uas: ignore UAS for Thinkplus chips

commit 0fb9703a3eade0bb84c635705d9c795345e55053 upstream.

The UAS mode of Thinkplus(0x17ef, 0x3899) is reported to influence
performance and trigger kernel panic on several platforms with the
following error message:

[   39.702439] xhci_hcd 0000:0c:00.3: ERROR Transfer event for disabled
               endpoint or incorrect stream ring
[   39.702442] xhci_hcd 0000:0c:00.3: @000000026c61f810 00000000 00000000
               1b000000 05038000

[  720.545894][13] Workqueue: usb_hub_wq hub_event
[  720.550971][13]  ffff88026c143c38 0000000000016300 ffff8802755bb900 ffff880
                    26cb80000
[  720.559673][13]  ffff88026c144000 ffff88026ca88100 0000000000000000 ffff880
                    26cb80000
[  720.568374][13]  ffff88026cb80000 ffff88026c143c50 ffffffff8186ae25 ffff880
                    26ca880f8
[  720.577076][13] Call Trace:
[  720.580201][13]  [<ffffffff8186ae25>] schedule+0x35/0x80
[  720.586137][13]  [<ffffffff8186b0ce>] schedule_preempt_disabled+0xe/0x10
[  720.593623][13]  [<ffffffff8186cb94>] __mutex_lock_slowpath+0x164/0x1e0
[  720.601012][13]  [<ffffffff8186cc3f>] mutex_lock+0x2f/0x40
[  720.607141][13]  [<ffffffff8162b8e9>] usb_disconnect+0x59/0x290

Falling back to USB mass storage can solve this problem, so ignore UAS
function of this chip.

Acked-by: Alan Stern <stern@rowland.harvard.edu>
Cc: stable <stable@kernel.org>
Signed-off-by: Hongling Zeng <zenghongling@kylinos.cn>
Link: https://lore.kernel.org/r/1663902249837086.19.seg@mailgw
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb-storage: Add Hiksemi USB3-FW to IGNORE_UAS

commit e00b488e813f0f1ad9f778e771b7cd2fe2877023 upstream.

The UAS mode of Hiksemi USB_HDD is reported to fail to work on several
platforms with the following error message, then after re-connecting the
device will be offlined and not working at all.

[  592.518442][ 2] sd 8:0:0:0: [sda] tag#17 uas_eh_abort_handler 0 uas-tag 18
                   inflight: CMD
[  592.527575][ 2] sd 8:0:0:0: [sda] tag#17 CDB: Write(10) 2a 00 03 6f 88 00 00
                   04 00 00
[  592.536330][ 2] sd 8:0:0:0: [sda] tag#0 uas_eh_abort_handler 0 uas-tag 1
                   inflight: CMD
[  592.545266][ 2] sd 8:0:0:0: [sda] tag#0 CDB: Write(10) 2a 00 07 44 1a 88 00
                   00 08 00

These disks have a broken uas implementation, the tag field of the status
iu-s is not set properly,so we need to fall-back to usb-storage.

Acked-by: Alan Stern <stern@rowland.harvard.edu>
Cc: stable <stable@kernel.org>
Signed-off-by: Hongling Zeng <zenghongling@kylinos.cn>
Link: https://lore.kernel.org/r/1663901185-21067-1-git-send-email-zenghongling@kylinos.cn
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

uas: add no-uas quirk for Hiksemi usb_disk

commit a625a4b8806cc1e928b7dd2cca1fee709c9de56e upstream.

The UAS mode of Hiksemi is reported to fail to work on several platforms
with the following error message, then after re-connecting the device will
be offlined and not working at all.

[  592.518442][ 2] sd 8:0:0:0: [sda] tag#17 uas_eh_abort_handler 0 uas-tag 18
                   inflight: CMD
[  592.527575][ 2] sd 8:0:0:0: [sda] tag#17 CDB: Write(10) 2a 00 03 6f 88 00 00
                   04 00 00
[  592.536330][ 2] sd 8:0:0:0: [sda] tag#0 uas_eh_abort_handler 0 uas-tag 1
                   inflight: CMD
[  592.545266][ 2] sd 8:0:0:0: [sda] tag#0 CDB: Write(10) 2a 00 07 44 1a 88 00
                   00 08 00

These disks have a broken uas implementation, the tag field of the status
iu-s is not set properly,so we need to fall-back to usb-storage.

Acked-by: Alan Stern <stern@rowland.harvard.edu>
Cc: stable <stable@kernel.org>
Signed-off-by: Hongling Zeng <zenghongling@kylinos.cn>
Link: https://lore.kernel.org/r/1663901173-21020-1-git-send-email-zenghongling@kylinos.cn
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

cgroup: cgroup_get_from_id() must check the looked-up kn is a directory

[ Upstream commit df02452f3df069a59bc9e69c84435bf115cb6e37 ]

cgroup has to be one kernfs dir, otherwise kernel panic is caused,
especially cgroup id is provide from userspace.

Reported-by: Marco Patalano <mpatalan@redhat.com>
Fixes: 6b658c4863c1 ("scsi: cgroup: Add cgroup_get_from_id()")
Cc: Muneendra <muneendra.kumar@broadcom.com>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Acked-by: Mukesh Ojha <quic_mojha@quicinc.com>
Cc: stable@vger.kernel.org # v5.14+
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

cgroup: reduce dependency on cgroup_mutex

[ Upstream commit be288169712f3dea0bc6b50c00b3ab53d85f1435 ]

Currently cgroup_get_from_path() and cgroup_get_from_id() grab
cgroup_mutex before traversing the default hierarchy to find the
kernfs_node corresponding to the path/id and then extract the linked
cgroup. Since cgroup_mutex is still held, it is guaranteed that the
cgroup will be alive and the reference can be taken on it.

However similar guarantee can be provided without depending on the
cgroup_mutex and potentially reducing avenues of cgroup_mutex contentions.
The kernfs_node's priv pointer is RCU protected pointer and with just
rcu read lock we can grab the reference on the cgroup without
cgroup_mutex. So, remove cgroup_mutex from them.

Signed-off-by: Shakeel Butt <shakeelb@google.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Stable-dep-of: df02452f3df0 ("cgroup: cgroup_get_from_id() must check the looked-up kn is a directory")
Signed-off-by: Sasha Levin <sashal@kernel.org>

ALSA: hda/realtek: fix speakers and micmute on HP 855 G8

[ Upstream commit 91502a9a0b0d5252cf3f32ebd898823c2f5aadab ]

There are several PCI ids associated with HP EliteBook 855 G8 Notebook
PC. Commit 0e68c4b11f1e6 ("ALSA: hda/realtek: fix mute/micmute LEDs for
HP 855 G8") covers 0x103c:0x8896, while this commit covers 0x103c:0x8895
which needs some additional work on top of the quirk from 0e68c4b11f1e6.

Note that the device can boot up with working speakers and micmute LED
without this patch, but the success rate would be quite low (order of
16 working boots across 709 boots) at least for the built-in drivers
scenario. This also means that there are some timing issues during early
boot and this patch is a workaround.

With this patch applied speakers and headphones are consistenly working,
as well as mute/micmute LEDs and the internal microphone.

Signed-off-by: Alexander Sergeyev <sergeev917@gmail.com>
Link: https://lore.kernel.org/r/20220114165050.ouw2nknuspclynro@localhost.localdomain
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Stable-dep-of: 496322302bf1 ("ALSA: hda/realtek: Add a quirk for HP OMEN 16 (8902) mute LED")
Signed-off-by: Sasha Levin <sashal@kernel.org>

ALSA: hda: Fix Nvidia dp infoframe

[ Upstream commit f89e409402e2aeb3bc3aa44d2b7a597959e4e6af ]

Nvidia HDA HW expects infoframe data bytes order same for both
HDMI and DP i.e infoframe data starts from 5th bytes offset. As
dp infoframe structure has 4th byte as valid infoframe data, use
hdmi infoframe structure for nvidia dp infoframe to match HW behvaior.

Signed-off-by: Mohan Kumar <mkumard@nvidia.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20220913065818.13015-1-mkumard@nvidia.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ALSA: hda: Fix hang at HD-audio codec unbinding due to refcount saturation

[ Upstream commit ead3d3c5b54f76da79c079e61bacb4279ec56965 ]

We fixed the potential deadlock at dynamic unbinding the HD-audio
codec at the commit 7206998f578d ("ALSA: hda: Fix potential deadlock
at codec unbinding"), but ironically, this caused another potential
deadlock.  The current code uses refcount_dec() and waits for the
pending task with wait_event for dropping the refcount to 0.  This
works fine when PCMs are assigned and actually waiting for the
refcount drop.

Meanwhile, when there was no PCM assigned, the refcount_dec() call
itself was supposed to drop to zero -- alas, it doesn't in reality;
refcount_dec() complains, spews kernel warning and it saturates
instead of dropping to 0, due to the nature of refcount_dec()
implementation.  This eventually blocks the wait_event() wakeup and
the code get stuck there.

For avoiding the problem, we call refcount_dec_and_test() and skips
the sync-wait if it already reaches to zero.

The patch does a slight code reshuffling to make sure to invoke other
disconnect calls before the sync-wait, too.

Fixes: 7206998f578d ("ALSA: hda: Fix potential deadlock at codec unbinding")
Reported-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/YxtflWQnslMHVlU7@intel.com
Link: https://lore.kernel.org/r/20220910142550.28494-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ALSA: hda: Do disconnect jacks at codec unbind

[ Upstream commit 37c4fd0db7c961145d9d1909ecab386fdf703c26 ]

The HD-audio codec driver remove may happen also at dynamically
unbinding during operation, hence it needs manual triggers of
snd_device_disconnect() calls, while it's missing for the jack objects
that are associated with the codec.

This patch adds the manual disconnection call for jacks when the
remove happens without card->shutdown (i.e. not under the full
removal).

Link: https://lore.kernel.org/r/20211117133040.20272-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Stable-dep-of: ead3d3c5b54f ("ALSA: hda: Fix hang at HD-audio codec unbinding due to refcount saturation")
Signed-off-by: Sasha Levin <sashal@kernel.org>

Linux 5.15.71

Link: https://lore.kernel.org/r/20220926100756.074519146@linuxfoundation.org
Tested-by: Shuah Khan <skhan@linuxfoundation.org>
Link: https://lore.kernel.org/r/20220926163551.791017156@linuxfoundation.org
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Bagas Sanjaya <bagasdotme@gmail.com>
Tested-by: Linux Kernel Functional Testing <lkft@linaro.org>
Tested-by: Sudip Mukherjee <sudip.mukherjee@codethink.co.uk>
Tested-by: Ron Economos <re@w6rz.net>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Tested-by: Kelsey Steele <kelseysteele@linux.microsoft.com>
Tested-by: Jon Hunter <jonathanh@nvidia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: use locality group preallocation for small closed files

commit a9f2a2931d0e197ab28c6007966053fdababd53f upstream.

Curently we don't use any preallocation when a file is already closed
when allocating blocks (from writeback code when converting delayed
allocation). However for small files, using locality group preallocation
is actually desirable as that is not specific to a particular file.
Rather it is a method to pack small files together to reduce
fragmentation and for that the fact the file is closed is actually even
stronger hint the file would benefit from packing. So change the logic
to allow locality group preallocation in this case.

Fixes: 196e402adf2e ("ext4: improve cr 0 / cr 1 group scanning")
CC: stable@kernel.org
Reported-and-tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Tested-by: Ojaswin Mujoo <ojaswin@linux.ibm.com>
Reviewed-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/all/0d81a7c2-46b7-6010-62a4-3e6cfc1628d6@i2se.com/
Link: https://lore.kernel.org/r/20220908092136.11770-4-jack@suse.cz
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: avoid unnecessary spreading of allocations among groups

commit 1940265ede6683f6317cba0d428ce6505eaca944 upstream.

mb_set_largest_free_order() updates lists containing groups with largest
chunk of free space of given order. The way it updates it leads to
always moving the group to the tail of the list. Thus allocations
looking for free space of given order effectively end up cycling through
all groups (and due to initialization in last to first order). This
spreads allocations among block groups which reduces performance for
rotating disks or low-end flash media. Change
mb_set_largest_free_order() to only update lists if the order of the
largest free chunk in the group changed.

Fixes: 196e402adf2e ("ext4: improve cr 0 / cr 1 group scanning")
CC: stable@kernel.org
Reported-and-tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Tested-by: Ojaswin Mujoo <ojaswin@linux.ibm.com>
Reviewed-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/all/0d81a7c2-46b7-6010-62a4-3e6cfc1628d6@i2se.com/
Link: https://lore.kernel.org/r/20220908092136.11770-2-jack@suse.cz
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: make mballoc try target group first even with mb_optimize_scan

commit 4fca50d440cc5d4dc570ad5484cc0b70b381bc2a upstream.

One of the side-effects of mb_optimize_scan was that the optimized
functions to select next group to try were called even before we tried
the goal group. As a result we no longer allocate files close to
corresponding inodes as well as we don't try to expand currently
allocated extent in the same group. This results in reaim regression
with workfile.disk workload of upto 8% with many clients on my test
machine:

                     baseline               mb_optimize_scan
Hmean     disk-1       2114.16 (   0.00%)     2099.37 (  -0.70%)
Hmean     disk-41     87794.43 (   0.00%)    83787.47 *  -4.56%*
Hmean     disk-81    148170.73 (   0.00%)   135527.05 *  -8.53%*
Hmean     disk-121   177506.11 (   0.00%)   166284.93 *  -6.32%*
Hmean     disk-161   220951.51 (   0.00%)   207563.39 *  -6.06%*
Hmean     disk-201   208722.74 (   0.00%)   203235.59 (  -2.63%)
Hmean     disk-241   222051.60 (   0.00%)   217705.51 (  -1.96%)
Hmean     disk-281   252244.17 (   0.00%)   241132.72 *  -4.41%*
Hmean     disk-321   255844.84 (   0.00%)   245412.84 *  -4.08%*

Also this is causing huge regression (time increased by a factor of 5 or
so) when untarring archive with lots of small files on some eMMC storage
cards.

Fix the problem by making sure we try goal group first.

Fixes: 196e402adf2e ("ext4: improve cr 0 / cr 1 group scanning")
CC: stable@kernel.org
Reported-and-tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Tested-by: Ojaswin Mujoo <ojaswin@linux.ibm.com>
Reviewed-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
Link: https://lore.kernel.org/all/20220727105123.ckwrhbilzrxqpt24@quack3/
Link: https://lore.kernel.org/all/0d81a7c2-46b7-6010-62a4-3e6cfc1628d6@i2se.com/
Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220908092136.11770-1-jack@suse.cz
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: limit the number of retries after discarding preallocations blocks

commit 80fa46d6b9e7b1527bfd2197d75431fd9c382161 upstream.

This patch avoids threads live-locking for hours when a large number
threads are competing over the last few free extents as they blocks
getting added and removed from preallocation pools.  From our bug
reporter:

   A reliable way for triggering this has multiple writers
   continuously write() to files when the filesystem is full, while
   small amounts of space are freed (e.g. by truncating a large file
   -1MiB at a time). In the local filesystem, this can be done by
   simply not checking the return code of write (0) and/or the error
   (ENOSPACE) that is set. Over NFS with an async mount, even clients
   with proper error checking will behave this way since the linux NFS
   client implementation will not propagate the server errors [the
   write syscalls immediately return success] until the file handle is
   closed. This leads to a situation where NFS clients send a
   continuous stream of WRITE rpcs which result in ERRNOSPACE -- but
   since the client isn't seeing this, the stream of writes continues
   at maximum network speed.

   When some space does appear, multiple writers will all attempt to
   claim it for their current write. For NFS, we may see dozens to
   hundreds of threads that do this.

   The real-world scenario of this is database backup tooling (in
   particular, github.com/mdkent/percona-xtrabackup) which may write
   large files (>1TiB) to NFS for safe keeping. Some temporary files
   are written, rewound, and read back -- all before closing the file
   handle (the temp file is actually unlinked, to trigger automatic
   deletion on close/crash.) An application like this operating on an
   async NFS mount will not see an error code until TiB have been
   written/read.

   The lockup was observed when running this database backup on large
   filesystems (64 TiB in this case) with a high number of block
   groups and no free space. Fragmentation is generally not a factor
   in this filesystem (~thousands of large files, mostly contiguous
   except for the parts written while the filesystem is at capacity.)

Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: fix bug in extents parsing when eh_entries == 0 and eh_depth > 0

commit 29a5b8a137ac8eb410cc823653a29ac0e7b7e1b0 upstream.

When walking through an inode extents, the ext4_ext_binsearch_idx() function
assumes that the extent header has been previously validated.  However, there
are no checks that verify that the number of entries (eh->eh_entries) is
non-zero when depth is > 0.  And this will lead to problems because the
EXT_FIRST_INDEX() and EXT_LAST_INDEX() will return garbage and result in this:

[  135.245946] ------------[ cut here ]------------
[  135.247579] kernel BUG at fs/ext4/extents.c:2258!
[  135.249045] invalid opcode: 0000 [#1] PREEMPT SMP
[  135.250320] CPU: 2 PID: 238 Comm: tmp118 Not tainted 5.19.0-rc8+ #4
[  135.252067] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.15.0-0-g2dd4b9b-rebuilt.opensuse.org 04/01/2014
[  135.255065] RIP: 0010:ext4_ext_map_blocks+0xc20/0xcb0
[  135.256475] Code:
[  135.261433] RSP: 0018:ffffc900005939f8 EFLAGS: 00010246
[  135.262847] RAX: 0000000000000024 RBX: ffffc90000593b70 RCX: 0000000000000023
[  135.264765] RDX: ffff8880038e5f10 RSI: 0000000000000003 RDI: ffff8880046e922c
[  135.266670] RBP: ffff8880046e9348 R08: 0000000000000001 R09: ffff888002ca580c
[  135.268576] R10: 0000000000002602 R11: 0000000000000000 R12: 0000000000000024
[  135.270477] R13: 0000000000000000 R14: 0000000000000024 R15: 0000000000000000
[  135.272394] FS:  00007fdabdc56740(0000) GS:ffff88807dd00000(0000) knlGS:0000000000000000
[  135.274510] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  135.276075] CR2: 00007ffc26bd4f00 CR3: 0000000006261004 CR4: 0000000000170ea0
[  135.277952] Call Trace:
[  135.278635]  <TASK>
[  135.279247]  ? preempt_count_add+0x6d/0xa0
[  135.280358]  ? percpu_counter_add_batch+0x55/0xb0
[  135.281612]  ? _raw_read_unlock+0x18/0x30
[  135.282704]  ext4_map_blocks+0x294/0x5a0
[  135.283745]  ? xa_load+0x6f/0xa0
[  135.284562]  ext4_mpage_readpages+0x3d6/0x770
[  135.285646]  read_pages+0x67/0x1d0
[  135.286492]  ? folio_add_lru+0x51/0x80
[  135.287441]  page_cache_ra_unbounded+0x124/0x170
[  135.288510]  filemap_get_pages+0x23d/0x5a0
[  135.289457]  ? path_openat+0xa72/0xdd0
[  135.290332]  filemap_read+0xbf/0x300
[  135.291158]  ? _raw_spin_lock_irqsave+0x17/0x40
[  135.292192]  new_sync_read+0x103/0x170
[  135.293014]  vfs_read+0x15d/0x180
[  135.293745]  ksys_read+0xa1/0xe0
[  135.294461]  do_syscall_64+0x3c/0x80
[  135.295284]  entry_SYSCALL_64_after_hwframe+0x46/0xb0

This patch simply adds an extra check in __ext4_ext_check(), verifying that
eh_entries is not 0 when eh_depth is > 0.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=215941
Link: https://bugzilla.kernel.org/show_bug.cgi?id=216283
Cc: Baokun Li <libaokun1@huawei.com>
Cc: stable@kernel.org
Signed-off-by: Luís Henriques <lhenriques@suse.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Baokun Li <libaokun1@huawei.com>
Link: https://lore.kernel.org/r/20220822094235.2690-1-lhenriques@suse.de
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: make directory inode spreading reflect flexbg size

commit 613c5a85898d1cd44e68f28d65eccf64a8ace9cf upstream.

Currently the Orlov inode allocator searches for free inodes for a
directory only in flex block groups with at most inodes_per_group/16
more directory inodes than average per flex block group. However with
growing size of flex block group this becomes unnecessarily strict.
Scale allowed difference from average directory count per flex block
group with flex block group size as we do with other metrics.

Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Tested-by: Ojaswin Mujoo <ojaswin@linux.ibm.com>
Cc: stable@kernel.org
Link: https://lore.kernel.org/all/0d81a7c2-46b7-6010-62a4-3e6cfc1628d6@i2se.com/
Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220908092136.11770-3-jack@suse.cz
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

devdax: Fix soft-reservation memory description

commit 67feaba413ec68daf4124e9870878899b4ed9a0e upstream.

The "hmem" platform-devices that are created to represent the
platform-advertised "Soft Reserved" memory ranges end up inserting a
resource that causes the iomem_resource tree to look like this:

340000000-43fffffff : hmem.0
  340000000-43fffffff : Soft Reserved
    340000000-43fffffff : dax0.0

This is because insert_resource() reparents ranges when they completely
intersect an existing range.

This matters because code that uses region_intersects() to scan for a
given IORES_DESC will only check that top-level 'hmem.0' resource and
not the 'Soft Reserved' descendant.

So, to support EINJ (via einj_error_inject()) to inject errors into
memory hosted by a dax-device, be sure to describe the memory as
IORES_DESC_SOFT_RESERVED. This is a follow-on to:

commit b13a3e5fd40b ("ACPI: APEI: Fix _EINJ vs EFI_MEMORY_SP")

...that fixed EINJ support for "Soft Reserved" ranges in the first
instance.

Fixes: 262b45ae3ab4 ("x86/efi: EFI soft reservation to E820 enumeration")
Reported-by: Ricardo Sandoval Torres <ricardo.sandoval.torres@intel.com>
Tested-by: Ricardo Sandoval Torres <ricardo.sandoval.torres@intel.com>
Cc: <stable@vger.kernel.org>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Omar Avelar <omar.avelar@intel.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Mark Gross <markgross@kernel.org>
Link: https://lore.kernel.org/r/166397075670.389916.7435722208896316387.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

NFSv4: Fixes for nfs4_inode_return_delegation()

commit 6e176d47160cec8bcaa28d9aa06926d72d54237c upstream.

We mustn't call nfs_wb_all() on anything other than a regular file.
Furthermore, we can exit early when we don't hold a delegation.

Reported-by: David Wysochanski <dwysocha@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Cc: Thorsten Leemhuis <regressions@leemhuis.info>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/amdgpu: don't register a dirty callback for non-atomic

[ Upstream commit abbc7a3dafb91b9d4ec56b70ec9a7520f8e13334 ]

Some asics still support non-atomic code paths.

Fixes: 66f99628eb2440 ("drm/amdgpu: use dirty framebuffer helper")
Reported-by: Arthur Marsh <arthur.marsh@internode.on.net>
Reviewed-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

i2c: mlxbf: Fix frequency calculation

[ Upstream commit 37f071ec327b04c83d47637c5e5c2199b39899ca ]

The i2c-mlxbf.c driver is currently broken because there is a bug
in the calculation of the frequency. core_f, core_r and core_od
are components read from hardware registers and are used to
compute the frequency used to compute different timing parameters.
The shifting mechanism used to get core_f, core_r and core_od is
wrong. Use FIELD_GET to mask and shift the bitfields properly.

Fixes: b5b5b32081cd206b (i2c: mlxbf: I2C SMBus driver for Mellanox BlueField SoC)
Reviewed-by: Khalil Blaiech <kblaiech@nvidia.com>
Signed-off-by: Asmaa Mnebhi <asmaa@nvidia.com>
Signed-off-by: Wolfram Sang <wsa@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

i2c: mlxbf: prevent stack overflow in mlxbf_i2c_smbus_start_transaction()

[ Upstream commit de24aceb07d426b6f1c59f33889d6a964770547b ]

memcpy() is called in a loop while 'operation->length' upper bound
is not checked and 'data_idx' also increments.

Fixes: b5b5b32081cd206b ("i2c: mlxbf: I2C SMBus driver for Mellanox BlueField SoC")
Reviewed-by: Khalil Blaiech <kblaiech@nvidia.com>
Signed-off-by: Asmaa Mnebhi <asmaa@nvidia.com>
Signed-off-by: Wolfram Sang <wsa@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

i2c: mlxbf: incorrect base address passed during io write

[ Upstream commit 2a5be6d1340c0fefcee8a6489cff7fd88a0d5b85 ]

Correct the base address used during io write.
This bug had no impact over the overall functionality of the read and write
transactions. MLXBF_I2C_CAUSE_OR_CLEAR=0x18 so writing to (smbus->io + 0x18)
instead of (mst_cause->ioi + 0x18) actually writes to the sc_low_timeout
register which just sets the timeout value before a read/write aborts.

Fixes: b5b5b32081cd206b (i2c: mlxbf: I2C SMBus driver for Mellanox BlueField SoC)
Reviewed-by: Khalil Blaiech <kblaiech@nvidia.com>
Signed-off-by: Asmaa Mnebhi <asmaa@nvidia.com>
Signed-off-by: Wolfram Sang <wsa@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

i2c: imx: If pm_runtime_get_sync() returned 1 device access is possible

[ Upstream commit 085aacaa73163f4b8a89dec24ecb32cfacd34017 ]

pm_runtime_get_sync() returning 1 also means the device is powered. So
resetting the chip registers in .remove() is possible and should be
done.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Fixes: d98bdd3a5b50 ("i2c: imx: Make sure to unregister adapter on remove()")
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: Oleksij Rempel <o.rempel@pengutronix.de>
Signed-off-by: Wolfram Sang <wsa@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

workqueue: don't skip lockdep work dependency in cancel_work_sync()

[ Upstream commit c0feea594e058223973db94c1c32a830c9807c86 ]

Like Hillf Danton mentioned

  syzbot should have been able to catch cancel_work_sync() in work context
  by checking lockdep_map in __flush_work() for both flush and cancel.

in [1], being unable to report an obvious deadlock scenario shown below is
broken. From locking dependency perspective, sync version of cancel request
should behave as if flush request, for it waits for completion of work if
that work has already started execution.

  ----------
  #include <linux/module.h>
  #include <linux/sched.h>
  static DEFINE_MUTEX(mutex);
  static void work_fn(struct work_struct *work)
  {
    schedule_timeout_uninterruptible(HZ / 5);
    mutex_lock(&mutex);
    mutex_unlock(&mutex);
  }
  static DECLARE_WORK(work, work_fn);
  static int __init test_init(void)
  {
    schedule_work(&work);
    schedule_timeout_uninterruptible(HZ / 10);
    mutex_lock(&mutex);
    cancel_work_sync(&work);
    mutex_unlock(&mutex);
    return -EINVAL;
  }
  module_init(test_init);
  MODULE_LICENSE("GPL");
  ----------

The check this patch restores was added by commit 0976dfc1d0cd80a4
("workqueue: Catch more locking problems with flush_work()").

Then, lockdep's crossrelease feature was added by commit b09be676e0ff25bd
("locking/lockdep: Implement the 'crossrelease' feature"). As a result,
this check was once removed by commit fd1a5b04dfb899f8 ("workqueue: Remove
now redundant lock acquisitions wrt. workqueue flushes").

But lockdep's crossrelease feature was removed by commit e966eaeeb623f099
("locking/lockdep: Remove the cross-release locking checks"). At this
point, this check should have been restored.

Then, commit d6e89786bed977f3 ("workqueue: skip lockdep wq dependency in
cancel_work_sync()") introduced a boolean flag in order to distinguish
flush_work() and cancel_work_sync(), for checking "struct workqueue_struct"
dependency when called from cancel_work_sync() was causing false positives.

Then, commit 87915adc3f0acdf0 ("workqueue: re-add lockdep dependencies for
flushing") tried to restore "struct work_struct" dependency check, but by
error checked this boolean flag. Like an example shown above indicates,
"struct work_struct" dependency needs to be checked for both flush_work()
and cancel_work_sync().

Link: https://lkml.kernel.org/r/20220504044800.4966-1-hdanton@sina.com
Reported-by: Hillf Danton <hdanton@sina.com>
Suggested-by: Lai Jiangshan <jiangshanlai@gmail.com>
Fixes: 87915adc3f0acdf0 ("workqueue: re-add lockdep dependencies for flushing")
Cc: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

fsdax: Fix infinite loop in dax_iomap_rw()

[ Upstream commit 17d9c15c9b9e7fb285f7ac5367dfb5f00ff575e3 ]

I got an infinite loop and a WARNING report when executing a tail command
in virtiofs.

  WARNING: CPU: 10 PID: 964 at fs/iomap/iter.c:34 iomap_iter+0x3a2/0x3d0
  Modules linked in:
  CPU: 10 PID: 964 Comm: tail Not tainted 5.19.0-rc7
  Call Trace:
  <TASK>
  dax_iomap_rw+0xea/0x620
  ? __this_cpu_preempt_check+0x13/0x20
  fuse_dax_read_iter+0x47/0x80
  fuse_file_read_iter+0xae/0xd0
  new_sync_read+0xfe/0x180
  ? 0xffffffff81000000
  vfs_read+0x14d/0x1a0
  ksys_read+0x6d/0xf0
  __x64_sys_read+0x1a/0x20
  do_syscall_64+0x3b/0x90
  entry_SYSCALL_64_after_hwframe+0x63/0xcd

The tail command will call read() with a count of 0. In this case,
iomap_iter() will report this WARNING, and always return 1 which casuing
the infinite loop in dax_iomap_rw().

Fixing by checking count whether is 0 in dax_iomap_rw().

Fixes: ca289e0b95af ("fsdax: switch dax_iomap_rw to use iomap_iter")
Signed-off-by: Li Jinlin <lijinlin3@huawei.com>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Link: https://lore.kernel.org/r/20220725032050.3873372-1-lijinlin3@huawei.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/rockchip: Fix return type of cdn_dp_connector_mode_valid

[ Upstream commit b0b9408f132623dc88e78adb5282f74e4b64bb57 ]

The mode_valid field in drm_connector_helper_funcs is expected to be of
type:
enum drm_mode_status (* mode_valid) (struct drm_connector *connector,
     struct drm_display_mode *mode);

The mismatched return type breaks forward edge kCFI since the underlying
function definition does not match the function hook definition.

The return type of cdn_dp_connector_mode_valid should be changed from
int to enum drm_mode_status.

Reported-by: Dan Carpenter <error27@gmail.com>
Link: https://github.com/ClangBuiltLinux/linux/issues/1703
Cc: llvm@lists.linux.dev
Signed-off-by: Nathan Huckleberry <nhuck@google.com>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20220913205555.155149-1-nhuck@google.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/amd/display: Mark dml30's UseMinimumDCFCLK() as noinline for stack usage

[ Upstream commit 41012d715d5d7b9751ae84b8fb255e404ac9c5d0 ]

This function consumes a lot of stack space and it blows up the size of
dml30_ModeSupportAndSystemConfigurationFull() with clang:

  drivers/gpu/drm/amd/amdgpu/../display/dc/dml/dcn30/display_mode_vba_30.c:3542:6: error: stack frame size (2200) exceeds limit (2048) in 'dml30_ModeSupportAndSystemConfigurationFull' [-Werror,-Wframe-larger-than]
  void dml30_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_lib)
       ^
  1 error generated.

Commit a0f7e7f759cf ("drm/amd/display: fix i386 frame size warning")
aimed to address this for i386 but it did not help x86_64.

To reduce the amount of stack space that
dml30_ModeSupportAndSystemConfigurationFull() uses, mark
UseMinimumDCFCLK() as noinline, using the _for_stack variant for
documentation. While this will increase the total amount of stack usage
between the two functions (1632 and 1304 bytes respectively), it will
make sure both stay below the limit of 2048 bytes for these files. The
aforementioned change does help reduce UseMinimumDCFCLK()'s stack usage
so it should not be reverted in favor of this change.

Link: https://github.com/ClangBuiltLinux/linux/issues/1681
Reported-by: "Sudip Mukherjee (Codethink)" <sudipm.mukherjee@gmail.com>
Tested-by: Maíra Canal <mairacanal@riseup.net>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/amd/display: Reduce number of arguments of dml31's CalculateFlipSchedule()

[ Upstream commit 21485d3da659b66c37d99071623af83ee1c6733d ]

Most of the arguments are identical between the two call sites and they
can be accessed through the 'struct vba_vars_st' pointer. This reduces
the total amount of stack space that
dml31_ModeSupportAndSystemConfigurationFull() uses by 112 bytes with
LLVM 16 (1976 -> 1864), helping clear up the following clang warning:

  drivers/gpu/drm/amd/amdgpu/../display/dc/dml/dcn31/display_mode_vba_31.c:3908:6: error: stack frame size (2216) exceeds limit (2048) in 'dml31_ModeSupportAndSystemConfigurationFull' [-Werror,-Wframe-larger-than]
  void dml31_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_lib)
      ^
  1 error generated.

Link: https://github.com/ClangBuiltLinux/linux/issues/1681
Reported-by: "Sudip Mukherjee (Codethink)" <sudipm.mukherjee@gmail.com>
Tested-by: Maíra Canal <mairacanal@riseup.net>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/amd/display: Reduce number of arguments of dml31's CalculateWatermarksAndDRAMSpeedChangeSupport()

[ Upstream commit 37934d4118e22bceb80141804391975078f31734 ]

Most of the arguments are identical between the two call sites and they
can be accessed through the 'struct vba_vars_st' pointer. This reduces
the total amount of stack space that
dml31_ModeSupportAndSystemConfigurationFull() uses by 240 bytes with
LLVM 16 (2216 -> 1976), helping clear up the following clang warning:

  drivers/gpu/drm/amd/amdgpu/../display/dc/dml/dcn31/display_mode_vba_31.c:3908:6: error: stack frame size (2216) exceeds limit (2048) in 'dml31_ModeSupportAndSystemConfigurationFull' [-Werror,-Wframe-larger-than]
  void dml31_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_lib)
      ^
  1 error generated.

Link: https://github.com/ClangBuiltLinux/linux/issues/1681
Reported-by: "Sudip Mukherjee (Codethink)" <sudipm.mukherjee@gmail.com>
Tested-by: Maíra Canal <mairacanal@riseup.net>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/amd/display: Limit user regamma to a valid value

[ Upstream commit 3601d620f22e37740cf73f8278eabf9f2aa19eb7 ]

[Why]
For HDR mode, we get total 512 tf_point and after switching to SDR mode
we actually get 400 tf_point and the rest of points(401~512) still use
dirty value from HDR mode. We should limit the rest of the points to max
value.

[How]
Limit the value when coordinates_x.x > 1, just like what we do in
translate_from_linear_space for other re-gamma build paths.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Pavle Kotarac <Pavle.Kotarac@amd.com>
Signed-off-by: Yao Wang1 <Yao.Wang1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/amdgpu: use dirty framebuffer helper

[ Upstream commit 66f99628eb24409cb8feb5061f78283c8b65f820 ]

Currently, we aren't handling DRM_IOCTL_MODE_DIRTYFB. So, use
drm_atomic_helper_dirtyfb() as the dirty callback in the amdgpu_fb_funcs
struct.

Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/amd/pm: disable BACO entry/exit completely on several sienna cichlid cards

[ Upstream commit 7c6fb61a400bf3218c6504cb2d48858f98822c9d ]

To avoid hardware intermittent failures.

Signed-off-by: Guchun Chen <guchun.chen@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/gma500: Fix BUG: sleeping function called from invalid context errors

[ Upstream commit 63e37a79f7bd939314997e29c2f5a9f0ef184281 ]

gma_crtc_page_flip() was holding the event_lock spinlock while calling
crtc_funcs->mode_set_base() which takes ww_mutex.

The only reason to hold event_lock is to clear gma_crtc->page_flip_event
on mode_set_base() errors.

Instead unlock it after setting gma_crtc->page_flip_event and on
errors re-take the lock and clear gma_crtc->page_flip_event it
it is still set.

This fixes the following WARN/stacktrace:

[  512.122953] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:870
[  512.123004] in_atomic(): 1, irqs_disabled(): 1, non_block: 0, pid: 1253, name: gnome-shell
[  512.123031] preempt_count: 1, expected: 0
[  512.123048] RCU nest depth: 0, expected: 0
[  512.123066] INFO: lockdep is turned off.
[  512.123080] irq event stamp: 0
[  512.123094] hardirqs last  enabled at (0): [<0000000000000000>] 0x0
[  512.123134] hardirqs last disabled at (0): [<ffffffff8d0ec28c>] copy_process+0x9fc/0x1de0
[  512.123176] softirqs last  enabled at (0): [<ffffffff8d0ec28c>] copy_process+0x9fc/0x1de0
[  512.123207] softirqs last disabled at (0): [<0000000000000000>] 0x0
[  512.123233] Preemption disabled at:
[  512.123241] [<0000000000000000>] 0x0
[  512.123275] CPU: 3 PID: 1253 Comm: gnome-shell Tainted: G        W         5.19.0+ #1
[  512.123304] Hardware name: Packard Bell dot s/SJE01_CT, BIOS V1.10 07/23/2013
[  512.123323] Call Trace:
[  512.123346]  <TASK>
[  512.123370]  dump_stack_lvl+0x5b/0x77
[  512.123412]  __might_resched.cold+0xff/0x13a
[  512.123458]  ww_mutex_lock+0x1e/0xa0
[  512.123495]  psb_gem_pin+0x2c/0x150 [gma500_gfx]
[  512.123601]  gma_pipe_set_base+0x76/0x240 [gma500_gfx]
[  512.123708]  gma_crtc_page_flip+0x95/0x130 [gma500_gfx]
[  512.123808]  drm_mode_page_flip_ioctl+0x57d/0x5d0
[  512.123897]  ? drm_mode_cursor2_ioctl+0x10/0x10
[  512.123936]  drm_ioctl_kernel+0xa1/0x150
[  512.123984]  drm_ioctl+0x21f/0x420
[  512.124025]  ? drm_mode_cursor2_ioctl+0x10/0x10
[  512.124070]  ? rcu_read_lock_bh_held+0xb/0x60
[  512.124104]  ? lock_release+0x1ef/0x2d0
[  512.124161]  __x64_sys_ioctl+0x8d/0xd0
[  512.124203]  do_syscall_64+0x58/0x80
[  512.124239]  ? do_syscall_64+0x67/0x80
[  512.124267]  ? trace_hardirqs_on_prepare+0x55/0xe0
[  512.124300]  ? do_syscall_64+0x67/0x80
[  512.124340]  ? rcu_read_lock_sched_held+0x10/0x80
[  512.124377]  entry_SYSCALL_64_after_hwframe+0x63/0xcd
[  512.124411] RIP: 0033:0x7fcc4a70740f
[  512.124442] Code: 00 48 89 44 24 18 31 c0 48 8d 44 24 60 c7 04 24 10 00 00 00 48 89 44 24 08 48 8d 44 24 20 48 89 44 24 10 b8 10 00 00 00 0f 05 <89> c2 3d 00 f0 ff ff 77 18 48 8b 44 24 18 64 48 2b 04 25 28 00 00
[  512.124470] RSP: 002b:00007ffda73f5390 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
[  512.124503] RAX: ffffffffffffffda RBX: 000055cc9e474500 RCX: 00007fcc4a70740f
[  512.124524] RDX: 00007ffda73f5420 RSI: 00000000c01864b0 RDI: 0000000000000009
[  512.124544] RBP: 00007ffda73f5420 R08: 000055cc9c0b0cb0 R09: 0000000000000034
[  512.124564] R10: 0000000000000000 R11: 0000000000000246 R12: 00000000c01864b0
[  512.124584] R13: 0000000000000009 R14: 000055cc9df484d0 R15: 000055cc9af5d0c0
[  512.124647]  </TASK>

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220906203852.527663-2-hdegoede@redhat.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

Drivers: hv: Never allocate anything besides framebuffer from framebuffer memory region

[ Upstream commit f0880e2cb7e1f8039a048fdd01ce45ab77247221 ]

Passed through PCI device sometimes misbehave on Gen1 VMs when Hyper-V
DRM driver is also loaded. Looking at IOMEM assignment, we can see e.g.

$ cat /proc/iomem
...
f8000000-fffbffff : PCI Bus 0000:00
  f8000000-fbffffff : 0000:00:08.0
    f8000000-f8001fff : bb8c4f33-2ba2-4808-9f7f-02f3b4da22fe
...
fe0000000-fffffffff : PCI Bus 0000:00
  fe0000000-fe07fffff : bb8c4f33-2ba2-4808-9f7f-02f3b4da22fe
    fe0000000-fe07fffff : 2ba2:00:02.0
      fe0000000-fe07fffff : mlx4_core

the interesting part is the 'f8000000' region as it is actually the
VM's framebuffer:

$ lspci -v
...
0000:00:08.0 VGA compatible controller: Microsoft Corporation Hyper-V virtual VGA (prog-if 00 [VGA controller])
Flags: bus master, fast devsel, latency 0, IRQ 11
Memory at f8000000 (32-bit, non-prefetchable) [size=64M]
...

 hv_vmbus: registering driver hyperv_drm
 hyperv_drm 5620e0c7-8062-4dce-aeb7-520c7ef76171: [drm] Synthvid Version major 3, minor 5
 hyperv_drm 0000:00:08.0: vgaarb: deactivate vga console
 hyperv_drm 0000:00:08.0: BAR 0: can't reserve [mem 0xf8000000-0xfbffffff]
 hyperv_drm 5620e0c7-8062-4dce-aeb7-520c7ef76171: [drm] Cannot request framebuffer, boot fb still active?

Note: "Cannot request framebuffer" is not a fatal error in
hyperv_setup_gen1() as the code assumes there's some other framebuffer
device there but we actually have some other PCI device (mlx4 in this
case) config space there!

The problem appears to be that vmbus_allocate_mmio() can use dedicated
framebuffer region to serve any MMIO request from any device. The
semantics one might assume of a parameter named "fb_overlap_ok"
aren't implemented because !fb_overlap_ok essentially has no effect.
The existing semantics are really "prefer_fb_overlap". This patch
implements the expected and needed semantics, which is to not allocate
from the frame buffer space when !fb_overlap_ok.

Note, Gen2 VMs are usually unaffected by the issue because
framebuffer region is already taken by EFI fb (in case kernel supports
it) but Gen1 VMs may have this region unclaimed by the time Hyper-V PCI
pass-through driver tries allocating MMIO space if Hyper-V DRM/FB drivers
load after it. Devices can be brought up in any sequence so let's
resolve the issue by always ignoring 'fb_mmio' region for non-FB
requests, even if the region is unclaimed.

Reviewed-by: Michael Kelley <mikelley@microsoft.com>
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Link: https://lore.kernel.org/r/20220827130345.1320254-4-vkuznets@redhat.com
Signed-off-by: Wei Liu <wei.liu@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/amd/amdgpu: fixing read wrong pf2vf data in SRIOV

commit 9a458402fb69bda886aa6cbe067311b6e3d9c52a upstream.

[Why]
This fixes 892deb48269c ("drm/amdgpu: Separate vf2pf work item init from virt data exchange").
we should read pf2vf data based at mman.fw_vram_usage_va after gmc
sw_init. commit 892deb48269c breaks this logic.

[How]
calling amdgpu_virt_exchange_data in amdgpu_virt_init_data_exchange to
set the right base in the right sequence.

v2:
call amdgpu_virt_init_data_exchange after gmc sw_init to make data
exchange workqueue run

v3:
clean up the code logic

v4:
add some comment and make the code more readable

Fixes: 892deb48269c ("drm/amdgpu: Separate vf2pf work item init from virt data exchange")
Signed-off-by: Jingwen Chen <Jingwen.Chen2@amd.com>
Reviewed-by: Horace Chen <horace.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

s390/dasd: fix Oops in dasd_alias_get_start_dev due to missing pavgroup

commit db7ba07108a48c0f95b74fabbfd5d63e924f992d upstream.

Fix Oops in dasd_alias_get_start_dev() function caused by the pavgroup
pointer being NULL.

The pavgroup pointer is checked on the entrance of the function but
without the lcu->lock being held. Therefore there is a race window
between dasd_alias_get_start_dev() and _lcu_update() which sets
pavgroup to NULL with the lcu->lock held.

Fix by checking the pavgroup pointer with lcu->lock held.

Cc: <stable@vger.kernel.org> # 2.6.25+
Fixes: 8e09f21574ea ("[S390] dasd: add hyper PAV support to DASD device driver, part 1")
Signed-off-by: Stefan Haberland <sth@linux.ibm.com>
Reviewed-by: Jan Hoeppner <hoeppner@linux.ibm.com>
Link: https://lore.kernel.org/r/20220919154931.4123002-2-sth@linux.ibm.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

serial: tegra-tcu: Use uart_xmit_advance(), fixes icount.tx accounting

commit 1d10cd4da593bc0196a239dcc54dac24b6b0a74e upstream.

Tx'ing does not correctly account Tx'ed characters into icount.tx.
Using uart_xmit_advance() fixes the problem.

Fixes: 2d908b38d409 ("serial: Add Tegra Combined UART driver")
Cc: <stable@vger.kernel.org> # serial: Create uart_xmit_advance()
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Link: https://lore.kernel.org/r/20220901143934.8850-4-ilpo.jarvinen@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

serial: tegra: Use uart_xmit_advance(), fixes icount.tx accounting

commit 754f68044c7dd6c52534ba3e0f664830285c4b15 upstream.

DMA complete & stop paths did not correctly account Tx'ed characters
into icount.tx. Using uart_xmit_advance() fixes the problem.

Fixes: e9ea096dd225 ("serial: tegra: add serial driver")
Cc: <stable@vger.kernel.org> # serial: Create uart_xmit_advance()
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Link: https://lore.kernel.org/r/20220901143934.8850-3-ilpo.jarvinen@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

serial: Create uart_xmit_advance()

commit e77cab77f2cb3a1ca2ba8df4af45bb35617ac16d upstream.

A very common pattern in the drivers is to advance xmit tail
index and do bookkeeping of Tx'ed characters. Create
uart_xmit_advance() to handle it.

Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Cc: stable <stable@kernel.org>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Link: https://lore.kernel.org/r/20220901143934.8850-2-ilpo.jarvinen@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

serial: fsl_lpuart: Reset prior to registration

commit 60f361722ad2ae5ee667d0b0545d40c42f754daf upstream.

Since commit bd5305dcabbc ("tty: serial: fsl_lpuart: do software reset
for imx7ulp and imx8qxp"), certain i.MX UARTs are reset after they've
already been registered.  Register state may thus be clobbered after
user space has begun to open and access the UART.

Avoid by performing the reset prior to registration.

Fixes: bd5305dcabbc ("tty: serial: fsl_lpuart: do software reset for imx7ulp and imx8qxp")
Cc: stable@vger.kernel.org # v5.15+
Cc: Fugang Duan <fugang.duan@nxp.com>
Cc: Sherry Sun <sherry.sun@nxp.com>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: https://lore.kernel.org/r/72fb646c1b0b11c989850c55f52f9ff343d1b2fa.1662884345.git.lukas@wunner.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

KVM: x86/mmu: Fold rmap_recycle into rmap_add

[ Upstream commit 68be1306caea8948738cab04014ca4506b590d38 ]

Consolidate rmap_recycle and rmap_add into a single function since they
are only ever called together (and only from one place). This has a nice
side effect of eliminating an extra kvm_vcpu_gfn_to_memslot(). In
addition it makes mmu_set_spte(), which is a very long function, a
little shorter.

No functional change intended.

Signed-off-by: David Matlack <dmatlack@google.com>
Message-Id: <20210813203504.2742757-3-dmatlack@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Stable-dep-of: 604f533262ae ("KVM: x86/mmu: add missing update to max_mmu_rmap_size")
Signed-off-by: Sasha Levin <sashal@kernel.org>

selftests: forwarding: add shebang for sch_red.sh

[ Upstream commit 83e4b196838d90799a8879e5054a3beecf9ed256 ]

RHEL/Fedora RPM build checks are stricter, and complain when executable
files don't have a shebang line, e.g.

*** WARNING: ./kselftests/net/forwarding/sch_red.sh is executable but has no shebang, removing executable bit

Fix it by adding shebang line.

Fixes: 6cf0291f9517 ("selftests: forwarding: Add a RED test for SW datapath")
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Link: https://lore.kernel.org/r/20220922024453.437757-1-liuhangbin@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bnxt: prevent skb UAF after handing over to PTP worker

[ Upstream commit c31f26c8f69f776759cbbdfb38e40ea91aa0dd65 ]

When reading the timestamp is required bnxt_tx_int() hands
over the ownership of the completed skb to the PTP worker.
The skb should not be used afterwards, as the worker may
run before the rest of our code and free the skb, leading
to a use-after-free.

Since dev_kfree_skb_any() accepts NULL make the loss of
ownership more obvious and set skb to NULL.

Fixes: 83bb623c968e ("bnxt_en: Transmit and retrieve packet timestamps")
Reviewed-by: Andy Gospodarek <gospo@broadcom.com>
Reviewed-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20220921201005.335390-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: sched: fix possible refcount leak in tc_new_tfilter()

[ Upstream commit c2e1cfefcac35e0eea229e148c8284088ce437b5 ]

tfilter_put need to be called to put the refount got by tp->ops->get to
avoid possible refcount leak when chain->tmplt_ops != NULL and
chain->tmplt_ops != tp->ops.

Fixes: 7d5509fa0d3d ("net: sched: extend proto ops with 'put' callback")
Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
Reviewed-by: Vlad Buslov <vladbu@nvidia.com>
Link: https://lore.kernel.org/r/20220921092734.31700-1-hbh25y@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: sunhme: Fix packet reception for len < RX_COPY_THRESHOLD

[ Upstream commit 878e2405710aacfeeb19364c300f38b7a9abfe8f ]

There is a separate receive path for small packets (under 256 bytes).
Instead of allocating a new dma-capable skb to be used for the next packet,
this path allocates a skb and copies the data into it (reusing the existing
sbk for the next packet). There are two bytes of junk data at the beginning
of every packet. I believe these are inserted in order to allow aligned DMA
and IP headers. We skip over them using skb_reserve. Before copying over
the data, we must use a barrier to ensure we see the whole packet. The
current code only synchronizes len bytes, starting from the beginning of
the packet, including the junk bytes. However, this leaves off the final
two bytes in the packet. Synchronize the whole packet.

To reproduce this problem, ping a HME with a payload size between 17 and
214

$ ping -s 17 <hme_address>

which will complain rather loudly about the data mismatch. Small packets
(below 60 bytes on the wire) do not have this issue. I suspect this is
related to the padding added to increase the minimum packet size.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Sean Anderson <seanga2@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20220920235018.1675956-1-seanga2@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bonding: fix NULL deref in bond_rr_gen_slave_id

[ Upstream commit 0e400d602f46360752e4b32ce842dba3808e15e6 ]

Fix a NULL dereference of the struct bonding.rr_tx_counter member because
if a bond is initially created with an initial mode != zero (Round Robin)
the memory required for the counter is never created and when the mode is
changed there is never any attempt to verify the memory is allocated upon
switching modes.

This causes the following Oops on an aarch64 machine:
    [  334.686773] Unable to handle kernel paging request at virtual address ffff2c91ac905000
    [  334.694703] Mem abort info:
    [  334.697486]   ESR = 0x0000000096000004
    [  334.701234]   EC = 0x25: DABT (current EL), IL = 32 bits
    [  334.706536]   SET = 0, FnV = 0
    [  334.709579]   EA = 0, S1PTW = 0
    [  334.712719]   FSC = 0x04: level 0 translation fault
    [  334.717586] Data abort info:
    [  334.720454]   ISV = 0, ISS = 0x00000004
    [  334.724288]   CM = 0, WnR = 0
    [  334.727244] swapper pgtable: 4k pages, 48-bit VAs, pgdp=000008044d662000
    [  334.733944] [ffff2c91ac905000] pgd=0000000000000000, p4d=0000000000000000
    [  334.740734] Internal error: Oops: 96000004 [#1] SMP
    [  334.745602] Modules linked in: bonding tls veth rfkill sunrpc arm_spe_pmu vfat fat acpi_ipmi ipmi_ssif ixgbe igb i40e mdio ipmi_devintf ipmi_msghandler arm_cmn arm_dsu_pmu cppc_cpufreq acpi_tad fuse zram crct10dif_ce ast ghash_ce sbsa_gwdt nvme drm_vram_helper drm_ttm_helper nvme_core ttm xgene_hwmon
    [  334.772217] CPU: 7 PID: 2214 Comm: ping Not tainted 6.0.0-rc4-00133-g64ae13ed4784 #4
    [  334.779950] Hardware name: GIGABYTE R272-P31-00/MP32-AR1-00, BIOS F18v (SCP: 1.08.20211002) 12/01/2021
    [  334.789244] pstate: 60400009 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
    [  334.796196] pc : bond_rr_gen_slave_id+0x40/0x124 [bonding]
    [  334.801691] lr : bond_xmit_roundrobin_slave_get+0x38/0xdc [bonding]
    [  334.807962] sp : ffff8000221733e0
    [  334.811265] x29: ffff8000221733e0 x28: ffffdbac8572d198 x27: ffff80002217357c
    [  334.818392] x26: 000000000000002a x25: ffffdbacb33ee000 x24: ffff07ff980fa000
    [  334.825519] x23: ffffdbacb2e398ba x22: ffff07ff98102000 x21: ffff07ff981029c0
    [  334.832646] x20: 0000000000000001 x19: ffff07ff981029c0 x18: 0000000000000014
    [  334.839773] x17: 0000000000000000 x16: ffffdbacb1004364 x15: 0000aaaabe2f5a62
    [  334.846899] x14: ffff07ff8e55d968 x13: ffff07ff8e55db30 x12: 0000000000000000
    [  334.854026] x11: ffffdbacb21532e8 x10: 0000000000000001 x9 : ffffdbac857178ec
    [  334.861153] x8 : ffff07ff9f6e5a28 x7 : 0000000000000000 x6 : 000000007c2b3742
    [  334.868279] x5 : ffff2c91ac905000 x4 : ffff2c91ac905000 x3 : ffff07ff9f554400
    [  334.875406] x2 : ffff2c91ac905000 x1 : 0000000000000001 x0 : ffff07ff981029c0
    [  334.882532] Call trace:
    [  334.884967]  bond_rr_gen_slave_id+0x40/0x124 [bonding]
    [  334.890109]  bond_xmit_roundrobin_slave_get+0x38/0xdc [bonding]
    [  334.896033]  __bond_start_xmit+0x128/0x3a0 [bonding]
    [  334.901001]  bond_start_xmit+0x54/0xb0 [bonding]
    [  334.905622]  dev_hard_start_xmit+0xb4/0x220
    [  334.909798]  __dev_queue_xmit+0x1a0/0x720
    [  334.913799]  arp_xmit+0x3c/0xbc
    [  334.916932]  arp_send_dst+0x98/0xd0
    [  334.920410]  arp_solicit+0xe8/0x230
    [  334.923888]  neigh_probe+0x60/0xb0
    [  334.927279]  __neigh_event_send+0x3b0/0x470
    [  334.931453]  neigh_resolve_output+0x70/0x90
    [  334.935626]  ip_finish_output2+0x158/0x514
    [  334.939714]  __ip_finish_output+0xac/0x1a4
    [  334.943800]  ip_finish_output+0x40/0xfc
    [  334.947626]  ip_output+0xf8/0x1a4
    [  334.950931]  ip_send_skb+0x5c/0x100
    [  334.954410]  ip_push_pending_frames+0x3c/0x60
    [  334.958758]  raw_sendmsg+0x458/0x6d0
    [  334.962325]  inet_sendmsg+0x50/0x80
    [  334.965805]  sock_sendmsg+0x60/0x6c
    [  334.969286]  __sys_sendto+0xc8/0x134
    [  334.972853]  __arm64_sys_sendto+0x34/0x4c
    [  334.976854]  invoke_syscall+0x78/0x100
    [  334.980594]  el0_svc_common.constprop.0+0x4c/0xf4
    [  334.985287]  do_el0_svc+0x38/0x4c
    [  334.988591]  el0_svc+0x34/0x10c
    [  334.991724]  el0t_64_sync_handler+0x11c/0x150
    [  334.996072]  el0t_64_sync+0x190/0x194
    [  334.999726] Code: b9001062 f9403c02 d53cd044 8b040042 (b8210040)
    [  335.005810] ---[ end trace 0000000000000000 ]---
    [  335.010416] Kernel panic - not syncing: Oops: Fatal exception in interrupt
    [  335.017279] SMP: stopping secondary CPUs
    [  335.021374] Kernel Offset: 0x5baca8eb0000 from 0xffff800008000000
    [  335.027456] PHYS_OFFSET: 0x80000000
    [  335.030932] CPU features: 0x0000,0085c029,19805c82
    [  335.035713] Memory Limit: none
    [  335.038756] Rebooting in 180 seconds..

The fix is to allocate the memory in bond_open() which is guaranteed
to be called before any packets are processed.

Fixes: 848ca9182a7d ("net: bonding: Use per-cpu rr_tx_counter")
CC: Jussi Maki <joamaki@gmail.com>
Signed-off-by: Jonathan Toppins <jtoppins@redhat.com>
Acked-by: Jay Vosburgh <jay.vosburgh@canonical.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net/smc: Stop the CLC flow if no link to map buffers on

[ Upstream commit e738455b2c6dcdab03e45d97de36476f93f557d2 ]

There might be a potential race between SMC-R buffer map and
link group termination.

smc_smcr_terminate_all()     | smc_connect_rdma()
--------------------------------------------------------------
                             | smc_conn_create()
for links in smcibdev        |
        schedule links down  |
                             | smc_buf_create()
                             |  \- smcr_buf_map_usable_links()
                             |      \- no usable links found,
                             |         (rmb->mr = NULL)
                             |
                             | smc_clc_send_confirm()
                             |  \- access conn->rmb_desc->mr[]->rkey
                             |     (panic)

During reboot and IB device module remove, all links will be set
down and no usable links remain in link groups. In such situation
smcr_buf_map_usable_links() should return an error and stop the
CLC flow accessing to uninitialized mr.

Fixes: b9247544c1bc ("net/smc: convert static link ID instances to support multiple links")
Signed-off-by: Wen Gu <guwen@linux.alibaba.com>
Link: https://lore.kernel.org/r/1663656189-32090-1-git-send-email-guwen@linux.alibaba.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/mediatek: dsi: Move mtk_dsi_stop() call back to mtk_dsi_poweroff()

[ Upstream commit 90144dd8b0d137d9e78ef34b3c418e51a49299ad ]

As the comment right before the mtk_dsi_stop() call advises,
mtk_dsi_stop() should only be called after
mtk_drm_crtc_atomic_disable(). That's because that function calls
drm_crtc_wait_one_vblank(), which requires the vblank irq to be enabled.

Previously mtk_dsi_stop(), being in mtk_dsi_poweroff() and guarded by a
refcount, would only be called at the end of
mtk_drm_crtc_atomic_disable(), through the call to mtk_crtc_ddp_hw_fini().
Commit cde7e2e35c28 ("drm/mediatek: Separate poweron/poweroff from
enable/disable and define new funcs") moved the mtk_dsi_stop() call to
mtk_output_dsi_disable(), causing it to be called before
mtk_drm_crtc_atomic_disable(), and consequently generating vblank
timeout warnings during suspend.

Move the mtk_dsi_stop() call back to mtk_dsi_poweroff() so that we have
a working vblank irq during mtk_drm_crtc_atomic_disable() and stop
getting vblank timeout warnings.

Fixes: cde7e2e35c28 ("drm/mediatek: Separate poweron/poweroff from enable/disable and define new funcs")
Signed-off-by: Nícolas F. R. A. Prado <nfraprado@collabora.com>
Tested-by: Hsin-Yi Wang <hsinyi@chromium.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Tested-by: Allen-KH Cheng <allen-kh.cheng@mediatek.com>
Link: http://lists.infradead.org/pipermail/linux-mediatek/2022-August/046713.html
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

perf tools: Honor namespace when synthesizing build-ids

[ Upstream commit 999e4eaa4b3691acf85d094836260ec4b66c74fd ]

It needs to enter the namespace before reading a file.

Fixes: 4183a8d70a288627 ("perf tools: Allow synthesizing the build id for kernel/modules/tasks in PERF_RECORD_MMAP2")
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/20220920222822.2171056-1-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

perf kcore_copy: Do not check /proc/modules is unchanged

[ Upstream commit 5b427df27b94aec1312cace48a746782a0925c53 ]

/proc/kallsyms and /proc/modules are compared before and after the copy
in order to ensure no changes during the copy.

However /proc/modules also might change due to reference counts changing
even though that does not make any difference.

Any modules loaded or unloaded should be visible in changes to kallsyms,
so it is not necessary to check /proc/modules also anyway.

Remove the comparison checking that /proc/modules is unchanged.

Fixes: fc1b691d7651d949 ("perf buildid-cache: Add ability to add kcore to the cache")
Reported-by: Daniel Dao <dqminh@cloudflare.com>
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Tested-by: Daniel Dao <dqminh@cloudflare.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Link: https://lore.kernel.org/r/20220914122429.8770-1-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

perf jit: Include program header in ELF files

[ Upstream commit babd04386b1df8c364cdaa39ac0e54349502e1e5 ]

The missing header makes it hard for programs like elfutils to open
these files.

Fixes: 2d86612aacb7805f ("perf symbol: Correct address for bss symbols")
Reviewed-by: Leo Yan <leo.yan@linaro.org>
Signed-off-by: Lieven Hey <lieven.hey@kdab.com>
Tested-by: Leo Yan <leo.yan@linaro.org>
Cc: Leo Yan <leo.yan@linaro.org>
Link: https://lore.kernel.org/r/20220915092910.711036-1-lieven.hey@kdab.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

perf stat: Fix BPF program section name

[ Upstream commit 0d77326c3369e255715ed2440a78894ccc98dd69 ]

It seems the recent libbpf got more strict about the section name.
I'm seeing a failure like this:

  $ sudo ./perf stat -a --bpf-counters --for-each-cgroup ^. sleep 1
  libbpf: prog 'on_cgrp_switch': missing BPF prog type, check ELF section name 'perf_events'
  libbpf: prog 'on_cgrp_switch': failed to load: -22
  libbpf: failed to load object 'bperf_cgroup_bpf'
  libbpf: failed to load BPF skeleton 'bperf_cgroup_bpf': -22
  Failed to load cgroup skeleton

The section name should be 'perf_event' (without the trailing 's').
Although it's related to the libbpf change, it'd be better fix the
section name in the first place.

Fixes: 944138f048f7d759 ("perf stat: Enable BPF counter with --for-each-cgroup")
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: bpf@vger.kernel.org
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/r/20220916184132.1161506-2-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

can: gs_usb: gs_can_open(): fix race dev->can.state condition

[ Upstream commit 5440428b3da65408dba0241985acb7a05258b85e ]

The dev->can.state is set to CAN_STATE_ERROR_ACTIVE, after the device
has been started. On busy networks the CAN controller might receive
CAN frame between and go into an error state before the dev->can.state
is assigned.

Assign dev->can.state before starting the controller to close the race
window.

Fixes: d08e973a77d1 ("can: gs_usb: Added support for the GS_USB CAN devices")
Link: https://lore.kernel.org/all/20220920195216.232481-1-mkl@pengutronix.de
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: sh_eth: Fix PHY state warning splat during system resume

[ Upstream commit 6a1dbfefdae4f7809b3e277cc76785dac0ac1cd0 ]

Since commit 744d23c71af39c7d ("net: phy: Warn about incorrect
mdio_bus_phy_resume() state"), a warning splat is printed during system
resume with Wake-on-LAN disabled:

WARNING: CPU: 0 PID: 626 at drivers/net/phy/phy_device.c:323 mdio_bus_phy_resume+0xbc/0xe4

As the Renesas SuperH Ethernet driver already calls phy_{stop,start}()
in its suspend/resume callbacks, it is sufficient to just mark the MAC
responsible for managing the power state of the PHY.

Fixes: fba863b816049b03 ("net: phy: make PHY PM ops a no-op if MAC driver manages PHY PM")
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Link: https://lore.kernel.org/r/c6e1331b9bef61225fa4c09db3ba3e2e7214ba2d.1663598886.git.geert+renesas@glider.be
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: ravb: Fix PHY state warning splat during system resume

[ Upstream commit 4924c0cdce75575295f8fa682851fb8e5d619dd2 ]

Since commit 744d23c71af39c7d ("net: phy: Warn about incorrect
mdio_bus_phy_resume() state"), a warning splat is printed during system
resume with Wake-on-LAN disabled:

        WARNING: CPU: 0 PID: 1197 at drivers/net/phy/phy_device.c:323 mdio_bus_phy_resume+0xbc/0xc8

As the Renesas Ethernet AVB driver already calls phy_{stop,start}() in
its suspend/resume callbacks, it is sufficient to just mark the MAC
responsible for managing the power state of the PHY.

Fixes: fba863b816049b03 ("net: phy: make PHY PM ops a no-op if MAC driver manages PHY PM")
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Link: https://lore.kernel.org/r/8ec796f47620980fdd0403e21bd8b7200b4fa1d4.1663598796.git.geert+renesas@glider.be
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: ebtables: fix memory leak when blob is malformed

[ Upstream commit 62ce44c4fff947eebdf10bb582267e686e6835c9 ]

The bug fix was incomplete, it "replaced" crash with a memory leak.
The old code had an assignment to "ret" embedded into the conditional,
restore this.

Fixes: 7997eff82828 ("netfilter: ebtables: reject blobs that don't provide all entry points")
Reported-and-tested-by: syzbot+a24c5252f3e3ab733464@syzkaller.appspotmail.com
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: nf_tables: fix percpu memory leak at nf_tables_addchain()

[ Upstream commit 9a4d6dd554b86e65581ef6b6638a39ae079b17ac ]

It seems to me that percpu memory for chain stats started leaking since
commit 3bc158f8d0330f0a ("netfilter: nf_tables: map basechain priority to
hardware priority") when nft_chain_offload_priority() returned an error.

Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Fixes: 3bc158f8d0330f0a ("netfilter: nf_tables: map basechain priority to hardware priority")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: nf_tables: fix nft_counters_enabled underflow at nf_tables_addchain()

[ Upstream commit 921ebde3c0d22c8cba74ce8eb3cc4626abff1ccd ]

syzbot is reporting underflow of nft_counters_enabled counter at
nf_tables_addchain() [1], for commit 43eb8949cfdffa76 ("netfilter:
nf_tables: do not leave chain stats enabled on error") missed that
nf_tables_chain_destroy() after nft_basechain_init() in the error path of
nf_tables_addchain() decrements the counter because nft_basechain_init()
makes nft_is_base_chain() return true by setting NFT_CHAIN_BASE flag.

Increment the counter immediately after returning from
nft_basechain_init().

Link:  https://syzkaller.appspot.com/bug?extid=b5d82a651b71cd8a75ab [1]
Reported-by: syzbot <syzbot+b5d82a651b71cd8a75ab@syzkaller.appspotmail.com>
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Tested-by: syzbot <syzbot+b5d82a651b71cd8a75ab@syzkaller.appspotmail.com>
Fixes: 43eb8949cfdffa76 ("netfilter: nf_tables: do not leave chain stats enabled on error")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net/sched: taprio: make qdisc_leaf() see the per-netdev-queue pfifo child qdiscs

[ Upstream commit 1461d212ab277d8bba1a753d33e9afe03d81f9d4 ]

taprio can only operate as root qdisc, and to that end, there exists the
following check in taprio_init(), just as in mqprio:

if (sch->parent != TC_H_ROOT)
return -EOPNOTSUPP;

And indeed, when we try to attach taprio to an mqprio child, it fails as
expected:

$ tc qdisc add dev swp0 root handle 1: mqprio num_tc 8 \
map 0 1 2 3 4 5 6 7 \
queues 1@0 1@1 1@2 1@3 1@4 1@5 1@6 1@7 hw 0
$ tc qdisc replace dev swp0 parent 1:2 taprio num_tc 8 \
map 0 1 2 3 4 5 6 7 \
queues 1@0 1@1 1@2 1@3 1@4 1@5 1@6 1@7 \
base-time 0 sched-entry S 0x7f 990000 sched-entry S 0x80 100000 \
flags 0x0 clockid CLOCK_TAI
Error: sch_taprio: Can only be attached as root qdisc.

(extack message added by me)

But when we try to attach a taprio child to a taprio root qdisc,
surprisingly it doesn't fail:

$ tc qdisc replace dev swp0 root handle 1: taprio num_tc 8 \
map 0 1 2 3 4 5 6 7 queues 1@0 1@1 1@2 1@3 1@4 1@5 1@6 1@7 \
base-time 0 sched-entry S 0x7f 990000 sched-entry S 0x80 100000 \
flags 0x0 clockid CLOCK_TAI
$ tc qdisc replace dev swp0 parent 1:2 taprio num_tc 8 \
map 0 1 2 3 4 5 6 7 \
queues 1@0 1@1 1@2 1@3 1@4 1@5 1@6 1@7 \
base-time 0 sched-entry S 0x7f 990000 sched-entry S 0x80 100000 \
flags 0x0 clockid CLOCK_TAI

This is because tc_modify_qdisc() behaves differently when mqprio is
root, vs when taprio is root.

In the mqprio case, it finds the parent qdisc through
p = qdisc_lookup(dev, TC_H_MAJ(clid)), and then the child qdisc through
q = qdisc_leaf(p, clid). This leaf qdisc q has handle 0, so it is
ignored according to the comment right below ("It may be default qdisc,
ignore it"). As a result, tc_modify_qdisc() goes through the
qdisc_create() code path, and this gives taprio_init() a chance to check
for sch_parent != TC_H_ROOT and error out.

Whereas in the taprio case, the returned q = qdisc_leaf(p, clid) is
different. It is not the default qdisc created for each netdev queue
(both taprio and mqprio call qdisc_create_dflt() and keep them in
a private q->qdiscs[], or priv->qdiscs[], respectively). Instead, taprio
makes qdisc_leaf() return the _root_ qdisc, aka itself.

When taprio does that, tc_modify_qdisc() goes through the qdisc_change()
code path, because the qdisc layer never finds out about the child qdisc
of the root. And through the ->change() ops, taprio has no reason to
check whether its parent is root or not, just through ->init(), which is
not called.

The problem is the taprio_leaf() implementation. Even though code wise,
it does the exact same thing as mqprio_leaf() which it is copied from,
it works with different input data. This is because mqprio does not
attach itself (the root) to each device TX queue, but one of the default
qdiscs from its private array.

In fact, since commit 13511704f8d7 ("net: taprio offload: enforce qdisc
to netdev queue mapping"), taprio does this too, but just for the full
offload case. So if we tried to attach a taprio child to a fully
offloaded taprio root qdisc, it would properly fail too; just not to a
software root taprio.

To fix the problem, stop looking at the Qdisc that's attached to the TX
queue, and instead, always return the default qdiscs that we've
allocated (and to which we privately enqueue and dequeue, in software
scheduling mode).

Since Qdisc_class_ops :: leaf  is only called from tc_modify_qdisc(),
the risk of unforeseen side effects introduced by this change is
minimal.

Fixes: 5a781ccbd19e ("tc: Add support for configuring the taprio scheduler")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net/sched: taprio: avoid disabling offload when it was never enabled

[ Upstream commit db46e3a88a09c5cf7e505664d01da7238cd56c92 ]

In an incredibly strange API design decision, qdisc->destroy() gets
called even if qdisc->init() never succeeded, not exclusively since
commit 87b60cfacf9f ("net_sched: fix error recovery at qdisc creation"),
but apparently also earlier (in the case of qdisc_create_dflt()).

The taprio qdisc does not fully acknowledge this when it attempts full
offload, because it starts off with q->flags = TAPRIO_FLAGS_INVALID in
taprio_init(), then it replaces q->flags with TCA_TAPRIO_ATTR_FLAGS
parsed from netlink (in taprio_change(), tail called from taprio_init()).

But in taprio_destroy(), we call taprio_disable_offload(), and this
determines what to do based on FULL_OFFLOAD_IS_ENABLED(q->flags).

But looking at the implementation of FULL_OFFLOAD_IS_ENABLED()
(a bitwise check of bit 1 in q->flags), it is invalid to call this macro
on q->flags when it contains TAPRIO_FLAGS_INVALID, because that is set
to U32_MAX, and therefore FULL_OFFLOAD_IS_ENABLED() will return true on
an invalid set of flags.

As a result, it is possible to crash the kernel if user space forces an
error between setting q->flags = TAPRIO_FLAGS_INVALID, and the calling
of taprio_enable_offload(). This is because drivers do not expect the
offload to be disabled when it was never enabled.

The error that we force here is to attach taprio as a non-root qdisc,
but instead as child of an mqprio root qdisc:

$ tc qdisc add dev swp0 root handle 1: \
mqprio num_tc 8 map 0 1 2 3 4 5 6 7 \
queues 1@0 1@1 1@2 1@3 1@4 1@5 1@6 1@7 hw 0
$ tc qdisc replace dev swp0 parent 1:1 \
taprio num_tc 8 map 0 1 2 3 4 5 6 7 \
queues 1@0 1@1 1@2 1@3 1@4 1@5 1@6 1@7 base-time 0 \
sched-entry S 0x7f 990000 sched-entry S 0x80 100000 \
flags 0x0 clockid CLOCK_TAI
Unable to handle kernel paging request at virtual address fffffffffffffff8
[fffffffffffffff8] pgd=0000000000000000, p4d=0000000000000000
Internal error: Oops: 96000004 [#1] PREEMPT SMP
Call trace:
 taprio_dump+0x27c/0x310
 vsc9959_port_setup_tc+0x1f4/0x460
 felix_port_setup_tc+0x24/0x3c
 dsa_slave_setup_tc+0x54/0x27c
 taprio_disable_offload.isra.0+0x58/0xe0
 taprio_destroy+0x80/0x104
 qdisc_create+0x240/0x470
 tc_modify_qdisc+0x1fc/0x6b0
 rtnetlink_rcv_msg+0x12c/0x390
 netlink_rcv_skb+0x5c/0x130
 rtnetlink_rcv+0x1c/0x2c

Fix this by keeping track of the operations we made, and undo the
offload only if we actually did it.

I've added "bool offloaded" inside a 4 byte hole between "int clockid"
and "atomic64_t picos_per_byte". Now the first cache line looks like
below:

$ pahole -C taprio_sched net/sched/sch_taprio.o
struct taprio_sched {
        struct Qdisc * *           qdiscs;               /*     0     8 */
        struct Qdisc *             root;                 /*     8     8 */
        u32                        flags;                /*    16     4 */
        enum tk_offsets            tk_offset;            /*    20     4 */
        int                        clockid;              /*    24     4 */
        bool                       offloaded;            /*    28     1 */

        /* XXX 3 bytes hole, try to pack */

        atomic64_t                 picos_per_byte;       /*    32     0 */

        /* XXX 8 bytes hole, try to pack */

        spinlock_t                 current_entry_lock;   /*    40     0 */

        /* XXX 8 bytes hole, try to pack */

        struct sched_entry *       current_entry;        /*    48     8 */
        struct sched_gate_list *   oper_sched;           /*    56     8 */
        /* --- cacheline 1 boundary (64 bytes) --- */

Fixes: 9c66d1564676 ("taprio: Add support for hardware offloading")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: enetc: deny offload of tc-based TSN features on VF interfaces

[ Upstream commit 5641c751fe2f92d3d9e8a8e03c1263ac8caa0b42 ]

TSN features on the ENETC (taprio, cbs, gate, police) are configured
through a mix of command BD ring messages and port registers:
enetc_port_rd(), enetc_port_wr().

Port registers are a region of the ENETC memory map which are only
accessible from the PCIe Physical Function. They are not accessible from
the Virtual Functions.

Moreover, attempting to access these registers crashes the kernel:

$ echo 1 > /sys/bus/pci/devices/0000\:00\:00.0/sriov_numvfs
pci 0000:00:01.0: [1957:ef00] type 00 class 0x020001
fsl_enetc_vf 0000:00:01.0: Adding to iommu group 15
fsl_enetc_vf 0000:00:01.0: enabling device (0000 -> 0002)
fsl_enetc_vf 0000:00:01.0 eno0vf0: renamed from eth0
$ tc qdisc replace dev eno0vf0 root taprio num_tc 8 map 0 1 2 3 4 5 6 7 \
queues 1@0 1@1 1@2 1@3 1@4 1@5 1@6 1@7 base-time 0 \
sched-entry S 0x7f 900000 sched-entry S 0x80 100000 flags 0x2
Unable to handle kernel paging request at virtual address ffff800009551a08
Internal error: Oops: 96000007 [#1] PREEMPT SMP
pc : enetc_setup_tc_taprio+0x170/0x47c
lr : enetc_setup_tc_taprio+0x16c/0x47c
Call trace:
 enetc_setup_tc_taprio+0x170/0x47c
 enetc_setup_tc+0x38/0x2dc
 taprio_change+0x43c/0x970
 taprio_init+0x188/0x1e0
 qdisc_create+0x114/0x470
 tc_modify_qdisc+0x1fc/0x6c0
 rtnetlink_rcv_msg+0x12c/0x390

Split enetc_setup_tc() into separate functions for the PF and for the
VF drivers. Also remove enetc_qos.o from being included into
enetc-vf.ko, since it serves absolutely no purpose there.

Fixes: 34c6adf1977b ("enetc: Configure the Time-Aware Scheduler via tc-taprio offload")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Link: https://lore.kernel.org/r/20220916133209.3351399-2-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: enetc: move enetc_set_psfp() out of the common enetc_set_features()

[ Upstream commit fed38e64d9b99d65a36c0dbadc3d3f8ddd9ea030 ]

The VF netdev driver shouldn't respond to changes in the NETIF_F_HW_TC
flag; only PFs should. Moreover, TSN-specific code should go to
enetc_qos.c, which should not be included in the VF driver.

Fixes: 79e499829f3f ("net: enetc: add hw tc hw offload features for PSPF capability")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Link: https://lore.kernel.org/r/20220916133209.3351399-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wireguard: netlink: avoid variable-sized memcpy on sockaddr

[ Upstream commit 26c013108c12b94bc023bf19198a4300596c98b1 ]

Doing a variable-sized memcpy is slower, and the compiler isn't smart
enough to turn this into a constant-size assignment.

Further, Kees' latest fortified memcpy will actually bark, because the
destination pointer is type sockaddr, not explicitly sockaddr_in or
sockaddr_in6, so it thinks there's an overflow:

    memcpy: detected field-spanning write (size 28) of single field
    "&endpoint.addr" at drivers/net/wireguard/netlink.c:446 (size 16)

Fix this by just assigning by using explicit casts for each checked
case.

Fixes: e7096c131e51 ("net: WireGuard secure network tunnel")
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Reported-by: syzbot+a448cda4dba2dac50de5@syzkaller.appspotmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wireguard: ratelimiter: disable timings test by default

[ Upstream commit 684dec3cf45da2b0848298efae4adf3b2aeafeda ]

A previous commit tried to make the ratelimiter timings test more
reliable but in the process made it less reliable on other
configurations. This is an impossible problem to solve without
increasingly ridiculous heuristics. And it's not even a problem that
actually needs to be solved in any comprehensive way, since this is only
ever used during development. So just cordon this off with a DEBUG_
ifdef, just like we do for the trie's randomized tests, so it can be
enabled while hacking on the code, and otherwise disabled in CI. In the
process we also revert 151c8e499f47.

Fixes: 151c8e499f47 ("wireguard: ratelimiter: use hrtimer in selftest")
Fixes: e7096c131e51 ("net: WireGuard secure network tunnel")
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: ipa: properly limit modem routing table use

[ Upstream commit cf412ec333250cb82bafe57169204e14a9f1c2ac ]

IPA can route packets between IPA-connected entities.  The AP and
modem are currently the only such entities supported, and no routing
is required to transfer packets between them.

The number of entries in each routing table is fixed, and defined at
initialization time.  Some of these entries are designated for use
by the modem, and the rest are available for the AP to use.  The AP
sends a QMI message to the modem which describes (among other
things) information about routing table memory available for the
modem to use.

Currently the QMI initialization packet gives wrong information in
its description of routing tables.  What *should* be supplied is the
maximum index that the modem can use for the routing table memory
located at a given location.  The current code instead supplies the
total *number* of routing table entries.  Furthermore, the modem is
granted the entire table, not just the subset it's supposed to use.

This patch fixes this.  First, the ipa_mem_bounds structure is
generalized so its "end" field can be interpreted either as a final
byte offset, or a final array index.  Second, the IPv4 and IPv6
(non-hashed and hashed) table information fields in the QMI
ipa_init_modem_driver_req structure are changed to be ipa_mem_bounds
rather than ipa_mem_array structures.  Third, we set the "end" value
for each routing table to be the last index, rather than setting the
"count" to be the number of indices.  Finally, instead of allowing
the modem to use all of a routing table's memory, it is limited to
just the portion meant to be used by the modem.  In all versions of
IPA currently supported, that is IPA_ROUTE_MODEM_COUNT (8) entries.

Update a few comments for clarity.

Fixes: 530f9216a9537 ("soc: qcom: ipa: AP/modem communications")
Signed-off-by: Alex Elder <elder@linaro.org>
Link: https://lore.kernel.org/r/20220913204602.1803004-1-elder@linaro.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

of: mdio: Add of_node_put() when breaking out of for_each_xx

[ Upstream commit 1c48709e6d9d353acaaac1d8e33474756b121d78 ]

In of_mdiobus_register(), we should call of_node_put() for 'child'
escaped out of for_each_available_child_of_node().

Fixes: 66bdede495c7 ("of_mdio: Fix broken PHY IRQ in case of probe deferral")
Co-developed-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: Liang He <windhl@126.com>
Link: https://lore.kernel.org/r/20220913125659.3331969-1-windhl@126.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/hisilicon: Add depends on MMU

[ Upstream commit d8a79c03054911c375a2252627a429c9bc4615b6 ]

The Kconfig symbol depended on MMU but was dropped by the commit
acad3fe650a5 ("drm/hisilicon: Removed the dependency on the mmu")
because it already had as a dependency ARM64 that already selects MMU.

But later, commit a0f25a6bb319 ("drm/hisilicon/hibmc: Allow to be built
if COMPILE_TEST is enabled") allowed the driver to be built for non-ARM64
when COMPILE_TEST is set but that could lead to unmet direct dependencies
and linking errors.

Prevent a kconfig warning when MMU is not enabled by making
DRM_HISI_HIBMC depend on MMU.

WARNING: unmet direct dependencies detected for DRM_TTM
  Depends on [n]: HAS_IOMEM [=y] && DRM [=m] && MMU [=n]
  Selected by [m]:
  - DRM_TTM_HELPER [=m] && HAS_IOMEM [=y] && DRM [=m]
  - DRM_HISI_HIBMC [=m] && HAS_IOMEM [=y] && DRM [=m] && PCI [=y] && (ARM64 || COMPILE_TEST [=y])

Fixes: acad3fe650a5 ("drm/hisilicon: Removed the dependency on the mmu")
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Xinliang Liu <xinliang.liu@linaro.org>
Cc: Tian Tao <tiantao6@hisilicon.com>
Cc: John Stultz <jstultz@google.com>
Cc: Xinwei Kong <kong.kongxinwei@hisilicon.com>
Cc: Chen Feng <puck.chen@hisilicon.com>
Cc: Christian Koenig <christian.koenig@amd.com>
Cc: Huang Rui <ray.huang@amd.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220531025557.29593-1-rdunlap@infradead.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/hisilicon/hibmc: Allow to be built if COMPILE_TEST is enabled

[ Upstream commit a0f25a6bb319aa05e04dcf51707c97c2881b4f47 ]

The commit feeb07d0ca5a ("drm/hisilicon/hibmc: Make CONFIG_DRM_HISI_HIBMC
depend on ARM64") made the driver Kconfig symbol to depend on ARM64 since
it only supports that architecture and loading the module on others would
lead to incorrect video modes being used.

But it also prevented the driver to be built on other architectures which
is useful to have compile test coverage when doing subsystem wide changes.

Make the dependency instead to be (ARM64 || COMPILE_TEST), so the driver
is buildable when the CONFIG_COMPILE_TEST option is enabled.

Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20211216210936.3329977-1-javierm@redhat.com
Stable-dep-of: d8a79c030549 ("drm/hisilicon: Add depends on MMU")
Signed-off-by: Sasha Levin <sashal@kernel.org>

sfc: fix null pointer dereference in efx_hard_start_xmit

[ Upstream commit 0a242eb2913a4aa3d6fbdb86559f27628e9466f3 ]

Trying to get the channel from the tx_queue variable here is wrong
because we can only be here if tx_queue is NULL, so we shouldn't
dereference it. As the above comment in the code says, this is very
unlikely to happen, but it's wrong anyway so let's fix it.

I hit this issue because of a different bug that caused tx_queue to be
NULL. If that happens, this is the error message that we get here:
  BUG: unable to handle kernel NULL pointer dereference at 0000000000000020
  [...]
  RIP: 0010:efx_hard_start_xmit+0x153/0x170 [sfc]

Fixes: 12804793b17c ("sfc: decouple TXQ type from label")
Reported-by: Tianhao Zhao <tizhao@redhat.com>
Signed-off-by: Íñigo Huguet <ihuguet@redhat.com>
Acked-by: Edward Cree <ecree.xilinx@gmail.com>
Link: https://lore.kernel.org/r/20220914111135.21038-1-ihuguet@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

sfc: fix TX channel offset when using legacy interrupts

[ Upstream commit f232af4295653afa4ade3230462b3be15ad16419 ]

In legacy interrupt mode the tx_channel_offset was hardcoded to 1, but
that's not correct if efx_sepparate_tx_channels is false. In that case,
the offset is 0 because the tx queues are in the single existing channel
at index 0, together with the rx queue.

Without this fix, as soon as you try to send any traffic, it tries to
get the tx queues from an uninitialized channel getting these errors:
  WARNING: CPU: 1 PID: 0 at drivers/net/ethernet/sfc/tx.c:540 efx_hard_start_xmit+0x12e/0x170 [sfc]
  [...]
  RIP: 0010:efx_hard_start_xmit+0x12e/0x170 [sfc]
  [...]
  Call Trace:
   <IRQ>
   dev_hard_start_xmit+0xd7/0x230
   sch_direct_xmit+0x9f/0x360
   __dev_queue_xmit+0x890/0xa40
  [...]
  BUG: unable to handle kernel NULL pointer dereference at 0000000000000020
  [...]
  RIP: 0010:efx_hard_start_xmit+0x153/0x170 [sfc]
  [...]
  Call Trace:
   <IRQ>
   dev_hard_start_xmit+0xd7/0x230
   sch_direct_xmit+0x9f/0x360
   __dev_queue_xmit+0x890/0xa40
  [...]

Fixes: c308dfd1b43e ("sfc: fix wrong tx channel offset with efx_separate_tx_channels")
Reported-by: Tianhao Zhao <tizhao@redhat.com>
Signed-off-by: Íñigo Huguet <ihuguet@redhat.com>
Acked-by: Edward Cree <ecree.xilinx@gmail.com>
Link: https://lore.kernel.org/r/20220914103648.16902-1-ihuguet@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

i40e: Fix set max_tx_rate when it is lower than 1 Mbps

[ Upstream commit 198eb7e1b81d8ba676d0f4f120c092032ae69a8e ]

While converting max_tx_rate from bytes to Mbps, this value was set to 0,
if the original value was lower than 125000 bytes (1 Mbps). This would
cause no transmission rate limiting to occur. This happened due to lack of
check of max_tx_rate against the 1 Mbps value for max_tx_rate and the
following division by 125000. Fix this issue by adding a helper
i40e_bw_bytes_to_mbits() which sets max_tx_rate to minimum usable value of
50 Mbps, if its value is less than 1 Mbps, otherwise do the required
conversion by dividing by 125000.

Fixes: 5ecae4120a6b ("i40e: Refactor VF BW rate limiting")
Signed-off-by: Michal Jaron <michalx.jaron@intel.com>
Signed-off-by: Andrii Staikov <andrii.staikov@intel.com>
Tested-by: Bharathi Sreenivas <bharathi.sreenivas@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

i40e: Fix VF set max MTU size

[ Upstream commit 372539def2824c43b6afe2403045b140f65c5acc ]

Max MTU sent to VF is set to 0 during memory allocation. It cause
that max MTU on VF is changed to IAVF_MAX_RXBUFFER and does not
depend on data from HW.

Set max_mtu field in virtchnl_vf_resource struct to inform
VF in GET_VF_RESOURCES msg what size should be max frame.

Fixes: dab86afdbbd1 ("i40e/i40evf: Change the way we limit the maximum frame size for Rx")
Signed-off-by: Michal Jaron <michalx.jaron@intel.com>
Signed-off-by: Mateusz Palczewski <mateusz.palczewski@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iavf: Fix set max MTU size with port VLAN and jumbo frames

[ Upstream commit 399c98c4dc50b7eb7e9f24da7ffdda6f025676ef ]

After setting port VLAN and MTU to 9000 on VF with ice driver there
was an iavf error
"PF returned error -5 (IAVF_ERR_PARAM) to our request 6".

During queue configuration, VF's max packet size was set to
IAVF_MAX_RXBUFFER but on ice max frame size was smaller by VLAN_HLEN
due to making some space for port VLAN as VF is not aware whether it's
in a port VLAN. This mismatch in sizes caused ice to reject queue
configuration with ERR_PARAM error. Proper max_mtu is sent from ice PF
to VF with GET_VF_RESOURCES msg but VF does not look at this.

In iavf change max_frame from IAVF_MAX_RXBUFFER to max_mtu
received from pf with GET_VF_RESOURCES msg to make vf's
max_frame_size dependent from pf. Add check if received max_mtu is
not in eligible range then set it to IAVF_MAX_RXBUFFER.

Fixes: dab86afdbbd1 ("i40e/i40evf: Change the way we limit the maximum frame size for Rx")
Signed-off-by: Michal Jaron <michalx.jaron@intel.com>
Signed-off-by: Mateusz Palczewski <mateusz.palczewski@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mlxbf_gige: clear MDIO gateway lock after read

[ Upstream commit 182447b12144b7be9b63a273d27c5a11bd54960a ]

The MDIO gateway (GW) lock in BlueField-2 GIGE logic is
set after read.  This patch adds logic to make sure the
lock is always cleared at the end of each MDIO transaction.

Fixes: f92e1869d74e ("Add Mellanox BlueField Gigabit Ethernet driver")
Reviewed-by: Asmaa Mnebhi <asmaa@nvidia.com>
Signed-off-by: David Thompson <davthompson@nvidia.com>
Link: https://lore.kernel.org/r/20220902164247.19862-1-davthompson@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iavf: Fix bad page state

[ Upstream commit 66039eb9015eee4f7ff0c99b83c65c7ecb3c8190 ]

Fix bad page state, free inappropriate page in handling dummy
descriptor. iavf_build_skb now has to check not only if rx_buffer is
NULL but also if size is zero, same thing in iavf_clean_rx_irq.
Without this patch driver would free page that will be used
by napi_build_skb.

Fixes: a9f49e006030 ("iavf: Fix handling of dummy receive descriptors")
Signed-off-by: Norbert Zulinski <norbertx.zulinski@intel.com>
Signed-off-by: Mateusz Palczewski <mateusz.palczewski@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

um: fix default console kernel parameter

[ Upstream commit 782b1f70f8a8b28571949d2ba43fe88b96d75ec3 ]

OpenWrt's UML with 5.15 was producing odd errors/warnings during preinit
part of the early userspace portion:

|[    0.000000] Kernel command line: ubd0=root.img root=98:0 console=tty
|[...]
|[    0.440000] random: jshn: uninitialized urandom read (4 bytes read)
|[    0.460000] random: jshn: uninitialized urandom read (4 bytes read)
|/etc/preinit: line 47: can't create /dev/tty: No such device or address
|/etc/preinit: line 48: can't create /dev/tty: No such device or address
|/etc/preinit: line 58: can't open /dev/tty: No such device or address
|[...] repeated many times

That "/dev/tty" came from the command line (which is automatically
added if no console= parameter was specified for the uml binary).

The TLDP project tells the following about the /dev/tty:
<https://tldp.org/HOWTO/Text-Terminal-HOWTO-7.html#ss7.3>
| /dev/tty stands for the controlling terminal (if any) for the current
| process.[...]
| /dev/tty is something like a link to the actually terminal device[..]

The "(if any)" is important here, since it's possible for processes to
not have a controlling terminal.

I think this was a simple typo and the author wanted tty0 there.

CC: Thomas Meyer <thomas@m3y3r.de>
Fixes: d7ffac33631b ("um: stdio_console: Make preferred console")
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
Signed-off-by: Richard Weinberger <richard@nod.at>
Signed-off-by: Sasha Levin <sashal@kernel.org>

MIPS: Loongson32: Fix PHY-mode being left unspecified

[ Upstream commit e9f3f8f488005f6da3cfb66070706770ecaef747 ]

commit 0060c8783330 ("net: stmmac: implement support for passive mode
converters via dt") has changed the plat->interface field semantics from
containing the PHY-mode to specifying the MAC-PCS interface mode. Due to
that the loongson32 platform code will leave the phylink interface
uninitialized with the PHY-mode intended by the means of the actual
platform setup. The commit-author most likely has just missed the
arch-specific code to fix. Let's mend the Loongson32 platform code then by
assigning the PHY-mode to the phy_interface field of the STMMAC platform
data.

Fixes: 0060c8783330 ("net: stmmac: implement support for passive mode converters via dt")
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Keguang Zhang <keguang.zhang@gmail.com>
Tested-by: Keguang Zhang <keguang.zhang@gmail.com>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

MIPS: lantiq: export clk_get_io() for lantiq_wdt.ko

[ Upstream commit 502550123bee6a2ffa438409b5b9aad4d6db3a8c ]

The lantiq WDT driver uses clk_get_io(), which is not exported,
so export it to fix a build error:

ERROR: modpost: "clk_get_io" [drivers/watchdog/lantiq_wdt.ko] undefined!

Fixes: 287e3f3f4e68 ("MIPS: lantiq: implement support for clkdev api")
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Reported-by: kernel test robot <lkp@intel.com>
Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Cc: John Crispin <john@phrozen.org>
Cc: linux-mips@vger.kernel.org
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/panel: simple: Fix innolux_g121i1_l01 bus_format

[ Upstream commit a7c48a0ab87ae52c087d663e83e56b8225ac4cce ]

innolux_g121i1_l01 sets bpc to 6, so use the corresponding bus format:
MEDIA_BUS_FMT_RGB666_1X7X3_SPWG.

Fixes: 4ae13e486866 ("drm/panel: simple: Add more properties to Innolux G121I1-L01")
Signed-off-by: Heiko Schocher <hs@denx.de>
Signed-off-by: Fabio Estevam <festevam@denx.de>
Signed-off-by: Marek Vasut <marex@denx.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20220826165021.1592532-1-festevam@denx.de
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: team: Unsync device addresses on ndo_stop

[ Upstream commit bd60234222b2fd5573526da7bcd422801f271f5f ]

Netdev drivers are expected to call dev_{uc,mc}_sync() in their
ndo_set_rx_mode method and dev_{uc,mc}_unsync() in their ndo_stop method.
This is mentioned in the kerneldoc for those dev_* functions.

The team driver calls dev_{uc,mc}_unsync() during ndo_uninit instead of
ndo_stop. This is ineffective because address lists (dev->{uc,mc}) have
already been emptied in unregister_netdevice_many() before ndo_uninit is
called. This mistake can result in addresses being leftover on former team
ports after a team device has been deleted; see test_LAG_cleanup() in the
last patch in this series.

Add unsync calls at their expected location, team_close().

v3:
* When adding or deleting a port, only sync/unsync addresses if the team
  device is up. In other cases, it is taken care of at the right time by
  ndo_open/ndo_set_rx_mode/ndo_stop.

Fixes: 3d249d4ca7d0 ("net: introduce ethernet teaming device")
Signed-off-by: Benjamin Poirier <bpoirier@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: bonding: Unsync device addresses on ndo_stop

[ Upstream commit 86247aba599e5b07d7e828e6edaaebb0ef2b1158 ]

Netdev drivers are expected to call dev_{uc,mc}_sync() in their
ndo_set_rx_mode method and dev_{uc,mc}_unsync() in their ndo_stop method.
This is mentioned in the kerneldoc for those dev_* functions.

The bonding driver calls dev_{uc,mc}_unsync() during ndo_uninit instead of
ndo_stop. This is ineffective because address lists (dev->{uc,mc}) have
already been emptied in unregister_netdevice_many() before ndo_uninit is
called. This mistake can result in addresses being leftover on former bond
slaves after a bond has been deleted; see test_LAG_cleanup() in the last
patch in this series.

Add unsync calls, via bond_hw_addr_flush(), at their expected location,
bond_close().
Add dev_mc_add() call to bond_open() to match the above change.

v3:
* When adding or deleting a slave, only sync/unsync, add/del addresses if
  the bond is up. In other cases, it is taken care of at the right time by
  ndo_open/ndo_set_rx_mode/ndo_stop.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Benjamin Poirier <bpoirier@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: bonding: Share lacpdu_mcast_addr definition

[ Upstream commit 1d9a143ee3408349700f44a9197b7ae0e4faae5d ]

There are already a few definitions of arrays containing
MULTICAST_LACPDU_ADDR and the next patch will add one more use. These all
contain the same constant data so define one common instance for all
bonding code.

Signed-off-by: Benjamin Poirier <bpoirier@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Stable-dep-of: 86247aba599e ("net: bonding: Unsync device addresses on ndo_stop")
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: mpt3sas: Fix return value check of dma_get_required_mask()

[ Upstream commit e0e0747de0ea3dd87cdbb0393311e17471a9baf1 ]

Fix the incorrect return value check of dma_get_required_mask().  Due to
this incorrect check, the driver was always setting the DMA mask to 63 bit.

Link: https://lore.kernel.org/r/20220913120538.18759-2-sreekanth.reddy@broadcom.com
Fixes: ba27c5cf286d ("scsi: mpt3sas: Don't change the DMA coherent mask after allocations")
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: Fix memory leak in __qlt_24xx_handle_abts()

[ Upstream commit 601be20fc6a1b762044d2398befffd6bf236cebf ]

Commit 8f394da36a36 ("scsi: qla2xxx: Drop TARGET_SCF_LOOKUP_LUN_FROM_TAG")
made the __qlt_24xx_handle_abts() function return early if
tcm_qla2xxx_find_cmd_by_tag() didn't find a command, but it missed to clean
up the allocated memory for the management command.

Link: https://lore.kernel.org/r/20220914024924.695604-1-rafaelmendsr@gmail.com
Fixes: 8f394da36a36 ("scsi: qla2xxx: Drop TARGET_SCF_LOOKUP_LUN_FROM_TAG")
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Rafael Mendonca <rafaelmendsr@gmail.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: phy: aquantia: wait for the suspend/resume operations to finish

[ Upstream commit ca2dccdeeb49a7e408112d681bf447984c845292 ]

The Aquantia datasheet notes that after issuing a Processor-Intensive
MDIO operation, like changing the low-power state of the device, the
driver should wait for the operation to finish before issuing a new MDIO
command.

The new aqr107_wait_processor_intensive_op() function is added which can
be used after these kind of MDIO operations. At the moment, we are only
adding it at the end of the suspend/resume calls.

The issue was identified on a board featuring the AQR113C PHY, on
which commands like 'ip link (..) up / down' issued without any delays
between them would render the link on the PHY to remain down.
The issue was easy to reproduce with a one-liner:
 $ ip link set dev ethX down; ip link set dev ethX up; \
 ip link set dev ethX down; ip link set dev ethX up;

Fixes: ac9e81c230eb ("net: phy: aquantia: add suspend / resume callbacks for AQR107 family")
Signed-off-by: Ioana Ciornei <ioana.ciornei@nxp.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20220906130451.1483448-1-ioana.ciornei@nxp.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: core: fix flow symmetric hash

[ Upstream commit 64ae13ed478428135cddc2f1113dff162d8112d4 ]

__flow_hash_consistentify() wrongly swaps ipv4 addresses in few cases.
This function is indirectly used by __skb_get_hash_symmetric(), which is
used to fanout packets in AF_PACKET.
Intrusion detection systems may be impacted by this issue.

__flow_hash_consistentify() computes the addresses difference then swaps
them if the difference is negative. In few cases src - dst and dst - src
are both negative.

The following snippet mimics __flow_hash_consistentify():

```
 #include <stdio.h>
 #include <stdint.h>

 int main(int argc, char** argv) {

     int diffs_d, diffd_s;
     uint32_t dst  = 0xb225a8c0; /* 178.37.168.192 --> 192.168.37.178 */
     uint32_t src  = 0x3225a8c0; /*  50.37.168.192 --> 192.168.37.50  */
     uint32_t dst2 = 0x3325a8c0; /*  51.37.168.192 --> 192.168.37.51  */

     diffs_d = src - dst;
     diffd_s = dst - src;

     printf("src:%08x dst:%08x, diff(s-d)=%d(0x%x) diff(d-s)=%d(0x%x)\n",
             src, dst, diffs_d, diffs_d, diffd_s, diffd_s);

     diffs_d = src - dst2;
     diffd_s = dst2 - src;

     printf("src:%08x dst:%08x, diff(s-d)=%d(0x%x) diff(d-s)=%d(0x%x)\n",
             src, dst2, diffs_d, diffs_d, diffd_s, diffd_s);

     return 0;
 }
```

Results:

src:3225a8c0 dst:b225a8c0, \
    diff(s-d)=-2147483648(0x80000000) \
    diff(d-s)=-2147483648(0x80000000)

src:3225a8c0 dst:3325a8c0, \
    diff(s-d)=-16777216(0xff000000) \
    diff(d-s)=16777216(0x1000000)

In the first case the addresses differences are always < 0, therefore
__flow_hash_consistentify() always swaps, thus dst->src and src->dst
packets have differents hashes.

Fixes: c3f8324188fa8 ("net: Add full IPv6 addresses to flow_keys")
Signed-off-by: Ludovic Cintrat <ludovic.cintrat@gatewatcher.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ipvlan: Fix out-of-bound bugs caused by unset skb->mac_header

[ Upstream commit 81225b2ea161af48e093f58e8dfee6d705b16af4 ]

If an AF_PACKET socket is used to send packets through ipvlan and the
default xmit function of the AF_PACKET socket is changed from
dev_queue_xmit() to packet_direct_xmit() via setsockopt() with the option
name of PACKET_QDISC_BYPASS, the skb->mac_header may not be reset and
remains as the initial value of 65535, this may trigger slab-out-of-bounds
bugs as following:

=================================================================
UG: KASAN: slab-out-of-bounds in ipvlan_xmit_mode_l2+0xdb/0x330 [ipvlan]
PU: 2 PID: 1768 Comm: raw_send Kdump: loaded Not tainted 6.0.0-rc4+ #6
ardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-1.fc33
all Trace:
print_address_description.constprop.0+0x1d/0x160
print_report.cold+0x4f/0x112
kasan_report+0xa3/0x130
ipvlan_xmit_mode_l2+0xdb/0x330 [ipvlan]
ipvlan_start_xmit+0x29/0xa0 [ipvlan]
__dev_direct_xmit+0x2e2/0x380
packet_direct_xmit+0x22/0x60
packet_snd+0x7c9/0xc40
sock_sendmsg+0x9a/0xa0
__sys_sendto+0x18a/0x230
__x64_sys_sendto+0x74/0x90
do_syscall_64+0x3b/0x90
entry_SYSCALL_64_after_hwframe+0x63/0xcd

The root cause is:
  1. packet_snd() only reset skb->mac_header when sock->type is SOCK_RAW
     and skb->protocol is not specified as in packet_parse_headers()

  2. packet_direct_xmit() doesn't reset skb->mac_header as dev_queue_xmit()

In this case, skb->mac_header is 65535 when ipvlan_xmit_mode_l2() is
called. So when ipvlan_xmit_mode_l2() gets mac header with eth_hdr() which
use "skb->head + skb->mac_header", out-of-bound access occurs.

This patch replaces eth_hdr() with skb_eth_hdr() in ipvlan_xmit_mode_l2()
and reset mac header in multicast to solve this out-of-bound bug.

Fixes: 2ad7bf363841 ("ipvlan: Initial check-in of the IPVLAN driver.")
Signed-off-by: Lu Wei <luwei32@huawei.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iavf: Fix cached head and tail value for iavf_get_tx_pending

[ Upstream commit 809f23c0423a43266e47a7dc67e95b5cb4d1cbfc ]

The underlying hardware may or may not allow reading of the head or tail
registers and it really makes no difference if we use the software
cached values. So, always used the software cached values.

Fixes: 9c6c12595b73 ("i40e: Detection and recovery of TX queue hung logic moved to service_task from tx_timeout")
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Co-developed-by: Norbert Zulinski <norbertx.zulinski@intel.com>
Signed-off-by: Norbert Zulinski <norbertx.zulinski@intel.com>
Signed-off-by: Mateusz Palczewski <mateusz.palczewski@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ice: Don't double unplug aux on peer initiated reset

[ Upstream commit 23c619190318376769ad7b61504c2ea0703fb783 ]

In the IDC callback that is accessed when the aux drivers request a reset,
the function to unplug the aux devices is called.  This function is also
called in the ice_prepare_for_reset function. This double call is causing
a "scheduling while atomic" BUG.

[  662.676430] ice 0000:4c:00.0 rocep76s0: cqp opcode = 0x1 maj_err_code = 0xffff min_err_code = 0x8003

[  662.676609] ice 0000:4c:00.0 rocep76s0: [Modify QP Cmd Error][op_code=8] status=-29 waiting=1 completion_err=1 maj=0xffff min=0x8003

[  662.815006] ice 0000:4c:00.0 rocep76s0: ICE OICR event notification: oicr = 0x10000003

[  662.815014] ice 0000:4c:00.0 rocep76s0: critical PE Error, GLPE_CRITERR=0x00011424

[  662.815017] ice 0000:4c:00.0 rocep76s0: Requesting a reset

[  662.815475] BUG: scheduling while atomic: swapper/37/0/0x00010002

[  662.815475] BUG: scheduling while atomic: swapper/37/0/0x00010002
[  662.815477] Modules linked in: rpcsec_gss_krb5 auth_rpcgss nfsv4 dns_resolver nfs lockd grace fscache netfs rfkill 8021q garp mrp stp llc vfat fat rpcrdma intel_rapl_msr intel_rapl_common sunrpc i10nm_edac rdma_ucm nfit ib_srpt libnvdimm ib_isert iscsi_target_mod x86_pkg_temp_thermal intel_powerclamp coretemp target_core_mod snd_hda_intel ib_iser snd_intel_dspcfg libiscsi snd_intel_sdw_acpi scsi_transport_iscsi kvm_intel iTCO_wdt rdma_cm snd_hda_codec kvm iw_cm ipmi_ssif iTCO_vendor_support snd_hda_core irqbypass crct10dif_pclmul crc32_pclmul ghash_clmulni_intel snd_hwdep snd_seq snd_seq_device rapl snd_pcm snd_timer isst_if_mbox_pci pcspkr isst_if_mmio irdma intel_uncore idxd acpi_ipmi joydev isst_if_common snd mei_me idxd_bus ipmi_si soundcore i2c_i801 mei ipmi_devintf i2c_smbus i2c_ismt ipmi_msghandler acpi_power_meter acpi_pad rv(OE) ib_uverbs ib_cm ib_core xfs libcrc32c ast i2c_algo_bit drm_vram_helper drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops drm_ttm_helpe
 r ttm
[  662.815546]  nvme nvme_core ice drm crc32c_intel i40e t10_pi wmi pinctrl_emmitsburg dm_mirror dm_region_hash dm_log dm_mod fuse
[  662.815557] Preemption disabled at:
[  662.815558] [<0000000000000000>] 0x0
[  662.815563] CPU: 37 PID: 0 Comm: swapper/37 Kdump: loaded Tainted: G S         OE     5.17.1 #2
[  662.815566] Hardware name: Intel Corporation D50DNP/D50DNP, BIOS SE5C6301.86B.6624.D18.2111021741 11/02/2021
[  662.815568] Call Trace:
[  662.815572]  <IRQ>
[  662.815574]  dump_stack_lvl+0x33/0x42
[  662.815581]  __schedule_bug.cold.147+0x7d/0x8a
[  662.815588]  __schedule+0x798/0x990
[  662.815595]  schedule+0x44/0xc0
[  662.815597]  schedule_preempt_disabled+0x14/0x20
[  662.815600]  __mutex_lock.isra.11+0x46c/0x490
[  662.815603]  ? __ibdev_printk+0x76/0xc0 [ib_core]
[  662.815633]  device_del+0x37/0x3d0
[  662.815639]  ice_unplug_aux_dev+0x1a/0x40 [ice]
[  662.815674]  ice_schedule_reset+0x3c/0xd0 [ice]
[  662.815693]  irdma_iidc_event_handler.cold.7+0xb6/0xd3 [irdma]
[  662.815712]  ? bitmap_find_next_zero_area_off+0x45/0xa0
[  662.815719]  ice_send_event_to_aux+0x54/0x70 [ice]
[  662.815741]  ice_misc_intr+0x21d/0x2d0 [ice]
[  662.815756]  __handle_irq_event_percpu+0x4c/0x180
[  662.815762]  handle_irq_event_percpu+0xf/0x40
[  662.815764]  handle_irq_event+0x34/0x60
[  662.815766]  handle_edge_irq+0x9a/0x1c0
[  662.815770]  __common_interrupt+0x62/0x100
[  662.815774]  common_interrupt+0xb4/0xd0
[  662.815779]  </IRQ>
[  662.815780]  <TASK>
[  662.815780]  asm_common_interrupt+0x1e/0x40
[  662.815785] RIP: 0010:cpuidle_enter_state+0xd6/0x380
[  662.815789] Code: 49 89 c4 0f 1f 44 00 00 31 ff e8 65 d7 95 ff 45 84 ff 74 12 9c 58 f6 c4 02 0f 85 64 02 00 00 31 ff e8 ae c5 9c ff fb 45 85 f6 <0f> 88 12 01 00 00 49 63 d6 4c 2b 24 24 48 8d 04 52 48 8d 04 82 49
[  662.815791] RSP: 0018:ff2c2c4f18edbe80 EFLAGS: 00000202
[  662.815793] RAX: ff280805df140000 RBX: 0000000000000002 RCX: 000000000000001f
[  662.815795] RDX: 0000009a52da2d08 RSI: ffffffff93f8240b RDI: ffffffff93f53ee7
[  662.815796] RBP: ff5e2bd11ff41928 R08: 0000000000000000 R09: 000000000002f8c0
[  662.815797] R10: 0000010c3f18e2cf R11: 000000000000000f R12: 0000009a52da2d08
[  662.815798] R13: ffffffff94ad7e20 R14: 0000000000000002 R15: 0000000000000000
[  662.815801]  cpuidle_enter+0x29/0x40
[  662.815803]  do_idle+0x261/0x2b0
[  662.815807]  cpu_startup_entry+0x19/0x20
[  662.815809]  start_secondary+0x114/0x150
[  662.815813]  secondary_startup_64_no_verify+0xd5/0xdb
[  662.815818]  </TASK>
[  662.815846] bad: scheduling from the idle thread!
[  662.815849] CPU: 37 PID: 0 Comm: swapper/37 Kdump: loaded Tainted: G S      W  OE     5.17.1 #2
[  662.815852] Hardware name: Intel Corporation D50DNP/D50DNP, BIOS SE5C6301.86B.6624.D18.2111021741 11/02/2021
[  662.815853] Call Trace:
[  662.815855]  <IRQ>
[  662.815856]  dump_stack_lvl+0x33/0x42
[  662.815860]  dequeue_task_idle+0x20/0x30
[  662.815863]  __schedule+0x1c3/0x990
[  662.815868]  schedule+0x44/0xc0
[  662.815871]  schedule_preempt_disabled+0x14/0x20
[  662.815873]  __mutex_lock.isra.11+0x3a8/0x490
[  662.815876]  ? __ibdev_printk+0x76/0xc0 [ib_core]
[  662.815904]  device_del+0x37/0x3d0
[  662.815909]  ice_unplug_aux_dev+0x1a/0x40 [ice]
[  662.815937]  ice_schedule_reset+0x3c/0xd0 [ice]
[  662.815961]  irdma_iidc_event_handler.cold.7+0xb6/0xd3 [irdma]
[  662.815979]  ? bitmap_find_next_zero_area_off+0x45/0xa0
[  662.815985]  ice_send_event_to_aux+0x54/0x70 [ice]
[  662.816011]  ice_misc_intr+0x21d/0x2d0 [ice]
[  662.816033]  __handle_irq_event_percpu+0x4c/0x180
[  662.816037]  handle_irq_event_percpu+0xf/0x40
[  662.816039]  handle_irq_event+0x34/0x60
[  662.816042]  handle_edge_irq+0x9a/0x1c0
[  662.816045]  __common_interrupt+0x62/0x100
[  662.816048]  common_interrupt+0xb4/0xd0
[  662.816052]  </IRQ>
[  662.816053]  <TASK>
[  662.816054]  asm_common_interrupt+0x1e/0x40
[  662.816057] RIP: 0010:cpuidle_enter_state+0xd6/0x380
[  662.816060] Code: 49 89 c4 0f 1f 44 00 00 31 ff e8 65 d7 95 ff 45 84 ff 74 12 9c 58 f6 c4 02 0f 85 64 02 00 00 31 ff e8 ae c5 9c ff fb 45 85 f6 <0f> 88 12 01 00 00 49 63 d6 4c 2b 24 24 48 8d 04 52 48 8d 04 82 49
[  662.816063] RSP: 0018:ff2c2c4f18edbe80 EFLAGS: 00000202
[  662.816065] RAX: ff280805df140000 RBX: 0000000000000002 RCX: 000000000000001f
[  662.816067] RDX: 0000009a52da2d08 RSI: ffffffff93f8240b RDI: ffffffff93f53ee7
[  662.816068] RBP: ff5e2bd11ff41928 R08: 0000000000000000 R09: 000000000002f8c0
[  662.816070] R10: 0000010c3f18e2cf R11: 000000000000000f R12: 0000009a52da2d08
[  662.816071] R13: ffffffff94ad7e20 R14: 0000000000000002 R15: 0000000000000000
[  662.816075]  cpuidle_enter+0x29/0x40
[  662.816077]  do_idle+0x261/0x2b0
[  662.816080]  cpu_startup_entry+0x19/0x20
[  662.816083]  start_secondary+0x114/0x150
[  662.816087]  secondary_startup_64_no_verify+0xd5/0xdb
[  662.816091]  </TASK>
[  662.816169] bad: scheduling from the idle thread!

The correct place to unplug the aux devices for a reset is in the
prepare_for_reset function, as this is a common place for all reset flows.
It also has built in protection from being called twice in a single reset
instance before the aux devices are replugged.

Fixes: f9f5301e7e2d4 ("ice: Register auxiliary device to provide RDMA")
Signed-off-by: Dave Ertman <david.m.ertman@intel.com>
Tested-by: Helena Anna Dubel <helena.anna.dubel@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: nfnetlink_osf: fix possible bogus match in nf_osf_find()

[ Upstream commit 559c36c5a8d730c49ef805a72b213d3bba155cc8 ]

nf_osf_find() incorrectly returns true on mismatch, this leads to
copying uninitialized memory area in nft_osf which can be used to leak
stale kernel stack data to userspace.

Fixes: 22c7652cdaa8 ("netfilter: nft_osf: Add version option support")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: nf_conntrack_irc: Tighten matching on DCC message

[ Upstream commit e8d5dfd1d8747b56077d02664a8838c71ced948e ]

CTCP messages should only be at the start of an IRC message, not
anywhere within it.

While the helper only decodes packes in the ORIGINAL direction, its
possible to make a client send a CTCP message back by empedding one into
a PING request.  As-is, thats enough to make the helper believe that it
saw a CTCP message.

Fixes: 869f37d8e48f ("[NETFILTER]: nf_conntrack/nf_nat: add IRC helper port")
Signed-off-by: David Leadbeater <dgl@dgl.cx>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: nf_conntrack_sip: fix ct_sip_walk_headers

[ Upstream commit 39aebedeaaa95757f5c1f2ddb5f43fdddbf478ca ]

ct_sip_next_header and ct_sip_get_header return an absolute
value of matchoff, not a shift from current dataoff.
So dataoff should be assigned matchoff, not incremented by it.

This issue can be seen in the scenario when there are multiple
Contact headers and the first one is using a hostname and other headers
use IP addresses. In this case, ct_sip_walk_headers will work as follows:

The first ct_sip_get_header call to will find the first Contact header
but will return -1 as the header uses a hostname. But matchoff will
be changed to the offset of this header. After that, dataoff should be
set to matchoff, so that the next ct_sip_get_header call find the next
Contact header. But instead of assigning dataoff to matchoff, it is
incremented by it, which is not correct, as matchoff is an absolute
value of the offset. So on the next call to the ct_sip_get_header,
dataoff will be incorrect, and the next Contact header may not be
found at all.

Fixes: 05e3ced297fe ("[NETFILTER]: nf_conntrack_sip: introduce SIP-URI parsing helper")
Signed-off-by: Igor Ryzhov <iryzhov@nfware.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: rockchip: Remove 'enable-active-low' from rk3399-puma

[ Upstream commit a994b34b9abb9c08ee09e835b4027ff2147f9d94 ]

The 'enable-active-low' property is not a valid one.

Only 'enable-active-high' is valid, and when this property is absent
the gpio regulator will act as active low by default.

Remove the invalid 'enable-active-low' property.

Fixes: 2c66fc34e945 ("arm64: dts: rockchip: add RK3399-Q7 (Puma) SoM")
Signed-off-by: Fabio Estevam <festevam@denx.de>
Link: https://lore.kernel.org/r/20220827175140.1696699-1-festevam@denx.de
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

dmaengine: ti: k3-udma-private: Fix refcount leak bug in of_xudma_dev_get()

[ Upstream commit f9fdb0b86f087c2b7f6c6168dd0985a3c1eda87e ]

We should call of_node_put() for the reference returned by
of_parse_phandle() in fail path or when it is not used anymore.
Here we only need to move the of_node_put() before the check.

Fixes: d70241913413 ("dmaengine: ti: k3-udma: Add glue layer for non DMAengine users")
Signed-off-by: Liang He <windhl@126.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@gmail.com>
Link: https://lore.kernel.org/r/20220720073234.1255474-1-windhl@126.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: rockchip: Set RK3399-Gru PCLK_EDP to 24 MHz

[ Upstream commit 8123437cf46ea5a0f6ca5cb3c528d8b6db97b9c2 ]

We've found the AUX channel to be less reliable with PCLK_EDP at a
higher rate (typically 25 MHz). This is especially important on systems
with PSR-enabled panels (like Gru-Kevin), since we make heavy, constant
use of AUX.

According to Rockchip, using any rate other than 24 MHz can cause
"problems between syncing the PHY an PCLK", which leads to all sorts of
unreliabilities around register operations.

Fixes: d67a38c5a623 ("arm64: dts: rockchip: move core edp from rk3399-kevin to shared chromebook")
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: zain wang <wzz@rock-chips.com>
Signed-off-by: Brian Norris <briannorris@chromium.org>
Link: https://lore.kernel.org/r/20220830131212.v2.1.I98d30623f13b785ca77094d0c0fd4339550553b6@changeid
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/mediatek: dsi: Add atomic {destroy,duplicate}_state, reset callbacks

[ Upstream commit eeda05b5e92f51d9a09646ecb493f0a1e872a6ef ]

Add callbacks for atomic_destroy_state, atomic_duplicate_state and
atomic_reset to restore functionality of the DSI driver: this solves
vblank timeouts when another bridge is present in the chain.

Tested bridge chain: DSI <=> ANX7625 => aux-bus panel

Fixes: 7f6335c6a258 ("drm/mediatek: Modify dsi funcs to atomic operations")
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Tested-by: Chen-Yu Tsai <wenst@chromium.org>
Reviewed-by: Nícolas F. R. A. Prado <nfraprado@collabora.com>
Tested-by: Nícolas F. R. A. Prado <nfraprado@collabora.com>
Link: https://patchwork.kernel.org/project/linux-mediatek/patch/20220721172727.14624-1-angelogioacchino.delregno@collabora.com/
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: rockchip: Fix typo in lisense text for PX30.Core

[ Upstream commit 4a00c43818dcc19be97250d4c3c4a1e2f1ed4f9d ]

Fix the Amarula Solutions typo mistake in lisense text added
in Engicam PX30.Core SoM dtsi.

Fixes: d92a7c331f53c ("arm64: dts: rockchip: Add Engicam PX30.Core SOM")
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Link: https://lore.kernel.org/r/20220822103524.266731-1-jagan@amarulasolutions.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: rockchip: Pull up wlan wake# on Gru-Bob

[ Upstream commit e5467359a725de90b6b8d0dd865500f6373828ca ]

The Gru-Bob board does not have a pull-up resistor on its
WLAN_HOST_WAKE# pin, but Kevin does. The production/vendor kernel
specified the pin configuration correctly as a pull-up, but this didn't
get ported correctly to upstream.

This means Bob's WLAN_HOST_WAKE# pin is floating, causing inconsistent
wakeup behavior.

Note that bt_host_wake_l has a similar dynamic, but apparently the
upstream choice was to redundantly configure both internal and external
pull-up on Kevin (see the "Kevin has an external pull up" comment in
rk3399-gru.dtsi). This doesn't cause any functional problem, although
it's perhaps wasteful.

Fixes: 8559bbeeb849 ("arm64: dts: rockchip: add Google Bob")
Signed-off-by: Brian Norris <briannorris@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20220822164453.1.I75c57b48b0873766ec993bdfb7bc1e63da5a1637@changeid
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

firmware: arm_scmi: Fix the asynchronous reset requests

[ Upstream commit b75c83d9b961fd3abf7310f8d36d5e6e9f573efb ]

SCMI Reset protocol specification allows the asynchronous reset request
only when an autonomous reset action is specified. Reset requests based
on explicit assert/deassert of signals should not be served
asynchronously.

Current implementation will instead issue an asynchronous request in any
case, as long as the reset domain had advertised to support asynchronous
resets.

Avoid requesting the asynchronous resets when the reset action is not
of the autonomous type, even if the target reset domain does, in general,
support the asynchronous requests.

Link: https://lore.kernel.org/r/20220817172731.1185305-6-cristian.marussi@arm.com
Fixes: 95a15d80aa0d ("firmware: arm_scmi: Add RESET protocol in SCMI v2.0")
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

firmware: arm_scmi: Harden accesses to the reset domains

[ Upstream commit e9076ffbcaed5da6c182b144ef9f6e24554af268 ]

Accessing reset domains descriptors by the index upon the SCMI drivers
requests through the SCMI reset operations interface can potentially
lead to out-of-bound violations if the SCMI driver misbehave.

Add an internal consistency check before any such domains descriptors
accesses.

Link: https://lore.kernel.org/r/20220817172731.1185305-5-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Stable-dep-of: b75c83d9b961 ("firmware: arm_scmi: Fix the asynchronous reset requests")
Signed-off-by: Sasha Levin <sashal@kernel.org>