]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
2 years agorpc.tlsclntd: Add the -2 option to the usage line
Rick Macklem [Sun, 22 May 2022 21:41:20 +0000 (14:41 -0700)]
rpc.tlsclntd: Add the -2 option to the usage line

Commit 72bf76d6b8c9 added the -2 option to use
TLS 1.2 NFS-over-TLS mounts.  It did not add -2
to the usage message.

This patch adds -2 to the usage message.

MFC after: 2 weeks

2 years agorpc.tlsservd: Add the -2 option to the usage line
Rick Macklem [Sun, 22 May 2022 21:36:20 +0000 (14:36 -0700)]
rpc.tlsservd: Add the -2 option to the usage line

Commit 0637b12b13be added the -2 option to allow
TLS 1.2 NFS-over-TLS mounts.  It did not add -2
to the usage message.

This patch adds -2 to the usage message.

MFC after: 2 weeks

2 years agorpc.tlsclntd: Add the -2 option to the man page
Rick Macklem [Sun, 22 May 2022 21:20:14 +0000 (14:20 -0700)]
rpc.tlsclntd: Add the -2 option to the man page

Since the KTLS now supports TLS1.3, the daemons default
to version 1.3, since the draft (to be an RFC someday)
requires TLS1.3.

However, since FreeBSD 13,0, 13,1 uses TLS1.2 for
NFS-over-TLS, the "-2" option is added to both daemons
for compatibility with FreeBSD 13.0, 13.1.

This patch updates the man pages for this.

This is a content change.

Reviewed by: pauamma_gundo.com
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D35290

2 years agorpc.tlsservd: Add the -2 option to the man page
Rick Macklem [Sun, 22 May 2022 21:17:06 +0000 (14:17 -0700)]
rpc.tlsservd: Add the -2 option to the man page

Since the KTLS now supports TLS1.3, the daemons default
to version 1.3, since the draft (to be an RFC someday)
requires TLS1.3.

However, since FreeBSD 13,0, 13,1 uses TLS1.2 for
NFS-over-TLS, the "-2" option is added to both daemons
for compatibility with FreeBSD 13.0, 13.1.

This patch updates the man pages for this.

This is a content change.

Reviewed by: pauamma_gundo.com
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D35290

2 years agorpc.tlsclntd: Update the man page for the -C option
Rick Macklem [Sun, 22 May 2022 20:58:59 +0000 (13:58 -0700)]
rpc.tlsclntd: Update the man page for the -C option

Commit f5b40aa0dea6 modifies the -C command line option
for rpc.tlsclntd to use the TLS 1.3 SSL_CTX_set_ciphersuites().

This patch updates the man page for this modified command
line option.

This is a content change.

Reviewed by: jhb, pauamma_gundo.com (same change for rpc.tlsservd.8)
MFC after: 2 weeks

2 years agorpc.tlsclntd: Modify the -C option to use SSL_CTX_set_ciphersuites
Rick Macklem [Sun, 22 May 2022 20:49:08 +0000 (13:49 -0700)]
rpc.tlsclntd: Modify the -C option to use SSL_CTX_set_ciphersuites

Commit 0b4f2ab0e913 fixes the krpc so that it can use TLS
version 1.3 for NFS-over-TLS, as required by
the draft (someday to be an RFC).
This patch replaces SSL_CTX_set_cipher_list() with
SSL_CTX_set_ciphersuites(), since that is the function
that is used for TLS1.3.

The man page will be updated in a separate commit.

MFC after:  2 weeks

2 years agorpc.tlsservd: Modify the -C option to use SSL_CTX_set_ciphersuites
Rick Macklem [Sun, 22 May 2022 20:44:31 +0000 (13:44 -0700)]
rpc.tlsservd: Modify the -C option to use SSL_CTX_set_ciphersuites

Commit 0b4f2ab0e913 fixes the krpc so that it can use TLS
version 1.3 for NFS-over-TLS, as required by
the draft (someday to be an RFC).
This patch replaces SSL_CTX_set_cipher_list() with
SSL_CTX_set_ciphersuites(), since that is the function
that is used for TLS1.3.

The man page has already been updated.

MFC after:  2 weeks

2 years agopam_exec: fix segfault when authtok is null
Yan Ka Chiu [Sun, 22 May 2022 16:33:02 +0000 (12:33 -0400)]
pam_exec: fix segfault when authtok is null

According to pam_exec(8), the `expose_authtok` option should be ignored
when the service function is `pam_sm_setcred`. Currently `pam_exec` only
prevent prompt for anth token when `expose_authtok` is set on
`pam_sm_setcred`. This subsequently led to segfault when there isn't an
existing auth token available.

Bug reported on this: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=263893

After reading https://reviews.freebsd.org/rS349556 I am not sure if the
default behaviour supposed to be simply not prompt for authentication
token, or is it to ignore the option entirely as stated in the man page.

This patch is therefore only adding an additional NULL check on the item
`pam_get_item` provide, and exit with `PAM_SYSTEM_ERR` when such item is
NULL.

MFC after: 1 week
Reviewed by: des, khng
Differential Revision: https://reviews.freebsd.org/D35169

2 years agopfctl: fix out-of-bounds access
Jessica Clarke [Sun, 22 May 2022 08:31:42 +0000 (10:31 +0200)]
pfctl: fix out-of-bounds access

If pfctl is called with "pfctl -a ''" we read outside of the anchoropt
buffer. Check that the buffer is sufficiently long to avoid that.

Maintain the existing (and desired, because it's used as such in
/etc/periodic/security/520.pfdenied) behaviour of treating "-a ''" as a
request for the root anchor (or no anchor specified).

PR: 264128
Reviewed by: kp

2 years agorpc.tlsservd: Update the man page for the -C option
Rick Macklem [Sat, 21 May 2022 22:44:29 +0000 (15:44 -0700)]
rpc.tlsservd: Update the man page for the -C option

Commit 712aac1389e8 adds a new -C command line option
to rpc.tlsservd.

This patch updates the man page for this new command
line option.

This is a content change.

Reviewed by: jhb, pauamma_gundo.com
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D35136

2 years agoboot/zfs: Extend zfsimpl.h and make it easier to use
Mark Johnston [Sat, 21 May 2022 18:22:30 +0000 (14:22 -0400)]
boot/zfs: Extend zfsimpl.h and make it easier to use

Some makefs(8) patches make use of zfsimpl.h (not zfsimpl.c though) to
provide definitions for various on-disk structures.  Most of this diff
simply adds new definitions that are useful.

Also reduce dependencies of the header:
- remove an unused list_node_t field to drop the sys/list.h dependency
- replace CTASSERT with _Static_assert
And fix the declaration of decode_embedded_bp_compressed().

No functional change intended.

Reviewed by: tsoome
MFC after: 2 weeks
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D35278

2 years agobsdinstall netconfig_ipv6: Replace dialog with bsddialog
Alfonso S. Siciliano [Sat, 21 May 2022 16:02:20 +0000 (18:02 +0200)]
bsdinstall netconfig_ipv6: Replace dialog with bsddialog

Replace (LGPL) dialog utility with (BSD-2-Clause) bsddialog utility.

Reviewed by: bapt
Differential Revision: https://reviews.freebsd.org/D35270

2 years agobsdinstall netconfig_ipv4: Replace dialog with bsddialog
Alfonso S. Siciliano [Sat, 21 May 2022 15:37:12 +0000 (17:37 +0200)]
bsdinstall netconfig_ipv4: Replace dialog with bsddialog

Replace (LGPL) dialog utility with (BSD-2-Clause) bsddialog utility.

Reviewed by: bapt
Differential Revision: https://reviews.freebsd.org/D35267

2 years agobsdinstall jail: Replace dialog with bsddialog
Alfonso S. Siciliano [Sat, 21 May 2022 15:10:35 +0000 (17:10 +0200)]
bsdinstall jail: Replace dialog with bsddialog

Replace (LGPL) dialog utility with (BSD-2-Clause) bsddialog utility.

Reviewed by: bapt
Differential Revision: https://reviews.freebsd.org/D35274

2 years agobsdinstall services: Improve mouse configuration
Alfonso S. Siciliano [Sat, 21 May 2022 14:37:53 +0000 (16:37 +0200)]
bsdinstall services: Improve mouse configuration

bsdinstall(8) has an option to enable moused support.
However, if it is not selected, moused is still started
through the configuration of nondefault devices.
So, automatically add the moused_nondefault_enable="NO"
setting to rc.conf unless moused support is selected.

PR: 227999
Reported by: bcran
Reviewed by: bapt
Differential Revision: https://reviews.freebsd.org/D35193

2 years agoSet mm before passing it to the UEFI firmware
Andrew Turner [Sat, 21 May 2022 10:45:41 +0000 (11:45 +0100)]
Set mm before passing it to the UEFI firmware

When reading the UEFI memory map we pass in a pointer to the memory to
hold the map. Unfortunately it wasn't initialised before the first use
so clang decided it was undefined behaviour so the entire loop was
removed. This leads to everything in bi_load after this to also be
removed as dead code.

The next function after bi_load in the binary is efi_copy_init. The
above caused us to enter efi_copy_init with a return address of the
start of the function. Because of this it would enter an infinite
loop of calling the function, allocating memory, then returning to
the start of the function.

PR: 264021

2 years agolibsa: mark head_errlog feature supported.
Toomas Soome [Fri, 20 May 2022 10:45:33 +0000 (13:45 +0300)]
libsa: mark head_errlog feature supported.

head_errlog is new format for errlog, but we do not really
use errlog, so we can just mark it supported, to enable reading
from pool.

MFC after: 1 week

2 years agoping: add missing test files to Makefile
Mike Karels [Sat, 21 May 2022 03:13:02 +0000 (22:13 -0500)]
ping: add missing test files to Makefile

Should fix ATF tests after 5af718a5927a.

2 years agoping: if -S srcaddr uses a numeric address, use that protocol
Mike Karels [Fri, 20 May 2022 14:16:01 +0000 (09:16 -0500)]
ping: if -S srcaddr uses a numeric address, use that protocol

The command "ping -S dotted.quad hostname" fails on dual-stack hosts
with the confusing message "ping: invalid source address: Name does
not resolve" because IPv6 is selected in preference.  If the argument
to -S is numeric (likely), select the corresponding address family,
as if -4 or -6 was specified.  Add tests that either IPv4 or IPv6 can
be forced via a -S parameter.

Reviewed by: asomers
Differential Revision: https://reviews.freebsd.org/D35271
MFC after: 1 week

2 years agorpc.tlsclntd: Add an option to force use of TLS version 1.2
Rick Macklem [Fri, 20 May 2022 21:57:42 +0000 (14:57 -0700)]
rpc.tlsclntd: Add an option to force use of TLS version 1.2

Commit 0b4f2ab0e913 fixes the krpc so that it can use TLS
version 1.3 for NFS-over-TLS, as required by
the draft (someday to be an RFC).
Since FreeBSD 13.0, 13.1 use TLS version 1.2 for
NFS-over-TLS mounts, this command line option
may be used so that NFS-over-TLS mounts to 13.0, 13.1
servers will still work.

Without the command line option, NFS-over-TLS mounts
will use TLS version 1.3.

The man page update will be a separate commit.

MFC after:  2 weeks

2 years agorpc.tlsservd: Add an option to allow TLS version 1.2
Rick Macklem [Fri, 20 May 2022 21:44:50 +0000 (14:44 -0700)]
rpc.tlsservd: Add an option to allow TLS version 1.2

Commit 0b4f2ab0e913 fixes the krpc so that it can use TLS
version 1.3 for NFS-over-TLS, as required by
the draft (someday to be an RFC).
Since FreeBSD 13.0, 13.1 use TLS version 1.2 for
NFS-over-TLS mounts, this command line option
may be used so that mounts from 13.0, 13.1 will still work.

Without the command line option, only TLS version 1.3
mounts are permitted.

The man page update will be a separate commit.

MFC after:  2 weeks

2 years agovfs: fix copy-pasto in previous
Mateusz Guzik [Fri, 20 May 2022 20:58:11 +0000 (20:58 +0000)]
vfs: fix copy-pasto in previous

Reported by: dchagin

2 years agoLinuxKPI 802.11 / iwlwifi / rtw88 : re-factor ieee802211_sta for MLO
Bjoern A. Zeeb [Fri, 20 May 2022 18:37:27 +0000 (18:37 +0000)]
LinuxKPI 802.11 / iwlwifi / rtw88 : re-factor ieee802211_sta for MLO

Working on an update for rtw88 baed on wireless-testing I ran into
a build issue with struct ieee802211_sta.  Some fields were factored
out into their own struct apparently preparing for MLO (Multi Link
Operation).

In order to be able to update one driver we have to adjust both
rtw88 and iwlwifi (and the ones still out-of-tree) now.

This is mostly a sed-replace job and no functional changes are
intended.

Sponsored by: The FreeBSD Foundation
MFC after: 3 days

2 years agovfs: call vn_truncate_locked from kern_truncate
Mateusz Guzik [Mon, 16 May 2022 01:28:22 +0000 (03:28 +0200)]
vfs: call vn_truncate_locked from kern_truncate

This fixes a bug where the syscall would not bump writecount.

PR: 263999

2 years agovfs: make sure truncate always calls NDFREE_*
Mateusz Guzik [Mon, 16 May 2022 01:16:17 +0000 (03:16 +0200)]
vfs: make sure truncate always calls NDFREE_*

While here convert it to NDFREE_NOTHING.

2 years agoSwitch to GCC 9 for the GCC tinderbox.
John Baldwin [Fri, 20 May 2022 17:16:18 +0000 (10:16 -0700)]
Switch to GCC 9 for the GCC tinderbox.

This also adds arm and riscv to the GCC tinderbox.

Reviewed by: emaste (older version)
Differential Revision: https://reviews.freebsd.org/D26202

2 years agovmm: Destroy character devices synchronously.
John Baldwin [Fri, 20 May 2022 16:53:43 +0000 (09:53 -0700)]
vmm: Destroy character devices synchronously.

This fixes a userland race where bhyveload or bhyve can fail to reuse
a VM name after bhyvectl --destroy has returned.

Reported by: Michael Dexter
Reviewed by: markj
Differential Revision: https://reviews.freebsd.org/D35186

2 years agosys/dts: Remove MIPS files
Brooks Davis [Fri, 20 May 2022 16:11:27 +0000 (17:11 +0100)]
sys/dts: Remove MIPS files

Reviewed by: manu, imp
Sponsored by: DARPA, AFRL
Differential Revision: https://reviews.freebsd.org/D35263

2 years agobitstring_test: Add regression tests for bit_ff(c|s)_area_at()
Mark Johnston [Fri, 20 May 2022 14:18:10 +0000 (10:18 -0400)]
bitstring_test: Add regression tests for bit_ff(c|s)_area_at()

Validate the cases where a match can be found immediately and where no
match can be found.  This extends the existing test cases and is enough
to catch the bug fixed in commit 6e7a585348d5 ("bitstring: fix ff_area()
when start!=0").

Reviewed by: dougm
MFC after: 1 week
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D35259

2 years agolibsa: Make the nvlist implementation more self-contained
Mark Johnston [Fri, 20 May 2022 14:17:35 +0000 (10:17 -0400)]
libsa: Make the nvlist implementation more self-contained

Move declarations into a new nvlist.h rather than putting everything in
libzfs.h.  This makes this nvlist code easier to reuse elsewhere.  In
particular, the nvlist implementation in sys/contrib/libnv does not
provide XDR encoding, but this is needed when reading from or writing to
ZFS pools.

Also:
- Remove references to boolean_t.  It has to be a 32-bit int here, so
  just reference the underlying type.
- Add includes needed when compiling the nvlist code outside of stand/.

No functional change intended.

Reviewed by: tsoome
MFC after: 1 week
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D35255

2 years agolibsa: Fix a bug in nvlist creation
Mark Johnston [Fri, 20 May 2022 14:15:00 +0000 (10:15 -0400)]
libsa: Fix a bug in nvlist creation

When adding an entry to an nvlist, the data buffer might need to be
resized.  When this happens, the XDR encoder's notion of the buffer size
also needs to be updated, otherwise the operation may erroneously fail.

Reviewed by: tsoome, imp
MFC after: 1 week
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D35254

2 years agocpuset: Fix the KASAN and KMSAN builds
Mark Johnston [Fri, 20 May 2022 14:11:31 +0000 (10:11 -0400)]
cpuset: Fix the KASAN and KMSAN builds

Rename the "copyin" and "copyout" fields of struct cpuset_copy_cb to
something less generic, since sanitizers define interceptors for
copyin() and copyout() using #define.

Reported by: syzbot+2db5d644097fc698fb6f@syzkaller.appspotmail.com
Fixes: 47a57144af25 ("cpuset: Byte swap cpuset for compat32 on big endian architectures")
Sponsored by: The FreeBSD Foundation

2 years agopf tests: extend ethernet dummynet test
Kristof Provost [Wed, 18 May 2022 15:29:35 +0000 (17:29 +0200)]
pf tests: extend ethernet dummynet test

Extend the existing ethernet dummynet test to also test dummynet on the
outbound direction.
This used to be a problem as traffic shaping wasn't done in the ethernet
code. It merely tagged the packet and left shaping up to the layer 3 pf
code. This works in the inbound direction, but not for outbound traffic
where we hit the L3 code first and only then the L2 code.

Sponsored by: Rubicon Communications, LLC ("Netgate")
Differential Revision: https://reviews.freebsd.org/D35258

2 years agopf: call dummynet directly from the ethernet code
Kristof Provost [Wed, 18 May 2022 15:49:28 +0000 (17:49 +0200)]
pf: call dummynet directly from the ethernet code

Until recently dummynet in ethernet rules did not send packets directly
to dummynet but instead marked them and left the interactions with
dummynet to the layer 3 pf code.
This worked fine for incoming packets (where we process ethernet rules
before layer 3 rules), but not for outbound packets (where the order of
operations is the reverse).

Dummynet does support handling layer 2 traffic, so send the packets
directly to dummynet.

The main limitation now is that pf does not inspect layer 4 (i.e.
TCP/UDP) so we don't have protocol information or port numbers. Dummynet
potentially uses this to separate traffic flows, which will not work for
ethernet dummynet rules. However, pipes (i.e. adding latency or
restricting bandwidth) will work exactly as expected.

Sponsored by: Rubicon Communications, LLC ("Netgate")
Differential Revision: https://reviews.freebsd.org/D35257

2 years agodummynet: do not store struct ifnet pointers
Kristof Provost [Thu, 19 May 2022 13:12:54 +0000 (15:12 +0200)]
dummynet: do not store struct ifnet pointers

The dn_pkt_tag tag contained a struct ifnet pointer. If we persist that
across NET_EPOCH boundaries (as we did in dummynet) we risk panics if
the interface is removed between the packet being enqueued and it being
dequeued.

Convert the pointer into an index/generation pair and restore it when
the packet is taken out of the queue.

Sponsored by: Rubicon Communications, LLC ("Netgate")
Differential Revision: https://reviews.freebsd.org/D35256

2 years agoRetire sv_transtrap
Dmitry Chagin [Fri, 20 May 2022 11:54:03 +0000 (14:54 +0300)]
Retire sv_transtrap

Call translate_traps directly from sendsig().

MFC after: 2 weeks

2 years agolinux(4): Retire unneeded translate_traps from arm64
Dmitry Chagin [Fri, 20 May 2022 11:19:51 +0000 (14:19 +0300)]
linux(4): Retire unneeded translate_traps from arm64

MFC after: 2 weeks

2 years ago[vlan + lagg] add IFNET_EVENT_UPDATE_BAUDRATE event
Andrey V. Elsukov [Thu, 19 May 2022 06:02:27 +0000 (08:02 +0200)]
[vlan + lagg] add IFNET_EVENT_UPDATE_BAUDRATE event

use it to update if_baudrate for vlan interfaces created on the LACP lagg.

Differential revision: https://reviews.freebsd.org/D33405

2 years agoLinuxKPI: 802.11 updates
Bjoern A. Zeeb [Wed, 18 May 2022 22:39:57 +0000 (22:39 +0000)]
LinuxKPI: 802.11 updates

Add more values to the ieee80211_min_mpdu_start_spacing enum with
two missing given we do not know how they are called.  Also update
the reference while here.

Add struct ieee80211_hdr_3addr and correct the comment on the
ieee80211_hdr to match the one in net80211 to avoid future confusion.

Sponsored by: The FreeBSD Foundation
MFC after: 3 days

2 years agoLinuxKPI: implement pcie_capability_set_word()
Bjoern A. Zeeb [Wed, 18 May 2022 22:36:31 +0000 (22:36 +0000)]
LinuxKPI: implement pcie_capability_set_word()

Implement pcie_capability_set_word() using the already available
read/write functions.
Also define the completion timeout disable value to our PCI one.
Both needed by a driver update.

Sponsored by: The FreeBSD Foundation
MFC after: 3 days
Reviewed by: manu, hselasky
Differential Revision: https://reviews.freebsd.org/D35250

2 years agocxgbe/iw_cxgbe: c4iw_zero_addr should not use IN_ZERONET.
Navdeep Parhar [Thu, 19 May 2022 19:25:30 +0000 (12:25 -0700)]
cxgbe/iw_cxgbe: c4iw_zero_addr should not use IN_ZERONET.

c4iw_zero_addr is supposed to check for all-zero addresses but was using
IN_ZERONET (which does something different) for IPv4 addresses.  Fix it
by simply checking for 0 as is done for IPv6 addresses.

Reported by: karels@
MFC after: 3 days
Sponsored by: Chelsio Communications

2 years agosrc.libnames: Replace redundant += with =
Cy Schubert [Wed, 18 May 2022 01:59:13 +0000 (18:59 -0700)]
src.libnames: Replace redundant += with =

The += for unique assignments is equivalent to =. Make these confusing
assignments simply assignments.

Reviewed by: emaste
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D35242

2 years agolinux(4): To improve readability use FUTEX_UNOWNED instead of 0
Dmitry Chagin [Thu, 19 May 2022 18:42:38 +0000 (21:42 +0300)]
linux(4): To improve readability use FUTEX_UNOWNED instead of 0

MFC after: 2 weeks

2 years agoFix the build after 47a57144
Dmitry Chagin [Thu, 19 May 2022 18:40:59 +0000 (21:40 +0300)]
Fix the build after 47a57144

2 years agolinux(4): Convert the native kernel signal codes into the Linux codes
Dmitry Chagin [Thu, 19 May 2022 16:56:50 +0000 (19:56 +0300)]
linux(4): Convert the native kernel signal codes into the Linux codes

MFC after: 2 weeks

2 years agolinux(4): Add kernel signal code definitions
Dmitry Chagin [Thu, 19 May 2022 16:55:56 +0000 (19:55 +0300)]
linux(4): Add kernel signal code definitions

In the next commit I'll convert the native signal codes into the Linux codes,
since they are not 1:1 mapped.

MFC after: 2 weeks

2 years agolinux(4): Move signal codes definitions to the appropriate header
Dmitry Chagin [Thu, 19 May 2022 16:55:20 +0000 (19:55 +0300)]
linux(4): Move signal codes definitions to the appropriate header

In the Linux the struct siginfo related bits are placed into the siginfo.h
header.

MFC after: 2 weeks

2 years agoarm64: Enable the floating-point exception traps
Dmitry Chagin [Thu, 19 May 2022 16:53:56 +0000 (19:53 +0300)]
arm64: Enable the floating-point exception traps

To enable it user-space needs to call feenableexcept().

FPE_FLTIDO has been added as the IDF bit can't be mapped to any existing
FPE code.

Reviewed by: andrew@
Differential revision: https://reviews.freebsd.org/D35247
MFC after: 2 weeks

2 years agolinux(4): Handle cas failure on ll/sc operations
Dmitry Chagin [Thu, 19 May 2022 16:52:18 +0000 (19:52 +0300)]
linux(4): Handle cas failure on ll/sc operations

Follow the 11a6ecd4. Check and handle the case when the ll/sc casu fails
even when the compare succeeds.

For more details PR/263825, https://reviews.freebsd.org/D35150.

Obtained from: Andrew@
MFC after: 2 weeks

2 years agokqueue: Trim trailing whitespace
Dmitry Chagin [Thu, 19 May 2022 16:52:02 +0000 (19:52 +0300)]
kqueue: Trim trailing whitespace

MFC after: 1 week

2 years agocpuset: Byte swap cpuset for compat32 on big endian architectures
Justin Hibbits [Thu, 12 May 2022 20:38:22 +0000 (15:38 -0500)]
cpuset: Byte swap cpuset for compat32 on big endian architectures

Summary:
BITSET uses long as its basic underlying type, which is dependent on the
compile type, meaning on 32-bit builds the basic type is 32 bits, but on
64-bit builds it's 64 bits.  On little endian architectures this doesn't
matter, because the LSB is always at the low bit, so the words get
effectively concatenated moving between 32-bit and 64-bit, but on
big-endian architectures it throws a wrench in, as setting bit 0 in
32-bit mode is equivalent to setting bit 32 in 64-bit mode.  To
demonstrate:

32-bit mode:

BIT_SET(foo, 0):        0x00000001

64-bit sees: 0x0000000100000000

cpuset is the only system interface that uses bitsets, so solve this
by swapping the integer sub-components at the copyin/copyout points.

Reviewed by: kib
MFC after: 3 days
Sponsored by: Juniper Networks, Inc.
Differential Revision: https://reviews.freebsd.org/D35225

2 years agomakefs: Remove execute permission from in-tree test scripts
Mark Johnston [Wed, 23 Mar 2022 17:32:17 +0000 (13:32 -0400)]
makefs: Remove execute permission from in-tree test scripts

Let the build system set it as needed.  No functional change intended.

MFC after: 1 week
Sponsored by: The FreeBSD Foundation

2 years agojexec man: Explain how PATH is adjusted.
Gleb Popov [Thu, 19 May 2022 05:59:41 +0000 (08:59 +0300)]
jexec man: Explain how PATH is adjusted.

Add a sentence that explains that PATH is set to `/bin:/usr/bin`.
I had to dive into `jexec.c` to find this out.

Reviewed by: markj
Differential Revision: https://reviews.freebsd.org/D35251

2 years agoUse getpagesize in gcore to find the page size
Andrew Turner [Fri, 13 May 2022 11:56:14 +0000 (12:56 +0100)]
Use getpagesize in gcore to find the page size

Reviewed by: markj
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D35194

2 years agoSupport LSE atomics in the arm64 casue* functions
Andrew Turner [Tue, 17 May 2022 13:52:30 +0000 (14:52 +0100)]
Support LSE atomics in the arm64 casue* functions

As with atomic(9) use the ARMv8.1 Large System Extension atomic
instructions to implement the userspace compare and swap functions.

Reviewed by: kib
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D35234

2 years agoHandle cas failure when the compare succeeds
Andrew Turner [Mon, 9 May 2022 14:28:56 +0000 (15:28 +0100)]
Handle cas failure when the compare succeeds

When locking a priority inherit mutex we perform a compare and swap
operation to try and acquire the mutex. This may fail even when the
compare succeeds.

Check and handle this case.

PR: 263825
Reviewed by: kib, markj
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D35150

2 years agoFix the name of Makefile.aarch64
Andrew Turner [Thu, 12 May 2022 12:50:59 +0000 (13:50 +0100)]
Fix the name of Makefile.aarch64

Makefile.arm64 relies on being included via Makefile.${MACHINE}. This
only works when arm64 is the only aarch64 architecture. Switch to
Makefile.aarch64 so downstream CheriBSD can use the file when building
for Morello which uses a differen ${MACHINE} value.

Sponsored by: Innovate UK

2 years agomlx4: Fix a memory leak bug.
Hans Petter Selasky [Thu, 19 May 2022 08:10:48 +0000 (10:10 +0200)]
mlx4: Fix a memory leak bug.

In function mlx4_opreq_action(), pointer "mailbox" is not released,
when mlx4_cmd_box() return and error, causing a memory leak bug.
Fix this issue by going to "out" label, mlx4_free_cmd_mailbox() can
free this pointer.

Linux commit:
febfd9d3c7f74063e8e630b15413ca91b567f963

PR: 264056
MFC after: 1 week
Sponsored by: NVIDIA Networking

2 years agoibcore: Fix possible memory leak in ib_mad_post_receive_mads()
Hans Petter Selasky [Thu, 19 May 2022 08:06:42 +0000 (10:06 +0200)]
ibcore: Fix possible memory leak in ib_mad_post_receive_mads()

If ib_dma_mapping_error() returns non-zero value,
ib_mad_post_receive_mads() will jump out of loops and return -ENOMEM
without freeing mad_priv. Fix this memory-leak problem by freeing
mad_priv in this case.

Linux commit:
a17f4bed811c60712d8131883cdba11a105d0161

PR: 264057
MFC after: 1 week
Sponsored by: NVIDIA Networking

2 years agozfs: merge openzfs/zfs@c0cf6ed67
Martin Matuska [Wed, 18 May 2022 22:54:40 +0000 (00:54 +0200)]
zfs: merge openzfs/zfs@c0cf6ed67

Notable upstream pull request merges:
  #10662 zvol_wait: Ignore locked zvols
  #12789 Improve log spacemap load time
  #12812 Improved zpool status output, list all affected datasets
  #13277 FreeBSD: Use NDFREE_PNBUF if available
  #13302 Make zfs_max_recordsize default to 16M
  #13311 Fix error handling in FreeBSD's get/putpages VOPs
  #13345 FreeBSD: Fix translation from ABD to physical pages
  #13373 zfs: holds: dequadratify
  #13375 Corrected edge case in uncompressed ARC->L2ARC handling
  #13388 Improve mg_aliquot math
  #13405 Reduce dbuf_find() lock contention
  #13406 FreeBSD: use zero_region instead of allocating a dedicated page

Obtained from: OpenZFS
OpenZFS commit: c0cf6ed6792e545fd614c2a88cb53756db7e03f8

2 years agoRemove accidentally committed diff file
Dimitry Andric [Wed, 18 May 2022 19:45:40 +0000 (21:45 +0200)]
Remove accidentally committed diff file

I accidentally added the upstream diff file from the llvm commit I
merged in 1a0329799c6d. Clean it out.

Noticed by: markj
Fixes: 1a0329799c6d
MFC after: 2 weeks

2 years agomakefs: sort cd9660 sysid entries in man page
Ed Maste [Wed, 18 May 2022 19:38:54 +0000 (15:38 -0400)]
makefs: sort cd9660 sysid entries in man page

"efi" was added at the end of the sysid list rather than in alpha
order.  Move it as a diff reduction against NetBSD.

2 years agoApply llvm fix for "Invalid PPC CTR loop!" error on powerpcspe
Dimitry Andric [Wed, 18 May 2022 09:02:03 +0000 (11:02 +0200)]
Apply llvm fix for "Invalid PPC CTR loop!" error on powerpcspe

Merge commit d9d15af7873f from llvm git (Qiu Chaofan):

  [PowerPC] Treat llvm.fmuladd intrinsic as using CTR

  This fixes bug 55463, similar to D78668. This is a temporary fix since
  we will switch to post-isel CTR loop determination in the future.

  Reviewed By: dim, shchenz

  Differential Revision: https://reviews.llvm.org/D125746

MFC after: 2 weeks

2 years agoloader.efi: add cross-reference to loader(8)
Ed Maste [Wed, 18 May 2022 14:33:20 +0000 (10:33 -0400)]
loader.efi: add cross-reference to loader(8)

The loader.efi man page describes UEFI-specific loader details, but not
general loader information.

MFC after: 3 days
Sponsored by: The FreeBSD Foundation

2 years agoAdd OFW support to arm64's IOMMU framework.
Ruslan Bukin [Wed, 18 May 2022 13:11:23 +0000 (14:11 +0100)]
Add OFW support to arm64's IOMMU framework.
This is needed to support non-PCI devices like memory-mapped
display controllers.
Split-out some initialization code from iommu_ctx_alloc() into
iommu_ctx_init() method so we could pass controller's MD-data
obtained from DTS to the driver prior to a CTX initialization.

Tested on Morello SoC.

Sponsored by: UKRI

2 years agoFix stream table entry (STE) initialization and removal.
Ruslan Bukin [Wed, 18 May 2022 12:42:37 +0000 (13:42 +0100)]
Fix stream table entry (STE) initialization and removal.
For PCI devices we have entire L1 descriptor for every session ID (SID),
but for non-PCI (e.g. Display Processing Unit DPU), a single L1
descriptor serves multiple SIDs.
So prevent re-initialization of L1 descriptor if already initialized.
Don't free entire L1 descriptor on every STE removal.

Sponsored by: UKRI

2 years agohidraw: Return string lengths for certain ioctls
Michael Gmelin [Tue, 17 May 2022 09:37:21 +0000 (11:37 +0200)]
hidraw: Return string lengths for certain ioctls

Make HIDIOCGRAWPHYS, HIDIOCGRAWNAME, and HIDIOCGRAWUNIQ return
the length of the copied out string (including the trailing NUL
character), so they behave like their Linux hidraw counterparts.

Reviewed by: wulf
Differential Revision: https://reviews.freebsd.org/D35233

2 years agobsd-family-tree: FreeBSD 13.1
Warner Losh [Wed, 18 May 2022 00:02:42 +0000 (18:02 -0600)]
bsd-family-tree: FreeBSD 13.1

Now that 13.1 has been announced, add today's date UTC as the release
date.

MFC After: 3 days

2 years agocxgbe: Deactivate upper layer drivers (like TOE) during detach.
John Baldwin [Tue, 17 May 2022 23:33:49 +0000 (16:33 -0700)]
cxgbe: Deactivate upper layer drivers (like TOE) during detach.

Reviewed by: np
Sponsored by: Chelsio Communications
Differential Revision: https://reviews.freebsd.org/D35237

2 years agonet80211: remove (optional) RSN from preobereq
Bjoern A. Zeeb [Tue, 12 Apr 2022 22:01:27 +0000 (22:01 +0000)]
net80211: remove (optional) RSN from preobereq

802.11i-2004 did not specify that RSN can be added to the probereq,
only to the proberesp (and others) yet it was added in what seems was
b032f27c365b992e9d8e42214183b39acfb8c6ac (multi-BSS support).
802.11-2020 does not list the IE either for probereq so remove it.

Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
Okayed by: adrian
Differential Revision: https://reviews.freebsd.org/D34893

2 years agoPrevent running sigwait tests in parallel
Olivier Cochard [Tue, 17 May 2022 21:28:56 +0000 (23:28 +0200)]
Prevent running sigwait tests in parallel

test_sig_discard_ign_* could not run at the same time.

Approved by: dchagin
Sponsored by: Netflix
Differential Revision: https://reviews.freebsd.org/D35236

2 years agoFix compiler warnings about zero-length arrays in inline bitops
Coleman Kane [Tue, 17 May 2022 20:07:39 +0000 (16:07 -0400)]
Fix compiler warnings about zero-length arrays in inline bitops

The compiler appears to be expanding the unused NULL pointer into a
zero-length array via the inline bitops code. When -Werror=array-bounds
is used, this causes a build failure. Recommended solution is allocate
temporary structures, fill with zeros (to avoid uninitialized data use
warnings), and pass the pointer to those to the inline calls.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Coleman Kane <ckane@colemankane.org>
Closes #13463
Closes #13465

2 years agopowerpc: enable supported sanitizers on powerpc64*
Piotr Kubaj [Tue, 17 May 2022 06:39:29 +0000 (08:39 +0200)]
powerpc: enable supported sanitizers on powerpc64*

1. Merge LLVM's 315d792130258a9b7250494be8d002ebb427b08f, adding support
for FreeBSD/powerpc64*.
2. Add sanitizer list to lib/libclang_rt/Makefile, taken from the list of
libraries that llvm-devel port builds.
3. powerpc64le supports the same sanitizers that powerpc64, but powerpc64le
also supports xray* sanitizers.
4. lib/libclang_rt/xray/Makefile hardcodes amd64-specific files, so that needs
to be conditionalized.
5. Sanitizers are not enabled for powerpc, because powerpc supports only
builtins and profile.

Reviewed by: dim
Differential Revision: https://reviews.freebsd.org/D35228
Relnotes: yes
MFC after: 3 days

2 years agotruss: add ppoll(2) argument decoding
Christian Weisgerber [Tue, 17 May 2022 18:20:53 +0000 (20:20 +0200)]
truss: add ppoll(2) argument decoding

PR: 264029
Approved by: emaste
MFC after: 3 days

2 years agosockbuf: improve sbcreatecontrol()
Gleb Smirnoff [Tue, 17 May 2022 17:10:42 +0000 (10:10 -0700)]
sockbuf: improve sbcreatecontrol()

o Constify memory pointer.  Make length unsigned.
o Make it never fail with M_WAITOK and assert that length is sane.

2 years agosockbuf: merge two versions of sbcreatecontrol() into one
Gleb Smirnoff [Tue, 17 May 2022 17:10:42 +0000 (10:10 -0700)]
sockbuf: merge two versions of sbcreatecontrol() into one

No functional change.

2 years agounix: garbage collect unp_dispose_mbuf() for brevity
Gleb Smirnoff [Tue, 17 May 2022 17:10:41 +0000 (10:10 -0700)]
unix: garbage collect unp_dispose_mbuf() for brevity

2 years agounix: fix mbuf leak on close of socket with data
Gleb Smirnoff [Tue, 17 May 2022 17:10:41 +0000 (10:10 -0700)]
unix: fix mbuf leak on close of socket with data

Fixes: 1f32cef47189403e9e70b1893c731c68b97b964e

2 years agopf: allocate krule->timestamp in pf_krule_alloc()
Kristof Provost [Tue, 17 May 2022 12:41:25 +0000 (14:41 +0200)]
pf: allocate krule->timestamp in pf_krule_alloc()

There are three calls which can allocate a new rule. DIOCADDRULE,
DIOCADDRULENV and DIOCCHANGERULE. The first two call pf_ioctl_addrule(),
but DIOCCHANGERULE does not. As a result rules created through
DIOCCHANGERULE do not have the timestamp per-cpu memory allocated, and
we panic when the rule is exported with pf_krule_to_nvrule().

Sponsored by: Rubicon Communications, LLC ("Netgate")

2 years agoLinuxKPI: Switch irq_work implementation back to standard taskqueue
Vladimir Kondratyev [Tue, 17 May 2022 12:10:20 +0000 (15:10 +0300)]
LinuxKPI: Switch irq_work implementation back to standard taskqueue

from fast one as dmabuf does not run callbacks from critical sections
since drm-kmod 5.7.
Consumers which requires fair irq_work like drm-kmod 5.4 must #define
LKPI_IRQ_WORK_USE_FAST_TQ somewhere in source file before linux/irq_work.h
inclusion to enable old behavior.

MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D35052

2 years agoLinuxKPI: Do not use forward declaration for struct llist_node
Vladimir Kondratyev [Tue, 17 May 2022 12:10:20 +0000 (15:10 +0300)]
LinuxKPI: Do not use forward declaration for struct llist_node

in linux/irq_work.h as irq_work structure contains full version of
llist_node rather than reference to it.

MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D35052

2 years agoLinuxKPI: IRQ work add TASKQUEUE_FAIL_IF_PENDING flag to taskqueue_enqueue
Vladimir Kondratyev [Tue, 17 May 2022 12:10:20 +0000 (15:10 +0300)]
LinuxKPI: IRQ work add TASKQUEUE_FAIL_IF_PENDING flag to taskqueue_enqueue

call to match Linux.

MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D35052

2 years agoLinuxKPI: Implement kthread_worker related functions
Vladimir Kondratyev [Tue, 17 May 2022 12:10:20 +0000 (15:10 +0300)]
LinuxKPI: Implement kthread_worker related functions

Kthread worker is a single thread workqueue which can be used in cases
where specific kthread association is necessary, for example, when it
should have RT priority or be assigned to certain cgroup.

This change implements Linux v4.9 interface which mostly hides kthread
internals from users thus allowing to use ordinary taskqueue(9) KPI.
As kthread worker prohibits enqueueing of already pending or canceling
tasks some minimal changes to taskqueue(9) were done.
taskqueue_enqueue_flags() was added to taskqueue KPI which accepts extra
flags parameter. It contains one or more of the following flags:

TASKQUEUE_FAIL_IF_PENDING - taskqueue_enqueue_flags() fails if the task
    is already scheduled to execution. EEXIST is returned and the
    ta_pending counter value remains unchanged.
TASKQUEUE_FAIL_IF_CANCELING - taskqueue_enqueue_flags() fails if the
    task is in the canceling state and ECANCELED is returned.

Required by: drm-kmod 5.10

MFC after: 1 week
Reviewed by: hselasky, Pau Amma (docs)
Differential Revision: https://reviews.freebsd.org/D35051

2 years agoLinuxKPI: Implement sched_set_fifo(_low) functions
Vladimir Kondratyev [Tue, 17 May 2022 12:10:20 +0000 (15:10 +0300)]
LinuxKPI: Implement sched_set_fifo(_low) functions

Required by: drm-kmod

MFC after: 1 week
Reviewed by: hselasky, manu
Differential Revision: https://reviews.freebsd.org/D35050

2 years agoLinuxKPI: Add some pollution required by drm-kmod to linux/sched.h
Vladimir Kondratyev [Tue, 17 May 2022 12:10:20 +0000 (15:10 +0300)]
LinuxKPI: Add some pollution required by drm-kmod to linux/sched.h

MFC after: 1 week
Reviewed by: hselasky, manu
Differential Revision: https://reviews.freebsd.org/D35050

2 years agoLinuxKPI: Add cpu_latency_qos_request_active stub to linux/pm_qos.h
Vladimir Kondratyev [Tue, 17 May 2022 12:10:20 +0000 (15:10 +0300)]
LinuxKPI: Add cpu_latency_qos_request_active stub to linux/pm_qos.h

Required by: drm-kmod 5.10

MFC after: 1 week
Reviewers: hselasky, manu
Differential Revision: https://reviews.freebsd.org/D35049

2 years agoLinuxKPI: Fix typo in cond_resched_lock
Vladimir Kondratyev [Tue, 17 May 2022 12:10:20 +0000 (15:10 +0300)]
LinuxKPI: Fix typo in cond_resched_lock

Lock must be released rather than acquired around mi_switch call.

MFC after: 1 week
Reviewed by: hselasky, manu
Differential Revision: https://reviews.freebsd.org/D35048

2 years agoLinuxKPI: Convert lkpi-shrinker lock to sx.
Vladimir Kondratyev [Tue, 17 May 2022 12:10:20 +0000 (15:10 +0300)]
LinuxKPI: Convert lkpi-shrinker lock to sx.

This fixes "might_sleep() with the following non-sleepable locks held:
exclusive sleep mutex lkpi-shrinker" warnings.

MFC after: 1 week
Reviewed by: hselasky, manu
Differential Revision: https://reviews.freebsd.org/D35047

2 years agoLinuxKPI: Allow lkpi_iic driver to be a child of drm device.
Vladimir Kondratyev [Tue, 17 May 2022 12:10:19 +0000 (15:10 +0300)]
LinuxKPI: Allow lkpi_iic driver to be a child of drm device.

i915kms exposes "AUX #/port #" I2C adapters as drm children

MFC after: 1 week
Reviewed by: hselasky, manu
Differential Revision: https://reviews.freebsd.org/D35046

2 years agoigc: Increase rx_buffer_size local variable to 32b
Kevin Bowling [Tue, 17 May 2022 03:15:34 +0000 (20:15 -0700)]
igc: Increase rx_buffer_size local variable to 32b

Apply 6987c47569b377f4b6eba9966afdedfb1b39fca8 to igc. This is not
expected to have any benefit on current parts with current observed PBA
sizes but will avoid surprises if they are increased in future chips.

Approved by: grehan
MFC after: 3 days
Differential Revision: https://reviews.freebsd.org/D35217

2 years agotests/unix_dgram: account for size of sender address in the filling cycle
Gleb Smirnoff [Tue, 17 May 2022 02:04:02 +0000 (19:04 -0700)]
tests/unix_dgram: account for size of sender address in the filling cycle

This fixes test failure with large net.local.dgram.recvspace values.

2 years agolinux: libzutil: zfs_strip_path: only strip known prefixes
наб [Tue, 3 May 2022 18:13:22 +0000 (20:13 +0200)]
linux: libzutil: zfs_strip_path: only strip known prefixes

This mirrors FreeBSD:
  # zpool create -o cachefile= testpsko media/testpsko
  # zpool create -o cachefile= testpsko2 $PWD/testpsko2
  $ ./zpool list -v
  NAME                                              SIZE  ALLOC   FREE
  filling                                          25.5T  6.85T  18.6T
    mirror-0                                       3.64T   500G  3.15T
      ata-HGST_HUS726T4TALE6L4_V6K2L4RR                -      -      -
      ata-HGST_HUS726T4TALE6L4_V6K2MHYR                -      -      -
    raidz1-1                                       21.8T  6.36T  15.5T
      ata-HGST_HUS728T8TALE6L4_VDKT237K                -      -      -
      ata-HGST_HUS728T8TALE6L4_VDGY075D                -      -      -
      ata-HGST_HUS728T8TALE6L4_VDKVRRJK                -      -      -
  cache                                                -      -      -
    nvme0n1p4                                      63.0G  12.8G  50.2G
  tarta-boot                                        240M  50.0M   190M
    mirror-0                                        240M  50.0M   190M
      tarta-boot                                       -      -      -
      tarta-boot-nvme                                  -      -      -
  tarta-zoot                                       55.5G  6.96G  48.5G
    mirror-0                                       55.5G  6.96G  48.5G
      tarta-zoot                                       -      -      -
      tarta-zoot-nvme                                  -      -      -
  testpsko                                         39.5G   744K  39.5G
    media/testpsko1                                39.5G   744K  39.5G
  testpsko2                                        39.5G   130K  39.5G
    /home/nabijaczleweli/store/code/zfs/testpsko2  39.5G   130K  39.5G

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes #13413
Closes #9771

2 years agolibzfs: constify zfs_strip_partition(), zfs_strip_path()
наб [Tue, 3 May 2022 17:56:12 +0000 (19:56 +0200)]
libzfs: constify zfs_strip_partition(), zfs_strip_path()

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes #13413

2 years agolibzfs: pool: zpool_vdev_name: use libzfs_envvar_is_set
наб [Tue, 3 May 2022 17:38:15 +0000 (19:38 +0200)]
libzfs: pool: zpool_vdev_name: use libzfs_envvar_is_set

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes #13413

2 years agozpool: max_width: monomorphise subtype iteration
наб [Tue, 3 May 2022 17:33:31 +0000 (19:33 +0200)]
zpool: max_width: monomorphise subtype iteration

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes #13413

2 years agomakefs: fix calculation of file sizes
Alan Somers [Mon, 16 May 2022 22:32:10 +0000 (16:32 -0600)]
makefs: fix calculation of file sizes

When a new FS image is created we need to calculate how much space each
file is going to consume.
Fix two bugs in that logic:

1) Count the space needed for indirect blocks for large files.
1) Normally the trailing data of a file is written to a block of frag
   size, 4 kB by default.

However for files that use indirect blocks a full block is allocated,
32kB by default.  Take that into account.

Adjust size calculations to match what is done in ffs_mkfs routine:

* Depending on the UFS version the superblock is stored at a different
  offset. Take that into account.
* Add the cylinder group block size.
* All of the above has to be aligned to the block size.

Finally, Remove "ncg" variable. It's always 1 and it was used to
multiply stuff.

PR: 229929
Reviewed by: mckusick
MFC after: 2 weeks
Sponsored by: Semihalf
Submitted by: Kornel Dulęba <mindal@semihalf.com>
Differential Revision: https://reviews.freebsd.org/D35131
Differential Revision: https://reviews.freebsd.org/D35132

2 years agoriscv: also enable includes, stats and stats_client in libclang_rt
Piotr Kubaj [Mon, 18 Apr 2022 18:26:40 +0000 (20:26 +0200)]
riscv: also enable includes, stats and stats_client in libclang_rt

Reviewed by: dim (earlier version)
Differential Revision: https://reviews.freebsd.org/D34735
MFC after: 3 days

2 years agolinux(4): Fix arm64 syscalls.master layout
Dmitry Chagin [Mon, 16 May 2022 19:53:38 +0000 (22:53 +0300)]
linux(4): Fix arm64 syscalls.master layout

MFC after: 2 weeks

2 years agolibsysdecode: Fix decoding of SCHED_ flags
Dmitry Chagin [Mon, 16 May 2022 18:57:48 +0000 (21:57 +0300)]
libsysdecode: Fix decoding of SCHED_ flags

Use sys/sched.h where SCHED_ flags are defined.

Reviewed by: emaste
Differential revision: https://reviews.freebsd.org/D35208
MFC after: 2 weeks

2 years agoiwlwifi: import two fixes from linux v5.18-rc7
Bjoern A. Zeeb [Mon, 16 May 2022 18:07:36 +0000 (18:07 +0000)]
iwlwifi: import two fixes from linux v5.18-rc7

Import two changes from torvalds/linux.git at tag: v5.18-rc7:
b20bdd9cc9740ac1f2138adab25ddd51245c67be
    iwlwifi: mvm: Don't fail if PPAG isn't supported
7635a1ad8d92dcc8247b53f949e37795154b5b6f
    iwlwifi: iwl-dbg: Use del_timer_sync() before freeing

Sponsored by: The FreeBSD Foundation
MFC after: 3 days