openbsd
3 years agoAllow for different (but POSIX compliant) behaviour of basename(3) and
dtucker [Wed, 4 Aug 2021 01:34:55 +0000 (01:34 +0000)]
Allow for different (but POSIX compliant) behaviour of basename(3) and
prevent a use-after-free in that case in the new sftp-compat code.

POSIX allows basename(3) to either return a pointer to static storage
or modify the passed string and return a pointer to that.  OpenBSD does
the former and works as is, but on other platforms "filename" points
into "tmp" which was just freed.  This makes the freeing of tmp
consistent with the other variable in the loop.

Pinpointed by the -portable Valgrind regress test.  ok djm@ deraadt@

3 years agoDocument X509_get_default_cert_dir_env(3)
schwarze [Tue, 3 Aug 2021 19:47:39 +0000 (19:47 +0000)]
Document X509_get_default_cert_dir_env(3)
and X509_get_default_cert_file_env(3).

LibreSSL itself does not call getenv(3), but a few application programs
including epic5, fetchmail, fossil, slic3r call these functions, so in
case programmers find them in existing code, telling them what they do
seems useful.

3 years agoDocument X509_get_default_cert_area(3).
schwarze [Tue, 3 Aug 2021 18:49:30 +0000 (18:49 +0000)]
Document X509_get_default_cert_area(3).
Put it into this page because this is the code actually using it.
Despite its name and include file, it is unrelated to X.509
and unrelated to certificates: it is just the default directory
containing the library configuration file, openssl.cnf(5).

3 years agoIncrease default data bytes limit for Child SAs to 4 GB.
tobhe [Tue, 3 Aug 2021 12:46:30 +0000 (12:46 +0000)]
Increase default data bytes limit for Child SAs to 4 GB.
Lower limits lead to excessive rekeying and lost data in high performance
setups without much benefit.

Brought up by mvs@
ok patrick@ sthen@

3 years agoMake sure VM_MAXUSER_ADDRESS is aligned to a page boundary, as MI code
semarie [Tue, 3 Aug 2021 04:45:54 +0000 (04:45 +0000)]
Make sure VM_MAXUSER_ADDRESS is aligned to a page boundary, as MI code
relies upon this; sparc64 turned out to be the only platform where this
was not the case.

help, diff, and commit message from Miod Vallat
"makes sense" kettenis@

3 years agosync
deraadt [Tue, 3 Aug 2021 04:43:30 +0000 (04:43 +0000)]
sync

3 years agoregression tests for scp SFTP protocol support; mostly by Jakub
djm [Tue, 3 Aug 2021 01:05:24 +0000 (01:05 +0000)]
regression tests for scp SFTP protocol support; mostly by Jakub
Jelen in GHPR#194 ok markus

3 years agosupport for using the SFTP protocol for file transfers in scp, via a
djm [Mon, 2 Aug 2021 23:38:27 +0000 (23:38 +0000)]
support for using the SFTP protocol for file transfers in scp, via a
new "-M sftp" option. Marked as experimental for now.

Some corner-cases exist, in particular there is no attempt to
provide bug-compatibility with scp's weird "double shell" quoting
rules.

Mostly by Jakub Jelen in GHPR#194 with some tweaks by me. ok markus@
Thanks jmc@ for improving the scp.1 bits.

3 years agoMP support seems to work fine now
kettenis [Mon, 2 Aug 2021 21:46:39 +0000 (21:46 +0000)]
MP support seems to work fine now

ok deraadt@

3 years agoDon't call rtm_ifchg() in trunk_port_state().
mvs [Mon, 2 Aug 2021 21:10:55 +0000 (21:10 +0000)]
Don't call rtm_ifchg() in trunk_port_state().

The preceding trunk_link_active() already produced RTM_IFINFO message when
trunk(4) state was changed. I such case we double RTM_IFINFO message or we
produce false message when trunk(4) state was not changed.

ok florian@

3 years agoAdd memory barrier (data FENCE) before making the SBI call to issue a
kettenis [Mon, 2 Aug 2021 19:07:29 +0000 (19:07 +0000)]
Add memory barrier (data FENCE) before making the SBI call to issue a
FENCE.I instruction on the remote harts.  According to the RISC-V
ISA documentation this is necessary to make remote harts observe earlier
stores.  Also issue the local FENCE.I first; this is what both FreeBSD and
Linux does.  Seems to fix at least some of the issue we've seen running
GENERIC.MP on MP machines.

ok jca@

3 years agorkpwm(4) device tree bindings were updated to not have clock names anymore,
patrick [Mon, 2 Aug 2021 18:22:41 +0000 (18:22 +0000)]
rkpwm(4) device tree bindings were updated to not have clock names anymore,
so simply look up the first clock instead of looking for a specific one by
name.  Allows rkpwm(4) to find the clock when using a recent device tree.

From Tomasz Bielecki
ok kettenis@

3 years agoShow if add_path was used when decoding the BGP message in MRT format.
claudio [Mon, 2 Aug 2021 16:51:39 +0000 (16:51 +0000)]
Show if add_path was used when decoding the BGP message in MRT format.
This may help to spot encoding errors a bit more easily.

3 years agoShuffle conditions around to help my brain.
claudio [Mon, 2 Aug 2021 16:42:13 +0000 (16:42 +0000)]
Shuffle conditions around to help my brain.

3 years agotweaks regarding X509_LOOKUP_by_subject(3):
schwarze [Mon, 2 Aug 2021 16:29:27 +0000 (16:29 +0000)]
tweaks regarding X509_LOOKUP_by_subject(3):
* document the X509_OBJECT output parameter
* more precision regarding return values
* clarify relationship with X509_LOOKUP_ctrl(3) for the dir lookup method

3 years agonew manual page X509_STORE_get_by_subject(3)
schwarze [Mon, 2 Aug 2021 16:21:11 +0000 (16:21 +0000)]
new manual page X509_STORE_get_by_subject(3)

3 years agoDon't call cpu_setperf() when reading hw.setperf.
tb [Mon, 2 Aug 2021 15:15:47 +0000 (15:15 +0000)]
Don't call cpu_setperf() when reading hw.setperf.

"makes perfect sense to me" chris
ok gnezdo jca

3 years agoFix incorrect status code for expired mails resulting in a misleading
eric [Mon, 2 Aug 2021 12:33:34 +0000 (12:33 +0000)]
Fix incorrect status code for expired mails resulting in a misleading
bounce report.

From Erik Brens

3 years agoAdd infrastructure to build llvm-objcopy and llvm-objdump.
patrick [Mon, 2 Aug 2021 08:50:51 +0000 (08:50 +0000)]
Add infrastructure to build llvm-objcopy and llvm-objdump.

ok kettenis@ sthen@

3 years agoLeave resolv.conf to resolvd if it learned nameservers
kn [Mon, 2 Aug 2021 01:29:26 +0000 (01:29 +0000)]
Leave resolv.conf to resolvd if it learned nameservers

During install (not upgrade) the installer parses resolv.conf looking for
nameservers, FQDN and domain information;  it also prompts for nameservers
if none were found.

If DHCP is used and nameservers were found, resolv.conf gets recreated from
scratch.

Under dhclient(8) this was fine, but resolvd(8) persists learned proposals
across such file modifications and resolv.conf will thus contain a mix
between managed and manual entries -- stale duplicates which users have to
remove manually.

Do not touch resolv.conf anymore in case resolvd knows better and thus
provide a clean file for installations.

OK florian

3 years ago%i is going out of vogue, use %d instead
job [Sun, 1 Aug 2021 22:29:49 +0000 (22:29 +0000)]
%i is going out of vogue, use %d instead

also clean up a comment

OK tb@

3 years agowhitespace
kettenis [Sun, 1 Aug 2021 19:04:37 +0000 (19:04 +0000)]
whitespace

3 years agostyle(9)
kettenis [Sun, 1 Aug 2021 17:42:56 +0000 (17:42 +0000)]
style(9)

3 years agodocument X509_STORE_load_mem(3) and X509_STORE_add_lookup(3)
schwarze [Sun, 1 Aug 2021 15:37:53 +0000 (15:37 +0000)]
document X509_STORE_load_mem(3) and X509_STORE_add_lookup(3)

3 years agosync
deraadt [Sun, 1 Aug 2021 12:43:30 +0000 (12:43 +0000)]
sync

3 years agoAllow trapv2 packages to be submitted over SNMPv3.
martijn [Sun, 1 Aug 2021 11:36:48 +0000 (11:36 +0000)]
Allow trapv2 packages to be submitted over SNMPv3.

This is a type-O that snuck in when merging traphandler into snmpe.

OK jmatthew@

3 years agoSet smi_application in usm's ber as well.
martijn [Sun, 1 Aug 2021 11:30:56 +0000 (11:30 +0000)]
Set smi_application in usm's ber as well.

OK jmatthew@

3 years agoDo not abuse the IMSG_CTL_SEND_REQUEST imsg to transition to
florian [Sun, 1 Aug 2021 09:07:03 +0000 (09:07 +0000)]
Do not abuse the IMSG_CTL_SEND_REQUEST imsg to transition to
REBOOTING. There will be a few more cases internal to dhcpleased that
have nothing to do with the control socket.
While here move requesting a new lease via a call to dhclient under
ifndef SMALL, nothing on the ramdisk uses this.

3 years agoKNF
job [Sun, 1 Aug 2021 00:40:13 +0000 (00:40 +0000)]
KNF

OK benno@

3 years agosync
deraadt [Sat, 31 Jul 2021 16:39:30 +0000 (16:39 +0000)]
sync

3 years agodocument X509_LOOKUP_mem(3) in X509_LOOKUP_hash_dir(3)
schwarze [Sat, 31 Jul 2021 14:54:33 +0000 (14:54 +0000)]
document X509_LOOKUP_mem(3) in X509_LOOKUP_hash_dir(3)
and add a new manual page X509_LOOKUP_new(3)

3 years agoremove mention of BeagleV Starlight
jsg [Sat, 31 Jul 2021 10:57:55 +0000 (10:57 +0000)]
remove mention of BeagleV Starlight

"The BeagleV-Starlight prototype will not be going into mass production"
https://forum.beagleboard.org/t/the-future-of-beaglev-community/30463

3 years agoWe have defines for alert levels - use them instead of magic numbers.
jsing [Sat, 31 Jul 2021 09:31:04 +0000 (09:31 +0000)]
We have defines for alert levels - use them instead of magic numbers.

3 years agodrm/ttm: add a check against null pointer dereference
jsg [Sat, 31 Jul 2021 06:37:43 +0000 (06:37 +0000)]
drm/ttm: add a check against null pointer dereference

From Zheyu Ma
5c3d753b872a706af9317fef4edfb6b640d8a71e in linux 5.10.y/5.10.55
9e5c772954406829e928dbe59891d08938ead04b in mainline linux

3 years agoFix displaying incorrect patterns on LUNA's wscons with 1bpp
aoyama [Sat, 31 Jul 2021 05:22:36 +0000 (05:22 +0000)]
Fix displaying incorrect patterns on LUNA's wscons with 1bpp
framebuffer when Backspace is typed.

Reading data from LUNA framebuffer's 'common write plane' is not
valid.  But on 1bpp framebuffer routine attempts to read from common
write plane in macro.  That causes displaying incorrect patterns.

This bug was found on nono's LUNA-88K emulation first, then inspected
on the real hardware after I fortunately got 1bpp framebuffer.

Spotted and investigated by Isaki and Sugahara of nono procject.
Tested on my LUNA-88K2.

3 years agoAdd newline for debug printf.
mglocker [Fri, 30 Jul 2021 18:56:01 +0000 (18:56 +0000)]
Add newline for debug printf.

3 years agoIncrease static number of nframes which enables very initial ISOC support,
mglocker [Fri, 30 Jul 2021 18:05:24 +0000 (18:05 +0000)]
Increase static number of nframes which enables very initial ISOC support,
e.g. uvideo(4) with lower resolutions.

In general we might need to re-write parts of the nframes handling in the
driver, since the NetBSD nframes transfer allocation doesn't match with our
USB stack.

With this we can at least start further testing and improvement for ISOC
support.

3 years agoAdd checks when reading VRPs with a maxLength via RTR
job [Fri, 30 Jul 2021 15:34:37 +0000 (15:34 +0000)]
Add checks when reading VRPs with a maxLength via RTR

Ensure the maxLength is greater than or equal to the length of
the accompanying prefix, and less than or equal to the length
(in bits) of an IP address in the address family (32 for IPv4
and 128 for IPv6). The same checks are applied when reading VRPs
from configuration on disk.

OK claudio@

3 years agoMove the explanations related to *ptree closer together and correct
schwarze [Fri, 30 Jul 2021 15:01:40 +0000 (15:01 +0000)]
Move the explanations related to *ptree closer together and correct
the lie that *ptree is set upon success - in some cases of success,
it is set to NULL, whereas in some cases of failure, a non-trivial
tree may be returned.

beck@ pointed out that statements related to *ptree were scattered
all over the place, and this patch works for him.

3 years agofix a formatting error and add some Xr;
jmc [Fri, 30 Jul 2021 14:28:13 +0000 (14:28 +0000)]
fix a formatting error and add some Xr;
from debian at helgefjell de

removed references to rlogin etc. as no longer relevant;
suggested by djm

ok djm dtucker

3 years agofix a formatting error and mark up known_hosts consistently;
jmc [Fri, 30 Jul 2021 14:25:01 +0000 (14:25 +0000)]
fix a formatting error and mark up known_hosts consistently;
issues reported by debian at helgefjell de

ok djm dtucker

3 years agoPrint iwx(4) firmware version numbers like Linux iwlwifi does it.
stsp [Fri, 30 Jul 2021 14:09:12 +0000 (14:09 +0000)]
Print iwx(4) firmware version numbers like Linux iwlwifi does it.

Starting with major version 35 the Linux driver prints the minor version
number in hexadecimal.

Same change was made for iwm(4) in CVS commit LCM6R5u9jeF8bcXB

3 years agoRemove an unused 320k bytes DMA memory allocation from iwx(4).
stsp [Fri, 30 Jul 2021 13:56:44 +0000 (13:56 +0000)]
Remove an unused 320k bytes DMA memory allocation from iwx(4).

This allocation was left over from code inherited from iwm(4) where
it is used for transferring firmware code to the device. Devices
supported by iwx(4) use an entirely different mechanism for loading
firmware and don't need this allocation at all.

Based on a patch by zxystd from the OpenIntelWireless project.

3 years agoUse inline function for SFENCE.VMA instruction. Fixes missing "memory"
kettenis [Fri, 30 Jul 2021 13:17:33 +0000 (13:17 +0000)]
Use inline function for SFENCE.VMA instruction.  Fixes missing "memory"
clobber in the inline assembly.

3 years agoCleanup ptrace-related code; still untested.
kettenis [Fri, 30 Jul 2021 13:10:12 +0000 (13:10 +0000)]
Cleanup ptrace-related code; still untested.

3 years agoFix the transfer abort function dwc2_abort_xfer() to work again with the
mglocker [Fri, 30 Jul 2021 12:33:27 +0000 (12:33 +0000)]
Fix the transfer abort function dwc2_abort_xfer() to work again with the
recently updated code.  There, sync the hardware specific parts with the
NetBSD driver.

3 years agoFix the show rib detail output. The add_path condition was totally wrong.
claudio [Fri, 30 Jul 2021 09:45:52 +0000 (09:45 +0000)]
Fix the show rib detail output. The add_path condition was totally wrong.

3 years agodocument locate.mklocatedb as well
espie [Fri, 30 Jul 2021 09:28:26 +0000 (09:28 +0000)]
document locate.mklocatedb as well
tweaks by schwarze@ okay millert@

3 years agoAdd /etc/dhcpleased.conf
bket [Fri, 30 Jul 2021 07:00:02 +0000 (07:00 +0000)]
Add /etc/dhcpleased.conf

OK florian@

3 years agoset the name of the allocating process in struct drm_framebuffer
jsg [Fri, 30 Jul 2021 04:04:25 +0000 (04:04 +0000)]
set the name of the allocating process in struct drm_framebuffer

3 years agowhitespace
tb [Fri, 30 Jul 2021 02:47:37 +0000 (02:47 +0000)]
whitespace

3 years agoDo bounds check before assignment
tb [Fri, 30 Jul 2021 02:46:53 +0000 (02:46 +0000)]
Do bounds check before assignment

As reported by alf (alf.schlichting lemarit com) on bugs@, a bounds check
didn't trigger because it was performed after assignment, so truncation
would happen for large ranges. Fix this and clean up and simplify the code
a bit.

ok deraadt millert

3 years agoAllow relocation R_386_GOTOFF to be used from .debug_info sections. Newer
pascal [Thu, 29 Jul 2021 18:23:41 +0000 (18:23 +0000)]
Allow relocation R_386_GOTOFF to be used from .debug_info sections.  Newer
GCC versions will sometimes emit it, and we have stumbled upon it a few times
in ports.

Upstream commit: https://reviews.llvm.org/D95994
GCC bug report: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98946

ok kettenis@

3 years agomanually tag the macro names that have false positives
schwarze [Thu, 29 Jul 2021 17:31:25 +0000 (17:31 +0000)]
manually tag the macro names that have false positives

3 years agoUpdate the list of firmware files used by iwx(4).
stsp [Thu, 29 Jul 2021 17:06:21 +0000 (17:06 +0000)]
Update the list of firmware files used by iwx(4).

3 years agoImprove the description of .Fl in multiple respects and in paricular
schwarze [Thu, 29 Jul 2021 16:23:43 +0000 (16:23 +0000)]
Improve the description of .Fl in multiple respects and in paricular
improve the .Fl examples in multiple respects, including better
demonstrating long options.
Prompted by a question from espie@.
Feedback and OK jmc@.

3 years agoTreat doas with arguments as a valid SUDO variable. Allows one to
anton [Thu, 29 Jul 2021 15:34:09 +0000 (15:34 +0000)]
Treat doas with arguments as a valid SUDO variable. Allows one to
specify SUDO="doas -n" which I do while running make regress.

ok dtucker@

3 years agoEnsure that the kill signal undergoing testing is not ignored.
anton [Thu, 29 Jul 2021 15:33:17 +0000 (15:33 +0000)]
Ensure that the kill signal undergoing testing is not ignored.

ok bluhm@

3 years agoFix device class.
visa [Thu, 29 Jul 2021 14:11:53 +0000 (14:11 +0000)]
Fix device class.

3 years agoFix a documentation bug i introduced that tb@ pointed out:
schwarze [Thu, 29 Jul 2021 12:39:47 +0000 (12:39 +0000)]
Fix a documentation bug i introduced that tb@ pointed out:
X509_policy_check(3) never returns 2.
If validation succeeds, it always returns 1.

3 years agoSwitch iwx(4) to -63 firmware images as shipped in iwx-firmware-20210512.
stsp [Thu, 29 Jul 2021 12:13:58 +0000 (12:13 +0000)]
Switch iwx(4) to -63 firmware images as shipped in iwx-firmware-20210512.

These images contain fixes which address fragattacks vulnerabilities:
https://www.intel.com/content/www/us/en/security-center/advisory/intel-sa-00473.html

Running fw_update(1) may be required before rebooting into a new kernel.
sysupgrade(8) will take care of this.

When reporting issues, please enable 'ifconfig iwx0 debug', reproduce the
error once more, and include the full dmesg in your bug report.

Tested:
ax200: stsp, kevlo, hrvoje, jmc, Mark Patruck, Ashton Fagg
ax201: kettenis, Fredrik Engberg, Eric Auge

ok kevlo@

3 years agoImplement support for the BAR_FRAME_RELEASE iwx(4) firmware notification.
stsp [Thu, 29 Jul 2021 12:01:45 +0000 (12:01 +0000)]
Implement support for the BAR_FRAME_RELEASE iwx(4) firmware notification.

New firmware will generate this notification when a block ack request is
received. Older firmware passed the block ack request frame to the driver.

ok kevlo@

3 years agoGet Tx queues working with new iwx(4) firmware.
stsp [Thu, 29 Jul 2021 12:01:04 +0000 (12:01 +0000)]
Get Tx queues working with new iwx(4) firmware.

ADD_STA command version >= 12 implies that firmware uses an internal AUX
station for scanning. We do not configure an AUX Tx queue in this case
and data queue indices assigned by firmware shift upwards accordingly.

ok kevlo@

3 years agoSupport the new iwx(4) firmware TLC_MNG_CONFIG command.
stsp [Thu, 29 Jul 2021 12:00:30 +0000 (12:00 +0000)]
Support the new iwx(4) firmware TLC_MNG_CONFIG command.

This command was expanded by a 4 byte max_tx_op field. As far as I can tell
the Linux driver makes no use of this field, so just initialize it to zero.
New firmware panics when we try to initialize Tx rate scaling otherwise.

ok kevlo@

3 years agoSupport the new iwx(4) firmware session protection command.
stsp [Thu, 29 Jul 2021 11:58:35 +0000 (11:58 +0000)]
Support the new iwx(4) firmware session protection command.

Required for having associations succeed with new firmware.

ok kevlo@

3 years agoAdd support for iwx(4) umac scan command version 14.
stsp [Thu, 29 Jul 2021 11:57:59 +0000 (11:57 +0000)]
Add support for iwx(4) umac scan command version 14.

ok kevlo@

3 years agoAdd support for iwx(4) "reduced" scan config command required by new firmware.
stsp [Thu, 29 Jul 2021 11:57:33 +0000 (11:57 +0000)]
Add support for iwx(4) "reduced" scan config command required by new firmware.

ok kevlo@

3 years agoHandle iwx(4) PHY context command version 3.
stsp [Thu, 29 Jul 2021 11:56:53 +0000 (11:56 +0000)]
Handle iwx(4) PHY context command version 3.

ok kevlo@

3 years agoNew iwx(4) firmware doesn't require an AUX station to be added.
stsp [Thu, 29 Jul 2021 11:56:21 +0000 (11:56 +0000)]
New iwx(4) firmware doesn't require an AUX station to be added.

ADD_STA command version >= 12 implies that firmware uses an internal
AUX station for scanning, and firmware panics if we try to add one.

ok kevlo@

3 years agoRemove support for the narrow command header on the iwx(4) command queue.
stsp [Thu, 29 Jul 2021 11:53:46 +0000 (11:53 +0000)]
Remove support for the narrow command header on the iwx(4) command queue.

This is related to the previous commit which fixed "BAD COMMAND" firmware
errors. We can no longer use old-style "narrow" commands on the command
queue with new firmware, and our current -48 firmware images don't seem
to care either way. We can simplify this code and align it with iwlwifi.

ok kevlo@

3 years agoFix "BAD COMMAND" errors with new iwx(4) firmware.
stsp [Thu, 29 Jul 2021 11:52:58 +0000 (11:52 +0000)]
Fix "BAD COMMAND" errors with new iwx(4) firmware.

Firmware API versions >= 50 reject old-style commands in group 0 with a
"BAD_COMMAND" firmware error. We must pretend that such commands were in
the LONG_GROUP instead in order for firmware to accept them.

ok kevlo@

3 years agoHandle iwx(4) firmware alive response version 5.
stsp [Thu, 29 Jul 2021 11:52:11 +0000 (11:52 +0000)]
Handle iwx(4) firmware alive response version 5.

ok kevlo@

3 years agoRecognize "phy integration" and "num stations" TLVs found in new iwx(4)
stsp [Thu, 29 Jul 2021 11:51:39 +0000 (11:51 +0000)]
Recognize "phy integration" and "num stations" TLVs found in new iwx(4)
firmware images. For now, we can simply ignore them while loading firmware.

ok kevlo@

3 years agoIgnore debug sections present in newer iwx(4) firmware images.
stsp [Thu, 29 Jul 2021 11:50:57 +0000 (11:50 +0000)]
Ignore debug sections present in newer iwx(4) firmware images.

ok kevlo@

3 years agoIncrease fixed limits for the number of iwx(4) command versions, channels,
stsp [Thu, 29 Jul 2021 11:50:37 +0000 (11:50 +0000)]
Increase fixed limits for the number of iwx(4) command versions, channels,
and microcode sections. Required for loading new firmware images.

ok kevlo@

3 years agoDeclare additional iwx(4) firmware capability and api support flags.
stsp [Thu, 29 Jul 2021 11:49:53 +0000 (11:49 +0000)]
Declare additional iwx(4) firmware capability and api support flags.

ok kevlo@

3 years agoDocument X509_STORE_set_verify_func(3), mostly using text from the
schwarze [Thu, 29 Jul 2021 10:13:45 +0000 (10:13 +0000)]
Document X509_STORE_set_verify_func(3), mostly using text from the
OpenSSL 1.1.1 branch, which is still under a free license, tweaked
by me.

While here, garbage collect the weird BUGS section.

3 years agoChange driver class from DV_DULL to DV_IFNET, so that it can be recognized
patrick [Thu, 29 Jul 2021 09:19:42 +0000 (09:19 +0000)]
Change driver class from DV_DULL to DV_IFNET, so that it can be recognized
as boot interface when doing netboot.  This makes auto install/upgrade work.

ok kettenis@ visa@

3 years agodocument X509_STORE_CTX_get0_parent_ctx(3)
schwarze [Thu, 29 Jul 2021 09:14:23 +0000 (09:14 +0000)]
document X509_STORE_CTX_get0_parent_ctx(3)

3 years agodocument X509_STORE_CTX_set_app_data(3) and X509_STORE_CTX_get_app_data(3)
schwarze [Thu, 29 Jul 2021 08:32:13 +0000 (08:32 +0000)]
document X509_STORE_CTX_set_app_data(3) and X509_STORE_CTX_get_app_data(3)

3 years agodrm/i915/gvt: Clear d3_entered on elsp cmd submission.
jsg [Thu, 29 Jul 2021 03:15:56 +0000 (03:15 +0000)]
drm/i915/gvt: Clear d3_entered on elsp cmd submission.

From Colin Xu
1df4fe5a8871f49d34d681ff5b7f93a84d50af4b in linux 5.10.y/5.10.54
c90b4503ccf42d9d367e843c223df44aa550e82a in mainline linux

3 years agodrm/amdgpu: update golden setting for sienna_cichlid
jsg [Thu, 29 Jul 2021 03:12:14 +0000 (03:12 +0000)]
drm/amdgpu: update golden setting for sienna_cichlid

From Likun Gao
fc31b5be1383e31ca046fdd6e11e0a9a0b3a01d5 in linux 5.10.y/5.10.54
3e94b5965e624f7e6d8dd18eb8f3bf2bb99ba30d in mainline linux

3 years agodrm: Return -ENOTTY for non-drm ioctls
jsg [Thu, 29 Jul 2021 03:09:17 +0000 (03:09 +0000)]
drm: Return -ENOTTY for non-drm ioctls

From Charles Baylis
69a603aa170e1c145b93d5d7efcca83a8b1268fe in linux 5.10.y/5.10.54
3abab27c322e0f2acf981595aa8040c9164dc9fb in mainline linux

3 years agoA slightly less precise match on the ifconfig output lines works
deraadt [Wed, 28 Jul 2021 20:48:36 +0000 (20:48 +0000)]
A slightly less precise match on the ifconfig output lines works
better for rdomain or metric tuned interfaces
from leon fischer

3 years agoadd format attribute to vaararg functions.
benno [Wed, 28 Jul 2021 19:39:50 +0000 (19:39 +0000)]
add format attribute to vaararg functions.
millert@ thinks its useful.

3 years agoreference veb(4) bridges in tpmr(4). While there improve the description.
benno [Wed, 28 Jul 2021 19:19:38 +0000 (19:19 +0000)]
reference veb(4) bridges in tpmr(4). While there improve the description.
With and ok jmc@

3 years agocleanup some md vs mi differences that accumulated
deraadt [Wed, 28 Jul 2021 17:39:22 +0000 (17:39 +0000)]
cleanup some md vs mi differences that accumulated

3 years agobump rpki-client version to 7.2
benno [Wed, 28 Jul 2021 15:45:34 +0000 (15:45 +0000)]
bump rpki-client version to 7.2

3 years agodocument X509_STORE_CTX_get0_policy_tree(3)
schwarze [Wed, 28 Jul 2021 14:48:09 +0000 (14:48 +0000)]
document X509_STORE_CTX_get0_policy_tree(3)
and X509_STORE_CTX_get_explicit_policy(3)

3 years agodocument X509_policy_tree_free(3)
schwarze [Wed, 28 Jul 2021 14:00:33 +0000 (14:00 +0000)]
document X509_policy_tree_free(3)

3 years agoconsisely explain the meaning of return values rather than merely
schwarze [Wed, 28 Jul 2021 13:47:21 +0000 (13:47 +0000)]
consisely explain the meaning of return values rather than merely
refering to child object names defined in the standard

3 years agoThe clocks for the Rockchip Ethernet controller need to be enabled much
patrick [Wed, 28 Jul 2021 13:43:11 +0000 (13:43 +0000)]
The clocks for the Rockchip Ethernet controller need to be enabled much
earlier.  So far we haven't noticed this, as we had the assumption that
all clocks are enabled anyway.  On the NanoPi R4S this does not seem to
be the case, so we need to bring the clock enable code closer to the
other bringup code.

ok kettenis@

3 years agoOn some RK3399 boards the firmware disables some of the clocks. Allow
patrick [Wed, 28 Jul 2021 13:39:39 +0000 (13:39 +0000)]
On some RK3399 boards the firmware disables some of the clocks.  Allow
some of those clocks to be enabled.

Noticed on the NanoPi R4S, where the Ethernet controller clocks were
surprisingly turned off.

ok kettenis@

3 years agoExplain the meaning of the policy_oids input argument, correct the
schwarze [Wed, 28 Jul 2021 13:39:20 +0000 (13:39 +0000)]
Explain the meaning of the policy_oids input argument, correct the
description of the *pexplicit_policy output argument and make it
less technical, and drop the mention of the expected_policy_set
because the library provides no accessor function for it.

3 years agoSwitch to an MI implementation of kmap_atomic() and implement
kettenis [Wed, 28 Jul 2021 13:28:04 +0000 (13:28 +0000)]
Switch to an MI implementation of kmap_atomic() and implement
kmap_atomic_prot().  Use this to unstub ttm_copy_io_ttm_page()
and ttm_copy_ttm_io_page().  This fixes suspend/resume of machines
with certain radeondrm(4) hardware.

Based on a diff from jsg@.  Tested by Edd Barrett and Alf Schlichting.
ok jsg@

3 years agoAdd RK3399 PCIe PHY reference clocks.
patrick [Wed, 28 Jul 2021 13:10:28 +0000 (13:10 +0000)]
Add RK3399 PCIe PHY reference clocks.

ok kettenis@

3 years agoAdd RK3399 Type-C PHY clocks.
patrick [Wed, 28 Jul 2021 13:04:46 +0000 (13:04 +0000)]
Add RK3399 Type-C PHY clocks.

ok kettenis@

3 years agoAdd more checks for ROA maxLength values
job [Wed, 28 Jul 2021 12:32:14 +0000 (12:32 +0000)]
Add more checks for ROA maxLength values

* maxLength may not be zero or negative
* maxLength can't be lower than the prefixlength of ip_addr
* maxLength can't be higher than the maximum prefix length

OK claudio@

3 years agoUse a more descriptive name for the raw sensor which reflects the number
anton [Wed, 28 Jul 2021 09:56:54 +0000 (09:56 +0000)]
Use a more descriptive name for the raw sensor which reflects the number
of battery levels the device can report.

3 years agoDelay installation of sensors until a device with battery support is
anton [Wed, 28 Jul 2021 09:55:58 +0000 (09:55 +0000)]
Delay installation of sensors until a device with battery support is
connected. Allows sensorsd(8) to pick up hotplugged devices.

Thanks to Laurence Tratt <laurie at tratt dot net> for the report.