openbsd
7 months agodrm/amdgpu: amdgpu_ttm_gart_bind set gtt bound flag
jsg [Thu, 4 Apr 2024 07:26:19 +0000 (07:26 +0000)]
drm/amdgpu: amdgpu_ttm_gart_bind set gtt bound flag

From Philip Yang
6fcd12cb90888ef2d8af8d4c04e913252eee4ef3 in linux-6.6.y/6.6.24
6c6064cbe58b43533e3451ad6a8ba9736c109ac3 in mainline linux

7 months agodrm/panel: do not return negative error codes from drm_panel_get_modes()
jsg [Thu, 4 Apr 2024 07:23:40 +0000 (07:23 +0000)]
drm/panel: do not return negative error codes from drm_panel_get_modes()

From Jani Nikula
a686732df6cecc3561595e6ce7a962a0121eed00 in linux-6.6.y/6.6.24
fc4e97726530241d96dd7db72eb65979217422c9 in mainline linux

7 months agodrm/probe-helper: warn about negative .get_modes()
jsg [Thu, 4 Apr 2024 07:20:55 +0000 (07:20 +0000)]
drm/probe-helper: warn about negative .get_modes()

From Jani Nikula
12bbe2c25cdb571190d523fd814e7696964a84d9 in linux-6.6.y/6.6.24
7af03e688792293ba33149fb8df619a8dff90e80 in mainline linux

7 months agoadd pgprot_decrypted() for 6.6.24 drm
jsg [Thu, 4 Apr 2024 05:43:43 +0000 (05:43 +0000)]
add pgprot_decrypted() for 6.6.24 drm

7 months agoadd cc_platform_has() for 6.6.24 drm
jsg [Thu, 4 Apr 2024 05:38:35 +0000 (05:38 +0000)]
add cc_platform_has() for 6.6.24 drm

7 months agotzset_basic: only call issetugid(2) if TZ has changed
millert [Thu, 4 Apr 2024 02:20:01 +0000 (02:20 +0000)]
tzset_basic: only call issetugid(2) if TZ has changed
If we are just going to return without parsing TZ, there is no need
to call issetugid(2) first.  We only need to call issetugid(2) the
first time TZ is checked or when the value of TZ has changed.
Previously, we called issetugid(2) for every call to the functions
described by localtime(3).  OK deraadt@

7 months agoStopping grabbing the kernel lock in kbind(2).
kettenis [Wed, 3 Apr 2024 22:21:48 +0000 (22:21 +0000)]
Stopping grabbing the kernel lock in kbind(2).

ok mpi@

7 months agoInitialize earlier macppc's and powerpc64's pmap_hash_lock
gkoehler [Wed, 3 Apr 2024 19:30:59 +0000 (19:30 +0000)]
Initialize earlier macppc's and powerpc64's pmap_hash_lock

At boot, the powerpc64 kernel was calling
  pmap_bootstrap -> pmap_kenter_pa -> mtx_enter(&pmap_hash_lock)
before it did
  pmap_init -> mtx_init(&pmap_hash_lock, IPL_HIGH)

Change from mtx_init to MUTEX_INITIALIZER.  This allows an option
WITNESS kernel to boot without warning of an uninitialized mutex.

Also change macppc's pmap_hash_lock from __ppc_lock_init to
PPC_LOCK_INITIALIZER, though WITNESS doesn't see this lock.

ok mpi@

7 months agopmap_virtual_space() and pmap_steal_memory() are mutually exclusive, so
miod [Wed, 3 Apr 2024 18:43:32 +0000 (18:43 +0000)]
pmap_virtual_space() and pmap_steal_memory() are mutually exclusive, so
make sure only one of them is prototyped and only one of them is implemented.
ok mpi@ kettenis@

7 months agoReuse the timeout from struct scsi_xfer rather than declaring one in
miod [Wed, 3 Apr 2024 18:41:38 +0000 (18:41 +0000)]
Reuse the timeout from struct scsi_xfer rather than declaring one in
struct ncr53c9x_ecb.

When the scsi_xfer timeout was introduced *cough* 23 years ago, SCSI drivers
were progressively updated to use this, but for some reason that driver got
missed.

ok dlg@ krw@

7 months agoChange ipmi(4)'s taskq from IPL_NONE to IPL_MPFLOOR
gkoehler [Wed, 3 Apr 2024 18:32:47 +0000 (18:32 +0000)]
Change ipmi(4)'s taskq from IPL_NONE to IPL_MPFLOOR

The IPL_NONE mutex didn't block interrupts that can cause a powerpc64
kernel with option WITNESS to "panic: acquiring blockable sleep lock
with spinlock or critical section held (kernel_lock) &kernel_lock".

ok mpi@

7 months agosync
deraadt [Wed, 3 Apr 2024 15:21:03 +0000 (15:21 +0000)]
sync

7 months agoSwitch one of the tcpmd5 passwords over to use the hex key instead.
claudio [Wed, 3 Apr 2024 08:58:55 +0000 (08:58 +0000)]
Switch one of the tcpmd5 passwords over to use the hex key instead.
This way both modes are tested.

7 months agoRewrite str2key() to not use strtoul() to convert 2 hexdigits into one
claudio [Wed, 3 Apr 2024 08:57:26 +0000 (08:57 +0000)]
Rewrite str2key() to not use strtoul() to convert 2 hexdigits into one
char. Instead use a simple function to do the conversion per nibble.
OK deraadt@ tb@

7 months agoSince ssh-agent(1) is only readable by root by now, use ssh(1) while
anton [Wed, 3 Apr 2024 06:01:11 +0000 (06:01 +0000)]
Since ssh-agent(1) is only readable by root by now, use ssh(1) while
generating data in tests.

7 months agoReading https://github.com/Geal/rust-syslog/issues/79, I came to a
deraadt [Wed, 3 Apr 2024 04:36:53 +0000 (04:36 +0000)]
Reading https://github.com/Geal/rust-syslog/issues/79, I came to a
related conclusion that our syslog_r should not stomp on errno.
The errno being returned from sendsyslog() isn't exactly compatible
with the what a legacy syslog_r() would do here anyways, and it is
better to just be void and non-stomping;
ok millert bluhm

7 months agoFix warning about DistributionPointName type
tb [Wed, 3 Apr 2024 04:20:13 +0000 (04:20 +0000)]
Fix warning about DistributionPointName type

The warning incorrectly mentioned GEN_OTHERNAME, while this is about
fullName vs nameRelativeToCRLIssuer. Also add a comment to indicate
that there's no enum or macros obviate the meaning of the magic 0.

ok claudio job

7 months agoAdd ci_cpuid_level and ci_vendor holding the per-CPU basic cpuid
guenther [Wed, 3 Apr 2024 02:01:21 +0000 (02:01 +0000)]
Add ci_cpuid_level and ci_vendor holding the per-CPU basic cpuid
level and a numeric mapping of the cpu vendor, both from CPUID(0).
Convert the general use of strcmp(cpu_vendor) to simple numeric
tests of ci_vendor.  Track the minimum of all ci_cpuid_level in the
cpuid_level global and continue to use that for what we vmm exposes.

AMD testing help matthieu@ krw@
ok miod@ deraadt@ cheloha@

7 months agoBackout previous commit (intending that libressl client rejects a supported
sthen [Tue, 2 Apr 2024 22:50:54 +0000 (22:50 +0000)]
Backout previous commit (intending that libressl client rejects a supported
groups extension from the server). It triggers 'CONNECT_CR_SRVR_HELLO:tlsv1
alert decode error' when connecting to a (modern) java server (tomcat 10.1.18
on openjdk 17.0.10).

"please revert" tb@

7 months agoImplement and enable TSO in vmx(4)
jan [Tue, 2 Apr 2024 20:59:48 +0000 (20:59 +0000)]
Implement and enable TSO in vmx(4)

Tested with IPv4/IPv6 and vlan(4).

ok jmatthew@

7 months agoAdd check to make sure EAP header length matches expected payload length.
tobhe [Tue, 2 Apr 2024 19:58:28 +0000 (19:58 +0000)]
Add check to make sure EAP header length matches expected payload length.

ok markus@ yasuoka@

7 months agoImplement SO_ACCEPTCONN in getsockopt(2)
claudio [Tue, 2 Apr 2024 14:23:15 +0000 (14:23 +0000)]
Implement SO_ACCEPTCONN in getsockopt(2)
Requested by robert@
OK mvs@ millert@ deraadt@

7 months agoOops, incorrect hex conversion spotted by claudio. While here try to
deraadt [Tue, 2 Apr 2024 12:22:38 +0000 (12:22 +0000)]
Oops, incorrect hex conversion spotted by claudio.  While here try to
improve how it reads a bit better.  Surprising the regression tests
didn't spot this error, maybe it fails to roundtrip the values.

7 months agoRemove wrong "temporary udp error" comment in filt_so{read,write}(). Not
mvs [Tue, 2 Apr 2024 12:21:39 +0000 (12:21 +0000)]
Remove wrong "temporary udp error" comment in filt_so{read,write}(). Not
only udp(4) sockets set and check `so_error'.

No functional changes.

ok bluhm

7 months agoRemove unused file
claudio [Tue, 2 Apr 2024 10:04:28 +0000 (10:04 +0000)]
Remove unused file

7 months agofor parse_ipqos(), use strtonum() instead of mostly idiomatic strtoul(),
deraadt [Tue, 2 Apr 2024 10:02:08 +0000 (10:02 +0000)]
for parse_ipqos(), use strtonum() instead of mostly idiomatic strtoul(),
but wow it's so gross.
ok djm

7 months agocan shortcut by returning strtonum() value directly; ok djm
deraadt [Tue, 2 Apr 2024 09:56:58 +0000 (09:56 +0000)]
can shortcut by returning strtonum() value directly; ok djm

7 months agorewrite convtime() to use a isdigit-scanner and strtonum() instead of
deraadt [Tue, 2 Apr 2024 09:52:14 +0000 (09:52 +0000)]
rewrite convtime() to use a isdigit-scanner and strtonum() instead of
strange strtoul can might be fooled by garage characters.
passes regress/usr.bin/ssh/unittests/misc
ok djm

7 months agoRemove unused ptr[3] char array in pkcs11_decode_hex.
claudio [Tue, 2 Apr 2024 09:48:24 +0000 (09:48 +0000)]
Remove unused ptr[3] char array in pkcs11_decode_hex.
OK deraadt@

7 months agoReplace non-idiomatic strtoul(, 16) to parse a region of 2-character
deraadt [Tue, 2 Apr 2024 09:32:28 +0000 (09:32 +0000)]
Replace non-idiomatic strtoul(, 16) to parse a region of 2-character
hex sequences with a low-level replacement designed just for the task.
ok djm

7 months agoUse strtonum() instead of severely non-idomatic strtoul()
deraadt [Tue, 2 Apr 2024 09:29:31 +0000 (09:29 +0000)]
Use strtonum() instead of severely non-idomatic strtoul()
In particular this will now reject trailing garbage, ie. '12garbage'.
ok djm

7 months agofix unintended sizeof(pointer) introduced in rev 1.41
jsg [Tue, 2 Apr 2024 08:51:06 +0000 (08:51 +0000)]
fix unintended sizeof(pointer) introduced in rev 1.41
found with noderef.cocci, ok deraadt@

7 months agoDelete the msyscall mechanism entirely, since mimmutable+pinsyscalls has
deraadt [Tue, 2 Apr 2024 08:39:16 +0000 (08:39 +0000)]
Delete the msyscall mechanism entirely, since mimmutable+pinsyscalls has
replaced it with a more strict mechanism, which happens to be lockless O(1)
rather than micro-lock O(1)+O(log N).  Also nop-out the sys_msyscall(2) guts,
but leave the syscall around for a bit longer so that people can build through
it, since ld.so(1) still wants to call it.

7 months agoremove useless whitespace; from Jia Tan
deraadt [Tue, 2 Apr 2024 08:27:22 +0000 (08:27 +0000)]
remove useless whitespace; from Jia Tan

7 months agoalso relink ssh-agent
deraadt [Tue, 2 Apr 2024 08:21:04 +0000 (08:21 +0000)]
also relink ssh-agent

7 months agosndiod: Migrate clients upon watchdog timeout
ratchov [Tue, 2 Apr 2024 05:32:10 +0000 (05:32 +0000)]
sndiod: Migrate clients upon watchdog timeout

When -F is used, fixes the difference in behavior between watchdog
timer expiry and other errors.

7 months agosndiod: Fix confusion between the slot and ctlslot pointers
ratchov [Tue, 2 Apr 2024 05:21:32 +0000 (05:21 +0000)]
sndiod: Fix confusion between the slot and ctlslot pointers

Fixes crash when the device is disconnected and the clients are not
migrated to another device.

7 months agoSimplify ecx_cms_sign_or_verify() using X509_ALGOR_set0_by_nid()
tb [Tue, 2 Apr 2024 04:04:07 +0000 (04:04 +0000)]
Simplify ecx_cms_sign_or_verify() using X509_ALGOR_set0_by_nid()

ok jsing

7 months agoinit var to fix missing symtab section path
jsg [Mon, 1 Apr 2024 22:49:04 +0000 (22:49 +0000)]
init var to fix missing symtab section path
found by smatch, ok mpi@

7 months agoAdvertise SMTPUTF8 and 8BITMIME extensions in EHLO.
millert [Mon, 1 Apr 2024 21:09:44 +0000 (21:09 +0000)]
Advertise SMTPUTF8 and 8BITMIME extensions in EHLO.
Fixes potential interoperability issues when the real MTA
supports those extensions.  From Arnt Gulbrandsen, OK benno@

7 months agopinsyscalls appeared in 7.5
benno [Mon, 1 Apr 2024 19:29:16 +0000 (19:29 +0000)]
pinsyscalls appeared in 7.5
ok deraadt

7 months agoThe regress/sys/arch/amd64/fpu/fenv.S utility issues "direct" syscalls
anton [Mon, 1 Apr 2024 18:52:34 +0000 (18:52 +0000)]
The regress/sys/arch/amd64/fpu/fenv.S utility issues "direct" syscalls
which does not go through libc. This is no longer possible due to
pinsyscalls(2). Borrow the PINSYSCALL asm macro from libc/arch/DEFS.h to
setup a .openbsd.syscalls section.

7 months agosync
deraadt [Mon, 1 Apr 2024 18:10:47 +0000 (18:10 +0000)]
sync

7 months agoalso create a relink kit for ssh-agent, since it is a long-running setgid
deraadt [Mon, 1 Apr 2024 15:50:17 +0000 (15:50 +0000)]
also create a relink kit for ssh-agent, since it is a long-running setgid
program carrying keys with some (not very powerful) communication channels.
solution for testing the binary from dtucker.  agreement from djm.
Will add it into /etc/rc in a few days.

7 months agonew-style relink kit for sshd. The old scheme created a Makefile by
deraadt [Mon, 1 Apr 2024 15:48:16 +0000 (15:48 +0000)]
new-style relink kit for sshd.  The old scheme created a Makefile by
concatenating two Makefiles and was incredibly fragile.  In the new way
a narrow-purposed install.sh script is created and shipped with the objects.
A recently commited /etc/rc script understands these files.

7 months agoThe previous unveil feature test assumed that having the
florian [Mon, 1 Apr 2024 14:24:30 +0000 (14:24 +0000)]
The previous unveil feature test assumed that having the
unistd.h header file was enough. The new feature tests also
requires that the required unveil system calls are
supported.

7 months agoAdd some #ifdef XML_TESTING to the libexpat tests.
bluhm [Mon, 1 Apr 2024 13:46:06 +0000 (13:46 +0000)]
Add some #ifdef XML_TESTING to the libexpat tests.

Upstream builds special libexpat objects for testing with an internal
flag and counter.  As OpenBSD does test the installed production
library, disable tests that depend on special options.  Use the
same define that activates the test code in the library.  Without
this, regress would not compile.

7 months agoUpdate libexpat to version 2.6.2.
bluhm [Mon, 1 Apr 2024 13:30:41 +0000 (13:30 +0000)]
Update libexpat to version 2.6.2.

The fix for CVE-2024-28757 has been applied earlier.  Relevant for
OpenBSD are bug fixes #839 #841, and other change #829.  No library
bump is necessary.

OK deraadt@

7 months agoMention that port can be specified as 0 in some situations, so that the
deraadt [Mon, 1 Apr 2024 12:40:18 +0000 (12:40 +0000)]
Mention that port can be specified as 0 in some situations, so that the
system will allocate a port.

7 months agoEnforce the pinsyscall rules on non-static/ld.so/libc.so text segments
deraadt [Mon, 1 Apr 2024 12:00:15 +0000 (12:00 +0000)]
Enforce the pinsyscall rules on non-static/ld.so/libc.so text segments
also. Previously this was handled by msyscall(2), but that was just
removed and anton fond out a piece of enforcement was missed..

7 months agoAdd clocks for the RK3588 PWM controller.
patrick [Mon, 1 Apr 2024 11:16:11 +0000 (11:16 +0000)]
Add clocks for the RK3588 PWM controller.

ok kettenis@

7 months agosync
deraadt [Mon, 1 Apr 2024 05:47:03 +0000 (05:47 +0000)]
sync

7 months agoDelete 108 lines of ASM from vmx_enter_guest() that predated lots
guenther [Mon, 1 Apr 2024 05:11:49 +0000 (05:11 +0000)]
Delete 108 lines of ASM from vmx_enter_guest() that predated lots
of later enhancements, removing the save/restore of flags, selectors,
and MSRs: flags are caller-saved and don't need restoring while
selectors and MSRs are auto-restored.  The FSBASE, GSBASE, and
KERNELGSBASE MSRs just need the correct values set with vmwrite()
in the "on new CPU?" block of vcpu_run_vmx().

Also, only rdmsr(MSR_MISC_ENABLE) once in vcpu_reset_regs_vmx(),
give symbolic names to the exit-load MSR slots, eliminate
VMX_NUM_MSR_STORE, and #if 0 the vc_vmx_msr_entry_load_{va,pa} code
and definitions as unused.

ok dv@

7 months agoEnable rge(4).
kettenis [Sun, 31 Mar 2024 19:17:53 +0000 (19:17 +0000)]
Enable rge(4).

7 months agoCombine route_cache() and rtalloc_mpath() in new route_mpath().
bluhm [Sun, 31 Mar 2024 15:53:12 +0000 (15:53 +0000)]
Combine route_cache() and rtalloc_mpath() in new route_mpath().

Fill and check the cache and call rtalloc_mpath() together.  Then
the caller of route_mpath() does not have to care about the uint32_t
*src pointer and just pass struct in_addr.  All the conversions are
done inside the functions.

A previous version of this diff was backed out.  There was an
additional rtisvalid() in rtalloc_mpath() that prevented packet
output via interfaces that were not up.  Now the route in the cache
has to be valid, but after new lookup, rtalloc_mpath() may return
invalid routes.  This generates less errors in userland an preserves
existing behavior.

OK sashan@

7 months agoAllow listen(2) only on sockets of type SOCK_STREAM or SOCK_SEQPACKET.
mvs [Sun, 31 Mar 2024 14:01:28 +0000 (14:01 +0000)]
Allow listen(2) only on sockets of type SOCK_STREAM or SOCK_SEQPACKET.
listen(2) man(1) page clearly prohibits sockets of other types.

Reported-by: syzbot+00450333592fcd38c6fe@syzkaller.appspotmail.com
ok bluhm

7 months agoMark `so_rcv' sockbuf of udp(4) sockets as SB_OWNLOCK.
mvs [Sun, 31 Mar 2024 13:50:00 +0000 (13:50 +0000)]
Mark `so_rcv' sockbuf of udp(4) sockets as SB_OWNLOCK.

sbappend*() and soreceive() of SB_MTXLOCK marked sockets uses `sb_mtx'
mutex(9) for protection, meanwhile buffer usage check and corresponding
sbwait() sleep still serialized by solock(). Mark udp(4) as SB_OWNLOCK
to avoid solock() serialization and rely to `sb_mtx' mutex(9). The
`sb_state' and `sb_flags' modifications must be protected by `sb_mtx'
too.

ok bluhm

7 months agoRegen
miod [Sun, 31 Mar 2024 10:14:46 +0000 (10:14 +0000)]
Regen

7 months agoFix /dev/bio major.
miod [Sun, 31 Mar 2024 10:14:35 +0000 (10:14 +0000)]
Fix /dev/bio major.

7 months agoRemove long dead isa-on-powerpc leftovers.
miod [Sun, 31 Mar 2024 09:49:33 +0000 (09:49 +0000)]
Remove long dead isa-on-powerpc leftovers.

7 months agoStop referring to dead-since-30-years gatherstats() in comments.
miod [Sun, 31 Mar 2024 07:23:29 +0000 (07:23 +0000)]
Stop referring to dead-since-30-years gatherstats() in comments.

7 months agoPrevent a recursion inside wakeup(9) when scheduler tracepoints are enabled.
mpi [Sat, 30 Mar 2024 13:33:20 +0000 (13:33 +0000)]
Prevent a recursion inside wakeup(9) when scheduler tracepoints are enabled.

Tracepoints like "sched:enqueue" and "sched:unsleep" were called from inside
the loop iterating over sleeping threads as part of wakeup_proc().  When such
tracepoints were enabled they could result in another wakeup(9) possibly
corrupting the sleepqueue.

Rewrite wakeup(9) in two stages, first dequeue threads from the sleepqueue then
call setrunnable() and possible tracepoints for each of them.

This requires moving unsleep() outside of setrunnable() because it messes with
the sleepqueue.

ok claudio@

7 months agoDocument that pmemrange control data are protected by `uvm.fpageqlock'.
mpi [Sat, 30 Mar 2024 12:47:47 +0000 (12:47 +0000)]
Document that pmemrange control data are protected by `uvm.fpageqlock'.

7 months agoregen
jsg [Sat, 30 Mar 2024 11:13:17 +0000 (11:13 +0000)]
regen

7 months agoadd ids seen on Milk-V Pioneer
jsg [Sat, 30 Mar 2024 11:12:20 +0000 (11:12 +0000)]
add ids seen on Milk-V Pioneer

but not the Cadence ppb, it seems unlikely they asked Cadence to use it
ok deraadt@ kettenis@

7 months agoUse illop1 as failure instruction in the locore gap file.
miod [Sat, 30 Mar 2024 10:43:38 +0000 (10:43 +0000)]
Use illop1 as failure instruction in the locore gap file.

7 months agoPrefer illop1 rather than nop for text padding in the ld.so linker script.
miod [Sat, 30 Mar 2024 10:42:26 +0000 (10:42 +0000)]
Prefer illop1 rather than nop for text padding in the ld.so linker script.

7 months agoProvide illop1 instruction encoding as TRAP.
miod [Sat, 30 Mar 2024 10:41:17 +0000 (10:41 +0000)]
Provide illop1 instruction encoding as TRAP.

7 months agoAdd missing LCRYPTO_ALIAS()
tb [Sat, 30 Mar 2024 10:10:58 +0000 (10:10 +0000)]
Add missing LCRYPTO_ALIAS()

HMAC_CTX_reset() and HMAC_Init() had missing LCRYPTO_ALIAS().

ok beck jsing

7 months agoFix LCRYPTO_UNUSED() in namespace builds
tb [Sat, 30 Mar 2024 10:09:43 +0000 (10:09 +0000)]
Fix LCRYPTO_UNUSED() in namespace builds

If namespace builds are enabled, static links don't work due to missing
_lcry_* symbols. Make LCRYPTO_UNUSED() match LCRYPTO_USED() with an extra
deprecated attribute. This way we can remove the !LIBRESSL_INTERNAL #ifdef
wrapping in public headers.

ok beck joshua

7 months agoInitialize quic_method in tlsexttest
tb [Sat, 30 Mar 2024 09:53:41 +0000 (09:53 +0000)]
Initialize quic_method in tlsexttest

This is only there to pretend a quic method was set on the SSL, but
apparently some compilers warn about an uninitialized variable.

from Christian Andersen

7 months agoReplace silly test1_exit() with exit()
tb [Sat, 30 Mar 2024 09:49:59 +0000 (09:49 +0000)]
Replace silly test1_exit() with exit()

Appeases some compiler.

from Christian Andersen

7 months agoRemove call to lh_FUNCTION_free()
tb [Sat, 30 Mar 2024 09:41:25 +0000 (09:41 +0000)]
Remove call to lh_FUNCTION_free()

This is a wrapper around lh_free() which is already tested.

From Christian Andersen

7 months agoRemove nonsensical comments. NFC
miod [Sat, 30 Mar 2024 08:44:20 +0000 (08:44 +0000)]
Remove nonsensical comments. NFC

7 months agoIn _malloc_init(), round up the region being mprotected RW to the malloc
miod [Sat, 30 Mar 2024 07:50:39 +0000 (07:50 +0000)]
In _malloc_init(), round up the region being mprotected RW to the malloc
page size, rather than relying upon mprotect to round up to the actual mmu
page size.

This repairs malloc operation on systems where the malloc page size
(1 << _MAX_PAGE_SHIFT) is larger than the mmu page size.

ok otto@

7 months agoBring back an inline db_enter(), which makes backtraces easier to grok
miod [Sat, 30 Mar 2024 07:45:21 +0000 (07:45 +0000)]
Bring back an inline db_enter(), which makes backtraces easier to grok
when sh*t hits the fan; per kettenis@ request and forgotten in previous
cleaning commit.

7 months agoImplement else branching logic including 'else if'.
mpi [Sat, 30 Mar 2024 07:41:45 +0000 (07:41 +0000)]
Implement else branching logic including 'else if'.

Statement lists for if & else conditions are now wrapped in a new
'struct bt_cond'.  Handling B_AC_TEST statements moved to stmt_eval()
to handle nested conditional statements.

From Christian Ludwig christian_ludwig at genua.de

7 months agoprogram relinking currently uses a Makefile.relink inside the re-link kit.
deraadt [Sat, 30 Mar 2024 07:35:01 +0000 (07:35 +0000)]
program relinking currently uses a Makefile.relink inside the re-link kit.
For sshd (the only relinked program at the moment), this file is created
in an extremely nasty way.  It'll be better if we have a proper clean
install.sh script, which I've built for sshd.  But let's first commit the
change to /etc/rc which will handle that in the near future.
ok djm

7 months agoin OpenSSH private key format, correct type for subsequent private
djm [Sat, 30 Mar 2024 05:56:22 +0000 (05:56 +0000)]
in OpenSSH private key format, correct type for subsequent private
keys in blob. From Jakub Jelen via GHPR430

7 months agoHide symbols in aes
joshua [Sat, 30 Mar 2024 05:14:12 +0000 (05:14 +0000)]
Hide symbols in aes

ok jsing

7 months agoHide symbols in camellia
joshua [Sat, 30 Mar 2024 04:58:12 +0000 (04:58 +0000)]
Hide symbols in camellia

ok tb

7 months agoRewrite RSA_padding_add_PKCS1_type_{1,2}() with CBS/CBB.
jsing [Sat, 30 Mar 2024 04:34:17 +0000 (04:34 +0000)]
Rewrite RSA_padding_add_PKCS1_type_{1,2}() with CBS/CBB.

ok tb@

7 months agoallow WAYLAND_DISPLAY to enable SSH_ASKPASS
djm [Sat, 30 Mar 2024 04:27:44 +0000 (04:27 +0000)]
allow WAYLAND_DISPLAY to enable SSH_ASKPASS

From dkg via GHPR479; ok dtucker@

7 months agoHide symbols in whrlpool
joshua [Sat, 30 Mar 2024 03:45:47 +0000 (03:45 +0000)]
Hide symbols in whrlpool

ok tb

7 months agoDo multiple padding add/check iterations.
jsing [Sat, 30 Mar 2024 02:20:39 +0000 (02:20 +0000)]
Do multiple padding add/check iterations.

This increases the chance of triggering bugs due to input and/or randomised
padding.

7 months agoHide symbols in CMS
joshua [Sat, 30 Mar 2024 01:53:05 +0000 (01:53 +0000)]
Hide symbols in CMS

ok jsing

7 months agouse void in function decl with no args; avoids non-ANSI smatch warnings
jsg [Sat, 30 Mar 2024 01:13:56 +0000 (01:13 +0000)]
use void in function decl with no args; avoids non-ANSI smatch warnings

7 months agoAdd initial regress for RSA padding.
jsing [Sat, 30 Mar 2024 00:36:14 +0000 (00:36 +0000)]
Add initial regress for RSA padding.

7 months agoRemove pointless guard in cms header
joshua [Sat, 30 Mar 2024 00:35:15 +0000 (00:35 +0000)]
Remove pointless guard in cms header

ok jsing

7 months agoTidy indentation.
jsing [Sat, 30 Mar 2024 00:34:40 +0000 (00:34 +0000)]
Tidy indentation.

Requested by tb@

7 months agoRemove -app2_1 flags being passed to dsa and rsa regress.
jsing [Sat, 30 Mar 2024 00:32:13 +0000 (00:32 +0000)]
Remove -app2_1 flags being passed to dsa and rsa regress.

Nothing does flag handling in these tests.

7 months agoUse SBI calls to reboot or power down the machine when the firmware
kettenis [Fri, 29 Mar 2024 22:11:34 +0000 (22:11 +0000)]
Use SBI calls to reboot or power down the machine when the firmware
supports them.

ok jca@

7 months agoSkip calculating timings if we can't determine the frequency of the clock.
kettenis [Fri, 29 Mar 2024 22:08:09 +0000 (22:08 +0000)]
Skip calculating timings if we can't determine the frequency of the clock.
This avoids programming the hardware with clearly wrong values.  On the
Milk-V Pioneer this means the I2C controller remains functional and
rebooting using the appropriate SBI call still works.

ok miod@

7 months agoDrop the first argument of intr_establish().
miod [Fri, 29 Mar 2024 21:29:33 +0000 (21:29 +0000)]
Drop the first argument of intr_establish().

Instead, require all callers to put the right value in the ih_pil field, and
have intr_establish() trust them rather than assigning this field again from
its first argument.

ok claudio@ kettenis@

7 months agoStore the physical address of each pcb in struct mdproc, and use this in
miod [Fri, 29 Mar 2024 21:27:53 +0000 (21:27 +0000)]
Store the physical address of each pcb in struct mdproc, and use this in
order to speed up window spills, rather than doing an inline pmap_extract
(well, pseg_get).

ok claudio@ kettenis@

7 months agoRemove truly unneeded includes (not included indirectly).
miod [Fri, 29 Mar 2024 21:26:38 +0000 (21:26 +0000)]
Remove truly unneeded includes (not included indirectly).

ok claudio@ kettenis@

7 months agoCheck for %otherwin being nonzero earlier in rft_user.
miod [Fri, 29 Mar 2024 21:25:55 +0000 (21:25 +0000)]
Check for %otherwin being nonzero earlier in rft_user.

There is one code path using it in %g2 and another using it in %g7.
There is no reason for them to use different registers, and fixing
this allows the check to be performed a bit earlier.

ok claudio@ kettenis@

7 months agoSimplify two more places where we don't need to check for unbiased stacks.
miod [Fri, 29 Mar 2024 21:23:17 +0000 (21:23 +0000)]
Simplify two more places where we don't need to check for unbiased stacks.

ok claudio@ kettenis@

7 months agoSTACKFRAME macro is always invoked on a well-formed 64-bit stack, no need
miod [Fri, 29 Mar 2024 21:20:03 +0000 (21:20 +0000)]
STACKFRAME macro is always invoked on a well-formed 64-bit stack, no need
to check for missing BIAS.

ok claudio@ kettenis@

7 months agoText faults should not invoke uvm_grow() since the fault address is not on
miod [Fri, 29 Mar 2024 21:19:30 +0000 (21:19 +0000)]
Text faults should not invoke uvm_grow() since the fault address is not on
stack. Remove duplicated "panic if uvm_fault() fails and we are in kernel mode"
blocks.

ok claudio@ kettenis@