kettenis [Tue, 4 Oct 2022 19:41:21 +0000 (19:41 +0000)]
On CPUs that support the Data Independent Timing feature, enable this feature
by default in both the kernel and userland. At this point, this feature has
only been seen on Apple's CPU cores, where turning it on has no measurable
impact on performance. Turning this feature on should help to mitigate
timing side-channel attacks.
ok deraadt@, beck@
miod [Tue, 4 Oct 2022 19:38:20 +0000 (19:38 +0000)]
Repair Apple-specific translation support broken by mistake in 1.88; reported
by Leonardo Moreno
kettenis [Tue, 4 Oct 2022 19:36:20 +0000 (19:36 +0000)]
Add a few more PSTATE bits.
ok deraadt@
kn [Tue, 4 Oct 2022 19:34:44 +0000 (19:34 +0000)]
sh(1) is not make(1), use newlines inside double quotes
sh(1) happily accepts newlines inside double quotes just like in scripts:
$ sh -c "echo foo
echo bar"
foo
bar
So no need to squash things into a single line as usually done inside make
targets where each makefile line is considered its own script unless
continued with trailing backslashes.
OK millert
espie [Tue, 4 Oct 2022 15:31:02 +0000 (15:31 +0000)]
fix an obvious thinko without serious consequences in the display of e
for rsa and friends.
okay tb@
millert [Tue, 4 Oct 2022 13:33:57 +0000 (13:33 +0000)]
Better path handling description, also document tzname, timezone daylight.
Explicitly mention that most programs do not need to call tzset() directly.
OK deraadt@ jmc@ benno@
kn [Tue, 4 Oct 2022 08:03:26 +0000 (08:03 +0000)]
Sort commands in help output, add help to manual
OK millert
Feedback OK jmc
kn [Tue, 4 Oct 2022 07:05:28 +0000 (07:05 +0000)]
Drop cpath promise unless file creation is allowed
OK millert
kn [Tue, 4 Oct 2022 07:03:27 +0000 (07:03 +0000)]
Only print prompt in interactive usage
Scripting tftp(1) makes it non-interactive, yet the prompt is still
printed and may mess up the shell's PS1:
$ echo put nonexistent | tftp localhost
tftp> tftp: open: nonexistent: No such file or directory
tftp> $
The fix seems easy and works as expected for multiple commands as well:
$ echo 'verbose\nput nonexistent' | ./obj/tftp localhost
Verbose mode on.
tftp: open: nonexistent: No such file or directory
$
OK millert
kn [Tue, 4 Oct 2022 07:01:38 +0000 (07:01 +0000)]
Unveil /tftpboot only if needed
Unless -t is used, this directory is not accessed in any way.
OK millert
kn [Tue, 4 Oct 2022 06:52:52 +0000 (06:52 +0000)]
Don't print device when passed as an argument
r1.11 "Don't print device name on failure" made it print unconditionally,
which contradicts what the manual says.
Report + diff from Brin Conway <bconway AT rcesoftware DOT com>, thanks.
From Brian Conway
deraadt [Tue, 4 Oct 2022 02:58:41 +0000 (02:58 +0000)]
sync
jca [Mon, 3 Oct 2022 21:02:02 +0000 (21:02 +0000)]
dapmic(4): support clean shutdown with the power button
Register an interrupt handler and look for nONKEY events.
Mask off all other events. Also clear the FAULT_LOG register at
startup, but don't attempt to acknowledge its contents at runtime. This
lets the user force a shutdown with a long power button press. Some
events in EVENT_* or FAULT_LOG may benefit from more graceful handling.
This makes it possible to cleanly shutdown my Hifive Unmatched using
the power button.
Thanks to kettenis@ who wrote sfgpio(4) to handle dapmic(4) interrupts
on the Unmatched. ok kettenis@
kettenis [Mon, 3 Oct 2022 19:32:22 +0000 (19:32 +0000)]
Reorganize the EFI code a bit. Move the efi.h header from dev/acpi to
dev/efi and rename the arm64 efi.c to efi_machdep.c, preparing the way
for MI EFI code and an amd64 implementation of EFI runtime support.
ok deraadt@, mlarkin@
bluhm [Mon, 3 Oct 2022 16:43:52 +0000 (16:43 +0000)]
System calls should not fail due to temporary memory shortage in
malloc(9) or pool_get(9).
Pass down a wait flag to pru_attach(). During syscall socket(2)
it is ok to wait, this logic was missing for internet pcb. Pfkey
and route sockets were already waiting.
sonewconn() must not wait when called during TCP 3-way handshake.
This logic has been preserved. Unix domain stream socket connect(2)
can wait until the other side has created the socket to accept.
OK mvs@
millert [Mon, 3 Oct 2022 15:34:39 +0000 (15:34 +0000)]
Allow TZ to contain absolutes paths starting with /usr/share/zoneinfo/
Other absolutes paths are still rejected.
bluhm [Mon, 3 Oct 2022 14:10:53 +0000 (14:10 +0000)]
Add a second membar producer into counters_zero(). Now it is
symmetric to counters_read().
OK jmatthew@
jsg [Mon, 3 Oct 2022 12:02:12 +0000 (12:02 +0000)]
drm/amdgpu: don't register a dirty callback for non-atomic
From Alex Deucher
21b0301f2234112fbe5cfc1e13968f0a12f0f2d2 in linux 5.15.y/5.15.71
abbc7a3dafb91b9d4ec56b70ec9a7520f8e13334 in mainline linux
jsg [Mon, 3 Oct 2022 11:59:53 +0000 (11:59 +0000)]
drm/amd/display: Mark dml30's UseMinimumDCFCLK() as noinline for stack usage
From Nathan Chancellor
1c26968caf181add92bbf556e3eb3c14ccbe7fa2 in linux 5.15.y/5.15.71
41012d715d5d7b9751ae84b8fb255e404ac9c5d0 in mainline linux
jsg [Mon, 3 Oct 2022 11:57:51 +0000 (11:57 +0000)]
drm/amd/display: Reduce number of arguments of dml31's CalculateFlipSchedule()
From Nathan Chancellor
492db4ffcff3c3bac1f36bbb100e45c9fe718e34 in linux 5.15.y/5.15.71
21485d3da659b66c37d99071623af83ee1c6733d in mainline linux
jsg [Mon, 3 Oct 2022 11:56:02 +0000 (11:56 +0000)]
drm/amd/display: Reduce number of arguments of dml31's CalculateWatermarksAndDRAMSpeedChangeSupport()
From Nathan Chancellor
9539cfc744933198318eb061ffce87964559ff87 in linux 5.15.y/5.15.71
37934d4118e22bceb80141804391975078f31734 in mainline linux
jsg [Mon, 3 Oct 2022 11:52:57 +0000 (11:52 +0000)]
drm/amd/display: Limit user regamma to a valid value
From Yao Wang1
a541c011181823494c53b15020fae5c3b45f5ec2 in linux 5.15.y/5.15.71
3601d620f22e37740cf73f8278eabf9f2aa19eb7 in mainline linux
jsg [Mon, 3 Oct 2022 11:50:34 +0000 (11:50 +0000)]
drm/amdgpu: use dirty framebuffer helper
From Hamza Mahfooz
33b128f790b6df7e44e4435c6b9387b257637d15 in linux 5.15.y/5.15.71
66f99628eb24409cb8feb5061f78283c8b65f820 in mainline linux
jsg [Mon, 3 Oct 2022 11:48:47 +0000 (11:48 +0000)]
drm/amd/pm: disable BACO entry/exit completely on several sienna cichlid cards
From Guchun Chen
f76d6f309a685d24a32703b4d5c5a71e76f4b793 in linux 5.15.y/5.15.71
7c6fb61a400bf3218c6504cb2d48858f98822c9d in mainline linux
jsg [Mon, 3 Oct 2022 11:46:17 +0000 (11:46 +0000)]
drm/amd/amdgpu: fixing read wrong pf2vf data in SRIOV
From Jingwen Chen
5f270b61ee8b01ab410e3e6ee4f4cc0b32300605 in linux 5.15.y/5.15.71
9a458402fb69bda886aa6cbe067311b6e3d9c52a in mainline linux
jsg [Mon, 3 Oct 2022 11:42:42 +0000 (11:42 +0000)]
drm/amdgpu: make sure to init common IP before gmc
From Alex Deucher
55653c548612d7ffbb21cf05ebff6d14f79cc913 in linux 5.15.y/5.15.71
a8671493d2074950553da3cf07d1be43185ef6c6 in mainline linux
jsg [Mon, 3 Oct 2022 11:39:40 +0000 (11:39 +0000)]
drm/amdgpu: Separate vf2pf work item init from virt data exchange
From Victor Skvortsov
25a90a11036b6fd385ee75536b78659fc481d529 in linux 5.15.y/5.15.71
892deb48269c65376f3eeb5b4c032ff2c2979bd7 in mainline linux
jsg [Mon, 3 Oct 2022 11:30:27 +0000 (11:30 +0000)]
add noinline_for_stack for 5.15.71 drm
jsg [Mon, 3 Oct 2022 10:18:28 +0000 (10:18 +0000)]
drm/amdgpu: move nbio sdma_doorbell_range() into sdma code for vega
From Alex Deucher
8c2bbfb0ded3f41e2985a7856da56f130288d926 in linux 5.15.y/5.15.70
e3163bc8ffdfdb405e10530b140135b2ee487f89 in mainline linux
jsg [Mon, 3 Oct 2022 10:16:12 +0000 (10:16 +0000)]
drm/amdgpu: move nbio ih_doorbell_range() into ih code for vega
From Alex Deucher
0a7d86f156fad2a4420404c1f71c9ee79970ab0c in linux 5.15.y/5.15.70
dc1d85cb790f2091eea074cee24a704b2d6c4a06 in mainline linux
jsg [Mon, 3 Oct 2022 10:13:23 +0000 (10:13 +0000)]
drm/amdgpu: Don't enable LTR if not supported
From Lijo Lazar
dcef16f64969eefb83c8711173fd025611f258f8 in linux 5.15.y/5.15.70
6c20490663553cd7e07d8de8af482012329ab9d6 in mainline linux
jsg [Mon, 3 Oct 2022 10:07:01 +0000 (10:07 +0000)]
define CONFIG_PCIEASPM, add ltr_path to struct pci_dev for 5.15.70 drm
jsg [Mon, 3 Oct 2022 05:39:59 +0000 (05:39 +0000)]
regen
jsg [Mon, 3 Oct 2022 05:39:07 +0000 (05:39 +0000)]
add id for Raphael, Ryzen 7000 GPU (gfx1036, GC 10.3.6?)
kettenis [Sun, 2 Oct 2022 19:00:46 +0000 (19:00 +0000)]
Fetch vendor and product IDs from the keyboard/touchpad controller.
To be used in a future diff to handle machines with touchbar a bit better.
ok robert@
jsing [Sun, 2 Oct 2022 16:40:56 +0000 (16:40 +0000)]
Dynamically link libssl for QUIC regress.
Now that the QUIC API is public, we can stop linking this statically.
jsing [Sun, 2 Oct 2022 16:39:39 +0000 (16:39 +0000)]
Remove unused headers.
jsing [Sun, 2 Oct 2022 16:38:23 +0000 (16:38 +0000)]
Revise for SSL_CTX_INTERNAL and SSL_INTERNAL removal.
jsing [Sun, 2 Oct 2022 16:36:41 +0000 (16:36 +0000)]
Get rid of SSL_CTX_INTERNAL and SSL_INTERNAL.
These are no longer necessary due to SSL_CTX and SSL now being fully
opaque. Merge SSL_CTX_INTERNAL back into SSL_CTX and SSL_INTERNAL back
into SSL.
Prompted by tb@
mpi [Sun, 2 Oct 2022 11:56:43 +0000 (11:56 +0000)]
Remove unused DEBUG_LINE define.
deraadt [Sat, 1 Oct 2022 23:49:38 +0000 (23:49 +0000)]
The syscall table generation awk script was also used by compat layers
in the past, but those compat layers are gone. Remove support for the
"config file"
ok miod millert
jsing [Sat, 1 Oct 2022 16:23:15 +0000 (16:23 +0000)]
Move handshake message handling functions from ssl_both.c to client/server.
Currently, ssl_both.c contains several functions that are used by both the
legacy client and legacy server. This interwines the client and server,
making it harder to make progressive changes. While it does deduplicate
some code, it also ends up with code that is conditioned on s->server and
forces the caller to pass in SSL3_ST_* values.
Move these functions from ssl_both.c into ssl_clnt.c and ssl_srvr.c,
renaming as appropriate and removing the s->server conditionals. Also move
the client and server function prototypes from ssl_locl.h into the .c
files, making them static in the process.
ok tb@
jsg [Fri, 30 Sep 2022 02:56:23 +0000 (02:56 +0000)]
re-enable POOL_DEBUG
ok deraadt@
jmc [Thu, 29 Sep 2022 16:52:31 +0000 (16:52 +0000)]
FILES:
- fix the crazy list width
- since there's only one item, -compact makes no sense
deraadt [Thu, 29 Sep 2022 04:10:27 +0000 (04:10 +0000)]
There no longer is any KVM_ET_* to keep in sync with UVM_ET_*, so
comment can be deleted.
jmc [Wed, 28 Sep 2022 20:27:12 +0000 (20:27 +0000)]
use Fn rather than Nm for swab(); from josiah frentsos
gnezdo [Wed, 28 Sep 2022 16:44:14 +0000 (16:44 +0000)]
Fix incorrect range check for size in setvbuf
From enh AT google.com:
The existing test is wrong for LP64, where size_t has twice as many
relevant bits as int, not just one. (Found by inspection by
rprichard.)
Looks good to deraadt@ and millert@
mbuhl [Wed, 28 Sep 2022 13:21:13 +0000 (13:21 +0000)]
Fix memory corruptions with sysv semaphores due to sleeps in copyin,
copyout and malloc. During a sleep another thread could delete the
semaphore (and possibly allocate another one at the same location
with different permissions) which would lead to an invalid access
after wake up. Therefore check the semaphore pointer, the sequence,
the permissions and some values in seminfo after each sleep.
OK bluhm@
Reported-by: syzbot+60ba811fe2e8a6b0f975@syzkaller.appspotmail.com
nicm [Wed, 28 Sep 2022 07:59:50 +0000 (07:59 +0000)]
Add scroll-top and scroll-bottom commands to scroll so cursor is at top
or bottom. From Anindya Mukherjee, GitHub issue 3334.
nicm [Wed, 28 Sep 2022 07:55:29 +0000 (07:55 +0000)]
Add a -T flag to capture-pane to stop at the last used cell instead of
the full width. Restore the previous behaviour by making it default to
off unless -J is used (the only time it matters). Fixes mosh unit tests;
GitHub issue 3339.
kn [Tue, 27 Sep 2022 13:30:36 +0000 (13:30 +0000)]
complete bootparamd -> rpc.bootparamd
reminded by jmc
kn [Tue, 27 Sep 2022 12:28:25 +0000 (12:28 +0000)]
Avoid escaping inside here documents
The delimiter can be quoted (single or double) to disable parameter, command
and arithmetic expansion inside the here document:
$ cat <<__EOT
echo $(echo foo)
__EOT
echo foo
$ cat <<'__EOT'
echo $(echo foo)
__EOT
echo $(echo foo)
Do the latter to be able to write the here document/file content exactly as
it would end up in output/rc.firsttime, making it easier to read.
To be more consistent and explicit, switch the remaining here documents with
pure plain text (no shell expansion, etc.) to quoted delimiters.
OK millert
kn [Tue, 27 Sep 2022 12:22:29 +0000 (12:22 +0000)]
simpler ftplist[0-9].o.o removal
We read /tmp/i/hosts line-wise to fill /mnt/etc/hosts and remove the tmp
file immediately afterwards, so just skip ftplist entries inside the loop
with a slightly easier to read ksh pattern rather than purge the tmp file
up-front with sed(1).
This is also a tiny bit more robust should the ftplist entries ever be added
with a tab as separator instead of a space and/or an alias since the sed
one-liner hardcodes a single space and expects no alias whereas ksh's read
takes any amount of whitespace between _addr and _hn while not caring about
optional aliases.
Comment is obvious so zap it.
OK millert
kn [Tue, 27 Sep 2022 11:52:29 +0000 (11:52 +0000)]
Zap .Nm bootparamd
Match rpc.{lock,stat}d(8) only having their proper name.
OK deraadt
kn [Tue, 27 Sep 2022 11:48:57 +0000 (11:48 +0000)]
fix passing explicit stage files
This fixes installboot regress on octeon; same diff as
macppc_installboot.c r1.6, powerpc64_installboot r1.7 and
octeon_installboot r1.8.
loongson was the last architecture requiring this fix. I don't have a
machine to test it myself (loongson isn't built anymore, anyway) but given
the same diff works on four other architectures, this should just work.
kn [Tue, 27 Sep 2022 11:42:16 +0000 (11:42 +0000)]
fix passing explicit stage files
This fixes installboot regress on octeon; same diff as
macppc_installboot.c r1.6 and powerpc64_installboot r1.7.
kn [Tue, 27 Sep 2022 11:31:46 +0000 (11:31 +0000)]
fix passing explicit stage files
This fixes installboot regress on powerpc64.
The exact same diff already landed for macppc; efi also has the same fix
for md_init() but without the string handling cleanup that entails.
macppc_installboot.c r1.6 "Fix passing explicit stage files":
Using `stage1' leads to a bit more cleanup since early MI installboot.c
handles `-r', i.e. write_filesystem() no longer has needs to do the
fileprefix() dance itself.
OK gkoehler
jmc [Tue, 27 Sep 2022 05:53:32 +0000 (05:53 +0000)]
- no more /usr/include/objc; confirmed by miod
- add /var/agentx; text from martijn
deraadt [Tue, 27 Sep 2022 03:01:42 +0000 (03:01 +0000)]
typing ^C and seeing "terminated by signal %d" is so ... I don't
have the words
ok florian
deraadt [Tue, 27 Sep 2022 02:39:24 +0000 (02:39 +0000)]
we are now working on 7.2-current
djm [Mon, 26 Sep 2022 22:18:40 +0000 (22:18 +0000)]
openssh-9.1
kettenis [Mon, 26 Sep 2022 15:49:59 +0000 (15:49 +0000)]
For framebuffers that don't start on a page boundary, we need to allow
mapping of all the pages used by the framebuffer, even those that are
only used partially. Adjust the check in simplefb_wsmmap() to allow that.
While there, also make sure we use a (rounded down) page aligned address
base address for the physical address we return.
Fixes X on the 16" Macbook Pro.
ok patrick@, deraadt@
martijn [Mon, 26 Sep 2022 08:48:52 +0000 (08:48 +0000)]
Fix a use after free in case mta_tls_init fails.
Found the hard way by renaud <at> allard <dot> it
OK eric@, gilles@, millert@
sdk [Mon, 26 Sep 2022 06:17:22 +0000 (06:17 +0000)]
Regenerate usbdevs{.h,_data.h} for Wacom One S (CTL-472)
sdk [Mon, 26 Sep 2022 06:14:21 +0000 (06:14 +0000)]
uwacom(4): Support for Wacom One S (CTL-472)
kn [Mon, 26 Sep 2022 00:29:55 +0000 (00:29 +0000)]
Drop incomplete archs lists from wsmoused(8) comment
OK deraadt
kettenis [Mon, 26 Sep 2022 00:20:14 +0000 (00:20 +0000)]
Hide error messages while extracting Apple firmware. Depending on the
model there might be no firmware and we want to avoid confusing users
with WARNING messages about patterns that were not matched.
ok deraadt@
kn [Sun, 25 Sep 2022 22:47:27 +0000 (22:47 +0000)]
Document savecore_flags
OK jmc deraadt
jmc [Sun, 25 Sep 2022 20:54:07 +0000 (20:54 +0000)]
add /etc/rpki/
tweak/ok claudio
stsp [Sun, 25 Sep 2022 08:15:43 +0000 (08:15 +0000)]
Give mfii(4) firmware more time to transition out of UNDEFINED state.
Prevents occasional failure to recover from firmware FAULT state where
the driver gave up too early: mfii0: firmware stuck in state 0
ok deraadt@
millert [Sat, 24 Sep 2022 17:08:32 +0000 (17:08 +0000)]
ascii_load_sockaddr: Remove old IPv6 address parsing for envelope files.
IPv6 addresses have been formatted as "[address]" in envelope files
for years. This was supposed to be removed after the 6.6 release
but got forgotten. Noticed by kn@, OK deraadt@ kn@
claudio [Sat, 24 Sep 2022 16:25:22 +0000 (16:25 +0000)]
rpki-client 8.1
claudio [Sat, 24 Sep 2022 16:24:34 +0000 (16:24 +0000)]
OpenBGPD 7.7
florian [Sat, 24 Sep 2022 16:07:26 +0000 (16:07 +0000)]
There are time zones that have minute offsets, display those
correctly. Pointed out by pjanzen@.
To display the offset, use ISO 8601, as suggested by David Goerger.
While here check if tm->tm_gmtoff changed which probably means that we
moved in or out of daylight savings time.
Input & OK millert, deraadt
jmc [Sat, 24 Sep 2022 15:22:59 +0000 (15:22 +0000)]
macro tweaks; from josiah frentsos
ok deraadt
jmc [Sat, 24 Sep 2022 15:21:21 +0000 (15:21 +0000)]
macro tweaks; from josiah frentsos
ok deraadt
millert [Sat, 24 Sep 2022 14:33:28 +0000 (14:33 +0000)]
Sync with https://github.com/JodaOrg/global-tz
Major changes:
o Palestine DST transitions are now Saturdays at 02:00.
jsg [Sat, 24 Sep 2022 13:30:21 +0000 (13:30 +0000)]
disable POOL_DEBUG for release
ok deraadt@
jsg [Sat, 24 Sep 2022 12:22:31 +0000 (12:22 +0000)]
add SH-B0 0x00000f50 socket 940 family 0Fh Opteron
we already had SH-B0 0x00000f40 socket 754 family 0Fh Athlon 64
SH-B0 is part of amd64_errata_set8[]. Used for:
Errata 89: Potential Deadlock With Locked Transactions
ok deraadt@
claudio [Sat, 24 Sep 2022 11:29:16 +0000 (11:29 +0000)]
Adjust dummy rde_generate_updates() to the new prototype.
bluhm [Fri, 23 Sep 2022 21:33:17 +0000 (21:33 +0000)]
Fix detection of duplicate sticky-address in pf.conf parser.
reported to FreeBSD by Franco Fichtner; from Kristof Provost
florian [Fri, 23 Sep 2022 19:37:23 +0000 (19:37 +0000)]
Sync to libunbound 1.16.3
florian [Fri, 23 Sep 2022 19:24:08 +0000 (19:24 +0000)]
Xr to correct man page; from Josiah Frentsos, thanks!
millert [Fri, 23 Sep 2022 17:29:22 +0000 (17:29 +0000)]
POSIX timzone specs may contain '.' so only reject names containing '../'.
Noted by pjanzen@ with input from deraadt@.
florian [Fri, 23 Sep 2022 16:58:33 +0000 (16:58 +0000)]
Since tzset(3) ignores arbitrary files, we no longer need rpath
and can depend on the /usr/share/zoneinfo bypass.
OK mestre, millert, deraadt
claudio [Fri, 23 Sep 2022 15:50:41 +0000 (15:50 +0000)]
Don't depend on RTLABEL_LEN but instead define our own ROUTELABEL_LEN.
With this bgpd.h no longer depends on net/route.h
OK tb@
claudio [Fri, 23 Sep 2022 15:49:20 +0000 (15:49 +0000)]
Implement a special update generator for add-path send all.
The generic add-path code up_generate_addpath() reevaluates everything
since this is the simplest way to select the announced paths. For add-path
all this is overkill since there is no dependency between prefixes and so
individual prefixes can be handled more efficently.
Extend rde_generate_updates() to pass the current newbest and oldbest
prefixes (for the selected best path) but now also include newpath and
oldpath (which is the prefix that is added/removed/modified).
If newpath or oldpath is set then a single prefix was altered and
up_generate_addpath_all() can just remove or add this prefix.
If newpath and oldpath are NULL than the full list based on newbest
needs to be inserted and any old path/prefix removed in the process.
This improves update generation performance on big route collectors using
add-path all substantially.
OK tb@
sthen [Fri, 23 Sep 2022 14:20:01 +0000 (14:20 +0000)]
merge unbound 1.16.3
krw [Fri, 23 Sep 2022 12:32:50 +0000 (12:32 +0000)]
Don't hardcode disk major device types inside DEBUG block. Use
findblkname() and DISKUNIT(). Allows all block devices, not just
sd* and vnd* to generate useful names in DEBUG output.
Cluebat by deraadt@
aoyama [Fri, 23 Sep 2022 02:35:46 +0000 (02:35 +0000)]
Delete unused variables originally come from mvme88k.
jsg [Fri, 23 Sep 2022 01:25:39 +0000 (01:25 +0000)]
only call printf the first time amd64_errata() is called
it may later be used from a resume path when we don't want to print
same change as amd64 amd64errata.c 1.11 by robert
ok robert@ deraadt@
deraadt [Thu, 22 Sep 2022 17:44:20 +0000 (17:44 +0000)]
remove dup line; from j@bitminer
robert [Thu, 22 Sep 2022 04:57:07 +0000 (04:57 +0000)]
use the always serializing RDTSCP instruction in tsc and usertc if available
tweaks from cheloha@; ok deraadt@, sthen@, cheloha@
robert [Thu, 22 Sep 2022 04:36:37 +0000 (04:36 +0000)]
Call amd64_errata() from cpu_fix_msrs() instead of identifycpu() so that
on resume, the errata is re-applied.
In addition make amd64_errata() print the information about the applied
errata only once for the first CPU.
input from jsg@ and deraadt@, ok deraadt@
tobhe [Wed, 21 Sep 2022 22:32:10 +0000 (22:32 +0000)]
Distinguish between retransmit ok and nothing to retransmit. This makes
sure ikes_retransmit_response events don't also increase the
ikes_msg_rcvd_busy counter.
ok markus@
dtucker [Wed, 21 Sep 2022 22:26:50 +0000 (22:26 +0000)]
Fix typo. From AlexanderStohr via github PR#343.
claudio [Wed, 21 Sep 2022 21:12:03 +0000 (21:12 +0000)]
The values for fib_priority are OS dependent. To help portability move
the RTP_BGP and similar defines all into kroute.c and export them via
kr_default_prio() and kr_check_prio().
OK tb@
millert [Wed, 21 Sep 2022 15:57:49 +0000 (15:57 +0000)]
tzset: ignore TZ if it contains an absolute path or issetugid().
Reading time zone files from user-controlled paths can result in
pledge(2) or unveil(2) violations. We also ignore files that contain
a '.' character to avoid paths containing ".." or hidden files.
Work with and OK deraadt@
tb [Wed, 21 Sep 2022 15:24:45 +0000 (15:24 +0000)]
Tweak symbols test in such a way that it would have caught the recent
Symbols.list mistake: undefine aliases (except _cfb block ciphers which
are aliases for historical reasons). Use -Wl,--no-allow-shlib-undefined.
claudio [Wed, 21 Sep 2022 10:39:17 +0000 (10:39 +0000)]
Adjust pathid_assign() to be much faster in the common case.
Use a per peer path_id_tx to assign to paths received from none add-path
enabled peers. This skips two extra walks of the RIB prefix list and is
a big speed-up when there are many regular sessions. If the session uses
add-path recv then the old way of assigning random path_ids needs to be
used.
With input and OK tb@