Date: Fri, 25 Nov 2022 09:41:28 -0700 From: Warner Losh <imp@bsdimp.com> To: Alexander Leidinger <netchild@freebsd.org> Cc: src-committers <src-committers@freebsd.org>, "<dev-commits-src-all@freebsd.org>" <dev-commits-src-all@freebsd.org>, dev-commits-src-main@freebsd.org Subject: Re: git: ff0c7816db69 - main - Remove UPDATING entries from old branches. Message-ID: <CANCZdfq%2BAVGWa91Cv80t60jKAmw0UwoTVNFeOGRjOhAPjsJH%2Bw@mail.gmail.com> In-Reply-To: <202211250923.2AP9NakT073087@gitrepo.freebsd.org> References: <202211250923.2AP9NakT073087@gitrepo.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
--00000000000021a3de05ee4e325b Content-Type: text/plain; charset="UTF-8" Please revert this. We keep older updating entries on purpose. You purged way too much. Let's chat about how much to remove in arch@. They are for more than just source updates, so your reasoning is wrong. They are also there for users updating their products which can have a larger leap in time. We've traditionally kept closer to 5-10 years here for that reason. Warner Warner On Fri, Nov 25, 2022, 2:23 AM Alexander Leidinger <netchild@freebsd.org> wrote: > The branch main has been updated by netchild: > > URL: > https://cgit.FreeBSD.org/src/commit/?id=ff0c7816db696d31adc437134dcad45a70ad5889 > > commit ff0c7816db696d31adc437134dcad45a70ad5889 > Author: Alexander Leidinger <netchild@FreeBSD.org> > AuthorDate: 2022-11-25 09:17:14 +0000 > Commit: Alexander Leidinger <netchild@FreeBSD.org> > CommitDate: 2022-11-25 09:17:14 +0000 > > Remove UPDATING entries from old branches. > > We only support updates from major version N to N+1: > stable/13 was branched on 20210122, remove all old entries from > stable/10 > branch point in 2013 to 20210122. > --- > UPDATING | 2175 > -------------------------------------------------------------- > 1 file changed, 2175 deletions(-) > > diff --git a/UPDATING b/UPDATING > index 3782692c2a49..83e97246b964 100644 > --- a/UPDATING > +++ b/UPDATING > @@ -275,2181 +275,6 @@ > https://lists.freebsd.org/archives/freebsd-current/2021-June/000153.html > Please update your drm-kmod port to after the __FreeBSD_version > 1400003 > update. > > -20210108: > - PC Card attachments for all devices have been removed. In the case > of > - wi and cmx, the entire drivers were removed because they were only > - PC Card devices. FreeBSD_version 1300134 should be used for this > - since it was bumped so recently. > - > -20210107: > - Transport-independent parts of HID support have been split off the > USB > - code in to separate subsystem. Kernel configs which include one of > - ums, ukbd, uhid, atp, wsp, wmt, uaudio, ugold or ucycom drivers > should > - be updated with adding of "device hid" line. > - > -20210105: > - ncurses installation has been modified to only keep the widechar > - enabled version. Incremental build is broken for that change, so > it > - requires a clean build. > - > -20201223: > - The FreeBSD project has migrated from Subversion to Git. Temporary > - instructions can be found at > - https://github.com/bsdimp/freebsd-git-docs/blob/main/src-cvt.md > - and other documents in that repo. > - > -20201216: > - The services database has been updated to cover more of the basic > - services expected in a modern system. The database is big enough > - that it will cause issues in mergemaster in Releases previous to > - 12.2 and 11.3, or in very old current systems from before r358154. > - > -20201215: > - Obsolete in-tree GDB 6.1.1 has been removed. GDB (including kgdb) > - may be installed from ports or packages. > - > -20201124: > - ping6 has been merged into ping. It can now be called as "ping > -6". > - See ping(8) for details. > - > -20201108: > - Default value of net.add_addr_allfibs has been changed to 0. > - If you have multi-fib configuration and rely on existence of all > - interface routes in every fib, you need to set the above sysctl to > 1. > - > -20201030: > - The internal pre-processor in the calendar(1) program has been > - extended to support more C pre-processor commands (e.g. #ifdef, > #else, > - and #undef) and to detect unbalanced conditional statements. > - Error messages have been extended to include the filename and line > - number if processing stops to help fixing malformed data files. > - > -20201026: > - All the data files for the calendar(1) program, except > calendar.freebsd, > - have been moved to the deskutils/calendar-data port, much like the > - jewish calendar entries were moved to deskutils/hebcal years ago. > After > - make delete-old-files, you need to install it to retain full > - functionality. calendar(1) will issue a reminder for files it can't > - find. > - > -20200923: > - LINT files are no longer generated. We now include the relevant > NOTES > - files. Note: This may cause conflicts with updating in some cases. > - find sys -name LINT\* -delete > - is suggested across this commit to remove the generated LINT files. > - > - If you have tried to update with generated files there, the svn > - command you want to un-auger the tree is > - cd sys/amd64/conf > - svn revert -R . > - and then do the above find from the top level. Substitute 'amd64' > - above with where the error message indicates a conflict. > - > -20200824: > - OpenZFS support has been integrated. Do not upgrade root pools > until > - the loader is updated to support zstd. Furthermore, we caution > against > - 'zpool upgrade' for the next few weeks. The change should be > transparent > - unless you want to use new features. > - > - Not all "NO_CLEAN" build scenarios work across these changes. Many > - scenarios have been tested and fixed, but rebuilding kernels > without > - rebuilding world may fail. > - > - The ZFS cache file has moved from /boot to /etc to match the > OpenZFS > - upstream default. A fallback to /boot has been added for mountroot. > - > - Pool auto import behavior at boot has been moved from the kernel > module > - to an explicit "zpool import -a" in one of the rc scripts enabled > by > - zfs_enable=YES. This means your non-root zpools won't auto import > until > - you upgrade your /etc/rc.d files. > - > -20200824: > - The resume code now notifies devd with the 'kernel' system > - rather than the old 'kern' subsystem to be consistent with > - other use. The old notification will be created as well, but > - will be removed prior to FreeBSD 14.0. > - > -20200821: > - r362275 changed the internal API between the kernel RPC and the > - NFS modules. As such, all the modules must be recompiled from > - sources. > - > -20200817: > - r364330 modified the internal API used between the NFS modules. > - As such, all the NFS modules must be re-compiled from sources. > - > -20200816: > - Clang, llvm, lld, lldb, compiler-rt, libc++, libunwind and openmp > have > - been upgraded to 11.0.0. Please see the 20141231 entry below for > - information about prerequisites and upgrading, if you are not > already > - using clang 3.5.0 or higher. > - > -20200810: > - r364092 modified the internal ABI used between the kernel NFS > - modules. As such, all of these modules need to be rebuilt > - from sources, so a version bump was done. > - > -20200807: > - Makefile.inc has been updated to work around the issue documented > in > - 20200729. It was a case where the optimization of using symbolic > links > - to point to binaries created a situation where we'd run new > binaries > - with old libraries starting midway through the installworld > process. > - > -20200729: > - r363679 has redefined some undefined behavior in regcomp(3); > notably, > - extraneous escapes of most ordinary characters will no longer be > - accepted. An exp-run has identified all of the problems with this > in > - ports, but other non-ports software may need extra escapes removed > to > - continue to function. > - > - Because of this change, installworld may encounter the following > error > - from rtld: Undefined symbol "regcomp@FBSD_1.6" -- It is > imperative that > - you do not halt installworld. Instead, let it run to completion > (whether > - successful or not) and run installworld once more. > - > -20200627: > - A new implementation of bc and dc has been imported in r362681. > This > - implementation corrects non-conformant behavior of the previous bc > - and adds GNU bc compatible options. It offers a number of > extensions, > - is much faster on large values, and has support for message > catalogs > - (a number of languages are already supported, contributions of > further > - languages welcome). The option WITHOUT_GH_BC can be used to build > the > - world with the previous versions of bc and dc. > - > -20200625: > - r362639 changed the internal API used between the NFS kernel > modules. > - As such, they all need to be rebuilt from sources. > - > -20200613: > - r362158 changed the arguments for VFS_CHECKEXP(). As such, any > - out of tree file systems need to be modified and rebuilt. > - Also, any file systems that are modules must be rebuilt. > - > -20200604: > - read(2) of a directory fd is now rejected by default. root may > - re-enable it for system root only on non-ZFS filesystems with the > - security.bsd.allow_read_dir sysctl(8) MIB if > - security.bsd.suser_enabled=1. > - > - It may be advised to setup aliases for grep to default to `-d > skip` if > - commonly non-recursively grepping a list that includes directories > and > - the potential for the resulting stderr output is not tolerable. > Example > - aliases are now installed, commented out, in /root/.cshrc and > - /root/.shrc. > - > -20200523: > - Clang, llvm, lld, lldb, compiler-rt, libc++, libunwind and openmp > have > - been upgraded to 10.0.1. Please see the 20141231 entry below for > - information about prerequisites and upgrading, if you are not > already > - using clang 3.5.0 or higher. > - > -20200512: > - Support for obsolete compilers has been removed from the build > system. > - Clang 6 and GCC 6.4 are the minimum supported versions. > - > -20200424: > - closefrom(2) has been moved under COMPAT12, and replaced in libc > with a > - stub that calls close_range(2). If using a custom kernel > configuration, > - you may want to ensure that the COMPAT_FREEBSD12 option is > included, as > - a slightly older -CURRENT userland and older FreeBSD userlands may > not > - be functional without closefrom(2). > - > -20200414: > - Upstream DTS from Linux 5.6 was merged and they now have the SID > - and THS (Secure ID controller and THermal Sensor) node present. > - The DTB overlays have now been removed from the tree for the H3/H5 > and > - A64 SoCs and the aw_sid and aw_thermal driver have been updated to > - deal with upstream DTS. If you are using those overlays you need to > - remove them from loader.conf and update the DTBs on the FAT > partition. > - > -20200310: > - Clang, llvm, lld, lldb, compiler-rt, libc++, libunwind and openmp > have > - been upgraded to 10.0.0. Please see the 20141231 entry below for > - information about prerequisites and upgrading, if you are not > already > - using clang 3.5.0 or higher. > - > -20200309: > - The amd(8) automount daemon has been removed from the source tree. > - As of FreeBSD 10.1 autofs(5) is the preferred tool for > automounting. > - amd is still available in the sysutils/am-utils port. > - > -20200301: > - Removed brooktree driver (bktr.4) from the tree. > - > -20200229: > - The WITH_GPL_DTC option has been removed. The BSD-licenced device > tree > - compiler in usr.bin/dtc is used on all architectures which use > dtc, and > - the GPL dtc is available (if needed) from the sysutils/dtc port. > - > -20200229: > - The WITHOUT_LLVM_LIBUNWIND option has been removed. LLVM's > libunwind > - is used by all supported CPU architectures. > - > -20200229: > - GCC 4.2.1 has been removed from the tree. The WITH_GCC, > - WITH_GCC_BOOTSTRAP, and WITH_GNUCXX options are no longer > available. > - Users who wish to build FreeBSD with GCC must use the external > toolchain > - ports or packages. > - > -20200220: > - ncurses has been updated to a newer version (6.2-20200215). Given > the ABI > - has changed, users will have to rebuild all the ports that are > linked to > - ncurses. > - > -20200217: > - The size of struct vnet and the magic cookie have changed. > - Users need to recompile libkvm and all modules using VIMAGE > - together with their new kernel. > - > -20200212: > - Defining the long deprecated NO_CTF, NO_DEBUG_FILES, NO_INSTALLLIB, > - NO_MAN, NO_PROFILE, and NO_WARNS variables is now an error. Update > - your Makefiles and scripts to define MK_<var>=no instead as > required. > - > - One exception to this is that program or library Makefiles should > - define MAN to empty rather than setting MK_MAN=no. > - > -20200108: > - Clang/LLVM is now the default compiler and LLD the default > - linker for riscv64. > - > -20200107: > - make universe no longer uses GCC 4.2.1 on any architectures. > - Architectures not supported by in-tree Clang/LLVM require an > - external toolchain package. > - > -20200104: > - GCC 4.2.1 is now not built by default, as part of the GCC 4.2.1 > - retirement plan. Specifically, the GCC, GCC_BOOTSTRAP, and GNUCXX > - options default to off for all supported CPU architectures. As a > - short-term transition aid they may be enabled via WITH_* options. > - GCC 4.2.1 is expected to be removed from the tree on 2020-03-31. > - > -20200102: > - Support for armv5 has been disconnected and is being removed. The > - machine combination MACHINE=arm MACHINE_ARCH=arm is no longer > valid. > - You must now use a MACHINE_ARCH of armv6 or armv7. The default > - MACHINE_ARCH for MACHINE=arm is now armv7. > - > -20191226: > - Clang/LLVM is now the default compiler for all powerpc > architectures. > - LLD is now the default linker for powerpc64. The change for > powerpc64 > - also includes a change to the ELFv2 ABI, incompatible with the > existing > - ABI. > - > -20191226: > - Kernel-loadable random(4) modules are no longer unloadable. > - > -20191222: > - Clang, llvm, lld, lldb, compiler-rt, libc++, libunwind and openmp > have > - been upgraded to 9.0.1. Please see the 20141231 entry below for > - information about prerequisites and upgrading, if you are not > already > - using clang 3.5.0 or higher. > - > -20191212: > - r355677 has modified the internal interface used between the > - NFS modules in the kernel. As such, they must all be upgraded > - simultaneously. I will do a version bump for this. > - > -20191205: > - The root certificates of the Mozilla CA Certificate Store have been > - imported into the base system and can be managed with the > certctl(8) > - utility. If you have installed the security/ca_root_nss port or > package > - with the ETCSYMLINK option (the default), be advised that there > may be > - differences between those included in the port and those included > in > - base due to differences in nss branch used as well as general > update > - frequency. Note also that certctl(8) cannot manage certs in the > - format used by the security/ca_root_nss port. > - > -20191120: > - The amd(8) automount daemon has been disabled by default, and will > be > - removed in the future. As of FreeBSD 10.1 the autofs(5) is > available > - for automounting. > - > -20191107: > - The nctgpio and wbwd drivers have been moved to the superio bus. > - If you have one of these drivers in a kernel configuration, then > - you should add device superio to it. If you use one of these > drivers > - as a module and you compile a custom set of modules, then you > should > - add superio to the set. > - > -20191021: > - KPIs for network drivers to access interface addresses have > changed. > - Users need to recompile NIC driver modules together with kernel. > - > -20191021: > - The net.link.tap.user_open sysctl no longer prevents user opening > of > - already created /dev/tapNN devices. Access is still controlled by > - node permissions, just like tun devices. The > net.link.tap.user_open > - sysctl is now used only to allow users to perform devfs cloning of > - tap devices, and the subsequent open may not succeed if the user > is not > - in the appropriate group. This sysctl may be deprecated/removed > - completely in the future. > - > -20191009: > - mips, powerpc, and sparc64 are no longer built as part of > - universe / tinderbox unless MAKE_OBSOLETE_GCC is defined. If > - not defined, mips, powerpc, and sparc64 builds will look for > - the xtoolchain binaries and if installed use them for universe > - builds. As llvm 9.0 becomes vetted for these architectures, they > - will be removed from the list. > - > -20191009: > - Clang, llvm, lld, lldb, compiler-rt, libc++, libunwind and openmp > have > - been upgraded to 9.0.0. Please see the 20141231 entry below for > - information about prerequisites and upgrading, if you are not > already > - using clang 3.5.0 or higher. > - > -20191003: > - The hpt27xx, hptmv, hptnr, and hptrr drivers have been removed from > - GENERIC. They are available as modules and can be loaded by adding > - to /boot/loader.conf hpt27xx_load="YES", hptmv_load="YES", > - hptnr_load="YES", or hptrr_load="YES", respectively. > - > -20190913: > - ntpd no longer by default locks its pages in memory, allowing them > - to be paged out by the kernel. Use rlimit memlock to restore > - historic BSD behaviour. For example, add "rlimit memlock 32" > - to ntp.conf to lock up to 32 MB of ntpd address space in memory. > - > -20190823: > - Several of ping6's options have been renamed for better consistency > - with ping. If you use any of -ARWXaghmrtwx, you must update your > - scripts. See ping6(8) for details. > - > -20190727: > - The vfs.fusefs.sync_unmount and vfs.fusefs.init_backgrounded > sysctls > - and the "-o sync_unmount" and "-o init_backgrounded" mount options > have > - been removed from mount_fusefs(8). You can safely remove them from > - your scripts, because they had no effect. > - > - The vfs.fusefs.fix_broken_io, vfs.fusefs.sync_resize, > - vfs.fusefs.refresh_size, vfs.fusefs.mmap_enable, > - vfs.fusefs.reclaim_revoked, and vfs.fusefs.data_cache_invalidate > - sysctls have been removed. If you felt the need to set any of > them to > - a non-default value, please tell asomers@FreeBSD.org why. > - > -20190713: > - Default permissions on the /var/account/acct file (and copies of it > - rotated by periodic daily scripts) are changed from 0644 to 0640 > - because the file contains sensitive information that should not be > - world-readable. If the /var/account directory must be created by > - rc.d/accounting, the mode used is now 0750. Admins who use the > - accounting feature are encouraged to change the mode of an existing > - /var/account directory to 0750 or 0700. > - > -20190620: > - Entropy collection and the /dev/random device are no longer > optional > - components. The "device random" option has been removed. > - Implementations of distilling algorithms can still be made loadable > - with "options RANDOM_LOADABLE" (e.g., random_fortuna.ko). > - > -20190612: > - Clang, llvm, lld, lldb, compiler-rt, libc++, libunwind and openmp > have > - been upgraded to 8.0.1. Please see the 20141231 entry below for > - information about prerequisites and upgrading, if you are not > already > - using clang 3.5.0 or higher. > - > -20190608: > - A fix was applied to i386 kernel modules to avoid panics with > - dpcpu or vnet. Users need to recompile i386 kernel modules > - having pcpu or vnet sections or they will refuse to load. > - > -20190513: > - User-wired pages now have their own counter, > - vm.stats.vm.v_user_wire_count. The vm.max_wired sysctl was renamed > - to vm.max_user_wired and changed from an unsigned int to an > unsigned > - long. bhyve VMs wired with the -S are now subject to the user > - wiring limit; the vm.max_user_wired sysctl may need to be tuned to > - avoid running into the limit. > - > -20190507: > - The IPSEC option has been removed from GENERIC. Users requiring > - ipsec(4) must now load the ipsec(4) kernel module. > - > -20190507: > - The tap(4) driver has been folded into tun(4), and the module has > been > - renamed to tuntap. You should update any kld_list="if_tap" or > - kld_list="if_tun" entries in /etc/rc.conf, if_tap_load="YES" or > - if_tun_load="YES" entries in /boot/loader.conf to load the > if_tuntap > - module instead, and "device tap" or "device tun" entries in kernel > - config files to select the tuntap device instead. > - > -20190418: > - The following knobs have been added related to tradeoffs between > - safe use of the random device and availability in the absence of > - entropy: > - > - kern.random.initial_seeding.bypass_before_seeding: tunable; set > - non-zero to bypass the random device prior to seeding, or zero to > - block random requests until the random device is initially seeded. > - For now, set to 1 (unsafe) by default to restore pre-r346250 boot > - availability properties. > - > - kern.random.initial_seeding.read_random_bypassed_before_seeding: > - read-only diagnostic sysctl that is set when bypass is enabled and > - read_random(9) is bypassed, to enable programmatic handling of this > - initial condition, if desired. > - > - kern.random.initial_seeding.arc4random_bypassed_before_seeding: > - Similar to the above, but for arc4random(9) initial seeding. > - > - kern.random.initial_seeding.disable_bypass_warnings: tunable; set > - non-zero to disable warnings in dmesg when the same conditions are > - met as for the diagnostic sysctls above. Defaults to zero, i.e., > - produce warnings in dmesg when the conditions are met. > - > -20190416: > - The loadable random module KPI has changed; the random_infra_init() > - routine now requires a 3rd function pointer for a bool (*)(void) > - method that returns true if the random device is seeded (and > - therefore unblocked). > - > -20190404: > - r345895 reverts r320698. This implies that an nfsuserd(8) daemon > - built from head sources between r320757 (July 6, 2017) and > - r338192 (Aug. 22, 2018) will not work unless the "-use-udpsock" > - is added to the command line. > - nfsuserd daemons built from head sources that are post-r338192 are > - not affected and should continue to work. > - > -20190320: > - The fuse(4) module has been renamed to fusefs(4) for consistency > with > - other filesystems. You should update any kld_load="fuse" entries > in > - /etc/rc.conf, fuse_load="YES" entries in /boot/loader.conf, and > - "options FUSE" entries in kernel config files. > - > -20190304: > - Clang, llvm, lld, lldb, compiler-rt and libc++ have been upgraded > to > - 8.0.0. Please see the 20141231 entry below for information about > - prerequisites and upgrading, if you are not already using clang > 3.5.0 > - or higher. > - > -20190226: > - geom_uzip(4) depends on the new module xz. If geom_uzip is > statically > - compiled into your custom kernel, add 'device xz' statement to the > - kernel config. > - > -20190219: > - drm and drm2 have been removed from the tree. Please see > - https://wiki.freebsd.org/Graphics for the latest information on > - migrating to the drm ports. > - > -20190131: > - Iflib is no longer unconditionally compiled into the kernel. > Drivers > - using iflib and statically compiled into the kernel, now require > - the 'device iflib' config option. For the same drivers loaded as > - modules on kernels not having 'device iflib', the iflib.ko module > - is loaded automatically. > - > -20190125: > - The IEEE80211_AMPDU_AGE and AH_SUPPORT_AR5416 kernel configuration > - options no longer exist since r343219 and r343427 respectively; > - nothing uses them, so they should be just removed from custom > - kernel config files. > - > -20181230: > - r342635 changes the way efibootmgr(8) works by requiring users to > add > - the -b (bootnum) parameter for commands where the bootnum was > previously > - specified with each option. For example 'efibootmgr -B 0001' is now > - 'efibootmgr -B -b 0001'. > - > -20181220: > - r342286 modifies the NFSv4 server so that it obeys > vfs.nfsd.nfs_privport > - in the same as it is applied to NFSv2 and 3. This implies that > NFSv4 > - servers that have vfs.nfsd.nfs_privport set will only allow mounts > - from clients using a reserved port. Since both the FreeBSD and > Linux > - NFSv4 clients use reserved ports by default, this should not affect > - most NFSv4 mounts. > - > -20181219: > - The XLP config has been removed. We can't support 64-bit atomics > in this > - kernel because it is running in 32-bit mode. XLP users must > transition > - to running a 64-bit kernel (XLP64 or XLPN32). > - > - The mips GXEMUL support has been removed from FreeBSD. MALTA* + > qemu is > - the preferred emulator today and we don't need two different ones. > - > - The old sibyte / swarm / Broadcom BCM1250 support has been > - removed from the mips port. > - > -20181211: > - Clang, llvm, lld, lldb, compiler-rt and libc++ have been upgraded > to > - 7.0.1. Please see the 20141231 entry below for information about > - prerequisites and upgrading, if you are not already using clang > 3.5.0 > - or higher. > - > -20181211: > - Remove the timed and netdate programs from the base tree. Setting > - the time with these daemons has been obsolete for over a decade. > - > -20181126: > - On amd64, arm64 and armv7 (architectures that install LLVM's ld.lld > - linker as /usr/bin/ld) GNU ld is no longer installed as ld.bfd, as > - it produces broken binaries when ifuncs are in use. Users needing > - GNU ld should install the binutils port or package. > - > -20181123: > - The BSD crtbegin and crtend code has been enabled by default. It > has > - had extensive testing on amd64, arm64, and i386. It can be disabled > - by building a world with -DWITHOUT_BSD_CRTBEGIN. > - > -20181115: > - The set of CTM commands (ctm, ctm_smail, ctm_rmail, ctm_dequeue) > - has been converted to a port (misc/ctm) and will be removed from > - FreeBSD-13. It is available as a package (ctm) for all supported > - FreeBSD versions. > - > -20181110: > - The default newsyslog.conf(5) file has been changed to only include > - files in /etc/newsyslog.conf.d/ and > /usr/local/etc/newsyslog.conf.d/ if > - the filenames end in '.conf' and do not begin with a '.'. > - > - You should check the configuration files in these two directories > match > - this naming convention. You can verify which configuration files > are > - being included using the command: > - $ newsyslog -Nrv > - > -20181015: > - Ports for the DRM modules have been simplified. Now, amd64 users > should > - just install the drm-kmod port. All others should install > - drm-legacy-kmod. > - > - Graphics hardware that's newer than about 2010 usually works with > - drm-kmod. For hardware older than 2013, however, some users will > need > - to use drm-legacy-kmod if drm-kmod doesn't work for them. Hardware > older > - than 2008 usually only works in drm-legacy-kmod. The graphics team > can > - only commit to hardware made since 2013 due to the complexity of > the > - market and difficulty to test all the older cards effectively. If > you > - have hardware supported by drm-kmod, you are strongly encouraged > to use > - that as you will get better support. > - > - Other than KPI chasing, drm-legacy-kmod will not be updated. As > outlined > - elsewhere, the drm and drm2 modules will be eliminated from the > src base > - soon (with a limited exception for arm). Please update to the > package > - asap and report any issues to x11@freebsd.org. > - > - Generally, anybody using the drm*-kmod packages should add > - WITHOUT_DRM_MODULE=t and WITHOUT_DRM2_MODULE=t to avoid nasty > - cross-threading surprises, especially with automatic driver > - loading from X11 startup. These will become the defaults in > 13-current > - shortly. > - > -20181012: > - The ixlv(4) driver has been renamed to iavf(4). As a consequence, > - custom kernel and module loading configuration files must be > updated > - accordingly. Moreover, interfaces previous presented as ixlvN to > the > - system are now exposed as iavfN and network configuration files > must > - be adjusted as necessary. > - > -20181009: > - OpenSSL has been updated to version 1.1.1. This update included > - additional various API changes throughout the base system. It is > - important to rebuild third-party software after upgrading. The > value > - of __FreeBSD_version has been bumped accordingly. > - > -20181006: > - The legacy DRM modules and drivers have now been added to the > loader's > - module blacklist, in favor of loading them with kld_list in > rc.conf(5). > - The module blacklist may be overridden with the loader.conf(5) > - 'module_blacklist' variable, but loading them via rc.conf(5) is > strongly > - encouraged. > - > -20181002: > - The cam(4) based nda(4) driver will be used over nvd(4) by default > on > - powerpc64. You may set 'options NVME_USE_NVD=1' in your kernel > conf or > - loader tunable 'hw.nvme.use_nvd=1' if you wish to use the existing > - driver. Make sure to edit /boot/etc/kboot.conf and fstab to use > the > - nda device name. > - > -20180913: > - Reproducible build mode is now on by default, in preparation for > - FreeBSD 12.0. This eliminates build metadata such as the user, > - host, and time from the kernel (and uname), unless the working tree > - corresponds to a modified checkout from a version control system. > - The previous behavior can be obtained by setting the /etc/src.conf > - knob WITHOUT_REPRODUCIBLE_BUILD. > - > -20180826: > - The Yarrow CSPRNG has been removed from the kernel as it has not > been > - supported by its designers since at least 2003. Fortuna has been > the > - default since FreeBSD-11. > - > -20180822: > - devctl freeze/thaw have gone into the tree, the rc scripts have > been > - updated to use them and devmatch has been changed. You should > update > - kernel, userland and rc scripts all at the same time. > - > -20180818: > - The default interpreter has been switched from 4th to Lua. > - LOADER_DEFAULT_INTERP, documented in build(7), will override the > default > - interpreter. If you have custom FORTH code you will need to set > - LOADER_DEFAULT_INTERP=4th (valid values are 4th, lua or simp) in > - src.conf for the build. This will create default hard links > between > - loader and loader_4th instead of loader and loader_lua, the new > default. > - If you are using UEFI it will create the proper hard link to > loader.efi. > - > - bhyve uses userboot.so. It remains 4th-only until some issues are > solved > - regarding coexisting with multiple versions of FreeBSD are > resolved. > - > -20180815: > - ls(1) now respects the COLORTERM environment variable used in other > - systems and software to indicate that a colored terminal is both > - supported and desired. If ls(1) is suddenly emitting colors, they > may > - be disabled again by either removing the unwanted COLORTERM from > your > - environment, or using `ls --color=never`. The ls(1) specific > CLICOLOR > - may not be observed in a future release. > - > -20180808: > - The default pager for most commands has been changed to "less". To > - restore the old behavior, set PAGER="more" and MANPAGER="more -s" > in > - your environment. > - > -20180731: > - The jedec_ts(4) driver has been removed. A superset of its > functionality > - is available in the jedec_dimm(4) driver, and the manpage for that > - driver includes migration instructions. If you have "device > jedec_ts" > - in your kernel configuration file, it must be removed. > - > -20180730: > - amd64/GENERIC now has EFI runtime services, EFIRT, enabled by > default. > - This should have no effect if the kernel is booted via BIOS/legacy > boot. > - EFIRT may be disabled via a loader tunable, efi.rt.disabled, if a > system > - has a buggy firmware that prevents a successful boot due to use of > - runtime services. > - > -20180727: > - Atmel AT91RM9200 and AT91SAM9, Cavium CNS 11xx and XScale > - support has been removed from the tree. These ports were > - obsolete and/or known to be broken for many years. > - > -20180723: > - loader.efi has been augmented to participate more fully in the > - UEFI boot manager protocol. loader.efi will now look at the > - BootXXXX environment variable to determine if a specific kernel > - or root partition was specified. XXXX is derived from BootCurrent. > - efibootmgr(8) manages these standard UEFI variables. > - > -20180720: > - zfsloader's functionality has now been folded into loader. > - zfsloader is no longer necessary once you've updated your > - boot blocks. For a transition period, we will install a > - hardlink for zfsloader to loader to allow a smooth transition > - until the boot blocks can be updated (hard link because old > - zfs boot blocks don't understand symlinks). > - > -20180719: > - ARM64 now have efifb support, if you want to have serial console > - on your arm64 board when an screen is connected and the bootloader > - setup a frame buffer for us to use, just add : > - boot_serial=YES > - boot_multicons=YES > - in /boot/loader.conf > - For Raspberry Pi 3 (RPI) users, this is needed even if you don't > have > - an screen connected as the firmware will setup a frame buffer are > that > - u-boot will expose as an EFI frame buffer. > - > -20180719: > - New uid:gid added, ntpd:ntpd (123:123). Be sure to run mergemaster > - or take steps to update /etc/passwd before doing installworld on > - existing systems. Do not skip the "mergemaster -Fp" step before > - installworld, as described in the update procedures near the bottom > - of this document. Also, rc.d/ntpd now starts ntpd(8) as user ntpd > - if the new mac_ntpd(4) policy is available, unless ntpd_flags or > - the ntp config file contain options that change file/dir locations. > - When such options (e.g., "statsdir" or "crypto") are used, ntpd can > - still be run as non-root by setting ntpd_user=ntpd in rc.conf, > after > - taking steps to ensure that all required files/dirs are accessible > - by the ntpd user. > - > -20180717: > - Big endian arm support has been removed. > - > -20180711: > - The static environment setup in kernel configs is no longer > mutually > - exclusive with the loader(8) environment by default. In order to > - restore the previous default behavior of disabling the loader(8) > - environment if a static environment is present, you must specify > - loader_env.disabled=1 in the static environment. > - > -20180705: > - The ABI of syscalls used by management tools like sockstat and > - netstat has been broken to allow 32-bit binaries to work on > - 64-bit kernels without modification. These programs will need > - to match the kernel in order to function. External programs may > - require minor modifications to accommodate a change of type in > - structures from pointers to 64-bit virtual addresses. > - > -20180702: > - On i386 and amd64 atomics are now inlined. Out of tree modules > using > - atomics will need to be rebuilt. > - > -20180701: > - The '%I' format in the kern.corefile sysctl limits the number of > - core files that a process can generate to the number stored in the > - debug.ncores sysctl. The '%I' format is replaced by the single > digit > - index. Previously, if all indexes were taken the kernel would > overwrite > - only a core file with the highest index in a filename. > - Currently the system will create a new core file if there is a free > - index or if all slots are taken it will overwrite the oldest one. > - > -20180630: > - Clang, llvm, lld, lldb, compiler-rt and libc++ have been upgraded > to > - 6.0.1. Please see the 20141231 entry below for information about > - prerequisites and upgrading, if you are not already using clang > 3.5.0 > - or higher. > - > -20180628: > - r335753 introduced a new quoting method. However, > etc/devd/devmatch.conf > - needed to be changed to work with it. This change was made with > r335763 > - and requires a mergemaster / etcupdate / etc to update the > installed > - file. > - > -20180612: > - r334930 changed the interface between the NFS modules, so they all > - need to be rebuilt. r335018 did a __FreeBSD_version bump for this. > - > -20180530: > - As of r334391 lld is the default amd64 system linker; it is > installed > - as /usr/bin/ld. Kernel build workarounds (see 20180510 entry) are > no > - longer necessary. > - > -20180530: > - The kernel / userland interface for devinfo changed, so you'll > - need a new kernel and userland as a pair for it to work (rebuilding > - lib/libdevinfo is all that's required). devinfo and devmatch will > - not work, but everything else will when there's a mismatch. > - > -20180523: > - The on-disk format for hwpmc callchain records has changed to > include > - threadid corresponding to a given record. This changes the field > offsets > - and thus requires that libpmcstat be rebuilt before using a kernel > - later than r334108. > - > -20180517: > - The vxge(4) driver has been removed. This driver was introduced > into > - HEAD one week before the Exar left the Ethernet market and is not > - known to be used. If you have device vxge in your kernel config > file > - it must be removed. > - > -20180510: > - The amd64 kernel now requires a ld that supports ifunc to produce a > - working kernel, either lld or a newer binutils. lld is built by > default > - on amd64, and the 'buildkernel' target uses it automatically. > However, > - it is not the default linker, so building the kernel the > traditional > - way requires LD=ld.lld on the command line (or > LD=/usr/local/bin/ld for > - binutils port/package). lld will soon be default, and this > requirement > - will go away. > - > - NOTE: As of r334391 lld is the default system linker on amd64, and > no > - workaround is necessary. > - > -20180508: > - The nxge(4) driver has been removed. This driver was for PCI-X 10g > - cards made by s2io/Neterion. The company was acquired by Exar and > - no longer sells or supports Ethernet products. If you have device > - nxge in your kernel config file it must be removed. > - > -20180504: > - The tz database (tzdb) has been updated to 2018e. This version > more > - correctly models time stamps in time zones with negative DST such > as > - Europe/Dublin (from 1971 on), Europe/Prague (1946/7), and > - Africa/Windhoek (1994/2017). This does not affect the UT offsets, > only > - time zone abbreviations and the tm_isdst flag. > - > -20180502: > - The ixgb(4) driver has been removed. This driver was for an early > and > - uncommon legacy PCI 10GbE for a single ASIC, Intel 82597EX. Intel > - quickly shifted to the long lived ixgbe family. If you have device > - ixgb in your kernel config file it must be removed. > - > -20180501: > - The lmc(4) driver has been removed. This was a WAN interface > - card that was already reportedly rare in 2003, and had an ambiguous > - license. If you have device lmc in your kernel config file it must > - be removed. > - > -20180413: > - Support for Arcnet networks has been removed. If you have device > - arcnet or device cm in your kernel config file they must be > - removed. > - > -20180411: > - Support for FDDI networks has been removed. If you have device > - fddi or device fpa in your kernel config file they must be > - removed. > - > -20180406: > - In addition to supporting RFC 3164 formatted messages, the > - syslogd(8) service is now capable of parsing RFC 5424 formatted > - log messages. The main benefit of using RFC 5424 is that clients > - may now send log messages with timestamps containing year numbers, > - microseconds and time zone offsets. > - > - Similarly, the syslog(3) C library function has been altered to > - send RFC 5424 formatted messages to the local system logging > - daemon. On systems using syslogd(8), this change should have no > - negative impact, as long as syslogd(8) and the C library are > - updated at the same time. On systems using a different system > - logging daemon, it may be necessary to make configuration > - adjustments, depending on the software used. > - > - When using syslog-ng, add the 'syslog-protocol' flag to local > - input sources to enable parsing of RFC 5424 formatted messages: > - > - source src { > - unix-dgram("/var/run/log" flags(syslog-protocol)); > - } > - > - When using rsyslog, disable the 'SysSock.UseSpecialParser' option > - of the 'imuxsock' module to let messages be processed by the > - regular RFC 3164/5424 parsing pipeline: > - > - module(load="imuxsock" SysSock.UseSpecialParser="off") > - > - Do note that these changes only affect communication between local > - applications and syslogd(8). The format that syslogd(8) uses to > - store messages on disk or forward messages to other systems > - remains unchanged. syslogd(8) still uses RFC 3164 for these > - purposes. Options to customize this behaviour will be added in the > - future. Utilities that process log files stored in /var/log are > - thus expected to continue to function as before. > - > - __FreeBSD_version has been incremented to 1200061 to denote this > - change. > - > -20180328: > - Support for token ring networks has been removed. If you > - have "device token" in your kernel config you should remove > - it. No device drivers supported token ring. > - > -20180323: > - makefs was modified to be able to tag ISO9660 El Torito boot > catalog > - entries as EFI instead of overloading the i386 tag as done > previously. > - The amd64 mkisoimages.sh script used to build amd64 ISO images for > - release was updated to use this. This may mean that makefs must be > - updated before "make cdrom" can be run in the release directory. > This > - should be as simple as: > - > - $ cd $SRCDIR/usr.sbin/makefs > - $ make depend all install > - > -20180212: > - FreeBSD boot loader enhanced with Lua scripting. It's purely > opt-in for > - now by building WITH_LOADER_LUA and WITHOUT_FORTH in /etc/src.conf. > - Co-existence for the transition period will come shortly. Booting > is a > - complex environment and test coverage for Lua-enabled loaders has > been > - thin, so it would be prudent to assume it might not work and make > - provisions for backup boot methods. > - > -20180211: > - devmatch functionality has been turned on in devd. It will > automatically > - load drivers for unattached devices. This may cause unexpected > drivers > - to be loaded. Please report any problems to current@ and > - imp@freebsd.org. > - > -20180114: > - Clang, llvm, lld, lldb, compiler-rt and libc++ have been upgraded > to > - 6.0.0. Please see the 20141231 entry below for information about > - prerequisites and upgrading, if you are not already using clang > 3.5.0 > - or higher. > - > -20180110: > - LLVM's lld linker is now used as the FreeBSD/amd64 bootstrap > linker. > - This means it is used to link the kernel and userland libraries and > - executables, but is not yet installed as /usr/bin/ld by default. > - > - To revert to ld.bfd as the bootstrap linker, in /etc/src.conf set > - WITHOUT_LLD_BOOTSTRAP=yes > - > -20180110: > - On i386, pmtimer has been removed. Its functionality has been > folded > - into apm. It was a no-op on ACPI in current for a while now (but > was > - still needed on i386 in FreeBSD 11 and earlier). Users may need to > - remove it from kernel config files. > - > -20180104: > - The use of RSS hash from the network card aka flowid has been > - disabled by default for lagg(4) as it's currently incompatible with > - the lacp and loadbalance protocols. > - > - This can be re-enabled by setting the following in loader.conf: > - net.link.lagg.default_use_flowid="1" > - > -20180102: > - The SW_WATCHDOG option is no longer necessary to enable the > - hardclock-based software watchdog if no hardware watchdog is > - configured. As before, SW_WATCHDOG will cause the software > - watchdog to be enabled even if a hardware watchdog is configured. > - > -20171215: > - r326887 fixes the issue described in the 20171214 UPDATING entry. > - r326888 flips the switch back to building GELI support always. > - > -20171214: > - r362593 broke ZFS + GELI support for reasons unknown. However, > - it also broke ZFS support generally, so GELI has been turned off > - by default as the lesser evil in r326857. If you boot off ZFS > and/or > - GELI, it might not be a good time to update. > - > -20171125: > - PowerPC users must update loader(8) by rebuilding world before > - installing a new kernel, as the protocol connecting them has > - changed. Without the update, loader metadata will not be passed > - successfully to the kernel and users will have to enter their > - root partition at the kernel mountroot prompt to continue booting. > - Newer versions of loader can boot old kernels without issue. > - > -20171110: > - The LOADER_FIREWIRE_SUPPORT build variable has been renamed to > - WITH/OUT_LOADER_FIREWIRE. LOADER_{NO_,}GELI_SUPPORT has been > renamed > - to WITH/OUT_LOADER_GELI. > - > -20171106: > - The naive and non-compliant support of posix_fallocate(2) in ZFS > - has been removed as of r325320. The system call now returns EINVAL > - when used on a ZFS file. Although the new behavior complies with > the > - standard, some consumers are not prepared to cope with it. > - One known victim is lld prior to r325420. > - > -20171102: > - Building in a FreeBSD src checkout will automatically create object > - directories now rather than store files in the current directory if > - 'make obj' was not ran. Calling 'make obj' is no longer necessary. > - This feature can be disabled by setting WITHOUT_AUTO_OBJ=yes in > - /etc/src-env.conf (not /etc/src.conf), or passing the option in the > - environment. > - > -20171101: > - The default MAKEOBJDIR has changed from /usr/obj/<srcdir> for > native > - builds, and /usr/obj/<arch>/<srcdir> for cross-builds, to a unified > - /usr/obj/<srcdir>/<arch>. This behavior can be changed to the old > - format by setting WITHOUT_UNIFIED_OBJDIR=yes in /etc/src-env.conf, > - the environment, or with -DWITHOUT_UNIFIED_OBJDIR when building. > - The UNIFIED_OBJDIR option is a transitional feature that will be > - removed for 12.0 release; please migrate to the new format for any > - tools by looking up the OBJDIR used by 'make -V .OBJDIR' means > rather > - than hardcoding paths. > - > -20171028: > - The native-xtools target no longer installs the files by default > to the > - OBJDIR. Use the native-xtools-install target with a DESTDIR to > install > - to ${DESTDIR}/${NXTP} where NXTP defaults to /nxb-bin. > *** 1218 LINES SKIPPED *** > --00000000000021a3de05ee4e325b Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable <div dir=3D"auto">Please revert this. We keep older updating entries on pur= pose. You purged way too much. Let's chat about how much to remove in a= rch@. They are for more than just source updates, so your reasoning is wron= g. They are also there for users updating their products which can have a l= arger leap in time. We've traditionally kept closer to 5-10 years here = for that reason.<div dir=3D"auto"><br></div><div dir=3D"auto">Warner<br><di= v dir=3D"auto"><br></div><div dir=3D"auto">Warner=C2=A0</div></div></div><b= r><div class=3D"gmail_quote"><div dir=3D"ltr" class=3D"gmail_attr">On Fri, = Nov 25, 2022, 2:23 AM Alexander Leidinger <<a href=3D"mailto:netchild@fr= eebsd.org">netchild@freebsd.org</a>> wrote:<br></div><blockquote class= =3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padd= ing-left:1ex">The branch main has been updated by netchild:<br> <br> URL: <a href=3D"https://cgit.FreeBSD.org/src/commit/?id=3Dff0c7816db696d31a= dc437134dcad45a70ad5889" rel=3D"noreferrer noreferrer" target=3D"_blank">ht= tps://cgit.FreeBSD.org/src/commit/?id=3Dff0c7816db696d31adc437134dcad45a70a= d5889</a><br> <br> commit ff0c7816db696d31adc437134dcad45a70ad5889<br> Author:=C2=A0 =C2=A0 =C2=A0Alexander Leidinger <netchild@FreeBSD.org>= <br> AuthorDate: 2022-11-25 09:17:14 +0000<br> Commit:=C2=A0 =C2=A0 =C2=A0Alexander Leidinger <netchild@FreeBSD.org>= <br> CommitDate: 2022-11-25 09:17:14 +0000<br> <br> =C2=A0 =C2=A0 Remove UPDATING entries from old branches.<br> <br> =C2=A0 =C2=A0 We only support updates from major version N to N+1:<br> =C2=A0 =C2=A0 stable/13 was branched on 20210122, remove all old entries fr= om stable/10<br> =C2=A0 =C2=A0 branch point in 2013 to 20210122.<br> ---<br> =C2=A0UPDATING | 2175 -----------------------------------------------------= ---------<br> =C2=A01 file changed, 2175 deletions(-)<br> <br> diff --git a/UPDATING b/UPDATING<br> index 3782692c2a49..83e97246b964 100644<br> --- a/UPDATING<br> +++ b/UPDATING<br> @@ -275,2181 +275,6 @@ <a href=3D"https://lists.freebsd.org/archives/freebs= d-current/2021-June/000153.html" rel=3D"noreferrer noreferrer" target=3D"_b= lank">https://lists.freebsd.org/archives/freebsd-current/2021-June/000153.h= tml</a><br> =C2=A0 =C2=A0 =C2=A0 =C2=A0 Please update your drm-kmod port to after the _= _FreeBSD_version 1400003<br> =C2=A0 =C2=A0 =C2=A0 =C2=A0 update.<br> <br> -20210108:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0PC Card attachments for all devices have been r= emoved. In the case of<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0wi and cmx, the entire drivers were removed bec= ause they were only<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0PC Card devices. FreeBSD_version 1300134 should= be used for this<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0since it was bumped so recently.<br> -<br> -20210107:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Transport-independent parts of HID support have= been split off the USB<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0code in to separate subsystem.=C2=A0 Kernel con= figs which include one of<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0ums, ukbd, uhid, atp, wsp, wmt, uaudio, ugold o= r ucycom drivers should<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0be updated with adding of "device hid"= ; line.<br> -<br> -20210105:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0ncurses installation has been modified to only = keep the widechar<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0enabled version.=C2=A0 Incremental build is bro= ken for that change, so it<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0requires a clean build.<br> -<br> -20201223:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The FreeBSD project has migrated from Subversio= n to Git. Temporary<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0instructions can be found at<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0<a href=3D"https://github.com/bsdimp/freebsd-gi= t-docs/blob/main/src-cvt.md" rel=3D"noreferrer noreferrer" target=3D"_blank= ">https://github.com/bsdimp/freebsd-git-docs/blob/main/src-cvt.md</a><br>; -=C2=A0 =C2=A0 =C2=A0 =C2=A0and other documents in that repo.<br> -<br> -20201216:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The services database has been updated to cover= more of the basic<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0services expected in a modern system. The datab= ase is big enough<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0that it will cause issues in mergemaster in Rel= eases previous to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A012.2 and 11.3, or in very old current systems f= rom before r358154.<br> -<br> -20201215:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Obsolete in-tree GDB 6.1.1 has been removed.=C2= =A0 GDB (including kgdb)<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0may be installed from ports or packages.<br> -<br> -20201124:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0ping6 has been merged into ping.=C2=A0 It can n= ow be called as "ping -6".<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0See ping(8) for details.<br> -<br> -20201108:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Default value of net.add_addr_allfibs has been = changed to 0.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0If you have multi-fib configuration and rely on= existence of all<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0interface routes in every fib, you need to set = the above sysctl to 1.<br> -<br> -20201030:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The internal pre-processor in the calendar(1) p= rogram has been<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0extended to support more C pre-processor comman= ds (e.g. #ifdef, #else,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0and #undef) and to detect unbalanced conditiona= l statements.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Error messages have been extended to include th= e filename and line<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0number if processing stops to help fixing malfo= rmed data files.<br> -<br> -20201026:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0All the data files for the calendar(1) program,= except calendar.freebsd,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0have been moved to the deskutils/calendar-data = port, much like the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0jewish calendar entries were moved to deskutils= /hebcal years ago. After<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0make delete-old-files, you need to install it t= o retain full<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0functionality. calendar(1) will issue a reminde= r for files it can't<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0find.<br> -<br> -20200923:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0LINT files are no longer generated. We now incl= ude the relevant NOTES<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0files. Note: This may cause conflicts with upda= ting in some cases.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0find sys -name LINT= \* -delete<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0is suggested across this commit to remove the g= enerated LINT files.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0If you have tried to update with generated file= s there, the svn<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0command you want to un-auger the tree is<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0cd sys/amd64/conf<b= r> -=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0svn revert -R .<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0and then do the above find from the top level. = Substitute 'amd64'<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0above with where the error message indicates a = conflict.<br> -<br> -20200824:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0OpenZFS support has been integrated. Do not upg= rade root pools until<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0the loader is updated to support zstd. Furtherm= ore, we caution against<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0'zpool upgrade' for the next few weeks.= The change should be transparent<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0unless you=C2=A0 want to use new features.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Not all "NO_CLEAN" build scenarios wo= rk across these changes. Many<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0scenarios have been tested and fixed, but rebui= lding kernels without<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0rebuilding world may fail.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The ZFS cache file has moved from /boot to /etc= to match the OpenZFS<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0upstream default. A fallback to /boot has been = added for mountroot.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Pool auto import behavior at boot has been move= d from the kernel module<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0to an explicit "zpool import -a" in o= ne of the rc scripts enabled by<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0zfs_enable=3DYES. This means your non-root zpoo= ls won't auto import until<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0you upgrade your /etc/rc.d files.<br> -<br> -20200824:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The resume code now notifies devd with the '= ;kernel' system<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0rather than the old 'kern' subsystem to= be consistent with<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0other use. The old notification will be created= as well, but<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0will be removed prior to FreeBSD 14.0.<br> -<br> -20200821:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r362275 changed the internal API between the ke= rnel RPC and the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0NFS modules. As such, all the modules must be r= ecompiled from<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0sources.<br> -<br> -20200817:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r364330 modified the internal API used between = the NFS modules.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0As such, all the NFS modules must be re-compile= d from sources.<br> -<br> -20200816:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Clang, llvm, lld, lldb, compiler-rt, libc++, li= bunwind and openmp have<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0been upgraded to 11.0.0.=C2=A0 Please see the 2= 0141231 entry below for<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0information about prerequisites and upgrading, = if you are not already<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0using clang 3.5.0 or higher.<br> -<br> -20200810:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r364092 modified the internal ABI used between = the kernel NFS<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0modules.=C2=A0 As such, all of these modules ne= ed to be rebuilt<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0from sources, so a version bump was done.<br> -<br> -20200807:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Makefile.inc has been updated to work around th= e issue documented in<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A020200729. It was a case where the optimization = of using symbolic links<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0to point to binaries created a situation where = we'd run new binaries<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0with old libraries starting midway through the = installworld process.<br> -<br> -20200729:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r363679 has redefined some undefined behavior i= n regcomp(3); notably,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0extraneous escapes of most ordinary characters = will no longer be<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0accepted.=C2=A0 An exp-run has identified all o= f the problems with this in<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0ports, but other non-ports software may need ex= tra escapes removed to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0continue to function.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Because of this change, installworld may encoun= ter the following error<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0from rtld: Undefined symbol "regcomp@FBSD_= 1.6" -- It is imperative that<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0you do not halt installworld. Instead, let it r= un to completion (whether<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0successful or not) and run installworld once mo= re.<br> -<br> -20200627:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0A new implementation of bc and dc has been impo= rted in r362681. This<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0implementation corrects non-conformant behavior= of the previous bc<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0and adds GNU bc compatible options. It offers a= number of extensions,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0is much faster on large values, and has support= for message catalogs<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0(a number of languages are already supported, c= ontributions of further<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0languages welcome). The option WITHOUT_GH_BC ca= n be used to build the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0world with the previous versions of bc and dc.<= br> -<br> -20200625:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r362639 changed the internal API used between t= he NFS kernel modules.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0As such, they all need to be rebuilt from sourc= es.<br> -<br> -20200613:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r362158 changed the arguments for VFS_CHECKEXP(= ).=C2=A0 As such, any<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0out of tree file systems need to be modified an= d rebuilt.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Also, any file systems that are modules must be= rebuilt.<br> -<br> -20200604:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0read(2) of a directory fd is now rejected by de= fault.=C2=A0 root may<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0re-enable it for system root only on non-ZFS fi= lesystems with the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0security.bsd.allow_read_dir sysctl(8) MIB if<br= > -=C2=A0 =C2=A0 =C2=A0 =C2=A0security.bsd.suser_enabled=3D1.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0It may be advised to setup aliases for grep to = default to `-d skip` if<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0commonly non-recursively grepping a list that i= ncludes directories and<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0the potential for the resulting stderr output i= s not tolerable.=C2=A0 Example<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0aliases are now installed, commented out, in /r= oot/.cshrc and<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0/root/.shrc.<br> -<br> -20200523:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Clang, llvm, lld, lldb, compiler-rt, libc++, li= bunwind and openmp have<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0been upgraded to 10.0.1.=C2=A0 Please see the 2= 0141231 entry below for<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0information about prerequisites and upgrading, = if you are not already<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0using clang 3.5.0 or higher.<br> -<br> -20200512:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Support for obsolete compilers has been removed= from the build system.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Clang 6 and GCC 6.4 are the minimum supported v= ersions.<br> -<br> -20200424:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0closefrom(2) has been moved under COMPAT12, and= replaced in libc with a<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0stub that calls close_range(2).=C2=A0 If using = a custom kernel configuration,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0you may want to ensure that the COMPAT_FREEBSD1= 2 option is included, as<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0a slightly older -CURRENT userland and older Fr= eeBSD userlands may not<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0be functional without closefrom(2).<br> -<br> -20200414:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Upstream DTS from Linux 5.6 was merged and they= now have the SID<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0and THS (Secure ID controller and THermal Senso= r) node present.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The DTB overlays have now been removed from the= tree for the H3/H5 and<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0A64 SoCs and the aw_sid and aw_thermal driver h= ave been updated to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0deal with upstream DTS. If you are using those = overlays you need to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0remove them from loader.conf and update the DTB= s on the FAT partition.<br> -<br> -20200310:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Clang, llvm, lld, lldb, compiler-rt, libc++, li= bunwind and openmp have<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0been upgraded to 10.0.0.=C2=A0 Please see the 2= 0141231 entry below for<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0information about prerequisites and upgrading, = if you are not already<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0using clang 3.5.0 or higher.<br> -<br> -20200309:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The amd(8) automount daemon has been removed fr= om the source tree.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0As of FreeBSD 10.1 autofs(5) is the preferred t= ool for automounting.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0amd is still available in the sysutils/am-utils= port.<br> -<br> -20200301:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Removed brooktree driver (bktr.4) from the tree= .<br> -<br> -20200229:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The WITH_GPL_DTC option has been removed.=C2=A0= The BSD-licenced device tree<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0compiler in usr.bin/dtc is used on all architec= tures which use dtc, and<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0the GPL dtc is available (if needed) from the s= ysutils/dtc port.<br> -<br> -20200229:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The WITHOUT_LLVM_LIBUNWIND option has been remo= ved.=C2=A0 LLVM's libunwind<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0is used by all supported CPU architectures.<br> -<br> -20200229:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0GCC 4.2.1 has been removed from the tree.=C2=A0= The WITH_GCC,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0WITH_GCC_BOOTSTRAP, and WITH_GNUCXX options are= no longer available.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Users who wish to build FreeBSD with GCC must u= se the external toolchain<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0ports or packages.<br> -<br> -20200220:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0ncurses has been updated to a newer version (6.= 2-20200215). Given the ABI<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0has changed, users will have to rebuild all the= ports that are linked to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0ncurses.<br> -<br> -20200217:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The size of struct vnet and the magic cookie ha= ve changed.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Users need to recompile libkvm and all modules = using VIMAGE<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0together with their new kernel.<br> -<br> -20200212:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Defining the long deprecated NO_CTF, NO_DEBUG_F= ILES, NO_INSTALLLIB,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0NO_MAN, NO_PROFILE, and NO_WARNS variables is n= ow an error.=C2=A0 Update<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0your Makefiles and scripts to define MK_<var= >=3Dno instead as required.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0One exception to this is that program or librar= y Makefiles should<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0define MAN to empty rather than setting MK_MAN= =3Dno.<br> -<br> -20200108:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Clang/LLVM is now the default compiler and LLD = the default<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0linker for riscv64.<br> -<br> -20200107:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0make universe no longer uses GCC 4.2.1 on any a= rchitectures.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Architectures not supported by in-tree Clang/LL= VM require an<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0external toolchain package.<br> -<br> -20200104:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0GCC 4.2.1 is now not built by default, as part = of the GCC 4.2.1<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0retirement plan.=C2=A0 Specifically, the GCC, G= CC_BOOTSTRAP, and GNUCXX<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0options default to off for all supported CPU ar= chitectures.=C2=A0 As a<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0short-term transition aid they may be enabled v= ia WITH_* options.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0GCC 4.2.1 is expected to be removed from the tr= ee on 2020-03-31.<br> -<br> -20200102:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Support for armv5 has been disconnected and is = being removed. The<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0machine combination MACHINE=3Darm MACHINE_ARCH= =3Darm is no longer valid.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0You must now use a MACHINE_ARCH of armv6 or arm= v7. The default<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0MACHINE_ARCH for MACHINE=3Darm is now armv7.<br= > -<br> -20191226:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Clang/LLVM is now the default compiler for all = powerpc architectures.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0LLD is now the default linker for powerpc64.=C2= =A0 The change for powerpc64<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0also includes a change to the ELFv2 ABI, incomp= atible with the existing<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0ABI.<br> -<br> -20191226:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Kernel-loadable random(4) modules are no longer= unloadable.<br> -<br> -20191222:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Clang, llvm, lld, lldb, compiler-rt, libc++, li= bunwind and openmp have<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0been upgraded to 9.0.1.=C2=A0 Please see the 20= 141231 entry below for<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0information about prerequisites and upgrading, = if you are not already<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0using clang 3.5.0 or higher.<br> -<br> -20191212:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r355677 has modified the internal interface use= d between the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0NFS modules in the kernel. As such, they must a= ll be upgraded<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0simultaneously. I will do a version bump for th= is.<br> -<br> -20191205:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The root certificates of the Mozilla CA Certifi= cate Store have been<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0imported into the base system and can be manage= d with the certctl(8)<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0utility.=C2=A0 If you have installed the securi= ty/ca_root_nss port or package<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0with the ETCSYMLINK option (the default), be ad= vised that there may be<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0differences between those included in the port = and those included in<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0base due to differences in nss branch used as w= ell as general update<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0frequency.=C2=A0 Note also that certctl(8) cann= ot manage certs in the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0format used by the security/ca_root_nss port.<b= r> -<br> -20191120:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The amd(8) automount daemon has been disabled b= y default, and will be<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0removed in the future.=C2=A0 As of FreeBSD 10.1= the autofs(5) is available<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0for automounting.<br> -<br> -20191107:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The nctgpio and wbwd drivers have been moved to= the superio bus.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0If you have one of these drivers in a kernel co= nfiguration, then<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0you should add device superio to it.=C2=A0 If y= ou use one of these drivers<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0as a module and you compile a custom set of mod= ules, then you should<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0add superio to the set.<br> -<br> -20191021:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0KPIs for network drivers to access interface ad= dresses have changed.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Users need to recompile NIC driver modules toge= ther with kernel.<br> -<br> -20191021:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The net.link.tap.user_open sysctl no longer pre= vents user opening of<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0already created /dev/tapNN devices.=C2=A0 Acces= s is still controlled by<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0node permissions, just like tun devices.=C2=A0 = The net.link.tap.user_open<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0sysctl is now used only to allow users to perfo= rm devfs cloning of<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0tap devices, and the subsequent open may not su= cceed if the user is not<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0in the appropriate group.=C2=A0 This sysctl may= be deprecated/removed<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0completely in the future.<br> -<br> -20191009:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0mips, powerpc, and sparc64 are no longer built = as part of<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0universe / tinderbox unless MAKE_OBSOLETE_GCC i= s defined. If<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0not defined, mips, powerpc, and sparc64 builds = will look for<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0the xtoolchain binaries and if installed use th= em for universe<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0builds. As llvm 9.0 becomes vetted for these ar= chitectures, they<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0will be removed from the list.<br> -<br> -20191009:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Clang, llvm, lld, lldb, compiler-rt, libc++, li= bunwind and openmp have<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0been upgraded to 9.0.0.=C2=A0 Please see the 20= 141231 entry below for<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0information about prerequisites and upgrading, = if you are not already<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0using clang 3.5.0 or higher.<br> -<br> -20191003:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The hpt27xx, hptmv, hptnr, and hptrr drivers ha= ve been removed from<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0GENERIC.=C2=A0 They are available as modules an= d can be loaded by adding<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0to /boot/loader.conf hpt27xx_load=3D"YES&q= uot;, hptmv_load=3D"YES",<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0hptnr_load=3D"YES", or hptrr_load=3D&= quot;YES", respectively.<br> -<br> -20190913:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0ntpd no longer by default locks its pages in me= mory, allowing them<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0to be paged out by the kernel. Use rlimit memlo= ck to restore<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0historic BSD behaviour. For example, add "= rlimit memlock 32"<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0to ntp.conf to lock up to 32 MB of ntpd address= space in memory.<br> -<br> -20190823:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Several of ping6's options have been rename= d for better consistency<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0with ping.=C2=A0 If you use any of -ARWXaghmrtw= x, you must update your<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0scripts.=C2=A0 See ping6(8) for details.<br> -<br> -20190727:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The vfs.fusefs.sync_unmount and vfs.fusefs.init= _backgrounded sysctls<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0and the "-o sync_unmount" and "-= o init_backgrounded" mount options have<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0been removed from mount_fusefs(8).=C2=A0 You ca= n safely remove them from<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0your scripts, because they had no effect.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The vfs.fusefs.fix_broken_io, vfs.fusefs.sync_r= esize,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0vfs.fusefs.refresh_size, vfs.fusefs.mmap_enable= ,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0vfs.fusefs.reclaim_revoked, and vfs.fusefs.data= _cache_invalidate<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0sysctls have been removed.=C2=A0 If you felt th= e need to set any of them to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0a non-default value, please tell asomers@FreeBS= D.org why.<br> -<br> -20190713:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Default permissions on the /var/account/acct fi= le (and copies of it<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0rotated by periodic daily scripts) are changed = from 0644 to 0640<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0because the file contains sensitive information= that should not be<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0world-readable.=C2=A0 If the /var/account direc= tory must be created by<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0rc.d/accounting, the mode used is now 0750.=C2= =A0 Admins who use the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0accounting feature are encouraged to change the= mode of an existing<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0/var/account directory to 0750 or 0700.<br> -<br> -20190620:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Entropy collection and the /dev/random device a= re no longer optional<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0components.=C2=A0 The "device random"= option has been removed.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Implementations of distilling algorithms can st= ill be made loadable<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0with "options RANDOM_LOADABLE" (e.g.,= random_fortuna.ko).<br> -<br> -20190612:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Clang, llvm, lld, lldb, compiler-rt, libc++, li= bunwind and openmp have<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0been upgraded to 8.0.1.=C2=A0 Please see the 20= 141231 entry below for<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0information about prerequisites and upgrading, = if you are not already<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0using clang 3.5.0 or higher.<br> -<br> -20190608:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0A fix was applied to i386 kernel modules to avo= id panics with<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0dpcpu or vnet.=C2=A0 Users need to recompile i3= 86 kernel modules<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0having pcpu or vnet sections or they will refus= e to load.<br> -<br> -20190513:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0User-wired pages now have their own counter,<br= > -=C2=A0 =C2=A0 =C2=A0 =C2=A0vm.stats.vm.v_user_wire_count.=C2=A0 The vm.max= _wired sysctl was renamed<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0to vm.max_user_wired and changed from an unsign= ed int to an unsigned<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0long.=C2=A0 bhyve VMs wired with the -S are now= subject to the user<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0wiring limit; the vm.max_user_wired sysctl may = need to be tuned to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0avoid running into the limit.<br> -<br> -20190507:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The IPSEC option has been removed from GENERIC.= =C2=A0 Users requiring<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0ipsec(4) must now load the ipsec(4) kernel modu= le.<br> -<br> -20190507:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The tap(4) driver has been folded into tun(4), = and the module has been<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0renamed to tuntap.=C2=A0 You should update any = kld_list=3D"if_tap" or<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0kld_list=3D"if_tun" entries in /etc/r= c.conf, if_tap_load=3D"YES" or<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0if_tun_load=3D"YES" entries in /boot/= loader.conf to load the if_tuntap<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0module instead, and "device tap" or &= quot;device tun" entries in kernel<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0config files to select the tuntap device instea= d.<br> -<br> -20190418:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The following knobs have been added related to = tradeoffs between<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0safe use of the random device and availability = in the absence of<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0entropy:<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0kern.random.initial_seeding.bypass_before_seedi= ng: tunable; set<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0non-zero to bypass the random device prior to s= eeding, or zero to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0block random requests until the random device i= s initially seeded.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0For now, set to 1 (unsafe) by default to restor= e pre-r346250 boot<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0availability properties.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0kern.random.initial_seeding.read_random_bypasse= d_before_seeding:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0read-only diagnostic sysctl that is set when by= pass is enabled and<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0read_random(9) is bypassed, to enable programma= tic handling of this<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0initial condition, if desired.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0kern.random.initial_seeding.arc4random_bypassed= _before_seeding:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Similar to the above, but for arc4random(9) ini= tial seeding.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0kern.random.initial_seeding.disable_bypass_warn= ings: tunable; set<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0non-zero to disable warnings in dmesg when the = same conditions are<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0met as for the diagnostic sysctls above.=C2=A0 = Defaults to zero, i.e.,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0produce warnings in dmesg when the conditions a= re met.<br> -<br> -20190416:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The loadable random module KPI has changed; the= random_infra_init()<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0routine now requires a 3rd function pointer for= a bool (*)(void)<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0method that returns true if the random device i= s seeded (and<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0therefore unblocked).<br> -<br> -20190404:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r345895 reverts r320698. This implies that an n= fsuserd(8) daemon<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0built from head sources between r320757 (July 6= , 2017) and<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r338192 (Aug. 22, 2018) will not work unless th= e "-use-udpsock"<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0is added to the command line.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0nfsuserd daemons built from head sources that a= re post-r338192 are<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0not affected and should continue to work.<br> -<br> -20190320:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The fuse(4) module has been renamed to fusefs(4= ) for consistency with<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0other filesystems.=C2=A0 You should update any = kld_load=3D"fuse" entries in<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0/etc/rc.conf, fuse_load=3D"YES" entri= es in /boot/loader.conf, and<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0"options FUSE" entries in kernel conf= ig files.<br> -<br> -20190304:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Clang, llvm, lld, lldb, compiler-rt and libc++ = have been upgraded to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A08.0.0.=C2=A0 Please see the 20141231 entry belo= w for information about<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0prerequisites and upgrading, if you are not alr= eady using clang 3.5.0<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0or higher.<br> -<br> -20190226:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0geom_uzip(4) depends on the new module xz.=C2= =A0 If geom_uzip is statically<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0compiled into your custom kernel, add 'devi= ce xz' statement to the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0kernel config.<br> -<br> -20190219:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0drm and drm2 have been removed from the tree. P= lease see<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0<a href=3D"https://wiki.freebsd.org/Graphics" r= el=3D"noreferrer noreferrer" target=3D"_blank">https://wiki.freebsd.org/Gra= phics</a> for the latest information on<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0migrating to the drm ports.<br> -<br> -20190131:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Iflib is no longer unconditionally compiled int= o the kernel.=C2=A0 Drivers<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0using iflib and statically compiled into the ke= rnel, now require<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0the 'device iflib' config option.=C2=A0= For the same drivers loaded as<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0modules on kernels not having 'device iflib= ', the iflib.ko module<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0is loaded automatically.<br> -<br> -20190125:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The IEEE80211_AMPDU_AGE and AH_SUPPORT_AR5416 k= ernel configuration<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0options no longer exist since r343219 and r3434= 27 respectively;<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0nothing uses them, so they should be just remov= ed from custom<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0kernel config files.<br> -<br> -20181230:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r342635 changes the way efibootmgr(8) works by = requiring users to add<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0the -b (bootnum) parameter for commands where t= he bootnum was previously<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0specified with each option. For example 'ef= ibootmgr -B 0001' is now<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0'efibootmgr -B -b 0001'.<br> -<br> -20181220:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r342286 modifies the NFSv4 server so that it ob= eys vfs.nfsd.nfs_privport<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0in the same as it is applied to NFSv2 and 3.=C2= =A0 This implies that NFSv4<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0servers that have vfs.nfsd.nfs_privport set wil= l only allow mounts<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0from clients using a reserved port. Since both = the FreeBSD and Linux<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0NFSv4 clients use reserved ports by default, th= is should not affect<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0most NFSv4 mounts.<br> -<br> -20181219:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The XLP config has been removed. We can't s= upport 64-bit atomics in this<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0kernel because it is running in 32-bit mode. XL= P users must transition<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0to running a 64-bit kernel (XLP64 or XLPN32).<b= r> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The mips GXEMUL support has been removed from F= reeBSD. MALTA* + qemu is<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0the preferred emulator today and we don't n= eed two different ones.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The old sibyte / swarm / Broadcom BCM1250 suppo= rt has been<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0removed from the mips port.<br> -<br> -20181211:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Clang, llvm, lld, lldb, compiler-rt and libc++ = have been upgraded to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A07.0.1.=C2=A0 Please see the 20141231 entry belo= w for information about<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0prerequisites and upgrading, if you are not alr= eady using clang 3.5.0<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0or higher.<br> -<br> -20181211:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Remove the timed and netdate programs from the = base tree.=C2=A0 Setting<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0the time with these daemons has been obsolete f= or over a decade.<br> -<br> -20181126:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0On amd64, arm64 and armv7 (architectures that i= nstall LLVM's ld.lld<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0linker as /usr/bin/ld) GNU ld is no longer inst= alled as ld.bfd, as<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0it produces broken binaries when ifuncs are in = use.=C2=A0 Users needing<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0GNU ld should install the binutils port or pack= age.<br> -<br> -20181123:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The BSD crtbegin and crtend code has been enabl= ed by default. It has<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0had extensive testing on amd64, arm64, and i386= . It can be disabled<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0by building a world with -DWITHOUT_BSD_CRTBEGIN= .<br> -<br> -20181115:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The set of CTM commands (ctm, ctm_smail, ctm_rm= ail, ctm_dequeue)<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0has been converted to a port (misc/ctm) and wil= l be removed from<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0FreeBSD-13.=C2=A0 It is available as a package = (ctm) for all supported<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0FreeBSD versions.<br> -<br> -20181110:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The default newsyslog.conf(5) file has been cha= nged to only include<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0files in /etc/newsyslog.conf.d/ and /usr/local/= etc/newsyslog.conf.d/ if<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0the filenames end in '.conf' and do not= begin with a '.'.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0You should check the configuration files in the= se two directories match<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0this naming convention. You can verify which co= nfiguration files are<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0being included using the command:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0$ newsyslog -Nrv<br= > -<br> -20181015:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Ports for the DRM modules have been simplified.= Now, amd64 users should<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0just install the drm-kmod port. All others shou= ld install<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0drm-legacy-kmod.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Graphics hardware that's newer than about 2= 010 usually works with<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0drm-kmod.=C2=A0 For hardware older than 2013, h= owever, some users will need<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0to use drm-legacy-kmod if drm-kmod doesn't = work for them. Hardware older<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0than 2008 usually only works in drm-legacy-kmod= . The graphics team can<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0only commit to hardware made since 2013 due to = the complexity of the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0market and difficulty to test all the older car= ds effectively. If you<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0have hardware supported by drm-kmod, you are st= rongly encouraged to use<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0that as you will get better support.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Other than KPI chasing, drm-legacy-kmod will no= t be updated. As outlined<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0elsewhere, the drm and drm2 modules will be eli= minated from the src base<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0soon (with a limited exception for arm). Please= update to the package<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0asap and report any issues to <a href=3D"mailto= :x11@freebsd.org" target=3D"_blank" rel=3D"noreferrer">x11@freebsd.org</a>.= <br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Generally, anybody using the drm*-kmod packages= should add<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0WITHOUT_DRM_MODULE=3Dt and WITHOUT_DRM2_MODULE= =3Dt to avoid nasty<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0cross-threading surprises, especially with auto= matic driver<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0loading from X11 startup. These will become the= defaults in 13-current<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0shortly.<br> -<br> -20181012:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The ixlv(4) driver has been renamed to iavf(4).= =C2=A0 As a consequence,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0custom kernel and module loading configuration = files must be updated<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0accordingly.=C2=A0 Moreover, interfaces previou= s presented as ixlvN to the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0system are now exposed as iavfN and network con= figuration files must<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0be adjusted as necessary.<br> -<br> -20181009:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0OpenSSL has been updated to version 1.1.1.=C2= =A0 This update included<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0additional various API changes throughout the b= ase system.=C2=A0 It is<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0important to rebuild third-party software after= upgrading.=C2=A0 The value<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0of __FreeBSD_version has been bumped accordingl= y.<br> -<br> -20181006:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The legacy DRM modules and drivers have now bee= n added to the loader's<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0module blacklist, in favor of loading them with= kld_list in rc.conf(5).<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The module blacklist may be overridden with the= loader.conf(5)<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0'module_blacklist' variable, but loadin= g them via rc.conf(5) is strongly<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0encouraged.<br> -<br> -20181002:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The cam(4) based nda(4) driver will be used ove= r nvd(4) by default on<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0powerpc64. You may set 'options NVME_USE_NV= D=3D1' in your kernel conf or<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0loader tunable 'hw.nvme.use_nvd=3D1' if= you wish to use the existing<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0driver.=C2=A0 Make sure to edit /boot/etc/kboot= .conf and fstab to use the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0nda device name.<br> -<br> -20180913:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Reproducible build mode is now on by default, i= n preparation for<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0FreeBSD 12.0.=C2=A0 This eliminates build metad= ata such as the user,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0host, and time from the kernel (and uname), unl= ess the working tree<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0corresponds to a modified checkout from a versi= on control system.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The previous behavior can be obtained by settin= g the /etc/src.conf<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0knob WITHOUT_REPRODUCIBLE_BUILD.<br> -<br> -20180826:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The Yarrow CSPRNG has been removed from the ker= nel as it has not been<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0supported by its designers since at least 2003.= Fortuna has been the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0default since FreeBSD-11.<br> -<br> -20180822:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0devctl freeze/thaw have gone into the tree, the= rc scripts have been<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0updated to use them and devmatch has been chang= ed.=C2=A0 You should update<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0kernel, userland and rc scripts all at the same= time.<br> -<br> -20180818:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The default interpreter has been switched from = 4th to Lua.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0LOADER_DEFAULT_INTERP, documented in build(7), = will override the default<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0interpreter.=C2=A0 If you have custom FORTH cod= e you will need to set<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0LOADER_DEFAULT_INTERP=3D4th (valid values are 4= th, lua or simp) in<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0src.conf for the build.=C2=A0 This will create = default hard links between<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0loader and loader_4th instead of loader and loa= der_lua, the new default.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0If you are using UEFI it will create the proper= hard link to loader.efi.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0bhyve uses userboot.so. It remains 4th-only unt= il some issues are solved<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0regarding coexisting with multiple versions of = FreeBSD are resolved.<br> -<br> -20180815:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0ls(1) now respects the COLORTERM environment va= riable used in other<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0systems and software to indicate that a colored= terminal is both<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0supported and desired.=C2=A0 If ls(1) is sudden= ly emitting colors, they may<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0be disabled again by either removing the unwant= ed COLORTERM from your<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0environment, or using `ls --color=3Dnever`.=C2= =A0 The ls(1) specific CLICOLOR<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0may not be observed in a future release.<br> -<br> -20180808:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The default pager for most commands has been ch= anged to "less".=C2=A0 To<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0restore the old behavior, set PAGER=3D"mor= e" and MANPAGER=3D"more -s" in<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0your environment.<br> -<br> -20180731:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The jedec_ts(4) driver has been removed. A supe= rset of its functionality<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0is available in the jedec_dimm(4) driver, and t= he manpage for that<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0driver includes migration instructions. If you = have "device jedec_ts"<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0in your kernel configuration file, it must be r= emoved.<br> -<br> -20180730:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0amd64/GENERIC now has EFI runtime services, EFI= RT, enabled by default.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0This should have no effect if the kernel is boo= ted via BIOS/legacy boot.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0EFIRT may be disabled via a loader tunable, efi= .rt.disabled, if a system<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0has a buggy firmware that prevents a successful= boot due to use of<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0runtime services.<br> -<br> -20180727:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Atmel AT91RM9200 and AT91SAM9, Cavium CNS 11xx = and XScale<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0support has been removed from the tree. These p= orts were<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0obsolete and/or known to be broken for many yea= rs.<br> -<br> -20180723:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0loader.efi has been augmented to participate mo= re fully in the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0UEFI boot manager protocol. loader.efi will now= look at the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0BootXXXX environment variable to determine if a= specific kernel<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0or root partition was specified. XXXX is derive= d from BootCurrent.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0efibootmgr(8) manages these standard UEFI varia= bles.<br> -<br> -20180720:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0zfsloader's functionality has now been fold= ed into loader.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0zfsloader is no longer necessary once you'v= e updated your<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0boot blocks. For a transition period, we will i= nstall a<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0hardlink for zfsloader to loader to allow a smo= oth transition<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0until the boot blocks can be updated (hard link= because old<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0zfs boot blocks don't understand symlinks).= <br> -<br> -20180719:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0ARM64 now have efifb support, if you want to ha= ve serial console<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0on your arm64 board when an screen is connected= and the bootloader<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0setup a frame buffer for us to use, just add :<= br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0boot_serial=3DYES<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0boot_multicons=3DYES<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0in /boot/loader.conf<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0For Raspberry Pi 3 (RPI) users, this is needed = even if you don't have<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0an screen connected as the firmware will setup = a frame buffer are that<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0u-boot will expose as an EFI frame buffer.<br> -<br> -20180719:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0New uid:gid added, ntpd:ntpd (123:123).=C2=A0 B= e sure to run mergemaster<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0or take steps to update /etc/passwd before doin= g installworld on<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0existing systems.=C2=A0 Do not skip the "m= ergemaster -Fp" step before<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0installworld, as described in the update proced= ures near the bottom<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0of this document.=C2=A0 Also, rc.d/ntpd now sta= rts ntpd(8) as user ntpd<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0if the new mac_ntpd(4) policy is available, unl= ess ntpd_flags or<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0the ntp config file contain options that change= file/dir locations.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0When such options (e.g., "statsdir" o= r "crypto") are used, ntpd can<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0still be run as non-root by setting ntpd_user= =3Dntpd in rc.conf, after<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0taking steps to ensure that all required files/= dirs are accessible<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0by the ntpd user.<br> -<br> -20180717:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Big endian arm support has been removed.<br> -<br> -20180711:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The static environment setup in kernel configs = is no longer mutually<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0exclusive with the loader(8) environment by def= ault.=C2=A0 In order to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0restore the previous default behavior of disabl= ing the loader(8)<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0environment if a static environment is present,= you must specify<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0loader_env.disabled=3D1 in the static environme= nt.<br> -<br> -20180705:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The ABI of syscalls used by management tools li= ke sockstat and<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0netstat has been broken to allow 32-bit binarie= s to work on<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A064-bit kernels without modification.=C2=A0 Thes= e programs will need<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0to match the kernel in order to function.=C2=A0= External programs may<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0require minor modifications to accommodate a ch= ange of type in<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0structures from pointers to 64-bit virtual addr= esses.<br> -<br> -20180702:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0On i386 and amd64 atomics are now inlined. Out = of tree modules using<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0atomics will need to be rebuilt.<br> -<br> -20180701:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The '%I' format in the kern.corefile sy= sctl limits the number of<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0core files that a process can generate to the n= umber stored in the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0debug.ncores sysctl. The '%I' format is= replaced by the single digit<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0index. Previously, if all indexes were taken th= e kernel would overwrite<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0only a core file with the highest index in a fi= lename.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Currently the system will create a new core fil= e if there is a free<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0index or if all slots are taken it will overwri= te the oldest one.<br> -<br> -20180630:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Clang, llvm, lld, lldb, compiler-rt and libc++ = have been upgraded to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A06.0.1.=C2=A0 Please see the 20141231 entry belo= w for information about<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0prerequisites and upgrading, if you are not alr= eady using clang 3.5.0<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0or higher.<br> -<br> -20180628:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r335753 introduced a new quoting method. Howeve= r, etc/devd/devmatch.conf<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0needed to be changed to work with it. This chan= ge was made with r335763<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0and requires a mergemaster / etcupdate / etc to= update the installed<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0file.<br> -<br> -20180612:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r334930 changed the interface between the NFS m= odules, so they all<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0need to be rebuilt.=C2=A0 r335018 did a __FreeB= SD_version bump for this.<br> -<br> -20180530:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0As of r334391 lld is the default amd64 system l= inker; it is installed<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0as /usr/bin/ld.=C2=A0 Kernel build workarounds = (see 20180510 entry) are no<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0longer necessary.<br> -<br> -20180530:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The kernel / userland interface for devinfo cha= nged, so you'll<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0need a new kernel and userland as a pair for it= to work (rebuilding<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0lib/libdevinfo is all that's required). dev= info and devmatch will<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0not work, but everything else will when there&#= 39;s a mismatch.<br> -<br> -20180523:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The on-disk format for hwpmc callchain records = has changed to include<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0threadid corresponding to a given record. This = changes the field offsets<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0and thus requires that libpmcstat be rebuilt be= fore using a kernel<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0later than r334108.<br> -<br> -20180517:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The vxge(4) driver has been removed.=C2=A0 This= driver was introduced into<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0HEAD one week before the Exar left the Ethernet= market and is not<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0known to be used.=C2=A0 If you have device vxge= in your kernel config file<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0it must be removed.<br> -<br> -20180510:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The amd64 kernel now requires a ld that support= s ifunc to produce a<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0working kernel, either lld or a newer binutils.= lld is built by default<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0on amd64, and the 'buildkernel' target = uses it automatically. However,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0it is not the default linker, so building the k= ernel the traditional<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0way requires LD=3Dld.lld on the command line (o= r LD=3D/usr/local/bin/ld for<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0binutils port/package). lld will soon be defaul= t, and this requirement<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0will go away.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0NOTE: As of r334391 lld is the default system l= inker on amd64, and no<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0workaround is necessary.<br> -<br> -20180508:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The nxge(4) driver has been removed.=C2=A0 This= driver was for PCI-X 10g<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0cards made by s2io/Neterion.=C2=A0 The company = was acquired by Exar and<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0no longer sells or supports Ethernet products.= =C2=A0 If you have device<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0nxge in your kernel config file it must be remo= ved.<br> -<br> -20180504:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The tz database (tzdb) has been updated to 2018= e.=C2=A0 This version more<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0correctly models time stamps in time zones with= negative DST such as<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Europe/Dublin (from 1971 on), Europe/Prague (19= 46/7), and<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Africa/Windhoek (1994/2017).=C2=A0 This does no= t affect the UT offsets, only<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0time zone abbreviations and the tm_isdst flag.<= br> -<br> -20180502:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The ixgb(4) driver has been removed.=C2=A0 This= driver was for an early and<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0uncommon legacy PCI 10GbE for a single ASIC, In= tel 82597EX. Intel<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0quickly shifted to the long lived ixgbe family.= =C2=A0 If you have device<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0ixgb in your kernel config file it must be remo= ved.<br> -<br> -20180501:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The lmc(4) driver has been removed.=C2=A0 This = was a WAN interface<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0card that was already reportedly rare in 2003, = and had an ambiguous<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0license.=C2=A0 If you have device lmc in your k= ernel config file it must<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0be removed.<br> -<br> -20180413:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Support for Arcnet networks has been removed.= =C2=A0 If you have device<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0arcnet or device cm in your kernel config file = they must be<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0removed.<br> -<br> -20180411:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Support for FDDI networks has been removed.=C2= =A0 If you have device<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0fddi or device fpa in your kernel config file t= hey must be<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0removed.<br> -<br> -20180406:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0In addition to supporting RFC 3164 formatted me= ssages, the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0syslogd(8) service is now capable of parsing RF= C 5424 formatted<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0log messages. The main benefit of using RFC 542= 4 is that clients<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0may now send log messages with timestamps conta= ining year numbers,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0microseconds and time zone offsets.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Similarly, the syslog(3) C library function has= been altered to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0send RFC 5424 formatted messages to the local s= ystem logging<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0daemon. On systems using syslogd(8), this chang= e should have no<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0negative impact, as long as syslogd(8) and the = C library are<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0updated at the same time. On systems using a di= fferent system<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0logging daemon, it may be necessary to make con= figuration<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0adjustments, depending on the software used.<br= > -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0When using syslog-ng, add the 'syslog-proto= col' flag to local<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0input sources to enable parsing of RFC 5424 for= matted messages:<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0source src {<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0unix-dgram("/var/run/log" flags(syslog-protocol));<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0}<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0When using rsyslog, disable the 'SysSock.Us= eSpecialParser' option<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0of the 'imuxsock' module to let message= s be processed by the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0regular RFC 3164/5424 parsing pipeline:<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0module(load=3D"= ;imuxsock" SysSock.UseSpecialParser=3D"off")<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Do note that these changes only affect communic= ation between local<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0applications and syslogd(8). The format that sy= slogd(8) uses to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0store messages on disk or forward messages to o= ther systems<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0remains unchanged. syslogd(8) still uses RFC 31= 64 for these<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0purposes. Options to customize this behaviour w= ill be added in the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0future. Utilities that process log files stored= in /var/log are<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0thus expected to continue to function as before= .<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0__FreeBSD_version has been incremented to 12000= 61 to denote this<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0change.<br> -<br> -20180328:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Support for token ring networks has been remove= d. If you<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0have "device token" in your kernel co= nfig you should remove<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0it. No device drivers supported token ring.<br> -<br> -20180323:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0makefs was modified to be able to tag ISO9660 E= l Torito boot catalog<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0entries as EFI instead of overloading the i386 = tag as done previously.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The amd64 mkisoimages.sh script used to build a= md64 ISO images for<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0release was updated to use this. This may mean = that makefs must be<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0updated before "make cdrom" can be ru= n in the release directory. This<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0should be as simple as:<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0$ cd $SRCDIR/usr.sb= in/makefs<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0$ make depend all i= nstall<br> -<br> -20180212:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0FreeBSD boot loader enhanced with Lua scripting= . It's purely opt-in for<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0now by building WITH_LOADER_LUA and WITHOUT_FOR= TH in /etc/src.conf.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Co-existence for the transition period will com= e shortly. Booting is a<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0complex environment and test coverage for Lua-e= nabled loaders has been<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0thin, so it would be prudent to assume it might= not work and make<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0provisions for backup boot methods.<br> -<br> -20180211:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0devmatch functionality has been turned on in de= vd. It will automatically<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0load drivers for unattached devices. This may c= ause unexpected drivers<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0to be loaded. Please report any problems to cur= rent@ and<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0<a href=3D"mailto:imp@freebsd.org" target=3D"_b= lank" rel=3D"noreferrer">imp@freebsd.org</a>.<br> -<br> -20180114:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Clang, llvm, lld, lldb, compiler-rt and libc++ = have been upgraded to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A06.0.0.=C2=A0 Please see the 20141231 entry belo= w for information about<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0prerequisites and upgrading, if you are not alr= eady using clang 3.5.0<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0or higher.<br> -<br> -20180110:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0LLVM's lld linker is now used as the FreeBS= D/amd64 bootstrap linker.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0This means it is used to link the kernel and us= erland libraries and<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0executables, but is not yet installed as /usr/b= in/ld by default.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0To revert to ld.bfd as the bootstrap linker, in= /etc/src.conf set<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0WITHOUT_LLD_BOOTSTRAP=3Dyes<br> -<br> -20180110:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0On i386, pmtimer has been removed. Its function= ality has been folded<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0into apm. It was a no-op on ACPI in current for= a while now (but was<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0still needed on i386 in FreeBSD 11 and earlier)= . Users may need to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0remove it from kernel config files.<br> -<br> -20180104:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The use of RSS hash from the network card aka f= lowid has been<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0disabled by default for lagg(4) as it's cur= rently incompatible with<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0the lacp and loadbalance protocols.<br> -<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0This can be re-enabled by setting the following= in loader.conf:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0net.link.lagg.default_use_flowid=3D"1"= ;<br> -<br> -20180102:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The SW_WATCHDOG option is no longer necessary t= o enable the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0hardclock-based software watchdog if no hardwar= e watchdog is<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0configured. As before, SW_WATCHDOG will cause t= he software<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0watchdog to be enabled even if a hardware watch= dog is configured.<br> -<br> -20171215:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r326887 fixes the issue described in the 201712= 14 UPDATING entry.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r326888 flips the switch back to building GELI = support always.<br> -<br> -20171214:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0r362593 broke ZFS + GELI support for reasons un= known. However,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0it also broke ZFS support generally, so GELI ha= s been turned off<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0by default as the lesser evil in r326857. If yo= u boot off ZFS and/or<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0GELI, it might not be a good time to update.<br= > -<br> -20171125:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0PowerPC users must update loader(8) by rebuildi= ng world before<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0installing a new kernel, as the protocol connec= ting them has<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0changed. Without the update, loader metadata wi= ll not be passed<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0successfully to the kernel and users will have = to enter their<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0root partition at the kernel mountroot prompt t= o continue booting.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Newer versions of loader can boot old kernels w= ithout issue.<br> -<br> -20171110:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The LOADER_FIREWIRE_SUPPORT build variable has = been renamed to<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0WITH/OUT_LOADER_FIREWIRE. LOADER_{NO_,}GELI_SUP= PORT has been renamed<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0to WITH/OUT_LOADER_GELI.<br> -<br> -20171106:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The naive and non-compliant support of posix_fa= llocate(2) in ZFS<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0has been removed as of r325320.=C2=A0 The syste= m call now returns EINVAL<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0when used on a ZFS file.=C2=A0 Although the new= behavior complies with the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0standard, some consumers are not prepared to co= pe with it.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0One known victim is lld prior to r325420.<br> -<br> -20171102:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0Building in a FreeBSD src checkout will automat= ically create object<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0directories now rather than store files in the = current directory if<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0'make obj' was not ran.=C2=A0 Calling &= #39;make obj' is no longer necessary.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0This feature can be disabled by setting WITHOUT= _AUTO_OBJ=3Dyes in<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0/etc/src-env.conf (not /etc/src.conf), or passi= ng the option in the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0environment.<br> -<br> -20171101:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The default MAKEOBJDIR has changed from /usr/ob= j/<srcdir> for native<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0builds, and /usr/obj/<arch>/<srcdir>= ; for cross-builds, to a unified<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0/usr/obj/<srcdir>/<arch>.=C2=A0 Thi= s behavior can be changed to the old<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0format by setting WITHOUT_UNIFIED_OBJDIR=3Dyes = in /etc/src-env.conf,<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0the environment, or with -DWITHOUT_UNIFIED_OBJD= IR when building.<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The UNIFIED_OBJDIR option is a transitional fea= ture that will be<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0removed for 12.0 release; please migrate to the= new format for any<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0tools by looking up the OBJDIR used by 'mak= e -V .OBJDIR' means rather<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0than hardcoding paths.<br> -<br> -20171028:<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0The native-xtools target no longer installs the= files by default to the<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0OBJDIR.=C2=A0 Use the native-xtools-install tar= get with a DESTDIR to install<br> -=C2=A0 =C2=A0 =C2=A0 =C2=A0to ${DESTDIR}/${NXTP} where NXTP defaults to /n= xb-bin.<br> *** 1218 LINES SKIPPED ***<br> </blockquote></div> --00000000000021a3de05ee4e325b--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CANCZdfq%2BAVGWa91Cv80t60jKAmw0UwoTVNFeOGRjOhAPjsJH%2Bw>