openbsd
17 months agosince apparently, the "example" conf is not free-form (adding comments
espie [Tue, 16 May 2023 09:02:50 +0000 (09:02 +0000)]
since apparently, the "example" conf is not free-form (adding comments
would break scripts, go figure), highlight the staging servers so that
people with non-standard configurations (challenge/response) get a chance
at figuring things out.

okay tb@

17 months agoClean up SHA-512 input handling and round macros.
jsing [Tue, 16 May 2023 07:04:57 +0000 (07:04 +0000)]
Clean up SHA-512 input handling and round macros.

Avoid reach around and initialisation outside of the macro, cleaning up
the call sites to remove the initialisation. Use a T2 variable to more
closely follow the documented algorithm and remove the gorgeous compound
statement X = Y += A + B + C.

There is no change to the clang generated assembly on aarch64.

ok tb@

17 months agouser: simplify memsave() to strsave()
millert [Mon, 15 May 2023 17:00:24 +0000 (17:00 +0000)]
user: simplify memsave() to strsave()
All callers of memsave() pass strlen(s) as the size argument.
We can eliminate the size argument and just use strdup(3) instead.
OK tb@

17 months agoImplement the TCP/IP layer for hardware TCP segmentation offload.
bluhm [Mon, 15 May 2023 16:34:56 +0000 (16:34 +0000)]
Implement the TCP/IP layer for hardware TCP segmentation offload.
If the driver of a network interface claims to support TSO, do not
chop the packet in software, but pass it down to the interface
layer.
Precalculate parts of the pseudo header checksum, but without the
packet length.  The length of all generated smaller packets is not
known yet.  Driver and hardware will use the mbuf packet header
field ph_mss to calculate it and update checksum.
Introduce separate flags IFCAP_TSOv4 and IFCAP_TSOv6 as hardware
might support ony one protocol family.  The old flag IFXF_TSO is
only relevant for large receive offload.  It is missnamed, but keep
that for now.
Note that drivers do not set TSO capabilites yet.  Also the ifconfig
flags and pseudo interfaces capabilities will be done separately.
So this commit should not change behavior.
heavily based on the work from jan@; OK sashan@

17 months agocalendar.canada: sync the entry for victoria day with that of calendar.holiday.
jmc [Mon, 15 May 2023 16:22:07 +0000 (16:22 +0000)]
calendar.canada: sync the entry for victoria day with that of calendar.holiday.
as seen this morning, Mon-Third* does not always work. Mon-2 works for now...

calendar.uk: victoria day is obscure enough in scotland that i don;t think it
warrants worth an entry

17 months agoPass the ESR as the trapno value; this might help in determining the detailed
kettenis [Mon, 15 May 2023 15:02:06 +0000 (15:02 +0000)]
Pass the ESR as the trapno value; this might help in determining the detailed
cause of a signal.

ok bluhm@, guenther@, deraadt@

17 months agoEnable BTI PLT entries by default.
kettenis [Mon, 15 May 2023 13:15:17 +0000 (13:15 +0000)]
Enable BTI PLT entries by default.

ok jsg@

17 months agocast to '(long long)' instead of '(long long int)'
op [Mon, 15 May 2023 12:03:04 +0000 (12:03 +0000)]
cast to '(long long)' instead of '(long long int)'

while here adjust the spacing in some of the touched lines.
requested by deraadt@, ok tb@

17 months agocast to '(long long)' instead of '(long long int)'
op [Mon, 15 May 2023 12:02:40 +0000 (12:02 +0000)]
cast to '(long long)' instead of '(long long int)'

requested by deraadt@, ok tb@

17 months agoAdd Juneteenth.
bentley [Mon, 15 May 2023 10:36:08 +0000 (10:36 +0000)]
Add Juneteenth.

ok phessler@ jmc@

17 months agoTurn on pointer-authentication on arm64 as well by default. This means
kettenis [Mon, 15 May 2023 09:58:06 +0000 (09:58 +0000)]
Turn on pointer-authentication on arm64 as well by default.  This means
we effectively enable -mbranch-protection=standard on arm64 now.

ok deraadt@

17 months agosync
deraadt [Mon, 15 May 2023 02:00:50 +0000 (02:00 +0000)]
sync

17 months agoRename arguments of X509_STORE_CTX_init()
tb [Sun, 14 May 2023 20:20:40 +0000 (20:20 +0000)]
Rename arguments of X509_STORE_CTX_init()

It is higly confusing to call the list of untrusted certs chain, when
you're later going to call X509_STORE_CTX_get0_chain() to get a completely
unrelated chain by the verifier. Other X509_STORE_CTX APIs call this list
of certs 'untrusted', so go with that. At the same time, rename the x509
into leaf, which is more explicit.

suggested by/ok jsing

17 months agoImprove emulation of a disktab entry (-t) by using the type
krw [Sun, 14 May 2023 18:34:02 +0000 (18:34 +0000)]
Improve emulation of a disktab entry (-t) by using the type
specified in the disktab entry (:dt:) when setting the default
disklabel's d_type. This will enable installboot(8) to know the
vnd device should be treated as a floppy disk.

Feedback & improved field packing from millert@

17 months agoFix X509error() and X509V3error()
tb [Sun, 14 May 2023 17:20:26 +0000 (17:20 +0000)]
Fix X509error() and X509V3error()

When v3err.c was merged into x509_err.c nearly three years ago, it was
overlooked that the code needed two distinct pairs of ERR_FUNC/ERR_REASON,
one for ERR_LIB_X509 and one for ERR_LIB_X509V3. The result is that the
reason strings for the X509_R_* codes would be overwritten by the ones for
X509V3_R_* with the same value while the reason strings for all X509V3_R_*
would be left undefined.

Fix this by an #undef/#define dance for ERR_LIB_X509V3 once we no longer
the ERR_FUNC/ERR_REASON pair for ERR_LIB_X509.

reported by job
ok jsing

17 months agoSend the linebuffer BIO to the attic
tb [Sun, 14 May 2023 16:36:25 +0000 (16:36 +0000)]
Send the linebuffer BIO to the attic

  *) On VMS, stdout may very well lead to a file that is written to
     in a record-oriented fashion.  That means that every write() will
     write a separate record, which will be read separately by the
     programs trying to read from it.  This can be very confusing.

     The solution is to put a BIO filter in the way that will buffer
     text until a linefeed is reached, and then write everything a
     line at a time, so every record written will be an actual line,
     not chunks of lines and not (usually doesn't happen, but I've
     seen it once) several lines in one record.  BIO_f_linebuffer() is
     the answer.

     Currently, it's a VMS-only method, because that's where it has
     been tested well enough.
     [Richard Levitte]

Yeah, no, we don't care about any of this and haven't compiled this file
since forever. Looks like tedu's chainsaw got blunt at some point...

17 months agoFix another mandoc -Tlint warning
tb [Sun, 14 May 2023 09:33:19 +0000 (09:33 +0000)]
Fix another mandoc -Tlint warning

With this the only -Tlint warnings are about Xr to undocumented functions:
EVP_CIPHER_CTX_copy, EVP_CIPHER_CTX_get_cipher_data, X509V3_EXT_get_nid.

17 months agoRephrase a sentence slightly to apease mandoc -Tlint
tb [Sun, 14 May 2023 09:29:37 +0000 (09:29 +0000)]
Rephrase a sentence slightly to apease mandoc -Tlint

17 months agoFix Xr as BN_is_prime(3) is in the attic
tb [Sun, 14 May 2023 09:27:15 +0000 (09:27 +0000)]
Fix Xr as BN_is_prime(3) is in the attic

17 months agoZap trailing comma
tb [Sun, 14 May 2023 08:03:57 +0000 (08:03 +0000)]
Zap trailing comma

17 months agoX509_policy_tree_level_count(3) is gone
tb [Sun, 14 May 2023 08:03:02 +0000 (08:03 +0000)]
X509_policy_tree_level_count(3) is gone

17 months agoadd missing #include <string.h>; ok tb@
op [Sun, 14 May 2023 07:26:25 +0000 (07:26 +0000)]
add missing #include <string.h>; ok tb@

17 months agogive softnet threads unique names by suffixing softnet with their index.
dlg [Sun, 14 May 2023 01:46:53 +0000 (01:46 +0000)]
give softnet threads unique names by suffixing softnet with their index.

ie, you'll see softnet0, softnet1, etc in top/ps/etc now instead
of just softnet on these threads.

this is done by wrapping the taskq and name up in a softnet struct.

ok patrick@ bluhm@ mvs@ kn@ sashan@

17 months agovmm(4)/vmd(8): switch to anonymous shared mappings.
dv [Sat, 13 May 2023 23:15:28 +0000 (23:15 +0000)]
vmm(4)/vmd(8): switch to anonymous shared mappings.

While splitting out emulated virtio network and block devices into
separate processes, I originally used named mappings via shm_mkstemp(3).
While this functionally achieved the desired result, it had two
unintended consequences:

1) tearing down a vm process and its child processes required
excessive locking as the guest memory was tied into the VFS layer.

2) it was observed by mlarkin@ that actions in other parts of the
VFS layer could cause some of the guest memory to flush to storage,
possibly filling /tmp.

This commit adds a new vmm(4) ioctl dedicated to allowing a process
request the kernel share a mapping of guest memory into its own vm
space. This requires an open fd to /dev/vmm (requiring root) and
both the "vmm" and "proc" pledge(2) promises. In addition, the caller
must know enough about the original memory ranges to reconstruct them
to make the vm's ranges.

Tested with help from Mischa Peters.

ok mlarkin@

17 months agoadd /usr/local/sbin to PATH;
jmc [Sat, 13 May 2023 18:34:49 +0000 (18:34 +0000)]
add /usr/local/sbin to PATH;
confirmed/ok kn

17 months agoIn the absence of the 'disktype' command line parameter always
krw [Sat, 13 May 2023 18:13:42 +0000 (18:13 +0000)]
In the absence of the 'disktype' command line parameter always
use the d_type value provided by the kernel when creating,
editing or printing a disklabel.

i.e. treat d_type as disk geometry is treated, reflecting
current reality as opposed to historical data.

17 months agodefault path includes /usr/local/sbin
kn [Sat, 13 May 2023 16:36:40 +0000 (16:36 +0000)]
default path includes /usr/local/sbin

Improve markup while here.
Feedback tb jmc
OK millert

17 months agoFinally remove the kernel lock from IPv6 neighbor discovery. ND6
bluhm [Sat, 13 May 2023 16:27:59 +0000 (16:27 +0000)]
Finally remove the kernel lock from IPv6 neighbor discovery.  ND6
entries in rt_llinfo are protected either by exclusive netlock or
the ND6 mutex.  The performance critical lookup path in nd6_resolve()
uses shared netlock, but is not lockless.  In contrast to ARP it
grabs the mutex also in the common case.
tested by Hrvoje Popovski; with and OK kn@

17 months agoInstead of implementing IPv4 header checksum creation everywhere,
bluhm [Sat, 13 May 2023 13:35:17 +0000 (13:35 +0000)]
Instead of implementing IPv4 header checksum creation everywhere,
introduce in_hdr_cksum_out().  It is used like in_proto_cksum_out().
OK claudio@

17 months agoPut back in the simplification of the aiodone daemon.
mpi [Sat, 13 May 2023 09:24:59 +0000 (09:24 +0000)]
Put back in the simplification of the aiodone daemon.

Previous "breakage" of the swap on arm64 has been found to be an issue
on one machine the rockpro/arm64 related to a deadlock built into the
sdmmc(4) stack interacting with swapping code both running under
KERNEL_LOCK().

This issue is easily reproducible on -current and entering swap when
building LLVM on a rockpro crashes the machine by memory corruption.

Tested by mlarkin@ on octeon & i386, by myself on amd64 & arm64 and by
sthen@ on i386 port bulk.

ok beck@ some time ago.

Previous commit message:

Simplify the aiodone daemon which is only used for async writes.

- Remove unused support for asynchronous read, including error conditions

- Grab the proper lock for each page that has been written to swap.  This
allows to enable an assertion in uvm_page_unbusy().

- Move the uvm_anon_release() call outside of uvm_page_unbusy() and
assert for the different anon cases.

ok beck@, kettenis@

17 months agoAssert that test->want != NULL at this point
tb [Sat, 13 May 2023 07:17:32 +0000 (07:17 +0000)]
Assert that test->want != NULL at this point

Should make coverity happier

17 months agoadd axppmic support to RAMDISK,
uaa [Fri, 12 May 2023 21:32:49 +0000 (21:32 +0000)]
add axppmic support to RAMDISK,
to support ethernet on OrangePi One Plus (Allwinner H6).

ok kettenis@

17 months agoBob points out that one error should be an X509V3error()
tb [Fri, 12 May 2023 19:02:10 +0000 (19:02 +0000)]
Bob points out that one error should be an X509V3error()

17 months agox509_utl.c: fix some style nits.
tb [Fri, 12 May 2023 18:39:44 +0000 (18:39 +0000)]
x509_utl.c: fix some style nits.

17 months agoAdd interface names in front of error messages.
jan [Fri, 12 May 2023 18:24:13 +0000 (18:24 +0000)]
Add interface names in front of error messages.

Suggested by Hrvoje Popovski.

ok phessler@

17 months agoFall back on loading the kernel from the EFI system partition if we're
kettenis [Fri, 12 May 2023 16:43:00 +0000 (16:43 +0000)]
Fall back on loading the kernel from the EFI system partition if we're
booting from a disk without a BSD disklabel.

ok kn@, caspar@

17 months agovmd(8): fix segfault on vm creation.
dv [Fri, 12 May 2023 16:18:17 +0000 (16:18 +0000)]
vmd(8): fix segfault on vm creation.

vm_instance was using the wrong vm instance for checking the
vm_kernel_path member. Switch to using the value from the parent
vm instance in the check for if a kernel is known.

Issue reported by kn@. OK mlarkin@, kn@.

17 months agovmd(8): fix console attach from vmctl(8).
dv [Fri, 12 May 2023 14:42:30 +0000 (14:42 +0000)]
vmd(8): fix console attach from vmctl(8).

Adding in the ability to override the boot kernel created an edge
case in the ipc message handling logic for the parent process (vmd)
when receiving a "start vm" request. Result was incorrectly responding
to the control process, and as a result the vmctl client, with a
bogus "start vm response" reply with an empty tty name.

This commit rewrites the logic of how vmd goes about processing the
"start vm" request with the aim of making it simpler to understand
while addressing the edge case.

Issue reported by kn@. OK mlarkin@.

17 months agoAdd example user profile btrace script.
claudio [Fri, 12 May 2023 14:34:00 +0000 (14:34 +0000)]
Add example user profile btrace script.
With and OK mpi@

17 months agobtrace(8) support to symbolize utrace addresses.
claudio [Fri, 12 May 2023 14:14:16 +0000 (14:14 +0000)]
btrace(8) support to symbolize utrace addresses.

This only works for a single static binary where everything was compiled
with -fno-omit-frame-pointer since the stack unwinder requires the
frame-pointer.

A possible btrace script to capture performace of a single process is:
profile:hz:100 / pid == $1 / {
        @[ustack] = count();
}
Then using btrace -p program uprofile.bt `pgrep program` will collect
the information for program.

This is far from perfect but should allow other people to play with this
and hopefully improve work.
OK mpi@

17 months agoRewrite string_to_hex() and hex_to_string() using CBB/CBS
tb [Fri, 12 May 2023 13:56:17 +0000 (13:56 +0000)]
Rewrite string_to_hex() and hex_to_string() using CBB/CBS

These helpers used to contain messy pointer bashing some with weird logic
for NUL termination. This can be written more safely and cleanly using
CBB/CBS, so do that. The result is nearly but not entirely identical to
code used elsewhere due to some strange semantics. Apart from errors pushed
on the stack due to out-of-memory conditions, care was taken to preserve
error codes.

ok jsing

17 months agoMake access to rt_llinfo consistent and remove needless initialisation.
bluhm [Fri, 12 May 2023 12:42:16 +0000 (12:42 +0000)]
Make access to rt_llinfo consistent and remove needless initialisation.
OK mvs@

17 months agoAccess rt_llinfo without checking RTF_LLINFO flag before. They are
bluhm [Fri, 12 May 2023 12:40:49 +0000 (12:40 +0000)]
Access rt_llinfo without checking RTF_LLINFO flag before.  They are
always set together with ARP mutex.
OK mvs@

17 months agoregen
jsg [Fri, 12 May 2023 11:43:03 +0000 (11:43 +0000)]
regen

17 months agomatch 6.1.28 drm and remove Intel 0x5698
jsg [Fri, 12 May 2023 11:42:22 +0000 (11:42 +0000)]
match 6.1.28 drm and remove Intel 0x5698

previously DG2-G11, Intel have now decided it is reserved for future use

17 months agodrm/amd/display (gcc13): fix enum mismatch
jsg [Fri, 12 May 2023 11:28:36 +0000 (11:28 +0000)]
drm/amd/display (gcc13): fix enum mismatch

From Jiri Slaby
010842e88269b00b58c6deced7049ff818d73834 in linux-6.1.y/6.1.28
545094d993f4639482018becda5f2a47d126f0ab in mainline linux

17 months agodrm/i915: Make intel_get_crtc_new_encoder() less oopsy
jsg [Fri, 12 May 2023 11:22:54 +0000 (11:22 +0000)]
drm/i915: Make intel_get_crtc_new_encoder() less oopsy

From Ville Syrjala
0fe6ef82e4f4764e8f556632e4cd93d78d448e99 in linux-6.1.y/6.1.28
631420b06597a33c72b6dcef78d1c2dea17f452d in mainline linux

17 months agodrm/amd/display: Fix potential null dereference
jsg [Fri, 12 May 2023 11:21:13 +0000 (11:21 +0000)]
drm/amd/display: Fix potential null dereference

From Igor Artemiev
b75aaebac265e3f29863699d9a929fdfba13d0a4 in linux-6.1.y/6.1.28
52f1783ff4146344342422c1cd94fcb4ce39b6fe in mainline linux

17 months agodrm/ttm/pool: Fix ttm_pool_alloc error path
jsg [Fri, 12 May 2023 11:19:12 +0000 (11:19 +0000)]
drm/ttm/pool: Fix ttm_pool_alloc error path

From Thomas Hellstrom
d2151c5d9dbe3f8fec4cae5f4784edce3ced3a7e in linux-6.1.y/6.1.28
379989e7cbdc7aa7496a00ee286ec146c7599cf0 in mainline linux

17 months agodrm/ttm: optimize pool allocations a bit v2
jsg [Fri, 12 May 2023 11:16:58 +0000 (11:16 +0000)]
drm/ttm: optimize pool allocations a bit v2

From Christian Koenig
5e5a4185c66f9478a8cb2f74eed1a5b5a5000b13 in linux-6.1.y/6.1.28
735c466465eba51deaee3012d8403c10fc7c8c03 in mainline linux

17 months agodrm/amdgpu: register a vga_switcheroo client for MacBooks with apple-gmux
jsg [Fri, 12 May 2023 11:14:00 +0000 (11:14 +0000)]
drm/amdgpu: register a vga_switcheroo client for MacBooks with apple-gmux

From Orlando Chamberlain
78e32896ecc790994034a35a2edef0b6ae043537 in linux-6.1.y/6.1.28
d37a3929ca0363ed1dce02b2772cd5bc547ca66d in mainline linux

17 months agodrm/probe-helper: Cancel previous job before starting new one
jsg [Fri, 12 May 2023 11:09:03 +0000 (11:09 +0000)]
drm/probe-helper: Cancel previous job before starting new one

From Dom Cobley
809a3fb8d8fce5801300ead8aaf52a4ee102e0db in linux-6.1.y/6.1.28
a8e47884f1906cd7440fafa056adc8817568e73e in mainline linux

17 months agodrm/i915/dg2: Drop one PCI ID
jsg [Fri, 12 May 2023 11:06:39 +0000 (11:06 +0000)]
drm/i915/dg2: Drop one PCI ID

From Matt Roper
46473f3bd14a8d6887d2ee56c78aff682bf5d596 in linux-6.1.y/6.1.28
3a38be31ec82920a871963c086393bc0ba26a655 in mainline linux

17 months agodrm/amd/pm: re-enable the gfx imu when smu resume
jsg [Fri, 12 May 2023 11:03:19 +0000 (11:03 +0000)]
drm/amd/pm: re-enable the gfx imu when smu resume

From Tim Huang
f455c9cb9eed246e398977d0d14ac57cd34cefeb in linux-6.1.y/6.1.28
f7f28f268b861c29dd18086bb636abedf0ff59ff in mainline linux

17 months agoswsmu/amdgpu_smu: Fix the wrong if-condition
jsg [Fri, 12 May 2023 11:01:48 +0000 (11:01 +0000)]
swsmu/amdgpu_smu: Fix the wrong if-condition

From Yu Songping
d78777c1d4de809d8bc9efc697cb636cbe8e32c2 in linux-6.1.y/6.1.28
484d7dcc709da46a5976c9530eeff931e9ecba82 in mainline linux

17 months agodrm/amd/display: fix a divided-by-zero error
jsg [Fri, 12 May 2023 11:00:22 +0000 (11:00 +0000)]
drm/amd/display: fix a divided-by-zero error

From Alex Hung
e43cf7abece2efaebe0a757909d258212d01c404 in linux-6.1.y/6.1.28
0b5dfe12755f87ec014bb4cc1930485026167430 in mainline linux

17 months agodrm/amd/display: fix PSR-SU/DSC interoperability support
jsg [Fri, 12 May 2023 10:57:58 +0000 (10:57 +0000)]
drm/amd/display: fix PSR-SU/DSC interoperability support

From Hamza Mahfooz
09c41688b6e5c038df5baacdb0f4e23c8a10ebdb in linux-6.1.y/6.1.28
13b90cf900ab69dd5cab3cc5035bc7614037e64e in mainline linux

17 months agodrm/amd/display: limit timing for single dimm memory
jsg [Fri, 12 May 2023 10:55:23 +0000 (10:55 +0000)]
drm/amd/display: limit timing for single dimm memory

From Daniel Miess
2abff94db2c6d692772fc07f39b4f9ee894c7197 in linux-6.1.y/6.1.28
1e994cc0956b8dabd1b1fef315bbd722733b8aa8 in mainline linux

17 months agodrm/amd/display: Remove stutter only configurations
jsg [Fri, 12 May 2023 10:53:36 +0000 (10:53 +0000)]
drm/amd/display: Remove stutter only configurations

From Nasir Osman
5e1574aa0639abf811375aaba1d6799d035dc97b in linux-6.1.y/6.1.28
71c4ca2d3b079d0ba4d9b3033641fea906cebfb6 in mainline linux

17 months agoadd apple-gmux.h for 6.1.28 drm
jsg [Fri, 12 May 2023 10:50:22 +0000 (10:50 +0000)]
add apple-gmux.h for 6.1.28 drm

17 months agoasn1oct: add a couple more tests
tb [Fri, 12 May 2023 10:43:28 +0000 (10:43 +0000)]
asn1oct: add a couple more tests

17 months agoReduce the number of SHA-512 C implementations from three to one.
jsing [Fri, 12 May 2023 10:10:55 +0000 (10:10 +0000)]
Reduce the number of SHA-512 C implementations from three to one.

We currently have three C implementations for SHA-512 - a version that is
optimised for CPUs with minimal registers (specifically i386), a regular
implementation and a semi-unrolled implementation. Testing on a ~15 year
old i386 CPU, the fastest version is actually the semi-unrolled version
(not to mention that we still currently have an i586 assembly
implementation that is used on i386 instead...).

More decent architectures do not seem to care between the regular and
semi-unrolled version, presumably since they are effectively doing the
same thing in hardware during execution.

Remove all except the semi-unrolled version.

ok tb@

17 months agoasn1oct: minor tweak in error message
tb [Fri, 12 May 2023 08:56:05 +0000 (08:56 +0000)]
asn1oct: minor tweak in error message

17 months agoAdd regress coverage for {s2i,i2s}_ASN1_OCTET_STRING
tb [Fri, 12 May 2023 08:28:05 +0000 (08:28 +0000)]
Add regress coverage for {s2i,i2s}_ASN1_OCTET_STRING

17 months agoprimility -> primality
jsg [Fri, 12 May 2023 08:18:13 +0000 (08:18 +0000)]
primility -> primality
ok tb@

17 months agoBe a bit more precise on how s2i_ASN1_OCTET_STRING handles colons
tb [Fri, 12 May 2023 08:12:09 +0000 (08:12 +0000)]
Be a bit more precise on how s2i_ASN1_OCTET_STRING handles colons

17 months agotest ChrootDirectory in Match block
djm [Fri, 12 May 2023 06:37:42 +0000 (06:37 +0000)]
test ChrootDirectory in Match block

17 months agobetter error messages
djm [Fri, 12 May 2023 06:36:27 +0000 (06:36 +0000)]
better error messages

17 months agostaggered deployment parallelism defaults to number of *online* CPUs
kn [Fri, 12 May 2023 00:06:53 +0000 (00:06 +0000)]
staggered deployment parallelism defaults to number of *online* CPUs

vmd looks at hw.ncpuonline not hw.ncpu[found], which can differ w/o SMT.

17 months agoTeach dump_tables about PKs and display non-readable pages as such
guenther [Thu, 11 May 2023 22:28:38 +0000 (22:28 +0000)]
Teach dump_tables about PKs and display non-readable pages as such

17 months agoPrint the manifestNumber when warning about unrecoverable message digest mismatches
job [Thu, 11 May 2023 20:13:30 +0000 (20:13 +0000)]
Print the manifestNumber when warning about unrecoverable message digest mismatches

This makes it easier to understand "when" certain error conditions arose.

OK tb@

17 months agoRetguard addition in commit 72pYktDvmJhq7OyF was based on the understanding
guenther [Thu, 11 May 2023 19:35:50 +0000 (19:35 +0000)]
Retguard addition in commit 72pYktDvmJhq7OyF was based on the understanding
that the removal of the off_t padding, amd64 syscalls no longer passed a 7th
or later argument.  We overlooked that syscall(2) bumps the arg count by one,
so six argument calls like SYS_sysctl still pass an argument on the stack.

So, repush the 7th argument so it's at the expected stack offset after the
retguard register is pushed.

problem reported and ok bluhm@

17 months agoStop sending a PHY context update in iwx_run_stop()
stsp [Thu, 11 May 2023 16:55:46 +0000 (16:55 +0000)]
Stop sending a PHY context update in iwx_run_stop()

Fixes firmware error iwx0: 0x20101A28 | ADVANCED_SYSASSERT which would
occur while tearing down the state of our association to the AP.

The problem occurred when we were trying to reconfigure the PHY context
in a way that is incompatible with the TLC configuration. Since we will
always call iwx_deauth() next, which removes the AP station and related
data such as TLC from firmware before updating the PHY context yet again
we can simply skip the redundant PHY context update in iwx_run_stop() to
prevent this firmware error.

A reliable way to reproduce the problem is an attempt to connect to an
11ac access point on a 5GHz channel while the wrong WPA key is set which
happened to me because mpi@ uses a French WPA passphrase.
Problem confirmed fixed by me with this simple reproducer.

Helpful hints provided by Johannes Berg at Linux/Intel, thanks!

17 months agoIn valid_cert() also skip the check for CERT_IP_INHERIT objects like
claudio [Thu, 11 May 2023 14:05:31 +0000 (14:05 +0000)]
In valid_cert() also skip the check for CERT_IP_INHERIT objects like
it is done for CERT_AS_INHERIT.

Without this inheritance of IP address resources does not work. Problem
noticed by Ties de Kock (tdekock (at) ripe.net)

OK job@ tb@ benno@

17 months agopools are always initialised, zap overcautious NULL check
kn [Thu, 11 May 2023 12:36:22 +0000 (12:36 +0000)]
pools are always initialised, zap overcautious NULL check

All pools are init'd after pfattach(), none is ever destroyed,
so struct pf_pool_limit's .pp always points to valid pools.

Drop a check for the impossible from twenty years ago.

OK sashan dlg

17 months agomi_syscall() can add the KTRC_CODE_SYSCALL flag to the syscall code.
bluhm [Thu, 11 May 2023 09:51:33 +0000 (09:51 +0000)]
mi_syscall() can add the KTRC_CODE_SYSCALL flag to the syscall code.
In ktrsyscall() mask the code, otherwise kdump(1) does not show the
correct mib of SYS_sysctl when called by syscall(2).
OK guenther@

17 months agotls_verify.c: give up on variable alignment in this file
tb [Thu, 11 May 2023 07:35:27 +0000 (07:35 +0000)]
tls_verify.c: give up on variable alignment in this file

The previous commit resulted in misalignment, which impacts my OCD worse
than no alignment at all. Alignment wasn't consistently done in this file
anyway. op tells me it won't affect current efforts in reducing the diff.

17 months agoDocument recent changes in primality testing
tb [Thu, 11 May 2023 05:45:33 +0000 (05:45 +0000)]
Document recent changes in primality testing

With input from beck and jsing

17 months agonat-to may fail to insert state due to conflict on chosen source
sashan [Wed, 10 May 2023 22:42:51 +0000 (22:42 +0000)]
nat-to may fail to insert state due to conflict on chosen source
port number. This is typically indicated by 'wire key attach failed on...'
message when pf(4) debugging is enabled. The problem is caused by
glitch in pf_get_sport() which fails to discover conflict in advance.
In order to fix it we must also calculate toeplitz hash in
pf_get_sport() to initialize look up key properly.

the bug has been kindly reported by  joosepm _von_ gmail _dot_ com

OK dlg@

17 months agosort the hardware list;
jmc [Wed, 10 May 2023 22:23:25 +0000 (22:23 +0000)]
sort the hardware list;

17 months agoUse is_pseudoprime instead of is_prime in bn_bpsw.c
tb [Wed, 10 May 2023 21:05:24 +0000 (21:05 +0000)]
Use is_pseudoprime instead of is_prime in bn_bpsw.c

This is more accurate and improves readability a bit. Apart from a comment
tweak this is sed + knfmt (which resulted in four wrapped lines).

Discussed with beck and jsing

17 months agoNew udl(4) device, reported by "S V" on tech@
miod [Wed, 10 May 2023 18:28:04 +0000 (18:28 +0000)]
New udl(4) device, reported by "S V" on tech@

17 months agoregen
miod [Wed, 10 May 2023 18:26:43 +0000 (18:26 +0000)]
regen

17 months agoNew udl(4) device, reported by "S V" on tech@
miod [Wed, 10 May 2023 18:26:05 +0000 (18:26 +0000)]
New udl(4) device, reported by "S V" on tech@

17 months agoAdd flag SDEV_UFI so umass_scsi_attach() can provide information
krw [Wed, 10 May 2023 15:28:26 +0000 (15:28 +0000)]
Add flag SDEV_UFI so umass_scsi_attach() can provide information
sufficient to get sdgetdisklabel() to correctly set d_type to
DTYPE_FLOPPY in the default disklabel.

installboot(8) in particular likes to know it is dealing with a
floppy.

ok miod@

17 months agoReset the io_time (io timeout value) when the timer expires. Without this
claudio [Wed, 10 May 2023 15:24:41 +0000 (15:24 +0000)]
Reset the io_time (io timeout value) when the timer expires. Without this
the next connect will immediatly timeout again which is not intended.
It does not matter in the other case since the connection is destroyed anyway.
Problem noticed by job@ and Jelle (jelle (at) luteijn.email)
OK job@ tb@

17 months agoswitch two ASN1_STRING_data() to ASN1_STRING_get0_data()
op [Wed, 10 May 2023 13:48:54 +0000 (13:48 +0000)]
switch two ASN1_STRING_data() to ASN1_STRING_get0_data()

and while here mark as const data.

This diff is actually from gilles@, in OpenSMTPD-portable bundled
libtls.

ok tb@, jsing@

17 months agoDisplay correct type name for Microsoft basic data GPT
krw [Wed, 10 May 2023 13:45:49 +0000 (13:45 +0000)]
Display correct type name for Microsoft basic data GPT
partitions. Inadvertantly lost in transition to new menu logic.

17 months agoMicrosoft GPT partition attributes can appear on partitions other
krw [Wed, 10 May 2023 12:59:47 +0000 (12:59 +0000)]
Microsoft GPT partition attributes can appear on partitions other
than Basic Data partitions. e.g. Windows Recovery and Windows
Reserved partitions.

So tweak verbiage to make clear they are MS attributes by
prepending 'MS' and always display them.

No functional change other than modified fdisk -v output.

17 months agoAdd Miller-Rabin test for random bases to BPSW
tb [Wed, 10 May 2023 12:21:55 +0000 (12:21 +0000)]
Add Miller-Rabin test for random bases to BPSW

The behavior of the BPSW primality test for numbers > 2^64 is not very
well understood. While there is no known composite that passes the test,
there are heuristics that indicate that there are likely infinitely many.
Therefore it seems appropriate to harden the test. Having a settable
number of MR rounds before doing a version of BPSW is also the approach
taken by Go's primality check in math/big.

This adds a new implementation of the old MR test that runs before running
the strong Lucas test. I like to imagine that it's slightly cleaner code.
We're effectively at about twice the cost of what we had a year ago. In
addition, it adds some non-determinism in case there actually are false
positives for the BPSW test.

The implementation is straightforward. It could easily be tweaked to use
the additional gcds in the "enhanced" MR test of FIPS 186-5, but as long
as we are only going to throw away the additional info, that's not worth
much.

This is a first step towards incorporating some of the considerations in
"A performant misuse-resistant API for Primality Testing" by Massimo and
Paterson. Further work will happen in tree. In particular, there are plans
to crank the number of Miller-Rabin tests considerably so as to have a
guaranteed baseline. The manual will be updated shortly.

positive feedback beck
ok jsing

17 months agoImplement TCP send offloading, for now in software only. This is
bluhm [Wed, 10 May 2023 12:07:16 +0000 (12:07 +0000)]
Implement TCP send offloading, for now in software only.  This is
meant as a fallback if network hardware does not support TSO.  Driver
support is still work in progress.  TCP output generates large
packets.  In IP output the packet is chopped to TCP maximum segment
size.  This reduces the CPU cycles used by pf.  The regular output
could be assisted by hardware later, but pf route-to and IPsec needs
the software fallback in general.
For performance comparison or to workaround possible bugs, sysctl
net.inet.tcp.tso=0 disables the feature.  netstat -s -p tcp shows
TSO counter with chopped and generated packets.
based on work from jan@
tested by jmc@ jan@ Hrvoje Popovski
OK jan@ claudio@

17 months agoRemove now-unused prototypes for ssh1 RSA functions. From lengyijun via
dtucker [Wed, 10 May 2023 10:04:20 +0000 (10:04 +0000)]
Remove now-unused prototypes for ssh1 RSA functions.  From lengyijun via
github PR#396.

17 months agocast off_t to long long int for printing
op [Wed, 10 May 2023 08:03:49 +0000 (08:03 +0000)]
cast off_t to long long int for printing

reduces the diff with opensmtpd-portable.
ok tb@

17 months agoconstify day, month and tz
op [Wed, 10 May 2023 08:02:10 +0000 (08:02 +0000)]
constify day, month and tz

spotted while diffing with -portable, where tz is marked as const.
ok tb@

17 months agoAs mmap(2) is no longer a LOCK syscall, do away with the extra
otto [Wed, 10 May 2023 07:58:06 +0000 (07:58 +0000)]
As mmap(2) is no longer a LOCK syscall, do away with the extra
unlock-lock dance it serves no real purpose any more. Confirmed
by a small performance increase in tests.  ok @tb

17 months agoAdd a hint on using letsencrypt-staging for testing
tb [Wed, 10 May 2023 07:34:57 +0000 (07:34 +0000)]
Add a hint on using letsencrypt-staging for testing

While florian rightly points out that this is likely to be overlooked, it
may help someone. Also doesn't affect the adjust-config-with-sed-ability
of the example file after input by a few.

no objection florian
ok espie millert

17 months agocast time_t to long long for printf, reduces the diff with -portable
op [Wed, 10 May 2023 07:20:20 +0000 (07:20 +0000)]
cast time_t to long long for printf, reduces the diff with -portable

ok millert@

17 months agoadd a few missing headers, reduces the diff with -portable
op [Wed, 10 May 2023 07:19:49 +0000 (07:19 +0000)]
add a few missing headers, reduces the diff with -portable

ok millert@

17 months agoremove trailing whitespace
op [Wed, 10 May 2023 07:19:08 +0000 (07:19 +0000)]
remove trailing whitespace

17 months agomacro line fix;
jmc [Tue, 9 May 2023 19:44:06 +0000 (19:44 +0000)]
macro line fix;