Index: head/UPDATING =================================================================== --- head/UPDATING (revision 305928) +++ head/UPDATING (revision 305929) @@ -1,1647 +1,1650 @@ 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 12.x IS SLOW: FreeBSD 12.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".) +20160918: + The backup_uses_rcs functionality has been removed from rc.subr. + 20160908: The queue(3) debugging macro, QUEUE_MACRO_DEBUG, has been split into two separate components, QUEUE_MACRO_DEBUG_TRACE and QUEUE_MACRO_DEBUG_TRASH. Define both for the original QUEUE_MACRO_DEBUG behavior. 20160824: r304787 changed some ioctl interfaces between the iSCSI userspace programs and the kernel. ctladm, ctld, iscsictl, and iscsid must be rebuilt to work with new kernels. __FreeBSD_version has been bumped to 1200005. 20160818: The UDP receive code has been updated to only treat incoming UDP packets that were addressed to an L2 broadcast address as L3 broadcast packets. It is not expected that this will affect any standards-conforming UDP application. The new behaviour can be disabled by setting the sysctl net.inet.udp.require_l2_bcast to 0. 20160818: Remove the openbsd_poll system call. __FreeBSD_version has been bumped because of this. 20160622: The libc stub for the pipe(2) system call has been replaced with a wrapper that calls the pipe2(2) system call and the pipe(2) system call is now only implemented by the kernels that include "options COMPAT_FREEBSD10" in their config file (this is the default). Users should ensure that this option is enabled in their kernel or upgrade userspace to r302092 before upgrading their kernel. 20160527: CAM will now strip leading spaces from SCSI disks' serial numbers. This will 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: head/etc/rc.subr =================================================================== --- head/etc/rc.subr (revision 305928) +++ head/etc/rc.subr (revision 305929) @@ -1,2170 +1,2124 @@ # $NetBSD: rc.subr,v 1.67 2006/10/07 11:25:15 elad Exp $ # $FreeBSD$ # # Copyright (c) 1997-2004 The NetBSD Foundation, Inc. # All rights reserved. # # This code is derived from software contributed to The NetBSD Foundation # by Luke Mewburn. # # Redistribution and use in source and binary forms, with or without # modification, are permitted provided that the following conditions # are met: # 1. Redistributions of source code must retain the above copyright # notice, this list of conditions and the following disclaimer. # 2. Redistributions in binary form must reproduce the above copyright # notice, this list of conditions and the following disclaimer in the # documentation and/or other materials provided with the distribution. # # THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS # ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED # TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR # PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS # BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR # CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF # SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS # INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN # CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) # ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE # POSSIBILITY OF SUCH DAMAGE. # # rc.subr # functions used by various rc scripts # : ${RC_PID:=$$}; export RC_PID # # Operating System dependent/independent variables # if [ -z "${_rc_subr_loaded}" ]; then _rc_subr_loaded="YES" SYSCTL="/sbin/sysctl" SYSCTL_N="${SYSCTL} -n" SYSCTL_W="${SYSCTL}" PROTECT="/usr/bin/protect" ID="/usr/bin/id" IDCMD="if [ -x $ID ]; then $ID -un; fi" PS="/bin/ps -ww" JID=`$PS -p $$ -o jid=` # # functions # --------- # list_vars pattern # List vars matching pattern. # list_vars() { set | { while read LINE; do var="${LINE%%=*}" case "$var" in "$LINE"|*[!a-zA-Z0-9_]*) continue ;; $1) echo $var esac done; } } # set_rcvar [var] [defval] [desc] # # Echo or define a rc.conf(5) variable name. Global variable # $rcvars is used. # # If no argument is specified, echo "${name}_enable". # # If only a var is specified, echo "${var}_enable". # # If var and defval are specified, the ${var} is defined as # rc.conf(5) variable and the default value is ${defvar}. An # optional argument $desc can also be specified to add a # description for that. # set_rcvar() { local _var case $# in 0) echo ${name}_enable ;; 1) echo ${1}_enable ;; *) debug "set_rcvar: \$$1=$2 is added" \ " as a rc.conf(5) variable." _var=$1 rcvars="${rcvars# } $_var" eval ${_var}_defval=\"$2\" shift 2 eval ${_var}_desc=\"$*\" ;; esac } # set_rcvar_obsolete oldvar [newvar] [msg] # Define obsolete variable. # Global variable $rcvars_obsolete is used. # set_rcvar_obsolete() { local _var _var=$1 debug "set_rcvar_obsolete: \$$1(old) -> \$$2(new) is defined" rcvars_obsolete="${rcvars_obsolete# } $1" eval ${1}_newvar=\"$2\" shift 2 eval ${_var}_obsolete_msg=\"$*\" } # # force_depend script [rcvar] # Force a service to start. Intended for use by services # to resolve dependency issues. # $1 - filename of script, in /etc/rc.d, to run # $2 - name of the script's rcvar (minus the _enable) # force_depend() { local _depend _dep_rcvar _depend="$1" _dep_rcvar="${2:-$1}_enable" [ -n "$rc_fast" ] && ! checkyesno always_force_depends && checkyesno $_dep_rcvar && return 0 /etc/rc.d/${_depend} forcestatus >/dev/null 2>&1 && return 0 info "${name} depends on ${_depend}, which will be forced to start." if ! /etc/rc.d/${_depend} forcestart; then warn "Unable to force ${_depend}. It may already be running." return 1 fi } # # checkyesno var # Test $1 variable, and warn if not set to YES or NO. # Return 0 if it's "yes" (et al), nonzero otherwise. # checkyesno() { eval _value=\$${1} debug "checkyesno: $1 is set to $_value." case $_value in # "yes", "true", "on", or "1" [Yy][Ee][Ss]|[Tt][Rr][Uu][Ee]|[Oo][Nn]|1) return 0 ;; # "no", "false", "off", or "0" [Nn][Oo]|[Ff][Aa][Ll][Ss][Ee]|[Oo][Ff][Ff]|0) return 1 ;; *) warn "\$${1} is not set properly - see rc.conf(5)." return 1 ;; esac } # # reverse_list list # print the list in reverse order # reverse_list() { _revlist= for _revfile; do _revlist="$_revfile $_revlist" done echo $_revlist } # stop_boot always # If booting directly to multiuser or $always is enabled, # send SIGTERM to the parent (/etc/rc) to abort the boot. # Otherwise just exit. # stop_boot() { local always case $1 in # "yes", "true", "on", or "1" [Yy][Ee][Ss]|[Tt][Rr][Uu][Ee]|[Oo][Nn]|1) always=true ;; *) always=false ;; esac if [ "$autoboot" = yes -o "$always" = true ]; then echo "ERROR: ABORTING BOOT (sending SIGTERM to parent)!" kill -TERM ${RC_PID} fi exit 1 } # # mount_critical_filesystems type # Go through the list of critical filesystems as provided in # the rc.conf(5) variable $critical_filesystems_${type}, checking # each one to see if it is mounted, and if it is not, mounting it. # mount_critical_filesystems() { eval _fslist=\$critical_filesystems_${1} for _fs in $_fslist; do mount | ( _ismounted=false while read what _on on _type type; do if [ $on = $_fs ]; then _ismounted=true fi done if $_ismounted; then : else mount $_fs >/dev/null 2>&1 fi ) done } # # check_pidfile pidfile procname [interpreter] # Parses the first line of pidfile for a PID, and ensures # that the process is running and matches procname. # Prints the matching PID upon success, nothing otherwise. # interpreter is optional; see _find_processes() for details. # check_pidfile() { _pidfile=$1 _procname=$2 _interpreter=$3 if [ -z "$_pidfile" -o -z "$_procname" ]; then err 3 'USAGE: check_pidfile pidfile procname [interpreter]' fi if [ ! -f $_pidfile ]; then debug "pid file ($_pidfile): not readable." return fi read _pid _junk < $_pidfile if [ -z "$_pid" ]; then debug "pid file ($_pidfile): no pid in file." return fi _find_processes $_procname ${_interpreter:-.} '-p '"$_pid" } # # check_process procname [interpreter] # Ensures that a process (or processes) named procname is running. # Prints a list of matching PIDs. # interpreter is optional; see _find_processes() for details. # check_process() { _procname=$1 _interpreter=$2 if [ -z "$_procname" ]; then err 3 'USAGE: check_process procname [interpreter]' fi _find_processes $_procname ${_interpreter:-.} '-ax' } # # _find_processes procname interpreter psargs # Search for procname in the output of ps generated by psargs. # Prints the PIDs of any matching processes, space separated. # # If interpreter == ".", check the following variations of procname # against the first word of each command: # procname # `basename procname` # `basename procname` + ":" # "(" + `basename procname` + ")" # "[" + `basename procname` + "]" # # If interpreter != ".", read the first line of procname, remove the # leading #!, normalise whitespace, append procname, and attempt to # match that against each command, either as is, or with extra words # at the end. As an alternative, to deal with interpreted daemons # using perl, the basename of the interpreter plus a colon is also # tried as the prefix to procname. # _find_processes() { if [ $# -ne 3 ]; then err 3 'USAGE: _find_processes procname interpreter psargs' fi _procname=$1 _interpreter=$2 _psargs=$3 _pref= if [ $_interpreter != "." ]; then # an interpreted script _script="${_chroot}${_chroot:+/}$_procname" if [ -r "$_script" ]; then read _interp < $_script # read interpreter name case "$_interp" in \#!*) _interp=${_interp#\#!} # strip #! set -- $_interp case $1 in */bin/env) shift # drop env to get real name ;; esac if [ $_interpreter != $1 ]; then warn "\$command_interpreter $_interpreter != $1" fi ;; *) warn "no shebang line in $_script" set -- $_interpreter ;; esac else warn "cannot read shebang line from $_script" set -- $_interpreter fi _interp="$* $_procname" # cleanup spaces, add _procname _interpbn=${1##*/} _fp_args='_argv' _fp_match='case "$_argv" in ${_interp}|"${_interp} "*|"[${_interpbn}]"|"${_interpbn}: ${_procname}"*)' else # a normal daemon _procnamebn=${_procname##*/} _fp_args='_arg0 _argv' _fp_match='case "$_arg0" in $_procname|$_procnamebn|${_procnamebn}:|"(${_procnamebn})"|"[${_procnamebn}]")' fi _proccheck="\ $PS 2>/dev/null -o pid= -o jid= -o command= $_psargs"' | while read _npid _jid '"$_fp_args"'; do '"$_fp_match"' if [ "$JID" -eq "$_jid" ]; then echo -n "$_pref$_npid"; _pref=" "; fi ;; esac done' # debug "in _find_processes: proccheck is ($_proccheck)." eval $_proccheck } # sort_lite [-b] [-n] [-k POS] [-t SEP] # A lite version of sort(1) (supporting a few options) that can be used # before the real sort(1) is available (e.g., in scripts that run prior # to mountcritremote). Requires only shell built-in functionality. # sort_lite() { local funcname=sort_lite local sort_sep="$IFS" sort_ignore_leading_space= local sort_field=0 sort_strict_fields= sort_numeric= local nitems=0 skip_leading=0 trim= local OPTIND flag while getopts bnk:t: flag; do case "$flag" in b) sort_ignore_leading_space=1 ;; n) sort_numeric=1 sort_ignore_leading_space=1 ;; k) sort_field="${OPTARG%%,*}" ;; # only up to first comma # NB: Unlike sort(1) only one POS allowed t) sort_sep="$OPTARG" if [ ${#sort_sep} -gt 1 ]; then echo "$funcname: multi-character tab \`$sort_sep'" >&2 return 1 fi sort_strict_fields=1 ;; \?) return 1 ;; esac done shift $(( $OPTIND - 1 )) # Create transformation pattern to trim leading text if desired case "$sort_field" in ""|[!0-9]*|*[!0-9.]*) echo "$funcname: invalid sort field \`$sort_field'" >&2 return 1 ;; *.*) skip_leading=${sort_field#*.} sort_field=${sort_field%%.*} while [ ${skip_leading:-0} -gt 1 ] 2> /dev/null; do trim="$trim?" skip_leading=$(( $skip_leading - 1 )) done esac # Copy input to series of local numbered variables # NB: IFS of NULL preserves leading whitespace local LINE while IFS= read -r LINE || [ "$LINE" ]; do nitems=$(( $nitems + 1 )) local src_$nitems="$LINE" done # # Sort numbered locals using insertion sort # local curitem curitem_orig curitem_mod curitem_haskey local dest dest_orig dest_mod dest_haskey local d gt n local i=1 while [ $i -le $nitems ]; do curitem_haskey=1 # Assume sort field (-k POS) exists eval curitem=\"\$src_$i\" curitem_mod="$curitem" # for modified comparison curitem_orig="$curitem" # for original comparison # Trim leading whitespace if desired if [ "$sort_ignore_leading_space" ]; then while case "$curitem_orig" in [$IFS]*) : ;; *) false; esac do curitem_orig="${curitem_orig#?}" done curitem_mod="$curitem_orig" fi # Shift modified comparison value if sort field (-k POS) is > 1 n=$sort_field while [ $n -gt 1 ]; do case "$curitem_mod" in *[$sort_sep]*) # Cut text up-to (and incl.) first separator curitem_mod="${curitem_mod#*[$sort_sep]}" # Skip NULLs unless strict field splitting [ "$sort_strict_fields" ] || [ "${curitem_mod%%[$sort_sep]*}" ] || [ $n -eq 2 ] || continue ;; *) # Asked for a field that doesn't exist curitem_haskey= break esac n=$(( $n - 1 )) done # Trim trailing words if sort field >= 1 [ $sort_field -ge 1 -a "$sort_numeric" ] && curitem_mod="${curitem_mod%%[$sort_sep]*}" # Apply optional trim (-k POS.TRIM) to cut leading characters curitem_mod="${curitem_mod#$trim}" # Determine the type of modified comparison to use initially # NB: Prefer numerical if requested but fallback to standard case "$curitem_mod" in ""|[!0-9]*) # NULL or begins with non-number gt=">" [ "$sort_numeric" ] && curitem_mod=0 ;; *) if [ "$sort_numeric" ]; then gt="-gt" curitem_mod="${curitem_mod%%[!0-9]*}" # NB: trailing non-digits removed # otherwise numeric comparison fails else gt=">" fi esac # If first time through, short-circuit below position-search if [ $i -le 1 ]; then d=0 else d=1 fi # # Find appropriate element position # while [ $d -gt 0 ] do dest_haskey=$curitem_haskey eval dest=\"\$dest_$d\" dest_mod="$dest" # for modified comparison dest_orig="$dest" # for original comparison # Trim leading whitespace if desired if [ "$sort_ignore_leading_space" ]; then while case "$dest_orig" in [$IFS]*) : ;; *) false; esac do dest_orig="${dest_orig#?}" done dest_mod="$dest_orig" fi # Shift modified value if sort field (-k POS) is > 1 n=$sort_field while [ $n -gt 1 ]; do case "$dest_mod" in *[$sort_sep]*) # Cut text up-to (and incl.) 1st sep dest_mod="${dest_mod#*[$sort_sep]}" # Skip NULLs unless strict fields [ "$sort_strict_fields" ] || [ "${dest_mod%%[$sort_sep]*}" ] || [ $n -eq 2 ] || continue ;; *) # Asked for a field that doesn't exist dest_haskey= break esac n=$(( $n - 1 )) done # Trim trailing words if sort field >= 1 [ $sort_field -ge 1 -a "$sort_numeric" ] && dest_mod="${dest_mod%%[$sort_sep]*}" # Apply optional trim (-k POS.TRIM), cut leading chars dest_mod="${dest_mod#$trim}" # Determine type of modified comparison to use # NB: Prefer numerical if requested, fallback to std case "$dest_mod" in ""|[!0-9]*) # NULL or begins with non-number gt=">" [ "$sort_numeric" ] && dest_mod=0 ;; *) if [ "$sort_numeric" ]; then gt="-gt" dest_mod="${dest_mod%%[!0-9]*}" # NB: kill trailing non-digits # for numeric comparison safety else gt=">" fi esac # Break if we've found the proper element position if [ "$curitem_haskey" -a "$dest_haskey" ]; then if [ "$dest_mod" = "$curitem_mod" ]; then [ "$dest_orig" ">" "$curitem_orig" ] && break elif [ "$dest_mod" $gt "$curitem_mod" ] \ 2> /dev/null then break fi else [ "$dest_orig" ">" "$curitem_orig" ] && break fi # Break if we've hit the end [ $d -ge $i ] && break d=$(( $d + 1 )) done # Shift remaining positions forward, making room for new item n=$i while [ $n -ge $d ]; do # Shift destination item forward one placement eval dest_$(( $n + 1 ))=\"\$dest_$n\" n=$(( $n - 1 )) done # Place the element if [ $i -eq 1 ]; then local dest_1="$curitem" else local dest_$d="$curitem" fi i=$(( $i + 1 )) done # Print sorted results d=1 while [ $d -le $nitems ]; do eval echo \"\$dest_$d\" d=$(( $d + 1 )) done } # # wait_for_pids pid [pid ...] # spins until none of the pids exist # wait_for_pids() { local _list _prefix _nlist _j _list="$@" if [ -z "$_list" ]; then return fi _prefix= while true; do _nlist=""; for _j in $_list; do if kill -0 $_j 2>/dev/null; then _nlist="${_nlist}${_nlist:+ }$_j" [ -n "$_prefix" ] && sleep 1 fi done if [ -z "$_nlist" ]; then break fi _list=$_nlist echo -n ${_prefix:-"Waiting for PIDS: "}$_list _prefix=", " pwait $_list 2>/dev/null done if [ -n "$_prefix" ]; then echo "." fi } # # get_pidfile_from_conf string file # # Takes a string to search for in the specified file. # Ignores lines with traditional comment characters. # # Example: # # if get_pidfile_from_conf string file; then # pidfile="$_pidfile_from_conf" # else # pidfile='appropriate default' # fi # get_pidfile_from_conf() { if [ -z "$1" -o -z "$2" ]; then err 3 "USAGE: get_pidfile_from_conf string file ($name)" fi local string file line string="$1" ; file="$2" if [ ! -s "$file" ]; then err 3 "get_pidfile_from_conf: $file does not exist ($name)" fi while read line; do case "$line" in *[#\;]*${string}*) continue ;; *${string}*) break ;; esac done < $file if [ -n "$line" ]; then line=${line#*/} _pidfile_from_conf="/${line%%[\"\;]*}" else return 1 fi } # # check_startmsgs # If rc_quiet is set (usually as a result of using faststart at # boot time) check if rc_startmsgs is enabled. # check_startmsgs() { if [ -n "$rc_quiet" ]; then checkyesno rc_startmsgs else return 0 fi } # # run_rc_command argument # Search for argument in the list of supported commands, which is: # "start stop restart rcvar status poll ${extra_commands}" # If there's a match, run ${argument}_cmd or the default method # (see below). # # If argument has a given prefix, then change the operation as follows: # Prefix Operation # ------ --------- # fast Skip the pid check, and set rc_fast=yes, rc_quiet=yes # force Set ${rcvar} to YES, and set rc_force=yes # one Set ${rcvar} to YES # quiet Don't output some diagnostics, and set rc_quiet=yes # # The following globals are used: # # Name Needed Purpose # ---- ------ ------- # name y Name of script. # # command n Full path to command. # Not needed if ${rc_arg}_cmd is set for # each keyword. # # command_args n Optional args/shell directives for command. # # command_interpreter n If not empty, command is interpreted, so # call check_{pidfile,process}() appropriately. # # desc n Description of script. # # extra_commands n List of extra commands supported. # # pidfile n If set, use check_pidfile $pidfile $command, # otherwise use check_process $command. # In either case, only check if $command is set. # # procname n Process name to check for instead of $command. # # rcvar n This is checked with checkyesno to determine # if the action should be run. # # ${name}_program n Full path to command. # Meant to be used in /etc/rc.conf to override # ${command}. # # ${name}_chroot n Directory to chroot to before running ${command} # Requires /usr to be mounted. # # ${name}_chdir n Directory to cd to before running ${command} # (if not using ${name}_chroot). # # ${name}_flags n Arguments to call ${command} with. # NOTE: $flags from the parent environment # can be used to override this. # # ${name}_env n Environment variables to run ${command} with. # # ${name}_fib n Routing table number to run ${command} with. # # ${name}_nice n Nice level to run ${command} at. # # ${name}_oomprotect n Don't kill ${command} when swap space is exhausted. # # ${name}_user n User to run ${command} as, using su(1) if not # using ${name}_chroot. # Requires /usr to be mounted. # # ${name}_group n Group to run chrooted ${command} as. # Requires /usr to be mounted. # # ${name}_groups n Comma separated list of supplementary groups # to run the chrooted ${command} with. # Requires /usr to be mounted. # # ${name}_prepend n Command added before ${command}. # # ${name}_login_class n Login class to use, else "daemon". # # ${rc_arg}_cmd n If set, use this as the method when invoked; # Otherwise, use default command (see below) # # ${rc_arg}_precmd n If set, run just before performing the # ${rc_arg}_cmd method in the default # operation (i.e, after checking for required # bits and process (non)existence). # If this completes with a non-zero exit code, # don't run ${rc_arg}_cmd. # # ${rc_arg}_postcmd n If set, run just after performing the # ${rc_arg}_cmd method, if that method # returned a zero exit code. # # required_dirs n If set, check for the existence of the given # directories before running a (re)start command. # # required_files n If set, check for the readability of the given # files before running a (re)start command. # # required_modules n If set, ensure the given kernel modules are # loaded before running a (re)start command. # The check and possible loads are actually # done after start_precmd so that the modules # aren't loaded in vain, should the precmd # return a non-zero status to indicate a error. # If a word in the list looks like "foo:bar", # "foo" is the KLD file name and "bar" is the # module name. If a word looks like "foo~bar", # "foo" is the KLD file name and "bar" is a # egrep(1) pattern matching the module name. # Otherwise the module name is assumed to be # the same as the KLD file name, which is most # common. See load_kld(). # # required_vars n If set, perform checkyesno on each of the # listed variables before running the default # (re)start command. # # Default behaviour for a given argument, if no override method is # provided: # # Argument Default behaviour # -------- ----------------- # start if !running && checkyesno ${rcvar} # ${command} # # stop if ${pidfile} # rc_pid=$(check_pidfile $pidfile $command) # else # rc_pid=$(check_process $command) # kill $sig_stop $rc_pid # wait_for_pids $rc_pid # ($sig_stop defaults to TERM.) # # reload Similar to stop, except use $sig_reload instead, # and doesn't wait_for_pids. # $sig_reload defaults to HUP. # Note that `reload' isn't provided by default, # it should be enabled via $extra_commands. # # restart Run `stop' then `start'. # # status Show if ${command} is running, etc. # # poll Wait for ${command} to exit. # # rcvar Display what rc.conf variable is used (if any). # # enabled Return true if the service is enabled. # # describe Show the service's description # # extracommands Show the service's extra commands # # Variables available to methods, and after run_rc_command() has # completed: # # Variable Purpose # -------- ------- # rc_arg Argument to command, after fast/force/one processing # performed # # rc_flags Flags to start the default command with. # Defaults to ${name}_flags, unless overridden # by $flags from the environment. # This variable may be changed by the precmd method. # # rc_pid PID of command (if appropriate) # # rc_fast Not empty if "fast" was provided (q.v.) # # rc_force Not empty if "force" was provided (q.v.) # # rc_quiet Not empty if "quiet" was provided # # run_rc_command() { _return=0 rc_arg=$1 if [ -z "$name" ]; then err 3 'run_rc_command: $name is not set.' fi # Don't repeat the first argument when passing additional command- # line arguments to the command subroutines. # shift 1 rc_extra_args="$*" _rc_prefix= case "$rc_arg" in fast*) # "fast" prefix; don't check pid rc_arg=${rc_arg#fast} rc_fast=yes rc_quiet=yes ;; force*) # "force" prefix; always run rc_force=yes _rc_prefix=force rc_arg=${rc_arg#${_rc_prefix}} if [ -n "${rcvar}" ]; then eval ${rcvar}=YES fi ;; one*) # "one" prefix; set ${rcvar}=yes _rc_prefix=one rc_arg=${rc_arg#${_rc_prefix}} if [ -n "${rcvar}" ]; then eval ${rcvar}=YES fi ;; quiet*) # "quiet" prefix; omit some messages _rc_prefix=quiet rc_arg=${rc_arg#${_rc_prefix}} rc_quiet=yes ;; esac eval _override_command=\$${name}_program command=${_override_command:-$command} _keywords="start stop restart rcvar enabled describe extracommands $extra_commands" rc_pid= _pidcmd= _procname=${procname:-${command}} # setup pid check command if [ -n "$_procname" ]; then if [ -n "$pidfile" ]; then _pidcmd='rc_pid=$(check_pidfile '"$pidfile $_procname $command_interpreter"')' else _pidcmd='rc_pid=$(check_process '"$_procname $command_interpreter"')' fi if [ -n "$_pidcmd" ]; then _keywords="${_keywords} status poll" fi fi if [ -z "$rc_arg" ]; then rc_usage $_keywords fi if [ "$rc_arg" = "enabled" ] ; then checkyesno ${rcvar} return $? fi if [ -n "$flags" ]; then # allow override from environment rc_flags=$flags else eval rc_flags=\$${name}_flags fi eval _chdir=\$${name}_chdir _chroot=\$${name}_chroot \ _nice=\$${name}_nice _user=\$${name}_user \ _group=\$${name}_group _groups=\$${name}_groups \ _fib=\$${name}_fib _env=\$${name}_env \ _prepend=\$${name}_prepend _login_class=\${${name}_login_class:-daemon} \ _oomprotect=\$${name}_oomprotect if [ -n "$_user" ]; then # unset $_user if running as that user if [ "$_user" = "$(eval $IDCMD)" ]; then unset _user fi fi [ -z "$autoboot" ] && eval $_pidcmd # determine the pid if necessary for _elem in $_keywords; do if [ "$_elem" != "$rc_arg" ]; then continue fi # if ${rcvar} is set, $1 is not "rcvar" and not "describe" # and ${rc_pid} is not set, then run # checkyesno ${rcvar} # and return if that failed # if [ -n "${rcvar}" -a "$rc_arg" != "rcvar" -a "$rc_arg" != "stop" \ -a "$rc_arg" != "describe" ] || [ -n "${rcvar}" -a "$rc_arg" = "stop" -a -z "${rc_pid}" ]; then if ! checkyesno ${rcvar}; then if [ -n "${rc_quiet}" ]; then return 0 fi echo -n "Cannot '${rc_arg}' $name. Set ${rcvar} to " echo -n "YES in /etc/rc.conf or use 'one${rc_arg}' " echo "instead of '${rc_arg}'." return 0 fi fi if [ $rc_arg = "start" -a -z "$rc_fast" -a -n "$rc_pid" ]; then if [ -z "$rc_quiet" ]; then echo 1>&2 "${name} already running? " \ "(pid=$rc_pid)." fi return 1 fi # if there's a custom ${XXX_cmd}, # run that instead of the default # eval _cmd=\$${rc_arg}_cmd \ _precmd=\$${rc_arg}_precmd \ _postcmd=\$${rc_arg}_postcmd if [ -n "$_cmd" ]; then _run_rc_precmd || return 1 _run_rc_doit "$_cmd $rc_extra_args" || return 1 _run_rc_postcmd return $_return fi case "$rc_arg" in # default operations... describe) if [ -n "$desc" ]; then echo "$desc" fi ;; extracommands) echo "$extra_commands" ;; status) _run_rc_precmd || return 1 if [ -n "$rc_pid" ]; then echo "${name} is running as pid $rc_pid." else echo "${name} is not running." return 1 fi _run_rc_postcmd ;; start) if [ ! -x "${_chroot}${_chroot:+/}${command}" ]; then warn "run_rc_command: cannot run $command" return 1 fi if ! _run_rc_precmd; then warn "failed precmd routine for ${name}" return 1 fi # setup the full command to run # check_startmsgs && echo "Starting ${name}." if [ -n "$_chroot" ]; then _doit="\ ${_nice:+nice -n $_nice }\ ${_fib:+setfib -F $_fib }\ ${_env:+env $_env }\ chroot ${_user:+-u $_user }${_group:+-g $_group }${_groups:+-G $_groups }\ $_chroot $command $rc_flags $command_args" else _doit="\ ${_chdir:+cd $_chdir && }\ ${_fib:+setfib -F $_fib }\ ${_env:+env $_env }\ $command $rc_flags $command_args" if [ -n "$_user" ]; then _doit="su -m $_user -c 'sh -c \"$_doit\"'" fi if [ -n "$_nice" ]; then if [ -z "$_user" ]; then _doit="sh -c \"$_doit\"" fi _doit="nice -n $_nice $_doit" fi if [ -n "$_prepend" ]; then _doit="$_prepend $_doit" fi fi # Prepend default limits _doit="limits -C $_login_class $_doit" # run the full command # if ! _run_rc_doit "$_doit"; then warn "failed to start ${name}" return 1 fi # finally, run postcmd # _run_rc_postcmd ;; stop) if [ -z "$rc_pid" ]; then [ -n "$rc_fast" ] && return 0 _run_rc_notrunning return 1 fi _run_rc_precmd || return 1 # send the signal to stop # echo "Stopping ${name}." _doit=$(_run_rc_killcmd "${sig_stop:-TERM}") _run_rc_doit "$_doit" || return 1 # wait for the command to exit, # and run postcmd. wait_for_pids $rc_pid _run_rc_postcmd ;; reload) if [ -z "$rc_pid" ]; then _run_rc_notrunning return 1 fi _run_rc_precmd || return 1 _doit=$(_run_rc_killcmd "${sig_reload:-HUP}") _run_rc_doit "$_doit" || return 1 _run_rc_postcmd ;; restart) # prevent restart being called more # than once by any given script # if ${_rc_restart_done:-false}; then return 0 fi _rc_restart_done=true _run_rc_precmd || return 1 # run those in a subshell to keep global variables ( run_rc_command ${_rc_prefix}stop $rc_extra_args ) ( run_rc_command ${_rc_prefix}start $rc_extra_args ) _return=$? [ $_return -ne 0 ] && [ -z "$rc_force" ] && return 1 _run_rc_postcmd ;; poll) _run_rc_precmd || return 1 if [ -n "$rc_pid" ]; then wait_for_pids $rc_pid fi _run_rc_postcmd ;; rcvar) echo -n "# $name" if [ -n "$desc" ]; then echo " : $desc" else echo "" fi echo "#" # Get unique vars in $rcvar $rcvars for _v in $rcvar $rcvars; do case $v in $_v\ *|\ *$_v|*\ $_v\ *) ;; *) v="${v# } $_v" ;; esac done # Display variables. for _v in $v; do if [ -z "$_v" ]; then continue fi eval _desc=\$${_v}_desc eval _defval=\$${_v}_defval _h="-" eval echo \"$_v=\\\"\$$_v\\\"\" # decode multiple lines of _desc while [ -n "$_desc" ]; do case $_desc in *^^*) echo "# $_h ${_desc%%^^*}" _desc=${_desc#*^^} _h=" " ;; *) echo "# $_h ${_desc}" break ;; esac done echo "# (default: \"$_defval\")" done echo "" ;; *) rc_usage $_keywords ;; esac # Apply protect(1) to the PID if ${name}_oomprotect is set. case "$rc_arg" in start) if [ -n "$_oomprotect" ]; then if [ -f "${PROTECT}" ]; then pid=$(check_process $command) case $_oomprotect in [Aa][Ll][Ll]) ${PROTECT} -i -p ${pid} ;; [Yy][Ee][Ss]) ${PROTECT} -p ${pid} ;; esac fi fi ;; esac return $_return done echo 1>&2 "$0: unknown directive '$rc_arg'." rc_usage $_keywords # not reached } # # Helper functions for run_rc_command: common code. # They use such global variables besides the exported rc_* ones: # # name R/W # ------------------ # _precmd R # _postcmd R # _return W # _run_rc_precmd() { check_required_before "$rc_arg" || return 1 if [ -n "$_precmd" ]; then debug "run_rc_command: ${rc_arg}_precmd: $_precmd $rc_extra_args" eval "$_precmd $rc_extra_args" _return=$? # If precmd failed and force isn't set, request exit. if [ $_return -ne 0 ] && [ -z "$rc_force" ]; then return 1 fi fi check_required_after "$rc_arg" || return 1 return 0 } _run_rc_postcmd() { if [ -n "$_postcmd" ]; then debug "run_rc_command: ${rc_arg}_postcmd: $_postcmd $rc_extra_args" eval "$_postcmd $rc_extra_args" _return=$? fi return 0 } _run_rc_doit() { debug "run_rc_command: doit: $*" eval "$@" _return=$? # If command failed and force isn't set, request exit. if [ $_return -ne 0 ] && [ -z "$rc_force" ]; then return 1 fi return 0 } _run_rc_notrunning() { local _pidmsg if [ -n "$pidfile" ]; then _pidmsg=" (check $pidfile)." else _pidmsg= fi echo 1>&2 "${name} not running?${_pidmsg}" } _run_rc_killcmd() { local _cmd _cmd="kill -$1 $rc_pid" if [ -n "$_user" ]; then _cmd="su -m ${_user} -c 'sh -c \"${_cmd}\"'" fi echo "$_cmd" } # # run_rc_script file arg # Start the script `file' with `arg', and correctly handle the # return value from the script. # If `file' ends with `.sh' and lives in /etc/rc.d, ignore it as it's # an old-style startup file. # If `file' ends with `.sh' and does not live in /etc/rc.d, it's sourced # into the current environment if $rc_fast_and_loose is set; otherwise # it is run as a child process. # If `file' appears to be a backup or scratch file, ignore it. # Otherwise if it is executable run as a child process. # run_rc_script() { _file=$1 _arg=$2 if [ -z "$_file" -o -z "$_arg" ]; then err 3 'USAGE: run_rc_script file arg' fi unset name command command_args command_interpreter \ extra_commands pidfile procname \ rcvar rcvars rcvars_obsolete required_dirs required_files \ required_vars eval unset ${_arg}_cmd ${_arg}_precmd ${_arg}_postcmd case "$_file" in /etc/rc.d/*.sh) # no longer allowed in the base warn "Ignoring old-style startup script $_file" ;; *[~#]|*.OLD|*.bak|*.orig|*,v) # scratch file; skip warn "Ignoring scratch file $_file" ;; *) # run in subshell if [ -x $_file ]; then if [ -n "$rc_fast_and_loose" ]; then set $_arg; . $_file else ( trap "echo Script $_file interrupted >&2 ; kill -QUIT $$" 3 trap "echo Script $_file interrupted >&2 ; exit 1" 2 trap "echo Script $_file running >&2" 29 set $_arg; . $_file ) fi fi ;; esac } # # load_rc_config [service] # Source in the configuration file(s) for a given service. # If no service is specified, only the global configuration # file(s) will be loaded. # load_rc_config() { local _name _rcvar_val _var _defval _v _msg _new _d _name=$1 if ${_rc_conf_loaded:-false}; then : else if [ -r /etc/defaults/rc.conf ]; then debug "Sourcing /etc/defaults/rc.conf" . /etc/defaults/rc.conf source_rc_confs elif [ -r /etc/rc.conf ]; then debug "Sourcing /etc/rc.conf (/etc/defaults/rc.conf doesn't exist)." . /etc/rc.conf fi _rc_conf_loaded=true fi # If a service name was specified, attempt to load # service-specific configuration if [ -n "$_name" ] ; then for _d in /etc ${local_startup}; do _d=${_d%/rc.d} if [ -f ${_d}/rc.conf.d/"$_name" ]; then debug "Sourcing ${_d}/rc.conf.d/$_name" . ${_d}/rc.conf.d/"$_name" elif [ -d ${_d}/rc.conf.d/"$_name" ] ; then local _rc for _rc in ${_d}/rc.conf.d/"$_name"/* ; do if [ -f "$_rc" ] ; then debug "Sourcing $_rc" . "$_rc" fi done fi done fi # Set defaults if defined. for _var in $rcvar $rcvars; do eval _defval=\$${_var}_defval if [ -n "$_defval" ]; then eval : \${$_var:=\$${_var}_defval} fi done # check obsolete rc.conf variables for _var in $rcvars_obsolete; do eval _v=\$$_var eval _msg=\$${_var}_obsolete_msg eval _new=\$${_var}_newvar case $_v in "") ;; *) if [ -z "$_new" ]; then _msg="Ignored." else eval $_new=\"\$$_var\" if [ -z "$_msg" ]; then _msg="Use \$$_new instead." fi fi warn "\$$_var is obsolete. $_msg" ;; esac done } # # load_rc_config_var name var # Read the rc.conf(5) var for name and set in the # current shell, using load_rc_config in a subshell to prevent # unwanted side effects from other variable assignments. # load_rc_config_var() { if [ $# -ne 2 ]; then err 3 'USAGE: load_rc_config_var name var' fi eval $(eval '( load_rc_config '$1' >/dev/null; if [ -n "${'$2'}" -o "${'$2'-UNSET}" != "UNSET" ]; then echo '$2'=\'\''${'$2'}\'\''; fi )' ) } # # rc_usage commands # Print a usage string for $0, with `commands' being a list of # valid commands. # rc_usage() { echo -n 1>&2 "Usage: $0 [fast|force|one|quiet](" _sep= for _elem; do echo -n 1>&2 "$_sep$_elem" _sep="|" done echo 1>&2 ")" exit 1 } # # err exitval message # Display message to stderr and log to the syslog, and exit with exitval. # err() { exitval=$1 shift if [ -x /usr/bin/logger ]; then logger "$0: ERROR: $*" fi echo 1>&2 "$0: ERROR: $*" exit $exitval } # # warn message # Display message to stderr and log to the syslog. # warn() { if [ -x /usr/bin/logger ]; then logger "$0: WARNING: $*" fi echo 1>&2 "$0: WARNING: $*" } # # info message # Display informational message to stdout and log to syslog. # info() { case ${rc_info} in [Yy][Ee][Ss]|[Tt][Rr][Uu][Ee]|[Oo][Nn]|1) if [ -x /usr/bin/logger ]; then logger "$0: INFO: $*" fi echo "$0: INFO: $*" ;; esac } # # debug message # If debugging is enabled in rc.conf output message to stderr. # BEWARE that you don't call any subroutine that itself calls this # function. # debug() { case ${rc_debug} in [Yy][Ee][Ss]|[Tt][Rr][Uu][Ee]|[Oo][Nn]|1) if [ -x /usr/bin/logger ]; then logger "$0: DEBUG: $*" fi echo 1>&2 "$0: DEBUG: $*" ;; esac } # # backup_file action file cur backup # Make a backup copy of `file' into `cur', and save the previous -# version of `cur' as `backup' or use rcs for archiving. +# version of `cur' as `backup'. # -# This routine checks the value of the backup_uses_rcs variable, -# which can be either YES or NO. -# # The `action' keyword can be one of the following: # # add `file' is now being backed up (and is possibly # being reentered into the backups system). `cur' -# is created and RCS files, if necessary, are -# created as well. +# is created. # # update `file' has changed and needs to be backed up. -# If `cur' exists, it is copied to to `back' or -# checked into RCS (if the repository file is old), -# and then `file' is copied to `cur'. Another RCS -# check in done here if RCS is being used. +# If `cur' exists, it is copied to to `back' +# and then `file' is copied to `cur'. # # remove `file' is no longer being tracked by the backups -# system. If RCS is not being used, `cur' is moved -# to `back', otherwise an empty file is checked in, -# and then `cur' is removed. +# system. `cur' is moved `back'. # # backup_file() { _action=$1 _file=$2 _cur=$3 _back=$4 - if checkyesno backup_uses_rcs; then - _msg0="backup archive" - _msg1="update" - - # ensure that history file is not locked - if [ -f $_cur,v ]; then - rcs -q -u -U -M $_cur - fi - - # ensure after switching to rcs that the - # current backup is not lost + case $_action in + add|update) if [ -f $_cur ]; then - # no archive, or current newer than archive - if [ ! -f $_cur,v -o $_cur -nt $_cur,v ]; then - ci -q -f -u -t-"$_msg0" -m"$_msg1" $_cur - rcs -q -kb -U $_cur - co -q -f -u $_cur - fi + cp -p $_cur $_back fi - - case $_action in - add|update) - cp -p $_file $_cur - ci -q -f -u -t-"$_msg0" -m"$_msg1" $_cur - rcs -q -kb -U $_cur - co -q -f -u $_cur - chown root:wheel $_cur $_cur,v - ;; - remove) - cp /dev/null $_cur - ci -q -f -u -t-"$_msg0" -m"$_msg1" $_cur - rcs -q -kb -U $_cur - chown root:wheel $_cur $_cur,v - rm $_cur - ;; - esac - else - case $_action in - add|update) - if [ -f $_cur ]; then - cp -p $_cur $_back - fi - cp -p $_file $_cur - chown root:wheel $_cur - ;; - remove) - mv -f $_cur $_back - ;; - esac - fi + cp -p $_file $_cur + chown root:wheel $_cur + ;; + remove) + mv -f $_cur $_back + ;; + esac } # make_symlink src link # Make a symbolic link 'link' to src from basedir. If the # directory in which link is to be created does not exist # a warning will be displayed and an error will be returned. # Returns 0 on success, 1 otherwise. # make_symlink() { local src link linkdir _me src="$1" link="$2" linkdir="`dirname $link`" _me="make_symlink()" if [ -z "$src" -o -z "$link" ]; then warn "$_me: requires two arguments." return 1 fi if [ ! -d "$linkdir" ]; then warn "$_me: the directory $linkdir does not exist." return 1 fi if ! ln -sf $src $link; then warn "$_me: unable to make a symbolic link from $link to $src" return 1 fi return 0 } # devfs_rulesets_from_file file # Reads a set of devfs commands from file, and creates # the specified rulesets with their rules. Returns non-zero # if there was an error. # devfs_rulesets_from_file() { local file _err _me _opts file="$1" _me="devfs_rulesets_from_file" _err=0 if [ -z "$file" ]; then warn "$_me: you must specify a file" return 1 fi if [ ! -e "$file" ]; then debug "$_me: no such file ($file)" return 0 fi # Disable globbing so that the rule patterns are not expanded # by accident with matching filesystem entries. _opts=$-; set -f debug "reading rulesets from file ($file)" { while read line do case $line in \#*) continue ;; \[*\]*) rulenum=`expr "$line" : "\[.*=\([0-9]*\)\]"` if [ -z "$rulenum" ]; then warn "$_me: cannot extract rule number ($line)" _err=1 break fi rulename=`expr "$line" : "\[\(.*\)=[0-9]*\]"` if [ -z "$rulename" ]; then warn "$_me: cannot extract rule name ($line)" _err=1 break; fi eval $rulename=\$rulenum debug "found ruleset: $rulename=$rulenum" if ! /sbin/devfs rule -s $rulenum delset; then _err=1 break fi ;; *) rulecmd="${line%%"\#*"}" # evaluate the command incase it includes # other rules if [ -n "$rulecmd" ]; then debug "adding rule ($rulecmd)" if ! eval /sbin/devfs rule -s $rulenum $rulecmd then _err=1 break fi fi ;; esac if [ $_err -ne 0 ]; then debug "error in $_me" break fi done } < $file case $_opts in *f*) ;; *) set +f ;; esac return $_err } # devfs_init_rulesets # Initializes rulesets from configuration files. Returns # non-zero if there was an error. # devfs_init_rulesets() { local file _me _me="devfs_init_rulesets" # Go through this only once if [ -n "$devfs_rulesets_init" ]; then debug "$_me: devfs rulesets already initialized" return fi for file in $devfs_rulesets; do if ! devfs_rulesets_from_file $file; then warn "$_me: could not read rules from $file" return 1 fi done devfs_rulesets_init=1 debug "$_me: devfs rulesets initialized" return 0 } # devfs_set_ruleset ruleset [dir] # Sets the default ruleset of dir to ruleset. The ruleset argument # must be a ruleset name as specified in devfs.rules(5) file. # Returns non-zero if it could not set it successfully. # devfs_set_ruleset() { local devdir rs _me [ -n "$1" ] && eval rs=\$$1 || rs= [ -n "$2" ] && devdir="-m "$2"" || devdir= _me="devfs_set_ruleset" if [ -z "$rs" ]; then warn "$_me: you must specify a ruleset number" return 1 fi debug "$_me: setting ruleset ($rs) on mount-point (${devdir#-m })" if ! /sbin/devfs $devdir ruleset $rs; then warn "$_me: unable to set ruleset $rs to ${devdir#-m }" return 1 fi return 0 } # devfs_apply_ruleset ruleset [dir] # Apply ruleset number $ruleset to the devfs mountpoint $dir. # The ruleset argument must be a ruleset name as specified # in a devfs.rules(5) file. Returns 0 on success or non-zero # if it could not apply the ruleset. # devfs_apply_ruleset() { local devdir rs _me [ -n "$1" ] && eval rs=\$$1 || rs= [ -n "$2" ] && devdir="-m "$2"" || devdir= _me="devfs_apply_ruleset" if [ -z "$rs" ]; then warn "$_me: you must specify a ruleset" return 1 fi debug "$_me: applying ruleset ($rs) to mount-point (${devdir#-m })" if ! /sbin/devfs $devdir rule -s $rs applyset; then warn "$_me: unable to apply ruleset $rs to ${devdir#-m }" return 1 fi return 0 } # devfs_domount dir [ruleset] # Mount devfs on dir. If ruleset is specified it is set # on the mount-point. It must also be a ruleset name as specified # in a devfs.rules(5) file. Returns 0 on success. # devfs_domount() { local devdir rs _me devdir="$1" [ -n "$2" ] && rs=$2 || rs= _me="devfs_domount()" if [ -z "$devdir" ]; then warn "$_me: you must specify a mount-point" return 1 fi debug "$_me: mount-point is ($devdir), ruleset is ($rs)" if ! mount -t devfs dev "$devdir"; then warn "$_me: Unable to mount devfs on $devdir" return 1 fi if [ -n "$rs" ]; then devfs_init_rulesets devfs_set_ruleset $rs $devdir devfs -m $devdir rule applyset fi return 0 } # Provide a function for normalizing the mounting of memory # filesystems. This should allow the rest of the code here to remain # as close as possible between 5-current and 4-stable. # $1 = size # $2 = mount point # $3 = (optional) extra mdmfs flags mount_md() { if [ -n "$3" ]; then flags="$3" fi /sbin/mdmfs $flags -s $1 md $2 } # Code common to scripts that need to load a kernel module # if it isn't in the kernel yet. Syntax: # load_kld [-e regex] [-m module] file # where -e or -m chooses the way to check if the module # is already loaded: # regex is egrep'd in the output from `kldstat -v', # module is passed to `kldstat -m'. # The default way is as though `-m file' were specified. load_kld() { local _loaded _mod _opt _re while getopts "e:m:" _opt; do case "$_opt" in e) _re="$OPTARG" ;; m) _mod="$OPTARG" ;; *) err 3 'USAGE: load_kld [-e regex] [-m module] file' ;; esac done shift $(($OPTIND - 1)) if [ $# -ne 1 ]; then err 3 'USAGE: load_kld [-e regex] [-m module] file' fi _mod=${_mod:-$1} _loaded=false if [ -n "$_re" ]; then if kldstat -v | egrep -q -e "$_re"; then _loaded=true fi else if kldstat -q -m "$_mod"; then _loaded=true fi fi if ! $_loaded; then if ! kldload "$1"; then warn "Unable to load kernel module $1" return 1 else info "$1 kernel module loaded." fi else debug "load_kld: $1 kernel module already loaded." fi return 0 } # ltr str src dst [var] # Change every $src in $str to $dst. # Useful when /usr is not yet mounted and we cannot use tr(1), sed(1) nor # awk(1). If var is non-NULL, set it to the result. ltr() { local _str _src _dst _out _com _var _str="$1" _src="$2" _dst="$3" _var="$4" _out="" local IFS="${_src}" for _com in ${_str}; do if [ -z "${_out}" ]; then _out="${_com}" else _out="${_out}${_dst}${_com}" fi done if [ -n "${_var}" ]; then setvar "${_var}" "${_out}" else echo "${_out}" fi } # Creates a list of providers for GELI encryption. geli_make_list() { local devices devices2 local provider mountpoint type options rest # Create list of GELI providers from fstab. while read provider mountpoint type options rest ; do case ":${options}" in :*noauto*) noauto=yes ;; *) noauto=no ;; esac case ":${provider}" in :#*) continue ;; *.eli) # Skip swap devices. if [ "${type}" = "swap" -o "${options}" = "sw" -o "${noauto}" = "yes" ]; then continue fi devices="${devices} ${provider}" ;; esac done < /etc/fstab # Append providers from geli_devices. devices="${devices} ${geli_devices}" for provider in ${devices}; do provider=${provider%.eli} provider=${provider#/dev/} devices2="${devices2} ${provider}" done echo ${devices2} } # Originally, root mount hold had to be released before mounting # the root filesystem. This delayed the boot, so it was changed # to only wait if the root device isn't readily available. This # can result in rc scripts executing before all the devices - such # as graid(8), or USB disks - can be accessed. This function can # be used to explicitly wait for root mount holds to be released. root_hold_wait() { local wait waited holders waited=0 while true; do holders="$(sysctl -n vfs.root_mount_hold)" if [ -z "${holders}" ]; then break; fi if [ ${waited} -eq 0 ]; then echo -n "Waiting ${root_hold_delay}s" \ "for the root mount holders: ${holders}" else echo -n . fi if [ ${waited} -ge ${root_hold_delay} ]; then echo break fi sleep 1 waited=$(($waited + 1)) done } # Find scripts in local_startup directories that use the old syntax # find_local_scripts_old() { zlist='' slist='' for dir in ${local_startup}; do if [ -d "${dir}" ]; then for file in ${dir}/[0-9]*.sh; do grep '^# PROVIDE:' $file >/dev/null 2>&1 && continue zlist="$zlist $file" done for file in ${dir}/[!0-9]*.sh; do grep '^# PROVIDE:' $file >/dev/null 2>&1 && continue slist="$slist $file" done fi done } find_local_scripts_new() { local_rc='' for dir in ${local_startup}; do if [ -d "${dir}" ]; then for file in `grep -l '^# PROVIDE:' ${dir}/* 2>/dev/null`; do case "$file" in *.sample) ;; *) if [ -x "$file" ]; then local_rc="${local_rc} ${file}" fi ;; esac done fi done } # check_required_{before|after} command # Check for things required by the command before and after its precmd, # respectively. The two separate functions are needed because some # conditions should prevent precmd from being run while other things # depend on precmd having already been run. # check_required_before() { local _f case "$1" in start) for _f in $required_vars; do if ! checkyesno $_f; then warn "\$${_f} is not enabled." if [ -z "$rc_force" ]; then return 1 fi fi done for _f in $required_dirs; do if [ ! -d "${_f}/." ]; then warn "${_f} is not a directory." if [ -z "$rc_force" ]; then return 1 fi fi done for _f in $required_files; do if [ ! -r "${_f}" ]; then warn "${_f} is not readable." if [ -z "$rc_force" ]; then return 1 fi fi done ;; esac return 0 } check_required_after() { local _f _args case "$1" in start) for _f in $required_modules; do case "${_f}" in *~*) _args="-e ${_f#*~} ${_f%%~*}" ;; *:*) _args="-m ${_f#*:} ${_f%%:*}" ;; *) _args="${_f}" ;; esac if ! load_kld ${_args}; then if [ -z "$rc_force" ]; then return 1 fi fi done ;; esac return 0 } # check_jail mib # Return true if security.jail.$mib exists and set to 1. check_jail() { local _mib _v _mib=$1 if _v=$(${SYSCTL_N} "security.jail.$_mib" 2> /dev/null); then case $_v in 1) return 0;; esac fi return 1 } # check_kern_features mib # Return existence of kern.features.* sysctl MIB as true or # false. The result will be cached in $_rc_cache_kern_features_ # namespace. "0" means the kern.features.X exists. check_kern_features() { local _v [ -n "$1" ] || return 1; eval _v=\$_rc_cache_kern_features_$1 [ -n "$_v" ] && return "$_v"; if ${SYSCTL_N} kern.features.$1 > /dev/null 2>&1; then eval _rc_cache_kern_features_$1=0 return 0 else eval _rc_cache_kern_features_$1=1 return 1 fi } # check_namevarlist var # Return "0" if ${name}_var is reserved in rc.subr. _rc_namevarlist="program chroot chdir env flags fib nice user group groups prepend" check_namevarlist() { local _v for _v in $_rc_namevarlist; do case $1 in $_v) return 0 ;; esac done return 1 } # _echoonce var msg mode # mode=0: Echo $msg if ${$var} is empty. # After doing echo, a string is set to ${$var}. # # mode=1: Echo $msg if ${$var} is a string with non-zero length. # _echoonce() { local _var _msg _mode eval _var=\$$1 _msg=$2 _mode=$3 case $_mode in 1) [ -n "$_var" ] && echo "$_msg" ;; *) [ -z "$_var" ] && echo -n "$_msg" && eval "$1=finished" ;; esac } fi # [ -z "${_rc_subr_loaded}" ] _rc_subr_loaded=: Index: head/share/man/man8/rc.subr.8 =================================================================== --- head/share/man/man8/rc.subr.8 (revision 305928) +++ head/share/man/man8/rc.subr.8 (revision 305929) @@ -1,915 +1,894 @@ .\" $NetBSD: rc.subr.8,v 1.12 2004/01/06 00:52:24 lukem Exp $ .\" .\" Copyright (c) 2002-2004 The NetBSD Foundation, Inc. .\" All rights reserved. .\" .\" This code is derived from software contributed to The NetBSD Foundation .\" by Luke Mewburn. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS .\" ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED .\" TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR .\" PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS .\" BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR .\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF .\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS .\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN .\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) .\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE .\" POSSIBILITY OF SUCH DAMAGE. .\" .\" $FreeBSD$ .\" -.Dd February 4, 2016 +.Dd September 18, 2016 .Dt RC.SUBR 8 .Os .Sh NAME .Nm rc.subr .Nd functions used by system shell scripts .Sh SYNOPSIS .Bl -item -compact .It .Ic .\& Pa /etc/rc.subr .Pp .It .Ic backup_file Ar action Ar file Ar current Ar backup .It .Ic checkyesno Ar var .It .Ic check_pidfile Ar pidfile Ar procname Op Ar interpreter .It .Ic check_process Ar procname Op Ar interpreter .It .Ic debug Ar message .It .Ic err Ar exitval Ar message .It .Ic force_depend Ar name .It .Ic info Ar message .It .Ic load_kld Oo Fl e Ar regex Oc Oo Fl m Ar module Oc Ar file .It .Ic load_rc_config Ar name .It .Ic load_rc_config_var Ar name Ar var .It .Ic mount_critical_filesystems Ar type .It .Ic rc_usage Ar command ... .It .Ic reverse_list Ar item ... .It .Ic run_rc_command Ar argument .It .Ic run_rc_script Ar file Ar argument .It .Ic wait_for_pids Op Ar pid ... .It .Ic warn Ar message .El .Sh DESCRIPTION The .Nm script contains commonly used shell script functions and variable definitions which are used by various scripts such as .Xr rc 8 . Scripts required by ports in .Pa /usr/local/etc/rc.d will also eventually be rewritten to make use of it. .Pp The .Nm functions were mostly imported from .Nx . .Pp They are accessed by sourcing .Pa /etc/rc.subr into the current shell. .Pp The following shell functions are available: .Bl -tag -width 4n .It Ic backup_file Ar action file current backup Make a backup copy of .Ar file into .Ar current . -If the -.Xr rc.conf 5 -variable -.Va backup_uses_rcs -is -.Dq Li YES , -use -.Xr rcs 1 -to archive the previous version of -.Ar current , -otherwise save the previous version of +Save the previous version of .Ar current as .Ar backup . .Pp The .Ar action argument may be one of the following: .Bl -tag -width ".Cm remove" .It Cm add .Ar file is now being backed up by or possibly re-entered into this backup mechanism. .Ar current -is created, and if necessary, the -.Xr rcs 1 -files are created as well. +is created. .It Cm update .Ar file has changed and needs to be backed up. If .Ar current exists, it is copied to .Ar backup -or checked into -.Xr rcs 1 -(if the repository file is old), and then .Ar file is copied to .Ar current . .It Cm remove .Ar file is no longer being tracked by this backup mechanism. -If -.Xr rcs 1 -is being used, an empty file is checked in and -.Ar current -is removed, -otherwise .Ar current is moved to .Ar backup . .El .It Ic checkyesno Ar var Return 0 if .Ar var is defined to .Dq Li YES , .Dq Li TRUE , .Dq Li ON , or .Ql 1 . Return 1 if .Ar var is defined to .Dq Li NO , .Dq Li FALSE , .Dq Li OFF , or .Ql 0 . Otherwise, warn that .Ar var is not set correctly. The values are case insensitive. .Sy Note : .Ar var should be a variable name, not its value; .Ic checkyesno will expand the variable by itself. .It Ic check_pidfile Ar pidfile procname Op Ar interpreter Parses the first word of the first line of .Ar pidfile for a PID, and ensures that the process with that PID is running and its first argument matches .Ar procname . Prints the matching PID if successful, otherwise nothing. If .Ar interpreter is provided, parse the first line of .Ar procname , ensure that the line is of the form: .Pp .Dl "#! interpreter [...]" .Pp and use .Ar interpreter with its optional arguments and .Ar procname appended as the process string to search for. .It Ic check_process Ar procname Op Ar interpreter Prints the PIDs of any processes that are running with a first argument that matches .Ar procname . .Ar interpreter is handled as per .Ic check_pidfile . .It Ic debug Ar message Display a debugging message to .Va stderr , log it to the system log using .Xr logger 1 , and return to the caller. The error message consists of the script name (from .Va $0 ) , followed by .Dq Li ": DEBUG: " , and then .Ar message . This function is intended to be used by developers as an aid to debugging scripts. It can be turned on or off by the .Xr rc.conf 5 variable .Va rc_debug . .It Ic err Ar exitval message Display an error message to .Va stderr , log it to the system log using .Xr logger 1 , and .Ic exit with an exit value of .Ar exitval . The error message consists of the script name (from .Va $0 ) , followed by .Dq Li ": ERROR: " , and then .Ar message . .It Ic force_depend Ar name Output an advisory message and force the .Ar name service to start. The .Ar name argument is the .Xr basename 1 component of the path to the script, usually .Pa /etc/rc.d/name . If the script fails for any reason it will output a warning and return with a return value of 1. If it was successful it will return 0. .It Ic info Ar message Display an informational message to .Va stdout , and log it to the system log using .Xr logger 1 . The message consists of the script name (from .Va $0 ) , followed by .Dq Li ": INFO: " , and then .Ar message . The display of this informational output can be turned on or off by the .Xr rc.conf 5 variable .Va rc_info . .It Ic load_kld Oo Fl e Ar regex Oc Oo Fl m Ar module Oc Ar file Load .Ar file as a kernel module unless it is already loaded. For the purpose of checking the module status, either the exact module name can be specified using .Fl m , or an .Xr egrep 1 regular expression matching the module name can be supplied via .Fl e . By default, the module is assumed to have the same name as .Ar file , which is not always the case. .It Ic load_rc_config Ar name Source in the configuration files for .Ar name . First, .Pa /etc/rc.conf is sourced if it has not yet been read in. Then, .Pa /etc/rc.conf.d/ Ns Ar name is sourced if it is an existing file. The latter may also contain other variable assignments to override .Ic run_rc_command arguments defined by the calling script, to provide an easy mechanism for an administrator to override the behaviour of a given .Xr rc.d 8 script without requiring the editing of that script. .It Ic load_rc_config_var Ar name Ar var Read the .Xr rc.conf 5 variable .Ar var for .Ar name and set in the current shell, using .Ic load_rc_config in a sub-shell to prevent unwanted side effects from other variable assignments. .It Ic mount_critical_filesystems Ar type Go through a list of critical file systems, as found in the .Xr rc.conf 5 variable .Va critical_filesystems_ Ns Ar type , mounting each one that is not currently mounted. .It Ic rc_usage Ar command ... Print a usage message for .Va $0 , with .Ar commands being the list of valid arguments prefixed by .Sm off .Dq Bq Li fast | force | one | quiet . .Sm on .It Ic reverse_list Ar item ... Print the list of .Ar items in reverse order. .It Ic run_rc_command Ar argument Run the .Ar argument method for the current .Xr rc.d 8 script, based on the settings of various shell variables. .Ic run_rc_command is extremely flexible, and allows fully functional .Xr rc.d 8 scripts to be implemented in a small amount of shell code. .Pp .Ar argument is searched for in the list of supported commands, which may be one of: .Bl -tag -width ".Cm restart" -offset indent .It Cm start Start the service. This should check that the service is to be started as specified by .Xr rc.conf 5 . Also checks if the service is already running and refuses to start if it is. This latter check is not performed by standard .Fx scripts if the system is starting directly to multi-user mode, to speed up the boot process. .It Cm stop If the service is to be started as specified by .Xr rc.conf 5 , stop the service. This should check that the service is running and complain if it is not. .It Cm restart Perform a .Cm stop then a .Cm start . Defaults to displaying the process ID of the program (if running). .It Cm enabled Return 0 if the service is enabled and 1 if it is not. This command does not print anything. .It Cm rcvar Display which .Xr rc.conf 5 variables are used to control the startup of the service (if any). .El .Pp If .Va pidfile or .Va procname is set, also support: .Bl -tag -width ".Cm restart" -offset indent .It Cm poll Wait for the command to exit. .It Cm status Show the status of the process. .El .Pp Other supported commands are listed in the optional variable .Va extra_commands . .Pp .Ar argument may have one of the following prefixes which alters its operation: .Bl -tag -width ".Li force" -offset indent .It Li fast Skip the check for an existing running process, and sets .Va rc_fast Ns = Ns Li YES . .It Li force Skip the checks for .Va rcvar being set to .Dq Li YES , and sets .Va rc_force Ns = Ns Li YES . This ignores .Ar argument Ns Va _precmd returning non-zero, and ignores any of the .Va required_* tests failing, and always returns a zero exit status. .It Li one Skip the checks for .Va rcvar being set to .Dq Li YES , but performs all the other prerequisite tests. .It Li quiet Inhibits some verbose diagnostics. Currently, this includes messages .Qq Starting ${name} (as checked by .Ic check_startmsgs inside .Nm ) and errors about usage of services that are not enabled in .Xr rc.conf 5 . This prefix also sets .Va rc_quiet Ns = Ns Li YES . .Em Please, note: .Va rc_quiet is not intended to completely mask all debug and warning messages, but only certain small classes of them. .El .Pp .Ic run_rc_command uses the following shell variables to control its behaviour. Unless otherwise stated, these are optional. .Bl -tag -width ".Va procname" -offset indent .It Va name The name of this script. This is not optional. .It Va rcvar The value of .Va rcvar is checked with .Ic checkyesno to determine if this method should be run. .It Va command Full path to the command. Not required if .Ar argument Ns Va _cmd is defined for each supported keyword. Can be overridden by .Va ${name}_program . .It Va command_args Optional arguments and/or shell directives for .Va command . .It Va command_interpreter .Va command is started with: .Pp .Dl "#! command_interpreter [...]" .Pp which results in its .Xr ps 1 command being: .Pp .Dl "command_interpreter [...] command" .Pp so use that string to find the PID(s) of the running command rather than .Va command . .It Va extra_commands Extra commands/keywords/arguments supported. .It Va pidfile Path to PID file. Used to determine the PID(s) of the running command. If .Va pidfile is set, use: .Pp .Dl "check_pidfile $pidfile $procname" .Pp to find the PID. Otherwise, if .Va command is set, use: .Pp .Dl "check_process $procname" .Pp to find the PID. .It Va procname Process name to check for. Defaults to the value of .Va command . .It Va required_dirs Check for the existence of the listed directories before running the .Cm start method. .It Va required_files Check for the readability of the listed files before running the .Cm start method. .It Va required_modules Ensure that the listed kernel modules are loaded before running the .Cm start method. This is done after invoking the commands from .Va start_precmd so that the missing modules are not loaded in vain if the preliminary commands indicate a error condition. A word in the list can have an optional .Dq Li : Ns Ar modname or .Dq Li ~ Ns Ar pattern suffix. The .Ar modname or .Ar pattern parameter is passed to .Ic load_kld through a .Fl m or .Fl e option, respectively. See the description of .Ic load_kld in this document for details. .It Va required_vars Perform .Ic checkyesno on each of the list variables before running the .Cm start method. .It Va ${name}_chdir Directory to .Ic cd to before running .Va command , if .Va ${name}_chroot is not provided. .It Va ${name}_chroot Directory to .Xr chroot 8 to before running .Va command . Only supported after .Pa /usr is mounted. .It Va ${name}_env A list of environment variables to run .Va command with. This will be passed as arguments to .Xr env 1 utility. .It Va ${name}_fib FIB .Pa Routing Table number to run .Va command with. See .Xr setfib 1 for more details. .It Va ${name}_flags Arguments to call .Va command with. This is usually set in .Xr rc.conf 5 , and not in the .Xr rc.d 8 script. The environment variable .Sq Ev flags can be used to override this. .It Va ${name}_nice .Xr nice 1 level to run .Va command as. Only supported after .Pa /usr is mounted. .It Va ${name}_oomprotect .Xr protect 1 .Va command from being killed when swap space is exhausted. If .Em YES is used, no child processes are protected. If .Em ALL , protect all child processes. .It Va ${name}_program Full path to the command. Overrides .Va command if both are set, but has no effect if .Va command is unset. As a rule, .Va command should be set in the script while .Va ${name}_program should be set in .Xr rc.conf 5 . .It Va ${name}_user User to run .Va command as, using .Xr chroot 8 if .Va ${name}_chroot is set, otherwise uses .Xr su 1 . Only supported after .Pa /usr is mounted. .It Va ${name}_group Group to run the chrooted .Va command as. .It Va ${name}_groups Comma separated list of supplementary groups to run the chrooted .Va command with. .It Va ${name}_prepend Commands to be prepended to .Va command . This is a generic version of .Va ${name}_env , .Va ${name}_fib , or .Va ${name}_nice . .It Ar argument Ns Va _cmd Shell commands which override the default method for .Ar argument . .It Ar argument Ns Va _precmd Shell commands to run just before running .Ar argument Ns Va _cmd or the default method for .Ar argument . If this returns a non-zero exit code, the main method is not performed. If the default method is being executed, this check is performed after the .Va required_* checks and process (non-)existence checks. .It Ar argument Ns Va _postcmd Shell commands to run if running .Ar argument Ns Va _cmd or the default method for .Ar argument returned a zero exit code. .It Va sig_stop Signal to send the processes to stop in the default .Cm stop method. Defaults to .Dv SIGTERM . .It Va sig_reload Signal to send the processes to reload in the default .Cm reload method. Defaults to .Dv SIGHUP . .El .Pp For a given method .Ar argument , if .Ar argument Ns Va _cmd is not defined, then a default method is provided by .Ic run_rc_command : .Bl -tag -width ".Sy Argument" -offset indent .It Sy Argument .Sy Default method .It Cm start If .Va command is not running and .Ic checkyesno Va rcvar succeeds, start .Va command . .It Cm stop Determine the PIDs of .Va command with .Ic check_pidfile or .Ic check_process (as appropriate), .Ic kill Va sig_stop those PIDs, and run .Ic wait_for_pids on those PIDs. .It Cm reload Similar to .Cm stop , except that it uses .Va sig_reload instead, and does not run .Ic wait_for_pids . Another difference from .Cm stop is that .Cm reload is not provided by default. It can be enabled via .Va extra_commands if appropriate: .Pp .Dl "extra_commands=reload" .It Cm restart Runs the .Cm stop method, then the .Cm start method. .It Cm status Show the PID of .Va command , or some other script specific status operation. .It Cm poll Wait for .Va command to exit. .It Cm rcvar Display which .Xr rc.conf 5 variable is used (if any). This method always works, even if the appropriate .Xr rc.conf 5 variable is set to .Dq Li NO . .El .Pp The following variables are available to the methods (such as .Ar argument Ns Va _cmd ) as well as after .Ic run_rc_command has completed: .Bl -tag -width ".Va rc_flags" -offset indent .It Va rc_arg Argument provided to .Ic run_rc_command , after fast and force processing has been performed. .It Va rc_flags Flags to start the default command with. Defaults to .Va ${name}_flags , unless overridden by the environment variable .Sq Ev flags . This variable may be changed by the .Ar argument Ns Va _precmd method. .It Va rc_pid PID of .Va command (if appropriate). .It Va rc_fast Not empty if .Dq Li fast prefix was used. .It Va rc_force Not empty if .Dq Li force prefix was used. .El .It Ic run_rc_script Ar file argument Start the script .Ar file with an argument of .Ar argument , and handle the return value from the script. .Pp Various shell variables are unset before .Ar file is started: .Bd -ragged -offset indent .Va name , .Va command , .Va command_args , .Va command_interpreter , .Va extra_commands , .Va pidfile , .Va rcvar , .Va required_dirs , .Va required_files , .Va required_vars , .Ar argument Ns Va _cmd , .Ar argument Ns Va _precmd . .Ar argument Ns Va _postcmd . .Ed .Pp The startup behaviour of .Ar file depends upon the following checks: .Bl -enum .It If .Ar file ends in .Pa .sh , it is sourced into the current shell. .It If .Ar file appears to be a backup or scratch file (e.g., with a suffix of .Pa ~ , # , .OLD , or .Pa .orig ) , ignore it. .It If .Ar file is not executable, ignore it. .It If the .Xr rc.conf 5 variable .Va rc_fast_and_loose is empty, source .Ar file in a sub shell, otherwise source .Ar file into the current shell. .El .It Ic stop_boot Op Ar always Prevent booting to multiuser mode. If the .Va autoboot variable is set to .Ql yes , or .Ic checkyesno Ar always indicates a truth value, then a .Dv SIGTERM signal is sent to the parent process, which is assumed to be .Xr rc 8 . Otherwise, the shell exits with a non-zero status. .It Ic wait_for_pids Op Ar pid ... Wait until all of the provided .Ar pids do not exist any more, printing the list of outstanding .Ar pids every two seconds. .It Ic warn Ar message Display a warning message to .Va stderr and log it to the system log using .Xr logger 1 . The warning message consists of the script name (from .Va $0 ) , followed by .Dq Li ": WARNING: " , and then .Ar message . .El .Sh FILES .Bl -tag -width ".Pa /etc/rc.subr" -compact .It Pa /etc/rc.subr The .Nm file resides in .Pa /etc . .El .Sh SEE ALSO .Xr rc.conf 5 , .Xr rc 8 .Sh HISTORY The .Nm script appeared in .Nx 1.3 . The .Xr rc.d 8 support functions appeared in .Nx 1.5 . The .Nm script first appeared in .Fx 5.0 .