platform/kernel/linux-starfive.git
2 years agoRevert "swiotlb: panic if nslabs is too small"
Yu Zhao [Wed, 31 Aug 2022 06:38:18 +0000 (00:38 -0600)]
Revert "swiotlb: panic if nslabs is too small"

This reverts commit 0bf28fc40d89b1a3e00d1b79473bad4e9ca20ad1.

Reasons:
  1. new panic()s shouldn't be added [1].
  2. It does no "cleanup" but breaks MIPS [2].

v2: properly solved the conflict [3] with
commit 20347fca71a38 ("swiotlb: split up the global swiotlb lock")
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
[1] https://lore.kernel.org/r/CAHk-=wit-DmhMfQErY29JSPjFgebx_Ld+pnerc4J2Ag990WwAA@mail.gmail.com/
[2] https://lore.kernel.org/r/20220820012031.1285979-1-yuzhao@google.com/
[3] https://lore.kernel.org/r/202208310701.LKr1WDCh-lkp@intel.com/

Fixes: 0bf28fc40d89b ("swiotlb: panic if nslabs is too small")
Signed-off-by: Yu Zhao <yuzhao@google.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
2 years agoRDMA/irdma: Report RNR NAK generation in device caps
Sindhu-Devale [Tue, 6 Sep 2022 22:32:44 +0000 (17:32 -0500)]
RDMA/irdma: Report RNR NAK generation in device caps

Report RNR NAK generation when device capabilities are queried

Fixes: b48c24c2d710 ("RDMA/irdma: Implement device supported verb APIs")
Signed-off-by: Sindhu-Devale <sindhu.devale@intel.com>
Signed-off-by: Shiraz Saleem <shiraz.saleem@intel.com>
Link: https://lore.kernel.org/r/20220906223244.1119-6-shiraz.saleem@intel.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
2 years agoRDMA/irdma: Use s/g array in post send only when its valid
Sindhu-Devale [Tue, 6 Sep 2022 22:32:43 +0000 (17:32 -0500)]
RDMA/irdma: Use s/g array in post send only when its valid

Send with invalidate verb call can pass in an
uninitialized s/g array with 0 sge's which is
filled into irdma WQE and causes a HW asynchronous
event.

Fix this by using the s/g array in irdma post send
only when its valid.

Fixes: 551c46e ("RDMA/irdma: Add user/kernel shared libraries")
Signed-off-by: Sindhu-Devale <sindhu.devale@intel.com>
Signed-off-by: Shiraz Saleem <shiraz.saleem@intel.com>
Link: https://lore.kernel.org/r/20220906223244.1119-5-shiraz.saleem@intel.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
2 years agoRDMA/irdma: Return correct WC error for bind operation failure
Sindhu-Devale [Tue, 6 Sep 2022 22:32:42 +0000 (17:32 -0500)]
RDMA/irdma: Return correct WC error for bind operation failure

When a QP and a MR on a local host are in different PDs, the HW generates
an asynchronous event (AE). The same AE is generated when a QP and a MW
are in different PDs during a bind operation. Return the more appropriate
IBV_WC_MW_BIND_ERR for the latter case by checking the OP type from the
CQE in error.

Fixes: 551c46edc769 ("RDMA/irdma: Add user/kernel shared libraries")
Signed-off-by: Sindhu-Devale <sindhu.devale@intel.com>
Signed-off-by: Shiraz Saleem <shiraz.saleem@intel.com>
Link: https://lore.kernel.org/r/20220906223244.1119-4-shiraz.saleem@intel.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
2 years agoRDMA/irdma: Return error on MR deregister CQP failure
Sindhu-Devale [Tue, 6 Sep 2022 22:32:41 +0000 (17:32 -0500)]
RDMA/irdma: Return error on MR deregister CQP failure

The MR deregister CQP can fail if an MW is bound to it.
Return an appropriate error for this case.

Fixes: b48c24c2d710 ("RDMA/irdma: Implement device supported verb APIs")
Signed-off-by: Sindhu-Devale <sindhu.devale@intel.com>
Signed-off-by: Shiraz Saleem <shiraz.saleem@intel.com>
Link: https://lore.kernel.org/r/20220906223244.1119-3-shiraz.saleem@intel.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
2 years agoRDMA/irdma: Report the correct max cqes from query device
Sindhu-Devale [Tue, 6 Sep 2022 22:32:40 +0000 (17:32 -0500)]
RDMA/irdma: Report the correct max cqes from query device

Report the correct max cqes available to an application taking
into account a reserved entry to detect overflow.

Fixes: b48c24c2d710 ("RDMA/irdma: Implement device supported verb APIs")
Signed-off-by: Sindhu-Devale <sindhu.devale@intel.com>
Signed-off-by: Shiraz Saleem <shiraz.saleem@intel.com>
Link: https://lore.kernel.org/r/20220906223244.1119-2-shiraz.saleem@intel.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
2 years agoefi/x86: libstub: remove unused variable
chen zhang [Wed, 7 Sep 2022 02:32:55 +0000 (10:32 +0800)]
efi/x86: libstub: remove unused variable

The variable "has_system_memory" is unused in function
‘adjust_memory_range_protection’, remove it.

Signed-off-by: chen zhang <chenzhang@kylinos.cn>
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
2 years agonvme: requeue aen after firmware activation
Keith Busch [Thu, 1 Sep 2022 15:30:39 +0000 (08:30 -0700)]
nvme: requeue aen after firmware activation

The driver prevents async event work while handling a processing paused
event, but someone needs to restart it after the controller returns to a
live state.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=216400
Signed-off-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>
2 years agonvmet: fix mar and mor off-by-one errors
Dennis Maisenbacher [Tue, 6 Sep 2022 07:39:28 +0000 (09:39 +0200)]
nvmet: fix mar and mor off-by-one errors

Maximum Active Resources (MAR) and Maximum Open Resources (MOR) are 0's
based vales where a value of 0xffffffff indicates that there is no limit.

Decrement the values that are returned by bdev_max_open_zones and
bdev_max_active_zones as the block layer helpers are not 0's based.
A 0 returned by the block layer helpers indicates no limit, thus convert
it to 0xffffffff (U32_MAX).

Fixes: aaf2e048af27 ("nvmet: add ZBD over ZNS backend support")
Suggested-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Dennis Maisenbacher <dennis.maisenbacher@wdc.com>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
2 years agoALSA: emu10k1: Fix out of bounds access in snd_emu10k1_pcm_channel_alloc()
Tasos Sahanidis [Wed, 7 Sep 2022 01:18:00 +0000 (04:18 +0300)]
ALSA: emu10k1: Fix out of bounds access in snd_emu10k1_pcm_channel_alloc()

The voice allocator sometimes begins allocating from near the end of the
array and then wraps around, however snd_emu10k1_pcm_channel_alloc()
accesses the newly allocated voices as if it never wrapped around.

This results in out of bounds access if the first voice has a high enough
index so that first_voice + requested_voice_count > NUM_G (64).
The more voices are requested, the more likely it is for this to occur.

This was initially discovered using PipeWire, however it can be reproduced
by calling aplay multiple times with 16 channels:
aplay -r 48000 -D plughw:CARD=Live,DEV=3 -c 16 /dev/zero

UBSAN: array-index-out-of-bounds in sound/pci/emu10k1/emupcm.c:127:40
index 65 is out of range for type 'snd_emu10k1_voice [64]'
CPU: 1 PID: 31977 Comm: aplay Tainted: G        W IOE      6.0.0-rc2-emu10k1+ #7
Hardware name: ASUSTEK COMPUTER INC P5W DH Deluxe/P5W DH Deluxe, BIOS 3002    07/22/2010
Call Trace:
<TASK>
dump_stack_lvl+0x49/0x63
dump_stack+0x10/0x16
ubsan_epilogue+0x9/0x3f
__ubsan_handle_out_of_bounds.cold+0x44/0x49
snd_emu10k1_playback_hw_params+0x3bc/0x420 [snd_emu10k1]
snd_pcm_hw_params+0x29f/0x600 [snd_pcm]
snd_pcm_common_ioctl+0x188/0x1410 [snd_pcm]
? exit_to_user_mode_prepare+0x35/0x170
? do_syscall_64+0x69/0x90
? syscall_exit_to_user_mode+0x26/0x50
? do_syscall_64+0x69/0x90
? exit_to_user_mode_prepare+0x35/0x170
snd_pcm_ioctl+0x27/0x40 [snd_pcm]
__x64_sys_ioctl+0x95/0xd0
do_syscall_64+0x5c/0x90
? do_syscall_64+0x69/0x90
? do_syscall_64+0x69/0x90
entry_SYSCALL_64_after_hwframe+0x63/0xcd

Signed-off-by: Tasos Sahanidis <tasos@tasossah.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/3707dcab-320a-62ff-63c0-73fc201ef756@tasossah.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2 years agoscsi: mpt3sas: Fix use-after-free warning
Sreekanth Reddy [Tue, 6 Sep 2022 13:49:08 +0000 (19:19 +0530)]
scsi: mpt3sas: Fix use-after-free warning

Fix the following use-after-free warning which is observed during
controller reset:

refcount_t: underflow; use-after-free.
WARNING: CPU: 23 PID: 5399 at lib/refcount.c:28 refcount_warn_saturate+0xa6/0xf0

Link: https://lore.kernel.org/r/20220906134908.1039-2-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2 years agorv/reactor: add __init/__exit annotations to module init/exit funcs
Xiu Jianfeng [Tue, 6 Sep 2022 14:12:10 +0000 (22:12 +0800)]
rv/reactor: add __init/__exit annotations to module init/exit funcs

Add missing __init/__exit annotations to module init/exit funcs.

Link: https://lkml.kernel.org/r/20220906141210.132607-1-xiujianfeng@huawei.com
Fixes: 135b881ea885 ("rv/reactor: Add the printk reactor")
Fixes: e88043c0ac16 ("rv/reactor: Add the panic reactor")
Signed-off-by: Xiu Jianfeng <xiujianfeng@huawei.com>
Acked-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
2 years agotracing: Fix to check event_mutex is held while accessing trigger list
Masami Hiramatsu (Google) [Sun, 4 Sep 2022 04:12:29 +0000 (13:12 +0900)]
tracing: Fix to check event_mutex is held while accessing trigger list

Since the check_user_trigger() is called outside of RCU
read lock, this list_for_each_entry_rcu() caused a suspicious
RCU usage warning.

 # echo hist:keys=pid > events/sched/sched_stat_runtime/trigger
 # cat events/sched/sched_stat_runtime/trigger
[   43.167032]
[   43.167418] =============================
[   43.167992] WARNING: suspicious RCU usage
[   43.168567] 5.19.0-rc5-00029-g19ebe4651abf #59 Not tainted
[   43.169283] -----------------------------
[   43.169863] kernel/trace/trace_events_trigger.c:145 RCU-list traversed in non-reader section!!
...

However, this file->triggers list is safe when it is accessed
under event_mutex is held.
To fix this warning, adds a lockdep_is_held check to the
list_for_each_entry_rcu().

Link: https://lkml.kernel.org/r/166226474977.223837.1992182913048377113.stgit@devnote2
Cc: stable@vger.kernel.org
Fixes: 7491e2c44278 ("tracing: Add a probe that attaches to trace events")
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
2 years agotracing: hold caller_addr to hardirq_{enable,disable}_ip
Yipeng Zou [Thu, 1 Sep 2022 10:45:14 +0000 (18:45 +0800)]
tracing: hold caller_addr to hardirq_{enable,disable}_ip

Currently, The arguments passing to lockdep_hardirqs_{on,off} was fixed
in CALLER_ADDR0.
The function trace_hardirqs_on_caller should have been intended to use
caller_addr to represent the address that caller wants to be traced.

For example, lockdep log in riscv showing the last {enabled,disabled} at
__trace_hardirqs_{on,off} all the time(if called by):
[   57.853175] hardirqs last  enabled at (2519): __trace_hardirqs_on+0xc/0x14
[   57.853848] hardirqs last disabled at (2520): __trace_hardirqs_off+0xc/0x14

After use trace_hardirqs_xx_caller, we can get more effective information:
[   53.781428] hardirqs last  enabled at (2595): restore_all+0xe/0x66
[   53.782185] hardirqs last disabled at (2596): ret_from_exception+0xa/0x10

Link: https://lkml.kernel.org/r/20220901104515.135162-2-zouyipeng@huawei.com
Cc: stable@vger.kernel.org
Fixes: c3bc8fd637a96 ("tracing: Centralize preemptirq tracepoints and unify their usage")
Signed-off-by: Yipeng Zou <zouyipeng@huawei.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
2 years agotracepoint: Allow trace events in modules with TAINT_TEST
Alison Schofield [Mon, 29 Aug 2022 17:10:48 +0000 (10:10 -0700)]
tracepoint: Allow trace events in modules with TAINT_TEST

Commit 2852ca7fba9f ("panic: Taint kernel if tests are run")
introduced a new taint type, TAINT_TEST, to signal that an
in-kernel test module has been loaded.

TAINT_TEST taint type defaults into a 'bad_taint' list for
kernel tracing and blocks the creation of trace events. This
causes a problem for CXL testing where loading the cxl_test
module makes all CXL modules out-of-tree, blocking any trace
events.

Trace events are in development for CXL at the moment and this
issue was found in test with v6.0-rc1.

Link: https://lkml.kernel.org/r/20220829171048.263065-1-alison.schofield@intel.com
Fixes: 2852ca7fba9f7 ("panic: Taint kernel if tests are run")
Reported-by: Ira Weiny <ira.weiny@intel.com>
Suggested-by: Dan Williams <dan.j.williams@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: David Gow <davidgow@google.com>
Signed-off-by: Alison Schofield <alison.schofield@intel.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
2 years agoMAINTAINERS: add scripts/tracing/ to TRACING
Lukas Bulwahn [Thu, 25 Aug 2022 11:59:27 +0000 (13:59 +0200)]
MAINTAINERS: add scripts/tracing/ to TRACING

The files in scripts/tracing/ belong to the TRACING subsystem.

Add a corresponding file entry for TRACING.

Link: https://lkml.kernel.org/r/20220825115927.20598-1-lukas.bulwahn@gmail.com
Signed-off-by: Lukas Bulwahn <lukas.bulwahn@gmail.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
2 years agoMAINTAINERS: Add Runtime Verification (RV) entry
Daniel Bristot de Oliveira [Tue, 23 Aug 2022 15:33:16 +0000 (17:33 +0200)]
MAINTAINERS: Add Runtime Verification (RV) entry

Add a Runtime Verification (RV) entry in the MAINTAINERS file
with Steven Rostedt and myself as maintainers.

Link: https://lkml.kernel.org/r/b24c13553b6947a8da16d884ca464e4233eb8fb7.1661268579.git.bristot@kernel.org
Cc: Joe Perches <joe@perches.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
2 years agorv/monitors: Make monitor's automata definition static
Daniel Bristot de Oliveira [Tue, 23 Aug 2022 15:20:29 +0000 (17:20 +0200)]
rv/monitors: Make monitor's automata definition static

Monitor's automata definition is only used locally, so make
them static for all existing monitors.

Link: https://lore.kernel.org/all/202208210332.gtHXje45-lkp@intel.com
Link: https://lore.kernel.org/all/202208210358.6HH3OrVs-lkp@intel.com
Link: https://lkml.kernel.org/r/a50e27c3738d6ef809f4201857229fed64799234.1661266564.git.bristot@kernel.org
Fixes: ccc319dcb450 ("rv/monitor: Add the wwnr monitor")
Fixes: 8812d21219b9 ("rv/monitor: Add the wip monitor skeleton created by dot2k")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
2 years agoafs: Return -EAGAIN, not -EREMOTEIO, when a file already locked
David Howells [Tue, 6 Sep 2022 21:09:11 +0000 (22:09 +0100)]
afs: Return -EAGAIN, not -EREMOTEIO, when a file already locked

When trying to get a file lock on an AFS file, the server may return
UAEAGAIN to indicate that the lock is already held.  This is currently
translated by the default path to -EREMOTEIO.

Translate it instead to -EAGAIN so that we know we can retry it.

Signed-off-by: David Howells <dhowells@redhat.com>
Reviewed-by: Jeffrey E Altman <jaltman@auristor.com>
cc: Marc Dionne <marc.dionne@auristor.com>
cc: linux-afs@lists.infradead.org
Link: https://lore.kernel.org/r/166075761334.3533338.2591992675160918098.stgit@warthog.procyon.org.uk/
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agoMerge tag 'for-linus' of git://git.armlinux.org.uk/~rmk/linux-arm
Linus Torvalds [Wed, 7 Sep 2022 01:27:05 +0000 (21:27 -0400)]
Merge tag 'for-linus' of git://git.armlinux.org.uk/~rmk/linux-arm

Pull ARM fix from Russell King:
 "Just one fix for now for the AMBA bus code from Isaac Manjarres"

* tag 'for-linus' of git://git.armlinux.org.uk/~rmk/linux-arm:
  ARM: 9229/1: amba: Fix use-after-free in amba_read_periphid()

2 years agoMerge tag 'erofs-for-6.0-rc5-fixes' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 6 Sep 2022 19:19:45 +0000 (15:19 -0400)]
Merge tag 'erofs-for-6.0-rc5-fixes' of git://git./linux/kernel/git/xiang/erofs

Pull erofs fixes from Gao Xiang:

 - Fix return codes in erofs_fscache_{meta_,}read_folio error paths

 - Fix potential wrong pcluster sizes for later non-4K lclusters

 - Fix in-memory pcluster use-after-free on UP platforms

* tag 'erofs-for-6.0-rc5-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs:
  erofs: fix pcluster use-after-free on UP platforms
  erofs: avoid the potentially wrong m_plen for big pcluster
  erofs: fix error return code in erofs_fscache_{meta_,}read_folio

2 years agodrm/i915: consider HAS_FLAT_CCS() in needs_ccs_pages
Matthew Auld [Mon, 5 Sep 2022 10:53:29 +0000 (11:53 +0100)]
drm/i915: consider HAS_FLAT_CCS() in needs_ccs_pages

Just move the HAS_FLAT_CCS() check into needs_ccs_pages. This also then
fixes i915_ttm_memcpy_allowed() which was incorrectly reporting true on
DG1, even though it doesn't have small-BAR or flat-CCS.

References: https://gitlab.freedesktop.org/drm/intel/-/issues/6605
Fixes: efeb3caf4341 ("drm/i915/ttm: disallow CPU fallback mode for ccs pages")
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Nirmoy Das <nirmoy.das@intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220905105329.41455-1-matthew.auld@intel.com
(cherry picked from commit 873fef8833ea794526b7f4179088e565078fe0e8)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
2 years agodrm/i915: Implement WaEdpLinkRateDataReload
Ville Syrjälä [Fri, 2 Sep 2022 07:03:18 +0000 (10:03 +0300)]
drm/i915: Implement WaEdpLinkRateDataReload

A lot of modern laptops use the Parade PS8461E MUX for eDP
switching. The MUX can operate in jitter cleaning mode or
redriver mode, the first one resulting in higher link
quality. The jitter cleaning mode needs to know the link
rate used and the MUX achieves this by snooping the
LINK_BW_SET, LINK_RATE_SELECT and SUPPORTED_LINK_RATES
DPCD accesses.

When the MUX is powered down (seems this can happen whenever
the display is turned off) it loses track of the snooped
link rates so when we do the LINK_RATE_SELECT write it no
longer knowns which link rate we're selecting, and thus it
falls back to the lower quality redriver mode. This results
in unstable high link rates (eg. usually 8.1Gbps link rate
no longer works correctly).

In order to avoid all that let's re-snoop SUPPORTED_LINK_RATES
from the sink at the start of every link training.

Unfortunately we don't have a way to detect the presence of
the MUX. It looks like the set of laptops equipped with this
MUX is fairly large and contains devices from multiple
manufacturers. It may also still be growing with new models.
So a quirk doesn't seem like a very easily maintainable
option, thus we shall attempt to do this unconditionally on
all machines that use LINK_RATE_SELECT. Hopefully this extra
DPCD read doesn't cause issues for any unaffected machine.
If that turns out to be the case we'll need to convert this
into a quirk in the future.

Cc: stable@vger.kernel.org
Cc: Jason A. Donenfeld <Jason@zx2c4.com>
Cc: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/6205
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220902070319.15395-1-ville.syrjala@linux.intel.com
Tested-by: Aaron Ma <aaron.ma@canonical.com>
Tested-by: Jason A. Donenfeld <Jason@zx2c4.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
(cherry picked from commit 25899c590cb5ba9b9f284c6ca8e7e9086793d641)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
2 years agodrm/i915/slpc: Let's fix the PCODE min freq table setup for SLPC
Rodrigo Vivi [Wed, 31 Aug 2022 21:45:38 +0000 (17:45 -0400)]
drm/i915/slpc: Let's fix the PCODE min freq table setup for SLPC

We need to inform PCODE of a desired ring frequencies so PCODE update
the memory frequencies to us. rps->min_freq and rps->max_freq are the
frequencies used in that request. However they were unset when SLPC was
enabled and PCODE never updated the memory freq.

v2 (as Suggested by Ashutosh): if SLPC is in use, let's pick the right
   frequencies from the get_ia_constants instead of the fake init of
   rps' min and max.

v3: don't forget the max <= min return

v4: Move all the freq conversion to intel_rps.c. And the max <= min
    check to where it belongs.

v5: (Ashutosh) Fix old comment s/50 HZ/50 MHz and add a doc explaining
    the "raw format"

Fixes: 7ba79a671568 ("drm/i915/guc/slpc: Gate Host RPS when SLPC is enabled")
Cc: <stable@vger.kernel.org> # v5.15+
Cc: Ashutosh Dixit <ashutosh.dixit@intel.com>
Tested-by: Sushma Venkatesh Reddy <sushma.venkatesh.reddy@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220831214538.143950-1-rodrigo.vivi@intel.com
(cherry picked from commit 018a7bdbb090b9155a6509a0d1a684db4afaa5b1)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
2 years agodrm/i915/bios: Copy the whole MIPI sequence block
Ville Syrjälä [Mon, 29 Aug 2022 13:58:34 +0000 (16:58 +0300)]
drm/i915/bios: Copy the whole MIPI sequence block

Turns out the MIPI sequence block version number and
new block size fields are considered part of the block
header and are not included in the reported new block size
field itself. Bump up the block size appropriately so that
we'll copy over the last five bytes of the block as well.

For this particular machine those last five bytes included
parts of the GPIO op for the backlight on sequence, causing
the backlight no longer to turn back on:

  Sequence 6 - MIPI_SEQ_BACKLIGHT_ON
  Delay: 20000 us
- GPIO index 0, number 0, set 0 (0x00)
+ GPIO index 1, number 70, set 1 (0x01)

Cc: stable@vger.kernel.org
Fixes: e163cfb4c96d ("drm/i915/bios: Make copies of VBT data blocks")
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/6652
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220829135834.8585-1-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
(cherry picked from commit a06289f3f72431f3777af95ea1226b5b0abdc426)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
2 years agobtrfs: fix the max chunk size and stripe length calculation
Qu Wenruo [Thu, 18 Aug 2022 07:06:44 +0000 (15:06 +0800)]
btrfs: fix the max chunk size and stripe length calculation

[BEHAVIOR CHANGE]
Since commit f6fca3917b4d ("btrfs: store chunk size in space-info
struct"), btrfs no longer can create larger data chunks than 1G:

  mkfs.btrfs -f -m raid1 -d raid0 $dev1 $dev2 $dev3 $dev4
  mount $dev1 $mnt

  btrfs balance start --full $mnt
  btrfs balance start --full $mnt
  umount $mnt

  btrfs ins dump-tree -t chunk $dev1 | grep "DATA|RAID0" -C 2

Before that offending commit, what we got is a 4G data chunk:

item 6 key (FIRST_CHUNK_TREE CHUNK_ITEM 9492758528) itemoff 15491 itemsize 176
length 4294967296 owner 2 stripe_len 65536 type DATA|RAID0
io_align 65536 io_width 65536 sector_size 4096
num_stripes 4 sub_stripes 1

Now what we got is only 1G data chunk:

item 6 key (FIRST_CHUNK_TREE CHUNK_ITEM 6271533056) itemoff 15491 itemsize 176
length 1073741824 owner 2 stripe_len 65536 type DATA|RAID0
io_align 65536 io_width 65536 sector_size 4096
num_stripes 4 sub_stripes 1

This will increase the number of data chunks by the number of devices,
not only increase system chunk usage, but also greatly increase mount
time.

Without a proper reason, we should not change the max chunk size.

[CAUSE]
Previously, we set max data chunk size to 10G, while max data stripe
length to 1G.

Commit f6fca3917b4d ("btrfs: store chunk size in space-info struct")
completely ignored the 10G limit, but use 1G max stripe limit instead,
causing above shrink in max data chunk size.

[FIX]
Fix the max data chunk size to 10G, and in decide_stripe_size_regular()
we limit stripe_size to 1G manually.

This should only affect data chunks, as for metadata chunks we always
set the max stripe size the same as max chunk size (256M or 1G
depending on fs size).

Now the same script result the same old result:

item 6 key (FIRST_CHUNK_TREE CHUNK_ITEM 9492758528) itemoff 15491 itemsize 176
length 4294967296 owner 2 stripe_len 65536 type DATA|RAID0
io_align 65536 io_width 65536 sector_size 4096
num_stripes 4 sub_stripes 1

Reported-by: Wang Yugui <wangyugui@e16-tech.com>
Fixes: f6fca3917b4d ("btrfs: store chunk size in space-info struct")
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2 years agoperf c2c: Prevent potential memory leak in c2c_he_zalloc()
Shang XiaoJing [Tue, 6 Sep 2022 03:29:06 +0000 (11:29 +0800)]
perf c2c: Prevent potential memory leak in c2c_he_zalloc()

Free allocated resources when zalloc() fails for members in c2c_he, to
prevent potential memory leak in c2c_he_zalloc().

Signed-off-by: Shang XiaoJing <shangxiaojing@huawei.com>
Reviewed-by: Leo Yan <leo.yan@linaro.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/20220906032906.21395-4-shangxiaojing@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2 years agoperf genelf: Switch deprecated openssl MD5_* functions to new EVP API
Zixuan Tan [Thu, 25 Aug 2022 17:00:58 +0000 (01:00 +0800)]
perf genelf: Switch deprecated openssl MD5_* functions to new EVP API

Switch to the flavored EVP API like in test-libcrypto.c, and remove the
bad gcc #pragma.

Inspired-by: 5b245985a6de5ac1 ("tools build: Switch to new openssl API for test-libcrypto")
Signed-off-by: Zixuan Tan <tanzixuan.me@gmail.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/CABwm_eTnARC1GwMD-JF176k8WXU1Z0+H190mvXn61yr369qt6g@mail.gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2 years agotools/perf: Fix out of bound access to cpu mask array
Athira Rajeev [Mon, 5 Sep 2022 14:19:29 +0000 (19:49 +0530)]
tools/perf: Fix out of bound access to cpu mask array

The cpu mask init code in "record__mmap_cpu_mask_init" function access
"bits" array part of "struct mmap_cpu_mask".  The size of this array is
the value from cpu__max_cpu().cpu.  This array is used to contain the
cpumask value for each cpu. While setting bit for each cpu, it calls
"set_bit" function which access index in "bits" array.

If we provide a command line option to -C which is greater than the
number of CPU's present in the system, the set_bit could access an array
member which is out-of the array size. This is because currently, there
is no boundary check for the CPU. This will result in seg fault:

<<>>
  ./perf record -C 12341234 ls
  Perf can support 2048 CPUs. Consider raising MAX_NR_CPUS
  Segmentation fault (core dumped)
<<>>

Debugging with gdb, points to function flow as below:

<<>>
  set_bit
  record__mmap_cpu_mask_init
  record__init_thread_default_masks
  record__init_thread_masks
  cmd_record
<<>>

Fix this by adding boundary check for the array.

After the patch:

<<>>
./perf record -C 12341234 ls
  Perf can support 2048 CPUs. Consider raising MAX_NR_CPUS
  Failed to initialize parallel data streaming masks
<<>>

With this fix, if -C is given a non-exsiting CPU, perf
record will fail with:

<<>>
  ./perf record -C 50 ls
  Failed to initialize parallel data streaming masks
<<>>

Reported-by: Nageswara R Sastry <rnsastry@linux.ibm.com>
Signed-off-by: Athira Jajeev <atrajeev@linux.vnet.ibm.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Tested-by: Nageswara R Sastry <rnsastry@linux.ibm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: linuxppc-dev@lists.ozlabs.org
Link: https://lore.kernel.org/r/20220905141929.7171-2-atrajeev@linux.vnet.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2 years agoperf affinity: Fix out of bound access to "sched_cpus" mask
Athira Rajeev [Mon, 5 Sep 2022 14:19:28 +0000 (19:49 +0530)]
perf affinity: Fix out of bound access to "sched_cpus" mask

The affinity code in "affinity_set" function access array named
"sched_cpus". The size for this array is allocated in affinity_setup
function which is nothing but value from get_cpu_set_size. This is used
to contain the cpumask value for each cpu.

While setting bit for each cpu, it calls "set_bit" function which access
index in sched_cpus array.  If we provide a command-line option to -C
which is more than the number of CPU's present in the system, the
set_bit could access an array member which is out-of the array size.
This is because currently, there is no boundary check for the CPU.  This
will result in seg fault:

<<>>
   ./perf stat -C 12323431 ls
  Perf can support 2048 CPUs. Consider raising MAX_NR_CPUS
  Segmentation fault (core dumped)
<<>>

Fix this by adding boundary check for the array.

After the fix from powerpc system:

<<>>
  ./perf stat -C 12323431 ls 1>out
  Perf can support 2048 CPUs. Consider raising MAX_NR_CPUS

   Performance counter stats for 'CPU(s) 12323431':

     <not supported> msec cpu-clock
     <not supported>      context-switches
     <not supported>      cpu-migrations
     <not supported>      page-faults
     <not supported>      cycles
     <not supported>      instructions
     <not supported>      branches
     <not supported>      branch-misses

         0.001192373 seconds time elapsed
<<>>

Reported-by: Nageswara R Sastry <rnsastry@linux.ibm.com>
Signed-off-by: Athira Jajeev <atrajeev@linux.vnet.ibm.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Tested-by: Nageswara R Sastry <rnsastry@linux.ibm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: linuxppc-dev@lists.ozlabs.org
Link: https://lore.kernel.org/r/20220905141929.7171-1-atrajeev@linux.vnet.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2 years agonet: usb: qmi_wwan: add Quectel RM520N
jerry.meng [Mon, 5 Sep 2022 01:24:52 +0000 (09:24 +0800)]
net: usb: qmi_wwan: add Quectel RM520N

add support for Quectel RM520N which is based on Qualcomm SDX62 chip.

0x0801: DIAG + NMEA + AT + MODEM + RMNET

T:  Bus=03 Lev=01 Prnt=01 Port=01 Cnt=02 Dev#= 10 Spd=480  MxCh= 0
D:  Ver= 2.10 Cls=00(>ifc ) Sub=00 Prot=00 MxPS=64 #Cfgs=  1
P:  Vendor=2c7c ProdID=0801 Rev= 5.04
S:  Manufacturer=Quectel
S:  Product=RM520N-GL
S:  SerialNumber=384af524
C:* #Ifs= 5 Cfg#= 1 Atr=a0 MxPwr=500mA
I:* If#= 0 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=30 Driver=option
E:  Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 1 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=40 Driver=option
E:  Ad=83(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 2 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=85(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=84(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=03(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 3 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
E:  Ad=87(I) Atr=03(Int.) MxPS=  10 Ivl=32ms
E:  Ad=86(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=04(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 4 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=ff Driver=qmi_wwan
E:  Ad=88(I) Atr=03(Int.) MxPS=   8 Ivl=32ms
E:  Ad=8e(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=0f(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms

Signed-off-by: jerry.meng <jerry-meng@foxmail.com>
Acked-by: Bjørn Mork <bjorn@mork.no>
Link: https://lore.kernel.org/r/tencent_E50CA8A206904897C2D20DDAE90731183C05@qq.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agodrm/ttm: update bulk move object of ghost BO
ZhenGuo Yin [Tue, 6 Sep 2022 08:46:19 +0000 (16:46 +0800)]
drm/ttm: update bulk move object of ghost BO

[Why]
Ghost BO is released with non-empty bulk move object. There is a
warning trace:
WARNING: CPU: 19 PID: 1582 at ttm/ttm_bo.c:366 ttm_bo_release+0x2e1/0x2f0 [amdttm]
Call Trace:
  amddma_resv_reserve_fences+0x10d/0x1f0 [amdkcl]
  amdttm_bo_put+0x28/0x30 [amdttm]
  amdttm_bo_move_accel_cleanup+0x126/0x200 [amdttm]
  amdgpu_bo_move+0x1a8/0x770 [amdgpu]
  ttm_bo_handle_move_mem+0xb0/0x140 [amdttm]
  amdttm_bo_validate+0xbf/0x100 [amdttm]

[How]
The resource of ghost BO should be moved to LRU directly, instead of
using bulk move. The bulk move object of ghost BO should set to NULL
before function ttm_bo_move_to_lru_tail_unlocked.

v2: set bulk move to NULL manually if no resource associated with ghost BO

Fixed: 5b951e487fd6bf5f ("drm/ttm: fix bulk move handling v2")
Signed-off-by: ZhenGuo Yin <zhenguo.yin@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220906084619.2545456-1-zhenguo.yin@amd.com
2 years agonet: dsa: qca8k: fix NULL pointer dereference for of_device_get_match_data
Christian Marangi [Sun, 4 Sep 2022 21:53:19 +0000 (23:53 +0200)]
net: dsa: qca8k: fix NULL pointer dereference for of_device_get_match_data

of_device_get_match_data is called on priv->dev before priv->dev is
actually set. Move of_device_get_match_data after priv->dev is correctly
set to fix this kernel panic.

Fixes: 3bb0844e7bcd ("net: dsa: qca8k: cache match data to speed up access")
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Reviewed-by: Vladimir Oltean <olteanv@gmail.com>
Link: https://lore.kernel.org/r/20220904215319.13070-1-ansuelsmth@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agotcp: fix early ETIMEDOUT after spurious non-SACK RTO
Neal Cardwell [Sat, 3 Sep 2022 12:10:23 +0000 (08:10 -0400)]
tcp: fix early ETIMEDOUT after spurious non-SACK RTO

Fix a bug reported and analyzed by Nagaraj Arankal, where the handling
of a spurious non-SACK RTO could cause a connection to fail to clear
retrans_stamp, causing a later RTO to very prematurely time out the
connection with ETIMEDOUT.

Here is the buggy scenario, expanding upon Nagaraj Arankal's excellent
report:

(*1) Send one data packet on a non-SACK connection

(*2) Because no ACK packet is received, the packet is retransmitted
     and we enter CA_Loss; but this retransmission is spurious.

(*3) The ACK for the original data is received. The transmitted packet
     is acknowledged.  The TCP timestamp is before the retrans_stamp,
     so tcp_may_undo() returns true, and tcp_try_undo_loss() returns
     true without changing state to Open (because tcp_is_sack() is
     false), and tcp_process_loss() returns without calling
     tcp_try_undo_recovery().  Normally after undoing a CA_Loss
     episode, tcp_fastretrans_alert() would see that the connection
     has returned to CA_Open and fall through and call
     tcp_try_to_open(), which would set retrans_stamp to 0.  However,
     for non-SACK connections we hold the connection in CA_Loss, so do
     not fall through to call tcp_try_to_open() and do not set
     retrans_stamp to 0. So retrans_stamp is (erroneously) still
     non-zero.

     At this point the first "retransmission event" has passed and
     been recovered from. Any future retransmission is a completely
     new "event". However, retrans_stamp is erroneously still
     set. (And we are still in CA_Loss, which is correct.)

(*4) After 16 minutes (to correspond with tcp_retries2=15), a new data
     packet is sent. Note: No data is transmitted between (*3) and
     (*4) and we disabled keep alives.

     The socket's timeout SHOULD be calculated from this point in
     time, but instead it's calculated from the prior "event" 16
     minutes ago (step (*2)).

(*5) Because no ACK packet is received, the packet is retransmitted.

(*6) At the time of the 2nd retransmission, the socket returns
     ETIMEDOUT, prematurely, because retrans_stamp is (erroneously)
     too far in the past (set at the time of (*2)).

This commit fixes this bug by ensuring that we reuse in
tcp_try_undo_loss() the same careful logic for non-SACK connections
that we have in tcp_try_undo_recovery(). To avoid duplicating logic,
we factor out that logic into a new
tcp_is_non_sack_preventing_reopen() helper and call that helper from
both undo functions.

Fixes: da34ac7626b5 ("tcp: only undo on partial ACKs in CA_Loss")
Reported-by: Nagaraj Arankal <nagaraj.p.arankal@hpe.com>
Link: https://lore.kernel.org/all/SJ0PR84MB1847BE6C24D274C46A1B9B0EB27A9@SJ0PR84MB1847.NAMPRD84.PROD.OUTLOOK.COM/
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20220903121023.866900-1-ncardwell.kernel@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoALSA: hda: Once again fix regression of page allocations with IOMMU
Takashi Iwai [Tue, 6 Sep 2022 09:03:19 +0000 (11:03 +0200)]
ALSA: hda: Once again fix regression of page allocations with IOMMU

The last fix for trying to recover the regression on AMD platforms,
unfortunately, leaded to yet another regression: it turned out that
IOMMUs don't like the usage of raw page allocations.

This is yet another attempt for addressing the log saga; at this time,
we re-use the existing buffer allocation mechanism with SG-pages
although we require only single pages.  The SG buffer allocation
itself was confirmed to work for stream buffers, so it's relatively
easy to adapt for other places.

The only problem is: although the HD-audio code is accessing the
address directly via dmab->address field, SG-pages don't set up it.
For the ease of adaption, we now set up the dmab->addr field from the
address of the first page as default, so that it can run with the
HD-audio driver code as-is without the excessive call of
snd_sgbuf_get_addr() multiple times; that's the only change in the
memalloc helper side.  The rest is nothing but a flip of the dma_type
field in the HD-audio side.

Fixes: a8d302a0b770 ("ALSA: memalloc: Revive x86-specific WC page allocations again")
Reported-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/CABXGCsO+kB2t5QyHY-rUe76npr1m0-5JOtt8g8SiHUo34ur7Ww@mail.gmail.com
Link: https://bugzilla.kernel.org/show_bug.cgi?id=216112
Link: https://bugzilla.kernel.org/show_bug.cgi?id=216363
Link: https://lore.kernel.org/r/20220906090319.23358-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2 years agoarm64/bti: Disable in kernel BTI when cross section thunks are broken
Mark Brown [Mon, 5 Sep 2022 14:22:55 +0000 (15:22 +0100)]
arm64/bti: Disable in kernel BTI when cross section thunks are broken

GCC does not insert a `bti c` instruction at the beginning of a function
when it believes that all callers reach the function through a direct
branch[1]. Unfortunately the logic it uses to determine this is not
sufficiently robust, for example not taking account of functions being
placed in different sections which may be loaded separately, so we may
still see thunks being generated to these functions. If that happens,
the first instruction in the callee function will result in a Branch
Target Exception due to the missing landing pad.

While this has currently only been observed in the case of modules
having their main code loaded sufficiently far from their init section
to require thunks it could potentially happen for other cases so the
safest thing is to disable BTI for the kernel when building with an
affected toolchain.

[1]: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106671

Reported-by: D Scott Phillips <scott@os.amperecomputing.com>
[Bits of the commit message are lifted from his report & workaround]
Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20220905142255.591990-1-broonie@kernel.org
Cc: <stable@vger.kernel.org> # v5.10+
Signed-off-by: Will Deacon <will@kernel.org>
2 years agoALSA: usb-audio: Fix an out-of-bounds bug in __snd_usb_parse_audio_interface()
Dongxiang Ke [Tue, 6 Sep 2022 02:49:28 +0000 (10:49 +0800)]
ALSA: usb-audio: Fix an out-of-bounds bug in __snd_usb_parse_audio_interface()

There may be a bad USB audio device with a USB ID of (0x04fa, 0x4201) and
the number of it's interfaces less than 4, an out-of-bounds read bug occurs
when parsing the interface descriptor for this device.

Fix this by checking the number of interfaces.

Signed-off-by: Dongxiang Ke <kdx.glider@gmail.com>
Link: https://lore.kernel.org/r/20220906024928.10951-1-kdx.glider@gmail.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2 years agoALSA: hda/tegra: Align BDL entry to 4KB boundary
Mohan Kumar [Mon, 5 Sep 2022 17:24:20 +0000 (22:54 +0530)]
ALSA: hda/tegra: Align BDL entry to 4KB boundary

AZA HW may send a burst read/write request crossing 4K memory boundary.
The 4KB boundary is not guaranteed by Tegra HDA HW. Make SW change to
include the flag AZX_DCAPS_4K_BDLE_BOUNDARY to align BDLE to 4K
boundary.

Signed-off-by: Mohan Kumar <mkumard@nvidia.com>
Link: https://lore.kernel.org/r/20220905172420.3801-1-mkumard@nvidia.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2 years agonvme-tcp: fix regression that causes sporadic requests to time out
Sagi Grimberg [Mon, 5 Sep 2022 15:07:06 +0000 (18:07 +0300)]
nvme-tcp: fix regression that causes sporadic requests to time out

When we queue requests, we strive to batch as much as possible and also
signal the network stack that more data is about to be sent over a socket
with MSG_SENDPAGE_NOTLAST. This flag looks at the pending requests queued
as well as queue->more_requests that is derived from the block layer
last-in-batch indication.

We set more_request=true when we flush the request directly from
.queue_rq submission context (in nvme_tcp_send_all), however this is
wrongly assuming that no other requests may be queued during the
execution of nvme_tcp_send_all.

Due to this, a race condition may happen where:

 1. request X is queued as !last-in-batch
 2. request X submission context calls nvme_tcp_send_all directly
 3. nvme_tcp_send_all is preempted and schedules to a different cpu
 4. request Y is queued as last-in-batch
 5. nvme_tcp_send_all context sends request X+Y, however signals for
    both MSG_SENDPAGE_NOTLAST because queue->more_requests=true.

==> none of the requests is pushed down to the wire as the network
stack is waiting for more data, both requests timeout.

To fix this, we eliminate queue->more_requests and only rely on
the queue req_list and send_list to be not-empty.

Fixes: 122e5b9f3d37 ("nvme-tcp: optimize network stack with setting msg flags according to batch size")
Reported-by: Jonathan Nicklin <jnicklin@blockbridge.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Tested-by: Jonathan Nicklin <jnicklin@blockbridge.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
2 years agonvme-tcp: fix UAF when detecting digest errors
Sagi Grimberg [Mon, 5 Sep 2022 10:54:17 +0000 (13:54 +0300)]
nvme-tcp: fix UAF when detecting digest errors

We should also bail from the io_work loop when we set rd_enabled to true,
so we don't attempt to read data from the socket when the TCP stream is
already out-of-sync or corrupted.

Fixes: 3f2304f8c6d6 ("nvme-tcp: add NVMe over TCP host driver")
Reported-by: Daniel Wagner <dwagner@suse.de>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Daniel Wagner <dwagner@suse.de>
Signed-off-by: Christoph Hellwig <hch@lst.de>
2 years agohwmon: (tps23861) fix byte order in resistance register
Alexandru Gagniuc [Mon, 5 Sep 2022 14:28:04 +0000 (09:28 -0500)]
hwmon: (tps23861) fix byte order in resistance register

The tps23861 registers are little-endian, and regmap_read_bulk() does
not do byte order conversion. On BE machines, the bytes were swapped,
and the interpretation of the resistance value was incorrect.

To make it work on both big and little-endian machines, use
le16_to_cpu() to convert the resitance register to host byte order.

Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Fixes: fff7b8ab22554 ("hwmon: add Texas Instruments TPS23861 driver")
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20220905142806.110598-1-mr.nuke.me@gmail.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
2 years agoMerge tag 'soc-fixes-6.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Linus Torvalds [Mon, 5 Sep 2022 21:44:48 +0000 (17:44 -0400)]
Merge tag 'soc-fixes-6.0-rc4' of git://git./linux/kernel/git/soc/soc

Pull ARM SoC fixes from Arnd Bergmann:
 "These are the expected fixes for the SoC tree. I have let the patches
  pile up a little too long, so this is bigger than I would have liked.

   - Minor build fixes for Broadcom STB and NXP i.MX8M SoCs as well\ as
     TEE firmware

   - Updates to the MAINTAINERS file for the PolarFire SoC

   - Minor DT fixes for Renesas White Hawk and Arm Versatile and Juno
     platforms

   - A fix for a missing dependnecy in the NXP DPIO driver

   - Broadcom BCA fixes to the newly added devicetree files

   - Multiple fixes for Microchip AT91 based SoCs, dealing with
     self-refresh timings and regulator settings in DT

   - Several DT fixes for NXP i.MX platforms, dealing with incorrect
     GPIO settings, extraneous nodes, and a wrong clock setting"

* tag 'soc-fixes-6.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc: (45 commits)
  soc: fsl: select FSL_GUTS driver for DPIO
  ARM: dts: at91: sama5d2_icp: don't keep vdd_other enabled all the time
  ARM: dts: at91: sama5d27_wlsom1: don't keep ldo2 enabled all the time
  ARM: dts: at91: sama7g5ek: specify proper regulator output ranges
  ARM: dts: at91: sama5d2_icp: specify proper regulator output ranges
  ARM: dts: at91: sama5d27_wlsom1: specify proper regulator output ranges
  ARM: at91: pm: fix DDR recalibration when resuming from backup and self-refresh
  ARM: at91: pm: fix self-refresh for sama7g5
  soc: brcmstb: pm-arm: Fix refcount leak and __iomem leak bugs
  ARM: configs: at91: remove CONFIG_MICROCHIP_PIT64B
  ARM: ixp4xx: fix typos in comments
  arm64: dts: renesas: r8a779g0: Fix HSCIF0 interrupt number
  tee: fix compiler warning in tee_shm_register()
  arm64: dts: freescale: verdin-imx8mp: fix atmel_mxt_ts reset polarity
  arm64: dts: freescale: verdin-imx8mm: fix atmel_mxt_ts reset polarity
  arm64: dts: imx8mp: Fix I2C5 GPIO assignment on i.MX8M Plus DHCOM
  arm64: dts: imx8mm-venice-gw7901: fix port/phy validation
  arm64: dts: verdin-imx8mm: add otg2 pd to usbphy
  soc: imx: gpcv2: Assert reset before ungating clock
  arm64: dts: ls1028a-qds-65bb: don't use in-band autoneg for 2500base-x
  ...

2 years agoio_uring/notif: Remove the unused function io_notif_complete()
Jiapeng Chong [Mon, 5 Sep 2022 02:04:36 +0000 (10:04 +0800)]
io_uring/notif: Remove the unused function io_notif_complete()

The function io_notif_complete() is defined in the notif.c file, but not
called elsewhere, so delete this unused function.

io_uring/notif.c:24:20: warning: unused function 'io_notif_complete' [-Wunused-function].

Link: https://bugzilla.openanolis.cn/show_bug.cgi?id=2047
Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Signed-off-by: Jiapeng Chong <jiapeng.chong@linux.alibaba.com>
Reviewed-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/20220905020436.51894-1-jiapeng.chong@linux.alibaba.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoerofs: fix pcluster use-after-free on UP platforms
Gao Xiang [Fri, 2 Sep 2022 04:57:10 +0000 (12:57 +0800)]
erofs: fix pcluster use-after-free on UP platforms

During stress testing with CONFIG_SMP disabled, KASAN reports as below:

==================================================================
BUG: KASAN: use-after-free in __mutex_lock+0xe5/0xc30
Read of size 8 at addr ffff8881094223f8 by task stress/7789

CPU: 0 PID: 7789 Comm: stress Not tainted 6.0.0-rc1-00002-g0d53d2e882f9 #3
Hardware name: Red Hat KVM, BIOS 0.5.1 01/01/2011
Call Trace:
 <TASK>
..
 __mutex_lock+0xe5/0xc30
..
 z_erofs_do_read_page+0x8ce/0x1560
..
 z_erofs_readahead+0x31c/0x580
..
Freed by task 7787
 kasan_save_stack+0x1e/0x40
 kasan_set_track+0x20/0x30
 kasan_set_free_info+0x20/0x40
 __kasan_slab_free+0x10c/0x190
 kmem_cache_free+0xed/0x380
 rcu_core+0x3d5/0xc90
 __do_softirq+0x12d/0x389

Last potentially related work creation:
 kasan_save_stack+0x1e/0x40
 __kasan_record_aux_stack+0x97/0xb0
 call_rcu+0x3d/0x3f0
 erofs_shrink_workstation+0x11f/0x210
 erofs_shrink_scan+0xdc/0x170
 shrink_slab.constprop.0+0x296/0x530
 drop_slab+0x1c/0x70
 drop_caches_sysctl_handler+0x70/0x80
 proc_sys_call_handler+0x20a/0x2f0
 vfs_write+0x555/0x6c0
 ksys_write+0xbe/0x160
 do_syscall_64+0x3b/0x90

The root cause is that erofs_workgroup_unfreeze() doesn't reset to
orig_val thus it causes a race that the pcluster reuses unexpectedly
before freeing.

Since UP platforms are quite rare now, such path becomes unnecessary.
Let's drop such specific-designed path directly instead.

Fixes: 73f5c66df3e2 ("staging: erofs: fix `erofs_workgroup_{try_to_freeze, unfreeze}'")
Reviewed-by: Yue Hu <huyue2@coolpad.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Link: https://lore.kernel.org/r/20220902045710.109530-1-hsiangkao@linux.alibaba.com
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
2 years agoerofs: avoid the potentially wrong m_plen for big pcluster
Yue Hu [Fri, 12 Aug 2022 06:01:50 +0000 (14:01 +0800)]
erofs: avoid the potentially wrong m_plen for big pcluster

Actually, 'compressedlcs' stores compressed block count rather than
lcluster count. Therefore, the number of bits for shifting the count
should be 'LOG_BLOCK_SIZE' rather than 'lclusterbits' although current
lcluster size is 4K.

The value of 'm_plen' will be wrong once we enable the non 4K-sized
lcluster.

Signed-off-by: Yue Hu <huyue2@coolpad.com>
Reviewed-by: Gao Xiang <hsiangkao@linux.alibaba.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Link: https://lore.kernel.org/r/20220812060150.8510-1-huyue2@coolpad.com
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
2 years agoerofs: fix error return code in erofs_fscache_{meta_,}read_folio
Sun Ke [Mon, 15 Aug 2022 03:48:29 +0000 (11:48 +0800)]
erofs: fix error return code in erofs_fscache_{meta_,}read_folio

If erofs_fscache_alloc_request fail and then goto out, it will return 0.
it should return a negative error code instead of 0.

Fixes: d435d53228dd ("erofs: change to use asynchronous io for fscache readpage/readahead")
Signed-off-by: Sun Ke <sunke32@huawei.com>
Reviewed-by: Jingbo Xu <jefflexu@linux.alibaba.com>
Reviewed-by: Gao Xiang <hsiangkao@linux.alibaba.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Link: https://lore.kernel.org/r/20220815034829.3940803-1-sunke32@huawei.com
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
2 years agoasm-generic: Conditionally enable do_softirq_own_stack() via Kconfig.
Sebastian Andrzej Siewior [Thu, 25 Aug 2022 08:25:05 +0000 (10:25 +0200)]
asm-generic: Conditionally enable do_softirq_own_stack() via Kconfig.

Remove the CONFIG_PREEMPT_RT symbol from the ifdef around
do_softirq_own_stack() and move it to Kconfig instead.

Enable softirq stacks based on SOFTIRQ_ON_OWN_STACK which depends on
HAVE_SOFTIRQ_ON_OWN_STACK and its default value is set to !PREEMPT_RT.
This ensures that softirq stacks are not used on PREEMPT_RT and avoids
a 'select' statement on an option which has a 'depends' statement.

Link: https://lore.kernel.org/YvN5E%2FPrHfUhggr7@linutronix.de
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
2 years agoMerge tag 'for-net-2022-09-02' of git://git.kernel.org/pub/scm/linux/kernel/git/bluet...
David S. Miller [Mon, 5 Sep 2022 13:43:18 +0000 (14:43 +0100)]
Merge tag 'for-net-2022-09-02' of git://git./linux/kernel/git/bluetooth/bluetooth

Luiz Augusto von Dentz says:

====================
bluetooth pull request for net:

 - Fix regression preventing ACL packet transmission
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agobtrfs: zoned: fix mounting with conventional zones
Johannes Thumshirn [Mon, 5 Sep 2022 12:38:24 +0000 (05:38 -0700)]
btrfs: zoned: fix mounting with conventional zones

Since commit 6a921de58992 ("btrfs: zoned: introduce
space_info->active_total_bytes"), we're only counting the bytes of a
block group on an active zone as usable for metadata writes. But on a
SMR drive, we don't have active zones and short circuit some of the
logic.

This leads to an error on mount, because we cannot reserve space for
metadata writes.

Fix this by also setting the BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE bit in the
block-group's runtime flag if the zone is a conventional zone.

Fixes: 6a921de58992 ("btrfs: zoned: introduce space_info->active_total_bytes")
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2 years agobtrfs: zoned: set pseudo max append zone limit in zone emulation mode
Shin'ichiro Kawasaki [Fri, 26 Aug 2022 07:42:15 +0000 (16:42 +0900)]
btrfs: zoned: set pseudo max append zone limit in zone emulation mode

The commit 7d7672bc5d10 ("btrfs: convert count_max_extents() to use
fs_info->max_extent_size") introduced a division by
fs_info->max_extent_size. This max_extent_size is initialized with max
zone append limit size of the device btrfs runs on. However, in zone
emulation mode, the device is not zoned then its zone append limit is
zero. This resulted in zero value of fs_info->max_extent_size and caused
zero division error.

Fix the error by setting non-zero pseudo value to max append zone limit
in zone emulation mode. Set the pseudo value based on max_segments as
suggested in the commit c2ae7b772ef4 ("btrfs: zoned: revive
max_zone_append_bytes").

Fixes: 7d7672bc5d10 ("btrfs: convert count_max_extents() to use fs_info->max_extent_size")
CC: stable@vger.kernel.org # 5.12+
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Naohiro Aota <naohiro.aota@wdc.com>
Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2 years agobtrfs: zoned: fix API misuse of zone finish waiting
Naohiro Aota [Wed, 31 Aug 2022 04:55:48 +0000 (13:55 +0900)]
btrfs: zoned: fix API misuse of zone finish waiting

The commit 2ce543f47843 ("btrfs: zoned: wait until zone is finished when
allocation didn't progress") implemented a zone finish waiting mechanism
to the write path of zoned mode. However, using
wait_var_event()/wake_up_all() on fs_info->zone_finish_wait is wrong and
wait_var_event() just hangs because no one ever wakes it up once it goes
into sleep.

Instead, we can simply use wait_on_bit_io() and clear_and_wake_up_bit()
on fs_info->flags with a proper barrier installed.

Fixes: 2ce543f47843 ("btrfs: zoned: wait until zone is finished when allocation didn't progress")
CC: stable@vger.kernel.org # 5.16+
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2 years agostmmac: intel: Simplify intel_eth_pci_remove()
Christophe JAILLET [Fri, 2 Sep 2022 16:26:56 +0000 (18:26 +0200)]
stmmac: intel: Simplify intel_eth_pci_remove()

There is no point to call pcim_iounmap_regions() in the remove function,
this frees a managed resource that would be release by the framework
anyway.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: mvpp2: debugfs: fix memory leak when using debugfs_lookup()
Greg Kroah-Hartman [Fri, 2 Sep 2022 13:41:11 +0000 (15:41 +0200)]
net: mvpp2: debugfs: fix memory leak when using debugfs_lookup()

When calling debugfs_lookup() the result must have dput() called on it,
otherwise the memory will leak over time.  Fix this up to be much
simpler logic and only create the root debugfs directory once when the
driver is first accessed.  That resolves the memory leak and makes
things more obvious as to what the intent is.

Cc: Marcin Wojtas <mw@semihalf.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Jakub Kicinski <kuba@kernel.org>
Cc: Paolo Abeni <pabeni@redhat.com>
Cc: netdev@vger.kernel.org
Cc: stable <stable@kernel.org>
Fixes: 21da57a23125 ("net: mvpp2: add a debugfs interface for the Header Parser")
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoALSA: hda/sigmatel: Fix unused variable warning for beep power change
Takashi Iwai [Mon, 5 Sep 2022 13:06:30 +0000 (15:06 +0200)]
ALSA: hda/sigmatel: Fix unused variable warning for beep power change

The newly added stac_check_power_status() caused a compile warning
when CONFIG_SND_HDA_INPUT_BEEP is disabled.  Fix it.

Fixes: 414d38ba8710 ("ALSA: hda/sigmatel: Keep power up while beep is enabled")
Reported-by: kernel test robot <lkp@intel.com>
Link: https://lore.kernel.org/r/20220905130630.2845-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2 years agonvmet: fix a use-after-free
Bart Van Assche [Fri, 12 Aug 2022 21:03:17 +0000 (14:03 -0700)]
nvmet: fix a use-after-free

Fix the following use-after-free complaint triggered by blktests nvme/004:

BUG: KASAN: user-memory-access in blk_mq_complete_request_remote+0xac/0x350
Read of size 4 at addr 0000607bd1835943 by task kworker/13:1/460
Workqueue: nvmet-wq nvme_loop_execute_work [nvme_loop]
Call Trace:
 show_stack+0x52/0x58
 dump_stack_lvl+0x49/0x5e
 print_report.cold+0x36/0x1e2
 kasan_report+0xb9/0xf0
 __asan_load4+0x6b/0x80
 blk_mq_complete_request_remote+0xac/0x350
 nvme_loop_queue_response+0x1df/0x275 [nvme_loop]
 __nvmet_req_complete+0x132/0x4f0 [nvmet]
 nvmet_req_complete+0x15/0x40 [nvmet]
 nvmet_execute_io_connect+0x18a/0x1f0 [nvmet]
 nvme_loop_execute_work+0x20/0x30 [nvme_loop]
 process_one_work+0x56e/0xa70
 worker_thread+0x2d1/0x640
 kthread+0x183/0x1c0
 ret_from_fork+0x1f/0x30

Cc: stable@vger.kernel.org
Fixes: a07b4970f464 ("nvmet: add a generic NVMe target")
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>
2 years agoALSA: pcm: oss: Fix race at SNDCTL_DSP_SYNC
Takashi Iwai [Mon, 5 Sep 2022 06:07:14 +0000 (08:07 +0200)]
ALSA: pcm: oss: Fix race at SNDCTL_DSP_SYNC

There is a small race window at snd_pcm_oss_sync() that is called from
OSS PCM SNDCTL_DSP_SYNC ioctl; namely the function calls
snd_pcm_oss_make_ready() at first, then takes the params_lock mutex
for the rest.  When the stream is set up again by another thread
between them, it leads to inconsistency, and may result in unexpected
results such as NULL dereference of OSS buffer as a fuzzer spotted
recently.

The fix is simply to cover snd_pcm_oss_make_ready() call into the same
params_lock mutex with snd_pcm_oss_make_ready_locked() variant.

Reported-and-tested-by: butt3rflyh4ck <butterflyhuangxx@gmail.com>
Reviewed-by: Jaroslav Kysela <perex@perex.cz>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/CAFcO6XN7JDM4xSXGhtusQfS2mSBcx50VJKwQpCq=WeLt57aaZA@mail.gmail.com
Link: https://lore.kernel.org/r/20220905060714.22549-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2 years agoMAINTAINERS: Update maintainers of HiSilicon RoCE
Wenpeng Liang [Mon, 5 Sep 2022 02:38:15 +0000 (10:38 +0800)]
MAINTAINERS: Update maintainers of HiSilicon RoCE

Weihang has moved to work in other technical areas, and Haoyue will
maintain this module instead of him.

Signed-off-by: Wenpeng Liang <liangwenpeng@huawei.com>
Link: https://lore.kernel.org/r/20220905023815.1477684-1-liangwenpeng@huawei.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
2 years agoRDMA/mlx5: Fix UMR cleanup on error flow of driver init
Maor Gottlieb [Mon, 29 Aug 2022 09:02:29 +0000 (12:02 +0300)]
RDMA/mlx5: Fix UMR cleanup on error flow of driver init

The cited commit removed from the cleanup flow of umr the checks
if the resources were created. This could lead to null-ptr-deref
in case that we had failure in mlx5_ib_stage_ib_reg_init stage.

Fix it by adding new state to the umr that can say if the resources
were created or not and check it in the umr cleanup flow before
destroying the resources.

Fixes: 04876c12c19e ("RDMA/mlx5: Move init and cleanup of UMR to umr.c")
Reviewed-by: Michael Guralnik <michaelgur@nvidia.com>
Signed-off-by: Maor Gottlieb <maorg@nvidia.com>
Link: https://lore.kernel.org/r/4cfa61386cf202e9ce330e8d228ce3b25a36326e.1661763459.git.leonro@nvidia.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
2 years agoRDMA/mlx5: Set local port to one when accessing counters
Chris Mi [Mon, 29 Aug 2022 09:02:28 +0000 (12:02 +0300)]
RDMA/mlx5: Set local port to one when accessing counters

When accessing Ports Performance Counters Register (PPCNT),
local port must be one if it is Function-Per-Port HCA that
HCA_CAP.num_ports is 1.

The offending patch can change the local port to other values
when accessing PPCNT after enabling switchdev mode. The following
syndrome will be printed:

 # cat /sys/class/infiniband/rdmap4s0f0/ports/2/counters/*
 # dmesg
 mlx5_core 0000:04:00.0: mlx5_cmd_check:756:(pid 12450): ACCESS_REG(0x805) op_mod(0x1) failed, status bad parameter(0x3), syndrome (0x1e5585)

Fix it by setting local port to one for Function-Per-Port HCA.

Fixes: 210b1f78076f ("IB/mlx5: When not in dual port RoCE mode, use provided port as native")
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Chris Mi <cmi@nvidia.com>
Link: https://lore.kernel.org/r/6c5086c295c76211169e58dbd610fb0402360bab.1661763459.git.leonro@nvidia.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
2 years agoRDMA/mlx5: Rely on RoCE fw cap instead of devlink when setting profile
Maher Sanalla [Mon, 29 Aug 2022 09:02:27 +0000 (12:02 +0300)]
RDMA/mlx5: Rely on RoCE fw cap instead of devlink when setting profile

When the RDMA auxiliary driver probes, it sets its profile based on
devlink driverinit value. The latter might not be in sync with FW yet
(In case devlink reload is not performed), thus causing a mismatch
between RDMA driver and FW. This results in the following FW syndrome
when the RDMA driver tries to adjust RoCE state, which fails the probe:

"0xC1F678 | modify_nic_vport_context: roce_en set on a vport that
doesn't support roce"

To prevent this, select the PF profile based on FW RoCE capability
instead of relying on devlink driverinit value.
To provide backward compatibility of the RoCE disable feature, on older
FW's where roce_rw is not set (FW RoCE capability is read-only), keep
the current behavior e.g., rely on devlink driverinit value.

Fixes: fbfa97b4d79f ("net/mlx5: Disable roce at HCA level")
Reviewed-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Michael Guralnik <michaelgur@nvidia.com>
Reviewed-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Maher Sanalla <msanalla@nvidia.com>
Link: https://lore.kernel.org/r/cb34ce9a1df4a24c135cb804db87f7d2418bd6cc.1661763459.git.leonro@nvidia.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
2 years agoIB/core: Fix a nested dead lock as part of ODP flow
Yishai Hadas [Wed, 24 Aug 2022 06:10:36 +0000 (09:10 +0300)]
IB/core: Fix a nested dead lock as part of ODP flow

Fix a nested dead lock as part of ODP flow by using mmput_async().

From the below call trace [1] can see that calling mmput() once we have
the umem_odp->umem_mutex locked as required by
ib_umem_odp_map_dma_and_lock() might trigger in the same task the
exit_mmap()->__mmu_notifier_release()->mlx5_ib_invalidate_range() which
may dead lock when trying to lock the same mutex.

Moving to use mmput_async() will solve the problem as the above
exit_mmap() flow will be called in other task and will be executed once
the lock will be available.

[1]
[64843.077665] task:kworker/u133:2  state:D stack:    0 pid:80906 ppid:
2 flags:0x00004000
[64843.077672] Workqueue: mlx5_ib_page_fault mlx5_ib_eqe_pf_action [mlx5_ib]
[64843.077719] Call Trace:
[64843.077722]  <TASK>
[64843.077724]  __schedule+0x23d/0x590
[64843.077729]  schedule+0x4e/0xb0
[64843.077735]  schedule_preempt_disabled+0xe/0x10
[64843.077740]  __mutex_lock.constprop.0+0x263/0x490
[64843.077747]  __mutex_lock_slowpath+0x13/0x20
[64843.077752]  mutex_lock+0x34/0x40
[64843.077758]  mlx5_ib_invalidate_range+0x48/0x270 [mlx5_ib]
[64843.077808]  __mmu_notifier_release+0x1a4/0x200
[64843.077816]  exit_mmap+0x1bc/0x200
[64843.077822]  ? walk_page_range+0x9c/0x120
[64843.077828]  ? __cond_resched+0x1a/0x50
[64843.077833]  ? mutex_lock+0x13/0x40
[64843.077839]  ? uprobe_clear_state+0xac/0x120
[64843.077860]  mmput+0x5f/0x140
[64843.077867]  ib_umem_odp_map_dma_and_lock+0x21b/0x580 [ib_core]
[64843.077931]  pagefault_real_mr+0x9a/0x140 [mlx5_ib]
[64843.077962]  pagefault_mr+0xb4/0x550 [mlx5_ib]
[64843.077992]  pagefault_single_data_segment.constprop.0+0x2ac/0x560
[mlx5_ib]
[64843.078022]  mlx5_ib_eqe_pf_action+0x528/0x780 [mlx5_ib]
[64843.078051]  process_one_work+0x22b/0x3d0
[64843.078059]  worker_thread+0x53/0x410
[64843.078065]  ? process_one_work+0x3d0/0x3d0
[64843.078073]  kthread+0x12a/0x150
[64843.078079]  ? set_kthread_struct+0x50/0x50
[64843.078085]  ret_from_fork+0x22/0x30
[64843.078093]  </TASK>

Fixes: 36f30e486dce ("IB/core: Improve ODP to use hmm_range_fault()")
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Yishai Hadas <yishaih@nvidia.com>
Link: https://lore.kernel.org/r/74d93541ea533ef7daec6f126deb1072500aeb16.1661251841.git.leonro@nvidia.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
2 years agosched/debug: fix dentry leak in update_sched_domain_debugfs
Greg Kroah-Hartman [Fri, 2 Sep 2022 12:31:07 +0000 (14:31 +0200)]
sched/debug: fix dentry leak in update_sched_domain_debugfs

Kuyo reports that the pattern of using debugfs_remove(debugfs_lookup())
leaks a dentry and with a hotplug stress test, the machine eventually
runs out of memory.

Fix this up by using the newly created debugfs_lookup_and_remove() call
instead which properly handles the dentry reference counting logic.

Cc: Major Chen <major.chen@samsung.com>
Cc: stable <stable@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ben Segall <bsegall@google.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Reported-by: Kuyo Chang <kuyo.chang@mediatek.com>
Tested-by: Kuyo Chang <kuyo.chang@mediatek.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20220902123107.109274-2-gregkh@linuxfoundation.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2 years agodebugfs: add debugfs_lookup_and_remove()
Greg Kroah-Hartman [Fri, 2 Sep 2022 14:59:15 +0000 (16:59 +0200)]
debugfs: add debugfs_lookup_and_remove()

There is a very common pattern of using
debugfs_remove(debufs_lookup(..)) which results in a dentry leak of the
dentry that was looked up.  Instead of having to open-code the correct
pattern of calling dput() on the dentry, create
debugfs_lookup_and_remove() to handle this pattern automatically and
properly without any memory leaks.

Cc: stable <stable@kernel.org>
Reported-by: Kuyo Chang <kuyo.chang@mediatek.com>
Tested-by: Kuyo Chang <kuyo.chang@mediatek.com>
Link: https://lore.kernel.org/r/YxIaQ8cSinDR881k@kroah.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2 years agodriver core: fix driver_set_override() issue with empty strings
Greg Kroah-Hartman [Thu, 1 Sep 2022 16:37:34 +0000 (18:37 +0200)]
driver core: fix driver_set_override() issue with empty strings

Python likes to send an empty string for some sysfs files, including the
driver_override field.  When commit 23d99baf9d72 ("PCI: Use
driver_set_override() instead of open-coding") moved the PCI core to use
the driver core function instead of hand-rolling their own handler, this
showed up as a regression from some userspace tools, like DPDK.

Fix this up by actually looking at the length of the string first
instead of trusting that userspace got it correct.

Fixes: 23d99baf9d72 ("PCI: Use driver_set_override() instead of open-coding")
Cc: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>
Cc: Andy Shevchenko <andy.shevchenko@gmail.com>
Cc: stable <stable@kernel.org>
Reported-by: Stephen Hemminger <stephen@networkplumber.org>
Tested-by: Huisong Li <lihuisong@huawei.com>
Reviewed-by: Stephen Hemminger <stephen@networkplumber.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20220901163734.3583106-1-gregkh@linuxfoundation.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2 years agoipv6: sr: fix out-of-bounds read when setting HMAC data.
David Lebrun [Fri, 2 Sep 2022 09:45:06 +0000 (10:45 +0100)]
ipv6: sr: fix out-of-bounds read when setting HMAC data.

The SRv6 layer allows defining HMAC data that can later be used to sign IPv6
Segment Routing Headers. This configuration is realised via netlink through
four attributes: SEG6_ATTR_HMACKEYID, SEG6_ATTR_SECRET, SEG6_ATTR_SECRETLEN and
SEG6_ATTR_ALGID. Because the SECRETLEN attribute is decoupled from the actual
length of the SECRET attribute, it is possible to provide invalid combinations
(e.g., secret = "", secretlen = 64). This case is not checked in the code and
with an appropriately crafted netlink message, an out-of-bounds read of up
to 64 bytes (max secret length) can occur past the skb end pointer and into
skb_shared_info:

Breakpoint 1, seg6_genl_sethmac (skb=<optimized out>, info=<optimized out>) at net/ipv6/seg6.c:208
208 memcpy(hinfo->secret, secret, slen);
(gdb) bt
 #0  seg6_genl_sethmac (skb=<optimized out>, info=<optimized out>) at net/ipv6/seg6.c:208
 #1  0xffffffff81e012e9 in genl_family_rcv_msg_doit (skb=skb@entry=0xffff88800b1f9f00, nlh=nlh@entry=0xffff88800b1b7600,
    extack=extack@entry=0xffffc90000ba7af0, ops=ops@entry=0xffffc90000ba7a80, hdrlen=4, net=0xffffffff84237580 <init_net>, family=<optimized out>,
    family=<optimized out>) at net/netlink/genetlink.c:731
 #2  0xffffffff81e01435 in genl_family_rcv_msg (extack=0xffffc90000ba7af0, nlh=0xffff88800b1b7600, skb=0xffff88800b1f9f00,
    family=0xffffffff82fef6c0 <seg6_genl_family>) at net/netlink/genetlink.c:775
 #3  genl_rcv_msg (skb=0xffff88800b1f9f00, nlh=0xffff88800b1b7600, extack=0xffffc90000ba7af0) at net/netlink/genetlink.c:792
 #4  0xffffffff81dfffc3 in netlink_rcv_skb (skb=skb@entry=0xffff88800b1f9f00, cb=cb@entry=0xffffffff81e01350 <genl_rcv_msg>)
    at net/netlink/af_netlink.c:2501
 #5  0xffffffff81e00919 in genl_rcv (skb=0xffff88800b1f9f00) at net/netlink/genetlink.c:803
 #6  0xffffffff81dff6ae in netlink_unicast_kernel (ssk=0xffff888010eec800, skb=0xffff88800b1f9f00, sk=0xffff888004aed000)
    at net/netlink/af_netlink.c:1319
 #7  netlink_unicast (ssk=ssk@entry=0xffff888010eec800, skb=skb@entry=0xffff88800b1f9f00, portid=portid@entry=0, nonblock=<optimized out>)
    at net/netlink/af_netlink.c:1345
 #8  0xffffffff81dff9a4 in netlink_sendmsg (sock=<optimized out>, msg=0xffffc90000ba7e48, len=<optimized out>) at net/netlink/af_netlink.c:1921
...
(gdb) p/x ((struct sk_buff *)0xffff88800b1f9f00)->head + ((struct sk_buff *)0xffff88800b1f9f00)->end
$1 = 0xffff88800b1b76c0
(gdb) p/x secret
$2 = 0xffff88800b1b76c0
(gdb) p slen
$3 = 64 '@'

The OOB data can then be read back from userspace by dumping HMAC state. This
commit fixes this by ensuring SECRETLEN cannot exceed the actual length of
SECRET.

Reported-by: Lucas Leong <wmliang.tw@gmail.com>
Tested: verified that EINVAL is correctly returned when secretlen > len(secret)
Fixes: 4f4853dc1c9c1 ("ipv6: sr: implement API to control SR HMAC structure")
Signed-off-by: David Lebrun <dlebrun@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'bonding-fixes'
David S. Miller [Mon, 5 Sep 2022 09:07:05 +0000 (10:07 +0100)]
Merge branch 'bonding-fixes'

Hangbin Liu says:

====================
bonding: fix lladdr finding and confirmation

This patch set fixed 3 issues when setting lladdr as bonding IPv6 target.
Please see each patch for the details.

v2: separate the patch to 3 parts
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agobonding: accept unsolicited NA message
Hangbin Liu [Tue, 30 Aug 2022 09:37:22 +0000 (17:37 +0800)]
bonding: accept unsolicited NA message

The unsolicited NA message with all-nodes multicast dest address should
be valid, as this also means the link could reach the target.

Also rename bond_validate_ns() to bond_validate_na().

Reported-by: LiLiang <liali@redhat.com>
Fixes: 5e1eeef69c0f ("bonding: NS target should accept link local address")
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agobonding: add all node mcast address when slave up
Hangbin Liu [Tue, 30 Aug 2022 09:37:21 +0000 (17:37 +0800)]
bonding: add all node mcast address when slave up

When a link is enslave to bond, it need to set the interface down first.
This makes the slave remove mac multicast address 33:33:00:00:00:01(The
IPv6 multicast address ff02::1 is kept even when the interface down). When
bond set the slave up, ipv6_mc_up() was not called due to commit c2edacf80e15
("bonding / ipv6: no addrconf for slaves separately from master").

This is not an issue before we adding the lladdr target feature for bonding,
as the mac multicast address will be added back when bond interface up and
join group ff02::1.

But after adding lladdr target feature for bonding. When user set a lladdr
target, the unsolicited NA message with all-nodes multicast dest will be
dropped as the slave interface never add 33:33:00:00:00:01 back.

Fix this by calling ipv6_mc_up() to add 33:33:00:00:00:01 back when
the slave interface up.

Reported-by: LiLiang <liali@redhat.com>
Fixes: 5e1eeef69c0f ("bonding: NS target should accept link local address")
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agobonding: use unspecified address if no available link local address
Hangbin Liu [Tue, 30 Aug 2022 09:37:20 +0000 (17:37 +0800)]
bonding: use unspecified address if no available link local address

When ns_ip6_target was set, the ipv6_dev_get_saddr() will be called to get
available source address and send IPv6 neighbor solicit message.

If the target is global address, ipv6_dev_get_saddr() will get any
available src address. But if the target is link local address,
ipv6_dev_get_saddr() will only get available address from our interface,
i.e. the corresponding bond interface.

But before bond interface up, all the address is tentative, while
ipv6_dev_get_saddr() will ignore tentative address. This makes we can't
find available link local src address, then bond_ns_send() will not be
called and no NS message was sent. Finally bond interface will keep in
down state.

Fix this by sending NS with unspecified address if there is no available
source address.

Reported-by: LiLiang <liali@redhat.com>
Fixes: 5e1eeef69c0f ("bonding: NS target should accept link local address")
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoLinux 6.0-rc4
Linus Torvalds [Sun, 4 Sep 2022 20:10:01 +0000 (13:10 -0700)]
Linux 6.0-rc4

2 years agoMerge tag 'powerpc-6.0-4' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc...
Linus Torvalds [Sun, 4 Sep 2022 18:33:22 +0000 (11:33 -0700)]
Merge tag 'powerpc-6.0-4' of git://git./linux/kernel/git/powerpc/linux

Pull powerpc fixes from Michael Ellerman:

 - Fix handling of PCI domains in /proc on 32-bit systems using the
   recently added support for numbering buses from zero for each domain.

 - A fix and a revert for some changes to use READ/WRITE_ONCE() which
   caused problems with KASAN enabled due to sanitisation calls being
   introduced in low-level paths that can't cope with it.

 - Fix build errors on 32-bit caused by the syscall table being
   misaligned sometimes.

 - Two fixes to get IBM Cell native machines booting again, which had
   bit-rotted while my QS22 was temporarily out of action.

 - Fix the papr_scm driver to not assume the order of events returned by
   the hypervisor is stable, and a related compile fix.

Thanks to Aneesh Kumar K.V, Christophe Leroy, Jordan Niethe, Kajol Jain,
Masahiro Yamada, Nathan Chancellor, Pali Rohár, Vaibhav Jain, and Zhouyi
Zhou.

* tag 'powerpc-6.0-4' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  powerpc/papr_scm: Ensure rc is always initialized in papr_scm_pmu_register()
  Revert "powerpc/irq: Don't open code irq_soft_mask helpers"
  powerpc: Fix hard_irq_disable() with sanitizer
  powerpc/rtas: Fix RTAS MSR[HV] handling for Cell
  Revert "powerpc: Remove unused FW_FEATURE_NATIVE references"
  powerpc: align syscall table for ppc32
  powerpc/pci: Enable PCI domains in /proc when PCI bus numbers are not unique
  powerpc/papr_scm: Fix nvdimm event mappings

2 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Sun, 4 Sep 2022 18:27:14 +0000 (11:27 -0700)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull kvm fixes from Paolo Bonzini:
 "s390:

   - PCI interpretation compile fixes

  RISC-V:

   - fix unused variable warnings in vcpu_timer.c

   - move extern sbi_ext declarations to a header

  x86:

   - check validity of argument to KVM_SET_MP_STATE

   - use guest's global_ctrl to completely disable guest PEBS

   - fix a memory leak on memory allocation failure

   - mask off unsupported and unknown bits of IA32_ARCH_CAPABILITIES

   - fix build failure with Clang integrated assembler

   - fix MSR interception

   - always flush TLBs when enabling dirty logging"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: x86: check validity of argument to KVM_SET_MP_STATE
  perf/x86/core: Completely disable guest PEBS via guest's global_ctrl
  KVM: x86: fix memoryleak in kvm_arch_vcpu_create()
  KVM: x86: Mask off unsupported and unknown bits of IA32_ARCH_CAPABILITIES
  KVM: s390: pci: Hook to access KVM lowlevel from VFIO
  riscv: kvm: move extern sbi_ext declarations to a header
  riscv: kvm: vcpu_timer: fix unused variable warnings
  KVM: selftests: Fix ambiguous mov in KVM_ASM_SAFE()
  KVM: selftests: Fix KVM_EXCEPTION_MAGIC build with Clang
  KVM: VMX: Heed the 'msr' argument in msr_write_intercepted()
  kvm: x86: mmu: Always flush TLBs when enabling dirty logging
  kvm: x86: mmu: Drop the need_remote_flush() function

2 years agoMakefile.extrawarn: re-enable -Wformat for clang; take 2
Nick Desaulniers [Thu, 1 Sep 2022 17:59:13 +0000 (10:59 -0700)]
Makefile.extrawarn: re-enable -Wformat for clang; take 2

-Wformat was recently re-enabled for builds with clang, then quickly
re-disabled, due to concerns stemming from the frequency of default
argument promotion related warning instances.

commit 258fafcd0683 ("Makefile.extrawarn: re-enable -Wformat for clang")
commit 21f9c8a13bb2 ("Revert "Makefile.extrawarn: re-enable -Wformat for clang"")

ISO WG14 has ratified N2562 to address default argument promotion
explicitly for printf, as part of the upcoming ISO C2X standard.

The behavior of clang was changed in clang-16 to not warn for the cited
cases in all language modes.

Add a version check, so that users of clang-16 now get the full effect
of -Wformat. For older clang versions, re-enable flags under the
-Wformat group that way users still get some useful checks related to
format strings, without noisy default argument promotion warnings. I
intentionally omitted -Wformat-y2k and -Wformat-security from being
re-enabled, which are also part of -Wformat in clang-16.

Link: https://github.com/ClangBuiltLinux/linux/issues/378
Link: https://github.com/llvm/llvm-project/issues/57102
Link: https://www.open-std.org/jtc1/sc22/wg14/www/docs/n2562.pdf
Suggested-by: Justin Stitt <jstitt007@gmail.com>
Suggested-by: Nathan Chancellor <nathan@kernel.org>
Suggested-by: Youngmin Nam <youngmin.nam@samsung.com>
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Reviewed-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agoMerge tag 'wireless-2022-09-03' of git://git.kernel.org/pub/scm/linux/kernel/git...
David S. Miller [Sun, 4 Sep 2022 10:23:11 +0000 (11:23 +0100)]
Merge tag 'wireless-2022-09-03' of git://git./linux/kernel/git/wireless/wireless

Johannes berg says:

====================
We have a handful of fixes:
 - fix DMA from stack in wilc1000 driver
 - fix crash on chip reset failure in mt7921e
 - fix for the reported warning on aggregation timer expiry
 - check packet lengths in hwsim virtio paths
 - fix compiler warnings/errors with AAD construction by
   using struct_group
 - fix Intel 4965 driver rate scale operation
 - release channel contexts correctly in mac80211 mlme code
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoALSA: hda/sigmatel: Keep power up while beep is enabled
Takashi Iwai [Sun, 4 Sep 2022 07:27:50 +0000 (09:27 +0200)]
ALSA: hda/sigmatel: Keep power up while beep is enabled

It seems that the beep playback doesn't work well on IDT codec devices
when the codec auto-pm is enabled.  Keep the power on while the beep
switch is enabled.

Link: https://bugzilla.suse.com/show_bug.cgi?id=1200544
Link: https://lore.kernel.org/r/20220904072750.26164-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2 years agoRDMA/siw: Pass a pointer to virt_to_page()
Linus Walleij [Fri, 2 Sep 2022 21:59:18 +0000 (23:59 +0200)]
RDMA/siw: Pass a pointer to virt_to_page()

Functions that work on a pointer to virtual memory such as
virt_to_pfn() and users of that function such as
virt_to_page() are supposed to pass a pointer to virtual
memory, ideally a (void *) or other pointer. However since
many architectures implement virt_to_pfn() as a macro,
this function becomes polymorphic and accepts both a
(unsigned long) and a (void *).

If we instead implement a proper virt_to_pfn(void *addr)
function the following happens (occurred on arch/arm):

drivers/infiniband/sw/siw/siw_qp_tx.c:32:23: warning: incompatible
  integer to pointer conversion passing 'dma_addr_t' (aka 'unsigned int')
  to parameter of type 'const void *' [-Wint-conversion]
drivers/infiniband/sw/siw/siw_qp_tx.c:32:37: warning: passing argument
  1 of 'virt_to_pfn' makes pointer from integer without a cast
  [-Wint-conversion]
drivers/infiniband/sw/siw/siw_qp_tx.c:538:36: warning: incompatible
  integer to pointer conversion passing 'unsigned long long'
  to parameter of type 'const void *' [-Wint-conversion]

Fix this with an explicit cast. In one case where the SIW
SGE uses an unaligned u64 we need a double cast modifying the
virtual address (va) to a platform-specific uintptr_t before
casting to a (void *).

Fixes: b9be6f18cf9e ("rdma/siw: transmit path")
Cc: linux-rdma@vger.kernel.org
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20220902215918.603761-1-linus.walleij@linaro.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>
2 years agoMerge tag 'gpio-fixes-for-v6.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 4 Sep 2022 04:27:27 +0000 (21:27 -0700)]
Merge tag 'gpio-fixes-for-v6.0-rc4' of git://git./linux/kernel/git/brgl/linux

Pull gpio fixes from Bartosz Golaszewski:
 "A a set of fixes from the GPIO subsystem.

  Most are small driver fixes except the realtek-otto driver patch which
  is pretty big but addresses a significant flaw that can cause the CPU
  to stay infinitely busy on uncleared ISR on some platforms.

  Summary:
   - MAINTAINERS update
   - fix resource leaks in gpio-mockup and gpio-pxa
   - add missing locking in gpio-pca953x
   - use 32-bit I/O in gpio-realtek-otto
   - make irq_chip structures immutable in four more drivers"

* tag 'gpio-fixes-for-v6.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/brgl/linux:
  gpio: ws16c48: Make irq_chip immutable
  gpio: 104-idio-16: Make irq_chip immutable
  gpio: 104-idi-48: Make irq_chip immutable
  gpio: 104-dio-48e: Make irq_chip immutable
  gpio: realtek-otto: switch to 32-bit I/O
  gpio: pca953x: Add mutex_lock for regcache sync in PM
  gpio: mockup: remove gpio debugfs when remove device
  gpio: pxa: use devres for the clock struct
  MAINTAINERS: rectify entry for XILINX GPIO DRIVER

2 years agogpio: ws16c48: Make irq_chip immutable
William Breathitt Gray [Fri, 2 Sep 2022 17:45:26 +0000 (13:45 -0400)]
gpio: ws16c48: Make irq_chip immutable

Kernel warns about mutable irq_chips:

    "not an immutable chip, please consider fixing!"

Make the struct irq_chip const, flag it as IRQCHIP_IMMUTABLE, add the
new helper functions, and call the appropriate gpiolib functions.

Signed-off-by: William Breathitt Gray <william.gray@linaro.org>
Signed-off-by: Bartosz Golaszewski <brgl@bgdev.pl>
2 years agogpio: 104-idio-16: Make irq_chip immutable
William Breathitt Gray [Fri, 2 Sep 2022 17:45:25 +0000 (13:45 -0400)]
gpio: 104-idio-16: Make irq_chip immutable

Kernel warns about mutable irq_chips:

    "not an immutable chip, please consider fixing!"

Make the struct irq_chip const, flag it as IRQCHIP_IMMUTABLE, add the
new helper functions, and call the appropriate gpiolib functions.

Signed-off-by: William Breathitt Gray <william.gray@linaro.org>
Signed-off-by: Bartosz Golaszewski <brgl@bgdev.pl>
2 years agogpio: 104-idi-48: Make irq_chip immutable
William Breathitt Gray [Fri, 2 Sep 2022 17:45:24 +0000 (13:45 -0400)]
gpio: 104-idi-48: Make irq_chip immutable

Kernel warns about mutable irq_chips:

    "not an immutable chip, please consider fixing!"

Make the struct irq_chip const, flag it as IRQCHIP_IMMUTABLE, add the
new helper functions, and call the appropriate gpiolib functions.

Signed-off-by: William Breathitt Gray <william.gray@linaro.org>
Signed-off-by: Bartosz Golaszewski <brgl@bgdev.pl>
2 years agogpio: 104-dio-48e: Make irq_chip immutable
William Breathitt Gray [Fri, 2 Sep 2022 17:45:23 +0000 (13:45 -0400)]
gpio: 104-dio-48e: Make irq_chip immutable

Kernel warns about mutable irq_chips:

    "not an immutable chip, please consider fixing!"

Make the struct irq_chip const, flag it as IRQCHIP_IMMUTABLE, add the
new helper functions, and call the appropriate gpiolib functions.

Signed-off-by: William Breathitt Gray <william.gray@linaro.org>
Signed-off-by: Bartosz Golaszewski <brgl@bgdev.pl>
2 years agoMerge tag 'for-linus-6.0-rc4-tag' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 3 Sep 2022 20:23:11 +0000 (13:23 -0700)]
Merge tag 'for-linus-6.0-rc4-tag' of git://git./linux/kernel/git/xen/tip

Pull xen fixes from Juergen Gross:

 - a minor fix for the Xen grant driver

 - a small series fixing a recently introduced problem in the Xen
   blkfront/blkback drivers with negotiation of feature usage

* tag 'for-linus-6.0-rc4-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip:
  xen/grants: prevent integer overflow in gnttab_dma_alloc_pages()
  xen-blkfront: Cache feature_persistent value before advertisement
  xen-blkfront: Advertise feature-persistent as user requested
  xen-blkback: Advertise feature-persistent as user requested

2 years agoMerge tag 'loongarch-fixes-6.0-2' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 3 Sep 2022 20:21:01 +0000 (13:21 -0700)]
Merge tag 'loongarch-fixes-6.0-2' of git://git./linux/kernel/git/chenhuacai/linux-loongson

Pull LoongArch fixes from Huacai Chen:
 "Fix several build errors or warnings, cleanup some code, and adjust
  arch_do_signal_or_restart() to adapt generic entry"

* tag 'loongarch-fixes-6.0-2' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson:
  LoongArch: mm: Remove the unneeded result variable
  LoongArch: Fix arch_remove_memory() undefined build error
  LoongArch: Fix section mismatch due to acpi_os_ioremap()
  LoongArch: Improve dump_tlb() output messages
  LoongArch: Adjust arch_do_signal_or_restart() to adapt generic entry
  LoongArch: Avoid orphan input sections

2 years agoMerge tag 's390-6.0-3' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
Linus Torvalds [Sat, 3 Sep 2022 20:17:33 +0000 (13:17 -0700)]
Merge tag 's390-6.0-3' of git://git./linux/kernel/git/s390/linux

Pull s390 fixes from Vasily Gorbik:

 - Update defconfigs

 - Fix linker script to align nospec tables correctly to avoid
   potentially unbootable kernel with some config options

 - Fix alignment check in prepare_hugepage_range() for 2GB hugepages to
   avoid BUG in __unmap_hugepage_range() for unaligned mappings later

 - Remove useless hugepage address alignment in hugetlb fault handling

* tag 's390-6.0-3' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390/hugetlb: fix prepare_hugepage_range() check for 2 GB hugepages
  s390: update defconfigs
  s390: fix nospec table alignments
  s390/mm: remove useless hugepage address alignment

2 years agoMerge tag 'input-for-v6.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor...
Linus Torvalds [Sat, 3 Sep 2022 20:09:46 +0000 (13:09 -0700)]
Merge tag 'input-for-v6.0-rc3' of git://git./linux/kernel/git/dtor/input

Pull input fixes from Dmitry Torokhov:

 - GT1158 ID added to Goodix touchscreen driver

 - Boeder Force Feedback Wheel USB added to iforce joystick driver

 - fixup for iforce driver to avoid hangups

 - fix autoloading of rk805-pwrkey driver.

* tag 'input-for-v6.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: iforce - add support for Boeder Force Feedback Wheel
  Input: iforce - wake up after clearing IFORCE_XMIT_RUNNING flag
  Input: goodix - add compatible string for GT1158
  MAINTAINERS: add include/dt-bindings/input to INPUT DRIVERS
  Input: rk805-pwrkey - fix module autoloading
  Input: goodix - add support for GT1158
  dt-bindings: input: touchscreen: add compatible string for Goodix GT1158

2 years agoMerge tag 'dt-fixes-for-palmer-6.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel...
Palmer Dabbelt [Sat, 3 Sep 2022 19:30:23 +0000 (12:30 -0700)]
Merge tag 'dt-fixes-for-palmer-6.0-rc4' of git://git./linux/kernel/git/conor/linux.git into fixes

Microchip RISC-V devicetree fixes for 6.0-rc4 (or later)

A fix for the warnings introduced in rc3 as part of fixing the console
spam from the L2's isr.

Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
* tag 'dt-fixes-for-palmer-6.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/conor/linux.git:
  riscv: dts: microchip: use an mpfs specific l2 compatible
  dt-bindings: riscv: sifive-l2: add a PolarFire SoC compatible

2 years agoMerge tag 'tty-6.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
Linus Torvalds [Sat, 3 Sep 2022 17:34:02 +0000 (10:34 -0700)]
Merge tag 'tty-6.0-rc4' of git://git./linux/kernel/git/gregkh/tty

Pull tty/serial driver fixes from Greg KH:
 "Here are some small tty/serial/vt driver fixes for 6.0-rc4 that
  resolve a number of reported issues:

   - n_gsm fixups for previous changes that caused problems

   - much-reported serdev crash fix that showed up in 6.0-rc1

   - vt font selection bugfix

   - kerneldoc build warning fixes

   - other tiny serial core fixes

  All of these have been in linux-next for a while with no reported
  problems"

* tag 'tty-6.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
  tty: n_gsm: avoid call of sleeping functions from atomic context
  tty: n_gsm: replace kicktimer with delayed_work
  tty: n_gsm: initialize more members at gsm_alloc_mux()
  tty: n_gsm: add sanity check for gsm->receive in gsm_receive_buf()
  tty: serial: atmel: Preserve previous USART mode if RS485 disabled
  tty: serial: lpuart: disable flow control while waiting for the transmit engine to complete
  tty: Fix lookahead_buf crash with serdev
  serial: fsl_lpuart: RS485 RTS polariy is inverse
  vt: Clear selection before changing the font
  serial: document start_rx member at struct uart_ops

2 years agoMerge tag 'staging-6.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Sat, 3 Sep 2022 17:32:17 +0000 (10:32 -0700)]
Merge tag 'staging-6.0-rc4' of git://git./linux/kernel/git/gregkh/staging

Pull staging driver fixes from Greg KH:
 "Here are three small staging driver fixes for 6.0-rc4 that resolve
  some reported problems and add some a device id:

   - new device id for r8188eu driver

   - use-after-free bugfixes for the rtl8712 driver

   - fix up firmware dependency problem for the r8188eu driver

  All of these have been in linux-next for a while with no reported
  problems"

* tag 'staging-6.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
  staging: rtl8712: fix use after free bugs
  staging: r8188eu: Add Rosewill USB-N150 Nano to device tables
  staging: r8188eu: add firmware dependency

2 years agoblock: don't add partitions if GD_SUPPRESS_PART_SCAN is set
Ming Lei [Tue, 23 Aug 2022 10:38:19 +0000 (18:38 +0800)]
block: don't add partitions if GD_SUPPRESS_PART_SCAN is set

Commit b9684a71fca7 ("block, loop: support partitions without scanning")
adds GD_SUPPRESS_PART_SCAN for replacing part function of
GENHD_FL_NO_PART. But looks blk_add_partitions() is missed, since
loop doesn't want to add partitions if GENHD_FL_NO_PART was set.
And it causes regression on libblockdev (as called from udisks) which
operates with the LO_FLAGS_PARTSCAN.

Fixes the issue by not adding partitions if GD_SUPPRESS_PART_SCAN is
set.

Fixes: b9684a71fca7 ("block, loop: support partitions without scanning")
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220823103819.395776-1-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2 years agoMerge tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 3 Sep 2022 17:27:25 +0000 (10:27 -0700)]
Merge tag 'clk-fixes-for-linus' of git://git./linux/kernel/git/clk/linux

Pull clk fixes from Stephen Boyd:
 "Here's a collection of primarily clk driver fixes, with a couple fixes
  to the core framework.

  We had to revert out a commit that affected boot on some devices that
  have the CLK_OPS_PARENT_ENABLE flag set. It isn't critical to have
  that fix so we'll try again next time.

  Driver side fixes include:

   - Plug an OF-node refcount bug in the TI clk driver

   - Fix the error handling in the raspberry pi firmware get_rate so
     that errors don't look like valid frequencies

   - Avoid going out of bounds in the raspberry pi driver too if the
     video firmware returns something we're not expecting"

* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
  Revert "clk: core: Honor CLK_OPS_PARENT_ENABLE for clk gate ops"
  clk: bcm: rpi: Show clock id limit in error case
  clk: bcm: rpi: Add missing newline
  clk: bcm: rpi: Prevent out-of-bounds access
  clk: bcm: rpi: Fix error handling of raspberrypi_fw_get_rate
  clk: core: Fix runtime PM sequence in clk_core_unprepare()
  clk: core: Honor CLK_OPS_PARENT_ENABLE for clk gate ops
  clk: ti: Fix missing of_node_get() ti_find_clock_provider()

2 years agoMerge tag 'hwmon-for-v6.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/groec...
Linus Torvalds [Sat, 3 Sep 2022 17:24:30 +0000 (10:24 -0700)]
Merge tag 'hwmon-for-v6.0-rc4' of git://git./linux/kernel/git/groeck/linux-staging

Pull hwmon fixes from Guenter Roeck:

 - Fix out of bounds access in gpio-fan driver

 - Fix VOUT margin caching in PMBus core

 - Avoid error message after -EPROBE_DEFER from devm_regulator_register()

* tag 'hwmon-for-v6.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  hwmon: (gpio-fan) Fix array out of bounds access
  hwmon: (pmbus) Fix vout margin caching
  hwmon: (pmbus) Use dev_err_probe() to filter -EPROBE_DEFER error messages

2 years agomm: pagewalk: Fix race between unmap and page walker
Steven Price [Fri, 2 Sep 2022 11:26:12 +0000 (12:26 +0100)]
mm: pagewalk: Fix race between unmap and page walker

The mmap lock protects the page walker from changes to the page tables
during the walk.  However a read lock is insufficient to protect those
areas which don't have a VMA as munmap() detaches the VMAs before
downgrading to a read lock and actually tearing down PTEs/page tables.

For users of walk_page_range() the solution is to simply call pte_hole()
immediately without checking the actual page tables when a VMA is not
present. We now never call __walk_page_range() without a valid vma.

For walk_page_range_novma() the locking requirements are tightened to
require the mmap write lock to be taken, and then walking the pgd
directly with 'no_vma' set.

This in turn means that all page walkers either have a valid vma, or
it's that special 'novma' case for page table debugging.  As a result,
all the odd '(!walk->vma && !walk->no_vma)' tests can be removed.

Fixes: dd2283f2605e ("mm: mmap: zap pages with read mmap_sem in munmap")
Reported-by: Jann Horn <jannh@google.com>
Signed-off-by: Steven Price <steven.price@arm.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Konstantin Khlebnikov <koct9i@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agowifi: use struct_group to copy addresses
Johannes Berg [Mon, 29 Aug 2022 09:46:38 +0000 (11:46 +0200)]
wifi: use struct_group to copy addresses

We sometimes copy all the addresses from the 802.11 header
for the AAD, which may cause complaints from fortify checks.
Use struct_group() to avoid the compiler warnings/errors.

Change-Id: Ic3ea389105e7813b22095b295079eecdabde5045
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2 years agowifi: mac80211_hwsim: check length for virtio packets
Soenke Huster [Fri, 2 Sep 2022 08:19:58 +0000 (10:19 +0200)]
wifi: mac80211_hwsim: check length for virtio packets

An invalid packet with a length shorter than the specified length in the
netlink header can lead to use-after-frees and slab-out-of-bounds in the
processing of the netlink attributes, such as the following:

  BUG: KASAN: slab-out-of-bounds in __nla_validate_parse+0x1258/0x2010
  Read of size 2 at addr ffff88800ac7952c by task kworker/0:1/12

  Workqueue: events hwsim_virtio_rx_work
  Call Trace:
   <TASK>
   dump_stack_lvl+0x45/0x5d
   print_report.cold+0x5e/0x5e5
   kasan_report+0xb1/0x1c0
   __nla_validate_parse+0x1258/0x2010
   __nla_parse+0x22/0x30
   hwsim_virtio_handle_cmd.isra.0+0x13f/0x2d0
   hwsim_virtio_rx_work+0x1b2/0x370
   process_one_work+0x8df/0x1530
   worker_thread+0x575/0x11a0
   kthread+0x29d/0x340
   ret_from_fork+0x22/0x30
 </TASK>

Discarding packets with an invalid length solves this.
Therefore, skb->len must be set at reception.

Change-Id: Ieaeb9a4c62d3beede274881a7c2722c6c6f477b6
Signed-off-by: Soenke Huster <soenke.huster@eknoes.de>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2 years agowifi: mac80211: fix locking in auth/assoc timeout
Johannes Berg [Fri, 2 Sep 2022 14:11:14 +0000 (16:11 +0200)]
wifi: mac80211: fix locking in auth/assoc timeout

If we hit an authentication or association timeout, we only
release the chanctx for the deflink, and the other link(s)
are released later by ieee80211_vif_set_links(), but we're
not locking this correctly.

Fix the locking here while releasing the channels and links.

Change-Id: I9e08c1a5434592bdc75253c1abfa6c788f9f39b1
Fixes: 81151ce462e5 ("wifi: mac80211: support MLO authentication/association with one link")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2 years agowifi: mac80211: mlme: release deflink channel in error case
Johannes Berg [Fri, 2 Sep 2022 14:11:15 +0000 (16:11 +0200)]
wifi: mac80211: mlme: release deflink channel in error case

In the prep_channel error case we didn't release the deflink
channel leaving it to be left around. Fix that.

Change-Id: If0dfd748125ec46a31fc6045a480dc28e03723d2
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2 years agowifi: mac80211: fix link warning in RX agg timer expiry
Mukesh Sisodiya [Fri, 2 Sep 2022 14:11:31 +0000 (16:11 +0200)]
wifi: mac80211: fix link warning in RX agg timer expiry

The rx data link pointer isn't set from the RX aggregation timer,
resulting in a later warning. Fix that by setting it to the first
valid link for now, with a FIXME to worry about statistics later,
it's not very important since it's just the timeout case.

Reported-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/498d714c-76be-9d04-26db-a1206878de5e@redhat.com
Fixes: 56057da4569b ("wifi: mac80211: rx: track link in RX data")
Signed-off-by: Mukesh Sisodiya <mukesh.sisodiya@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2 years agoLoongArch: mm: Remove the unneeded result variable
ye xingchen [Fri, 26 Aug 2022 07:29:03 +0000 (07:29 +0000)]
LoongArch: mm: Remove the unneeded result variable

Return the value pa_to_nid() directly instead of storing it in another
redundant variable.

Reported-by: Zeal Robot <zealci@zte.com.cn>
Signed-off-by: ye xingchen <ye.xingchen@zte.com.cn>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
2 years agoLoongArch: Fix arch_remove_memory() undefined build error
Yupeng Li [Wed, 31 Aug 2022 05:40:17 +0000 (13:40 +0800)]
LoongArch: Fix arch_remove_memory() undefined build error

The kernel build error when unslected CONFIG_MEMORY_HOTREMOVE because
arch_remove_memory() is needed by mm/memory_hotplug.c but undefined.

Some build error messages like:

 LD      vmlinux.o
 MODPOST vmlinux.symvers
 MODINFO modules.builtin.modinfo
 GEN     modules.builtin
 LD      .tmp_vmlinux.kallsyms1
loongarch64-linux-gnu-ld: mm/memory_hotplug.o: in function `.L242':
memory_hotplug.c:(.ref.text+0x930): undefined reference to `arch_remove_memory'
make: *** [Makefile:1169:vmlinux] 错误 1

Removed CONFIG_MEMORY_HOTREMOVE requirement and rearrange the file refer
to the definitions of other platform architectures.

Signed-off-by: Yupeng Li <liyupeng@zbhlos.com>
Signed-off-by: Caicai <caizp2008@163.com>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
2 years agoLoongArch: Fix section mismatch due to acpi_os_ioremap()
Huacai Chen [Fri, 2 Sep 2022 14:33:42 +0000 (22:33 +0800)]
LoongArch: Fix section mismatch due to acpi_os_ioremap()

Now acpi_os_ioremap() is marked with __init because it calls memblock_
is_memory() which is also marked with __init in the !ARCH_KEEP_MEMBLOCK
case. However, acpi_os_ioremap() is called by ordinary functions such
as acpi_os_{read, write}_memory() and causes section mismatch warnings:

WARNING: modpost: vmlinux.o: section mismatch in reference: acpi_os_read_memory (section: .text) -> acpi_os_ioremap (section: .init.text)
WARNING: modpost: vmlinux.o: section mismatch in reference: acpi_os_write_memory (section: .text) -> acpi_os_ioremap (section: .init.text)

Fix these warnings by selecting ARCH_KEEP_MEMBLOCK unconditionally and
removing the __init modifier of acpi_os_ioremap(). This can also give a
chance to track "memory" and "reserved" memblocks after early boot.

Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>