openbsd
2 months agonc(1): remove useless function pointer dereferencing
jan [Fri, 2 Aug 2024 21:08:47 +0000 (21:08 +0000)]
nc(1): remove useless function pointer dereferencing

ok tb@

2 months agoFix previous
tb [Fri, 2 Aug 2024 16:02:35 +0000 (16:02 +0000)]
Fix previous

Arguably the want_protocol entries in various of these tests are incorrect
but I'll leave that for another day.

2 months agoAdjust tls regress for protocol parsing fixes
tb [Fri, 2 Aug 2024 15:02:22 +0000 (15:02 +0000)]
Adjust tls regress for protocol parsing fixes

This mostly reverts what was done by beck in Tallinn and adjust tlstest
to add new test cases and now failing connection tests.

2 months agolibtls: fix legacy protocol parsing
tb [Fri, 2 Aug 2024 15:00:01 +0000 (15:00 +0000)]
libtls: fix legacy protocol parsing

Redefining TLS_PROTOCOL_TLSv1_0 and TLS_PROTOCOL_TLSv1_1 to be the same
as TLS_PROTOCOL_TLSv1_2 had undesired side effects, as witnessed in the
accompanying regress tests. The protocol string all:tlsv1.0 would disable
TLSv1.2 (so only enable TLSv1.3) and tlsv1.2:!tlsv1.1 would disable all
protocols.

It makes more sense to ignore any setting of TLSv1.0 and TLSv1.1, so if
you request 'tlsv1.1' you get no protocol, but 'all:!tlsv1.1' will enable
the two supported protocols TLSv1.3 and TLSv1.2.

Restore the defines to their original values and adjust the parsing code
to set/unset them.

Issue reported by Kenjiro Nakayama
Fixes https://github.com/libressl/openbsd/issues/151

with/ok jsing

2 months agoregen
mvs [Fri, 2 Aug 2024 14:35:56 +0000 (14:35 +0000)]
regen

2 months agoPush kernel lock down to sysctl(2).
mvs [Fri, 2 Aug 2024 14:34:45 +0000 (14:34 +0000)]
Push kernel lock down to sysctl(2).

Unlock few obvious immutable or read-only variables from "kern.*" and
"hw.*" paths. Keep the rest variables locked as before, include pages
wiring. Use new sysctl_vs{,un}lock() functions introduced for thar
purpose.

In kern.* path:

 - KERN_OSTYPE, KERN_OSRELEASE, KERN_OSVERSION, KERN_VERSION -
   immutable;
 - KERN_NUMVNODES - read-only access to integer;
 - KERN_MBSTAT - read-only access to per-CPU counters;

In hw.* path:

 - HW_MACHINE, HW_MODEL, HW_NCPUONLINE, HW_PHYSMEM, HW_VENDOR,
   HW_PRODUCT, HW_VERSION, HW_SERIALNO, HW_UUID, HW_PHYSMEM64 -
   immutable;
 - HW_USERMEM and HW_USERMEM64 - `physmem' is immutable, uvmexp.wired
   is mutable but integer; read-only access to localy stored difference
   between `physmem' and uvmexp.wired;
 - `hw_vars' - read-only access to integers; some of them like
   HW_BYTEORDER and HW_PAGESIZE are immutable;

ok bluhm kettenis

2 months agofree class->indexes in CRYPTO_cleanup_all_ex_data()
tb [Fri, 2 Aug 2024 14:02:33 +0000 (14:02 +0000)]
free class->indexes in CRYPTO_cleanup_all_ex_data()

ok jsing

2 months agoBackout previous. It was incompleted and too early.
yasuoka [Fri, 2 Aug 2024 12:43:55 +0000 (12:43 +0000)]
Backout previous.  It was incompleted and too early.

2 months agoRewrite CRYPTO_EX_DATA.
jsing [Fri, 2 Aug 2024 10:48:54 +0000 (10:48 +0000)]
Rewrite CRYPTO_EX_DATA.

CRYPTO_EX_DATA exists as a way to allow an application to attach data to
various components in libcrypto and libssl. The general idea is that there
are various "classes" (e.g. RSA) and an application can get an "index"
(which can have new/dup/free functions provided). The application can then
use the index to store a pointer to some form of data within that class,
for later retrieval.

However, even by OpenSSL standards, this is an insane API. The current
implementation allows for data to be set without calling new, indexes
can be used without allocation, new can be called without actually getting
an index and dup can be called either after new or without new (see regress
and RSA_get_ex_new_index(3)/CRYPTO_set_ex_data(3) for more details). On
top of this, the previous "overhaul" of the code was written to be
infinitely extensible.

For now, the rewrite intends to maintain the existing behaviour - once we
bed this down we can attempt to ratchet the API requirements and require
some sort of sensible sequence. The only intentional change is that there
is now a hard limit on the number of indexes that can be allocated
(previously there was none, relying only on ENOMEM).

ok tb@

2 months agoSwitch off unreferenced power resources in DVACT_POWERDOWN instead of
kettenis [Fri, 2 Aug 2024 09:28:35 +0000 (09:28 +0000)]
Switch off unreferenced power resources in DVACT_POWERDOWN instead of
DVACT_SUSPEND.  Fixes an issue with (un)hibernate where we don't attach
xhci(4) controllers, which would make is turn the xhci(4) off inadvertedly
because its power resource wasn't referenced.

ok mlarkin@

3 months agosync
tb [Fri, 2 Aug 2024 05:00:35 +0000 (05:00 +0000)]
sync

3 months agobump minor after yesterday's symbol addition
tb [Fri, 2 Aug 2024 04:59:55 +0000 (04:59 +0000)]
bump minor after yesterday's symbol addition

requested by kettenis

3 months agoPut exported symbols of libz under our own control
tb [Fri, 2 Aug 2024 04:59:23 +0000 (04:59 +0000)]
Put exported symbols of libz under our own control

This adds a Symbols.list containing the exported symbols like in other
libraries in base. If upstream adds new API this won't necessarily need
a bump on our side.

help/ok kettenis

3 months agogetentropy(2) is in POSIX-2024, so adjust the #include visibility,
guenther [Fri, 2 Aug 2024 01:53:21 +0000 (01:53 +0000)]
getentropy(2) is in POSIX-2024, so adjust the #include visibility,
change the "too much" error to EINVAL, add GETENTROPY_MAX to
<limits.h> (via sys/syslimits.h), and update the manpage.

ok deraadt@

3 months agoAdd qcgpio at fdt.
mglocker [Thu, 1 Aug 2024 21:55:48 +0000 (21:55 +0000)]
Add qcgpio at fdt.

ok kettenis@

3 months agoRun socket splice idle timeout without kernel lock.
bluhm [Thu, 1 Aug 2024 17:19:01 +0000 (17:19 +0000)]
Run socket splice idle timeout without kernel lock.

OK mvs@

3 months agoBump libsndio pkg-config version to 1.10.0
ratchov [Thu, 1 Aug 2024 16:25:34 +0000 (16:25 +0000)]
Bump libsndio pkg-config version to 1.10.0

ok tb@

3 months agobump posix spec from 2008 -> 2024;
jmc [Thu, 1 Aug 2024 14:44:34 +0000 (14:44 +0000)]
bump posix spec from 2008 -> 2024;
no documented changes.

3 months agosndiod: Fix insufficent check of input from clients
ratchov [Thu, 1 Aug 2024 14:36:27 +0000 (14:36 +0000)]
sndiod: Fix insufficent check of input from clients

Fixes possible sndiod(8) crashes caused by a global table overread
triggered by the client.

Found and analysed by Henry Ford <henryfordkjv at gmail.com>, thanks!

ok blumh, help from millert, mlarkin

3 months agobump posix spec 2008 -> 2024;
jmc [Thu, 1 Aug 2024 14:30:17 +0000 (14:30 +0000)]
bump posix spec 2008 -> 2024;

no change noted from 2008 spec, but i've added -h to the list
of extensions. that was an omission from our 2008 notes,
rather than a change in the spec.

3 months agobump posix spec 2008 -> 2024;
jmc [Thu, 1 Aug 2024 14:08:38 +0000 (14:08 +0000)]
bump posix spec 2008 -> 2024;

3 months agobump posix spec 2008 -> 2024;
jmc [Thu, 1 Aug 2024 14:08:07 +0000 (14:08 +0000)]
bump posix spec 2008 -> 2024;

3 months agoAdd device tree mapping for Samsung Galaxy Book4 Edge.
mglocker [Thu, 1 Aug 2024 11:53:03 +0000 (11:53 +0000)]
Add device tree mapping for Samsung Galaxy Book4 Edge.

ok kettenis@

3 months agovirtio: Fix dmamap_sync calls
sf [Thu, 1 Aug 2024 11:13:19 +0000 (11:13 +0000)]
virtio: Fix dmamap_sync calls

Add some missing bus_dmamap_sync calls, noticed with SEV and based on
an earlier diff by hshoexer@.

Some of the required syncing is done in virtio_check_vq(). Make sure
to use that function instead of calling call the virtqueue done
function directly from device specific drivers.

For viogpu this means that we cannot poll with virtio_dequeue() but
must use virtio_check_vq() instead. To make this more clear, rename
viogpu_vq_wait() into viogpu_vq_done(). While there, set the DRIVER_OK
flag even earlier. It must be set before using any virtqueue.

ok kettenis@

3 months agoRemove a comment that was committed by accident
tb [Thu, 1 Aug 2024 06:04:22 +0000 (06:04 +0000)]
Remove a comment that was committed by accident

3 months agosync with src
tb [Thu, 1 Aug 2024 04:03:10 +0000 (04:03 +0000)]
sync with src

3 months agosync with upstream
tb [Thu, 1 Aug 2024 04:02:26 +0000 (04:02 +0000)]
sync with upstream

This adds a new public API, deflateUsed(), so is technically a minor bump.
Nothing will be using this anytime soon, so no shared library bump.

discussed with deraadt during c2k24

3 months agoPossible to read the core file of a process.
asou [Thu, 1 Aug 2024 01:43:08 +0000 (01:43 +0000)]
Possible to read the core file of a process.

3 months agoUse arc4random_uniform(n) instead of "% n".
yasuoka [Thu, 1 Aug 2024 00:58:14 +0000 (00:58 +0000)]
Use arc4random_uniform(n) instead of "% n".

3 months agoChange the routing entry of the pool addresses from blackhole to
yasuoka [Thu, 1 Aug 2024 00:05:16 +0000 (00:05 +0000)]
Change the routing entry of the pool addresses from blackhole to
reject.  This is better because the routing cache can be purged by the
reject.

3 months agobump posix spec 2008 -> 2024;
jmc [Wed, 31 Jul 2024 17:09:23 +0000 (17:09 +0000)]
bump posix spec 2008 -> 2024;

3 months agoadd a random amount of time (up to 4 seconds) to the grace login time.
dlg [Wed, 31 Jul 2024 12:00:18 +0000 (12:00 +0000)]
add a random amount of time (up to 4 seconds) to the grace login time.

ok deraadt@ djm@

3 months agoAdd ufshci(4) at fdt support.
mglocker [Wed, 31 Jul 2024 10:07:33 +0000 (10:07 +0000)]
Add ufshci(4) at fdt support.

This lets me boot the Samsung Galaxy Book4 Edge in DT mode with ufshci(4),
based on a WIP DTB file.

ok kettenis@

3 months agoquote the book title (%T) reference, to prevent formatters (in this case
jmc [Wed, 31 Jul 2024 05:36:13 +0000 (05:36 +0000)]
quote the book title (%T) reference, to prevent formatters (in this case
groff) stumbling over literal words that match macro names;

3 months agoMake the Samsung Galaxy Book4 Edge (x1e80100) boot in ACPI mode;
mglocker [Tue, 30 Jul 2024 19:47:05 +0000 (19:47 +0000)]
Make the Samsung Galaxy Book4 Edge (x1e80100) boot in ACPI mode;
Until now the kernel did panic during early boot because we couldn't
respond to the battery requests sent to the 0xa1 "bus".  This (dummy)
driver registers a 0xa1 region space very early, and responds with a
static zero value to satisfy the battery requests on 0xa1, which let us
boot successfully to multi-user.

In the future this machine should be operated in FDT mode, and an improved
AML parsing interface might make this workaround obsolete also for RAMDISK.

Help and OK kettenis@, deraadt@

3 months agobump posix ref 2008 -> 2024;
jmc [Tue, 30 Jul 2024 19:16:49 +0000 (19:16 +0000)]
bump posix ref 2008 -> 2024;

3 months agofold the "timespec" info into the options list (since "time_arg" is already
jmc [Tue, 30 Jul 2024 13:58:02 +0000 (13:58 +0000)]
fold the "timespec" info into the options list (since "time_arg" is already
there). move the information of "time_arg" matching touch -t to the "time_arg"
description, instead of "timespec".

ok millert

3 months agobump posix spec from 2008 to 2024; ok millert
jmc [Tue, 30 Jul 2024 13:55:11 +0000 (13:55 +0000)]
bump posix spec from 2008 to 2024; ok millert

3 months agoExports the statistics when PIPEXDSESSION. Found by ymatsui at iij.
yasuoka [Tue, 30 Jul 2024 13:41:15 +0000 (13:41 +0000)]
Exports the statistics when PIPEXDSESSION.  Found by ymatsui at iij.

ok mvs

3 months agosched_yield() is not strong enough to overflow the recv buffer on some
claudio [Tue, 30 Jul 2024 13:28:27 +0000 (13:28 +0000)]
sched_yield() is not strong enough to overflow the recv buffer on some
systems. Use a proper sleep using usleep(100) instead.

3 months agoOn arm64, check whether the CPU has the BT feature to determine the expected
kettenis [Tue, 30 Jul 2024 09:07:00 +0000 (09:07 +0000)]
On arm64, check whether the CPU has the BT feature to determine the expected
outcome of the test.

3 months agoPopulate most of the remaining hwcap and hwcap2 flags based on the detected
kettenis [Tue, 30 Jul 2024 08:59:33 +0000 (08:59 +0000)]
Populate most of the remaining hwcap and hwcap2 flags based on the detected
CPU features.

ok naddy@

3 months agoRecognize the features defines for POSIX 1003.1-2024, giving new values
guenther [Tue, 30 Jul 2024 05:57:31 +0000 (05:57 +0000)]
Recognize the features defines for POSIX 1003.1-2024, giving new values
for __POSIX_VISIBLE and __XPG_VISIBLE.  ok millert@

3 months agobump posix spec from 2008 -> 2024:
jmc [Mon, 29 Jul 2024 20:25:55 +0000 (20:25 +0000)]
bump posix spec from 2008 -> 2024:
- "now" is a permissable timespec
- split the at and batch notes in STANDARDS
- note batch behaviour differences
- note default queue name differences

ok millert

3 months agoSet the target ACPI to S5 when we're powering down the machine. Prevents
kettenis [Mon, 29 Jul 2024 18:43:11 +0000 (18:43 +0000)]
Set the target ACPI to S5 when we're powering down the machine.  Prevents
us from trying to put devices into the D3 power state which some hardware
doesn't appreciate.

ok ratchov@, deraadt@

3 months agoMove the signal related kqueue filters to kern_event.c.
claudio [Mon, 29 Jul 2024 12:42:53 +0000 (12:42 +0000)]
Move the signal related kqueue filters to kern_event.c.

Since proc and signal filters share the same klist it makes sense
to keep them together.
OK mvs@

3 months agoUse shared net lock instead of exclusive when frag6 calls icmp6_error().
bluhm [Mon, 29 Jul 2024 12:41:30 +0000 (12:41 +0000)]
Use shared net lock instead of exclusive when frag6 calls icmp6_error().

OK mvs@ a while ago as part of a larger diff

3 months agoadd SBZ
patrick [Mon, 29 Jul 2024 12:32:02 +0000 (12:32 +0000)]
add SBZ

3 months agoRemove `sb_sel' from sobuf_print() output, no sense to print "...".
mvs [Mon, 29 Jul 2024 10:35:22 +0000 (10:35 +0000)]
Remove `sb_sel' from sobuf_print() output, no sense to print "...".

ok bluhm

3 months agoSync with sys/proc.h after P_CONTINUED -> PS_CONTINUED change.
claudio [Mon, 29 Jul 2024 09:50:30 +0000 (09:50 +0000)]
Sync with sys/proc.h after P_CONTINUED -> PS_CONTINUED change.
OK mpi@

3 months agoReplace per thread P_CONTINUED with per process PS_CONTINUED flag
claudio [Mon, 29 Jul 2024 09:49:49 +0000 (09:49 +0000)]
Replace per thread P_CONTINUED with per process PS_CONTINUED flag

dowait6() can only look at per process state so switch this over.
Right now SIGCONT handling in ptsignal is recursive and not quite
right but this is a step in the right direction. It fixes dowait6()
handling for multithreaded processes where the main thread exited.

OK mpi@

3 months agotypo: regresss -> regress
tb [Mon, 29 Jul 2024 06:05:31 +0000 (06:05 +0000)]
typo: regresss -> regress

3 months agodrm/amdgpu: Fix signedness bug in sdma_v4_0_process_trap_irq()
jsg [Mon, 29 Jul 2024 01:26:15 +0000 (01:26 +0000)]
drm/amdgpu: Fix signedness bug in sdma_v4_0_process_trap_irq()

From Dan Carpenter
4edb0a84e6b32e75dc9bd6dd085b2c2ff19ec287 in linux-6.6.y/6.6.43
6769a23697f17f9bf9365ca8ed62fe37e361a05a in mainline linux

3 months agoregen
jsg [Mon, 29 Jul 2024 01:18:32 +0000 (01:18 +0000)]
regen

3 months agoadd "Strix Point" APU device id, and another Navi 31
jsg [Mon, 29 Jul 2024 01:17:53 +0000 (01:17 +0000)]
add "Strix Point" APU device id, and another Navi 31

150e rev c1 is Radeon 880M/890M
150e rev c4 is Radeon 880M/890M
150e rev c6 is Radeon 890M
744a rev 00 is Radeon Pro W7900 Dual Slot

found in AMD Software: Adrenalin Edition 24.7.1

3 months agoThe dash must not come first in the getopt(3) string
kn [Sun, 28 Jul 2024 21:44:42 +0000 (21:44 +0000)]
The dash must not come first in the getopt(3) string

Broken/sorted in previous -u addition, fix/move it to the end.
By tb

3 months agopwd_mkdb: limit db entries to _PW_BUF_LEN to match libc
millert [Sun, 28 Jul 2024 19:13:26 +0000 (19:13 +0000)]
pwd_mkdb: limit db entries to _PW_BUF_LEN to match libc

Otherwise, it is possible to create a passwd(5) entry that is too
large for getpwent(3), which ignores database entries larger than
_PW_BUF_LEN.  This adds a check in db_store() so that we do not
store an entry larger than getpwent(3) can read.  Callers of
pwd_mkdb(8), typically via pw_mkdb(3), already check for failure.

In most cases, the checks in chpass(1) will prevent a user from
creating an entry that is too large by changing their gecos field.
However, it is only when storing the db record that we know the
true size.

OK deraadt@

3 months agoblock SIGHUP in the same places where SIGCHLD is blocked to protect
deraadt [Sun, 28 Jul 2024 15:31:22 +0000 (15:31 +0000)]
block SIGHUP in the same places where SIGCHLD is blocked to protect
the process lists, because the SIGHUP handler looks at them (and
it is very difficult to rewrite the that handler a different way)
ok millert

3 months agoSupport "-u name" to remove variable from environment
kn [Sun, 28 Jul 2024 10:08:44 +0000 (10:08 +0000)]
Support "-u name" to remove variable from environment

OK aisha millert
Feedback jmc

3 months agoAllow for another TEMPerHUM version. From Sebastiaan Indesteege.
miod [Sat, 27 Jul 2024 17:31:49 +0000 (17:31 +0000)]
Allow for another TEMPerHUM version. From Sebastiaan Indesteege.

3 months agoUse ".arch_extension fp" for the signal trampoline since it saves and
kettenis [Sat, 27 Jul 2024 13:31:26 +0000 (13:31 +0000)]
Use ".arch_extension fp" for the signal trampoline since it saves and
restores the FPU registers (running in userland).  And add an
".arch_extension nofp" afterwards such that the remaining kernel code
can't touch the FPU.

Needed to build with clang18

ok jsg@

3 months agosync
deraadt [Sat, 27 Jul 2024 03:26:12 +0000 (03:26 +0000)]
sync

3 months agoshorten strings by removing pointless information
deraadt [Sat, 27 Jul 2024 03:26:04 +0000 (03:26 +0000)]
shorten strings by removing pointless information
ok jsg

3 months agoSort and regularize all the ktrstruct() macros
guenther [Sat, 27 Jul 2024 02:10:26 +0000 (02:10 +0000)]
Sort and regularize all the ktrstruct() macros

ok deraadt@

3 months agoTrace struct itimerval
guenther [Fri, 26 Jul 2024 19:16:31 +0000 (19:16 +0000)]
Trace struct itimerval

ok deraadt@ claudio@

3 months agoOn AMD vmm(4) set SVM_INTERCEPT_INVLPGA in intercept1.
bluhm [Fri, 26 Jul 2024 15:59:04 +0000 (15:59 +0000)]
On AMD vmm(4) set SVM_INTERCEPT_INVLPGA in intercept1.

According to the AMD64 Architecture Programmer's Manual volume 2
the intercept SVM_INTERCEPT_INVLPGA needs to be set in vmcb.intercept1
(vector 3, offest 00Ch) instead of intercept2 (vector 4, offset
010h).  SVM_INTERCEPT_INVLPGA is bit 26, so before vcpu_reset_regs_svm()
was actually setting an intercept for CR10, which does not exist.

from hshoexer@; OK mlarkin@

3 months agoMark ipsecflowinfo immutable.
yasuoka [Fri, 26 Jul 2024 15:51:09 +0000 (15:51 +0000)]
Mark ipsecflowinfo immutable.

ok mvs

3 months agoIn pipex_l2tp_input(), check if ipsecflowinfo is not changed instead
yasuoka [Fri, 26 Jul 2024 15:45:31 +0000 (15:45 +0000)]
In pipex_l2tp_input(), check if ipsecflowinfo is not changed instead
of updating it blindly.

ok mvs

3 months agodocument the reduced logingrace penalty
naddy [Fri, 26 Jul 2024 15:24:49 +0000 (15:24 +0000)]
document the reduced logingrace penalty

3 months agoRun UDP input on multiple CPU in parallel.
bluhm [Fri, 26 Jul 2024 14:38:20 +0000 (14:38 +0000)]
Run UDP input on multiple CPU in parallel.

The socket layer of UDP has been made fully MP safe.  UDP output
is MP safe for a while.  mvs@ has fixed the missing pieces in socket
splicing recently.  This means that complete UDP stack can be
processed by multiple threads now.  Activate multi processing for
udp_input() when called with IPv4 or IPv6 packets.

Usually IP processing runs on multiple softnet threads with shared
net lock.  From there local packets are queued and processed by one
thread with exclusive net lock.  If the PR_MPINPUT flag is set,
protocol input is called directly from IP input on multiple threads,
with shared net lock and no additional queueing.

tested by Hrvoje Popovski; OK mvs@

3 months agoMark X509at_get_attr{,_count}() and X509at_delete_attr() as unused
tb [Fri, 26 Jul 2024 13:34:56 +0000 (13:34 +0000)]
Mark X509at_get_attr{,_count}() and X509at_delete_attr() as unused

ok jsing

3 months agoDisable X509at_get_attr{,_count}() and X509at_delete_attr()
tb [Fri, 26 Jul 2024 13:33:39 +0000 (13:33 +0000)]
Disable X509at_get_attr{,_count}() and X509at_delete_attr()

These are (not so) thin wrappers around the stack API and only make
things unreadable by adding an unneccesary layer of indirection and
repeating checks already present in the stack API. X509at_delete_attr()
is a masterpiece.

ok jsing

3 months agoInline last user of X509at_get_attr()
tb [Fri, 26 Jul 2024 13:30:40 +0000 (13:30 +0000)]
Inline last user of X509at_get_attr()

ok jsing

3 months agoInline trivial X509at_* calls in x509_req
tb [Fri, 26 Jul 2024 13:24:39 +0000 (13:24 +0000)]
Inline trivial X509at_* calls in x509_req

ok jsing

3 months agoInline a few trivial X509at_* calls in cms
tb [Fri, 26 Jul 2024 13:23:52 +0000 (13:23 +0000)]
Inline a few trivial X509at_* calls in cms

ok jsing

3 months agovirtio: add/fix feature bits
sf [Fri, 26 Jul 2024 07:55:23 +0000 (07:55 +0000)]
virtio: add/fix feature bits

There was a off-by-one in unused vioblk feature defines. Fix this.

Add missing feature bits from virtio 1.2 so that they are displayed
with VIRTIO_DEBUG.

ok jan@

3 months agovio: Don't request csum offload if not negotiated
sf [Fri, 26 Jul 2024 06:29:01 +0000 (06:29 +0000)]
vio: Don't request csum offload if not negotiated

The standard says "A driver MUST NOT enable an offload for which the
appropriate feature has not been negotiated."

ok jan@

3 months ago- document "ca" in the action ... relay section
jmc [Fri, 26 Jul 2024 06:24:52 +0000 (06:24 +0000)]
- document "ca" in the action ... relay section
from philipp plus openbsd at bureaucracy de

- within the "ca" section, refer to "listen on" and "action ... relay",
which is the actual syntax

ok op

3 months agodrm/radeon: check bo_va->bo is non-NULL before using it
jsg [Fri, 26 Jul 2024 03:42:02 +0000 (03:42 +0000)]
drm/radeon: check bo_va->bo is non-NULL before using it

From Pierre-Eric Pelloux-Prayer
f13c96e0e325a057c03f8a47734adb360e112efe in linux-6.6.y/6.6.42
6fb15dcbcf4f212930350eaee174bb60ed40a536 in mainline linux

3 months agoPut opening brace on correct line
tb [Fri, 26 Jul 2024 03:40:43 +0000 (03:40 +0000)]
Put opening brace on correct line

3 months agodrm/amd/display: Fix refresh rate range for some panel
jsg [Fri, 26 Jul 2024 03:39:34 +0000 (03:39 +0000)]
drm/amd/display: Fix refresh rate range for some panel

From Tom Chung
607731f315d15558696e9b542121d40411df37fd in linux-6.6.y/6.6.42
9ef1548aeaa8858e7aee2152bf95cc71cdcd6dff in mainline linux

3 months agodrm/amd/display: Account for cursor prefetch BW in DML1 mode support
jsg [Fri, 26 Jul 2024 03:37:12 +0000 (03:37 +0000)]
drm/amd/display: Account for cursor prefetch BW in DML1 mode support

From Alvin Lee
6ecf239f71ef7f6e9f8ea46b9c7ef4df2afd96c9 in linux-6.6.y/6.6.42
074b3a886713f69d98d30bb348b1e4cb3ce52b22 in mainline linux

3 months agodrm/amd/display: Add refresh rate range check
jsg [Fri, 26 Jul 2024 03:34:59 +0000 (03:34 +0000)]
drm/amd/display: Add refresh rate range check

From Tom Chung
531e7852fb62bfc6a15cc1df85f62720f704d537 in linux-6.6.y/6.6.42
74ad26b36d303ac233eccadc5c3a8d7ee4709f31 in mainline linux

3 months agodrm/amdgpu: Indicate CU havest info to CP
jsg [Fri, 26 Jul 2024 03:32:20 +0000 (03:32 +0000)]
drm/amdgpu: Indicate CU havest info to CP

From Harish Kasiviswanathan
86a6a3964f600d458412295ac8d705b6b9a8efdb in linux-6.6.y/6.6.42
49c9ffabde555c841392858d8b9e6cf58998a50c in mainline linux

3 months agodrm: panel-orientation-quirks: Add quirk for Aya Neo KUN
jsg [Fri, 26 Jul 2024 03:29:47 +0000 (03:29 +0000)]
drm: panel-orientation-quirks: Add quirk for Aya Neo KUN

From Tobias Jakobi
97f71343f794c6ecbb3a2d55b4a04464e924b139 in linux-6.6.y/6.6.42
f74fb5df429ebc6a614dc5aa9e44d7194d402e5a in mainline linux

3 months agosync
deraadt [Fri, 26 Jul 2024 01:38:59 +0000 (01:38 +0000)]
sync

3 months agouse the fp target attribute with fpu_save()/fpu_load()
jsg [Fri, 26 Jul 2024 00:23:57 +0000 (00:23 +0000)]
use the fp target attribute with fpu_save()/fpu_load()

otherwise clang 18 errors when inline assembly uses floating point
registers with nofp in -march

ok kettenis@

3 months agoreduce logingrace penalty.
djm [Thu, 25 Jul 2024 23:44:01 +0000 (23:44 +0000)]
reduce logingrace penalty.

A single forgotton login that times out should be below the penalty
threshold.

ok deraadt/claudio

3 months agoFix proxy multiplexing (-O proxy) bug
djm [Thu, 25 Jul 2024 22:40:08 +0000 (22:40 +0000)]
Fix proxy multiplexing (-O proxy) bug

If a mux started with ControlPersist then later has a forwarding added using
mux proxy connection and the forwarding was used, then when the mux proxy
session terminates, the mux master process will send a channel close to the
server with a bad channel ID and crash the connection.

This was caused by my stupidly reusing c->remote_id for mux channel
associations when I should have just added another member to struct channel.

ok markus@

3 months agoExplicitly request for battery status updates as the firmware doesn't seem
kettenis [Thu, 25 Jul 2024 20:21:40 +0000 (20:21 +0000)]
Explicitly request for battery status updates as the firmware doesn't seem
to send us regular updates itself.  Estamate the remaining battery (charge)
time by using the reported rate.

ok phessler@

3 months agodocument deflateUsed()
tb [Thu, 25 Jul 2024 17:29:51 +0000 (17:29 +0000)]
document deflateUsed()

text mostly from upstream, the function will soon be added to libz.

ok jmc

3 months agofix a small markup mistake (missing newline)
tb [Thu, 25 Jul 2024 17:25:59 +0000 (17:25 +0000)]
fix a small markup mistake (missing newline)

ok jmc (as part of a larger diff)

3 months agomention MODPY_PYBUILD=pdm
sthen [Thu, 25 Jul 2024 15:36:56 +0000 (15:36 +0000)]
mention MODPY_PYBUILD=pdm

3 months agofix double space; from crystal kolipe
jmc [Thu, 25 Jul 2024 13:40:55 +0000 (13:40 +0000)]
fix double space; from crystal kolipe

3 months agoSilence a false positive warning for gcc
tb [Thu, 25 Jul 2024 08:44:39 +0000 (08:44 +0000)]
Silence a false positive warning for gcc

The code path gcc is whining about ensures that the struct auth in
question is initialized, but the pile of garbage that is gcc's use
of uninitialized warnings can't figure that one out. Enough time on
this was wasted during the last few releases that silencing gcc with
annoying workarounds may be the lesser evil.

ok claudio

3 months agovirtio: Allow more verbose debugging
sf [Thu, 25 Jul 2024 08:35:40 +0000 (08:35 +0000)]
virtio: Allow more verbose debugging

If VIRTIO_DEBUG is set to 2, dump the whole virtqueues.

3 months agoFix test23. It assumed the place of the attribute is unchanged. But now,
yasuoka [Thu, 25 Jul 2024 07:24:40 +0000 (07:24 +0000)]
Fix test23.  It assumed the place of the attribute is unchanged.  But now,
the message-authenticator is located at first always.  found anton

3 months agosync
miod [Thu, 25 Jul 2024 05:33:06 +0000 (05:33 +0000)]
sync

3 months agoIf the CPU cores implement FEAT_IDST, emulate access to the CPU ID
kettenis [Wed, 24 Jul 2024 21:24:18 +0000 (21:24 +0000)]
If the CPU cores implement FEAT_IDST, emulate access to the CPU ID
registers from userland and set HWCAP_CPUID.  This will allow detection
of features to be introduced into the architecture in the future without
allocating new HWCAP_xxx or HWCAP2_xxx bits.  We provide the same
sanitized view of the CPU ID registers as is currently available through
sysctl(2).

Note that this introduces an unconditional read of ID_AA64MMFR2_EL1.  This
is known to cause problems on older versions of QEMU.  If this turns out
to be a problem in cases where updating QEMU is not an option, we'll have
to implement a workaround.

Also note that since we don't emulate the CPU ID registers on older core,
this means that microarchitectural optimizations keyed of reads of MIDR_EL1
are not possible on OpenBSD.  I don't think that is a real problem.

ok jca@

3 months agovmm(4): drop checks for EPT or RVI in some functions.
dv [Wed, 24 Jul 2024 21:04:12 +0000 (21:04 +0000)]
vmm(4): drop checks for EPT or RVI in some functions.

vmm(4) only supports using nested paging on Intel or AMD and no
longer supports HLAT or shadow paging.

ok mlarkin@