drm/amdkfd: Fixed kfd_process cleanup on module exit.

Handle case when module is unloaded (kfd_exit) before a process space
(mm_struct) is released.

v2: Fixed potential race conditions by removing all kfd_process from
the process table first, then working on releasing the resources.

v3: Fixed loop element access / synchronization.  Fixed extra empty lines.

Signed-off-by: David Belanger <david.belanger@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: disconnect MPCC only on OTG change

[Why]
Framedrops are observed while playing Vp9 and Av1 10 bit
video on 8k resolution using VSR while playback controls
are disappeared/appeared

[How]
Now ODM 2 to 1 is disabled for 5k or greater resolutions on VSR.

Cc: stable@vger.kernel.org
Cc: Mario Limonciello <mario.limonciello@amd.com>
Reviewed-by: Alvin Lee <Alvin.Lee2@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Ayush Gupta <ayugupta@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix DP MST sinks removal issue

[Why]
In USB4 DP tunneling, it's possible to have this scenario that
the path becomes unavailable and CM tears down the path a little bit late.
So, in this case, the HPD is high but fails to read any DPCD register.
That causes the link connection type to be set to sst.
And not all sinks are removed behind the MST branch.

[How]
Restore the link connection type if it fails to read DPCD register.

Cc: stable@vger.kernel.org
Cc: Mario Limonciello <mario.limonciello@amd.com>
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Cruise Hung <Cruise.Hung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Do not set DRR on pipe Commit

[WHY]
Writing to DRR registers such as OTG_V_TOTAL_MIN on the same frame as a
pipe commit can cause underflow.

Cc: stable@vger.kernel.org
Cc: Mario Limonciello <mario.limonciello@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Wesley Chalmers <Wesley.Chalmers@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Remove OTG DIV register write for Virtual signals.

[WHY]
Hot plugging and then hot unplugging leads to k1 and k2 values to
change, as signal is detected as a virtual signal on hot unplug. Writing
these values to OTG_PIXEL_RATE_DIV register might cause primary display
to blank (known hw bug).

[HOW]
No longer write k1 and k2 values to register if signal is virtual, we
have safe guards in place in the case that k1 and k2 is unassigned so
that an unknown value is not written to the register either.

Cc: stable@vger.kernel.org
Cc: Mario Limonciello <mario.limonciello@amd.com>
Reviewed-by: Samson Tam <Samson.Tam@amd.com>
Reviewed-by: Alvin Lee <Alvin.Lee2@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Saaem Rizvi <SyedSaaem.Rizvi@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Merge tag 'linux-kselftest-fixes-6.3-rc3' of git://git./linux/kernel/git/shuah/linux-kselftest

Pull kselftest fixes from Shuah Khan:
 "A fix to amd-pstate test Makefile and a fix to LLVM build for x86 in
  kselftest common lib.mk"

* tag 'linux-kselftest-fixes-6.3-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest:
  selftests: fix LLVM build for i386 and x86_64
  selftests: amd-pstate: fix TEST_FILES

Merge branch 'md-fixes' of https://git./linux/kernel/git/song/md into block-6.3

Pull MD fixes from Song:

"This set contains two fixes for old issues (by Neil) and one fix
 for 6.3 (by Xiao)."

* 'md-fixes' of https://git.kernel.org/pub/scm/linux/kernel/git/song/md:
  md: select BLOCK_LEGACY_AUTOLOAD
  md: avoid signed overflow in slot_store()
  md: Free resources in __md_stop

md: select BLOCK_LEGACY_AUTOLOAD

When BLOCK_LEGACY_AUTOLOAD is not enable, mdadm is not able to
activate new arrays unless "CREATE names=yes" appears in
mdadm.conf

As this is a regression we need to always enable BLOCK_LEGACY_AUTOLOAD
for when MD is selected - at least until mdadm is updated and the
updates widely available.

Cc: stable@vger.kernel.org # v5.18+
Fixes: fbdee71bb5d8 ("block: deprecate autoloading based on dev_t")
Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Song Liu <song@kernel.org>

block: count 'ios' and 'sectors' when io is done for bio-based device

While using iostat for raid, I observed very strange 'await'
occasionally, and turns out it's due to that 'ios' and 'sectors' is
counted in bdev_start_io_acct(), while 'nsecs' is counted in
bdev_end_io_acct(). I'm not sure why they are ccounted like that
but I think this behaviour is obviously wrong because user will get
wrong disk stats.

Fix the problem by counting 'ios' and 'sectors' when io is done, like
what rq-based device does.

Fixes: 394ffa503bc4 ("blk: introduce generic io stat accounting help function")
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20230223091226.1135678-1-yukuai1@huaweicloud.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: sunvdc: add check for mdesc_grab() returning NULL

In vdc_port_probe(), we should check the return value of mdesc_grab() as
it may return NULL, which can cause potential NPD bug.

Fixes: 43fdf27470b2 ("[SPARC64]: Abstract out mdesc accesses for better MD update handling.")
Signed-off-by: Liang He <windhl@126.com>
Link: https://lore.kernel.org/r/20230315062032.1741692-1-windhl@126.com
[axboe: style cleanup]
Signed-off-by: Jens Axboe <axboe@kernel.dk>

nvmet: avoid potential UAF in nvmet_req_complete()

An nvme target ->queue_response() operation implementation may free the
request passed as argument. Such implementation potentially could result
in a use after free of the request pointer when percpu_ref_put() is
called in nvmet_req_complete().

Avoid such problem by using a local variable to save the sq pointer
before calling __nvmet_req_complete(), thus avoiding dereferencing the
req pointer after that function call.

Fixes: a07b4970f464 ("nvmet: add a generic NVMe target")
Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvme-trace: show more opcode names

We have more commands to show in the trace. Sync up.

Signed-off-by: Minwoo Im <minwoo.im.dev@gmail.com>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvme-tcp: add nvme-tcp pdu size build protection

Make sure that we don't somehow mess up the wire structures in the spec.

Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kkch@nvidia.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvme-tcp: fix opcode reporting in the timeout handler

For non in-capsule writes we reuse the request pdu space for a h2cdata
pdu in order to avoid over allocating space (either preallocate or
dynamically upon receving an r2t pdu). However if the request times out
the core expects to find the opcode in the start of the request, which
we override.

In order to prevent that, without sacrificing additional 24 bytes per
request, we just use the tail of the command pdu space instead (last
24 bytes from the 72 bytes command pdu). That should make the command
opcode always available, and we get away from allocating more space.

If in the future we would need the last 24 bytes of the nvme command
available we would need to allocate a dedicated space for it in the
request, but until then we can avoid doing so.

Reported-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kkch@nvidia.com>
Tested-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvme-pci: add NVME_QUIRK_BOGUS_NID for Lexar NM620

Added a quirk to fix Lexar NM620 1TB SSD reporting duplicate NGUIDs.

Signed-off-by: Philipp Geulen <p.geulen@js-elektronik.de>
Reviewed-by: Chaitanya Kulkarni <kkch@nvidia.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvme-pci: add NVME_QUIRK_BOGUS_NID for Netac NV3000

Added a quirk to fix the Netac NV3000 SSD reporting duplicate NGUIDs.

Cc: <stable@vger.kernel.org>
Signed-off-by: Elmer Miroslav Mosher Golovin <miroslav@mishamosher.com>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvme-pci: fixing memory leak in probe teardown path

In case the nvme_probe teardown path is triggered the ctrl ref count does
not reach 0 thus creating a memory leak upon failure of nvme_probe.

Signed-off-by: Irvin Cote <irvincoteg@gmail.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

nvme: fix handling single range discard request

When investigating one customer report on warning in nvme_setup_discard,
we observed the controller(nvme/tcp) actually exposes
queue_max_discard_segments(req->q) == 1.

Obviously the current code can't handle this situation, since contiguity
merge like normal RW request is taken.

Fix the issue by building range from request sector/nr_sectors directly.

Fixes: b35ba01ea697 ("nvme: support ranged discard requests")
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

MAINTAINERS: repair malformed T: entries in NVM EXPRESS DRIVERS

The T: entries shall be composed of a SCM tree type (git, hg, quilt, stgit
or topgit) and location.

Add the SCM tree type to the T: entry, and reorder the file entries in
alphabetical order.

Fixes: b508fc354f6d ("nvme: update maintainers information")
Signed-off-by: Lukas Bulwahn <lukas.bulwahn@gmail.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>

io_uring/sqpoll: Do not set PF_NO_SETAFFINITY on sqpoll threads

Users may specify a CPU where the sqpoll thread would run. This may
conflict with cpuset operations because of strict PF_NO_SETAFFINITY
requirement. That flag is unnecessary for polling "kernel" threads, see
the reasoning in commit 01e68ce08a30 ("io_uring/io-wq: stop setting
PF_NO_SETAFFINITY on io-wq workers"). Drop the flag on poll threads too.

Fixes: 01e68ce08a30 ("io_uring/io-wq: stop setting PF_NO_SETAFFINITY on io-wq workers")
Link: https://lore.kernel.org/all/20230314162559.pnyxdllzgw7jozgx@blackpad/
Signed-off-by: Michal Koutný <mkoutny@suse.com>
Link: https://lore.kernel.org/r/20230314183332.25834-1-mkoutny@suse.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: null_blk: cleanup null_queue_rq()

Use a local struct request pointer variable to avoid having to
dereference struct blk_mq_queue_data multiple times. While at it, also
fix the function argument indentation and remove a useless "else" after
a return.

Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Pankaj Raghav <p.raghav@samsung.com>
Link: https://lore.kernel.org/r/20230314041106.19173-2-damien.lemoal@opensource.wdc.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block: null_blk: Fix handling of fake timeout request

When injecting a fake timeout into the null_blk driver using
fail_io_timeout, the request timeout handler does not execute
blk_mq_complete_request(), so the complete callback is never executed
for a timedout request.

The null_blk driver also has a driver-specific fake timeout mechanism
which does not have this problem. Fix the problem with fail_io_timeout
by using the same meachanism as null_blk internal timeout feature, using
the fake_timeout field of null_blk commands.

Reported-by: Akinobu Mita <akinobu.mita@gmail.com>
Fixes: de3510e52b0a ("null_blk: fix command timeout completion handling")
Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20230314041106.19173-2-damien.lemoal@opensource.wdc.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

powerpc/mm: Fix false detection of read faults

To support detection of read faults with Radix execute-only memory, the
vma_is_accessible() check in access_error() (which checks for PROT_NONE)
was replaced with a check to see if VM_READ was missing, and if so,
returns true to assert the fault was caused by a bad read.

This is incorrect, as it ignores that both VM_WRITE and VM_EXEC imply
read on powerpc, as defined in protection_map[].  This causes mappings
containing VM_WRITE or VM_EXEC without VM_READ to misreport the cause of
page faults, since the MMU is still allowing reads.

Correct this by restoring the original vma_is_accessible() check for
PROT_NONE mappings, and adding a separate check for Radix PROT_EXEC-only
mappings.

Fixes: 395cac7752b9 ("powerpc/mm: Support execute-only memory on the Radix MMU")
Reported-by: Michal Suchánek <msuchanek@suse.de>
Link: https://lore.kernel.org/r/20230308152702.GR19419@kitsune.suse.cz
Tested-by: Benjamin Gray <bgray@linux.ibm.com>
Signed-off-by: Russell Currey <ruscur@russell.cc>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20230310050834.63105-1-ruscur@russell.cc

drm/meson: dw-hdmi: Fix devm_regulator_*get_enable*() conversion again

devm_regulator_get_enable_optional() returns -ENODEV if requested
optional regulator is not present. Adjust code for that, because in the
67d0a30128c9 I've incorrectly assumed that it also returns 0 when
regulator is not present.

Reported-by: Ricardo Cañuelo <ricardo.canuelo@collabora.com>
Fixes: 67d0a30128c9 ("drm/meson: dw-hdmi: Fix devm_regulator_*get_enable*() conversion")
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Acked-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Acked-by: Neil Armstrong <neil.armstrong@linaro.org>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230309152446.104913-1-m.szyprowski@samsung.com

drm/bridge: Fix returned array size name for atomic_get_input_bus_fmts kdoc

The returned array size for input formats is set through
atomic_get_input_bus_fmts()'s 'num_input_fmts' argument, so use
'num_input_fmts' to represent the array size in the function's kdoc,
not 'num_output_fmts'.

Fixes: 91ea83306bfa ("drm/bridge: Fix the bridge kernel doc")
Fixes: f32df58acc68 ("drm/bridge: Add the necessary bits to support bus format negotiation")
Signed-off-by: Liu Ying <victor.liu@nxp.com>
Reviewed-by: Robert Foss <rfoss@kernel.org>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20230314055035.3731179-1-victor.liu@nxp.com

Merge branch 'mtk_eth_soc-SGMII-fixes'

Daniel Golle says:

====================
net: ethernet: mtk_eth_soc: minor SGMII fixes

This small series brings two minor fixes for the SGMII unit found in
MediaTek's router SoCs.

The first patch resets the PCS internal state machine on major
configuration changes, just like it is also done in MediaTek's SDK.

The second patch makes sure we only write values and restart AN if
actually needed, thus preventing unnesseray loss of an existing link
in some cases.

Both patches have previously been submitted as part of the series
"net: ethernet: mtk_eth_soc: various enhancements" which grew a bit
too big and it has correctly been criticized that some of the patches
should rather go as fixes to net-next.

This new series tries to address this.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: only write values if needed

Only restart auto-negotiation and write link timer if actually
necessary. This prevents losing the link in case of minor
changes.

Fixes: 7e538372694b ("net: ethernet: mediatek: Re-add support SGMII")
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Tested-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: reset PCS state

Reset the internal PCS state machine when changing interface mode.
This prevents confusing the state machine when changing interface
modes, e.g. from SGMII to 2500Base-X or vice-versa.

Fixes: 7e538372694b ("net: ethernet: mediatek: Re-add support SGMII")
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Tested-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: usb: smsc75xx: Limit packet length to skb->len

Packet length retrieved from skb data may be larger than
the actual socket buffer length (up to 9026 bytes). In such
case the cloned skb passed up the network stack will leak
kernel memory contents.

Fixes: d0cad871703b ("smsc75xx: SMSC LAN75xx USB gigabit ethernet adapter driver")
Signed-off-by: Szymon Heidrich <szymon.heidrich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'net-smc-fixes'

Wenjia Zhang says:

====================
net/smc: Fixes 2023-03-01

The 1st patch solves the problem that CLC message initialization was
not properly reversed in error handling path. And the 2nd one fixes
the possible deadlock triggered by cancel_delayed_work_sync().
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

net/smc: Fix device de-init sequence

CLC message initialization was not properly reversed in error handling path.

Reported-and-suggested-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Stefan Raspl <raspl@linux.ibm.com>
Signed-off-by: Wenjia Zhang <wenjia@linux.ibm.com>
Reviewed-by: Tony Lu <tonylu@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/smc: fix deadlock triggered by cancel_delayed_work_syn()

The following LOCKDEP was detected:
Workqueue: events smc_lgr_free_work [smc]
WARNING: possible circular locking dependency detected
6.1.0-20221027.rc2.git8.56bc5b569087.300.fc36.s390x+debug #1 Not tainted
------------------------------------------------------
kworker/3:0/176251 is trying to acquire lock:
00000000f1467148 ((wq_completion)smc_tx_wq-00000000#2){+.+.}-{0:0},
at: __flush_workqueue+0x7a/0x4f0
but task is already holding lock:
0000037fffe97dc8 ((work_completion)(&(&lgr->free_work)->work)){+.+.}-{0:0},
at: process_one_work+0x232/0x730
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #4 ((work_completion)(&(&lgr->free_work)->work)){+.+.}-{0:0}:
       __lock_acquire+0x58e/0xbd8
       lock_acquire.part.0+0xe2/0x248
       lock_acquire+0xac/0x1c8
       __flush_work+0x76/0xf0
       __cancel_work_timer+0x170/0x220
       __smc_lgr_terminate.part.0+0x34/0x1c0 [smc]
       smc_connect_rdma+0x15e/0x418 [smc]
       __smc_connect+0x234/0x480 [smc]
       smc_connect+0x1d6/0x230 [smc]
       __sys_connect+0x90/0xc0
       __do_sys_socketcall+0x186/0x370
       __do_syscall+0x1da/0x208
       system_call+0x82/0xb0
-> #3 (smc_client_lgr_pending){+.+.}-{3:3}:
       __lock_acquire+0x58e/0xbd8
       lock_acquire.part.0+0xe2/0x248
       lock_acquire+0xac/0x1c8
       __mutex_lock+0x96/0x8e8
       mutex_lock_nested+0x32/0x40
       smc_connect_rdma+0xa4/0x418 [smc]
       __smc_connect+0x234/0x480 [smc]
       smc_connect+0x1d6/0x230 [smc]
       __sys_connect+0x90/0xc0
       __do_sys_socketcall+0x186/0x370
       __do_syscall+0x1da/0x208
       system_call+0x82/0xb0
-> #2 (sk_lock-AF_SMC){+.+.}-{0:0}:
       __lock_acquire+0x58e/0xbd8
       lock_acquire.part.0+0xe2/0x248
       lock_acquire+0xac/0x1c8
       lock_sock_nested+0x46/0xa8
       smc_tx_work+0x34/0x50 [smc]
       process_one_work+0x30c/0x730
       worker_thread+0x62/0x420
       kthread+0x138/0x150
       __ret_from_fork+0x3c/0x58
       ret_from_fork+0xa/0x40
-> #1 ((work_completion)(&(&smc->conn.tx_work)->work)){+.+.}-{0:0}:
       __lock_acquire+0x58e/0xbd8
       lock_acquire.part.0+0xe2/0x248
       lock_acquire+0xac/0x1c8
       process_one_work+0x2bc/0x730
       worker_thread+0x62/0x420
       kthread+0x138/0x150
       __ret_from_fork+0x3c/0x58
       ret_from_fork+0xa/0x40
-> #0 ((wq_completion)smc_tx_wq-00000000#2){+.+.}-{0:0}:
       check_prev_add+0xd8/0xe88
       validate_chain+0x70c/0xb20
       __lock_acquire+0x58e/0xbd8
       lock_acquire.part.0+0xe2/0x248
       lock_acquire+0xac/0x1c8
       __flush_workqueue+0xaa/0x4f0
       drain_workqueue+0xaa/0x158
       destroy_workqueue+0x44/0x2d8
       smc_lgr_free+0x9e/0xf8 [smc]
       process_one_work+0x30c/0x730
       worker_thread+0x62/0x420
       kthread+0x138/0x150
       __ret_from_fork+0x3c/0x58
       ret_from_fork+0xa/0x40
other info that might help us debug this:
Chain exists of:
  (wq_completion)smc_tx_wq-00000000#2
     --> smc_client_lgr_pending
  --> (work_completion)(&(&lgr->free_work)->work)
 Possible unsafe locking scenario:
       CPU0                    CPU1
       ----                    ----
  lock((work_completion)(&(&lgr->free_work)->work));
                   lock(smc_client_lgr_pending);
                   lock((work_completion)
(&(&lgr->free_work)->work));
  lock((wq_completion)smc_tx_wq-00000000#2);
 *** DEADLOCK ***
2 locks held by kworker/3:0/176251:
 #0: 0000000080183548
((wq_completion)events){+.+.}-{0:0},
at: process_one_work+0x232/0x730
 #1: 0000037fffe97dc8
((work_completion)
 (&(&lgr->free_work)->work)){+.+.}-{0:0},
at: process_one_work+0x232/0x730
stack backtrace:
CPU: 3 PID: 176251 Comm: kworker/3:0 Not tainted
Hardware name: IBM 8561 T01 701 (z/VM 7.2.0)
Call Trace:
 [<000000002983c3e4>] dump_stack_lvl+0xac/0x100
 [<0000000028b477ae>] check_noncircular+0x13e/0x160
 [<0000000028b48808>] check_prev_add+0xd8/0xe88
 [<0000000028b49cc4>] validate_chain+0x70c/0xb20
 [<0000000028b4bd26>] __lock_acquire+0x58e/0xbd8
 [<0000000028b4cf6a>] lock_acquire.part.0+0xe2/0x248
 [<0000000028b4d17c>] lock_acquire+0xac/0x1c8
 [<0000000028addaaa>] __flush_workqueue+0xaa/0x4f0
 [<0000000028addf9a>] drain_workqueue+0xaa/0x158
 [<0000000028ae303c>] destroy_workqueue+0x44/0x2d8
 [<000003ff8029af26>] smc_lgr_free+0x9e/0xf8 [smc]
 [<0000000028adf3d4>] process_one_work+0x30c/0x730
 [<0000000028adf85a>] worker_thread+0x62/0x420
 [<0000000028aeac50>] kthread+0x138/0x150
 [<0000000028a63914>] __ret_from_fork+0x3c/0x58
 [<00000000298503da>] ret_from_fork+0xa/0x40
INFO: lockdep is turned off.
===================================================================

This deadlock occurs because cancel_delayed_work_sync() waits for
the work(&lgr->free_work) to finish, while the &lgr->free_work
waits for the work(lgr->tx_wq), which needs the sk_lock-AF_SMC, that
is already used under the mutex_lock.

The solution is to use cancel_delayed_work() instead, which kills
off a pending work.

Fixes: a52bcc919b14 ("net/smc: improve termination processing")
Signed-off-by: Wenjia Zhang <wenjia@linux.ibm.com>
Reviewed-by: Jan Karcher <jaka@linux.ibm.com>
Reviewed-by: Karsten Graul <kgraul@linux.ibm.com>
Reviewed-by: Tony Lu <tonylu@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlxsw: spectrum: Fix incorrect parsing depth after reload

Spectrum ASICs have a configurable limit on how deep into the packet
they parse. By default, the limit is 96 bytes.

There are several cases where this parsing depth is not enough and there
is a need to increase it. For example, timestamping of PTP packets and a
FIB multipath hash policy that requires hashing on inner fields. The
driver therefore maintains a reference count that reflects the number of
consumers that require an increased parsing depth.

During reload_down() the parsing depth reference count does not
necessarily drop to zero, but the parsing depth itself is restored to
the default during reload_up() when the firmware is reset. It is
therefore possible to end up in situations where the driver thinks that
the parsing depth was increased (reference count is non-zero), when it
is not.

Fix by making sure that all the consumers that increase the parsing
depth reference count also decrease it during reload_down().
Specifically, make sure that when the routing code is de-initialized it
drops the reference count if it was increased because of a FIB multipath
hash policy that requires hashing on inner fields.

Add a warning if the reference count is not zero after the driver was
de-initialized and explicitly reset it to zero during initialization for
good measures.

Fixes: 2d91f0803b84 ("mlxsw: spectrum: Add infrastructure for parsing configuration")
Reported-by: Maksym Yaremchuk <maksymy@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: Petr Machata <petrm@nvidia.com>
Link: https://lore.kernel.org/r/9c35e1b3e6c1d8f319a2449d14e2b86373f3b3ba.1678727526.git.petrm@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

veth: rely on rtnl_dereference() instead of on rcu_dereference() in veth_set_xdp_features()

Fix the following kernel warning in veth_set_xdp_features routine
relying on rtnl_dereference() instead of on rcu_dereference():

=============================
WARNING: suspicious RCU usage
6.3.0-rc1-00144-g064d70527aaa #149 Not tainted
-----------------------------
drivers/net/veth.c:1265 suspicious rcu_dereference_check() usage!

other info that might help us debug this:

rcu_scheduler_active = 2, debug_locks = 1
1 lock held by ip/135:
(net/core/rtnetlink.c:6172)

stack backtrace:
CPU: 1 PID: 135 Comm: ip Not tainted 6.3.0-rc1-00144-g064d70527aaa #149
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1
04/01/2014
Call Trace:
 <TASK>
dump_stack_lvl (lib/dump_stack.c:107)
lockdep_rcu_suspicious (include/linux/context_tracking.h:152)
veth_set_xdp_features (drivers/net/veth.c:1265 (discriminator 9))
veth_newlink (drivers/net/veth.c:1892)
? veth_set_features (drivers/net/veth.c:1774)
? kasan_save_stack (mm/kasan/common.c:47)
? kasan_save_stack (mm/kasan/common.c:46)
? kasan_set_track (mm/kasan/common.c:52)
? alloc_netdev_mqs (include/linux/slab.h:737)
? rcu_read_lock_sched_held (kernel/rcu/update.c:125)
? trace_kmalloc (include/trace/events/kmem.h:54)
? __xdp_rxq_info_reg (net/core/xdp.c:188)
? alloc_netdev_mqs (net/core/dev.c:10657)
? rtnl_create_link (net/core/rtnetlink.c:3312)
rtnl_newlink_create (net/core/rtnetlink.c:3440)
? rtnl_link_get_net_capable.constprop.0 (net/core/rtnetlink.c:3391)
__rtnl_newlink (net/core/rtnetlink.c:3657)
? lock_downgrade (kernel/locking/lockdep.c:5321)
? rtnl_link_unregister (net/core/rtnetlink.c:3487)
rtnl_newlink (net/core/rtnetlink.c:3671)
rtnetlink_rcv_msg (net/core/rtnetlink.c:6174)
? rtnl_link_fill (net/core/rtnetlink.c:6070)
? mark_usage (kernel/locking/lockdep.c:4914)
? mark_usage (kernel/locking/lockdep.c:4914)
netlink_rcv_skb (net/netlink/af_netlink.c:2574)
? rtnl_link_fill (net/core/rtnetlink.c:6070)
? netlink_ack (net/netlink/af_netlink.c:2551)
? lock_acquire (kernel/locking/lockdep.c:467)
? net_generic (include/linux/rcupdate.h:805)
? netlink_deliver_tap (include/linux/rcupdate.h:805)
netlink_unicast (net/netlink/af_netlink.c:1340)
? netlink_attachskb (net/netlink/af_netlink.c:1350)
netlink_sendmsg (net/netlink/af_netlink.c:1942)
? netlink_unicast (net/netlink/af_netlink.c:1861)
? netlink_unicast (net/netlink/af_netlink.c:1861)
sock_sendmsg (net/socket.c:727)
____sys_sendmsg (net/socket.c:2501)
? kernel_sendmsg (net/socket.c:2448)
? __copy_msghdr (net/socket.c:2428)
___sys_sendmsg (net/socket.c:2557)
? mark_usage (kernel/locking/lockdep.c:4914)
? do_recvmmsg (net/socket.c:2544)
? lock_acquire (kernel/locking/lockdep.c:467)
? find_held_lock (kernel/locking/lockdep.c:5159)
? __lock_release (kernel/locking/lockdep.c:5345)
? __might_fault (mm/memory.c:5625)
? lock_downgrade (kernel/locking/lockdep.c:5321)
? __fget_light (include/linux/atomic/atomic-arch-fallback.h:227)
__sys_sendmsg (include/linux/file.h:31)
? __sys_sendmsg_sock (net/socket.c:2572)
? rseq_get_rseq_cs (kernel/rseq.c:275)
? lockdep_hardirqs_on_prepare.part.0 (kernel/locking/lockdep.c:4263)
do_syscall_64 (arch/x86/entry/common.c:50)
entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:120)
RIP: 0033:0x7f0d1aadeb17
Code: 0f 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b9 0f 1f 00 f3 0f 1e
fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 2e 00 00 00 0f 05 <48> 3d 00
f0 ff ff 77 51 c3 48 83 ec 28 89 54 24 1c 48 89 74 24 10

Fixes: fccca038f300 ("veth: take into account device reconfiguration for xdp_features flag")
Suggested-by: Eric Dumazet <edumazet@google.com>
Reported-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Link: https://lore.kernel.org/netdev/cover.1678364612.git.lorenzo@kernel.org/T/#me4c9d8e985ec7ebee981cfdb5bc5ec651ef4035d
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Reported-by: syzbot+c3d0d9c42d59ff644ea6@syzkaller.appspotmail.com
Reviewed-by: Eric Dumazet <edumazet@google.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Link: https://lore.kernel.org/r/dfd6a9a7d85e9113063165e1f47b466b90ad7b8a.1678748579.git.lorenzo@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

nfc: st-nci: Fix use after free bug in ndlc_remove due to race condition

This bug influences both st_nci_i2c_remove and st_nci_spi_remove.
Take st_nci_i2c_remove as an example.

In st_nci_i2c_probe, it called ndlc_probe and bound &ndlc->sm_work
with llt_ndlc_sm_work.

When it calls ndlc_recv or timeout handler, it will finally call
schedule_work to start the work.

When we call st_nci_i2c_remove to remove the driver, there
may be a sequence as follows:

Fix it by finishing the work before cleanup in ndlc_remove

CPU0                  CPU1

                    |llt_ndlc_sm_work
st_nci_i2c_remove   |
  ndlc_remove       |
     st_nci_remove  |
     nci_free_device|
     kfree(ndev)    |
//free ndlc->ndev   |
                    |llt_ndlc_rcv_queue
                    |nci_recv_frame
                    |//use ndlc->ndev

Fixes: 35630df68d60 ("NFC: st21nfcb: Add driver for STMicroelectronics ST21NFCB NFC chip")
Signed-off-by: Zheng Wang <zyytlz.wz@163.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20230312160837.2040857-1-zyytlz.wz@163.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'tcp-fix-bind-regression-for-dual-stack-wildcard-address'

Kuniyuki Iwashima says:

====================
tcp: Fix bind() regression for dual-stack wildcard address.

The first patch fixes the regression reported in [0], and the second
patch adds a test for similar cases to catch future regression.

[0]: https://lore.kernel.org/netdev/e21bf153-80b0-9ec0-15ba-e04a4ad42c34@redhat.com/
====================

Link: https://lore.kernel.org/r/20230312031904.4674-1-kuniyu@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftest: Add test for bind() conflicts.

The test checks if (IPv4, IPv6) address pair properly conflict or not.

  * IPv4
    * 0.0.0.0
    * 127.0.0.1

  * IPv6
    * ::
    * ::1

If the IPv6 address is [::], the second bind() always fails.

Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

tcp: Fix bind() conflict check for dual-stack wildcard address.

Paul Holzinger reported [0] that commit 5456262d2baa ("net: Fix
incorrect address comparison when searching for a bind2 bucket")
introduced a bind() regression.  Paul also gave a nice repro that
calls two types of bind() on the same port, both of which now
succeed, but the second call should fail:

  bind(fd1, ::, port) + bind(fd2, 127.0.0.1, port)

The cited commit added address family tests in three functions to
fix the uninit-value KMSAN report. [1]  However, the test added to
inet_bind2_bucket_match_addr_any() removed a necessary conflict
check; the dual-stack wildcard address no longer conflicts with
an IPv4 non-wildcard address.

If tb->family is AF_INET6 and sk->sk_family is AF_INET in
inet_bind2_bucket_match_addr_any(), we still need to check
if tb has the dual-stack wildcard address.

Note that the IPv4 wildcard address does not conflict with
IPv6 non-wildcard addresses.

[0]: https://lore.kernel.org/netdev/e21bf153-80b0-9ec0-15ba-e04a4ad42c34@redhat.com/
[1]: https://lore.kernel.org/netdev/CAG_fn=Ud3zSW7AZWXc+asfMhZVL5ETnvuY44Pmyv4NPv-ijN-A@mail.gmail.com/

Fixes: 5456262d2baa ("net: Fix incorrect address comparison when searching for a bind2 bucket")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reported-by: Paul Holzinger <pholzing@redhat.com>
Link: https://lore.kernel.org/netdev/CAG_fn=Ud3zSW7AZWXc+asfMhZVL5ETnvuY44Pmyv4NPv-ijN-A@mail.gmail.com/
Reviewed-by: Eric Dumazet <edumazet@google.com>
Tested-by: Paul Holzinger <pholzing@redhat.com>
Reviewed-by: Martin KaFai Lau <martin.lau@kernel.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: phy: smsc: bail out in lan87xx_read_status if genphy_read_status fails

If genphy_read_status fails then further access to the PHY may result
in unpredictable behavior. To prevent this bail out immediately if
genphy_read_status fails.

Fixes: 4223dbffed9f ("net: phy: smsc: Re-enable EDPD mode for LAN87xx")
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Link: https://lore.kernel.org/r/026aa4f2-36f5-1c10-ab9f-cdb17dda6ac4@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: tunnels: annotate lockless accesses to dev->needed_headroom

IP tunnels can apparently update dev->needed_headroom
in their xmit path.

This patch takes care of three tunnels xmit, and also the
core LL_RESERVED_SPACE() and LL_RESERVED_SPACE_EXTRA()
helpers.

More changes might be needed for completeness.

BUG: KCSAN: data-race in ip_tunnel_xmit / ip_tunnel_xmit

read to 0xffff88815b9da0ec of 2 bytes by task 888 on cpu 1:
ip_tunnel_xmit+0x1270/0x1730 net/ipv4/ip_tunnel.c:803
__gre_xmit net/ipv4/ip_gre.c:469 [inline]
ipgre_xmit+0x516/0x570 net/ipv4/ip_gre.c:661
__netdev_start_xmit include/linux/netdevice.h:4881 [inline]
netdev_start_xmit include/linux/netdevice.h:4895 [inline]
xmit_one net/core/dev.c:3580 [inline]
dev_hard_start_xmit+0x127/0x400 net/core/dev.c:3596
__dev_queue_xmit+0x1007/0x1eb0 net/core/dev.c:4246
dev_queue_xmit include/linux/netdevice.h:3051 [inline]
neigh_direct_output+0x17/0x20 net/core/neighbour.c:1623
neigh_output include/net/neighbour.h:546 [inline]
ip_finish_output2+0x740/0x840 net/ipv4/ip_output.c:228
ip_finish_output+0xf4/0x240 net/ipv4/ip_output.c:316
NF_HOOK_COND include/linux/netfilter.h:291 [inline]
ip_output+0xe5/0x1b0 net/ipv4/ip_output.c:430
dst_output include/net/dst.h:444 [inline]
ip_local_out+0x64/0x80 net/ipv4/ip_output.c:126
iptunnel_xmit+0x34a/0x4b0 net/ipv4/ip_tunnel_core.c:82
ip_tunnel_xmit+0x1451/0x1730 net/ipv4/ip_tunnel.c:813
__gre_xmit net/ipv4/ip_gre.c:469 [inline]
ipgre_xmit+0x516/0x570 net/ipv4/ip_gre.c:661
__netdev_start_xmit include/linux/netdevice.h:4881 [inline]
netdev_start_xmit include/linux/netdevice.h:4895 [inline]
xmit_one net/core/dev.c:3580 [inline]
dev_hard_start_xmit+0x127/0x400 net/core/dev.c:3596
__dev_queue_xmit+0x1007/0x1eb0 net/core/dev.c:4246
dev_queue_xmit include/linux/netdevice.h:3051 [inline]
neigh_direct_output+0x17/0x20 net/core/neighbour.c:1623
neigh_output include/net/neighbour.h:546 [inline]
ip_finish_output2+0x740/0x840 net/ipv4/ip_output.c:228
ip_finish_output+0xf4/0x240 net/ipv4/ip_output.c:316
NF_HOOK_COND include/linux/netfilter.h:291 [inline]
ip_output+0xe5/0x1b0 net/ipv4/ip_output.c:430
dst_output include/net/dst.h:444 [inline]
ip_local_out+0x64/0x80 net/ipv4/ip_output.c:126
iptunnel_xmit+0x34a/0x4b0 net/ipv4/ip_tunnel_core.c:82
ip_tunnel_xmit+0x1451/0x1730 net/ipv4/ip_tunnel.c:813
__gre_xmit net/ipv4/ip_gre.c:469 [inline]
ipgre_xmit+0x516/0x570 net/ipv4/ip_gre.c:661
__netdev_start_xmit include/linux/netdevice.h:4881 [inline]
netdev_start_xmit include/linux/netdevice.h:4895 [inline]
xmit_one net/core/dev.c:3580 [inline]
dev_hard_start_xmit+0x127/0x400 net/core/dev.c:3596
__dev_queue_xmit+0x1007/0x1eb0 net/core/dev.c:4246
dev_queue_xmit include/linux/netdevice.h:3051 [inline]
neigh_direct_output+0x17/0x20 net/core/neighbour.c:1623
neigh_output include/net/neighbour.h:546 [inline]
ip_finish_output2+0x740/0x840 net/ipv4/ip_output.c:228
ip_finish_output+0xf4/0x240 net/ipv4/ip_output.c:316
NF_HOOK_COND include/linux/netfilter.h:291 [inline]
ip_output+0xe5/0x1b0 net/ipv4/ip_output.c:430
dst_output include/net/dst.h:444 [inline]
ip_local_out+0x64/0x80 net/ipv4/ip_output.c:126
iptunnel_xmit+0x34a/0x4b0 net/ipv4/ip_tunnel_core.c:82
ip_tunnel_xmit+0x1451/0x1730 net/ipv4/ip_tunnel.c:813
__gre_xmit net/ipv4/ip_gre.c:469 [inline]
ipgre_xmit+0x516/0x570 net/ipv4/ip_gre.c:661
__netdev_start_xmit include/linux/netdevice.h:4881 [inline]
netdev_start_xmit include/linux/netdevice.h:4895 [inline]
xmit_one net/core/dev.c:3580 [inline]
dev_hard_start_xmit+0x127/0x400 net/core/dev.c:3596
__dev_queue_xmit+0x1007/0x1eb0 net/core/dev.c:4246
dev_queue_xmit include/linux/netdevice.h:3051 [inline]
neigh_direct_output+0x17/0x20 net/core/neighbour.c:1623
neigh_output include/net/neighbour.h:546 [inline]
ip_finish_output2+0x740/0x840 net/ipv4/ip_output.c:228
ip_finish_output+0xf4/0x240 net/ipv4/ip_output.c:316
NF_HOOK_COND include/linux/netfilter.h:291 [inline]
ip_output+0xe5/0x1b0 net/ipv4/ip_output.c:430
dst_output include/net/dst.h:444 [inline]
ip_local_out+0x64/0x80 net/ipv4/ip_output.c:126
iptunnel_xmit+0x34a/0x4b0 net/ipv4/ip_tunnel_core.c:82
ip_tunnel_xmit+0x1451/0x1730 net/ipv4/ip_tunnel.c:813
__gre_xmit net/ipv4/ip_gre.c:469 [inline]
ipgre_xmit+0x516/0x570 net/ipv4/ip_gre.c:661
__netdev_start_xmit include/linux/netdevice.h:4881 [inline]
netdev_start_xmit include/linux/netdevice.h:4895 [inline]
xmit_one net/core/dev.c:3580 [inline]
dev_hard_start_xmit+0x127/0x400 net/core/dev.c:3596
__dev_queue_xmit+0x1007/0x1eb0 net/core/dev.c:4246
dev_queue_xmit include/linux/netdevice.h:3051 [inline]
neigh_direct_output+0x17/0x20 net/core/neighbour.c:1623
neigh_output include/net/neighbour.h:546 [inline]
ip_finish_output2+0x740/0x840 net/ipv4/ip_output.c:228
ip_finish_output+0xf4/0x240 net/ipv4/ip_output.c:316
NF_HOOK_COND include/linux/netfilter.h:291 [inline]
ip_output+0xe5/0x1b0 net/ipv4/ip_output.c:430
dst_output include/net/dst.h:444 [inline]
ip_local_out+0x64/0x80 net/ipv4/ip_output.c:126
iptunnel_xmit+0x34a/0x4b0 net/ipv4/ip_tunnel_core.c:82
ip_tunnel_xmit+0x1451/0x1730 net/ipv4/ip_tunnel.c:813
__gre_xmit net/ipv4/ip_gre.c:469 [inline]
ipgre_xmit+0x516/0x570 net/ipv4/ip_gre.c:661
__netdev_start_xmit include/linux/netdevice.h:4881 [inline]
netdev_start_xmit include/linux/netdevice.h:4895 [inline]
xmit_one net/core/dev.c:3580 [inline]
dev_hard_start_xmit+0x127/0x400 net/core/dev.c:3596
__dev_queue_xmit+0x1007/0x1eb0 net/core/dev.c:4246
dev_queue_xmit include/linux/netdevice.h:3051 [inline]
neigh_direct_output+0x17/0x20 net/core/neighbour.c:1623
neigh_output include/net/neighbour.h:546 [inline]
ip_finish_output2+0x740/0x840 net/ipv4/ip_output.c:228
ip_finish_output+0xf4/0x240 net/ipv4/ip_output.c:316
NF_HOOK_COND include/linux/netfilter.h:291 [inline]
ip_output+0xe5/0x1b0 net/ipv4/ip_output.c:430
dst_output include/net/dst.h:444 [inline]
ip_local_out+0x64/0x80 net/ipv4/ip_output.c:126
iptunnel_xmit+0x34a/0x4b0 net/ipv4/ip_tunnel_core.c:82
ip_tunnel_xmit+0x1451/0x1730 net/ipv4/ip_tunnel.c:813
__gre_xmit net/ipv4/ip_gre.c:469 [inline]
ipgre_xmit+0x516/0x570 net/ipv4/ip_gre.c:661
__netdev_start_xmit include/linux/netdevice.h:4881 [inline]
netdev_start_xmit include/linux/netdevice.h:4895 [inline]
xmit_one net/core/dev.c:3580 [inline]
dev_hard_start_xmit+0x127/0x400 net/core/dev.c:3596
__dev_queue_xmit+0x1007/0x1eb0 net/core/dev.c:4246

write to 0xffff88815b9da0ec of 2 bytes by task 2379 on cpu 0:
ip_tunnel_xmit+0x1294/0x1730 net/ipv4/ip_tunnel.c:804
__gre_xmit net/ipv4/ip_gre.c:469 [inline]
ipgre_xmit+0x516/0x570 net/ipv4/ip_gre.c:661
__netdev_start_xmit include/linux/netdevice.h:4881 [inline]
netdev_start_xmit include/linux/netdevice.h:4895 [inline]
xmit_one net/core/dev.c:3580 [inline]
dev_hard_start_xmit+0x127/0x400 net/core/dev.c:3596
__dev_queue_xmit+0x1007/0x1eb0 net/core/dev.c:4246
dev_queue_xmit include/linux/netdevice.h:3051 [inline]
neigh_direct_output+0x17/0x20 net/core/neighbour.c:1623
neigh_output include/net/neighbour.h:546 [inline]
ip6_finish_output2+0x9bc/0xc50 net/ipv6/ip6_output.c:134
__ip6_finish_output net/ipv6/ip6_output.c:195 [inline]
ip6_finish_output+0x39a/0x4e0 net/ipv6/ip6_output.c:206
NF_HOOK_COND include/linux/netfilter.h:291 [inline]
ip6_output+0xeb/0x220 net/ipv6/ip6_output.c:227
dst_output include/net/dst.h:444 [inline]
NF_HOOK include/linux/netfilter.h:302 [inline]
mld_sendpack+0x438/0x6a0 net/ipv6/mcast.c:1820
mld_send_cr net/ipv6/mcast.c:2121 [inline]
mld_ifc_work+0x519/0x7b0 net/ipv6/mcast.c:2653
process_one_work+0x3e6/0x750 kernel/workqueue.c:2390
worker_thread+0x5f2/0xa10 kernel/workqueue.c:2537
kthread+0x1ac/0x1e0 kernel/kthread.c:376
ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:308

value changed: 0x0dd4 -> 0x0e14

Reported by Kernel Concurrency Sanitizer on:
CPU: 0 PID: 2379 Comm: kworker/0:0 Not tainted 6.3.0-rc1-syzkaller-00002-g8ca09d5fa354-dirty #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 03/02/2023
Workqueue: mld mld_ifc_work

Fixes: 8eb30be0352d ("ipv6: Create ip6_tnl_xmit")
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20230310191109.2384387-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

ice: avoid bonding causing auxiliary plug/unplug under RTNL lock

RDMA is not supported in ice on a PF that has been added to a bonded
interface. To enforce this, when an interface enters a bond, we unplug
the auxiliary device that supports RDMA functionality.  This unplug
currently happens in the context of handling the netdev bonding event.
This event is sent to the ice driver under RTNL context.  This is causing
a deadlock where the RDMA driver is waiting for the RTNL lock to complete
the removal.

Defer the unplugging/re-plugging of the auxiliary device to the service
task so that it is not performed under the RTNL lock context.

Cc: stable@vger.kernel.org # 6.1.x
Reported-by: Jaroslav Pulchart <jaroslav.pulchart@gooddata.com>
Link: https://lore.kernel.org/netdev/CAK8fFZ6A_Gphw_3-QMGKEFQk=sfCw1Qmq0TVZK3rtAi7vb621A@mail.gmail.com/
Fixes: 5cb1ebdbc434 ("ice: Fix race condition during interface enslave")
Fixes: 4eace75e0853 ("RDMA/irdma: Report the correct link speed")
Signed-off-by: Dave Ertman <david.m.ertman@intel.com>
Tested-by: Arpana Arland <arpanax.arland@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Link: https://lore.kernel.org/r/20230310194833.3074601-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

cifs: use DFS root session instead of tcon ses

Use DFS root session whenever possible to get new DFS referrals
otherwise we might end up with an IPC tcon (tcon->ses->tcon_ipc) that
doesn't respond to them.  It should be safe accessing
@ses->dfs_root_ses directly in cifs_inval_name_dfs_link_error() as it
has same lifetime as of @tcon.

Signed-off-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Cc: stable@vger.kernel.org # 6.2
Signed-off-by: Steve French <stfrench@microsoft.com>

cifs: return DFS root session id in DebugData

Return the DFS root session id in /proc/fs/cifs/DebugData to make it
easier to track which IPC tcon was used to get new DFS referrals for a
specific connection, and aids in debugging.

A simple output of it would be

  Sessions:
  1) Address: 192.168.1.13 Uses: 1 Capability: 0x300067   Session Status: 1
  Security type: RawNTLMSSP  SessionId: 0xd80000000009
  User: 0 Cred User: 0
  DFS root session id: 0x128006c000035

Signed-off-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Cc: stable@vger.kernel.org # 6.2
Signed-off-by: Steve French <stfrench@microsoft.com>

sched_getaffinity: don't assume 'cpumask_size()' is fully initialized

The getaffinity() system call uses 'cpumask_size()' to decide how big
the CPU mask is - so far so good.  It is indeed the allocation size of a
cpumask.

But the code also assumes that the whole allocation is initialized
without actually doing so itself.  That's wrong, because we might have
fixed-size allocations (making copying and clearing more efficient), but
not all of it is then necessarily used if 'nr_cpu_ids' is smaller.

Having checked other users of 'cpumask_size()', they all seem to be ok,
either using it purely for the allocation size, or explicitly zeroing
the cpumask before using the size in bytes to copy it.

See for example the ublk_ctrl_get_queue_affinity() function that uses
the proper 'zalloc_cpumask_var()' to make sure that the whole mask is
cleared, whether the storage is on the stack or if it was an external
allocation.

Fix this by just zeroing the allocation before using it.  Do the same
for the compat version of sched_getaffinity(), which had the same logic.

Also, for consistency, make sched_getaffinity() use 'cpumask_bits()' to
access the bits.  For a cpumask_var_t, it ends up being a pointer to the
same data either way, but it's just a good idea to treat it like you
would a 'cpumask_t'.  The compat case already did that.

Reported-by: Ryan Roberts <ryan.roberts@arm.com>
Link: https://lore.kernel.org/lkml/7d026744-6bd6-6827-0471-b5e8eae0be3f@arm.com/
Cc: Yury Norov <yury.norov@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

RISC-V: mm: Support huge page in vmalloc_fault()

Since RISC-V supports ioremap() with huge page (pud/pmd) mapping,
However, vmalloc_fault() assumes that the vmalloc range is limited
to pte mappings. To complete the vmalloc_fault() function by adding
huge page support.

Fixes: 310f541a027b ("riscv: Enable HAVE_ARCH_HUGE_VMAP for 64BIT")
Cc: stable@vger.kernel.org
Signed-off-by: Dylan Jhong <dylan@andestech.com>
Reviewed-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Link: https://lore.kernel.org/r/20230310075021.3919290-1-dylan@andestech.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

cifs: fix use-after-free bug in refresh_cache_worker()

The UAF bug occurred because we were putting DFS root sessions in
cifs_umount() while DFS cache refresher was being executed.

Make DFS root sessions have same lifetime as DFS tcons so we can avoid
the use-after-free bug is DFS cache refresher and other places that
require IPCs to get new DFS referrals on.  Also, get rid of mount
group handling in DFS cache as we no longer need it.

This fixes below use-after-free bug catched by KASAN

[ 379.946955] BUG: KASAN: use-after-free in __refresh_tcon.isra.0+0x10b/0xc10 [cifs]
[ 379.947642] Read of size 8 at addr ffff888018f57030 by task kworker/u4:3/56
[ 379.948096]
[ 379.948208] CPU: 0 PID: 56 Comm: kworker/u4:3 Not tainted 6.2.0-rc7-lku #23
[ 379.948661] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS
rel-1.16.0-0-gd239552-rebuilt.opensuse.org 04/01/2014
[ 379.949368] Workqueue: cifs-dfscache refresh_cache_worker [cifs]
[ 379.949942] Call Trace:
[ 379.950113] <TASK>
[ 379.950260] dump_stack_lvl+0x50/0x67
[ 379.950510] print_report+0x16a/0x48e
[ 379.950759] ? __virt_addr_valid+0xd8/0x160
[ 379.951040] ? __phys_addr+0x41/0x80
[ 379.951285] kasan_report+0xdb/0x110
[ 379.951533] ? __refresh_tcon.isra.0+0x10b/0xc10 [cifs]
[ 379.952056] ? __refresh_tcon.isra.0+0x10b/0xc10 [cifs]
[ 379.952585] __refresh_tcon.isra.0+0x10b/0xc10 [cifs]
[ 379.953096] ? __pfx___refresh_tcon.isra.0+0x10/0x10 [cifs]
[ 379.953637] ? __pfx___mutex_lock+0x10/0x10
[ 379.953915] ? lock_release+0xb6/0x720
[ 379.954167] ? __pfx_lock_acquire+0x10/0x10
[ 379.954443] ? refresh_cache_worker+0x34e/0x6d0 [cifs]
[ 379.954960] ? __pfx_wb_workfn+0x10/0x10
[ 379.955239] refresh_cache_worker+0x4ad/0x6d0 [cifs]
[ 379.955755] ? __pfx_refresh_cache_worker+0x10/0x10 [cifs]
[ 379.956323] ? __pfx_lock_acquired+0x10/0x10
[ 379.956615] ? read_word_at_a_time+0xe/0x20
[ 379.956898] ? lockdep_hardirqs_on_prepare+0x12/0x220
[ 379.957235] process_one_work+0x535/0x990
[ 379.957509] ? __pfx_process_one_work+0x10/0x10
[ 379.957812] ? lock_acquired+0xb7/0x5f0
[ 379.958069] ? __list_add_valid+0x37/0xd0
[ 379.958341] ? __list_add_valid+0x37/0xd0
[ 379.958611] worker_thread+0x8e/0x630
[ 379.958861] ? __pfx_worker_thread+0x10/0x10
[ 379.959148] kthread+0x17d/0x1b0
[ 379.959369] ? __pfx_kthread+0x10/0x10
[ 379.959630] ret_from_fork+0x2c/0x50
[ 379.959879] </TASK>

Signed-off-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Cc: stable@vger.kernel.org # 6.2
Signed-off-by: Steve French <stfrench@microsoft.com>

cifs: set DFS root session in cifs_get_smb_ses()

Set the DFS root session pointer earlier when creating a new SMB
session to prevent racing with smb2_reconnect(), cifs_reconnect_tcon()
and DFS cache refresher.

Signed-off-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Cc: stable@vger.kernel.org # 6.2
Signed-off-by: Steve French <stfrench@microsoft.com>

blk-mq: fix "bad unlock balance detected" on q->srcu in __blk_mq_run_dispatch_ops

The 'q' parameter of the macro __blk_mq_run_dispatch_ops may not be one
local variable, such as, it is rq->q, then request queue pointed by
this variable could be changed to another queue in case of
BLK_MQ_F_TAG_QUEUE_SHARED after 'dispatch_ops' returns, then
'bad unlock balance' is triggered.

Fixes the issue by adding one local variable for doing srcu lock/unlock.

Fixes: 2a904d00855f ("blk-mq: remove hctx_lock and hctx_unlock")
Cc: Marco Patalano <mpatalan@redhat.com>
Signed-off-by: Chris Leech <cleech@redhat.com>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20230310010913.1014789-1-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

loop: Fix use-after-free issues

do_req_filebacked() calls blk_mq_complete_request() synchronously or
asynchronously when using asynchronous I/O unless memory allocation fails.
Hence, modify loop_handle_cmd() such that it does not dereference 'cmd' nor
'rq' after do_req_filebacked() finished unless we are sure that the request
has not yet been completed. This patch fixes the following kernel crash:

Unable to handle kernel NULL pointer dereference at virtual address 0000000000000054
Call trace:
 css_put.42938+0x1c/0x1ac
 loop_process_work+0xc8c/0xfd4
 loop_rootcg_workfn+0x24/0x34
 process_one_work+0x244/0x558
 worker_thread+0x400/0x8fc
 kthread+0x16c/0x1e0
 ret_from_fork+0x10/0x20

Cc: Christoph Hellwig <hch@lst.de>
Cc: Ming Lei <ming.lei@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Dan Schatzberg <schatzberg.dan@gmail.com>
Fixes: c74d40e8b5e2 ("loop: charge i/o to mem and blk cg")
Fixes: bc07c10a3603 ("block: loop: support DIO & AIO")
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20230314182155.80625-1-bvanassche@acm.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>

Merge tag 'mm-hotfixes-stable-2023-03-14-16-51' of git://git./linux/kernel/git/akpm/mm

Pull misc fixes from Andrew Morton:
 "Eleven hotfixes.

  Four of these are cc:stable and the remainder address post-6.2 issues
  or aren't considered suitable for backporting.

  Seven of these fixes are for MM"

* tag 'mm-hotfixes-stable-2023-03-14-16-51' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm:
  mm/damon/paddr: fix folio_nr_pages() after folio_put() in damon_pa_mark_accessed_or_deactivate()
  mm/damon/paddr: fix folio_size() call after folio_put() in damon_pa_young()
  ocfs2: fix data corruption after failed write
  migrate_pages: try migrate in batch asynchronously firstly
  migrate_pages: move split folios processing out of migrate_pages_batch()
  migrate_pages: fix deadlock in batched migration
  .mailmap: add Alexandre Ghiti personal email address
  mailmap: correct Dikshita Agarwal's Qualcomm email address
  mailmap: updates for Jarkko Sakkinen
  mm/userfaultfd: propagate uffd-wp bit when PTE-mapping the huge zeropage
  mm: teach mincore_hugetlb about pte markers

Merge tag 'trace-v6.3-rc1' of git://git./linux/kernel/git/trace/linux-trace

Pull tracing fixes from Steven Rostedt:

 - Do not allow histogram values to have modifies. They can cause a NULL
   pointer dereference if they do.

 - Warn if hist_field_name() is passed a NULL. Prevent the NULL pointer
   dereference mentioned above.

 - Fix invalid address look up race in lookup_rec()

 - Define ftrace_stub_graph conditionally to prevent linker errors

 - Always check if RCU is watching at all tracepoint locations

* tag 'trace-v6.3-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  tracing: Make tracepoint lockdep check actually test something
  ftrace,kcfi: Define ftrace_stub_graph conditionally
  ftrace: Fix invalid address access in lookup_rec() when index is 0
  tracing: Check field value in hist_field_name()
  tracing: Do not let histogram values have some modifiers

Merge tag 'zstd-linus-v6.3-rc3' of https://github.com/terrelln/linux

Pull zstd fixes from Nick Terrell:
 "A small number of fixes for zstd-v1.5.2.

  I'm not pulling in zstd-v1.5.4 from upstream this release because it
  didn't have any time to bake in linux-next, but I'm aiming for the
  next update in v6.4"

* tag 'zstd-linus-v6.3-rc3' of https://github.com/terrelln/linux:
  zstd: Fix definition of assert()
  lib: zstd: Backport fix for in-place decompression
  lib: zstd: Fix -Wstringop-overflow warning

Merge tag 'clk-fixes-for-linus' of git://git./linux/kernel/git/clk/linux

Pull clk fixes from Stephen Boyd:
 "A collection of clk driver fixes, and a couple OF clk patches to fix
  regressions seen in the last few weeks. The fwnode patch broke the
  build for one driver that isn't always compiled, so I waited over the
  weekend to be certain no more build issues came up.

   - Mark the firmware node (fwnode) that matches the compatible in
     CLK_OF_DECLARE() as initialized to fix a regression on u8500 SoCs
     after fw_devlink stopped checking parent nodes in
     of_link_to_phandle()

   - Remove a couple MODULE_LICENSE macros in non-modules

   - Update the maintainers file for Microchip clk drivers

   - Use 'select' instead of 'depend on' for the REGMAP config to fix
     Kconfig issues

   - Use div_u64() for portable 64-bit division in K210 clk driver"

* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
  clk: Avoid invalid function names in CLK_OF_DECLARE()
  clk: k210: remove an implicit 64-bit division
  MAINTAINERS: add missing clock driver coverage for Microchip FPGAs
  clk: HI655X: select REGMAP instead of depending on it
  kbuild, clk: remove MODULE_LICENSE in non-modules
  kbuild, clk: bcm2835: remove MODULE_LICENSE in non-modules
  clk: Mark a fwnode as initialized when using CLK_OF_DECLARE() macro

cifs: generate signkey for the channel that's reconnecting

Before my changes to how multichannel reconnects work, the
primary channel was always used to do a non-binding session
setup. With my changes, that is not the case anymore.
Missed this place where channel at index 0 was forcibly
updated with the signing key.

Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Reviewed-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Cc: stable@vger.kernel.org
Signed-off-by: Steve French <stfrench@microsoft.com>

cifs: Fix smb2_set_path_size()

If cifs_get_writable_path() finds a writable file, smb2_compound_op()
must use that file's FID and not the COMPOUND_FID.

Cc: stable@vger.kernel.org
Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>

tracing: Make tracepoint lockdep check actually test something

A while ago where the trace events had the following:

   rcu_read_lock_sched_notrace();
   rcu_dereference_sched(...);
   rcu_read_unlock_sched_notrace();

If the tracepoint is enabled, it could trigger RCU issues if called in
the wrong place. And this warning was only triggered if lockdep was
enabled. If the tracepoint was never enabled with lockdep, the bug would
not be caught. To handle this, the above sequence was done when lockdep
was enabled regardless if the tracepoint was enabled or not (although the
always enabled code really didn't do anything, it would still trigger a
warning).

But a lot has changed since that lockdep code was added. One is, that
sequence no longer triggers any warning. Another is, the tracepoint when
enabled doesn't even do that sequence anymore.

The main check we care about today is whether RCU is "watching" or not.
So if lockdep is enabled, always check if rcu_is_watching() which will
trigger a warning if it is not (tracepoints require RCU to be watching).

Note, that old sequence did add a bit of overhead when lockdep was enabled,
and with the latest kernel updates, would cause the system to slow down
enough to trigger kernel "stalled" warnings.

Link: http://lore.kernel.org/lkml/20140806181801.GA4605@redhat.com
Link: http://lore.kernel.org/lkml/20140807175204.C257CAC5@viggo.jf.intel.com
Link: https://lore.kernel.org/lkml/20230307184645.521db5c9@gandalf.local.home/
Link: https://lore.kernel.org/linux-trace-kernel/20230310172856.77406446@gandalf.local.home
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: "Paul E. McKenney" <paulmck@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Joel Fernandes <joel@joelfernandes.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Paul E. McKenney <paulmck@kernel.org>
Fixes: e6753f23d961 ("tracepoint: Make rcuidle tracepoint callers use SRCU")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>

ACPI: tools: pfrut: Check if the input of level and type is in the right numeric range

The user provides arbitrary non-numeic value to level and type,
which could bring unexpected behavior. In this case the expected
behavior would be to throw an error.

 pfrut -h
usage: pfrut [OPTIONS]
code injection:
-l, --load
-s, --stage
-a, --activate
-u, --update [stage and activate]
-q, --query
-d, --revid
update telemetry:
-G, --getloginfo
-T, --type(0:execution, 1:history)
-L, --level(0, 1, 2, 4)
-R, --read
-D, --revid log

 pfrut -T A
 pfrut -G
log_level:0
log_type:0
log_revid:2
max_data_size:65536
chunk1_size:0
chunk2_size:1530
rollover_cnt:0
reset_cnt:17

Fix this by restricting the input to be in the expected range.

Reported-by: Hariganesh Govindarajulu <hariganesh.govindarajulu@intel.com>
Suggested-by: "Rafael J. Wysocki" <rafael.j.wysocki@intel.com>
Signed-off-by: Chen Yu <yu.c.chen@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

ACPI: PPTT: Fix to avoid sleep in the atomic context when PPTT is absent

Commit 0c80f9e165f8 ("ACPI: PPTT: Leave the table mapped for the runtime usage")
enabled to map PPTT once on the first invocation of acpi_get_pptt() and
never unmapped the same allowing it to be used at runtime with out the
hassle of mapping and unmapping the table. This was needed to fetch LLC
information from the PPTT in the cpuhotplug path which is executed in
the atomic context as the acpi_get_table() might sleep waiting for a
mutex.

However it missed to handle the case when there is no PPTT on the system
which results in acpi_get_pptt() being called from all the secondary
CPUs attempting to fetch the LLC information in the atomic context
without knowing the absence of PPTT resulting in the splat like below:

 | BUG: sleeping function called from invalid context at kernel/locking/semaphore.c:164
 | in_atomic(): 1, irqs_disabled(): 1, non_block: 0, pid: 0, name: swapper/1
 | preempt_count: 1, expected: 0
 | RCU nest depth: 0, expected: 0
 | no locks held by swapper/1/0.
 | irq event stamp: 0
 | hardirqs last  enabled at (0): 0x0
 | hardirqs last disabled at (0): copy_process+0x61c/0x1b40
 | softirqs last  enabled at (0): copy_process+0x61c/0x1b40
 | softirqs last disabled at (0): 0x0
 | CPU: 1 PID: 0 Comm: swapper/1 Not tainted 6.3.0-rc1 #1
 | Call trace:
 |  dump_backtrace+0xac/0x138
 |  show_stack+0x30/0x48
 |  dump_stack_lvl+0x60/0xb0
 |  dump_stack+0x18/0x28
 |  __might_resched+0x160/0x270
 |  __might_sleep+0x58/0xb0
 |  down_timeout+0x34/0x98
 |  acpi_os_wait_semaphore+0x7c/0xc0
 |  acpi_ut_acquire_mutex+0x58/0x108
 |  acpi_get_table+0x40/0xe8
 |  acpi_get_pptt+0x48/0xa0
 |  acpi_get_cache_info+0x38/0x140
 |  init_cache_level+0xf4/0x118
 |  detect_cache_attributes+0x2e4/0x640
 |  update_siblings_masks+0x3c/0x330
 |  store_cpu_topology+0x88/0xf0
 |  secondary_start_kernel+0xd0/0x168
 |  __secondary_switched+0xb8/0xc0

Update acpi_get_pptt() to consider the fact that PPTT is once checked and
is not available on the system and return NULL avoiding any attempts to
fetch PPTT and thereby avoiding any possible sleep waiting for a mutex
in the atomic context.

Fixes: 0c80f9e165f8 ("ACPI: PPTT: Leave the table mapped for the runtime usage")
Reported-by: Aishwarya TCV <aishwarya.tcv@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Tested-by: Pierre Gondois <pierre.gondois@arm.com>
Cc: 6.0+ <stable@vger.kernel.org> # 6.0+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Merge tag 'docs-6.3-fixes' of git://git.lwn.net/linux

Pull documentation fixes from Jonathan Corbet:
 "A handful of fixes and minor documentation updates"

* tag 'docs-6.3-fixes' of git://git.lwn.net/linux:
  docs: vfio: fix header path
  docs: process: typo fix
  docs/mm: hugetlbfs_reserv: fix a reference to a file that doesn't exist
  docs/mm: Physical Memory: fix a reference to a file that doesn't exist
  docs: rebasing-and-merging: Drop wrong statement about git
  docs: programming-language: add Rust programming language section
  docs: programming-language: remove mention of the Intel compiler
  docs: Correct missing "d_" prefix for dentry_operations member d_weak_revalidate
  sched/doc: supplement CPU capacity with RISC-V

pm-graph: sleepgraph: Avoid crashing on binary data in device names

A regression has occurred in the hid-sensor code where a device
name string has not been initialized to 0, and ends up without
a NULL char and is printed with %s. This includes random binary
data in the device name, which makes its way into the ftrace output
and ends up crashing sleepgraph because it expects the ftrace output
to be ASCII only.

For example: "HID-SENSOR-INT-020b?.39.auto" ends up in ftrace instead
of "HID-SENSOR-INT-020b.39.auto". It causes this crash in sleepgraph:

  File "/usr/bin/sleepgraph", line 5579, in executeSuspend
    for line in fp:
  File "/usr/lib/python3.10/codecs.py", line 322, in decode
    (result, consumed) = self._buffer_decode(data, self.errors, final)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xff in position
1568: invalid start byte

The issue is present in 6.3-rc1 and is described in full here:
https://bugzilla.kernel.org/show_bug.cgi?id=217169

A separate fix has been submitted to have this issue repaired, but
it has also exposed a larger bug in sleepgraph, since nothing should
make sleepgraph crash. Sleepgraph needs to be able to handle binary
data showing up in ftrace gracefully.

Modify the ftrace processing code to treat it as potentially binary
and to filter out binary data and leave just the ASCII.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=217169
Fixes: 98c062e82451 ("HID: hid-sensor-custom: Allow more custom iio sensors")
Signed-off-by: Todd Brandt <todd.e.brandt@linux.intel.com>
[ rjw: Subject and changelog edits ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

docs: vfio: fix header path

The text points to a different header file, fix by changing
the path to "uapi".

Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Yi Liu <yi.l.liu@intel.com>
Link: https://lore.kernel.org/r/20230310095857.985814-1-jiri@resnulli.us
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

docs: process: typo fix

In the second paragraph of section "Respond to review comments", there is
a spelling mistake: "aganst" should be "against".

Signed-off-by: Xujun Leng <lengxujun2007@126.com>
Link: https://lore.kernel.org/r/20230312071423.3042-1-lengxujun2007@126.com
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

ALSA: hda/ca0132: fixup buffer overrun at tuning_ctl_set()

tuning_ctl_set() might have buffer overrun at (X) if it didn't break
from loop by matching (A).

static int tuning_ctl_set(...)
{
for (i = 0; i < TUNING_CTLS_COUNT; i++)
(A) if (nid == ca0132_tuning_ctls[i].nid)
break;

snd_hda_power_up(...);
(X) dspio_set_param(..., ca0132_tuning_ctls[i].mid, ...);
snd_hda_power_down(...);                ^

return 1;
}

We will get below error by cppcheck

sound/pci/hda/patch_ca0132.c:4229:2: note: After for loop, i has value 12
 for (i = 0; i < TUNING_CTLS_COUNT; i++)
 ^
sound/pci/hda/patch_ca0132.c:4234:43: note: Array index out of bounds
 dspio_set_param(codec, ca0132_tuning_ctls[i].mid, 0x20,
                                           ^
This patch cares non match case.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/87sfe9eap7.wl-kuninori.morimoto.gx@renesas.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>

ALSA: asihpi: check pao in control_message()

control_message() might be called with pao = NULL.
Here indicates control_message() as sample.

(B) static void control_message(struct hpi_adapter_obj *pao, ...)
{                                                   ^^^
struct hpi_hw_obj *phw = pao->priv;
...                      ^^^
}

(A) void _HPI_6205(struct hpi_adapter_obj *pao, ...)
{                                      ^^^
...
case HPI_OBJ_CONTROL:
(B) control_message(pao, phm, phr);
break;          ^^^
...
}

void HPI_6205(...)
{
...
(A) _HPI_6205(NULL, phm, phr);
...       ^^^^
}

Therefore, We will get too many warning via cppcheck, like below

sound/pci/asihpi/hpi6205.c:238:27: warning: Possible null pointer dereference: pao [nullPointer]
 struct hpi_hw_obj *phw = pao->priv;
                          ^
sound/pci/asihpi/hpi6205.c:433:13: note: Calling function '_HPI_6205', 1st argument 'NULL' value is 0
  _HPI_6205(NULL, phm, phr);
            ^
sound/pci/asihpi/hpi6205.c:401:20: note: Calling function 'control_message', 1st argument 'pao' value is 0
   control_message(pao, phm, phr);
                   ^
Set phr->error like many functions doing, and don't call _HPI_6205()
with NULL.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/87ttypeaqz.wl-kuninori.morimoto.gx@renesas.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>

net: hsr: Don't log netdev_err message on unknown prp dst node

If no frames has been exchanged with a node for HSR_NODE_FORGET_TIME, the
node will be deleted from the node_db list. If a frame is sent to the node
after it is deleted, a netdev_err message for each slave interface is
produced. This should not happen with dan nodes because of supervision
frames, but can happen often with san nodes, which clutters the kernel
log. Since the hsr protocol does not support sans, this is only relevant
for the prp protocol.

Signed-off-by: Kristian Overskeid <koverskeid@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

block: do not reverse request order when flushing plug list

Commit 26fed4ac4eab ("block: flush plug based on hardware and software
queue order") changed flushing of plug list to submit requests one
device at a time. However while doing that it also started using
list_add_tail() instead of list_add() used previously thus effectively
submitting requests in reverse order. Also when forming a rq_list with
remaining requests (in case two or more devices are used), we
effectively reverse the ordering of the plug list for each device we
process. Submitting requests in reverse order has negative impact on
performance for rotational disks (when BFQ is not in use). We observe
10-25% regression in random 4k write throughput, as well as ~20%
regression in MariaDB OLTP benchmark on rotational storage on btrfs
filesystem.

Fix the problem by preserving ordering of the plug list when inserting
requests into the queuelist as well as by appending to requeue_list
instead of prepending to it.

Fixes: 26fed4ac4eab ("block: flush plug based on hardware and software queue order")
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20230313093002.11756-1-jack@suse.cz
Signed-off-by: Jens Axboe <axboe@kernel.dk>

drm/amdgpu/vcn: Disable indirect SRAM on Vangogh broken BIOSes

The VCN firmware loading path enables the indirect SRAM mode if it's
advertised as supported. We might have some cases of FW issues that
prevents this mode to working properly though, ending-up in a failed
probe. An example below, observed in the Steam Deck:

[...]
[drm] failed to load ucode VCN0_RAM(0x3A)
[drm] psp gfx command LOAD_IP_FW(0x6) failed and response status is (0xFFFF0000)
amdgpu 0000:04:00.0: [drm:amdgpu_ring_test_helper [amdgpu]] *ERROR* ring vcn_dec_0 test failed (-110)
[drm:amdgpu_device_init.cold [amdgpu]] *ERROR* hw_init of IP block <vcn_v3_0> failed -110
amdgpu 0000:04:00.0: amdgpu: amdgpu_device_ip_init failed
amdgpu 0000:04:00.0: amdgpu: Fatal error during GPU init
[...]

Disabling the VCN block circumvents this, but it's a very invasive
workaround that turns off the entire feature. So, let's add a quirk
on VCN loading that checks for known problematic BIOSes on Vangogh,
so we can proactively disable the indirect SRAM mode and allow the
HW proper probe and VCN IP block to work fine.

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/2385
Fixes: 82132ecc5432 ("drm/amdgpu: enable Vangogh VCN indirect sram mode")
Cc: stable@vger.kernel.org
Cc: James Zhu <James.Zhu@amd.com>
Cc: Leo Liu <leo.liu@amd.com>
Signed-off-by: Guilherme G. Piccoli <gpiccoli@igalia.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/nv: fix codec array for SR_IOV

Copy paste error.

Fixes: 384334120b66 ("drm/amdgpu/nv: don't expose AV1 if VCN0 is harvested")
Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Link: https://bugzilla.openanolis.cn/show_bug.cgi?id=4454
Cc: Jiapeng Chong <jiapeng.chong@linux.alibaba.com>
Acked-by: Guchun Chen <guchun.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Write to correct dirty_rect

When FB_DAMAGE_CLIPS are provided in a non-MPO scenario, the loop does
not use the counter i. This causes the fill_dc_dity_rect() to always
fill dirty_rects[0], causing graphical artifacts when a damage clip
aware DRM client sends more than 1 damage clip.

Instead, use the flip_addrs->dirty_rect_count which is incremented by
fill_dc_dirty_rect() on a successful fill.

Fixes: 30ebe41582d1 ("drm/amd/display: add FB_DAMAGE_CLIPS support")
Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/2453
Signed-off-by: Benjamin Cheng <ben@bcheng.me>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.1.x

drm/amdgpu: move poll enabled/disable into non DC path

Some amd asics having reliable hotplug support don't call
drm_kms_helper_poll_init in driver init sequence. However,
due to the unified suspend/resume path for all asics, because
the output_poll_work->func is not set for these asics, a warning
arrives when suspending.

[   90.656049]  <TASK>
[   90.656050]  ? console_unlock+0x4d/0x100
[   90.656053]  ? __irq_work_queue_local+0x27/0x60
[   90.656056]  ? irq_work_queue+0x2b/0x50
[   90.656057]  ? __wake_up_klogd+0x40/0x60
[   90.656059]  __cancel_work_timer+0xed/0x180
[   90.656061]  drm_kms_helper_poll_disable.cold+0x1f/0x2c [drm_kms_helper]
[   90.656072]  amdgpu_device_suspend+0x81/0x170 [amdgpu]
[   90.656180]  amdgpu_pmops_runtime_suspend+0xb5/0x1b0 [amdgpu]
[   90.656269]  pci_pm_runtime_suspend+0x61/0x1b0

drm_kms_helper_poll_enable/disable is valid when poll_init is called in
amdgpu code, which is only used in non DC path. So move such codes into
non-DC path code to get rid of such warnings.

v1: introduce use_kms_poll flag in amdgpu as the poll stuff check
v2: use dc_enabled as the flag to simply code
v3: move code into non DC path instead of relying on any flag

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/2411
Fixes: a4e771729a51 ("drm/probe_helper: sort out poll_running vs poll_enabled")
Reported-by: Bert Karwatzki <spasswolf@web.de>
Suggested-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Suggested-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Guchun Chen <guchun.chen@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

drm/amd/display: Fix HDCP failing to enable after suspend

[Why]
On resume some displays are not ready for HDCP, so they will fail if we
start the hdcp authentintication too soon.

Add a delay so that the displays can be ready before we start.

NOTE: Previoulsy this delay was set to 3 seconds but it was causing
issues with compliance, 2 seconds should enough for compliance and the
s3 resume case.

[How]
Change the Delay to 2 seconds.

Reviewed-by: Aurabindo Pillai <Aurabindo.Pillai@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: fix potential kgd_mem UAFs

kgd_mem pointers returned by kfd_process_device_translate_handle are
only guaranteed to be valid while p->mutex is held. As soon as the mutex
is unlocked, another thread can free the BO.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/vcn: custom video info caps for sriov

for sriov, we added a new flag to indicate av1 support,
this will override the original caps info.

Signed-off-by: Jane Jian <Jane.Jian@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pm: Fix sienna cichlid incorrect OD volage after resume

Always setup overdrive tables after resume. Preserve only some
user-defined settings in user_overdrive_table if they're set.

Copy restored user_overdrive_table into od_table to get correct
values.

On cold boot, BTC was triggered and GfxVfCurve was calibrated. We
got VfCurve settings (a). On resuming back, BTC will be triggered
again and GfxVfCurve will be recalibrated. VfCurve settings (b)
got may be different from those of cold boot.  So if we reuse
those VfCurve settings (a) got on cold boot on suspend, we can
run into discrepencies.

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1897
Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/2276
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Błażej Szczygieł <mumei6102@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

drm/amd/pm: bump SMU 13.0.4 driver_if header version

Align the SMU driver interface version with PMFW to
suppress the version mismatch message on driver loading.

Signed-off-by: Tim Huang <tim.huang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 6.1.x

drm/amdkfd: fix a potential double free in pqm_create_queue

Set *q to NULL on errors, otherwise pqm_create_queue would free it
again.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Get prange->offset after svm_range_vram_node_new

During miration to vram prange->offset is valid after vram buffer is located,
either use old one or allocate a new one. Move svm_range_vram_node_new before
migrate for each vma to get valid prange->offset.

v2: squash in warning fix

Fixes: b4ee9606378b ("drm/amdkfd: Fix BO offset for multi-VMA page migration")
Signed-off-by: Xiaogang Chen <Xiaogang.Chen@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdkfd: Fix BO offset for multi-VMA page migration

svm_migrate_ram_to_vram migrates a prange from sys ram to vram. The prange may
cross multiple vma. Need remember current dst vram offset in the TTM resource for
each migration.

v2: squash in warning fix (Alex)

Signed-off-by: Xiaogang Chen <Xiaogang.Chen@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

x86/PVH: obtain VGA console info in Dom0

A new platform-op was added to Xen to allow obtaining the same VGA
console information PV Dom0 is handed. Invoke the new function and have
the output data processed by xen_init_vga().

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Juergen Gross <jgross@suse.com>
Link: https://lore.kernel.org/r/8f315e92-7bda-c124-71cc-478ab9c5e610@suse.com
Signed-off-by: Juergen Gross <jgross@suse.com>

KVM: selftests: Sync KVM exit reasons in selftests

Add missing KVM_EXIT_* reasons in KVM selftests from
include/uapi/linux/kvm.h

Signed-off-by: Vipin Sharma <vipinsh@google.com>
Message-Id: <20230204014547.583711-5-vipinsh@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: selftests: Add macro to generate KVM exit reason strings

Add and use a macro to generate the KVM exit reason strings array
instead of relying on developers to correctly copy+paste+edit each
string.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230204014547.583711-4-vipinsh@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: selftests: Print expected and actual exit reason in KVM exit reason assert

Print what KVM exit reason a test was expecting and what it actually
got int TEST_ASSERT_KVM_EXIT_REASON().

Signed-off-by: Vipin Sharma <vipinsh@google.com>
Message-Id: <20230204014547.583711-3-vipinsh@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: selftests: Make vCPU exit reason test assertion common

Make TEST_ASSERT_KVM_EXIT_REASON() macro and replace all exit reason
test assert statements with it.

No functional changes intended.

Signed-off-by: Vipin Sharma <vipinsh@google.com>
Reviewed-by: David Matlack <dmatlack@google.com>
Message-Id: <20230204014547.583711-2-vipinsh@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: selftests: Add EVTCHNOP_send slow path test to xen_shinfo_test

When kvm_xen_evtchn_send() takes the slow path because the shinfo GPC
needs to be revalidated, it used to violate the SRCU vs. kvm->lock
locking rules and potentially cause a deadlock.

Now that lockdep is learning to catch such things, make sure that code
path is exercised by the selftest.

Link: https://lore.kernel.org/all/20230113124606.10221-2-dwmw2@infradead.org
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230204024151.1373296-5-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: selftests: Use enum for test numbers in xen_shinfo_test

The xen_shinfo_test started off with very few iterations, and the numbers
we used in GUEST_SYNC() were precisely mapped to the RUNSTATE_xxx values
anyway to start with.

It has since grown quite a few more tests, and it's kind of awful to be
handling them all as bare numbers. Especially when I want to add a new
test in the middle. Define an enum for the test stages, and use it both
in the guest code and the host switch statement.

No functional change, if I can count to 24.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230204024151.1373296-4-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: selftests: Add helpers to make Xen-style VMCALL/VMMCALL hypercalls

Add wrappers to do hypercalls using VMCALL/VMMCALL and Xen's register ABI
(as opposed to full Xen-style hypercalls through a hypervisor provided
page).  Using the common helpers dedups a pile of code, and uses the
native hypercall instruction when running on AMD.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230204024151.1373296-3-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: selftests: Move the guts of kvm_hypercall() to a separate macro

Extract the guts of kvm_hypercall() to a macro so that Xen hypercalls,
which have a different register ABI, can reuse the VMCALL vs. VMMCALL
logic.

No functional change intended.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230204024151.1373296-2-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: SVM: WARN if GATag generation drops VM or vCPU ID information

WARN if generating a GATag given a VM ID and vCPU ID doesn't yield the
same IDs when pulling the IDs back out of the tag.  Don't bother adding
error handling to callers, this is very much a paranoid sanity check as
KVM fully controls the VM ID and is supposed to reject too-big vCPU IDs.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Reviewed-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Tested-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Message-Id: <20230207002156.521736-4-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: SVM: Modify AVIC GATag to support max number of 512 vCPUs

Define AVIC_VCPU_ID_MASK based on AVIC_PHYSICAL_MAX_INDEX, i.e. the mask
that effectively controls the largest guest physical APIC ID supported by
x2AVIC, instead of hardcoding the number of bits to 8 (and the number of
VM bits to 24).

The AVIC GATag is programmed into the AMD IOMMU IRTE to provide a
reference back to KVM in case the IOMMU cannot inject an interrupt into a
non-running vCPU.  In such a case, the IOMMU notifies software by creating
a GALog entry with the corresponded GATag, and KVM then uses the GATag to
find the correct VM+vCPU to kick.  Dropping bit 8 from the GATag results
in kicking the wrong vCPU when targeting vCPUs with x2APIC ID > 255.

Fixes: 4d1d7942e36a ("KVM: SVM: Introduce logic to (de)activate x2AVIC mode")
Cc: stable@vger.kernel.org
Reported-by: Alejandro Jimenez <alejandro.j.jimenez@oracle.com>
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Co-developed-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Reviewed-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Tested-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Message-Id: <20230207002156.521736-3-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: SVM: Fix a benign off-by-one bug in AVIC physical table mask

Define the "physical table max index mask" as bits 8:0, not 9:0.  x2AVIC
currently supports a max of 512 entries, i.e. the max index is 511, and
the inputs to GENMASK_ULL() are inclusive.  The bug is benign as bit 9 is
reserved and never set by KVM, i.e. KVM is just clearing bits that are
guaranteed to be zero.

Note, as of this writing, APM "Rev. 3.39-October 2022" incorrectly states
that bits 11:8 are reserved in Table B-1. VMCB Layout, Control Area.  I.e.
that table wasn't updated when x2AVIC support was added.

Opportunistically fix the comment for the max AVIC ID to align with the
code, and clean up comment formatting too.

Fixes: 4d1d7942e36a ("KVM: SVM: Introduce logic to (de)activate x2AVIC mode")
Cc: stable@vger.kernel.org
Cc: Alejandro Jimenez <alejandro.j.jimenez@oracle.com>
Cc: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Reviewed-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Tested-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Message-Id: <20230207002156.521736-2-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

selftests: KVM: skip hugetlb tests if huge pages are not available

Right now, if KVM memory stress tests are run with hugetlb sources but hugetlb is
not available (either in the kernel or because /proc/sys/vm/nr_hugepages is 0)
the test will fail with a memory allocation error.

This makes it impossible to add tests that default to hugetlb-backed memory,
because on a machine with a default configuration they will fail.  Therefore,
check HugePages_Total as well and, if zero, direct the user to enable hugepages
in procfs.  Furthermore, return KSFT_SKIP whenever hugetlb is not available.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: VMX: Use tabs instead of spaces for indentation

Code indentation should use tabs where possible and miss a '*'.

Signed-off-by: Rong Tao <rongtao@cestc.cn>
Message-Id: <tencent_A492CB3F9592578451154442830EA1B02C07@qq.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: VMX: Fix indentation coding style issue

Code indentation should use tabs where possible.

Signed-off-by: Rong Tao <rongtao@cestc.cn>
Message-Id: <tencent_31E6ACADCB6915E157CF5113C41803212107@qq.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: nVMX: remove unnecessary #ifdef

nested_vmx_check_controls() has already run by the time KVM checks host state,
so the "host address space size" exit control can only be set on x86-64 hosts.
Simplify the condition at the cost of adding some dead code to 32-bit kernels.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

KVM: nVMX: add missing consistency checks for CR0 and CR4

The effective values of the guest CR0 and CR4 registers may differ from
those included in the VMCS12.  In particular, disabling EPT forces
CR4.PAE=1 and disabling unrestricted guest mode forces CR0.PG=CR0.PE=1.

Therefore, checks on these bits cannot be delegated to the processor
and must be performed by KVM.

Reported-by: Reima ISHII <ishiir@g.ecc.u-tokyo.ac.jp>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Merge tag 'kvmarm-fixes-6.3-1' of git://git./linux/kernel/git/kvmarm/kvmarm into HEAD

KVM/arm64 fixes for 6.3, part #1

A single patch to address a rather annoying bug w.r.t. guest timer
offsetting. Effectively the synchronization of timer offsets between
vCPUs was broken, leading to inconsistent timer reads within the VM.

powerpc/pseries: RTAS work area requires GENERIC_ALLOCATOR

The RTAS work area allocator uses code that is built by
GENERIC_ALLOCATOR, so the PSERIES Kconfig should select the
required Kconfig symbol to fix multiple build errors.

powerpc64-linux-ld: arch/powerpc/platforms/pseries/rtas-work-area.o: in function `.rtas_work_area_allocator_init':
rtas-work-area.c:(.init.text+0x288): undefined reference to `.gen_pool_create'
powerpc64-linux-ld: rtas-work-area.c:(.init.text+0x2dc): undefined reference to `.gen_pool_set_algo'
powerpc64-linux-ld: rtas-work-area.c:(.init.text+0x310): undefined reference to `.gen_pool_add_owner'
powerpc64-linux-ld: rtas-work-area.c:(.init.text+0x43c): undefined reference to `.gen_pool_destroy'
powerpc64-linux-ld: arch/powerpc/platforms/pseries/rtas-work-area.o:(.toc+0x0): undefined reference to `gen_pool_first_fit_order_align'
powerpc64-linux-ld: arch/powerpc/platforms/pseries/rtas-work-area.o: in function `.__rtas_work_area_alloc':
rtas-work-area.c:(.ref.text+0x14c): undefined reference to `.gen_pool_alloc_algo_owner'
powerpc64-linux-ld: rtas-work-area.c:(.ref.text+0x238): undefined reference to `.gen_pool_alloc_algo_owner'
powerpc64-linux-ld: arch/powerpc/platforms/pseries/rtas-work-area.o: in function `.rtas_work_area_free':
rtas-work-area.c:(.ref.text+0x44c): undefined reference to `.gen_pool_free_owner'

Fixes: 43033bc62d34 ("powerpc/pseries: add RTAS work area allocator")
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Reviewed-by: Nathan Lynch <nathanl@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20230223070116.660-2-rdunlap@infradead.org

net: ipa: fix a surprising number of bad offsets

A recent commit eliminated a hack that adjusted the offset used for
many GSI registers.  It became possible because we now specify all
GSI register offsets explicitly for every version of IPA.

Unfortunately, a large number of register offsets were *not* updated
as they should have been in that commit.  For IPA v4.5+, the offset
for every GSI register *except* the two inter-EE interrupt masking
registers were supposed to have been reduced by 0xd000.

Tested-by: Luca Weiss <luca.weiss@fairphone.com>
Tested-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org> # SM8350-HDK
Fixes: 59b12b1d27f3 ("net: ipa: kill gsi->virt_raw")
Signed-off-by: Alex Elder <elder@linaro.org>
Link: https://lore.kernel.org/r/20230310193709.1477102-1-elder@linaro.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: dsa: mt7530: set PLL frequency and trgmii only when trgmii is used

As my testing on the MCM MT7530 switch on MT7621 SoC shows, setting the PLL
frequency does not affect MII modes other than trgmii on port 5 and port 6.
So the assumption is that the operation here called "setting the PLL
frequency" actually sets the frequency of the TRGMII TX clock.

Make it so that it and the rest of the trgmii setup run only when the
trgmii mode is used.

Tested rgmii and trgmii modes of port 6 on MCM MT7530 on MT7621AT Unielec
U7621-06 and standalone MT7530 on MT7623NI Bananapi BPI-R2.

Fixes: b8f126a8d543 ("net-next: dsa: add dsa support for Mediatek MT7530 switch")
Tested-by: Arınç ÜNAL <arinc.unal@arinc9.com>
Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com>
Link: https://lore.kernel.org/r/20230310073338.5836-2-arinc.unal@arinc9.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: dsa: mt7530: remove now incorrect comment regarding port 5

Remove now incorrect comment regarding port 5 as GMAC5. This is supposed to
be supported since commit 38f790a80560 ("net: dsa: mt7530: Add support for
port 5") under mt7530_setup_port5().

Fixes: 38f790a80560 ("net: dsa: mt7530: Add support for port 5")
Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com>
Link: https://lore.kernel.org/r/20230310073338.5836-1-arinc.unal@arinc9.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>