptr; ptr = in_ptr; if ((off < 0) || (len <= 0)) return (NULL); /* find the desired start location */ while ((m != NULL) && (off > 0)) { if (off < SCTP_BUF_LEN(m)) break; off -= SCTP_BUF_LEN(m); m = SCTP_BUF_NEXT(m); } if (m == NULL) return (NULL); /* is the current mbuf large enough (eg. contiguous)? */ if ((SCTP_BUF_LEN(m) - off) >= len) { return (mtod(m, caddr_t)+off); } else { /* else, it spans more than one mbuf, so save a temp copy... */ while ((m != NULL) && (len > 0)) { count = min(SCTP_BUF_LEN(m) - off, len); bcopy(mtod(m, caddr_t)+off, ptr, count); len -= count; ptr += count; off = 0; m = SCTP_BUF_NEXT(m); } if ((m == NULL) && (len > 0)) return (NULL); else return ((caddr_t)in_ptr); } } struct sctp_paramhdr * sctp_get_next_param(struct mbuf *m, int offset, struct sctp_paramhdr *pull, int pull_limit) { /* This just provides a typed signature to Peter's Pull routine */ return ((struct sctp_paramhdr *)sctp_m_getptr(m, offset, pull_limit, (uint8_t *) pull)); } struct mbuf * sctp_add_pad_tombuf(struct mbuf *m, int padlen) { struct mbuf *m_last; caddr_t dp; if (padlen > 3) { return (NULL); } if (padlen <= M_TRAILINGSPACE(m)) { /* * The easy way. We hope the majority of the time we hit * here :) */ m_last = m; } else { /* Hard way we must grow the mbuf chain */ m_last = sctp_get_mbuf_for_msg(padlen, 0, M_NOWAIT, 1, MT_DATA); if (m_last == NULL) { return (NULL); } SCTP_BUF_LEN(m_last) = 0; SCTP_BUF_NEXT(m_last) = NULL; SCTP_BUF_NEXT(m) = m_last; } dp = mtod(m_last, caddr_t)+SCTP_BUF_LEN(m_last); SCTP_BUF_LEN(m_last) += padlen; memset(dp, 0, padlen); return (m_last); } struct mbuf * sctp_pad_lastmbuf(struct mbuf *m, int padval, struct mbuf *last_mbuf) { /* find the last mbuf in chain and pad it */ struct mbuf *m_at; if (last_mbuf != NULL) { return (sctp_add_pad_tombuf(last_mbuf, padval)); } else { for (m_at = m; m_at; m_at = SCTP_BUF_NEXT(m_at)) { if (SCTP_BUF_NEXT(m_at) == NULL) { return (sctp_add_pad_tombuf(m_at, padval)); } } } return (NULL); } static void sctp_notify_assoc_change(uint16_t state, struct sctp_tcb *stcb, uint16_t error, struct sctp_abort_chunk *abort, uint8_t from_peer, int so_locked #if !defined(__APPLE__) && !defined(SCTP_SO_LOCK_TESTING) SCTP_UNUSED #endif ) { struct mbuf *m_notify; struct sctp_assoc_change *sac; struct sctp_queued_to_read *control; unsigned int notif_len; uint16_t abort_len; unsigned int i; #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) struct socket *so; #endif if (stcb == NULL) { return; } if (sctp_stcb_is_feature_on(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_RECVASSOCEVNT)) { notif_len = (unsigned int)sizeof(struct sctp_assoc_change); if (abort != NULL) { abort_len = ntohs(abort->ch.chunk_length); } else { abort_len = 0; } if ((state == SCTP_COMM_UP) || (state == SCTP_RESTART)) { notif_len += SCTP_ASSOC_SUPPORTS_MAX; } else if ((state == SCTP_COMM_LOST) || (state == SCTP_CANT_STR_ASSOC)) { notif_len += abort_len; } m_notify = sctp_get_mbuf_for_msg(notif_len, 0, M_NOWAIT, 1, MT_DATA); if (m_notify == NULL) { /* Retry with smaller value. */ notif_len = (unsigned int)sizeof(struct sctp_assoc_change); m_notify = sctp_get_mbuf_for_msg(notif_len, 0, M_NOWAIT, 1, MT_DATA); if (m_notify == NULL) { goto set_error; } } SCTP_BUF_NEXT(m_notify) = NULL; sac = mtod(m_notify, struct sctp_assoc_change *); memset(sac, 0, notif_len); sac->sac_type = SCTP_ASSOC_CHANGE; sac->sac_flags = 0; sac->sac_length = sizeof(struct sctp_assoc_change); sac->sac_state = state; sac->sac_error = error; /* XXX verify these stream counts */ sac->sac_outbound_streams = stcb->asoc.streamoutcnt; sac->sac_inbound_streams = stcb->asoc.streamincnt; sac->sac_assoc_id = sctp_get_associd(stcb); if (notif_len > sizeof(struct sctp_assoc_change)) { if ((state == SCTP_COMM_UP) || (state == SCTP_RESTART)) { i = 0; if (stcb->asoc.prsctp_supported == 1) { sac->sac_info[i++] = SCTP_ASSOC_SUPPORTS_PR; } if (stcb->asoc.auth_supported == 1) { sac->sac_info[i++] = SCTP_ASSOC_SUPPORTS_AUTH; } if (stcb->asoc.asconf_supported == 1) { sac->sac_info[i++] = SCTP_ASSOC_SUPPORTS_ASCONF; } if (stcb->asoc.idata_supported == 1) { sac->sac_info[i++] = SCTP_ASSOC_SUPPORTS_INTERLEAVING; } sac->sac_info[i++] = SCTP_ASSOC_SUPPORTS_MULTIBUF; if (stcb->asoc.reconfig_supported == 1) { sac->sac_info[i++] = SCTP_ASSOC_SUPPORTS_RE_CONFIG; } sac->sac_length += i; } else if ((state == SCTP_COMM_LOST) || (state == SCTP_CANT_STR_ASSOC)) { memcpy(sac->sac_info, abort, abort_len); sac->sac_length += abort_len; } } SCTP_BUF_LEN(m_notify) = sac->sac_length; control = sctp_build_readq_entry(stcb, stcb->asoc.primary_destination, 0, 0, stcb->asoc.context, 0, 0, 0, m_notify); if (control != NULL) { control->length = SCTP_BUF_LEN(m_notify); /* not that we need this */ control->tail_mbuf = m_notify; control->spec_flags = M_NOTIFICATION; sctp_add_to_readq(stcb->sctp_ep, stcb, control, &stcb->sctp_socket->so_rcv, 1, SCTP_READ_LOCK_NOT_HELD, so_locked); } else { sctp_m_freem(m_notify); } } /* * For 1-to-1 style sockets, we send up and error when an ABORT * comes in. */ set_error: if (((stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) && ((state == SCTP_COMM_LOST) || (state == SCTP_CANT_STR_ASSOC))) { SOCK_LOCK(stcb->sctp_socket); if (from_peer) { if (SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_COOKIE_WAIT) { SCTP_LTRACE_ERR_RET(NULL, stcb, NULL, SCTP_FROM_SCTPUTIL, ECONNREFUSED); stcb->sctp_socket->so_error = ECONNREFUSED; } else { SCTP_LTRACE_ERR_RET(NULL, stcb, NULL, SCTP_FROM_SCTPUTIL, ECONNRESET); stcb->sctp_socket->so_error = ECONNRESET; } } else { if ((SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_COOKIE_WAIT) || (SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_COOKIE_ECHOED)) { SCTP_LTRACE_ERR_RET(NULL, stcb, NULL, SCTP_FROM_SCTPUTIL, ETIMEDOUT); stcb->sctp_socket->so_error = ETIMEDOUT; } else { SCTP_LTRACE_ERR_RET(NULL, stcb, NULL, SCTP_FROM_SCTPUTIL, ECONNABORTED); stcb->sctp_socket->so_error = ECONNABORTED; } } } /* Wake ANY sleepers */ #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) so = SCTP_INP_SO(stcb->sctp_ep); if (!so_locked) { atomic_add_int(&stcb->asoc.refcnt, 1); SCTP_TCB_UNLOCK(stcb); SCTP_SOCKET_LOCK(so, 1); SCTP_TCB_LOCK(stcb); atomic_subtract_int(&stcb->asoc.refcnt, 1); if (stcb->asoc.state & SCTP_STATE_CLOSED_SOCKET) { SCTP_SOCKET_UNLOCK(so, 1); return; } } #endif if (((stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) && ((state == SCTP_COMM_LOST) || (state == SCTP_CANT_STR_ASSOC))) { socantrcvmore_locked(stcb->sctp_socket); } sorwakeup(stcb->sctp_socket); sowwakeup(stcb->sctp_socket); #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) if (!so_locked) { SCTP_SOCKET_UNLOCK(so, 1); } #endif } static void sctp_notify_peer_addr_change(struct sctp_tcb *stcb, uint32_t state, struct sockaddr *sa, uint32_t error, int so_locked #if !defined(__APPLE__) && !defined(SCTP_SO_LOCK_TESTING) SCTP_UNUSED #endif ) { struct mbuf *m_notify; struct sctp_paddr_change *spc; struct sctp_queued_to_read *control; if ((stcb == NULL) || sctp_stcb_is_feature_off(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_RECVPADDREVNT)) { /* event not enabled */ return; } m_notify = sctp_get_mbuf_for_msg(sizeof(struct sctp_paddr_change), 0, M_NOWAIT, 1, MT_DATA); if (m_notify == NULL) return; SCTP_BUF_LEN(m_notify) = 0; spc = mtod(m_notify, struct sctp_paddr_change *); memset(spc, 0, sizeof(struct sctp_paddr_change)); spc->spc_type = SCTP_PEER_ADDR_CHANGE; spc->spc_flags = 0; spc->spc_length = sizeof(struct sctp_paddr_change); switch (sa->sa_family) { #ifdef INET case AF_INET: #ifdef INET6 if (sctp_is_feature_on(stcb->sctp_ep, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4)) { in6_sin_2_v4mapsin6((struct sockaddr_in *)sa, (struct sockaddr_in6 *)&spc->spc_aaddr); } else { memcpy(&spc->spc_aaddr, sa, sizeof(struct sockaddr_in)); } #else memcpy(&spc->spc_aaddr, sa, sizeof(struct sockaddr_in)); #endif break; #endif #ifdef INET6 case AF_INET6: { struct sockaddr_in6 *sin6; memcpy(&spc->spc_aaddr, sa, sizeof(struct sockaddr_in6)); sin6 = (struct sockaddr_in6 *)&spc->spc_aaddr; if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr)) { if (sin6->sin6_scope_id == 0) { /* recover scope_id for user */ (void)sa6_recoverscope(sin6); } else { /* clear embedded scope_id for user */ in6_clearscope(&sin6->sin6_addr); } } break; } #endif default: /* TSNH */ break; } spc->spc_state = state; spc->spc_error = error; spc->spc_assoc_id = sctp_get_associd(stcb); SCTP_BUF_LEN(m_notify) = sizeof(struct sctp_paddr_change); SCTP_BUF_NEXT(m_notify) = NULL; /* append to socket */ control = sctp_build_readq_entry(stcb, stcb->asoc.primary_destination, 0, 0, stcb->asoc.context, 0, 0, 0, m_notify); if (control == NULL) { /* no memory */ sctp_m_freem(m_notify); return; } control->length = SCTP_BUF_LEN(m_notify); control->spec_flags = M_NOTIFICATION; /* not that we need this */ control->tail_mbuf = m_notify; sctp_add_to_readq(stcb->sctp_ep, stcb, control, &stcb->sctp_socket->so_rcv, 1, SCTP_READ_LOCK_NOT_HELD, so_locked); } static void sctp_notify_send_failed(struct sctp_tcb *stcb, uint8_t sent, uint32_t error, struct sctp_tmit_chunk *chk, int so_locked #if !defined(__APPLE__) && !defined(SCTP_SO_LOCK_TESTING) SCTP_UNUSED #endif ) { struct mbuf *m_notify; struct sctp_send_failed *ssf; struct sctp_send_failed_event *ssfe; struct sctp_queued_to_read *control; int length; if ((stcb == NULL) || (sctp_stcb_is_feature_off(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_RECVSENDFAILEVNT) && sctp_stcb_is_feature_off(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_RECVNSENDFAILEVNT))) { /* event not enabled */ return; } if (sctp_stcb_is_feature_on(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_RECVNSENDFAILEVNT)) { length = sizeof(struct sctp_send_failed_event); } else { length = sizeof(struct sctp_send_failed); } m_notify = sctp_get_mbuf_for_msg(length, 0, M_NOWAIT, 1, MT_DATA); if (m_notify == NULL) /* no space left */ return; SCTP_BUF_LEN(m_notify) = 0; if (sctp_stcb_is_feature_on(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_RECVNSENDFAILEVNT)) { ssfe = mtod(m_notify, struct sctp_send_failed_event *); memset(ssfe, 0, length); ssfe->ssfe_type = SCTP_SEND_FAILED_EVENT; if (sent) { ssfe->ssfe_flags = SCTP_DATA_SENT; } else { ssfe->ssfe_flags = SCTP_DATA_UNSENT; } length += chk->send_size; length -= sizeof(struct sctp_data_chunk); ssfe->ssfe_length = length; ssfe->ssfe_error = error; /* not exactly what the user sent in, but should be close :) */ ssfe->ssfe_info.snd_sid = chk->rec.data.stream_number; ssfe->ssfe_info.snd_flags = chk->rec.data.rcv_flags; ssfe->ssfe_info.snd_ppid = chk->rec.data.payloadtype; ssfe->ssfe_info.snd_context = chk->rec.data.context; ssfe->ssfe_info.snd_assoc_id = sctp_get_associd(stcb); ssfe->ssfe_assoc_id = sctp_get_associd(stcb); SCTP_BUF_LEN(m_notify) = sizeof(struct sctp_send_failed_event); } else { ssf = mtod(m_notify, struct sctp_send_failed *); memset(ssf, 0, length); ssf->ssf_type = SCTP_SEND_FAILED; if (sent) { ssf->ssf_flags = SCTP_DATA_SENT; } else { ssf->ssf_flags = SCTP_DATA_UNSENT; } length += chk->send_size; length -= sizeof(struct sctp_data_chunk); ssf->ssf_length = length; ssf->ssf_error = error; /* not exactly what the user sent in, but should be close :) */ bzero(&ssf->ssf_info, sizeof(ssf->ssf_info)); ssf->ssf_info.sinfo_stream = chk->rec.data.stream_number; ssf->ssf_info.sinfo_ssn = chk->rec.data.stream_seq; ssf->ssf_info.sinfo_flags = chk->rec.data.rcv_flags; ssf->ssf_info.sinfo_ppid = chk->rec.data.payloadtype; ssf->ssf_info.sinfo_context = chk->rec.data.context; ssf->ssf_info.sinfo_assoc_id = sctp_get_associd(stcb); ssf->ssf_assoc_id = sctp_get_associd(stcb); SCTP_BUF_LEN(m_notify) = sizeof(struct sctp_send_failed); } if (chk->data) { /* * trim off the sctp chunk header(it should be there) */ if (chk->send_size >= sizeof(struct sctp_data_chunk)) { m_adj(chk->data, sizeof(struct sctp_data_chunk)); sctp_mbuf_crush(chk->data); chk->send_size -= sizeof(struct sctp_data_chunk); } } SCTP_BUF_NEXT(m_notify) = chk->data; /* Steal off the mbuf */ chk->data = NULL; /* * For this case, we check the actual socket buffer, since the assoc * is going away we don't want to overfill the socket buffer for a * non-reader */ if (sctp_sbspace_failedmsgs(&stcb->sctp_socket->so_rcv) < SCTP_BUF_LEN(m_notify)) { sctp_m_freem(m_notify); return; } /* append to socket */ control = sctp_build_readq_entry(stcb, stcb->asoc.primary_destination, 0, 0, stcb->asoc.context, 0, 0, 0, m_notify); if (control == NULL) { /* no memory */ sctp_m_freem(m_notify); return; } control->spec_flags = M_NOTIFICATION; sctp_add_to_readq(stcb->sctp_ep, stcb, control, &stcb->sctp_socket->so_rcv, 1, SCTP_READ_LOCK_NOT_HELD, so_locked); } static void sctp_notify_send_failed2(struct sctp_tcb *stcb, uint32_t error, struct sctp_stream_queue_pending *sp, int so_locked #if !defined(__APPLE__) && !defined(SCTP_SO_LOCK_TESTING) SCTP_UNUSED #endif ) { struct mbuf *m_notify; struct sctp_send_failed *ssf; struct sctp_send_failed_event *ssfe; struct sctp_queued_to_read *control; int length; if ((stcb == NULL) || (sctp_stcb_is_feature_off(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_RECVSENDFAILEVNT) && sctp_stcb_is_feature_off(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_RECVNSENDFAILEVNT))) { /* event not enabled */ return; } if (sctp_stcb_is_feature_on(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_RECVNSENDFAILEVNT)) { length = sizeof(struct sctp_send_failed_event); } else { length = sizeof(struct sctp_send_failed); } m_notify = sctp_get_mbuf_for_msg(length, 0, M_NOWAIT, 1, MT_DATA); if (m_notify == NULL) { /* no space left */ return; } SCTP_BUF_LEN(m_notify) = 0; if (sctp_stcb_is_feature_on(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_RECVNSENDFAILEVNT)) { ssfe = mtod(m_notify, struct sctp_send_failed_event *); memset(ssfe, 0, length); ssfe->ssfe_type = SCTP_SEND_FAILED_EVENT; ssfe->ssfe_flags = SCTP_DATA_UNSENT; length += sp->length; ssfe->ssfe_length = length; ssfe->ssfe_error = error; /* not exactly what the user sent in, but should be close :) */ ssfe->ssfe_info.snd_sid = sp->stream; if (sp->some_taken) { ssfe->ssfe_info.snd_flags = SCTP_DATA_LAST_FRAG; } else { ssfe->ssfe_info.snd_flags = SCTP_DATA_NOT_FRAG; } ssfe->ssfe_info.snd_ppid = sp->ppid; ssfe->ssfe_info.snd_context = sp->context; ssfe->ssfe_info.snd_assoc_id = sctp_get_associd(stcb); ssfe->ssfe_assoc_id = sctp_get_associd(stcb); SCTP_BUF_LEN(m_notify) = sizeof(struct sctp_send_failed_event); } else { ssf = mtod(m_notify, struct sctp_send_failed *); memset(ssf, 0, length); ssf->ssf_type = SCTP_SEND_FAILED; ssf->ssf_flags = SCTP_DATA_UNSENT; length += sp->length; ssf->ssf_length = length; ssf->ssf_error = error; /* not exactly what the user sent in, but should be close :) */ ssf->ssf_info.sinfo_stream = sp->stream; ssf->ssf_info.sinfo_ssn = 0; if (sp->some_taken) { ssf->ssf_info.sinfo_flags = SCTP_DATA_LAST_FRAG; } else { ssf->ssf_info.sinfo_flags = SCTP_DATA_NOT_FRAG; } ssf->ssf_info.sinfo_ppid = sp->ppid; ssf->ssf_info.sinfo_context = sp->context; ssf->ssf_info.sinfo_assoc_id = sctp_get_associd(stcb); ssf->ssf_assoc_id = sctp_get_associd(stcb); SCTP_BUF_LEN(m_notify) = sizeof(struct sctp_send_failed); } SCTP_BUF_NEXT(m_notify) = sp->data; /* Steal off the mbuf */ sp->data = NULL; /* * For this case, we check the actual socket buffer, since the assoc * is going away we don't want to overfill the socket buffer for a * non-reader */ if (sctp_sbspace_failedmsgs(&stcb->sctp_socket->so_rcv) < SCTP_BUF_LEN(m_notify)) { sctp_m_freem(m_notify); return; } /* append to socket */ control = sctp_build_readq_entry(stcb, stcb->asoc.primary_destination, 0, 0, stcb->asoc.context, 0, 0, 0, m_notify); if (control == NULL) { /* no memory */ sctp_m_freem(m_notify); return; } control->spec_flags = M_NOTIFICATION; sctp_add_to_readq(stcb->sctp_ep, stcb, control, &stcb->sctp_socket->so_rcv, 1, SCTP_READ_LOCK_NOT_HELD, so_locked); } static void sctp_notify_adaptation_layer(struct sctp_tcb *stcb) { struct mbuf *m_notify; struct sctp_adaptation_event *sai; struct sctp_queued_to_read *control; if ((stcb == NULL) || sctp_stcb_is_feature_off(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_ADAPTATIONEVNT)) { /* event not enabled */ return; } m_notify = sctp_get_mbuf_for_msg(sizeof(struct sctp_adaption_event), 0, M_NOWAIT, 1, MT_DATA); if (m_notify == NULL) /* no space left */ return; SCTP_BUF_LEN(m_notify) = 0; sai = mtod(m_notify, struct sctp_adaptation_event *); memset(sai, 0, sizeof(struct sctp_adaptation_event)); sai->sai_type = SCTP_ADAPTATION_INDICATION; sai->sai_flags = 0; sai->sai_length = sizeof(struct sctp_adaptation_event); sai->sai_adaptation_ind = stcb->asoc.peers_adaptation; sai->sai_assoc_id = sctp_get_associd(stcb); SCTP_BUF_LEN(m_notify) = sizeof(struct sctp_adaptation_event); SCTP_BUF_NEXT(m_notify) = NULL; /* append to socket */ control = sctp_build_readq_entry(stcb, stcb->asoc.primary_destination, 0, 0, stcb->asoc.context, 0, 0, 0, m_notify); if (control == NULL) { /* no memory */ sctp_m_freem(m_notify); return; } control->length = SCTP_BUF_LEN(m_notify); control->spec_flags = M_NOTIFICATION; /* not that we need this */ control->tail_mbuf = m_notify; sctp_add_to_readq(stcb->sctp_ep, stcb, control, &stcb->sctp_socket->so_rcv, 1, SCTP_READ_LOCK_NOT_HELD, SCTP_SO_NOT_LOCKED); } /* This always must be called with the read-queue LOCKED in the INP */ static void sctp_notify_partial_delivery_indication(struct sctp_tcb *stcb, uint32_t error, uint32_t val, int so_locked #if !defined(__APPLE__) && !defined(SCTP_SO_LOCK_TESTING) SCTP_UNUSED #endif ) { struct mbuf *m_notify; struct sctp_pdapi_event *pdapi; struct sctp_queued_to_read *control; struct sockbuf *sb; if ((stcb == NULL) || sctp_stcb_is_feature_off(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_PDAPIEVNT)) { /* event not enabled */ return; } if (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_SOCKET_CANT_READ) { return; } m_notify = sctp_get_mbuf_for_msg(sizeof(struct sctp_pdapi_event), 0, M_NOWAIT, 1, MT_DATA); if (m_notify == NULL) /* no space left */ return; SCTP_BUF_LEN(m_notify) = 0; pdapi = mtod(m_notify, struct sctp_pdapi_event *); memset(pdapi, 0, sizeof(struct sctp_pdapi_event)); pdapi->pdapi_type = SCTP_PARTIAL_DELIVERY_EVENT; pdapi->pdapi_flags = 0; pdapi->pdapi_length = sizeof(struct sctp_pdapi_event); pdapi->pdapi_indication = error; pdapi->pdapi_stream = (val >> 16); pdapi->pdapi_seq = (val & 0x0000ffff); pdapi->pdapi_assoc_id = sctp_get_associd(stcb); SCTP_BUF_LEN(m_notify) = sizeof(struct sctp_pdapi_event); SCTP_BUF_NEXT(m_notify) = NULL; control = sctp_build_readq_entry(stcb, stcb->asoc.primary_destination, 0, 0, stcb->asoc.context, 0, 0, 0, m_notify); if (control == NULL) { /* no memory */ sctp_m_freem(m_notify); return; } control->spec_flags = M_NOTIFICATION; control->length = SCTP_BUF_LEN(m_notify); /* not that we need this */ control->tail_mbuf = m_notify; control->held_length = 0; control->length = 0; sb = &stcb->sctp_socket->so_rcv; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_SB_LOGGING_ENABLE) { sctp_sblog(sb, control->do_not_ref_stcb ? NULL : stcb, SCTP_LOG_SBALLOC, SCTP_BUF_LEN(m_notify)); } sctp_sballoc(stcb, sb, m_notify); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_SB_LOGGING_ENABLE) { sctp_sblog(sb, control->do_not_ref_stcb ? NULL : stcb, SCTP_LOG_SBRESULT, 0); } atomic_add_int(&control->length, SCTP_BUF_LEN(m_notify)); control->end_added = 1; if (stcb->asoc.control_pdapi) TAILQ_INSERT_AFTER(&stcb->sctp_ep->read_queue, stcb->asoc.control_pdapi, control, next); else { /* we really should not see this case */ TAILQ_INSERT_TAIL(&stcb->sctp_ep->read_queue, control, next); } if (stcb->sctp_ep && stcb->sctp_socket) { /* This should always be the case */ #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) struct socket *so; so = SCTP_INP_SO(stcb->sctp_ep); if (!so_locked) { atomic_add_int(&stcb->asoc.refcnt, 1); SCTP_TCB_UNLOCK(stcb); SCTP_SOCKET_LOCK(so, 1); SCTP_TCB_LOCK(stcb); atomic_subtract_int(&stcb->asoc.refcnt, 1); if (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) { SCTP_SOCKET_UNLOCK(so, 1); return; } } #endif sctp_sorwakeup(stcb->sctp_ep, stcb->sctp_socket); #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) if (!so_locked) { SCTP_SOCKET_UNLOCK(so, 1); } #endif } } static void sctp_notify_shutdown_event(struct sctp_tcb *stcb) { struct mbuf *m_notify; struct sctp_shutdown_event *sse; struct sctp_queued_to_read *control; /* * For TCP model AND UDP connected sockets we will send an error up * when an SHUTDOWN completes */ if ((stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) { /* mark socket closed for read/write and wakeup! */ #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) struct socket *so; so = SCTP_INP_SO(stcb->sctp_ep); atomic_add_int(&stcb->asoc.refcnt, 1); SCTP_TCB_UNLOCK(stcb); SCTP_SOCKET_LOCK(so, 1); SCTP_TCB_LOCK(stcb); atomic_subtract_int(&stcb->asoc.refcnt, 1); if (stcb->asoc.state & SCTP_STATE_CLOSED_SOCKET) { SCTP_SOCKET_UNLOCK(so, 1); return; } #endif socantsendmore(stcb->sctp_socket); #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) SCTP_SOCKET_UNLOCK(so, 1); #endif } if (sctp_stcb_is_feature_off(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_RECVSHUTDOWNEVNT)) { /* event not enabled */ return; } m_notify = sctp_get_mbuf_for_msg(sizeof(struct sctp_shutdown_event), 0, M_NOWAIT, 1, MT_DATA); if (m_notify == NULL) /* no space left */ return; sse = mtod(m_notify, struct sctp_shutdown_event *); memset(sse, 0, sizeof(struct sctp_shutdown_event)); sse->sse_type = SCTP_SHUTDOWN_EVENT; sse->sse_flags = 0; sse->sse_length = sizeof(struct sctp_shutdown_event); sse->sse_assoc_id = sctp_get_associd(stcb); SCTP_BUF_LEN(m_notify) = sizeof(struct sctp_shutdown_event); SCTP_BUF_NEXT(m_notify) = NULL; /* append to socket */ control = sctp_build_readq_entry(stcb, stcb->asoc.primary_destination, 0, 0, stcb->asoc.context, 0, 0, 0, m_notify); if (control == NULL) { /* no memory */ sctp_m_freem(m_notify); return; } control->spec_flags = M_NOTIFICATION; control->length = SCTP_BUF_LEN(m_notify); /* not that we need this */ control->tail_mbuf = m_notify; sctp_add_to_readq(stcb->sctp_ep, stcb, control, &stcb->sctp_socket->so_rcv, 1, SCTP_READ_LOCK_NOT_HELD, SCTP_SO_NOT_LOCKED); } static void sctp_notify_sender_dry_event(struct sctp_tcb *stcb, int so_locked #if !defined(__APPLE__) && !defined(SCTP_SO_LOCK_TESTING) SCTP_UNUSED #endif ) { struct mbuf *m_notify; struct sctp_sender_dry_event *event; struct sctp_queued_to_read *control; if ((stcb == NULL) || sctp_stcb_is_feature_off(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_DRYEVNT)) { /* event not enabled */ return; } m_notify = sctp_get_mbuf_for_msg(sizeof(struct sctp_sender_dry_event), 0, M_NOWAIT, 1, MT_DATA); if (m_notify == NULL) { /* no space left */ return; } SCTP_BUF_LEN(m_notify) = 0; event = mtod(m_notify, struct sctp_sender_dry_event *); memset(event, 0, sizeof(struct sctp_sender_dry_event)); event->sender_dry_type = SCTP_SENDER_DRY_EVENT; event->sender_dry_flags = 0; event->sender_dry_length = sizeof(struct sctp_sender_dry_event); event->sender_dry_assoc_id = sctp_get_associd(stcb); SCTP_BUF_LEN(m_notify) = sizeof(struct sctp_sender_dry_event); SCTP_BUF_NEXT(m_notify) = NULL; /* append to socket */ control = sctp_build_readq_entry(stcb, stcb->asoc.primary_destination, 0, 0, stcb->asoc.context, 0, 0, 0, m_notify); if (control == NULL) { /* no memory */ sctp_m_freem(m_notify); return; } control->length = SCTP_BUF_LEN(m_notify); control->spec_flags = M_NOTIFICATION; /* not that we need this */ control->tail_mbuf = m_notify; sctp_add_to_readq(stcb->sctp_ep, stcb, control, &stcb->sctp_socket->so_rcv, 1, SCTP_READ_LOCK_NOT_HELD, so_locked); } void sctp_notify_stream_reset_add(struct sctp_tcb *stcb, uint16_t numberin, uint16_t numberout, int flag) { struct mbuf *m_notify; struct sctp_queued_to_read *control; struct sctp_stream_change_event *stradd; if ((stcb == NULL) || (sctp_stcb_is_feature_off(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_STREAM_CHANGEEVNT))) { /* event not enabled */ return; } if ((stcb->asoc.peer_req_out) && flag) { /* Peer made the request, don't tell the local user */ stcb->asoc.peer_req_out = 0; return; } stcb->asoc.peer_req_out = 0; m_notify = sctp_get_mbuf_for_msg(sizeof(struct sctp_stream_change_event), 0, M_NOWAIT, 1, MT_DATA); if (m_notify == NULL) /* no space left */ return; SCTP_BUF_LEN(m_notify) = 0; stradd = mtod(m_notify, struct sctp_stream_change_event *); memset(stradd, 0, sizeof(struct sctp_stream_change_event)); stradd->strchange_type = SCTP_STREAM_CHANGE_EVENT; stradd->strchange_flags = flag; stradd->strchange_length = sizeof(struct sctp_stream_change_event); stradd->strchange_assoc_id = sctp_get_associd(stcb); stradd->strchange_instrms = numberin; stradd->strchange_outstrms = numberout; SCTP_BUF_LEN(m_notify) = sizeof(struct sctp_stream_change_event); SCTP_BUF_NEXT(m_notify) = NULL; if (sctp_sbspace(&stcb->asoc, &stcb->sctp_socket->so_rcv) < SCTP_BUF_LEN(m_notify)) { /* no space */ sctp_m_freem(m_notify); return; } /* append to socket */ control = sctp_build_readq_entry(stcb, stcb->asoc.primary_destination, 0, 0, stcb->asoc.context, 0, 0, 0, m_notify); if (control == NULL) { /* no memory */ sctp_m_freem(m_notify); return; } control->spec_flags = M_NOTIFICATION; control->length = SCTP_BUF_LEN(m_notify); /* not that we need this */ control->tail_mbuf = m_notify; sctp_add_to_readq(stcb->sctp_ep, stcb, control, &stcb->sctp_socket->so_rcv, 1, SCTP_READ_LOCK_NOT_HELD, SCTP_SO_NOT_LOCKED); } void sctp_notify_stream_reset_tsn(struct sctp_tcb *stcb, uint32_t sending_tsn, uint32_t recv_tsn, int flag) { struct mbuf *m_notify; struct sctp_queued_to_read *control; struct sctp_assoc_reset_event *strasoc; if ((stcb == NULL) || (sctp_stcb_is_feature_off(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_ASSOC_RESETEVNT))) { /* event not enabled */ return; } m_notify = sctp_get_mbuf_for_msg(sizeof(struct sctp_assoc_reset_event), 0, M_NOWAIT, 1, MT_DATA); if (m_notify == NULL) /* no space left */ return; SCTP_BUF_LEN(m_notify) = 0; strasoc = mtod(m_notify, struct sctp_assoc_reset_event *); memset(strasoc, 0, sizeof(struct sctp_assoc_reset_event)); strasoc->assocreset_type = SCTP_ASSOC_RESET_EVENT; strasoc->assocreset_flags = flag; strasoc->assocreset_length = sizeof(struct sctp_assoc_reset_event); strasoc->assocreset_assoc_id = sctp_get_associd(stcb); strasoc->assocreset_local_tsn = sending_tsn; strasoc->assocreset_remote_tsn = recv_tsn; SCTP_BUF_LEN(m_notify) = sizeof(struct sctp_assoc_reset_event); SCTP_BUF_NEXT(m_notify) = NULL; if (sctp_sbspace(&stcb->asoc, &stcb->sctp_socket->so_rcv) < SCTP_BUF_LEN(m_notify)) { /* no space */ sctp_m_freem(m_notify); return; } /* append to socket */ control = sctp_build_readq_entry(stcb, stcb->asoc.primary_destination, 0, 0, stcb->asoc.context, 0, 0, 0, m_notify); if (control == NULL) { /* no memory */ sctp_m_freem(m_notify); return; } control->spec_flags = M_NOTIFICATION; control->length = SCTP_BUF_LEN(m_notify); /* not that we need this */ control->tail_mbuf = m_notify; sctp_add_to_readq(stcb->sctp_ep, stcb, control, &stcb->sctp_socket->so_rcv, 1, SCTP_READ_LOCK_NOT_HELD, SCTP_SO_NOT_LOCKED); } static void sctp_notify_stream_reset(struct sctp_tcb *stcb, int number_entries, uint16_t * list, int flag) { struct mbuf *m_notify; struct sctp_queued_to_read *control; struct sctp_stream_reset_event *strreset; int len; if ((stcb == NULL) || (sctp_stcb_is_feature_off(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_STREAM_RESETEVNT))) { /* event not enabled */ return; } m_notify = sctp_get_mbuf_for_msg(MCLBYTES, 0, M_NOWAIT, 1, MT_DATA); if (m_notify == NULL) /* no space left */ return; SCTP_BUF_LEN(m_notify) = 0; len = sizeof(struct sctp_stream_reset_event) + (number_entries * sizeof(uint16_t)); if (len > M_TRAILINGSPACE(m_notify)) { /* never enough room */ sctp_m_freem(m_notify); return; } strreset = mtod(m_notify, struct sctp_stream_reset_event *); memset(strreset, 0, len); strreset->strreset_type = SCTP_STREAM_RESET_EVENT; strreset->strreset_flags = flag; strreset->strreset_length = len; strreset->strreset_assoc_id = sctp_get_associd(stcb); if (number_entries) { int i; for (i = 0; i < number_entries; i++) { strreset->strreset_stream_list[i] = ntohs(list[i]); } } SCTP_BUF_LEN(m_notify) = len; SCTP_BUF_NEXT(m_notify) = NULL; if (sctp_sbspace(&stcb->asoc, &stcb->sctp_socket->so_rcv) < SCTP_BUF_LEN(m_notify)) { /* no space */ sctp_m_freem(m_notify); return; } /* append to socket */ control = sctp_build_readq_entry(stcb, stcb->asoc.primary_destination, 0, 0, stcb->asoc.context, 0, 0, 0, m_notify); if (control == NULL) { /* no memory */ sctp_m_freem(m_notify); return; } control->spec_flags = M_NOTIFICATION; control->length = SCTP_BUF_LEN(m_notify); /* not that we need this */ control->tail_mbuf = m_notify; sctp_add_to_readq(stcb->sctp_ep, stcb, control, &stcb->sctp_socket->so_rcv, 1, SCTP_READ_LOCK_NOT_HELD, SCTP_SO_NOT_LOCKED); } static void sctp_notify_remote_error(struct sctp_tcb *stcb, uint16_t error, struct sctp_error_chunk *chunk) { struct mbuf *m_notify; struct sctp_remote_error *sre; struct sctp_queued_to_read *control; unsigned int notif_len; uint16_t chunk_len; if ((stcb == NULL) || sctp_stcb_is_feature_off(stcb->sctp_ep, stcb, SCTP_PCB_FLAGS_RECVPEERERR)) { return; } if (chunk != NULL) { chunk_len = ntohs(chunk->ch.chunk_length); } else { chunk_len = 0; } notif_len = (unsigned int)(sizeof(struct sctp_remote_error) + chunk_len); m_notify = sctp_get_mbuf_for_msg(notif_len, 0, M_NOWAIT, 1, MT_DATA); if (m_notify == NULL) { /* Retry with smaller value. */ notif_len = (unsigned int)sizeof(struct sctp_remote_error); m_notify = sctp_get_mbuf_for_msg(notif_len, 0, M_NOWAIT, 1, MT_DATA); if (m_notify == NULL) { return; } } SCTP_BUF_NEXT(m_notify) = NULL; sre = mtod(m_notify, struct sctp_remote_error *); memset(sre, 0, notif_len); sre->sre_type = SCTP_REMOTE_ERROR; sre->sre_flags = 0; sre->sre_length = sizeof(struct sctp_remote_error); sre->sre_error = error; sre->sre_assoc_id = sctp_get_associd(stcb); if (notif_len > sizeof(struct sctp_remote_error)) { memcpy(sre->sre_data, chunk, chunk_len); sre->sre_length += chunk_len; } SCTP_BUF_LEN(m_notify) = sre->sre_length; control = sctp_build_readq_entry(stcb, stcb->asoc.primary_destination, 0, 0, stcb->asoc.context, 0, 0, 0, m_notify); if (control != NULL) { control->length = SCTP_BUF_LEN(m_notify); /* not that we need this */ control->tail_mbuf = m_notify; control->spec_flags = M_NOTIFICATION; sctp_add_to_readq(stcb->sctp_ep, stcb, control, &stcb->sctp_socket->so_rcv, 1, SCTP_READ_LOCK_NOT_HELD, SCTP_SO_NOT_LOCKED); } else { sctp_m_freem(m_notify); } } void sctp_ulp_notify(uint32_t notification, struct sctp_tcb *stcb, uint32_t error, void *data, int so_locked #if !defined(__APPLE__) && !defined(SCTP_SO_LOCK_TESTING) SCTP_UNUSED #endif ) { if ((stcb == NULL) || (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) || (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE) || (stcb->asoc.state & SCTP_STATE_CLOSED_SOCKET)) { /* If the socket is gone we are out of here */ return; } if (stcb->sctp_socket->so_rcv.sb_state & SBS_CANTRCVMORE) { return; } if ((stcb->asoc.state & SCTP_STATE_COOKIE_WAIT) || (stcb->asoc.state & SCTP_STATE_COOKIE_ECHOED)) { if ((notification == SCTP_NOTIFY_INTERFACE_DOWN) || (notification == SCTP_NOTIFY_INTERFACE_UP) || (notification == SCTP_NOTIFY_INTERFACE_CONFIRMED)) { /* Don't report these in front states */ return; } } switch (notification) { case SCTP_NOTIFY_ASSOC_UP: if (stcb->asoc.assoc_up_sent == 0) { sctp_notify_assoc_change(SCTP_COMM_UP, stcb, error, NULL, 0, so_locked); stcb->asoc.assoc_up_sent = 1; } if (stcb->asoc.adaptation_needed && (stcb->asoc.adaptation_sent == 0)) { sctp_notify_adaptation_layer(stcb); } if (stcb->asoc.auth_supported == 0) { sctp_ulp_notify(SCTP_NOTIFY_NO_PEER_AUTH, stcb, 0, NULL, so_locked); } break; case SCTP_NOTIFY_ASSOC_DOWN: sctp_notify_assoc_change(SCTP_SHUTDOWN_COMP, stcb, error, NULL, 0, so_locked); break; case SCTP_NOTIFY_INTERFACE_DOWN: { struct sctp_nets *net; net = (struct sctp_nets *)data; sctp_notify_peer_addr_change(stcb, SCTP_ADDR_UNREACHABLE, (struct sockaddr *)&net->ro._l_addr, error, so_locked); break; } case SCTP_NOTIFY_INTERFACE_UP: { struct sctp_nets *net; net = (struct sctp_nets *)data; sctp_notify_peer_addr_change(stcb, SCTP_ADDR_AVAILABLE, (struct sockaddr *)&net->ro._l_addr, error, so_locked); break; } case SCTP_NOTIFY_INTERFACE_CONFIRMED: { struct sctp_nets *net; net = (struct sctp_nets *)data; sctp_notify_peer_addr_change(stcb, SCTP_ADDR_CONFIRMED, (struct sockaddr *)&net->ro._l_addr, error, so_locked); break; } case SCTP_NOTIFY_SPECIAL_SP_FAIL: sctp_notify_send_failed2(stcb, error, (struct sctp_stream_queue_pending *)data, so_locked); break; case SCTP_NOTIFY_SENT_DG_FAIL: sctp_notify_send_failed(stcb, 1, error, (struct sctp_tmit_chunk *)data, so_locked); break; case SCTP_NOTIFY_UNSENT_DG_FAIL: sctp_notify_send_failed(stcb, 0, error, (struct sctp_tmit_chunk *)data, so_locked); break; case SCTP_NOTIFY_PARTIAL_DELVIERY_INDICATION: { uint32_t val; val = *((uint32_t *) data); sctp_notify_partial_delivery_indication(stcb, error, val, so_locked); break; } case SCTP_NOTIFY_ASSOC_LOC_ABORTED: if (((stcb->asoc.state & SCTP_STATE_MASK) == SCTP_STATE_COOKIE_WAIT) || ((stcb->asoc.state & SCTP_STATE_MASK) == SCTP_STATE_COOKIE_ECHOED)) { sctp_notify_assoc_change(SCTP_CANT_STR_ASSOC, stcb, error, data, 0, so_locked); } else { sctp_notify_assoc_change(SCTP_COMM_LOST, stcb, error, data, 0, so_locked); } break; case SCTP_NOTIFY_ASSOC_REM_ABORTED: if (((stcb->asoc.state & SCTP_STATE_MASK) == SCTP_STATE_COOKIE_WAIT) || ((stcb->asoc.state & SCTP_STATE_MASK) == SCTP_STATE_COOKIE_ECHOED)) { sctp_notify_assoc_change(SCTP_CANT_STR_ASSOC, stcb, error, data, 1, so_locked); } else { sctp_notify_assoc_change(SCTP_COMM_LOST, stcb, error, data, 1, so_locked); } break; case SCTP_NOTIFY_ASSOC_RESTART: sctp_notify_assoc_change(SCTP_RESTART, stcb, error, NULL, 0, so_locked); if (stcb->asoc.auth_supported == 0) { sctp_ulp_notify(SCTP_NOTIFY_NO_PEER_AUTH, stcb, 0, NULL, so_locked); } break; case SCTP_NOTIFY_STR_RESET_SEND: sctp_notify_stream_reset(stcb, error, ((uint16_t *) data), SCTP_STREAM_RESET_OUTGOING_SSN); break; case SCTP_NOTIFY_STR_RESET_RECV: sctp_notify_stream_reset(stcb, error, ((uint16_t *) data), SCTP_STREAM_RESET_INCOMING); break; case SCTP_NOTIFY_STR_RESET_FAILED_OUT: sctp_notify_stream_reset(stcb, error, ((uint16_t *) data), (SCTP_STREAM_RESET_OUTGOING_SSN | SCTP_STREAM_RESET_FAILED)); break; case SCTP_NOTIFY_STR_RESET_DENIED_OUT: sctp_notify_stream_reset(stcb, error, ((uint16_t *) data), (SCTP_STREAM_RESET_OUTGOING_SSN | SCTP_STREAM_RESET_DENIED)); break; case SCTP_NOTIFY_STR_RESET_FAILED_IN: sctp_notify_stream_reset(stcb, error, ((uint16_t *) data), (SCTP_STREAM_RESET_INCOMING | SCTP_STREAM_RESET_FAILED)); break; case SCTP_NOTIFY_STR_RESET_DENIED_IN: sctp_notify_stream_reset(stcb, error, ((uint16_t *) data), (SCTP_STREAM_RESET_INCOMING | SCTP_STREAM_RESET_DENIED)); break; case SCTP_NOTIFY_ASCONF_ADD_IP: sctp_notify_peer_addr_change(stcb, SCTP_ADDR_ADDED, data, error, so_locked); break; case SCTP_NOTIFY_ASCONF_DELETE_IP: sctp_notify_peer_addr_change(stcb, SCTP_ADDR_REMOVED, data, error, so_locked); break; case SCTP_NOTIFY_ASCONF_SET_PRIMARY: sctp_notify_peer_addr_change(stcb, SCTP_ADDR_MADE_PRIM, data, error, so_locked); break; case SCTP_NOTIFY_PEER_SHUTDOWN: sctp_notify_shutdown_event(stcb); break; case SCTP_NOTIFY_AUTH_NEW_KEY: sctp_notify_authentication(stcb, SCTP_AUTH_NEW_KEY, error, (uint16_t) (uintptr_t) data, so_locked); break; case SCTP_NOTIFY_AUTH_FREE_KEY: sctp_notify_authentication(stcb, SCTP_AUTH_FREE_KEY, error, (uint16_t) (uintptr_t) data, so_locked); break; case SCTP_NOTIFY_NO_PEER_AUTH: sctp_notify_authentication(stcb, SCTP_AUTH_NO_AUTH, error, (uint16_t) (uintptr_t) data, so_locked); break; case SCTP_NOTIFY_SENDER_DRY: sctp_notify_sender_dry_event(stcb, so_locked); break; case SCTP_NOTIFY_REMOTE_ERROR: sctp_notify_remote_error(stcb, error, data); break; default: SCTPDBG(SCTP_DEBUG_UTIL1, "%s: unknown notification %xh (%u)\n", __func__, notification, notification); break; } /* end switch */ } void sctp_report_all_outbound(struct sctp_tcb *stcb, uint16_t error, int holds_lock, int so_locked #if !defined(__APPLE__) && !defined(SCTP_SO_LOCK_TESTING) SCTP_UNUSED #endif ) { struct sctp_association *asoc; struct sctp_stream_out *outs; struct sctp_tmit_chunk *chk, *nchk; struct sctp_stream_queue_pending *sp, *nsp; int i; if (stcb == NULL) { return; } asoc = &stcb->asoc; if (asoc->state & SCTP_STATE_ABOUT_TO_BE_FREED) { /* already being freed */ return; } if ((stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) || (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE) || (asoc->state & SCTP_STATE_CLOSED_SOCKET)) { return; } /* now through all the gunk freeing chunks */ if (holds_lock == 0) { SCTP_TCB_SEND_LOCK(stcb); } /* sent queue SHOULD be empty */ TAILQ_FOREACH_SAFE(chk, &asoc->sent_queue, sctp_next, nchk) { TAILQ_REMOVE(&asoc->sent_queue, chk, sctp_next); asoc->sent_queue_cnt--; if (chk->sent != SCTP_DATAGRAM_NR_ACKED) { if (asoc->strmout[chk->rec.data.stream_number].chunks_on_queues > 0) { asoc->strmout[chk->rec.data.stream_number].chunks_on_queues--; #ifdef INVARIANTS } else { panic("No chunks on the queues for sid %u.", chk->rec.data.stream_number); #endif } } if (chk->data != NULL) { sctp_free_bufspace(stcb, asoc, chk, 1); sctp_ulp_notify(SCTP_NOTIFY_SENT_DG_FAIL, stcb, error, chk, so_locked); if (chk->data) { sctp_m_freem(chk->data); chk->data = NULL; } } sctp_free_a_chunk(stcb, chk, so_locked); /* sa_ignore FREED_MEMORY */ } /* pending send queue SHOULD be empty */ TAILQ_FOREACH_SAFE(chk, &asoc->send_queue, sctp_next, nchk) { TAILQ_REMOVE(&asoc->send_queue, chk, sctp_next); asoc->send_queue_cnt--; if (asoc->strmout[chk->rec.data.stream_number].chunks_on_queues > 0) { asoc->strmout[chk->rec.data.stream_number].chunks_on_queues--; #ifdef INVARIANTS } else { panic("No chunks on the queues for sid %u.", chk->rec.data.stream_number); #endif } if (chk->data != NULL) { sctp_free_bufspace(stcb, asoc, chk, 1); sctp_ulp_notify(SCTP_NOTIFY_UNSENT_DG_FAIL, stcb, error, chk, so_locked); if (chk->data) { sctp_m_freem(chk->data); chk->data = NULL; } } sctp_free_a_chunk(stcb, chk, so_locked); /* sa_ignore FREED_MEMORY */ } for (i = 0; i < asoc->streamoutcnt; i++) { /* For each stream */ outs = &asoc->strmout[i]; /* clean up any sends there */ asoc->locked_on_sending = NULL; TAILQ_FOREACH_SAFE(sp, &outs->outqueue, next, nsp) { asoc->stream_queue_cnt--; TAILQ_REMOVE(&outs->outqueue, sp, next); sctp_free_spbufspace(stcb, asoc, sp); if (sp->data) { sctp_ulp_notify(SCTP_NOTIFY_SPECIAL_SP_FAIL, stcb, error, (void *)sp, so_locked); if (sp->data) { sctp_m_freem(sp->data); sp->data = NULL; sp->tail_mbuf = NULL; sp->length = 0; } } if (sp->net) { sctp_free_remote_addr(sp->net); sp->net = NULL; } /* Free the chunk */ sctp_free_a_strmoq(stcb, sp, so_locked); /* sa_ignore FREED_MEMORY */ } } if (holds_lock == 0) { SCTP_TCB_SEND_UNLOCK(stcb); } } void sctp_abort_notification(struct sctp_tcb *stcb, uint8_t from_peer, uint16_t error, struct sctp_abort_chunk *abort, int so_locked #if !defined(__APPLE__) && !defined(SCTP_SO_LOCK_TESTING) SCTP_UNUSED #endif ) { if (stcb == NULL) { return; } if ((stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL) || ((stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) && (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_CONNECTED))) { stcb->sctp_ep->sctp_flags |= SCTP_PCB_FLAGS_WAS_ABORTED; } if ((stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) || (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE) || (stcb->asoc.state & SCTP_STATE_CLOSED_SOCKET)) { return; } /* Tell them we lost the asoc */ sctp_report_all_outbound(stcb, error, 1, so_locked); if (from_peer) { sctp_ulp_notify(SCTP_NOTIFY_ASSOC_REM_ABORTED, stcb, error, abort, so_locked); } else { sctp_ulp_notify(SCTP_NOTIFY_ASSOC_LOC_ABORTED, stcb, error, abort, so_locked); } } void sctp_abort_association(struct sctp_inpcb *inp, struct sctp_tcb *stcb, struct mbuf *m, int iphlen, struct sockaddr *src, struct sockaddr *dst, struct sctphdr *sh, struct mbuf *op_err, uint8_t mflowtype, uint32_t mflowid, uint32_t vrf_id, uint16_t port) { uint32_t vtag; #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) struct socket *so; #endif vtag = 0; if (stcb != NULL) { /* We have a TCB to abort, send notification too */ vtag = stcb->asoc.peer_vtag; sctp_abort_notification(stcb, 0, 0, NULL, SCTP_SO_NOT_LOCKED); /* get the assoc vrf id and table id */ vrf_id = stcb->asoc.vrf_id; stcb->asoc.state |= SCTP_STATE_WAS_ABORTED; } sctp_send_abort(m, iphlen, src, dst, sh, vtag, op_err, mflowtype, mflowid, inp->fibnum, vrf_id, port); if (stcb != NULL) { /* Ok, now lets free it */ #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) so = SCTP_INP_SO(inp); atomic_add_int(&stcb->asoc.refcnt, 1); SCTP_TCB_UNLOCK(stcb); SCTP_SOCKET_LOCK(so, 1); SCTP_TCB_LOCK(stcb); atomic_subtract_int(&stcb->asoc.refcnt, 1); #endif SCTP_STAT_INCR_COUNTER32(sctps_aborted); if ((SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_OPEN) || (SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_SHUTDOWN_RECEIVED)) { SCTP_STAT_DECR_GAUGE32(sctps_currestab); } (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTPUTIL + SCTP_LOC_4); #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) SCTP_SOCKET_UNLOCK(so, 1); #endif } } #ifdef SCTP_ASOCLOG_OF_TSNS void sctp_print_out_track_log(struct sctp_tcb *stcb) { #ifdef NOSIY_PRINTS int i; SCTP_PRINTF("Last ep reason:%x\n", stcb->sctp_ep->last_abort_code); SCTP_PRINTF("IN bound TSN log-aaa\n"); if ((stcb->asoc.tsn_in_at == 0) && (stcb->asoc.tsn_in_wrapped == 0)) { SCTP_PRINTF("None rcvd\n"); goto none_in; } if (stcb->asoc.tsn_in_wrapped) { for (i = stcb->asoc.tsn_in_at; i < SCTP_TSN_LOG_SIZE; i++) { SCTP_PRINTF("TSN:%x strm:%d seq:%d flags:%x sz:%d\n", stcb->asoc.in_tsnlog[i].tsn, stcb->asoc.in_tsnlog[i].strm, stcb->asoc.in_tsnlog[i].seq, stcb->asoc.in_tsnlog[i].flgs, stcb->asoc.in_tsnlog[i].sz); } } if (stcb->asoc.tsn_in_at) { for (i = 0; i < stcb->asoc.tsn_in_at; i++) { SCTP_PRINTF("TSN:%x strm:%d seq:%d flags:%x sz:%d\n", stcb->asoc.in_tsnlog[i].tsn, stcb->asoc.in_tsnlog[i].strm, stcb->asoc.in_tsnlog[i].seq, stcb->asoc.in_tsnlog[i].flgs, stcb->asoc.in_tsnlog[i].sz); } } none_in: SCTP_PRINTF("OUT bound TSN log-aaa\n"); if ((stcb->asoc.tsn_out_at == 0) && (stcb->asoc.tsn_out_wrapped == 0)) { SCTP_PRINTF("None sent\n"); } if (stcb->asoc.tsn_out_wrapped) { for (i = stcb->asoc.tsn_out_at; i < SCTP_TSN_LOG_SIZE; i++) { SCTP_PRINTF("TSN:%x strm:%d seq:%d flags:%x sz:%d\n", stcb->asoc.out_tsnlog[i].tsn, stcb->asoc.out_tsnlog[i].strm, stcb->asoc.out_tsnlog[i].seq, stcb->asoc.out_tsnlog[i].flgs, stcb->asoc.out_tsnlog[i].sz); } } if (stcb->asoc.tsn_out_at) { for (i = 0; i < stcb->asoc.tsn_out_at; i++) { SCTP_PRINTF("TSN:%x strm:%d seq:%d flags:%x sz:%d\n", stcb->asoc.out_tsnlog[i].tsn, stcb->asoc.out_tsnlog[i].strm, stcb->asoc.out_tsnlog[i].seq, stcb->asoc.out_tsnlog[i].flgs, stcb->asoc.out_tsnlog[i].sz); } } #endif } #endif void sctp_abort_an_association(struct sctp_inpcb *inp, struct sctp_tcb *stcb, struct mbuf *op_err, int so_locked #if !defined(__APPLE__) && !defined(SCTP_SO_LOCK_TESTING) SCTP_UNUSED #endif ) { #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) struct socket *so; #endif #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) so = SCTP_INP_SO(inp); #endif if (stcb == NULL) { /* Got to have a TCB */ if (inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) { if (LIST_EMPTY(&inp->sctp_asoc_list)) { sctp_inpcb_free(inp, SCTP_FREE_SHOULD_USE_ABORT, SCTP_CALLED_DIRECTLY_NOCMPSET); } } return; } else { stcb->asoc.state |= SCTP_STATE_WAS_ABORTED; } /* notify the ulp */ if ((inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) { sctp_abort_notification(stcb, 0, 0, NULL, so_locked); } /* notify the peer */ sctp_send_abort_tcb(stcb, op_err, so_locked); SCTP_STAT_INCR_COUNTER32(sctps_aborted); if ((SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_OPEN) || (SCTP_GET_STATE(&stcb->asoc) == SCTP_STATE_SHUTDOWN_RECEIVED)) { SCTP_STAT_DECR_GAUGE32(sctps_currestab); } /* now free the asoc */ #ifdef SCTP_ASOCLOG_OF_TSNS sctp_print_out_track_log(stcb); #endif #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) if (!so_locked) { atomic_add_int(&stcb->asoc.refcnt, 1); SCTP_TCB_UNLOCK(stcb); SCTP_SOCKET_LOCK(so, 1); SCTP_TCB_LOCK(stcb); atomic_subtract_int(&stcb->asoc.refcnt, 1); } #endif (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTPUTIL + SCTP_LOC_5); #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) if (!so_locked) { SCTP_SOCKET_UNLOCK(so, 1); } #endif } void sctp_handle_ootb(struct mbuf *m, int iphlen, int offset, struct sockaddr *src, struct sockaddr *dst, struct sctphdr *sh, struct sctp_inpcb *inp, struct mbuf *cause, uint8_t mflowtype, uint32_t mflowid, uint16_t fibnum, uint32_t vrf_id, uint16_t port) { struct sctp_chunkhdr *ch, chunk_buf; unsigned int chk_length; int contains_init_chunk; SCTP_STAT_INCR_COUNTER32(sctps_outoftheblue); /* Generate a TO address for future reference */ if (inp && (inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE)) { if (LIST_EMPTY(&inp->sctp_asoc_list)) { sctp_inpcb_free(inp, SCTP_FREE_SHOULD_USE_ABORT, SCTP_CALLED_DIRECTLY_NOCMPSET); } } contains_init_chunk = 0; ch = (struct sctp_chunkhdr *)sctp_m_getptr(m, offset, sizeof(*ch), (uint8_t *) & chunk_buf); while (ch != NULL) { chk_length = ntohs(ch->chunk_length); if (chk_length < sizeof(*ch)) { /* break to abort land */ break; } switch (ch->chunk_type) { case SCTP_INIT: contains_init_chunk = 1; break; case SCTP_PACKET_DROPPED: /* we don't respond to pkt-dropped */ return; case SCTP_ABORT_ASSOCIATION: /* we don't respond with an ABORT to an ABORT */ return; case SCTP_SHUTDOWN_COMPLETE: /* * we ignore it since we are not waiting for it and * peer is gone */ return; case SCTP_SHUTDOWN_ACK: sctp_send_shutdown_complete2(src, dst, sh, mflowtype, mflowid, fibnum, vrf_id, port); return; default: break; } offset += SCTP_SIZE32(chk_length); ch = (struct sctp_chunkhdr *)sctp_m_getptr(m, offset, sizeof(*ch), (uint8_t *) & chunk_buf); } if ((SCTP_BASE_SYSCTL(sctp_blackhole) == 0) || ((SCTP_BASE_SYSCTL(sctp_blackhole) == 1) && (contains_init_chunk == 0))) { sctp_send_abort(m, iphlen, src, dst, sh, 0, cause, mflowtype, mflowid, fibnum, vrf_id, port); } } /* * check the inbound datagram to make sure there is not an abort inside it, * if there is return 1, else return 0. */ int sctp_is_there_an_abort_here(struct mbuf *m, int iphlen, uint32_t * vtagfill) { struct sctp_chunkhdr *ch; struct sctp_init_chunk *init_chk, chunk_buf; int offset; unsigned int chk_length; offset = iphlen + sizeof(struct sctphdr); ch = (struct sctp_chunkhdr *)sctp_m_getptr(m, offset, sizeof(*ch), (uint8_t *) & chunk_buf); while (ch != NULL) { chk_length = ntohs(ch->chunk_length); if (chk_length < sizeof(*ch)) { /* packet is probably corrupt */ break; } /* we seem to be ok, is it an abort? */ if (ch->chunk_type == SCTP_ABORT_ASSOCIATION) { /* yep, tell them */ return (1); } if (ch->chunk_type == SCTP_INITIATION) { /* need to update the Vtag */ init_chk = (struct sctp_init_chunk *)sctp_m_getptr(m, offset, sizeof(*init_chk), (uint8_t *) & chunk_buf); if (init_chk != NULL) { *vtagfill = ntohl(init_chk->init.initiate_tag); } } /* Nope, move to the next chunk */ offset += SCTP_SIZE32(chk_length); ch = (struct sctp_chunkhdr *)sctp_m_getptr(m, offset, sizeof(*ch), (uint8_t *) & chunk_buf); } return (0); } /* * currently (2/02), ifa_addr embeds scope_id's and don't have sin6_scope_id * set (i.e. it's 0) so, create this function to compare link local scopes */ #ifdef INET6 uint32_t sctp_is_same_scope(struct sockaddr_in6 *addr1, struct sockaddr_in6 *addr2) { struct sockaddr_in6 a, b; /* save copies */ a = *addr1; b = *addr2; if (a.sin6_scope_id == 0) if (sa6_recoverscope(&a)) { /* can't get scope, so can't match */ return (0); } if (b.sin6_scope_id == 0) if (sa6_recoverscope(&b)) { /* can't get scope, so can't match */ return (0); } if (a.sin6_scope_id != b.sin6_scope_id) return (0); return (1); } /* * returns a sockaddr_in6 with embedded scope recovered and removed */ struct sockaddr_in6 * sctp_recover_scope(struct sockaddr_in6 *addr, struct sockaddr_in6 *store) { /* check and strip embedded scope junk */ if (addr->sin6_family == AF_INET6) { if (IN6_IS_SCOPE_LINKLOCAL(&addr->sin6_addr)) { if (addr->sin6_scope_id == 0) { *store = *addr; if (!sa6_recoverscope(store)) { /* use the recovered scope */ addr = store; } } else { /* else, return the original "to" addr */ in6_clearscope(&addr->sin6_addr); } } } return (addr); } #endif /* * are the two addresses the same? currently a "scopeless" check returns: 1 * if same, 0 if not */ int sctp_cmpaddr(struct sockaddr *sa1, struct sockaddr *sa2) { /* must be valid */ if (sa1 == NULL || sa2 == NULL) return (0); /* must be the same family */ if (sa1->sa_family != sa2->sa_family) return (0); switch (sa1->sa_family) { #ifdef INET6 case AF_INET6: { /* IPv6 addresses */ struct sockaddr_in6 *sin6_1, *sin6_2; sin6_1 = (struct sockaddr_in6 *)sa1; sin6_2 = (struct sockaddr_in6 *)sa2; return (SCTP6_ARE_ADDR_EQUAL(sin6_1, sin6_2)); } #endif #ifdef INET case AF_INET: { /* IPv4 addresses */ struct sockaddr_in *sin_1, *sin_2; sin_1 = (struct sockaddr_in *)sa1; sin_2 = (struct sockaddr_in *)sa2; return (sin_1->sin_addr.s_addr == sin_2->sin_addr.s_addr); } #endif default: /* we don't do these... */ return (0); } } void sctp_print_address(struct sockaddr *sa) { #ifdef INET6 char ip6buf[INET6_ADDRSTRLEN]; #endif switch (sa->sa_family) { #ifdef INET6 case AF_INET6: { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)sa; SCTP_PRINTF("IPv6 address: %s:port:%d scope:%u\n", ip6_sprintf(ip6buf, &sin6->sin6_addr), ntohs(sin6->sin6_port), sin6->sin6_scope_id); break; } #endif #ifdef INET case AF_INET: { struct sockaddr_in *sin; unsigned char *p; sin = (struct sockaddr_in *)sa; p = (unsigned char *)&sin->sin_addr; SCTP_PRINTF("IPv4 address: %u.%u.%u.%u:%d\n", p[0], p[1], p[2], p[3], ntohs(sin->sin_port)); break; } #endif default: SCTP_PRINTF("?\n"); break; } } void sctp_pull_off_control_to_new_inp(struct sctp_inpcb *old_inp, struct sctp_inpcb *new_inp, struct sctp_tcb *stcb, int waitflags) { /* * go through our old INP and pull off any control structures that * belong to stcb and move then to the new inp. */ struct socket *old_so, *new_so; struct sctp_queued_to_read *control, *nctl; struct sctp_readhead tmp_queue; struct mbuf *m; int error = 0; old_so = old_inp->sctp_socket; new_so = new_inp->sctp_socket; TAILQ_INIT(&tmp_queue); error = sblock(&old_so->so_rcv, waitflags); if (error) { /* * Gak, can't get sblock, we have a problem. data will be * left stranded.. and we don't dare look at it since the * other thread may be reading something. Oh well, its a * screwed up app that does a peeloff OR a accept while * reading from the main socket... actually its only the * peeloff() case, since I think read will fail on a * listening socket.. */ return; } /* lock the socket buffers */ SCTP_INP_READ_LOCK(old_inp); TAILQ_FOREACH_SAFE(control, &old_inp->read_queue, next, nctl) { /* Pull off all for out target stcb */ if (control->stcb == stcb) { /* remove it we want it */ TAILQ_REMOVE(&old_inp->read_queue, control, next); TAILQ_INSERT_TAIL(&tmp_queue, control, next); m = control->data; while (m) { if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_SB_LOGGING_ENABLE) { sctp_sblog(&old_so->so_rcv, control->do_not_ref_stcb ? NULL : stcb, SCTP_LOG_SBFREE, SCTP_BUF_LEN(m)); } sctp_sbfree(control, stcb, &old_so->so_rcv, m); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_SB_LOGGING_ENABLE) { sctp_sblog(&old_so->so_rcv, control->do_not_ref_stcb ? NULL : stcb, SCTP_LOG_SBRESULT, 0); } m = SCTP_BUF_NEXT(m); } } } SCTP_INP_READ_UNLOCK(old_inp); /* Remove the sb-lock on the old socket */ sbunlock(&old_so->so_rcv); /* Now we move them over to the new socket buffer */ SCTP_INP_READ_LOCK(new_inp); TAILQ_FOREACH_SAFE(control, &tmp_queue, next, nctl) { TAILQ_INSERT_TAIL(&new_inp->read_queue, control, next); m = control->data; while (m) { if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_SB_LOGGING_ENABLE) { sctp_sblog(&new_so->so_rcv, control->do_not_ref_stcb ? NULL : stcb, SCTP_LOG_SBALLOC, SCTP_BUF_LEN(m)); } sctp_sballoc(stcb, &new_so->so_rcv, m); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_SB_LOGGING_ENABLE) { sctp_sblog(&new_so->so_rcv, control->do_not_ref_stcb ? NULL : stcb, SCTP_LOG_SBRESULT, 0); } m = SCTP_BUF_NEXT(m); } } SCTP_INP_READ_UNLOCK(new_inp); } void sctp_wakeup_the_read_socket(struct sctp_inpcb *inp, struct sctp_tcb *stcb, int so_locked #if !defined(__APPLE__) && !defined(SCTP_SO_LOCK_TESTING) SCTP_UNUSED #endif ) { if ((inp != NULL) && (inp->sctp_socket != NULL)) { if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_ZERO_COPY_ACTIVE)) { SCTP_ZERO_COPY_EVENT(inp, inp->sctp_socket); } else { #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) struct socket *so; so = SCTP_INP_SO(inp); if (!so_locked) { if (stcb) { atomic_add_int(&stcb->asoc.refcnt, 1); SCTP_TCB_UNLOCK(stcb); } SCTP_SOCKET_LOCK(so, 1); if (stcb) { SCTP_TCB_LOCK(stcb); atomic_subtract_int(&stcb->asoc.refcnt, 1); } if (inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) { SCTP_SOCKET_UNLOCK(so, 1); return; } } #endif sctp_sorwakeup(inp, inp->sctp_socket); #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) if (!so_locked) { SCTP_SOCKET_UNLOCK(so, 1); } #endif } } } void sctp_add_to_readq(struct sctp_inpcb *inp, struct sctp_tcb *stcb, struct sctp_queued_to_read *control, struct sockbuf *sb, int end, int inp_read_lock_held, int so_locked #if !defined(__APPLE__) && !defined(SCTP_SO_LOCK_TESTING) SCTP_UNUSED #endif ) { /* * Here we must place the control on the end of the socket read * queue AND increment sb_cc so that select will work properly on * read. */ struct mbuf *m, *prev = NULL; if (inp == NULL) { /* Gak, TSNH!! */ #ifdef INVARIANTS panic("Gak, inp NULL on add_to_readq"); #endif return; } if (inp_read_lock_held == 0) SCTP_INP_READ_LOCK(inp); if (inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_CANT_READ) { sctp_free_remote_addr(control->whoFrom); if (control->data) { sctp_m_freem(control->data); control->data = NULL; } sctp_free_a_readq(stcb, control); if (inp_read_lock_held == 0) SCTP_INP_READ_UNLOCK(inp); return; } if (!(control->spec_flags & M_NOTIFICATION)) { atomic_add_int(&inp->total_recvs, 1); if (!control->do_not_ref_stcb) { atomic_add_int(&stcb->total_recvs, 1); } } m = control->data; control->held_length = 0; control->length = 0; while (m) { if (SCTP_BUF_LEN(m) == 0) { /* Skip mbufs with NO length */ if (prev == NULL) { /* First one */ control->data = sctp_m_free(m); m = control->data; } else { SCTP_BUF_NEXT(prev) = sctp_m_free(m); m = SCTP_BUF_NEXT(prev); } if (m == NULL) { control->tail_mbuf = prev; } continue; } prev = m; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_SB_LOGGING_ENABLE) { sctp_sblog(sb, control->do_not_ref_stcb ? NULL : stcb, SCTP_LOG_SBALLOC, SCTP_BUF_LEN(m)); } sctp_sballoc(stcb, sb, m); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_SB_LOGGING_ENABLE) { sctp_sblog(sb, control->do_not_ref_stcb ? NULL : stcb, SCTP_LOG_SBRESULT, 0); } atomic_add_int(&control->length, SCTP_BUF_LEN(m)); m = SCTP_BUF_NEXT(m); } if (prev != NULL) { control->tail_mbuf = prev; } else { /* Everything got collapsed out?? */ sctp_free_remote_addr(control->whoFrom); sctp_free_a_readq(stcb, control); if (inp_read_lock_held == 0) SCTP_INP_READ_UNLOCK(inp); return; } if (end) { control->end_added = 1; } TAILQ_INSERT_TAIL(&inp->read_queue, control, next); control->on_read_q = 1; if (inp_read_lock_held == 0) SCTP_INP_READ_UNLOCK(inp); if (inp && inp->sctp_socket) { sctp_wakeup_the_read_socket(inp, stcb, so_locked); } } /*************HOLD THIS COMMENT FOR PATCH FILE OF *************ALTERNATE ROUTING CODE */ /*************HOLD THIS COMMENT FOR END OF PATCH FILE OF *************ALTERNATE ROUTING CODE */ struct mbuf * sctp_generate_cause(uint16_t code, char *info) { struct mbuf *m; struct sctp_gen_error_cause *cause; size_t info_len; uint16_t len; if ((code == 0) || (info == NULL)) { return (NULL); } info_len = strlen(info); if (info_len > (SCTP_MAX_CAUSE_LENGTH - sizeof(struct sctp_paramhdr))) { return (NULL); } len = (uint16_t) (sizeof(struct sctp_paramhdr) + info_len); m = sctp_get_mbuf_for_msg(len, 0, M_NOWAIT, 1, MT_DATA); if (m != NULL) { SCTP_BUF_LEN(m) = len; cause = mtod(m, struct sctp_gen_error_cause *); cause->code = htons(code); cause->length = htons(len); memcpy(cause->info, info, info_len); } return (m); } struct mbuf * sctp_generate_no_user_data_cause(uint32_t tsn) { struct mbuf *m; struct sctp_error_no_user_data *no_user_data_cause; uint16_t len; len = (uint16_t) sizeof(struct sctp_error_no_user_data); m = sctp_get_mbuf_for_msg(len, 0, M_NOWAIT, 1, MT_DATA); if (m != NULL) { SCTP_BUF_LEN(m) = len; no_user_data_cause = mtod(m, struct sctp_error_no_user_data *); no_user_data_cause->cause.code = htons(SCTP_CAUSE_NO_USER_DATA); no_user_data_cause->cause.length = htons(len); no_user_data_cause->tsn = tsn; /* tsn is passed in as NBO */ } return (m); } #ifdef SCTP_MBCNT_LOGGING void sctp_free_bufspace(struct sctp_tcb *stcb, struct sctp_association *asoc, struct sctp_tmit_chunk *tp1, int chk_cnt) { if (tp1->data == NULL) { return; } asoc->chunks_on_out_queue -= chk_cnt; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_MBCNT_LOGGING_ENABLE) { sctp_log_mbcnt(SCTP_LOG_MBCNT_DECREASE, asoc->total_output_queue_size, tp1->book_size, 0, tp1->mbcnt); } if (asoc->total_output_queue_size >= tp1->book_size) { atomic_add_int(&asoc->total_output_queue_size, -tp1->book_size); } else { asoc->total_output_queue_size = 0; } if (stcb->sctp_socket && (((stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) || ((stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE)))) { if (stcb->sctp_socket->so_snd.sb_cc >= tp1->book_size) { stcb->sctp_socket->so_snd.sb_cc -= tp1->book_size; } else { stcb->sctp_socket->so_snd.sb_cc = 0; } } } #endif int sctp_release_pr_sctp_chunk(struct sctp_tcb *stcb, struct sctp_tmit_chunk *tp1, uint8_t sent, int so_locked #if !defined(__APPLE__) && !defined(SCTP_SO_LOCK_TESTING) SCTP_UNUSED #endif ) { struct sctp_stream_out *strq; struct sctp_tmit_chunk *chk = NULL, *tp2; struct sctp_stream_queue_pending *sp; uint16_t stream = 0, seq = 0; uint8_t foundeom = 0; int ret_sz = 0; int notdone; int do_wakeup_routine = 0; stream = tp1->rec.data.stream_number; seq = tp1->rec.data.stream_seq; if (sent || !(tp1->rec.data.rcv_flags & SCTP_DATA_FIRST_FRAG)) { stcb->asoc.abandoned_sent[0]++; stcb->asoc.abandoned_sent[PR_SCTP_POLICY(tp1->flags)]++; stcb->asoc.strmout[stream].abandoned_sent[0]++; #if defined(SCTP_DETAILED_STR_STATS) stcb->asoc.strmout[stream].abandoned_sent[PR_SCTP_POLICY(tp1->flags)]++; #endif } else { stcb->asoc.abandoned_unsent[0]++; stcb->asoc.abandoned_unsent[PR_SCTP_POLICY(tp1->flags)]++; stcb->asoc.strmout[stream].abandoned_unsent[0]++; #if defined(SCTP_DETAILED_STR_STATS) stcb->asoc.strmout[stream].abandoned_unsent[PR_SCTP_POLICY(tp1->flags)]++; #endif } do { ret_sz += tp1->book_size; if (tp1->data != NULL) { if (tp1->sent < SCTP_DATAGRAM_RESEND) { sctp_flight_size_decrease(tp1); sctp_total_flight_decrease(stcb, tp1); } sctp_free_bufspace(stcb, &stcb->asoc, tp1, 1); stcb->asoc.peers_rwnd += tp1->send_size; stcb->asoc.peers_rwnd += SCTP_BASE_SYSCTL(sctp_peer_chunk_oh); if (sent) { sctp_ulp_notify(SCTP_NOTIFY_SENT_DG_FAIL, stcb, 0, tp1, so_locked); } else { sctp_ulp_notify(SCTP_NOTIFY_UNSENT_DG_FAIL, stcb, 0, tp1, so_locked); } if (tp1->data) { sctp_m_freem(tp1->data); tp1->data = NULL; } do_wakeup_routine = 1; if (PR_SCTP_BUF_ENABLED(tp1->flags)) { stcb->asoc.sent_queue_cnt_removeable--; } } tp1->sent = SCTP_FORWARD_TSN_SKIP; if ((tp1->rec.data.rcv_flags & SCTP_DATA_NOT_FRAG) == SCTP_DATA_NOT_FRAG) { /* not frag'ed we ae done */ notdone = 0; foundeom = 1; } else if (tp1->rec.data.rcv_flags & SCTP_DATA_LAST_FRAG) { /* end of frag, we are done */ notdone = 0; foundeom = 1; } else { /* * Its a begin or middle piece, we must mark all of * it */ notdone = 1; tp1 = TAILQ_NEXT(tp1, sctp_next); } } while (tp1 && notdone); if (foundeom == 0) { /* * The multi-part message was scattered across the send and * sent queue. */ TAILQ_FOREACH_SAFE(tp1, &stcb->asoc.send_queue, sctp_next, tp2) { if ((tp1->rec.data.stream_number != stream) || (tp1->rec.data.stream_seq != seq)) { break; } /* * save to chk in case we have some on stream out * queue. If so and we have an un-transmitted one we * don't have to fudge the TSN. */ chk = tp1; ret_sz += tp1->book_size; sctp_free_bufspace(stcb, &stcb->asoc, tp1, 1); if (sent) { sctp_ulp_notify(SCTP_NOTIFY_SENT_DG_FAIL, stcb, 0, tp1, so_locked); } else { sctp_ulp_notify(SCTP_NOTIFY_UNSENT_DG_FAIL, stcb, 0, tp1, so_locked); } if (tp1->data) { sctp_m_freem(tp1->data); tp1->data = NULL; } /* No flight involved here book the size to 0 */ tp1->book_size = 0; if (tp1->rec.data.rcv_flags & SCTP_DATA_LAST_FRAG) { foundeom = 1; } do_wakeup_routine = 1; tp1->sent = SCTP_FORWARD_TSN_SKIP; TAILQ_REMOVE(&stcb->asoc.send_queue, tp1, sctp_next); /* * on to the sent queue so we can wait for it to be * passed by. */ TAILQ_INSERT_TAIL(&stcb->asoc.sent_queue, tp1, sctp_next); stcb->asoc.send_queue_cnt--; stcb->asoc.sent_queue_cnt++; } } if (foundeom == 0) { /* * Still no eom found. That means there is stuff left on the * stream out queue.. yuck. */ SCTP_TCB_SEND_LOCK(stcb); strq = &stcb->asoc.strmout[stream]; sp = TAILQ_FIRST(&strq->outqueue); if (sp != NULL) { sp->discard_rest = 1; /* * We may need to put a chunk on the queue that * holds the TSN that would have been sent with the * LAST bit. */ if (chk == NULL) { /* Yep, we have to */ sctp_alloc_a_chunk(stcb, chk); if (chk == NULL) { /* * we are hosed. All we can do is * nothing.. which will cause an * abort if the peer is paying * attention. */ goto oh_well; } memset(chk, 0, sizeof(*chk)); chk->rec.data.rcv_flags = 0; chk->sent = SCTP_FORWARD_TSN_SKIP; chk->asoc = &stcb->asoc; if (stcb->asoc.idata_supported == 0) { if (sp->sinfo_flags & SCTP_UNORDERED) { chk->rec.data.stream_seq = 0; } else { chk->rec.data.stream_seq = strq->next_mid_ordered; } } else { if (sp->sinfo_flags & SCTP_UNORDERED) { chk->rec.data.stream_seq = strq->next_mid_unordered; } else { chk->rec.data.stream_seq = strq->next_mid_ordered; } } chk->rec.data.stream_number = sp->stream; chk->rec.data.payloadtype = sp->ppid; chk->rec.data.context = sp->context; chk->flags = sp->act_flags; chk->whoTo = NULL; chk->rec.data.TSN_seq = atomic_fetchadd_int(&stcb->asoc.sending_seq, 1); strq->chunks_on_queues++; TAILQ_INSERT_TAIL(&stcb->asoc.sent_queue, chk, sctp_next); stcb->asoc.sent_queue_cnt++; stcb->asoc.pr_sctp_cnt++; } chk->rec.data.rcv_flags |= SCTP_DATA_LAST_FRAG; if (stcb->asoc.idata_supported == 0) { if ((sp->sinfo_flags & SCTP_UNORDERED) == 0) { strq->next_mid_ordered++; } } else { if (sp->sinfo_flags & SCTP_UNORDERED) { strq->next_mid_unordered++; } else { strq->next_mid_ordered++; } } oh_well: if (sp->data) { /* * Pull any data to free up the SB and allow * sender to "add more" while we will throw * away :-) */ sctp_free_spbufspace(stcb, &stcb->asoc, sp); ret_sz += sp->length; do_wakeup_routine = 1; sp->some_taken = 1; sctp_m_freem(sp->data); sp->data = NULL; sp->tail_mbuf = NULL; sp->length = 0; } } SCTP_TCB_SEND_UNLOCK(stcb); } if (do_wakeup_routine) { #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) struct socket *so; so = SCTP_INP_SO(stcb->sctp_ep); if (!so_locked) { atomic_add_int(&stcb->asoc.refcnt, 1); SCTP_TCB_UNLOCK(stcb); SCTP_SOCKET_LOCK(so, 1); SCTP_TCB_LOCK(stcb); atomic_subtract_int(&stcb->asoc.refcnt, 1); if (stcb->asoc.state & SCTP_STATE_CLOSED_SOCKET) { /* assoc was freed while we were unlocked */ SCTP_SOCKET_UNLOCK(so, 1); return (ret_sz); } } #endif sctp_sowwakeup(stcb->sctp_ep, stcb->sctp_socket); #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) if (!so_locked) { SCTP_SOCKET_UNLOCK(so, 1); } #endif } return (ret_sz); } /* * checks to see if the given address, sa, is one that is currently known by * the kernel note: can't distinguish the same address on multiple interfaces * and doesn't handle multiple addresses with different zone/scope id's note: * ifa_ifwithaddr() compares the entire sockaddr struct */ struct sctp_ifa * sctp_find_ifa_in_ep(struct sctp_inpcb *inp, struct sockaddr *addr, int holds_lock) { struct sctp_laddr *laddr; if (holds_lock == 0) { SCTP_INP_RLOCK(inp); } LIST_FOREACH(laddr, &inp->sctp_addr_list, sctp_nxt_addr) { if (laddr->ifa == NULL) continue; if (addr->sa_family != laddr->ifa->address.sa.sa_family) continue; #ifdef INET if (addr->sa_family == AF_INET) { if (((struct sockaddr_in *)addr)->sin_addr.s_addr == laddr->ifa->address.sin.sin_addr.s_addr) { /* found him. */ if (holds_lock == 0) { SCTP_INP_RUNLOCK(inp); } return (laddr->ifa); break; } } #endif #ifdef INET6 if (addr->sa_family == AF_INET6) { if (SCTP6_ARE_ADDR_EQUAL((struct sockaddr_in6 *)addr, &laddr->ifa->address.sin6)) { /* found him. */ if (holds_lock == 0) { SCTP_INP_RUNLOCK(inp); } return (laddr->ifa); break; } } #endif } if (holds_lock == 0) { SCTP_INP_RUNLOCK(inp); } return (NULL); } uint32_t sctp_get_ifa_hash_val(struct sockaddr *addr) { switch (addr->sa_family) { #ifdef INET case AF_INET: { struct sockaddr_in *sin; sin = (struct sockaddr_in *)addr; return (sin->sin_addr.s_addr ^ (sin->sin_addr.s_addr >> 16)); } #endif #ifdef INET6 case AF_INET6: { struct sockaddr_in6 *sin6; uint32_t hash_of_addr; sin6 = (struct sockaddr_in6 *)addr; hash_of_addr = (sin6->sin6_addr.s6_addr32[0] + sin6->sin6_addr.s6_addr32[1] + sin6->sin6_addr.s6_addr32[2] + sin6->sin6_addr.s6_addr32[3]); hash_of_addr = (hash_of_addr ^ (hash_of_addr >> 16)); return (hash_of_addr); } #endif default: break; } return (0); } struct sctp_ifa * sctp_find_ifa_by_addr(struct sockaddr *addr, uint32_t vrf_id, int holds_lock) { struct sctp_ifa *sctp_ifap; struct sctp_vrf *vrf; struct sctp_ifalist *hash_head; uint32_t hash_of_addr; if (holds_lock == 0) SCTP_IPI_ADDR_RLOCK(); vrf = sctp_find_vrf(vrf_id); if (vrf == NULL) { if (holds_lock == 0) SCTP_IPI_ADDR_RUNLOCK(); return (NULL); } hash_of_addr = sctp_get_ifa_hash_val(addr); hash_head = &vrf->vrf_addr_hash[(hash_of_addr & vrf->vrf_addr_hashmark)]; if (hash_head == NULL) { SCTP_PRINTF("hash_of_addr:%x mask:%x table:%x - ", hash_of_addr, (uint32_t) vrf->vrf_addr_hashmark, (uint32_t) (hash_of_addr & vrf->vrf_addr_hashmark)); sctp_print_address(addr); SCTP_PRINTF("No such bucket for address\n"); if (holds_lock == 0) SCTP_IPI_ADDR_RUNLOCK(); return (NULL); } LIST_FOREACH(sctp_ifap, hash_head, next_bucket) { if (addr->sa_family != sctp_ifap->address.sa.sa_family) continue; #ifdef INET if (addr->sa_family == AF_INET) { if (((struct sockaddr_in *)addr)->sin_addr.s_addr == sctp_ifap->address.sin.sin_addr.s_addr) { /* found him. */ if (holds_lock == 0) SCTP_IPI_ADDR_RUNLOCK(); return (sctp_ifap); break; } } #endif #ifdef INET6 if (addr->sa_family == AF_INET6) { if (SCTP6_ARE_ADDR_EQUAL((struct sockaddr_in6 *)addr, &sctp_ifap->address.sin6)) { /* found him. */ if (holds_lock == 0) SCTP_IPI_ADDR_RUNLOCK(); return (sctp_ifap); break; } } #endif } if (holds_lock == 0) SCTP_IPI_ADDR_RUNLOCK(); return (NULL); } static void sctp_user_rcvd(struct sctp_tcb *stcb, uint32_t * freed_so_far, int hold_rlock, uint32_t rwnd_req) { /* User pulled some data, do we need a rwnd update? */ int r_unlocked = 0; uint32_t dif, rwnd; struct socket *so = NULL; if (stcb == NULL) return; atomic_add_int(&stcb->asoc.refcnt, 1); if (stcb->asoc.state & (SCTP_STATE_ABOUT_TO_BE_FREED | SCTP_STATE_SHUTDOWN_RECEIVED | SCTP_STATE_SHUTDOWN_ACK_SENT)) { /* Pre-check If we are freeing no update */ goto no_lock; } SCTP_INP_INCR_REF(stcb->sctp_ep); if ((stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) || (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE)) { goto out; } so = stcb->sctp_socket; if (so == NULL) { goto out; } atomic_add_int(&stcb->freed_by_sorcv_sincelast, *freed_so_far); /* Have you have freed enough to look */ *freed_so_far = 0; /* Yep, its worth a look and the lock overhead */ /* Figure out what the rwnd would be */ rwnd = sctp_calc_rwnd(stcb, &stcb->asoc); if (rwnd >= stcb->asoc.my_last_reported_rwnd) { dif = rwnd - stcb->asoc.my_last_reported_rwnd; } else { dif = 0; } if (dif >= rwnd_req) { if (hold_rlock) { SCTP_INP_READ_UNLOCK(stcb->sctp_ep); r_unlocked = 1; } if (stcb->asoc.state & SCTP_STATE_ABOUT_TO_BE_FREED) { /* * One last check before we allow the guy possibly * to get in. There is a race, where the guy has not * reached the gate. In that case */ goto out; } SCTP_TCB_LOCK(stcb); if (stcb->asoc.state & SCTP_STATE_ABOUT_TO_BE_FREED) { /* No reports here */ SCTP_TCB_UNLOCK(stcb); goto out; } SCTP_STAT_INCR(sctps_wu_sacks_sent); sctp_send_sack(stcb, SCTP_SO_LOCKED); sctp_chunk_output(stcb->sctp_ep, stcb, SCTP_OUTPUT_FROM_USR_RCVD, SCTP_SO_LOCKED); /* make sure no timer is running */ sctp_timer_stop(SCTP_TIMER_TYPE_RECV, stcb->sctp_ep, stcb, NULL, SCTP_FROM_SCTPUTIL + SCTP_LOC_6); SCTP_TCB_UNLOCK(stcb); } else { /* Update how much we have pending */ stcb->freed_by_sorcv_sincelast = dif; } out: if (so && r_unlocked && hold_rlock) { SCTP_INP_READ_LOCK(stcb->sctp_ep); } SCTP_INP_DECR_REF(stcb->sctp_ep); no_lock: atomic_add_int(&stcb->asoc.refcnt, -1); return; } int sctp_sorecvmsg(struct socket *so, struct uio *uio, struct mbuf **mp, struct sockaddr *from, int fromlen, int *msg_flags, struct sctp_sndrcvinfo *sinfo, int filling_sinfo) { /* * MSG flags we will look at MSG_DONTWAIT - non-blocking IO. * MSG_PEEK - Look don't touch :-D (only valid with OUT mbuf copy * mp=NULL thus uio is the copy method to userland) MSG_WAITALL - ?? * On the way out we may send out any combination of: * MSG_NOTIFICATION MSG_EOR * */ struct sctp_inpcb *inp = NULL; int my_len = 0; int cp_len = 0, error = 0; struct sctp_queued_to_read *control = NULL, *ctl = NULL, *nxt = NULL; struct mbuf *m = NULL; struct sctp_tcb *stcb = NULL; int wakeup_read_socket = 0; int freecnt_applied = 0; int out_flags = 0, in_flags = 0; int block_allowed = 1; uint32_t freed_so_far = 0; uint32_t copied_so_far = 0; int in_eeor_mode = 0; int no_rcv_needed = 0; uint32_t rwnd_req = 0; int hold_sblock = 0; int hold_rlock = 0; ssize_t slen = 0; uint32_t held_length = 0; int sockbuf_lock = 0; if (uio == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); return (EINVAL); } if (msg_flags) { in_flags = *msg_flags; if (in_flags & MSG_PEEK) SCTP_STAT_INCR(sctps_read_peeks); } else { in_flags = 0; } slen = uio->uio_resid; /* Pull in and set up our int flags */ if (in_flags & MSG_OOB) { /* Out of band's NOT supported */ return (EOPNOTSUPP); } if ((in_flags & MSG_PEEK) && (mp != NULL)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); return (EINVAL); } if ((in_flags & (MSG_DONTWAIT | MSG_NBIO )) || SCTP_SO_IS_NBIO(so)) { block_allowed = 0; } /* setup the endpoint */ inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { SCTP_LTRACE_ERR_RET(NULL, NULL, NULL, SCTP_FROM_SCTPUTIL, EFAULT); return (EFAULT); } rwnd_req = (SCTP_SB_LIMIT_RCV(so) >> SCTP_RWND_HIWAT_SHIFT); /* Must be at least a MTU's worth */ if (rwnd_req < SCTP_MIN_RWND) rwnd_req = SCTP_MIN_RWND; in_eeor_mode = sctp_is_feature_on(inp, SCTP_PCB_FLAGS_EXPLICIT_EOR); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_RECV_RWND_LOGGING_ENABLE) { sctp_misc_ints(SCTP_SORECV_ENTER, rwnd_req, in_eeor_mode, so->so_rcv.sb_cc, (uint32_t) uio->uio_resid); } if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_RECV_RWND_LOGGING_ENABLE) { sctp_misc_ints(SCTP_SORECV_ENTERPL, rwnd_req, block_allowed, so->so_rcv.sb_cc, (uint32_t) uio->uio_resid); } error = sblock(&so->so_rcv, (block_allowed ? SBL_WAIT : 0)); if (error) { goto release_unlocked; } sockbuf_lock = 1; restart: restart_nosblocks: if (hold_sblock == 0) { SOCKBUF_LOCK(&so->so_rcv); hold_sblock = 1; } if ((inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) || (inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE)) { goto out; } if ((so->so_rcv.sb_state & SBS_CANTRCVMORE) && (so->so_rcv.sb_cc == 0)) { if (so->so_error) { error = so->so_error; if ((in_flags & MSG_PEEK) == 0) so->so_error = 0; goto out; } else { if (so->so_rcv.sb_cc == 0) { /* indicate EOF */ error = 0; goto out; } } } - if ((so->so_rcv.sb_cc <= held_length) && block_allowed) { - /* we need to wait for data */ + if (so->so_rcv.sb_cc <= held_length) { + if (so->so_error) { + error = so->so_error; + if ((in_flags & MSG_PEEK) == 0) { + so->so_error = 0; + } + goto out; + } if ((so->so_rcv.sb_cc == 0) && ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL))) { if ((inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) == 0) { /* * For active open side clear flags for * re-use passive open is blocked by * connect. */ if (inp->sctp_flags & SCTP_PCB_FLAGS_WAS_ABORTED) { /* * You were aborted, passive side * always hits here */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, ECONNRESET); error = ECONNRESET; } so->so_state &= ~(SS_ISCONNECTING | SS_ISDISCONNECTING | SS_ISCONFIRMING | SS_ISCONNECTED); if (error == 0) { if ((inp->sctp_flags & SCTP_PCB_FLAGS_WAS_CONNECTED) == 0) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, ENOTCONN); error = ENOTCONN; } } goto out; } } - error = sbwait(&so->so_rcv); - if (error) { - goto out; - } - held_length = 0; - goto restart_nosblocks; - } else if (so->so_rcv.sb_cc == 0) { - if (so->so_error) { - error = so->so_error; - if ((in_flags & MSG_PEEK) == 0) - so->so_error = 0; - } else { - if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) || - (inp->sctp_flags & SCTP_PCB_FLAGS_IN_TCPPOOL)) { - if ((inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) == 0) { - /* - * For active open side clear flags - * for re-use passive open is - * blocked by connect. - */ - if (inp->sctp_flags & SCTP_PCB_FLAGS_WAS_ABORTED) { - /* - * You were aborted, passive - * side always hits here - */ - SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, ECONNRESET); - error = ECONNRESET; - } - so->so_state &= ~(SS_ISCONNECTING | - SS_ISDISCONNECTING | - SS_ISCONFIRMING | - SS_ISCONNECTED); - if (error == 0) { - if ((inp->sctp_flags & SCTP_PCB_FLAGS_WAS_CONNECTED) == 0) { - SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, ENOTCONN); - error = ENOTCONN; - } - } - goto out; - } + if (block_allowed) { + error = sbwait(&so->so_rcv); + if (error) { + goto out; } + held_length = 0; + goto restart_nosblocks; + } else { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EWOULDBLOCK); error = EWOULDBLOCK; + goto out; } - goto out; } if (hold_sblock == 1) { SOCKBUF_UNLOCK(&so->so_rcv); hold_sblock = 0; } /* we possibly have data we can read */ /* sa_ignore FREED_MEMORY */ control = TAILQ_FIRST(&inp->read_queue); if (control == NULL) { /* * This could be happening since the appender did the * increment but as not yet did the tailq insert onto the * read_queue */ if (hold_rlock == 0) { SCTP_INP_READ_LOCK(inp); } control = TAILQ_FIRST(&inp->read_queue); if ((control == NULL) && (so->so_rcv.sb_cc != 0)) { #ifdef INVARIANTS panic("Huh, its non zero and nothing on control?"); #endif so->so_rcv.sb_cc = 0; } SCTP_INP_READ_UNLOCK(inp); hold_rlock = 0; goto restart; } if ((control->length == 0) && (control->do_not_ref_stcb)) { /* * Clean up code for freeing assoc that left behind a * pdapi.. maybe a peer in EEOR that just closed after * sending and never indicated a EOR. */ if (hold_rlock == 0) { hold_rlock = 1; SCTP_INP_READ_LOCK(inp); } control->held_length = 0; if (control->data) { /* Hmm there is data here .. fix */ struct mbuf *m_tmp; int cnt = 0; m_tmp = control->data; while (m_tmp) { cnt += SCTP_BUF_LEN(m_tmp); if (SCTP_BUF_NEXT(m_tmp) == NULL) { control->tail_mbuf = m_tmp; control->end_added = 1; } m_tmp = SCTP_BUF_NEXT(m_tmp); } control->length = cnt; } else { /* remove it */ TAILQ_REMOVE(&inp->read_queue, control, next); /* Add back any hiddend data */ sctp_free_remote_addr(control->whoFrom); sctp_free_a_readq(stcb, control); } if (hold_rlock) { hold_rlock = 0; SCTP_INP_READ_UNLOCK(inp); } goto restart; } if ((control->length == 0) && (control->end_added == 1)) { /* * Do we also need to check for (control->pdapi_aborted == * 1)? */ if (hold_rlock == 0) { hold_rlock = 1; SCTP_INP_READ_LOCK(inp); } TAILQ_REMOVE(&inp->read_queue, control, next); if (control->data) { #ifdef INVARIANTS panic("control->data not null but control->length == 0"); #else SCTP_PRINTF("Strange, data left in the control buffer. Cleaning up.\n"); sctp_m_freem(control->data); control->data = NULL; #endif } if (control->aux_data) { sctp_m_free(control->aux_data); control->aux_data = NULL; } #ifdef INVARIANTS if (control->on_strm_q) { panic("About to free ctl:%p so:%p and its in %d", control, so, control->on_strm_q); } #endif sctp_free_remote_addr(control->whoFrom); sctp_free_a_readq(stcb, control); if (hold_rlock) { hold_rlock = 0; SCTP_INP_READ_UNLOCK(inp); } goto restart; } if (control->length == 0) { if ((sctp_is_feature_on(inp, SCTP_PCB_FLAGS_FRAG_INTERLEAVE)) && (filling_sinfo)) { /* find a more suitable one then this */ ctl = TAILQ_NEXT(control, next); while (ctl) { if ((ctl->stcb != control->stcb) && (ctl->length) && (ctl->some_taken || (ctl->spec_flags & M_NOTIFICATION) || ((ctl->do_not_ref_stcb == 0) && (ctl->stcb->asoc.strmin[ctl->sinfo_stream].delivery_started == 0))) ) { /*- * If we have a different TCB next, and there is data * present. If we have already taken some (pdapi), OR we can * ref the tcb and no delivery as started on this stream, we * take it. Note we allow a notification on a different * assoc to be delivered.. */ control = ctl; goto found_one; } else if ((sctp_is_feature_on(inp, SCTP_PCB_FLAGS_INTERLEAVE_STRMS)) && (ctl->length) && ((ctl->some_taken) || ((ctl->do_not_ref_stcb == 0) && ((ctl->spec_flags & M_NOTIFICATION) == 0) && (ctl->stcb->asoc.strmin[ctl->sinfo_stream].delivery_started == 0)))) { /*- * If we have the same tcb, and there is data present, and we * have the strm interleave feature present. Then if we have * taken some (pdapi) or we can refer to tht tcb AND we have * not started a delivery for this stream, we can take it. * Note we do NOT allow a notificaiton on the same assoc to * be delivered. */ control = ctl; goto found_one; } ctl = TAILQ_NEXT(ctl, next); } } /* * if we reach here, not suitable replacement is available * fragment interleave is NOT on. So stuff the sb_cc * into the our held count, and its time to sleep again. */ held_length = so->so_rcv.sb_cc; control->held_length = so->so_rcv.sb_cc; goto restart; } /* Clear the held length since there is something to read */ control->held_length = 0; if (hold_rlock) { SCTP_INP_READ_UNLOCK(inp); hold_rlock = 0; } found_one: /* * If we reach here, control has a some data for us to read off. * Note that stcb COULD be NULL. */ control->some_taken++; if (hold_sblock) { SOCKBUF_UNLOCK(&so->so_rcv); hold_sblock = 0; } stcb = control->stcb; if (stcb) { if ((control->do_not_ref_stcb == 0) && (stcb->asoc.state & SCTP_STATE_ABOUT_TO_BE_FREED)) { if (freecnt_applied == 0) stcb = NULL; } else if (control->do_not_ref_stcb == 0) { /* you can't free it on me please */ /* * The lock on the socket buffer protects us so the * free code will stop. But since we used the * socketbuf lock and the sender uses the tcb_lock * to increment, we need to use the atomic add to * the refcnt */ if (freecnt_applied) { #ifdef INVARIANTS panic("refcnt already incremented"); #else SCTP_PRINTF("refcnt already incremented?\n"); #endif } else { atomic_add_int(&stcb->asoc.refcnt, 1); freecnt_applied = 1; } /* * Setup to remember how much we have not yet told * the peer our rwnd has opened up. Note we grab the * value from the tcb from last time. Note too that * sack sending clears this when a sack is sent, * which is fine. Once we hit the rwnd_req, we then * will go to the sctp_user_rcvd() that will not * lock until it KNOWs it MUST send a WUP-SACK. */ freed_so_far = stcb->freed_by_sorcv_sincelast; stcb->freed_by_sorcv_sincelast = 0; } } if (stcb && ((control->spec_flags & M_NOTIFICATION) == 0) && control->do_not_ref_stcb == 0) { stcb->asoc.strmin[control->sinfo_stream].delivery_started = 1; } /* First lets get off the sinfo and sockaddr info */ if ((sinfo != NULL) && (filling_sinfo != 0)) { sinfo->sinfo_stream = control->sinfo_stream; sinfo->sinfo_ssn = (uint16_t) control->sinfo_ssn; sinfo->sinfo_flags = control->sinfo_flags; sinfo->sinfo_ppid = control->sinfo_ppid; sinfo->sinfo_context = control->sinfo_context; sinfo->sinfo_timetolive = control->sinfo_timetolive; sinfo->sinfo_tsn = control->sinfo_tsn; sinfo->sinfo_cumtsn = control->sinfo_cumtsn; sinfo->sinfo_assoc_id = control->sinfo_assoc_id; nxt = TAILQ_NEXT(control, next); if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_EXT_RCVINFO) || sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVNXTINFO)) { struct sctp_extrcvinfo *s_extra; s_extra = (struct sctp_extrcvinfo *)sinfo; if ((nxt) && (nxt->length)) { s_extra->serinfo_next_flags = SCTP_NEXT_MSG_AVAIL; if (nxt->sinfo_flags & SCTP_UNORDERED) { s_extra->serinfo_next_flags |= SCTP_NEXT_MSG_IS_UNORDERED; } if (nxt->spec_flags & M_NOTIFICATION) { s_extra->serinfo_next_flags |= SCTP_NEXT_MSG_IS_NOTIFICATION; } s_extra->serinfo_next_aid = nxt->sinfo_assoc_id; s_extra->serinfo_next_length = nxt->length; s_extra->serinfo_next_ppid = nxt->sinfo_ppid; s_extra->serinfo_next_stream = nxt->sinfo_stream; if (nxt->tail_mbuf != NULL) { if (nxt->end_added) { s_extra->serinfo_next_flags |= SCTP_NEXT_MSG_ISCOMPLETE; } } } else { /* * we explicitly 0 this, since the memcpy * got some other things beyond the older * sinfo_ that is on the control's structure * :-D */ nxt = NULL; s_extra->serinfo_next_flags = SCTP_NO_NEXT_MSG; s_extra->serinfo_next_aid = 0; s_extra->serinfo_next_length = 0; s_extra->serinfo_next_ppid = 0; s_extra->serinfo_next_stream = 0; } } /* * update off the real current cum-ack, if we have an stcb. */ if ((control->do_not_ref_stcb == 0) && stcb) sinfo->sinfo_cumtsn = stcb->asoc.cumulative_tsn; /* * mask off the high bits, we keep the actual chunk bits in * there. */ sinfo->sinfo_flags &= 0x00ff; if ((control->sinfo_flags >> 8) & SCTP_DATA_UNORDERED) { sinfo->sinfo_flags |= SCTP_UNORDERED; } } #ifdef SCTP_ASOCLOG_OF_TSNS { int index, newindex; struct sctp_pcbtsn_rlog *entry; do { index = inp->readlog_index; newindex = index + 1; if (newindex >= SCTP_READ_LOG_SIZE) { newindex = 0; } } while (atomic_cmpset_int(&inp->readlog_index, index, newindex) == 0); entry = &inp->readlog[index]; entry->vtag = control->sinfo_assoc_id; entry->strm = control->sinfo_stream; entry->seq = control->sinfo_ssn; entry->sz = control->length; entry->flgs = control->sinfo_flags; } #endif if ((fromlen > 0) && (from != NULL)) { union sctp_sockstore store; size_t len; switch (control->whoFrom->ro._l_addr.sa.sa_family) { #ifdef INET6 case AF_INET6: len = sizeof(struct sockaddr_in6); store.sin6 = control->whoFrom->ro._l_addr.sin6; store.sin6.sin6_port = control->port_from; break; #endif #ifdef INET case AF_INET: #ifdef INET6 if (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_NEEDS_MAPPED_V4)) { len = sizeof(struct sockaddr_in6); in6_sin_2_v4mapsin6(&control->whoFrom->ro._l_addr.sin, &store.sin6); store.sin6.sin6_port = control->port_from; } else { len = sizeof(struct sockaddr_in); store.sin = control->whoFrom->ro._l_addr.sin; store.sin.sin_port = control->port_from; } #else len = sizeof(struct sockaddr_in); store.sin = control->whoFrom->ro._l_addr.sin; store.sin.sin_port = control->port_from; #endif break; #endif default: len = 0; break; } memcpy(from, &store, min((size_t)fromlen, len)); #ifdef INET6 { struct sockaddr_in6 lsa6, *from6; from6 = (struct sockaddr_in6 *)from; sctp_recover_scope_mac(from6, (&lsa6)); } #endif } /* now copy out what data we can */ if (mp == NULL) { /* copy out each mbuf in the chain up to length */ get_more_data: m = control->data; while (m) { /* Move out all we can */ cp_len = (int)uio->uio_resid; my_len = (int)SCTP_BUF_LEN(m); if (cp_len > my_len) { /* not enough in this buf */ cp_len = my_len; } if (hold_rlock) { SCTP_INP_READ_UNLOCK(inp); hold_rlock = 0; } if (cp_len > 0) error = uiomove(mtod(m, char *), cp_len, uio); /* re-read */ if (inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) { goto release; } if ((control->do_not_ref_stcb == 0) && stcb && stcb->asoc.state & SCTP_STATE_ABOUT_TO_BE_FREED) { no_rcv_needed = 1; } if (error) { /* error we are out of here */ goto release; } SCTP_INP_READ_LOCK(inp); hold_rlock = 1; if (cp_len == SCTP_BUF_LEN(m)) { if ((SCTP_BUF_NEXT(m) == NULL) && (control->end_added)) { out_flags |= MSG_EOR; if ((control->do_not_ref_stcb == 0) && (control->stcb != NULL) && ((control->spec_flags & M_NOTIFICATION) == 0)) control->stcb->asoc.strmin[control->sinfo_stream].delivery_started = 0; } if (control->spec_flags & M_NOTIFICATION) { out_flags |= MSG_NOTIFICATION; } /* we ate up the mbuf */ if (in_flags & MSG_PEEK) { /* just looking */ m = SCTP_BUF_NEXT(m); copied_so_far += cp_len; } else { /* dispose of the mbuf */ if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_SB_LOGGING_ENABLE) { sctp_sblog(&so->so_rcv, control->do_not_ref_stcb ? NULL : stcb, SCTP_LOG_SBFREE, SCTP_BUF_LEN(m)); } sctp_sbfree(control, stcb, &so->so_rcv, m); if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_SB_LOGGING_ENABLE) { sctp_sblog(&so->so_rcv, control->do_not_ref_stcb ? NULL : stcb, SCTP_LOG_SBRESULT, 0); } copied_so_far += cp_len; freed_so_far += cp_len; freed_so_far += MSIZE; atomic_subtract_int(&control->length, cp_len); control->data = sctp_m_free(m); m = control->data; /* * been through it all, must hold sb * lock ok to null tail */ if (control->data == NULL) { #ifdef INVARIANTS if ((control->end_added == 0) || (TAILQ_NEXT(control, next) == NULL)) { /* * If the end is not * added, OR the * next is NOT null * we MUST have the * lock. */ if (mtx_owned(&inp->inp_rdata_mtx) == 0) { panic("Hmm we don't own the lock?"); } } #endif control->tail_mbuf = NULL; #ifdef INVARIANTS if ((control->end_added) && ((out_flags & MSG_EOR) == 0)) { panic("end_added, nothing left and no MSG_EOR"); } #endif } } } else { /* Do we need to trim the mbuf? */ if (control->spec_flags & M_NOTIFICATION) { out_flags |= MSG_NOTIFICATION; } if ((in_flags & MSG_PEEK) == 0) { SCTP_BUF_RESV_UF(m, cp_len); SCTP_BUF_LEN(m) -= cp_len; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_SB_LOGGING_ENABLE) { sctp_sblog(&so->so_rcv, control->do_not_ref_stcb ? NULL : stcb, SCTP_LOG_SBFREE, cp_len); } atomic_subtract_int(&so->so_rcv.sb_cc, cp_len); if ((control->do_not_ref_stcb == 0) && stcb) { atomic_subtract_int(&stcb->asoc.sb_cc, cp_len); } copied_so_far += cp_len; freed_so_far += cp_len; freed_so_far += MSIZE; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_SB_LOGGING_ENABLE) { sctp_sblog(&so->so_rcv, control->do_not_ref_stcb ? NULL : stcb, SCTP_LOG_SBRESULT, 0); } atomic_subtract_int(&control->length, cp_len); } else { copied_so_far += cp_len; } } if ((out_flags & MSG_EOR) || (uio->uio_resid == 0)) { break; } if (((stcb) && (in_flags & MSG_PEEK) == 0) && (control->do_not_ref_stcb == 0) && (freed_so_far >= rwnd_req)) { sctp_user_rcvd(stcb, &freed_so_far, hold_rlock, rwnd_req); } } /* end while(m) */ /* * At this point we have looked at it all and we either have * a MSG_EOR/or read all the user wants... * control->length == 0. */ if ((out_flags & MSG_EOR) && ((in_flags & MSG_PEEK) == 0)) { /* we are done with this control */ if (control->length == 0) { if (control->data) { #ifdef INVARIANTS panic("control->data not null at read eor?"); #else SCTP_PRINTF("Strange, data left in the control buffer .. invarients would panic?\n"); sctp_m_freem(control->data); control->data = NULL; #endif } done_with_control: if (hold_rlock == 0) { SCTP_INP_READ_LOCK(inp); hold_rlock = 1; } TAILQ_REMOVE(&inp->read_queue, control, next); /* Add back any hiddend data */ if (control->held_length) { held_length = 0; control->held_length = 0; wakeup_read_socket = 1; } if (control->aux_data) { sctp_m_free(control->aux_data); control->aux_data = NULL; } no_rcv_needed = control->do_not_ref_stcb; sctp_free_remote_addr(control->whoFrom); control->data = NULL; #ifdef INVARIANTS if (control->on_strm_q) { panic("About to free ctl:%p so:%p and its in %d", control, so, control->on_strm_q); } #endif sctp_free_a_readq(stcb, control); control = NULL; if ((freed_so_far >= rwnd_req) && (no_rcv_needed == 0)) sctp_user_rcvd(stcb, &freed_so_far, hold_rlock, rwnd_req); } else { /* * The user did not read all of this * message, turn off the returned MSG_EOR * since we are leaving more behind on the * control to read. */ #ifdef INVARIANTS if (control->end_added && (control->data == NULL) && (control->tail_mbuf == NULL)) { panic("Gak, control->length is corrupt?"); } #endif no_rcv_needed = control->do_not_ref_stcb; out_flags &= ~MSG_EOR; } } if (out_flags & MSG_EOR) { goto release; } if ((uio->uio_resid == 0) || ((in_eeor_mode) && (copied_so_far >= (uint32_t) max(so->so_rcv.sb_lowat, 1)))) { goto release; } /* * If I hit here the receiver wants more and this message is * NOT done (pd-api). So two questions. Can we block? if not * we are done. Did the user NOT set MSG_WAITALL? */ if (block_allowed == 0) { goto release; } /* * We need to wait for more data a few things: - We don't * sbunlock() so we don't get someone else reading. - We * must be sure to account for the case where what is added * is NOT to our control when we wakeup. */ /* * Do we need to tell the transport a rwnd update might be * needed before we go to sleep? */ if (((stcb) && (in_flags & MSG_PEEK) == 0) && ((freed_so_far >= rwnd_req) && (control->do_not_ref_stcb == 0) && (no_rcv_needed == 0))) { sctp_user_rcvd(stcb, &freed_so_far, hold_rlock, rwnd_req); } wait_some_more: if (so->so_rcv.sb_state & SBS_CANTRCVMORE) { goto release; } if (inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE) goto release; if (hold_rlock == 1) { SCTP_INP_READ_UNLOCK(inp); hold_rlock = 0; } if (hold_sblock == 0) { SOCKBUF_LOCK(&so->so_rcv); hold_sblock = 1; } if ((copied_so_far) && (control->length == 0) && (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_FRAG_INTERLEAVE))) { goto release; } if (so->so_rcv.sb_cc <= control->held_length) { error = sbwait(&so->so_rcv); if (error) { goto release; } control->held_length = 0; } if (hold_sblock) { SOCKBUF_UNLOCK(&so->so_rcv); hold_sblock = 0; } if (control->length == 0) { /* still nothing here */ if (control->end_added == 1) { /* he aborted, or is done i.e.did a shutdown */ out_flags |= MSG_EOR; if (control->pdapi_aborted) { if ((control->do_not_ref_stcb == 0) && ((control->spec_flags & M_NOTIFICATION) == 0)) control->stcb->asoc.strmin[control->sinfo_stream].delivery_started = 0; out_flags |= MSG_TRUNC; } else { if ((control->do_not_ref_stcb == 0) && ((control->spec_flags & M_NOTIFICATION) == 0)) control->stcb->asoc.strmin[control->sinfo_stream].delivery_started = 0; } goto done_with_control; } if (so->so_rcv.sb_cc > held_length) { control->held_length = so->so_rcv.sb_cc; held_length = 0; } goto wait_some_more; } else if (control->data == NULL) { /* * we must re-sync since data is probably being * added */ SCTP_INP_READ_LOCK(inp); if ((control->length > 0) && (control->data == NULL)) { /* * big trouble.. we have the lock and its * corrupt? */ #ifdef INVARIANTS panic("Impossible data==NULL length !=0"); #endif out_flags |= MSG_EOR; out_flags |= MSG_TRUNC; control->length = 0; SCTP_INP_READ_UNLOCK(inp); goto done_with_control; } SCTP_INP_READ_UNLOCK(inp); /* We will fall around to get more data */ } goto get_more_data; } else { /*- * Give caller back the mbuf chain, * store in uio_resid the length */ wakeup_read_socket = 0; if ((control->end_added == 0) || (TAILQ_NEXT(control, next) == NULL)) { /* Need to get rlock */ if (hold_rlock == 0) { SCTP_INP_READ_LOCK(inp); hold_rlock = 1; } } if (control->end_added) { out_flags |= MSG_EOR; if ((control->do_not_ref_stcb == 0) && (control->stcb != NULL) && ((control->spec_flags & M_NOTIFICATION) == 0)) control->stcb->asoc.strmin[control->sinfo_stream].delivery_started = 0; } if (control->spec_flags & M_NOTIFICATION) { out_flags |= MSG_NOTIFICATION; } uio->uio_resid = control->length; *mp = control->data; m = control->data; while (m) { if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_SB_LOGGING_ENABLE) { sctp_sblog(&so->so_rcv, control->do_not_ref_stcb ? NULL : stcb, SCTP_LOG_SBFREE, SCTP_BUF_LEN(m)); } sctp_sbfree(control, stcb, &so->so_rcv, m); freed_so_far += SCTP_BUF_LEN(m); freed_so_far += MSIZE; if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_SB_LOGGING_ENABLE) { sctp_sblog(&so->so_rcv, control->do_not_ref_stcb ? NULL : stcb, SCTP_LOG_SBRESULT, 0); } m = SCTP_BUF_NEXT(m); } control->data = control->tail_mbuf = NULL; control->length = 0; if (out_flags & MSG_EOR) { /* Done with this control */ goto done_with_control; } } release: if (hold_rlock == 1) { SCTP_INP_READ_UNLOCK(inp); hold_rlock = 0; } if (hold_sblock == 1) { SOCKBUF_UNLOCK(&so->so_rcv); hold_sblock = 0; } sbunlock(&so->so_rcv); sockbuf_lock = 0; release_unlocked: if (hold_sblock) { SOCKBUF_UNLOCK(&so->so_rcv); hold_sblock = 0; } if ((stcb) && (in_flags & MSG_PEEK) == 0) { if ((freed_so_far >= rwnd_req) && (control && (control->do_not_ref_stcb == 0)) && (no_rcv_needed == 0)) sctp_user_rcvd(stcb, &freed_so_far, hold_rlock, rwnd_req); } out: if (msg_flags) { *msg_flags = out_flags; } if (((out_flags & MSG_EOR) == 0) && ((in_flags & MSG_PEEK) == 0) && (sinfo) && (sctp_is_feature_on(inp, SCTP_PCB_FLAGS_EXT_RCVINFO) || sctp_is_feature_on(inp, SCTP_PCB_FLAGS_RECVNXTINFO))) { struct sctp_extrcvinfo *s_extra; s_extra = (struct sctp_extrcvinfo *)sinfo; s_extra->serinfo_next_flags = SCTP_NO_NEXT_MSG; } if (hold_rlock == 1) { SCTP_INP_READ_UNLOCK(inp); } if (hold_sblock) { SOCKBUF_UNLOCK(&so->so_rcv); } if (sockbuf_lock) { sbunlock(&so->so_rcv); } if (freecnt_applied) { /* * The lock on the socket buffer protects us so the free * code will stop. But since we used the socketbuf lock and * the sender uses the tcb_lock to increment, we need to use * the atomic add to the refcnt. */ if (stcb == NULL) { #ifdef INVARIANTS panic("stcb for refcnt has gone NULL?"); goto stage_left; #else goto stage_left; #endif } atomic_add_int(&stcb->asoc.refcnt, -1); /* Save the value back for next time */ stcb->freed_by_sorcv_sincelast = freed_so_far; } if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_RECV_RWND_LOGGING_ENABLE) { if (stcb) { sctp_misc_ints(SCTP_SORECV_DONE, freed_so_far, (uint32_t) ((uio) ? (slen - uio->uio_resid) : slen), stcb->asoc.my_rwnd, so->so_rcv.sb_cc); } else { sctp_misc_ints(SCTP_SORECV_DONE, freed_so_far, (uint32_t) ((uio) ? (slen - uio->uio_resid) : slen), 0, so->so_rcv.sb_cc); } } stage_left: if (wakeup_read_socket) { sctp_sorwakeup(inp, so); } return (error); } #ifdef SCTP_MBUF_LOGGING struct mbuf * sctp_m_free(struct mbuf *m) { if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_MBUF_LOGGING_ENABLE) { sctp_log_mb(m, SCTP_MBUF_IFREE); } return (m_free(m)); } void sctp_m_freem(struct mbuf *mb) { while (mb != NULL) mb = sctp_m_free(mb); } #endif int sctp_dynamic_set_primary(struct sockaddr *sa, uint32_t vrf_id) { /* * Given a local address. For all associations that holds the * address, request a peer-set-primary. */ struct sctp_ifa *ifa; struct sctp_laddr *wi; ifa = sctp_find_ifa_by_addr(sa, vrf_id, 0); if (ifa == NULL) { SCTP_LTRACE_ERR_RET(NULL, NULL, NULL, SCTP_FROM_SCTPUTIL, EADDRNOTAVAIL); return (EADDRNOTAVAIL); } /* * Now that we have the ifa we must awaken the iterator with this * message. */ wi = SCTP_ZONE_GET(SCTP_BASE_INFO(ipi_zone_laddr), struct sctp_laddr); if (wi == NULL) { SCTP_LTRACE_ERR_RET(NULL, NULL, NULL, SCTP_FROM_SCTPUTIL, ENOMEM); return (ENOMEM); } /* Now incr the count and int wi structure */ SCTP_INCR_LADDR_COUNT(); bzero(wi, sizeof(*wi)); (void)SCTP_GETTIME_TIMEVAL(&wi->start_time); wi->ifa = ifa; wi->action = SCTP_SET_PRIM_ADDR; atomic_add_int(&ifa->refcount, 1); /* Now add it to the work queue */ SCTP_WQ_ADDR_LOCK(); /* * Should this really be a tailq? As it is we will process the * newest first :-0 */ LIST_INSERT_HEAD(&SCTP_BASE_INFO(addr_wq), wi, sctp_nxt_addr); SCTP_WQ_ADDR_UNLOCK(); sctp_timer_start(SCTP_TIMER_TYPE_ADDR_WQ, (struct sctp_inpcb *)NULL, (struct sctp_tcb *)NULL, (struct sctp_nets *)NULL); return (0); } int sctp_soreceive(struct socket *so, struct sockaddr **psa, struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, int *flagsp) { int error, fromlen; uint8_t sockbuf[256]; struct sockaddr *from; struct sctp_extrcvinfo sinfo; int filling_sinfo = 1; struct sctp_inpcb *inp; inp = (struct sctp_inpcb *)so->so_pcb; /* pickup the assoc we are reading from */ if (inp == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); return (EINVAL); } if ((sctp_is_feature_off(inp, SCTP_PCB_FLAGS_RECVDATAIOEVNT) && sctp_is_feature_off(inp, SCTP_PCB_FLAGS_RECVRCVINFO) && sctp_is_feature_off(inp, SCTP_PCB_FLAGS_RECVNXTINFO)) || (controlp == NULL)) { /* user does not want the sndrcv ctl */ filling_sinfo = 0; } if (psa) { from = (struct sockaddr *)sockbuf; fromlen = sizeof(sockbuf); from->sa_len = 0; } else { from = NULL; fromlen = 0; } if (filling_sinfo) { memset(&sinfo, 0, sizeof(struct sctp_extrcvinfo)); } error = sctp_sorecvmsg(so, uio, mp0, from, fromlen, flagsp, (struct sctp_sndrcvinfo *)&sinfo, filling_sinfo); if (controlp != NULL) { /* copy back the sinfo in a CMSG format */ if (filling_sinfo) *controlp = sctp_build_ctl_nchunk(inp, (struct sctp_sndrcvinfo *)&sinfo); else *controlp = NULL; } if (psa) { /* copy back the address info */ if (from && from->sa_len) { *psa = sodupsockaddr(from, M_NOWAIT); } else { *psa = NULL; } } return (error); } int sctp_connectx_helper_add(struct sctp_tcb *stcb, struct sockaddr *addr, int totaddr, int *error) { int added = 0; int i; struct sctp_inpcb *inp; struct sockaddr *sa; size_t incr = 0; #ifdef INET struct sockaddr_in *sin; #endif #ifdef INET6 struct sockaddr_in6 *sin6; #endif sa = addr; inp = stcb->sctp_ep; *error = 0; for (i = 0; i < totaddr; i++) { switch (sa->sa_family) { #ifdef INET case AF_INET: incr = sizeof(struct sockaddr_in); sin = (struct sockaddr_in *)sa; if ((sin->sin_addr.s_addr == INADDR_ANY) || (sin->sin_addr.s_addr == INADDR_BROADCAST) || IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) { SCTP_LTRACE_ERR_RET(NULL, stcb, NULL, SCTP_FROM_SCTPUTIL, EINVAL); (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTPUTIL + SCTP_LOC_7); *error = EINVAL; goto out_now; } if (sctp_add_remote_addr(stcb, sa, NULL, stcb->asoc.port, SCTP_DONOT_SETSCOPE, SCTP_ADDR_IS_CONFIRMED)) { /* assoc gone no un-lock */ SCTP_LTRACE_ERR_RET(NULL, stcb, NULL, SCTP_FROM_SCTPUTIL, ENOBUFS); (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTPUTIL + SCTP_LOC_8); *error = ENOBUFS; goto out_now; } added++; break; #endif #ifdef INET6 case AF_INET6: incr = sizeof(struct sockaddr_in6); sin6 = (struct sockaddr_in6 *)sa; if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) || IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) { SCTP_LTRACE_ERR_RET(NULL, stcb, NULL, SCTP_FROM_SCTPUTIL, EINVAL); (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTPUTIL + SCTP_LOC_9); *error = EINVAL; goto out_now; } if (sctp_add_remote_addr(stcb, sa, NULL, stcb->asoc.port, SCTP_DONOT_SETSCOPE, SCTP_ADDR_IS_CONFIRMED)) { /* assoc gone no un-lock */ SCTP_LTRACE_ERR_RET(NULL, stcb, NULL, SCTP_FROM_SCTPUTIL, ENOBUFS); (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTPUTIL + SCTP_LOC_10); *error = ENOBUFS; goto out_now; } added++; break; #endif default: break; } sa = (struct sockaddr *)((caddr_t)sa + incr); } out_now: return (added); } struct sctp_tcb * sctp_connectx_helper_find(struct sctp_inpcb *inp, struct sockaddr *addr, unsigned int *totaddr, unsigned int *num_v4, unsigned int *num_v6, int *error, unsigned int limit, int *bad_addr) { struct sockaddr *sa; struct sctp_tcb *stcb = NULL; unsigned int incr, at, i; at = incr = 0; sa = addr; *error = *num_v6 = *num_v4 = 0; /* account and validate addresses */ for (i = 0; i < *totaddr; i++) { switch (sa->sa_family) { #ifdef INET case AF_INET: if (sa->sa_len != incr) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; *bad_addr = 1; return (NULL); } (*num_v4) += 1; incr = (unsigned int)sizeof(struct sockaddr_in); break; #endif #ifdef INET6 case AF_INET6: { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)sa; if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { /* Must be non-mapped for connectx */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; *bad_addr = 1; return (NULL); } if (sa->sa_len != incr) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; *bad_addr = 1; return (NULL); } (*num_v6) += 1; incr = (unsigned int)sizeof(struct sockaddr_in6); break; } #endif default: *totaddr = i; /* we are done */ break; } if (i == *totaddr) { break; } SCTP_INP_INCR_REF(inp); stcb = sctp_findassociation_ep_addr(&inp, sa, NULL, NULL, NULL); if (stcb != NULL) { /* Already have or am bring up an association */ return (stcb); } else { SCTP_INP_DECR_REF(inp); } if ((at + incr) > limit) { *totaddr = i; break; } sa = (struct sockaddr *)((caddr_t)sa + incr); } return ((struct sctp_tcb *)NULL); } /* * sctp_bindx(ADD) for one address. * assumes all arguments are valid/checked by caller. */ void sctp_bindx_add_address(struct socket *so, struct sctp_inpcb *inp, struct sockaddr *sa, sctp_assoc_t assoc_id, uint32_t vrf_id, int *error, void *p) { struct sockaddr *addr_touse; #if defined(INET) && defined(INET6) struct sockaddr_in sin; #endif /* see if we're bound all already! */ if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; return; } addr_touse = sa; #ifdef INET6 if (sa->sa_family == AF_INET6) { #ifdef INET struct sockaddr_in6 *sin6; #endif if (sa->sa_len != sizeof(struct sockaddr_in6)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; return; } if ((inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) == 0) { /* can only bind v6 on PF_INET6 sockets */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; return; } #ifdef INET sin6 = (struct sockaddr_in6 *)addr_touse; if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { if ((inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) && SCTP_IPV6_V6ONLY(inp)) { /* can't bind v4-mapped on PF_INET sockets */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; return; } in6_sin6_2_sin(&sin, sin6); addr_touse = (struct sockaddr *)&sin; } #endif } #endif #ifdef INET if (sa->sa_family == AF_INET) { if (sa->sa_len != sizeof(struct sockaddr_in)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; return; } if ((inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) && SCTP_IPV6_V6ONLY(inp)) { /* can't bind v4 on PF_INET sockets */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; return; } } #endif if (inp->sctp_flags & SCTP_PCB_FLAGS_UNBOUND) { if (p == NULL) { /* Can't get proc for Net/Open BSD */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; return; } *error = sctp_inpcb_bind(so, addr_touse, NULL, p); return; } /* * No locks required here since bind and mgmt_ep_sa all do their own * locking. If we do something for the FIX: below we may need to * lock in that case. */ if (assoc_id == 0) { /* add the address */ struct sctp_inpcb *lep; struct sockaddr_in *lsin = (struct sockaddr_in *)addr_touse; /* validate the incoming port */ if ((lsin->sin_port != 0) && (lsin->sin_port != inp->sctp_lport)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; return; } else { /* user specified 0 port, set it to existing port */ lsin->sin_port = inp->sctp_lport; } lep = sctp_pcb_findep(addr_touse, 1, 0, vrf_id); if (lep != NULL) { /* * We must decrement the refcount since we have the * ep already and are binding. No remove going on * here. */ SCTP_INP_DECR_REF(lep); } if (lep == inp) { /* already bound to it.. ok */ return; } else if (lep == NULL) { ((struct sockaddr_in *)addr_touse)->sin_port = 0; *error = sctp_addr_mgmt_ep_sa(inp, addr_touse, SCTP_ADD_IP_ADDRESS, vrf_id, NULL); } else { *error = EADDRINUSE; } if (*error) return; } else { /* * FIX: decide whether we allow assoc based bindx */ } } /* * sctp_bindx(DELETE) for one address. * assumes all arguments are valid/checked by caller. */ void sctp_bindx_delete_address(struct sctp_inpcb *inp, struct sockaddr *sa, sctp_assoc_t assoc_id, uint32_t vrf_id, int *error) { struct sockaddr *addr_touse; #if defined(INET) && defined(INET6) struct sockaddr_in sin; #endif /* see if we're bound all already! */ if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; return; } addr_touse = sa; #ifdef INET6 if (sa->sa_family == AF_INET6) { #ifdef INET struct sockaddr_in6 *sin6; #endif if (sa->sa_len != sizeof(struct sockaddr_in6)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; return; } if ((inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) == 0) { /* can only bind v6 on PF_INET6 sockets */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; return; } #ifdef INET sin6 = (struct sockaddr_in6 *)addr_touse; if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { if ((inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) && SCTP_IPV6_V6ONLY(inp)) { /* can't bind mapped-v4 on PF_INET sockets */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; return; } in6_sin6_2_sin(&sin, sin6); addr_touse = (struct sockaddr *)&sin; } #endif } #endif #ifdef INET if (sa->sa_family == AF_INET) { if (sa->sa_len != sizeof(struct sockaddr_in)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; return; } if ((inp->sctp_flags & SCTP_PCB_FLAGS_BOUND_V6) && SCTP_IPV6_V6ONLY(inp)) { /* can't bind v4 on PF_INET sockets */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTPUTIL, EINVAL); *error = EINVAL; return; } } #endif /* * No lock required mgmt_ep_sa does its own locking. If the FIX: * below is ever changed we may need to lock before calling * association level binding. */ if (assoc_id == 0) { /* delete the address */ *error = sctp_addr_mgmt_ep_sa(inp, addr_touse, SCTP_DEL_IP_ADDRESS, vrf_id, NULL); } else { /* * FIX: decide whether we allow assoc based bindx */ } } /* * returns the valid local address count for an assoc, taking into account * all scoping rules */ int sctp_local_addr_count(struct sctp_tcb *stcb) { int loopback_scope; #if defined(INET) int ipv4_local_scope, ipv4_addr_legal; #endif #if defined (INET6) int local_scope, site_scope, ipv6_addr_legal; #endif struct sctp_vrf *vrf; struct sctp_ifn *sctp_ifn; struct sctp_ifa *sctp_ifa; int count = 0; /* Turn on all the appropriate scopes */ loopback_scope = stcb->asoc.scope.loopback_scope; #if defined(INET) ipv4_local_scope = stcb->asoc.scope.ipv4_local_scope; ipv4_addr_legal = stcb->asoc.scope.ipv4_addr_legal; #endif #if defined(INET6) local_scope = stcb->asoc.scope.local_scope; site_scope = stcb->asoc.scope.site_scope; ipv6_addr_legal = stcb->asoc.scope.ipv6_addr_legal; #endif SCTP_IPI_ADDR_RLOCK(); vrf = sctp_find_vrf(stcb->asoc.vrf_id); if (vrf == NULL) { /* no vrf, no addresses */ SCTP_IPI_ADDR_RUNLOCK(); return (0); } if (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) { /* * bound all case: go through all ifns on the vrf */ LIST_FOREACH(sctp_ifn, &vrf->ifnlist, next_ifn) { if ((loopback_scope == 0) && SCTP_IFN_IS_IFT_LOOP(sctp_ifn)) { continue; } LIST_FOREACH(sctp_ifa, &sctp_ifn->ifalist, next_ifa) { if (sctp_is_addr_restricted(stcb, sctp_ifa)) continue; switch (sctp_ifa->address.sa.sa_family) { #ifdef INET case AF_INET: if (ipv4_addr_legal) { struct sockaddr_in *sin; sin = &sctp_ifa->address.sin; if (sin->sin_addr.s_addr == 0) { /* * skip unspecified * addrs */ continue; } if (prison_check_ip4(stcb->sctp_ep->ip_inp.inp.inp_cred, &sin->sin_addr) != 0) { continue; } if ((ipv4_local_scope == 0) && (IN4_ISPRIVATE_ADDRESS(&sin->sin_addr))) { continue; } /* count this one */ count++; } else { continue; } break; #endif #ifdef INET6 case AF_INET6: if (ipv6_addr_legal) { struct sockaddr_in6 *sin6; sin6 = &sctp_ifa->address.sin6; if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { continue; } if (prison_check_ip6(stcb->sctp_ep->ip_inp.inp.inp_cred, &sin6->sin6_addr) != 0) { continue; } if (IN6_IS_ADDR_LINKLOCAL(&sin6->sin6_addr)) { if (local_scope == 0) continue; if (sin6->sin6_scope_id == 0) { if (sa6_recoverscope(sin6) != 0) /* * * bad * * li * nk * * loc * al * * add * re * ss * */ continue; } } if ((site_scope == 0) && (IN6_IS_ADDR_SITELOCAL(&sin6->sin6_addr))) { continue; } /* count this one */ count++; } break; #endif default: /* TSNH */ break; } } } } else { /* * subset bound case */ struct sctp_laddr *laddr; LIST_FOREACH(laddr, &stcb->sctp_ep->sctp_addr_list, sctp_nxt_addr) { if (sctp_is_addr_restricted(stcb, laddr->ifa)) { continue; } /* count this one */ count++; } } SCTP_IPI_ADDR_RUNLOCK(); return (count); } #if defined(SCTP_LOCAL_TRACE_BUF) void sctp_log_trace(uint32_t subsys, const char *str SCTP_UNUSED, uint32_t a, uint32_t b, uint32_t c, uint32_t d, uint32_t e, uint32_t f) { uint32_t saveindex, newindex; do { saveindex = SCTP_BASE_SYSCTL(sctp_log).index; if (saveindex >= SCTP_MAX_LOGGING_SIZE) { newindex = 1; } else { newindex = saveindex + 1; } } while (atomic_cmpset_int(&SCTP_BASE_SYSCTL(sctp_log).index, saveindex, newindex) == 0); if (saveindex >= SCTP_MAX_LOGGING_SIZE) { saveindex = 0; } SCTP_BASE_SYSCTL(sctp_log).entry[saveindex].timestamp = SCTP_GET_CYCLECOUNT; SCTP_BASE_SYSCTL(sctp_log).entry[saveindex].subsys = subsys; SCTP_BASE_SYSCTL(sctp_log).entry[saveindex].params[0] = a; SCTP_BASE_SYSCTL(sctp_log).entry[saveindex].params[1] = b; SCTP_BASE_SYSCTL(sctp_log).entry[saveindex].params[2] = c; SCTP_BASE_SYSCTL(sctp_log).entry[saveindex].params[3] = d; SCTP_BASE_SYSCTL(sctp_log).entry[saveindex].params[4] = e; SCTP_BASE_SYSCTL(sctp_log).entry[saveindex].params[5] = f; } #endif static void sctp_recv_udp_tunneled_packet(struct mbuf *m, int off, struct inpcb *inp, const struct sockaddr *sa SCTP_UNUSED, void *ctx SCTP_UNUSED) { struct ip *iph; #ifdef INET6 struct ip6_hdr *ip6; #endif struct mbuf *sp, *last; struct udphdr *uhdr; uint16_t port; if ((m->m_flags & M_PKTHDR) == 0) { /* Can't handle one that is not a pkt hdr */ goto out; } /* Pull the src port */ iph = mtod(m, struct ip *); uhdr = (struct udphdr *)((caddr_t)iph + off); port = uhdr->uh_sport; /* * Split out the mbuf chain. Leave the IP header in m, place the * rest in the sp. */ sp = m_split(m, off, M_NOWAIT); if (sp == NULL) { /* Gak, drop packet, we can't do a split */ goto out; } if (sp->m_pkthdr.len < sizeof(struct udphdr) + sizeof(struct sctphdr)) { /* Gak, packet can't have an SCTP header in it - too small */ m_freem(sp); goto out; } /* Now pull up the UDP header and SCTP header together */ sp = m_pullup(sp, sizeof(struct udphdr) + sizeof(struct sctphdr)); if (sp == NULL) { /* Gak pullup failed */ goto out; } /* Trim out the UDP header */ m_adj(sp, sizeof(struct udphdr)); /* Now reconstruct the mbuf chain */ for (last = m; last->m_next; last = last->m_next); last->m_next = sp; m->m_pkthdr.len += sp->m_pkthdr.len; /* * The CSUM_DATA_VALID flags indicates that the HW checked the UDP * checksum and it was valid. Since CSUM_DATA_VALID == * CSUM_SCTP_VALID this would imply that the HW also verified the * SCTP checksum. Therefore, clear the bit. */ SCTPDBG(SCTP_DEBUG_CRCOFFLOAD, "sctp_recv_udp_tunneled_packet(): Packet of length %d received on %s with csum_flags 0x%b.\n", m->m_pkthdr.len, if_name(m->m_pkthdr.rcvif), (int)m->m_pkthdr.csum_flags, CSUM_BITS); m->m_pkthdr.csum_flags &= ~CSUM_DATA_VALID; iph = mtod(m, struct ip *); switch (iph->ip_v) { #ifdef INET case IPVERSION: iph->ip_len = htons(ntohs(iph->ip_len) - sizeof(struct udphdr)); sctp_input_with_port(m, off, port); break; #endif #ifdef INET6 case IPV6_VERSION >> 4: ip6 = mtod(m, struct ip6_hdr *); ip6->ip6_plen = htons(ntohs(ip6->ip6_plen) - sizeof(struct udphdr)); sctp6_input_with_port(&m, &off, port); break; #endif default: goto out; break; } return; out: m_freem(m); } #ifdef INET static void sctp_recv_icmp_tunneled_packet(int cmd, struct sockaddr *sa, void *vip, void *ctx SCTP_UNUSED) { struct ip *outer_ip, *inner_ip; struct sctphdr *sh; struct icmp *icmp; struct udphdr *udp; struct sctp_inpcb *inp; struct sctp_tcb *stcb; struct sctp_nets *net; struct sctp_init_chunk *ch; struct sockaddr_in src, dst; uint8_t type, code; inner_ip = (struct ip *)vip; icmp = (struct icmp *)((caddr_t)inner_ip - (sizeof(struct icmp) - sizeof(struct ip))); outer_ip = (struct ip *)((caddr_t)icmp - sizeof(struct ip)); if (ntohs(outer_ip->ip_len) < sizeof(struct ip) + 8 + (inner_ip->ip_hl << 2) + sizeof(struct udphdr) + 8) { return; } udp = (struct udphdr *)((caddr_t)inner_ip + (inner_ip->ip_hl << 2)); sh = (struct sctphdr *)(udp + 1); memset(&src, 0, sizeof(struct sockaddr_in)); src.sin_family = AF_INET; src.sin_len = sizeof(struct sockaddr_in); src.sin_port = sh->src_port; src.sin_addr = inner_ip->ip_src; memset(&dst, 0, sizeof(struct sockaddr_in)); dst.sin_family = AF_INET; dst.sin_len = sizeof(struct sockaddr_in); dst.sin_port = sh->dest_port; dst.sin_addr = inner_ip->ip_dst; /* * 'dst' holds the dest of the packet that failed to be sent. 'src' * holds our local endpoint address. Thus we reverse the dst and the * src in the lookup. */ inp = NULL; net = NULL; stcb = sctp_findassociation_addr_sa((struct sockaddr *)&dst, (struct sockaddr *)&src, &inp, &net, 1, SCTP_DEFAULT_VRFID); if ((stcb != NULL) && (net != NULL) && - (inp != NULL) && - (inp->sctp_socket != NULL)) { + (inp != NULL)) { /* Check the UDP port numbers */ if ((udp->uh_dport != net->port) || (udp->uh_sport != htons(SCTP_BASE_SYSCTL(sctp_udp_tunneling_port)))) { SCTP_TCB_UNLOCK(stcb); return; } /* Check the verification tag */ if (ntohl(sh->v_tag) != 0) { /* * This must be the verification tag used for * sending out packets. We don't consider packets * reflecting the verification tag. */ if (ntohl(sh->v_tag) != stcb->asoc.peer_vtag) { SCTP_TCB_UNLOCK(stcb); return; } } else { if (ntohs(outer_ip->ip_len) >= sizeof(struct ip) + 8 + (inner_ip->ip_hl << 2) + 8 + 20) { /* * In this case we can check if we got an * INIT chunk and if the initiate tag * matches. */ ch = (struct sctp_init_chunk *)(sh + 1); if ((ch->ch.chunk_type != SCTP_INITIATION) || (ntohl(ch->init.initiate_tag) != stcb->asoc.my_vtag)) { SCTP_TCB_UNLOCK(stcb); return; } } else { SCTP_TCB_UNLOCK(stcb); return; } } type = icmp->icmp_type; code = icmp->icmp_code; if ((type == ICMP_UNREACH) && (code == ICMP_UNREACH_PORT)) { code = ICMP_UNREACH_PROTOCOL; } sctp_notify(inp, stcb, net, type, code, ntohs(inner_ip->ip_len), ntohs(icmp->icmp_nextmtu)); } else { if ((stcb == NULL) && (inp != NULL)) { /* reduce ref-count */ SCTP_INP_WLOCK(inp); SCTP_INP_DECR_REF(inp); SCTP_INP_WUNLOCK(inp); } if (stcb) { SCTP_TCB_UNLOCK(stcb); } } return; } #endif #ifdef INET6 static void sctp_recv_icmp6_tunneled_packet(int cmd, struct sockaddr *sa, void *d, void *ctx SCTP_UNUSED) { struct ip6ctlparam *ip6cp; struct sctp_inpcb *inp; struct sctp_tcb *stcb; struct sctp_nets *net; struct sctphdr sh; struct udphdr udp; struct sockaddr_in6 src, dst; uint8_t type, code; ip6cp = (struct ip6ctlparam *)d; /* * XXX: We assume that when IPV6 is non NULL, M and OFF are valid. */ if (ip6cp->ip6c_m == NULL) { return; } /* * Check if we can safely examine the ports and the verification tag * of the SCTP common header. */ if (ip6cp->ip6c_m->m_pkthdr.len < ip6cp->ip6c_off + sizeof(struct udphdr) + offsetof(struct sctphdr, checksum)) { return; } /* Copy out the UDP header. */ memset(&udp, 0, sizeof(struct udphdr)); m_copydata(ip6cp->ip6c_m, ip6cp->ip6c_off, sizeof(struct udphdr), (caddr_t)&udp); /* Copy out the port numbers and the verification tag. */ memset(&sh, 0, sizeof(struct sctphdr)); m_copydata(ip6cp->ip6c_m, ip6cp->ip6c_off + sizeof(struct udphdr), sizeof(uint16_t) + sizeof(uint16_t) + sizeof(uint32_t), (caddr_t)&sh); memset(&src, 0, sizeof(struct sockaddr_in6)); src.sin6_family = AF_INET6; src.sin6_len = sizeof(struct sockaddr_in6); src.sin6_port = sh.src_port; src.sin6_addr = ip6cp->ip6c_ip6->ip6_src; if (in6_setscope(&src.sin6_addr, ip6cp->ip6c_m->m_pkthdr.rcvif, NULL) != 0) { return; } memset(&dst, 0, sizeof(struct sockaddr_in6)); dst.sin6_family = AF_INET6; dst.sin6_len = sizeof(struct sockaddr_in6); dst.sin6_port = sh.dest_port; dst.sin6_addr = ip6cp->ip6c_ip6->ip6_dst; if (in6_setscope(&dst.sin6_addr, ip6cp->ip6c_m->m_pkthdr.rcvif, NULL) != 0) { return; } inp = NULL; net = NULL; stcb = sctp_findassociation_addr_sa((struct sockaddr *)&dst, (struct sockaddr *)&src, &inp, &net, 1, SCTP_DEFAULT_VRFID); if ((stcb != NULL) && (net != NULL) && - (inp != NULL) && - (inp->sctp_socket != NULL)) { + (inp != NULL)) { /* Check the UDP port numbers */ if ((udp.uh_dport != net->port) || (udp.uh_sport != htons(SCTP_BASE_SYSCTL(sctp_udp_tunneling_port)))) { SCTP_TCB_UNLOCK(stcb); return; } /* Check the verification tag */ if (ntohl(sh.v_tag) != 0) { /* * This must be the verification tag used for * sending out packets. We don't consider packets * reflecting the verification tag. */ if (ntohl(sh.v_tag) != stcb->asoc.peer_vtag) { SCTP_TCB_UNLOCK(stcb); return; } } else { if (ip6cp->ip6c_m->m_pkthdr.len >= ip6cp->ip6c_off + sizeof(struct udphdr) + sizeof(struct sctphdr) + sizeof(struct sctp_chunkhdr) + offsetof(struct sctp_init, a_rwnd)) { /* * In this case we can check if we got an * INIT chunk and if the initiate tag * matches. */ uint32_t initiate_tag; uint8_t chunk_type; m_copydata(ip6cp->ip6c_m, ip6cp->ip6c_off + sizeof(struct udphdr) + sizeof(struct sctphdr), sizeof(uint8_t), (caddr_t)&chunk_type); m_copydata(ip6cp->ip6c_m, ip6cp->ip6c_off + sizeof(struct udphdr) + sizeof(struct sctphdr) + sizeof(struct sctp_chunkhdr), sizeof(uint32_t), (caddr_t)&initiate_tag); if ((chunk_type != SCTP_INITIATION) || (ntohl(initiate_tag) != stcb->asoc.my_vtag)) { SCTP_TCB_UNLOCK(stcb); return; } } else { SCTP_TCB_UNLOCK(stcb); return; } } type = ip6cp->ip6c_icmp6->icmp6_type; code = ip6cp->ip6c_icmp6->icmp6_code; if ((type == ICMP6_DST_UNREACH) && (code == ICMP6_DST_UNREACH_NOPORT)) { type = ICMP6_PARAM_PROB; code = ICMP6_PARAMPROB_NEXTHEADER; } sctp6_notify(inp, stcb, net, type, code, (uint16_t) ntohl(ip6cp->ip6c_icmp6->icmp6_mtu)); } else { if ((stcb == NULL) && (inp != NULL)) { /* reduce inp's ref-count */ SCTP_INP_WLOCK(inp); SCTP_INP_DECR_REF(inp); SCTP_INP_WUNLOCK(inp); } if (stcb) { SCTP_TCB_UNLOCK(stcb); } } } #endif void sctp_over_udp_stop(void) { /* * This function assumes sysctl caller holds sctp_sysctl_info_lock() * for writting! */ #ifdef INET if (SCTP_BASE_INFO(udp4_tun_socket) != NULL) { soclose(SCTP_BASE_INFO(udp4_tun_socket)); SCTP_BASE_INFO(udp4_tun_socket) = NULL; } #endif #ifdef INET6 if (SCTP_BASE_INFO(udp6_tun_socket) != NULL) { soclose(SCTP_BASE_INFO(udp6_tun_socket)); SCTP_BASE_INFO(udp6_tun_socket) = NULL; } #endif } int sctp_over_udp_start(void) { uint16_t port; int ret; #ifdef INET struct sockaddr_in sin; #endif #ifdef INET6 struct sockaddr_in6 sin6; #endif /* * This function assumes sysctl caller holds sctp_sysctl_info_lock() * for writting! */ port = SCTP_BASE_SYSCTL(sctp_udp_tunneling_port); if (ntohs(port) == 0) { /* Must have a port set */ return (EINVAL); } #ifdef INET if (SCTP_BASE_INFO(udp4_tun_socket) != NULL) { /* Already running -- must stop first */ return (EALREADY); } #endif #ifdef INET6 if (SCTP_BASE_INFO(udp6_tun_socket) != NULL) { /* Already running -- must stop first */ return (EALREADY); } #endif #ifdef INET if ((ret = socreate(PF_INET, &SCTP_BASE_INFO(udp4_tun_socket), SOCK_DGRAM, IPPROTO_UDP, curthread->td_ucred, curthread))) { sctp_over_udp_stop(); return (ret); } /* Call the special UDP hook. */ if ((ret = udp_set_kernel_tunneling(SCTP_BASE_INFO(udp4_tun_socket), sctp_recv_udp_tunneled_packet, sctp_recv_icmp_tunneled_packet, NULL))) { sctp_over_udp_stop(); return (ret); } /* Ok, we have a socket, bind it to the port. */ memset(&sin, 0, sizeof(struct sockaddr_in)); sin.sin_len = sizeof(struct sockaddr_in); sin.sin_family = AF_INET; sin.sin_port = htons(port); if ((ret = sobind(SCTP_BASE_INFO(udp4_tun_socket), (struct sockaddr *)&sin, curthread))) { sctp_over_udp_stop(); return (ret); } #endif #ifdef INET6 if ((ret = socreate(PF_INET6, &SCTP_BASE_INFO(udp6_tun_socket), SOCK_DGRAM, IPPROTO_UDP, curthread->td_ucred, curthread))) { sctp_over_udp_stop(); return (ret); } /* Call the special UDP hook. */ if ((ret = udp_set_kernel_tunneling(SCTP_BASE_INFO(udp6_tun_socket), sctp_recv_udp_tunneled_packet, sctp_recv_icmp6_tunneled_packet, NULL))) { sctp_over_udp_stop(); return (ret); } /* Ok, we have a socket, bind it to the port. */ memset(&sin6, 0, sizeof(struct sockaddr_in6)); sin6.sin6_len = sizeof(struct sockaddr_in6); sin6.sin6_family = AF_INET6; sin6.sin6_port = htons(port); if ((ret = sobind(SCTP_BASE_INFO(udp6_tun_socket), (struct sockaddr *)&sin6, curthread))) { sctp_over_udp_stop(); return (ret); } #endif return (0); } Index: projects/vnet/sys/netinet/tcp_subr.c =================================================================== --- projects/vnet/sys/netinet/tcp_subr.c (revision 302157) +++ projects/vnet/sys/netinet/tcp_subr.c (revision 302158) @@ -1,3008 +1,3010 @@ /*- * 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. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)tcp_subr.c 8.2 (Berkeley) 5/24/95 */ #include __FBSDID("$FreeBSD$"); #include "opt_compat.h" #include "opt_inet.h" #include "opt_inet6.h" #include "opt_ipsec.h" #include "opt_tcpdebug.h" #include #include #include #include #include #include #include #include #include #include #include #include #ifdef INET6 #include #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef INET6 #include #include #include #include #include #include #endif #ifdef TCP_RFC7413 #include #endif #include #include #include #include #include #include #include #ifdef INET6 #include #endif #include #ifdef TCPPCAP #include #endif #ifdef TCPDEBUG #include #endif #ifdef INET6 #include #endif #ifdef TCP_OFFLOAD #include #endif #ifdef IPSEC #include #include #ifdef INET6 #include #endif #include #include #endif /*IPSEC*/ #include #include #include VNET_DEFINE(int, tcp_mssdflt) = TCP_MSS; #ifdef INET6 VNET_DEFINE(int, tcp_v6mssdflt) = TCP6_MSS; #endif struct rwlock tcp_function_lock; static int sysctl_net_inet_tcp_mss_check(SYSCTL_HANDLER_ARGS) { int error, new; new = V_tcp_mssdflt; error = sysctl_handle_int(oidp, &new, 0, req); if (error == 0 && req->newptr) { if (new < TCP_MINMSS) error = EINVAL; else V_tcp_mssdflt = new; } return (error); } SYSCTL_PROC(_net_inet_tcp, TCPCTL_MSSDFLT, mssdflt, CTLFLAG_VNET | CTLTYPE_INT | CTLFLAG_RW, &VNET_NAME(tcp_mssdflt), 0, &sysctl_net_inet_tcp_mss_check, "I", "Default TCP Maximum Segment Size"); #ifdef INET6 static int sysctl_net_inet_tcp_mss_v6_check(SYSCTL_HANDLER_ARGS) { int error, new; new = V_tcp_v6mssdflt; error = sysctl_handle_int(oidp, &new, 0, req); if (error == 0 && req->newptr) { if (new < TCP_MINMSS) error = EINVAL; else V_tcp_v6mssdflt = new; } return (error); } SYSCTL_PROC(_net_inet_tcp, TCPCTL_V6MSSDFLT, v6mssdflt, CTLFLAG_VNET | CTLTYPE_INT | CTLFLAG_RW, &VNET_NAME(tcp_v6mssdflt), 0, &sysctl_net_inet_tcp_mss_v6_check, "I", "Default TCP Maximum Segment Size for IPv6"); #endif /* INET6 */ /* * Minimum MSS we accept and use. This prevents DoS attacks where * we are forced to a ridiculous low MSS like 20 and send hundreds * of packets instead of one. The effect scales with the available * bandwidth and quickly saturates the CPU and network interface * with packet generation and sending. Set to zero to disable MINMSS * checking. This setting prevents us from sending too small packets. */ VNET_DEFINE(int, tcp_minmss) = TCP_MINMSS; SYSCTL_INT(_net_inet_tcp, OID_AUTO, minmss, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_minmss), 0, "Minimum TCP Maximum Segment Size"); VNET_DEFINE(int, tcp_do_rfc1323) = 1; SYSCTL_INT(_net_inet_tcp, TCPCTL_DO_RFC1323, rfc1323, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_do_rfc1323), 0, "Enable rfc1323 (high performance TCP) extensions"); static int tcp_log_debug = 0; SYSCTL_INT(_net_inet_tcp, OID_AUTO, log_debug, CTLFLAG_RW, &tcp_log_debug, 0, "Log errors caused by incoming TCP segments"); static int tcp_tcbhashsize; SYSCTL_INT(_net_inet_tcp, OID_AUTO, tcbhashsize, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &tcp_tcbhashsize, 0, "Size of TCP control-block hashtable"); static int do_tcpdrain = 1; SYSCTL_INT(_net_inet_tcp, OID_AUTO, do_tcpdrain, CTLFLAG_RW, &do_tcpdrain, 0, "Enable tcp_drain routine for extra help when low on mbufs"); SYSCTL_UINT(_net_inet_tcp, OID_AUTO, pcbcount, CTLFLAG_VNET | CTLFLAG_RD, &VNET_NAME(tcbinfo.ipi_count), 0, "Number of active PCBs"); static VNET_DEFINE(int, icmp_may_rst) = 1; #define V_icmp_may_rst VNET(icmp_may_rst) SYSCTL_INT(_net_inet_tcp, OID_AUTO, icmp_may_rst, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(icmp_may_rst), 0, "Certain ICMP unreachable messages may abort connections in SYN_SENT"); static VNET_DEFINE(int, tcp_isn_reseed_interval) = 0; #define V_tcp_isn_reseed_interval VNET(tcp_isn_reseed_interval) SYSCTL_INT(_net_inet_tcp, OID_AUTO, isn_reseed_interval, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_isn_reseed_interval), 0, "Seconds between reseeding of ISN secret"); static int tcp_soreceive_stream; SYSCTL_INT(_net_inet_tcp, OID_AUTO, soreceive_stream, CTLFLAG_RDTUN, &tcp_soreceive_stream, 0, "Using soreceive_stream for TCP sockets"); #ifdef TCP_SIGNATURE static int tcp_sig_checksigs = 1; SYSCTL_INT(_net_inet_tcp, OID_AUTO, signature_verify_input, CTLFLAG_RW, &tcp_sig_checksigs, 0, "Verify RFC2385 digests on inbound traffic"); #endif VNET_DEFINE(uma_zone_t, sack_hole_zone); #define V_sack_hole_zone VNET(sack_hole_zone) VNET_DEFINE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST+1]); static struct inpcb *tcp_notify(struct inpcb *, int); static struct inpcb *tcp_mtudisc_notify(struct inpcb *, int); static void tcp_mtudisc(struct inpcb *, int); static char * tcp_log_addr(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr); static struct tcp_function_block tcp_def_funcblk = { "default", tcp_output, tcp_do_segment, tcp_default_ctloutput, NULL, NULL, NULL, NULL, NULL, NULL, 0, 0 }; int t_functions_inited = 0; struct tcp_funchead t_functions; static struct tcp_function_block *tcp_func_set_ptr = &tcp_def_funcblk; static void init_tcp_functions(void) { if (t_functions_inited == 0) { TAILQ_INIT(&t_functions); rw_init_flags(&tcp_function_lock, "tcp_func_lock" , 0); t_functions_inited = 1; } } static struct tcp_function_block * find_tcp_functions_locked(struct tcp_function_set *fs) { struct tcp_function *f; struct tcp_function_block *blk=NULL; TAILQ_FOREACH(f, &t_functions, tf_next) { if (strcmp(f->tf_fb->tfb_tcp_block_name, fs->function_set_name) == 0) { blk = f->tf_fb; break; } } return(blk); } static struct tcp_function_block * find_tcp_fb_locked(struct tcp_function_block *blk, struct tcp_function **s) { struct tcp_function_block *rblk=NULL; struct tcp_function *f; TAILQ_FOREACH(f, &t_functions, tf_next) { if (f->tf_fb == blk) { rblk = blk; if (s) { *s = f; } break; } } return (rblk); } struct tcp_function_block * find_and_ref_tcp_functions(struct tcp_function_set *fs) { struct tcp_function_block *blk; rw_rlock(&tcp_function_lock); blk = find_tcp_functions_locked(fs); if (blk) refcount_acquire(&blk->tfb_refcnt); rw_runlock(&tcp_function_lock); return(blk); } struct tcp_function_block * find_and_ref_tcp_fb(struct tcp_function_block *blk) { struct tcp_function_block *rblk; rw_rlock(&tcp_function_lock); rblk = find_tcp_fb_locked(blk, NULL); if (rblk) refcount_acquire(&rblk->tfb_refcnt); rw_runlock(&tcp_function_lock); return(rblk); } static int sysctl_net_inet_default_tcp_functions(SYSCTL_HANDLER_ARGS) { int error=ENOENT; struct tcp_function_set fs; struct tcp_function_block *blk; memset(&fs, 0, sizeof(fs)); rw_rlock(&tcp_function_lock); blk = find_tcp_fb_locked(tcp_func_set_ptr, NULL); if (blk) { /* Found him */ strcpy(fs.function_set_name, blk->tfb_tcp_block_name); fs.pcbcnt = blk->tfb_refcnt; } rw_runlock(&tcp_function_lock); error = sysctl_handle_string(oidp, fs.function_set_name, sizeof(fs.function_set_name), req); /* Check for error or no change */ if (error != 0 || req->newptr == NULL) return(error); rw_wlock(&tcp_function_lock); blk = find_tcp_functions_locked(&fs); if ((blk == NULL) || (blk->tfb_flags & TCP_FUNC_BEING_REMOVED)) { error = ENOENT; goto done; } tcp_func_set_ptr = blk; done: rw_wunlock(&tcp_function_lock); return (error); } SYSCTL_PROC(_net_inet_tcp, OID_AUTO, functions_default, CTLTYPE_STRING | CTLFLAG_RW, NULL, 0, sysctl_net_inet_default_tcp_functions, "A", "Set/get the default TCP functions"); static int sysctl_net_inet_list_available(SYSCTL_HANDLER_ARGS) { int error, cnt, linesz; struct tcp_function *f; char *buffer, *cp; size_t bufsz, outsz; cnt = 0; rw_rlock(&tcp_function_lock); TAILQ_FOREACH(f, &t_functions, tf_next) { cnt++; } rw_runlock(&tcp_function_lock); bufsz = (cnt+2) * (TCP_FUNCTION_NAME_LEN_MAX + 12) + 1; buffer = malloc(bufsz, M_TEMP, M_WAITOK); error = 0; cp = buffer; linesz = snprintf(cp, bufsz, "\n%-32s%c %s\n", "Stack", 'D', "PCB count"); cp += linesz; bufsz -= linesz; outsz = linesz; rw_rlock(&tcp_function_lock); TAILQ_FOREACH(f, &t_functions, tf_next) { linesz = snprintf(cp, bufsz, "%-32s%c %u\n", f->tf_fb->tfb_tcp_block_name, (f->tf_fb == tcp_func_set_ptr) ? '*' : ' ', f->tf_fb->tfb_refcnt); if (linesz >= bufsz) { error = EOVERFLOW; break; } cp += linesz; bufsz -= linesz; outsz += linesz; } rw_runlock(&tcp_function_lock); if (error == 0) error = sysctl_handle_string(oidp, buffer, outsz + 1, req); free(buffer, M_TEMP); return (error); } SYSCTL_PROC(_net_inet_tcp, OID_AUTO, functions_available, CTLTYPE_STRING|CTLFLAG_RD, NULL, 0, sysctl_net_inet_list_available, "A", "list available TCP Function sets"); /* * Target size of TCP PCB hash tables. Must be a power of two. * * Note that this can be overridden by the kernel environment * variable net.inet.tcp.tcbhashsize */ #ifndef TCBHASHSIZE #define TCBHASHSIZE 0 #endif /* * XXX * Callouts should be moved into struct tcp directly. They are currently * separate because the tcpcb structure is exported to userland for sysctl * parsing purposes, which do not know about callouts. */ struct tcpcb_mem { struct tcpcb tcb; struct tcp_timer tt; struct cc_var ccv; struct osd osd; }; static VNET_DEFINE(uma_zone_t, tcpcb_zone); #define V_tcpcb_zone VNET(tcpcb_zone) MALLOC_DEFINE(M_TCPLOG, "tcplog", "TCP address and flags print buffers"); MALLOC_DEFINE(M_TCPFUNCTIONS, "tcpfunc", "TCP function set memory"); static struct mtx isn_mtx; #define ISN_LOCK_INIT() mtx_init(&isn_mtx, "isn_mtx", NULL, MTX_DEF) #define ISN_LOCK() mtx_lock(&isn_mtx) #define ISN_UNLOCK() mtx_unlock(&isn_mtx) /* * TCP initialization. */ static void tcp_zone_change(void *tag) { uma_zone_set_max(V_tcbinfo.ipi_zone, maxsockets); uma_zone_set_max(V_tcpcb_zone, maxsockets); tcp_tw_zone_change(); } static int tcp_inpcb_init(void *mem, int size, int flags) { struct inpcb *inp = mem; INP_LOCK_INIT(inp, "inp", "tcpinp"); return (0); } /* * Take a value and get the next power of 2 that doesn't overflow. * Used to size the tcp_inpcb hash buckets. */ static int maketcp_hashsize(int size) { int hashsize; /* * auto tune. * get the next power of 2 higher than maxsockets. */ hashsize = 1 << fls(size); /* catch overflow, and just go one power of 2 smaller */ if (hashsize < size) { hashsize = 1 << (fls(size) - 1); } return (hashsize); } int register_tcp_functions(struct tcp_function_block *blk, int wait) { struct tcp_function_block *lblk; struct tcp_function *n; struct tcp_function_set fs; if (t_functions_inited == 0) { init_tcp_functions(); } if ((blk->tfb_tcp_output == NULL) || (blk->tfb_tcp_do_segment == NULL) || (blk->tfb_tcp_ctloutput == NULL) || (strlen(blk->tfb_tcp_block_name) == 0)) { /* * These functions are required and you * need a name. */ return (EINVAL); } if (blk->tfb_tcp_timer_stop_all || blk->tfb_tcp_timer_activate || blk->tfb_tcp_timer_active || blk->tfb_tcp_timer_stop) { /* * If you define one timer function you * must have them all. */ if ((blk->tfb_tcp_timer_stop_all == NULL) || (blk->tfb_tcp_timer_activate == NULL) || (blk->tfb_tcp_timer_active == NULL) || (blk->tfb_tcp_timer_stop == NULL)) { return (EINVAL); } } n = malloc(sizeof(struct tcp_function), M_TCPFUNCTIONS, wait); if (n == NULL) { return (ENOMEM); } n->tf_fb = blk; strcpy(fs.function_set_name, blk->tfb_tcp_block_name); rw_wlock(&tcp_function_lock); lblk = find_tcp_functions_locked(&fs); if (lblk) { /* Duplicate name space not allowed */ rw_wunlock(&tcp_function_lock); free(n, M_TCPFUNCTIONS); return (EALREADY); } refcount_init(&blk->tfb_refcnt, 0); blk->tfb_flags = 0; TAILQ_INSERT_TAIL(&t_functions, n, tf_next); rw_wunlock(&tcp_function_lock); return(0); } int deregister_tcp_functions(struct tcp_function_block *blk) { struct tcp_function_block *lblk; struct tcp_function *f; int error=ENOENT; if (strcmp(blk->tfb_tcp_block_name, "default") == 0) { /* You can't un-register the default */ return (EPERM); } rw_wlock(&tcp_function_lock); if (blk == tcp_func_set_ptr) { /* You can't free the current default */ rw_wunlock(&tcp_function_lock); return (EBUSY); } if (blk->tfb_refcnt) { /* Still tcb attached, mark it. */ blk->tfb_flags |= TCP_FUNC_BEING_REMOVED; rw_wunlock(&tcp_function_lock); return (EBUSY); } lblk = find_tcp_fb_locked(blk, &f); if (lblk) { /* Found */ TAILQ_REMOVE(&t_functions, f, tf_next); f->tf_fb = NULL; free(f, M_TCPFUNCTIONS); error = 0; } rw_wunlock(&tcp_function_lock); return (error); } void tcp_init(void) { const char *tcbhash_tuneable; int hashsize; tcbhash_tuneable = "net.inet.tcp.tcbhashsize"; if (hhook_head_register(HHOOK_TYPE_TCP, HHOOK_TCP_EST_IN, &V_tcp_hhh[HHOOK_TCP_EST_IN], HHOOK_NOWAIT|HHOOK_HEADISINVNET) != 0) printf("%s: WARNING: unable to register helper hook\n", __func__); if (hhook_head_register(HHOOK_TYPE_TCP, HHOOK_TCP_EST_OUT, &V_tcp_hhh[HHOOK_TCP_EST_OUT], HHOOK_NOWAIT|HHOOK_HEADISINVNET) != 0) printf("%s: WARNING: unable to register helper hook\n", __func__); hashsize = TCBHASHSIZE; TUNABLE_INT_FETCH(tcbhash_tuneable, &hashsize); if (hashsize == 0) { /* * Auto tune the hash size based on maxsockets. * A perfect hash would have a 1:1 mapping * (hashsize = maxsockets) however it's been * suggested that O(2) average is better. */ hashsize = maketcp_hashsize(maxsockets / 4); /* * Our historical default is 512, * do not autotune lower than this. */ if (hashsize < 512) hashsize = 512; if (bootverbose && IS_DEFAULT_VNET(curvnet)) printf("%s: %s auto tuned to %d\n", __func__, tcbhash_tuneable, hashsize); } /* * We require a hashsize to be a power of two. * Previously if it was not a power of two we would just reset it * back to 512, which could be a nasty surprise if you did not notice * the error message. * Instead what we do is clip it to the closest power of two lower * than the specified hash value. */ if (!powerof2(hashsize)) { int oldhashsize = hashsize; hashsize = maketcp_hashsize(hashsize); /* prevent absurdly low value */ if (hashsize < 16) hashsize = 16; printf("%s: WARNING: TCB hash size not a power of 2, " "clipped from %d to %d.\n", __func__, oldhashsize, hashsize); } in_pcbinfo_init(&V_tcbinfo, "tcp", &V_tcb, hashsize, hashsize, "tcp_inpcb", tcp_inpcb_init, NULL, 0, IPI_HASHFIELDS_4TUPLE); /* * These have to be type stable for the benefit of the timers. */ V_tcpcb_zone = uma_zcreate("tcpcb", sizeof(struct tcpcb_mem), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0); uma_zone_set_max(V_tcpcb_zone, maxsockets); uma_zone_set_warning(V_tcpcb_zone, "kern.ipc.maxsockets limit reached"); tcp_tw_init(); syncache_init(); tcp_hc_init(); TUNABLE_INT_FETCH("net.inet.tcp.sack.enable", &V_tcp_do_sack); V_sack_hole_zone = uma_zcreate("sackhole", sizeof(struct sackhole), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0); /* Skip initialization of globals for non-default instances. */ if (!IS_DEFAULT_VNET(curvnet)) return; tcp_reass_global_init(); /* XXX virtualize those bellow? */ tcp_delacktime = TCPTV_DELACK; tcp_keepinit = TCPTV_KEEP_INIT; tcp_keepidle = TCPTV_KEEP_IDLE; tcp_keepintvl = TCPTV_KEEPINTVL; tcp_maxpersistidle = TCPTV_KEEP_IDLE; tcp_msl = TCPTV_MSL; tcp_rexmit_min = TCPTV_MIN; if (tcp_rexmit_min < 1) tcp_rexmit_min = 1; tcp_persmin = TCPTV_PERSMIN; tcp_persmax = TCPTV_PERSMAX; tcp_rexmit_slop = TCPTV_CPU_VAR; tcp_finwait2_timeout = TCPTV_FINWAIT2_TIMEOUT; tcp_tcbhashsize = hashsize; /* Setup the tcp function block list */ init_tcp_functions(); register_tcp_functions(&tcp_def_funcblk, M_WAITOK); if (tcp_soreceive_stream) { #ifdef INET tcp_usrreqs.pru_soreceive = soreceive_stream; #endif #ifdef INET6 tcp6_usrreqs.pru_soreceive = soreceive_stream; #endif /* INET6 */ } #ifdef INET6 #define TCP_MINPROTOHDR (sizeof(struct ip6_hdr) + sizeof(struct tcphdr)) #else /* INET6 */ #define TCP_MINPROTOHDR (sizeof(struct tcpiphdr)) #endif /* INET6 */ if (max_protohdr < TCP_MINPROTOHDR) max_protohdr = TCP_MINPROTOHDR; if (max_linkhdr + TCP_MINPROTOHDR > MHLEN) panic("tcp_init"); #undef TCP_MINPROTOHDR ISN_LOCK_INIT(); EVENTHANDLER_REGISTER(shutdown_pre_sync, tcp_fini, NULL, SHUTDOWN_PRI_DEFAULT); EVENTHANDLER_REGISTER(maxsockets_change, tcp_zone_change, NULL, EVENTHANDLER_PRI_ANY); #ifdef TCPPCAP tcp_pcap_init(); #endif #ifdef TCP_RFC7413 tcp_fastopen_init(); #endif } #ifdef VIMAGE static void tcp_destroy(void *unused __unused) { int error, n; /* * All our processes are gone, all our sockets should be cleaned * up, which means, we should be past the tcp_discardcb() calls. * Sleep to let all tcpcb timers really disappear and cleanup. */ - do { - pause("tcpdes", hz/10); + for (;;) { INP_LIST_RLOCK(&V_tcbinfo); n = V_tcbinfo.ipi_count; INP_LIST_RUNLOCK(&V_tcbinfo); - } while (n != 0); + if (n == 0) + break; + pause("tcpdes", hz / 10); + } tcp_hc_destroy(); syncache_destroy(); tcp_tw_destroy(); in_pcbinfo_destroy(&V_tcbinfo); /* tcp_discardcb() clears the sack_holes up. */ uma_zdestroy(V_sack_hole_zone); uma_zdestroy(V_tcpcb_zone); #ifdef TCP_RFC7413 /* * Cannot free the zone until all tcpcbs are released as we attach * the allocations to them. */ tcp_fastopen_destroy(); #endif error = hhook_head_deregister(V_tcp_hhh[HHOOK_TCP_EST_IN]); if (error != 0) { printf("%s: WARNING: unable to deregister helper hook " "type=%d, id=%d: error %d returned\n", __func__, HHOOK_TYPE_TCP, HHOOK_TCP_EST_IN, error); } error = hhook_head_deregister(V_tcp_hhh[HHOOK_TCP_EST_OUT]); if (error != 0) { printf("%s: WARNING: unable to deregister helper hook " "type=%d, id=%d: error %d returned\n", __func__, HHOOK_TYPE_TCP, HHOOK_TCP_EST_OUT, error); } } VNET_SYSUNINIT(tcp, SI_SUB_PROTO_DOMAIN, SI_ORDER_FOURTH, tcp_destroy, NULL); #endif void tcp_fini(void *xtp) { } /* * Fill in the IP and TCP headers for an outgoing packet, given the tcpcb. * tcp_template used to store this data in mbufs, but we now recopy it out * of the tcpcb each time to conserve mbufs. */ void tcpip_fillheaders(struct inpcb *inp, void *ip_ptr, void *tcp_ptr) { struct tcphdr *th = (struct tcphdr *)tcp_ptr; INP_WLOCK_ASSERT(inp); #ifdef INET6 if ((inp->inp_vflag & INP_IPV6) != 0) { struct ip6_hdr *ip6; ip6 = (struct ip6_hdr *)ip_ptr; ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) | (inp->inp_flow & IPV6_FLOWINFO_MASK); ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) | (IPV6_VERSION & IPV6_VERSION_MASK); ip6->ip6_nxt = IPPROTO_TCP; ip6->ip6_plen = htons(sizeof(struct tcphdr)); ip6->ip6_src = inp->in6p_laddr; ip6->ip6_dst = inp->in6p_faddr; } #endif /* INET6 */ #if defined(INET6) && defined(INET) else #endif #ifdef INET { struct ip *ip; ip = (struct ip *)ip_ptr; ip->ip_v = IPVERSION; ip->ip_hl = 5; ip->ip_tos = inp->inp_ip_tos; ip->ip_len = 0; ip->ip_id = 0; ip->ip_off = 0; ip->ip_ttl = inp->inp_ip_ttl; ip->ip_sum = 0; ip->ip_p = IPPROTO_TCP; ip->ip_src = inp->inp_laddr; ip->ip_dst = inp->inp_faddr; } #endif /* INET */ th->th_sport = inp->inp_lport; th->th_dport = inp->inp_fport; th->th_seq = 0; th->th_ack = 0; th->th_x2 = 0; th->th_off = 5; th->th_flags = 0; th->th_win = 0; th->th_urp = 0; th->th_sum = 0; /* in_pseudo() is called later for ipv4 */ } /* * Create template to be used to send tcp packets on a connection. * Allocates an mbuf and fills in a skeletal tcp/ip header. The only * use for this function is in keepalives, which use tcp_respond. */ struct tcptemp * tcpip_maketemplate(struct inpcb *inp) { struct tcptemp *t; t = malloc(sizeof(*t), M_TEMP, M_NOWAIT); if (t == NULL) return (NULL); tcpip_fillheaders(inp, (void *)&t->tt_ipgen, (void *)&t->tt_t); return (t); } /* * Send a single message to the TCP at address specified by * the given TCP/IP header. If m == NULL, then we make a copy * of the tcpiphdr at th and send directly to the addressed host. * This is used to force keep alive messages out using the TCP * template for a connection. If flags are given then we send * a message back to the TCP which originated the segment th, * and discard the mbuf containing it and any other attached mbufs. * * In any case the ack and sequence number of the transmitted * segment are as specified by the parameters. * * NOTE: If m != NULL, then th must point to *inside* the mbuf. */ void tcp_respond(struct tcpcb *tp, void *ipgen, struct tcphdr *th, struct mbuf *m, tcp_seq ack, tcp_seq seq, int flags) { struct tcpopt to; struct inpcb *inp; struct ip *ip; struct mbuf *optm; struct tcphdr *nth; u_char *optp; #ifdef INET6 struct ip6_hdr *ip6; int isipv6; #endif /* INET6 */ int optlen, tlen, win; bool incl_opts; KASSERT(tp != NULL || m != NULL, ("tcp_respond: tp and m both NULL")); #ifdef INET6 isipv6 = ((struct ip *)ipgen)->ip_v == (IPV6_VERSION >> 4); ip6 = ipgen; #endif /* INET6 */ ip = ipgen; if (tp != NULL) { inp = tp->t_inpcb; KASSERT(inp != NULL, ("tcp control block w/o inpcb")); INP_WLOCK_ASSERT(inp); } else inp = NULL; incl_opts = false; win = 0; if (tp != NULL) { if (!(flags & TH_RST)) { win = sbspace(&inp->inp_socket->so_rcv); if (win > (long)TCP_MAXWIN << tp->rcv_scale) win = (long)TCP_MAXWIN << tp->rcv_scale; } if ((tp->t_flags & TF_NOOPT) == 0) incl_opts = true; } if (m == NULL) { m = m_gethdr(M_NOWAIT, MT_DATA); if (m == NULL) return; m->m_data += max_linkhdr; #ifdef INET6 if (isipv6) { bcopy((caddr_t)ip6, mtod(m, caddr_t), sizeof(struct ip6_hdr)); ip6 = mtod(m, struct ip6_hdr *); nth = (struct tcphdr *)(ip6 + 1); } else #endif /* INET6 */ { bcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip)); ip = mtod(m, struct ip *); nth = (struct tcphdr *)(ip + 1); } bcopy((caddr_t)th, (caddr_t)nth, sizeof(struct tcphdr)); flags = TH_ACK; } else if (!M_WRITABLE(m)) { struct mbuf *n; /* Can't reuse 'm', allocate a new mbuf. */ n = m_gethdr(M_NOWAIT, MT_DATA); if (n == NULL) { m_freem(m); return; } if (!m_dup_pkthdr(n, m, M_NOWAIT)) { m_freem(m); m_freem(n); return; } n->m_data += max_linkhdr; /* m_len is set later */ #define xchg(a,b,type) { type t; t=a; a=b; b=t; } #ifdef INET6 if (isipv6) { bcopy((caddr_t)ip6, mtod(n, caddr_t), sizeof(struct ip6_hdr)); ip6 = mtod(n, struct ip6_hdr *); xchg(ip6->ip6_dst, ip6->ip6_src, struct in6_addr); nth = (struct tcphdr *)(ip6 + 1); } else #endif /* INET6 */ { bcopy((caddr_t)ip, mtod(n, caddr_t), sizeof(struct ip)); ip = mtod(n, struct ip *); xchg(ip->ip_dst.s_addr, ip->ip_src.s_addr, uint32_t); nth = (struct tcphdr *)(ip + 1); } bcopy((caddr_t)th, (caddr_t)nth, sizeof(struct tcphdr)); xchg(nth->th_dport, nth->th_sport, uint16_t); th = nth; m_freem(m); m = n; } else { /* * reuse the mbuf. * XXX MRT We inherit the FIB, which is lucky. */ m_freem(m->m_next); m->m_next = NULL; m->m_data = (caddr_t)ipgen; /* m_len is set later */ #ifdef INET6 if (isipv6) { xchg(ip6->ip6_dst, ip6->ip6_src, struct in6_addr); nth = (struct tcphdr *)(ip6 + 1); } else #endif /* INET6 */ { xchg(ip->ip_dst.s_addr, ip->ip_src.s_addr, uint32_t); nth = (struct tcphdr *)(ip + 1); } if (th != nth) { /* * this is usually a case when an extension header * exists between the IPv6 header and the * TCP header. */ nth->th_sport = th->th_sport; nth->th_dport = th->th_dport; } xchg(nth->th_dport, nth->th_sport, uint16_t); #undef xchg } tlen = 0; #ifdef INET6 if (isipv6) tlen = sizeof (struct ip6_hdr) + sizeof (struct tcphdr); #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET tlen = sizeof (struct tcpiphdr); #endif #ifdef INVARIANTS m->m_len = 0; KASSERT(M_TRAILINGSPACE(m) >= tlen, ("Not enough trailing space for message (m=%p, need=%d, have=%ld)", m, tlen, (long)M_TRAILINGSPACE(m))); #endif m->m_len = tlen; to.to_flags = 0; if (incl_opts) { /* Make sure we have room. */ if (M_TRAILINGSPACE(m) < TCP_MAXOLEN) { m->m_next = m_get(M_NOWAIT, MT_DATA); if (m->m_next) { optp = mtod(m->m_next, u_char *); optm = m->m_next; } else incl_opts = false; } else { optp = (u_char *) (nth + 1); optm = m; } } if (incl_opts) { /* Timestamps. */ if (tp->t_flags & TF_RCVD_TSTMP) { to.to_tsval = tcp_ts_getticks() + tp->ts_offset; to.to_tsecr = tp->ts_recent; to.to_flags |= TOF_TS; } #ifdef TCP_SIGNATURE /* TCP-MD5 (RFC2385). */ if (tp->t_flags & TF_SIGNATURE) to.to_flags |= TOF_SIGNATURE; #endif /* Add the options. */ tlen += optlen = tcp_addoptions(&to, optp); /* Update m_len in the correct mbuf. */ optm->m_len += optlen; } else optlen = 0; #ifdef INET6 if (isipv6) { ip6->ip6_flow = 0; ip6->ip6_vfc = IPV6_VERSION; ip6->ip6_nxt = IPPROTO_TCP; ip6->ip6_plen = htons(tlen - sizeof(*ip6)); } #endif #if defined(INET) && defined(INET6) else #endif #ifdef INET { ip->ip_len = htons(tlen); ip->ip_ttl = V_ip_defttl; if (V_path_mtu_discovery) ip->ip_off |= htons(IP_DF); } #endif m->m_pkthdr.len = tlen; m->m_pkthdr.rcvif = NULL; #ifdef MAC if (inp != NULL) { /* * Packet is associated with a socket, so allow the * label of the response to reflect the socket label. */ INP_WLOCK_ASSERT(inp); mac_inpcb_create_mbuf(inp, m); } else { /* * Packet is not associated with a socket, so possibly * update the label in place. */ mac_netinet_tcp_reply(m); } #endif nth->th_seq = htonl(seq); nth->th_ack = htonl(ack); nth->th_x2 = 0; nth->th_off = (sizeof (struct tcphdr) + optlen) >> 2; nth->th_flags = flags; if (tp != NULL) nth->th_win = htons((u_short) (win >> tp->rcv_scale)); else nth->th_win = htons((u_short)win); nth->th_urp = 0; #ifdef TCP_SIGNATURE if (to.to_flags & TOF_SIGNATURE) { tcp_signature_compute(m, 0, 0, optlen, to.to_signature, IPSEC_DIR_OUTBOUND); } #endif m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum); #ifdef INET6 if (isipv6) { m->m_pkthdr.csum_flags = CSUM_TCP_IPV6; nth->th_sum = in6_cksum_pseudo(ip6, tlen - sizeof(struct ip6_hdr), IPPROTO_TCP, 0); ip6->ip6_hlim = in6_selecthlim(tp != NULL ? tp->t_inpcb : NULL, NULL); } #endif /* INET6 */ #if defined(INET6) && defined(INET) else #endif #ifdef INET { m->m_pkthdr.csum_flags = CSUM_TCP; nth->th_sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, htons((u_short)(tlen - sizeof(struct ip) + ip->ip_p))); } #endif /* INET */ #ifdef TCPDEBUG if (tp == NULL || (inp->inp_socket->so_options & SO_DEBUG)) tcp_trace(TA_OUTPUT, 0, tp, mtod(m, void *), th, 0); #endif TCP_PROBE3(debug__output, tp, th, mtod(m, const char *)); if (flags & TH_RST) TCP_PROBE5(accept__refused, NULL, NULL, mtod(m, const char *), tp, nth); TCP_PROBE5(send, NULL, tp, mtod(m, const char *), tp, nth); #ifdef INET6 if (isipv6) (void) ip6_output(m, NULL, NULL, 0, NULL, NULL, inp); #endif /* INET6 */ #if defined(INET) && defined(INET6) else #endif #ifdef INET (void) ip_output(m, NULL, NULL, 0, NULL, inp); #endif } /* * Create a new TCP control block, making an * empty reassembly queue and hooking it to the argument * protocol control block. The `inp' parameter must have * come from the zone allocator set up in tcp_init(). */ struct tcpcb * tcp_newtcpcb(struct inpcb *inp) { struct tcpcb_mem *tm; struct tcpcb *tp; #ifdef INET6 int isipv6 = (inp->inp_vflag & INP_IPV6) != 0; #endif /* INET6 */ tm = uma_zalloc(V_tcpcb_zone, M_NOWAIT | M_ZERO); if (tm == NULL) return (NULL); tp = &tm->tcb; /* Initialise cc_var struct for this tcpcb. */ tp->ccv = &tm->ccv; tp->ccv->type = IPPROTO_TCP; tp->ccv->ccvc.tcp = tp; rw_rlock(&tcp_function_lock); tp->t_fb = tcp_func_set_ptr; refcount_acquire(&tp->t_fb->tfb_refcnt); rw_runlock(&tcp_function_lock); if (tp->t_fb->tfb_tcp_fb_init) { (*tp->t_fb->tfb_tcp_fb_init)(tp); } /* * Use the current system default CC algorithm. */ CC_LIST_RLOCK(); KASSERT(!STAILQ_EMPTY(&cc_list), ("cc_list is empty!")); CC_ALGO(tp) = CC_DEFAULT(); CC_LIST_RUNLOCK(); if (CC_ALGO(tp)->cb_init != NULL) if (CC_ALGO(tp)->cb_init(tp->ccv) > 0) { if (tp->t_fb->tfb_tcp_fb_fini) (*tp->t_fb->tfb_tcp_fb_fini)(tp); refcount_release(&tp->t_fb->tfb_refcnt); uma_zfree(V_tcpcb_zone, tm); return (NULL); } tp->osd = &tm->osd; if (khelp_init_osd(HELPER_CLASS_TCP, tp->osd)) { if (tp->t_fb->tfb_tcp_fb_fini) (*tp->t_fb->tfb_tcp_fb_fini)(tp); refcount_release(&tp->t_fb->tfb_refcnt); uma_zfree(V_tcpcb_zone, tm); return (NULL); } #ifdef VIMAGE tp->t_vnet = inp->inp_vnet; #endif tp->t_timers = &tm->tt; /* LIST_INIT(&tp->t_segq); */ /* XXX covered by M_ZERO */ tp->t_maxseg = #ifdef INET6 isipv6 ? V_tcp_v6mssdflt : #endif /* INET6 */ V_tcp_mssdflt; /* Set up our timeouts. */ callout_init(&tp->t_timers->tt_rexmt, 1); callout_init(&tp->t_timers->tt_persist, 1); callout_init(&tp->t_timers->tt_keep, 1); callout_init(&tp->t_timers->tt_2msl, 1); callout_init(&tp->t_timers->tt_delack, 1); if (V_tcp_do_rfc1323) tp->t_flags = (TF_REQ_SCALE|TF_REQ_TSTMP); if (V_tcp_do_sack) tp->t_flags |= TF_SACK_PERMIT; TAILQ_INIT(&tp->snd_holes); /* * The tcpcb will hold a reference on its inpcb until tcp_discardcb() * is called. */ in_pcbref(inp); /* Reference for tcpcb */ tp->t_inpcb = inp; /* * Init srtt to TCPTV_SRTTBASE (0), so we can tell that we have no * rtt estimate. Set rttvar so that srtt + 4 * rttvar gives * reasonable initial retransmit time. */ tp->t_srtt = TCPTV_SRTTBASE; tp->t_rttvar = ((TCPTV_RTOBASE - TCPTV_SRTTBASE) << TCP_RTTVAR_SHIFT) / 4; tp->t_rttmin = tcp_rexmit_min; tp->t_rxtcur = TCPTV_RTOBASE; tp->snd_cwnd = TCP_MAXWIN << TCP_MAX_WINSHIFT; tp->snd_ssthresh = TCP_MAXWIN << TCP_MAX_WINSHIFT; tp->t_rcvtime = ticks; /* * 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; inp->inp_ppcb = tp; #ifdef TCPPCAP /* * Init the TCP PCAP queues. */ tcp_pcap_tcpcb_init(tp); #endif return (tp); /* XXX */ } /* * Switch the congestion control algorithm back to NewReno for any active * control blocks using an algorithm which is about to go away. * This ensures the CC framework can allow the unload to proceed without leaving * any dangling pointers which would trigger a panic. * Returning non-zero would inform the CC framework that something went wrong * and it would be unsafe to allow the unload to proceed. However, there is no * way for this to occur with this implementation so we always return zero. */ int tcp_ccalgounload(struct cc_algo *unload_algo) { struct cc_algo *tmpalgo; struct inpcb *inp; struct tcpcb *tp; VNET_ITERATOR_DECL(vnet_iter); /* * Check all active control blocks across all network stacks and change * any that are using "unload_algo" back to NewReno. If "unload_algo" * requires cleanup code to be run, call it. */ VNET_LIST_RLOCK(); VNET_FOREACH(vnet_iter) { CURVNET_SET(vnet_iter); INP_INFO_WLOCK(&V_tcbinfo); /* * New connections already part way through being initialised * with the CC algo we're removing will not race with this code * because the INP_INFO_WLOCK is held during initialisation. We * therefore don't enter the loop below until the connection * list has stabilised. */ LIST_FOREACH(inp, &V_tcb, inp_list) { INP_WLOCK(inp); /* Important to skip tcptw structs. */ if (!(inp->inp_flags & INP_TIMEWAIT) && (tp = intotcpcb(inp)) != NULL) { /* * By holding INP_WLOCK here, we are assured * that the connection is not currently * executing inside the CC module's functions * i.e. it is safe to make the switch back to * NewReno. */ if (CC_ALGO(tp) == unload_algo) { tmpalgo = CC_ALGO(tp); /* NewReno does not require any init. */ CC_ALGO(tp) = &newreno_cc_algo; if (tmpalgo->cb_destroy != NULL) tmpalgo->cb_destroy(tp->ccv); } } INP_WUNLOCK(inp); } INP_INFO_WUNLOCK(&V_tcbinfo); CURVNET_RESTORE(); } VNET_LIST_RUNLOCK(); return (0); } /* * Drop a TCP connection, reporting * the specified error. If connection is synchronized, * then send a RST to peer. */ struct tcpcb * tcp_drop(struct tcpcb *tp, int errno) { struct socket *so = tp->t_inpcb->inp_socket; INP_INFO_LOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(tp->t_inpcb); if (TCPS_HAVERCVDSYN(tp->t_state)) { tcp_state_change(tp, TCPS_CLOSED); (void) tp->t_fb->tfb_tcp_output(tp); TCPSTAT_INC(tcps_drops); } else TCPSTAT_INC(tcps_conndrops); if (errno == ETIMEDOUT && tp->t_softerror) errno = tp->t_softerror; so->so_error = errno; return (tcp_close(tp)); } void tcp_discardcb(struct tcpcb *tp) { struct inpcb *inp = tp->t_inpcb; struct socket *so = inp->inp_socket; #ifdef INET6 int isipv6 = (inp->inp_vflag & INP_IPV6) != 0; #endif /* INET6 */ int released; INP_WLOCK_ASSERT(inp); /* * Make sure that all of our timers are stopped before we delete the * PCB. * * If stopping a timer fails, we schedule a discard function in same * callout, and the last discard function called will take care of * deleting the tcpcb. */ tp->t_timers->tt_draincnt = 0; tcp_timer_stop(tp, TT_REXMT); tcp_timer_stop(tp, TT_PERSIST); tcp_timer_stop(tp, TT_KEEP); tcp_timer_stop(tp, TT_2MSL); tcp_timer_stop(tp, TT_DELACK); if (tp->t_fb->tfb_tcp_timer_stop_all) { /* * Call the stop-all function of the methods, * this function should call the tcp_timer_stop() * method with each of the function specific timeouts. * That stop will be called via the tfb_tcp_timer_stop() * which should use the async drain function of the * callout system (see tcp_var.h). */ tp->t_fb->tfb_tcp_timer_stop_all(tp); } /* * If we got enough samples through the srtt filter, * save the rtt and rttvar in the routing entry. * 'Enough' is arbitrarily defined as 4 rtt samples. * 4 samples is enough for the srtt filter to converge * to within enough % of the correct value; fewer samples * and we could save a bogus rtt. The danger is not high * as tcp quickly recovers from everything. * XXX: Works very well but needs some more statistics! */ if (tp->t_rttupdated >= 4) { struct hc_metrics_lite metrics; u_long ssthresh; bzero(&metrics, sizeof(metrics)); /* * Update the ssthresh always when the conditions below * are satisfied. This gives us better new start value * for the congestion avoidance for new connections. * ssthresh is only set if packet loss occurred on a session. * * XXXRW: 'so' may be NULL here, and/or socket buffer may be * being torn down. Ideally this code would not use 'so'. */ ssthresh = tp->snd_ssthresh; if (ssthresh != 0 && ssthresh < so->so_snd.sb_hiwat / 2) { /* * convert the limit from user data bytes to * packets then to packet data bytes. */ ssthresh = (ssthresh + tp->t_maxseg / 2) / tp->t_maxseg; if (ssthresh < 2) ssthresh = 2; ssthresh *= (u_long)(tp->t_maxseg + #ifdef INET6 (isipv6 ? sizeof (struct ip6_hdr) + sizeof (struct tcphdr) : #endif sizeof (struct tcpiphdr) #ifdef INET6 ) #endif ); } else ssthresh = 0; metrics.rmx_ssthresh = ssthresh; metrics.rmx_rtt = tp->t_srtt; metrics.rmx_rttvar = tp->t_rttvar; metrics.rmx_cwnd = tp->snd_cwnd; metrics.rmx_sendpipe = 0; metrics.rmx_recvpipe = 0; tcp_hc_update(&inp->inp_inc, &metrics); } /* free the reassembly queue, if any */ tcp_reass_flush(tp); #ifdef TCP_OFFLOAD /* Disconnect offload device, if any. */ if (tp->t_flags & TF_TOE) tcp_offload_detach(tp); #endif tcp_free_sackholes(tp); #ifdef TCPPCAP /* Free the TCP PCAP queues. */ tcp_pcap_drain(&(tp->t_inpkts)); tcp_pcap_drain(&(tp->t_outpkts)); #endif /* Allow the CC algorithm to clean up after itself. */ if (CC_ALGO(tp)->cb_destroy != NULL) CC_ALGO(tp)->cb_destroy(tp->ccv); khelp_destroy_osd(tp->osd); CC_ALGO(tp) = NULL; inp->inp_ppcb = NULL; if (tp->t_timers->tt_draincnt == 0) { /* We own the last reference on tcpcb, let's free it. */ if (tp->t_fb->tfb_tcp_fb_fini) (*tp->t_fb->tfb_tcp_fb_fini)(tp); refcount_release(&tp->t_fb->tfb_refcnt); tp->t_inpcb = NULL; uma_zfree(V_tcpcb_zone, tp); released = in_pcbrele_wlocked(inp); KASSERT(!released, ("%s: inp %p should not have been released " "here", __func__, inp)); } } void tcp_timer_discard(void *ptp) { struct inpcb *inp; struct tcpcb *tp; tp = (struct tcpcb *)ptp; CURVNET_SET(tp->t_vnet); INP_INFO_RLOCK(&V_tcbinfo); inp = tp->t_inpcb; KASSERT(inp != NULL, ("%s: tp %p tp->t_inpcb == NULL", __func__, tp)); INP_WLOCK(inp); KASSERT((tp->t_timers->tt_flags & TT_STOPPED) != 0, ("%s: tcpcb has to be stopped here", __func__)); tp->t_timers->tt_draincnt--; if (tp->t_timers->tt_draincnt == 0) { /* We own the last reference on this tcpcb, let's free it. */ if (tp->t_fb->tfb_tcp_fb_fini) (*tp->t_fb->tfb_tcp_fb_fini)(tp); refcount_release(&tp->t_fb->tfb_refcnt); tp->t_inpcb = NULL; uma_zfree(V_tcpcb_zone, tp); if (in_pcbrele_wlocked(inp)) { INP_INFO_RUNLOCK(&V_tcbinfo); CURVNET_RESTORE(); return; } } INP_WUNLOCK(inp); INP_INFO_RUNLOCK(&V_tcbinfo); CURVNET_RESTORE(); } /* * Attempt to close a TCP control block, marking it as dropped, and freeing * the socket if we hold the only reference. */ struct tcpcb * tcp_close(struct tcpcb *tp) { struct inpcb *inp = tp->t_inpcb; struct socket *so; INP_INFO_LOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(inp); #ifdef TCP_OFFLOAD if (tp->t_state == TCPS_LISTEN) tcp_offload_listen_stop(tp); #endif #ifdef TCP_RFC7413 /* * This releases the TFO pending counter resource for TFO listen * sockets as well as passively-created TFO sockets that transition * from SYN_RECEIVED to CLOSED. */ if (tp->t_tfo_pending) { tcp_fastopen_decrement_counter(tp->t_tfo_pending); tp->t_tfo_pending = NULL; } #endif in_pcbdrop(inp); TCPSTAT_INC(tcps_closed); TCPSTATES_DEC(tp->t_state); KASSERT(inp->inp_socket != NULL, ("tcp_close: inp_socket NULL")); so = inp->inp_socket; soisdisconnected(so); if (inp->inp_flags & INP_SOCKREF) { KASSERT(so->so_state & SS_PROTOREF, ("tcp_close: !SS_PROTOREF")); inp->inp_flags &= ~INP_SOCKREF; INP_WUNLOCK(inp); ACCEPT_LOCK(); SOCK_LOCK(so); so->so_state &= ~SS_PROTOREF; sofree(so); return (NULL); } return (tp); } void tcp_drain(void) { VNET_ITERATOR_DECL(vnet_iter); if (!do_tcpdrain) return; VNET_LIST_RLOCK_NOSLEEP(); VNET_FOREACH(vnet_iter) { CURVNET_SET(vnet_iter); struct inpcb *inpb; struct tcpcb *tcpb; /* * Walk the tcpbs, if existing, and flush the reassembly queue, * if there is one... * XXX: The "Net/3" implementation doesn't imply that the TCP * reassembly queue should be flushed, but in a situation * where we're really low on mbufs, this is potentially * useful. */ INP_INFO_WLOCK(&V_tcbinfo); LIST_FOREACH(inpb, V_tcbinfo.ipi_listhead, inp_list) { if (inpb->inp_flags & INP_TIMEWAIT) continue; INP_WLOCK(inpb); if ((tcpb = intotcpcb(inpb)) != NULL) { tcp_reass_flush(tcpb); tcp_clean_sackreport(tcpb); } INP_WUNLOCK(inpb); } INP_INFO_WUNLOCK(&V_tcbinfo); CURVNET_RESTORE(); } VNET_LIST_RUNLOCK_NOSLEEP(); } /* * Notify a tcp user of an asynchronous error; * store error as soft error, but wake up user * (for now, won't do anything until can select for soft error). * * Do not wake up user since there currently is no mechanism for * reporting soft errors (yet - a kqueue filter may be added). */ static struct inpcb * tcp_notify(struct inpcb *inp, int error) { struct tcpcb *tp; INP_INFO_LOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(inp); if ((inp->inp_flags & INP_TIMEWAIT) || (inp->inp_flags & INP_DROPPED)) return (inp); tp = intotcpcb(inp); KASSERT(tp != NULL, ("tcp_notify: tp == NULL")); /* * Ignore some errors if we are hooked up. * If connection hasn't completed, has retransmitted several times, * and receives a second error, give up now. This is better * than waiting a long time to establish a connection that * can never complete. */ if (tp->t_state == TCPS_ESTABLISHED && (error == EHOSTUNREACH || error == ENETUNREACH || error == EHOSTDOWN)) { if (inp->inp_route.ro_rt) { RTFREE(inp->inp_route.ro_rt); inp->inp_route.ro_rt = (struct rtentry *)NULL; } return (inp); } else if (tp->t_state < TCPS_ESTABLISHED && tp->t_rxtshift > 3 && tp->t_softerror) { tp = tcp_drop(tp, error); if (tp != NULL) return (inp); else return (NULL); } else { tp->t_softerror = error; return (inp); } #if 0 wakeup( &so->so_timeo); sorwakeup(so); sowwakeup(so); #endif } static int tcp_pcblist(SYSCTL_HANDLER_ARGS) { int error, i, m, n, pcb_count; struct inpcb *inp, **inp_list; inp_gen_t gencnt; struct xinpgen xig; /* * The process of preparing the TCB list is too time-consuming and * resource-intensive to repeat twice on every request. */ if (req->oldptr == NULL) { n = V_tcbinfo.ipi_count + counter_u64_fetch(V_tcps_states[TCPS_SYN_RECEIVED]); n += imax(n / 8, 10); req->oldidx = 2 * (sizeof xig) + n * sizeof(struct xtcpcb); return (0); } if (req->newptr != NULL) return (EPERM); /* * OK, now we're committed to doing something. */ INP_LIST_RLOCK(&V_tcbinfo); gencnt = V_tcbinfo.ipi_gencnt; n = V_tcbinfo.ipi_count; INP_LIST_RUNLOCK(&V_tcbinfo); m = counter_u64_fetch(V_tcps_states[TCPS_SYN_RECEIVED]); error = sysctl_wire_old_buffer(req, 2 * (sizeof xig) + (n + m) * sizeof(struct xtcpcb)); if (error != 0) return (error); xig.xig_len = sizeof xig; xig.xig_count = n + m; xig.xig_gen = gencnt; xig.xig_sogen = so_gencnt; error = SYSCTL_OUT(req, &xig, sizeof xig); if (error) return (error); error = syncache_pcblist(req, m, &pcb_count); if (error) return (error); inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK); INP_INFO_WLOCK(&V_tcbinfo); for (inp = LIST_FIRST(V_tcbinfo.ipi_listhead), i = 0; inp != NULL && i < n; inp = LIST_NEXT(inp, inp_list)) { INP_WLOCK(inp); if (inp->inp_gencnt <= gencnt) { /* * XXX: This use of cr_cansee(), introduced with * TCP state changes, is not quite right, but for * now, better than nothing. */ if (inp->inp_flags & INP_TIMEWAIT) { if (intotw(inp) != NULL) error = cr_cansee(req->td->td_ucred, intotw(inp)->tw_cred); else error = EINVAL; /* Skip this inp. */ } else error = cr_canseeinpcb(req->td->td_ucred, inp); if (error == 0) { in_pcbref(inp); inp_list[i++] = inp; } } INP_WUNLOCK(inp); } INP_INFO_WUNLOCK(&V_tcbinfo); n = i; error = 0; for (i = 0; i < n; i++) { inp = inp_list[i]; INP_RLOCK(inp); if (inp->inp_gencnt <= gencnt) { struct xtcpcb xt; void *inp_ppcb; bzero(&xt, sizeof(xt)); xt.xt_len = sizeof xt; /* XXX should avoid extra copy */ bcopy(inp, &xt.xt_inp, sizeof *inp); inp_ppcb = inp->inp_ppcb; if (inp_ppcb == NULL) bzero((char *) &xt.xt_tp, sizeof xt.xt_tp); else if (inp->inp_flags & INP_TIMEWAIT) { bzero((char *) &xt.xt_tp, sizeof xt.xt_tp); xt.xt_tp.t_state = TCPS_TIME_WAIT; } else { bcopy(inp_ppcb, &xt.xt_tp, sizeof xt.xt_tp); if (xt.xt_tp.t_timers) tcp_timer_to_xtimer(&xt.xt_tp, xt.xt_tp.t_timers, &xt.xt_timer); } if (inp->inp_socket != NULL) sotoxsocket(inp->inp_socket, &xt.xt_socket); else { bzero(&xt.xt_socket, sizeof xt.xt_socket); xt.xt_socket.xso_protocol = IPPROTO_TCP; } xt.xt_inp.inp_gencnt = inp->inp_gencnt; INP_RUNLOCK(inp); error = SYSCTL_OUT(req, &xt, sizeof xt); } else INP_RUNLOCK(inp); } INP_INFO_RLOCK(&V_tcbinfo); for (i = 0; i < n; i++) { inp = inp_list[i]; INP_RLOCK(inp); if (!in_pcbrele_rlocked(inp)) INP_RUNLOCK(inp); } INP_INFO_RUNLOCK(&V_tcbinfo); 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. */ INP_LIST_RLOCK(&V_tcbinfo); xig.xig_gen = V_tcbinfo.ipi_gencnt; xig.xig_sogen = so_gencnt; xig.xig_count = V_tcbinfo.ipi_count + pcb_count; INP_LIST_RUNLOCK(&V_tcbinfo); error = SYSCTL_OUT(req, &xig, sizeof xig); } free(inp_list, M_TEMP); return (error); } SYSCTL_PROC(_net_inet_tcp, TCPCTL_PCBLIST, pcblist, CTLTYPE_OPAQUE | CTLFLAG_RD, NULL, 0, tcp_pcblist, "S,xtcpcb", "List of active TCP connections"); #ifdef INET static int tcp_getcred(SYSCTL_HANDLER_ARGS) { struct xucred xuc; struct sockaddr_in addrs[2]; 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); inp = in_pcblookup(&V_tcbinfo, addrs[1].sin_addr, addrs[1].sin_port, addrs[0].sin_addr, addrs[0].sin_port, INPLOOKUP_RLOCKPCB, NULL); if (inp != NULL) { 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_tcp, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW|CTLFLAG_PRISON, 0, 0, tcp_getcred, "S,xucred", "Get the xucred of a TCP connection"); #endif /* INET */ #ifdef INET6 static int tcp6_getcred(SYSCTL_HANDLER_ARGS) { struct xucred xuc; struct sockaddr_in6 addrs[2]; struct inpcb *inp; int error; #ifdef INET int mapped = 0; #endif error = priv_check(req->td, PRIV_NETINET_GETCRED); if (error) return (error); 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); } if (IN6_IS_ADDR_V4MAPPED(&addrs[0].sin6_addr)) { #ifdef INET if (IN6_IS_ADDR_V4MAPPED(&addrs[1].sin6_addr)) mapped = 1; else #endif return (EINVAL); } #ifdef INET if (mapped == 1) inp = in_pcblookup(&V_tcbinfo, *(struct in_addr *)&addrs[1].sin6_addr.s6_addr[12], addrs[1].sin6_port, *(struct in_addr *)&addrs[0].sin6_addr.s6_addr[12], addrs[0].sin6_port, INPLOOKUP_RLOCKPCB, NULL); else #endif inp = in6_pcblookup(&V_tcbinfo, &addrs[1].sin6_addr, addrs[1].sin6_port, &addrs[0].sin6_addr, addrs[0].sin6_port, INPLOOKUP_RLOCKPCB, NULL); if (inp != NULL) { 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_inet6_tcp6, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW|CTLFLAG_PRISON, 0, 0, tcp6_getcred, "S,xucred", "Get the xucred of a TCP6 connection"); #endif /* INET6 */ #ifdef INET void tcp_ctlinput(int cmd, struct sockaddr *sa, void *vip) { struct ip *ip = vip; struct tcphdr *th; struct in_addr faddr; struct inpcb *inp; struct tcpcb *tp; struct inpcb *(*notify)(struct inpcb *, int) = tcp_notify; struct icmp *icp; struct in_conninfo inc; tcp_seq icmp_tcp_seq; int mtu; faddr = ((struct sockaddr_in *)sa)->sin_addr; if (sa->sa_family != AF_INET || faddr.s_addr == INADDR_ANY) return; if (cmd == PRC_MSGSIZE) notify = tcp_mtudisc_notify; else if (V_icmp_may_rst && (cmd == PRC_UNREACH_ADMIN_PROHIB || cmd == PRC_UNREACH_PORT || cmd == PRC_TIMXCEED_INTRANS) && ip) notify = tcp_drop_syn_sent; else if (PRC_IS_REDIRECT(cmd)) { /* signal EHOSTDOWN, as it flushes the cached route */ in_pcbnotifyall(&V_tcbinfo, faddr, EHOSTDOWN, 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. */ else if (cmd == PRC_HOSTDEAD) ip = NULL; else if ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0) return; if (ip == NULL) { in_pcbnotifyall(&V_tcbinfo, faddr, inetctlerrmap[cmd], notify); return; } icp = (struct icmp *)((caddr_t)ip - offsetof(struct icmp, icmp_ip)); th = (struct tcphdr *)((caddr_t)ip + (ip->ip_hl << 2)); INP_INFO_RLOCK(&V_tcbinfo); inp = in_pcblookup(&V_tcbinfo, faddr, th->th_dport, ip->ip_src, th->th_sport, INPLOOKUP_WLOCKPCB, NULL); if (inp != NULL) { if (!(inp->inp_flags & INP_TIMEWAIT) && !(inp->inp_flags & INP_DROPPED) && !(inp->inp_socket == NULL)) { icmp_tcp_seq = ntohl(th->th_seq); tp = intotcpcb(inp); if (SEQ_GEQ(icmp_tcp_seq, tp->snd_una) && SEQ_LT(icmp_tcp_seq, tp->snd_max)) { if (cmd == PRC_MSGSIZE) { /* * MTU discovery: * If we got a needfrag set the MTU * in the route to the suggested new * value (if given) and then notify. */ mtu = ntohs(icp->icmp_nextmtu); /* * If no alternative MTU was * proposed, try the next smaller * one. */ if (!mtu) mtu = ip_next_mtu( ntohs(ip->ip_len), 1); if (mtu < V_tcp_minmss + sizeof(struct tcpiphdr)) mtu = V_tcp_minmss + sizeof(struct tcpiphdr); /* * Only process the offered MTU if it * is smaller than the current one. */ if (mtu < tp->t_maxseg + sizeof(struct tcpiphdr)) { bzero(&inc, sizeof(inc)); inc.inc_faddr = faddr; inc.inc_fibnum = inp->inp_inc.inc_fibnum; tcp_hc_updatemtu(&inc, mtu); tcp_mtudisc(inp, mtu); } } else inp = (*notify)(inp, inetctlerrmap[cmd]); } } if (inp != NULL) INP_WUNLOCK(inp); } else { bzero(&inc, sizeof(inc)); inc.inc_fport = th->th_dport; inc.inc_lport = th->th_sport; inc.inc_faddr = faddr; inc.inc_laddr = ip->ip_src; syncache_unreach(&inc, th); } INP_INFO_RUNLOCK(&V_tcbinfo); } #endif /* INET */ #ifdef INET6 void tcp6_ctlinput(int cmd, struct sockaddr *sa, void *d) { struct tcphdr th; struct inpcb *(*notify)(struct inpcb *, int) = tcp_notify; struct ip6_hdr *ip6; struct mbuf *m; struct ip6ctlparam *ip6cp = NULL; const struct sockaddr_in6 *sa6_src = NULL; int off; struct tcp_portonly { u_int16_t th_sport; u_int16_t th_dport; } *thp; if (sa->sa_family != AF_INET6 || sa->sa_len != sizeof(struct sockaddr_in6)) return; if (cmd == PRC_MSGSIZE) notify = tcp_mtudisc_notify; else if (!PRC_IS_REDIRECT(cmd) && ((unsigned)cmd >= PRC_NCMDS || 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; sa6_src = ip6cp->ip6c_src; } else { m = NULL; ip6 = NULL; off = 0; /* fool gcc */ sa6_src = &sa6_any; } if (ip6 != NULL) { struct in_conninfo inc; /* * 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(*thp)) return; bzero(&th, sizeof(th)); m_copydata(m, off, sizeof(*thp), (caddr_t)&th); in6_pcbnotify(&V_tcbinfo, sa, th.th_dport, (struct sockaddr *)ip6cp->ip6c_src, th.th_sport, cmd, NULL, notify); bzero(&inc, sizeof(inc)); inc.inc_fport = th.th_dport; inc.inc_lport = th.th_sport; inc.inc6_faddr = ((struct sockaddr_in6 *)sa)->sin6_addr; inc.inc6_laddr = ip6cp->ip6c_src->sin6_addr; inc.inc_flags |= INC_ISIPV6; INP_INFO_RLOCK(&V_tcbinfo); syncache_unreach(&inc, &th); INP_INFO_RUNLOCK(&V_tcbinfo); } else in6_pcbnotify(&V_tcbinfo, sa, 0, (const struct sockaddr *)sa6_src, 0, cmd, NULL, notify); } #endif /* INET6 */ /* * Following is where TCP initial sequence number generation occurs. * * There are two places where we must use initial sequence numbers: * 1. In SYN-ACK packets. * 2. In SYN packets. * * All ISNs for SYN-ACK packets are generated by the syncache. See * tcp_syncache.c for details. * * The ISNs in SYN packets must be monotonic; TIME_WAIT recycling * depends on this property. In addition, these ISNs should be * unguessable so as to prevent connection hijacking. To satisfy * the requirements of this situation, the algorithm outlined in * RFC 1948 is used, with only small modifications. * * Implementation details: * * Time is based off the system timer, and is corrected so that it * increases by one megabyte per second. This allows for proper * recycling on high speed LANs while still leaving over an hour * before rollover. * * As reading the *exact* system time is too expensive to be done * whenever setting up a TCP connection, we increment the time * offset in two ways. First, a small random positive increment * is added to isn_offset for each connection that is set up. * Second, the function tcp_isn_tick fires once per clock tick * and increments isn_offset as necessary so that sequence numbers * are incremented at approximately ISN_BYTES_PER_SECOND. The * random positive increments serve only to ensure that the same * exact sequence number is never sent out twice (as could otherwise * happen when a port is recycled in less than the system tick * interval.) * * net.inet.tcp.isn_reseed_interval controls the number of seconds * between seeding of isn_secret. This is normally set to zero, * as reseeding should not be necessary. * * Locking of the global variables isn_secret, isn_last_reseed, isn_offset, * isn_offset_old, and isn_ctx is performed using the TCP pcbinfo lock. In * general, this means holding an exclusive (write) lock. */ #define ISN_BYTES_PER_SECOND 1048576 #define ISN_STATIC_INCREMENT 4096 #define ISN_RANDOM_INCREMENT (4096 - 1) static VNET_DEFINE(u_char, isn_secret[32]); static VNET_DEFINE(int, isn_last); static VNET_DEFINE(int, isn_last_reseed); static VNET_DEFINE(u_int32_t, isn_offset); static VNET_DEFINE(u_int32_t, isn_offset_old); #define V_isn_secret VNET(isn_secret) #define V_isn_last VNET(isn_last) #define V_isn_last_reseed VNET(isn_last_reseed) #define V_isn_offset VNET(isn_offset) #define V_isn_offset_old VNET(isn_offset_old) tcp_seq tcp_new_isn(struct tcpcb *tp) { MD5_CTX isn_ctx; u_int32_t md5_buffer[4]; tcp_seq new_isn; u_int32_t projected_offset; INP_WLOCK_ASSERT(tp->t_inpcb); ISN_LOCK(); /* Seed if this is the first use, reseed if requested. */ if ((V_isn_last_reseed == 0) || ((V_tcp_isn_reseed_interval > 0) && (((u_int)V_isn_last_reseed + (u_int)V_tcp_isn_reseed_interval*hz) < (u_int)ticks))) { read_random(&V_isn_secret, sizeof(V_isn_secret)); V_isn_last_reseed = ticks; } /* Compute the md5 hash and return the ISN. */ MD5Init(&isn_ctx); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_fport, sizeof(u_short)); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_lport, sizeof(u_short)); #ifdef INET6 if ((tp->t_inpcb->inp_vflag & INP_IPV6) != 0) { MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->in6p_faddr, sizeof(struct in6_addr)); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->in6p_laddr, sizeof(struct in6_addr)); } else #endif { MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_faddr, sizeof(struct in_addr)); MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_laddr, sizeof(struct in_addr)); } MD5Update(&isn_ctx, (u_char *) &V_isn_secret, sizeof(V_isn_secret)); MD5Final((u_char *) &md5_buffer, &isn_ctx); new_isn = (tcp_seq) md5_buffer[0]; V_isn_offset += ISN_STATIC_INCREMENT + (arc4random() & ISN_RANDOM_INCREMENT); if (ticks != V_isn_last) { projected_offset = V_isn_offset_old + ISN_BYTES_PER_SECOND / hz * (ticks - V_isn_last); if (SEQ_GT(projected_offset, V_isn_offset)) V_isn_offset = projected_offset; V_isn_offset_old = V_isn_offset; V_isn_last = ticks; } new_isn += V_isn_offset; ISN_UNLOCK(); return (new_isn); } /* * When a specific ICMP unreachable message is received and the * connection state is SYN-SENT, drop the connection. This behavior * is controlled by the icmp_may_rst sysctl. */ struct inpcb * tcp_drop_syn_sent(struct inpcb *inp, int errno) { struct tcpcb *tp; INP_INFO_RLOCK_ASSERT(&V_tcbinfo); INP_WLOCK_ASSERT(inp); if ((inp->inp_flags & INP_TIMEWAIT) || (inp->inp_flags & INP_DROPPED)) return (inp); tp = intotcpcb(inp); if (tp->t_state != TCPS_SYN_SENT) return (inp); tp = tcp_drop(tp, errno); if (tp != NULL) return (inp); else return (NULL); } /* * When `need fragmentation' ICMP is received, update our idea of the MSS * based on the new value. Also nudge TCP to send something, since we * know the packet we just sent was dropped. * This duplicates some code in the tcp_mss() function in tcp_input.c. */ static struct inpcb * tcp_mtudisc_notify(struct inpcb *inp, int error) { tcp_mtudisc(inp, -1); return (inp); } static void tcp_mtudisc(struct inpcb *inp, int mtuoffer) { struct tcpcb *tp; struct socket *so; INP_WLOCK_ASSERT(inp); if ((inp->inp_flags & INP_TIMEWAIT) || (inp->inp_flags & INP_DROPPED)) return; tp = intotcpcb(inp); KASSERT(tp != NULL, ("tcp_mtudisc: tp == NULL")); tcp_mss_update(tp, -1, mtuoffer, NULL, NULL); so = inp->inp_socket; SOCKBUF_LOCK(&so->so_snd); /* If the mss is larger than the socket buffer, decrease the mss. */ if (so->so_snd.sb_hiwat < tp->t_maxseg) tp->t_maxseg = so->so_snd.sb_hiwat; SOCKBUF_UNLOCK(&so->so_snd); TCPSTAT_INC(tcps_mturesent); tp->t_rtttime = 0; tp->snd_nxt = tp->snd_una; tcp_free_sackholes(tp); tp->snd_recover = tp->snd_max; if (tp->t_flags & TF_SACK_PERMIT) EXIT_FASTRECOVERY(tp->t_flags); tp->t_fb->tfb_tcp_output(tp); } #ifdef INET /* * Look-up the routing entry to the peer of this inpcb. If no route * is found and it cannot be allocated, then return 0. This routine * is called by TCP routines that access the rmx structure and by * tcp_mss_update to get the peer/interface MTU. */ u_long tcp_maxmtu(struct in_conninfo *inc, struct tcp_ifcap *cap) { struct nhop4_extended nh4; struct ifnet *ifp; u_long maxmtu = 0; KASSERT(inc != NULL, ("tcp_maxmtu with NULL in_conninfo pointer")); if (inc->inc_faddr.s_addr != INADDR_ANY) { if (fib4_lookup_nh_ext(inc->inc_fibnum, inc->inc_faddr, NHR_REF, 0, &nh4) != 0) return (0); ifp = nh4.nh_ifp; maxmtu = nh4.nh_mtu; /* Report additional interface capabilities. */ if (cap != NULL) { if (ifp->if_capenable & IFCAP_TSO4 && ifp->if_hwassist & CSUM_TSO) { cap->ifcap |= CSUM_TSO; cap->tsomax = ifp->if_hw_tsomax; cap->tsomaxsegcount = ifp->if_hw_tsomaxsegcount; cap->tsomaxsegsize = ifp->if_hw_tsomaxsegsize; } } fib4_free_nh_ext(inc->inc_fibnum, &nh4); } return (maxmtu); } #endif /* INET */ #ifdef INET6 u_long tcp_maxmtu6(struct in_conninfo *inc, struct tcp_ifcap *cap) { struct nhop6_extended nh6; struct in6_addr dst6; uint32_t scopeid; struct ifnet *ifp; u_long maxmtu = 0; KASSERT(inc != NULL, ("tcp_maxmtu6 with NULL in_conninfo pointer")); if (!IN6_IS_ADDR_UNSPECIFIED(&inc->inc6_faddr)) { in6_splitscope(&inc->inc6_faddr, &dst6, &scopeid); if (fib6_lookup_nh_ext(inc->inc_fibnum, &dst6, scopeid, 0, 0, &nh6) != 0) return (0); ifp = nh6.nh_ifp; maxmtu = nh6.nh_mtu; /* Report additional interface capabilities. */ if (cap != NULL) { if (ifp->if_capenable & IFCAP_TSO6 && ifp->if_hwassist & CSUM_TSO) { cap->ifcap |= CSUM_TSO; cap->tsomax = ifp->if_hw_tsomax; cap->tsomaxsegcount = ifp->if_hw_tsomaxsegcount; cap->tsomaxsegsize = ifp->if_hw_tsomaxsegsize; } } fib6_free_nh_ext(inc->inc_fibnum, &nh6); } return (maxmtu); } #endif /* INET6 */ /* * Calculate effective SMSS per RFC5681 definition for a given TCP * connection at its current state, taking into account SACK and etc. */ u_int tcp_maxseg(const struct tcpcb *tp) { u_int optlen; if (tp->t_flags & TF_NOOPT) return (tp->t_maxseg); /* * Here we have a simplified code from tcp_addoptions(), * without a proper loop, and having most of paddings hardcoded. * We might make mistakes with padding here in some edge cases, * but this is harmless, since result of tcp_maxseg() is used * only in cwnd and ssthresh estimations. */ #define PAD(len) ((((len) / 4) + !!((len) % 4)) * 4) if (TCPS_HAVEESTABLISHED(tp->t_state)) { if (tp->t_flags & TF_RCVD_TSTMP) optlen = TCPOLEN_TSTAMP_APPA; else optlen = 0; #ifdef TCP_SIGNATURE if (tp->t_flags & TF_SIGNATURE) optlen += PAD(TCPOLEN_SIGNATURE); #endif if ((tp->t_flags & TF_SACK_PERMIT) && tp->rcv_numsacks > 0) { optlen += TCPOLEN_SACKHDR; optlen += tp->rcv_numsacks * TCPOLEN_SACK; optlen = PAD(optlen); } } else { if (tp->t_flags & TF_REQ_TSTMP) optlen = TCPOLEN_TSTAMP_APPA; else optlen = PAD(TCPOLEN_MAXSEG); if (tp->t_flags & TF_REQ_SCALE) optlen += PAD(TCPOLEN_WINDOW); #ifdef TCP_SIGNATURE if (tp->t_flags & TF_SIGNATURE) optlen += PAD(TCPOLEN_SIGNATURE); #endif if (tp->t_flags & TF_SACK_PERMIT) optlen += PAD(TCPOLEN_SACK_PERMITTED); } #undef PAD optlen = min(optlen, TCP_MAXOLEN); return (tp->t_maxseg - optlen); } #ifdef IPSEC /* compute ESP/AH header size for TCP, including outer IP header. */ size_t ipsec_hdrsiz_tcp(struct tcpcb *tp) { struct inpcb *inp; struct mbuf *m; size_t hdrsiz; struct ip *ip; #ifdef INET6 struct ip6_hdr *ip6; #endif struct tcphdr *th; if ((tp == NULL) || ((inp = tp->t_inpcb) == NULL) || (!key_havesp(IPSEC_DIR_OUTBOUND))) return (0); m = m_gethdr(M_NOWAIT, MT_DATA); if (!m) return (0); #ifdef INET6 if ((inp->inp_vflag & INP_IPV6) != 0) { ip6 = mtod(m, struct ip6_hdr *); th = (struct tcphdr *)(ip6 + 1); m->m_pkthdr.len = m->m_len = sizeof(struct ip6_hdr) + sizeof(struct tcphdr); tcpip_fillheaders(inp, ip6, th); hdrsiz = ipsec_hdrsiz(m, IPSEC_DIR_OUTBOUND, inp); } else #endif /* INET6 */ { ip = mtod(m, struct ip *); th = (struct tcphdr *)(ip + 1); m->m_pkthdr.len = m->m_len = sizeof(struct tcpiphdr); tcpip_fillheaders(inp, ip, th); hdrsiz = ipsec_hdrsiz(m, IPSEC_DIR_OUTBOUND, inp); } m_free(m); return (hdrsiz); } #endif /* IPSEC */ #ifdef TCP_SIGNATURE /* * Callback function invoked by m_apply() to digest TCP segment data * contained within an mbuf chain. */ static int tcp_signature_apply(void *fstate, void *data, u_int len) { MD5Update(fstate, (u_char *)data, len); return (0); } /* * XXX The key is retrieved from the system's PF_KEY SADB, by keying a * search with the destination IP address, and a 'magic SPI' to be * determined by the application. This is hardcoded elsewhere to 1179 */ struct secasvar * tcp_get_sav(struct mbuf *m, u_int direction) { union sockaddr_union dst; struct secasvar *sav; struct ip *ip; #ifdef INET6 struct ip6_hdr *ip6; char ip6buf[INET6_ADDRSTRLEN]; #endif /* Extract the destination from the IP header in the mbuf. */ bzero(&dst, sizeof(union sockaddr_union)); ip = mtod(m, struct ip *); #ifdef INET6 ip6 = NULL; /* Make the compiler happy. */ #endif switch (ip->ip_v) { #ifdef INET case IPVERSION: dst.sa.sa_len = sizeof(struct sockaddr_in); dst.sa.sa_family = AF_INET; dst.sin.sin_addr = (direction == IPSEC_DIR_INBOUND) ? ip->ip_src : ip->ip_dst; break; #endif #ifdef INET6 case (IPV6_VERSION >> 4): ip6 = mtod(m, struct ip6_hdr *); dst.sa.sa_len = sizeof(struct sockaddr_in6); dst.sa.sa_family = AF_INET6; dst.sin6.sin6_addr = (direction == IPSEC_DIR_INBOUND) ? ip6->ip6_src : ip6->ip6_dst; break; #endif default: return (NULL); /* NOTREACHED */ break; } /* Look up an SADB entry which matches the address of the peer. */ sav = KEY_ALLOCSA(&dst, IPPROTO_TCP, htonl(TCP_SIG_SPI)); if (sav == NULL) { ipseclog((LOG_ERR, "%s: SADB lookup failed for %s\n", __func__, (ip->ip_v == IPVERSION) ? inet_ntoa(dst.sin.sin_addr) : #ifdef INET6 (ip->ip_v == (IPV6_VERSION >> 4)) ? ip6_sprintf(ip6buf, &dst.sin6.sin6_addr) : #endif "(unsupported)")); } return (sav); } /* * Compute TCP-MD5 hash of a TCP segment. (RFC2385) * * Parameters: * m pointer to head of mbuf chain * len length of TCP segment data, excluding options * optlen length of TCP segment options * buf pointer to storage for computed MD5 digest * sav pointer to security assosiation * * We do this over ip, tcphdr, segment data, and the key in the SADB. * When called from tcp_input(), we can be sure that th_sum has been * zeroed out and verified already. * * Releases reference to SADB key before return. * * Return 0 if successful, otherwise return -1. * */ int tcp_signature_do_compute(struct mbuf *m, int len, int optlen, u_char *buf, struct secasvar *sav) { #ifdef INET struct ippseudo ippseudo; #endif MD5_CTX ctx; int doff; struct ip *ip; #ifdef INET struct ipovly *ipovly; #endif struct tcphdr *th; #ifdef INET6 struct ip6_hdr *ip6; struct in6_addr in6; uint32_t plen; uint16_t nhdr; #endif u_short savecsum; KASSERT(m != NULL, ("NULL mbuf chain")); KASSERT(buf != NULL, ("NULL signature pointer")); /* Extract the destination from the IP header in the mbuf. */ ip = mtod(m, struct ip *); #ifdef INET6 ip6 = NULL; /* Make the compiler happy. */ #endif MD5Init(&ctx); /* * Step 1: Update MD5 hash with IP(v6) pseudo-header. * * XXX The ippseudo header MUST be digested in network byte order, * or else we'll fail the regression test. Assume all fields we've * been doing arithmetic on have been in host byte order. * XXX One cannot depend on ipovly->ih_len here. When called from * tcp_output(), the underlying ip_len member has not yet been set. */ switch (ip->ip_v) { #ifdef INET case IPVERSION: ipovly = (struct ipovly *)ip; ippseudo.ippseudo_src = ipovly->ih_src; ippseudo.ippseudo_dst = ipovly->ih_dst; ippseudo.ippseudo_pad = 0; ippseudo.ippseudo_p = IPPROTO_TCP; ippseudo.ippseudo_len = htons(len + sizeof(struct tcphdr) + optlen); MD5Update(&ctx, (char *)&ippseudo, sizeof(struct ippseudo)); th = (struct tcphdr *)((u_char *)ip + sizeof(struct ip)); doff = sizeof(struct ip) + sizeof(struct tcphdr) + optlen; break; #endif #ifdef INET6 /* * RFC 2385, 2.0 Proposal * For IPv6, the pseudo-header is as described in RFC 2460, namely the * 128-bit source IPv6 address, 128-bit destination IPv6 address, zero- * extended next header value (to form 32 bits), and 32-bit segment * length. * Note: Upper-Layer Packet Length comes before Next Header. */ case (IPV6_VERSION >> 4): in6 = ip6->ip6_src; in6_clearscope(&in6); MD5Update(&ctx, (char *)&in6, sizeof(struct in6_addr)); in6 = ip6->ip6_dst; in6_clearscope(&in6); MD5Update(&ctx, (char *)&in6, sizeof(struct in6_addr)); plen = htonl(len + sizeof(struct tcphdr) + optlen); MD5Update(&ctx, (char *)&plen, sizeof(uint32_t)); nhdr = 0; MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); nhdr = IPPROTO_TCP; MD5Update(&ctx, (char *)&nhdr, sizeof(uint8_t)); th = (struct tcphdr *)((u_char *)ip6 + sizeof(struct ip6_hdr)); doff = sizeof(struct ip6_hdr) + sizeof(struct tcphdr) + optlen; break; #endif default: KEY_FREESAV(&sav); return (-1); /* NOTREACHED */ break; } /* * Step 2: Update MD5 hash with TCP header, excluding options. * The TCP checksum must be set to zero. */ savecsum = th->th_sum; th->th_sum = 0; MD5Update(&ctx, (char *)th, sizeof(struct tcphdr)); th->th_sum = savecsum; /* * Step 3: Update MD5 hash with TCP segment data. * Use m_apply() to avoid an early m_pullup(). */ if (len > 0) m_apply(m, doff, len, tcp_signature_apply, &ctx); /* * Step 4: Update MD5 hash with shared secret. */ MD5Update(&ctx, sav->key_auth->key_data, _KEYLEN(sav->key_auth)); MD5Final(buf, &ctx); key_sa_recordxfer(sav, m); KEY_FREESAV(&sav); return (0); } /* * Compute TCP-MD5 hash of a TCP segment. (RFC2385) * * Return 0 if successful, otherwise return -1. */ int tcp_signature_compute(struct mbuf *m, int _unused, int len, int optlen, u_char *buf, u_int direction) { struct secasvar *sav; if ((sav = tcp_get_sav(m, direction)) == NULL) return (-1); return (tcp_signature_do_compute(m, len, optlen, buf, sav)); } /* * Verify the TCP-MD5 hash of a TCP segment. (RFC2385) * * Parameters: * m pointer to head of mbuf chain * len length of TCP segment data, excluding options * optlen length of TCP segment options * buf pointer to storage for computed MD5 digest * direction direction of flow (IPSEC_DIR_INBOUND or OUTBOUND) * * Return 1 if successful, otherwise return 0. */ int tcp_signature_verify(struct mbuf *m, int off0, int tlen, int optlen, struct tcpopt *to, struct tcphdr *th, u_int tcpbflag) { char tmpdigest[TCP_SIGLEN]; if (tcp_sig_checksigs == 0) return (1); if ((tcpbflag & TF_SIGNATURE) == 0) { if ((to->to_flags & TOF_SIGNATURE) != 0) { /* * If this socket is not expecting signature but * the segment contains signature just fail. */ TCPSTAT_INC(tcps_sig_err_sigopt); TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } /* Signature is not expected, and not present in segment. */ return (1); } /* * If this socket is expecting signature but the segment does not * contain any just fail. */ if ((to->to_flags & TOF_SIGNATURE) == 0) { TCPSTAT_INC(tcps_sig_err_nosigopt); TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } if (tcp_signature_compute(m, off0, tlen, optlen, &tmpdigest[0], IPSEC_DIR_INBOUND) == -1) { TCPSTAT_INC(tcps_sig_err_buildsig); TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } if (bcmp(to->to_signature, &tmpdigest[0], TCP_SIGLEN) != 0) { TCPSTAT_INC(tcps_sig_rcvbadsig); return (0); } TCPSTAT_INC(tcps_sig_rcvgoodsig); return (1); } #endif /* TCP_SIGNATURE */ static int sysctl_drop(SYSCTL_HANDLER_ARGS) { /* addrs[0] is a foreign socket, addrs[1] is a local one. */ struct sockaddr_storage addrs[2]; struct inpcb *inp; struct tcpcb *tp; struct tcptw *tw; struct sockaddr_in *fin, *lin; #ifdef INET6 struct sockaddr_in6 *fin6, *lin6; #endif int error; inp = NULL; fin = lin = NULL; #ifdef INET6 fin6 = lin6 = NULL; #endif error = 0; if (req->oldptr != NULL || req->oldlen != 0) return (EINVAL); if (req->newptr == NULL) return (EPERM); if (req->newlen < sizeof(addrs)) return (ENOMEM); error = SYSCTL_IN(req, &addrs, sizeof(addrs)); if (error) return (error); switch (addrs[0].ss_family) { #ifdef INET6 case AF_INET6: fin6 = (struct sockaddr_in6 *)&addrs[0]; lin6 = (struct sockaddr_in6 *)&addrs[1]; if (fin6->sin6_len != sizeof(struct sockaddr_in6) || lin6->sin6_len != sizeof(struct sockaddr_in6)) return (EINVAL); if (IN6_IS_ADDR_V4MAPPED(&fin6->sin6_addr)) { if (!IN6_IS_ADDR_V4MAPPED(&lin6->sin6_addr)) return (EINVAL); in6_sin6_2_sin_in_sock((struct sockaddr *)&addrs[0]); in6_sin6_2_sin_in_sock((struct sockaddr *)&addrs[1]); fin = (struct sockaddr_in *)&addrs[0]; lin = (struct sockaddr_in *)&addrs[1]; break; } error = sa6_embedscope(fin6, V_ip6_use_defzone); if (error) return (error); error = sa6_embedscope(lin6, V_ip6_use_defzone); if (error) return (error); break; #endif #ifdef INET case AF_INET: fin = (struct sockaddr_in *)&addrs[0]; lin = (struct sockaddr_in *)&addrs[1]; if (fin->sin_len != sizeof(struct sockaddr_in) || lin->sin_len != sizeof(struct sockaddr_in)) return (EINVAL); break; #endif default: return (EINVAL); } INP_INFO_RLOCK(&V_tcbinfo); switch (addrs[0].ss_family) { #ifdef INET6 case AF_INET6: inp = in6_pcblookup(&V_tcbinfo, &fin6->sin6_addr, fin6->sin6_port, &lin6->sin6_addr, lin6->sin6_port, INPLOOKUP_WLOCKPCB, NULL); break; #endif #ifdef INET case AF_INET: inp = in_pcblookup(&V_tcbinfo, fin->sin_addr, fin->sin_port, lin->sin_addr, lin->sin_port, INPLOOKUP_WLOCKPCB, NULL); break; #endif } if (inp != NULL) { if (inp->inp_flags & INP_TIMEWAIT) { /* * XXXRW: There currently exists a state where an * inpcb is present, but its timewait state has been * discarded. For now, don't allow dropping of this * type of inpcb. */ tw = intotw(inp); if (tw != NULL) tcp_twclose(tw, 0); else INP_WUNLOCK(inp); } else if (!(inp->inp_flags & INP_DROPPED) && !(inp->inp_socket->so_options & SO_ACCEPTCONN)) { tp = intotcpcb(inp); tp = tcp_drop(tp, ECONNABORTED); if (tp != NULL) INP_WUNLOCK(inp); } else INP_WUNLOCK(inp); } else error = ESRCH; INP_INFO_RUNLOCK(&V_tcbinfo); return (error); } SYSCTL_PROC(_net_inet_tcp, TCPCTL_DROP, drop, CTLFLAG_VNET | CTLTYPE_STRUCT | CTLFLAG_WR | CTLFLAG_SKIP, NULL, 0, sysctl_drop, "", "Drop TCP connection"); /* * Generate a standardized TCP log line for use throughout the * tcp subsystem. Memory allocation is done with M_NOWAIT to * allow use in the interrupt context. * * NB: The caller MUST free(s, M_TCPLOG) the returned string. * NB: The function may return NULL if memory allocation failed. * * Due to header inclusion and ordering limitations the struct ip * and ip6_hdr pointers have to be passed as void pointers. */ char * tcp_log_vain(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr) { /* Is logging enabled? */ if (tcp_log_in_vain == 0) return (NULL); return (tcp_log_addr(inc, th, ip4hdr, ip6hdr)); } char * tcp_log_addrs(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr) { /* Is logging enabled? */ if (tcp_log_debug == 0) return (NULL); return (tcp_log_addr(inc, th, ip4hdr, ip6hdr)); } static char * tcp_log_addr(struct in_conninfo *inc, struct tcphdr *th, void *ip4hdr, const void *ip6hdr) { char *s, *sp; size_t size; struct ip *ip; #ifdef INET6 const struct ip6_hdr *ip6; ip6 = (const struct ip6_hdr *)ip6hdr; #endif /* INET6 */ ip = (struct ip *)ip4hdr; /* * The log line looks like this: * "TCP: [1.2.3.4]:50332 to [1.2.3.4]:80 tcpflags 0x2" */ size = sizeof("TCP: []:12345 to []:12345 tcpflags 0x2<>") + sizeof(PRINT_TH_FLAGS) + 1 + #ifdef INET6 2 * INET6_ADDRSTRLEN; #else 2 * INET_ADDRSTRLEN; #endif /* INET6 */ s = malloc(size, M_TCPLOG, M_ZERO|M_NOWAIT); if (s == NULL) return (NULL); strcat(s, "TCP: ["); sp = s + strlen(s); if (inc && ((inc->inc_flags & INC_ISIPV6) == 0)) { inet_ntoa_r(inc->inc_faddr, sp); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(inc->inc_fport)); sp = s + strlen(s); inet_ntoa_r(inc->inc_laddr, sp); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(inc->inc_lport)); #ifdef INET6 } else if (inc) { ip6_sprintf(sp, &inc->inc6_faddr); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(inc->inc_fport)); sp = s + strlen(s); ip6_sprintf(sp, &inc->inc6_laddr); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(inc->inc_lport)); } else if (ip6 && th) { ip6_sprintf(sp, &ip6->ip6_src); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(th->th_sport)); sp = s + strlen(s); ip6_sprintf(sp, &ip6->ip6_dst); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(th->th_dport)); #endif /* INET6 */ #ifdef INET } else if (ip && th) { inet_ntoa_r(ip->ip_src, sp); sp = s + strlen(s); sprintf(sp, "]:%i to [", ntohs(th->th_sport)); sp = s + strlen(s); inet_ntoa_r(ip->ip_dst, sp); sp = s + strlen(s); sprintf(sp, "]:%i", ntohs(th->th_dport)); #endif /* INET */ } else { free(s, M_TCPLOG); return (NULL); } sp = s + strlen(s); if (th) sprintf(sp, " tcpflags 0x%b", th->th_flags, PRINT_TH_FLAGS); if (*(s + size - 1) != '\0') panic("%s: string too long", __func__); return (s); } /* * A subroutine which makes it easy to track TCP state changes with DTrace. * This function shouldn't be called for t_state initializations that don't * correspond to actual TCP state transitions. */ void tcp_state_change(struct tcpcb *tp, int newstate) { #if defined(KDTRACE_HOOKS) int pstate = tp->t_state; #endif TCPSTATES_DEC(tp->t_state); TCPSTATES_INC(newstate); tp->t_state = newstate; TCP_PROBE6(state__change, NULL, tp, NULL, tp, NULL, pstate); } Index: projects/vnet/sys/netinet6/sctp6_usrreq.c =================================================================== --- projects/vnet/sys/netinet6/sctp6_usrreq.c (revision 302157) +++ projects/vnet/sys/netinet6/sctp6_usrreq.c (revision 302158) @@ -1,1202 +1,1201 @@ /*- * Copyright (c) 2001-2007, by Cisco Systems, Inc. All rights reserved. * Copyright (c) 2008-2012, by Randall Stewart. All rights reserved. * Copyright (c) 2008-2012, by Michael Tuexen. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * a) Redistributions of source code must retain the above copyright notice, * this list of conditions and the following disclaimer. * * b) Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in * the documentation and/or other materials provided with the distribution. * * c) Neither the name of Cisco Systems, Inc. nor the names of its * contributors may be used to endorse or promote products derived * from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF * THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #ifdef INET6 #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef IPSEC #include #include #endif /* IPSEC */ extern struct protosw inetsw[]; int sctp6_input_with_port(struct mbuf **i_pak, int *offp, uint16_t port) { struct mbuf *m; int iphlen; uint32_t vrf_id; uint8_t ecn_bits; struct sockaddr_in6 src, dst; struct ip6_hdr *ip6; struct sctphdr *sh; struct sctp_chunkhdr *ch; int length, offset; #if !defined(SCTP_WITH_NO_CSUM) uint8_t compute_crc; #endif uint32_t mflowid; uint8_t mflowtype; uint16_t fibnum; iphlen = *offp; if (SCTP_GET_PKT_VRFID(*i_pak, vrf_id)) { SCTP_RELEASE_PKT(*i_pak); return (IPPROTO_DONE); } m = SCTP_HEADER_TO_CHAIN(*i_pak); #ifdef SCTP_MBUF_LOGGING /* Log in any input mbufs */ if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_MBUF_LOGGING_ENABLE) { sctp_log_mbc(m, SCTP_MBUF_INPUT); } #endif #ifdef SCTP_PACKET_LOGGING if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_LAST_PACKET_TRACING) { sctp_packet_log(m); } #endif SCTPDBG(SCTP_DEBUG_CRCOFFLOAD, "sctp6_input(): Packet of length %d received on %s with csum_flags 0x%b.\n", m->m_pkthdr.len, if_name(m->m_pkthdr.rcvif), (int)m->m_pkthdr.csum_flags, CSUM_BITS); mflowid = m->m_pkthdr.flowid; mflowtype = M_HASHTYPE_GET(m); fibnum = M_GETFIB(m); SCTP_STAT_INCR(sctps_recvpackets); SCTP_STAT_INCR_COUNTER64(sctps_inpackets); /* Get IP, SCTP, and first chunk header together in the first mbuf. */ offset = iphlen + sizeof(struct sctphdr) + sizeof(struct sctp_chunkhdr); ip6 = mtod(m, struct ip6_hdr *); IP6_EXTHDR_GET(sh, struct sctphdr *, m, iphlen, (int)(sizeof(struct sctphdr) + sizeof(struct sctp_chunkhdr))); if (sh == NULL) { SCTP_STAT_INCR(sctps_hdrops); return (IPPROTO_DONE); } ch = (struct sctp_chunkhdr *)((caddr_t)sh + sizeof(struct sctphdr)); offset -= sizeof(struct sctp_chunkhdr); memset(&src, 0, sizeof(struct sockaddr_in6)); src.sin6_family = AF_INET6; src.sin6_len = sizeof(struct sockaddr_in6); src.sin6_port = sh->src_port; src.sin6_addr = ip6->ip6_src; if (in6_setscope(&src.sin6_addr, m->m_pkthdr.rcvif, NULL) != 0) { goto out; } memset(&dst, 0, sizeof(struct sockaddr_in6)); dst.sin6_family = AF_INET6; dst.sin6_len = sizeof(struct sockaddr_in6); dst.sin6_port = sh->dest_port; dst.sin6_addr = ip6->ip6_dst; if (in6_setscope(&dst.sin6_addr, m->m_pkthdr.rcvif, NULL) != 0) { goto out; } length = ntohs(ip6->ip6_plen) + iphlen; /* Validate mbuf chain length with IP payload length. */ if (SCTP_HEADER_LEN(m) != length) { SCTPDBG(SCTP_DEBUG_INPUT1, "sctp6_input() length:%d reported length:%d\n", length, SCTP_HEADER_LEN(m)); SCTP_STAT_INCR(sctps_hdrops); goto out; } if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { goto out; } ecn_bits = ((ntohl(ip6->ip6_flow) >> 20) & 0x000000ff); #if defined(SCTP_WITH_NO_CSUM) SCTP_STAT_INCR(sctps_recvnocrc); #else if (m->m_pkthdr.csum_flags & CSUM_SCTP_VALID) { SCTP_STAT_INCR(sctps_recvhwcrc); compute_crc = 0; } else { SCTP_STAT_INCR(sctps_recvswcrc); compute_crc = 1; } #endif sctp_common_input_processing(&m, iphlen, offset, length, (struct sockaddr *)&src, (struct sockaddr *)&dst, sh, ch, #if !defined(SCTP_WITH_NO_CSUM) compute_crc, #endif ecn_bits, mflowtype, mflowid, fibnum, vrf_id, port); out: if (m) { sctp_m_freem(m); } return (IPPROTO_DONE); } int sctp6_input(struct mbuf **i_pak, int *offp, int proto SCTP_UNUSED) { return (sctp6_input_with_port(i_pak, offp, 0)); } void sctp6_notify(struct sctp_inpcb *inp, struct sctp_tcb *stcb, struct sctp_nets *net, uint8_t icmp6_type, uint8_t icmp6_code, uint16_t next_mtu) { #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) struct socket *so; #endif int timer_stopped; switch (icmp6_type) { case ICMP6_DST_UNREACH: if ((icmp6_code == ICMP6_DST_UNREACH_NOROUTE) || (icmp6_code == ICMP6_DST_UNREACH_ADMIN) || (icmp6_code == ICMP6_DST_UNREACH_BEYONDSCOPE) || (icmp6_code == ICMP6_DST_UNREACH_ADDR)) { /* Mark the net unreachable. */ if (net->dest_state & SCTP_ADDR_REACHABLE) { /* Ok that destination is not reachable */ net->dest_state &= ~SCTP_ADDR_REACHABLE; net->dest_state &= ~SCTP_ADDR_PF; sctp_ulp_notify(SCTP_NOTIFY_INTERFACE_DOWN, stcb, 0, (void *)net, SCTP_SO_NOT_LOCKED); } } SCTP_TCB_UNLOCK(stcb); break; case ICMP6_PARAM_PROB: /* Treat it like an ABORT. */ if (icmp6_code == ICMP6_PARAMPROB_NEXTHEADER) { sctp_abort_notification(stcb, 1, 0, NULL, SCTP_SO_NOT_LOCKED); #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) so = SCTP_INP_SO(inp); atomic_add_int(&stcb->asoc.refcnt, 1); SCTP_TCB_UNLOCK(stcb); SCTP_SOCKET_LOCK(so, 1); SCTP_TCB_LOCK(stcb); atomic_subtract_int(&stcb->asoc.refcnt, 1); #endif (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_2); #if defined(__APPLE__) || defined(SCTP_SO_LOCK_TESTING) SCTP_SOCKET_UNLOCK(so, 1); #endif } else { SCTP_TCB_UNLOCK(stcb); } break; case ICMP6_PACKET_TOO_BIG: if (SCTP_OS_TIMER_PENDING(&net->pmtu_timer.timer)) { timer_stopped = 1; sctp_timer_stop(SCTP_TIMER_TYPE_PATHMTURAISE, inp, stcb, net, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_1); } else { timer_stopped = 0; } /* Update the path MTU. */ if (net->mtu > next_mtu) { net->mtu = next_mtu; if (net->port) { net->mtu -= sizeof(struct udphdr); } } /* Update the association MTU */ if (stcb->asoc.smallest_mtu > next_mtu) { sctp_pathmtu_adjustment(stcb, next_mtu); } /* Finally, start the PMTU timer if it was running before. */ if (timer_stopped) { sctp_timer_start(SCTP_TIMER_TYPE_PATHMTURAISE, inp, stcb, net); } SCTP_TCB_UNLOCK(stcb); break; default: SCTP_TCB_UNLOCK(stcb); break; } } void sctp6_ctlinput(int cmd, struct sockaddr *pktdst, void *d) { struct ip6ctlparam *ip6cp; struct sctp_inpcb *inp; struct sctp_tcb *stcb; struct sctp_nets *net; struct sctphdr sh; struct sockaddr_in6 src, dst; if (pktdst->sa_family != AF_INET6 || pktdst->sa_len != sizeof(struct sockaddr_in6)) { return; } if ((unsigned)cmd >= PRC_NCMDS) { return; } if (PRC_IS_REDIRECT(cmd)) { d = NULL; } else if (inet6ctlerrmap[cmd] == 0) { return; } /* If the parameter is from icmp6, decode it. */ if (d != NULL) { ip6cp = (struct ip6ctlparam *)d; } else { ip6cp = (struct ip6ctlparam *)NULL; } if (ip6cp != NULL) { /* * XXX: We assume that when IPV6 is non NULL, M and OFF are * valid. */ if (ip6cp->ip6c_m == NULL) { return; } /* * Check if we can safely examine the ports and the * verification tag of the SCTP common header. */ if (ip6cp->ip6c_m->m_pkthdr.len < (int32_t) (ip6cp->ip6c_off + offsetof(struct sctphdr, checksum))) { return; } /* Copy out the port numbers and the verification tag. */ bzero(&sh, sizeof(sh)); m_copydata(ip6cp->ip6c_m, ip6cp->ip6c_off, sizeof(uint16_t) + sizeof(uint16_t) + sizeof(uint32_t), (caddr_t)&sh); memset(&src, 0, sizeof(struct sockaddr_in6)); src.sin6_family = AF_INET6; src.sin6_len = sizeof(struct sockaddr_in6); src.sin6_port = sh.src_port; src.sin6_addr = ip6cp->ip6c_ip6->ip6_src; if (in6_setscope(&src.sin6_addr, ip6cp->ip6c_m->m_pkthdr.rcvif, NULL) != 0) { return; } memset(&dst, 0, sizeof(struct sockaddr_in6)); dst.sin6_family = AF_INET6; dst.sin6_len = sizeof(struct sockaddr_in6); dst.sin6_port = sh.dest_port; dst.sin6_addr = ip6cp->ip6c_ip6->ip6_dst; if (in6_setscope(&dst.sin6_addr, ip6cp->ip6c_m->m_pkthdr.rcvif, NULL) != 0) { return; } inp = NULL; net = NULL; stcb = sctp_findassociation_addr_sa((struct sockaddr *)&dst, (struct sockaddr *)&src, &inp, &net, 1, SCTP_DEFAULT_VRFID); if ((stcb != NULL) && (net != NULL) && - (inp != NULL) && - (inp->sctp_socket != NULL)) { + (inp != NULL)) { /* Check the verification tag */ if (ntohl(sh.v_tag) != 0) { /* * This must be the verification tag used * for sending out packets. We don't * consider packets reflecting the * verification tag. */ if (ntohl(sh.v_tag) != stcb->asoc.peer_vtag) { SCTP_TCB_UNLOCK(stcb); return; } } else { if (ip6cp->ip6c_m->m_pkthdr.len >= ip6cp->ip6c_off + sizeof(struct sctphdr) + sizeof(struct sctp_chunkhdr) + offsetof(struct sctp_init, a_rwnd)) { /* * In this case we can check if we * got an INIT chunk and if the * initiate tag matches. */ uint32_t initiate_tag; uint8_t chunk_type; m_copydata(ip6cp->ip6c_m, ip6cp->ip6c_off + sizeof(struct sctphdr), sizeof(uint8_t), (caddr_t)&chunk_type); m_copydata(ip6cp->ip6c_m, ip6cp->ip6c_off + sizeof(struct sctphdr) + sizeof(struct sctp_chunkhdr), sizeof(uint32_t), (caddr_t)&initiate_tag); if ((chunk_type != SCTP_INITIATION) || (ntohl(initiate_tag) != stcb->asoc.my_vtag)) { SCTP_TCB_UNLOCK(stcb); return; } } else { SCTP_TCB_UNLOCK(stcb); return; } } sctp6_notify(inp, stcb, net, ip6cp->ip6c_icmp6->icmp6_type, ip6cp->ip6c_icmp6->icmp6_code, (uint16_t) ntohl(ip6cp->ip6c_icmp6->icmp6_mtu)); } else { if ((stcb == NULL) && (inp != NULL)) { /* reduce inp's ref-count */ SCTP_INP_WLOCK(inp); SCTP_INP_DECR_REF(inp); SCTP_INP_WUNLOCK(inp); } if (stcb) { SCTP_TCB_UNLOCK(stcb); } } } } /* * this routine can probably be collasped into the one in sctp_userreq.c * since they do the same thing and now we lookup with a sockaddr */ static int sctp6_getcred(SYSCTL_HANDLER_ARGS) { struct xucred xuc; struct sockaddr_in6 addrs[2]; struct sctp_inpcb *inp; struct sctp_nets *net; struct sctp_tcb *stcb; int error; uint32_t vrf_id; vrf_id = SCTP_DEFAULT_VRFID; error = priv_check(req->td, PRIV_NETINET_GETCRED); if (error) return (error); if (req->newlen != sizeof(addrs)) { SCTP_LTRACE_ERR_RET(NULL, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } if (req->oldlen != sizeof(struct ucred)) { SCTP_LTRACE_ERR_RET(NULL, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } error = SYSCTL_IN(req, addrs, sizeof(addrs)); if (error) return (error); stcb = sctp_findassociation_addr_sa(sin6tosa(&addrs[1]), sin6tosa(&addrs[0]), &inp, &net, 1, vrf_id); if (stcb == NULL || inp == NULL || inp->sctp_socket == NULL) { if ((inp != NULL) && (stcb == NULL)) { /* reduce ref-count */ SCTP_INP_WLOCK(inp); SCTP_INP_DECR_REF(inp); goto cred_can_cont; } SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, ENOENT); error = ENOENT; goto out; } SCTP_TCB_UNLOCK(stcb); /* * We use the write lock here, only since in the error leg we need * it. If we used RLOCK, then we would have to * wlock/decr/unlock/rlock. Which in theory could create a hole. * Better to use higher wlock. */ SCTP_INP_WLOCK(inp); cred_can_cont: error = cr_canseesocket(req->td->td_ucred, inp->sctp_socket); if (error) { SCTP_INP_WUNLOCK(inp); goto out; } cru2x(inp->sctp_socket->so_cred, &xuc); SCTP_INP_WUNLOCK(inp); error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred)); out: return (error); } SYSCTL_PROC(_net_inet6_sctp6, OID_AUTO, getcred, CTLTYPE_OPAQUE | CTLFLAG_RW, 0, 0, sctp6_getcred, "S,ucred", "Get the ucred of a SCTP6 connection"); /* This is the same as the sctp_abort() could be made common */ static void sctp6_abort(struct socket *so) { struct sctp_inpcb *inp; uint32_t flags; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return; } sctp_must_try_again: flags = inp->sctp_flags; #ifdef SCTP_LOG_CLOSING sctp_log_closing(inp, NULL, 17); #endif if (((flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) && (atomic_cmpset_int(&inp->sctp_flags, flags, (flags | SCTP_PCB_FLAGS_SOCKET_GONE | SCTP_PCB_FLAGS_CLOSE_IP)))) { #ifdef SCTP_LOG_CLOSING sctp_log_closing(inp, NULL, 16); #endif sctp_inpcb_free(inp, SCTP_FREE_SHOULD_USE_ABORT, SCTP_CALLED_AFTER_CMPSET_OFCLOSE); SOCK_LOCK(so); SCTP_SB_CLEAR(so->so_snd); /* * same for the rcv ones, they are only here for the * accounting/select. */ SCTP_SB_CLEAR(so->so_rcv); /* Now null out the reference, we are completely detached. */ so->so_pcb = NULL; SOCK_UNLOCK(so); } else { flags = inp->sctp_flags; if ((flags & SCTP_PCB_FLAGS_SOCKET_GONE) == 0) { goto sctp_must_try_again; } } return; } static int sctp6_attach(struct socket *so, int proto SCTP_UNUSED, struct thread *p SCTP_UNUSED) { struct in6pcb *inp6; int error; struct sctp_inpcb *inp; uint32_t vrf_id = SCTP_DEFAULT_VRFID; inp = (struct sctp_inpcb *)so->so_pcb; if (inp != NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { error = SCTP_SORESERVE(so, SCTP_BASE_SYSCTL(sctp_sendspace), SCTP_BASE_SYSCTL(sctp_recvspace)); if (error) return (error); } error = sctp_inpcb_alloc(so, vrf_id); if (error) return (error); inp = (struct sctp_inpcb *)so->so_pcb; SCTP_INP_WLOCK(inp); inp->sctp_flags |= SCTP_PCB_FLAGS_BOUND_V6; /* I'm v6! */ inp6 = (struct in6pcb *)inp; inp6->inp_vflag |= INP_IPV6; inp6->in6p_hops = -1; /* use kernel default */ inp6->in6p_cksum = -1; /* just to be sure */ #ifdef INET /* * 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. */ inp6->inp_ip_ttl = MODULE_GLOBAL(ip_defttl); #endif /* * Hmm what about the IPSEC stuff that is missing here but in * sctp_attach()? */ SCTP_INP_WUNLOCK(inp); return (0); } static int sctp6_bind(struct socket *so, struct sockaddr *addr, struct thread *p) { struct sctp_inpcb *inp; struct in6pcb *inp6; int error; inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } if (addr) { switch (addr->sa_family) { #ifdef INET case AF_INET: if (addr->sa_len != sizeof(struct sockaddr_in)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } break; #endif #ifdef INET6 case AF_INET6: if (addr->sa_len != sizeof(struct sockaddr_in6)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } break; #endif default: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } } inp6 = (struct in6pcb *)inp; inp6->inp_vflag &= ~INP_IPV4; inp6->inp_vflag |= INP_IPV6; if ((addr != NULL) && (SCTP_IPV6_V6ONLY(inp6) == 0)) { switch (addr->sa_family) { #ifdef INET case AF_INET: /* binding v4 addr to v6 socket, so reset flags */ inp6->inp_vflag |= INP_IPV4; inp6->inp_vflag &= ~INP_IPV6; break; #endif #ifdef INET6 case AF_INET6: { struct sockaddr_in6 *sin6_p; sin6_p = (struct sockaddr_in6 *)addr; if (IN6_IS_ADDR_UNSPECIFIED(&sin6_p->sin6_addr)) { inp6->inp_vflag |= INP_IPV4; } #ifdef INET if (IN6_IS_ADDR_V4MAPPED(&sin6_p->sin6_addr)) { struct sockaddr_in sin; in6_sin6_2_sin(&sin, sin6_p); inp6->inp_vflag |= INP_IPV4; inp6->inp_vflag &= ~INP_IPV6; error = sctp_inpcb_bind(so, (struct sockaddr *)&sin, NULL, p); return (error); } #endif break; } #endif default: break; } } else if (addr != NULL) { struct sockaddr_in6 *sin6_p; /* IPV6_V6ONLY socket */ #ifdef INET if (addr->sa_family == AF_INET) { /* can't bind v4 addr to v6 only socket! */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } #endif sin6_p = (struct sockaddr_in6 *)addr; if (IN6_IS_ADDR_V4MAPPED(&sin6_p->sin6_addr)) { /* can't bind v4-mapped addrs either! */ /* NOTE: we don't support SIIT */ SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } } error = sctp_inpcb_bind(so, addr, NULL, p); return (error); } static void sctp6_close(struct socket *so) { sctp_close(so); } /* This could be made common with sctp_detach() since they are identical */ static int sctp6_disconnect(struct socket *so) { return (sctp_disconnect(so)); } int sctp_sendm(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr, struct mbuf *control, struct thread *p); static int sctp6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *addr, struct mbuf *control, struct thread *p) { struct sctp_inpcb *inp; struct in6pcb *inp6; #ifdef INET struct sockaddr_in6 *sin6; #endif /* INET */ /* No SPL needed since sctp_output does this */ inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { if (control) { SCTP_RELEASE_PKT(control); control = NULL; } SCTP_RELEASE_PKT(m); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } inp6 = (struct in6pcb *)inp; /* * For the TCP model we may get a NULL addr, if we are a connected * socket thats ok. */ if ((inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) && (addr == NULL)) { goto connected_type; } if (addr == NULL) { SCTP_RELEASE_PKT(m); if (control) { SCTP_RELEASE_PKT(control); control = NULL; } SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EDESTADDRREQ); return (EDESTADDRREQ); } #ifdef INET sin6 = (struct sockaddr_in6 *)addr; if (SCTP_IPV6_V6ONLY(inp6)) { /* * if IPV6_V6ONLY flag, we discard datagrams destined to a * v4 addr or v4-mapped addr */ if (addr->sa_family == AF_INET) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } } if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { struct sockaddr_in sin; /* convert v4-mapped into v4 addr and send */ in6_sin6_2_sin(&sin, sin6); return (sctp_sendm(so, flags, m, (struct sockaddr *)&sin, control, p)); } #endif /* INET */ connected_type: /* now what about control */ if (control) { if (inp->control) { SCTP_PRINTF("huh? control set?\n"); SCTP_RELEASE_PKT(inp->control); inp->control = NULL; } inp->control = control; } /* Place the data */ if (inp->pkt) { SCTP_BUF_NEXT(inp->pkt_last) = m; inp->pkt_last = m; } else { inp->pkt_last = inp->pkt = m; } if ( /* FreeBSD and MacOSX uses a flag passed */ ((flags & PRUS_MORETOCOME) == 0) ) { /* * note with the current version this code will only be used * by OpenBSD, NetBSD and FreeBSD have methods for * re-defining sosend() to use sctp_sosend(). One can * optionaly switch back to this code (by changing back the * defininitions but this is not advisable. */ int ret; ret = sctp_output(inp, inp->pkt, addr, inp->control, p, flags); inp->pkt = NULL; inp->control = NULL; return (ret); } else { return (0); } } static int sctp6_connect(struct socket *so, struct sockaddr *addr, struct thread *p) { uint32_t vrf_id; int error = 0; struct sctp_inpcb *inp; struct sctp_tcb *stcb; #ifdef INET struct in6pcb *inp6; struct sockaddr_in6 *sin6; union sctp_sockstore store; #endif #ifdef INET inp6 = (struct in6pcb *)so->so_pcb; #endif inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, ECONNRESET); return (ECONNRESET); /* I made the same as TCP since we are * not setup? */ } if (addr == NULL) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } switch (addr->sa_family) { #ifdef INET case AF_INET: if (addr->sa_len != sizeof(struct sockaddr_in)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } break; #endif #ifdef INET6 case AF_INET6: if (addr->sa_len != sizeof(struct sockaddr_in6)) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } break; #endif default: SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } vrf_id = inp->def_vrf_id; SCTP_ASOC_CREATE_LOCK(inp); SCTP_INP_RLOCK(inp); if ((inp->sctp_flags & SCTP_PCB_FLAGS_UNBOUND) == SCTP_PCB_FLAGS_UNBOUND) { /* Bind a ephemeral port */ SCTP_INP_RUNLOCK(inp); error = sctp6_bind(so, NULL, p); if (error) { SCTP_ASOC_CREATE_UNLOCK(inp); return (error); } SCTP_INP_RLOCK(inp); } if ((inp->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) && (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED)) { /* We are already connected AND the TCP model */ SCTP_INP_RUNLOCK(inp); SCTP_ASOC_CREATE_UNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EADDRINUSE); return (EADDRINUSE); } #ifdef INET sin6 = (struct sockaddr_in6 *)addr; if (SCTP_IPV6_V6ONLY(inp6)) { /* * if IPV6_V6ONLY flag, ignore connections destined to a v4 * addr or v4-mapped addr */ if (addr->sa_family == AF_INET) { SCTP_INP_RUNLOCK(inp); SCTP_ASOC_CREATE_UNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { SCTP_INP_RUNLOCK(inp); SCTP_ASOC_CREATE_UNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } } if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) { /* convert v4-mapped into v4 addr */ in6_sin6_2_sin(&store.sin, sin6); addr = &store.sa; } #endif /* INET */ /* Now do we connect? */ if (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) { stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb) { SCTP_TCB_UNLOCK(stcb); } SCTP_INP_RUNLOCK(inp); } else { SCTP_INP_RUNLOCK(inp); SCTP_INP_WLOCK(inp); SCTP_INP_INCR_REF(inp); SCTP_INP_WUNLOCK(inp); stcb = sctp_findassociation_ep_addr(&inp, addr, NULL, NULL, NULL); if (stcb == NULL) { SCTP_INP_WLOCK(inp); SCTP_INP_DECR_REF(inp); SCTP_INP_WUNLOCK(inp); } } if (stcb != NULL) { /* Already have or am bring up an association */ SCTP_ASOC_CREATE_UNLOCK(inp); SCTP_TCB_UNLOCK(stcb); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EALREADY); return (EALREADY); } /* We are GOOD to go */ stcb = sctp_aloc_assoc(inp, addr, &error, 0, vrf_id, inp->sctp_ep.pre_open_stream_count, inp->sctp_ep.port, p); SCTP_ASOC_CREATE_UNLOCK(inp); if (stcb == NULL) { /* Gak! no memory */ return (error); } if (stcb->sctp_ep->sctp_flags & SCTP_PCB_FLAGS_TCPTYPE) { stcb->sctp_ep->sctp_flags |= SCTP_PCB_FLAGS_CONNECTED; /* Set the connected flag so we can queue data */ soisconnecting(so); } stcb->asoc.state = SCTP_STATE_COOKIE_WAIT; (void)SCTP_GETTIME_TIMEVAL(&stcb->asoc.time_entered); /* initialize authentication parameters for the assoc */ sctp_initialize_auth_params(inp, stcb); sctp_send_initiate(inp, stcb, SCTP_SO_LOCKED); SCTP_TCB_UNLOCK(stcb); return (error); } static int sctp6_getaddr(struct socket *so, struct sockaddr **addr) { struct sockaddr_in6 *sin6; struct sctp_inpcb *inp; uint32_t vrf_id; struct sctp_ifa *sctp_ifa; int error; /* * Do the malloc first in case it blocks. */ SCTP_MALLOC_SONAME(sin6, struct sockaddr_in6 *, sizeof(*sin6)); if (sin6 == NULL) return (ENOMEM); sin6->sin6_family = AF_INET6; sin6->sin6_len = sizeof(*sin6); inp = (struct sctp_inpcb *)so->so_pcb; if (inp == NULL) { SCTP_FREE_SONAME(sin6); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, ECONNRESET); return (ECONNRESET); } SCTP_INP_RLOCK(inp); sin6->sin6_port = inp->sctp_lport; if (inp->sctp_flags & SCTP_PCB_FLAGS_BOUNDALL) { /* For the bound all case you get back 0 */ if (inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) { struct sctp_tcb *stcb; struct sockaddr_in6 *sin_a6; struct sctp_nets *net; int fnd; stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb == NULL) { SCTP_INP_RUNLOCK(inp); SCTP_FREE_SONAME(sin6); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, ENOENT); return (ENOENT); } fnd = 0; sin_a6 = NULL; TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { sin_a6 = (struct sockaddr_in6 *)&net->ro._l_addr; if (sin_a6 == NULL) /* this will make coverity happy */ continue; if (sin_a6->sin6_family == AF_INET6) { fnd = 1; break; } } if ((!fnd) || (sin_a6 == NULL)) { /* punt */ SCTP_INP_RUNLOCK(inp); SCTP_FREE_SONAME(sin6); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, ENOENT); return (ENOENT); } vrf_id = inp->def_vrf_id; sctp_ifa = sctp_source_address_selection(inp, stcb, (sctp_route_t *) & net->ro, net, 0, vrf_id); if (sctp_ifa) { sin6->sin6_addr = sctp_ifa->address.sin6.sin6_addr; } } else { /* For the bound all case you get back 0 */ memset(&sin6->sin6_addr, 0, sizeof(sin6->sin6_addr)); } } else { /* Take the first IPv6 address in the list */ struct sctp_laddr *laddr; int fnd = 0; LIST_FOREACH(laddr, &inp->sctp_addr_list, sctp_nxt_addr) { if (laddr->ifa->address.sa.sa_family == AF_INET6) { struct sockaddr_in6 *sin_a; sin_a = &laddr->ifa->address.sin6; sin6->sin6_addr = sin_a->sin6_addr; fnd = 1; break; } } if (!fnd) { SCTP_FREE_SONAME(sin6); SCTP_INP_RUNLOCK(inp); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, ENOENT); return (ENOENT); } } SCTP_INP_RUNLOCK(inp); /* Scoping things for v6 */ if ((error = sa6_recoverscope(sin6)) != 0) { SCTP_FREE_SONAME(sin6); return (error); } (*addr) = (struct sockaddr *)sin6; return (0); } static int sctp6_peeraddr(struct socket *so, struct sockaddr **addr) { struct sockaddr_in6 *sin6; int fnd; struct sockaddr_in6 *sin_a6; struct sctp_inpcb *inp; struct sctp_tcb *stcb; struct sctp_nets *net; int error; /* Do the malloc first in case it blocks. */ SCTP_MALLOC_SONAME(sin6, struct sockaddr_in6 *, sizeof *sin6); if (sin6 == NULL) return (ENOMEM); sin6->sin6_family = AF_INET6; sin6->sin6_len = sizeof(*sin6); inp = (struct sctp_inpcb *)so->so_pcb; if ((inp == NULL) || ((inp->sctp_flags & SCTP_PCB_FLAGS_CONNECTED) == 0)) { /* UDP type and listeners will drop out here */ SCTP_FREE_SONAME(sin6); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, ENOTCONN); return (ENOTCONN); } SCTP_INP_RLOCK(inp); stcb = LIST_FIRST(&inp->sctp_asoc_list); if (stcb) { SCTP_TCB_LOCK(stcb); } SCTP_INP_RUNLOCK(inp); if (stcb == NULL) { SCTP_FREE_SONAME(sin6); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, ECONNRESET); return (ECONNRESET); } fnd = 0; TAILQ_FOREACH(net, &stcb->asoc.nets, sctp_next) { sin_a6 = (struct sockaddr_in6 *)&net->ro._l_addr; if (sin_a6->sin6_family == AF_INET6) { fnd = 1; sin6->sin6_port = stcb->rport; sin6->sin6_addr = sin_a6->sin6_addr; break; } } SCTP_TCB_UNLOCK(stcb); if (!fnd) { /* No IPv4 address */ SCTP_FREE_SONAME(sin6); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, ENOENT); return (ENOENT); } if ((error = sa6_recoverscope(sin6)) != 0) { SCTP_FREE_SONAME(sin6); SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, error); return (error); } *addr = (struct sockaddr *)sin6; return (0); } static int sctp6_in6getaddr(struct socket *so, struct sockaddr **nam) { struct in6pcb *inp6 = sotoin6pcb(so); int error; if (inp6 == NULL) { SCTP_LTRACE_ERR_RET(NULL, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } /* allow v6 addresses precedence */ error = sctp6_getaddr(so, nam); #ifdef INET if (error) { struct sockaddr_in6 *sin6; /* try v4 next if v6 failed */ error = sctp_ingetaddr(so, nam); if (error) { return (error); } SCTP_MALLOC_SONAME(sin6, struct sockaddr_in6 *, sizeof *sin6); if (sin6 == NULL) { SCTP_FREE_SONAME(*nam); return (ENOMEM); } in6_sin_2_v4mapsin6((struct sockaddr_in *)*nam, sin6); SCTP_FREE_SONAME(*nam); *nam = (struct sockaddr *)sin6; } #endif return (error); } static int sctp6_getpeeraddr(struct socket *so, struct sockaddr **nam) { struct in6pcb *inp6 = sotoin6pcb(so); int error; if (inp6 == NULL) { SCTP_LTRACE_ERR_RET(NULL, NULL, NULL, SCTP_FROM_SCTP6_USRREQ, EINVAL); return (EINVAL); } /* allow v6 addresses precedence */ error = sctp6_peeraddr(so, nam); #ifdef INET if (error) { struct sockaddr_in6 *sin6; /* try v4 next if v6 failed */ error = sctp_peeraddr(so, nam); if (error) { return (error); } SCTP_MALLOC_SONAME(sin6, struct sockaddr_in6 *, sizeof *sin6); if (sin6 == NULL) { SCTP_FREE_SONAME(*nam); return (ENOMEM); } in6_sin_2_v4mapsin6((struct sockaddr_in *)*nam, sin6); SCTP_FREE_SONAME(*nam); *nam = (struct sockaddr *)sin6; } #endif return (error); } struct pr_usrreqs sctp6_usrreqs = { .pru_abort = sctp6_abort, .pru_accept = sctp_accept, .pru_attach = sctp6_attach, .pru_bind = sctp6_bind, .pru_connect = sctp6_connect, .pru_control = in6_control, .pru_close = sctp6_close, .pru_detach = sctp6_close, .pru_sopoll = sopoll_generic, .pru_flush = sctp_flush, .pru_disconnect = sctp6_disconnect, .pru_listen = sctp_listen, .pru_peeraddr = sctp6_getpeeraddr, .pru_send = sctp6_send, .pru_shutdown = sctp_shutdown, .pru_sockaddr = sctp6_in6getaddr, .pru_sosend = sctp_sosend, .pru_soreceive = sctp_soreceive }; #endif Index: projects/vnet/sys/netpfil/pf/pf.c =================================================================== --- projects/vnet/sys/netpfil/pf/pf.c (revision 302157) +++ projects/vnet/sys/netpfil/pf/pf.c (revision 302158) @@ -1,6644 +1,6645 @@ /*- * Copyright (c) 2001 Daniel Hartmeier * Copyright (c) 2002 - 2008 Henning Brauer * Copyright (c) 2012 Gleb Smirnoff * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * * - Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * - 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 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 HOLDERS 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. * * Effort sponsored in part by the Defense Advanced Research Projects * Agency (DARPA) and Air Force Research Laboratory, Air Force * Materiel Command, USAF, under agreement number F30602-01-2-0537. * * $OpenBSD: pf.c,v 1.634 2009/02/27 12:37:45 henning Exp $ */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include "opt_bpf.h" #include "opt_pf.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 /* XXX: only for DIR_IN/DIR_OUT */ #ifdef INET6 #include #include #include #include #include #include #include #endif /* INET6 */ #include #include #define DPFPRINTF(n, x) if (V_pf_status.debug >= (n)) printf x /* * Global variables */ /* state tables */ VNET_DEFINE(struct pf_altqqueue, pf_altqs[2]); VNET_DEFINE(struct pf_palist, pf_pabuf); VNET_DEFINE(struct pf_altqqueue *, pf_altqs_active); VNET_DEFINE(struct pf_altqqueue *, pf_altqs_inactive); VNET_DEFINE(struct pf_kstatus, pf_status); VNET_DEFINE(u_int32_t, ticket_altqs_active); VNET_DEFINE(u_int32_t, ticket_altqs_inactive); VNET_DEFINE(int, altqs_inactive_open); VNET_DEFINE(u_int32_t, ticket_pabuf); VNET_DEFINE(MD5_CTX, pf_tcp_secret_ctx); #define V_pf_tcp_secret_ctx VNET(pf_tcp_secret_ctx) VNET_DEFINE(u_char, pf_tcp_secret[16]); #define V_pf_tcp_secret VNET(pf_tcp_secret) VNET_DEFINE(int, pf_tcp_secret_init); #define V_pf_tcp_secret_init VNET(pf_tcp_secret_init) VNET_DEFINE(int, pf_tcp_iss_off); #define V_pf_tcp_iss_off VNET(pf_tcp_iss_off) /* * Queue for pf_intr() sends. */ static MALLOC_DEFINE(M_PFTEMP, "pf_temp", "pf(4) temporary allocations"); struct pf_send_entry { STAILQ_ENTRY(pf_send_entry) pfse_next; struct mbuf *pfse_m; enum { PFSE_IP, PFSE_IP6, PFSE_ICMP, PFSE_ICMP6, } pfse_type; struct { int type; int code; int mtu; } icmpopts; }; STAILQ_HEAD(pf_send_head, pf_send_entry); static VNET_DEFINE(struct pf_send_head, pf_sendqueue); #define V_pf_sendqueue VNET(pf_sendqueue) static struct mtx pf_sendqueue_mtx; MTX_SYSINIT(pf_sendqueue_mtx, &pf_sendqueue_mtx, "pf send queue", MTX_DEF); #define PF_SENDQ_LOCK() mtx_lock(&pf_sendqueue_mtx) #define PF_SENDQ_UNLOCK() mtx_unlock(&pf_sendqueue_mtx) /* * Queue for pf_overload_task() tasks. */ struct pf_overload_entry { SLIST_ENTRY(pf_overload_entry) next; struct pf_addr addr; sa_family_t af; uint8_t dir; struct pf_rule *rule; }; SLIST_HEAD(pf_overload_head, pf_overload_entry); static VNET_DEFINE(struct pf_overload_head, pf_overloadqueue); #define V_pf_overloadqueue VNET(pf_overloadqueue) static VNET_DEFINE(struct task, pf_overloadtask); #define V_pf_overloadtask VNET(pf_overloadtask) static struct mtx pf_overloadqueue_mtx; MTX_SYSINIT(pf_overloadqueue_mtx, &pf_overloadqueue_mtx, "pf overload/flush queue", MTX_DEF); #define PF_OVERLOADQ_LOCK() mtx_lock(&pf_overloadqueue_mtx) #define PF_OVERLOADQ_UNLOCK() mtx_unlock(&pf_overloadqueue_mtx) VNET_DEFINE(struct pf_rulequeue, pf_unlinked_rules); struct mtx pf_unlnkdrules_mtx; MTX_SYSINIT(pf_unlnkdrules_mtx, &pf_unlnkdrules_mtx, "pf unlinked rules", MTX_DEF); static VNET_DEFINE(uma_zone_t, pf_sources_z); #define V_pf_sources_z VNET(pf_sources_z) uma_zone_t pf_mtag_z; VNET_DEFINE(uma_zone_t, pf_state_z); VNET_DEFINE(uma_zone_t, pf_state_key_z); VNET_DEFINE(uint64_t, pf_stateid[MAXCPU]); #define PFID_CPUBITS 8 #define PFID_CPUSHIFT (sizeof(uint64_t) * NBBY - PFID_CPUBITS) #define PFID_CPUMASK ((uint64_t)((1 << PFID_CPUBITS) - 1) << PFID_CPUSHIFT) #define PFID_MAXID (~PFID_CPUMASK) CTASSERT((1 << PFID_CPUBITS) >= MAXCPU); static void pf_src_tree_remove_state(struct pf_state *); static void pf_init_threshold(struct pf_threshold *, u_int32_t, u_int32_t); static void pf_add_threshold(struct pf_threshold *); static int pf_check_threshold(struct pf_threshold *); static void pf_change_ap(struct mbuf *, struct pf_addr *, u_int16_t *, u_int16_t *, u_int16_t *, struct pf_addr *, u_int16_t, u_int8_t, sa_family_t); static int pf_modulate_sack(struct mbuf *, int, struct pf_pdesc *, struct tcphdr *, struct pf_state_peer *); static void pf_change_icmp(struct pf_addr *, u_int16_t *, struct pf_addr *, struct pf_addr *, u_int16_t, u_int16_t *, u_int16_t *, u_int16_t *, u_int16_t *, u_int8_t, sa_family_t); static void pf_send_tcp(struct mbuf *, const struct pf_rule *, sa_family_t, const struct pf_addr *, const struct pf_addr *, u_int16_t, u_int16_t, u_int32_t, u_int32_t, u_int8_t, u_int16_t, u_int16_t, u_int8_t, int, u_int16_t, struct ifnet *); static void pf_send_icmp(struct mbuf *, u_int8_t, u_int8_t, sa_family_t, struct pf_rule *); static void pf_detach_state(struct pf_state *); static int pf_state_key_attach(struct pf_state_key *, struct pf_state_key *, struct pf_state *); static void pf_state_key_detach(struct pf_state *, int); static int pf_state_key_ctor(void *, int, void *, int); static u_int32_t pf_tcp_iss(struct pf_pdesc *); static int pf_test_rule(struct pf_rule **, struct pf_state **, int, struct pfi_kif *, struct mbuf *, int, struct pf_pdesc *, struct pf_rule **, struct pf_ruleset **, struct inpcb *); static int pf_create_state(struct pf_rule *, struct pf_rule *, struct pf_rule *, struct pf_pdesc *, struct pf_src_node *, struct pf_state_key *, struct pf_state_key *, struct mbuf *, int, u_int16_t, u_int16_t, int *, struct pfi_kif *, struct pf_state **, int, u_int16_t, u_int16_t, int); static int pf_test_fragment(struct pf_rule **, int, struct pfi_kif *, struct mbuf *, void *, struct pf_pdesc *, struct pf_rule **, struct pf_ruleset **); static int pf_tcp_track_full(struct pf_state_peer *, struct pf_state_peer *, struct pf_state **, struct pfi_kif *, struct mbuf *, int, struct pf_pdesc *, u_short *, int *); static int pf_tcp_track_sloppy(struct pf_state_peer *, struct pf_state_peer *, struct pf_state **, struct pf_pdesc *, u_short *); static int pf_test_state_tcp(struct pf_state **, int, struct pfi_kif *, struct mbuf *, int, void *, struct pf_pdesc *, u_short *); static int pf_test_state_udp(struct pf_state **, int, struct pfi_kif *, struct mbuf *, int, void *, struct pf_pdesc *); static int pf_test_state_icmp(struct pf_state **, int, struct pfi_kif *, struct mbuf *, int, void *, struct pf_pdesc *, u_short *); static int pf_test_state_other(struct pf_state **, int, struct pfi_kif *, struct mbuf *, struct pf_pdesc *); static u_int8_t pf_get_wscale(struct mbuf *, int, u_int16_t, sa_family_t); static u_int16_t pf_get_mss(struct mbuf *, int, u_int16_t, sa_family_t); static u_int16_t pf_calc_mss(struct pf_addr *, sa_family_t, int, u_int16_t); static int pf_check_proto_cksum(struct mbuf *, int, int, u_int8_t, sa_family_t); static void pf_print_state_parts(struct pf_state *, struct pf_state_key *, struct pf_state_key *); static int pf_addr_wrap_neq(struct pf_addr_wrap *, struct pf_addr_wrap *); static struct pf_state *pf_find_state(struct pfi_kif *, struct pf_state_key_cmp *, u_int); static int pf_src_connlimit(struct pf_state **); static void pf_overload_task(void *v, int pending); static int pf_insert_src_node(struct pf_src_node **, struct pf_rule *, struct pf_addr *, sa_family_t); static u_int pf_purge_expired_states(u_int, int); static void pf_purge_unlinked_rules(void); static int pf_mtag_uminit(void *, int, int); static void pf_mtag_free(struct m_tag *); #ifdef INET static void pf_route(struct mbuf **, struct pf_rule *, int, struct ifnet *, struct pf_state *, struct pf_pdesc *); #endif /* INET */ #ifdef INET6 static void pf_change_a6(struct pf_addr *, u_int16_t *, struct pf_addr *, u_int8_t); static void pf_route6(struct mbuf **, struct pf_rule *, int, struct ifnet *, struct pf_state *, struct pf_pdesc *); #endif /* INET6 */ int in4_cksum(struct mbuf *m, u_int8_t nxt, int off, int len); extern int pf_end_threads; VNET_DEFINE(struct pf_limit, pf_limits[PF_LIMIT_MAX]); #define PACKET_LOOPED(pd) ((pd)->pf_mtag && \ (pd)->pf_mtag->flags & PF_PACKET_LOOPED) #define STATE_LOOKUP(i, k, d, s, pd) \ do { \ (s) = pf_find_state((i), (k), (d)); \ if ((s) == NULL) \ return (PF_DROP); \ if (PACKET_LOOPED(pd)) \ return (PF_PASS); \ if ((d) == PF_OUT && \ (((s)->rule.ptr->rt == PF_ROUTETO && \ (s)->rule.ptr->direction == PF_OUT) || \ ((s)->rule.ptr->rt == PF_REPLYTO && \ (s)->rule.ptr->direction == PF_IN)) && \ (s)->rt_kif != NULL && \ (s)->rt_kif != (i)) \ return (PF_PASS); \ } while (0) #define BOUND_IFACE(r, k) \ ((r)->rule_flag & PFRULE_IFBOUND) ? (k) : V_pfi_all #define STATE_INC_COUNTERS(s) \ do { \ counter_u64_add(s->rule.ptr->states_cur, 1); \ counter_u64_add(s->rule.ptr->states_tot, 1); \ if (s->anchor.ptr != NULL) { \ counter_u64_add(s->anchor.ptr->states_cur, 1); \ counter_u64_add(s->anchor.ptr->states_tot, 1); \ } \ if (s->nat_rule.ptr != NULL) { \ counter_u64_add(s->nat_rule.ptr->states_cur, 1);\ counter_u64_add(s->nat_rule.ptr->states_tot, 1);\ } \ } while (0) #define STATE_DEC_COUNTERS(s) \ do { \ if (s->nat_rule.ptr != NULL) \ counter_u64_add(s->nat_rule.ptr->states_cur, -1);\ if (s->anchor.ptr != NULL) \ counter_u64_add(s->anchor.ptr->states_cur, -1); \ counter_u64_add(s->rule.ptr->states_cur, -1); \ } while (0) static MALLOC_DEFINE(M_PFHASH, "pf_hash", "pf(4) hash header structures"); VNET_DEFINE(struct pf_keyhash *, pf_keyhash); VNET_DEFINE(struct pf_idhash *, pf_idhash); VNET_DEFINE(struct pf_srchash *, pf_srchash); SYSCTL_NODE(_net, OID_AUTO, pf, CTLFLAG_RW, 0, "pf(4)"); u_long pf_hashmask; u_long pf_srchashmask; static u_long pf_hashsize; static u_long pf_srchashsize; SYSCTL_ULONG(_net_pf, OID_AUTO, states_hashsize, CTLFLAG_RDTUN, &pf_hashsize, 0, "Size of pf(4) states hashtable"); SYSCTL_ULONG(_net_pf, OID_AUTO, source_nodes_hashsize, CTLFLAG_RDTUN, &pf_srchashsize, 0, "Size of pf(4) source nodes hashtable"); VNET_DEFINE(void *, pf_swi_cookie); VNET_DEFINE(uint32_t, pf_hashseed); #define V_pf_hashseed VNET(pf_hashseed) int pf_addr_cmp(struct pf_addr *a, struct pf_addr *b, sa_family_t af) { switch (af) { #ifdef INET case AF_INET: if (a->addr32[0] > b->addr32[0]) return (1); if (a->addr32[0] < b->addr32[0]) return (-1); break; #endif /* INET */ #ifdef INET6 case AF_INET6: if (a->addr32[3] > b->addr32[3]) return (1); if (a->addr32[3] < b->addr32[3]) return (-1); if (a->addr32[2] > b->addr32[2]) return (1); if (a->addr32[2] < b->addr32[2]) return (-1); if (a->addr32[1] > b->addr32[1]) return (1); if (a->addr32[1] < b->addr32[1]) return (-1); if (a->addr32[0] > b->addr32[0]) return (1); if (a->addr32[0] < b->addr32[0]) return (-1); break; #endif /* INET6 */ default: panic("%s: unknown address family %u", __func__, af); } return (0); } static __inline uint32_t pf_hashkey(struct pf_state_key *sk) { uint32_t h; h = murmur3_32_hash32((uint32_t *)sk, sizeof(struct pf_state_key_cmp)/sizeof(uint32_t), V_pf_hashseed); return (h & pf_hashmask); } static __inline uint32_t pf_hashsrc(struct pf_addr *addr, sa_family_t af) { uint32_t h; switch (af) { case AF_INET: h = murmur3_32_hash32((uint32_t *)&addr->v4, sizeof(addr->v4)/sizeof(uint32_t), V_pf_hashseed); break; case AF_INET6: h = murmur3_32_hash32((uint32_t *)&addr->v6, sizeof(addr->v6)/sizeof(uint32_t), V_pf_hashseed); break; default: panic("%s: unknown address family %u", __func__, af); } return (h & pf_srchashmask); } #ifdef ALTQ static int pf_state_hash(struct pf_state *s) { u_int32_t hv = (intptr_t)s / sizeof(*s); hv ^= crc32(&s->src, sizeof(s->src)); hv ^= crc32(&s->dst, sizeof(s->dst)); if (hv == 0) hv = 1; return (hv); } #endif #ifdef INET6 void pf_addrcpy(struct pf_addr *dst, struct pf_addr *src, sa_family_t af) { switch (af) { #ifdef INET case AF_INET: dst->addr32[0] = src->addr32[0]; break; #endif /* INET */ case AF_INET6: dst->addr32[0] = src->addr32[0]; dst->addr32[1] = src->addr32[1]; dst->addr32[2] = src->addr32[2]; dst->addr32[3] = src->addr32[3]; break; } } #endif /* INET6 */ static void pf_init_threshold(struct pf_threshold *threshold, u_int32_t limit, u_int32_t seconds) { threshold->limit = limit * PF_THRESHOLD_MULT; threshold->seconds = seconds; threshold->count = 0; threshold->last = time_uptime; } static void pf_add_threshold(struct pf_threshold *threshold) { u_int32_t t = time_uptime, diff = t - threshold->last; if (diff >= threshold->seconds) threshold->count = 0; else threshold->count -= threshold->count * diff / threshold->seconds; threshold->count += PF_THRESHOLD_MULT; threshold->last = t; } static int pf_check_threshold(struct pf_threshold *threshold) { return (threshold->count > threshold->limit); } static int pf_src_connlimit(struct pf_state **state) { struct pf_overload_entry *pfoe; int bad = 0; PF_STATE_LOCK_ASSERT(*state); (*state)->src_node->conn++; (*state)->src.tcp_est = 1; pf_add_threshold(&(*state)->src_node->conn_rate); if ((*state)->rule.ptr->max_src_conn && (*state)->rule.ptr->max_src_conn < (*state)->src_node->conn) { counter_u64_add(V_pf_status.lcounters[LCNT_SRCCONN], 1); bad++; } if ((*state)->rule.ptr->max_src_conn_rate.limit && pf_check_threshold(&(*state)->src_node->conn_rate)) { counter_u64_add(V_pf_status.lcounters[LCNT_SRCCONNRATE], 1); bad++; } if (!bad) return (0); /* Kill this state. */ (*state)->timeout = PFTM_PURGE; (*state)->src.state = (*state)->dst.state = TCPS_CLOSED; if ((*state)->rule.ptr->overload_tbl == NULL) return (1); /* Schedule overloading and flushing task. */ pfoe = malloc(sizeof(*pfoe), M_PFTEMP, M_NOWAIT); if (pfoe == NULL) return (1); /* too bad :( */ bcopy(&(*state)->src_node->addr, &pfoe->addr, sizeof(pfoe->addr)); pfoe->af = (*state)->key[PF_SK_WIRE]->af; pfoe->rule = (*state)->rule.ptr; pfoe->dir = (*state)->direction; PF_OVERLOADQ_LOCK(); SLIST_INSERT_HEAD(&V_pf_overloadqueue, pfoe, next); PF_OVERLOADQ_UNLOCK(); taskqueue_enqueue(taskqueue_swi, &V_pf_overloadtask); return (1); } static void pf_overload_task(void *v, int pending) { struct pf_overload_head queue; struct pfr_addr p; struct pf_overload_entry *pfoe, *pfoe1; uint32_t killed = 0; CURVNET_SET((struct vnet *)v); PF_OVERLOADQ_LOCK(); queue = V_pf_overloadqueue; SLIST_INIT(&V_pf_overloadqueue); PF_OVERLOADQ_UNLOCK(); bzero(&p, sizeof(p)); SLIST_FOREACH(pfoe, &queue, next) { counter_u64_add(V_pf_status.lcounters[LCNT_OVERLOAD_TABLE], 1); if (V_pf_status.debug >= PF_DEBUG_MISC) { printf("%s: blocking address ", __func__); pf_print_host(&pfoe->addr, 0, pfoe->af); printf("\n"); } p.pfra_af = pfoe->af; switch (pfoe->af) { #ifdef INET case AF_INET: p.pfra_net = 32; p.pfra_ip4addr = pfoe->addr.v4; break; #endif #ifdef INET6 case AF_INET6: p.pfra_net = 128; p.pfra_ip6addr = pfoe->addr.v6; break; #endif } PF_RULES_WLOCK(); pfr_insert_kentry(pfoe->rule->overload_tbl, &p, time_second); PF_RULES_WUNLOCK(); } /* * Remove those entries, that don't need flushing. */ SLIST_FOREACH_SAFE(pfoe, &queue, next, pfoe1) if (pfoe->rule->flush == 0) { SLIST_REMOVE(&queue, pfoe, pf_overload_entry, next); free(pfoe, M_PFTEMP); } else counter_u64_add( V_pf_status.lcounters[LCNT_OVERLOAD_FLUSH], 1); /* If nothing to flush, return. */ if (SLIST_EMPTY(&queue)) { CURVNET_RESTORE(); return; } for (int i = 0; i <= pf_hashmask; i++) { struct pf_idhash *ih = &V_pf_idhash[i]; struct pf_state_key *sk; struct pf_state *s; PF_HASHROW_LOCK(ih); LIST_FOREACH(s, &ih->states, entry) { sk = s->key[PF_SK_WIRE]; SLIST_FOREACH(pfoe, &queue, next) if (sk->af == pfoe->af && ((pfoe->rule->flush & PF_FLUSH_GLOBAL) || pfoe->rule == s->rule.ptr) && ((pfoe->dir == PF_OUT && PF_AEQ(&pfoe->addr, &sk->addr[1], sk->af)) || (pfoe->dir == PF_IN && PF_AEQ(&pfoe->addr, &sk->addr[0], sk->af)))) { s->timeout = PFTM_PURGE; s->src.state = s->dst.state = TCPS_CLOSED; killed++; } } PF_HASHROW_UNLOCK(ih); } SLIST_FOREACH_SAFE(pfoe, &queue, next, pfoe1) free(pfoe, M_PFTEMP); if (V_pf_status.debug >= PF_DEBUG_MISC) printf("%s: %u states killed", __func__, killed); CURVNET_RESTORE(); } /* * Can return locked on failure, so that we can consistently * allocate and insert a new one. */ struct pf_src_node * pf_find_src_node(struct pf_addr *src, struct pf_rule *rule, sa_family_t af, int returnlocked) { struct pf_srchash *sh; struct pf_src_node *n; counter_u64_add(V_pf_status.scounters[SCNT_SRC_NODE_SEARCH], 1); sh = &V_pf_srchash[pf_hashsrc(src, af)]; PF_HASHROW_LOCK(sh); LIST_FOREACH(n, &sh->nodes, entry) if (n->rule.ptr == rule && n->af == af && ((af == AF_INET && n->addr.v4.s_addr == src->v4.s_addr) || (af == AF_INET6 && bcmp(&n->addr, src, sizeof(*src)) == 0))) break; if (n != NULL) { n->states++; PF_HASHROW_UNLOCK(sh); } else if (returnlocked == 0) PF_HASHROW_UNLOCK(sh); return (n); } static int pf_insert_src_node(struct pf_src_node **sn, struct pf_rule *rule, struct pf_addr *src, sa_family_t af) { KASSERT((rule->rule_flag & PFRULE_RULESRCTRACK || rule->rpool.opts & PF_POOL_STICKYADDR), ("%s for non-tracking rule %p", __func__, rule)); if (*sn == NULL) *sn = pf_find_src_node(src, rule, af, 1); if (*sn == NULL) { struct pf_srchash *sh = &V_pf_srchash[pf_hashsrc(src, af)]; PF_HASHROW_ASSERT(sh); if (!rule->max_src_nodes || counter_u64_fetch(rule->src_nodes) < rule->max_src_nodes) (*sn) = uma_zalloc(V_pf_sources_z, M_NOWAIT | M_ZERO); else counter_u64_add(V_pf_status.lcounters[LCNT_SRCNODES], 1); if ((*sn) == NULL) { PF_HASHROW_UNLOCK(sh); return (-1); } pf_init_threshold(&(*sn)->conn_rate, rule->max_src_conn_rate.limit, rule->max_src_conn_rate.seconds); (*sn)->af = af; (*sn)->rule.ptr = rule; PF_ACPY(&(*sn)->addr, src, af); LIST_INSERT_HEAD(&sh->nodes, *sn, entry); (*sn)->creation = time_uptime; (*sn)->ruletype = rule->action; (*sn)->states = 1; if ((*sn)->rule.ptr != NULL) counter_u64_add((*sn)->rule.ptr->src_nodes, 1); PF_HASHROW_UNLOCK(sh); counter_u64_add(V_pf_status.scounters[SCNT_SRC_NODE_INSERT], 1); } else { if (rule->max_src_states && (*sn)->states >= rule->max_src_states) { counter_u64_add(V_pf_status.lcounters[LCNT_SRCSTATES], 1); return (-1); } } return (0); } void pf_unlink_src_node(struct pf_src_node *src) { PF_HASHROW_ASSERT(&V_pf_srchash[pf_hashsrc(&src->addr, src->af)]); LIST_REMOVE(src, entry); if (src->rule.ptr) counter_u64_add(src->rule.ptr->src_nodes, -1); } u_int pf_free_src_nodes(struct pf_src_node_list *head) { struct pf_src_node *sn, *tmp; u_int count = 0; LIST_FOREACH_SAFE(sn, head, entry, tmp) { uma_zfree(V_pf_sources_z, sn); count++; } counter_u64_add(V_pf_status.scounters[SCNT_SRC_NODE_REMOVALS], count); return (count); } void pf_mtag_initialize() { pf_mtag_z = uma_zcreate("pf mtags", sizeof(struct m_tag) + sizeof(struct pf_mtag), NULL, NULL, pf_mtag_uminit, NULL, UMA_ALIGN_PTR, 0); } /* Per-vnet data storage structures initialization. */ void pf_initialize() { struct pf_keyhash *kh; struct pf_idhash *ih; struct pf_srchash *sh; u_int i; if (pf_hashsize == 0 || !powerof2(pf_hashsize)) pf_hashsize = PF_HASHSIZ; if (pf_srchashsize == 0 || !powerof2(pf_srchashsize)) pf_srchashsize = PF_HASHSIZ / 4; V_pf_hashseed = arc4random(); /* States and state keys storage. */ V_pf_state_z = uma_zcreate("pf states", sizeof(struct pf_state), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0); V_pf_limits[PF_LIMIT_STATES].zone = V_pf_state_z; uma_zone_set_max(V_pf_state_z, PFSTATE_HIWAT); uma_zone_set_warning(V_pf_state_z, "PF states limit reached"); V_pf_state_key_z = uma_zcreate("pf state keys", sizeof(struct pf_state_key), pf_state_key_ctor, NULL, NULL, NULL, UMA_ALIGN_PTR, 0); V_pf_keyhash = malloc(pf_hashsize * sizeof(struct pf_keyhash), M_PFHASH, M_WAITOK | M_ZERO); V_pf_idhash = malloc(pf_hashsize * sizeof(struct pf_idhash), M_PFHASH, M_WAITOK | M_ZERO); pf_hashmask = pf_hashsize - 1; for (i = 0, kh = V_pf_keyhash, ih = V_pf_idhash; i <= pf_hashmask; i++, kh++, ih++) { mtx_init(&kh->lock, "pf_keyhash", NULL, MTX_DEF | MTX_DUPOK); mtx_init(&ih->lock, "pf_idhash", NULL, MTX_DEF); } /* Source nodes. */ V_pf_sources_z = uma_zcreate("pf source nodes", sizeof(struct pf_src_node), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0); V_pf_limits[PF_LIMIT_SRC_NODES].zone = V_pf_sources_z; uma_zone_set_max(V_pf_sources_z, PFSNODE_HIWAT); uma_zone_set_warning(V_pf_sources_z, "PF source nodes limit reached"); V_pf_srchash = malloc(pf_srchashsize * sizeof(struct pf_srchash), M_PFHASH, M_WAITOK|M_ZERO); pf_srchashmask = pf_srchashsize - 1; for (i = 0, sh = V_pf_srchash; i <= pf_srchashmask; i++, sh++) mtx_init(&sh->lock, "pf_srchash", NULL, MTX_DEF); /* ALTQ */ TAILQ_INIT(&V_pf_altqs[0]); TAILQ_INIT(&V_pf_altqs[1]); TAILQ_INIT(&V_pf_pabuf); V_pf_altqs_active = &V_pf_altqs[0]; V_pf_altqs_inactive = &V_pf_altqs[1]; /* Send & overload+flush queues. */ STAILQ_INIT(&V_pf_sendqueue); SLIST_INIT(&V_pf_overloadqueue); TASK_INIT(&V_pf_overloadtask, 0, pf_overload_task, curvnet); /* Unlinked, but may be referenced rules. */ TAILQ_INIT(&V_pf_unlinked_rules); } void pf_mtag_cleanup() { uma_zdestroy(pf_mtag_z); } void pf_cleanup() { struct pf_keyhash *kh; struct pf_idhash *ih; struct pf_srchash *sh; struct pf_send_entry *pfse, *next; u_int i; for (i = 0, kh = V_pf_keyhash, ih = V_pf_idhash; i <= pf_hashmask; i++, kh++, ih++) { KASSERT(LIST_EMPTY(&kh->keys), ("%s: key hash not empty", __func__)); KASSERT(LIST_EMPTY(&ih->states), ("%s: id hash not empty", __func__)); mtx_destroy(&kh->lock); mtx_destroy(&ih->lock); } free(V_pf_keyhash, M_PFHASH); free(V_pf_idhash, M_PFHASH); for (i = 0, sh = V_pf_srchash; i <= pf_srchashmask; i++, sh++) { KASSERT(LIST_EMPTY(&sh->nodes), ("%s: source node hash not empty", __func__)); mtx_destroy(&sh->lock); } free(V_pf_srchash, M_PFHASH); STAILQ_FOREACH_SAFE(pfse, &V_pf_sendqueue, pfse_next, next) { m_freem(pfse->pfse_m); free(pfse, M_PFTEMP); } uma_zdestroy(V_pf_sources_z); uma_zdestroy(V_pf_state_z); uma_zdestroy(V_pf_state_key_z); } static int pf_mtag_uminit(void *mem, int size, int how) { struct m_tag *t; t = (struct m_tag *)mem; t->m_tag_cookie = MTAG_ABI_COMPAT; t->m_tag_id = PACKET_TAG_PF; t->m_tag_len = sizeof(struct pf_mtag); t->m_tag_free = pf_mtag_free; return (0); } static void pf_mtag_free(struct m_tag *t) { uma_zfree(pf_mtag_z, t); } struct pf_mtag * pf_get_mtag(struct mbuf *m) { struct m_tag *mtag; if ((mtag = m_tag_find(m, PACKET_TAG_PF, NULL)) != NULL) return ((struct pf_mtag *)(mtag + 1)); mtag = uma_zalloc(pf_mtag_z, M_NOWAIT); if (mtag == NULL) return (NULL); bzero(mtag + 1, sizeof(struct pf_mtag)); m_tag_prepend(m, mtag); return ((struct pf_mtag *)(mtag + 1)); } static int pf_state_key_attach(struct pf_state_key *skw, struct pf_state_key *sks, struct pf_state *s) { struct pf_keyhash *khs, *khw, *kh; struct pf_state_key *sk, *cur; struct pf_state *si, *olds = NULL; int idx; KASSERT(s->refs == 0, ("%s: state not pristine", __func__)); KASSERT(s->key[PF_SK_WIRE] == NULL, ("%s: state has key", __func__)); KASSERT(s->key[PF_SK_STACK] == NULL, ("%s: state has key", __func__)); /* * We need to lock hash slots of both keys. To avoid deadlock * we always lock the slot with lower address first. Unlock order * isn't important. * * We also need to lock ID hash slot before dropping key * locks. On success we return with ID hash slot locked. */ if (skw == sks) { khs = khw = &V_pf_keyhash[pf_hashkey(skw)]; PF_HASHROW_LOCK(khs); } else { khs = &V_pf_keyhash[pf_hashkey(sks)]; khw = &V_pf_keyhash[pf_hashkey(skw)]; if (khs == khw) { PF_HASHROW_LOCK(khs); } else if (khs < khw) { PF_HASHROW_LOCK(khs); PF_HASHROW_LOCK(khw); } else { PF_HASHROW_LOCK(khw); PF_HASHROW_LOCK(khs); } } #define KEYS_UNLOCK() do { \ if (khs != khw) { \ PF_HASHROW_UNLOCK(khs); \ PF_HASHROW_UNLOCK(khw); \ } else \ PF_HASHROW_UNLOCK(khs); \ } while (0) /* * First run: start with wire key. */ sk = skw; kh = khw; idx = PF_SK_WIRE; keyattach: LIST_FOREACH(cur, &kh->keys, entry) if (bcmp(cur, sk, sizeof(struct pf_state_key_cmp)) == 0) break; if (cur != NULL) { /* Key exists. Check for same kif, if none, add to key. */ TAILQ_FOREACH(si, &cur->states[idx], key_list[idx]) { struct pf_idhash *ih = &V_pf_idhash[PF_IDHASH(si)]; PF_HASHROW_LOCK(ih); if (si->kif == s->kif && si->direction == s->direction) { if (sk->proto == IPPROTO_TCP && si->src.state >= TCPS_FIN_WAIT_2 && si->dst.state >= TCPS_FIN_WAIT_2) { /* * New state matches an old >FIN_WAIT_2 * state. We can't drop key hash locks, * thus we can't unlink it properly. * * As a workaround we drop it into * TCPS_CLOSED state, schedule purge * ASAP and push it into the very end * of the slot TAILQ, so that it won't * conflict with our new state. */ si->src.state = si->dst.state = TCPS_CLOSED; si->timeout = PFTM_PURGE; olds = si; } else { if (V_pf_status.debug >= PF_DEBUG_MISC) { printf("pf: %s key attach " "failed on %s: ", (idx == PF_SK_WIRE) ? "wire" : "stack", s->kif->pfik_name); pf_print_state_parts(s, (idx == PF_SK_WIRE) ? sk : NULL, (idx == PF_SK_STACK) ? sk : NULL); printf(", existing: "); pf_print_state_parts(si, (idx == PF_SK_WIRE) ? sk : NULL, (idx == PF_SK_STACK) ? sk : NULL); printf("\n"); } PF_HASHROW_UNLOCK(ih); KEYS_UNLOCK(); uma_zfree(V_pf_state_key_z, sk); if (idx == PF_SK_STACK) pf_detach_state(s); return (EEXIST); /* collision! */ } } PF_HASHROW_UNLOCK(ih); } uma_zfree(V_pf_state_key_z, sk); s->key[idx] = cur; } else { LIST_INSERT_HEAD(&kh->keys, sk, entry); s->key[idx] = sk; } stateattach: /* List is sorted, if-bound states before floating. */ if (s->kif == V_pfi_all) TAILQ_INSERT_TAIL(&s->key[idx]->states[idx], s, key_list[idx]); else TAILQ_INSERT_HEAD(&s->key[idx]->states[idx], s, key_list[idx]); if (olds) { TAILQ_REMOVE(&s->key[idx]->states[idx], olds, key_list[idx]); TAILQ_INSERT_TAIL(&s->key[idx]->states[idx], olds, key_list[idx]); olds = NULL; } /* * Attach done. See how should we (or should not?) * attach a second key. */ if (sks == skw) { s->key[PF_SK_STACK] = s->key[PF_SK_WIRE]; idx = PF_SK_STACK; sks = NULL; goto stateattach; } else if (sks != NULL) { /* * Continue attaching with stack key. */ sk = sks; kh = khs; idx = PF_SK_STACK; sks = NULL; goto keyattach; } PF_STATE_LOCK(s); KEYS_UNLOCK(); KASSERT(s->key[PF_SK_WIRE] != NULL && s->key[PF_SK_STACK] != NULL, ("%s failure", __func__)); return (0); #undef KEYS_UNLOCK } static void pf_detach_state(struct pf_state *s) { struct pf_state_key *sks = s->key[PF_SK_STACK]; struct pf_keyhash *kh; if (sks != NULL) { kh = &V_pf_keyhash[pf_hashkey(sks)]; PF_HASHROW_LOCK(kh); if (s->key[PF_SK_STACK] != NULL) pf_state_key_detach(s, PF_SK_STACK); /* * If both point to same key, then we are done. */ if (sks == s->key[PF_SK_WIRE]) { pf_state_key_detach(s, PF_SK_WIRE); PF_HASHROW_UNLOCK(kh); return; } PF_HASHROW_UNLOCK(kh); } if (s->key[PF_SK_WIRE] != NULL) { kh = &V_pf_keyhash[pf_hashkey(s->key[PF_SK_WIRE])]; PF_HASHROW_LOCK(kh); if (s->key[PF_SK_WIRE] != NULL) pf_state_key_detach(s, PF_SK_WIRE); PF_HASHROW_UNLOCK(kh); } } static void pf_state_key_detach(struct pf_state *s, int idx) { struct pf_state_key *sk = s->key[idx]; #ifdef INVARIANTS struct pf_keyhash *kh = &V_pf_keyhash[pf_hashkey(sk)]; PF_HASHROW_ASSERT(kh); #endif TAILQ_REMOVE(&sk->states[idx], s, key_list[idx]); s->key[idx] = NULL; if (TAILQ_EMPTY(&sk->states[0]) && TAILQ_EMPTY(&sk->states[1])) { LIST_REMOVE(sk, entry); uma_zfree(V_pf_state_key_z, sk); } } static int pf_state_key_ctor(void *mem, int size, void *arg, int flags) { struct pf_state_key *sk = mem; bzero(sk, sizeof(struct pf_state_key_cmp)); TAILQ_INIT(&sk->states[PF_SK_WIRE]); TAILQ_INIT(&sk->states[PF_SK_STACK]); return (0); } struct pf_state_key * pf_state_key_setup(struct pf_pdesc *pd, struct pf_addr *saddr, struct pf_addr *daddr, u_int16_t sport, u_int16_t dport) { struct pf_state_key *sk; sk = uma_zalloc(V_pf_state_key_z, M_NOWAIT); if (sk == NULL) return (NULL); PF_ACPY(&sk->addr[pd->sidx], saddr, pd->af); PF_ACPY(&sk->addr[pd->didx], daddr, pd->af); sk->port[pd->sidx] = sport; sk->port[pd->didx] = dport; sk->proto = pd->proto; sk->af = pd->af; return (sk); } struct pf_state_key * pf_state_key_clone(struct pf_state_key *orig) { struct pf_state_key *sk; sk = uma_zalloc(V_pf_state_key_z, M_NOWAIT); if (sk == NULL) return (NULL); bcopy(orig, sk, sizeof(struct pf_state_key_cmp)); return (sk); } int pf_state_insert(struct pfi_kif *kif, struct pf_state_key *skw, struct pf_state_key *sks, struct pf_state *s) { struct pf_idhash *ih; struct pf_state *cur; int error; KASSERT(TAILQ_EMPTY(&sks->states[0]) && TAILQ_EMPTY(&sks->states[1]), ("%s: sks not pristine", __func__)); KASSERT(TAILQ_EMPTY(&skw->states[0]) && TAILQ_EMPTY(&skw->states[1]), ("%s: skw not pristine", __func__)); KASSERT(s->refs == 0, ("%s: state not pristine", __func__)); s->kif = kif; if (s->id == 0 && s->creatorid == 0) { /* XXX: should be atomic, but probability of collision low */ if ((s->id = V_pf_stateid[curcpu]++) == PFID_MAXID) V_pf_stateid[curcpu] = 1; s->id |= (uint64_t )curcpu << PFID_CPUSHIFT; s->id = htobe64(s->id); s->creatorid = V_pf_status.hostid; } /* Returns with ID locked on success. */ if ((error = pf_state_key_attach(skw, sks, s)) != 0) return (error); ih = &V_pf_idhash[PF_IDHASH(s)]; PF_HASHROW_ASSERT(ih); LIST_FOREACH(cur, &ih->states, entry) if (cur->id == s->id && cur->creatorid == s->creatorid) break; if (cur != NULL) { PF_HASHROW_UNLOCK(ih); if (V_pf_status.debug >= PF_DEBUG_MISC) { printf("pf: state ID collision: " "id: %016llx creatorid: %08x\n", (unsigned long long)be64toh(s->id), ntohl(s->creatorid)); } pf_detach_state(s); return (EEXIST); } LIST_INSERT_HEAD(&ih->states, s, entry); /* One for keys, one for ID hash. */ refcount_init(&s->refs, 2); counter_u64_add(V_pf_status.fcounters[FCNT_STATE_INSERT], 1); if (pfsync_insert_state_ptr != NULL) pfsync_insert_state_ptr(s); /* Returns locked. */ return (0); } /* * Find state by ID: returns with locked row on success. */ struct pf_state * pf_find_state_byid(uint64_t id, uint32_t creatorid) { struct pf_idhash *ih; struct pf_state *s; counter_u64_add(V_pf_status.fcounters[FCNT_STATE_SEARCH], 1); ih = &V_pf_idhash[(be64toh(id) % (pf_hashmask + 1))]; PF_HASHROW_LOCK(ih); LIST_FOREACH(s, &ih->states, entry) if (s->id == id && s->creatorid == creatorid) break; if (s == NULL) PF_HASHROW_UNLOCK(ih); return (s); } /* * Find state by key. * Returns with ID hash slot locked on success. */ static struct pf_state * pf_find_state(struct pfi_kif *kif, struct pf_state_key_cmp *key, u_int dir) { struct pf_keyhash *kh; struct pf_state_key *sk; struct pf_state *s; int idx; counter_u64_add(V_pf_status.fcounters[FCNT_STATE_SEARCH], 1); kh = &V_pf_keyhash[pf_hashkey((struct pf_state_key *)key)]; PF_HASHROW_LOCK(kh); LIST_FOREACH(sk, &kh->keys, entry) if (bcmp(sk, key, sizeof(struct pf_state_key_cmp)) == 0) break; if (sk == NULL) { PF_HASHROW_UNLOCK(kh); return (NULL); } idx = (dir == PF_IN ? PF_SK_WIRE : PF_SK_STACK); /* List is sorted, if-bound states before floating ones. */ TAILQ_FOREACH(s, &sk->states[idx], key_list[idx]) if (s->kif == V_pfi_all || s->kif == kif) { PF_STATE_LOCK(s); PF_HASHROW_UNLOCK(kh); if (s->timeout >= PFTM_MAX) { /* * State is either being processed by * pf_unlink_state() in an other thread, or * is scheduled for immediate expiry. */ PF_STATE_UNLOCK(s); return (NULL); } return (s); } PF_HASHROW_UNLOCK(kh); return (NULL); } struct pf_state * pf_find_state_all(struct pf_state_key_cmp *key, u_int dir, int *more) { struct pf_keyhash *kh; struct pf_state_key *sk; struct pf_state *s, *ret = NULL; int idx, inout = 0; counter_u64_add(V_pf_status.fcounters[FCNT_STATE_SEARCH], 1); kh = &V_pf_keyhash[pf_hashkey((struct pf_state_key *)key)]; PF_HASHROW_LOCK(kh); LIST_FOREACH(sk, &kh->keys, entry) if (bcmp(sk, key, sizeof(struct pf_state_key_cmp)) == 0) break; if (sk == NULL) { PF_HASHROW_UNLOCK(kh); return (NULL); } switch (dir) { case PF_IN: idx = PF_SK_WIRE; break; case PF_OUT: idx = PF_SK_STACK; break; case PF_INOUT: idx = PF_SK_WIRE; inout = 1; break; default: panic("%s: dir %u", __func__, dir); } second_run: TAILQ_FOREACH(s, &sk->states[idx], key_list[idx]) { if (more == NULL) { PF_HASHROW_UNLOCK(kh); return (s); } if (ret) (*more)++; else ret = s; } if (inout == 1) { inout = 0; idx = PF_SK_STACK; goto second_run; } PF_HASHROW_UNLOCK(kh); return (ret); } /* END state table stuff */ static void pf_send(struct pf_send_entry *pfse) { PF_SENDQ_LOCK(); STAILQ_INSERT_TAIL(&V_pf_sendqueue, pfse, pfse_next); PF_SENDQ_UNLOCK(); swi_sched(V_pf_swi_cookie, 0); } void pf_intr(void *v) { struct pf_send_head queue; struct pf_send_entry *pfse, *next; CURVNET_SET((struct vnet *)v); PF_SENDQ_LOCK(); queue = V_pf_sendqueue; STAILQ_INIT(&V_pf_sendqueue); PF_SENDQ_UNLOCK(); STAILQ_FOREACH_SAFE(pfse, &queue, pfse_next, next) { switch (pfse->pfse_type) { #ifdef INET case PFSE_IP: ip_output(pfse->pfse_m, NULL, NULL, 0, NULL, NULL); break; case PFSE_ICMP: icmp_error(pfse->pfse_m, pfse->icmpopts.type, pfse->icmpopts.code, 0, pfse->icmpopts.mtu); break; #endif /* INET */ #ifdef INET6 case PFSE_IP6: ip6_output(pfse->pfse_m, NULL, NULL, 0, NULL, NULL, NULL); break; case PFSE_ICMP6: icmp6_error(pfse->pfse_m, pfse->icmpopts.type, pfse->icmpopts.code, pfse->icmpopts.mtu); break; #endif /* INET6 */ default: panic("%s: unknown type", __func__); } free(pfse, M_PFTEMP); } CURVNET_RESTORE(); } void pf_purge_thread(void *unused __unused) { VNET_ITERATOR_DECL(vnet_iter); u_int idx = 0; for (;;) { PF_RULES_RLOCK(); rw_sleep(pf_purge_thread, &pf_rules_lock, 0, "pftm", hz / 10); PF_RULES_RUNLOCK(); VNET_LIST_RLOCK(); VNET_FOREACH(vnet_iter) { CURVNET_SET(vnet_iter); if (pf_end_threads) { + pf_end_threads++; wakeup(pf_purge_thread); kproc_exit(0); } /* Process 1/interval fraction of the state table every run. */ idx = pf_purge_expired_states(idx, pf_hashmask / (V_pf_default_rule.timeout[PFTM_INTERVAL] * 10)); /* Purge other expired types every PFTM_INTERVAL seconds. */ if (idx == 0) { /* * Order is important: * - states and src nodes reference rules * - states and rules reference kifs */ pf_purge_expired_fragments(); pf_purge_expired_src_nodes(); pf_purge_unlinked_rules(); pfi_kif_purge(); } CURVNET_RESTORE(); } VNET_LIST_RUNLOCK(); } /* not reached */ } void pf_unload_vnet_purge(void) { /* * To cleanse up all kifs and rules we need * two runs: first one clears reference flags, * then pf_purge_expired_states() doesn't * raise them, and then second run frees. */ pf_purge_unlinked_rules(); pfi_kif_purge(); /* * Now purge everything. */ pf_purge_expired_states(0, pf_hashmask); pf_purge_expired_fragments(); pf_purge_expired_src_nodes(); /* * Now all kifs & rules should be unreferenced, * thus should be successfully freed. */ pf_purge_unlinked_rules(); pfi_kif_purge(); } u_int32_t pf_state_expires(const struct pf_state *state) { u_int32_t timeout; u_int32_t start; u_int32_t end; u_int32_t states; /* handle all PFTM_* > PFTM_MAX here */ if (state->timeout == PFTM_PURGE) return (time_uptime); KASSERT(state->timeout != PFTM_UNLINKED, ("pf_state_expires: timeout == PFTM_UNLINKED")); KASSERT((state->timeout < PFTM_MAX), ("pf_state_expires: timeout > PFTM_MAX")); timeout = state->rule.ptr->timeout[state->timeout]; if (!timeout) timeout = V_pf_default_rule.timeout[state->timeout]; start = state->rule.ptr->timeout[PFTM_ADAPTIVE_START]; if (start) { end = state->rule.ptr->timeout[PFTM_ADAPTIVE_END]; states = counter_u64_fetch(state->rule.ptr->states_cur); } else { start = V_pf_default_rule.timeout[PFTM_ADAPTIVE_START]; end = V_pf_default_rule.timeout[PFTM_ADAPTIVE_END]; states = V_pf_status.states; } if (end && states > start && start < end) { if (states < end) return (state->expire + timeout * (end - states) / (end - start)); else return (time_uptime); } return (state->expire + timeout); } void pf_purge_expired_src_nodes() { struct pf_src_node_list freelist; struct pf_srchash *sh; struct pf_src_node *cur, *next; int i; LIST_INIT(&freelist); for (i = 0, sh = V_pf_srchash; i <= pf_srchashmask; i++, sh++) { PF_HASHROW_LOCK(sh); LIST_FOREACH_SAFE(cur, &sh->nodes, entry, next) if (cur->states == 0 && cur->expire <= time_uptime) { pf_unlink_src_node(cur); LIST_INSERT_HEAD(&freelist, cur, entry); } else if (cur->rule.ptr != NULL) cur->rule.ptr->rule_flag |= PFRULE_REFS; PF_HASHROW_UNLOCK(sh); } pf_free_src_nodes(&freelist); V_pf_status.src_nodes = uma_zone_get_cur(V_pf_sources_z); } static void pf_src_tree_remove_state(struct pf_state *s) { struct pf_src_node *sn; struct pf_srchash *sh; uint32_t timeout; timeout = s->rule.ptr->timeout[PFTM_SRC_NODE] ? s->rule.ptr->timeout[PFTM_SRC_NODE] : V_pf_default_rule.timeout[PFTM_SRC_NODE]; if (s->src_node != NULL) { sn = s->src_node; sh = &V_pf_srchash[pf_hashsrc(&sn->addr, sn->af)]; PF_HASHROW_LOCK(sh); if (s->src.tcp_est) --sn->conn; if (--sn->states == 0) sn->expire = time_uptime + timeout; PF_HASHROW_UNLOCK(sh); } if (s->nat_src_node != s->src_node && s->nat_src_node != NULL) { sn = s->nat_src_node; sh = &V_pf_srchash[pf_hashsrc(&sn->addr, sn->af)]; PF_HASHROW_LOCK(sh); if (--sn->states == 0) sn->expire = time_uptime + timeout; PF_HASHROW_UNLOCK(sh); } s->src_node = s->nat_src_node = NULL; } /* * Unlink and potentilly free a state. Function may be * called with ID hash row locked, but always returns * unlocked, since it needs to go through key hash locking. */ int pf_unlink_state(struct pf_state *s, u_int flags) { struct pf_idhash *ih = &V_pf_idhash[PF_IDHASH(s)]; if ((flags & PF_ENTER_LOCKED) == 0) PF_HASHROW_LOCK(ih); else PF_HASHROW_ASSERT(ih); if (s->timeout == PFTM_UNLINKED) { /* * State is being processed * by pf_unlink_state() in * an other thread. */ PF_HASHROW_UNLOCK(ih); return (0); /* XXXGL: undefined actually */ } if (s->src.state == PF_TCPS_PROXY_DST) { /* XXX wire key the right one? */ pf_send_tcp(NULL, s->rule.ptr, s->key[PF_SK_WIRE]->af, &s->key[PF_SK_WIRE]->addr[1], &s->key[PF_SK_WIRE]->addr[0], s->key[PF_SK_WIRE]->port[1], s->key[PF_SK_WIRE]->port[0], s->src.seqhi, s->src.seqlo + 1, TH_RST|TH_ACK, 0, 0, 0, 1, s->tag, NULL); } LIST_REMOVE(s, entry); pf_src_tree_remove_state(s); if (pfsync_delete_state_ptr != NULL) pfsync_delete_state_ptr(s); STATE_DEC_COUNTERS(s); s->timeout = PFTM_UNLINKED; PF_HASHROW_UNLOCK(ih); pf_detach_state(s); refcount_release(&s->refs); return (pf_release_state(s)); } void pf_free_state(struct pf_state *cur) { KASSERT(cur->refs == 0, ("%s: %p has refs", __func__, cur)); KASSERT(cur->timeout == PFTM_UNLINKED, ("%s: timeout %u", __func__, cur->timeout)); pf_normalize_tcp_cleanup(cur); uma_zfree(V_pf_state_z, cur); counter_u64_add(V_pf_status.fcounters[FCNT_STATE_REMOVALS], 1); } /* * Called only from pf_purge_thread(), thus serialized. */ static u_int pf_purge_expired_states(u_int i, int maxcheck) { struct pf_idhash *ih; struct pf_state *s; V_pf_status.states = uma_zone_get_cur(V_pf_state_z); /* * Go through hash and unlink states that expire now. */ while (maxcheck > 0) { ih = &V_pf_idhash[i]; relock: PF_HASHROW_LOCK(ih); LIST_FOREACH(s, &ih->states, entry) { if (pf_state_expires(s) <= time_uptime) { V_pf_status.states -= pf_unlink_state(s, PF_ENTER_LOCKED); goto relock; } s->rule.ptr->rule_flag |= PFRULE_REFS; if (s->nat_rule.ptr != NULL) s->nat_rule.ptr->rule_flag |= PFRULE_REFS; if (s->anchor.ptr != NULL) s->anchor.ptr->rule_flag |= PFRULE_REFS; s->kif->pfik_flags |= PFI_IFLAG_REFS; if (s->rt_kif) s->rt_kif->pfik_flags |= PFI_IFLAG_REFS; } PF_HASHROW_UNLOCK(ih); /* Return when we hit end of hash. */ if (++i > pf_hashmask) { V_pf_status.states = uma_zone_get_cur(V_pf_state_z); return (0); } maxcheck--; } V_pf_status.states = uma_zone_get_cur(V_pf_state_z); return (i); } static void pf_purge_unlinked_rules() { struct pf_rulequeue tmpq; struct pf_rule *r, *r1; /* * If we have overloading task pending, then we'd * better skip purging this time. There is a tiny * probability that overloading task references * an already unlinked rule. */ PF_OVERLOADQ_LOCK(); if (!SLIST_EMPTY(&V_pf_overloadqueue)) { PF_OVERLOADQ_UNLOCK(); return; } PF_OVERLOADQ_UNLOCK(); /* * Do naive mark-and-sweep garbage collecting of old rules. * Reference flag is raised by pf_purge_expired_states() * and pf_purge_expired_src_nodes(). * * To avoid LOR between PF_UNLNKDRULES_LOCK/PF_RULES_WLOCK, * use a temporary queue. */ TAILQ_INIT(&tmpq); PF_UNLNKDRULES_LOCK(); TAILQ_FOREACH_SAFE(r, &V_pf_unlinked_rules, entries, r1) { if (!(r->rule_flag & PFRULE_REFS)) { TAILQ_REMOVE(&V_pf_unlinked_rules, r, entries); TAILQ_INSERT_TAIL(&tmpq, r, entries); } else r->rule_flag &= ~PFRULE_REFS; } PF_UNLNKDRULES_UNLOCK(); if (!TAILQ_EMPTY(&tmpq)) { PF_RULES_WLOCK(); TAILQ_FOREACH_SAFE(r, &tmpq, entries, r1) { TAILQ_REMOVE(&tmpq, r, entries); pf_free_rule(r); } PF_RULES_WUNLOCK(); } } void pf_print_host(struct pf_addr *addr, u_int16_t p, sa_family_t af) { switch (af) { #ifdef INET case AF_INET: { u_int32_t a = ntohl(addr->addr32[0]); printf("%u.%u.%u.%u", (a>>24)&255, (a>>16)&255, (a>>8)&255, a&255); if (p) { p = ntohs(p); printf(":%u", p); } break; } #endif /* INET */ #ifdef INET6 case AF_INET6: { u_int16_t b; u_int8_t i, curstart, curend, maxstart, maxend; curstart = curend = maxstart = maxend = 255; for (i = 0; i < 8; i++) { if (!addr->addr16[i]) { if (curstart == 255) curstart = i; curend = i; } else { if ((curend - curstart) > (maxend - maxstart)) { maxstart = curstart; maxend = curend; } curstart = curend = 255; } } if ((curend - curstart) > (maxend - maxstart)) { maxstart = curstart; maxend = curend; } for (i = 0; i < 8; i++) { if (i >= maxstart && i <= maxend) { if (i == 0) printf(":"); if (i == maxend) printf(":"); } else { b = ntohs(addr->addr16[i]); printf("%x", b); if (i < 7) printf(":"); } } if (p) { p = ntohs(p); printf("[%u]", p); } break; } #endif /* INET6 */ } } void pf_print_state(struct pf_state *s) { pf_print_state_parts(s, NULL, NULL); } static void pf_print_state_parts(struct pf_state *s, struct pf_state_key *skwp, struct pf_state_key *sksp) { struct pf_state_key *skw, *sks; u_int8_t proto, dir; /* Do our best to fill these, but they're skipped if NULL */ skw = skwp ? skwp : (s ? s->key[PF_SK_WIRE] : NULL); sks = sksp ? sksp : (s ? s->key[PF_SK_STACK] : NULL); proto = skw ? skw->proto : (sks ? sks->proto : 0); dir = s ? s->direction : 0; switch (proto) { case IPPROTO_IPV4: printf("IPv4"); break; case IPPROTO_IPV6: printf("IPv6"); break; case IPPROTO_TCP: printf("TCP"); break; case IPPROTO_UDP: printf("UDP"); break; case IPPROTO_ICMP: printf("ICMP"); break; case IPPROTO_ICMPV6: printf("ICMPv6"); break; default: printf("%u", proto); break; } switch (dir) { case PF_IN: printf(" in"); break; case PF_OUT: printf(" out"); break; } if (skw) { printf(" wire: "); pf_print_host(&skw->addr[0], skw->port[0], skw->af); printf(" "); pf_print_host(&skw->addr[1], skw->port[1], skw->af); } if (sks) { printf(" stack: "); if (sks != skw) { pf_print_host(&sks->addr[0], sks->port[0], sks->af); printf(" "); pf_print_host(&sks->addr[1], sks->port[1], sks->af); } else printf("-"); } if (s) { if (proto == IPPROTO_TCP) { printf(" [lo=%u high=%u win=%u modulator=%u", s->src.seqlo, s->src.seqhi, s->src.max_win, s->src.seqdiff); if (s->src.wscale && s->dst.wscale) printf(" wscale=%u", s->src.wscale & PF_WSCALE_MASK); printf("]"); printf(" [lo=%u high=%u win=%u modulator=%u", s->dst.seqlo, s->dst.seqhi, s->dst.max_win, s->dst.seqdiff); if (s->src.wscale && s->dst.wscale) printf(" wscale=%u", s->dst.wscale & PF_WSCALE_MASK); printf("]"); } printf(" %u:%u", s->src.state, s->dst.state); } } void pf_print_flags(u_int8_t f) { if (f) printf(" "); if (f & TH_FIN) printf("F"); if (f & TH_SYN) printf("S"); if (f & TH_RST) printf("R"); if (f & TH_PUSH) printf("P"); if (f & TH_ACK) printf("A"); if (f & TH_URG) printf("U"); if (f & TH_ECE) printf("E"); if (f & TH_CWR) printf("W"); } #define PF_SET_SKIP_STEPS(i) \ do { \ while (head[i] != cur) { \ head[i]->skip[i].ptr = cur; \ head[i] = TAILQ_NEXT(head[i], entries); \ } \ } while (0) void pf_calc_skip_steps(struct pf_rulequeue *rules) { struct pf_rule *cur, *prev, *head[PF_SKIP_COUNT]; int i; cur = TAILQ_FIRST(rules); prev = cur; for (i = 0; i < PF_SKIP_COUNT; ++i) head[i] = cur; while (cur != NULL) { if (cur->kif != prev->kif || cur->ifnot != prev->ifnot) PF_SET_SKIP_STEPS(PF_SKIP_IFP); if (cur->direction != prev->direction) PF_SET_SKIP_STEPS(PF_SKIP_DIR); if (cur->af != prev->af) PF_SET_SKIP_STEPS(PF_SKIP_AF); if (cur->proto != prev->proto) PF_SET_SKIP_STEPS(PF_SKIP_PROTO); if (cur->src.neg != prev->src.neg || pf_addr_wrap_neq(&cur->src.addr, &prev->src.addr)) PF_SET_SKIP_STEPS(PF_SKIP_SRC_ADDR); if (cur->src.port[0] != prev->src.port[0] || cur->src.port[1] != prev->src.port[1] || cur->src.port_op != prev->src.port_op) PF_SET_SKIP_STEPS(PF_SKIP_SRC_PORT); if (cur->dst.neg != prev->dst.neg || pf_addr_wrap_neq(&cur->dst.addr, &prev->dst.addr)) PF_SET_SKIP_STEPS(PF_SKIP_DST_ADDR); if (cur->dst.port[0] != prev->dst.port[0] || cur->dst.port[1] != prev->dst.port[1] || cur->dst.port_op != prev->dst.port_op) PF_SET_SKIP_STEPS(PF_SKIP_DST_PORT); prev = cur; cur = TAILQ_NEXT(cur, entries); } for (i = 0; i < PF_SKIP_COUNT; ++i) PF_SET_SKIP_STEPS(i); } static int pf_addr_wrap_neq(struct pf_addr_wrap *aw1, struct pf_addr_wrap *aw2) { if (aw1->type != aw2->type) return (1); switch (aw1->type) { case PF_ADDR_ADDRMASK: case PF_ADDR_RANGE: if (PF_ANEQ(&aw1->v.a.addr, &aw2->v.a.addr, AF_INET6)) return (1); if (PF_ANEQ(&aw1->v.a.mask, &aw2->v.a.mask, AF_INET6)) return (1); return (0); case PF_ADDR_DYNIFTL: return (aw1->p.dyn->pfid_kt != aw2->p.dyn->pfid_kt); case PF_ADDR_NOROUTE: case PF_ADDR_URPFFAILED: return (0); case PF_ADDR_TABLE: return (aw1->p.tbl != aw2->p.tbl); default: printf("invalid address type: %d\n", aw1->type); return (1); } } /** * Checksum updates are a little complicated because the checksum in the TCP/UDP * header isn't always a full checksum. In some cases (i.e. output) it's a * pseudo-header checksum, which is a partial checksum over src/dst IP * addresses, protocol number and length. * * That means we have the following cases: * * Input or forwarding: we don't have TSO, the checksum fields are full * checksums, we need to update the checksum whenever we change anything. * * Output (i.e. the checksum is a pseudo-header checksum): * x The field being updated is src/dst address or affects the length of * the packet. We need to update the pseudo-header checksum (note that this * checksum is not ones' complement). * x Some other field is being modified (e.g. src/dst port numbers): We * don't have to update anything. **/ u_int16_t pf_cksum_fixup(u_int16_t cksum, u_int16_t old, u_int16_t new, u_int8_t udp) { u_int32_t l; if (udp && !cksum) return (0x0000); l = cksum + old - new; l = (l >> 16) + (l & 65535); l = l & 65535; if (udp && !l) return (0xFFFF); return (l); } u_int16_t pf_proto_cksum_fixup(struct mbuf *m, u_int16_t cksum, u_int16_t old, u_int16_t new, u_int8_t udp) { if (m->m_pkthdr.csum_flags & (CSUM_DELAY_DATA | CSUM_DELAY_DATA_IPV6)) return (cksum); return (pf_cksum_fixup(cksum, old, new, udp)); } static void pf_change_ap(struct mbuf *m, struct pf_addr *a, u_int16_t *p, u_int16_t *ic, u_int16_t *pc, struct pf_addr *an, u_int16_t pn, u_int8_t u, sa_family_t af) { struct pf_addr ao; u_int16_t po = *p; PF_ACPY(&ao, a, af); PF_ACPY(a, an, af); if (m->m_pkthdr.csum_flags & (CSUM_DELAY_DATA | CSUM_DELAY_DATA_IPV6)) *pc = ~*pc; *p = pn; switch (af) { #ifdef INET case AF_INET: *ic = pf_cksum_fixup(pf_cksum_fixup(*ic, ao.addr16[0], an->addr16[0], 0), ao.addr16[1], an->addr16[1], 0); *p = pn; *pc = pf_cksum_fixup(pf_cksum_fixup(*pc, ao.addr16[0], an->addr16[0], u), ao.addr16[1], an->addr16[1], u); *pc = pf_proto_cksum_fixup(m, *pc, po, pn, u); break; #endif /* INET */ #ifdef INET6 case AF_INET6: *pc = pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( pf_cksum_fixup(pf_cksum_fixup(*pc, ao.addr16[0], an->addr16[0], u), ao.addr16[1], an->addr16[1], u), ao.addr16[2], an->addr16[2], u), ao.addr16[3], an->addr16[3], u), ao.addr16[4], an->addr16[4], u), ao.addr16[5], an->addr16[5], u), ao.addr16[6], an->addr16[6], u), ao.addr16[7], an->addr16[7], u); *pc = pf_proto_cksum_fixup(m, *pc, po, pn, u); break; #endif /* INET6 */ } if (m->m_pkthdr.csum_flags & (CSUM_DELAY_DATA | CSUM_DELAY_DATA_IPV6)) { *pc = ~*pc; if (! *pc) *pc = 0xffff; } } /* Changes a u_int32_t. Uses a void * so there are no align restrictions */ void pf_change_a(void *a, u_int16_t *c, u_int32_t an, u_int8_t u) { u_int32_t ao; memcpy(&ao, a, sizeof(ao)); memcpy(a, &an, sizeof(u_int32_t)); *c = pf_cksum_fixup(pf_cksum_fixup(*c, ao / 65536, an / 65536, u), ao % 65536, an % 65536, u); } void pf_change_proto_a(struct mbuf *m, void *a, u_int16_t *c, u_int32_t an, u_int8_t udp) { u_int32_t ao; memcpy(&ao, a, sizeof(ao)); memcpy(a, &an, sizeof(u_int32_t)); *c = pf_proto_cksum_fixup(m, pf_proto_cksum_fixup(m, *c, ao / 65536, an / 65536, udp), ao % 65536, an % 65536, udp); } #ifdef INET6 static void pf_change_a6(struct pf_addr *a, u_int16_t *c, struct pf_addr *an, u_int8_t u) { struct pf_addr ao; PF_ACPY(&ao, a, AF_INET6); PF_ACPY(a, an, AF_INET6); *c = pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( pf_cksum_fixup(pf_cksum_fixup(*c, ao.addr16[0], an->addr16[0], u), ao.addr16[1], an->addr16[1], u), ao.addr16[2], an->addr16[2], u), ao.addr16[3], an->addr16[3], u), ao.addr16[4], an->addr16[4], u), ao.addr16[5], an->addr16[5], u), ao.addr16[6], an->addr16[6], u), ao.addr16[7], an->addr16[7], u); } #endif /* INET6 */ static void pf_change_icmp(struct pf_addr *ia, u_int16_t *ip, struct pf_addr *oa, struct pf_addr *na, u_int16_t np, u_int16_t *pc, u_int16_t *h2c, u_int16_t *ic, u_int16_t *hc, u_int8_t u, sa_family_t af) { struct pf_addr oia, ooa; PF_ACPY(&oia, ia, af); if (oa) PF_ACPY(&ooa, oa, af); /* Change inner protocol port, fix inner protocol checksum. */ if (ip != NULL) { u_int16_t oip = *ip; u_int32_t opc; if (pc != NULL) opc = *pc; *ip = np; if (pc != NULL) *pc = pf_cksum_fixup(*pc, oip, *ip, u); *ic = pf_cksum_fixup(*ic, oip, *ip, 0); if (pc != NULL) *ic = pf_cksum_fixup(*ic, opc, *pc, 0); } /* Change inner ip address, fix inner ip and icmp checksums. */ PF_ACPY(ia, na, af); switch (af) { #ifdef INET case AF_INET: { u_int32_t oh2c = *h2c; *h2c = pf_cksum_fixup(pf_cksum_fixup(*h2c, oia.addr16[0], ia->addr16[0], 0), oia.addr16[1], ia->addr16[1], 0); *ic = pf_cksum_fixup(pf_cksum_fixup(*ic, oia.addr16[0], ia->addr16[0], 0), oia.addr16[1], ia->addr16[1], 0); *ic = pf_cksum_fixup(*ic, oh2c, *h2c, 0); break; } #endif /* INET */ #ifdef INET6 case AF_INET6: *ic = pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( pf_cksum_fixup(pf_cksum_fixup(*ic, oia.addr16[0], ia->addr16[0], u), oia.addr16[1], ia->addr16[1], u), oia.addr16[2], ia->addr16[2], u), oia.addr16[3], ia->addr16[3], u), oia.addr16[4], ia->addr16[4], u), oia.addr16[5], ia->addr16[5], u), oia.addr16[6], ia->addr16[6], u), oia.addr16[7], ia->addr16[7], u); break; #endif /* INET6 */ } /* Outer ip address, fix outer ip or icmpv6 checksum, if necessary. */ if (oa) { PF_ACPY(oa, na, af); switch (af) { #ifdef INET case AF_INET: *hc = pf_cksum_fixup(pf_cksum_fixup(*hc, ooa.addr16[0], oa->addr16[0], 0), ooa.addr16[1], oa->addr16[1], 0); break; #endif /* INET */ #ifdef INET6 case AF_INET6: *ic = pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( pf_cksum_fixup(pf_cksum_fixup(*ic, ooa.addr16[0], oa->addr16[0], u), ooa.addr16[1], oa->addr16[1], u), ooa.addr16[2], oa->addr16[2], u), ooa.addr16[3], oa->addr16[3], u), ooa.addr16[4], oa->addr16[4], u), ooa.addr16[5], oa->addr16[5], u), ooa.addr16[6], oa->addr16[6], u), ooa.addr16[7], oa->addr16[7], u); break; #endif /* INET6 */ } } } /* * Need to modulate the sequence numbers in the TCP SACK option * (credits to Krzysztof Pfaff for report and patch) */ static int pf_modulate_sack(struct mbuf *m, int off, struct pf_pdesc *pd, struct tcphdr *th, struct pf_state_peer *dst) { int hlen = (th->th_off << 2) - sizeof(*th), thoptlen = hlen; u_int8_t opts[TCP_MAXOLEN], *opt = opts; int copyback = 0, i, olen; struct sackblk sack; #define TCPOLEN_SACKLEN (TCPOLEN_SACK + 2) if (hlen < TCPOLEN_SACKLEN || !pf_pull_hdr(m, off + sizeof(*th), opts, hlen, NULL, NULL, pd->af)) return 0; while (hlen >= TCPOLEN_SACKLEN) { olen = opt[1]; switch (*opt) { case TCPOPT_EOL: /* FALLTHROUGH */ case TCPOPT_NOP: opt++; hlen--; break; case TCPOPT_SACK: if (olen > hlen) olen = hlen; if (olen >= TCPOLEN_SACKLEN) { for (i = 2; i + TCPOLEN_SACK <= olen; i += TCPOLEN_SACK) { memcpy(&sack, &opt[i], sizeof(sack)); pf_change_proto_a(m, &sack.start, &th->th_sum, htonl(ntohl(sack.start) - dst->seqdiff), 0); pf_change_proto_a(m, &sack.end, &th->th_sum, htonl(ntohl(sack.end) - dst->seqdiff), 0); memcpy(&opt[i], &sack, sizeof(sack)); } copyback = 1; } /* FALLTHROUGH */ default: if (olen < 2) olen = 2; hlen -= olen; opt += olen; } } if (copyback) m_copyback(m, off + sizeof(*th), thoptlen, (caddr_t)opts); return (copyback); } static void pf_send_tcp(struct mbuf *replyto, const struct pf_rule *r, sa_family_t af, const struct pf_addr *saddr, const struct pf_addr *daddr, u_int16_t sport, u_int16_t dport, u_int32_t seq, u_int32_t ack, u_int8_t flags, u_int16_t win, u_int16_t mss, u_int8_t ttl, int tag, u_int16_t rtag, struct ifnet *ifp) { struct pf_send_entry *pfse; struct mbuf *m; int len, tlen; #ifdef INET struct ip *h = NULL; #endif /* INET */ #ifdef INET6 struct ip6_hdr *h6 = NULL; #endif /* INET6 */ struct tcphdr *th; char *opt; struct pf_mtag *pf_mtag; len = 0; th = NULL; /* maximum segment size tcp option */ tlen = sizeof(struct tcphdr); if (mss) tlen += 4; switch (af) { #ifdef INET case AF_INET: len = sizeof(struct ip) + tlen; break; #endif /* INET */ #ifdef INET6 case AF_INET6: len = sizeof(struct ip6_hdr) + tlen; break; #endif /* INET6 */ default: panic("%s: unsupported af %d", __func__, af); } /* Allocate outgoing queue entry, mbuf and mbuf tag. */ pfse = malloc(sizeof(*pfse), M_PFTEMP, M_NOWAIT); if (pfse == NULL) return; m = m_gethdr(M_NOWAIT, MT_DATA); if (m == NULL) { free(pfse, M_PFTEMP); return; } #ifdef MAC mac_netinet_firewall_send(m); #endif if ((pf_mtag = pf_get_mtag(m)) == NULL) { free(pfse, M_PFTEMP); m_freem(m); return; } if (tag) m->m_flags |= M_SKIP_FIREWALL; pf_mtag->tag = rtag; if (r != NULL && r->rtableid >= 0) M_SETFIB(m, r->rtableid); #ifdef ALTQ if (r != NULL && r->qid) { pf_mtag->qid = r->qid; /* add hints for ecn */ pf_mtag->hdr = mtod(m, struct ip *); } #endif /* ALTQ */ m->m_data += max_linkhdr; m->m_pkthdr.len = m->m_len = len; m->m_pkthdr.rcvif = NULL; bzero(m->m_data, len); switch (af) { #ifdef INET case AF_INET: h = mtod(m, struct ip *); /* IP header fields included in the TCP checksum */ h->ip_p = IPPROTO_TCP; h->ip_len = htons(tlen); h->ip_src.s_addr = saddr->v4.s_addr; h->ip_dst.s_addr = daddr->v4.s_addr; th = (struct tcphdr *)((caddr_t)h + sizeof(struct ip)); break; #endif /* INET */ #ifdef INET6 case AF_INET6: h6 = mtod(m, struct ip6_hdr *); /* IP header fields included in the TCP checksum */ h6->ip6_nxt = IPPROTO_TCP; h6->ip6_plen = htons(tlen); memcpy(&h6->ip6_src, &saddr->v6, sizeof(struct in6_addr)); memcpy(&h6->ip6_dst, &daddr->v6, sizeof(struct in6_addr)); th = (struct tcphdr *)((caddr_t)h6 + sizeof(struct ip6_hdr)); break; #endif /* INET6 */ } /* TCP header */ th->th_sport = sport; th->th_dport = dport; th->th_seq = htonl(seq); th->th_ack = htonl(ack); th->th_off = tlen >> 2; th->th_flags = flags; th->th_win = htons(win); if (mss) { opt = (char *)(th + 1); opt[0] = TCPOPT_MAXSEG; opt[1] = 4; HTONS(mss); bcopy((caddr_t)&mss, (caddr_t)(opt + 2), 2); } switch (af) { #ifdef INET case AF_INET: /* TCP checksum */ th->th_sum = in_cksum(m, len); /* Finish the IP header */ h->ip_v = 4; h->ip_hl = sizeof(*h) >> 2; h->ip_tos = IPTOS_LOWDELAY; h->ip_off = htons(V_path_mtu_discovery ? IP_DF : 0); h->ip_len = htons(len); h->ip_ttl = ttl ? ttl : V_ip_defttl; h->ip_sum = 0; pfse->pfse_type = PFSE_IP; break; #endif /* INET */ #ifdef INET6 case AF_INET6: /* TCP checksum */ th->th_sum = in6_cksum(m, IPPROTO_TCP, sizeof(struct ip6_hdr), tlen); h6->ip6_vfc |= IPV6_VERSION; h6->ip6_hlim = IPV6_DEFHLIM; pfse->pfse_type = PFSE_IP6; break; #endif /* INET6 */ } pfse->pfse_m = m; pf_send(pfse); } static int pf_ieee8021q_setpcp(struct mbuf *m, u_int8_t prio) { struct m_tag *mtag; KASSERT(prio <= PF_PRIO_MAX, ("%s with invalid pcp", __func__)); mtag = m_tag_locate(m, MTAG_8021Q, MTAG_8021Q_PCP_OUT, NULL); if (mtag == NULL) { mtag = m_tag_alloc(MTAG_8021Q, MTAG_8021Q_PCP_OUT, sizeof(uint8_t), M_NOWAIT); if (mtag == NULL) return (ENOMEM); m_tag_prepend(m, mtag); } *(uint8_t *)(mtag + 1) = prio; return (0); } static int pf_match_ieee8021q_pcp(u_int8_t prio, struct mbuf *m) { struct m_tag *mtag; u_int8_t mpcp; mtag = m_tag_locate(m, MTAG_8021Q, MTAG_8021Q_PCP_IN, NULL); if (mtag == NULL) return (0); if (prio == PF_PRIO_ZERO) prio = 0; mpcp = *(uint8_t *)(mtag + 1); return (mpcp == prio); } static void pf_send_icmp(struct mbuf *m, u_int8_t type, u_int8_t code, sa_family_t af, struct pf_rule *r) { struct pf_send_entry *pfse; struct mbuf *m0; struct pf_mtag *pf_mtag; /* Allocate outgoing queue entry, mbuf and mbuf tag. */ pfse = malloc(sizeof(*pfse), M_PFTEMP, M_NOWAIT); if (pfse == NULL) return; if ((m0 = m_copypacket(m, M_NOWAIT)) == NULL) { free(pfse, M_PFTEMP); return; } if ((pf_mtag = pf_get_mtag(m0)) == NULL) { free(pfse, M_PFTEMP); return; } /* XXX: revisit */ m0->m_flags |= M_SKIP_FIREWALL; if (r->rtableid >= 0) M_SETFIB(m0, r->rtableid); #ifdef ALTQ if (r->qid) { pf_mtag->qid = r->qid; /* add hints for ecn */ pf_mtag->hdr = mtod(m0, struct ip *); } #endif /* ALTQ */ switch (af) { #ifdef INET case AF_INET: pfse->pfse_type = PFSE_ICMP; break; #endif /* INET */ #ifdef INET6 case AF_INET6: pfse->pfse_type = PFSE_ICMP6; break; #endif /* INET6 */ } pfse->pfse_m = m0; pfse->icmpopts.type = type; pfse->icmpopts.code = code; pf_send(pfse); } /* * Return 1 if the addresses a and b match (with mask m), otherwise return 0. * If n is 0, they match if they are equal. If n is != 0, they match if they * are different. */ int pf_match_addr(u_int8_t n, struct pf_addr *a, struct pf_addr *m, struct pf_addr *b, sa_family_t af) { int match = 0; switch (af) { #ifdef INET case AF_INET: if ((a->addr32[0] & m->addr32[0]) == (b->addr32[0] & m->addr32[0])) match++; break; #endif /* INET */ #ifdef INET6 case AF_INET6: if (((a->addr32[0] & m->addr32[0]) == (b->addr32[0] & m->addr32[0])) && ((a->addr32[1] & m->addr32[1]) == (b->addr32[1] & m->addr32[1])) && ((a->addr32[2] & m->addr32[2]) == (b->addr32[2] & m->addr32[2])) && ((a->addr32[3] & m->addr32[3]) == (b->addr32[3] & m->addr32[3]))) match++; break; #endif /* INET6 */ } if (match) { if (n) return (0); else return (1); } else { if (n) return (1); else return (0); } } /* * Return 1 if b <= a <= e, otherwise return 0. */ int pf_match_addr_range(struct pf_addr *b, struct pf_addr *e, struct pf_addr *a, sa_family_t af) { switch (af) { #ifdef INET case AF_INET: if ((a->addr32[0] < b->addr32[0]) || (a->addr32[0] > e->addr32[0])) return (0); break; #endif /* INET */ #ifdef INET6 case AF_INET6: { int i; /* check a >= b */ for (i = 0; i < 4; ++i) if (a->addr32[i] > b->addr32[i]) break; else if (a->addr32[i] < b->addr32[i]) return (0); /* check a <= e */ for (i = 0; i < 4; ++i) if (a->addr32[i] < e->addr32[i]) break; else if (a->addr32[i] > e->addr32[i]) return (0); break; } #endif /* INET6 */ } return (1); } static int pf_match(u_int8_t op, u_int32_t a1, u_int32_t a2, u_int32_t p) { switch (op) { case PF_OP_IRG: return ((p > a1) && (p < a2)); case PF_OP_XRG: return ((p < a1) || (p > a2)); case PF_OP_RRG: return ((p >= a1) && (p <= a2)); case PF_OP_EQ: return (p == a1); case PF_OP_NE: return (p != a1); case PF_OP_LT: return (p < a1); case PF_OP_LE: return (p <= a1); case PF_OP_GT: return (p > a1); case PF_OP_GE: return (p >= a1); } return (0); /* never reached */ } int pf_match_port(u_int8_t op, u_int16_t a1, u_int16_t a2, u_int16_t p) { NTOHS(a1); NTOHS(a2); NTOHS(p); return (pf_match(op, a1, a2, p)); } static int pf_match_uid(u_int8_t op, uid_t a1, uid_t a2, uid_t u) { if (u == UID_MAX && op != PF_OP_EQ && op != PF_OP_NE) return (0); return (pf_match(op, a1, a2, u)); } static int pf_match_gid(u_int8_t op, gid_t a1, gid_t a2, gid_t g) { if (g == GID_MAX && op != PF_OP_EQ && op != PF_OP_NE) return (0); return (pf_match(op, a1, a2, g)); } int pf_match_tag(struct mbuf *m, struct pf_rule *r, int *tag, int mtag) { if (*tag == -1) *tag = mtag; return ((!r->match_tag_not && r->match_tag == *tag) || (r->match_tag_not && r->match_tag != *tag)); } int pf_tag_packet(struct mbuf *m, struct pf_pdesc *pd, int tag) { KASSERT(tag > 0, ("%s: tag %d", __func__, tag)); if (pd->pf_mtag == NULL && ((pd->pf_mtag = pf_get_mtag(m)) == NULL)) return (ENOMEM); pd->pf_mtag->tag = tag; return (0); } #define PF_ANCHOR_STACKSIZE 32 struct pf_anchor_stackframe { struct pf_ruleset *rs; struct pf_rule *r; /* XXX: + match bit */ struct pf_anchor *child; }; /* * XXX: We rely on malloc(9) returning pointer aligned addresses. */ #define PF_ANCHORSTACK_MATCH 0x00000001 #define PF_ANCHORSTACK_MASK (PF_ANCHORSTACK_MATCH) #define PF_ANCHOR_MATCH(f) ((uintptr_t)(f)->r & PF_ANCHORSTACK_MATCH) #define PF_ANCHOR_RULE(f) (struct pf_rule *) \ ((uintptr_t)(f)->r & ~PF_ANCHORSTACK_MASK) #define PF_ANCHOR_SET_MATCH(f) do { (f)->r = (void *) \ ((uintptr_t)(f)->r | PF_ANCHORSTACK_MATCH); \ } while (0) void pf_step_into_anchor(struct pf_anchor_stackframe *stack, int *depth, struct pf_ruleset **rs, int n, struct pf_rule **r, struct pf_rule **a, int *match) { struct pf_anchor_stackframe *f; PF_RULES_RASSERT(); if (match) *match = 0; if (*depth >= PF_ANCHOR_STACKSIZE) { printf("%s: anchor stack overflow on %s\n", __func__, (*r)->anchor->name); *r = TAILQ_NEXT(*r, entries); return; } else if (*depth == 0 && a != NULL) *a = *r; f = stack + (*depth)++; f->rs = *rs; f->r = *r; if ((*r)->anchor_wildcard) { struct pf_anchor_node *parent = &(*r)->anchor->children; if ((f->child = RB_MIN(pf_anchor_node, parent)) == NULL) { *r = NULL; return; } *rs = &f->child->ruleset; } else { f->child = NULL; *rs = &(*r)->anchor->ruleset; } *r = TAILQ_FIRST((*rs)->rules[n].active.ptr); } int pf_step_out_of_anchor(struct pf_anchor_stackframe *stack, int *depth, struct pf_ruleset **rs, int n, struct pf_rule **r, struct pf_rule **a, int *match) { struct pf_anchor_stackframe *f; struct pf_rule *fr; int quick = 0; PF_RULES_RASSERT(); do { if (*depth <= 0) break; f = stack + *depth - 1; fr = PF_ANCHOR_RULE(f); if (f->child != NULL) { struct pf_anchor_node *parent; /* * This block traverses through * a wildcard anchor. */ parent = &fr->anchor->children; if (match != NULL && *match) { /* * If any of "*" matched, then * "foo/ *" matched, mark frame * appropriately. */ PF_ANCHOR_SET_MATCH(f); *match = 0; } f->child = RB_NEXT(pf_anchor_node, parent, f->child); if (f->child != NULL) { *rs = &f->child->ruleset; *r = TAILQ_FIRST((*rs)->rules[n].active.ptr); if (*r == NULL) continue; else break; } } (*depth)--; if (*depth == 0 && a != NULL) *a = NULL; *rs = f->rs; if (PF_ANCHOR_MATCH(f) || (match != NULL && *match)) quick = fr->quick; *r = TAILQ_NEXT(fr, entries); } while (*r == NULL); return (quick); } #ifdef INET6 void pf_poolmask(struct pf_addr *naddr, struct pf_addr *raddr, struct pf_addr *rmask, struct pf_addr *saddr, sa_family_t af) { switch (af) { #ifdef INET case AF_INET: naddr->addr32[0] = (raddr->addr32[0] & rmask->addr32[0]) | ((rmask->addr32[0] ^ 0xffffffff ) & saddr->addr32[0]); break; #endif /* INET */ case AF_INET6: naddr->addr32[0] = (raddr->addr32[0] & rmask->addr32[0]) | ((rmask->addr32[0] ^ 0xffffffff ) & saddr->addr32[0]); naddr->addr32[1] = (raddr->addr32[1] & rmask->addr32[1]) | ((rmask->addr32[1] ^ 0xffffffff ) & saddr->addr32[1]); naddr->addr32[2] = (raddr->addr32[2] & rmask->addr32[2]) | ((rmask->addr32[2] ^ 0xffffffff ) & saddr->addr32[2]); naddr->addr32[3] = (raddr->addr32[3] & rmask->addr32[3]) | ((rmask->addr32[3] ^ 0xffffffff ) & saddr->addr32[3]); break; } } void pf_addr_inc(struct pf_addr *addr, sa_family_t af) { switch (af) { #ifdef INET case AF_INET: addr->addr32[0] = htonl(ntohl(addr->addr32[0]) + 1); break; #endif /* INET */ case AF_INET6: if (addr->addr32[3] == 0xffffffff) { addr->addr32[3] = 0; if (addr->addr32[2] == 0xffffffff) { addr->addr32[2] = 0; if (addr->addr32[1] == 0xffffffff) { addr->addr32[1] = 0; addr->addr32[0] = htonl(ntohl(addr->addr32[0]) + 1); } else addr->addr32[1] = htonl(ntohl(addr->addr32[1]) + 1); } else addr->addr32[2] = htonl(ntohl(addr->addr32[2]) + 1); } else addr->addr32[3] = htonl(ntohl(addr->addr32[3]) + 1); break; } } #endif /* INET6 */ int pf_socket_lookup(int direction, struct pf_pdesc *pd, struct mbuf *m) { struct pf_addr *saddr, *daddr; u_int16_t sport, dport; struct inpcbinfo *pi; struct inpcb *inp; pd->lookup.uid = UID_MAX; pd->lookup.gid = GID_MAX; switch (pd->proto) { case IPPROTO_TCP: if (pd->hdr.tcp == NULL) return (-1); sport = pd->hdr.tcp->th_sport; dport = pd->hdr.tcp->th_dport; pi = &V_tcbinfo; break; case IPPROTO_UDP: if (pd->hdr.udp == NULL) return (-1); sport = pd->hdr.udp->uh_sport; dport = pd->hdr.udp->uh_dport; pi = &V_udbinfo; break; default: return (-1); } if (direction == PF_IN) { saddr = pd->src; daddr = pd->dst; } else { u_int16_t p; p = sport; sport = dport; dport = p; saddr = pd->dst; daddr = pd->src; } switch (pd->af) { #ifdef INET case AF_INET: inp = in_pcblookup_mbuf(pi, saddr->v4, sport, daddr->v4, dport, INPLOOKUP_RLOCKPCB, NULL, m); if (inp == NULL) { inp = in_pcblookup_mbuf(pi, saddr->v4, sport, daddr->v4, dport, INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB, NULL, m); if (inp == NULL) return (-1); } break; #endif /* INET */ #ifdef INET6 case AF_INET6: inp = in6_pcblookup_mbuf(pi, &saddr->v6, sport, &daddr->v6, dport, INPLOOKUP_RLOCKPCB, NULL, m); if (inp == NULL) { inp = in6_pcblookup_mbuf(pi, &saddr->v6, sport, &daddr->v6, dport, INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB, NULL, m); if (inp == NULL) return (-1); } break; #endif /* INET6 */ default: return (-1); } INP_RLOCK_ASSERT(inp); pd->lookup.uid = inp->inp_cred->cr_uid; pd->lookup.gid = inp->inp_cred->cr_groups[0]; INP_RUNLOCK(inp); return (1); } static u_int8_t pf_get_wscale(struct mbuf *m, int off, u_int16_t th_off, sa_family_t af) { int hlen; u_int8_t hdr[60]; u_int8_t *opt, optlen; u_int8_t wscale = 0; hlen = th_off << 2; /* hlen <= sizeof(hdr) */ if (hlen <= sizeof(struct tcphdr)) return (0); if (!pf_pull_hdr(m, off, hdr, hlen, NULL, NULL, af)) return (0); opt = hdr + sizeof(struct tcphdr); hlen -= sizeof(struct tcphdr); while (hlen >= 3) { switch (*opt) { case TCPOPT_EOL: case TCPOPT_NOP: ++opt; --hlen; break; case TCPOPT_WINDOW: wscale = opt[2]; if (wscale > TCP_MAX_WINSHIFT) wscale = TCP_MAX_WINSHIFT; wscale |= PF_WSCALE_FLAG; /* FALLTHROUGH */ default: optlen = opt[1]; if (optlen < 2) optlen = 2; hlen -= optlen; opt += optlen; break; } } return (wscale); } static u_int16_t pf_get_mss(struct mbuf *m, int off, u_int16_t th_off, sa_family_t af) { int hlen; u_int8_t hdr[60]; u_int8_t *opt, optlen; u_int16_t mss = V_tcp_mssdflt; hlen = th_off << 2; /* hlen <= sizeof(hdr) */ if (hlen <= sizeof(struct tcphdr)) return (0); if (!pf_pull_hdr(m, off, hdr, hlen, NULL, NULL, af)) return (0); opt = hdr + sizeof(struct tcphdr); hlen -= sizeof(struct tcphdr); while (hlen >= TCPOLEN_MAXSEG) { switch (*opt) { case TCPOPT_EOL: case TCPOPT_NOP: ++opt; --hlen; break; case TCPOPT_MAXSEG: bcopy((caddr_t)(opt + 2), (caddr_t)&mss, 2); NTOHS(mss); /* FALLTHROUGH */ default: optlen = opt[1]; if (optlen < 2) optlen = 2; hlen -= optlen; opt += optlen; break; } } return (mss); } static u_int16_t pf_calc_mss(struct pf_addr *addr, sa_family_t af, int rtableid, u_int16_t offer) { #ifdef INET struct nhop4_basic nh4; #endif /* INET */ #ifdef INET6 struct nhop6_basic nh6; struct in6_addr dst6; uint32_t scopeid; #endif /* INET6 */ int hlen = 0; uint16_t mss = 0; switch (af) { #ifdef INET case AF_INET: hlen = sizeof(struct ip); if (fib4_lookup_nh_basic(rtableid, addr->v4, 0, 0, &nh4) == 0) mss = nh4.nh_mtu - hlen - sizeof(struct tcphdr); break; #endif /* INET */ #ifdef INET6 case AF_INET6: hlen = sizeof(struct ip6_hdr); in6_splitscope(&addr->v6, &dst6, &scopeid); if (fib6_lookup_nh_basic(rtableid, &dst6, scopeid, 0,0,&nh6)==0) mss = nh6.nh_mtu - hlen - sizeof(struct tcphdr); break; #endif /* INET6 */ } mss = max(V_tcp_mssdflt, mss); mss = min(mss, offer); mss = max(mss, 64); /* sanity - at least max opt space */ return (mss); } static u_int32_t pf_tcp_iss(struct pf_pdesc *pd) { MD5_CTX ctx; u_int32_t digest[4]; if (V_pf_tcp_secret_init == 0) { read_random(&V_pf_tcp_secret, sizeof(V_pf_tcp_secret)); MD5Init(&V_pf_tcp_secret_ctx); MD5Update(&V_pf_tcp_secret_ctx, V_pf_tcp_secret, sizeof(V_pf_tcp_secret)); V_pf_tcp_secret_init = 1; } ctx = V_pf_tcp_secret_ctx; MD5Update(&ctx, (char *)&pd->hdr.tcp->th_sport, sizeof(u_short)); MD5Update(&ctx, (char *)&pd->hdr.tcp->th_dport, sizeof(u_short)); if (pd->af == AF_INET6) { MD5Update(&ctx, (char *)&pd->src->v6, sizeof(struct in6_addr)); MD5Update(&ctx, (char *)&pd->dst->v6, sizeof(struct in6_addr)); } else { MD5Update(&ctx, (char *)&pd->src->v4, sizeof(struct in_addr)); MD5Update(&ctx, (char *)&pd->dst->v4, sizeof(struct in_addr)); } MD5Final((u_char *)digest, &ctx); V_pf_tcp_iss_off += 4096; #define ISN_RANDOM_INCREMENT (4096 - 1) return (digest[0] + (arc4random() & ISN_RANDOM_INCREMENT) + V_pf_tcp_iss_off); #undef ISN_RANDOM_INCREMENT } static int pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, struct pfi_kif *kif, struct mbuf *m, int off, struct pf_pdesc *pd, struct pf_rule **am, struct pf_ruleset **rsm, struct inpcb *inp) { struct pf_rule *nr = NULL; struct pf_addr * const saddr = pd->src; struct pf_addr * const daddr = pd->dst; sa_family_t af = pd->af; struct pf_rule *r, *a = NULL; struct pf_ruleset *ruleset = NULL; struct pf_src_node *nsn = NULL; struct tcphdr *th = pd->hdr.tcp; struct pf_state_key *sk = NULL, *nk = NULL; u_short reason; int rewrite = 0, hdrlen = 0; int tag = -1, rtableid = -1; int asd = 0; int match = 0; int state_icmp = 0; u_int16_t sport = 0, dport = 0; u_int16_t bproto_sum = 0, bip_sum = 0; u_int8_t icmptype = 0, icmpcode = 0; struct pf_anchor_stackframe anchor_stack[PF_ANCHOR_STACKSIZE]; PF_RULES_RASSERT(); if (inp != NULL) { INP_LOCK_ASSERT(inp); pd->lookup.uid = inp->inp_cred->cr_uid; pd->lookup.gid = inp->inp_cred->cr_groups[0]; pd->lookup.done = 1; } switch (pd->proto) { case IPPROTO_TCP: sport = th->th_sport; dport = th->th_dport; hdrlen = sizeof(*th); break; case IPPROTO_UDP: sport = pd->hdr.udp->uh_sport; dport = pd->hdr.udp->uh_dport; hdrlen = sizeof(*pd->hdr.udp); break; #ifdef INET case IPPROTO_ICMP: if (pd->af != AF_INET) break; sport = dport = pd->hdr.icmp->icmp_id; hdrlen = sizeof(*pd->hdr.icmp); icmptype = pd->hdr.icmp->icmp_type; icmpcode = pd->hdr.icmp->icmp_code; if (icmptype == ICMP_UNREACH || icmptype == ICMP_SOURCEQUENCH || icmptype == ICMP_REDIRECT || icmptype == ICMP_TIMXCEED || icmptype == ICMP_PARAMPROB) state_icmp++; break; #endif /* INET */ #ifdef INET6 case IPPROTO_ICMPV6: if (af != AF_INET6) break; sport = dport = pd->hdr.icmp6->icmp6_id; hdrlen = sizeof(*pd->hdr.icmp6); icmptype = pd->hdr.icmp6->icmp6_type; icmpcode = pd->hdr.icmp6->icmp6_code; if (icmptype == ICMP6_DST_UNREACH || icmptype == ICMP6_PACKET_TOO_BIG || icmptype == ICMP6_TIME_EXCEEDED || icmptype == ICMP6_PARAM_PROB) state_icmp++; break; #endif /* INET6 */ default: sport = dport = hdrlen = 0; break; } r = TAILQ_FIRST(pf_main_ruleset.rules[PF_RULESET_FILTER].active.ptr); /* check packet for BINAT/NAT/RDR */ if ((nr = pf_get_translation(pd, m, off, direction, kif, &nsn, &sk, &nk, saddr, daddr, sport, dport, anchor_stack)) != NULL) { KASSERT(sk != NULL, ("%s: null sk", __func__)); KASSERT(nk != NULL, ("%s: null nk", __func__)); if (pd->ip_sum) bip_sum = *pd->ip_sum; switch (pd->proto) { case IPPROTO_TCP: bproto_sum = th->th_sum; pd->proto_sum = &th->th_sum; if (PF_ANEQ(saddr, &nk->addr[pd->sidx], af) || nk->port[pd->sidx] != sport) { pf_change_ap(m, saddr, &th->th_sport, pd->ip_sum, &th->th_sum, &nk->addr[pd->sidx], nk->port[pd->sidx], 0, af); pd->sport = &th->th_sport; sport = th->th_sport; } if (PF_ANEQ(daddr, &nk->addr[pd->didx], af) || nk->port[pd->didx] != dport) { pf_change_ap(m, daddr, &th->th_dport, pd->ip_sum, &th->th_sum, &nk->addr[pd->didx], nk->port[pd->didx], 0, af); dport = th->th_dport; pd->dport = &th->th_dport; } rewrite++; break; case IPPROTO_UDP: bproto_sum = pd->hdr.udp->uh_sum; pd->proto_sum = &pd->hdr.udp->uh_sum; if (PF_ANEQ(saddr, &nk->addr[pd->sidx], af) || nk->port[pd->sidx] != sport) { pf_change_ap(m, saddr, &pd->hdr.udp->uh_sport, pd->ip_sum, &pd->hdr.udp->uh_sum, &nk->addr[pd->sidx], nk->port[pd->sidx], 1, af); sport = pd->hdr.udp->uh_sport; pd->sport = &pd->hdr.udp->uh_sport; } if (PF_ANEQ(daddr, &nk->addr[pd->didx], af) || nk->port[pd->didx] != dport) { pf_change_ap(m, daddr, &pd->hdr.udp->uh_dport, pd->ip_sum, &pd->hdr.udp->uh_sum, &nk->addr[pd->didx], nk->port[pd->didx], 1, af); dport = pd->hdr.udp->uh_dport; pd->dport = &pd->hdr.udp->uh_dport; } rewrite++; break; #ifdef INET case IPPROTO_ICMP: nk->port[0] = nk->port[1]; if (PF_ANEQ(saddr, &nk->addr[pd->sidx], AF_INET)) pf_change_a(&saddr->v4.s_addr, pd->ip_sum, nk->addr[pd->sidx].v4.s_addr, 0); if (PF_ANEQ(daddr, &nk->addr[pd->didx], AF_INET)) pf_change_a(&daddr->v4.s_addr, pd->ip_sum, nk->addr[pd->didx].v4.s_addr, 0); if (nk->port[1] != pd->hdr.icmp->icmp_id) { pd->hdr.icmp->icmp_cksum = pf_cksum_fixup( pd->hdr.icmp->icmp_cksum, sport, nk->port[1], 0); pd->hdr.icmp->icmp_id = nk->port[1]; pd->sport = &pd->hdr.icmp->icmp_id; } m_copyback(m, off, ICMP_MINLEN, (caddr_t)pd->hdr.icmp); break; #endif /* INET */ #ifdef INET6 case IPPROTO_ICMPV6: nk->port[0] = nk->port[1]; if (PF_ANEQ(saddr, &nk->addr[pd->sidx], AF_INET6)) pf_change_a6(saddr, &pd->hdr.icmp6->icmp6_cksum, &nk->addr[pd->sidx], 0); if (PF_ANEQ(daddr, &nk->addr[pd->didx], AF_INET6)) pf_change_a6(daddr, &pd->hdr.icmp6->icmp6_cksum, &nk->addr[pd->didx], 0); rewrite++; break; #endif /* INET */ default: switch (af) { #ifdef INET case AF_INET: if (PF_ANEQ(saddr, &nk->addr[pd->sidx], AF_INET)) pf_change_a(&saddr->v4.s_addr, pd->ip_sum, nk->addr[pd->sidx].v4.s_addr, 0); if (PF_ANEQ(daddr, &nk->addr[pd->didx], AF_INET)) pf_change_a(&daddr->v4.s_addr, pd->ip_sum, nk->addr[pd->didx].v4.s_addr, 0); break; #endif /* INET */ #ifdef INET6 case AF_INET6: if (PF_ANEQ(saddr, &nk->addr[pd->sidx], AF_INET6)) PF_ACPY(saddr, &nk->addr[pd->sidx], af); if (PF_ANEQ(daddr, &nk->addr[pd->didx], AF_INET6)) PF_ACPY(saddr, &nk->addr[pd->didx], af); break; #endif /* INET */ } break; } if (nr->natpass) r = NULL; pd->nat_rule = nr; } while (r != NULL) { r->evaluations++; if (pfi_kif_match(r->kif, kif) == r->ifnot) r = r->skip[PF_SKIP_IFP].ptr; else if (r->direction && r->direction != direction) r = r->skip[PF_SKIP_DIR].ptr; else if (r->af && r->af != af) r = r->skip[PF_SKIP_AF].ptr; else if (r->proto && r->proto != pd->proto) r = r->skip[PF_SKIP_PROTO].ptr; else if (PF_MISMATCHAW(&r->src.addr, saddr, af, r->src.neg, kif, M_GETFIB(m))) r = r->skip[PF_SKIP_SRC_ADDR].ptr; /* tcp/udp only. port_op always 0 in other cases */ else if (r->src.port_op && !pf_match_port(r->src.port_op, r->src.port[0], r->src.port[1], sport)) r = r->skip[PF_SKIP_SRC_PORT].ptr; else if (PF_MISMATCHAW(&r->dst.addr, daddr, af, r->dst.neg, NULL, M_GETFIB(m))) r = r->skip[PF_SKIP_DST_ADDR].ptr; /* tcp/udp only. port_op always 0 in other cases */ else if (r->dst.port_op && !pf_match_port(r->dst.port_op, r->dst.port[0], r->dst.port[1], dport)) r = r->skip[PF_SKIP_DST_PORT].ptr; /* icmp only. type always 0 in other cases */ else if (r->type && r->type != icmptype + 1) r = TAILQ_NEXT(r, entries); /* icmp only. type always 0 in other cases */ else if (r->code && r->code != icmpcode + 1) r = TAILQ_NEXT(r, entries); else if (r->tos && !(r->tos == pd->tos)) r = TAILQ_NEXT(r, entries); else if (r->rule_flag & PFRULE_FRAGMENT) r = TAILQ_NEXT(r, entries); else if (pd->proto == IPPROTO_TCP && (r->flagset & th->th_flags) != r->flags) r = TAILQ_NEXT(r, entries); /* tcp/udp only. uid.op always 0 in other cases */ else if (r->uid.op && (pd->lookup.done || (pd->lookup.done = pf_socket_lookup(direction, pd, m), 1)) && !pf_match_uid(r->uid.op, r->uid.uid[0], r->uid.uid[1], pd->lookup.uid)) r = TAILQ_NEXT(r, entries); /* tcp/udp only. gid.op always 0 in other cases */ else if (r->gid.op && (pd->lookup.done || (pd->lookup.done = pf_socket_lookup(direction, pd, m), 1)) && !pf_match_gid(r->gid.op, r->gid.gid[0], r->gid.gid[1], pd->lookup.gid)) r = TAILQ_NEXT(r, entries); else if (r->prio && !pf_match_ieee8021q_pcp(r->prio, m)) r = TAILQ_NEXT(r, entries); else if (r->prob && r->prob <= arc4random()) r = TAILQ_NEXT(r, entries); else if (r->match_tag && !pf_match_tag(m, r, &tag, pd->pf_mtag ? pd->pf_mtag->tag : 0)) r = TAILQ_NEXT(r, entries); else if (r->os_fingerprint != PF_OSFP_ANY && (pd->proto != IPPROTO_TCP || !pf_osfp_match( pf_osfp_fingerprint(pd, m, off, th), r->os_fingerprint))) r = TAILQ_NEXT(r, entries); else { if (r->tag) tag = r->tag; if (r->rtableid >= 0) rtableid = r->rtableid; if (r->anchor == NULL) { match = 1; *rm = r; *am = a; *rsm = ruleset; if ((*rm)->quick) break; r = TAILQ_NEXT(r, entries); } else pf_step_into_anchor(anchor_stack, &asd, &ruleset, PF_RULESET_FILTER, &r, &a, &match); } if (r == NULL && pf_step_out_of_anchor(anchor_stack, &asd, &ruleset, PF_RULESET_FILTER, &r, &a, &match)) break; } r = *rm; a = *am; ruleset = *rsm; REASON_SET(&reason, PFRES_MATCH); if (r->log || (nr != NULL && nr->log)) { if (rewrite) m_copyback(m, off, hdrlen, pd->hdr.any); PFLOG_PACKET(kif, m, af, direction, reason, r->log ? r : nr, a, ruleset, pd, 1); } if ((r->action == PF_DROP) && ((r->rule_flag & PFRULE_RETURNRST) || (r->rule_flag & PFRULE_RETURNICMP) || (r->rule_flag & PFRULE_RETURN))) { /* undo NAT changes, if they have taken place */ if (nr != NULL) { PF_ACPY(saddr, &sk->addr[pd->sidx], af); PF_ACPY(daddr, &sk->addr[pd->didx], af); if (pd->sport) *pd->sport = sk->port[pd->sidx]; if (pd->dport) *pd->dport = sk->port[pd->didx]; if (pd->proto_sum) *pd->proto_sum = bproto_sum; if (pd->ip_sum) *pd->ip_sum = bip_sum; m_copyback(m, off, hdrlen, pd->hdr.any); } if (pd->proto == IPPROTO_TCP && ((r->rule_flag & PFRULE_RETURNRST) || (r->rule_flag & PFRULE_RETURN)) && !(th->th_flags & TH_RST)) { u_int32_t ack = ntohl(th->th_seq) + pd->p_len; int len = 0; #ifdef INET struct ip *h4; #endif #ifdef INET6 struct ip6_hdr *h6; #endif switch (af) { #ifdef INET case AF_INET: h4 = mtod(m, struct ip *); len = ntohs(h4->ip_len) - off; break; #endif #ifdef INET6 case AF_INET6: h6 = mtod(m, struct ip6_hdr *); len = ntohs(h6->ip6_plen) - (off - sizeof(*h6)); break; #endif } if (pf_check_proto_cksum(m, off, len, IPPROTO_TCP, af)) REASON_SET(&reason, PFRES_PROTCKSUM); else { if (th->th_flags & TH_SYN) ack++; if (th->th_flags & TH_FIN) ack++; pf_send_tcp(m, r, af, pd->dst, pd->src, th->th_dport, th->th_sport, ntohl(th->th_ack), ack, TH_RST|TH_ACK, 0, 0, r->return_ttl, 1, 0, kif->pfik_ifp); } } else if (pd->proto != IPPROTO_ICMP && af == AF_INET && r->return_icmp) pf_send_icmp(m, r->return_icmp >> 8, r->return_icmp & 255, af, r); else if (pd->proto != IPPROTO_ICMPV6 && af == AF_INET6 && r->return_icmp6) pf_send_icmp(m, r->return_icmp6 >> 8, r->return_icmp6 & 255, af, r); } if (r->action == PF_DROP) goto cleanup; if (tag > 0 && pf_tag_packet(m, pd, tag)) { REASON_SET(&reason, PFRES_MEMORY); goto cleanup; } if (rtableid >= 0) M_SETFIB(m, rtableid); if (!state_icmp && (r->keep_state || nr != NULL || (pd->flags & PFDESC_TCP_NORM))) { int action; action = pf_create_state(r, nr, a, pd, nsn, nk, sk, m, off, sport, dport, &rewrite, kif, sm, tag, bproto_sum, bip_sum, hdrlen); if (action != PF_PASS) return (action); } else { if (sk != NULL) uma_zfree(V_pf_state_key_z, sk); if (nk != NULL) uma_zfree(V_pf_state_key_z, nk); } /* copy back packet headers if we performed NAT operations */ if (rewrite) m_copyback(m, off, hdrlen, pd->hdr.any); if (*sm != NULL && !((*sm)->state_flags & PFSTATE_NOSYNC) && direction == PF_OUT && pfsync_defer_ptr != NULL && pfsync_defer_ptr(*sm, m)) /* * We want the state created, but we dont * want to send this in case a partner * firewall has to know about it to allow * replies through it. */ return (PF_DEFER); return (PF_PASS); cleanup: if (sk != NULL) uma_zfree(V_pf_state_key_z, sk); if (nk != NULL) uma_zfree(V_pf_state_key_z, nk); return (PF_DROP); } static int pf_create_state(struct pf_rule *r, struct pf_rule *nr, struct pf_rule *a, struct pf_pdesc *pd, struct pf_src_node *nsn, struct pf_state_key *nk, struct pf_state_key *sk, struct mbuf *m, int off, u_int16_t sport, u_int16_t dport, int *rewrite, struct pfi_kif *kif, struct pf_state **sm, int tag, u_int16_t bproto_sum, u_int16_t bip_sum, int hdrlen) { struct pf_state *s = NULL; struct pf_src_node *sn = NULL; struct tcphdr *th = pd->hdr.tcp; u_int16_t mss = V_tcp_mssdflt; u_short reason; /* check maximums */ if (r->max_states && (counter_u64_fetch(r->states_cur) >= r->max_states)) { counter_u64_add(V_pf_status.lcounters[LCNT_STATES], 1); REASON_SET(&reason, PFRES_MAXSTATES); return (PF_DROP); } /* src node for filter rule */ if ((r->rule_flag & PFRULE_SRCTRACK || r->rpool.opts & PF_POOL_STICKYADDR) && pf_insert_src_node(&sn, r, pd->src, pd->af) != 0) { REASON_SET(&reason, PFRES_SRCLIMIT); goto csfailed; } /* src node for translation rule */ if (nr != NULL && (nr->rpool.opts & PF_POOL_STICKYADDR) && pf_insert_src_node(&nsn, nr, &sk->addr[pd->sidx], pd->af)) { REASON_SET(&reason, PFRES_SRCLIMIT); goto csfailed; } s = uma_zalloc(V_pf_state_z, M_NOWAIT | M_ZERO); if (s == NULL) { REASON_SET(&reason, PFRES_MEMORY); goto csfailed; } s->rule.ptr = r; s->nat_rule.ptr = nr; s->anchor.ptr = a; STATE_INC_COUNTERS(s); if (r->allow_opts) s->state_flags |= PFSTATE_ALLOWOPTS; if (r->rule_flag & PFRULE_STATESLOPPY) s->state_flags |= PFSTATE_SLOPPY; s->log = r->log & PF_LOG_ALL; s->sync_state = PFSYNC_S_NONE; if (nr != NULL) s->log |= nr->log & PF_LOG_ALL; switch (pd->proto) { case IPPROTO_TCP: s->src.seqlo = ntohl(th->th_seq); s->src.seqhi = s->src.seqlo + pd->p_len + 1; if ((th->th_flags & (TH_SYN|TH_ACK)) == TH_SYN && r->keep_state == PF_STATE_MODULATE) { /* Generate sequence number modulator */ if ((s->src.seqdiff = pf_tcp_iss(pd) - s->src.seqlo) == 0) s->src.seqdiff = 1; pf_change_proto_a(m, &th->th_seq, &th->th_sum, htonl(s->src.seqlo + s->src.seqdiff), 0); *rewrite = 1; } else s->src.seqdiff = 0; if (th->th_flags & TH_SYN) { s->src.seqhi++; s->src.wscale = pf_get_wscale(m, off, th->th_off, pd->af); } s->src.max_win = MAX(ntohs(th->th_win), 1); if (s->src.wscale & PF_WSCALE_MASK) { /* Remove scale factor from initial window */ int win = s->src.max_win; win += 1 << (s->src.wscale & PF_WSCALE_MASK); s->src.max_win = (win - 1) >> (s->src.wscale & PF_WSCALE_MASK); } if (th->th_flags & TH_FIN) s->src.seqhi++; s->dst.seqhi = 1; s->dst.max_win = 1; s->src.state = TCPS_SYN_SENT; s->dst.state = TCPS_CLOSED; s->timeout = PFTM_TCP_FIRST_PACKET; break; case IPPROTO_UDP: s->src.state = PFUDPS_SINGLE; s->dst.state = PFUDPS_NO_TRAFFIC; s->timeout = PFTM_UDP_FIRST_PACKET; break; case IPPROTO_ICMP: #ifdef INET6 case IPPROTO_ICMPV6: #endif s->timeout = PFTM_ICMP_FIRST_PACKET; break; default: s->src.state = PFOTHERS_SINGLE; s->dst.state = PFOTHERS_NO_TRAFFIC; s->timeout = PFTM_OTHER_FIRST_PACKET; } if (r->rt && r->rt != PF_FASTROUTE) { if (pf_map_addr(pd->af, r, pd->src, &s->rt_addr, NULL, &sn)) { REASON_SET(&reason, PFRES_MAPFAILED); pf_src_tree_remove_state(s); STATE_DEC_COUNTERS(s); uma_zfree(V_pf_state_z, s); goto csfailed; } s->rt_kif = r->rpool.cur->kif; } s->creation = time_uptime; s->expire = time_uptime; if (sn != NULL) s->src_node = sn; if (nsn != NULL) { /* XXX We only modify one side for now. */ PF_ACPY(&nsn->raddr, &nk->addr[1], pd->af); s->nat_src_node = nsn; } if (pd->proto == IPPROTO_TCP) { if ((pd->flags & PFDESC_TCP_NORM) && pf_normalize_tcp_init(m, off, pd, th, &s->src, &s->dst)) { REASON_SET(&reason, PFRES_MEMORY); pf_src_tree_remove_state(s); STATE_DEC_COUNTERS(s); uma_zfree(V_pf_state_z, s); return (PF_DROP); } if ((pd->flags & PFDESC_TCP_NORM) && s->src.scrub && pf_normalize_tcp_stateful(m, off, pd, &reason, th, s, &s->src, &s->dst, rewrite)) { /* This really shouldn't happen!!! */ DPFPRINTF(PF_DEBUG_URGENT, ("pf_normalize_tcp_stateful failed on first pkt")); pf_normalize_tcp_cleanup(s); pf_src_tree_remove_state(s); STATE_DEC_COUNTERS(s); uma_zfree(V_pf_state_z, s); return (PF_DROP); } } s->direction = pd->dir; /* * sk/nk could already been setup by pf_get_translation(). */ if (nr == NULL) { KASSERT((sk == NULL && nk == NULL), ("%s: nr %p sk %p, nk %p", __func__, nr, sk, nk)); sk = pf_state_key_setup(pd, pd->src, pd->dst, sport, dport); if (sk == NULL) goto csfailed; nk = sk; } else KASSERT((sk != NULL && nk != NULL), ("%s: nr %p sk %p, nk %p", __func__, nr, sk, nk)); /* Swap sk/nk for PF_OUT. */ if (pf_state_insert(BOUND_IFACE(r, kif), (pd->dir == PF_IN) ? sk : nk, (pd->dir == PF_IN) ? nk : sk, s)) { if (pd->proto == IPPROTO_TCP) pf_normalize_tcp_cleanup(s); REASON_SET(&reason, PFRES_STATEINS); pf_src_tree_remove_state(s); STATE_DEC_COUNTERS(s); uma_zfree(V_pf_state_z, s); return (PF_DROP); } else *sm = s; if (tag > 0) s->tag = tag; if (pd->proto == IPPROTO_TCP && (th->th_flags & (TH_SYN|TH_ACK)) == TH_SYN && r->keep_state == PF_STATE_SYNPROXY) { s->src.state = PF_TCPS_PROXY_SRC; /* undo NAT changes, if they have taken place */ if (nr != NULL) { struct pf_state_key *skt = s->key[PF_SK_WIRE]; if (pd->dir == PF_OUT) skt = s->key[PF_SK_STACK]; PF_ACPY(pd->src, &skt->addr[pd->sidx], pd->af); PF_ACPY(pd->dst, &skt->addr[pd->didx], pd->af); if (pd->sport) *pd->sport = skt->port[pd->sidx]; if (pd->dport) *pd->dport = skt->port[pd->didx]; if (pd->proto_sum) *pd->proto_sum = bproto_sum; if (pd->ip_sum) *pd->ip_sum = bip_sum; m_copyback(m, off, hdrlen, pd->hdr.any); } s->src.seqhi = htonl(arc4random()); /* Find mss option */ int rtid = M_GETFIB(m); mss = pf_get_mss(m, off, th->th_off, pd->af); mss = pf_calc_mss(pd->src, pd->af, rtid, mss); mss = pf_calc_mss(pd->dst, pd->af, rtid, mss); s->src.mss = mss; pf_send_tcp(NULL, r, pd->af, pd->dst, pd->src, th->th_dport, th->th_sport, s->src.seqhi, ntohl(th->th_seq) + 1, TH_SYN|TH_ACK, 0, s->src.mss, 0, 1, 0, NULL); REASON_SET(&reason, PFRES_SYNPROXY); return (PF_SYNPROXY_DROP); } return (PF_PASS); csfailed: if (sk != NULL) uma_zfree(V_pf_state_key_z, sk); if (nk != NULL) uma_zfree(V_pf_state_key_z, nk); if (sn != NULL) { struct pf_srchash *sh; sh = &V_pf_srchash[pf_hashsrc(&sn->addr, sn->af)]; PF_HASHROW_LOCK(sh); if (--sn->states == 0 && sn->expire == 0) { pf_unlink_src_node(sn); uma_zfree(V_pf_sources_z, sn); counter_u64_add( V_pf_status.scounters[SCNT_SRC_NODE_REMOVALS], 1); } PF_HASHROW_UNLOCK(sh); } if (nsn != sn && nsn != NULL) { struct pf_srchash *sh; sh = &V_pf_srchash[pf_hashsrc(&nsn->addr, nsn->af)]; PF_HASHROW_LOCK(sh); if (--nsn->states == 0 && nsn->expire == 0) { pf_unlink_src_node(nsn); uma_zfree(V_pf_sources_z, nsn); counter_u64_add( V_pf_status.scounters[SCNT_SRC_NODE_REMOVALS], 1); } PF_HASHROW_UNLOCK(sh); } return (PF_DROP); } static int pf_test_fragment(struct pf_rule **rm, int direction, struct pfi_kif *kif, struct mbuf *m, void *h, struct pf_pdesc *pd, struct pf_rule **am, struct pf_ruleset **rsm) { struct pf_rule *r, *a = NULL; struct pf_ruleset *ruleset = NULL; sa_family_t af = pd->af; u_short reason; int tag = -1; int asd = 0; int match = 0; struct pf_anchor_stackframe anchor_stack[PF_ANCHOR_STACKSIZE]; PF_RULES_RASSERT(); r = TAILQ_FIRST(pf_main_ruleset.rules[PF_RULESET_FILTER].active.ptr); while (r != NULL) { r->evaluations++; if (pfi_kif_match(r->kif, kif) == r->ifnot) r = r->skip[PF_SKIP_IFP].ptr; else if (r->direction && r->direction != direction) r = r->skip[PF_SKIP_DIR].ptr; else if (r->af && r->af != af) r = r->skip[PF_SKIP_AF].ptr; else if (r->proto && r->proto != pd->proto) r = r->skip[PF_SKIP_PROTO].ptr; else if (PF_MISMATCHAW(&r->src.addr, pd->src, af, r->src.neg, kif, M_GETFIB(m))) r = r->skip[PF_SKIP_SRC_ADDR].ptr; else if (PF_MISMATCHAW(&r->dst.addr, pd->dst, af, r->dst.neg, NULL, M_GETFIB(m))) r = r->skip[PF_SKIP_DST_ADDR].ptr; else if (r->tos && !(r->tos == pd->tos)) r = TAILQ_NEXT(r, entries); else if (r->os_fingerprint != PF_OSFP_ANY) r = TAILQ_NEXT(r, entries); else if (pd->proto == IPPROTO_UDP && (r->src.port_op || r->dst.port_op)) r = TAILQ_NEXT(r, entries); else if (pd->proto == IPPROTO_TCP && (r->src.port_op || r->dst.port_op || r->flagset)) r = TAILQ_NEXT(r, entries); else if ((pd->proto == IPPROTO_ICMP || pd->proto == IPPROTO_ICMPV6) && (r->type || r->code)) r = TAILQ_NEXT(r, entries); else if (r->prio && !pf_match_ieee8021q_pcp(r->prio, m)) r = TAILQ_NEXT(r, entries); else if (r->prob && r->prob <= (arc4random() % (UINT_MAX - 1) + 1)) r = TAILQ_NEXT(r, entries); else if (r->match_tag && !pf_match_tag(m, r, &tag, pd->pf_mtag ? pd->pf_mtag->tag : 0)) r = TAILQ_NEXT(r, entries); else { if (r->anchor == NULL) { match = 1; *rm = r; *am = a; *rsm = ruleset; if ((*rm)->quick) break; r = TAILQ_NEXT(r, entries); } else pf_step_into_anchor(anchor_stack, &asd, &ruleset, PF_RULESET_FILTER, &r, &a, &match); } if (r == NULL && pf_step_out_of_anchor(anchor_stack, &asd, &ruleset, PF_RULESET_FILTER, &r, &a, &match)) break; } r = *rm; a = *am; ruleset = *rsm; REASON_SET(&reason, PFRES_MATCH); if (r->log) PFLOG_PACKET(kif, m, af, direction, reason, r, a, ruleset, pd, 1); if (r->action != PF_PASS) return (PF_DROP); if (tag > 0 && pf_tag_packet(m, pd, tag)) { REASON_SET(&reason, PFRES_MEMORY); return (PF_DROP); } return (PF_PASS); } static int pf_tcp_track_full(struct pf_state_peer *src, struct pf_state_peer *dst, struct pf_state **state, struct pfi_kif *kif, struct mbuf *m, int off, struct pf_pdesc *pd, u_short *reason, int *copyback) { struct tcphdr *th = pd->hdr.tcp; u_int16_t win = ntohs(th->th_win); u_int32_t ack, end, seq, orig_seq; u_int8_t sws, dws; int ackskew; if (src->wscale && dst->wscale && !(th->th_flags & TH_SYN)) { sws = src->wscale & PF_WSCALE_MASK; dws = dst->wscale & PF_WSCALE_MASK; } else sws = dws = 0; /* * Sequence tracking algorithm from Guido van Rooij's paper: * http://www.madison-gurkha.com/publications/tcp_filtering/ * tcp_filtering.ps */ orig_seq = seq = ntohl(th->th_seq); if (src->seqlo == 0) { /* First packet from this end. Set its state */ if ((pd->flags & PFDESC_TCP_NORM || dst->scrub) && src->scrub == NULL) { if (pf_normalize_tcp_init(m, off, pd, th, src, dst)) { REASON_SET(reason, PFRES_MEMORY); return (PF_DROP); } } /* Deferred generation of sequence number modulator */ if (dst->seqdiff && !src->seqdiff) { /* use random iss for the TCP server */ while ((src->seqdiff = arc4random() - seq) == 0) ; ack = ntohl(th->th_ack) - dst->seqdiff; pf_change_proto_a(m, &th->th_seq, &th->th_sum, htonl(seq + src->seqdiff), 0); pf_change_proto_a(m, &th->th_ack, &th->th_sum, htonl(ack), 0); *copyback = 1; } else { ack = ntohl(th->th_ack); } end = seq + pd->p_len; if (th->th_flags & TH_SYN) { end++; if (dst->wscale & PF_WSCALE_FLAG) { src->wscale = pf_get_wscale(m, off, th->th_off, pd->af); if (src->wscale & PF_WSCALE_FLAG) { /* Remove scale factor from initial * window */ sws = src->wscale & PF_WSCALE_MASK; win = ((u_int32_t)win + (1 << sws) - 1) >> sws; dws = dst->wscale & PF_WSCALE_MASK; } else { /* fixup other window */ dst->max_win <<= dst->wscale & PF_WSCALE_MASK; /* in case of a retrans SYN|ACK */ dst->wscale = 0; } } } if (th->th_flags & TH_FIN) end++; src->seqlo = seq; if (src->state < TCPS_SYN_SENT) src->state = TCPS_SYN_SENT; /* * May need to slide the window (seqhi may have been set by * the crappy stack check or if we picked up the connection * after establishment) */ if (src->seqhi == 1 || SEQ_GEQ(end + MAX(1, dst->max_win << dws), src->seqhi)) src->seqhi = end + MAX(1, dst->max_win << dws); if (win > src->max_win) src->max_win = win; } else { ack = ntohl(th->th_ack) - dst->seqdiff; if (src->seqdiff) { /* Modulate sequence numbers */ pf_change_proto_a(m, &th->th_seq, &th->th_sum, htonl(seq + src->seqdiff), 0); pf_change_proto_a(m, &th->th_ack, &th->th_sum, htonl(ack), 0); *copyback = 1; } end = seq + pd->p_len; if (th->th_flags & TH_SYN) end++; if (th->th_flags & TH_FIN) end++; } if ((th->th_flags & TH_ACK) == 0) { /* Let it pass through the ack skew check */ ack = dst->seqlo; } else if ((ack == 0 && (th->th_flags & (TH_ACK|TH_RST)) == (TH_ACK|TH_RST)) || /* broken tcp stacks do not set ack */ (dst->state < TCPS_SYN_SENT)) { /* * Many stacks (ours included) will set the ACK number in an * FIN|ACK if the SYN times out -- no sequence to ACK. */ ack = dst->seqlo; } if (seq == end) { /* Ease sequencing restrictions on no data packets */ seq = src->seqlo; end = seq; } ackskew = dst->seqlo - ack; /* * Need to demodulate the sequence numbers in any TCP SACK options * (Selective ACK). We could optionally validate the SACK values * against the current ACK window, either forwards or backwards, but * I'm not confident that SACK has been implemented properly * everywhere. It wouldn't surprise me if several stacks accidentally * SACK too far backwards of previously ACKed data. There really aren't * any security implications of bad SACKing unless the target stack * doesn't validate the option length correctly. Someone trying to * spoof into a TCP connection won't bother blindly sending SACK * options anyway. */ if (dst->seqdiff && (th->th_off << 2) > sizeof(struct tcphdr)) { if (pf_modulate_sack(m, off, pd, th, dst)) *copyback = 1; } #define MAXACKWINDOW (0xffff + 1500) /* 1500 is an arbitrary fudge factor */ if (SEQ_GEQ(src->seqhi, end) && /* Last octet inside other's window space */ SEQ_GEQ(seq, src->seqlo - (dst->max_win << dws)) && /* Retrans: not more than one window back */ (ackskew >= -MAXACKWINDOW) && /* Acking not more than one reassembled fragment backwards */ (ackskew <= (MAXACKWINDOW << sws)) && /* Acking not more than one window forward */ ((th->th_flags & TH_RST) == 0 || orig_seq == src->seqlo || (orig_seq == src->seqlo + 1) || (orig_seq + 1 == src->seqlo) || (pd->flags & PFDESC_IP_REAS) == 0)) { /* Require an exact/+1 sequence match on resets when possible */ if (dst->scrub || src->scrub) { if (pf_normalize_tcp_stateful(m, off, pd, reason, th, *state, src, dst, copyback)) return (PF_DROP); } /* update max window */ if (src->max_win < win) src->max_win = win; /* synchronize sequencing */ if (SEQ_GT(end, src->seqlo)) src->seqlo = end; /* slide the window of what the other end can send */ if (SEQ_GEQ(ack + (win << sws), dst->seqhi)) dst->seqhi = ack + MAX((win << sws), 1); /* update states */ if (th->th_flags & TH_SYN) if (src->state < TCPS_SYN_SENT) src->state = TCPS_SYN_SENT; if (th->th_flags & TH_FIN) if (src->state < TCPS_CLOSING) src->state = TCPS_CLOSING; if (th->th_flags & TH_ACK) { if (dst->state == TCPS_SYN_SENT) { dst->state = TCPS_ESTABLISHED; if (src->state == TCPS_ESTABLISHED && (*state)->src_node != NULL && pf_src_connlimit(state)) { REASON_SET(reason, PFRES_SRCLIMIT); return (PF_DROP); } } else if (dst->state == TCPS_CLOSING) dst->state = TCPS_FIN_WAIT_2; } if (th->th_flags & TH_RST) src->state = dst->state = TCPS_TIME_WAIT; /* update expire time */ (*state)->expire = time_uptime; if (src->state >= TCPS_FIN_WAIT_2 && dst->state >= TCPS_FIN_WAIT_2) (*state)->timeout = PFTM_TCP_CLOSED; else if (src->state >= TCPS_CLOSING && dst->state >= TCPS_CLOSING) (*state)->timeout = PFTM_TCP_FIN_WAIT; else if (src->state < TCPS_ESTABLISHED || dst->state < TCPS_ESTABLISHED) (*state)->timeout = PFTM_TCP_OPENING; else if (src->state >= TCPS_CLOSING || dst->state >= TCPS_CLOSING) (*state)->timeout = PFTM_TCP_CLOSING; else (*state)->timeout = PFTM_TCP_ESTABLISHED; /* Fall through to PASS packet */ } else if ((dst->state < TCPS_SYN_SENT || dst->state >= TCPS_FIN_WAIT_2 || src->state >= TCPS_FIN_WAIT_2) && SEQ_GEQ(src->seqhi + MAXACKWINDOW, end) && /* Within a window forward of the originating packet */ SEQ_GEQ(seq, src->seqlo - MAXACKWINDOW)) { /* Within a window backward of the originating packet */ /* * This currently handles three situations: * 1) Stupid stacks will shotgun SYNs before their peer * replies. * 2) When PF catches an already established stream (the * firewall rebooted, the state table was flushed, routes * changed...) * 3) Packets get funky immediately after the connection * closes (this should catch Solaris spurious ACK|FINs * that web servers like to spew after a close) * * This must be a little more careful than the above code * since packet floods will also be caught here. We don't * update the TTL here to mitigate the damage of a packet * flood and so the same code can handle awkward establishment * and a loosened connection close. * In the establishment case, a correct peer response will * validate the connection, go through the normal state code * and keep updating the state TTL. */ if (V_pf_status.debug >= PF_DEBUG_MISC) { printf("pf: loose state match: "); pf_print_state(*state); pf_print_flags(th->th_flags); printf(" seq=%u (%u) ack=%u len=%u ackskew=%d " "pkts=%llu:%llu dir=%s,%s\n", seq, orig_seq, ack, pd->p_len, ackskew, (unsigned long long)(*state)->packets[0], (unsigned long long)(*state)->packets[1], pd->dir == PF_IN ? "in" : "out", pd->dir == (*state)->direction ? "fwd" : "rev"); } if (dst->scrub || src->scrub) { if (pf_normalize_tcp_stateful(m, off, pd, reason, th, *state, src, dst, copyback)) return (PF_DROP); } /* update max window */ if (src->max_win < win) src->max_win = win; /* synchronize sequencing */ if (SEQ_GT(end, src->seqlo)) src->seqlo = end; /* slide the window of what the other end can send */ if (SEQ_GEQ(ack + (win << sws), dst->seqhi)) dst->seqhi = ack + MAX((win << sws), 1); /* * Cannot set dst->seqhi here since this could be a shotgunned * SYN and not an already established connection. */ if (th->th_flags & TH_FIN) if (src->state < TCPS_CLOSING) src->state = TCPS_CLOSING; if (th->th_flags & TH_RST) src->state = dst->state = TCPS_TIME_WAIT; /* Fall through to PASS packet */ } else { if ((*state)->dst.state == TCPS_SYN_SENT && (*state)->src.state == TCPS_SYN_SENT) { /* Send RST for state mismatches during handshake */ if (!(th->th_flags & TH_RST)) pf_send_tcp(NULL, (*state)->rule.ptr, pd->af, pd->dst, pd->src, th->th_dport, th->th_sport, ntohl(th->th_ack), 0, TH_RST, 0, 0, (*state)->rule.ptr->return_ttl, 1, 0, kif->pfik_ifp); src->seqlo = 0; src->seqhi = 1; src->max_win = 1; } else if (V_pf_status.debug >= PF_DEBUG_MISC) { printf("pf: BAD state: "); pf_print_state(*state); pf_print_flags(th->th_flags); printf(" seq=%u (%u) ack=%u len=%u ackskew=%d " "pkts=%llu:%llu dir=%s,%s\n", seq, orig_seq, ack, pd->p_len, ackskew, (unsigned long long)(*state)->packets[0], (unsigned long long)(*state)->packets[1], pd->dir == PF_IN ? "in" : "out", pd->dir == (*state)->direction ? "fwd" : "rev"); printf("pf: State failure on: %c %c %c %c | %c %c\n", SEQ_GEQ(src->seqhi, end) ? ' ' : '1', SEQ_GEQ(seq, src->seqlo - (dst->max_win << dws)) ? ' ': '2', (ackskew >= -MAXACKWINDOW) ? ' ' : '3', (ackskew <= (MAXACKWINDOW << sws)) ? ' ' : '4', SEQ_GEQ(src->seqhi + MAXACKWINDOW, end) ?' ' :'5', SEQ_GEQ(seq, src->seqlo - MAXACKWINDOW) ?' ' :'6'); } REASON_SET(reason, PFRES_BADSTATE); return (PF_DROP); } return (PF_PASS); } static int pf_tcp_track_sloppy(struct pf_state_peer *src, struct pf_state_peer *dst, struct pf_state **state, struct pf_pdesc *pd, u_short *reason) { struct tcphdr *th = pd->hdr.tcp; if (th->th_flags & TH_SYN) if (src->state < TCPS_SYN_SENT) src->state = TCPS_SYN_SENT; if (th->th_flags & TH_FIN) if (src->state < TCPS_CLOSING) src->state = TCPS_CLOSING; if (th->th_flags & TH_ACK) { if (dst->state == TCPS_SYN_SENT) { dst->state = TCPS_ESTABLISHED; if (src->state == TCPS_ESTABLISHED && (*state)->src_node != NULL && pf_src_connlimit(state)) { REASON_SET(reason, PFRES_SRCLIMIT); return (PF_DROP); } } else if (dst->state == TCPS_CLOSING) { dst->state = TCPS_FIN_WAIT_2; } else if (src->state == TCPS_SYN_SENT && dst->state < TCPS_SYN_SENT) { /* * Handle a special sloppy case where we only see one * half of the connection. If there is a ACK after * the initial SYN without ever seeing a packet from * the destination, set the connection to established. */ dst->state = src->state = TCPS_ESTABLISHED; if ((*state)->src_node != NULL && pf_src_connlimit(state)) { REASON_SET(reason, PFRES_SRCLIMIT); return (PF_DROP); } } else if (src->state == TCPS_CLOSING && dst->state == TCPS_ESTABLISHED && dst->seqlo == 0) { /* * Handle the closing of half connections where we * don't see the full bidirectional FIN/ACK+ACK * handshake. */ dst->state = TCPS_CLOSING; } } if (th->th_flags & TH_RST) src->state = dst->state = TCPS_TIME_WAIT; /* update expire time */ (*state)->expire = time_uptime; if (src->state >= TCPS_FIN_WAIT_2 && dst->state >= TCPS_FIN_WAIT_2) (*state)->timeout = PFTM_TCP_CLOSED; else if (src->state >= TCPS_CLOSING && dst->state >= TCPS_CLOSING) (*state)->timeout = PFTM_TCP_FIN_WAIT; else if (src->state < TCPS_ESTABLISHED || dst->state < TCPS_ESTABLISHED) (*state)->timeout = PFTM_TCP_OPENING; else if (src->state >= TCPS_CLOSING || dst->state >= TCPS_CLOSING) (*state)->timeout = PFTM_TCP_CLOSING; else (*state)->timeout = PFTM_TCP_ESTABLISHED; return (PF_PASS); } static int pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, struct mbuf *m, int off, void *h, struct pf_pdesc *pd, u_short *reason) { struct pf_state_key_cmp key; struct tcphdr *th = pd->hdr.tcp; int copyback = 0; struct pf_state_peer *src, *dst; struct pf_state_key *sk; bzero(&key, sizeof(key)); key.af = pd->af; key.proto = IPPROTO_TCP; if (direction == PF_IN) { /* wire side, straight */ PF_ACPY(&key.addr[0], pd->src, key.af); PF_ACPY(&key.addr[1], pd->dst, key.af); key.port[0] = th->th_sport; key.port[1] = th->th_dport; } else { /* stack side, reverse */ PF_ACPY(&key.addr[1], pd->src, key.af); PF_ACPY(&key.addr[0], pd->dst, key.af); key.port[1] = th->th_sport; key.port[0] = th->th_dport; } STATE_LOOKUP(kif, &key, direction, *state, pd); if (direction == (*state)->direction) { src = &(*state)->src; dst = &(*state)->dst; } else { src = &(*state)->dst; dst = &(*state)->src; } sk = (*state)->key[pd->didx]; if ((*state)->src.state == PF_TCPS_PROXY_SRC) { if (direction != (*state)->direction) { REASON_SET(reason, PFRES_SYNPROXY); return (PF_SYNPROXY_DROP); } if (th->th_flags & TH_SYN) { if (ntohl(th->th_seq) != (*state)->src.seqlo) { REASON_SET(reason, PFRES_SYNPROXY); return (PF_DROP); } pf_send_tcp(NULL, (*state)->rule.ptr, pd->af, pd->dst, pd->src, th->th_dport, th->th_sport, (*state)->src.seqhi, ntohl(th->th_seq) + 1, TH_SYN|TH_ACK, 0, (*state)->src.mss, 0, 1, 0, NULL); REASON_SET(reason, PFRES_SYNPROXY); return (PF_SYNPROXY_DROP); } else if (!(th->th_flags & TH_ACK) || (ntohl(th->th_ack) != (*state)->src.seqhi + 1) || (ntohl(th->th_seq) != (*state)->src.seqlo + 1)) { REASON_SET(reason, PFRES_SYNPROXY); return (PF_DROP); } else if ((*state)->src_node != NULL && pf_src_connlimit(state)) { REASON_SET(reason, PFRES_SRCLIMIT); return (PF_DROP); } else (*state)->src.state = PF_TCPS_PROXY_DST; } if ((*state)->src.state == PF_TCPS_PROXY_DST) { if (direction == (*state)->direction) { if (((th->th_flags & (TH_SYN|TH_ACK)) != TH_ACK) || (ntohl(th->th_ack) != (*state)->src.seqhi + 1) || (ntohl(th->th_seq) != (*state)->src.seqlo + 1)) { REASON_SET(reason, PFRES_SYNPROXY); return (PF_DROP); } (*state)->src.max_win = MAX(ntohs(th->th_win), 1); if ((*state)->dst.seqhi == 1) (*state)->dst.seqhi = htonl(arc4random()); pf_send_tcp(NULL, (*state)->rule.ptr, pd->af, &sk->addr[pd->sidx], &sk->addr[pd->didx], sk->port[pd->sidx], sk->port[pd->didx], (*state)->dst.seqhi, 0, TH_SYN, 0, (*state)->src.mss, 0, 0, (*state)->tag, NULL); REASON_SET(reason, PFRES_SYNPROXY); return (PF_SYNPROXY_DROP); } else if (((th->th_flags & (TH_SYN|TH_ACK)) != (TH_SYN|TH_ACK)) || (ntohl(th->th_ack) != (*state)->dst.seqhi + 1)) { REASON_SET(reason, PFRES_SYNPROXY); return (PF_DROP); } else { (*state)->dst.max_win = MAX(ntohs(th->th_win), 1); (*state)->dst.seqlo = ntohl(th->th_seq); pf_send_tcp(NULL, (*state)->rule.ptr, pd->af, pd->dst, pd->src, th->th_dport, th->th_sport, ntohl(th->th_ack), ntohl(th->th_seq) + 1, TH_ACK, (*state)->src.max_win, 0, 0, 0, (*state)->tag, NULL); pf_send_tcp(NULL, (*state)->rule.ptr, pd->af, &sk->addr[pd->sidx], &sk->addr[pd->didx], sk->port[pd->sidx], sk->port[pd->didx], (*state)->src.seqhi + 1, (*state)->src.seqlo + 1, TH_ACK, (*state)->dst.max_win, 0, 0, 1, 0, NULL); (*state)->src.seqdiff = (*state)->dst.seqhi - (*state)->src.seqlo; (*state)->dst.seqdiff = (*state)->src.seqhi - (*state)->dst.seqlo; (*state)->src.seqhi = (*state)->src.seqlo + (*state)->dst.max_win; (*state)->dst.seqhi = (*state)->dst.seqlo + (*state)->src.max_win; (*state)->src.wscale = (*state)->dst.wscale = 0; (*state)->src.state = (*state)->dst.state = TCPS_ESTABLISHED; REASON_SET(reason, PFRES_SYNPROXY); return (PF_SYNPROXY_DROP); } } if (((th->th_flags & (TH_SYN|TH_ACK)) == TH_SYN) && dst->state >= TCPS_FIN_WAIT_2 && src->state >= TCPS_FIN_WAIT_2) { if (V_pf_status.debug >= PF_DEBUG_MISC) { printf("pf: state reuse "); pf_print_state(*state); pf_print_flags(th->th_flags); printf("\n"); } /* XXX make sure it's the same direction ?? */ (*state)->src.state = (*state)->dst.state = TCPS_CLOSED; pf_unlink_state(*state, PF_ENTER_LOCKED); *state = NULL; return (PF_DROP); } if ((*state)->state_flags & PFSTATE_SLOPPY) { if (pf_tcp_track_sloppy(src, dst, state, pd, reason) == PF_DROP) return (PF_DROP); } else { if (pf_tcp_track_full(src, dst, state, kif, m, off, pd, reason, ©back) == PF_DROP) return (PF_DROP); } /* translate source/destination address, if necessary */ if ((*state)->key[PF_SK_WIRE] != (*state)->key[PF_SK_STACK]) { struct pf_state_key *nk = (*state)->key[pd->didx]; if (PF_ANEQ(pd->src, &nk->addr[pd->sidx], pd->af) || nk->port[pd->sidx] != th->th_sport) pf_change_ap(m, pd->src, &th->th_sport, pd->ip_sum, &th->th_sum, &nk->addr[pd->sidx], nk->port[pd->sidx], 0, pd->af); if (PF_ANEQ(pd->dst, &nk->addr[pd->didx], pd->af) || nk->port[pd->didx] != th->th_dport) pf_change_ap(m, pd->dst, &th->th_dport, pd->ip_sum, &th->th_sum, &nk->addr[pd->didx], nk->port[pd->didx], 0, pd->af); copyback = 1; } /* Copyback sequence modulation or stateful scrub changes if needed */ if (copyback) m_copyback(m, off, sizeof(*th), (caddr_t)th); return (PF_PASS); } static int pf_test_state_udp(struct pf_state **state, int direction, struct pfi_kif *kif, struct mbuf *m, int off, void *h, struct pf_pdesc *pd) { struct pf_state_peer *src, *dst; struct pf_state_key_cmp key; struct udphdr *uh = pd->hdr.udp; bzero(&key, sizeof(key)); key.af = pd->af; key.proto = IPPROTO_UDP; if (direction == PF_IN) { /* wire side, straight */ PF_ACPY(&key.addr[0], pd->src, key.af); PF_ACPY(&key.addr[1], pd->dst, key.af); key.port[0] = uh->uh_sport; key.port[1] = uh->uh_dport; } else { /* stack side, reverse */ PF_ACPY(&key.addr[1], pd->src, key.af); PF_ACPY(&key.addr[0], pd->dst, key.af); key.port[1] = uh->uh_sport; key.port[0] = uh->uh_dport; } STATE_LOOKUP(kif, &key, direction, *state, pd); if (direction == (*state)->direction) { src = &(*state)->src; dst = &(*state)->dst; } else { src = &(*state)->dst; dst = &(*state)->src; } /* update states */ if (src->state < PFUDPS_SINGLE) src->state = PFUDPS_SINGLE; if (dst->state == PFUDPS_SINGLE) dst->state = PFUDPS_MULTIPLE; /* update expire time */ (*state)->expire = time_uptime; if (src->state == PFUDPS_MULTIPLE && dst->state == PFUDPS_MULTIPLE) (*state)->timeout = PFTM_UDP_MULTIPLE; else (*state)->timeout = PFTM_UDP_SINGLE; /* translate source/destination address, if necessary */ if ((*state)->key[PF_SK_WIRE] != (*state)->key[PF_SK_STACK]) { struct pf_state_key *nk = (*state)->key[pd->didx]; if (PF_ANEQ(pd->src, &nk->addr[pd->sidx], pd->af) || nk->port[pd->sidx] != uh->uh_sport) pf_change_ap(m, pd->src, &uh->uh_sport, pd->ip_sum, &uh->uh_sum, &nk->addr[pd->sidx], nk->port[pd->sidx], 1, pd->af); if (PF_ANEQ(pd->dst, &nk->addr[pd->didx], pd->af) || nk->port[pd->didx] != uh->uh_dport) pf_change_ap(m, pd->dst, &uh->uh_dport, pd->ip_sum, &uh->uh_sum, &nk->addr[pd->didx], nk->port[pd->didx], 1, pd->af); m_copyback(m, off, sizeof(*uh), (caddr_t)uh); } return (PF_PASS); } static int pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, struct mbuf *m, int off, void *h, struct pf_pdesc *pd, u_short *reason) { struct pf_addr *saddr = pd->src, *daddr = pd->dst; u_int16_t icmpid = 0, *icmpsum; u_int8_t icmptype; int state_icmp = 0; struct pf_state_key_cmp key; bzero(&key, sizeof(key)); switch (pd->proto) { #ifdef INET case IPPROTO_ICMP: icmptype = pd->hdr.icmp->icmp_type; icmpid = pd->hdr.icmp->icmp_id; icmpsum = &pd->hdr.icmp->icmp_cksum; if (icmptype == ICMP_UNREACH || icmptype == ICMP_SOURCEQUENCH || icmptype == ICMP_REDIRECT || icmptype == ICMP_TIMXCEED || icmptype == ICMP_PARAMPROB) state_icmp++; break; #endif /* INET */ #ifdef INET6 case IPPROTO_ICMPV6: icmptype = pd->hdr.icmp6->icmp6_type; icmpid = pd->hdr.icmp6->icmp6_id; icmpsum = &pd->hdr.icmp6->icmp6_cksum; if (icmptype == ICMP6_DST_UNREACH || icmptype == ICMP6_PACKET_TOO_BIG || icmptype == ICMP6_TIME_EXCEEDED || icmptype == ICMP6_PARAM_PROB) state_icmp++; break; #endif /* INET6 */ } if (!state_icmp) { /* * ICMP query/reply message not related to a TCP/UDP packet. * Search for an ICMP state. */ key.af = pd->af; key.proto = pd->proto; key.port[0] = key.port[1] = icmpid; if (direction == PF_IN) { /* wire side, straight */ PF_ACPY(&key.addr[0], pd->src, key.af); PF_ACPY(&key.addr[1], pd->dst, key.af); } else { /* stack side, reverse */ PF_ACPY(&key.addr[1], pd->src, key.af); PF_ACPY(&key.addr[0], pd->dst, key.af); } STATE_LOOKUP(kif, &key, direction, *state, pd); (*state)->expire = time_uptime; (*state)->timeout = PFTM_ICMP_ERROR_REPLY; /* translate source/destination address, if necessary */ if ((*state)->key[PF_SK_WIRE] != (*state)->key[PF_SK_STACK]) { struct pf_state_key *nk = (*state)->key[pd->didx]; switch (pd->af) { #ifdef INET case AF_INET: if (PF_ANEQ(pd->src, &nk->addr[pd->sidx], AF_INET)) pf_change_a(&saddr->v4.s_addr, pd->ip_sum, nk->addr[pd->sidx].v4.s_addr, 0); if (PF_ANEQ(pd->dst, &nk->addr[pd->didx], AF_INET)) pf_change_a(&daddr->v4.s_addr, pd->ip_sum, nk->addr[pd->didx].v4.s_addr, 0); if (nk->port[0] != pd->hdr.icmp->icmp_id) { pd->hdr.icmp->icmp_cksum = pf_cksum_fixup( pd->hdr.icmp->icmp_cksum, icmpid, nk->port[pd->sidx], 0); pd->hdr.icmp->icmp_id = nk->port[pd->sidx]; } m_copyback(m, off, ICMP_MINLEN, (caddr_t )pd->hdr.icmp); break; #endif /* INET */ #ifdef INET6 case AF_INET6: if (PF_ANEQ(pd->src, &nk->addr[pd->sidx], AF_INET6)) pf_change_a6(saddr, &pd->hdr.icmp6->icmp6_cksum, &nk->addr[pd->sidx], 0); if (PF_ANEQ(pd->dst, &nk->addr[pd->didx], AF_INET6)) pf_change_a6(daddr, &pd->hdr.icmp6->icmp6_cksum, &nk->addr[pd->didx], 0); m_copyback(m, off, sizeof(struct icmp6_hdr), (caddr_t )pd->hdr.icmp6); break; #endif /* INET6 */ } } return (PF_PASS); } else { /* * ICMP error message in response to a TCP/UDP packet. * Extract the inner TCP/UDP header and search for that state. */ struct pf_pdesc pd2; bzero(&pd2, sizeof pd2); #ifdef INET struct ip h2; #endif /* INET */ #ifdef INET6 struct ip6_hdr h2_6; int terminal = 0; #endif /* INET6 */ int ipoff2 = 0; int off2 = 0; pd2.af = pd->af; /* Payload packet is from the opposite direction. */ pd2.sidx = (direction == PF_IN) ? 1 : 0; pd2.didx = (direction == PF_IN) ? 0 : 1; switch (pd->af) { #ifdef INET case AF_INET: /* offset of h2 in mbuf chain */ ipoff2 = off + ICMP_MINLEN; if (!pf_pull_hdr(m, ipoff2, &h2, sizeof(h2), NULL, reason, pd2.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: ICMP error message too short " "(ip)\n")); return (PF_DROP); } /* * ICMP error messages don't refer to non-first * fragments */ if (h2.ip_off & htons(IP_OFFMASK)) { REASON_SET(reason, PFRES_FRAG); return (PF_DROP); } /* offset of protocol header that follows h2 */ off2 = ipoff2 + (h2.ip_hl << 2); pd2.proto = h2.ip_p; pd2.src = (struct pf_addr *)&h2.ip_src; pd2.dst = (struct pf_addr *)&h2.ip_dst; pd2.ip_sum = &h2.ip_sum; break; #endif /* INET */ #ifdef INET6 case AF_INET6: ipoff2 = off + sizeof(struct icmp6_hdr); if (!pf_pull_hdr(m, ipoff2, &h2_6, sizeof(h2_6), NULL, reason, pd2.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: ICMP error message too short " "(ip6)\n")); return (PF_DROP); } pd2.proto = h2_6.ip6_nxt; pd2.src = (struct pf_addr *)&h2_6.ip6_src; pd2.dst = (struct pf_addr *)&h2_6.ip6_dst; pd2.ip_sum = NULL; off2 = ipoff2 + sizeof(h2_6); do { switch (pd2.proto) { case IPPROTO_FRAGMENT: /* * ICMPv6 error messages for * non-first fragments */ REASON_SET(reason, PFRES_FRAG); return (PF_DROP); case IPPROTO_AH: case IPPROTO_HOPOPTS: case IPPROTO_ROUTING: case IPPROTO_DSTOPTS: { /* get next header and header length */ struct ip6_ext opt6; if (!pf_pull_hdr(m, off2, &opt6, sizeof(opt6), NULL, reason, pd2.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: ICMPv6 short opt\n")); return (PF_DROP); } if (pd2.proto == IPPROTO_AH) off2 += (opt6.ip6e_len + 2) * 4; else off2 += (opt6.ip6e_len + 1) * 8; pd2.proto = opt6.ip6e_nxt; /* goto the next header */ break; } default: terminal++; break; } } while (!terminal); break; #endif /* INET6 */ } switch (pd2.proto) { case IPPROTO_TCP: { struct tcphdr th; u_int32_t seq; struct pf_state_peer *src, *dst; u_int8_t dws; int copyback = 0; /* * Only the first 8 bytes of the TCP header can be * expected. Don't access any TCP header fields after * th_seq, an ackskew test is not possible. */ if (!pf_pull_hdr(m, off2, &th, 8, NULL, reason, pd2.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: ICMP error message too short " "(tcp)\n")); return (PF_DROP); } key.af = pd2.af; key.proto = IPPROTO_TCP; PF_ACPY(&key.addr[pd2.sidx], pd2.src, key.af); PF_ACPY(&key.addr[pd2.didx], pd2.dst, key.af); key.port[pd2.sidx] = th.th_sport; key.port[pd2.didx] = th.th_dport; STATE_LOOKUP(kif, &key, direction, *state, pd); if (direction == (*state)->direction) { src = &(*state)->dst; dst = &(*state)->src; } else { src = &(*state)->src; dst = &(*state)->dst; } if (src->wscale && dst->wscale) dws = dst->wscale & PF_WSCALE_MASK; else dws = 0; /* Demodulate sequence number */ seq = ntohl(th.th_seq) - src->seqdiff; if (src->seqdiff) { pf_change_a(&th.th_seq, icmpsum, htonl(seq), 0); copyback = 1; } if (!((*state)->state_flags & PFSTATE_SLOPPY) && (!SEQ_GEQ(src->seqhi, seq) || !SEQ_GEQ(seq, src->seqlo - (dst->max_win << dws)))) { if (V_pf_status.debug >= PF_DEBUG_MISC) { printf("pf: BAD ICMP %d:%d ", icmptype, pd->hdr.icmp->icmp_code); pf_print_host(pd->src, 0, pd->af); printf(" -> "); pf_print_host(pd->dst, 0, pd->af); printf(" state: "); pf_print_state(*state); printf(" seq=%u\n", seq); } REASON_SET(reason, PFRES_BADSTATE); return (PF_DROP); } else { if (V_pf_status.debug >= PF_DEBUG_MISC) { printf("pf: OK ICMP %d:%d ", icmptype, pd->hdr.icmp->icmp_code); pf_print_host(pd->src, 0, pd->af); printf(" -> "); pf_print_host(pd->dst, 0, pd->af); printf(" state: "); pf_print_state(*state); printf(" seq=%u\n", seq); } } /* translate source/destination address, if necessary */ if ((*state)->key[PF_SK_WIRE] != (*state)->key[PF_SK_STACK]) { struct pf_state_key *nk = (*state)->key[pd->didx]; if (PF_ANEQ(pd2.src, &nk->addr[pd2.sidx], pd2.af) || nk->port[pd2.sidx] != th.th_sport) pf_change_icmp(pd2.src, &th.th_sport, daddr, &nk->addr[pd2.sidx], nk->port[pd2.sidx], NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, pd2.af); if (PF_ANEQ(pd2.dst, &nk->addr[pd2.didx], pd2.af) || nk->port[pd2.didx] != th.th_dport) pf_change_icmp(pd2.dst, &th.th_dport, saddr, &nk->addr[pd2.didx], nk->port[pd2.didx], NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, pd2.af); copyback = 1; } if (copyback) { switch (pd2.af) { #ifdef INET case AF_INET: m_copyback(m, off, ICMP_MINLEN, (caddr_t )pd->hdr.icmp); m_copyback(m, ipoff2, sizeof(h2), (caddr_t )&h2); break; #endif /* INET */ #ifdef INET6 case AF_INET6: m_copyback(m, off, sizeof(struct icmp6_hdr), (caddr_t )pd->hdr.icmp6); m_copyback(m, ipoff2, sizeof(h2_6), (caddr_t )&h2_6); break; #endif /* INET6 */ } m_copyback(m, off2, 8, (caddr_t)&th); } return (PF_PASS); break; } case IPPROTO_UDP: { struct udphdr uh; if (!pf_pull_hdr(m, off2, &uh, sizeof(uh), NULL, reason, pd2.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: ICMP error message too short " "(udp)\n")); return (PF_DROP); } key.af = pd2.af; key.proto = IPPROTO_UDP; PF_ACPY(&key.addr[pd2.sidx], pd2.src, key.af); PF_ACPY(&key.addr[pd2.didx], pd2.dst, key.af); key.port[pd2.sidx] = uh.uh_sport; key.port[pd2.didx] = uh.uh_dport; STATE_LOOKUP(kif, &key, direction, *state, pd); /* translate source/destination address, if necessary */ if ((*state)->key[PF_SK_WIRE] != (*state)->key[PF_SK_STACK]) { struct pf_state_key *nk = (*state)->key[pd->didx]; if (PF_ANEQ(pd2.src, &nk->addr[pd2.sidx], pd2.af) || nk->port[pd2.sidx] != uh.uh_sport) pf_change_icmp(pd2.src, &uh.uh_sport, daddr, &nk->addr[pd2.sidx], nk->port[pd2.sidx], &uh.uh_sum, pd2.ip_sum, icmpsum, pd->ip_sum, 1, pd2.af); if (PF_ANEQ(pd2.dst, &nk->addr[pd2.didx], pd2.af) || nk->port[pd2.didx] != uh.uh_dport) pf_change_icmp(pd2.dst, &uh.uh_dport, saddr, &nk->addr[pd2.didx], nk->port[pd2.didx], &uh.uh_sum, pd2.ip_sum, icmpsum, pd->ip_sum, 1, pd2.af); switch (pd2.af) { #ifdef INET case AF_INET: m_copyback(m, off, ICMP_MINLEN, (caddr_t )pd->hdr.icmp); m_copyback(m, ipoff2, sizeof(h2), (caddr_t)&h2); break; #endif /* INET */ #ifdef INET6 case AF_INET6: m_copyback(m, off, sizeof(struct icmp6_hdr), (caddr_t )pd->hdr.icmp6); m_copyback(m, ipoff2, sizeof(h2_6), (caddr_t )&h2_6); break; #endif /* INET6 */ } m_copyback(m, off2, sizeof(uh), (caddr_t)&uh); } return (PF_PASS); break; } #ifdef INET case IPPROTO_ICMP: { struct icmp iih; if (!pf_pull_hdr(m, off2, &iih, ICMP_MINLEN, NULL, reason, pd2.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: ICMP error message too short i" "(icmp)\n")); return (PF_DROP); } key.af = pd2.af; key.proto = IPPROTO_ICMP; PF_ACPY(&key.addr[pd2.sidx], pd2.src, key.af); PF_ACPY(&key.addr[pd2.didx], pd2.dst, key.af); key.port[0] = key.port[1] = iih.icmp_id; STATE_LOOKUP(kif, &key, direction, *state, pd); /* translate source/destination address, if necessary */ if ((*state)->key[PF_SK_WIRE] != (*state)->key[PF_SK_STACK]) { struct pf_state_key *nk = (*state)->key[pd->didx]; if (PF_ANEQ(pd2.src, &nk->addr[pd2.sidx], pd2.af) || nk->port[pd2.sidx] != iih.icmp_id) pf_change_icmp(pd2.src, &iih.icmp_id, daddr, &nk->addr[pd2.sidx], nk->port[pd2.sidx], NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, AF_INET); if (PF_ANEQ(pd2.dst, &nk->addr[pd2.didx], pd2.af) || nk->port[pd2.didx] != iih.icmp_id) pf_change_icmp(pd2.dst, &iih.icmp_id, saddr, &nk->addr[pd2.didx], nk->port[pd2.didx], NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, AF_INET); m_copyback(m, off, ICMP_MINLEN, (caddr_t)pd->hdr.icmp); m_copyback(m, ipoff2, sizeof(h2), (caddr_t)&h2); m_copyback(m, off2, ICMP_MINLEN, (caddr_t)&iih); } return (PF_PASS); break; } #endif /* INET */ #ifdef INET6 case IPPROTO_ICMPV6: { struct icmp6_hdr iih; if (!pf_pull_hdr(m, off2, &iih, sizeof(struct icmp6_hdr), NULL, reason, pd2.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: ICMP error message too short " "(icmp6)\n")); return (PF_DROP); } key.af = pd2.af; key.proto = IPPROTO_ICMPV6; PF_ACPY(&key.addr[pd2.sidx], pd2.src, key.af); PF_ACPY(&key.addr[pd2.didx], pd2.dst, key.af); key.port[0] = key.port[1] = iih.icmp6_id; STATE_LOOKUP(kif, &key, direction, *state, pd); /* translate source/destination address, if necessary */ if ((*state)->key[PF_SK_WIRE] != (*state)->key[PF_SK_STACK]) { struct pf_state_key *nk = (*state)->key[pd->didx]; if (PF_ANEQ(pd2.src, &nk->addr[pd2.sidx], pd2.af) || nk->port[pd2.sidx] != iih.icmp6_id) pf_change_icmp(pd2.src, &iih.icmp6_id, daddr, &nk->addr[pd2.sidx], nk->port[pd2.sidx], NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, AF_INET6); if (PF_ANEQ(pd2.dst, &nk->addr[pd2.didx], pd2.af) || nk->port[pd2.didx] != iih.icmp6_id) pf_change_icmp(pd2.dst, &iih.icmp6_id, saddr, &nk->addr[pd2.didx], nk->port[pd2.didx], NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, AF_INET6); m_copyback(m, off, sizeof(struct icmp6_hdr), (caddr_t)pd->hdr.icmp6); m_copyback(m, ipoff2, sizeof(h2_6), (caddr_t)&h2_6); m_copyback(m, off2, sizeof(struct icmp6_hdr), (caddr_t)&iih); } return (PF_PASS); break; } #endif /* INET6 */ default: { key.af = pd2.af; key.proto = pd2.proto; PF_ACPY(&key.addr[pd2.sidx], pd2.src, key.af); PF_ACPY(&key.addr[pd2.didx], pd2.dst, key.af); key.port[0] = key.port[1] = 0; STATE_LOOKUP(kif, &key, direction, *state, pd); /* translate source/destination address, if necessary */ if ((*state)->key[PF_SK_WIRE] != (*state)->key[PF_SK_STACK]) { struct pf_state_key *nk = (*state)->key[pd->didx]; if (PF_ANEQ(pd2.src, &nk->addr[pd2.sidx], pd2.af)) pf_change_icmp(pd2.src, NULL, daddr, &nk->addr[pd2.sidx], 0, NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, pd2.af); if (PF_ANEQ(pd2.dst, &nk->addr[pd2.didx], pd2.af)) pf_change_icmp(pd2.dst, NULL, saddr, &nk->addr[pd2.didx], 0, NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, pd2.af); switch (pd2.af) { #ifdef INET case AF_INET: m_copyback(m, off, ICMP_MINLEN, (caddr_t)pd->hdr.icmp); m_copyback(m, ipoff2, sizeof(h2), (caddr_t)&h2); break; #endif /* INET */ #ifdef INET6 case AF_INET6: m_copyback(m, off, sizeof(struct icmp6_hdr), (caddr_t )pd->hdr.icmp6); m_copyback(m, ipoff2, sizeof(h2_6), (caddr_t )&h2_6); break; #endif /* INET6 */ } } return (PF_PASS); break; } } } } static int pf_test_state_other(struct pf_state **state, int direction, struct pfi_kif *kif, struct mbuf *m, struct pf_pdesc *pd) { struct pf_state_peer *src, *dst; struct pf_state_key_cmp key; bzero(&key, sizeof(key)); key.af = pd->af; key.proto = pd->proto; if (direction == PF_IN) { PF_ACPY(&key.addr[0], pd->src, key.af); PF_ACPY(&key.addr[1], pd->dst, key.af); key.port[0] = key.port[1] = 0; } else { PF_ACPY(&key.addr[1], pd->src, key.af); PF_ACPY(&key.addr[0], pd->dst, key.af); key.port[1] = key.port[0] = 0; } STATE_LOOKUP(kif, &key, direction, *state, pd); if (direction == (*state)->direction) { src = &(*state)->src; dst = &(*state)->dst; } else { src = &(*state)->dst; dst = &(*state)->src; } /* update states */ if (src->state < PFOTHERS_SINGLE) src->state = PFOTHERS_SINGLE; if (dst->state == PFOTHERS_SINGLE) dst->state = PFOTHERS_MULTIPLE; /* update expire time */ (*state)->expire = time_uptime; if (src->state == PFOTHERS_MULTIPLE && dst->state == PFOTHERS_MULTIPLE) (*state)->timeout = PFTM_OTHER_MULTIPLE; else (*state)->timeout = PFTM_OTHER_SINGLE; /* translate source/destination address, if necessary */ if ((*state)->key[PF_SK_WIRE] != (*state)->key[PF_SK_STACK]) { struct pf_state_key *nk = (*state)->key[pd->didx]; KASSERT(nk, ("%s: nk is null", __func__)); KASSERT(pd, ("%s: pd is null", __func__)); KASSERT(pd->src, ("%s: pd->src is null", __func__)); KASSERT(pd->dst, ("%s: pd->dst is null", __func__)); switch (pd->af) { #ifdef INET case AF_INET: if (PF_ANEQ(pd->src, &nk->addr[pd->sidx], AF_INET)) pf_change_a(&pd->src->v4.s_addr, pd->ip_sum, nk->addr[pd->sidx].v4.s_addr, 0); if (PF_ANEQ(pd->dst, &nk->addr[pd->didx], AF_INET)) pf_change_a(&pd->dst->v4.s_addr, pd->ip_sum, nk->addr[pd->didx].v4.s_addr, 0); break; #endif /* INET */ #ifdef INET6 case AF_INET6: if (PF_ANEQ(pd->src, &nk->addr[pd->sidx], AF_INET)) PF_ACPY(pd->src, &nk->addr[pd->sidx], pd->af); if (PF_ANEQ(pd->dst, &nk->addr[pd->didx], AF_INET)) PF_ACPY(pd->dst, &nk->addr[pd->didx], pd->af); #endif /* INET6 */ } } return (PF_PASS); } /* * ipoff and off are measured from the start of the mbuf chain. * h must be at "ipoff" on the mbuf chain. */ void * pf_pull_hdr(struct mbuf *m, int off, void *p, int len, u_short *actionp, u_short *reasonp, sa_family_t af) { switch (af) { #ifdef INET case AF_INET: { struct ip *h = mtod(m, struct ip *); u_int16_t fragoff = (ntohs(h->ip_off) & IP_OFFMASK) << 3; if (fragoff) { if (fragoff >= len) ACTION_SET(actionp, PF_PASS); else { ACTION_SET(actionp, PF_DROP); REASON_SET(reasonp, PFRES_FRAG); } return (NULL); } if (m->m_pkthdr.len < off + len || ntohs(h->ip_len) < off + len) { ACTION_SET(actionp, PF_DROP); REASON_SET(reasonp, PFRES_SHORT); return (NULL); } break; } #endif /* INET */ #ifdef INET6 case AF_INET6: { struct ip6_hdr *h = mtod(m, struct ip6_hdr *); if (m->m_pkthdr.len < off + len || (ntohs(h->ip6_plen) + sizeof(struct ip6_hdr)) < (unsigned)(off + len)) { ACTION_SET(actionp, PF_DROP); REASON_SET(reasonp, PFRES_SHORT); return (NULL); } break; } #endif /* INET6 */ } m_copydata(m, off, len, p); return (p); } #ifdef RADIX_MPATH static int pf_routable_oldmpath(struct pf_addr *addr, sa_family_t af, struct pfi_kif *kif, int rtableid) { struct radix_node_head *rnh; struct sockaddr_in *dst; int ret = 1; int check_mpath; #ifdef INET6 struct sockaddr_in6 *dst6; struct route_in6 ro; #else struct route ro; #endif struct radix_node *rn; struct rtentry *rt; struct ifnet *ifp; check_mpath = 0; /* XXX: stick to table 0 for now */ rnh = rt_tables_get_rnh(0, af); if (rnh != NULL && rn_mpath_capable(rnh)) check_mpath = 1; bzero(&ro, sizeof(ro)); switch (af) { case AF_INET: dst = satosin(&ro.ro_dst); dst->sin_family = AF_INET; dst->sin_len = sizeof(*dst); dst->sin_addr = addr->v4; break; #ifdef INET6 case AF_INET6: /* * Skip check for addresses with embedded interface scope, * as they would always match anyway. */ if (IN6_IS_SCOPE_EMBED(&addr->v6)) goto out; dst6 = (struct sockaddr_in6 *)&ro.ro_dst; dst6->sin6_family = AF_INET6; dst6->sin6_len = sizeof(*dst6); dst6->sin6_addr = addr->v6; break; #endif /* INET6 */ default: return (0); } /* Skip checks for ipsec interfaces */ if (kif != NULL && kif->pfik_ifp->if_type == IFT_ENC) goto out; switch (af) { #ifdef INET6 case AF_INET6: in6_rtalloc_ign(&ro, 0, rtableid); break; #endif #ifdef INET case AF_INET: in_rtalloc_ign((struct route *)&ro, 0, rtableid); break; #endif } if (ro.ro_rt != NULL) { /* No interface given, this is a no-route check */ if (kif == NULL) goto out; if (kif->pfik_ifp == NULL) { ret = 0; goto out; } /* Perform uRPF check if passed input interface */ ret = 0; rn = (struct radix_node *)ro.ro_rt; do { rt = (struct rtentry *)rn; ifp = rt->rt_ifp; if (kif->pfik_ifp == ifp) ret = 1; rn = rn_mpath_next(rn); } while (check_mpath == 1 && rn != NULL && ret == 0); } else ret = 0; out: if (ro.ro_rt != NULL) RTFREE(ro.ro_rt); return (ret); } #endif int pf_routable(struct pf_addr *addr, sa_family_t af, struct pfi_kif *kif, int rtableid) { #ifdef INET struct nhop4_basic nh4; #endif #ifdef INET6 struct nhop6_basic nh6; #endif struct ifnet *ifp; #ifdef RADIX_MPATH struct radix_node_head *rnh; /* XXX: stick to table 0 for now */ rnh = rt_tables_get_rnh(0, af); if (rnh != NULL && rn_mpath_capable(rnh)) return (pf_routable_oldmpath(addr, af, kif, rtableid)); #endif /* * Skip check for addresses with embedded interface scope, * as they would always match anyway. */ if (af == AF_INET6 && IN6_IS_SCOPE_EMBED(&addr->v6)) return (1); if (af != AF_INET && af != AF_INET6) return (0); /* Skip checks for ipsec interfaces */ if (kif != NULL && kif->pfik_ifp->if_type == IFT_ENC) return (1); ifp = NULL; switch (af) { #ifdef INET6 case AF_INET6: if (fib6_lookup_nh_basic(rtableid, &addr->v6, 0, 0, 0, &nh6)!=0) return (0); ifp = nh6.nh_ifp; break; #endif #ifdef INET case AF_INET: if (fib4_lookup_nh_basic(rtableid, addr->v4, 0, 0, &nh4) != 0) return (0); ifp = nh4.nh_ifp; break; #endif } /* No interface given, this is a no-route check */ if (kif == NULL) return (1); if (kif->pfik_ifp == NULL) return (0); /* Perform uRPF check if passed input interface */ if (kif->pfik_ifp == ifp) return (1); return (0); } #ifdef INET static void pf_route(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, struct pf_state *s, struct pf_pdesc *pd) { struct mbuf *m0, *m1; struct sockaddr_in dst; struct ip *ip; struct ifnet *ifp = NULL; struct pf_addr naddr; struct pf_src_node *sn = NULL; int error = 0; uint16_t ip_len, ip_off; KASSERT(m && *m && r && oifp, ("%s: invalid parameters", __func__)); KASSERT(dir == PF_IN || dir == PF_OUT, ("%s: invalid direction", __func__)); if ((pd->pf_mtag == NULL && ((pd->pf_mtag = pf_get_mtag(*m)) == NULL)) || pd->pf_mtag->routed++ > 3) { m0 = *m; *m = NULL; goto bad_locked; } if (r->rt == PF_DUPTO) { if ((m0 = m_dup(*m, M_NOWAIT)) == NULL) { if (s) PF_STATE_UNLOCK(s); return; } } else { if ((r->rt == PF_REPLYTO) == (r->direction == dir)) { if (s) PF_STATE_UNLOCK(s); return; } m0 = *m; } ip = mtod(m0, struct ip *); bzero(&dst, sizeof(dst)); dst.sin_family = AF_INET; dst.sin_len = sizeof(dst); dst.sin_addr = ip->ip_dst; if (r->rt == PF_FASTROUTE) { struct nhop4_basic nh4; if (s) PF_STATE_UNLOCK(s); if (fib4_lookup_nh_basic(M_GETFIB(m0), ip->ip_dst, 0, m0->m_pkthdr.flowid, &nh4) != 0) { KMOD_IPSTAT_INC(ips_noroute); error = EHOSTUNREACH; goto bad; } ifp = nh4.nh_ifp; dst.sin_addr = nh4.nh_addr; } else { if (TAILQ_EMPTY(&r->rpool.list)) { DPFPRINTF(PF_DEBUG_URGENT, ("%s: TAILQ_EMPTY(&r->rpool.list)\n", __func__)); goto bad_locked; } if (s == NULL) { pf_map_addr(AF_INET, r, (struct pf_addr *)&ip->ip_src, &naddr, NULL, &sn); if (!PF_AZERO(&naddr, AF_INET)) dst.sin_addr.s_addr = naddr.v4.s_addr; ifp = r->rpool.cur->kif ? r->rpool.cur->kif->pfik_ifp : NULL; } else { if (!PF_AZERO(&s->rt_addr, AF_INET)) dst.sin_addr.s_addr = s->rt_addr.v4.s_addr; ifp = s->rt_kif ? s->rt_kif->pfik_ifp : NULL; PF_STATE_UNLOCK(s); } } if (ifp == NULL) goto bad; if (oifp != ifp) { if (pf_test(PF_OUT, ifp, &m0, NULL) != PF_PASS) goto bad; else if (m0 == NULL) goto done; if (m0->m_len < sizeof(struct ip)) { DPFPRINTF(PF_DEBUG_URGENT, ("%s: m0->m_len < sizeof(struct ip)\n", __func__)); goto bad; } ip = mtod(m0, struct ip *); } if (ifp->if_flags & IFF_LOOPBACK) m0->m_flags |= M_SKIP_FIREWALL; ip_len = ntohs(ip->ip_len); ip_off = ntohs(ip->ip_off); /* Copied from FreeBSD 10.0-CURRENT ip_output. */ m0->m_pkthdr.csum_flags |= CSUM_IP; if (m0->m_pkthdr.csum_flags & CSUM_DELAY_DATA & ~ifp->if_hwassist) { in_delayed_cksum(m0); m0->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; } #ifdef SCTP if (m0->m_pkthdr.csum_flags & CSUM_SCTP & ~ifp->if_hwassist) { sctp_delayed_cksum(m, (uint32_t)(ip->ip_hl << 2)); m0->m_pkthdr.csum_flags &= ~CSUM_SCTP; } #endif /* * If small enough for interface, or the interface will take * care of the fragmentation for us, we can just send directly. */ if (ip_len <= ifp->if_mtu || (m0->m_pkthdr.csum_flags & ifp->if_hwassist & CSUM_TSO) != 0) { ip->ip_sum = 0; if (m0->m_pkthdr.csum_flags & CSUM_IP & ~ifp->if_hwassist) { ip->ip_sum = in_cksum(m0, ip->ip_hl << 2); m0->m_pkthdr.csum_flags &= ~CSUM_IP; } m_clrprotoflags(m0); /* Avoid confusing lower layers. */ error = (*ifp->if_output)(ifp, m0, sintosa(&dst), NULL); goto done; } /* Balk when DF bit is set or the interface didn't support TSO. */ if ((ip_off & IP_DF) || (m0->m_pkthdr.csum_flags & CSUM_TSO)) { error = EMSGSIZE; KMOD_IPSTAT_INC(ips_cantfrag); if (r->rt != PF_DUPTO) { icmp_error(m0, ICMP_UNREACH, ICMP_UNREACH_NEEDFRAG, 0, ifp->if_mtu); goto done; } else goto bad; } error = ip_fragment(ip, &m0, ifp->if_mtu, ifp->if_hwassist); if (error) goto bad; for (; m0; m0 = m1) { m1 = m0->m_nextpkt; m0->m_nextpkt = NULL; if (error == 0) { m_clrprotoflags(m0); error = (*ifp->if_output)(ifp, m0, sintosa(&dst), NULL); } else m_freem(m0); } if (error == 0) KMOD_IPSTAT_INC(ips_fragmented); done: if (r->rt != PF_DUPTO) *m = NULL; return; bad_locked: if (s) PF_STATE_UNLOCK(s); bad: m_freem(m0); goto done; } #endif /* INET */ #ifdef INET6 static void pf_route6(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, struct pf_state *s, struct pf_pdesc *pd) { struct mbuf *m0; struct sockaddr_in6 dst; struct ip6_hdr *ip6; struct ifnet *ifp = NULL; struct pf_addr naddr; struct pf_src_node *sn = NULL; KASSERT(m && *m && r && oifp, ("%s: invalid parameters", __func__)); KASSERT(dir == PF_IN || dir == PF_OUT, ("%s: invalid direction", __func__)); if ((pd->pf_mtag == NULL && ((pd->pf_mtag = pf_get_mtag(*m)) == NULL)) || pd->pf_mtag->routed++ > 3) { m0 = *m; *m = NULL; goto bad_locked; } if (r->rt == PF_DUPTO) { if ((m0 = m_dup(*m, M_NOWAIT)) == NULL) { if (s) PF_STATE_UNLOCK(s); return; } } else { if ((r->rt == PF_REPLYTO) == (r->direction == dir)) { if (s) PF_STATE_UNLOCK(s); return; } m0 = *m; } ip6 = mtod(m0, struct ip6_hdr *); bzero(&dst, sizeof(dst)); dst.sin6_family = AF_INET6; dst.sin6_len = sizeof(dst); dst.sin6_addr = ip6->ip6_dst; /* Cheat. XXX why only in the v6 case??? */ if (r->rt == PF_FASTROUTE) { if (s) PF_STATE_UNLOCK(s); m0->m_flags |= M_SKIP_FIREWALL; ip6_output(m0, NULL, NULL, 0, NULL, NULL, NULL); *m = NULL; return; } if (TAILQ_EMPTY(&r->rpool.list)) { DPFPRINTF(PF_DEBUG_URGENT, ("%s: TAILQ_EMPTY(&r->rpool.list)\n", __func__)); goto bad_locked; } if (s == NULL) { pf_map_addr(AF_INET6, r, (struct pf_addr *)&ip6->ip6_src, &naddr, NULL, &sn); if (!PF_AZERO(&naddr, AF_INET6)) PF_ACPY((struct pf_addr *)&dst.sin6_addr, &naddr, AF_INET6); ifp = r->rpool.cur->kif ? r->rpool.cur->kif->pfik_ifp : NULL; } else { if (!PF_AZERO(&s->rt_addr, AF_INET6)) PF_ACPY((struct pf_addr *)&dst.sin6_addr, &s->rt_addr, AF_INET6); ifp = s->rt_kif ? s->rt_kif->pfik_ifp : NULL; } if (s) PF_STATE_UNLOCK(s); if (ifp == NULL) goto bad; if (oifp != ifp) { if (pf_test6(PF_FWD, ifp, &m0, NULL) != PF_PASS) goto bad; else if (m0 == NULL) goto done; if (m0->m_len < sizeof(struct ip6_hdr)) { DPFPRINTF(PF_DEBUG_URGENT, ("%s: m0->m_len < sizeof(struct ip6_hdr)\n", __func__)); goto bad; } ip6 = mtod(m0, struct ip6_hdr *); } if (ifp->if_flags & IFF_LOOPBACK) m0->m_flags |= M_SKIP_FIREWALL; if (m0->m_pkthdr.csum_flags & CSUM_DELAY_DATA_IPV6 & ~ifp->if_hwassist) { uint32_t plen = m0->m_pkthdr.len - sizeof(*ip6); in6_delayed_cksum(m0, plen, sizeof(struct ip6_hdr)); m0->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA_IPV6; } /* * If the packet is too large for the outgoing interface, * send back an icmp6 error. */ if (IN6_IS_SCOPE_EMBED(&dst.sin6_addr)) dst.sin6_addr.s6_addr16[1] = htons(ifp->if_index); if ((u_long)m0->m_pkthdr.len <= ifp->if_mtu) nd6_output_ifp(ifp, ifp, m0, &dst, NULL); else { in6_ifstat_inc(ifp, ifs6_in_toobig); if (r->rt != PF_DUPTO) icmp6_error(m0, ICMP6_PACKET_TOO_BIG, 0, ifp->if_mtu); else goto bad; } done: if (r->rt != PF_DUPTO) *m = NULL; return; bad_locked: if (s) PF_STATE_UNLOCK(s); bad: m_freem(m0); goto done; } #endif /* INET6 */ /* * FreeBSD supports cksum offloads for the following drivers. * em(4), fxp(4), ixgb(4), lge(4), ndis(4), nge(4), re(4), * ti(4), txp(4), xl(4) * * CSUM_DATA_VALID | CSUM_PSEUDO_HDR : * network driver performed cksum including pseudo header, need to verify * csum_data * CSUM_DATA_VALID : * network driver performed cksum, needs to additional pseudo header * cksum computation with partial csum_data(i.e. lack of H/W support for * pseudo header, for instance hme(4), sk(4) and possibly gem(4)) * * After validating the cksum of packet, set both flag CSUM_DATA_VALID and * CSUM_PSEUDO_HDR in order to avoid recomputation of the cksum in upper * TCP/UDP layer. * Also, set csum_data to 0xffff to force cksum validation. */ static int pf_check_proto_cksum(struct mbuf *m, int off, int len, u_int8_t p, sa_family_t af) { u_int16_t sum = 0; int hw_assist = 0; struct ip *ip; if (off < sizeof(struct ip) || len < sizeof(struct udphdr)) return (1); if (m->m_pkthdr.len < off + len) return (1); switch (p) { case IPPROTO_TCP: if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) { if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR) { sum = m->m_pkthdr.csum_data; } else { ip = mtod(m, struct ip *); sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, htonl((u_short)len + m->m_pkthdr.csum_data + IPPROTO_TCP)); } sum ^= 0xffff; ++hw_assist; } break; case IPPROTO_UDP: if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) { if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR) { sum = m->m_pkthdr.csum_data; } else { ip = mtod(m, struct ip *); sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, htonl((u_short)len + m->m_pkthdr.csum_data + IPPROTO_UDP)); } sum ^= 0xffff; ++hw_assist; } break; case IPPROTO_ICMP: #ifdef INET6 case IPPROTO_ICMPV6: #endif /* INET6 */ break; default: return (1); } if (!hw_assist) { switch (af) { case AF_INET: if (p == IPPROTO_ICMP) { if (m->m_len < off) return (1); m->m_data += off; m->m_len -= off; sum = in_cksum(m, len); m->m_data -= off; m->m_len += off; } else { if (m->m_len < sizeof(struct ip)) return (1); sum = in4_cksum(m, p, off, len); } break; #ifdef INET6 case AF_INET6: if (m->m_len < sizeof(struct ip6_hdr)) return (1); sum = in6_cksum(m, p, off, len); break; #endif /* INET6 */ default: return (1); } } if (sum) { switch (p) { case IPPROTO_TCP: { KMOD_TCPSTAT_INC(tcps_rcvbadsum); break; } case IPPROTO_UDP: { KMOD_UDPSTAT_INC(udps_badsum); break; } #ifdef INET case IPPROTO_ICMP: { KMOD_ICMPSTAT_INC(icps_checksum); break; } #endif #ifdef INET6 case IPPROTO_ICMPV6: { KMOD_ICMP6STAT_INC(icp6s_checksum); break; } #endif /* INET6 */ } return (1); } else { if (p == IPPROTO_TCP || p == IPPROTO_UDP) { m->m_pkthdr.csum_flags |= (CSUM_DATA_VALID | CSUM_PSEUDO_HDR); m->m_pkthdr.csum_data = 0xffff; } } return (0); } #ifdef INET int pf_test(int dir, struct ifnet *ifp, struct mbuf **m0, struct inpcb *inp) { struct pfi_kif *kif; u_short action, reason = 0, log = 0; struct mbuf *m = *m0; struct ip *h = NULL; struct m_tag *ipfwtag; struct pf_rule *a = NULL, *r = &V_pf_default_rule, *tr, *nr; struct pf_state *s = NULL; struct pf_ruleset *ruleset = NULL; struct pf_pdesc pd; int off, dirndx, pqid = 0; M_ASSERTPKTHDR(m); if (!V_pf_status.running) return (PF_PASS); memset(&pd, 0, sizeof(pd)); kif = (struct pfi_kif *)ifp->if_pf_kif; if (kif == NULL) { DPFPRINTF(PF_DEBUG_URGENT, ("pf_test: kif == NULL, if_xname %s\n", ifp->if_xname)); return (PF_DROP); } if (kif->pfik_flags & PFI_IFLAG_SKIP) return (PF_PASS); if (m->m_flags & M_SKIP_FIREWALL) return (PF_PASS); pd.pf_mtag = pf_find_mtag(m); PF_RULES_RLOCK(); if (ip_divert_ptr != NULL && ((ipfwtag = m_tag_locate(m, MTAG_IPFW_RULE, 0, NULL)) != NULL)) { struct ipfw_rule_ref *rr = (struct ipfw_rule_ref *)(ipfwtag+1); if (rr->info & IPFW_IS_DIVERT && rr->rulenum == 0) { if (pd.pf_mtag == NULL && ((pd.pf_mtag = pf_get_mtag(m)) == NULL)) { action = PF_DROP; goto done; } pd.pf_mtag->flags |= PF_PACKET_LOOPED; m_tag_delete(m, ipfwtag); } if (pd.pf_mtag && pd.pf_mtag->flags & PF_FASTFWD_OURS_PRESENT) { m->m_flags |= M_FASTFWD_OURS; pd.pf_mtag->flags &= ~PF_FASTFWD_OURS_PRESENT; } } else if (pf_normalize_ip(m0, dir, kif, &reason, &pd) != PF_PASS) { /* We do IP header normalization and packet reassembly here */ action = PF_DROP; goto done; } m = *m0; /* pf_normalize messes with m0 */ h = mtod(m, struct ip *); off = h->ip_hl << 2; if (off < (int)sizeof(struct ip)) { action = PF_DROP; REASON_SET(&reason, PFRES_SHORT); log = 1; goto done; } pd.src = (struct pf_addr *)&h->ip_src; pd.dst = (struct pf_addr *)&h->ip_dst; pd.sport = pd.dport = NULL; pd.ip_sum = &h->ip_sum; pd.proto_sum = NULL; pd.proto = h->ip_p; pd.dir = dir; pd.sidx = (dir == PF_IN) ? 0 : 1; pd.didx = (dir == PF_IN) ? 1 : 0; pd.af = AF_INET; pd.tos = h->ip_tos; pd.tot_len = ntohs(h->ip_len); /* handle fragments that didn't get reassembled by normalization */ if (h->ip_off & htons(IP_MF | IP_OFFMASK)) { action = pf_test_fragment(&r, dir, kif, m, h, &pd, &a, &ruleset); goto done; } switch (h->ip_p) { case IPPROTO_TCP: { struct tcphdr th; pd.hdr.tcp = &th; if (!pf_pull_hdr(m, off, &th, sizeof(th), &action, &reason, AF_INET)) { log = action != PF_PASS; goto done; } pd.p_len = pd.tot_len - off - (th.th_off << 2); if ((th.th_flags & TH_ACK) && pd.p_len == 0) pqid = 1; action = pf_normalize_tcp(dir, kif, m, 0, off, h, &pd); if (action == PF_DROP) goto done; action = pf_test_state_tcp(&s, dir, kif, m, off, h, &pd, &reason); if (action == PF_PASS) { if (pfsync_update_state_ptr != NULL) pfsync_update_state_ptr(s); r = s->rule.ptr; a = s->anchor.ptr; log = s->log; } else if (s == NULL) action = pf_test_rule(&r, &s, dir, kif, m, off, &pd, &a, &ruleset, inp); break; } case IPPROTO_UDP: { struct udphdr uh; pd.hdr.udp = &uh; if (!pf_pull_hdr(m, off, &uh, sizeof(uh), &action, &reason, AF_INET)) { log = action != PF_PASS; goto done; } if (uh.uh_dport == 0 || ntohs(uh.uh_ulen) > m->m_pkthdr.len - off || ntohs(uh.uh_ulen) < sizeof(struct udphdr)) { action = PF_DROP; REASON_SET(&reason, PFRES_SHORT); goto done; } action = pf_test_state_udp(&s, dir, kif, m, off, h, &pd); if (action == PF_PASS) { if (pfsync_update_state_ptr != NULL) pfsync_update_state_ptr(s); r = s->rule.ptr; a = s->anchor.ptr; log = s->log; } else if (s == NULL) action = pf_test_rule(&r, &s, dir, kif, m, off, &pd, &a, &ruleset, inp); break; } case IPPROTO_ICMP: { struct icmp ih; pd.hdr.icmp = &ih; if (!pf_pull_hdr(m, off, &ih, ICMP_MINLEN, &action, &reason, AF_INET)) { log = action != PF_PASS; goto done; } action = pf_test_state_icmp(&s, dir, kif, m, off, h, &pd, &reason); if (action == PF_PASS) { if (pfsync_update_state_ptr != NULL) pfsync_update_state_ptr(s); r = s->rule.ptr; a = s->anchor.ptr; log = s->log; } else if (s == NULL) action = pf_test_rule(&r, &s, dir, kif, m, off, &pd, &a, &ruleset, inp); break; } #ifdef INET6 case IPPROTO_ICMPV6: { action = PF_DROP; DPFPRINTF(PF_DEBUG_MISC, ("pf: dropping IPv4 packet with ICMPv6 payload\n")); goto done; } #endif default: action = pf_test_state_other(&s, dir, kif, m, &pd); if (action == PF_PASS) { if (pfsync_update_state_ptr != NULL) pfsync_update_state_ptr(s); r = s->rule.ptr; a = s->anchor.ptr; log = s->log; } else if (s == NULL) action = pf_test_rule(&r, &s, dir, kif, m, off, &pd, &a, &ruleset, inp); break; } done: PF_RULES_RUNLOCK(); if (action == PF_PASS && h->ip_hl > 5 && !((s && s->state_flags & PFSTATE_ALLOWOPTS) || r->allow_opts)) { action = PF_DROP; REASON_SET(&reason, PFRES_IPOPTIONS); log = r->log; DPFPRINTF(PF_DEBUG_MISC, ("pf: dropping packet with ip options\n")); } if (s && s->tag > 0 && pf_tag_packet(m, &pd, s->tag)) { action = PF_DROP; REASON_SET(&reason, PFRES_MEMORY); } if (r->rtableid >= 0) M_SETFIB(m, r->rtableid); if (r->scrub_flags & PFSTATE_SETPRIO) { if (pd.tos & IPTOS_LOWDELAY) pqid = 1; if (pf_ieee8021q_setpcp(m, r->set_prio[pqid])) { action = PF_DROP; REASON_SET(&reason, PFRES_MEMORY); log = 1; DPFPRINTF(PF_DEBUG_MISC, ("pf: failed to allocate 802.1q mtag\n")); } } #ifdef ALTQ if (action == PF_PASS && r->qid) { if (pd.pf_mtag == NULL && ((pd.pf_mtag = pf_get_mtag(m)) == NULL)) { action = PF_DROP; REASON_SET(&reason, PFRES_MEMORY); } else { if (s != NULL) pd.pf_mtag->qid_hash = pf_state_hash(s); if (pqid || (pd.tos & IPTOS_LOWDELAY)) pd.pf_mtag->qid = r->pqid; else pd.pf_mtag->qid = r->qid; /* Add hints for ecn. */ pd.pf_mtag->hdr = h; } } #endif /* ALTQ */ /* * connections redirected to loopback should not match sockets * bound specifically to loopback due to security implications, * see tcp_input() and in_pcblookup_listen(). */ if (dir == PF_IN && action == PF_PASS && (pd.proto == IPPROTO_TCP || pd.proto == IPPROTO_UDP) && s != NULL && s->nat_rule.ptr != NULL && (s->nat_rule.ptr->action == PF_RDR || s->nat_rule.ptr->action == PF_BINAT) && (ntohl(pd.dst->v4.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) m->m_flags |= M_SKIP_FIREWALL; if (action == PF_PASS && r->divert.port && ip_divert_ptr != NULL && !PACKET_LOOPED(&pd)) { ipfwtag = m_tag_alloc(MTAG_IPFW_RULE, 0, sizeof(struct ipfw_rule_ref), M_NOWAIT | M_ZERO); if (ipfwtag != NULL) { ((struct ipfw_rule_ref *)(ipfwtag+1))->info = ntohs(r->divert.port); ((struct ipfw_rule_ref *)(ipfwtag+1))->rulenum = dir; if (s) PF_STATE_UNLOCK(s); m_tag_prepend(m, ipfwtag); if (m->m_flags & M_FASTFWD_OURS) { if (pd.pf_mtag == NULL && ((pd.pf_mtag = pf_get_mtag(m)) == NULL)) { action = PF_DROP; REASON_SET(&reason, PFRES_MEMORY); log = 1; DPFPRINTF(PF_DEBUG_MISC, ("pf: failed to allocate tag\n")); } else { pd.pf_mtag->flags |= PF_FASTFWD_OURS_PRESENT; m->m_flags &= ~M_FASTFWD_OURS; } } ip_divert_ptr(*m0, dir == PF_IN ? DIR_IN : DIR_OUT); *m0 = NULL; return (action); } else { /* XXX: ipfw has the same behaviour! */ action = PF_DROP; REASON_SET(&reason, PFRES_MEMORY); log = 1; DPFPRINTF(PF_DEBUG_MISC, ("pf: failed to allocate divert tag\n")); } } if (log) { struct pf_rule *lr; if (s != NULL && s->nat_rule.ptr != NULL && s->nat_rule.ptr->log & PF_LOG_ALL) lr = s->nat_rule.ptr; else lr = r; PFLOG_PACKET(kif, m, AF_INET, dir, reason, lr, a, ruleset, &pd, (s == NULL)); } kif->pfik_bytes[0][dir == PF_OUT][action != PF_PASS] += pd.tot_len; kif->pfik_packets[0][dir == PF_OUT][action != PF_PASS]++; if (action == PF_PASS || r->action == PF_DROP) { dirndx = (dir == PF_OUT); r->packets[dirndx]++; r->bytes[dirndx] += pd.tot_len; if (a != NULL) { a->packets[dirndx]++; a->bytes[dirndx] += pd.tot_len; } if (s != NULL) { if (s->nat_rule.ptr != NULL) { s->nat_rule.ptr->packets[dirndx]++; s->nat_rule.ptr->bytes[dirndx] += pd.tot_len; } if (s->src_node != NULL) { s->src_node->packets[dirndx]++; s->src_node->bytes[dirndx] += pd.tot_len; } if (s->nat_src_node != NULL) { s->nat_src_node->packets[dirndx]++; s->nat_src_node->bytes[dirndx] += pd.tot_len; } dirndx = (dir == s->direction) ? 0 : 1; s->packets[dirndx]++; s->bytes[dirndx] += pd.tot_len; } tr = r; nr = (s != NULL) ? s->nat_rule.ptr : pd.nat_rule; if (nr != NULL && r == &V_pf_default_rule) tr = nr; if (tr->src.addr.type == PF_ADDR_TABLE) pfr_update_stats(tr->src.addr.p.tbl, (s == NULL) ? pd.src : &s->key[(s->direction == PF_IN)]-> addr[(s->direction == PF_OUT)], pd.af, pd.tot_len, dir == PF_OUT, r->action == PF_PASS, tr->src.neg); if (tr->dst.addr.type == PF_ADDR_TABLE) pfr_update_stats(tr->dst.addr.p.tbl, (s == NULL) ? pd.dst : &s->key[(s->direction == PF_IN)]-> addr[(s->direction == PF_IN)], pd.af, pd.tot_len, dir == PF_OUT, r->action == PF_PASS, tr->dst.neg); } switch (action) { case PF_SYNPROXY_DROP: m_freem(*m0); case PF_DEFER: *m0 = NULL; action = PF_PASS; break; case PF_DROP: m_freem(*m0); *m0 = NULL; break; default: /* pf_route() returns unlocked. */ if (r->rt) { pf_route(m0, r, dir, kif->pfik_ifp, s, &pd); return (action); } break; } if (s) PF_STATE_UNLOCK(s); return (action); } #endif /* INET */ #ifdef INET6 int pf_test6(int dir, struct ifnet *ifp, struct mbuf **m0, struct inpcb *inp) { struct pfi_kif *kif; u_short action, reason = 0, log = 0; struct mbuf *m = *m0, *n = NULL; struct m_tag *mtag; struct ip6_hdr *h = NULL; struct pf_rule *a = NULL, *r = &V_pf_default_rule, *tr, *nr; struct pf_state *s = NULL; struct pf_ruleset *ruleset = NULL; struct pf_pdesc pd; int off, terminal = 0, dirndx, rh_cnt = 0, pqid = 0; int fwdir = dir; M_ASSERTPKTHDR(m); /* Detect packet forwarding. * If the input interface is different from the output interface we're * forwarding. * We do need to be careful about bridges. If the * net.link.bridge.pfil_bridge sysctl is set we can be filtering on a * bridge, so if the input interface is a bridge member and the output * interface is its bridge or a member of the same bridge we're not * actually forwarding but bridging. */ if (dir == PF_OUT && m->m_pkthdr.rcvif && ifp != m->m_pkthdr.rcvif && (m->m_pkthdr.rcvif->if_bridge == NULL || (m->m_pkthdr.rcvif->if_bridge != ifp->if_softc && m->m_pkthdr.rcvif->if_bridge != ifp->if_bridge))) fwdir = PF_FWD; if (!V_pf_status.running) return (PF_PASS); memset(&pd, 0, sizeof(pd)); pd.pf_mtag = pf_find_mtag(m); if (pd.pf_mtag && pd.pf_mtag->flags & PF_TAG_GENERATED) return (PF_PASS); kif = (struct pfi_kif *)ifp->if_pf_kif; if (kif == NULL) { DPFPRINTF(PF_DEBUG_URGENT, ("pf_test6: kif == NULL, if_xname %s\n", ifp->if_xname)); return (PF_DROP); } if (kif->pfik_flags & PFI_IFLAG_SKIP) return (PF_PASS); if (m->m_flags & M_SKIP_FIREWALL) return (PF_PASS); PF_RULES_RLOCK(); /* We do IP header normalization and packet reassembly here */ if (pf_normalize_ip6(m0, dir, kif, &reason, &pd) != PF_PASS) { action = PF_DROP; goto done; } m = *m0; /* pf_normalize messes with m0 */ h = mtod(m, struct ip6_hdr *); #if 1 /* * we do not support jumbogram yet. if we keep going, zero ip6_plen * will do something bad, so drop the packet for now. */ if (htons(h->ip6_plen) == 0) { action = PF_DROP; REASON_SET(&reason, PFRES_NORM); /*XXX*/ goto done; } #endif pd.src = (struct pf_addr *)&h->ip6_src; pd.dst = (struct pf_addr *)&h->ip6_dst; pd.sport = pd.dport = NULL; pd.ip_sum = NULL; pd.proto_sum = NULL; pd.dir = dir; pd.sidx = (dir == PF_IN) ? 0 : 1; pd.didx = (dir == PF_IN) ? 1 : 0; pd.af = AF_INET6; pd.tos = 0; pd.tot_len = ntohs(h->ip6_plen) + sizeof(struct ip6_hdr); off = ((caddr_t)h - m->m_data) + sizeof(struct ip6_hdr); pd.proto = h->ip6_nxt; do { switch (pd.proto) { case IPPROTO_FRAGMENT: action = pf_test_fragment(&r, dir, kif, m, h, &pd, &a, &ruleset); if (action == PF_DROP) REASON_SET(&reason, PFRES_FRAG); goto done; case IPPROTO_ROUTING: { struct ip6_rthdr rthdr; if (rh_cnt++) { DPFPRINTF(PF_DEBUG_MISC, ("pf: IPv6 more than one rthdr\n")); action = PF_DROP; REASON_SET(&reason, PFRES_IPOPTIONS); log = 1; goto done; } if (!pf_pull_hdr(m, off, &rthdr, sizeof(rthdr), NULL, &reason, pd.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: IPv6 short rthdr\n")); action = PF_DROP; REASON_SET(&reason, PFRES_SHORT); log = 1; goto done; } if (rthdr.ip6r_type == IPV6_RTHDR_TYPE_0) { DPFPRINTF(PF_DEBUG_MISC, ("pf: IPv6 rthdr0\n")); action = PF_DROP; REASON_SET(&reason, PFRES_IPOPTIONS); log = 1; goto done; } /* FALLTHROUGH */ } case IPPROTO_AH: case IPPROTO_HOPOPTS: case IPPROTO_DSTOPTS: { /* get next header and header length */ struct ip6_ext opt6; if (!pf_pull_hdr(m, off, &opt6, sizeof(opt6), NULL, &reason, pd.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: IPv6 short opt\n")); action = PF_DROP; log = 1; goto done; } if (pd.proto == IPPROTO_AH) off += (opt6.ip6e_len + 2) * 4; else off += (opt6.ip6e_len + 1) * 8; pd.proto = opt6.ip6e_nxt; /* goto the next header */ break; } default: terminal++; break; } } while (!terminal); /* if there's no routing header, use unmodified mbuf for checksumming */ if (!n) n = m; switch (pd.proto) { case IPPROTO_TCP: { struct tcphdr th; pd.hdr.tcp = &th; if (!pf_pull_hdr(m, off, &th, sizeof(th), &action, &reason, AF_INET6)) { log = action != PF_PASS; goto done; } pd.p_len = pd.tot_len - off - (th.th_off << 2); action = pf_normalize_tcp(dir, kif, m, 0, off, h, &pd); if (action == PF_DROP) goto done; action = pf_test_state_tcp(&s, dir, kif, m, off, h, &pd, &reason); if (action == PF_PASS) { if (pfsync_update_state_ptr != NULL) pfsync_update_state_ptr(s); r = s->rule.ptr; a = s->anchor.ptr; log = s->log; } else if (s == NULL) action = pf_test_rule(&r, &s, dir, kif, m, off, &pd, &a, &ruleset, inp); break; } case IPPROTO_UDP: { struct udphdr uh; pd.hdr.udp = &uh; if (!pf_pull_hdr(m, off, &uh, sizeof(uh), &action, &reason, AF_INET6)) { log = action != PF_PASS; goto done; } if (uh.uh_dport == 0 || ntohs(uh.uh_ulen) > m->m_pkthdr.len - off || ntohs(uh.uh_ulen) < sizeof(struct udphdr)) { action = PF_DROP; REASON_SET(&reason, PFRES_SHORT); goto done; } action = pf_test_state_udp(&s, dir, kif, m, off, h, &pd); if (action == PF_PASS) { if (pfsync_update_state_ptr != NULL) pfsync_update_state_ptr(s); r = s->rule.ptr; a = s->anchor.ptr; log = s->log; } else if (s == NULL) action = pf_test_rule(&r, &s, dir, kif, m, off, &pd, &a, &ruleset, inp); break; } case IPPROTO_ICMP: { action = PF_DROP; DPFPRINTF(PF_DEBUG_MISC, ("pf: dropping IPv6 packet with ICMPv4 payload\n")); goto done; } case IPPROTO_ICMPV6: { struct icmp6_hdr ih; pd.hdr.icmp6 = &ih; if (!pf_pull_hdr(m, off, &ih, sizeof(ih), &action, &reason, AF_INET6)) { log = action != PF_PASS; goto done; } action = pf_test_state_icmp(&s, dir, kif, m, off, h, &pd, &reason); if (action == PF_PASS) { if (pfsync_update_state_ptr != NULL) pfsync_update_state_ptr(s); r = s->rule.ptr; a = s->anchor.ptr; log = s->log; } else if (s == NULL) action = pf_test_rule(&r, &s, dir, kif, m, off, &pd, &a, &ruleset, inp); break; } default: action = pf_test_state_other(&s, dir, kif, m, &pd); if (action == PF_PASS) { if (pfsync_update_state_ptr != NULL) pfsync_update_state_ptr(s); r = s->rule.ptr; a = s->anchor.ptr; log = s->log; } else if (s == NULL) action = pf_test_rule(&r, &s, dir, kif, m, off, &pd, &a, &ruleset, inp); break; } done: PF_RULES_RUNLOCK(); if (n != m) { m_freem(n); n = NULL; } /* handle dangerous IPv6 extension headers. */ if (action == PF_PASS && rh_cnt && !((s && s->state_flags & PFSTATE_ALLOWOPTS) || r->allow_opts)) { action = PF_DROP; REASON_SET(&reason, PFRES_IPOPTIONS); log = r->log; DPFPRINTF(PF_DEBUG_MISC, ("pf: dropping packet with dangerous v6 headers\n")); } if (s && s->tag > 0 && pf_tag_packet(m, &pd, s->tag)) { action = PF_DROP; REASON_SET(&reason, PFRES_MEMORY); } if (r->rtableid >= 0) M_SETFIB(m, r->rtableid); if (r->scrub_flags & PFSTATE_SETPRIO) { if (pd.tos & IPTOS_LOWDELAY) pqid = 1; if (pf_ieee8021q_setpcp(m, r->set_prio[pqid])) { action = PF_DROP; REASON_SET(&reason, PFRES_MEMORY); log = 1; DPFPRINTF(PF_DEBUG_MISC, ("pf: failed to allocate 802.1q mtag\n")); } } #ifdef ALTQ if (action == PF_PASS && r->qid) { if (pd.pf_mtag == NULL && ((pd.pf_mtag = pf_get_mtag(m)) == NULL)) { action = PF_DROP; REASON_SET(&reason, PFRES_MEMORY); } else { if (s != NULL) pd.pf_mtag->qid_hash = pf_state_hash(s); if (pd.tos & IPTOS_LOWDELAY) pd.pf_mtag->qid = r->pqid; else pd.pf_mtag->qid = r->qid; /* Add hints for ecn. */ pd.pf_mtag->hdr = h; } } #endif /* ALTQ */ if (dir == PF_IN && action == PF_PASS && (pd.proto == IPPROTO_TCP || pd.proto == IPPROTO_UDP) && s != NULL && s->nat_rule.ptr != NULL && (s->nat_rule.ptr->action == PF_RDR || s->nat_rule.ptr->action == PF_BINAT) && IN6_IS_ADDR_LOOPBACK(&pd.dst->v6)) m->m_flags |= M_SKIP_FIREWALL; /* XXX: Anybody working on it?! */ if (r->divert.port) printf("pf: divert(9) is not supported for IPv6\n"); if (log) { struct pf_rule *lr; if (s != NULL && s->nat_rule.ptr != NULL && s->nat_rule.ptr->log & PF_LOG_ALL) lr = s->nat_rule.ptr; else lr = r; PFLOG_PACKET(kif, m, AF_INET6, dir, reason, lr, a, ruleset, &pd, (s == NULL)); } kif->pfik_bytes[1][dir == PF_OUT][action != PF_PASS] += pd.tot_len; kif->pfik_packets[1][dir == PF_OUT][action != PF_PASS]++; if (action == PF_PASS || r->action == PF_DROP) { dirndx = (dir == PF_OUT); r->packets[dirndx]++; r->bytes[dirndx] += pd.tot_len; if (a != NULL) { a->packets[dirndx]++; a->bytes[dirndx] += pd.tot_len; } if (s != NULL) { if (s->nat_rule.ptr != NULL) { s->nat_rule.ptr->packets[dirndx]++; s->nat_rule.ptr->bytes[dirndx] += pd.tot_len; } if (s->src_node != NULL) { s->src_node->packets[dirndx]++; s->src_node->bytes[dirndx] += pd.tot_len; } if (s->nat_src_node != NULL) { s->nat_src_node->packets[dirndx]++; s->nat_src_node->bytes[dirndx] += pd.tot_len; } dirndx = (dir == s->direction) ? 0 : 1; s->packets[dirndx]++; s->bytes[dirndx] += pd.tot_len; } tr = r; nr = (s != NULL) ? s->nat_rule.ptr : pd.nat_rule; if (nr != NULL && r == &V_pf_default_rule) tr = nr; if (tr->src.addr.type == PF_ADDR_TABLE) pfr_update_stats(tr->src.addr.p.tbl, (s == NULL) ? pd.src : &s->key[(s->direction == PF_IN)]->addr[0], pd.af, pd.tot_len, dir == PF_OUT, r->action == PF_PASS, tr->src.neg); if (tr->dst.addr.type == PF_ADDR_TABLE) pfr_update_stats(tr->dst.addr.p.tbl, (s == NULL) ? pd.dst : &s->key[(s->direction == PF_IN)]->addr[1], pd.af, pd.tot_len, dir == PF_OUT, r->action == PF_PASS, tr->dst.neg); } switch (action) { case PF_SYNPROXY_DROP: m_freem(*m0); case PF_DEFER: *m0 = NULL; action = PF_PASS; break; case PF_DROP: m_freem(*m0); *m0 = NULL; break; default: /* pf_route6() returns unlocked. */ if (r->rt) { pf_route6(m0, r, dir, kif->pfik_ifp, s, &pd); return (action); } break; } if (s) PF_STATE_UNLOCK(s); /* If reassembled packet passed, create new fragments. */ if (action == PF_PASS && *m0 && fwdir == PF_FWD && (mtag = m_tag_find(m, PF_REASSEMBLED, NULL)) != NULL) action = pf_refragment6(ifp, m0, mtag); return (action); } #endif /* INET6 */ Index: projects/vnet/sys/netpfil/pf/pf_if.c =================================================================== --- projects/vnet/sys/netpfil/pf/pf_if.c (revision 302157) +++ projects/vnet/sys/netpfil/pf/pf_if.c (revision 302158) @@ -1,922 +1,922 @@ /*- * Copyright (c) 2001 Daniel Hartmeier * Copyright (c) 2003 Cedric Berger * Copyright (c) 2005 Henning Brauer * Copyright (c) 2005 Ryan McBride * Copyright (c) 2012 Gleb Smirnoff * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * * - Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * - 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 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 HOLDERS 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. * * $OpenBSD: pf_if.c,v 1.54 2008/06/14 16:55:28 mk Exp $ */ #include __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include #include #include #include #include #include #include #include #include #include #include #include VNET_DEFINE(struct pfi_kif *, pfi_all); static VNET_DEFINE(long, pfi_update); #define V_pfi_update VNET(pfi_update) #define PFI_BUFFER_MAX 0x10000 VNET_DECLARE(int, pf_vnet_active); #define V_pf_vnet_active VNET(pf_vnet_active) static VNET_DEFINE(struct pfr_addr *, pfi_buffer); static VNET_DEFINE(int, pfi_buffer_cnt); static VNET_DEFINE(int, pfi_buffer_max); #define V_pfi_buffer VNET(pfi_buffer) #define V_pfi_buffer_cnt VNET(pfi_buffer_cnt) #define V_pfi_buffer_max VNET(pfi_buffer_max) eventhandler_tag pfi_attach_cookie; eventhandler_tag pfi_detach_cookie; eventhandler_tag pfi_attach_group_cookie; eventhandler_tag pfi_change_group_cookie; eventhandler_tag pfi_detach_group_cookie; eventhandler_tag pfi_ifaddr_event_cookie; static void pfi_attach_ifnet(struct ifnet *); static void pfi_attach_ifgroup(struct ifg_group *); static void pfi_kif_update(struct pfi_kif *); static void pfi_dynaddr_update(struct pfi_dynaddr *dyn); static void pfi_table_update(struct pfr_ktable *, struct pfi_kif *, int, int); static void pfi_instance_add(struct ifnet *, int, int); static void pfi_address_add(struct sockaddr *, int, int); static int pfi_if_compare(struct pfi_kif *, struct pfi_kif *); static int pfi_skip_if(const char *, struct pfi_kif *); static int pfi_unmask(void *); static void pfi_attach_ifnet_event(void * __unused, struct ifnet *); static void pfi_detach_ifnet_event(void * __unused, struct ifnet *); static void pfi_attach_group_event(void *, struct ifg_group *); static void pfi_change_group_event(void *, char *); static void pfi_detach_group_event(void *, struct ifg_group *); static void pfi_ifaddr_event(void * __unused, struct ifnet *); RB_HEAD(pfi_ifhead, pfi_kif); static RB_PROTOTYPE(pfi_ifhead, pfi_kif, pfik_tree, pfi_if_compare); static RB_GENERATE(pfi_ifhead, pfi_kif, pfik_tree, pfi_if_compare); static VNET_DEFINE(struct pfi_ifhead, pfi_ifs); #define V_pfi_ifs VNET(pfi_ifs) #define PFI_BUFFER_MAX 0x10000 MALLOC_DEFINE(PFI_MTYPE, "pf_ifnet", "pf(4) interface database"); LIST_HEAD(pfi_list, pfi_kif); static VNET_DEFINE(struct pfi_list, pfi_unlinked_kifs); #define V_pfi_unlinked_kifs VNET(pfi_unlinked_kifs) static struct mtx pfi_unlnkdkifs_mtx; MTX_SYSINIT(pfi_unlnkdkifs_mtx, &pfi_unlnkdkifs_mtx, "pf unlinked interfaces", MTX_DEF); void pfi_initialize_vnet(void) { struct ifg_group *ifg; struct ifnet *ifp; struct pfi_kif *kif; V_pfi_buffer_max = 64; V_pfi_buffer = malloc(V_pfi_buffer_max * sizeof(*V_pfi_buffer), PFI_MTYPE, M_WAITOK); kif = malloc(sizeof(*kif), PFI_MTYPE, M_WAITOK); PF_RULES_WLOCK(); V_pfi_all = pfi_kif_attach(kif, IFG_ALL); PF_RULES_WUNLOCK(); IFNET_RLOCK(); TAILQ_FOREACH(ifg, &V_ifg_head, ifg_next) pfi_attach_ifgroup(ifg); TAILQ_FOREACH(ifp, &V_ifnet, if_link) pfi_attach_ifnet(ifp); IFNET_RUNLOCK(); } void pfi_initialize(void) { pfi_attach_cookie = EVENTHANDLER_REGISTER(ifnet_arrival_event, pfi_attach_ifnet_event, NULL, EVENTHANDLER_PRI_ANY); pfi_detach_cookie = EVENTHANDLER_REGISTER(ifnet_departure_event, pfi_detach_ifnet_event, NULL, EVENTHANDLER_PRI_ANY); pfi_attach_group_cookie = EVENTHANDLER_REGISTER(group_attach_event, pfi_attach_group_event, curvnet, EVENTHANDLER_PRI_ANY); pfi_change_group_cookie = EVENTHANDLER_REGISTER(group_change_event, pfi_change_group_event, curvnet, EVENTHANDLER_PRI_ANY); pfi_detach_group_cookie = EVENTHANDLER_REGISTER(group_detach_event, pfi_detach_group_event, curvnet, EVENTHANDLER_PRI_ANY); pfi_ifaddr_event_cookie = EVENTHANDLER_REGISTER(ifaddr_event, pfi_ifaddr_event, NULL, EVENTHANDLER_PRI_ANY); } void pfi_cleanup_vnet(void) { struct pfi_kif *kif; PF_RULES_WASSERT(); V_pfi_all = NULL; while ((kif = RB_MIN(pfi_ifhead, &V_pfi_ifs))) { RB_REMOVE(pfi_ifhead, &V_pfi_ifs, kif); if (kif->pfik_group) kif->pfik_group->ifg_pf_kif = NULL; if (kif->pfik_ifp) kif->pfik_ifp->if_pf_kif = NULL; free(kif, PFI_MTYPE); } mtx_lock(&pfi_unlnkdkifs_mtx); while ((kif = LIST_FIRST(&V_pfi_unlinked_kifs))) { LIST_REMOVE(kif, pfik_list); free(kif, PFI_MTYPE); } mtx_unlock(&pfi_unlnkdkifs_mtx); free(V_pfi_buffer, PFI_MTYPE); } void pfi_cleanup(void) { EVENTHANDLER_DEREGISTER(ifnet_arrival_event, pfi_attach_cookie); EVENTHANDLER_DEREGISTER(ifnet_departure_event, pfi_detach_cookie); EVENTHANDLER_DEREGISTER(group_attach_event, pfi_attach_group_cookie); EVENTHANDLER_DEREGISTER(group_change_event, pfi_change_group_cookie); EVENTHANDLER_DEREGISTER(group_detach_event, pfi_detach_group_cookie); EVENTHANDLER_DEREGISTER(ifaddr_event, pfi_ifaddr_event_cookie); } struct pfi_kif * pfi_kif_find(const char *kif_name) { struct pfi_kif_cmp s; PF_RULES_ASSERT(); bzero(&s, sizeof(s)); strlcpy(s.pfik_name, kif_name, sizeof(s.pfik_name)); return (RB_FIND(pfi_ifhead, &V_pfi_ifs, (struct pfi_kif *)&s)); } struct pfi_kif * pfi_kif_attach(struct pfi_kif *kif, const char *kif_name) { struct pfi_kif *kif1; PF_RULES_WASSERT(); KASSERT(kif != NULL, ("%s: null kif", __func__)); kif1 = pfi_kif_find(kif_name); if (kif1 != NULL) { free(kif, PFI_MTYPE); return (kif1); } bzero(kif, sizeof(*kif)); strlcpy(kif->pfik_name, kif_name, sizeof(kif->pfik_name)); /* * It seems that the value of time_second is in unintialzied state * when pf sets interface statistics clear time in boot phase if pf * was statically linked to kernel. Instead of setting the bogus * time value have pfi_get_ifaces handle this case. In * pfi_get_ifaces it uses time_second if it sees the time is 0. */ kif->pfik_tzero = time_second > 1 ? time_second : 0; TAILQ_INIT(&kif->pfik_dynaddrs); RB_INSERT(pfi_ifhead, &V_pfi_ifs, kif); return (kif); } void pfi_kif_ref(struct pfi_kif *kif) { PF_RULES_WASSERT(); kif->pfik_rulerefs++; } void pfi_kif_unref(struct pfi_kif *kif) { PF_RULES_WASSERT(); KASSERT(kif->pfik_rulerefs > 0, ("%s: %p has zero refs", __func__, kif)); kif->pfik_rulerefs--; if (kif->pfik_rulerefs > 0) return; /* kif referencing an existing ifnet or group should exist. */ if (kif->pfik_ifp != NULL || kif->pfik_group != NULL || kif == V_pfi_all) return; RB_REMOVE(pfi_ifhead, &V_pfi_ifs, kif); kif->pfik_flags |= PFI_IFLAG_REFS; mtx_lock(&pfi_unlnkdkifs_mtx); LIST_INSERT_HEAD(&V_pfi_unlinked_kifs, kif, pfik_list); mtx_unlock(&pfi_unlnkdkifs_mtx); } void pfi_kif_purge(void) { struct pfi_kif *kif, *kif1; /* * Do naive mark-and-sweep garbage collecting of old kifs. * Reference flag is raised by pf_purge_expired_states(). */ mtx_lock(&pfi_unlnkdkifs_mtx); LIST_FOREACH_SAFE(kif, &V_pfi_unlinked_kifs, pfik_list, kif1) { if (!(kif->pfik_flags & PFI_IFLAG_REFS)) { LIST_REMOVE(kif, pfik_list); free(kif, PFI_MTYPE); } else kif->pfik_flags &= ~PFI_IFLAG_REFS; } mtx_unlock(&pfi_unlnkdkifs_mtx); } int pfi_kif_match(struct pfi_kif *rule_kif, struct pfi_kif *packet_kif) { struct ifg_list *p; if (rule_kif == NULL || rule_kif == packet_kif) return (1); if (rule_kif->pfik_group != NULL) /* XXXGL: locking? */ TAILQ_FOREACH(p, &packet_kif->pfik_ifp->if_groups, ifgl_next) if (p->ifgl_group == rule_kif->pfik_group) return (1); return (0); } static void pfi_attach_ifnet(struct ifnet *ifp) { struct pfi_kif *kif; kif = malloc(sizeof(*kif), PFI_MTYPE, M_WAITOK); PF_RULES_WLOCK(); V_pfi_update++; kif = pfi_kif_attach(kif, ifp->if_xname); kif->pfik_ifp = ifp; ifp->if_pf_kif = kif; pfi_kif_update(kif); PF_RULES_WUNLOCK(); } static void pfi_attach_ifgroup(struct ifg_group *ifg) { struct pfi_kif *kif; kif = malloc(sizeof(*kif), PFI_MTYPE, M_WAITOK); PF_RULES_WLOCK(); V_pfi_update++; kif = pfi_kif_attach(kif, ifg->ifg_group); kif->pfik_group = ifg; ifg->ifg_pf_kif = kif; PF_RULES_WUNLOCK(); } int pfi_match_addr(struct pfi_dynaddr *dyn, struct pf_addr *a, sa_family_t af) { switch (af) { #ifdef INET case AF_INET: switch (dyn->pfid_acnt4) { case 0: return (0); case 1: return (PF_MATCHA(0, &dyn->pfid_addr4, &dyn->pfid_mask4, a, AF_INET)); default: return (pfr_match_addr(dyn->pfid_kt, a, AF_INET)); } break; #endif /* INET */ #ifdef INET6 case AF_INET6: switch (dyn->pfid_acnt6) { case 0: return (0); case 1: return (PF_MATCHA(0, &dyn->pfid_addr6, &dyn->pfid_mask6, a, AF_INET6)); default: return (pfr_match_addr(dyn->pfid_kt, a, AF_INET6)); } break; #endif /* INET6 */ default: return (0); } } int pfi_dynaddr_setup(struct pf_addr_wrap *aw, sa_family_t af) { struct pfi_dynaddr *dyn; char tblname[PF_TABLE_NAME_SIZE]; struct pf_ruleset *ruleset = NULL; struct pfi_kif *kif; int rv = 0; PF_RULES_WASSERT(); KASSERT(aw->type == PF_ADDR_DYNIFTL, ("%s: type %u", __func__, aw->type)); KASSERT(aw->p.dyn == NULL, ("%s: dyn is %p", __func__, aw->p.dyn)); if ((dyn = malloc(sizeof(*dyn), PFI_MTYPE, M_NOWAIT | M_ZERO)) == NULL) return (ENOMEM); if ((kif = malloc(sizeof(*kif), PFI_MTYPE, M_NOWAIT)) == NULL) { free(dyn, PFI_MTYPE); return (ENOMEM); } if (!strcmp(aw->v.ifname, "self")) dyn->pfid_kif = pfi_kif_attach(kif, IFG_ALL); else dyn->pfid_kif = pfi_kif_attach(kif, aw->v.ifname); pfi_kif_ref(dyn->pfid_kif); dyn->pfid_net = pfi_unmask(&aw->v.a.mask); if (af == AF_INET && dyn->pfid_net == 32) dyn->pfid_net = 128; strlcpy(tblname, aw->v.ifname, sizeof(tblname)); if (aw->iflags & PFI_AFLAG_NETWORK) strlcat(tblname, ":network", sizeof(tblname)); if (aw->iflags & PFI_AFLAG_BROADCAST) strlcat(tblname, ":broadcast", sizeof(tblname)); if (aw->iflags & PFI_AFLAG_PEER) strlcat(tblname, ":peer", sizeof(tblname)); if (aw->iflags & PFI_AFLAG_NOALIAS) strlcat(tblname, ":0", sizeof(tblname)); if (dyn->pfid_net != 128) snprintf(tblname + strlen(tblname), sizeof(tblname) - strlen(tblname), "/%d", dyn->pfid_net); if ((ruleset = pf_find_or_create_ruleset(PF_RESERVED_ANCHOR)) == NULL) { rv = ENOMEM; goto _bad; } if ((dyn->pfid_kt = pfr_attach_table(ruleset, tblname)) == NULL) { rv = ENOMEM; goto _bad; } dyn->pfid_kt->pfrkt_flags |= PFR_TFLAG_ACTIVE; dyn->pfid_iflags = aw->iflags; dyn->pfid_af = af; TAILQ_INSERT_TAIL(&dyn->pfid_kif->pfik_dynaddrs, dyn, entry); aw->p.dyn = dyn; pfi_kif_update(dyn->pfid_kif); return (0); _bad: if (dyn->pfid_kt != NULL) pfr_detach_table(dyn->pfid_kt); if (ruleset != NULL) pf_remove_if_empty_ruleset(ruleset); if (dyn->pfid_kif != NULL) pfi_kif_unref(dyn->pfid_kif); free(dyn, PFI_MTYPE); return (rv); } static void pfi_kif_update(struct pfi_kif *kif) { struct ifg_list *ifgl; struct pfi_dynaddr *p; PF_RULES_WASSERT(); /* update all dynaddr */ TAILQ_FOREACH(p, &kif->pfik_dynaddrs, entry) pfi_dynaddr_update(p); /* again for all groups kif is member of */ if (kif->pfik_ifp != NULL) { IF_ADDR_RLOCK(kif->pfik_ifp); TAILQ_FOREACH(ifgl, &kif->pfik_ifp->if_groups, ifgl_next) pfi_kif_update((struct pfi_kif *) ifgl->ifgl_group->ifg_pf_kif); IF_ADDR_RUNLOCK(kif->pfik_ifp); } } static void pfi_dynaddr_update(struct pfi_dynaddr *dyn) { struct pfi_kif *kif; struct pfr_ktable *kt; PF_RULES_WASSERT(); KASSERT(dyn && dyn->pfid_kif && dyn->pfid_kt, ("%s: bad argument", __func__)); kif = dyn->pfid_kif; kt = dyn->pfid_kt; if (kt->pfrkt_larg != V_pfi_update) { /* this table needs to be brought up-to-date */ pfi_table_update(kt, kif, dyn->pfid_net, dyn->pfid_iflags); kt->pfrkt_larg = V_pfi_update; } pfr_dynaddr_update(kt, dyn); } static void pfi_table_update(struct pfr_ktable *kt, struct pfi_kif *kif, int net, int flags) { int e, size2 = 0; struct ifg_member *ifgm; V_pfi_buffer_cnt = 0; if (kif->pfik_ifp != NULL) pfi_instance_add(kif->pfik_ifp, net, flags); else if (kif->pfik_group != NULL) { IFNET_RLOCK_NOSLEEP(); TAILQ_FOREACH(ifgm, &kif->pfik_group->ifg_members, ifgm_next) pfi_instance_add(ifgm->ifgm_ifp, net, flags); IFNET_RUNLOCK_NOSLEEP(); } if ((e = pfr_set_addrs(&kt->pfrkt_t, V_pfi_buffer, V_pfi_buffer_cnt, &size2, NULL, NULL, NULL, 0, PFR_TFLAG_ALLMASK))) printf("%s: cannot set %d new addresses into table %s: %d\n", __func__, V_pfi_buffer_cnt, kt->pfrkt_name, e); } static void pfi_instance_add(struct ifnet *ifp, int net, int flags) { struct ifaddr *ia; int got4 = 0, got6 = 0; int net2, af; IF_ADDR_RLOCK(ifp); TAILQ_FOREACH(ia, &ifp->if_addrhead, ifa_list) { if (ia->ifa_addr == NULL) continue; af = ia->ifa_addr->sa_family; if (af != AF_INET && af != AF_INET6) continue; /* * XXX: For point-to-point interfaces, (ifname:0) and IPv4, * jump over addresses without a proper route to work * around a problem with ppp not fully removing the * address used during IPCP. */ if ((ifp->if_flags & IFF_POINTOPOINT) && !(ia->ifa_flags & IFA_ROUTE) && (flags & PFI_AFLAG_NOALIAS) && (af == AF_INET)) continue; if ((flags & PFI_AFLAG_BROADCAST) && af == AF_INET6) continue; if ((flags & PFI_AFLAG_BROADCAST) && !(ifp->if_flags & IFF_BROADCAST)) continue; if ((flags & PFI_AFLAG_PEER) && !(ifp->if_flags & IFF_POINTOPOINT)) continue; if ((flags & PFI_AFLAG_NETWORK) && af == AF_INET6 && IN6_IS_ADDR_LINKLOCAL( &((struct sockaddr_in6 *)ia->ifa_addr)->sin6_addr)) continue; if (flags & PFI_AFLAG_NOALIAS) { if (af == AF_INET && got4) continue; if (af == AF_INET6 && got6) continue; } if (af == AF_INET) got4 = 1; else if (af == AF_INET6) got6 = 1; net2 = net; if (net2 == 128 && (flags & PFI_AFLAG_NETWORK)) { if (af == AF_INET) net2 = pfi_unmask(&((struct sockaddr_in *) ia->ifa_netmask)->sin_addr); else if (af == AF_INET6) net2 = pfi_unmask(&((struct sockaddr_in6 *) ia->ifa_netmask)->sin6_addr); } if (af == AF_INET && net2 > 32) net2 = 32; if (flags & PFI_AFLAG_BROADCAST) pfi_address_add(ia->ifa_broadaddr, af, net2); else if (flags & PFI_AFLAG_PEER) pfi_address_add(ia->ifa_dstaddr, af, net2); else pfi_address_add(ia->ifa_addr, af, net2); } IF_ADDR_RUNLOCK(ifp); } static void pfi_address_add(struct sockaddr *sa, int af, int net) { struct pfr_addr *p; int i; if (V_pfi_buffer_cnt >= V_pfi_buffer_max) { int new_max = V_pfi_buffer_max * 2; if (new_max > PFI_BUFFER_MAX) { printf("%s: address buffer full (%d/%d)\n", __func__, V_pfi_buffer_cnt, PFI_BUFFER_MAX); return; } p = malloc(new_max * sizeof(*V_pfi_buffer), PFI_MTYPE, M_NOWAIT); if (p == NULL) { printf("%s: no memory to grow buffer (%d/%d)\n", __func__, V_pfi_buffer_cnt, PFI_BUFFER_MAX); return; } memcpy(p, V_pfi_buffer, V_pfi_buffer_max * sizeof(*V_pfi_buffer)); /* no need to zero buffer */ free(V_pfi_buffer, PFI_MTYPE); V_pfi_buffer = p; V_pfi_buffer_max = new_max; } if (af == AF_INET && net > 32) net = 128; p = V_pfi_buffer + V_pfi_buffer_cnt++; bzero(p, sizeof(*p)); p->pfra_af = af; p->pfra_net = net; if (af == AF_INET) p->pfra_ip4addr = ((struct sockaddr_in *)sa)->sin_addr; else if (af == AF_INET6) { p->pfra_ip6addr = ((struct sockaddr_in6 *)sa)->sin6_addr; if (IN6_IS_SCOPE_EMBED(&p->pfra_ip6addr)) p->pfra_ip6addr.s6_addr16[1] = 0; } /* mask network address bits */ if (net < 128) ((caddr_t)p)[p->pfra_net/8] &= ~(0xFF >> (p->pfra_net%8)); for (i = (p->pfra_net+7)/8; i < sizeof(p->pfra_u); i++) ((caddr_t)p)[i] = 0; } void pfi_dynaddr_remove(struct pfi_dynaddr *dyn) { KASSERT(dyn->pfid_kif != NULL, ("%s: null pfid_kif", __func__)); KASSERT(dyn->pfid_kt != NULL, ("%s: null pfid_kt", __func__)); TAILQ_REMOVE(&dyn->pfid_kif->pfik_dynaddrs, dyn, entry); pfi_kif_unref(dyn->pfid_kif); pfr_detach_table(dyn->pfid_kt); free(dyn, PFI_MTYPE); } void pfi_dynaddr_copyout(struct pf_addr_wrap *aw) { KASSERT(aw->type == PF_ADDR_DYNIFTL, ("%s: type %u", __func__, aw->type)); if (aw->p.dyn == NULL || aw->p.dyn->pfid_kif == NULL) return; aw->p.dyncnt = aw->p.dyn->pfid_acnt4 + aw->p.dyn->pfid_acnt6; } static int pfi_if_compare(struct pfi_kif *p, struct pfi_kif *q) { return (strncmp(p->pfik_name, q->pfik_name, IFNAMSIZ)); } void pfi_update_status(const char *name, struct pf_status *pfs) { struct pfi_kif *p; struct pfi_kif_cmp key; struct ifg_member p_member, *ifgm; TAILQ_HEAD(, ifg_member) ifg_members; int i, j, k; strlcpy(key.pfik_name, name, sizeof(key.pfik_name)); p = RB_FIND(pfi_ifhead, &V_pfi_ifs, (struct pfi_kif *)&key); if (p == NULL) return; if (p->pfik_group != NULL) { bcopy(&p->pfik_group->ifg_members, &ifg_members, sizeof(ifg_members)); } else { /* build a temporary list for p only */ bzero(&p_member, sizeof(p_member)); p_member.ifgm_ifp = p->pfik_ifp; TAILQ_INIT(&ifg_members); TAILQ_INSERT_TAIL(&ifg_members, &p_member, ifgm_next); } if (pfs) { bzero(pfs->pcounters, sizeof(pfs->pcounters)); bzero(pfs->bcounters, sizeof(pfs->bcounters)); } TAILQ_FOREACH(ifgm, &ifg_members, ifgm_next) { if (ifgm->ifgm_ifp == NULL || ifgm->ifgm_ifp->if_pf_kif == NULL) continue; p = (struct pfi_kif *)ifgm->ifgm_ifp->if_pf_kif; /* just clear statistics */ if (pfs == NULL) { bzero(p->pfik_packets, sizeof(p->pfik_packets)); bzero(p->pfik_bytes, sizeof(p->pfik_bytes)); p->pfik_tzero = time_second; continue; } for (i = 0; i < 2; i++) for (j = 0; j < 2; j++) for (k = 0; k < 2; k++) { pfs->pcounters[i][j][k] += p->pfik_packets[i][j][k]; pfs->bcounters[i][j] += p->pfik_bytes[i][j][k]; } } } void pfi_get_ifaces(const char *name, struct pfi_kif *buf, int *size) { struct pfi_kif *p, *nextp; int n = 0; for (p = RB_MIN(pfi_ifhead, &V_pfi_ifs); p; p = nextp) { nextp = RB_NEXT(pfi_ifhead, &V_pfi_ifs, p); if (pfi_skip_if(name, p)) continue; if (*size <= n++) break; if (!p->pfik_tzero) p->pfik_tzero = time_second; bcopy(p, buf++, sizeof(*buf)); nextp = RB_NEXT(pfi_ifhead, &V_pfi_ifs, p); } *size = n; } static int pfi_skip_if(const char *filter, struct pfi_kif *p) { int n; if (filter == NULL || !*filter) return (0); if (!strcmp(p->pfik_name, filter)) return (0); /* exact match */ n = strlen(filter); if (n < 1 || n >= IFNAMSIZ) return (1); /* sanity check */ if (filter[n-1] >= '0' && filter[n-1] <= '9') return (1); /* only do exact match in that case */ if (strncmp(p->pfik_name, filter, n)) return (1); /* prefix doesn't match */ return (p->pfik_name[n] < '0' || p->pfik_name[n] > '9'); } int pfi_set_flags(const char *name, int flags) { struct pfi_kif *p; RB_FOREACH(p, pfi_ifhead, &V_pfi_ifs) { if (pfi_skip_if(name, p)) continue; p->pfik_flags |= flags; } return (0); } int pfi_clear_flags(const char *name, int flags) { struct pfi_kif *p; RB_FOREACH(p, pfi_ifhead, &V_pfi_ifs) { if (pfi_skip_if(name, p)) continue; p->pfik_flags &= ~flags; } return (0); } /* from pf_print_state.c */ static int pfi_unmask(void *addr) { struct pf_addr *m = addr; int i = 31, j = 0, b = 0; u_int32_t tmp; while (j < 4 && m->addr32[j] == 0xffffffff) { b += 32; j++; } if (j < 4) { tmp = ntohl(m->addr32[j]); for (i = 31; tmp & (1 << i); --i) b++; } return (b); } static void pfi_attach_ifnet_event(void *arg __unused, struct ifnet *ifp) { CURVNET_SET(ifp->if_vnet); if (V_pf_vnet_active == 0) { - /* Avoid teardown race in the least expensie way. */ + /* Avoid teardown race in the least expensive way. */ CURVNET_RESTORE(); return; } pfi_attach_ifnet(ifp); #ifdef ALTQ PF_RULES_WLOCK(); pf_altq_ifnet_event(ifp, 0); PF_RULES_WUNLOCK(); #endif CURVNET_RESTORE(); } static void pfi_detach_ifnet_event(void *arg __unused, struct ifnet *ifp) { struct pfi_kif *kif = (struct pfi_kif *)ifp->if_pf_kif; if (kif == NULL) return; CURVNET_SET(ifp->if_vnet); if (V_pf_vnet_active == 0) { - /* Avoid teardown race in the least expensie way. */ + /* Avoid teardown race in the least expensive way. */ CURVNET_RESTORE(); return; } PF_RULES_WLOCK(); V_pfi_update++; pfi_kif_update(kif); kif->pfik_ifp = NULL; ifp->if_pf_kif = NULL; #ifdef ALTQ pf_altq_ifnet_event(ifp, 1); #endif PF_RULES_WUNLOCK(); CURVNET_RESTORE(); } static void pfi_attach_group_event(void *arg , struct ifg_group *ifg) { CURVNET_SET((struct vnet *)arg); if (V_pf_vnet_active == 0) { - /* Avoid teardown race in the least expensie way. */ + /* Avoid teardown race in the least expensive way. */ CURVNET_RESTORE(); return; } pfi_attach_ifgroup(ifg); CURVNET_RESTORE(); } static void pfi_change_group_event(void *arg, char *gname) { struct pfi_kif *kif; CURVNET_SET((struct vnet *)arg); if (V_pf_vnet_active == 0) { - /* Avoid teardown race in the least expensie way. */ + /* Avoid teardown race in the least expensive way. */ CURVNET_RESTORE(); return; } kif = malloc(sizeof(*kif), PFI_MTYPE, M_WAITOK); PF_RULES_WLOCK(); V_pfi_update++; kif = pfi_kif_attach(kif, gname); pfi_kif_update(kif); PF_RULES_WUNLOCK(); CURVNET_RESTORE(); } static void pfi_detach_group_event(void *arg, struct ifg_group *ifg) { struct pfi_kif *kif = (struct pfi_kif *)ifg->ifg_pf_kif; if (kif == NULL) return; CURVNET_SET((struct vnet *)arg); if (V_pf_vnet_active == 0) { - /* Avoid teardown race in the least expensie way. */ + /* Avoid teardown race in the least expensive way. */ CURVNET_RESTORE(); return; } PF_RULES_WLOCK(); V_pfi_update++; kif->pfik_group = NULL; ifg->ifg_pf_kif = NULL; PF_RULES_WUNLOCK(); CURVNET_RESTORE(); } static void pfi_ifaddr_event(void *arg __unused, struct ifnet *ifp) { if (ifp->if_pf_kif == NULL) return; CURVNET_SET(ifp->if_vnet); if (V_pf_vnet_active == 0) { - /* Avoid teardown race in the least expensie way. */ + /* Avoid teardown race in the least expensive way. */ CURVNET_RESTORE(); return; } PF_RULES_WLOCK(); if (ifp && ifp->if_pf_kif) { V_pfi_update++; pfi_kif_update(ifp->if_pf_kif); } PF_RULES_WUNLOCK(); CURVNET_RESTORE(); } Index: projects/vnet/sys/sys/param.h =================================================================== --- projects/vnet/sys/sys/param.h (revision 302157) +++ projects/vnet/sys/sys/param.h (revision 302158) @@ -1,363 +1,363 @@ /*- * Copyright (c) 1982, 1986, 1989, 1993 * The Regents of the University of California. All rights reserved. * (c) UNIX System Laboratories, Inc. * All or some portions of this file are derived from material licensed * to the University of California by American Telephone and Telegraph * Co. or Unix System Laboratories, Inc. and are reproduced herein with * the permission of UNIX System Laboratories, 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. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)param.h 8.3 (Berkeley) 4/4/95 * $FreeBSD$ */ #ifndef _SYS_PARAM_H_ #define _SYS_PARAM_H_ #include #define BSD 199506 /* System version (year & month). */ #define BSD4_3 1 #define BSD4_4 1 /* * __FreeBSD_version numbers are documented in the Porter's Handbook. * If you bump the version for any reason, you should update the documentation * there. * Currently this lives here in the doc/ repository: * * head/en_US.ISO8859-1/books/porters-handbook/versions/chapter.xml * * scheme is: Rxx * 'R' is in the range 0 to 4 if this is a release branch or * x.0-CURRENT before RELENG_*_0 is created, otherwise 'R' is * in the range 5 to 9. */ #undef __FreeBSD_version -#define __FreeBSD_version 1100118 /* Master, propagated to newvers */ +#define __FreeBSD_version 1100120 /* Master, propagated to newvers */ /* * __FreeBSD_kernel__ indicates that this system uses the kernel of FreeBSD, * which by definition is always true on FreeBSD. This macro is also defined * on other systems that use the kernel of FreeBSD, such as GNU/kFreeBSD. * * It is tempting to use this macro in userland code when we want to enable * kernel-specific routines, and in fact it's fine to do this in code that * is part of FreeBSD itself. However, be aware that as presence of this * macro is still not widespread (e.g. older FreeBSD versions, 3rd party * compilers, etc), it is STRONGLY DISCOURAGED to check for this macro in * external applications without also checking for __FreeBSD__ as an * alternative. */ #undef __FreeBSD_kernel__ #define __FreeBSD_kernel__ #ifdef _KERNEL #define P_OSREL_SIGWAIT 700000 #define P_OSREL_SIGSEGV 700004 #define P_OSREL_MAP_ANON 800104 #define P_OSREL_MAP_FSTRICT 1100036 #define P_OSREL_SHUTDOWN_ENOTCONN 1100077 #define P_OSREL_MAJOR(x) ((x) / 100000) #endif #ifndef LOCORE #include #endif /* * Machine-independent constants (some used in following include files). * Redefined constants are from POSIX 1003.1 limits file. * * MAXCOMLEN should be >= sizeof(ac_comm) (see ) */ #include #define MAXCOMLEN 19 /* max command name remembered */ #define MAXINTERP PATH_MAX /* max interpreter file name length */ #define MAXLOGNAME 33 /* max login name length (incl. NUL) */ #define MAXUPRC CHILD_MAX /* max simultaneous processes */ #define NCARGS ARG_MAX /* max bytes for an exec function */ #define NGROUPS (NGROUPS_MAX+1) /* max number groups */ #define NOFILE OPEN_MAX /* max open files per process */ #define NOGROUP 65535 /* marker for empty group set member */ #define MAXHOSTNAMELEN 256 /* max hostname size */ #define SPECNAMELEN 63 /* max length of devicename */ /* More types and definitions used throughout the kernel. */ #ifdef _KERNEL #include #include #ifndef LOCORE #include #include #endif #ifndef FALSE #define FALSE 0 #endif #ifndef TRUE #define TRUE 1 #endif #endif #ifndef _KERNEL /* Signals. */ #include #endif /* Machine type dependent parameters. */ #include #ifndef _KERNEL #include #endif #ifndef DEV_BSHIFT #define DEV_BSHIFT 9 /* log2(DEV_BSIZE) */ #endif #define DEV_BSIZE (1<>PAGE_SHIFT) #endif /* * btodb() is messy and perhaps slow because `bytes' may be an off_t. We * want to shift an unsigned type to avoid sign extension and we don't * want to widen `bytes' unnecessarily. Assume that the result fits in * a daddr_t. */ #ifndef btodb #define btodb(bytes) /* calculates (bytes / DEV_BSIZE) */ \ (sizeof (bytes) > sizeof(long) \ ? (daddr_t)((unsigned long long)(bytes) >> DEV_BSHIFT) \ : (daddr_t)((unsigned long)(bytes) >> DEV_BSHIFT)) #endif #ifndef dbtob #define dbtob(db) /* calculates (db * DEV_BSIZE) */ \ ((off_t)(db) << DEV_BSHIFT) #endif #define PRIMASK 0x0ff #define PCATCH 0x100 /* OR'd with pri for tsleep to check signals */ #define PDROP 0x200 /* OR'd with pri to stop re-entry of interlock mutex */ #define NZERO 0 /* default "nice" */ #define NBBY 8 /* number of bits in a byte */ #define NBPW sizeof(int) /* number of bytes per word (integer) */ #define CMASK 022 /* default file mask: S_IWGRP|S_IWOTH */ #define NODEV (dev_t)(-1) /* non-existent device */ /* * File system parameters and macros. * * MAXBSIZE - Filesystems are made out of blocks of at most MAXBSIZE bytes * per block. MAXBSIZE may be made larger without effecting * any existing filesystems as long as it does not exceed MAXPHYS, * and may be made smaller at the risk of not being able to use * filesystems which require a block size exceeding MAXBSIZE. * * MAXBCACHEBUF - Maximum size of a buffer in the buffer cache. This must * be >= MAXBSIZE and can be set differently for different * architectures by defining it in . * Making this larger allows NFS to do larger reads/writes. * * BKVASIZE - Nominal buffer space per buffer, in bytes. BKVASIZE is the * minimum KVM memory reservation the kernel is willing to make. * Filesystems can of course request smaller chunks. Actual * backing memory uses a chunk size of a page (PAGE_SIZE). * The default value here can be overridden on a per-architecture * basis by defining it in . This should * probably be done to increase its value, when MAXBCACHEBUF is * defined as a larger value in . * * If you make BKVASIZE too small you risk seriously fragmenting * the buffer KVM map which may slow things down a bit. If you * make it too big the kernel will not be able to optimally use * the KVM memory reserved for the buffer cache and will wind * up with too-few buffers. * * The default is 16384, roughly 2x the block size used by a * normal UFS filesystem. */ #define MAXBSIZE 65536 /* must be power of 2 */ #ifndef MAXBCACHEBUF #define MAXBCACHEBUF MAXBSIZE /* must be a power of 2 >= MAXBSIZE */ #endif #ifndef BKVASIZE #define BKVASIZE 16384 /* must be power of 2 */ #endif #define BKVAMASK (BKVASIZE-1) /* * MAXPATHLEN defines the longest permissible path length after expanding * symbolic links. It is used to allocate a temporary buffer from the buffer * pool in which to do the name expansion, hence should be a power of two, * and must be less than or equal to MAXBSIZE. MAXSYMLINKS defines the * maximum number of symbolic links that may be expanded in a path name. * It should be set high enough to allow all legitimate uses, but halt * infinite loops reasonably quickly. */ #define MAXPATHLEN PATH_MAX #define MAXSYMLINKS 32 /* Bit map related macros. */ #define setbit(a,i) (((unsigned char *)(a))[(i)/NBBY] |= 1<<((i)%NBBY)) #define clrbit(a,i) (((unsigned char *)(a))[(i)/NBBY] &= ~(1<<((i)%NBBY))) #define isset(a,i) \ (((const unsigned char *)(a))[(i)/NBBY] & (1<<((i)%NBBY))) #define isclr(a,i) \ ((((const unsigned char *)(a))[(i)/NBBY] & (1<<((i)%NBBY))) == 0) /* Macros for counting and rounding. */ #ifndef howmany #define howmany(x, y) (((x)+((y)-1))/(y)) #endif #define nitems(x) (sizeof((x)) / sizeof((x)[0])) #define rounddown(x, y) (((x)/(y))*(y)) #define rounddown2(x, y) ((x)&(~((y)-1))) /* if y is power of two */ #define roundup(x, y) ((((x)+((y)-1))/(y))*(y)) /* to any y */ #define roundup2(x, y) (((x)+((y)-1))&(~((y)-1))) /* if y is powers of two */ #define powerof2(x) ((((x)-1)&(x))==0) /* Macros for min/max. */ #define MIN(a,b) (((a)<(b))?(a):(b)) #define MAX(a,b) (((a)>(b))?(a):(b)) #ifdef _KERNEL /* * Basic byte order function prototypes for non-inline functions. */ #ifndef LOCORE #ifndef _BYTEORDER_PROTOTYPED #define _BYTEORDER_PROTOTYPED __BEGIN_DECLS __uint32_t htonl(__uint32_t); __uint16_t htons(__uint16_t); __uint32_t ntohl(__uint32_t); __uint16_t ntohs(__uint16_t); __END_DECLS #endif #endif #ifndef lint #ifndef _BYTEORDER_FUNC_DEFINED #define _BYTEORDER_FUNC_DEFINED #define htonl(x) __htonl(x) #define htons(x) __htons(x) #define ntohl(x) __ntohl(x) #define ntohs(x) __ntohs(x) #endif /* !_BYTEORDER_FUNC_DEFINED */ #endif /* lint */ #endif /* _KERNEL */ /* * Scale factor for scaled integers used to count %cpu time and load avgs. * * The number of CPU `tick's that map to a unique `%age' can be expressed * by the formula (1 / (2 ^ (FSHIFT - 11))). The maximum load average that * can be calculated (assuming 32 bits) can be closely approximated using * the formula (2 ^ (2 * (16 - FSHIFT))) for (FSHIFT < 15). * * For the scheduler to maintain a 1:1 mapping of CPU `tick' to `%age', * FSHIFT must be at least 11; this gives us a maximum load avg of ~1024. */ #define FSHIFT 11 /* bits to right of fixed binary point */ #define FSCALE (1<> (PAGE_SHIFT - DEV_BSHIFT)) #define ctodb(db) /* calculates pages to devblks */ \ ((db) << (PAGE_SHIFT - DEV_BSHIFT)) /* * Old spelling of __containerof(). */ #define member2struct(s, m, x) \ ((struct s *)(void *)((char *)(x) - offsetof(struct s, m))) /* * Access a variable length array that has been declared as a fixed * length array. */ #define __PAST_END(array, offset) (((__typeof__(*(array)) *)(array))[offset]) #endif /* _SYS_PARAM_H_ */ Index: projects/vnet/sys/sys/socketvar.h =================================================================== --- projects/vnet/sys/sys/socketvar.h (revision 302157) +++ projects/vnet/sys/sys/socketvar.h (revision 302158) @@ -1,419 +1,422 @@ /*- * Copyright (c) 1982, 1986, 1990, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)socketvar.h 8.3 (Berkeley) 2/19/95 * * $FreeBSD$ */ #ifndef _SYS_SOCKETVAR_H_ #define _SYS_SOCKETVAR_H_ #include /* for TAILQ macros */ #include /* for struct selinfo */ #include #include #include #include #include #include #ifdef _KERNEL #include #include #endif struct vnet; /* * Kernel structure per socket. * Contains send and receive buffer queues, * handle on protocol and pointer to protocol * private data and error information. */ typedef u_quad_t so_gen_t; struct socket; /*- * Locking key to struct socket: * (a) constant after allocation, no locking required. * (b) locked by SOCK_LOCK(so). * (c) locked by SOCKBUF_LOCK(&so->so_rcv). * (e) locked by ACCEPT_LOCK(). * (f) not locked since integer reads/writes are atomic. * (g) used only as a sleep/wakeup address, no value. * (h) locked by global mutex so_global_mtx. */ struct socket { int so_count; /* (b) reference count */ short so_type; /* (a) generic type, see socket.h */ short so_options; /* from socket call, see socket.h */ short so_linger; /* time to linger while closing */ short so_state; /* (b) internal state flags SS_* */ int so_qstate; /* (e) internal state flags SQ_* */ void *so_pcb; /* protocol control block */ struct vnet *so_vnet; /* (a) network stack instance */ struct protosw *so_proto; /* (a) protocol handle */ /* * Variables for connection queuing. * Socket where accepts occur is so_head in all subsidiary sockets. * If so_head is 0, socket is not related to an accept. * For head socket so_incomp queues partially completed connections, * while so_comp is a queue of connections ready to be accepted. * If a connection is aborted and it has so_head set, then * it has to be pulled out of either so_incomp or so_comp. * We allow connections to queue up based on current queue lengths * and limit on number of queued connections for this socket. */ struct socket *so_head; /* (e) back pointer to listen socket */ TAILQ_HEAD(, socket) so_incomp; /* (e) queue of partial unaccepted connections */ TAILQ_HEAD(, socket) so_comp; /* (e) queue of complete unaccepted connections */ TAILQ_ENTRY(socket) so_list; /* (e) list of unaccepted connections */ u_int so_qlen; /* (e) number of unaccepted connections */ u_int so_incqlen; /* (e) number of unaccepted incomplete connections */ u_int so_qlimit; /* (e) max number queued connections */ short so_timeo; /* (g) connection timeout */ u_short so_error; /* (f) error affecting connection */ struct sigio *so_sigio; /* [sg] information for async I/O or out of band data (SIGURG) */ u_long so_oobmark; /* (c) chars to oob mark */ struct sockbuf so_rcv, so_snd; struct ucred *so_cred; /* (a) user credentials */ struct label *so_label; /* (b) MAC label for socket */ struct label *so_peerlabel; /* (b) cached MAC label for peer */ /* NB: generation count must not be first. */ so_gen_t so_gencnt; /* (h) generation count */ void *so_emuldata; /* (b) private data for emulators */ struct so_accf { struct accept_filter *so_accept_filter; void *so_accept_filter_arg; /* saved filter args */ char *so_accept_filter_str; /* saved user args */ } *so_accf; struct osd osd; /* Object Specific extensions */ /* * so_fibnum, so_user_cookie and friends can be used to attach * some user-specified metadata to a socket, which then can be * used by the kernel for various actions. * so_user_cookie is used by ipfw/dummynet. */ int so_fibnum; /* routing domain for this socket */ uint32_t so_user_cookie; + + void *so_pspare[2]; /* packet pacing / general use */ + int so_ispare[2]; /* packet pacing / general use */ }; /* * Global accept mutex to serialize access to accept queues and * fields associated with multiple sockets. This allows us to * avoid defining a lock order between listen and accept sockets * until such time as it proves to be a good idea. */ extern struct mtx accept_mtx; #define ACCEPT_LOCK_ASSERT() mtx_assert(&accept_mtx, MA_OWNED) #define ACCEPT_UNLOCK_ASSERT() mtx_assert(&accept_mtx, MA_NOTOWNED) #define ACCEPT_LOCK() mtx_lock(&accept_mtx) #define ACCEPT_UNLOCK() mtx_unlock(&accept_mtx) /* * Per-socket mutex: we reuse the receive socket buffer mutex for space * efficiency. This decision should probably be revisited as we optimize * locking for the socket code. */ #define SOCK_MTX(_so) SOCKBUF_MTX(&(_so)->so_rcv) #define SOCK_LOCK(_so) SOCKBUF_LOCK(&(_so)->so_rcv) #define SOCK_OWNED(_so) SOCKBUF_OWNED(&(_so)->so_rcv) #define SOCK_UNLOCK(_so) SOCKBUF_UNLOCK(&(_so)->so_rcv) #define SOCK_LOCK_ASSERT(_so) SOCKBUF_LOCK_ASSERT(&(_so)->so_rcv) /* * Socket state bits stored in so_qstate. */ #define SQ_INCOMP 0x0800 /* unaccepted, incomplete connection */ #define SQ_COMP 0x1000 /* unaccepted, complete connection */ /* * Externalized form of struct socket used by the sysctl(3) interface. */ struct xsocket { size_t xso_len; /* length of this structure */ struct socket *xso_so; /* makes a convenient handle sometimes */ short so_type; short so_options; short so_linger; short so_state; caddr_t so_pcb; /* another convenient handle */ int xso_protocol; int xso_family; u_int so_qlen; u_int so_incqlen; u_int so_qlimit; short so_timeo; u_short so_error; pid_t so_pgid; u_long so_oobmark; struct xsockbuf so_rcv, so_snd; uid_t so_uid; /* XXX */ }; #ifdef _KERNEL /* * Macros for sockets and socket buffering. */ /* * Flags to sblock(). */ #define SBL_WAIT 0x00000001 /* Wait if not immediately available. */ #define SBL_NOINTR 0x00000002 /* Force non-interruptible sleep. */ #define SBL_VALID (SBL_WAIT | SBL_NOINTR) /* * Do we need to notify the other side when I/O is possible? */ #define sb_notify(sb) (((sb)->sb_flags & (SB_WAIT | SB_SEL | SB_ASYNC | \ SB_UPCALL | SB_AIO | SB_KNOTE)) != 0) /* do we have to send all at once on a socket? */ #define sosendallatonce(so) \ ((so)->so_proto->pr_flags & PR_ATOMIC) /* can we read something from so? */ #define soreadabledata(so) \ (sbavail(&(so)->so_rcv) >= (so)->so_rcv.sb_lowat || \ !TAILQ_EMPTY(&(so)->so_comp) || (so)->so_error) #define soreadable(so) \ (soreadabledata(so) || ((so)->so_rcv.sb_state & SBS_CANTRCVMORE)) /* can we write something to so? */ #define sowriteable(so) \ ((sbspace(&(so)->so_snd) >= (so)->so_snd.sb_lowat && \ (((so)->so_state&SS_ISCONNECTED) || \ ((so)->so_proto->pr_flags&PR_CONNREQUIRED)==0)) || \ ((so)->so_snd.sb_state & SBS_CANTSENDMORE) || \ (so)->so_error) /* * soref()/sorele() ref-count the socket structure. Note that you must * still explicitly close the socket, but the last ref count will free * the structure. */ #define soref(so) do { \ SOCK_LOCK_ASSERT(so); \ ++(so)->so_count; \ } while (0) #define sorele(so) do { \ ACCEPT_LOCK_ASSERT(); \ SOCK_LOCK_ASSERT(so); \ if ((so)->so_count <= 0) \ panic("sorele"); \ if (--(so)->so_count == 0) \ sofree(so); \ else { \ SOCK_UNLOCK(so); \ ACCEPT_UNLOCK(); \ } \ } while (0) /* * In sorwakeup() and sowwakeup(), acquire the socket buffer lock to * avoid a non-atomic test-and-wakeup. However, sowakeup is * responsible for releasing the lock if it is called. We unlock only * if we don't call into sowakeup. If any code is introduced that * directly invokes the underlying sowakeup() primitives, it must * maintain the same semantics. */ #define sorwakeup_locked(so) do { \ SOCKBUF_LOCK_ASSERT(&(so)->so_rcv); \ if (sb_notify(&(so)->so_rcv)) \ sowakeup((so), &(so)->so_rcv); \ else \ SOCKBUF_UNLOCK(&(so)->so_rcv); \ } while (0) #define sorwakeup(so) do { \ SOCKBUF_LOCK(&(so)->so_rcv); \ sorwakeup_locked(so); \ } while (0) #define sowwakeup_locked(so) do { \ SOCKBUF_LOCK_ASSERT(&(so)->so_snd); \ if (sb_notify(&(so)->so_snd)) \ sowakeup((so), &(so)->so_snd); \ else \ SOCKBUF_UNLOCK(&(so)->so_snd); \ } while (0) #define sowwakeup(so) do { \ SOCKBUF_LOCK(&(so)->so_snd); \ sowwakeup_locked(so); \ } while (0) struct accept_filter { char accf_name[16]; int (*accf_callback) (struct socket *so, void *arg, int waitflag); void * (*accf_create) (struct socket *so, char *arg); void (*accf_destroy) (struct socket *so); SLIST_ENTRY(accept_filter) accf_next; }; #ifdef MALLOC_DECLARE MALLOC_DECLARE(M_ACCF); MALLOC_DECLARE(M_PCB); MALLOC_DECLARE(M_SONAME); #endif /* * Socket specific helper hook point identifiers * Do not leave holes in the sequence, hook registration is a loop. */ #define HHOOK_SOCKET_OPT 0 #define HHOOK_SOCKET_CREATE 1 #define HHOOK_SOCKET_RCV 2 #define HHOOK_SOCKET_SND 3 #define HHOOK_FILT_SOREAD 4 #define HHOOK_FILT_SOWRITE 5 #define HHOOK_SOCKET_CLOSE 6 #define HHOOK_SOCKET_LAST HHOOK_SOCKET_CLOSE struct socket_hhook_data { struct socket *so; struct mbuf *m; void *hctx; /* hook point specific data*/ int status; }; extern int maxsockets; extern u_long sb_max; extern so_gen_t so_gencnt; struct file; struct filedesc; struct mbuf; struct sockaddr; struct ucred; struct uio; /* 'which' values for socket upcalls. */ #define SO_RCV 1 #define SO_SND 2 /* Return values for socket upcalls. */ #define SU_OK 0 #define SU_ISCONNECTED 1 /* * From uipc_socket and friends */ int getsockaddr(struct sockaddr **namp, caddr_t uaddr, size_t len); int getsock_cap(struct thread *td, int fd, cap_rights_t *rightsp, struct file **fpp, u_int *fflagp); void soabort(struct socket *so); int soaccept(struct socket *so, struct sockaddr **nam); void soaio_enqueue(struct task *task); void soaio_rcv(void *context, int pending); void soaio_snd(void *context, int pending); int socheckuid(struct socket *so, uid_t uid); int sobind(struct socket *so, struct sockaddr *nam, struct thread *td); int sobindat(int fd, struct socket *so, struct sockaddr *nam, struct thread *td); int soclose(struct socket *so); int soconnect(struct socket *so, struct sockaddr *nam, struct thread *td); int soconnectat(int fd, struct socket *so, struct sockaddr *nam, struct thread *td); int soconnect2(struct socket *so1, struct socket *so2); int socreate(int dom, struct socket **aso, int type, int proto, struct ucred *cred, struct thread *td); int sodisconnect(struct socket *so); struct sockaddr *sodupsockaddr(const struct sockaddr *sa, int mflags); void sofree(struct socket *so); void sohasoutofband(struct socket *so); int solisten(struct socket *so, int backlog, struct thread *td); void solisten_proto(struct socket *so, int backlog); int solisten_proto_check(struct socket *so); struct socket * sonewconn(struct socket *head, int connstatus); int sopoll(struct socket *so, int events, struct ucred *active_cred, struct thread *td); int sopoll_generic(struct socket *so, int events, struct ucred *active_cred, struct thread *td); int soreceive(struct socket *so, struct sockaddr **paddr, struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, int *flagsp); int soreceive_stream(struct socket *so, struct sockaddr **paddr, struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, int *flagsp); int soreceive_dgram(struct socket *so, struct sockaddr **paddr, struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, int *flagsp); int soreceive_generic(struct socket *so, struct sockaddr **paddr, struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, int *flagsp); int soreserve(struct socket *so, u_long sndcc, u_long rcvcc); void sorflush(struct socket *so); int sosend(struct socket *so, struct sockaddr *addr, struct uio *uio, struct mbuf *top, struct mbuf *control, int flags, struct thread *td); int sosend_dgram(struct socket *so, struct sockaddr *addr, struct uio *uio, struct mbuf *top, struct mbuf *control, int flags, struct thread *td); int sosend_generic(struct socket *so, struct sockaddr *addr, struct uio *uio, struct mbuf *top, struct mbuf *control, int flags, struct thread *td); int soshutdown(struct socket *so, int how); void sotoxsocket(struct socket *so, struct xsocket *xso); void soupcall_clear(struct socket *so, int which); void soupcall_set(struct socket *so, int which, int (*func)(struct socket *, void *, int), void *arg); void sowakeup(struct socket *so, struct sockbuf *sb); void sowakeup_aio(struct socket *so, struct sockbuf *sb); int selsocket(struct socket *so, int events, struct timeval *tv, struct thread *td); /* * Accept filter functions (duh). */ int accept_filt_add(struct accept_filter *filt); int accept_filt_del(char *name); struct accept_filter *accept_filt_get(char *name); #ifdef ACCEPT_FILTER_MOD #ifdef SYSCTL_DECL SYSCTL_DECL(_net_inet_accf); #endif int accept_filt_generic_mod_event(module_t mod, int event, void *data); #endif #endif /* _KERNEL */ #endif /* !_SYS_SOCKETVAR_H_ */ Index: projects/vnet/sys/vm/vm_page.c =================================================================== --- projects/vnet/sys/vm/vm_page.c (revision 302157) +++ projects/vnet/sys/vm/vm_page.c (revision 302158) @@ -1,3940 +1,3958 @@ /*- * Copyright (c) 1991 Regents of the University of California. * All rights reserved. * Copyright (c) 1998 Matthew Dillon. All Rights Reserved. * * This code is derived from software contributed to Berkeley by * The Mach Operating System project at Carnegie-Mellon University. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * from: @(#)vm_page.c 7.4 (Berkeley) 5/7/91 */ /*- * Copyright (c) 1987, 1990 Carnegie-Mellon University. * All rights reserved. * * Authors: Avadis Tevanian, Jr., Michael Wayne Young * * Permission to use, copy, modify and distribute this software and * its documentation is hereby granted, provided that both the copyright * notice and this permission notice appear in all copies of the * software, derivative works or modified versions, and any portions * thereof, and that both notices appear in supporting documentation. * * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. * * Carnegie Mellon requests users of this software to return to * * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU * School of Computer Science * Carnegie Mellon University * Pittsburgh PA 15213-3890 * * any improvements or extensions that they make and grant Carnegie the * rights to redistribute these changes. */ /* * GENERAL RULES ON VM_PAGE MANIPULATION * * - A page queue lock is required when adding or removing a page from a * page queue regardless of other locks or the busy state of a page. * * * In general, no thread besides the page daemon can acquire or * hold more than one page queue lock at a time. * * * The page daemon can acquire and hold any pair of page queue * locks in any order. * * - The object lock is required when inserting or removing * pages from an object (vm_page_insert() or vm_page_remove()). * */ /* * Resident memory management module. */ #include __FBSDID("$FreeBSD$"); #include "opt_vm.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 /* * Associated with page of user-allocatable memory is a * page structure. */ struct vm_domain vm_dom[MAXMEMDOM]; struct mtx_padalign vm_page_queue_free_mtx; struct mtx_padalign pa_lock[PA_LOCK_COUNT]; vm_page_t vm_page_array; long vm_page_array_size; long first_page; int vm_page_zero_count; static int boot_pages = UMA_BOOT_PAGES; SYSCTL_INT(_vm, OID_AUTO, boot_pages, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &boot_pages, 0, "number of pages allocated for bootstrapping the VM system"); static int pa_tryrelock_restart; SYSCTL_INT(_vm, OID_AUTO, tryrelock_restart, CTLFLAG_RD, &pa_tryrelock_restart, 0, "Number of tryrelock restarts"); static TAILQ_HEAD(, vm_page) blacklist_head; static int sysctl_vm_page_blacklist(SYSCTL_HANDLER_ARGS); SYSCTL_PROC(_vm, OID_AUTO, page_blacklist, CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, 0, sysctl_vm_page_blacklist, "A", "Blacklist pages"); /* Is the page daemon waiting for free pages? */ static int vm_pageout_pages_needed; static uma_zone_t fakepg_zone; static struct vnode *vm_page_alloc_init(vm_page_t m); static void vm_page_cache_turn_free(vm_page_t m); static void vm_page_clear_dirty_mask(vm_page_t m, vm_page_bits_t pagebits); static void vm_page_enqueue(uint8_t queue, vm_page_t m); static void vm_page_free_wakeup(void); static void vm_page_init_fakepg(void *dummy); static int vm_page_insert_after(vm_page_t m, vm_object_t object, vm_pindex_t pindex, vm_page_t mpred); static void vm_page_insert_radixdone(vm_page_t m, vm_object_t object, vm_page_t mpred); static int vm_page_reclaim_run(int req_class, u_long npages, vm_page_t m_run, vm_paddr_t high); SYSINIT(vm_page, SI_SUB_VM, SI_ORDER_SECOND, vm_page_init_fakepg, NULL); static void vm_page_init_fakepg(void *dummy) { fakepg_zone = uma_zcreate("fakepg", sizeof(struct vm_page), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE | UMA_ZONE_VM); } /* Make sure that u_long is at least 64 bits when PAGE_SIZE is 32K. */ #if PAGE_SIZE == 32768 #ifdef CTASSERT CTASSERT(sizeof(u_long) >= 8); #endif #endif /* * Try to acquire a physical address lock while a pmap is locked. If we * fail to trylock we unlock and lock the pmap directly and cache the * locked pa in *locked. The caller should then restart their loop in case * the virtual to physical mapping has changed. */ int vm_page_pa_tryrelock(pmap_t pmap, vm_paddr_t pa, vm_paddr_t *locked) { vm_paddr_t lockpa; lockpa = *locked; *locked = pa; if (lockpa) { PA_LOCK_ASSERT(lockpa, MA_OWNED); if (PA_LOCKPTR(pa) == PA_LOCKPTR(lockpa)) return (0); PA_UNLOCK(lockpa); } if (PA_TRYLOCK(pa)) return (0); PMAP_UNLOCK(pmap); atomic_add_int(&pa_tryrelock_restart, 1); PA_LOCK(pa); PMAP_LOCK(pmap); return (EAGAIN); } /* * vm_set_page_size: * * Sets the page size, perhaps based upon the memory * size. Must be called before any use of page-size * dependent functions. */ void vm_set_page_size(void) { if (vm_cnt.v_page_size == 0) vm_cnt.v_page_size = PAGE_SIZE; if (((vm_cnt.v_page_size - 1) & vm_cnt.v_page_size) != 0) panic("vm_set_page_size: page size not a power of two"); } /* * vm_page_blacklist_next: * * Find the next entry in the provided string of blacklist * addresses. Entries are separated by space, comma, or newline. * If an invalid integer is encountered then the rest of the * string is skipped. Updates the list pointer to the next * character, or NULL if the string is exhausted or invalid. */ static vm_paddr_t vm_page_blacklist_next(char **list, char *end) { vm_paddr_t bad; char *cp, *pos; if (list == NULL || *list == NULL) return (0); if (**list =='\0') { *list = NULL; return (0); } /* * If there's no end pointer then the buffer is coming from * the kenv and we know it's null-terminated. */ if (end == NULL) end = *list + strlen(*list); /* Ensure that strtoq() won't walk off the end */ if (*end != '\0') { if (*end == '\n' || *end == ' ' || *end == ',') *end = '\0'; else { printf("Blacklist not terminated, skipping\n"); *list = NULL; return (0); } } for (pos = *list; *pos != '\0'; pos = cp) { bad = strtoq(pos, &cp, 0); if (*cp == '\0' || *cp == ' ' || *cp == ',' || *cp == '\n') { if (bad == 0) { if (++cp < end) continue; else break; } } else break; if (*cp == '\0' || ++cp >= end) *list = NULL; else *list = cp; return (trunc_page(bad)); } printf("Garbage in RAM blacklist, skipping\n"); *list = NULL; return (0); } /* * vm_page_blacklist_check: * * Iterate through the provided string of blacklist addresses, pulling * each entry out of the physical allocator free list and putting it * onto a list for reporting via the vm.page_blacklist sysctl. */ static void vm_page_blacklist_check(char *list, char *end) { vm_paddr_t pa; vm_page_t m; char *next; int ret; next = list; while (next != NULL) { if ((pa = vm_page_blacklist_next(&next, end)) == 0) continue; m = vm_phys_paddr_to_vm_page(pa); if (m == NULL) continue; mtx_lock(&vm_page_queue_free_mtx); ret = vm_phys_unfree_page(m); mtx_unlock(&vm_page_queue_free_mtx); if (ret == TRUE) { TAILQ_INSERT_TAIL(&blacklist_head, m, listq); if (bootverbose) printf("Skipping page with pa 0x%jx\n", (uintmax_t)pa); } } } /* * vm_page_blacklist_load: * * Search for a special module named "ram_blacklist". It'll be a * plain text file provided by the user via the loader directive * of the same name. */ static void vm_page_blacklist_load(char **list, char **end) { void *mod; u_char *ptr; u_int len; mod = NULL; ptr = NULL; mod = preload_search_by_type("ram_blacklist"); if (mod != NULL) { ptr = preload_fetch_addr(mod); len = preload_fetch_size(mod); } *list = ptr; if (ptr != NULL) *end = ptr + len; else *end = NULL; return; } static int sysctl_vm_page_blacklist(SYSCTL_HANDLER_ARGS) { vm_page_t m; struct sbuf sbuf; int error, first; first = 1; error = sysctl_wire_old_buffer(req, 0); if (error != 0) return (error); sbuf_new_for_sysctl(&sbuf, NULL, 128, req); TAILQ_FOREACH(m, &blacklist_head, listq) { sbuf_printf(&sbuf, "%s%#jx", first ? "" : ",", (uintmax_t)m->phys_addr); first = 0; } error = sbuf_finish(&sbuf); sbuf_delete(&sbuf); return (error); } static void vm_page_domain_init(struct vm_domain *vmd) { struct vm_pagequeue *pq; int i; *__DECONST(char **, &vmd->vmd_pagequeues[PQ_INACTIVE].pq_name) = "vm inactive pagequeue"; *__DECONST(u_int **, &vmd->vmd_pagequeues[PQ_INACTIVE].pq_vcnt) = &vm_cnt.v_inactive_count; *__DECONST(char **, &vmd->vmd_pagequeues[PQ_ACTIVE].pq_name) = "vm active pagequeue"; *__DECONST(u_int **, &vmd->vmd_pagequeues[PQ_ACTIVE].pq_vcnt) = &vm_cnt.v_active_count; vmd->vmd_page_count = 0; vmd->vmd_free_count = 0; vmd->vmd_segs = 0; vmd->vmd_oom = FALSE; vmd->vmd_pass = 0; for (i = 0; i < PQ_COUNT; i++) { pq = &vmd->vmd_pagequeues[i]; TAILQ_INIT(&pq->pq_pl); mtx_init(&pq->pq_mutex, pq->pq_name, "vm pagequeue", MTX_DEF | MTX_DUPOK); } } /* * vm_page_startup: * * Initializes the resident memory module. * * Allocates memory for the page cells, and * for the object/offset-to-page hash table headers. * Each page cell is initialized and placed on the free list. */ vm_offset_t vm_page_startup(vm_offset_t vaddr) { vm_offset_t mapped; vm_paddr_t page_range; vm_paddr_t new_end; int i; vm_paddr_t pa; vm_paddr_t last_pa; char *list, *listend; vm_paddr_t end; vm_paddr_t biggestsize; vm_paddr_t low_water, high_water; int biggestone; biggestsize = 0; biggestone = 0; vaddr = round_page(vaddr); for (i = 0; phys_avail[i + 1]; i += 2) { phys_avail[i] = round_page(phys_avail[i]); phys_avail[i + 1] = trunc_page(phys_avail[i + 1]); } low_water = phys_avail[0]; high_water = phys_avail[1]; for (i = 0; i < vm_phys_nsegs; i++) { if (vm_phys_segs[i].start < low_water) low_water = vm_phys_segs[i].start; if (vm_phys_segs[i].end > high_water) high_water = vm_phys_segs[i].end; } for (i = 0; phys_avail[i + 1]; i += 2) { vm_paddr_t size = phys_avail[i + 1] - phys_avail[i]; if (size > biggestsize) { biggestone = i; biggestsize = size; } if (phys_avail[i] < low_water) low_water = phys_avail[i]; if (phys_avail[i + 1] > high_water) high_water = phys_avail[i + 1]; } end = phys_avail[biggestone+1]; /* * Initialize the page and queue locks. */ mtx_init(&vm_page_queue_free_mtx, "vm page free queue", NULL, MTX_DEF); for (i = 0; i < PA_LOCK_COUNT; i++) mtx_init(&pa_lock[i], "vm page", NULL, MTX_DEF); for (i = 0; i < vm_ndomains; i++) vm_page_domain_init(&vm_dom[i]); /* * Allocate memory for use when boot strapping the kernel memory * allocator. * * CTFLAG_RDTUN doesn't work during the early boot process, so we must * manually fetch the value. */ TUNABLE_INT_FETCH("vm.boot_pages", &boot_pages); new_end = end - (boot_pages * UMA_SLAB_SIZE); new_end = trunc_page(new_end); mapped = pmap_map(&vaddr, new_end, end, VM_PROT_READ | VM_PROT_WRITE); bzero((void *)mapped, end - new_end); uma_startup((void *)mapped, boot_pages); #if defined(__aarch64__) || defined(__amd64__) || defined(__arm__) || \ defined(__i386__) || defined(__mips__) /* * Allocate a bitmap to indicate that a random physical page * needs to be included in a minidump. * * The amd64 port needs this to indicate which direct map pages * need to be dumped, via calls to dump_add_page()/dump_drop_page(). * * However, i386 still needs this workspace internally within the * minidump code. In theory, they are not needed on i386, but are * included should the sf_buf code decide to use them. */ last_pa = 0; for (i = 0; dump_avail[i + 1] != 0; i += 2) if (dump_avail[i + 1] > last_pa) last_pa = dump_avail[i + 1]; page_range = last_pa / PAGE_SIZE; vm_page_dump_size = round_page(roundup2(page_range, NBBY) / NBBY); new_end -= vm_page_dump_size; vm_page_dump = (void *)(uintptr_t)pmap_map(&vaddr, new_end, new_end + vm_page_dump_size, VM_PROT_READ | VM_PROT_WRITE); bzero((void *)vm_page_dump, vm_page_dump_size); #endif #ifdef __amd64__ /* * Request that the physical pages underlying the message buffer be * included in a crash dump. Since the message buffer is accessed * through the direct map, they are not automatically included. */ pa = DMAP_TO_PHYS((vm_offset_t)msgbufp->msg_ptr); last_pa = pa + round_page(msgbufsize); while (pa < last_pa) { dump_add_page(pa); pa += PAGE_SIZE; } #endif /* * Compute the number of pages of memory that will be available for * use (taking into account the overhead of a page structure per * page). */ first_page = low_water / PAGE_SIZE; #ifdef VM_PHYSSEG_SPARSE page_range = 0; for (i = 0; i < vm_phys_nsegs; i++) { page_range += atop(vm_phys_segs[i].end - vm_phys_segs[i].start); } for (i = 0; phys_avail[i + 1] != 0; i += 2) page_range += atop(phys_avail[i + 1] - phys_avail[i]); #elif defined(VM_PHYSSEG_DENSE) page_range = high_water / PAGE_SIZE - first_page; #else #error "Either VM_PHYSSEG_DENSE or VM_PHYSSEG_SPARSE must be defined." #endif end = new_end; /* * Reserve an unmapped guard page to trap access to vm_page_array[-1]. */ vaddr += PAGE_SIZE; /* * Initialize the mem entry structures now, and put them in the free * queue. */ new_end = trunc_page(end - page_range * sizeof(struct vm_page)); mapped = pmap_map(&vaddr, new_end, end, VM_PROT_READ | VM_PROT_WRITE); vm_page_array = (vm_page_t) mapped; #if VM_NRESERVLEVEL > 0 /* * Allocate memory for the reservation management system's data * structures. */ new_end = vm_reserv_startup(&vaddr, new_end, high_water); #endif #if defined(__aarch64__) || defined(__amd64__) || defined(__mips__) /* * pmap_map on arm64, amd64, and mips can come out of the direct-map, * not kvm like i386, so the pages must be tracked for a crashdump to * include this data. This includes the vm_page_array and the early * UMA bootstrap pages. */ for (pa = new_end; pa < phys_avail[biggestone + 1]; pa += PAGE_SIZE) dump_add_page(pa); #endif phys_avail[biggestone + 1] = new_end; /* * Add physical memory segments corresponding to the available * physical pages. */ for (i = 0; phys_avail[i + 1] != 0; i += 2) vm_phys_add_seg(phys_avail[i], phys_avail[i + 1]); /* * Clear all of the page structures */ bzero((caddr_t) vm_page_array, page_range * sizeof(struct vm_page)); for (i = 0; i < page_range; i++) vm_page_array[i].order = VM_NFREEORDER; vm_page_array_size = page_range; /* * Initialize the physical memory allocator. */ vm_phys_init(); /* * Add every available physical page that is not blacklisted to * the free lists. */ vm_cnt.v_page_count = 0; vm_cnt.v_free_count = 0; for (i = 0; phys_avail[i + 1] != 0; i += 2) { pa = phys_avail[i]; last_pa = phys_avail[i + 1]; while (pa < last_pa) { vm_phys_add_page(pa); pa += PAGE_SIZE; } } TAILQ_INIT(&blacklist_head); vm_page_blacklist_load(&list, &listend); vm_page_blacklist_check(list, listend); list = kern_getenv("vm.blacklist"); vm_page_blacklist_check(list, NULL); freeenv(list); #if VM_NRESERVLEVEL > 0 /* * Initialize the reservation management system. */ vm_reserv_init(); #endif return (vaddr); } void vm_page_reference(vm_page_t m) { vm_page_aflag_set(m, PGA_REFERENCED); } /* * vm_page_busy_downgrade: * * Downgrade an exclusive busy page into a single shared busy page. */ void vm_page_busy_downgrade(vm_page_t m) { u_int x; vm_page_assert_xbusied(m); for (;;) { x = m->busy_lock; x &= VPB_BIT_WAITERS; if (atomic_cmpset_rel_int(&m->busy_lock, VPB_SINGLE_EXCLUSIVER | x, VPB_SHARERS_WORD(1) | x)) break; } } /* * vm_page_sbusied: * * Return a positive value if the page is shared busied, 0 otherwise. */ int vm_page_sbusied(vm_page_t m) { u_int x; x = m->busy_lock; return ((x & VPB_BIT_SHARED) != 0 && x != VPB_UNBUSIED); } /* * vm_page_sunbusy: * * Shared unbusy a page. */ void vm_page_sunbusy(vm_page_t m) { u_int x; vm_page_assert_sbusied(m); for (;;) { x = m->busy_lock; if (VPB_SHARERS(x) > 1) { if (atomic_cmpset_int(&m->busy_lock, x, x - VPB_ONE_SHARER)) break; continue; } if ((x & VPB_BIT_WAITERS) == 0) { KASSERT(x == VPB_SHARERS_WORD(1), ("vm_page_sunbusy: invalid lock state")); if (atomic_cmpset_int(&m->busy_lock, VPB_SHARERS_WORD(1), VPB_UNBUSIED)) break; continue; } KASSERT(x == (VPB_SHARERS_WORD(1) | VPB_BIT_WAITERS), ("vm_page_sunbusy: invalid lock state for waiters")); vm_page_lock(m); if (!atomic_cmpset_int(&m->busy_lock, x, VPB_UNBUSIED)) { vm_page_unlock(m); continue; } wakeup(m); vm_page_unlock(m); break; } } /* * vm_page_busy_sleep: * * Sleep and release the page lock, using the page pointer as wchan. * This is used to implement the hard-path of busying mechanism. * * The given page must be locked. */ void vm_page_busy_sleep(vm_page_t m, const char *wmesg) { u_int x; vm_page_lock_assert(m, MA_OWNED); x = m->busy_lock; if (x == VPB_UNBUSIED) { vm_page_unlock(m); return; } if ((x & VPB_BIT_WAITERS) == 0 && !atomic_cmpset_int(&m->busy_lock, x, x | VPB_BIT_WAITERS)) { vm_page_unlock(m); return; } msleep(m, vm_page_lockptr(m), PVM | PDROP, wmesg, 0); } /* * vm_page_trysbusy: * * Try to shared busy a page. * If the operation succeeds 1 is returned otherwise 0. * The operation never sleeps. */ int vm_page_trysbusy(vm_page_t m) { u_int x; for (;;) { x = m->busy_lock; if ((x & VPB_BIT_SHARED) == 0) return (0); if (atomic_cmpset_acq_int(&m->busy_lock, x, x + VPB_ONE_SHARER)) return (1); } } static void +vm_page_xunbusy_locked(vm_page_t m) +{ + + vm_page_assert_xbusied(m); + vm_page_assert_locked(m); + + atomic_store_rel_int(&m->busy_lock, VPB_UNBUSIED); + /* There is a waiter, do wakeup() instead of vm_page_flash(). */ + wakeup(m); +} + +static void vm_page_xunbusy_maybelocked(vm_page_t m) { bool lockacq; vm_page_assert_xbusied(m); + /* + * Fast path for unbusy. If it succeeds, we know that there + * are no waiters, so we do not need a wakeup. + */ + if (atomic_cmpset_rel_int(&m->busy_lock, VPB_SINGLE_EXCLUSIVER, + VPB_UNBUSIED)) + return; + lockacq = !mtx_owned(vm_page_lockptr(m)); if (lockacq) vm_page_lock(m); - vm_page_flash(m); - atomic_store_rel_int(&m->busy_lock, VPB_UNBUSIED); + vm_page_xunbusy_locked(m); if (lockacq) vm_page_unlock(m); } /* * vm_page_xunbusy_hard: * * Called after the first try the exclusive unbusy of a page failed. * It is assumed that the waiters bit is on. */ void vm_page_xunbusy_hard(vm_page_t m) { vm_page_assert_xbusied(m); vm_page_lock(m); - atomic_store_rel_int(&m->busy_lock, VPB_UNBUSIED); - wakeup(m); + vm_page_xunbusy_locked(m); vm_page_unlock(m); } /* * vm_page_flash: * * Wakeup anyone waiting for the page. * The ownership bits do not change. * * The given page must be locked. */ void vm_page_flash(vm_page_t m) { u_int x; vm_page_lock_assert(m, MA_OWNED); for (;;) { x = m->busy_lock; if ((x & VPB_BIT_WAITERS) == 0) return; if (atomic_cmpset_int(&m->busy_lock, x, x & (~VPB_BIT_WAITERS))) break; } wakeup(m); } /* * Keep page from being freed by the page daemon * much of the same effect as wiring, except much lower * overhead and should be used only for *very* temporary * holding ("wiring"). */ void vm_page_hold(vm_page_t mem) { vm_page_lock_assert(mem, MA_OWNED); mem->hold_count++; } void vm_page_unhold(vm_page_t mem) { vm_page_lock_assert(mem, MA_OWNED); KASSERT(mem->hold_count >= 1, ("vm_page_unhold: hold count < 0!!!")); --mem->hold_count; if (mem->hold_count == 0 && (mem->flags & PG_UNHOLDFREE) != 0) vm_page_free_toq(mem); } /* * vm_page_unhold_pages: * * Unhold each of the pages that is referenced by the given array. */ void vm_page_unhold_pages(vm_page_t *ma, int count) { struct mtx *mtx, *new_mtx; mtx = NULL; for (; count != 0; count--) { /* * Avoid releasing and reacquiring the same page lock. */ new_mtx = vm_page_lockptr(*ma); if (mtx != new_mtx) { if (mtx != NULL) mtx_unlock(mtx); mtx = new_mtx; mtx_lock(mtx); } vm_page_unhold(*ma); ma++; } if (mtx != NULL) mtx_unlock(mtx); } vm_page_t PHYS_TO_VM_PAGE(vm_paddr_t pa) { vm_page_t m; #ifdef VM_PHYSSEG_SPARSE m = vm_phys_paddr_to_vm_page(pa); if (m == NULL) m = vm_phys_fictitious_to_vm_page(pa); return (m); #elif defined(VM_PHYSSEG_DENSE) long pi; pi = atop(pa); if (pi >= first_page && (pi - first_page) < vm_page_array_size) { m = &vm_page_array[pi - first_page]; return (m); } return (vm_phys_fictitious_to_vm_page(pa)); #else #error "Either VM_PHYSSEG_DENSE or VM_PHYSSEG_SPARSE must be defined." #endif } /* * vm_page_getfake: * * Create a fictitious page with the specified physical address and * memory attribute. The memory attribute is the only the machine- * dependent aspect of a fictitious page that must be initialized. */ vm_page_t vm_page_getfake(vm_paddr_t paddr, vm_memattr_t memattr) { vm_page_t m; m = uma_zalloc(fakepg_zone, M_WAITOK | M_ZERO); vm_page_initfake(m, paddr, memattr); return (m); } void vm_page_initfake(vm_page_t m, vm_paddr_t paddr, vm_memattr_t memattr) { if ((m->flags & PG_FICTITIOUS) != 0) { /* * The page's memattr might have changed since the * previous initialization. Update the pmap to the * new memattr. */ goto memattr; } m->phys_addr = paddr; m->queue = PQ_NONE; /* Fictitious pages don't use "segind". */ m->flags = PG_FICTITIOUS; /* Fictitious pages don't use "order" or "pool". */ m->oflags = VPO_UNMANAGED; m->busy_lock = VPB_SINGLE_EXCLUSIVER; m->wire_count = 1; pmap_page_init(m); memattr: pmap_page_set_memattr(m, memattr); } /* * vm_page_putfake: * * Release a fictitious page. */ void vm_page_putfake(vm_page_t m) { KASSERT((m->oflags & VPO_UNMANAGED) != 0, ("managed %p", m)); KASSERT((m->flags & PG_FICTITIOUS) != 0, ("vm_page_putfake: bad page %p", m)); uma_zfree(fakepg_zone, m); } /* * vm_page_updatefake: * * Update the given fictitious page to the specified physical address and * memory attribute. */ void vm_page_updatefake(vm_page_t m, vm_paddr_t paddr, vm_memattr_t memattr) { KASSERT((m->flags & PG_FICTITIOUS) != 0, ("vm_page_updatefake: bad page %p", m)); m->phys_addr = paddr; pmap_page_set_memattr(m, memattr); } /* * vm_page_free: * * Free a page. */ void vm_page_free(vm_page_t m) { m->flags &= ~PG_ZERO; vm_page_free_toq(m); } /* * vm_page_free_zero: * * Free a page to the zerod-pages queue */ void vm_page_free_zero(vm_page_t m) { m->flags |= PG_ZERO; vm_page_free_toq(m); } /* * Unbusy and handle the page queueing for a page from the VOP_GETPAGES() * array which was optionally read ahead or behind. */ void vm_page_readahead_finish(vm_page_t m) { /* We shouldn't put invalid pages on queues. */ KASSERT(m->valid != 0, ("%s: %p is invalid", __func__, m)); /* * Since the page is not the actually needed one, whether it should * be activated or deactivated is not obvious. Empirical results * have shown that deactivating the page is usually the best choice, * unless the page is wanted by another thread. */ vm_page_lock(m); if ((m->busy_lock & VPB_BIT_WAITERS) != 0) vm_page_activate(m); else vm_page_deactivate(m); vm_page_unlock(m); vm_page_xunbusy(m); } /* * vm_page_sleep_if_busy: * * Sleep and release the page queues lock if the page is busied. * Returns TRUE if the thread slept. * * The given page must be unlocked and object containing it must * be locked. */ int vm_page_sleep_if_busy(vm_page_t m, const char *msg) { vm_object_t obj; vm_page_lock_assert(m, MA_NOTOWNED); VM_OBJECT_ASSERT_WLOCKED(m->object); if (vm_page_busied(m)) { /* * The page-specific object must be cached because page * identity can change during the sleep, causing the * re-lock of a different object. * It is assumed that a reference to the object is already * held by the callers. */ obj = m->object; vm_page_lock(m); VM_OBJECT_WUNLOCK(obj); vm_page_busy_sleep(m, msg); VM_OBJECT_WLOCK(obj); return (TRUE); } return (FALSE); } /* * vm_page_dirty_KBI: [ internal use only ] * * Set all bits in the page's dirty field. * * The object containing the specified page must be locked if the * call is made from the machine-independent layer. * * See vm_page_clear_dirty_mask(). * * This function should only be called by vm_page_dirty(). */ void vm_page_dirty_KBI(vm_page_t m) { /* These assertions refer to this operation by its public name. */ KASSERT((m->flags & PG_CACHED) == 0, ("vm_page_dirty: page in cache!")); KASSERT(m->valid == VM_PAGE_BITS_ALL, ("vm_page_dirty: page is invalid!")); m->dirty = VM_PAGE_BITS_ALL; } /* * vm_page_insert: [ internal use only ] * * Inserts the given mem entry into the object and object list. * * The object must be locked. */ int vm_page_insert(vm_page_t m, vm_object_t object, vm_pindex_t pindex) { vm_page_t mpred; VM_OBJECT_ASSERT_WLOCKED(object); mpred = vm_radix_lookup_le(&object->rtree, pindex); return (vm_page_insert_after(m, object, pindex, mpred)); } /* * vm_page_insert_after: * * Inserts the page "m" into the specified object at offset "pindex". * * The page "mpred" must immediately precede the offset "pindex" within * the specified object. * * The object must be locked. */ static int vm_page_insert_after(vm_page_t m, vm_object_t object, vm_pindex_t pindex, vm_page_t mpred) { vm_page_t msucc; VM_OBJECT_ASSERT_WLOCKED(object); KASSERT(m->object == NULL, ("vm_page_insert_after: page already inserted")); if (mpred != NULL) { KASSERT(mpred->object == object, ("vm_page_insert_after: object doesn't contain mpred")); KASSERT(mpred->pindex < pindex, ("vm_page_insert_after: mpred doesn't precede pindex")); msucc = TAILQ_NEXT(mpred, listq); } else msucc = TAILQ_FIRST(&object->memq); if (msucc != NULL) KASSERT(msucc->pindex > pindex, ("vm_page_insert_after: msucc doesn't succeed pindex")); /* * Record the object/offset pair in this page */ m->object = object; m->pindex = pindex; /* * Now link into the object's ordered list of backed pages. */ if (vm_radix_insert(&object->rtree, m)) { m->object = NULL; m->pindex = 0; return (1); } vm_page_insert_radixdone(m, object, mpred); return (0); } /* * vm_page_insert_radixdone: * * Complete page "m" insertion into the specified object after the * radix trie hooking. * * The page "mpred" must precede the offset "m->pindex" within the * specified object. * * The object must be locked. */ static void vm_page_insert_radixdone(vm_page_t m, vm_object_t object, vm_page_t mpred) { VM_OBJECT_ASSERT_WLOCKED(object); KASSERT(object != NULL && m->object == object, ("vm_page_insert_radixdone: page %p has inconsistent object", m)); if (mpred != NULL) { KASSERT(mpred->object == object, ("vm_page_insert_after: object doesn't contain mpred")); KASSERT(mpred->pindex < m->pindex, ("vm_page_insert_after: mpred doesn't precede pindex")); } if (mpred != NULL) TAILQ_INSERT_AFTER(&object->memq, mpred, m, listq); else TAILQ_INSERT_HEAD(&object->memq, m, listq); /* * Show that the object has one more resident page. */ object->resident_page_count++; /* * Hold the vnode until the last page is released. */ if (object->resident_page_count == 1 && object->type == OBJT_VNODE) vhold(object->handle); /* * Since we are inserting a new and possibly dirty page, * update the object's OBJ_MIGHTBEDIRTY flag. */ if (pmap_page_is_write_mapped(m)) vm_object_set_writeable_dirty(object); } /* * vm_page_remove: * * Removes the given mem entry from the object/offset-page * table and the object page list, but do not invalidate/terminate * the backing store. * * The object must be locked. The page must be locked if it is managed. */ void vm_page_remove(vm_page_t m) { vm_object_t object; if ((m->oflags & VPO_UNMANAGED) == 0) vm_page_assert_locked(m); if ((object = m->object) == NULL) return; VM_OBJECT_ASSERT_WLOCKED(object); if (vm_page_xbusied(m)) vm_page_xunbusy_maybelocked(m); /* * Now remove from the object's list of backed pages. */ vm_radix_remove(&object->rtree, m->pindex); TAILQ_REMOVE(&object->memq, m, listq); /* * And show that the object has one fewer resident page. */ object->resident_page_count--; /* * The vnode may now be recycled. */ if (object->resident_page_count == 0 && object->type == OBJT_VNODE) vdrop(object->handle); m->object = NULL; } /* * vm_page_lookup: * * Returns the page associated with the object/offset * pair specified; if none is found, NULL is returned. * * The object must be locked. */ vm_page_t vm_page_lookup(vm_object_t object, vm_pindex_t pindex) { VM_OBJECT_ASSERT_LOCKED(object); return (vm_radix_lookup(&object->rtree, pindex)); } /* * vm_page_find_least: * * Returns the page associated with the object with least pindex * greater than or equal to the parameter pindex, or NULL. * * The object must be locked. */ vm_page_t vm_page_find_least(vm_object_t object, vm_pindex_t pindex) { vm_page_t m; VM_OBJECT_ASSERT_LOCKED(object); if ((m = TAILQ_FIRST(&object->memq)) != NULL && m->pindex < pindex) m = vm_radix_lookup_ge(&object->rtree, pindex); return (m); } /* * Returns the given page's successor (by pindex) within the object if it is * resident; if none is found, NULL is returned. * * The object must be locked. */ vm_page_t vm_page_next(vm_page_t m) { vm_page_t next; VM_OBJECT_ASSERT_LOCKED(m->object); if ((next = TAILQ_NEXT(m, listq)) != NULL && next->pindex != m->pindex + 1) next = NULL; return (next); } /* * Returns the given page's predecessor (by pindex) within the object if it is * resident; if none is found, NULL is returned. * * The object must be locked. */ vm_page_t vm_page_prev(vm_page_t m) { vm_page_t prev; VM_OBJECT_ASSERT_LOCKED(m->object); if ((prev = TAILQ_PREV(m, pglist, listq)) != NULL && prev->pindex != m->pindex - 1) prev = NULL; return (prev); } /* * Uses the page mnew as a replacement for an existing page at index * pindex which must be already present in the object. * * The existing page must not be on a paging queue. */ vm_page_t vm_page_replace(vm_page_t mnew, vm_object_t object, vm_pindex_t pindex) { vm_page_t mold; VM_OBJECT_ASSERT_WLOCKED(object); KASSERT(mnew->object == NULL, ("vm_page_replace: page already in object")); /* * This function mostly follows vm_page_insert() and * vm_page_remove() without the radix, object count and vnode * dance. Double check such functions for more comments. */ mnew->object = object; mnew->pindex = pindex; mold = vm_radix_replace(&object->rtree, mnew); KASSERT(mold->queue == PQ_NONE, ("vm_page_replace: mold is on a paging queue")); /* Keep the resident page list in sorted order. */ TAILQ_INSERT_AFTER(&object->memq, mold, mnew, listq); TAILQ_REMOVE(&object->memq, mold, listq); mold->object = NULL; vm_page_xunbusy_maybelocked(mold); /* * The object's resident_page_count does not change because we have * swapped one page for another, but OBJ_MIGHTBEDIRTY. */ if (pmap_page_is_write_mapped(mnew)) vm_object_set_writeable_dirty(object); return (mold); } /* * vm_page_rename: * * Move the given memory entry from its * current object to the specified target object/offset. * * Note: swap associated with the page must be invalidated by the move. We * have to do this for several reasons: (1) we aren't freeing the * page, (2) we are dirtying the page, (3) the VM system is probably * moving the page from object A to B, and will then later move * the backing store from A to B and we can't have a conflict. * * Note: we *always* dirty the page. It is necessary both for the * fact that we moved it, and because we may be invalidating * swap. If the page is on the cache, we have to deactivate it * or vm_page_dirty() will panic. Dirty pages are not allowed * on the cache. * * The objects must be locked. */ int vm_page_rename(vm_page_t m, vm_object_t new_object, vm_pindex_t new_pindex) { vm_page_t mpred; vm_pindex_t opidx; VM_OBJECT_ASSERT_WLOCKED(new_object); mpred = vm_radix_lookup_le(&new_object->rtree, new_pindex); KASSERT(mpred == NULL || mpred->pindex != new_pindex, ("vm_page_rename: pindex already renamed")); /* * Create a custom version of vm_page_insert() which does not depend * by m_prev and can cheat on the implementation aspects of the * function. */ opidx = m->pindex; m->pindex = new_pindex; if (vm_radix_insert(&new_object->rtree, m)) { m->pindex = opidx; return (1); } /* * The operation cannot fail anymore. The removal must happen before * the listq iterator is tainted. */ m->pindex = opidx; vm_page_lock(m); vm_page_remove(m); /* Return back to the new pindex to complete vm_page_insert(). */ m->pindex = new_pindex; m->object = new_object; vm_page_unlock(m); vm_page_insert_radixdone(m, new_object, mpred); vm_page_dirty(m); return (0); } /* * Convert all of the given object's cached pages that have a * pindex within the given range into free pages. If the value * zero is given for "end", then the range's upper bound is * infinity. If the given object is backed by a vnode and it * transitions from having one or more cached pages to none, the * vnode's hold count is reduced. */ void vm_page_cache_free(vm_object_t object, vm_pindex_t start, vm_pindex_t end) { vm_page_t m; boolean_t empty; mtx_lock(&vm_page_queue_free_mtx); if (__predict_false(vm_radix_is_empty(&object->cache))) { mtx_unlock(&vm_page_queue_free_mtx); return; } while ((m = vm_radix_lookup_ge(&object->cache, start)) != NULL) { if (end != 0 && m->pindex >= end) break; vm_radix_remove(&object->cache, m->pindex); vm_page_cache_turn_free(m); } empty = vm_radix_is_empty(&object->cache); mtx_unlock(&vm_page_queue_free_mtx); if (object->type == OBJT_VNODE && empty) vdrop(object->handle); } /* * Returns the cached page that is associated with the given * object and offset. If, however, none exists, returns NULL. * * The free page queue must be locked. */ static inline vm_page_t vm_page_cache_lookup(vm_object_t object, vm_pindex_t pindex) { mtx_assert(&vm_page_queue_free_mtx, MA_OWNED); return (vm_radix_lookup(&object->cache, pindex)); } /* * Remove the given cached page from its containing object's * collection of cached pages. * * The free page queue must be locked. */ static void vm_page_cache_remove(vm_page_t m) { mtx_assert(&vm_page_queue_free_mtx, MA_OWNED); KASSERT((m->flags & PG_CACHED) != 0, ("vm_page_cache_remove: page %p is not cached", m)); vm_radix_remove(&m->object->cache, m->pindex); m->object = NULL; vm_cnt.v_cache_count--; } /* * Transfer all of the cached pages with offset greater than or * equal to 'offidxstart' from the original object's cache to the * new object's cache. However, any cached pages with offset * greater than or equal to the new object's size are kept in the * original object. Initially, the new object's cache must be * empty. Offset 'offidxstart' in the original object must * correspond to offset zero in the new object. * * The new object must be locked. */ void vm_page_cache_transfer(vm_object_t orig_object, vm_pindex_t offidxstart, vm_object_t new_object) { vm_page_t m; /* * Insertion into an object's collection of cached pages * requires the object to be locked. In contrast, removal does * not. */ VM_OBJECT_ASSERT_WLOCKED(new_object); KASSERT(vm_radix_is_empty(&new_object->cache), ("vm_page_cache_transfer: object %p has cached pages", new_object)); mtx_lock(&vm_page_queue_free_mtx); while ((m = vm_radix_lookup_ge(&orig_object->cache, offidxstart)) != NULL) { /* * Transfer all of the pages with offset greater than or * equal to 'offidxstart' from the original object's * cache to the new object's cache. */ if ((m->pindex - offidxstart) >= new_object->size) break; vm_radix_remove(&orig_object->cache, m->pindex); /* Update the page's object and offset. */ m->object = new_object; m->pindex -= offidxstart; if (vm_radix_insert(&new_object->cache, m)) vm_page_cache_turn_free(m); } mtx_unlock(&vm_page_queue_free_mtx); } /* * Returns TRUE if a cached page is associated with the given object and * offset, and FALSE otherwise. * * The object must be locked. */ boolean_t vm_page_is_cached(vm_object_t object, vm_pindex_t pindex) { vm_page_t m; /* * Insertion into an object's collection of cached pages requires the * object to be locked. Therefore, if the object is locked and the * object's collection is empty, there is no need to acquire the free * page queues lock in order to prove that the specified page doesn't * exist. */ VM_OBJECT_ASSERT_WLOCKED(object); if (__predict_true(vm_object_cache_is_empty(object))) return (FALSE); mtx_lock(&vm_page_queue_free_mtx); m = vm_page_cache_lookup(object, pindex); mtx_unlock(&vm_page_queue_free_mtx); return (m != NULL); } /* * vm_page_alloc: * * Allocate and return a page that is associated with the specified * object and offset pair. By default, this page is exclusive busied. * * The caller must always specify an allocation class. * * allocation classes: * VM_ALLOC_NORMAL normal process request * VM_ALLOC_SYSTEM system *really* needs a page * VM_ALLOC_INTERRUPT interrupt time request * * optional allocation flags: * VM_ALLOC_COUNT(number) the number of additional pages that the caller * intends to allocate * VM_ALLOC_IFCACHED return page only if it is cached * VM_ALLOC_IFNOTCACHED return NULL, do not reactivate if the page * is cached * VM_ALLOC_NOBUSY do not exclusive busy the page * VM_ALLOC_NODUMP do not include the page in a kernel core dump * VM_ALLOC_NOOBJ page is not associated with an object and * should not be exclusive busy * VM_ALLOC_SBUSY shared busy the allocated page * VM_ALLOC_WIRED wire the allocated page * VM_ALLOC_ZERO prefer a zeroed page * * This routine may not sleep. */ vm_page_t vm_page_alloc(vm_object_t object, vm_pindex_t pindex, int req) { struct vnode *vp = NULL; vm_object_t m_object; vm_page_t m, mpred; int flags, req_class; mpred = 0; /* XXX: pacify gcc */ KASSERT((object != NULL) == ((req & VM_ALLOC_NOOBJ) == 0) && (object != NULL || (req & VM_ALLOC_SBUSY) == 0) && ((req & (VM_ALLOC_NOBUSY | VM_ALLOC_SBUSY)) != (VM_ALLOC_NOBUSY | VM_ALLOC_SBUSY)), ("vm_page_alloc: inconsistent object(%p)/req(%x)", (void *)object, req)); if (object != NULL) VM_OBJECT_ASSERT_WLOCKED(object); req_class = req & VM_ALLOC_CLASS_MASK; /* * The page daemon is allowed to dig deeper into the free page list. */ if (curproc == pageproc && req_class != VM_ALLOC_INTERRUPT) req_class = VM_ALLOC_SYSTEM; if (object != NULL) { mpred = vm_radix_lookup_le(&object->rtree, pindex); KASSERT(mpred == NULL || mpred->pindex != pindex, ("vm_page_alloc: pindex already allocated")); } /* * The page allocation request can came from consumers which already * hold the free page queue mutex, like vm_page_insert() in * vm_page_cache(). */ mtx_lock_flags(&vm_page_queue_free_mtx, MTX_RECURSE); if (vm_cnt.v_free_count + vm_cnt.v_cache_count > vm_cnt.v_free_reserved || (req_class == VM_ALLOC_SYSTEM && vm_cnt.v_free_count + vm_cnt.v_cache_count > vm_cnt.v_interrupt_free_min) || (req_class == VM_ALLOC_INTERRUPT && vm_cnt.v_free_count + vm_cnt.v_cache_count > 0)) { /* * Allocate from the free queue if the number of free pages * exceeds the minimum for the request class. */ if (object != NULL && (m = vm_page_cache_lookup(object, pindex)) != NULL) { if ((req & VM_ALLOC_IFNOTCACHED) != 0) { mtx_unlock(&vm_page_queue_free_mtx); return (NULL); } if (vm_phys_unfree_page(m)) vm_phys_set_pool(VM_FREEPOOL_DEFAULT, m, 0); #if VM_NRESERVLEVEL > 0 else if (!vm_reserv_reactivate_page(m)) #else else #endif panic("vm_page_alloc: cache page %p is missing" " from the free queue", m); } else if ((req & VM_ALLOC_IFCACHED) != 0) { mtx_unlock(&vm_page_queue_free_mtx); return (NULL); #if VM_NRESERVLEVEL > 0 } else if (object == NULL || (object->flags & (OBJ_COLORED | OBJ_FICTITIOUS)) != OBJ_COLORED || (m = vm_reserv_alloc_page(object, pindex, mpred)) == NULL) { #else } else { #endif m = vm_phys_alloc_pages(object != NULL ? VM_FREEPOOL_DEFAULT : VM_FREEPOOL_DIRECT, 0); #if VM_NRESERVLEVEL > 0 if (m == NULL && vm_reserv_reclaim_inactive()) { m = vm_phys_alloc_pages(object != NULL ? VM_FREEPOOL_DEFAULT : VM_FREEPOOL_DIRECT, 0); } #endif } } else { /* * Not allocatable, give up. */ mtx_unlock(&vm_page_queue_free_mtx); atomic_add_int(&vm_pageout_deficit, max((u_int)req >> VM_ALLOC_COUNT_SHIFT, 1)); pagedaemon_wakeup(); return (NULL); } /* * At this point we had better have found a good page. */ KASSERT(m != NULL, ("vm_page_alloc: missing page")); KASSERT(m->queue == PQ_NONE, ("vm_page_alloc: page %p has unexpected queue %d", m, m->queue)); KASSERT(m->wire_count == 0, ("vm_page_alloc: page %p is wired", m)); KASSERT(m->hold_count == 0, ("vm_page_alloc: page %p is held", m)); KASSERT(!vm_page_sbusied(m), ("vm_page_alloc: page %p is busy", m)); KASSERT(m->dirty == 0, ("vm_page_alloc: page %p is dirty", m)); KASSERT(pmap_page_get_memattr(m) == VM_MEMATTR_DEFAULT, ("vm_page_alloc: page %p has unexpected memattr %d", m, pmap_page_get_memattr(m))); if ((m->flags & PG_CACHED) != 0) { KASSERT((m->flags & PG_ZERO) == 0, ("vm_page_alloc: cached page %p is PG_ZERO", m)); KASSERT(m->valid != 0, ("vm_page_alloc: cached page %p is invalid", m)); if (m->object == object && m->pindex == pindex) vm_cnt.v_reactivated++; else m->valid = 0; m_object = m->object; vm_page_cache_remove(m); if (m_object->type == OBJT_VNODE && vm_object_cache_is_empty(m_object)) vp = m_object->handle; } else { KASSERT(m->valid == 0, ("vm_page_alloc: free page %p is valid", m)); vm_phys_freecnt_adj(m, -1); if ((m->flags & PG_ZERO) != 0) vm_page_zero_count--; } mtx_unlock(&vm_page_queue_free_mtx); /* * Initialize the page. Only the PG_ZERO flag is inherited. */ flags = 0; if ((req & VM_ALLOC_ZERO) != 0) flags = PG_ZERO; flags &= m->flags; if ((req & VM_ALLOC_NODUMP) != 0) flags |= PG_NODUMP; m->flags = flags; m->aflags = 0; m->oflags = object == NULL || (object->flags & OBJ_UNMANAGED) != 0 ? VPO_UNMANAGED : 0; m->busy_lock = VPB_UNBUSIED; if ((req & (VM_ALLOC_NOBUSY | VM_ALLOC_NOOBJ | VM_ALLOC_SBUSY)) == 0) m->busy_lock = VPB_SINGLE_EXCLUSIVER; if ((req & VM_ALLOC_SBUSY) != 0) m->busy_lock = VPB_SHARERS_WORD(1); if (req & VM_ALLOC_WIRED) { /* * The page lock is not required for wiring a page until that * page is inserted into the object. */ atomic_add_int(&vm_cnt.v_wire_count, 1); m->wire_count = 1; } m->act_count = 0; if (object != NULL) { if (vm_page_insert_after(m, object, pindex, mpred)) { /* See the comment below about hold count. */ if (vp != NULL) vdrop(vp); pagedaemon_wakeup(); if (req & VM_ALLOC_WIRED) { atomic_subtract_int(&vm_cnt.v_wire_count, 1); m->wire_count = 0; } m->object = NULL; m->oflags = VPO_UNMANAGED; m->busy_lock = VPB_UNBUSIED; vm_page_free(m); return (NULL); } /* Ignore device objects; the pager sets "memattr" for them. */ if (object->memattr != VM_MEMATTR_DEFAULT && (object->flags & OBJ_FICTITIOUS) == 0) pmap_page_set_memattr(m, object->memattr); } else m->pindex = pindex; /* * The following call to vdrop() must come after the above call * to vm_page_insert() in case both affect the same object and * vnode. Otherwise, the affected vnode's hold count could * temporarily become zero. */ if (vp != NULL) vdrop(vp); /* * Don't wakeup too often - wakeup the pageout daemon when * we would be nearly out of memory. */ if (vm_paging_needed()) pagedaemon_wakeup(); return (m); } static void vm_page_alloc_contig_vdrop(struct spglist *lst) { while (!SLIST_EMPTY(lst)) { vdrop((struct vnode *)SLIST_FIRST(lst)-> plinks.s.pv); SLIST_REMOVE_HEAD(lst, plinks.s.ss); } } /* * vm_page_alloc_contig: * * Allocate a contiguous set of physical pages of the given size "npages" * from the free lists. All of the physical pages must be at or above * the given physical address "low" and below the given physical address * "high". The given value "alignment" determines the alignment of the * first physical page in the set. If the given value "boundary" is * non-zero, then the set of physical pages cannot cross any physical * address boundary that is a multiple of that value. Both "alignment" * and "boundary" must be a power of two. * * If the specified memory attribute, "memattr", is VM_MEMATTR_DEFAULT, * then the memory attribute setting for the physical pages is configured * to the object's memory attribute setting. Otherwise, the memory * attribute setting for the physical pages is configured to "memattr", * overriding the object's memory attribute setting. However, if the * object's memory attribute setting is not VM_MEMATTR_DEFAULT, then the * memory attribute setting for the physical pages cannot be configured * to VM_MEMATTR_DEFAULT. * * The caller must always specify an allocation class. * * allocation classes: * VM_ALLOC_NORMAL normal process request * VM_ALLOC_SYSTEM system *really* needs a page * VM_ALLOC_INTERRUPT interrupt time request * * optional allocation flags: * VM_ALLOC_NOBUSY do not exclusive busy the page * VM_ALLOC_NODUMP do not include the page in a kernel core dump * VM_ALLOC_NOOBJ page is not associated with an object and * should not be exclusive busy * VM_ALLOC_SBUSY shared busy the allocated page * VM_ALLOC_WIRED wire the allocated page * VM_ALLOC_ZERO prefer a zeroed page * * This routine may not sleep. */ vm_page_t vm_page_alloc_contig(vm_object_t object, vm_pindex_t pindex, int req, u_long npages, vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary, vm_memattr_t memattr) { struct vnode *drop; struct spglist deferred_vdrop_list; vm_page_t m, m_tmp, m_ret; u_int flags; int req_class; KASSERT((object != NULL) == ((req & VM_ALLOC_NOOBJ) == 0) && (object != NULL || (req & VM_ALLOC_SBUSY) == 0) && ((req & (VM_ALLOC_NOBUSY | VM_ALLOC_SBUSY)) != (VM_ALLOC_NOBUSY | VM_ALLOC_SBUSY)), ("vm_page_alloc: inconsistent object(%p)/req(%x)", (void *)object, req)); if (object != NULL) { VM_OBJECT_ASSERT_WLOCKED(object); KASSERT(object->type == OBJT_PHYS, ("vm_page_alloc_contig: object %p isn't OBJT_PHYS", object)); } KASSERT(npages > 0, ("vm_page_alloc_contig: npages is zero")); req_class = req & VM_ALLOC_CLASS_MASK; /* * The page daemon is allowed to dig deeper into the free page list. */ if (curproc == pageproc && req_class != VM_ALLOC_INTERRUPT) req_class = VM_ALLOC_SYSTEM; SLIST_INIT(&deferred_vdrop_list); mtx_lock(&vm_page_queue_free_mtx); if (vm_cnt.v_free_count + vm_cnt.v_cache_count >= npages + vm_cnt.v_free_reserved || (req_class == VM_ALLOC_SYSTEM && vm_cnt.v_free_count + vm_cnt.v_cache_count >= npages + vm_cnt.v_interrupt_free_min) || (req_class == VM_ALLOC_INTERRUPT && vm_cnt.v_free_count + vm_cnt.v_cache_count >= npages)) { #if VM_NRESERVLEVEL > 0 retry: if (object == NULL || (object->flags & OBJ_COLORED) == 0 || (m_ret = vm_reserv_alloc_contig(object, pindex, npages, low, high, alignment, boundary)) == NULL) #endif m_ret = vm_phys_alloc_contig(npages, low, high, alignment, boundary); } else { mtx_unlock(&vm_page_queue_free_mtx); atomic_add_int(&vm_pageout_deficit, npages); pagedaemon_wakeup(); return (NULL); } if (m_ret != NULL) for (m = m_ret; m < &m_ret[npages]; m++) { drop = vm_page_alloc_init(m); if (drop != NULL) { /* * Enqueue the vnode for deferred vdrop(). */ m->plinks.s.pv = drop; SLIST_INSERT_HEAD(&deferred_vdrop_list, m, plinks.s.ss); } } else { #if VM_NRESERVLEVEL > 0 if (vm_reserv_reclaim_contig(npages, low, high, alignment, boundary)) goto retry; #endif } mtx_unlock(&vm_page_queue_free_mtx); if (m_ret == NULL) return (NULL); /* * Initialize the pages. Only the PG_ZERO flag is inherited. */ flags = 0; if ((req & VM_ALLOC_ZERO) != 0) flags = PG_ZERO; if ((req & VM_ALLOC_NODUMP) != 0) flags |= PG_NODUMP; if ((req & VM_ALLOC_WIRED) != 0) atomic_add_int(&vm_cnt.v_wire_count, npages); if (object != NULL) { if (object->memattr != VM_MEMATTR_DEFAULT && memattr == VM_MEMATTR_DEFAULT) memattr = object->memattr; } for (m = m_ret; m < &m_ret[npages]; m++) { m->aflags = 0; m->flags = (m->flags | PG_NODUMP) & flags; m->busy_lock = VPB_UNBUSIED; if (object != NULL) { if ((req & (VM_ALLOC_NOBUSY | VM_ALLOC_SBUSY)) == 0) m->busy_lock = VPB_SINGLE_EXCLUSIVER; if ((req & VM_ALLOC_SBUSY) != 0) m->busy_lock = VPB_SHARERS_WORD(1); } if ((req & VM_ALLOC_WIRED) != 0) m->wire_count = 1; /* Unmanaged pages don't use "act_count". */ m->oflags = VPO_UNMANAGED; if (object != NULL) { if (vm_page_insert(m, object, pindex)) { vm_page_alloc_contig_vdrop( &deferred_vdrop_list); if (vm_paging_needed()) pagedaemon_wakeup(); if ((req & VM_ALLOC_WIRED) != 0) atomic_subtract_int(&vm_cnt.v_wire_count, npages); for (m_tmp = m, m = m_ret; m < &m_ret[npages]; m++) { if ((req & VM_ALLOC_WIRED) != 0) m->wire_count = 0; if (m >= m_tmp) { m->object = NULL; m->oflags |= VPO_UNMANAGED; } m->busy_lock = VPB_UNBUSIED; vm_page_free(m); } return (NULL); } } else m->pindex = pindex; if (memattr != VM_MEMATTR_DEFAULT) pmap_page_set_memattr(m, memattr); pindex++; } vm_page_alloc_contig_vdrop(&deferred_vdrop_list); if (vm_paging_needed()) pagedaemon_wakeup(); return (m_ret); } /* * Initialize a page that has been freshly dequeued from a freelist. * The caller has to drop the vnode returned, if it is not NULL. * * This function may only be used to initialize unmanaged pages. * * To be called with vm_page_queue_free_mtx held. */ static struct vnode * vm_page_alloc_init(vm_page_t m) { struct vnode *drop; vm_object_t m_object; KASSERT(m->queue == PQ_NONE, ("vm_page_alloc_init: page %p has unexpected queue %d", m, m->queue)); KASSERT(m->wire_count == 0, ("vm_page_alloc_init: page %p is wired", m)); KASSERT(m->hold_count == 0, ("vm_page_alloc_init: page %p is held", m)); KASSERT(!vm_page_sbusied(m), ("vm_page_alloc_init: page %p is busy", m)); KASSERT(m->dirty == 0, ("vm_page_alloc_init: page %p is dirty", m)); KASSERT(pmap_page_get_memattr(m) == VM_MEMATTR_DEFAULT, ("vm_page_alloc_init: page %p has unexpected memattr %d", m, pmap_page_get_memattr(m))); mtx_assert(&vm_page_queue_free_mtx, MA_OWNED); drop = NULL; if ((m->flags & PG_CACHED) != 0) { KASSERT((m->flags & PG_ZERO) == 0, ("vm_page_alloc_init: cached page %p is PG_ZERO", m)); m->valid = 0; m_object = m->object; vm_page_cache_remove(m); if (m_object->type == OBJT_VNODE && vm_object_cache_is_empty(m_object)) drop = m_object->handle; } else { KASSERT(m->valid == 0, ("vm_page_alloc_init: free page %p is valid", m)); vm_phys_freecnt_adj(m, -1); if ((m->flags & PG_ZERO) != 0) vm_page_zero_count--; } return (drop); } /* * vm_page_alloc_freelist: * * Allocate a physical page from the specified free page list. * * The caller must always specify an allocation class. * * allocation classes: * VM_ALLOC_NORMAL normal process request * VM_ALLOC_SYSTEM system *really* needs a page * VM_ALLOC_INTERRUPT interrupt time request * * optional allocation flags: * VM_ALLOC_COUNT(number) the number of additional pages that the caller * intends to allocate * VM_ALLOC_WIRED wire the allocated page * VM_ALLOC_ZERO prefer a zeroed page * * This routine may not sleep. */ vm_page_t vm_page_alloc_freelist(int flind, int req) { struct vnode *drop; vm_page_t m; u_int flags; int req_class; req_class = req & VM_ALLOC_CLASS_MASK; /* * The page daemon is allowed to dig deeper into the free page list. */ if (curproc == pageproc && req_class != VM_ALLOC_INTERRUPT) req_class = VM_ALLOC_SYSTEM; /* * Do not allocate reserved pages unless the req has asked for it. */ mtx_lock_flags(&vm_page_queue_free_mtx, MTX_RECURSE); if (vm_cnt.v_free_count + vm_cnt.v_cache_count > vm_cnt.v_free_reserved || (req_class == VM_ALLOC_SYSTEM && vm_cnt.v_free_count + vm_cnt.v_cache_count > vm_cnt.v_interrupt_free_min) || (req_class == VM_ALLOC_INTERRUPT && vm_cnt.v_free_count + vm_cnt.v_cache_count > 0)) m = vm_phys_alloc_freelist_pages(flind, VM_FREEPOOL_DIRECT, 0); else { mtx_unlock(&vm_page_queue_free_mtx); atomic_add_int(&vm_pageout_deficit, max((u_int)req >> VM_ALLOC_COUNT_SHIFT, 1)); pagedaemon_wakeup(); return (NULL); } if (m == NULL) { mtx_unlock(&vm_page_queue_free_mtx); return (NULL); } drop = vm_page_alloc_init(m); mtx_unlock(&vm_page_queue_free_mtx); /* * Initialize the page. Only the PG_ZERO flag is inherited. */ m->aflags = 0; flags = 0; if ((req & VM_ALLOC_ZERO) != 0) flags = PG_ZERO; m->flags &= flags; if ((req & VM_ALLOC_WIRED) != 0) { /* * The page lock is not required for wiring a page that does * not belong to an object. */ atomic_add_int(&vm_cnt.v_wire_count, 1); m->wire_count = 1; } /* Unmanaged pages don't use "act_count". */ m->oflags = VPO_UNMANAGED; if (drop != NULL) vdrop(drop); if (vm_paging_needed()) pagedaemon_wakeup(); return (m); } #define VPSC_ANY 0 /* No restrictions. */ #define VPSC_NORESERV 1 /* Skip reservations; implies VPSC_NOSUPER. */ #define VPSC_NOSUPER 2 /* Skip superpages. */ /* * vm_page_scan_contig: * * Scan vm_page_array[] between the specified entries "m_start" and * "m_end" for a run of contiguous physical pages that satisfy the * specified conditions, and return the lowest page in the run. The * specified "alignment" determines the alignment of the lowest physical * page in the run. If the specified "boundary" is non-zero, then the * run of physical pages cannot span a physical address that is a * multiple of "boundary". * * "m_end" is never dereferenced, so it need not point to a vm_page * structure within vm_page_array[]. * * "npages" must be greater than zero. "m_start" and "m_end" must not * span a hole (or discontiguity) in the physical address space. Both * "alignment" and "boundary" must be a power of two. */ vm_page_t vm_page_scan_contig(u_long npages, vm_page_t m_start, vm_page_t m_end, u_long alignment, vm_paddr_t boundary, int options) { struct mtx *m_mtx, *new_mtx; vm_object_t object; vm_paddr_t pa; vm_page_t m, m_run; #if VM_NRESERVLEVEL > 0 int level; #endif int m_inc, order, run_ext, run_len; KASSERT(npages > 0, ("npages is 0")); KASSERT(powerof2(alignment), ("alignment is not a power of 2")); KASSERT(powerof2(boundary), ("boundary is not a power of 2")); m_run = NULL; run_len = 0; m_mtx = NULL; for (m = m_start; m < m_end && run_len < npages; m += m_inc) { KASSERT((m->flags & (PG_FICTITIOUS | PG_MARKER)) == 0, ("page %p is PG_FICTITIOUS or PG_MARKER", m)); /* * If the current page would be the start of a run, check its * physical address against the end, alignment, and boundary * conditions. If it doesn't satisfy these conditions, either * terminate the scan or advance to the next page that * satisfies the failed condition. */ if (run_len == 0) { KASSERT(m_run == NULL, ("m_run != NULL")); if (m + npages > m_end) break; pa = VM_PAGE_TO_PHYS(m); if ((pa & (alignment - 1)) != 0) { m_inc = atop(roundup2(pa, alignment) - pa); continue; } if (rounddown2(pa ^ (pa + ptoa(npages) - 1), boundary) != 0) { m_inc = atop(roundup2(pa, boundary) - pa); continue; } } else KASSERT(m_run != NULL, ("m_run == NULL")); /* * Avoid releasing and reacquiring the same page lock. */ new_mtx = vm_page_lockptr(m); if (m_mtx != new_mtx) { if (m_mtx != NULL) mtx_unlock(m_mtx); m_mtx = new_mtx; mtx_lock(m_mtx); } m_inc = 1; retry: if (m->wire_count != 0 || m->hold_count != 0) run_ext = 0; #if VM_NRESERVLEVEL > 0 else if ((level = vm_reserv_level(m)) >= 0 && (options & VPSC_NORESERV) != 0) { run_ext = 0; /* Advance to the end of the reservation. */ pa = VM_PAGE_TO_PHYS(m); m_inc = atop(roundup2(pa + 1, vm_reserv_size(level)) - pa); } #endif else if ((object = m->object) != NULL) { /* * The page is considered eligible for relocation if * and only if it could be laundered or reclaimed by * the page daemon. */ if (!VM_OBJECT_TRYRLOCK(object)) { mtx_unlock(m_mtx); VM_OBJECT_RLOCK(object); mtx_lock(m_mtx); if (m->object != object) { /* * The page may have been freed. */ VM_OBJECT_RUNLOCK(object); goto retry; } else if (m->wire_count != 0 || m->hold_count != 0) { run_ext = 0; goto unlock; } } KASSERT((m->flags & PG_UNHOLDFREE) == 0, ("page %p is PG_UNHOLDFREE", m)); /* Don't care: PG_NODUMP, PG_WINATCFLS, PG_ZERO. */ if (object->type != OBJT_DEFAULT && object->type != OBJT_SWAP && object->type != OBJT_VNODE) run_ext = 0; else if ((m->flags & PG_CACHED) != 0 || m != vm_page_lookup(object, m->pindex)) { /* * The page is cached or recently converted * from cached to free. */ #if VM_NRESERVLEVEL > 0 if (level >= 0) { /* * The page is reserved. Extend the * current run by one page. */ run_ext = 1; } else #endif if ((order = m->order) < VM_NFREEORDER) { /* * The page is enqueued in the * physical memory allocator's cache/ * free page queues. Moreover, it is * the first page in a power-of-two- * sized run of contiguous cache/free * pages. Add these pages to the end * of the current run, and jump * ahead. */ run_ext = 1 << order; m_inc = 1 << order; } else run_ext = 0; #if VM_NRESERVLEVEL > 0 } else if ((options & VPSC_NOSUPER) != 0 && (level = vm_reserv_level_iffullpop(m)) >= 0) { run_ext = 0; /* Advance to the end of the superpage. */ pa = VM_PAGE_TO_PHYS(m); m_inc = atop(roundup2(pa + 1, vm_reserv_size(level)) - pa); #endif } else if (object->memattr == VM_MEMATTR_DEFAULT && m->queue != PQ_NONE && !vm_page_busied(m)) { /* * The page is allocated but eligible for * relocation. Extend the current run by one * page. */ KASSERT(pmap_page_get_memattr(m) == VM_MEMATTR_DEFAULT, ("page %p has an unexpected memattr", m)); KASSERT((m->oflags & (VPO_SWAPINPROG | VPO_SWAPSLEEP | VPO_UNMANAGED)) == 0, ("page %p has unexpected oflags", m)); /* Don't care: VPO_NOSYNC. */ run_ext = 1; } else run_ext = 0; unlock: VM_OBJECT_RUNLOCK(object); #if VM_NRESERVLEVEL > 0 } else if (level >= 0) { /* * The page is reserved but not yet allocated. In * other words, it is still cached or free. Extend * the current run by one page. */ run_ext = 1; #endif } else if ((order = m->order) < VM_NFREEORDER) { /* * The page is enqueued in the physical memory * allocator's cache/free page queues. Moreover, it * is the first page in a power-of-two-sized run of * contiguous cache/free pages. Add these pages to * the end of the current run, and jump ahead. */ run_ext = 1 << order; m_inc = 1 << order; } else { /* * Skip the page for one of the following reasons: (1) * It is enqueued in the physical memory allocator's * cache/free page queues. However, it is not the * first page in a run of contiguous cache/free pages. * (This case rarely occurs because the scan is * performed in ascending order.) (2) It is not * reserved, and it is transitioning from free to * allocated. (Conversely, the transition from * allocated to free for managed pages is blocked by * the page lock.) (3) It is allocated but not * contained by an object and not wired, e.g., * allocated by Xen's balloon driver. */ run_ext = 0; } /* * Extend or reset the current run of pages. */ if (run_ext > 0) { if (run_len == 0) m_run = m; run_len += run_ext; } else { if (run_len > 0) { m_run = NULL; run_len = 0; } } } if (m_mtx != NULL) mtx_unlock(m_mtx); if (run_len >= npages) return (m_run); return (NULL); } /* * vm_page_reclaim_run: * * Try to relocate each of the allocated virtual pages within the * specified run of physical pages to a new physical address. Free the * physical pages underlying the relocated virtual pages. A virtual page * is relocatable if and only if it could be laundered or reclaimed by * the page daemon. Whenever possible, a virtual page is relocated to a * physical address above "high". * * Returns 0 if every physical page within the run was already free or * just freed by a successful relocation. Otherwise, returns a non-zero * value indicating why the last attempt to relocate a virtual page was * unsuccessful. * * "req_class" must be an allocation class. */ static int vm_page_reclaim_run(int req_class, u_long npages, vm_page_t m_run, vm_paddr_t high) { struct mtx *m_mtx, *new_mtx; struct spglist free; vm_object_t object; vm_paddr_t pa; vm_page_t m, m_end, m_new; int error, order, req; KASSERT((req_class & VM_ALLOC_CLASS_MASK) == req_class, ("req_class is not an allocation class")); SLIST_INIT(&free); error = 0; m = m_run; m_end = m_run + npages; m_mtx = NULL; for (; error == 0 && m < m_end; m++) { KASSERT((m->flags & (PG_FICTITIOUS | PG_MARKER)) == 0, ("page %p is PG_FICTITIOUS or PG_MARKER", m)); /* * Avoid releasing and reacquiring the same page lock. */ new_mtx = vm_page_lockptr(m); if (m_mtx != new_mtx) { if (m_mtx != NULL) mtx_unlock(m_mtx); m_mtx = new_mtx; mtx_lock(m_mtx); } retry: if (m->wire_count != 0 || m->hold_count != 0) error = EBUSY; else if ((object = m->object) != NULL) { /* * The page is relocated if and only if it could be * laundered or reclaimed by the page daemon. */ if (!VM_OBJECT_TRYWLOCK(object)) { mtx_unlock(m_mtx); VM_OBJECT_WLOCK(object); mtx_lock(m_mtx); if (m->object != object) { /* * The page may have been freed. */ VM_OBJECT_WUNLOCK(object); goto retry; } else if (m->wire_count != 0 || m->hold_count != 0) { error = EBUSY; goto unlock; } } KASSERT((m->flags & PG_UNHOLDFREE) == 0, ("page %p is PG_UNHOLDFREE", m)); /* Don't care: PG_NODUMP, PG_WINATCFLS, PG_ZERO. */ if (object->type != OBJT_DEFAULT && object->type != OBJT_SWAP && object->type != OBJT_VNODE) error = EINVAL; else if ((m->flags & PG_CACHED) != 0 || m != vm_page_lookup(object, m->pindex)) { /* * The page is cached or recently converted * from cached to free. */ VM_OBJECT_WUNLOCK(object); goto cached; } else if (object->memattr != VM_MEMATTR_DEFAULT) error = EINVAL; else if (m->queue != PQ_NONE && !vm_page_busied(m)) { KASSERT(pmap_page_get_memattr(m) == VM_MEMATTR_DEFAULT, ("page %p has an unexpected memattr", m)); KASSERT((m->oflags & (VPO_SWAPINPROG | VPO_SWAPSLEEP | VPO_UNMANAGED)) == 0, ("page %p has unexpected oflags", m)); /* Don't care: VPO_NOSYNC. */ if (m->valid != 0) { /* * First, try to allocate a new page * that is above "high". Failing * that, try to allocate a new page * that is below "m_run". Allocate * the new page between the end of * "m_run" and "high" only as a last * resort. */ req = req_class | VM_ALLOC_NOOBJ; if ((m->flags & PG_NODUMP) != 0) req |= VM_ALLOC_NODUMP; if (trunc_page(high) != ~(vm_paddr_t)PAGE_MASK) { m_new = vm_page_alloc_contig( NULL, 0, req, 1, round_page(high), ~(vm_paddr_t)0, PAGE_SIZE, 0, VM_MEMATTR_DEFAULT); } else m_new = NULL; if (m_new == NULL) { pa = VM_PAGE_TO_PHYS(m_run); m_new = vm_page_alloc_contig( NULL, 0, req, 1, 0, pa - 1, PAGE_SIZE, 0, VM_MEMATTR_DEFAULT); } if (m_new == NULL) { pa += ptoa(npages); m_new = vm_page_alloc_contig( NULL, 0, req, 1, pa, high, PAGE_SIZE, 0, VM_MEMATTR_DEFAULT); } if (m_new == NULL) { error = ENOMEM; goto unlock; } KASSERT(m_new->wire_count == 0, ("page %p is wired", m)); /* * Replace "m" with the new page. For * vm_page_replace(), "m" must be busy * and dequeued. Finally, change "m" * as if vm_page_free() was called. */ if (object->ref_count != 0) pmap_remove_all(m); m_new->aflags = m->aflags; KASSERT(m_new->oflags == VPO_UNMANAGED, ("page %p is managed", m)); m_new->oflags = m->oflags & VPO_NOSYNC; pmap_copy_page(m, m_new); m_new->valid = m->valid; m_new->dirty = m->dirty; m->flags &= ~PG_ZERO; vm_page_xbusy(m); vm_page_remque(m); vm_page_replace_checked(m_new, object, m->pindex, m); m->valid = 0; vm_page_undirty(m); /* * The new page must be deactivated * before the object is unlocked. */ new_mtx = vm_page_lockptr(m_new); if (m_mtx != new_mtx) { mtx_unlock(m_mtx); m_mtx = new_mtx; mtx_lock(m_mtx); } vm_page_deactivate(m_new); } else { m->flags &= ~PG_ZERO; vm_page_remque(m); vm_page_remove(m); KASSERT(m->dirty == 0, ("page %p is dirty", m)); } SLIST_INSERT_HEAD(&free, m, plinks.s.ss); } else error = EBUSY; unlock: VM_OBJECT_WUNLOCK(object); } else { cached: mtx_lock(&vm_page_queue_free_mtx); order = m->order; if (order < VM_NFREEORDER) { /* * The page is enqueued in the physical memory * allocator's cache/free page queues. * Moreover, it is the first page in a power- * of-two-sized run of contiguous cache/free * pages. Jump ahead to the last page within * that run, and continue from there. */ m += (1 << order) - 1; } #if VM_NRESERVLEVEL > 0 else if (vm_reserv_is_page_free(m)) order = 0; #endif mtx_unlock(&vm_page_queue_free_mtx); if (order == VM_NFREEORDER) error = EINVAL; } } if (m_mtx != NULL) mtx_unlock(m_mtx); if ((m = SLIST_FIRST(&free)) != NULL) { mtx_lock(&vm_page_queue_free_mtx); do { SLIST_REMOVE_HEAD(&free, plinks.s.ss); vm_phys_freecnt_adj(m, 1); #if VM_NRESERVLEVEL > 0 if (!vm_reserv_free_page(m)) #else if (true) #endif vm_phys_free_pages(m, 0); } while ((m = SLIST_FIRST(&free)) != NULL); vm_page_zero_idle_wakeup(); vm_page_free_wakeup(); mtx_unlock(&vm_page_queue_free_mtx); } return (error); } #define NRUNS 16 CTASSERT(powerof2(NRUNS)); #define RUN_INDEX(count) ((count) & (NRUNS - 1)) #define MIN_RECLAIM 8 /* * vm_page_reclaim_contig: * * Reclaim allocated, contiguous physical memory satisfying the specified * conditions by relocating the virtual pages using that physical memory. * Returns true if reclamation is successful and false otherwise. Since * relocation requires the allocation of physical pages, reclamation may * fail due to a shortage of cache/free pages. When reclamation fails, * callers are expected to perform VM_WAIT before retrying a failed * allocation operation, e.g., vm_page_alloc_contig(). * * The caller must always specify an allocation class through "req". * * allocation classes: * VM_ALLOC_NORMAL normal process request * VM_ALLOC_SYSTEM system *really* needs a page * VM_ALLOC_INTERRUPT interrupt time request * * The optional allocation flags are ignored. * * "npages" must be greater than zero. Both "alignment" and "boundary" * must be a power of two. */ bool vm_page_reclaim_contig(int req, u_long npages, vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary) { vm_paddr_t curr_low; vm_page_t m_run, m_runs[NRUNS]; u_long count, reclaimed; int error, i, options, req_class; KASSERT(npages > 0, ("npages is 0")); KASSERT(powerof2(alignment), ("alignment is not a power of 2")); KASSERT(powerof2(boundary), ("boundary is not a power of 2")); req_class = req & VM_ALLOC_CLASS_MASK; /* * The page daemon is allowed to dig deeper into the free page list. */ if (curproc == pageproc && req_class != VM_ALLOC_INTERRUPT) req_class = VM_ALLOC_SYSTEM; /* * Return if the number of cached and free pages cannot satisfy the * requested allocation. */ count = vm_cnt.v_free_count + vm_cnt.v_cache_count; if (count < npages + vm_cnt.v_free_reserved || (count < npages + vm_cnt.v_interrupt_free_min && req_class == VM_ALLOC_SYSTEM) || (count < npages && req_class == VM_ALLOC_INTERRUPT)) return (false); /* * Scan up to three times, relaxing the restrictions ("options") on * the reclamation of reservations and superpages each time. */ for (options = VPSC_NORESERV;;) { /* * Find the highest runs that satisfy the given constraints * and restrictions, and record them in "m_runs". */ curr_low = low; count = 0; for (;;) { m_run = vm_phys_scan_contig(npages, curr_low, high, alignment, boundary, options); if (m_run == NULL) break; curr_low = VM_PAGE_TO_PHYS(m_run) + ptoa(npages); m_runs[RUN_INDEX(count)] = m_run; count++; } /* * Reclaim the highest runs in LIFO (descending) order until * the number of reclaimed pages, "reclaimed", is at least * MIN_RECLAIM. Reset "reclaimed" each time because each * reclamation is idempotent, and runs will (likely) recur * from one scan to the next as restrictions are relaxed. */ reclaimed = 0; for (i = 0; count > 0 && i < NRUNS; i++) { count--; m_run = m_runs[RUN_INDEX(count)]; error = vm_page_reclaim_run(req_class, npages, m_run, high); if (error == 0) { reclaimed += npages; if (reclaimed >= MIN_RECLAIM) return (true); } } /* * Either relax the restrictions on the next scan or return if * the last scan had no restrictions. */ if (options == VPSC_NORESERV) options = VPSC_NOSUPER; else if (options == VPSC_NOSUPER) options = VPSC_ANY; else if (options == VPSC_ANY) return (reclaimed != 0); } } /* * vm_wait: (also see VM_WAIT macro) * * Sleep until free pages are available for allocation. * - Called in various places before memory allocations. */ void vm_wait(void) { mtx_lock(&vm_page_queue_free_mtx); if (curproc == pageproc) { vm_pageout_pages_needed = 1; msleep(&vm_pageout_pages_needed, &vm_page_queue_free_mtx, PDROP | PSWP, "VMWait", 0); } else { if (!vm_pageout_wanted) { vm_pageout_wanted = true; wakeup(&vm_pageout_wanted); } vm_pages_needed = true; msleep(&vm_cnt.v_free_count, &vm_page_queue_free_mtx, PDROP | PVM, "vmwait", 0); } } /* * vm_waitpfault: (also see VM_WAITPFAULT macro) * * Sleep until free pages are available for allocation. * - Called only in vm_fault so that processes page faulting * can be easily tracked. * - Sleeps at a lower priority than vm_wait() so that vm_wait()ing * processes will be able to grab memory first. Do not change * this balance without careful testing first. */ void vm_waitpfault(void) { mtx_lock(&vm_page_queue_free_mtx); if (!vm_pageout_wanted) { vm_pageout_wanted = true; wakeup(&vm_pageout_wanted); } vm_pages_needed = true; msleep(&vm_cnt.v_free_count, &vm_page_queue_free_mtx, PDROP | PUSER, "pfault", 0); } struct vm_pagequeue * vm_page_pagequeue(vm_page_t m) { return (&vm_phys_domain(m)->vmd_pagequeues[m->queue]); } /* * vm_page_dequeue: * * Remove the given page from its current page queue. * * The page must be locked. */ void vm_page_dequeue(vm_page_t m) { struct vm_pagequeue *pq; vm_page_assert_locked(m); KASSERT(m->queue < PQ_COUNT, ("vm_page_dequeue: page %p is not queued", m)); pq = vm_page_pagequeue(m); vm_pagequeue_lock(pq); m->queue = PQ_NONE; TAILQ_REMOVE(&pq->pq_pl, m, plinks.q); vm_pagequeue_cnt_dec(pq); vm_pagequeue_unlock(pq); } /* * vm_page_dequeue_locked: * * Remove the given page from its current page queue. * * The page and page queue must be locked. */ void vm_page_dequeue_locked(vm_page_t m) { struct vm_pagequeue *pq; vm_page_lock_assert(m, MA_OWNED); pq = vm_page_pagequeue(m); vm_pagequeue_assert_locked(pq); m->queue = PQ_NONE; TAILQ_REMOVE(&pq->pq_pl, m, plinks.q); vm_pagequeue_cnt_dec(pq); } /* * vm_page_enqueue: * * Add the given page to the specified page queue. * * The page must be locked. */ static void vm_page_enqueue(uint8_t queue, vm_page_t m) { struct vm_pagequeue *pq; vm_page_lock_assert(m, MA_OWNED); KASSERT(queue < PQ_COUNT, ("vm_page_enqueue: invalid queue %u request for page %p", queue, m)); pq = &vm_phys_domain(m)->vmd_pagequeues[queue]; vm_pagequeue_lock(pq); m->queue = queue; TAILQ_INSERT_TAIL(&pq->pq_pl, m, plinks.q); vm_pagequeue_cnt_inc(pq); vm_pagequeue_unlock(pq); } /* * vm_page_requeue: * * Move the given page to the tail of its current page queue. * * The page must be locked. */ void vm_page_requeue(vm_page_t m) { struct vm_pagequeue *pq; vm_page_lock_assert(m, MA_OWNED); KASSERT(m->queue != PQ_NONE, ("vm_page_requeue: page %p is not queued", m)); pq = vm_page_pagequeue(m); vm_pagequeue_lock(pq); TAILQ_REMOVE(&pq->pq_pl, m, plinks.q); TAILQ_INSERT_TAIL(&pq->pq_pl, m, plinks.q); vm_pagequeue_unlock(pq); } /* * vm_page_requeue_locked: * * Move the given page to the tail of its current page queue. * * The page queue must be locked. */ void vm_page_requeue_locked(vm_page_t m) { struct vm_pagequeue *pq; KASSERT(m->queue != PQ_NONE, ("vm_page_requeue_locked: page %p is not queued", m)); pq = vm_page_pagequeue(m); vm_pagequeue_assert_locked(pq); TAILQ_REMOVE(&pq->pq_pl, m, plinks.q); TAILQ_INSERT_TAIL(&pq->pq_pl, m, plinks.q); } /* * vm_page_activate: * * Put the specified page on the active list (if appropriate). * Ensure that act_count is at least ACT_INIT but do not otherwise * mess with it. * * The page must be locked. */ void vm_page_activate(vm_page_t m) { int queue; vm_page_lock_assert(m, MA_OWNED); if ((queue = m->queue) != PQ_ACTIVE) { if (m->wire_count == 0 && (m->oflags & VPO_UNMANAGED) == 0) { if (m->act_count < ACT_INIT) m->act_count = ACT_INIT; if (queue != PQ_NONE) vm_page_dequeue(m); vm_page_enqueue(PQ_ACTIVE, m); } else KASSERT(queue == PQ_NONE, ("vm_page_activate: wired page %p is queued", m)); } else { if (m->act_count < ACT_INIT) m->act_count = ACT_INIT; } } /* * vm_page_free_wakeup: * * Helper routine for vm_page_free_toq() and vm_page_cache(). This * routine is called when a page has been added to the cache or free * queues. * * The page queues must be locked. */ static inline void vm_page_free_wakeup(void) { mtx_assert(&vm_page_queue_free_mtx, MA_OWNED); /* * if pageout daemon needs pages, then tell it that there are * some free. */ if (vm_pageout_pages_needed && vm_cnt.v_cache_count + vm_cnt.v_free_count >= vm_cnt.v_pageout_free_min) { wakeup(&vm_pageout_pages_needed); vm_pageout_pages_needed = 0; } /* * wakeup processes that are waiting on memory if we hit a * high water mark. And wakeup scheduler process if we have * lots of memory. this process will swapin processes. */ if (vm_pages_needed && !vm_page_count_min()) { vm_pages_needed = false; wakeup(&vm_cnt.v_free_count); } } /* * Turn a cached page into a free page, by changing its attributes. * Keep the statistics up-to-date. * * The free page queue must be locked. */ static void vm_page_cache_turn_free(vm_page_t m) { mtx_assert(&vm_page_queue_free_mtx, MA_OWNED); m->object = NULL; m->valid = 0; KASSERT((m->flags & PG_CACHED) != 0, ("vm_page_cache_turn_free: page %p is not cached", m)); m->flags &= ~PG_CACHED; vm_cnt.v_cache_count--; vm_phys_freecnt_adj(m, 1); } /* * vm_page_free_toq: * * Returns the given page to the free list, * disassociating it with any VM object. * * The object must be locked. The page must be locked if it is managed. */ void vm_page_free_toq(vm_page_t m) { if ((m->oflags & VPO_UNMANAGED) == 0) { vm_page_lock_assert(m, MA_OWNED); KASSERT(!pmap_page_is_mapped(m), ("vm_page_free_toq: freeing mapped page %p", m)); } else KASSERT(m->queue == PQ_NONE, ("vm_page_free_toq: unmanaged page %p is queued", m)); PCPU_INC(cnt.v_tfree); if (vm_page_sbusied(m)) panic("vm_page_free: freeing busy page %p", m); /* * Unqueue, then remove page. Note that we cannot destroy * the page here because we do not want to call the pager's * callback routine until after we've put the page on the * appropriate free queue. */ vm_page_remque(m); vm_page_remove(m); /* * If fictitious remove object association and * return, otherwise delay object association removal. */ if ((m->flags & PG_FICTITIOUS) != 0) { return; } m->valid = 0; vm_page_undirty(m); if (m->wire_count != 0) panic("vm_page_free: freeing wired page %p", m); if (m->hold_count != 0) { m->flags &= ~PG_ZERO; KASSERT((m->flags & PG_UNHOLDFREE) == 0, ("vm_page_free: freeing PG_UNHOLDFREE page %p", m)); m->flags |= PG_UNHOLDFREE; } else { /* * Restore the default memory attribute to the page. */ if (pmap_page_get_memattr(m) != VM_MEMATTR_DEFAULT) pmap_page_set_memattr(m, VM_MEMATTR_DEFAULT); /* * Insert the page into the physical memory allocator's * cache/free page queues. */ mtx_lock(&vm_page_queue_free_mtx); vm_phys_freecnt_adj(m, 1); #if VM_NRESERVLEVEL > 0 if (!vm_reserv_free_page(m)) #else if (TRUE) #endif vm_phys_free_pages(m, 0); if ((m->flags & PG_ZERO) != 0) ++vm_page_zero_count; else vm_page_zero_idle_wakeup(); vm_page_free_wakeup(); mtx_unlock(&vm_page_queue_free_mtx); } } /* * vm_page_wire: * * Mark this page as wired down by yet * another map, removing it from paging queues * as necessary. * * If the page is fictitious, then its wire count must remain one. * * The page must be locked. */ void vm_page_wire(vm_page_t m) { /* * Only bump the wire statistics if the page is not already wired, * and only unqueue the page if it is on some queue (if it is unmanaged * it is already off the queues). */ vm_page_lock_assert(m, MA_OWNED); if ((m->flags & PG_FICTITIOUS) != 0) { KASSERT(m->wire_count == 1, ("vm_page_wire: fictitious page %p's wire count isn't one", m)); return; } if (m->wire_count == 0) { KASSERT((m->oflags & VPO_UNMANAGED) == 0 || m->queue == PQ_NONE, ("vm_page_wire: unmanaged page %p is queued", m)); vm_page_remque(m); atomic_add_int(&vm_cnt.v_wire_count, 1); } m->wire_count++; KASSERT(m->wire_count != 0, ("vm_page_wire: wire_count overflow m=%p", m)); } /* * vm_page_unwire: * * Release one wiring of the specified page, potentially allowing it to be * paged out. Returns TRUE if the number of wirings transitions to zero and * FALSE otherwise. * * Only managed pages belonging to an object can be paged out. If the number * of wirings transitions to zero and the page is eligible for page out, then * the page is added to the specified paging queue (unless PQ_NONE is * specified). * * If a page is fictitious, then its wire count must always be one. * * A managed page must be locked. */ boolean_t vm_page_unwire(vm_page_t m, uint8_t queue) { KASSERT(queue < PQ_COUNT || queue == PQ_NONE, ("vm_page_unwire: invalid queue %u request for page %p", queue, m)); if ((m->oflags & VPO_UNMANAGED) == 0) vm_page_assert_locked(m); if ((m->flags & PG_FICTITIOUS) != 0) { KASSERT(m->wire_count == 1, ("vm_page_unwire: fictitious page %p's wire count isn't one", m)); return (FALSE); } if (m->wire_count > 0) { m->wire_count--; if (m->wire_count == 0) { atomic_subtract_int(&vm_cnt.v_wire_count, 1); if ((m->oflags & VPO_UNMANAGED) == 0 && m->object != NULL && queue != PQ_NONE) { if (queue == PQ_INACTIVE) m->flags &= ~PG_WINATCFLS; vm_page_enqueue(queue, m); } return (TRUE); } else return (FALSE); } else panic("vm_page_unwire: page %p's wire count is zero", m); } /* * Move the specified page to the inactive queue. * * Many pages placed on the inactive queue should actually go * into the cache, but it is difficult to figure out which. What * we do instead, if the inactive target is well met, is to put * clean pages at the head of the inactive queue instead of the tail. * This will cause them to be moved to the cache more quickly and * if not actively re-referenced, reclaimed more quickly. If we just * stick these pages at the end of the inactive queue, heavy filesystem * meta-data accesses can cause an unnecessary paging load on memory bound * processes. This optimization causes one-time-use metadata to be * reused more quickly. * * Normally noreuse is FALSE, resulting in LRU operation. noreuse is set * to TRUE if we want this page to be 'as if it were placed in the cache', * except without unmapping it from the process address space. In * practice this is implemented by inserting the page at the head of the * queue, using a marker page to guide FIFO insertion ordering. * * The page must be locked. */ static inline void _vm_page_deactivate(vm_page_t m, boolean_t noreuse) { struct vm_pagequeue *pq; int queue; vm_page_assert_locked(m); /* * Ignore if the page is already inactive, unless it is unlikely to be * reactivated. */ if ((queue = m->queue) == PQ_INACTIVE && !noreuse) return; if (m->wire_count == 0 && (m->oflags & VPO_UNMANAGED) == 0) { pq = &vm_phys_domain(m)->vmd_pagequeues[PQ_INACTIVE]; /* Avoid multiple acquisitions of the inactive queue lock. */ if (queue == PQ_INACTIVE) { vm_pagequeue_lock(pq); vm_page_dequeue_locked(m); } else { if (queue != PQ_NONE) vm_page_dequeue(m); m->flags &= ~PG_WINATCFLS; vm_pagequeue_lock(pq); } m->queue = PQ_INACTIVE; if (noreuse) TAILQ_INSERT_BEFORE(&vm_phys_domain(m)->vmd_inacthead, m, plinks.q); else TAILQ_INSERT_TAIL(&pq->pq_pl, m, plinks.q); vm_pagequeue_cnt_inc(pq); vm_pagequeue_unlock(pq); } } /* * Move the specified page to the inactive queue. * * The page must be locked. */ void vm_page_deactivate(vm_page_t m) { _vm_page_deactivate(m, FALSE); } /* * Move the specified page to the inactive queue with the expectation * that it is unlikely to be reused. * * The page must be locked. */ void vm_page_deactivate_noreuse(vm_page_t m) { _vm_page_deactivate(m, TRUE); } /* * vm_page_try_to_cache: * * Returns 0 on failure, 1 on success */ int vm_page_try_to_cache(vm_page_t m) { vm_page_lock_assert(m, MA_OWNED); VM_OBJECT_ASSERT_WLOCKED(m->object); if (m->dirty || m->hold_count || m->wire_count || (m->oflags & VPO_UNMANAGED) != 0 || vm_page_busied(m)) return (0); pmap_remove_all(m); if (m->dirty) return (0); vm_page_cache(m); return (1); } /* * vm_page_try_to_free() * * Attempt to free the page. If we cannot free it, we do nothing. * 1 is returned on success, 0 on failure. */ int vm_page_try_to_free(vm_page_t m) { vm_page_lock_assert(m, MA_OWNED); if (m->object != NULL) VM_OBJECT_ASSERT_WLOCKED(m->object); if (m->dirty || m->hold_count || m->wire_count || (m->oflags & VPO_UNMANAGED) != 0 || vm_page_busied(m)) return (0); pmap_remove_all(m); if (m->dirty) return (0); vm_page_free(m); return (1); } /* * vm_page_cache * * Put the specified page onto the page cache queue (if appropriate). * * The object and page must be locked. */ void vm_page_cache(vm_page_t m) { vm_object_t object; boolean_t cache_was_empty; vm_page_lock_assert(m, MA_OWNED); object = m->object; VM_OBJECT_ASSERT_WLOCKED(object); if (vm_page_busied(m) || (m->oflags & VPO_UNMANAGED) || m->hold_count || m->wire_count) panic("vm_page_cache: attempting to cache busy page"); KASSERT(!pmap_page_is_mapped(m), ("vm_page_cache: page %p is mapped", m)); KASSERT(m->dirty == 0, ("vm_page_cache: page %p is dirty", m)); if (m->valid == 0 || object->type == OBJT_DEFAULT || (object->type == OBJT_SWAP && !vm_pager_has_page(object, m->pindex, NULL, NULL))) { /* * Hypothesis: A cache-eligible page belonging to a * default object or swap object but without a backing * store must be zero filled. */ vm_page_free(m); return; } KASSERT((m->flags & PG_CACHED) == 0, ("vm_page_cache: page %p is already cached", m)); /* * Remove the page from the paging queues. */ vm_page_remque(m); /* * Remove the page from the object's collection of resident * pages. */ vm_radix_remove(&object->rtree, m->pindex); TAILQ_REMOVE(&object->memq, m, listq); object->resident_page_count--; /* * Restore the default memory attribute to the page. */ if (pmap_page_get_memattr(m) != VM_MEMATTR_DEFAULT) pmap_page_set_memattr(m, VM_MEMATTR_DEFAULT); /* * Insert the page into the object's collection of cached pages * and the physical memory allocator's cache/free page queues. */ m->flags &= ~PG_ZERO; mtx_lock(&vm_page_queue_free_mtx); cache_was_empty = vm_radix_is_empty(&object->cache); if (vm_radix_insert(&object->cache, m)) { mtx_unlock(&vm_page_queue_free_mtx); if (object->type == OBJT_VNODE && object->resident_page_count == 0) vdrop(object->handle); m->object = NULL; vm_page_free(m); return; } /* * The above call to vm_radix_insert() could reclaim the one pre- * existing cached page from this object, resulting in a call to * vdrop(). */ if (!cache_was_empty) cache_was_empty = vm_radix_is_singleton(&object->cache); m->flags |= PG_CACHED; vm_cnt.v_cache_count++; PCPU_INC(cnt.v_tcached); #if VM_NRESERVLEVEL > 0 if (!vm_reserv_free_page(m)) { #else if (TRUE) { #endif vm_phys_free_pages(m, 0); } vm_page_free_wakeup(); mtx_unlock(&vm_page_queue_free_mtx); /* * Increment the vnode's hold count if this is the object's only * cached page. Decrement the vnode's hold count if this was * the object's only resident page. */ if (object->type == OBJT_VNODE) { if (cache_was_empty && object->resident_page_count != 0) vhold(object->handle); else if (!cache_was_empty && object->resident_page_count == 0) vdrop(object->handle); } } /* * vm_page_advise * * Deactivate or do nothing, as appropriate. * * The object and page must be locked. */ void vm_page_advise(vm_page_t m, int advice) { vm_page_assert_locked(m); VM_OBJECT_ASSERT_WLOCKED(m->object); if (advice == MADV_FREE) /* * Mark the page clean. This will allow the page to be freed * up by the system. However, such pages are often reused * quickly by malloc() so we do not do anything that would * cause a page fault if we can help it. * * Specifically, we do not try to actually free the page now * nor do we try to put it in the cache (which would cause a * page fault on reuse). * * But we do make the page as freeable as we can without * actually taking the step of unmapping it. */ m->dirty = 0; else if (advice != MADV_DONTNEED) return; /* * Clear any references to the page. Otherwise, the page daemon will * immediately reactivate the page. */ vm_page_aflag_clear(m, PGA_REFERENCED); if (advice != MADV_FREE && m->dirty == 0 && pmap_is_modified(m)) vm_page_dirty(m); /* * Place clean pages at the head of the inactive queue rather than the * tail, thus defeating the queue's LRU operation and ensuring that the * page will be reused quickly. */ _vm_page_deactivate(m, m->dirty == 0); } /* * Grab a page, waiting until we are waken up due to the page * changing state. We keep on waiting, if the page continues * to be in the object. If the page doesn't exist, first allocate it * and then conditionally zero it. * * This routine may sleep. * * The object must be locked on entry. The lock will, however, be released * and reacquired if the routine sleeps. */ vm_page_t vm_page_grab(vm_object_t object, vm_pindex_t pindex, int allocflags) { vm_page_t m; int sleep; VM_OBJECT_ASSERT_WLOCKED(object); KASSERT((allocflags & VM_ALLOC_SBUSY) == 0 || (allocflags & VM_ALLOC_IGN_SBUSY) != 0, ("vm_page_grab: VM_ALLOC_SBUSY/VM_ALLOC_IGN_SBUSY mismatch")); retrylookup: if ((m = vm_page_lookup(object, pindex)) != NULL) { sleep = (allocflags & VM_ALLOC_IGN_SBUSY) != 0 ? vm_page_xbusied(m) : vm_page_busied(m); if (sleep) { if ((allocflags & VM_ALLOC_NOWAIT) != 0) return (NULL); /* * Reference the page before unlocking and * sleeping so that the page daemon is less * likely to reclaim it. */ vm_page_aflag_set(m, PGA_REFERENCED); vm_page_lock(m); VM_OBJECT_WUNLOCK(object); vm_page_busy_sleep(m, "pgrbwt"); VM_OBJECT_WLOCK(object); goto retrylookup; } else { if ((allocflags & VM_ALLOC_WIRED) != 0) { vm_page_lock(m); vm_page_wire(m); vm_page_unlock(m); } if ((allocflags & (VM_ALLOC_NOBUSY | VM_ALLOC_SBUSY)) == 0) vm_page_xbusy(m); if ((allocflags & VM_ALLOC_SBUSY) != 0) vm_page_sbusy(m); return (m); } } m = vm_page_alloc(object, pindex, allocflags); if (m == NULL) { if ((allocflags & VM_ALLOC_NOWAIT) != 0) return (NULL); VM_OBJECT_WUNLOCK(object); VM_WAIT; VM_OBJECT_WLOCK(object); goto retrylookup; } else if (m->valid != 0) return (m); if (allocflags & VM_ALLOC_ZERO && (m->flags & PG_ZERO) == 0) pmap_zero_page(m); return (m); } /* * Mapping function for valid or dirty bits in a page. * * Inputs are required to range within a page. */ vm_page_bits_t vm_page_bits(int base, int size) { int first_bit; int last_bit; KASSERT( base + size <= PAGE_SIZE, ("vm_page_bits: illegal base/size %d/%d", base, size) ); if (size == 0) /* handle degenerate case */ return (0); first_bit = base >> DEV_BSHIFT; last_bit = (base + size - 1) >> DEV_BSHIFT; return (((vm_page_bits_t)2 << last_bit) - ((vm_page_bits_t)1 << first_bit)); } /* * vm_page_set_valid_range: * * Sets portions of a page valid. The arguments are expected * to be DEV_BSIZE aligned but if they aren't the bitmap is inclusive * of any partial chunks touched by the range. The invalid portion of * such chunks will be zeroed. * * (base + size) must be less then or equal to PAGE_SIZE. */ void vm_page_set_valid_range(vm_page_t m, int base, int size) { int endoff, frag; VM_OBJECT_ASSERT_WLOCKED(m->object); if (size == 0) /* handle degenerate case */ return; /* * If the base is not DEV_BSIZE aligned and the valid * bit is clear, we have to zero out a portion of the * first block. */ if ((frag = rounddown2(base, DEV_BSIZE)) != base && (m->valid & (1 << (base >> DEV_BSHIFT))) == 0) pmap_zero_page_area(m, frag, base - frag); /* * If the ending offset is not DEV_BSIZE aligned and the * valid bit is clear, we have to zero out a portion of * the last block. */ endoff = base + size; if ((frag = rounddown2(endoff, DEV_BSIZE)) != endoff && (m->valid & (1 << (endoff >> DEV_BSHIFT))) == 0) pmap_zero_page_area(m, endoff, DEV_BSIZE - (endoff & (DEV_BSIZE - 1))); /* * Assert that no previously invalid block that is now being validated * is already dirty. */ KASSERT((~m->valid & vm_page_bits(base, size) & m->dirty) == 0, ("vm_page_set_valid_range: page %p is dirty", m)); /* * Set valid bits inclusive of any overlap. */ m->valid |= vm_page_bits(base, size); } /* * Clear the given bits from the specified page's dirty field. */ static __inline void vm_page_clear_dirty_mask(vm_page_t m, vm_page_bits_t pagebits) { uintptr_t addr; #if PAGE_SIZE < 16384 int shift; #endif /* * If the object is locked and the page is neither exclusive busy nor * write mapped, then the page's dirty field cannot possibly be * set by a concurrent pmap operation. */ VM_OBJECT_ASSERT_WLOCKED(m->object); if (!vm_page_xbusied(m) && !pmap_page_is_write_mapped(m)) m->dirty &= ~pagebits; else { /* * The pmap layer can call vm_page_dirty() without * holding a distinguished lock. The combination of * the object's lock and an atomic operation suffice * to guarantee consistency of the page dirty field. * * For PAGE_SIZE == 32768 case, compiler already * properly aligns the dirty field, so no forcible * alignment is needed. Only require existence of * atomic_clear_64 when page size is 32768. */ addr = (uintptr_t)&m->dirty; #if PAGE_SIZE == 32768 atomic_clear_64((uint64_t *)addr, pagebits); #elif PAGE_SIZE == 16384 atomic_clear_32((uint32_t *)addr, pagebits); #else /* PAGE_SIZE <= 8192 */ /* * Use a trick to perform a 32-bit atomic on the * containing aligned word, to not depend on the existence * of atomic_clear_{8, 16}. */ shift = addr & (sizeof(uint32_t) - 1); #if BYTE_ORDER == BIG_ENDIAN shift = (sizeof(uint32_t) - sizeof(m->dirty) - shift) * NBBY; #else shift *= NBBY; #endif addr &= ~(sizeof(uint32_t) - 1); atomic_clear_32((uint32_t *)addr, pagebits << shift); #endif /* PAGE_SIZE */ } } /* * vm_page_set_validclean: * * Sets portions of a page valid and clean. The arguments are expected * to be DEV_BSIZE aligned but if they aren't the bitmap is inclusive * of any partial chunks touched by the range. The invalid portion of * such chunks will be zero'd. * * (base + size) must be less then or equal to PAGE_SIZE. */ void vm_page_set_validclean(vm_page_t m, int base, int size) { vm_page_bits_t oldvalid, pagebits; int endoff, frag; VM_OBJECT_ASSERT_WLOCKED(m->object); if (size == 0) /* handle degenerate case */ return; /* * If the base is not DEV_BSIZE aligned and the valid * bit is clear, we have to zero out a portion of the * first block. */ if ((frag = rounddown2(base, DEV_BSIZE)) != base && (m->valid & ((vm_page_bits_t)1 << (base >> DEV_BSHIFT))) == 0) pmap_zero_page_area(m, frag, base - frag); /* * If the ending offset is not DEV_BSIZE aligned and the * valid bit is clear, we have to zero out a portion of * the last block. */ endoff = base + size; if ((frag = rounddown2(endoff, DEV_BSIZE)) != endoff && (m->valid & ((vm_page_bits_t)1 << (endoff >> DEV_BSHIFT))) == 0) pmap_zero_page_area(m, endoff, DEV_BSIZE - (endoff & (DEV_BSIZE - 1))); /* * Set valid, clear dirty bits. If validating the entire * page we can safely clear the pmap modify bit. We also * use this opportunity to clear the VPO_NOSYNC flag. If a process * takes a write fault on a MAP_NOSYNC memory area the flag will * be set again. * * We set valid bits inclusive of any overlap, but we can only * clear dirty bits for DEV_BSIZE chunks that are fully within * the range. */ oldvalid = m->valid; pagebits = vm_page_bits(base, size); m->valid |= pagebits; #if 0 /* NOT YET */ if ((frag = base & (DEV_BSIZE - 1)) != 0) { frag = DEV_BSIZE - frag; base += frag; size -= frag; if (size < 0) size = 0; } pagebits = vm_page_bits(base, size & (DEV_BSIZE - 1)); #endif if (base == 0 && size == PAGE_SIZE) { /* * The page can only be modified within the pmap if it is * mapped, and it can only be mapped if it was previously * fully valid. */ if (oldvalid == VM_PAGE_BITS_ALL) /* * Perform the pmap_clear_modify() first. Otherwise, * a concurrent pmap operation, such as * pmap_protect(), could clear a modification in the * pmap and set the dirty field on the page before * pmap_clear_modify() had begun and after the dirty * field was cleared here. */ pmap_clear_modify(m); m->dirty = 0; m->oflags &= ~VPO_NOSYNC; } else if (oldvalid != VM_PAGE_BITS_ALL) m->dirty &= ~pagebits; else vm_page_clear_dirty_mask(m, pagebits); } void vm_page_clear_dirty(vm_page_t m, int base, int size) { vm_page_clear_dirty_mask(m, vm_page_bits(base, size)); } /* * vm_page_set_invalid: * * Invalidates DEV_BSIZE'd chunks within a page. Both the * valid and dirty bits for the effected areas are cleared. */ void vm_page_set_invalid(vm_page_t m, int base, int size) { vm_page_bits_t bits; vm_object_t object; object = m->object; VM_OBJECT_ASSERT_WLOCKED(object); if (object->type == OBJT_VNODE && base == 0 && IDX_TO_OFF(m->pindex) + size >= object->un_pager.vnp.vnp_size) bits = VM_PAGE_BITS_ALL; else bits = vm_page_bits(base, size); if (object->ref_count != 0 && m->valid == VM_PAGE_BITS_ALL && bits != 0) pmap_remove_all(m); KASSERT((bits == 0 && m->valid == VM_PAGE_BITS_ALL) || !pmap_page_is_mapped(m), ("vm_page_set_invalid: page %p is mapped", m)); m->valid &= ~bits; m->dirty &= ~bits; } /* * vm_page_zero_invalid() * * The kernel assumes that the invalid portions of a page contain * garbage, but such pages can be mapped into memory by user code. * When this occurs, we must zero out the non-valid portions of the * page so user code sees what it expects. * * Pages are most often semi-valid when the end of a file is mapped * into memory and the file's size is not page aligned. */ void vm_page_zero_invalid(vm_page_t m, boolean_t setvalid) { int b; int i; VM_OBJECT_ASSERT_WLOCKED(m->object); /* * Scan the valid bits looking for invalid sections that * must be zeroed. Invalid sub-DEV_BSIZE'd areas ( where the * valid bit may be set ) have already been zeroed by * vm_page_set_validclean(). */ for (b = i = 0; i <= PAGE_SIZE / DEV_BSIZE; ++i) { if (i == (PAGE_SIZE / DEV_BSIZE) || (m->valid & ((vm_page_bits_t)1 << i))) { if (i > b) { pmap_zero_page_area(m, b << DEV_BSHIFT, (i - b) << DEV_BSHIFT); } b = i + 1; } } /* * setvalid is TRUE when we can safely set the zero'd areas * as being valid. We can do this if there are no cache consistancy * issues. e.g. it is ok to do with UFS, but not ok to do with NFS. */ if (setvalid) m->valid = VM_PAGE_BITS_ALL; } /* * vm_page_is_valid: * * Is (partial) page valid? Note that the case where size == 0 * will return FALSE in the degenerate case where the page is * entirely invalid, and TRUE otherwise. */ int vm_page_is_valid(vm_page_t m, int base, int size) { vm_page_bits_t bits; VM_OBJECT_ASSERT_LOCKED(m->object); bits = vm_page_bits(base, size); return (m->valid != 0 && (m->valid & bits) == bits); } /* * vm_page_ps_is_valid: * * Returns TRUE if the entire (super)page is valid and FALSE otherwise. */ boolean_t vm_page_ps_is_valid(vm_page_t m) { int i, npages; VM_OBJECT_ASSERT_LOCKED(m->object); npages = atop(pagesizes[m->psind]); /* * The physically contiguous pages that make up a superpage, i.e., a * page with a page size index ("psind") greater than zero, will * occupy adjacent entries in vm_page_array[]. */ for (i = 0; i < npages; i++) { if (m[i].valid != VM_PAGE_BITS_ALL) return (FALSE); } return (TRUE); } /* * Set the page's dirty bits if the page is modified. */ void vm_page_test_dirty(vm_page_t m) { VM_OBJECT_ASSERT_WLOCKED(m->object); if (m->dirty != VM_PAGE_BITS_ALL && pmap_is_modified(m)) vm_page_dirty(m); } void vm_page_lock_KBI(vm_page_t m, const char *file, int line) { mtx_lock_flags_(vm_page_lockptr(m), 0, file, line); } void vm_page_unlock_KBI(vm_page_t m, const char *file, int line) { mtx_unlock_flags_(vm_page_lockptr(m), 0, file, line); } int vm_page_trylock_KBI(vm_page_t m, const char *file, int line) { return (mtx_trylock_flags_(vm_page_lockptr(m), 0, file, line)); } #if defined(INVARIANTS) || defined(INVARIANT_SUPPORT) void vm_page_assert_locked_KBI(vm_page_t m, const char *file, int line) { vm_page_lock_assert_KBI(m, MA_OWNED, file, line); } void vm_page_lock_assert_KBI(vm_page_t m, int a, const char *file, int line) { mtx_assert_(vm_page_lockptr(m), a, file, line); } #endif #ifdef INVARIANTS void vm_page_object_lock_assert(vm_page_t m) { /* * Certain of the page's fields may only be modified by the * holder of the containing object's lock or the exclusive busy. * holder. Unfortunately, the holder of the write busy is * not recorded, and thus cannot be checked here. */ if (m->object != NULL && !vm_page_xbusied(m)) VM_OBJECT_ASSERT_WLOCKED(m->object); } void vm_page_assert_pga_writeable(vm_page_t m, uint8_t bits) { if ((bits & PGA_WRITEABLE) == 0) return; /* * The PGA_WRITEABLE flag can only be set if the page is * managed, is exclusively busied or the object is locked. * Currently, this flag is only set by pmap_enter(). */ KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("PGA_WRITEABLE on unmanaged page")); if (!vm_page_xbusied(m)) VM_OBJECT_ASSERT_LOCKED(m->object); } #endif #include "opt_ddb.h" #ifdef DDB #include #include DB_SHOW_COMMAND(page, vm_page_print_page_info) { db_printf("vm_cnt.v_free_count: %d\n", vm_cnt.v_free_count); db_printf("vm_cnt.v_cache_count: %d\n", vm_cnt.v_cache_count); db_printf("vm_cnt.v_inactive_count: %d\n", vm_cnt.v_inactive_count); db_printf("vm_cnt.v_active_count: %d\n", vm_cnt.v_active_count); db_printf("vm_cnt.v_wire_count: %d\n", vm_cnt.v_wire_count); db_printf("vm_cnt.v_free_reserved: %d\n", vm_cnt.v_free_reserved); db_printf("vm_cnt.v_free_min: %d\n", vm_cnt.v_free_min); db_printf("vm_cnt.v_free_target: %d\n", vm_cnt.v_free_target); db_printf("vm_cnt.v_inactive_target: %d\n", vm_cnt.v_inactive_target); } DB_SHOW_COMMAND(pageq, vm_page_print_pageq_info) { int dom; db_printf("pq_free %d pq_cache %d\n", vm_cnt.v_free_count, vm_cnt.v_cache_count); for (dom = 0; dom < vm_ndomains; dom++) { db_printf( "dom %d page_cnt %d free %d pq_act %d pq_inact %d pass %d\n", dom, vm_dom[dom].vmd_page_count, vm_dom[dom].vmd_free_count, vm_dom[dom].vmd_pagequeues[PQ_ACTIVE].pq_cnt, vm_dom[dom].vmd_pagequeues[PQ_INACTIVE].pq_cnt, vm_dom[dom].vmd_pass); } } DB_SHOW_COMMAND(pginfo, vm_page_print_pginfo) { vm_page_t m; boolean_t phys; if (!have_addr) { db_printf("show pginfo addr\n"); return; } phys = strchr(modif, 'p') != NULL; if (phys) m = PHYS_TO_VM_PAGE(addr); else m = (vm_page_t)addr; db_printf( "page %p obj %p pidx 0x%jx phys 0x%jx q %d hold %d wire %d\n" " af 0x%x of 0x%x f 0x%x act %d busy %x valid 0x%x dirty 0x%x\n", m, m->object, (uintmax_t)m->pindex, (uintmax_t)m->phys_addr, m->queue, m->hold_count, m->wire_count, m->aflags, m->oflags, m->flags, m->act_count, m->busy_lock, m->valid, m->dirty); } #endif /* DDB */ Index: projects/vnet/sys/vm/vm_page.h =================================================================== --- projects/vnet/sys/vm/vm_page.h (revision 302157) +++ projects/vnet/sys/vm/vm_page.h (revision 302158) @@ -1,701 +1,702 @@ /*- * Copyright (c) 1991, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * The Mach Operating System project at Carnegie-Mellon University. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * from: @(#)vm_page.h 8.2 (Berkeley) 12/13/93 * * * Copyright (c) 1987, 1990 Carnegie-Mellon University. * All rights reserved. * * Authors: Avadis Tevanian, Jr., Michael Wayne Young * * Permission to use, copy, modify and distribute this software and * its documentation is hereby granted, provided that both the copyright * notice and this permission notice appear in all copies of the * software, derivative works or modified versions, and any portions * thereof, and that both notices appear in supporting documentation. * * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. * * Carnegie Mellon requests users of this software to return to * * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU * School of Computer Science * Carnegie Mellon University * Pittsburgh PA 15213-3890 * * any improvements or extensions that they make and grant Carnegie the * rights to redistribute these changes. * * $FreeBSD$ */ /* * Resident memory system definitions. */ #ifndef _VM_PAGE_ #define _VM_PAGE_ #include /* * Management of resident (logical) pages. * * A small structure is kept for each resident * page, indexed by page number. Each structure * is an element of several collections: * * A radix tree used to quickly * perform object/offset lookups * * A list of all pages for a given object, * so they can be quickly deactivated at * time of deallocation. * * An ordered list of pages due for pageout. * * In addition, the structure contains the object * and offset to which this page belongs (for pageout), * and sundry status bits. * * In general, operations on this structure's mutable fields are * synchronized using either one of or a combination of the lock on the * object that the page belongs to (O), the pool lock for the page (P), * or the lock for either the free or paging queue (Q). If a field is * annotated below with two of these locks, then holding either lock is * sufficient for read access, but both locks are required for write * access. * * In contrast, the synchronization of accesses to the page's * dirty field is machine dependent (M). In the * machine-independent layer, the lock on the object that the * page belongs to must be held in order to operate on the field. * However, the pmap layer is permitted to set all bits within * the field without holding that lock. If the underlying * architecture does not support atomic read-modify-write * operations on the field's type, then the machine-independent * layer uses a 32-bit atomic on the aligned 32-bit word that * contains the dirty field. In the machine-independent layer, * the implementation of read-modify-write operations on the * field is encapsulated in vm_page_clear_dirty_mask(). */ #if PAGE_SIZE == 4096 #define VM_PAGE_BITS_ALL 0xffu typedef uint8_t vm_page_bits_t; #elif PAGE_SIZE == 8192 #define VM_PAGE_BITS_ALL 0xffffu typedef uint16_t vm_page_bits_t; #elif PAGE_SIZE == 16384 #define VM_PAGE_BITS_ALL 0xffffffffu typedef uint32_t vm_page_bits_t; #elif PAGE_SIZE == 32768 #define VM_PAGE_BITS_ALL 0xfffffffffffffffflu typedef uint64_t vm_page_bits_t; #endif struct vm_page { union { TAILQ_ENTRY(vm_page) q; /* page queue or free list (Q) */ struct { SLIST_ENTRY(vm_page) ss; /* private slists */ void *pv; } s; struct { u_long p; u_long v; } memguard; } plinks; TAILQ_ENTRY(vm_page) listq; /* pages in same object (O) */ vm_object_t object; /* which object am I in (O,P) */ vm_pindex_t pindex; /* offset into object (O,P) */ vm_paddr_t phys_addr; /* physical address of page */ struct md_page md; /* machine dependent stuff */ u_int wire_count; /* wired down maps refs (P) */ volatile u_int busy_lock; /* busy owners lock */ uint16_t hold_count; /* page hold count (P) */ uint16_t flags; /* page PG_* flags (P) */ uint8_t aflags; /* access is atomic */ uint8_t oflags; /* page VPO_* flags (O) */ uint8_t queue; /* page queue index (P,Q) */ int8_t psind; /* pagesizes[] index (O) */ int8_t segind; uint8_t order; /* index of the buddy queue */ uint8_t pool; u_char act_count; /* page usage count (P) */ /* NOTE that these must support one bit per DEV_BSIZE in a page */ /* so, on normal X86 kernels, they must be at least 8 bits wide */ vm_page_bits_t valid; /* map of valid DEV_BSIZE chunks (O) */ vm_page_bits_t dirty; /* map of dirty DEV_BSIZE chunks (M) */ }; /* * Page flags stored in oflags: * * Access to these page flags is synchronized by the lock on the object * containing the page (O). * * Note: VPO_UNMANAGED (used by OBJT_DEVICE, OBJT_PHYS and OBJT_SG) * indicates that the page is not under PV management but * otherwise should be treated as a normal page. Pages not * under PV management cannot be paged out via the * object/vm_page_t because there is no knowledge of their pte * mappings, and such pages are also not on any PQ queue. * */ #define VPO_UNUSED01 0x01 /* --available-- */ #define VPO_SWAPSLEEP 0x02 /* waiting for swap to finish */ #define VPO_UNMANAGED 0x04 /* no PV management for page */ #define VPO_SWAPINPROG 0x08 /* swap I/O in progress on page */ #define VPO_NOSYNC 0x10 /* do not collect for syncer */ /* * Busy page implementation details. * The algorithm is taken mostly by rwlock(9) and sx(9) locks implementation, * even if the support for owner identity is removed because of size * constraints. Checks on lock recursion are then not possible, while the * lock assertions effectiveness is someway reduced. */ #define VPB_BIT_SHARED 0x01 #define VPB_BIT_EXCLUSIVE 0x02 #define VPB_BIT_WAITERS 0x04 #define VPB_BIT_FLAGMASK \ (VPB_BIT_SHARED | VPB_BIT_EXCLUSIVE | VPB_BIT_WAITERS) #define VPB_SHARERS_SHIFT 3 #define VPB_SHARERS(x) \ (((x) & ~VPB_BIT_FLAGMASK) >> VPB_SHARERS_SHIFT) #define VPB_SHARERS_WORD(x) ((x) << VPB_SHARERS_SHIFT | VPB_BIT_SHARED) #define VPB_ONE_SHARER (1 << VPB_SHARERS_SHIFT) #define VPB_SINGLE_EXCLUSIVER VPB_BIT_EXCLUSIVE #define VPB_UNBUSIED VPB_SHARERS_WORD(0) #define PQ_NONE 255 #define PQ_INACTIVE 0 #define PQ_ACTIVE 1 #define PQ_COUNT 2 TAILQ_HEAD(pglist, vm_page); SLIST_HEAD(spglist, vm_page); struct vm_pagequeue { struct mtx pq_mutex; struct pglist pq_pl; int pq_cnt; u_int * const pq_vcnt; const char * const pq_name; } __aligned(CACHE_LINE_SIZE); struct vm_domain { struct vm_pagequeue vmd_pagequeues[PQ_COUNT]; u_int vmd_page_count; u_int vmd_free_count; long vmd_segs; /* bitmask of the segments */ boolean_t vmd_oom; int vmd_pass; /* local pagedaemon pass */ int vmd_oom_seq; int vmd_last_active_scan; struct vm_page vmd_marker; /* marker for pagedaemon private use */ struct vm_page vmd_inacthead; /* marker for LRU-defeating insertions */ }; extern struct vm_domain vm_dom[MAXMEMDOM]; #define vm_pagequeue_assert_locked(pq) mtx_assert(&(pq)->pq_mutex, MA_OWNED) #define vm_pagequeue_lock(pq) mtx_lock(&(pq)->pq_mutex) #define vm_pagequeue_unlock(pq) mtx_unlock(&(pq)->pq_mutex) #ifdef _KERNEL static __inline void vm_pagequeue_cnt_add(struct vm_pagequeue *pq, int addend) { #ifdef notyet vm_pagequeue_assert_locked(pq); #endif pq->pq_cnt += addend; atomic_add_int(pq->pq_vcnt, addend); } #define vm_pagequeue_cnt_inc(pq) vm_pagequeue_cnt_add((pq), 1) #define vm_pagequeue_cnt_dec(pq) vm_pagequeue_cnt_add((pq), -1) #endif /* _KERNEL */ extern struct mtx_padalign vm_page_queue_free_mtx; extern struct mtx_padalign pa_lock[]; #if defined(__arm__) #define PDRSHIFT PDR_SHIFT #elif !defined(PDRSHIFT) #define PDRSHIFT 21 #endif #define pa_index(pa) ((pa) >> PDRSHIFT) #define PA_LOCKPTR(pa) ((struct mtx *)(&pa_lock[pa_index(pa) % PA_LOCK_COUNT])) #define PA_LOCKOBJPTR(pa) ((struct lock_object *)PA_LOCKPTR((pa))) #define PA_LOCK(pa) mtx_lock(PA_LOCKPTR(pa)) #define PA_TRYLOCK(pa) mtx_trylock(PA_LOCKPTR(pa)) #define PA_UNLOCK(pa) mtx_unlock(PA_LOCKPTR(pa)) #define PA_UNLOCK_COND(pa) \ do { \ if ((pa) != 0) { \ PA_UNLOCK((pa)); \ (pa) = 0; \ } \ } while (0) #define PA_LOCK_ASSERT(pa, a) mtx_assert(PA_LOCKPTR(pa), (a)) #ifdef KLD_MODULE #define vm_page_lock(m) vm_page_lock_KBI((m), LOCK_FILE, LOCK_LINE) #define vm_page_unlock(m) vm_page_unlock_KBI((m), LOCK_FILE, LOCK_LINE) #define vm_page_trylock(m) vm_page_trylock_KBI((m), LOCK_FILE, LOCK_LINE) #else /* !KLD_MODULE */ #define vm_page_lockptr(m) (PA_LOCKPTR(VM_PAGE_TO_PHYS((m)))) #define vm_page_lock(m) mtx_lock(vm_page_lockptr((m))) #define vm_page_unlock(m) mtx_unlock(vm_page_lockptr((m))) #define vm_page_trylock(m) mtx_trylock(vm_page_lockptr((m))) #endif #if defined(INVARIANTS) #define vm_page_assert_locked(m) \ vm_page_assert_locked_KBI((m), __FILE__, __LINE__) #define vm_page_lock_assert(m, a) \ vm_page_lock_assert_KBI((m), (a), __FILE__, __LINE__) #else #define vm_page_assert_locked(m) #define vm_page_lock_assert(m, a) #endif /* * The vm_page's aflags are updated using atomic operations. To set or clear * these flags, the functions vm_page_aflag_set() and vm_page_aflag_clear() * must be used. Neither these flags nor these functions are part of the KBI. * * PGA_REFERENCED may be cleared only if the page is locked. It is set by * both the MI and MD VM layers. However, kernel loadable modules should not * directly set this flag. They should call vm_page_reference() instead. * * PGA_WRITEABLE is set exclusively on managed pages by pmap_enter(). * When it does so, the object must be locked, or the page must be * exclusive busied. The MI VM layer must never access this flag * directly. Instead, it should call pmap_page_is_write_mapped(). * * PGA_EXECUTABLE may be set by pmap routines, and indicates that a page has * at least one executable mapping. It is not consumed by the MI VM layer. */ #define PGA_WRITEABLE 0x01 /* page may be mapped writeable */ #define PGA_REFERENCED 0x02 /* page has been referenced */ #define PGA_EXECUTABLE 0x04 /* page may be mapped executable */ /* * Page flags. If changed at any other time than page allocation or * freeing, the modification must be protected by the vm_page lock. */ #define PG_CACHED 0x0001 /* page is cached */ #define PG_FICTITIOUS 0x0004 /* physical page doesn't exist */ #define PG_ZERO 0x0008 /* page is zeroed */ #define PG_MARKER 0x0010 /* special queue marker page */ #define PG_WINATCFLS 0x0040 /* flush dirty page on inactive q */ #define PG_NODUMP 0x0080 /* don't include this page in a dump */ #define PG_UNHOLDFREE 0x0100 /* delayed free of a held page */ /* * Misc constants. */ #define ACT_DECLINE 1 #define ACT_ADVANCE 3 #define ACT_INIT 5 #define ACT_MAX 64 #ifdef _KERNEL #include #include /* * Each pageable resident page falls into one of four lists: * * free * Available for allocation now. * * cache * Almost available for allocation. Still associated with * an object, but clean and immediately freeable. * * The following lists are LRU sorted: * * inactive * Low activity, candidates for reclamation. * This is the list of pages that should be * paged out next. * * active * Pages that are "active" i.e. they have been * recently referenced. * */ extern int vm_page_zero_count; extern vm_page_t vm_page_array; /* First resident page in table */ extern long vm_page_array_size; /* number of vm_page_t's */ extern long first_page; /* first physical page number */ #define VM_PAGE_TO_PHYS(entry) ((entry)->phys_addr) /* * PHYS_TO_VM_PAGE() returns the vm_page_t object that represents a memory * page to which the given physical address belongs. The correct vm_page_t * object is returned for addresses that are not page-aligned. */ vm_page_t PHYS_TO_VM_PAGE(vm_paddr_t pa); /* * Page allocation parameters for vm_page for the functions * vm_page_alloc(), vm_page_grab(), vm_page_alloc_contig() and * vm_page_alloc_freelist(). Some functions support only a subset * of the flags, and ignore others, see the flags legend. * * Bits 0 - 1 define class. * Bits 2 - 15 dedicated for flags. * Legend: * (a) - vm_page_alloc() supports the flag. * (c) - vm_page_alloc_contig() supports the flag. * (f) - vm_page_alloc_freelist() supports the flag. * (g) - vm_page_grab() supports the flag. * Bits above 15 define the count of additional pages that the caller * intends to allocate. */ #define VM_ALLOC_NORMAL 0 #define VM_ALLOC_INTERRUPT 1 #define VM_ALLOC_SYSTEM 2 #define VM_ALLOC_CLASS_MASK 3 #define VM_ALLOC_WIRED 0x0020 /* (acfg) Allocate non pageable page */ #define VM_ALLOC_ZERO 0x0040 /* (acfg) Try to obtain a zeroed page */ #define VM_ALLOC_NOOBJ 0x0100 /* (acg) No associated object */ #define VM_ALLOC_NOBUSY 0x0200 /* (acg) Do not busy the page */ #define VM_ALLOC_IFCACHED 0x0400 /* (ag) Fail if page is not cached */ #define VM_ALLOC_IFNOTCACHED 0x0800 /* (ag) Fail if page is cached */ #define VM_ALLOC_IGN_SBUSY 0x1000 /* (g) Ignore shared busy flag */ #define VM_ALLOC_NODUMP 0x2000 /* (ag) don't include in dump */ #define VM_ALLOC_SBUSY 0x4000 /* (acg) Shared busy the page */ #define VM_ALLOC_NOWAIT 0x8000 /* (g) Do not sleep, return NULL */ #define VM_ALLOC_COUNT_SHIFT 16 #define VM_ALLOC_COUNT(count) ((count) << VM_ALLOC_COUNT_SHIFT) #ifdef M_NOWAIT static inline int malloc2vm_flags(int malloc_flags) { int pflags; KASSERT((malloc_flags & M_USE_RESERVE) == 0 || (malloc_flags & M_NOWAIT) != 0, ("M_USE_RESERVE requires M_NOWAIT")); pflags = (malloc_flags & M_USE_RESERVE) != 0 ? VM_ALLOC_INTERRUPT : VM_ALLOC_SYSTEM; if ((malloc_flags & M_ZERO) != 0) pflags |= VM_ALLOC_ZERO; if ((malloc_flags & M_NODUMP) != 0) pflags |= VM_ALLOC_NODUMP; return (pflags); } #endif void vm_page_busy_downgrade(vm_page_t m); void vm_page_busy_sleep(vm_page_t m, const char *msg); void vm_page_flash(vm_page_t m); void vm_page_hold(vm_page_t mem); void vm_page_unhold(vm_page_t mem); void vm_page_free(vm_page_t m); void vm_page_free_zero(vm_page_t m); void vm_page_activate (vm_page_t); void vm_page_advise(vm_page_t m, int advice); vm_page_t vm_page_alloc (vm_object_t, vm_pindex_t, int); vm_page_t vm_page_alloc_contig(vm_object_t object, vm_pindex_t pindex, int req, u_long npages, vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary, vm_memattr_t memattr); vm_page_t vm_page_alloc_freelist(int, int); vm_page_t vm_page_grab (vm_object_t, vm_pindex_t, int); void vm_page_cache(vm_page_t); void vm_page_cache_free(vm_object_t, vm_pindex_t, vm_pindex_t); void vm_page_cache_transfer(vm_object_t, vm_pindex_t, vm_object_t); int vm_page_try_to_cache (vm_page_t); int vm_page_try_to_free (vm_page_t); void vm_page_deactivate (vm_page_t); void vm_page_deactivate_noreuse(vm_page_t); void vm_page_dequeue(vm_page_t m); void vm_page_dequeue_locked(vm_page_t m); vm_page_t vm_page_find_least(vm_object_t, vm_pindex_t); vm_page_t vm_page_getfake(vm_paddr_t paddr, vm_memattr_t memattr); void vm_page_initfake(vm_page_t m, vm_paddr_t paddr, vm_memattr_t memattr); int vm_page_insert (vm_page_t, vm_object_t, vm_pindex_t); boolean_t vm_page_is_cached(vm_object_t object, vm_pindex_t pindex); vm_page_t vm_page_lookup (vm_object_t, vm_pindex_t); vm_page_t vm_page_next(vm_page_t m); int vm_page_pa_tryrelock(pmap_t, vm_paddr_t, vm_paddr_t *); struct vm_pagequeue *vm_page_pagequeue(vm_page_t m); vm_page_t vm_page_prev(vm_page_t m); boolean_t vm_page_ps_is_valid(vm_page_t m); void vm_page_putfake(vm_page_t m); void vm_page_readahead_finish(vm_page_t m); bool vm_page_reclaim_contig(int req, u_long npages, vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary); void vm_page_reference(vm_page_t m); void vm_page_remove (vm_page_t); int vm_page_rename (vm_page_t, vm_object_t, vm_pindex_t); vm_page_t vm_page_replace(vm_page_t mnew, vm_object_t object, vm_pindex_t pindex); void vm_page_requeue(vm_page_t m); void vm_page_requeue_locked(vm_page_t m); int vm_page_sbusied(vm_page_t m); vm_page_t vm_page_scan_contig(u_long npages, vm_page_t m_start, vm_page_t m_end, u_long alignment, vm_paddr_t boundary, int options); void vm_page_set_valid_range(vm_page_t m, int base, int size); int vm_page_sleep_if_busy(vm_page_t m, const char *msg); vm_offset_t vm_page_startup(vm_offset_t vaddr); void vm_page_sunbusy(vm_page_t m); int vm_page_trysbusy(vm_page_t m); void vm_page_unhold_pages(vm_page_t *ma, int count); boolean_t vm_page_unwire(vm_page_t m, uint8_t queue); void vm_page_updatefake(vm_page_t m, vm_paddr_t paddr, vm_memattr_t memattr); void vm_page_wire (vm_page_t); void vm_page_xunbusy_hard(vm_page_t m); void vm_page_set_validclean (vm_page_t, int, int); void vm_page_clear_dirty (vm_page_t, int, int); void vm_page_set_invalid (vm_page_t, int, int); int vm_page_is_valid (vm_page_t, int, int); void vm_page_test_dirty (vm_page_t); vm_page_bits_t vm_page_bits(int base, int size); void vm_page_zero_invalid(vm_page_t m, boolean_t setvalid); void vm_page_free_toq(vm_page_t m); void vm_page_zero_idle_wakeup(void); void vm_page_dirty_KBI(vm_page_t m); void vm_page_lock_KBI(vm_page_t m, const char *file, int line); void vm_page_unlock_KBI(vm_page_t m, const char *file, int line); int vm_page_trylock_KBI(vm_page_t m, const char *file, int line); #if defined(INVARIANTS) || defined(INVARIANT_SUPPORT) void vm_page_assert_locked_KBI(vm_page_t m, const char *file, int line); void vm_page_lock_assert_KBI(vm_page_t m, int a, const char *file, int line); #endif #define vm_page_assert_sbusied(m) \ KASSERT(vm_page_sbusied(m), \ ("vm_page_assert_sbusied: page %p not shared busy @ %s:%d", \ (m), __FILE__, __LINE__)) #define vm_page_assert_unbusied(m) \ KASSERT(!vm_page_busied(m), \ ("vm_page_assert_unbusied: page %p busy @ %s:%d", \ (m), __FILE__, __LINE__)) #define vm_page_assert_xbusied(m) \ KASSERT(vm_page_xbusied(m), \ ("vm_page_assert_xbusied: page %p not exclusive busy @ %s:%d", \ (m), __FILE__, __LINE__)) #define vm_page_busied(m) \ ((m)->busy_lock != VPB_UNBUSIED) #define vm_page_sbusy(m) do { \ if (!vm_page_trysbusy(m)) \ panic("%s: page %p failed shared busying", __func__, \ (m)); \ } while (0) #define vm_page_tryxbusy(m) \ (atomic_cmpset_acq_int(&(m)->busy_lock, VPB_UNBUSIED, \ VPB_SINGLE_EXCLUSIVER)) #define vm_page_xbusied(m) \ (((m)->busy_lock & VPB_SINGLE_EXCLUSIVER) != 0) #define vm_page_xbusy(m) do { \ if (!vm_page_tryxbusy(m)) \ panic("%s: page %p failed exclusive busying", __func__, \ (m)); \ } while (0) +/* Note: page m's lock must not be owned by the caller. */ #define vm_page_xunbusy(m) do { \ if (!atomic_cmpset_rel_int(&(m)->busy_lock, \ VPB_SINGLE_EXCLUSIVER, VPB_UNBUSIED)) \ vm_page_xunbusy_hard(m); \ } while (0) #ifdef INVARIANTS void vm_page_object_lock_assert(vm_page_t m); #define VM_PAGE_OBJECT_LOCK_ASSERT(m) vm_page_object_lock_assert(m) void vm_page_assert_pga_writeable(vm_page_t m, uint8_t bits); #define VM_PAGE_ASSERT_PGA_WRITEABLE(m, bits) \ vm_page_assert_pga_writeable(m, bits) #else #define VM_PAGE_OBJECT_LOCK_ASSERT(m) (void)0 #define VM_PAGE_ASSERT_PGA_WRITEABLE(m, bits) (void)0 #endif /* * We want to use atomic updates for the aflags field, which is 8 bits wide. * However, not all architectures support atomic operations on 8-bit * destinations. In order that we can easily use a 32-bit operation, we * require that the aflags field be 32-bit aligned. */ CTASSERT(offsetof(struct vm_page, aflags) % sizeof(uint32_t) == 0); /* * Clear the given bits in the specified page. */ static inline void vm_page_aflag_clear(vm_page_t m, uint8_t bits) { uint32_t *addr, val; /* * The PGA_REFERENCED flag can only be cleared if the page is locked. */ if ((bits & PGA_REFERENCED) != 0) vm_page_assert_locked(m); /* * Access the whole 32-bit word containing the aflags field with an * atomic update. Parallel non-atomic updates to the other fields * within this word are handled properly by the atomic update. */ addr = (void *)&m->aflags; KASSERT(((uintptr_t)addr & (sizeof(uint32_t) - 1)) == 0, ("vm_page_aflag_clear: aflags is misaligned")); val = bits; #if BYTE_ORDER == BIG_ENDIAN val <<= 24; #endif atomic_clear_32(addr, val); } /* * Set the given bits in the specified page. */ static inline void vm_page_aflag_set(vm_page_t m, uint8_t bits) { uint32_t *addr, val; VM_PAGE_ASSERT_PGA_WRITEABLE(m, bits); /* * Access the whole 32-bit word containing the aflags field with an * atomic update. Parallel non-atomic updates to the other fields * within this word are handled properly by the atomic update. */ addr = (void *)&m->aflags; KASSERT(((uintptr_t)addr & (sizeof(uint32_t) - 1)) == 0, ("vm_page_aflag_set: aflags is misaligned")); val = bits; #if BYTE_ORDER == BIG_ENDIAN val <<= 24; #endif atomic_set_32(addr, val); } /* * vm_page_dirty: * * Set all bits in the page's dirty field. * * The object containing the specified page must be locked if the * call is made from the machine-independent layer. * * See vm_page_clear_dirty_mask(). */ static __inline void vm_page_dirty(vm_page_t m) { /* Use vm_page_dirty_KBI() under INVARIANTS to save memory. */ #if defined(KLD_MODULE) || defined(INVARIANTS) vm_page_dirty_KBI(m); #else m->dirty = VM_PAGE_BITS_ALL; #endif } /* * vm_page_remque: * * If the given page is in a page queue, then remove it from that page * queue. * * The page must be locked. */ static inline void vm_page_remque(vm_page_t m) { if (m->queue != PQ_NONE) vm_page_dequeue(m); } /* * vm_page_undirty: * * Set page to not be dirty. Note: does not clear pmap modify bits */ static __inline void vm_page_undirty(vm_page_t m) { VM_PAGE_OBJECT_LOCK_ASSERT(m); m->dirty = 0; } static inline void vm_page_replace_checked(vm_page_t mnew, vm_object_t object, vm_pindex_t pindex, vm_page_t mold) { vm_page_t mret; mret = vm_page_replace(mnew, object, pindex); KASSERT(mret == mold, ("invalid page replacement, mold=%p, mret=%p", mold, mret)); /* Unused if !INVARIANTS. */ (void)mold; (void)mret; } #endif /* _KERNEL */ #endif /* !_VM_PAGE_ */ Index: projects/vnet/sys/x86/acpica/acpi_wakeup.c =================================================================== --- projects/vnet/sys/x86/acpica/acpi_wakeup.c (revision 302157) +++ projects/vnet/sys/x86/acpica/acpi_wakeup.c (revision 302158) @@ -1,418 +1,418 @@ /*- * Copyright (c) 2001 Takanori Watanabe * Copyright (c) 2001-2012 Mitsuru IWASAKI * Copyright (c) 2003 Peter Wemm * Copyright (c) 2008-2012 Jung-uk Kim * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #if defined(__amd64__) #define DEV_APIC #else #include "opt_apic.h" #endif #ifdef __i386__ #include "opt_npx.h" #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef DEV_APIC #include #include #endif #ifdef SMP #include #include #endif #include #include #include "acpi_wakecode.h" #include "acpi_wakedata.h" /* Make sure the code is less than a page and leave room for the stack. */ CTASSERT(sizeof(wakecode) < PAGE_SIZE - 1024); extern int acpi_resume_beep; extern int acpi_reset_video; #ifdef SMP extern struct susppcb **susppcbs; static cpuset_t suspcpus; #else static struct susppcb **susppcbs; #endif static void *acpi_alloc_wakeup_handler(void); static void acpi_stop_beep(void *); #ifdef SMP static int acpi_wakeup_ap(struct acpi_softc *, int); static void acpi_wakeup_cpus(struct acpi_softc *); #endif #ifdef __amd64__ #define ACPI_PAGETABLES 3 #else #define ACPI_PAGETABLES 0 #endif #define WAKECODE_VADDR(sc) \ ((sc)->acpi_wakeaddr + (ACPI_PAGETABLES * PAGE_SIZE)) #define WAKECODE_PADDR(sc) \ ((sc)->acpi_wakephys + (ACPI_PAGETABLES * PAGE_SIZE)) #define WAKECODE_FIXUP(offset, type, val) do { \ type *addr; \ addr = (type *)(WAKECODE_VADDR(sc) + offset); \ *addr = val; \ } while (0) static void acpi_stop_beep(void *arg) { if (acpi_resume_beep != 0) timer_spkr_release(); } #ifdef SMP static int acpi_wakeup_ap(struct acpi_softc *sc, int cpu) { struct pcb *pcb; int vector = (WAKECODE_PADDR(sc) >> 12) & 0xff; int apic_id = cpu_apic_ids[cpu]; int ms; pcb = &susppcbs[cpu]->sp_pcb; WAKECODE_FIXUP(wakeup_pcb, struct pcb *, pcb); WAKECODE_FIXUP(wakeup_gdt, uint16_t, pcb->pcb_gdt.rd_limit); WAKECODE_FIXUP(wakeup_gdt + 2, uint64_t, pcb->pcb_gdt.rd_base); ipi_startup(apic_id, vector); /* Wait up to 5 seconds for it to resume. */ for (ms = 0; ms < 5000; ms++) { if (!CPU_ISSET(cpu, &suspended_cpus)) return (1); /* return SUCCESS */ DELAY(1000); } return (0); /* return FAILURE */ } #define WARMBOOT_TARGET 0 #define WARMBOOT_OFF (KERNBASE + 0x0467) #define WARMBOOT_SEG (KERNBASE + 0x0469) #define CMOS_REG (0x70) #define CMOS_DATA (0x71) #define BIOS_RESET (0x0f) #define BIOS_WARM (0x0a) static void acpi_wakeup_cpus(struct acpi_softc *sc) { uint32_t mpbioswarmvec; int cpu; u_char mpbiosreason; /* save the current value of the warm-start vector */ mpbioswarmvec = *((uint32_t *)WARMBOOT_OFF); outb(CMOS_REG, BIOS_RESET); mpbiosreason = inb(CMOS_DATA); /* setup a vector to our boot code */ *((volatile u_short *)WARMBOOT_OFF) = WARMBOOT_TARGET; *((volatile u_short *)WARMBOOT_SEG) = WAKECODE_PADDR(sc) >> 4; outb(CMOS_REG, BIOS_RESET); outb(CMOS_DATA, BIOS_WARM); /* 'warm-start' */ /* Wake up each AP. */ for (cpu = 1; cpu < mp_ncpus; cpu++) { if (!CPU_ISSET(cpu, &suspcpus)) continue; if (acpi_wakeup_ap(sc, cpu) == 0) { /* restore the warmstart vector */ *(uint32_t *)WARMBOOT_OFF = mpbioswarmvec; panic("acpi_wakeup: failed to resume AP #%d (PHY #%d)", cpu, cpu_apic_ids[cpu]); } } /* restore the warmstart vector */ *(uint32_t *)WARMBOOT_OFF = mpbioswarmvec; outb(CMOS_REG, BIOS_RESET); outb(CMOS_DATA, mpbiosreason); } #endif int acpi_sleep_machdep(struct acpi_softc *sc, int state) { ACPI_STATUS status; struct pcb *pcb; if (sc->acpi_wakeaddr == 0ul) return (-1); /* couldn't alloc wake memory */ #ifdef SMP suspcpus = all_cpus; CPU_CLR(PCPU_GET(cpuid), &suspcpus); #endif if (acpi_resume_beep != 0) timer_spkr_acquire(); AcpiSetFirmwareWakingVector(WAKECODE_PADDR(sc), 0); intr_suspend(); pcb = &susppcbs[0]->sp_pcb; if (savectx(pcb)) { #ifdef __amd64__ fpususpend(susppcbs[0]->sp_fpususpend); #elif defined(DEV_NPX) npxsuspend(susppcbs[0]->sp_fpususpend); #endif #ifdef SMP if (!CPU_EMPTY(&suspcpus) && suspend_cpus(suspcpus) == 0) { device_printf(sc->acpi_dev, "Failed to suspend APs\n"); return (0); /* couldn't sleep */ } #endif WAKECODE_FIXUP(resume_beep, uint8_t, (acpi_resume_beep != 0)); WAKECODE_FIXUP(reset_video, uint8_t, (acpi_reset_video != 0)); #ifndef __amd64__ WAKECODE_FIXUP(wakeup_cr4, register_t, pcb->pcb_cr4); #endif WAKECODE_FIXUP(wakeup_pcb, struct pcb *, pcb); WAKECODE_FIXUP(wakeup_gdt, uint16_t, pcb->pcb_gdt.rd_limit); WAKECODE_FIXUP(wakeup_gdt + 2, uint64_t, pcb->pcb_gdt.rd_base); /* Call ACPICA to enter the desired sleep state */ if (state == ACPI_STATE_S4 && sc->acpi_s4bios) status = AcpiEnterSleepStateS4bios(); else status = AcpiEnterSleepState(state); if (ACPI_FAILURE(status)) { device_printf(sc->acpi_dev, "AcpiEnterSleepState failed - %s\n", AcpiFormatException(status)); return (0); /* couldn't sleep */ } for (;;) ia32_pause(); } else { #ifdef __amd64__ fpuresume(susppcbs[0]->sp_fpususpend); #elif defined(DEV_NPX) npxresume(susppcbs[0]->sp_fpususpend); #endif } return (1); /* wakeup successfully */ } int acpi_wakeup_machdep(struct acpi_softc *sc, int state, int sleep_result, int intr_enabled) { if (sleep_result == -1) return (sleep_result); if (!intr_enabled) { /* Wakeup MD procedures in interrupt disabled context */ if (sleep_result == 1) { pmap_init_pat(); initializecpu(); PCPU_SET(switchtime, 0); PCPU_SET(switchticks, ticks); #ifdef DEV_APIC lapic_xapic_mode(); #endif #ifdef SMP if (!CPU_EMPTY(&suspcpus)) acpi_wakeup_cpus(sc); #endif } #ifdef SMP if (!CPU_EMPTY(&suspcpus)) restart_cpus(suspcpus); #endif mca_resume(); #ifdef __amd64__ if (vmm_resume_p != NULL) vmm_resume_p(); #endif intr_resume(/*suspend_cancelled*/false); AcpiSetFirmwareWakingVector(0, 0); } else { /* Wakeup MD procedures in interrupt enabled context */ if (sleep_result == 1 && mem_range_softc.mr_op != NULL && mem_range_softc.mr_op->reinit != NULL) mem_range_softc.mr_op->reinit(&mem_range_softc); } return (sleep_result); } static void * acpi_alloc_wakeup_handler(void) { void *wakeaddr; int i; /* * Specify the region for our wakeup code. We want it in the low 1 MB * region, excluding real mode IVT (0-0x3ff), BDA (0x400-0x4ff), EBDA * (less than 128KB, below 0xa0000, must be excluded by SMAP and DSDT), * and ROM area (0xa0000 and above). The temporary page tables must be * page-aligned. */ wakeaddr = contigmalloc((ACPI_PAGETABLES + 1) * PAGE_SIZE, M_DEVBUF, - M_WAITOK, 0x500, 0xa0000, PAGE_SIZE, 0ul); + M_NOWAIT, 0x500, 0xa0000, PAGE_SIZE, 0ul); if (wakeaddr == NULL) { printf("%s: can't alloc wake memory\n", __func__); return (NULL); } if (EVENTHANDLER_REGISTER(power_resume, acpi_stop_beep, NULL, EVENTHANDLER_PRI_LAST) == NULL) { printf("%s: can't register event handler\n", __func__); contigfree(wakeaddr, (ACPI_PAGETABLES + 1) * PAGE_SIZE, M_DEVBUF); return (NULL); } susppcbs = malloc(mp_ncpus * sizeof(*susppcbs), M_DEVBUF, M_WAITOK); for (i = 0; i < mp_ncpus; i++) { susppcbs[i] = malloc(sizeof(**susppcbs), M_DEVBUF, M_WAITOK); susppcbs[i]->sp_fpususpend = alloc_fpusave(M_WAITOK); } return (wakeaddr); } void acpi_install_wakeup_handler(struct acpi_softc *sc) { static void *wakeaddr = NULL; #ifdef __amd64__ uint64_t *pt4, *pt3, *pt2; int i; #endif if (wakeaddr != NULL) return; wakeaddr = acpi_alloc_wakeup_handler(); if (wakeaddr == NULL) return; sc->acpi_wakeaddr = (vm_offset_t)wakeaddr; sc->acpi_wakephys = vtophys(wakeaddr); bcopy(wakecode, (void *)WAKECODE_VADDR(sc), sizeof(wakecode)); /* Patch GDT base address, ljmp targets. */ WAKECODE_FIXUP((bootgdtdesc + 2), uint32_t, WAKECODE_PADDR(sc) + bootgdt); WAKECODE_FIXUP((wakeup_sw32 + 2), uint32_t, WAKECODE_PADDR(sc) + wakeup_32); #ifdef __amd64__ WAKECODE_FIXUP((wakeup_sw64 + 1), uint32_t, WAKECODE_PADDR(sc) + wakeup_64); WAKECODE_FIXUP(wakeup_pagetables, uint32_t, sc->acpi_wakephys); #endif /* Save pointers to some global data. */ WAKECODE_FIXUP(wakeup_ret, void *, resumectx); #ifndef __amd64__ #if defined(PAE) || defined(PAE_TABLES) WAKECODE_FIXUP(wakeup_cr3, register_t, vtophys(kernel_pmap->pm_pdpt)); #else WAKECODE_FIXUP(wakeup_cr3, register_t, vtophys(kernel_pmap->pm_pdir)); #endif #else /* Build temporary page tables below realmode code. */ pt4 = wakeaddr; pt3 = pt4 + (PAGE_SIZE) / sizeof(uint64_t); pt2 = pt3 + (PAGE_SIZE) / sizeof(uint64_t); /* Create the initial 1GB replicated page tables */ for (i = 0; i < 512; i++) { /* * Each slot of the level 4 pages points * to the same level 3 page */ pt4[i] = (uint64_t)(sc->acpi_wakephys + PAGE_SIZE); pt4[i] |= PG_V | PG_RW | PG_U; /* * Each slot of the level 3 pages points * to the same level 2 page */ pt3[i] = (uint64_t)(sc->acpi_wakephys + (2 * PAGE_SIZE)); pt3[i] |= PG_V | PG_RW | PG_U; /* The level 2 page slots are mapped with 2MB pages for 1GB. */ pt2[i] = i * (2 * 1024 * 1024); pt2[i] |= PG_V | PG_RW | PG_PS | PG_U; } #endif if (bootverbose) device_printf(sc->acpi_dev, "wakeup code va %#jx pa %#jx\n", (uintmax_t)sc->acpi_wakeaddr, (uintmax_t)sc->acpi_wakephys); } Index: projects/vnet/tools/tools/makeroot/makeroot.sh =================================================================== --- projects/vnet/tools/tools/makeroot/makeroot.sh (revision 302157) +++ projects/vnet/tools/tools/makeroot/makeroot.sh (revision 302158) @@ -1,242 +1,246 @@ #!/bin/sh -e #- # Copyright (c) 2012-2013 SRI International # Copyright (c) 2012 Robert N. M. Watson # All rights reserved. # # 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. # # $FreeBSD$ usage() { cat <&2 usage: makeroot.sh [-B byte-order] [-d] [-e ] [-f ] [-k [-K ]] [-p [-g ]] [-s ] EOF exit 1 } warn() { echo `basename $0` "$@" 1>&2 } err() { ret=$1 shift warn "$@" exit $ret } atexit() { if [ -z "${DEBUG}" ]; then rm -rf ${tmpdir} else warn "temp directory left at ${tmpdir}" fi } DEBUG= # Allow duplicate manifest entries when not file list is given because the # FreeBSD METALOG still includes it. DUPFLAG=-D EXTRAS= FILELIST= GROUP= KEYDIR= KEYUSERS= PASSWD= while getopts "B:de:f:g:K:k:l:p:s:" opt; do case "$opt" in B) BFLAG="-B ${OPTARG}" ;; d) DEBUG=1 ;; e) EXTRAS="${EXTRAS} ${OPTARG}" ;; f) FILELIST="${OPTARG}"; DUPFLAG= ;; g) GROUP="${OPTARG}" ;; K) KEYUSERS="${KEYUSERS} ${OPTARG}" ;; k) KEYDIR="${OPTARG}" ;; l) LABEL="${OPTARG}" ;; p) PASSWD="${OPTARG}" ;; s) SIZE="${OPTARG}" ;; *) usage ;; esac done shift $(($OPTIND - 1)) if [ $# -ne 2 ]; then usage; fi IMGFILE=$(realpath $(dirname $1))/$(basename $1) BSDROOT=$2 DBDIR=${BSDROOT}/etc if [ ! -r ${BSDROOT}/METALOG ]; then err 1 "${BSDROOT} does not contain a METALOG" fi if [ -n "${GROUP}" -a -z "${PASSWD}" ]; then warn "-g requires -p" usage fi if [ -n "${KEYUSERS}" -a -z "${KEYDIR}" ]; then warn "-K requires -k" usage fi if [ -n "${KEYDIR}" -a -z "${KEYUSERS}" ]; then KEYUSERS=root fi tmpdir=`mktemp -d /tmp/makeroot.XXXXX` if [ -z "${tmpdir}" -o ! -d "${tmpdir}" ]; then err 1 "failed to create tmpdir" fi trap atexit EXIT manifest=${tmpdir}/manifest echo "#mtree 2.0" > ${manifest} if [ -n "${PASSWD}" ]; then cp ${PASSWD} ${tmpdir}/master.passwd pwd_mkdb -d ${tmpdir} -p ${tmpdir}/master.passwd if [ -z "${GROUP}" ]; then cp ${DBDIR}/group ${tmpdir} else cp ${GROUP} ${tmpdir} fi cat <> ${tmpdir}/passwd.mtree ./etc/group type=file uname=root gname=wheel mode=0644 contents=${tmpdir}/group ./etc/master.passwd type=file uname=root gname=wheel mode=0600 contents=${tmpdir}/master.passwd ./etc/passwd type=file mode=0644 uname=root gname=wheel contents=${tmpdir}/passwd ./etc/pwd.db type=file mode=0644 uname=root gname=wheel contents=${tmpdir}/pwd.db ./etc/spwd.db type=file mode=0600 uname=root gname=wheel contents=${tmpdir}/spwd.db EOF EXTRAS="${EXTRAS} ${tmpdir}/passwd.mtree" DBDIR=${tmpdir} fi if [ -n "${FILELIST}" ]; then # build manifest from root manifest and FILELIST (echo .; grep -v ^# ${FILELIST} | while read path; do # Print each included path and all its sub-paths with a ./ # prepended. The "sort -u" will then discard all the # duplicate directory entries. This ensures that we # extract the permissions for each unlisted directory # from the METALOG. path="/${path}" while [ -n "${path}" ]; do echo ".${path}" path="${path%/*}" done done) | sort -u ${BSDROOT}/METALOG - | \ awk ' !/ type=/ { file = $1 } / type=/ { if ($1 == file) {print} }' >> ${manifest} elif [ -n "${EXTRAS}" ]; then # Start with all the files in BSDROOT/METALOG except those in # one of the EXTRAS manifests. grep -h type=file ${EXTRAS} | cut -d' ' -f1 | \ sort -u ${BSDROOT}/METALOG - | awk ' !/ type=/ { file = $1 } / type=/ { if ($1 != file) {print} }' >> ${manifest} else sort -u ${BSDROOT}/METALOG >> ${manifest} fi # For each extras file, add contents keys relative to the directory the # manifest lives in for each file line that does not have one. Adjust # contents keys relative to ./ to be relative to the same directory. for eman in ${EXTRAS}; do if [ ! -f ${eman} ]; then err 1 "${eman} is not a regular file" fi extradir=`realpath ${eman}`; extradir=`dirname ${extradir}` awk '{ if ($0 !~ /type=file/) { print } else { if ($0 !~ /contents=/) { printf ("%s contents=%s\n", $0, $1) } else { print } } }' ${eman} | \ sed -e "s|contents=\./|contents=${extradir}/|" >> ${manifest} done # /etc/rcorder.start allows the startup order to be stable even if # not all startup scripts are installed. In theory it should be # unnecessary, but dependencies in rc.d appear to be under recorded. # This is a hack local to beri/cheribsd. # echo /etc/rc.d/FIRST > ${tmpdir}/rcorder.start rcorder -s nostart ${BSDROOT}/etc/rc.d/* | sed -e "s:^${BSDROOT}::" | \ grep -v LAST | grep -v FIRST >> \ ${tmpdir}/rcorder.start echo /etc/rc.d/LAST >> ${tmpdir}/rcorder.start echo "./etc/rcorder.start type=file mode=644 uname=root gname=wheel" \ "contents=${tmpdir}/rcorder.start" >> ${manifest} # Add all public keys in KEYDIR to roots' authorized_keys file. if [ -n "${KEYDIR}" ]; then cat ${KEYDIR}/*.pub > ${tmpdir}/authorized_keys if [ ! -s ${tmpdir}/authorized_keys ]; then err 1 "no keys found in ${KEYDIR}" fi for user in ${KEYUSERS}; do userdir=`awk -F: "{if (\\\$1 == \"${user}\") {print \\\$9; exit} }" ${DBDIR}/master.passwd` gid=`awk -F: "{if (\\\$1 == \"${user}\") {print \\\$4; exit} }" ${DBDIR}/master.passwd` group=`awk -F: "{if (\\\$3 == \"${gid}\") {print \\\$1; exit} }" ${DBDIR}/group` if [ -z "${userdir}" ]; then err 1 "${user}: not found in ${DBDIR}/master.passwd" fi echo ".${userdir}/.ssh type=dir mode=700 uname=${user} gname=${group}" >> ${manifest} echo ".${userdir}/.ssh/authorized_keys type=file mode=600 uname=${user} gname=${group} contents=${tmpdir}/authorized_keys" >> ${manifest} done fi if [ -n "${LABEL}" ]; then LABELFLAG="-o label=${LABEL}" fi if [ -n "${SIZE}" ]; then SIZEFLAG="-s ${SIZE}" fi +# Zero out subsecond component of time= keywords as they are currently not +# supported by makefs +sed -i '' -E 's/(time=[0-9]*)\.[0-9]*/\1.0/' ${manifest} + cd ${BSDROOT}; makefs ${DUPFLAG} -N ${DBDIR} ${SIZEFLAG} ${BFLAG} \ -t ffs ${LABELFLAG} -f 256 ${IMGFILE} ${manifest} Index: projects/vnet/usr.sbin/bsdinstall/partedit/partedit_x86.c =================================================================== --- projects/vnet/usr.sbin/bsdinstall/partedit/partedit_x86.c (revision 302157) +++ projects/vnet/usr.sbin/bsdinstall/partedit/partedit_x86.c (revision 302158) @@ -1,150 +1,150 @@ /*- * Copyright (c) 2011 Nathan Whitehorn * 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 #include #include #include "partedit.h" static const char * x86_bootmethod(void) { static char fw[255] = ""; size_t len = sizeof(fw); int error; if (strlen(fw) == 0) { error = sysctlbyname("machdep.bootmethod", fw, &len, NULL, -1); if (error != 0) return ("BIOS"); } return (fw); } const char * default_scheme(void) { if (strcmp(x86_bootmethod(), "UEFI") == 0) return ("GPT"); else return ("MBR"); } int is_scheme_bootable(const char *part_type) { if (strcmp(part_type, "GPT") == 0) return (1); if (strcmp(x86_bootmethod(), "BIOS") == 0) { if (strcmp(part_type, "BSD") == 0) return (1); if (strcmp(part_type, "MBR") == 0) return (1); } return (0); } int is_fs_bootable(const char *part_type, const char *fs) { if (strcmp(fs, "freebsd-ufs") == 0) return (1); if (strcmp(fs, "freebsd-zfs") == 0 && strcmp(part_type, "GPT") == 0 && strcmp(x86_bootmethod(), "BIOS") == 0) return (1); return (0); } size_t bootpart_size(const char *scheme) { /* No partcode except for GPT */ if (strcmp(scheme, "GPT") != 0) return (0); if (strcmp(x86_bootmethod(), "BIOS") == 0) return (512*1024); else - return (800*1024); + return (200*1024*1024); return (0); } const char * bootpart_type(const char *scheme) { if (strcmp(x86_bootmethod(), "UEFI") == 0) return ("efi"); return ("freebsd-boot"); } const char * bootcode_path(const char *part_type) { if (strcmp(x86_bootmethod(), "UEFI") == 0) return (NULL); if (strcmp(part_type, "GPT") == 0) return ("/boot/pmbr"); if (strcmp(part_type, "MBR") == 0) return ("/boot/mbr"); if (strcmp(part_type, "BSD") == 0) return ("/boot/boot"); return (NULL); } const char * partcode_path(const char *part_type, const char *fs_type) { if (strcmp(part_type, "GPT") == 0) { if (strcmp(x86_bootmethod(), "UEFI") == 0) return ("/boot/boot1.efifat"); else if (strcmp(fs_type, "zfs") == 0) return ("/boot/gptzfsboot"); else return ("/boot/gptboot"); } /* No partcode except for GPT */ return (NULL); } Index: projects/vnet =================================================================== --- projects/vnet (revision 302157) +++ projects/vnet (revision 302158) Property changes on: projects/vnet ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head:r302120-302157