1
0
mirror of https://git.FreeBSD.org/src.git synced 2024-11-26 07:55:01 +00:00
freebsd/RELNOTES
Warner Losh 350f9ac5b3 Remove $FreeBSD$: one-line bare tag
Remove /^\s*\$FreeBSD\$$\n/

Similar commit in main:
(cherry picked from commit 05248206f7)
2023-08-23 11:43:33 -06:00

423 lines
16 KiB
Plaintext

Release notes for FreeBSD 13.0.
This file describes new user-visible features, changes and updates relevant to
users of binary FreeBSD releases. Each entry should describe the change in no
more than several sentences and should reference manual pages where an
interested user can find more information. Entries should wrap after 80
columns. Each entry should begin with one or more commit IDs on one line,
specified as a comma separated list and/or range, followed by a colon and a
newline. Entries should be separated by a newline.
Changes to this file should not be MFCed.
0644746d5091:
Add a new "syskrb5" mount option for Kerberized NFSv4.1/4.2 mounts.
Without this patch, a Kerberized NFSv4.1/4.2 mount must provide
a Kerberos credential for the client at mount time.
This patch uses a feature of NFSv4.1/4.2 called SP4_NONE, which
allows the state maintenance operations to be performed by any
authentication mechanism, so that these operations may be done via
AUTH_SYS instead of RPCSEC_GSS (KerberosV). As such, no Kerberos
credential is required at mount time.
See mount_nfs(8).
b4805d577787 and many others:
Add support so that nfsd(8), nfsuserd(8), mountd(8), gssd(8)
and rpc.tlsservd(8) can be run in an appropriately configured
vnet prison. The vnet prison must be on its own file system,
have the "allow.nfsd" jail parameter set on it and enforce_statfs
cannot be set to "0". Use of UDP and pNFS server configurations
are not permitted. (ie. The nfsd command line options "-u", "-p"
and "-m" are not supported.)
See jail(8), nfsd(8) and mountd(8).
68e86d5265bc,e58dfd0de589,59f5a5cb724e,6e272a78de36,4c4a4fd4a649,ba2ae2cca63a:
sendmail has been updated to the latest upstream version (8.17.1).
225443828ec6..c44d097dcf92:
bhyve now supports more than 16 vCPUs in a guest. By default
bhyve permits each guest to create the same number of vCPUs as
the count of physical CPUs on the host. This limit can be
adjusted via the loader tunable hw.vmm.maxcpu.
1462dc95f796:
Kernel TLS offload now supports receive-side offload of TLS 1.3.
3ee882bf21af:
Change handling of the lowest address on an IPv4 (sub)net so that
packets are not sent as a broadcast unless this has been set as the
broadcast address. This makes the lowest address usable for a host.
The old behavior can be restored with the net.inet.ip.broadcast_lowest
sysctl. For more information, see
https://datatracker.ietf.org/doc/draft-schoen-intarea-lowest-address/.
33ff39796ffe,8719e8a951b7:
A new rc(8) service script zfskeys allows for automatic decryption
of ZFS datasets encrypted with ZFS native encryption during boot.
See the rc.conf(5) manual page for more information.
b7a2cf0d9102 - eae02d959363:
Upgrade bhyve's emulation to version 1.4 of the NVMe specification
0a6760a1de32, 3f3676a71266, 580c04df4db6:
Add WiFi 6 support.
various:
Add support for the HiFive Unmatched RISC-V board.
9fb6e613373c:
Add a sysctl called vfs.nfsd.srvmaxio that can be used to
increase the NFS server's maximum I/O size from 128Kbytes
to any power of 2 up to 1Mbyte. It can only be set when
the nfsd threads are not running and will normally require
an increase in kern.ipc.maxsockbuf to at least the value
recommended by the console log message generated when
setting vfs.nfsd.srvmaxio is first attempted.
9ec7dbf46b0a:
Add a new NFSv4.1/4.2 mount option "nconnect" that can
be used to specify the number of TCP connections that
will be used for the mount, up to a maximum of 16.
The first (default) TCP connection will be used for
all RPCs that consist of small RPC messages.
The RPCs that can consist of large RPC messages
(Read/Readdir/ReaddirPlus/Write) will be sent on the
additional TCP connections in a round robin fashion.
If either the NFS client or NFS server have multiple
network interfaces aggregated together or a network
interface that uses multiple queues, this can increase
NFS performance for the mount.
various:
One True Awk has been updated to the latest from upstream
(20210215). All the FreeBSD patches, but one, have now been
either up streamed or discarded. Notable changes include:
o Locale is no longer used for ranges
o Various bugs fixed
o Better compatibility with gawk and mawk
The one FreeBSD change, likely to be removed in FreeBSD 14, is that
we still allow hex numbers, prefixed with 0x, to be parsed and
interpreted as hex numbers while all other awks (including one
true awk now) interpret them as 0 in line with awk's historic
behavior.
8a04edfdcbd2:
Change the default minor version used for an NFSv4 mount
to the highest minor version supported by the NFSv4 server.
This default can be overridden by using the "minorversion"
mount option.
2c76eebca71b, 59f6f5e23c1a:
Add two daemons rpc.tlsclntd(8) and rpc.tlsservd(8) that provide
support for NFS-over-TLS as described in the Internet Draft titled
"Towards Remote Procedure Call Encryption By Default".
These daemons are only built when WITH_OPENSSL_KTLS is specified
and are only tested on amd64 at this time.
They use KTLS to encrypt/decrypt all NFS RPC message traffic, plus
optional verification of machine identity via X.509 certificates.
f76393a6305b6:
Add AES-GCM support to armv8crypto(4) providing accelerated
support for KTLS, IPsec, and other crypto API consumers.
074a91f746bd:
The aesni(4) and armv8crypto(4) devices are now included in
GENERIC on amd64, i386, and arm64.
2e1c94aa1fd5:
Add support for enforcing W^X mapping policy for user
processes. The policy is not enforced by default but can be
enabled by setting the kern.elf32.allow_wx and
kern.elf64.allow_wx sysctls to 0. Individual binaries can be
exempted from the policy by elfctl(1) via the wxneeded
feature.
4979620ece98:
Add AES-XTS support to armv8crypto(4) providing accelerated
software support for the default GELI cipher on arm64 systems.
022ca2fc7fe0:
Add aio_writev(2) and aio_readv(2), vectored analogues of aio_write(2)
and aio_read(2).
92bbfe1f0d1f:
The fusefs(5) protocol has been updated to 7.28. Support for
FUSE_COPY_FILE_RANGE and FUSE_LSEEK is added.
r368667:
GDB 6.1.1 was removed. Users of crashinfo(8) should install the
gdb package or devel/gdb port.
r368559:
The hme(4) driver was removed.
r367660:
Fixes the case where gssd will not startup because /usr is a separate
local file system that is not yet mounted. It does not fix the case
where /usr is a separately mounted remote file system (such as NFS).
This latter case can be fixed by adding mountcritremote to the
REQUIRED line. Unfortunately doing so implies that all Kerberized
NFS mounts in /etc/fstab will need the "late" mount option.
This was not done, since the requirement for "late" would introduce
a POLA violation.
r367423:
This commit added a new startup scripts variable called
nfsv4_server_only which uses the -R option on mountd added by r367026.
When nfsv4_server_only is set to "YES" in /etc/rc.conf, the NFS server
only handles NFSv4 and does not register with rpcbind. As such, rpcbind
does not need to be running. Useful for sites which consider rpcbind a
security issue.
r366267:
Kernel option ACPI_DMAR was renamed to IOMMU. amd64's IOMMU subsystem
was split out from amd64 DMAR support and is now generic, i.e., it can
be used by all architectures.
r364896:
A series of commits ending with r364896 added NFS over TLS
to the kernel. This is believed to be compatible with
the Internet Draft titled "Towards Remote Procedure Call Encryption
By Default" (expected to soon become an RFC).
The mount_nfs(8) and exports(5) man pages describe the mount and
export option(s) related to NFS over TLS.
For NFS over TLS to work, the rpctlscd(8) { client } or rpctlssd(8)
{ server } must be running on a kernel built with "options KERN_TLS"
on an architecture where PMAP_HAS_DMAP != 0.
r364725:
Changes to one obscure devd event generated on resume need to
be documented. The old form will still be generated in 13, but not
in 14.
r363679:
Applications using regex(3), e.g. sed/grep, will no longer accept
redundant escapes for most ordinary characters.
r363253:
SCTP support has been removed from GENERIC kernel configurations.
The SCTP stack is now built as sctp.ko and can be dynamically loaded.
r363233:
Merge sendmail 8.16.1: See contrib/sendmail/RELEASE_NOTES for details.
r363180:
The safexcel(4) crypto offload driver has been added.
r363084:
nc(1) now implements SCTP mode, enabled by specifying the --sctp option.
r362681:
A new implementation of bc and dc has been imported. It offers
better standards compliance, performance, localization and comes
with extensive test cases that are optionally installed.
Use WITHOUT_GH_BC=yes to build and install the world with the
previous version instead of the new one, if required.
r362158, r362163:
struct export_args has changed so that the "user" specified for
the -maproot and -mapall exports(5) options may be in more than
16 groups.
r361884:
sed(1) has learned about hex escapes (e.g. \x27) and will now do the
right thing with them, removing the need for printf magic or obnoxious
escaping in many scenarios.
r361238, r361798, r361799:
ZFS will now unconditionally reject read(2) of a directory with EISDIR.
Additionally, read(2) of a directory is now rejected with EISDIR by
default and may be re-enabled for non-ZFS filesystems that allow it with
the sysctl(8) MIB 'security.bsd.allow_read_dir'.
Aliases for grep to default to '-d skip' may be desired if commonly
non-recursively grepping a list that includes directories and the
possibility of EISDIR errors in stderr is not tolerable. Example
aliases, commented out, have been installed in /root/.cshrc and
/root/.shrc.
r361066:
Add exec.prepare and exec.release hooks for jail(8) and jail.conf(5).
exec.prepare runs before mounts, so can be used to populate new jails.
exec.release runs after unmounts, so can be used to remove ephemeral
jails.
r360920,r360923,r360924,r360927,r360928,r360931,r360933,r360936:
Remove support for ARC4, Blowfish, Cast, DES, Triple DES, MD5,
MD5-KPDK, MD5-HMAC, SHA1-KPDK, and Skipjack algorithms from
the kernel open cryptographic framework (OCF).
r360562:
Remove support for ARC4, Blowfish, Cast, DES, Triple DES,
MD5-HMAC, and Skipjack algorithms from /dev/crypto.
r360557:
Remove support for DES, Triple DES, Blowfish, Cast, and
Camellia ciphers from IPsec(4). Remove support for MD5-HMAC,
Keyed MD5, Keyed SHA1, and RIPEMD160-HMAC from IPsec(4).
r359945:
Remove support for Triple DES, Blowfish, and MD5 HMAC from
geli(4).
r359786-r359787:
Remove support for DES, Triple DES, and RC4 from in-kernel GSS
authentication.
r357627:
remove elf2aout.
r357560-r357565:
init(8), service(8), and cron(8) will now adopt user/class environment
variables (excluding PATH, by default, which will be overwritten) by
default. Notably, environment variables for all cron jobs and rc
services can now be set via login.conf(5).
r357455:
sparc64 has been removed from FreeBSD.
r355677:
Adds support for NFSv4.2 (RFC-7862) and Extended Attributes
(RFC-8276) to the NFS client and server.
NFSv4.2 is comprised of several optional features that can be supported
in addition to NFSv4.1. This patch adds the following optional features:
- posix_fadvise(POSIX_FADV_WILLNEED/POSIX_FADV_DONTNEED)
- posix_fallocate()
- intra server file range copying via the copy_file_range(2) syscall
--> Avoiding data tranfer over the wire to/from the NFS client.
- lseek(SEEK_DATA/SEEK_HOLE)
- Extended attribute syscalls for "user" namespace attributes as defined
by RFC-8276.
For the client, NFSv4.2 is only used if the mount command line option
minorversion=2 is specified.
For the server, two new sysctls called vfs.nfsd.server_min_minorversion4
and vfs.nfsd.server_max_minorversion4 have been added that allow
sysadmins to limit the minor versions of NFSv4 supported by the nfsd
server.
Setting vfs.nfsd.server_max_minorversion4 to 0 or 1 will disable NFSv4.2
on the server.
r356263:
armv5 support has been removed from FreeBSD.
r354517:
iwm(4) now supports most Intel 9260, 9460 and 9560 Wi-Fi devices.
r354269:
sqlite3 is updated to sqlite3-3.30.1.
r352668:
cron(8) now supports the -n (suppress mail on succesful run) and -q
(suppress logging of command execution) options in the crontab format.
See the crontab(5) manpage for details.
r352304:
ntpd is no longer by default locked in memory. rlimit memlock 32
or rlimit memlock 0 can be used to restore this behaviour.
r351863:
rc.subr(8) now honors ${name}_env in all rc(8) scripts. Previously,
environment variables set by a user via ${name}_env were ignored
if the service defined a custom *_cmd variable to control the behavior
of the run_rc_command function, e.g., start_cmd, instead of relying on
the variables like command and command_args,
r351770,r352920,r352922,r352923:
dd(1) now supports conv=fsync, conv=fdatasync, oflag=fsync, oflag=sync,
and iflag=fullblock flags, compatible with illumos and GNU.
r351522:
Add kernel-side support for in-kernel Transport Layer Security
(KTLS). KTLS permits using sendfile(2) over sockets using
TLS.
r351397:
WPA is updated from 2.8 to 2.9.
r351361:
Add probes for lockmgr(9) to the lockstat DTrace provider, add
corresponding lockstat(1) events, and document the new probes in
dtrace_lockstat.4.
r351356:
Intel RST is a new 'feature' that remaps NVMe devices from
their normal location to part of the AHCI bar space. This
will eliminate the need to set the BIOS SATA setting from RST
to AHCI causing the nvme drive to be erased before FreeBSD
will see the nvme drive. FreeBSD will now be able to see the
nvme drive now in the default config.
r351201, r351372:
Add a vop_stdioctl() call, so that file systems that do not support
holes will have a trivial implementation of lseek(SEEK_DATA/SEEK_HOLE).
The algorithm appears to be compatible with the POSIX draft and
the implementation in Linux for the case of a file system that
does not support holes. Prior to this patch, lseek(2) would reply
-1 with errno set to ENOTTY for SEEK_DATA/SEEK_HOLE on files in
file systems that do not support holes.
r351372 maps ENOTTY to EINVAL for lseek(SEEK_DATA/SEEK_HOLE) for
any other cases, such as a ENOTTY return from vn_bmap_seekhole().
r350665:
The fuse driver has been renamed to fusefs(5) and been substantially
rewritten. The new driver includes many bug fixes and performance
enhancements, as well as the following user-visible features:
* Optional kernel-side permissions checks (-o default_permissions)
* mknod(2), socket(2), and pipe(2) support
* server side locking with fcntl(2)
* FUSE operations are now interruptible when mounted with -o intr
* server side handling of UTIME_NOW during utimensat(2)
* mount options may be updated with "mount -u"
* fusefs file system may now be exported over NFS
* RLIMIT_FSIZE support
* support for fuse file systems using protocols as old as 7.4
FUSE file system developers should also take note of the following new
features:
* The protocol level has been raised from 7.8 to 7.23
* kqueue support on /dev/fuse
* server-initiated cache invalidation via FUSE_NOTIFY_REPLY
r350471:
gnop(8) can now configure a delay to be applied to read and write
request delays. See the -d, -q and -x parameters.
r350315, r350316:
Adds a Linux compatible copy_file_range(2) syscall.
r350307:
libcap_random(3) has been removed. Applications can use native
APIs to get random data in capability mode.
r349529,r349530:
Add support for using unmapped mbufs with sendfile(2).
r349352:
nand(4) and related components have been removed.
r349349:
The UEFI loader now supports HTTP boot.
r349335:
bhyve(8) now implements a High Definition Audio (HDA) driver, allowing
guests to play to and record audio data from the host.
r349286:
swapon(8) can now erase a swap device immediately before enabling it,
similar to newfs(8)'s -E option. This behaviour can be specified by
adding -E to swapon(8)'s command-line parameters, or by adding the
"trimonce" option to a swap device's /etc/fstab entry.
r347908-r347923:
The following network drivers have been removed: bm(4), cs(4), de(4),
ed(4), ep(4), ex(4), fe(4), pcn(4), sf(4), sn(4), tl(4), tx(4), txp(4),
vx(4), wb(4), xe(4).
r347532:
Wired page accounting has been split into kernel wirings and user
wirings (e.g., by mlock(2)). Kernel wirings no long count towards
the global limit, which is renamed to vm.max_user_wired. bhyve -S
allocates user-wired memory and is now subject to that limit.