openbsd
2 years agoUse generic PNP ID to match HID-over-I2C devices.
kettenis [Thu, 8 Sep 2022 06:33:45 +0000 (06:33 +0000)]
Use generic PNP ID to match HID-over-I2C devices.

ok patrick@, drahn@

2 years agoAdd a few more pin mappings based on what is in some Linux device tree
kettenis [Thu, 8 Sep 2022 06:32:32 +0000 (06:32 +0000)]
Add a few more pin mappings based on what is in some Linux device tree
updates that are floating around.  Remove some #ifdef'ed out debug code
that was left behind.

ok patrick@, drahn@

2 years agoriscv64: cpu_initclocks: install tb_timecounter before cpu_startclock()
cheloha [Thu, 8 Sep 2022 03:09:40 +0000 (03:09 +0000)]
riscv64: cpu_initclocks: install tb_timecounter before cpu_startclock()

In the future, the clock interrupt code will need a real timecounter
to work correctly.  Nudge the tc_init(9) call for tb_timecounter up
before cpu_startclock().

2 years agomacppc: cpu_initclocks: install tb_timecounter before cpu_startclock()
cheloha [Thu, 8 Sep 2022 03:06:33 +0000 (03:06 +0000)]
macppc: cpu_initclocks: install tb_timecounter before cpu_startclock()

In the future, the clock interrupt code will need a real timecounter
to work correctly.  Nudge the tc_init(9) call for tb_timecounter up
before cpu_startclock().

2 years agorecognise Realtek ALC245
jsg [Thu, 8 Sep 2022 01:35:39 +0000 (01:35 +0000)]
recognise Realtek ALC245
from Neel Chauhan

2 years agomatch on Intel 600 Series LP HD Audio (0x51c8)
jsg [Thu, 8 Sep 2022 01:28:46 +0000 (01:28 +0000)]
match on Intel 600 Series LP HD Audio (0x51c8)

HP Spectre x360 16-f0013dx has subclass audio not hd audio
so azalia does not match

from Neel Chauhan

2 years agossl_cipher_process_rulestr: don't read outside rule_str buffer
millert [Wed, 7 Sep 2022 21:34:22 +0000 (21:34 +0000)]
ssl_cipher_process_rulestr: don't read outside rule_str buffer
If rule_str ended in a "-", "l" was incremented one byte past the
end of the buffer.  This resulted in an out-of-bounds read when "l"
is dereferenced at the end of the loop.  OK tb@

2 years agoAdd EVP test coverage for RC2.
tb [Wed, 7 Sep 2022 21:25:21 +0000 (21:25 +0000)]
Add EVP test coverage for RC2.

From Joshua Sing

2 years agoAdd output length validation for EVP
tb [Wed, 7 Sep 2022 21:17:32 +0000 (21:17 +0000)]
Add output length validation for EVP

From Joshua Sing

2 years agoAdd a few more PNP IDs to handle variants of the Qualcomm dual role
kettenis [Wed, 7 Sep 2022 20:06:23 +0000 (20:06 +0000)]
Add a few more PNP IDs to handle variants of the Qualcomm dual role
controller based on AML dumps found online and some further detective
work looking at Windows drivers by jsg@.

ok jsg@, millert@

2 years agovmm(4): allow reading MSR_TSC on Intel hosts.
dv [Wed, 7 Sep 2022 18:44:09 +0000 (18:44 +0000)]
vmm(4): allow reading MSR_TSC on Intel hosts.

Add's MSR_TSC to the read bitmap for msr access. This was added to
AMD/SVM in Feb 2020, but never added to Intel VMX. Some guests use
rdmsr instead of rdtsc, so this prevents a #GP exception.

ok mlarkin@

2 years agoOn the Lenovo X13s the named components don't have the single mapping
patrick [Wed, 7 Sep 2022 18:25:08 +0000 (18:25 +0000)]
On the Lenovo X13s the named components don't have the single mapping
flag set, which we so far had expected to be there.  This is allowed by
the IORT spec.  In that case simply make use of the first mapping if
there is a single one.

With feedback from and ok kettenis@

2 years agosync with arm64.html
jsg [Wed, 7 Sep 2022 10:30:03 +0000 (10:30 +0000)]
sync with arm64.html

2 years agoFix passing explicit stage files
kn [Wed, 7 Sep 2022 10:21:03 +0000 (10:21 +0000)]
Fix passing explicit stage files

Every platform ought to set `stages', `stage1' and optionally `stage2'
in md_init(), otherwise passing explicit files results won't work as
`stages' is zero-initialised and no default path is set:

# installboot -v sd0 /root/BOOTAA64.EFI
usage: installboot [-npv] [-r root] disk [stage1]

This is correct synopsis and ought to work, but efi_installboot.c has an
empty md_init().  Set stage bits to fix this:

# ./obj/installboot -nv sd0 /root/BOOTAA64.EFI
Using / as root
would install bootstrap on /dev/rsd0c
using first-stage /root/BOOTAA64.EFI
would copy /root/BOOTAA64.EFI to /tmp/installboot.2bGhLGT1eF/efi/boot/bootaa64.efi
would write /tmp/installboot.2bGhLGT1eF/efi/boot/startup.nsh

This makes regress/usr.sbin/installboot pass on armv7, arm64 and riscv64
(while being lucky or carrying miod's fix for the kernel disklabel race
manifesting on vnd).

2 years agoRemove the .SILENT bits again after feedback from bluhm
kn [Wed, 7 Sep 2022 10:07:39 +0000 (10:07 +0000)]
Remove the .SILENT bits again after feedback from bluhm

2 years agoMention bsd.rd and install72.img
kn [Wed, 7 Sep 2022 09:52:40 +0000 (09:52 +0000)]
Mention bsd.rd and install72.img

OK miod jsg

2 years agoansi
jsg [Wed, 7 Sep 2022 05:36:59 +0000 (05:36 +0000)]
ansi

2 years agoGBe -> GbE
jsg [Wed, 7 Sep 2022 01:22:08 +0000 (01:22 +0000)]
GBe -> GbE

2 years agoremove history of when each du(1) flag was added
jsg [Wed, 7 Sep 2022 00:46:20 +0000 (00:46 +0000)]
remove history of when each du(1) flag was added
ok jmc@ schwarze@

2 years agoStarFire -> StarFive
jsg [Wed, 7 Sep 2022 00:44:03 +0000 (00:44 +0000)]
StarFire -> StarFive

2 years agoDocument support for booting from RAID 1C on sparc64
kn [Tue, 6 Sep 2022 22:13:31 +0000 (22:13 +0000)]
Document support for booting from RAID 1C on sparc64

The bootloader (ofwboot) got support some time ago, but installboot(8) had
to be fixed as well for multi-chunk volumes (1C requires at least two).

As of today, the latest sparc64 install72.img can perform a default install
onto a passphrase-protected 1C volume across four chunk disks, resulting in
a bootable system without requiring manual invervention, just like amd64.

2 years agoCrank (sparse) disk size from 64M to 128M to make arm64 happy again
kn [Tue, 6 Sep 2022 20:54:37 +0000 (20:54 +0000)]
Crank (sparse) disk size from 64M to 128M to make arm64 happy again

2 years agoAdd VERBOSE=No default, .IGNORE and .SILENT REGRESS_CLEANUP
kn [Tue, 6 Sep 2022 20:30:00 +0000 (20:30 +0000)]
Add VERBOSE=No default, .IGNORE and .SILENT REGRESS_CLEANUP

Tiny bit less noise.  The many REGRESS_SETUP_ONCE targets could be merged
into a single one and then be added to .SILENT unless VERBOSE=yes to focus
on actual installboot output and ditch the dd/fdisk/disklabel noise by default.

2 years agokeep noise down, discard usage text for expected failure cases
kn [Tue, 6 Sep 2022 19:40:10 +0000 (19:40 +0000)]
keep noise down, discard usage text for expected failure cases

2 years agoUse multi-chunk RAID 1C with keydisk on softraid enabled archs
kn [Tue, 6 Sep 2022 19:35:21 +0000 (19:35 +0000)]
Use multi-chunk RAID 1C with keydisk on softraid enabled archs

Keydisks requires special treatment and should not be touched by installboot.
In real setups they are probably offline (temporary USB stick, only present
during boot to unlock), but when they are around (keydisk chunk on same disk
as crypto volume)...

installboot currently installs treats keydisk like any other chunk, which
must be fixed.

Thus change the tests into a should-work-but-currently-broken setup so an
installboot fix can be immediately tested against it.

2 years agoi386/lapic.c: sync with amd64/lapic.c
cheloha [Tue, 6 Sep 2022 17:26:27 +0000 (17:26 +0000)]
i386/lapic.c: sync with amd64/lapic.c

In anticipation of merging the new clock interrupt code after release,
let's synchronize the lapic timer parts of i386's lapic.c with those
of amd64's lapic.c.  The two files will need similar changes to switch
to the new code, so the more alike they are the better.

Tested by mlarkin@ on an ESXi VM.  Tested by me on my amd64 laptop
running in x86 compatibility mode.

ok mlarkin@

2 years agoAdd a table-driven test for RC2 in ECB mode to replace the old one.
tb [Tue, 6 Sep 2022 15:36:25 +0000 (15:36 +0000)]
Add a table-driven test for RC2 in ECB mode to replace the old one.

From Joshua Sing

2 years agoFix spacing/indent, no object change
kn [Tue, 6 Sep 2022 15:22:11 +0000 (15:22 +0000)]
Fix spacing/indent, no object change

2 years agoDelete some unreferenced SMD, ST506 and archaic SCSI #define's
krw [Tue, 6 Sep 2022 14:14:44 +0000 (14:14 +0000)]
Delete some unreferenced SMD, ST506 and archaic SCSI #define's
that use d_drivedata.

Since nothing else in the tree refers to d_drivedata other than
/etc/disktab parsing and no /etc/disktab entry utilizes the
:d[0-4]: attributes, stop disklabel(8) both printing "drivedata:
0" and paying attention to any "drivedata: ..." lines in ascii
labels being read.

ok jsg@ miod@

2 years agoAttach aplmbox(4) early.
kettenis [Tue, 6 Sep 2022 13:52:50 +0000 (13:52 +0000)]
Attach aplmbox(4) early.

2 years agoProperly free memory in filemode
job [Tue, 6 Sep 2022 11:16:51 +0000 (11:16 +0000)]
Properly free memory in filemode

OK tb@

2 years agoRemove more mkerr.pl remnants, missed in previous
kn [Tue, 6 Sep 2022 10:22:31 +0000 (10:22 +0000)]
Remove more mkerr.pl remnants, missed in previous

Noticed by jsg
Feedback OK jsg

2 years agoReplace aes_{ccm,xts}_cleanup usage with NULL.
jsing [Tue, 6 Sep 2022 06:38:26 +0000 (06:38 +0000)]
Replace aes_{ccm,xts}_cleanup usage with NULL.

Only change in generated assembly is due to line numbers.

2 years agoStop casting a size_t to a long and then passing it as a size_t.
jsing [Tue, 6 Sep 2022 06:17:11 +0000 (06:17 +0000)]
Stop casting a size_t to a long and then passing it as a size_t.

These cipher implementations take a size_t length argument, so stop
casting it to a long.

Found by Coverity.

ok tb@

2 years agosync
gkoehler [Mon, 5 Sep 2022 21:52:53 +0000 (21:52 +0000)]
sync

2 years agoFix copy-paste error
tb [Mon, 5 Sep 2022 21:36:46 +0000 (21:36 +0000)]
Fix copy-paste error

2 years agoPlug leaks due to not resetting the EVP_CIPHER_CTX
tb [Mon, 5 Sep 2022 21:34:23 +0000 (21:34 +0000)]
Plug leaks due to not resetting the EVP_CIPHER_CTX

2 years agoTwo more missing long long casts
tb [Mon, 5 Sep 2022 21:12:08 +0000 (21:12 +0000)]
Two more missing long long casts

2 years agoMore long long casts for printing time_t
tb [Mon, 5 Sep 2022 21:08:08 +0000 (21:08 +0000)]
More long long casts for printing time_t

2 years agoConvert from %i to %d
tb [Mon, 5 Sep 2022 21:06:31 +0000 (21:06 +0000)]
Convert from %i to %d

2 years agocast time_t to long long before printing for portability
tb [Mon, 5 Sep 2022 20:53:01 +0000 (20:53 +0000)]
cast time_t to long long before printing for portability

2 years agoRemove some inactive code, comments, and DPRINTFs. Minor style changes.
mglocker [Mon, 5 Sep 2022 20:11:44 +0000 (20:11 +0000)]
Remove some inactive code, comments, and DPRINTFs.  Minor style changes.

2 years agoDo not perform out-of-bounds arrray accesses for bogus relocation type values.
miod [Mon, 5 Sep 2022 20:09:24 +0000 (20:09 +0000)]
Do not perform out-of-bounds arrray accesses for bogus relocation type values.

2 years agoUpdate to most recent spec
job [Mon, 5 Sep 2022 20:08:26 +0000 (20:08 +0000)]
Update to most recent spec

2 years agoUse exact fdisk arguments from "(W)hole disk" answer
kn [Mon, 5 Sep 2022 19:32:50 +0000 (19:32 +0000)]
Use exact fdisk arguments from "(W)hole disk" answer

Not all architectures did what the installer does with empty disks.

2 years agoDon't ignore memory blocks smaller than 64KB. Some EFI implementations
kettenis [Mon, 5 Sep 2022 19:18:56 +0000 (19:18 +0000)]
Don't ignore memory blocks smaller than 64KB.  Some EFI implementations
(such as the one on the x13s) allocate memory with the EfiBootSevicesData
type in a semi-random fashion.  Ignoring small regions with that type
results in different memory maps between boots of the same kernel.
This causes problems with upcoming hibernate support.

The decision to ignore small regions was made because we do this on
amd64 to work arounmd broken BIOS implementations and because of fears
that we would run out of physical memory segments in UVM.  We have some
reasons to believe that we can trust the EFI memory map on arm64 and the
risk of running out of physical memory segments is mitigated by the fact
that we try to merge memory regions before loading them into UVM.  If
for some reason we see a significant drop in physical memory on certain
machines, we should probably increase the number of items in the array
we use to store memory regions and/or increase the number of physical
memory segments in UVM.

ok mlarkin@

2 years agoReset provider in each iteration
tb [Mon, 5 Sep 2022 18:07:04 +0000 (18:07 +0000)]
Reset provider in each iteration

If a providerAS sets an afiLimit, subsequent providerAS that don't set it
would erroneously inherit that limit. Zero out the provider at the top of
the loop to avoid this problem.

ok job

2 years agoPledge sendmmsg and recvmmsg with stdio similar to their non-loop
mbuhl [Mon, 5 Sep 2022 16:37:47 +0000 (16:37 +0000)]
Pledge sendmmsg and recvmmsg with stdio similar to their non-loop
counter parts. Input deraadt.
ok bluhm

2 years agoMove mld6 address variables from data to stack memory to make them
bluhm [Mon, 5 Sep 2022 15:47:39 +0000 (15:47 +0000)]
Move mld6 address variables from data to stack memory to make them
MP safe.  Due to the KAME scope address hack, the link-local all
nodes and all routers IPv6 addresses cannot be const.
OK benno@

2 years agoUse shared netlock in soreceive(). The UDP and IP divert layer
bluhm [Mon, 5 Sep 2022 14:56:08 +0000 (14:56 +0000)]
Use shared netlock in soreceive().  The UDP and IP divert layer
provide locking of the PCB.  If that is possible, use shared instead
of exclusive netlock in soreceive().  The PCB mutex provides a per
socket lock against multiple soreceive() running in parallel.
Release and regrab both locks in sosleep_nsec().
OK mvs@

2 years agodrm/amd/display: avoid doing vm_init multiple time
jsg [Mon, 5 Sep 2022 14:22:58 +0000 (14:22 +0000)]
drm/amd/display: avoid doing vm_init multiple time

From Charlene Liu
85dd24ff77c15f5fc0e16bc3f288f5d1ac93e02b in linux 5.15.y/5.15.65
5544a7b5a07480192eb5fd3536462faed2c21528 in mainline linux

2 years agodrm/amdgpu: Increase tlb flush timeout for sriov
jsg [Mon, 5 Sep 2022 14:21:08 +0000 (14:21 +0000)]
drm/amdgpu: Increase tlb flush timeout for sriov

From Dusica Milinkovic
898467ac9bcb56090f0b651088a36331ff779e4c in linux 5.15.y/5.15.65
373008bfc9cdb0f050258947fa5a095f0657e1bc in mainline linux

2 years agodrm/amd/display: Fix pixel clock programming
jsg [Mon, 5 Sep 2022 14:18:34 +0000 (14:18 +0000)]
drm/amd/display: Fix pixel clock programming

From Ilya Bakoulin
4df54c493e76f3cfa0bc8c54c5c43faa3b6e78d4 in linux 5.15.y/5.15.65
04fb918bf421b299feaee1006e82921d7d381f18 in mainline linux

2 years agodrm/amd/pm: add missing ->fini_microcode interface for Sienna Cichlid
jsg [Mon, 5 Sep 2022 14:16:36 +0000 (14:16 +0000)]
drm/amd/pm: add missing ->fini_microcode interface for Sienna Cichlid

From Evan Quan
a89e753d5a9f3b321f4a3098e2755c5aabcff0af in linux 5.15.y/5.15.65
0a2d922a5618377cdf8fa476351362733ef55342 in mainline linux

2 years agodrm/amd/display: clear optc underflow before turn off odm clock
jsg [Mon, 5 Sep 2022 14:15:03 +0000 (14:15 +0000)]
drm/amd/display: clear optc underflow before turn off odm clock

From Fudong Wang
5ee30bcfdb32526233d2572f3d9ec371928679f1 in linux 5.15.y/5.15.65
b2a93490201300a749ad261b5c5d05cb50179c44 in mainline linux

2 years agodrm/amd/display: For stereo keep "FLIP_ANY_FRAME"
jsg [Mon, 5 Sep 2022 14:12:51 +0000 (14:12 +0000)]
drm/amd/display: For stereo keep "FLIP_ANY_FRAME"

From Alvin Lee
e407e04a93d7b1209669220fab5eb69ab7b09baf in linux 5.15.y/5.15.65
84ef99c728079dfd21d6bc70b4c3e4af20602b3c in mainline linux

2 years agodrm/amd/display: Fix HDMI VSIF V3 incorrect issue
jsg [Mon, 5 Sep 2022 14:11:02 +0000 (14:11 +0000)]
drm/amd/display: Fix HDMI VSIF V3 incorrect issue

From Leo Ma
2cddd3d0b049a5e0666f93ea8c0d6ba6cc4bbec4 in linux 5.15.y/5.15.65
0591183699fceeafb4c4141072d47775de83ecfb in mainline linux

2 years agodrm/amd/display: Avoid MPC infinite loop
jsg [Mon, 5 Sep 2022 14:09:09 +0000 (14:09 +0000)]
drm/amd/display: Avoid MPC infinite loop

From Josip Pavic
0c8abeceee0f58ad3fdc66e1de0c0e02b962653b in linux 5.15.y/5.15.65
8de297dc046c180651c0500f8611663ae1c3828a in mainline linux

2 years agodrm/i915/gt: Skip TLB invalidations once wedged
jsg [Mon, 5 Sep 2022 14:06:39 +0000 (14:06 +0000)]
drm/i915/gt: Skip TLB invalidations once wedged

From Chris Wilson
4676773ea117624ced48fdf9806d5e0dc051bd7f in linux 5.15.y/5.15.65
e5a95c83ed1492c0f442b448b20c90c8faaf702b in mainline linux

2 years agodrm/bridge: Add stubs for devm_drm_of_get_bridge when OF is disabled
jsg [Mon, 5 Sep 2022 14:04:28 +0000 (14:04 +0000)]
drm/bridge: Add stubs for devm_drm_of_get_bridge when OF is disabled

From Maxime Ripard
f8b07c05b69969c41efafda7022d6cb184d61bf9 in linux 5.15.y/5.15.65
59050d783848d9b62e9d8fb6ce0cd00771c2bf87 in mainline linux

2 years agoDon't leak cert in aspa_parse()
tb [Mon, 5 Sep 2022 12:25:32 +0000 (12:25 +0000)]
Don't leak cert in aspa_parse()

ok job

2 years agosync
sthen [Mon, 5 Sep 2022 11:21:08 +0000 (11:21 +0000)]
sync

2 years agoFix -r on multi-chunk softraid volumes
kn [Mon, 5 Sep 2022 11:12:20 +0000 (11:12 +0000)]
Fix -r on multi-chunk softraid volumes

Running installboot(8) on softraid(4) volumes means installing stages on
every softraid chunk.

The overall idea is the same, but MD implementations differ.

sparc64_softraid.c's sr_install_bootblk() reuses sparc64_installboot.c's
md_installboot() for this.

For sparc64, md_installboot() does the copy of stage 2, usually
/usr/mdec/ofwboot to /ofwboot, so when `-r root' is passed, it prefixes the
file path with "root".

For single-disk installations (plain-disk and single-chunk softraid) this is
fine, but as soon as multiple chunks are used, md_installboot() currently
prefixes the path each time, obviously resulting in invalid paths starting
with the second run.

Other architectures do reuse md_installboot() as well but either don't do
such a copy or implement the prefixing differently -- plus they must support
softraid in the firt place to be able to hit this type of bug.

With this fixed, regress/usr.sbin/installboot finally passes on sparc64 and
installboot no longer fails at the end of a fresh installation onto softraid
with multiple chunks.

"looks correct" miod

2 years agoAdd missing prototypes for route_attach() and route_detach().
mvs [Mon, 5 Sep 2022 10:31:25 +0000 (10:31 +0000)]
Add missing prototypes for route_attach() and route_detach().

2 years agoboot_*.8: reference installboot.8, rectify FFS note on sparc64
kn [Mon, 5 Sep 2022 10:29:27 +0000 (10:29 +0000)]
boot_*.8: reference installboot.8, rectify FFS note on sparc64

On sparc64, installboot(8) does not deal with any file system; its -p is a
NOOP and according to CVS history the old sys/arch/sparc64/installboot/
never prepared a file system, either.

Remove this single misleading mention alltogether and rely on the now
consistently cross-linked MI installboot(8) for details, if any.

This syncs the list of currently existing MD boot_*(8/*) pages with the list
of usr.sbin/installboot/*_installboot.c files.

boot_alpha(8/alpha) already references installboot(8) but still uses
sys/arch/alpha/stand/installboot/ instead of usr.sbin/installboot/.

boot_luna88k(8/luna88k) exists but luna88k does not use installboot at all.

installboot(8) has additional support for efi (armv7, arm64, riscv64),
landisk, loongson, octeon and powerpc64 but those architectures currently
lack their own MD boot_*(8/*) manual.

OK miod

2 years agoFix file specification, zap partition bits, rectify disk naming errors
kn [Mon, 5 Sep 2022 10:03:50 +0000 (10:03 +0000)]
Fix file specification, zap partition bits, rectify disk naming errors

In boot_macppc(8), the current format a) claims to support specific disk/MBR
partitions which ofwboot clearly does not parse as such and b) lacks markup
to make clear which optional parts can (not) be omitted in what way.

Fixing both turns
promdev:partition/filename options
into
[[promdev:]filename] [-acds]

Same goes for boot(8/macppc), which has been apparently has been copied
from boot(8/amd64) without accounting for all platform specific details.

On amd64, biosboot(8) sees disks as 'sd' which (accidentially?) matches the
kernel driver's sd(4) name;  it also supports specific disklabel(5) slices,
so amd64 can do 'sd0a:/bsd'.

On macppc, disks show up as 'ide', 'cd' or 'hd' which stems from
OpenFirmware alone, not matching the wd(4) driver's name.  Also, ofwboot
always boots off the 'a' label, so macppc can only do 'hd:/bsd'.

Found while installing OpenBSD/macppc from CD inside QEMU but failing to
boot from disk inside QEMU and consulting our manual pages for help wrt.
specifying a boot device.  Turns out QEMU's OpenBIOS simply cannot boot from
MSDOS filesystems (thanks gkoehler), so it wouldn't work no matter what
boot file specification I'd use.

tests/agreement gkoehler
Feedback OK miod
OK tobhe
OK jmc on a previous boot_macppc.8 diff

2 years agoCleanup USB methods so that they align to what we do in other HC drivers,
mglocker [Mon, 5 Sep 2022 09:00:33 +0000 (09:00 +0000)]
Cleanup USB methods so that they align to what we do in other HC drivers,
like xhci(4) and ehci(4).

ok mpi@

2 years agoAdd minimal tooling for separating tests into dedicated programs.
anton [Mon, 5 Sep 2022 05:34:25 +0000 (05:34 +0000)]
Add minimal tooling for separating tests into dedicated programs.
Used to flag the newly added test_bind_unix_socket() as expected to
fail.

2 years agoRemove dead code.
jsing [Sun, 4 Sep 2022 15:56:51 +0000 (15:56 +0000)]
Remove dead code.

No change in generated assembly.

2 years agoRemove dead code.
jsing [Sun, 4 Sep 2022 15:54:42 +0000 (15:54 +0000)]
Remove dead code.

No change in generated assembly.

2 years agoAdd bounds checks for various EVP cipher implementations.
jsing [Sun, 4 Sep 2022 15:45:25 +0000 (15:45 +0000)]
Add bounds checks for various EVP cipher implementations.

The EVP cipher API uses size_t, however a number of the underlying
implementations use long in their API. This means that an input with
size > LONG_MAX will go negative.

Found by Coverity, hiding under a large pile of macros.

ok tb@

2 years agoRemove dead code.
jsing [Sun, 4 Sep 2022 13:55:39 +0000 (13:55 +0000)]
Remove dead code.

Only change to generated assembly is due to line numbers.

2 years agoExpand the looney M_do_cipher macro.
jsing [Sun, 4 Sep 2022 13:34:13 +0000 (13:34 +0000)]
Expand the looney M_do_cipher macro.

Only change in generated assembly is due to line numbers.

2 years agoRearrange some functions.
jsing [Sun, 4 Sep 2022 13:17:18 +0000 (13:17 +0000)]
Rearrange some functions.

Pull the init_key and ctrl (if present) functions up to the top. This
improves readability and allows for the removal of function prototypes.

No functional change.

2 years agoMake unveil tests less chatty when things work as expected.
anton [Sun, 4 Sep 2022 10:10:20 +0000 (10:10 +0000)]
Make unveil tests less chatty when things work as expected.

2 years agoRemove now unused EVP_C_DATA macro.
jsing [Sun, 4 Sep 2022 09:56:30 +0000 (09:56 +0000)]
Remove now unused EVP_C_DATA macro.

2 years agoRemove now unused BLOCK_CIPHER_* macros.
jsing [Sun, 4 Sep 2022 09:52:22 +0000 (09:52 +0000)]
Remove now unused BLOCK_CIPHER_* macros.

2 years agoExpand BLOCK_CIPHER_* macros.
jsing [Sun, 4 Sep 2022 09:48:23 +0000 (09:48 +0000)]
Expand BLOCK_CIPHER_* macros.

No change in generated assembly.

2 years agoUse pru_send function to check socket splicing compatibility. Only
bluhm [Sun, 4 Sep 2022 09:04:27 +0000 (09:04 +0000)]
Use pru_send function to check socket splicing compatibility.  Only
checking socket type is not sufficient as it could splice together
unix and inet sockets resulting in crashes.  As splicing is about
sending, the same send function looks like a good criteria.
Reported-by: syzbot+fc6901d63d858d5dd00a@syzkaller.appspotmail.com
Reported-by: syzbot+0e026f1bf8b259c6395e@syzkaller.appspotmail.com
OK gnezdo@

2 years agoMechanically expand BLOCK_CIPHER_* macros.
jsing [Sun, 4 Sep 2022 08:57:32 +0000 (08:57 +0000)]
Mechanically expand BLOCK_CIPHER_* macros.

No change in generated assembly.

2 years agoMechanically expand BLOCK_CIPHER_* macros.
jsing [Sun, 4 Sep 2022 08:54:16 +0000 (08:54 +0000)]
Mechanically expand BLOCK_CIPHER_* macros.

No change in generated assembly.

2 years agoImprove periodic USB transfers (device intr, isoc) used for input
mglocker [Sun, 4 Sep 2022 08:42:39 +0000 (08:42 +0000)]
Improve periodic USB transfers (device intr, isoc) used for input
devices, audio, and video.  It's still not perfect, and will need
further improvements.

High level, the diff contains following changes:

* Sync up with the Linux code base, which did re-work the periodic
  scheduling code path.
* Run the driver in IPL_VM instead of IPL_USB to prioritize us before
  lower/equal interrupts (same what NetBSD does).
* Add two new flags to our USB stack required by the updated driver
  code:
  - 'multi' flag in the usbd_hub structure to keep track whether
    a hub has one Transaction Translator for all ports (single TT)
    or one Transaction Translator per port (multi TT).
  - 'hcpriv' pointer in the usbd_tt structure for the HC driver to
    allocate memory for the scheduling depending on single or multi TT.

"go for it" kettenis@

2 years agoUse simpler disklabel template
kn [Sun, 4 Sep 2022 08:24:09 +0000 (08:24 +0000)]
Use simpler disklabel template

2 years agoMechanically expand BLOCK_CIPHER_* ciphers.
jsing [Sun, 4 Sep 2022 08:18:07 +0000 (08:18 +0000)]
Mechanically expand BLOCK_CIPHER_* ciphers.

This includes the wonderful BLOCK_CIPHER_ecb_loop - a for loop in a macro.

No change in generated assembly.

2 years agoDocument disklabel(8) -T's new RAID template
kn [Sun, 4 Sep 2022 08:06:40 +0000 (08:06 +0000)]
Document disklabel(8) -T's new RAID template

OK jmc

2 years agoRemove now unused IMPLEMENT_CFBR macro.
jsing [Sun, 4 Sep 2022 08:06:02 +0000 (08:06 +0000)]
Remove now unused IMPLEMENT_CFBR macro.

2 years agoMake ssl_create_cipher_list() have a single exit
tb [Sun, 4 Sep 2022 07:55:32 +0000 (07:55 +0000)]
Make ssl_create_cipher_list() have a single exit

This simplifies memory management and makes it easier to see the leak
that were introduced in the previous commit. Sprinkle a few malloc
errors for consistency.

CID 278396

with/ok jsing

2 years agoMechanically expand IMPLEMENT_CFBR macros.
jsing [Sun, 4 Sep 2022 07:54:59 +0000 (07:54 +0000)]
Mechanically expand IMPLEMENT_CFBR macros.

Only change to generated assembly is due to the use of EVPerror().

2 years agoFix regression in previous commit.
ajacoutot [Sun, 4 Sep 2022 06:57:13 +0000 (06:57 +0000)]
Fix regression in previous commit.
Moving the configuration check to its own configtest function means rc_pre
exit code will now be that of the && list (sasyncd_flags).
Just do what isakmpd does and return 0.

from lucas at sexy dot is, thanks!

2 years agospelling
jsg [Sun, 4 Sep 2022 06:49:11 +0000 (06:49 +0000)]
spelling

2 years agoProvide PCKBC_AUX to fix compilation
tb [Sat, 3 Sep 2022 23:12:36 +0000 (23:12 +0000)]
Provide PCKBC_AUX to fix compilation

from mlarkin

2 years agoMove PRU_PEERADDR request to (*pru_peeraddr)().
mvs [Sat, 3 Sep 2022 22:43:38 +0000 (22:43 +0000)]
Move PRU_PEERADDR request to (*pru_peeraddr)().

Introduce in{,6}_peeraddr() and use them for inet and inet6 sockets,
except tcp(4) case.

Also remove *_usrreq() handlers.

ok bluhm@

2 years agoInitialize TCP mutex forgotten in previous commit.
bluhm [Sat, 3 Sep 2022 22:11:09 +0000 (22:11 +0000)]
Initialize TCP mutex forgotten in previous commit.
found by Hrvoje Popovski with witness; OK mvs@

2 years agoProperly free() crl & auth tree in parser process
job [Sat, 3 Sep 2022 21:24:02 +0000 (21:24 +0000)]
Properly free() crl & auth tree in parser process

OK claudio@

2 years agoregen
mbuhl [Sat, 3 Sep 2022 21:16:51 +0000 (21:16 +0000)]
regen

2 years agoKeep _fw2 in local md_consoleinfo() scope
kn [Sat, 3 Sep 2022 21:16:05 +0000 (21:16 +0000)]
Keep _fw2 in local md_consoleinfo() scope

Missed in previous
OK kettenis

2 years agoMake recvmmsg and sendmmsg look more alike. change the flag type
mbuhl [Sat, 3 Sep 2022 21:13:48 +0000 (21:13 +0000)]
Make recvmmsg and sendmmsg look more alike. change the flag type
to int like other flag parameters, NetBSD uses unsigned int, FreeBSD
and Linux do int.
OK bluhm@

2 years agoRemove now unused IMPLEMENT_BLOCK_CIPHER macro.
jsing [Sat, 3 Sep 2022 20:12:24 +0000 (20:12 +0000)]
Remove now unused IMPLEMENT_BLOCK_CIPHER macro.

Also remove various comments noting that it cannot be used for certain
block ciphers (which kinda defeats the purpose of having a generic
implementation in the first place).

2 years agoMove the daemon() call in the parent process from after forking the
benno [Sat, 3 Sep 2022 20:07:31 +0000 (20:07 +0000)]
Move the daemon() call in the parent process from after forking the
children to just before.  That way the parent disasociates from its
controling terminal and shell, but not from its children.

Remove the dup2() bits that were copied from daemon() to solve the
problem that the children still had the stdio fds open. This is now
done in the parent earlier.

Remove the setsid() and setpgid(). It is unclear what their intent
was, but they dont seem to make sense, as daemon() covers this as well
and there seems to be no reason the cildren procs need to do that.

ok claudio@ bluhm@

2 years agoMechanically expand IMPLEMENT_BLOCK_CIPHER macro.
jsing [Sat, 3 Sep 2022 20:06:43 +0000 (20:06 +0000)]
Mechanically expand IMPLEMENT_BLOCK_CIPHER macro.

Only change to generated assembly is due to EVPerror()'s use of line
numbers.