openbsd
2 months agoSupport describing ABI changes for static libraries too.
guenther [Thu, 15 Aug 2024 01:25:13 +0000 (01:25 +0000)]
Support describing ABI changes for static libraries too.
Try the -S option

2 months agoadapt to EVP_PKEY conversion
djm [Thu, 15 Aug 2024 00:52:23 +0000 (00:52 +0000)]
adapt to EVP_PKEY conversion

2 months agoConvert RSA and ECDSA key to the libcrypto EVP_PKEY API. DSA remains
djm [Thu, 15 Aug 2024 00:51:51 +0000 (00:51 +0000)]
Convert RSA and ECDSA key to the libcrypto EVP_PKEY API. DSA remains
unconverted as it will be removed within six months.

Based on patches originally from Dmitry Belyavskiy, but significantly
reworked based on feedback from Bob Beck, Joel Sing and especially
Theo Buehler (apologies to anyone I've missed).

ok tb@

2 months ago'newercnt' no longer does anything that 'refcnt' doesn't; eliminate
guenther [Thu, 15 Aug 2024 00:47:44 +0000 (00:47 +0000)]
'newercnt' no longer does anything that 'refcnt' doesn't; eliminate
the former in favor of the latter.

ok millert@

2 months agoAdd 'bgpctl show rib filtered' support.
claudio [Wed, 14 Aug 2024 19:10:51 +0000 (19:10 +0000)]
Add 'bgpctl show rib filtered' support.
OK tb@

2 months agoIntroduce 'rde rib Loc-RIB include filtered' a feature that includes
claudio [Wed, 14 Aug 2024 19:09:51 +0000 (19:09 +0000)]
Introduce 'rde rib Loc-RIB include filtered' a feature that includes
filtered prefixes in the Loc-RIB

This includes filtered prefixes into the Loc-RIB but they are marked
ineligible so nothing will select them but it is possible to show them
in bgpctl. So 'bppctl show rib filtered' will return all prefixes filtered
out by the input filters.

OK tb@

2 months agoImplement bounce buffering for AMD SEV in amd64 bus dma.
bluhm [Wed, 14 Aug 2024 18:31:33 +0000 (18:31 +0000)]
Implement bounce buffering for AMD SEV in amd64 bus dma.

When running as SEV guest, as indicated by variable cpu_sev_guestmode,
allocate additional pages for each segment on dma map creation.
These pages are mapped with the PMAP_NOCRYPT attribute, i.e. the
crypt bit is not set in the PTE.  Thus, these pages are shared with
the hypervisor.
When the map is loaded with actual pages, the address in the
descriptor is replaced by the corresponding bounce buffer.  Using
bus_dmamap_sync(), data is copied from the encrypted pages used by
guest drivers to the unencrypted bounce buffers shared with the
hypervisor, and vice versa.
If the kernel is not running in SEV guest mode, which means as
normal host or non-SEV guest, no bounce buffers are used.

from hshoexer@; based on ancient code of mickey@; OK kettenis@

2 months agoPush kernel lock down to net_sysctl().
mvs [Wed, 14 Aug 2024 17:52:47 +0000 (17:52 +0000)]
Push kernel lock down to net_sysctl().

All except PF_MPLS paths are mp-safe:
- net_link_sysctl() and following net_ifiq_sysctl() only return
  EOPNOTSUPP;
- uipc_sysctl() - mp-safe atomic access to integers;
- bpf_sysctl() - mp-safe atomic access to integers;
- pflow_sysctl() - returns statistics from per-CPU counters;
- pipex_sysctl() - mp-safe atomic access to integer;

Push kernel lock down to mpls_sysctl(). sysctl_int_bounded() do copying
with local variable, so context switch is safe. No need to wire memory
or take `sysctl_lock' rwlock(9).

Keep protocols locked as they was include pages wiring. Copying will not
sleep - no network slowdown while doing it with net lock held.

ok bluhm

2 months agoresdebug and no_tld_query are unused
florian [Wed, 14 Aug 2024 17:38:57 +0000 (17:38 +0000)]
resdebug and no_tld_query are unused

2 months agosortlist is unused
florian [Wed, 14 Aug 2024 17:37:40 +0000 (17:37 +0000)]
sortlist is unused

2 months agolwserver is unused, whatever that is / was.
florian [Wed, 14 Aug 2024 17:36:48 +0000 (17:36 +0000)]
lwserver is unused, whatever that is / was.

2 months agoReorder calloc arguments
tobias [Wed, 14 Aug 2024 15:42:18 +0000 (15:42 +0000)]
Reorder calloc arguments

The first argument should be the amount, the second argument should be the
element size. Fixing this also silences some gcc compiler warnings for
portable.

Spotted with Benny Baumann (BenBE at geshi dot org).

ok djm@

2 months agoExtend sshbuf validation
tobias [Wed, 14 Aug 2024 15:40:30 +0000 (15:40 +0000)]
Extend sshbuf validation

Multiple sshbuf structs can be linked through a parent/child relationship.
Make sure that a single sshbuf cannot be its own parent. If this would ever
happen, it would result in reference counting issues.

This is a cheap way of testing this with very little overhead. It does not
detect A->B->A linkages though for performance reason and the fact that it
takes a programming error for this to occur anyway.

Authored with Benny Baumann (BenBE at geshi dot org).

ok djm@

2 months agoUse freezero for better readability
tobias [Wed, 14 Aug 2024 15:37:11 +0000 (15:37 +0000)]
Use freezero for better readability

It has the same meaning as the current pair of calling explicit_bzero
and free. Spotted with Benny Baumann (BenBE at geshi dot org).

ok djm@

2 months agoFix typo in comment
tobias [Wed, 14 Aug 2024 15:35:23 +0000 (15:35 +0000)]
Fix typo in comment

Spotted with Benny Baumann (BenBE at geshi dot org).

ok djm@

2 months agomention apple machines needing activation after erase
jsg [Wed, 14 Aug 2024 15:34:39 +0000 (15:34 +0000)]
mention apple machines needing activation after erase
ok patrick@ kettenis@ krw@

2 months agoIntroduce qwz(4), a work-in-progress port of the Linux ath12k driver.
patrick [Wed, 14 Aug 2024 14:40:45 +0000 (14:40 +0000)]
Introduce qwz(4), a work-in-progress port of the Linux ath12k driver.

This driver is not working yet, it's a copy of qwx(4) which bit-by-bit will be
adjusted to work on the newer generation of chips.  Even though this is only a
minor bump over the previous generation, the changes are large enough that
bringup, debugging and long-term maintenance would suffer trying to squash them
into a single driver.  This can be reconsidered once we have reached a stable
state.

2 months agoMake sysctl_int() and sysctl_int_lower() mp-safe and unlock KERN_HOSTID.
mvs [Wed, 14 Aug 2024 13:54:08 +0000 (13:54 +0000)]
Make sysctl_int() and sysctl_int_lower() mp-safe and unlock KERN_HOSTID.

The only difference between sysctl_int() and sysctl_int_bounded()
is the range check, so sysctl_int() is just sysctl_int_bounded(...,
INT_MIN, INT_MAX). sysctl_int() is not the fast path, so this useless
check is not significant.

Mp-safe sysctl_int() is meaningless for sysctl_int_lower(), so rework it
in the sysctl_int_bounded() style. This time all affected paths are
kernel locked, but this doesn't make sysctl_int_lower() worse.

Change `hostid' type to the type of int. It only stored but never used
within kernel, userland accesses it through sysctl_int(). Nothing
changes, but variable becomes consistent with sysctl_int().

ok bluhm

2 months agoIn a qcspmi_cmd_read() error case, the register is only returning the error
mglocker [Wed, 14 Aug 2024 10:54:58 +0000 (10:54 +0000)]
In a qcspmi_cmd_read() error case, the register is only returning the error
bit, not SPMI_STATUS_DONE.  Therefore, catch the error at the right place,
and print the specific error type.

ok patrick@

2 months agominiroot now works with USB Type-A ports on Apple machines
jsg [Wed, 14 Aug 2024 10:50:20 +0000 (10:50 +0000)]
miniroot now works with USB Type-A ports on Apple machines
discussed with kettenis@ and confirmed on M1 Mac mini

2 months agoMessages from modules were leaked. Call imsg_free() for them.
yasuoka [Wed, 14 Aug 2024 07:06:50 +0000 (07:06 +0000)]
Messages from modules were leaked.  Call imsg_free() for them.

2 months agoDon't abort the query when it is requesting to the next module.
yasuoka [Wed, 14 Aug 2024 07:04:54 +0000 (07:04 +0000)]
Don't abort the query when it is requesting to the next module.

2 months agotimingsafe_bcmp() is better than timingsafe_memcmp() for the purpose.
yasuoka [Wed, 14 Aug 2024 04:50:31 +0000 (04:50 +0000)]
timingsafe_bcmp() is better than timingsafe_memcmp() for the purpose.

2 months agoInitialize dae objects properly.
yasuoka [Wed, 14 Aug 2024 04:47:08 +0000 (04:47 +0000)]
Initialize dae objects properly.

2 months agoDon't abort the query when a duplicated query is received. Also tweak
yasuoka [Wed, 14 Aug 2024 04:44:07 +0000 (04:44 +0000)]
Don't abort the query when a duplicated query is received.  Also tweak
the log message.

2 months agoBe more aggressive about disabling interrupts in both DVACT_QUIESCE and
deraadt [Tue, 13 Aug 2024 22:32:58 +0000 (22:32 +0000)]
Be more aggressive about disabling interrupts in both DVACT_QUIESCE and
DVACT_SUSPEND, because we are seeing false interrupts during S0 resume.
Other problems remain, and ratchov has some clues he'll dig into in the
coming weeks.  For now, this is a good enough improvement.
ok ratchov

2 months agouse ACPI _WAK upon resume. Such a huge oversight it is surprising that
deraadt [Tue, 13 Aug 2024 22:31:16 +0000 (22:31 +0000)]
use ACPI _WAK upon resume.  Such a huge oversight it is surprising that
we've never noticed it before, maybe systems which depend on _WAK are rare.
Still, this may improve S3 resume on some rare machines.
ok kettenis

2 months agoFor AMD SEV use correct command mask in ccp(4) driver.
bluhm [Tue, 13 Aug 2024 20:48:00 +0000 (20:48 +0000)]
For AMD SEV use correct command mask in ccp(4) driver.

According to AMD SEV API specification, the mask for command ID is
supposed to be of length 10 (Bits [25:16]), rather than length 6.
This ensures that all commands are properly processed by the PSP.

from Aaron Debebe; OK hshoexer@

2 months agoFix a bug in .Ql handling that has been present since the beginning (2017).
schwarze [Tue, 13 Aug 2024 12:43:55 +0000 (12:43 +0000)]
Fix a bug in .Ql handling that has been present since the beginning (2017).

Since the .Ql macro action uses an output prefix of "'`" and an output
suffix of "`'", md_post_raw() would decrement the code_blocks state variable
even though md_pre_raw() had earlier neglected to increment it, hence
leaving the variable in an invalid negative state.  That in turn could
result in corrupt output in a variety of ways.

Fix this by checking in md_pre_raw() whether the prefix *contains* a
backtick rather than only checking whether it *starts* with a backtick.
For consistency, apply the same change to md_post_raw() even though
there was no bug in that function: all *suffixes* containing a backtick
actually contain it in the leading position.

Thanks to job@ for reporting this bug.  He noticed a particularly nasty
kind of output corruption: having .Ql in an input file would result
in ASCII_NBRSP (0x31) sneaking through into the output stream if later,
unrelated parts of the same input file directly or indirectly used
the \~ escape sequence, for example by using the .Ex macro.

2 months agoSync full virtqueue on device reset
sf [Tue, 13 Aug 2024 08:47:28 +0000 (08:47 +0000)]
Sync full virtqueue on device reset

We initialize the whole virtqueue and must make sure that the device
sees this even for the areas that are normally only written by the
device. Otherwise there may be an assertion fail during ifconfig up, as
found by bluhm@ with hshoexer@'s bounce buffer diff.

OK bluhm@

2 months agozap extra word;
jmc [Tue, 13 Aug 2024 05:52:09 +0000 (05:52 +0000)]
zap extra word;

2 months agodrm/i915/pxp/mtl: Update pxp-firmware packet size
jsg [Tue, 13 Aug 2024 05:13:34 +0000 (05:13 +0000)]
drm/i915/pxp/mtl: Update pxp-firmware packet size

From Alan Previn
c14d446e25fe00a9fd29d317b07bd221fd6f49db in mainline linux

2 months agodrm/i915/mtl: Wake GT before sending H2G message
jsg [Tue, 13 Aug 2024 04:04:53 +0000 (04:04 +0000)]
drm/i915/mtl: Wake GT before sending H2G message

From Vinay Belgaumkar
6d46d09a0d7dd412c5b76f74f89fe4448ba2117e in mainline linux

2 months agodrm/i915/mtl: Don't set PIPE_CONTROL_FLUSH_L3
jsg [Tue, 13 Aug 2024 03:58:13 +0000 (03:58 +0000)]
drm/i915/mtl: Don't set PIPE_CONTROL_FLUSH_L3

From Vinay Belgaumkar
5fde104ea8587c547732a06d9d6473f6e7e4ac1c in mainline linux

2 months agodrm/i915: Fix possible int overflow in skl_ddi_calculate_wrpll()
jsg [Tue, 13 Aug 2024 02:19:41 +0000 (02:19 +0000)]
drm/i915: Fix possible int overflow in skl_ddi_calculate_wrpll()

From Nikita Zhandarovich
6c0473c9ac502df8d580c15f20188efe708e8394 in linux-6.6.y/6.6.45
5b511572660190db1dc8ba412efd0be0d3781ab6 in mainline linux

2 months agodrm/i915/hdcp: Fix HDCP2_STREAM_STATUS macro
jsg [Tue, 13 Aug 2024 02:16:08 +0000 (02:16 +0000)]
drm/i915/hdcp: Fix HDCP2_STREAM_STATUS macro

From Suraj Kandpal
9541f99b2651adaeb1425474096999e689a60497 in linux-6.6.y/6.6.45
555069117390a5d581863bc797fb546bb4417c31 in mainline linux

2 months agoi915/perf: Remove code to update PWR_CLK_STATE for gen12
jsg [Tue, 13 Aug 2024 02:14:36 +0000 (02:14 +0000)]
i915/perf: Remove code to update PWR_CLK_STATE for gen12

From Umesh Nerlige Ramappa
582b6c7dd019c49a239db5f721146139130251c4 in linux-6.6.y/6.6.45
4bc14b9cfaa2149d41baef2f2620e9f82d9847d7 in mainline linux

2 months agodrm/i915: Prevent potential null-ptr-deref in engine_init_common
jsg [Tue, 13 Aug 2024 00:58:47 +0000 (00:58 +0000)]
drm/i915: Prevent potential null-ptr-deref in engine_init_common

From Nirmoy Das
f1cdb599ce0ef909343d6c8e7d372defbaa382b9 in mainline linux

2 months agosync
jsg [Tue, 13 Aug 2024 00:41:22 +0000 (00:41 +0000)]
sync

2 months agodrm/i915/mtl: Remove the 'force_probe' requirement for Meteor Lake
jsg [Tue, 13 Aug 2024 00:39:26 +0000 (00:39 +0000)]
drm/i915/mtl: Remove the 'force_probe' requirement for Meteor Lake

From Radhakrishna Sripada
213c43676beb5f5a63cb27a0c8e8e71035b08445 in mainline linux

2 months agoforce primary and console on Meteor Lake
jsg [Tue, 13 Aug 2024 00:36:32 +0000 (00:36 +0000)]
force primary and console on Meteor Lake

The GOP framebuffer doesn't match the PCI BAR ranges efifb tests.
discussed with kettenis@

2 months agodrm/i915: Annotate more of the BIOS fb takeover failure paths
jsg [Tue, 13 Aug 2024 00:29:51 +0000 (00:29 +0000)]
drm/i915: Annotate more of the BIOS fb takeover failure paths

From Ville Syrjala
074146f4578388b1b7f7832368d1581fceb1441e in mainline linux

2 months agodrm/i915: Try to relocate the BIOS fb to the start of ggtt
jsg [Tue, 13 Aug 2024 00:28:09 +0000 (00:28 +0000)]
drm/i915: Try to relocate the BIOS fb to the start of ggtt

From Ville Syrjala
a8153627520a2d468680bb7686fd404c222b13ca in mainline linux

2 months agodrm/i915: Tweak BIOS fb reuse check
jsg [Tue, 13 Aug 2024 00:25:13 +0000 (00:25 +0000)]
drm/i915: Tweak BIOS fb reuse check

From Ville Syrjala
ea5e150ac2cf88f586299c06244d5a4d473c041e in mainline linux

2 months agodrm/i915/fbdev: Fix smem_start for LMEMBAR stolen objects
jsg [Tue, 13 Aug 2024 00:23:25 +0000 (00:23 +0000)]
drm/i915/fbdev: Fix smem_start for LMEMBAR stolen objects

From Ville Syrjala
f1ee98cff3d86271491b08315fcdfa4c3f097e1e in mainline linux

2 months agodrm/i915: Simplify intel_initial_plane_config() calling convention
jsg [Tue, 13 Aug 2024 00:20:08 +0000 (00:20 +0000)]
drm/i915: Simplify intel_initial_plane_config() calling convention

From Ville Syrjala
30865e4abb799547299a9cf39c86fe943ee2913a in mainline linux

2 months agodrm/i915: Split the smem and lmem plane readout apart
jsg [Tue, 13 Aug 2024 00:18:24 +0000 (00:18 +0000)]
drm/i915: Split the smem and lmem plane readout apart

From Ville Syrjala
6bfdb06d1efafaa289f16ff5e5dfb4b02327525e in mainline linux

2 months agodrm/i915: s/phys_base/dma_addr/
jsg [Tue, 13 Aug 2024 00:16:20 +0000 (00:16 +0000)]
drm/i915: s/phys_base/dma_addr/

From Ville Syrjala
27fbcaf7ca2470dbc112a5ea2759ad6408581c79 in mainline linux

2 months agodrm/i915: Fix MTL initial plane readout
jsg [Tue, 13 Aug 2024 00:14:48 +0000 (00:14 +0000)]
drm/i915: Fix MTL initial plane readout

From Ville Syrjala
f46fb69489f57141bb7f8308893ec3e8ba664462 in mainline linux

2 months agodrm/i915: Fix region start during initial plane readout
jsg [Tue, 13 Aug 2024 00:12:42 +0000 (00:12 +0000)]
drm/i915: Fix region start during initial plane readout

From Ville Syrjala
d74f3a930c1d075a876a22b44723556455526881 in mainline linux

2 months agodrm/i915: Fix PTE decode during initial plane readout
jsg [Tue, 13 Aug 2024 00:10:51 +0000 (00:10 +0000)]
drm/i915: Fix PTE decode during initial plane readout

From Ville Syrjala
6b757e1d420cf01bc135212922906b8616b779d5 in mainline linux

2 months agodrm/i915: Rename the DSM/GSM registers
jsg [Tue, 13 Aug 2024 00:08:07 +0000 (00:08 +0000)]
drm/i915: Rename the DSM/GSM registers

From Ville Syrjala
f8ae1d5291c3e06d494cf59005c23d883f1afc0e in mainline linux

2 months agodrm/i915: Disable the "binder"
jsg [Tue, 13 Aug 2024 00:06:00 +0000 (00:06 +0000)]
drm/i915: Disable the "binder"

From Ville Syrjala
be5e8dc84f61ea79f4980fe4fb74fe9a0d0d2c33 in mainline linux

2 months agodrm/i915: Bypass LMEMBAR/GTTMMADR for MTL stolen memory access
jsg [Tue, 13 Aug 2024 00:03:54 +0000 (00:03 +0000)]
drm/i915: Bypass LMEMBAR/GTTMMADR for MTL stolen memory access

From Ville Syrjala
c08c364102d07288610734de34111a666e730ae7 in mainline linux

2 months agodrm/i915: Remove ad-hoc lmem/stolen debugs
jsg [Tue, 13 Aug 2024 00:00:37 +0000 (00:00 +0000)]
drm/i915: Remove ad-hoc lmem/stolen debugs

From Ville Syrjala
8f7cf0a215b34ef12b4b862f260a592b4ab37a52 in mainline linux

2 months agodrm/i915: Print memory region info during probe
jsg [Mon, 12 Aug 2024 23:57:43 +0000 (23:57 +0000)]
drm/i915: Print memory region info during probe

From Ville Syrjala
2ab1fe53e298fd2ca6491c73a8d306cc042cd2f0 in mainline linux

2 months agodrm/i915: Use struct resource for memory region IO as well
jsg [Mon, 12 Aug 2024 23:55:26 +0000 (23:55 +0000)]
drm/i915: Use struct resource for memory region IO as well

From Ville Syrjala
3c0fa9f4ec220d4a6fac1b80c231c38cf6d0934d in mainline linux

2 months agodrm/i915: Enable GGTT updates with binder in MTL
jsg [Mon, 12 Aug 2024 23:52:04 +0000 (23:52 +0000)]
drm/i915: Enable GGTT updates with binder in MTL

From Nirmoy Das
799d794f75598353c8e5854fc9c57cc46d236c4e in mainline linux

2 months agodrm/i915: Toggle binder context ready status
jsg [Mon, 12 Aug 2024 23:50:01 +0000 (23:50 +0000)]
drm/i915: Toggle binder context ready status

From Nirmoy Das
a2ae29629230588d50dfcba306decad7e4f690f3 in mainline linux

2 months agodrm/i915: Implement GGTT update method with MI_UPDATE_GTT
jsg [Mon, 12 Aug 2024 23:47:44 +0000 (23:47 +0000)]
drm/i915: Implement GGTT update method with MI_UPDATE_GTT

From Nirmoy Das
8a7f77fabac16e284cc47191fe033770012bf48d in mainline linux

2 months agodrm/i915: Parameterize binder context creation
jsg [Mon, 12 Aug 2024 23:44:39 +0000 (23:44 +0000)]
drm/i915: Parameterize binder context creation

From Nirmoy Das
3f5f62883631a987964102bc5044f7bf62c26323 in mainline linux

2 months agodrm/i915: Implement for_each_sgt_daddr_next
jsg [Mon, 12 Aug 2024 23:42:15 +0000 (23:42 +0000)]
drm/i915: Implement for_each_sgt_daddr_next

From Nirmoy Das
0e514878486053363f8b2a806525fe67ae692827 in mainline linux

2 months agodrm/i915: Create a kernel context for GGTT updates
jsg [Mon, 12 Aug 2024 23:39:18 +0000 (23:39 +0000)]
drm/i915: Create a kernel context for GGTT updates

From Nirmoy Das
b352749936806c9d5ed6a6021d84c1df4d1df3da in mainline linux

2 months agodrm/i915: Lift runtime-pm acquire callbacks out of intel_wakeref.mutex
jsg [Mon, 12 Aug 2024 23:36:45 +0000 (23:36 +0000)]
drm/i915: Lift runtime-pm acquire callbacks out of intel_wakeref.mutex

From Chris Wilson
4cd64e9d2c7206db05e7162d0258b455726b7ec5 in mainline linux

2 months agofflush.3: zap trailing whitespace
tb [Mon, 12 Aug 2024 22:37:57 +0000 (22:37 +0000)]
fflush.3: zap trailing whitespace

2 months agoFix grammar in history section
tb [Mon, 12 Aug 2024 21:27:57 +0000 (21:27 +0000)]
Fix grammar in history section

ok guenther

2 months agosync libc, libssl and libtls bumps
tb [Mon, 12 Aug 2024 21:04:46 +0000 (21:04 +0000)]
sync libc, libssl and libtls bumps

2 months agoGive libtls the same bump as libssl
tb [Mon, 12 Aug 2024 21:02:24 +0000 (21:02 +0000)]
Give libtls the same bump as libssl

2 months agoBump minor after symbol addition
tb [Mon, 12 Aug 2024 21:01:54 +0000 (21:01 +0000)]
Bump minor after symbol addition

2 months agoAdd SSL_CTX_set1_cert_store() and SSL_CIPHER_get_handshake_digest() to
tb [Mon, 12 Aug 2024 21:01:34 +0000 (21:01 +0000)]
Add SSL_CTX_set1_cert_store() and SSL_CIPHER_get_handshake_digest() to
Symbols.list

2 months agoExpose SSL_CTX_set1_cert_store() and SSL_CIPHER_get_handshake_digest()
tb [Mon, 12 Aug 2024 21:01:00 +0000 (21:01 +0000)]
Expose SSL_CTX_set1_cert_store() and SSL_CIPHER_get_handshake_digest()

2 months agoAdd <stdio_ext.h> with the seven APIs needed to have gnulib operate
guenther [Mon, 12 Aug 2024 20:56:55 +0000 (20:56 +0000)]
Add <stdio_ext.h> with the seven APIs needed to have gnulib operate
without poking directly into the FILE structure.

Repeated testing, "nope, need a few more" feedback, and ok tb@

2 months agoMake exit(), fclose(), fflush(), and freopen() comply with POSIX-2008
guenther [Mon, 12 Aug 2024 20:53:09 +0000 (20:53 +0000)]
Make exit(), fclose(), fflush(), and freopen() comply with POSIX-2008
requirements for setting the underlying file position when flushing
read-mode streams, and make an fseek()-after-fflush() not change the
underlying file position.

Much testing, review, and assistance from tb@
ok tb@ millert@

2 months agoxkbcomp 1.7.0 moved its data files from lib/X11 to share/X11
matthieu [Mon, 12 Aug 2024 19:40:17 +0000 (19:40 +0000)]
xkbcomp 1.7.0 moved its data files from lib/X11 to share/X11

2 months agoConditionally compile kqueue poll debug printfs, mainly preventing them
anton [Mon, 12 Aug 2024 19:32:05 +0000 (19:32 +0000)]
Conditionally compile kqueue poll debug printfs, mainly preventing them
from being included in the ramdisk kernel.

Looks sensible to deraadt@

2 months agoother growth has happened and I'm out of current culling options,
deraadt [Mon, 12 Aug 2024 18:43:41 +0000 (18:43 +0000)]
other growth has happened and I'm out of current culling options,
so mpi(4) gets removed from the floppy.

2 months agoRevert the guts of rev 1.14. This ends up attempting to turn of power
kettenis [Mon, 12 Aug 2024 17:24:58 +0000 (17:24 +0000)]
Revert the guts of rev 1.14.  This ends up attempting to turn of power
resources for devices that aren't present.  And that makes us run AML that
clearly hasn't been tested by the vendors.  And on my test laptop this no
longer seems to make a meaningful difference in the power consumption
during suspend-to-idle.

ok deraadt@

2 months agoPrepare bpf_sysctl() for upcoming net_sysctl() unlocking.
mvs [Mon, 12 Aug 2024 17:02:58 +0000 (17:02 +0000)]
Prepare bpf_sysctl() for upcoming net_sysctl() unlocking.

Both NET_BPF_MAXBUFSIZE and NET_BPF_BUFSIZE (`bpf_maxbufsize' and
`bpf_bufsize' respectively) are atomically accessed integers. No locks
required to modify them.

ok bluhm

2 months agoAdd -CRLfile option to 'cms' sub command
job [Mon, 12 Aug 2024 15:34:58 +0000 (15:34 +0000)]
Add -CRLfile option to 'cms' sub command

This option allows to verify certs in a CMS object against additional
CRLs.

Ported from work by Tom Harrison from APNIC

OK tb@

2 months agoRun network protocol timer without kernel lock.
bluhm [Mon, 12 Aug 2024 11:25:27 +0000 (11:25 +0000)]
Run network protocol timer without kernel lock.

Mark slow and fast protocol timeouts as MP safe.  This means they
run on a spearate thread without holding the kernel lock.
IGMP and MLD6 cannot run in parallel, they use exclusive net lock
to protect themselves.  As a performance optimization global variables
are used to skip igmp_fasttimo() and mld6_fasttimeo() if no multicast
is active.  These global variables use atomic operations and memory
barriers to work lockless.
IPv6 fragment timeout protects itself with a mutex.
TCP timers also run without kernel lock now.  The whole TCP stack
holds exclusive net lock, so additional kernel lock is useless.

OK mvs@

2 months agoimplement the report response for proc-filters too
op [Mon, 12 Aug 2024 09:32:44 +0000 (09:32 +0000)]
implement the report response for proc-filters too

Reported by renegm on GitHub:
<https://github.com/OpenSMTPD/OpenSMTPD/issues/1257>

ok gilles@

2 months agoPrint min-version of a RTR session.
claudio [Mon, 12 Aug 2024 09:05:28 +0000 (09:05 +0000)]
Print min-version of a RTR session.
OK tb@

2 months agoAdd 'min-version' RTR config option and default to RTR version 1 by default.
claudio [Mon, 12 Aug 2024 09:04:23 +0000 (09:04 +0000)]
Add 'min-version' RTR config option and default to RTR version 1 by default.

The min-version pins a minimal required version for rtr session. This is
needed if specific PDUs are required and it ensures that the session is
not suddenly downgraded. This is important for ASPA where a minimum
version of 2 is required. Only then the ASPA PDUs are transmitted.

By default a RTR version of 1 is used but setting min-version to 2
will enable draft-ietf-sidrops-8210bis-14 support and enforce it
at the same time. Right now defaulting to version 2 is not possible
since draft-ietf-sidrops-8210bis failed to progress for a too long
time resulting in split eco system with various incompatible RTR
version 2 implementations.

OK tb@

2 months agotry harder to leave a gap on the tx ring.
dlg [Mon, 12 Aug 2024 06:47:11 +0000 (06:47 +0000)]
try harder to leave a gap on the tx ring.

i think before this change we could overwrite entries on the ring,
which can confuse the chip and the tx completion code. i think.
it's funny how much a comparison hurts my brain.

ok patrick@
tested by and ok kevlo@

2 months agoGet rid of inet_aton and simplify.
florian [Mon, 12 Aug 2024 06:22:36 +0000 (06:22 +0000)]
Get rid of inet_aton and simplify.

localhost is just a weird spelling of 127.0.0.1.
OK bluhm

2 months agoGet rid of inet_aton.
florian [Mon, 12 Aug 2024 06:19:24 +0000 (06:19 +0000)]
Get rid of inet_aton.

Input & OK bluhm

2 months agoremove unused vars in local functions
jsg [Mon, 12 Aug 2024 02:16:21 +0000 (02:16 +0000)]
remove unused vars in local functions

2 months agouse INTEL_INFO for early platform tests in inteldrm_attach()
jsg [Mon, 12 Aug 2024 01:49:49 +0000 (01:49 +0000)]
use INTEL_INFO for early platform tests in inteldrm_attach()

runtime IS_PLATFORM tests are not setup until

inteldrm_attachhook
i915_driver_probe
i915_driver_early_probe
intel_device_info_runtime_init_early
intel_device_info_subplatform_init

2 months agoAdd missing refcnt.h header to fix libkvm build.
mvs [Sun, 11 Aug 2024 21:07:05 +0000 (21:07 +0000)]
Add missing refcnt.h header to fix libkvm build.

Reported and ok tb.

2 months agoEven though US-ASCII (= ANSI X3.4-1986) only defines 128 characters,
schwarze [Sun, 11 Aug 2024 18:24:43 +0000 (18:24 +0000)]
Even though US-ASCII (= ANSI X3.4-1986) only defines 128 characters,
the POSIX standard explicitly requires in section 6.2 that "the POSIX
locale shall contain 256 single-byte characters", see:
https://pubs.opengroup.org/onlinepubs/9799919799/basedefs/V1_chap06.html#tag_06_02

So the current behaviour of treating non-ASCII bytes in an LC_CTYPE=POSIX
input stream as if they were characters is not a POSIX violation, but
actually required by the standard - and not just for awk(1), but for
utility programs in general and even for library functions in general.
Consequently, delete the wrong sentence i added to the STANDARDS section
last year.

Thanks to millert@ and jmc@ for making me realize my mistake.
OK millert@ jmc@

2 months agodelete correct ACPI_SST_WAKING, pointed out by kettenis
deraadt [Sun, 11 Aug 2024 17:30:28 +0000 (17:30 +0000)]
delete correct ACPI_SST_WAKING, pointed out by kettenis

2 months agoMake exit1() wait sysctl(2) `allprocess' loops.
mvs [Sun, 11 Aug 2024 15:10:53 +0000 (15:10 +0000)]
Make exit1() wait sysctl(2) `allprocess' loops.

Regardless on wired userland memory, KERN_FILE_BYPID and KERN_FILE_BYUID
`allprocess' loops have netlock provided sleep points, so concurrent
process exit(1) could crash kernel.

The main exit1() problem is that process teardown begins while process
is still linked to `allprocess' list, and current code doesn't allow to
unlink it first. Wait for concurrent sysctl(2) `allprocess' loops
between PS_EXITING bit setting and list unlinking. Both KERN_FILE_BYPID
and KERN_FILE_BYUID loops do PS_EXITING check and won't deal with dying
process. Concurrent exit1() thread will wait loops keeping process
linked to `allprocess' list.

Tested with i386 dpb(1) run.
Stress tests and ok bluhm.

2 months agoAdd include path for crypto_arch.h.
jsing [Sun, 11 Aug 2024 13:05:43 +0000 (13:05 +0000)]
Add include path for crypto_arch.h.

2 months agoAdd include path for crypto_arch.h.
jsing [Sun, 11 Aug 2024 13:04:46 +0000 (13:04 +0000)]
Add include path for crypto_arch.h.

2 months agoProvide and use crypto_arch.h.
jsing [Sun, 11 Aug 2024 13:02:39 +0000 (13:02 +0000)]
Provide and use crypto_arch.h.

Provide a per architecture crypto_arch.h - this will be used in a similar
manner to bn_arch.h and will allow for architecture specific #defines and
static inline functions. Move the HAVE_AES_* and HAVE_RC4_* defines here.

ok tb@

2 months agowhitespace, no binary change
florian [Sun, 11 Aug 2024 11:56:08 +0000 (11:56 +0000)]
whitespace, no binary change

2 months agoMention H6 & H616 chipsets. ok patrick@
dtucker [Sun, 11 Aug 2024 08:25:12 +0000 (08:25 +0000)]
Mention H6 & H616 chipsets.  ok patrick@

2 months agorework DESCRIPTION for both dhcpleased(8) and slaacd(8) to more accurately
jmc [Sun, 11 Aug 2024 06:07:37 +0000 (06:07 +0000)]
rework DESCRIPTION for both dhcpleased(8) and slaacd(8) to more accurately
reflect how they work, to emphasize that they do nothing unless auto conf
is set, and to more closely match each other;

prompted by a diff from andrew christopher hawk on tech, who noted a wonky
sentence;

help/feedback deraadt florian; ok deraadt

2 months agoforgot the suspend-time duplication mentioned in 1.25
deraadt [Sun, 11 Aug 2024 05:28:02 +0000 (05:28 +0000)]
forgot the suspend-time duplication mentioned in 1.25

2 months agoput ACPI_SST_WAKING in gosleep() where kettenis asked for
deraadt [Sun, 11 Aug 2024 05:21:52 +0000 (05:21 +0000)]
put ACPI_SST_WAKING in gosleep() where kettenis asked for