A large tx latency issue was discovered during testing when only QBV was
enabled. The issue occurs because gtxoffset was not set when QBV is
active, it was only set when launch time is active.
The patch "igc: Correct the launchtime offset" only sets gtxoffset when
the launchtime_enable field is set by the user. Enabling launchtime_enable
ultimately sets the register IGC_TXQCTL_QUEUE_MODE_LAUNCHT (referred to as
LaunchT in the SW user manual).
Section 7.5.2.6 of the IGC i225/6 SW User Manual Rev 1.2.4 states:
"The latency between transmission scheduling (launch time) and the
time the packet is transmitted to the network is listed in Table 7-61."
However, the patch misinterprets the phrase "launch time" in that section
by assuming it specifically refers to the LaunchT register, whereas it
actually denotes the generic term for when a packet is released from the
internal buffer to the MAC transmit logic.
This launch time, as per that section, also implicitly refers to the QBV
gate open time, where a packet waits in the buffer for the QBV gate to
open. Therefore, latency applies whenever QBV is in use. TSN features such
as QBU and QAV reuse QBV, making the latency universal to TSN features.
Discussed with i226 HW owner (Shalev, Avi) and we were in agreement that
the term "launch time" used in Section 7.5.2.6 is not clear and can be
easily misinterpreted. Avi will update this section to:
"When TQAVCTRL.TRANSMIT_MODE = TSN, the latency between transmission
scheduling and the time the packet is transmitted to the network is listed
in Table 7-61."
Fix this issue by using igc_tsn_is_tx_mode_in_tsn() as a condition to
write to gtxoffset, aligning with the newly updated SW User Manual.
Tested:
1. Enrol taprio on talker board
base-time 0
cycle-time 1000000
flags 0x2
index 0 cmd S gatemask 0x1 interval1
index 0 cmd S gatemask 0x1 interval2
Note:
interval1 = interval for a 64 bytes packet to go through
interval2 = cycle-time - interval1
2. Take tcpdump on listener board
3. Use udp tai app on talker to send packets to listener
Following the "igc: Fix TX Hang issue when QBV Gate is close" changes,
remaining issues with the reset adapter logic in igc_tsn_offload_apply()
have been observed:
1. The reset adapter logics for i225 and i226 differ, although they should
be the same according to the guidelines in I225/6 HW Design Section
7.5.2.1 on software initialization during tx mode changes.
2. The i225 resets adapter every time, even though tx mode doesn't change.
This occurs solely based on the condition igc_is_device_id_i225() when
calling schedule_work().
3. i226 doesn't reset adapter for tsn->legacy tx mode changes. It only
resets adapter for legacy->tsn tx mode transitions.
4. qbv_count introduced in the patch is actually not needed; in this
context, a non-zero value of qbv_count is used to indicate if tx mode
was unconditionally set to tsn in igc_tsn_enable_offload(). This could
be replaced by checking the existing register
IGC_TQAVCTRL_TRANSMIT_MODE_TSN bit.
This patch resolves all issues and enters schedule_work() to reset the
adapter only when changing tx mode. It also removes reliance on qbv_count.
qbv_count field will be removed in a future patch.
Test ran:
1. Verify reset adapter behaviour in i225/6:
a) Enrol a new GCL
Reset adapter observed (tx mode change legacy->tsn)
b) Enrol a new GCL without deleting qdisc
No reset adapter observed (tx mode remain tsn->tsn)
c) Delete qdisc
Reset adapter observed (tx mode change tsn->legacy)
2. Tested scenario from "igc: Fix TX Hang issue when QBV Gate is closed"
to confirm it remains resolved.
Fixes: 175c241288c0 ("igc: Fix TX Hang issue when QBV Gate is closed") Signed-off-by: Faizal Rahim <faizal.abdul.rahim@linux.intel.com> Reviewed-by: Simon Horman <horms@kernel.org> Acked-by: Vinicius Costa Gomes <vinicius.gomes@intel.com> Tested-by: Mor Bar-Gabay <morx.bar.gabay@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When user issues these cmds:
1. Either a) or b)
a) mqprio with hardware offload disabled
b) taprio with txtime-assist feature enabled
2. etf
3. tc qdisc delete
4. taprio with base time in the past
At step 4, qbv_config_change_errors wrongly increased by 1.
Excerpt from IEEE 802.1Q-2018 8.6.9.3.1:
"If AdminBaseTime specifies a time in the past, and the current schedule
is running, then: Increment ConfigChangeError counter"
qbv_config_change_errors should only increase if base time is in the past
and no taprio is active. In user perspective, taprio was not active when
first triggered at step 4. However, i225/6 reuses qbv for etf, so qbv is
enabled with a dummy schedule at step 2 where it enters
igc_tsn_enable_offload() and qbv_count got incremented to 1. At step 4, it
enters igc_tsn_enable_offload() again, qbv_count is incremented to 2.
Because taprio is running, tc_setup_type is TC_SETUP_QDISC_ETF and
qbv_count > 1, qbv_config_change_errors value got incremented.
This issue happens due to reliance on qbv_count field where a non-zero
value indicates that taprio is running. But qbv_count increases
regardless if taprio is triggered by user or by other tsn feature. It does
not align with qbv_config_change_errors expectation where it is only
concerned with taprio triggered by user.
Fixing this by relocating the qbv_config_change_errors logic to
igc_save_qbv_schedule(), eliminating reliance on qbv_count and its
inaccuracies from i225/6's multiple uses of qbv feature for other TSN
features.
The new function created: igc_tsn_is_taprio_activated_by_user() uses
taprio_offload_enable field to indicate that the current running taprio
was triggered by user, instead of triggered by non-qbv feature like etf.
Fixes: ae4fe4698300 ("igc: Add qbv_config_change_errors counter") Signed-off-by: Faizal Rahim <faizal.abdul.rahim@linux.intel.com> Reviewed-by: Simon Horman <horms@kernel.org> Acked-by: Vinicius Costa Gomes <vinicius.gomes@intel.com> Tested-by: Mor Bar-Gabay <morx.bar.gabay@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Testing uncovered that even when the taprio gate is closed, some packets
still transmit.
According to i225/6 hardware errata [1], traffic might overflow the
planned QBV window. This happens because MAC maintains an internal buffer,
primarily for supporting half duplex retries. Therefore, even when the
gate closes, residual MAC data in the buffer may still transmit.
To mitigate this for i226, reduce the MAC's internal buffer from 192 bytes
to the recommended 88 bytes by modifying the RETX_CTL register value.
Note that for TC3, gate should open for 500us and close for another
500us.
3. Take tcpdump log on Board B.
4. Send udp packets via UDP tai app from Board A to Board B.
5. Analyze tcpdump log via wireshark log on Board B. Ensure that the
total time from the first to the last packet received during one cycle
for TC3 does not exceed 500us.
Fixes: 43546211738e ("igc: Add new device ID's") Signed-off-by: Faizal Rahim <faizal.abdul.rahim@linux.intel.com> Acked-by: Vinicius Costa Gomes <vinicius.gomes@intel.com> Tested-by: Mor Bar-Gabay <morx.bar.gabay@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
In __extent_writepage_io(), we call btrfs_set_range_writeback() ->
folio_start_writeback(), which clears PAGECACHE_TAG_DIRTY mark from the
mapping xarray if the folio is not dirty. This worked fine before commit 97713b1a2ced ("btrfs: do not clear page dirty inside
extent_write_locked_range()").
After the commit, however, the folio is still dirty at this point, so the
mapping DIRTY tag is not cleared anymore. Then, __extent_writepage_io()
calls btrfs_folio_clear_dirty() to clear the folio's dirty flag. That
results in the page being unlocked with a "strange" state. The page is not
PageDirty, but the mapping tag is set as PAGECACHE_TAG_DIRTY.
This strange state looks like causing a hang with a call trace below when
running fstests generic/091 on a null_blk device. It is waiting for a folio
lock.
While I don't have an exact relation between this hang and the strange
state, fixing the state also fixes the hang. And, that state is worth
fixing anyway.
This commit reorders btrfs_folio_clear_dirty() and
btrfs_set_range_writeback() in __extent_writepage_io(), so that the
PAGECACHE_TAG_DIRTY tag is properly removed from the xarray.
Daniel Hodges reported a kernel verifier crash when playing with sched-ext.
Further investigation shows that the crash is due to invalid memory access
in stacksafe(). More specifically, it is the following code:
The 'i' iterates old->allocated_stack.
If cur->allocated_stack < old->allocated_stack the out-of-bound
access will happen.
To fix the issue add 'i >= cur->allocated_stack' check such that if
the condition is true, stacksafe() should fail. Otherwise,
cur->stack[spi].slot_type[i % BPF_REG_SIZE] memory access is legal.
Fixes: 2793a8b015f7 ("bpf: exact states comparison for iterator convergence checks") Cc: Eduard Zingerman <eddyz87@gmail.com> Reported-by: Daniel Hodges <hodgesd@meta.com> Acked-by: Eduard Zingerman <eddyz87@gmail.com> Signed-off-by: Yonghong Song <yonghong.song@linux.dev> Link: https://lore.kernel.org/r/20240812214847.213612-1-yonghong.song@linux.dev Signed-off-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The commit f7866c358733 ("bpf: Fix null pointer dereference in resolve_prog_type() for BPF_PROG_TYPE_EXT")
fixed a NULL pointer dereference panic, but didn't fix the issue that
fails to update attached freplace prog to prog_array map.
Since commit 1c123c567fb1 ("bpf: Resolve fext program type when checking map compatibility"),
freplace prog and its target prog are able to tail call each other.
And the commit 3aac1ead5eb6 ("bpf: Move prog->aux->linked_prog and trampoline into bpf_link on attach")
sets prog->aux->dst_prog as NULL after attaching freplace prog to its
target prog.
After loading freplace the prog_array's owner type is BPF_PROG_TYPE_SCHED_CLS.
Then, after attaching freplace its prog->aux->dst_prog is NULL.
Then, while updating freplace in prog_array the bpf_prog_map_compatible()
incorrectly returns false because resolve_prog_type() returns
BPF_PROG_TYPE_EXT instead of BPF_PROG_TYPE_SCHED_CLS.
After this patch the resolve_prog_type() returns BPF_PROG_TYPE_SCHED_CLS
and update to prog_array can succeed.
Fixes: f7866c358733 ("bpf: Fix null pointer dereference in resolve_prog_type() for BPF_PROG_TYPE_EXT") Cc: Toke Høiland-Jørgensen <toke@redhat.com> Cc: Martin KaFai Lau <martin.lau@kernel.org> Acked-by: Yonghong Song <yonghong.song@linux.dev> Signed-off-by: Leon Hwang <leon.hwang@linux.dev> Link: https://lore.kernel.org/r/20240728114612.48486-2-leon.hwang@linux.dev Signed-off-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
After we switch tmpfs dir operations from simple_dir_operations to
simple_offset_dir_operations, every rename happened will fill new dentry
to dest dir's maple tree(&SHMEM_I(inode)->dir_offsets->mt) with a free
key starting with octx->newx_offset, and then set newx_offset equals to
free key + 1. This will lead to infinite readdir combine with rename
happened at the same time, which fail generic/736 in xfstests(detail show
as below).
1. create 5000 files(1 2 3...) under one dir
2. call readdir(man 3 readdir) once, and get one entry
3. rename(entry, "TEMPFILE"), then rename("TEMPFILE", entry)
4. loop 2~3, until readdir return nothing or we loop too many
times(tmpfs break test with the second condition)
We choose the same logic what commit 9b378f6ad48cf ("btrfs: fix infinite
directory reads") to fix it, record the last_index when we open dir, and
do not emit the entry which index >= last_index. The file->private_data
now used in offset dir can use directly to do this, and we also update
the last_index when we llseek the dir file.
Fixes: a2e459555c5f ("shmem: stable directory offsets") Signed-off-by: yangerkun <yangerkun@huawei.com> Link: https://lore.kernel.org/r/20240731043835.1828697-1-yangerkun@huawei.com Reviewed-by: Chuck Lever <chuck.lever@oracle.com>
[brauner: only update last_index after seek when offset is zero like Jan suggested] Signed-off-by: Christian Brauner <brauner@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When struct file_lease was split out from struct file_lock, the name of
the file_lock slab cache was copied to the new slab cache for
file_lease. This name conflict causes confusion in /proc/slabinfo and
/sys/kernel/slab. In particular, it caused failures in drgn's test case
for slab cache merging.
As in commit 4e527d5841e2 ("iomap: fault in smaller chunks for non-large
folio mappings"), we can see a performance loss for filesystems
which have not yet been converted to large folios.
Fixes: c38f4e96e605 ("netfs: Provide func to copy data to pagecache for buffered write") Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org> Link: https://lore.kernel.org/r/20240527201735.1898381-1-willy@infradead.org Reviewed-by: Jeff Layton <jlayton@kernel.org> Signed-off-by: Christian Brauner <brauner@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The return value uv_set_shared() and uv_remove_shared() (which are
wrappers around the share() function) is not always checked. The system
integrity of a protected guest depends on the Share and Unshare UVCs
being successful. This means that any caller that fails to check the
return value will compromise the security of the protected guest.
No code path that would lead to such violation of the security
guarantees is currently exercised, since all the areas that are shared
never get unshared during the lifetime of the system. This might
change and become an issue in the future.
The Share and Unshare UVCs can only fail in case of hypervisor
misbehaviour (either a bug or malicious behaviour). In such cases there
is no reasonable way forward, and the system needs to panic.
This patch replaces the return at the end of the share() function with
a panic, to guarantee system integrity.
Fixes: 5abb9351dfd9 ("s390/uv: introduce guest side ultravisor code") Signed-off-by: Claudio Imbrenda <imbrenda@linux.ibm.com> Reviewed-by: Christian Borntraeger <borntraeger@linux.ibm.com> Reviewed-by: Steffen Eiden <seiden@linux.ibm.com> Reviewed-by: Janosch Frank <frankja@linux.ibm.com> Link: https://lore.kernel.org/r/20240801112548.85303-1-imbrenda@linux.ibm.com
Message-ID: <20240801112548.85303-1-imbrenda@linux.ibm.com>
[frankja@linux.ibm.com: Fixed up patch subject] Signed-off-by: Janosch Frank <frankja@linux.ibm.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
It's currently possible to create pidfds for kthreads but it is unclear
what that is supposed to mean. Until we have use-cases for it and we
figured out what behavior we want block the creation of pidfds for
kthreads.
This needs to be set as well if the IB uses atomics.
Reviewed-by: Leo Liu <leo.liu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit c6c2e8b6a427d4fecc7c36cffccb908185afcab2) Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This needs to be set as well if the IB uses atomics.
Reviewed-by: Leo Liu <leo.liu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 35c628774e50b3784c59e8ca7973f03bcb067132) Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[why & how]
Cursor gets clipped off in the middle of the screen with hw
rotation 180. Fix a miscalculation of cursor offset when it's
placed near the edges in the pipe split case.
Cursor bugs with hw rotation were reported on AMD issue
tracker:
https://gitlab.freedesktop.org/drm/amd/-/issues/2247
The issues on rotation 270 was fixed by:
https://lore.kernel.org/amd-gfx/20221118125935.4013669-22-Brian.Chang@amd.com/
that partially addressed the rotation 180 too. So, this patch is the
final bits for rotation 180.
Reported-by: Xaver Hugl <xaver.hugl@gmail.com> Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/2247 Reviewed-by: Harry Wentland <harry.wentland@amd.com> Fixes: 9d84c7ef8a87 ("drm/amd/display: Correct cursor position on horizontal mirror") Signed-off-by: Melissa Wen <mwen@igalia.com> Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com> Signed-off-by: Tom Chung <chiahsuan.chung@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 1fd2cf090096af8a25bf85564341cfc21cec659d) Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[Why]
Tiled display cannot synchronize properly after S3.
The fix for commit 5f0c74915815 ("drm/amd/display: Fix for otg
synchronization logic") is not enable in DCN321, which causes
the otg is excluded from synchronization.
[How]
Enable otg synchronization logic in dcn321.
Fixes: 5f0c74915815 ("drm/amd/display: Fix for otg synchronization logic") Cc: Mario Limonciello <mario.limonciello@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Alvin Lee <alvin.lee2@amd.com> Signed-off-by: Loan Chen <lo-an.chen@amd.com> Signed-off-by: Tom Chung <chiahsuan.chung@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit d6ed53712f583423db61fbb802606759e023bf7b) Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To be able to get to the lowest power state when suspending systems with
DCN3.5+, we must be in IPS before the display hardware is put into
D3cold. So, to ensure that the system always reaches the lowest power
state while suspending, force systems that support IPS to enter idle
optimizations before entering D3cold.
Reviewed-by: Roman Li <roman.li@amd.com> Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 237193e21b29d4aa0617ffeea3d6f49e72999708) Cc: stable@vger.kernel.org # 6.10+ Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[why & how]
When the commit 9d84c7ef8a87 ("drm/amd/display: Correct cursor position
on horizontal mirror") was introduced, it used the wrong calculation for
the position copy for X. This commit uses the correct calculation for that
based on the original patch.
Fixes: 9d84c7ef8a87 ("drm/amd/display: Correct cursor position on horizontal mirror") Cc: Mario Limonciello <mario.limonciello@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Acked-by: Wayne Lin <wayne.lin@amd.com> Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com> Signed-off-by: Tom Chung <chiahsuan.chung@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 8f9b23abbae5ffcd64856facd26a86b67195bc2f) Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Although there are several patches improving the extent map shrinker,
there are still reports of too frequent shrinker behavior, taking too
much CPU for the kswapd process.
So let's only enable extent shrinker for now, until we got more
comprehensive understanding and a better solution.
[REPORT]
There is a corruption report that btrfs refused to mount a fs that has
overlapping dev extents:
BTRFS error (device sdc): dev extent devid 4 physical offset 14263979671552 overlap with previous dev extent end 14263980982272
BTRFS error (device sdc): failed to verify dev extents against chunks: -117
BTRFS error (device sdc): open_ctree failed
[CAUSE]
The direct cause is very obvious, there is a bad dev extent item with
incorrect length.
With btrfs check reporting two overlapping extents, the second one shows
some clue on the cause:
ERROR: dev extent devid 4 offset 14263979671552 len 6488064 overlap with previous dev extent end 14263980982272
ERROR: dev extent devid 13 offset 2257707008000 len 6488064 overlap with previous dev extent end 2257707270144
ERROR: errors found in extent allocation tree or chunk allocation
The second one looks like a bitflip happened during new chunk
allocation:
hex(2257707008000) = 0x20da9d30000
hex(2257707270144) = 0x20da9d70000
diff = 0x00000040000
So it looks like a bitflip happened during new dev extent allocation,
resulting the second overlap.
Currently we only do the dev-extent verification at mount time, but if the
corruption is caused by memory bitflip, we really want to catch it before
writing the corruption to the storage.
Furthermore the dev extent items has the following key definition:
(<device id> DEV_EXTENT <physical offset>)
Thus we can not just rely on the generic key order check to make sure
there is no overlapping.
[ENHANCEMENT]
Introduce dedicated dev extent checks, including:
- Fixed member checks
* chunk_tree should always be BTRFS_CHUNK_TREE_OBJECTID (3)
* chunk_objectid should always be
BTRFS_FIRST_CHUNK_CHUNK_TREE_OBJECTID (256)
- Alignment checks
* chunk_offset should be aligned to sectorsize
* length should be aligned to sectorsize
* key.offset should be aligned to sectorsize
- Overlap checks
If the previous key is also a dev-extent item, with the same
device id, make sure we do not overlap with the previous dev extent.
Reported: Stefan N <stefannnau@gmail.com> Link: https://lore.kernel.org/linux-btrfs/CA+W5K0rSO3koYTo=nzxxTm1-Pdu1HYgVxEpgJ=aGc7d=E8mGEg@mail.gmail.com/ CC: stable@vger.kernel.org # 5.10+ Reviewed-by: Anand Jain <anand.jain@oracle.com> Signed-off-by: Qu Wenruo <wqu@suse.com> Reviewed-by: David Sterba <dsterba@suse.com> Signed-off-by: David Sterba <dsterba@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
__btrfs_add_free_space_zoned() references and modifies bg's alloc_offset,
ro, and zone_unusable, but without taking the lock. It is mostly safe
because they monotonically increase (at least for now) and this function is
mostly called by a transaction commit, which is serialized by itself.
Still, taking the lock is a safer and correct option and I'm going to add a
change to reset zone_unusable while a block group is still alive. So, add
locking around the operations.
Fixes: 169e0da91a21 ("btrfs: zoned: track unusable bytes for zones") CC: stable@vger.kernel.org # 5.15+ Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com> Reviewed-by: David Sterba <dsterba@suse.com> Signed-off-by: David Sterba <dsterba@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently the extent map shrinker can be run by any task when attempting
to allocate memory and there's enough memory pressure to trigger it.
To avoid too much latency we stop iterating over extent maps and removing
them once the task needs to reschedule. This logic was introduced in commit b3ebb9b7e92a ("btrfs: stop extent map shrinker if reschedule is needed").
While that solved high latency problems for some use cases, it's still
not enough because with a too high number of tasks entering the extent map
shrinker code, either due to memory allocations or because they are a
kswapd task, we end up having a very high level of contention on some
spin locks, namely:
1) The fs_info->fs_roots_radix_lock spin lock, which we need to find
roots to iterate over their inodes;
2) The spin lock of the xarray used to track open inodes for a root
(struct btrfs_root::inodes) - on 6.10 kernels and below, it used to
be a red black tree and the spin lock was root->inode_lock;
3) The fs_info->delayed_iput_lock spin lock since the shrinker adds
delayed iputs (calls btrfs_add_delayed_iput()).
Instead of allowing the extent map shrinker to be run by any task, make
it run only by kswapd tasks. This still solves the problem of running
into OOM situations due to an unbounded extent map creation, which is
simple to trigger by direct IO writes, as described in the changelog
of commit 956a17d9d050 ("btrfs: add a shrinker for extent maps"), and
by a similar case when doing buffered IO on files with a very large
number of holes (keeping the file open and creating many holes, whose
extent maps are only released when the file is closed).
Reported-by: kzd <kzd@56709.net> Link: https://bugzilla.kernel.org/show_bug.cgi?id=219121 Reported-by: Octavia Togami <octavia.togami@gmail.com> Link: https://lore.kernel.org/linux-btrfs/CAHPNGSSt-a4ZZWrtJdVyYnJFscFjP9S7rMcvEMaNSpR556DdLA@mail.gmail.com/ Fixes: 956a17d9d050 ("btrfs: add a shrinker for extent maps") CC: stable@vger.kernel.org # 6.10+ Tested-by: kzd <kzd@56709.net> Tested-by: Octavia Togami <octavia.togami@gmail.com> Signed-off-by: Filipe Manana <fdmanana@suse.com> Reviewed-by: David Sterba <dsterba@suse.com> Signed-off-by: David Sterba <dsterba@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In the patch 78c52d9eb6b7 ("btrfs: check for refs on snapshot delete
resume") I added some code to handle file systems that had been
corrupted by a bug that incorrectly skipped updating the drop progress
key while dropping a snapshot. This code would check to see if we had
already deleted our reference for a child block, and skip the deletion
if we had already.
Unfortunately there is a bug, as the check would only check the on-disk
references. I made an incorrect assumption that blocks in an already
deleted snapshot that was having the deletion resume on mount wouldn't
be modified.
If we have 2 pending deleted snapshots that share blocks, we can easily
modify the rules for a block. Take the following example
subvolume a exists, and subvolume b is a snapshot of subvolume a. They
share references to block 1. Block 1 will have 2 full references, one
for subvolume a and one for subvolume b, and it belongs to subvolume a
(btrfs_header_owner(block 1) == subvolume a).
When deleting subvolume a, we will drop our full reference for block 1,
and because we are the owner we will drop our full reference for all of
block 1's children, convert block 1 to FULL BACKREF, and add a shared
reference to all of block 1's children.
Then we will start the snapshot deletion of subvolume b. We look up the
extent info for block 1, which checks delayed refs and tells us that
FULL BACKREF is set, so sets parent to the bytenr of block 1. However
because this is a resumed snapshot deletion, we call into
check_ref_exists(). Because check_ref_exists() only looks at the disk,
it doesn't find the shared backref for the child of block 1, and thus
returns 0 and we skip deleting the reference for the child of block 1
and continue. This orphans the child of block 1.
The fix is to lookup the delayed refs, similar to what we do in
btrfs_lookup_extent_info(). However we only care about whether the
reference exists or not. If we fail to find our reference on disk, go
look up the bytenr in the delayed refs, and if it exists look for an
existing ref in the delayed ref head. If that exists then we know we
can delete the reference safely and carry on. If it doesn't exist we
know we have to skip over this block.
This bug has existed since I introduced this fix, however requires
having multiple deleted snapshots pending when we unmount. We noticed
this in production because our shutdown path stops the container on the
system, which deletes a bunch of subvolumes, and then reboots the box.
This gives us plenty of opportunities to hit this issue. Looking at the
history we've seen this occasionally in production, but we had a big
spike recently thanks to faster machines getting jobs with multiple
subvolumes in the job.
Chris Mason wrote a reproducer which does the following
On the second loop this would fail consistently, with my patch it has
been running for hours and hasn't failed.
I also used dm-log-writes to capture the state of the failure so I could
debug the problem. Using the existing failure case to test my patch
validated that it fixes the problem.
Fixes: 78c52d9eb6b7 ("btrfs: check for refs on snapshot delete resume") CC: stable@vger.kernel.org # 5.4+ Reviewed-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Josef Bacik <josef@toxicpanda.com> Signed-off-by: David Sterba <dsterba@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If we a find that an extent is shared but its end offset is not sector
size aligned, then we don't clone it and issue write operations instead.
This is because the reflink (remap_file_range) operation does not allow
to clone unaligned ranges, except if the end offset of the range matches
the i_size of the source and destination files (and the start offset is
sector size aligned).
While this is not incorrect because send can only guarantee that a file
has the same data in the source and destination snapshots, it's not
optimal and generates confusion and surprising behaviour for users.
For example, running this test:
$ cat test.sh
#!/bin/bash
DEV=/dev/sdi
MNT=/mnt/sdi
mkfs.btrfs -f $DEV
mount $DEV $MNT
# Use a file size not aligned to any possible sector size.
file_size=$((1 * 1024 * 1024 + 5)) # 1MB + 5 bytes
dd if=/dev/random of=$MNT/foo bs=$file_size count=1
cp --reflink=always $MNT/foo $MNT/bar
[REPORT]
There is a bug report that kernel is rejecting a mismatching inode mode
and its dir item:
[ 1881.553937] BTRFS critical (device dm-0): inode mode mismatch with
dir: inode mode=040700 btrfs type=2 dir type=0
[CAUSE]
It looks like the inode mode is correct, while the dir item type
0 is BTRFS_FT_UNKNOWN, which should not be generated by btrfs at all.
This may be caused by a memory bit flip.
[ENHANCEMENT]
Although tree-checker is not able to do any cross-leaf verification, for
this particular case we can at least reject any dir type with
BTRFS_FT_UNKNOWN.
So here we enhance the dir type check from [0, BTRFS_FT_MAX), to
(0, BTRFS_FT_MAX).
Although the existing corruption can not be fixed just by such enhanced
checking, it should prevent the same 0x2->0x0 bitflip for dir type to
reach disk in the future.
When handling a numa page fault, task_numa_fault() should be called by a
process that restores the page table of the faulted folio to avoid
duplicated stats counting. Commit b99a342d4f11 ("NUMA balancing: reduce
TLB flush via delaying mapping on hint page fault") restructured
do_numa_page() and did not avoid task_numa_fault() call in the second page
table check after a numa migration failure. Fix it by making all
!pte_same() return immediately.
This issue can cause task_numa_fault() being called more than necessary
and lead to unexpected numa balancing results (It is hard to tell whether
the issue will cause positive or negative performance impact due to
duplicated numa fault counting).
Link: https://lkml.kernel.org/r/20240809145906.1513458-2-ziy@nvidia.com Fixes: b99a342d4f11 ("NUMA balancing: reduce TLB flush via delaying mapping on hint page fault") Signed-off-by: Zi Yan <ziy@nvidia.com> Reported-by: "Huang, Ying" <ying.huang@intel.com> Closes: https://lore.kernel.org/linux-mm/87zfqfw0yw.fsf@yhuang6-desk2.ccr.corp.intel.com/ Acked-by: David Hildenbrand <david@redhat.com> Cc: Baolin Wang <baolin.wang@linux.alibaba.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Yang Shi <shy828301@gmail.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The __vmap_pages_range_noflush() assumes its argument pages** contains
pages with the same page shift. However, since commit e9c3cda4d86e ("mm,
vmalloc: fix high order __GFP_NOFAIL allocations"), if gfp_flags includes
__GFP_NOFAIL with high order in vm_area_alloc_pages() and page allocation
failed for high order, the pages** may contain two different page shifts
(high order and order-0). This could lead __vmap_pages_range_noflush() to
perform incorrect mappings, potentially resulting in memory corruption.
Users might encounter this as follows (vmap_allow_huge = true, 2M is for
PMD_SIZE):
We can remove the fallback code because if a high-order allocation fails,
__vmalloc_node_range_noprof() will retry with order-0. Therefore, it is
unnecessary to fallback to order-0 here. Therefore, fix this by removing
the fallback code.
Link: https://lkml.kernel.org/r/20240808122019.3361-1-hailong.liu@oppo.com Fixes: e9c3cda4d86e ("mm, vmalloc: fix high order __GFP_NOFAIL allocations") Signed-off-by: Hailong Liu <hailong.liu@oppo.com> Reported-by: Tangquan Zheng <zhengtangquan@oppo.com> Reviewed-by: Baoquan He <bhe@redhat.com> Reviewed-by: Uladzislau Rezki (Sony) <urezki@gmail.com> Acked-by: Barry Song <baohua@kernel.org> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When handling a numa page fault, task_numa_fault() should be called by a
process that restores the page table of the faulted folio to avoid
duplicated stats counting. Commit c5b5a3dd2c1f ("mm: thp: refactor NUMA
fault handling") restructured do_huge_pmd_numa_page() and did not avoid
task_numa_fault() call in the second page table check after a numa
migration failure. Fix it by making all !pmd_same() return immediately.
This issue can cause task_numa_fault() being called more than necessary
and lead to unexpected numa balancing results (It is hard to tell whether
the issue will cause positive or negative performance impact due to
duplicated numa fault counting).
Link: https://lkml.kernel.org/r/20240809145906.1513458-3-ziy@nvidia.com Fixes: c5b5a3dd2c1f ("mm: thp: refactor NUMA fault handling") Reported-by: "Huang, Ying" <ying.huang@intel.com> Closes: https://lore.kernel.org/linux-mm/87zfqfw0yw.fsf@yhuang6-desk2.ccr.corp.intel.com/ Signed-off-by: Zi Yan <ziy@nvidia.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Baolin Wang <baolin.wang@linux.alibaba.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Yang Shi <shy828301@gmail.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In several cases we are freeing pages which were not allocated using
common page allocators. For such cases, in order to keep allocation
accounting correct, we should clear the page tag to indicate that the page
being freed is expected to not have a valid allocation tag. Introduce
clear_page_tag_ref() helper function to be used for this.
Link: https://lkml.kernel.org/r/20240813150758.855881-1-surenb@google.com Fixes: d224eb0287fb ("codetag: debug: mark codetags for reserved pages as empty") Signed-off-by: Suren Baghdasaryan <surenb@google.com> Suggested-by: David Hildenbrand <david@redhat.com> Acked-by: David Hildenbrand <david@redhat.com> Reviewed-by: Pasha Tatashin <pasha.tatashin@soleen.com> Cc: Kees Cook <keescook@chromium.org> Cc: Kent Overstreet <kent.overstreet@linux.dev> Cc: Sourav Panda <souravpanda@google.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: <stable@vger.kernel.org> [6.10] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[1] mentions that memfd_secret is only supported on arm64, riscv, x86 and
x86_64 for now. It doesn't support other architectures. I found the
build error on arm and decided to send the fix as it was creating noise on
KernelCI:
memfd_secret.c: In function 'memfd_secret':
memfd_secret.c:42:24: error: '__NR_memfd_secret' undeclared (first use in this function);
did you mean 'memfd_secret'?
42 | return syscall(__NR_memfd_secret, flags);
| ^~~~~~~~~~~~~~~~~
| memfd_secret
Hence I'm adding condition that memfd_secret should only be compiled on
supported architectures.
Also check in run_vmtests script if memfd_secret binary is present before
executing it.
The memory_failure_cpu structure is a per-cpu structure. Access to its
content requires the use of get_cpu_var() to lock in the current CPU and
disable preemption. The use of a regular spinlock_t for locking purpose
is fine for a non-RT kernel.
Since the integration of RT spinlock support into the v5.15 kernel, a
spinlock_t in a RT kernel becomes a sleeping lock and taking a sleeping
lock in a preemption disabled context is illegal resulting in the
following kind of warning.
Fix it by using a raw_spinlock_t for locking instead.
Also move the pr_err() out of the lock critical section and after
put_cpu_ptr() to avoid indeterminate latency and the possibility of sleep
with this call.
[longman@redhat.com: don't hold percpu ref across pr_err(), per Miaohe] Link: https://lkml.kernel.org/r/20240807181130.1122660-1-longman@redhat.com Link: https://lkml.kernel.org/r/20240806164107.1044956-1-longman@redhat.com Fixes: 0f383b6dc96e ("locking/spinlock: Provide RT variant") Signed-off-by: Waiman Long <longman@redhat.com> Acked-by: Miaohe Lin <linmiaohe@huawei.com> Cc: "Huang, Ying" <ying.huang@intel.com> Cc: Juri Lelli <juri.lelli@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Naoya Horiguchi <nao.horiguchi@gmail.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
During CMA activation, pages in CMA area are prepared and then freed
without being allocated. This triggers warnings when memory allocation
debug config (CONFIG_MEM_ALLOC_PROFILING_DEBUG) is enabled. Fix this by
marking these pages not tagged before freeing them.
Link: https://lkml.kernel.org/r/20240813150758.855881-2-surenb@google.com Fixes: d224eb0287fb ("codetag: debug: mark codetags for reserved pages as empty") Signed-off-by: Suren Baghdasaryan <surenb@google.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: Kees Cook <keescook@chromium.org> Cc: Kent Overstreet <kent.overstreet@linux.dev> Cc: Pasha Tatashin <pasha.tatashin@soleen.com> Cc: Sourav Panda <souravpanda@google.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: <stable@vger.kernel.org> [6.10] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When avc_add_xperms_decision() fails, the information recorded by the new
avc node is incomplete. In this case, the new avc node should be released
instead of replacing the old avc node.
Cc: stable@vger.kernel.org Fixes: fa1aa143ac4a ("selinux: extended permissions for ioctls") Suggested-by: Stephen Smalley <stephen.smalley.work@gmail.com> Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com> Acked-by: Stephen Smalley <stephen.smalley.work@gmail.com> Signed-off-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The count increases only when a node is successfully added to
the linked list.
Cc: stable@vger.kernel.org Fixes: fa1aa143ac4a ("selinux: extended permissions for ioctls") Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com> Acked-by: Stephen Smalley <stephen.smalley.work@gmail.com> Signed-off-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This happened because fscache_cookie_state_machine() was slow and was
still running while another process invoked fscache_unuse_cookie();
this led to a fscache_cookie_lru_do_one() call, setting the
FSCACHE_COOKIE_DO_LRU_DISCARD flag, which was picked up by
fscache_cookie_state_machine(), withdrawing the cookie via
cachefiles_withdraw_cookie(), clearing cookie->cache_priv.
At the same time, yet another process invoked
cachefiles_prepare_write(), which found a NULL pointer in this code
line:
During cachefiles_prepare_write(), the "n_accesses" counter is
non-zero (via fscache_begin_operation()). The cookie must not be
withdrawn until it drops to zero.
The counter is checked by fscache_cookie_state_machine() before
switching to FSCACHE_COOKIE_STATE_RELINQUISHING and
FSCACHE_COOKIE_STATE_WITHDRAWING (in "case
FSCACHE_COOKIE_STATE_FAILED"), but not for
FSCACHE_COOKIE_STATE_LRU_DISCARDING ("case
FSCACHE_COOKIE_STATE_ACTIVE").
This patch adds the missing check. With a non-zero access counter,
the function returns and the next fscache_end_cookie_access() call
will queue another fscache_cookie_state_machine() call to handle the
still-pending FSCACHE_COOKIE_DO_LRU_DISCARD.
Fixes: 12bb21a29c19 ("fscache: Implement cookie user counting and resource pinning") Signed-off-by: Max Kellermann <max.kellermann@ionos.com> Signed-off-by: David Howells <dhowells@redhat.com> Link: https://lore.kernel.org/r/20240729162002.3436763-2-dhowells@redhat.com
cc: Jeff Layton <jlayton@kernel.org>
cc: netfs@lists.linux.dev
cc: linux-fsdevel@vger.kernel.org
cc: stable@vger.kernel.org Signed-off-by: Christian Brauner <brauner@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
wpa_supplicant 2.11 sends since 1efdba5fdc2c ("Handle PMKSA flush in the
driver for SAE/OWE offload cases") SSID based PMKSA del commands.
brcmfmac is not prepared and tries to dereference the NULL bssid and
pmkid pointers in cfg80211_pmksa. PMKID_V3 operations support SSID based
updates so copy the SSID.
Fixes: a96202acaea4 ("wifi: brcmfmac: cfg80211: Add support for PMKID_V3 operations") Cc: stable@vger.kernel.org # 6.4.x Signed-off-by: Janne Grunau <j@jannau.net> Reviewed-by: Neal Gompa <neal@gompa.dev> Acked-by: Arend van Spriel <arend.vanspriel@broadcom.com> Signed-off-by: Kalle Valo <kvalo@kernel.org> Link: https://patch.msgid.link/20240803-brcmfmac_pmksa_del_ssid-v1-1-4e85f19135e1@jannau.net Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
After napi_complete_done() is called when NAPI is polling in the current
process context, another NAPI may be scheduled and start running in
softirq on another CPU and may ring the doorbell before the current CPU
does. When combined with unnecessary rings when there is no need to arm
the CQ, it triggers error paths in the hardware.
This patch fixes this by calling napi_complete_done() after doorbell
rings. It limits the number of unnecessary rings when there is
no need to arm. MANA hardware specifies that there must be one doorbell
ring every 8 CQ wraparounds. This driver guarantees one doorbell ring as
soon as the number of consumed CQEs exceeds 4 CQ wraparounds. In practical
workloads, the 4 CQ wraparounds proves to be big enough that it rarely
exceeds this limit before all the napi weight is consumed.
To implement this, add a per-CQ counter cq->work_done_since_doorbell,
and make sure the CQ is armed as soon as passing 4 wraparounds of the CQ.
Cc: stable@vger.kernel.org Fixes: e1b5683ff62e ("net: mana: Move NAPI from EQ to CQ") Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com> Signed-off-by: Long Li <longli@microsoft.com> Link: https://patch.msgid.link/1723219138-29887-1-git-send-email-longli@linuxonhyperv.com Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit a0821ca14bb8 ("media: atomisp: Remove test pattern generator (TPG)
support") broke BYT support because it removed a seemingly unused field
from struct sh_css_sp_config and a seemingly unused value from enum
ia_css_input_mode.
But these are part of the ABI between the kernel and firmware on ISP2400
and this part of the TPG support removal changes broke ISP2400 support.
ISP2401 support was not affected because on ISP2401 only a part of
struct sh_css_sp_config is used.
Restore the removed field and enum value to fix this.
Fixes: a0821ca14bb8 ("media: atomisp: Remove test pattern generator (TPG) support") Cc: stable@vger.kernel.org Signed-off-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The MANA driver's RX buffer alloc_size is passed into napi_build_skb() to
create SKB. skb_shinfo(skb) is located at the end of skb, and its alignment
is affected by the alloc_size passed into napi_build_skb(). The size needs
to be aligned properly for better performance and atomic operations.
Otherwise, on ARM64 CPU, for certain MTU settings like 4000, atomic
operations may panic on the skb_shinfo(skb)->dataref due to alignment fault.
To fix this bug, add proper alignment to the alloc_size calculation.
Cc: stable@vger.kernel.org Fixes: 80f6215b450e ("net: mana: Add support for jumbo frame") Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com> Reviewed-by: Long Li <longli@microsoft.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
read_balance() will avoid reading from slow disks as much as possible,
however, if valid data only lands in slow disks, and a new normal disk
is still in recovery, unrecovered data can be read:
raid1_read_request
read_balance
raid1_should_read_first
-> return false
choose_best_rdev
-> normal disk is not recovered, return -1
choose_bb_rdev
-> missing the checking of recovery, return the normal disk
-> read unrecovered data
Root cause is that the checking of recovery is missing in
choose_bb_rdev(). Hence add such checking to fix the problem.
Also fix similar problem in choose_slow_rdev().
Cc: stable@vger.kernel.org Fixes: 9f3ced792203 ("md/raid1: factor out choose_bb_rdev() from read_balance()") Fixes: dfa8ecd167c1 ("md/raid1: factor out choose_slow_rdev() from read_balance()") Reported-and-tested-by: Mateusz Jończyk <mat.jonczyk@o2.pl> Closes: https://lore.kernel.org/all/9952f532-2554-44bf-b906-4880b2e88e3a@o2.pl/ Signed-off-by: Yu Kuai <yukuai3@huawei.com> Link: https://lore.kernel.org/r/20240803091137.3197008-1-yukuai1@huaweicloud.com Signed-off-by: Song Liu <song@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We recently made GUP's common page table walking code to also walk hugetlb
VMAs without most hugetlb special-casing, preparing for the future of
having less hugetlb-specific page table walking code in the codebase.
Turns out that we missed one page table locking detail: page table locking
for hugetlb folios that are not mapped using a single PMD/PUD.
Assume we have hugetlb folio that spans multiple PTEs (e.g., 64 KiB
hugetlb folios on arm64 with 4 KiB base page size). GUP, as it walks the
page tables, will perform a pte_offset_map_lock() to grab the PTE table
lock.
However, hugetlb that concurrently modifies these page tables would
actually grab the mm->page_table_lock: with USE_SPLIT_PTE_PTLOCKS, the
locks would differ. Something similar can happen right now with hugetlb
folios that span multiple PMDs when USE_SPLIT_PMD_PTLOCKS.
This issue can be reproduced [1], for example triggering:
Let's make huge_pte_lockptr() effectively use the same PT locks as any
core-mm page table walker would. Add ptep_lockptr() to obtain the PTE
page table lock using a pte pointer -- unfortunately we cannot convert
pte_lockptr() because virt_to_page() doesn't work with kmap'ed page tables
we can have with CONFIG_HIGHPTE.
Handle CONFIG_PGTABLE_LEVELS correctly by checking in reverse order, such
that when e.g., CONFIG_PGTABLE_LEVELS==2 with
PGDIR_SIZE==P4D_SIZE==PUD_SIZE==PMD_SIZE will work as expected. Document
why that works.
There is one ugly case: powerpc 8xx, whereby we have an 8 MiB hugetlb
folio being mapped using two PTE page tables. While hugetlb wants to take
the PMD table lock, core-mm would grab the PTE table lock of one of both
PTE page tables. In such corner cases, we have to make sure that both
locks match, which is (fortunately!) currently guaranteed for 8xx as it
does not support SMP and consequently doesn't use split PT locks.
Link: https://lkml.kernel.org/r/20240801204748.99107-1-david@redhat.com Fixes: 9cb28da54643 ("mm/gup: handle hugetlb in the generic follow_page_mask code") Signed-off-by: David Hildenbrand <david@redhat.com> Acked-by: Peter Xu <peterx@redhat.com> Reviewed-by: Baolin Wang <baolin.wang@linux.alibaba.com> Tested-by: Baolin Wang <baolin.wang@linux.alibaba.com> Cc: Peter Xu <peterx@redhat.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Muchun Song <muchun.song@linux.dev> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Unaccepted memory is considered unusable free memory, which is not counted
as free on the zone watermark check. This causes get_page_from_freelist()
to accept more memory to hit the high watermark, but it creates problems
in the reclaim path.
The reclaim path encounters a failed zone watermark check and attempts to
reclaim memory. This is usually successful, but if there is little or no
reclaimable memory, it can result in endless reclaim with little to no
progress. This can occur early in the boot process, just after start of
the init process when the only reclaimable memory is the page cache of the
init executable and its libraries.
Make unaccepted memory free from watermark check point of view. This way
unaccepted memory will never be the trigger of memory reclaim. Accept
more memory in the get_page_from_freelist() if needed.
Link: https://lkml.kernel.org/r/20240809114854.3745464-2-kirill.shutemov@linux.intel.com Fixes: dcdfdd40fa82 ("mm: Add support for unaccepted memory") Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Reported-by: Jianxiong Gao <jxgao@google.com> Acked-by: David Hildenbrand <david@redhat.com> Tested-by: Jianxiong Gao <jxgao@google.com> Cc: Borislav Petkov <bp@alien8.de> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Mike Rapoport (Microsoft) <rppt@kernel.org> Cc: Tom Lendacky <thomas.lendacky@amd.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: <stable@vger.kernel.org> [6.5+] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
is_madv_discard did its check wrong. MADV_ flags are not bitwise,
they're normal sequential numbers. So, for instance:
behavior & (/* ... */ | MADV_REMOVE)
tagged both MADV_REMOVE and MADV_RANDOM (bit 0 set) as discard
operations.
As a result the kernel could erroneously block certain madvises (e.g
MADV_RANDOM or MADV_HUGEPAGE) on sealed VMAs due to them sharing bits
with blocked MADV operations (e.g REMOVE or WIPEONFORK).
This is obviously incorrect, so use a switch statement instead.
Link: https://lkml.kernel.org/r/20240807173336.2523757-1-pedro.falcato@gmail.com Link: https://lkml.kernel.org/r/20240807173336.2523757-2-pedro.falcato@gmail.com Fixes: 8be7258aad44 ("mseal: add mseal syscall") Signed-off-by: Pedro Falcato <pedro.falcato@gmail.com> Tested-by: Jeff Xu <jeffxu@chromium.org> Reviewed-by: Jeff Xu <jeffxu@chromium.org> Cc: Kees Cook <kees@kernel.org> Cc: Liam R. Howlett <Liam.Howlett@oracle.com> Cc: Shuah Khan <shuah@kernel.org> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The regressing commit is new in 6.10. It assumed that anytime event->prog
is set bpf_overflow_handler() should be invoked to execute the attached bpf
program. This assumption is false for tracing events, and as a result the
regressing commit broke bpftrace by invoking the bpf handler with garbage
inputs on overflow.
Prior to the regression the overflow handlers formed a chain (of length 0,
1, or 2) and perf_event_set_bpf_handler() (the !tracing case) added
bpf_overflow_handler() to that chain, while perf_event_attach_bpf_prog()
(the tracing case) did not. Both set event->prog. The chain of overflow
handlers was replaced by a single overflow handler slot and a fixed call to
bpf_overflow_handler() when appropriate. This modifies the condition there
to check event->prog->type == BPF_PROG_TYPE_PERF_EVENT, restoring the
previous behavior and fixing bpftrace.
The dd task would get stuck in an infinite loop in the kernel. What would
happen is the following:
When ring_buffer_read_page() returns -1 (no data) then a check is made to
see if the buffer is empty (as happens when the page is not full), it will
call wait_on_pipe() to wait until the ring buffer has data. When it is it
will try again to read data (unless O_NONBLOCK is set).
The issue happens when there's a reader and the file descriptor is closed.
The wait_on_pipe() will return when that is the case. But this loop will
continue to try again and wait_on_pipe() will again return immediately and
the loop will continue and never stop.
Simply check if the file was closed before looping and exit out if it is.
Add the missing geni_icc_disable() call before returning in the
geni_i2c_runtime_resume() function.
Commit 9ba48db9f77c ("i2c: qcom-geni: Add missing
geni_icc_disable in geni_i2c_runtime_resume") by Gaosheng missed
disabling the interconnect in one case.
copy_fd_bitmaps(new, old, count) is expected to copy the first
count/BITS_PER_LONG bits from old->full_fds_bits[] and fill
the rest with zeroes. What it does is copying enough words
(BITS_TO_LONGS(count/BITS_PER_LONG)), then memsets the rest.
That works fine, *if* all bits past the cutoff point are
clear. Otherwise we are risking garbage from the last word
we'd copied.
For most of the callers that is true - expand_fdtable() has
count equal to old->max_fds, so there's no open descriptors
past count, let alone fully occupied words in ->open_fds[],
which is what bits in ->full_fds_bits[] correspond to.
The other caller (dup_fd()) passes sane_fdtable_size(old_fdt, max_fds),
which is the smallest multiple of BITS_PER_LONG that covers all
opened descriptors below max_fds. In the common case (copying on
fork()) max_fds is ~0U, so all opened descriptors will be below
it and we are fine, by the same reasons why the call in expand_fdtable()
is safe.
Unfortunately, there is a case where max_fds is less than that
and where we might, indeed, end up with junk in ->full_fds_bits[] -
close_range(from, to, CLOSE_RANGE_UNSHARE) with
* descriptor table being currently shared
* 'to' being above the current capacity of descriptor table
* 'from' being just under some chunk of opened descriptors.
In that case we end up with observably wrong behaviour - e.g. spawn
a child with CLONE_FILES, get all descriptors in range 0..127 open,
then close_range(64, ~0U, CLOSE_RANGE_UNSHARE) and watch dup(0) ending
up with descriptor #128, despite #64 being observably not open.
The minimally invasive fix would be to deal with that in dup_fd().
If this proves to add measurable overhead, we can go that way, but
let's try to fix copy_fd_bitmaps() first.
* new helper: bitmap_copy_and_expand(to, from, bits_to_copy, size).
* make copy_fd_bitmaps() take the bitmap size in words, rather than
bits; it's 'count' argument is always a multiple of BITS_PER_LONG,
so we are not losing any information, and that way we can use the
same helper for all three bitmaps - compiler will see that count
is a multiple of BITS_PER_LONG for the large ones, so it'll generate
plain memcpy()+memset().
Reproducer added to tools/testing/selftests/core/close_range_test.c
Cc: stable@vger.kernel.org Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The inode reclaiming process(See function prune_icache_sb) collects all
reclaimable inodes and mark them with I_FREEING flag at first, at that
time, other processes will be stuck if they try getting these inodes
(See function find_inode_fast), then the reclaiming process destroy the
inodes by function dispose_list(). Some filesystems(eg. ext4 with
ea_inode feature, ubifs with xattr) may do inode lookup in the inode
evicting callback function, if the inode lookup is operated under the
inode lru traversing context, deadlock problems may happen.
Case 1: In function ext4_evict_inode(), the ea inode lookup could happen
if ea_inode feature is enabled, the lookup process will be stuck
under the evicting context like this:
1. File A has inode i_reg and an ea inode i_ea
2. getfattr(A, xattr_buf) // i_ea is added into lru // lru->i_ea
3. Then, following three processes running like this:
PA PB
echo 2 > /proc/sys/vm/drop_caches
shrink_slab
prune_dcache_sb
// i_reg is added into lru, lru->i_ea->i_reg
prune_icache_sb
list_lru_walk_one
inode_lru_isolate
i_ea->i_state |= I_FREEING // set inode state
inode_lru_isolate
__iget(i_reg)
spin_unlock(&i_reg->i_lock)
spin_unlock(lru_lock)
rm file A
i_reg->nlink = 0
iput(i_reg) // i_reg->nlink is 0, do evict
ext4_evict_inode
ext4_xattr_delete_inode
ext4_xattr_inode_dec_ref_all
ext4_xattr_inode_iget
ext4_iget(i_ea->i_ino)
iget_locked
find_inode_fast
__wait_on_freeing_inode(i_ea) ----→ AA deadlock
dispose_list // cannot be executed by prune_icache_sb
wake_up_bit(&i_ea->i_state)
Case 2: In deleted inode writing function ubifs_jnl_write_inode(), file
deleting process holds BASEHD's wbuf->io_mutex while getting the
xattr inode, which could race with inode reclaiming process(The
reclaiming process could try locking BASEHD's wbuf->io_mutex in
inode evicting function), then an ABBA deadlock problem would
happen as following:
1. File A has inode ia and a xattr(with inode ixa), regular file B has
inode ib and a xattr.
2. getfattr(A, xattr_buf) // ixa is added into lru // lru->ixa
3. Then, following three processes running like this:
PA PB PC
echo 2 > /proc/sys/vm/drop_caches
shrink_slab
prune_dcache_sb
// ib and ia are added into lru, lru->ixa->ib->ia
prune_icache_sb
list_lru_walk_one
inode_lru_isolate
ixa->i_state |= I_FREEING // set inode state
inode_lru_isolate
__iget(ib)
spin_unlock(&ib->i_lock)
spin_unlock(lru_lock)
rm file B
ib->nlink = 0
rm file A
iput(ia)
ubifs_evict_inode(ia)
ubifs_jnl_delete_inode(ia)
ubifs_jnl_write_inode(ia)
make_reservation(BASEHD) // Lock wbuf->io_mutex
ubifs_iget(ixa->i_ino)
iget_locked
find_inode_fast
__wait_on_freeing_inode(ixa)
| iput(ib) // ib->nlink is 0, do evict
| ubifs_evict_inode
| ubifs_jnl_delete_inode(ib)
↓ ubifs_jnl_write_inode
ABBA deadlock ←-----make_reservation(BASEHD)
dispose_list // cannot be executed by prune_icache_sb
wake_up_bit(&ixa->i_state)
Fix the possible deadlock by using new inode state flag I_LRU_ISOLATING
to pin the inode in memory while inode_lru_isolate() reclaims its pages
instead of using ordinary inode reference. This way inode deletion
cannot be triggered from inode_lru_isolate() thus avoiding the deadlock.
evict() is made to wait for I_LRU_ISOLATING to be cleared before
proceeding with inode cleanup.
kmalloc is unreliable when allocating more than 8 pages of memory. It may
fail when there is plenty of free memory but the memory is fragmented.
Zdenek Kabelac observed such failure in his tests.
This commit changes kmalloc to kvmalloc - kvmalloc will fall back to
vmalloc if the large allocation fails.
If the dm_resume method is called on a device that is not suspended, the
method will suspend the device briefly, before resuming it (so that the
table will be swapped).
However, there was a bug that the return value of dm_suspended_md was not
checked. dm_suspended_md may return an error when it is interrupted by a
signal. In this case, do_resume would call dm_swap_table, which would
return -EINVAL.
This commit fixes the logic, so that error returned by dm_suspend is
checked and the resume operation is undone.
Currently, only acpi_early_node_map[0] was initialized to NUMA_NO_NODE.
To ensure all the values were properly initialized, switch to initialize
all of them to NUMA_NO_NODE.
Fixes: e18962491696 ("arm64: numa: rework ACPI NUMA initialization") Cc: <stable@vger.kernel.org> # 4.19.x Reported-by: Andrew Jones <ajones@ventanamicro.com> Suggested-by: Andrew Jones <ajones@ventanamicro.com> Signed-off-by: Haibo Xu <haibo1.xu@intel.com> Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com> Reviewed-by: Sunil V L <sunilvl@ventanamicro.com> Reviewed-by: Andrew Jones <ajones@ventanamicro.com> Acked-by: Catalin Marinas <catalin.marinas@arm.com> Acked-by: Lorenzo Pieralisi <lpieralisi@kernel.org> Reviewed-by: Hanjun Guo <guohanjun@huawei.com> Link: https://lore.kernel.org/r/853d7f74aa243f6f5999e203246f0d1ae92d2b61.1722828421.git.haibo1.xu@intel.com Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 60fa6ae6e6d0 ("ACPI: EC: Install address space handler at the
namespace root") caused _REG methods for EC operation regions outside
the EC device scope to be evaluated which on some systems leads to the
evaluation of _REG methods in the scopes of device objects representing
devices that are not present and not functional according to the _STA
return values. Some of those device objects represent EC "alternatives"
and if _REG is evaluated for their operation regions, the platform
firmware may be confused and the platform may start to behave
incorrectly.
To avoid this problem, only evaluate _REG for EC operation regions
located in the scopes of device objects representing known-to-be-present
devices.
For this purpose, partially revert commit 60fa6ae6e6d0 and trigger the
evaluation of _REG for EC operation regions from acpi_bus_attach() for
the known-valid devices.
A subsequent change will need to pass a depth argument to
acpi_execute_reg_methods(), so prepare that function for it.
No intentional functional changes.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Hans de Goede <hdegoede@redhat.com> Cc: All applicable <stable@vger.kernel.org> Link: https://patch.msgid.link/8451567.NyiUUSuA9g@rjwysocki.net Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The problem arises because during __pm_runtime_resume(), the spinlock
&dev->power.lock is acquired before rpm_resume() is called. Later,
rpm_resume() invokes acpi_subsys_runtime_resume(), which relies on
mutexes, triggering the error.
To address this issue, devices on ACPI are now marked as not IRQ-safe,
considering the dependency of acpi_subsys_runtime_resume() on mutexes.
Fixes: bd2fdedbf2ba ("i2c: tegra: Add the ACPI support") Cc: <stable@vger.kernel.org> # v5.17+ Co-developed-by: Michael van der Westhuizen <rmikey@meta.com> Signed-off-by: Michael van der Westhuizen <rmikey@meta.com> Signed-off-by: Breno Leitao <leitao@debian.org> Reviewed-by: Dmitry Osipenko <digetx@gmail.com> Reviewed-by: Andy Shevchenko <andy@kernel.org> Signed-off-by: Andi Shyti <andi.shyti@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Mandatory locking is enforced for cached writes, which violates
default posix semantics, and also it is enforced inconsistently.
This apparently breaks recent versions of libreoffice, but can
also be demonstrated by opening a file twice from the same
client, locking it from handle one and writing to it from
handle two (which fails, returning EACCES).
Since there was already a mount option "forcemandatorylock"
(which defaults to off), with this change only when the user
intentionally specifies "forcemandatorylock" on mount will we
break posix semantics on write to a locked range (ie we will
only fail the write in this case, if the user mounts with
"forcemandatorylock").
Fixes: 85160e03a79e ("CIFS: Implement caching mechanism for mandatory brlocks") Cc: stable@vger.kernel.org Cc: Pavel Shilovsky <piastryyy@gmail.com> Reported-by: abartlet@samba.org Reported-by: Kevin Ottens <kevin.ottens@enioka.com> Reviewed-by: David Howells <dhowells@redhat.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Otherwise when the tracer changes syscall number to -1, the kernel fails
to initialize a0 with -ENOSYS and subsequently fails to return the error
code of the failed syscall to userspace. For example, it will break
strace syscall tampering.
Fixes: 52449c17bdd1 ("riscv: entry: set a0 = -ENOSYS only when syscall != -1") Reported-by: "Dmitry V. Levin" <ldv@strace.io> Reviewed-by: Björn Töpel <bjorn@rivosinc.com> Cc: stable@vger.kernel.org Signed-off-by: Celeste Liu <CoelacanthusHex@gmail.com> Link: https://lore.kernel.org/r/20240627142338.5114-2-CoelacanthusHex@gmail.com Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
With XIP kernel, kernel_map.size is set to be only the size of data part of
the kernel. This is inconsistent with "normal" kernel, who sets it to be
the size of the entire kernel.
More importantly, XIP kernel fails to boot if CONFIG_DEBUG_VIRTUAL is
enabled, because there are checks on virtual addresses with the assumption
that kernel_map.size is the size of the entire kernel (these checks are in
arch/riscv/mm/physaddr.c).
Change XIP's kernel_map.size to be the size of the entire kernel.
Trusted keys unseal the key blob on load, but keep the sealed payload in
the blob field so that every subsequent read (export) will simply
convert this field to hex and send it to userspace.
With DCP-based trusted keys, we decrypt the blob encryption key (BEK)
in the Kernel due hardware limitations and then decrypt the blob payload.
BEK decryption is done in-place which means that the trusted key blob
field is modified and it consequently holds the BEK in plain text.
Every subsequent read of that key thus send the plain text BEK instead
of the encrypted BEK to userspace.
This issue only occurs when importing a trusted DCP-based key and
then exporting it again. This should rarely happen as the common use cases
are to either create a new trusted key and export it, or import a key
blob and then just use it without exporting it again.
Fix this by performing BEK decryption and encryption in a dedicated
buffer. Further always wipe the plain text BEK buffer to prevent leaking
the key via uninitialized memory.
Cc: stable@vger.kernel.org # v6.10+ Fixes: 2e8a0f40a39c ("KEYS: trusted: Introduce NXP DCP-backed trusted keys") Signed-off-by: David Gstir <david@sigma-star.at> Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org> Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The DCP trusted key type uses the wrong helper function to store
the blob's payload length which can lead to the wrong byte order
being used in case this would ever run on big endian architectures.
Fix by using correct helper function.
Cc: stable@vger.kernel.org # v6.10+ Fixes: 2e8a0f40a39c ("KEYS: trusted: Introduce NXP DCP-backed trusted keys") Suggested-by: Richard Weinberger <richard@nod.at> Reported-by: kernel test robot <lkp@intel.com> Closes: https://lore.kernel.org/oe-kbuild-all/202405240610.fj53EK0q-lkp@intel.com/ Signed-off-by: David Gstir <david@sigma-star.at> Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org> Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Instead of clearing the "updated" flag for each cooling device
affected by the trip point crossing in bang_bang_control() and
walking all thermal instances to run thermal_cdev_update() for all
of the affected cooling devices, call __thermal_cdev_update()
directly for each of them.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Peter Kästle <peter@piie.net> Reviewed-by: Zhang Rui <rui.zhang@intel.com> Cc: 6.10+ <stable@vger.kernel.org> # 6.10+ Link: https://patch.msgid.link/13583081.uLZWGnKmhe@rjwysocki.net Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We might run into a SIE validity if gisa has been disabled either via using
kernel parameter "kvm.use_gisa=0" or by setting the related sysfs
attribute to N (echo N >/sys/module/kvm/parameters/use_gisa).
The validity is caused by an invalid value in the SIE control block's
gisa designation. That happens because we pass the uninitialized gisa
origin to virt_to_phys() before writing it to the gisa designation.
To fix this we return 0 in kvm_s390_get_gisa_desc() if the origin is 0.
kvm_s390_get_gisa_desc() is used to determine which gisa designation to
set in the SIE control block. A value of 0 in the gisa designation disables
gisa usage.
The issue surfaces in the host kernel with the following kernel message as
soon a new kvm guest start is attemted.
Extent Space Efficient (ESE) or thin provisioned volumes need to be
formatted on demand during usual IO processing.
The dasd_ese_needs_format function checks for error codes that signal
the non existence of a proper track format.
The check for incorrect length is to imprecise since other error cases
leading to transport of insufficient data also have this flag set.
This might lead to data corruption in certain error cases for example
during a storage server warmstart.
Fix by removing the check for incorrect length and replacing by
explicitly checking for invalid track format in transport mode.
Also remove the check for file protected since this is not a valid
ESE handling case.
Cc: stable@vger.kernel.org # 5.3+ Fixes: 5e2b17e712cf ("s390/dasd: Add dynamic formatting support for ESE volumes") Reviewed-by: Jan Hoeppner <hoeppner@linux.ibm.com> Signed-off-by: Stefan Haberland <sth@linux.ibm.com> Link: https://lore.kernel.org/r/20240812125733.126431-3-sth@linux.ibm.com Signed-off-by: Jens Axboe <axboe@kernel.dk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The recent addition of a sanity check for a too low start tick time
seems breaking some applications that uses aloop with a certain slave
timer setup. They may have the initial resolution 0, hence it's
treated as if it were a too low value.
Relax and skip the check for the slave timer instance for addressing
the regression.
Wrong calibration data order cause sound too low in some device.
Fix wrong calibrated data order, add calibration data converssion
by get_unaligned_be32() after reading from UEFI.
I noticed that when we do discrete host router NVM upgrade and it gets
hot-removed from the PCIe side as a result of NVM firmware authentication,
if there is another host connected with enabled paths we hang in tearing
them down. This is due to fact that the Thunderbolt networking driver
also tries to cleanup the paths and ends up blocking in
tb_disconnect_xdomain_paths() waiting for the domain lock.
However, at this point we already cleaned the paths in tb_stop() so
there is really no need for tb_disconnect_xdomain_paths() to do that
anymore. Furthermore it already checks if the XDomain is unplugged and
bails out early so take advantage of that and mark the XDomain as
unplugged when we remove the parent router.
Cc: stable@vger.kernel.org Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
re-enumerating full-speed devices after a failed address device command
can trigger a NULL pointer dereference.
Full-speed devices may need to reconfigure the endpoint 0 Max Packet Size
value during enumeration. Usb core calls usb_ep0_reinit() in this case,
which ends up calling xhci_configure_endpoint().
On Panther point xHC the xhci_configure_endpoint() function will
additionally check and reserve bandwidth in software. Other hosts do
this in hardware
If xHC address device command fails then a new xhci_virt_device structure
is allocated as part of re-enabling the slot, but the bandwidth table
pointers are not set up properly here.
This triggers the NULL pointer dereference the next time usb_ep0_reinit()
is called and xhci_configure_endpoint() tries to check and reserve
bandwidth
[46710.713538] usb 3-1: new full-speed USB device number 5 using xhci_hcd
[46710.713699] usb 3-1: Device not responding to setup address.
[46710.917684] usb 3-1: Device not responding to setup address.
[46711.125536] usb 3-1: device not accepting address 5, error -71
[46711.125594] BUG: kernel NULL pointer dereference, address: 0000000000000008
[46711.125600] #PF: supervisor read access in kernel mode
[46711.125603] #PF: error_code(0x0000) - not-present page
[46711.125606] PGD 0 P4D 0
[46711.125610] Oops: Oops: 0000 [#1] PREEMPT SMP PTI
[46711.125615] CPU: 1 PID: 25760 Comm: kworker/1:2 Not tainted 6.10.3_2 #1
[46711.125620] Hardware name: Gigabyte Technology Co., Ltd.
[46711.125623] Workqueue: usb_hub_wq hub_event [usbcore]
[46711.125668] RIP: 0010:xhci_reserve_bandwidth (drivers/usb/host/xhci.c
Fix this by making sure bandwidth table pointers are set up correctly
after a failed address device command, and additionally by avoiding
checking for bandwidth in cases like this where no actual endpoints are
added or removed, i.e. only context for default control endpoint 0 is
evaluated.
Reported-by: Karel Balej <balejk@matfyz.cz> Closes: https://lore.kernel.org/linux-usb/D3CKQQAETH47.1MUO22RTCH2O3@matfyz.cz/ Cc: stable@vger.kernel.org Fixes: 651aaf36a7d7 ("usb: xhci: Handle USB transaction error on address command") Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com> Link: https://lore.kernel.org/r/20240815141117.2702314-2-mathias.nyman@linux.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If xhci_mem_init() fails, it calls into xhci_mem_cleanup() to mop
up the damage. If it fails early enough, before xhci->interrupters
is allocated but after xhci->max_interrupters has been set, which
happens in most (all?) cases, things get uglier, as xhci_mem_cleanup()
unconditionally derefences xhci->interrupters. With prejudice.
Gate the interrupt freeing loop with a check on xhci->interrupters
being non-NULL.
Found while debugging a DMA allocation issue that led the XHCI driver
on this exact path.
Fixes: c99b38c41234 ("xhci: add support to allocate several interrupters") Cc: Mathias Nyman <mathias.nyman@linux.intel.com> Cc: Wesley Cheng <quic_wcheng@quicinc.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Marc Zyngier <maz@kernel.org> Cc: stable@vger.kernel.org # 6.8+ Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com> Link: https://lore.kernel.org/r/20240809124408.505786-2-mathias.nyman@linux.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Ensure, as the driver probes the device, that all endpoints that the
driver may attempt to access exist and are of the correct type.
All XillyUSB devices must have a Bulk IN and Bulk OUT endpoint at
address 1. This is verified in xillyusb_setup_base_eps().
On top of that, a XillyUSB device may have additional Bulk OUT
endpoints. The information about these endpoints' addresses is deduced
from a data structure (the IDT) that the driver fetches from the device
while probing it. These endpoints are checked in setup_channels().
A XillyUSB device never has more than one IN endpoint, as all data
towards the host is multiplexed in this single Bulk IN endpoint. This is
why setup_channels() only checks OUT endpoints.
Reported-by: syzbot+eac39cba052f2e750dbe@syzkaller.appspotmail.com Cc: stable <stable@kernel.org> Closes: https://lore.kernel.org/all/0000000000001d44a6061f7a54ee@google.com/T/ Fixes: a53d1202aef1 ("char: xillybus: Add driver for XillyUSB (Xillybus variant for USB)"). Signed-off-by: Eli Billauer <eli.billauer@gmail.com> Link: https://lore.kernel.org/r/20240816070200.50695-2-eli.billauer@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Triggered by a kref decrement, destroy_workqueue() may be called from
within a work item for destroying its own workqueue. This illegal
situation is averted by adding a module-global workqueue for exclusive
use of the offending work item. Other work items continue to be queued
on per-device workqueues to ensure performance.
fuse_notify_store(), unlike fuse_do_readpage(), does not enable page
zeroing (because it can be used to change partial page contents).
So fuse_notify_store() must be more careful to fully initialize page
contents (including parts of the page that are beyond end-of-file)
before marking the page uptodate.
The current code can leave beyond-EOF page contents uninitialized, which
makes these uninitialized page contents visible to userspace via mmap().
This is an information leak, but only affects systems which do not
enable init-on-alloc (via CONFIG_INIT_ON_ALLOC_DEFAULT_ON=y or the
corresponding kernel command line parameter).
Revert the patch that removes the deprecated use of PG_private_2 in
netfslib for the moment as Ceph is actually still using this to track
data copied to the cache.
Fixes: ae678317b95e ("netfs: Remove deprecated use of PG_private_2 as a second writeback flag") Reported-by: Max Kellermann <max.kellermann@ionos.com> Signed-off-by: David Howells <dhowells@redhat.com>
cc: Ilya Dryomov <idryomov@gmail.com>
cc: Xiubo Li <xiubli@redhat.com>
cc: Jeff Layton <jlayton@kernel.org>
cc: Matthew Wilcox <willy@infradead.org>
cc: ceph-devel@vger.kernel.org
cc: netfs@lists.linux.dev
cc: linux-fsdevel@vger.kernel.org
cc: linux-mm@kvack.org
https: //lore.kernel.org/r/3575457.1722355300@warthog.procyon.org.uk Signed-off-by: Christian Brauner <brauner@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Unfortunately it appears that vma_is_initial_heap() is currently broken
for applications that do not currently have any heap allocated, e.g.
brk == start_brk. The breakage is such that it will cause SELinux to
check for the process/execheap permission on memory regions that cross
brk/start_brk even when there is no heap.
The proper fix would be to correct vma_is_initial_heap(), but as there
are multiple callers I am hesitant to unilaterally modify the helper
out of concern that I would end up breaking some other subsystem. The
mm developers have been made aware of the situation and hopefully they
will have a fix at some point in the future, but we need a fix soon so
we are simply going to revert our use of vma_is_initial_heap() in favor
of our old logic/code which works as expected, even in the face of a
zero size heap. We can return to using vma_is_initial_heap() at some
point in the future when it is fixed.
Cc: stable@vger.kernel.org Reported-by: Marc Reisner <reisner.marc@gmail.com> Closes: https://lore.kernel.org/all/ZrPmoLKJEf1wiFmM@marcreisner.com Fixes: 68df1baf158f ("selinux: use vma_is_initial_stack() and vma_is_initial_heap()") Signed-off-by: Paul Moore <paul@paul-moore.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
During tcpm_init() stage, if the VBUS is still present after
tcpm_reset_port(), then we assume that VBUS will off and goto safe0v
after a specific discharge time. Following a TCPM_VBUS_EVENT event if
VBUS reach to off state. TCPM_VBUS_EVENT event may be set during
PORT_RESET handling stage. If pd_events reset to 0 after TCPM_VBUS_EVENT
set, we will lost this VBUS event. Then the port state machine may stuck
at one state.
This reverts commit 0e6b6dedf168 ("Revert "ACPI: EC: Evaluate orphan
_REG under EC device") because the problem addressed by it will be
addressed differently in what follows.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Hans de Goede <hdegoede@redhat.com> Cc: All applicable <stable@vger.kernel.org> Link: https://patch.msgid.link/3236716.5fSG56mABF@rjwysocki.net Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In RS485 mode, the RTS pin is driven high by hardware when the transmitter
is operating. This behaviour cannot be changed. This means that the driver
should claim that it supports SER_RS485_RTS_ON_SEND and not
SER_RS485_RTS_AFTER_SEND.
Otherwise, when configuring the port with the SER_RS485_RTS_ON_SEND, one
get the following warning:
With "earlycon initcall_debug=1 loglevel=8" in bootargs, kernel
sometimes boot hang. It is because normal console still is not ready,
but runtime suspend is called, so early console putchar will hang
in waiting TRDE set in UARTSTAT.
The lpuart driver has auto suspend delay set to 3000ms, but during
uart_add_one_port, a child device serial ctrl will added and probed with
its pm runtime enabled(see serial_ctrl.c).
The runtime suspend call path is:
device_add
|-> bus_probe_device
|->device_initial_probe
|->__device_attach
|-> pm_runtime_get_sync(dev->parent);
|-> pm_request_idle(dev);
|-> pm_runtime_put(dev->parent);
So in the end, before normal console ready, the lpuart get runtime
suspended. And earlycon putchar will hang.
To address the issue, mark last busy just after pm_runtime_enable,
three seconds is long enough to switch from bootconsole to normal
console.
Commit 6e20753da6bc ("tty: vt: conmakehash: cope with abs_srctree no
longer in env") included <linux/limits.h>, which invoked another
(wrong) patch that tried to address a build error on macOS.
According to the specification [1], the correct header to use PATH_MAX
is <limits.h>.
The minimal fix would be to replace <linux/limits.h> with <limits.h>.
However, the following commits seem questionable to me:
- 3bd85c6c97b2 ("tty: vt: conmakehash: Don't mention the full path of the input in output")
- 6e20753da6bc ("tty: vt: conmakehash: cope with abs_srctree no longer in env")
These commits made too many efforts to cope with a comment header in
drivers/tty/vt/consolemap_deftbl.c:
/*
* Do not edit this file; it was automatically generated by
*
* conmakehash drivers/tty/vt/cp437.uni > [this file]
*
*/
With this commit, the header part of the generate C file will be
simplified as follows:
/*
* Automatically generated file; Do not edit.
*/
BTW, another series of excessive efforts for a comment header can be
seen in the following:
- 5ef6dc08cfde ("lib/build_OID_registry: don't mention the full path of the script in output")
- 2fe29fe94563 ("lib/build_OID_registry: avoid non-destructive substitution for Perl < 5.13.2 compat")
It breaks the build on arm64 and arm systems, and trying to unwind the
ifdef mess to fix it up was not simple at all. So revert it and wait
for a fixed change to come back.
Cc: Jerry Zuo <jerry.zuo@amd.com> Cc: Zaeem Mohamed <zaeem.mohamed@amd.com> Cc: Wayne Lin <wayne.lin@amd.com> Cc: Daniel Wheeler <daniel.wheeler@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: Kevin Holm <kevin@holm.dev> Reported-by: Guenter Roeck <linux@roeck-us.net> Link: https://lore.kernel.org/r/b27c5434-f1b1-4697-985b-91bb3e9a22df@roeck-us.net Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Sense data can be in either fixed format or descriptor format.
SAT-6 revision 1, "10.4.6 Control mode page", defines the D_SENSE bit:
"The SATL shall support this bit as defined in SPC-5 with the following
exception: if the D_ SENSE bit is set to zero (i.e., fixed format sense
data), then the SATL should return fixed format sense data for ATA
PASS-THROUGH commands."
The libata SATL has always kept D_SENSE set to zero by default. (It is
however possible to change the value using a MODE SELECT SG_IO command.)
Failed ATA PASS-THROUGH commands correctly respected the D_SENSE bit,
however, successful ATA PASS-THROUGH commands incorrectly returned the
sense data in descriptor format (regardless of the D_SENSE bit).
Commit 28ab9769117c ("ata: libata-scsi: Honor the D_SENSE bit for
CK_COND=1 and no error") fixed this bug for successful ATA PASS-THROUGH
commands.
However, after commit 28ab9769117c ("ata: libata-scsi: Honor the D_SENSE
bit for CK_COND=1 and no error"), there were bug reports that hdparm,
hddtemp, and udisks were no longer working as expected.
These applications incorrectly assume the returned sense data is in
descriptor format, without even looking at the RESPONSE CODE field in the
returned sense data (to see which format the returned sense data is in).
Considering that there will be broken versions of these applications around
roughly forever, we are stuck with being bug compatible with older kernels.
Cc: stable@vger.kernel.org # 4.19+ Reported-by: Stephan Eisvogel <eisvogel@seitics.de> Reported-by: Christian Heusel <christian@heusel.eu> Closes: https://lore.kernel.org/linux-ide/0bf3f2f0-0fc6-4ba5-a420-c0874ef82d64@heusel.eu/ Fixes: 28ab9769117c ("ata: libata-scsi: Honor the D_SENSE bit for CK_COND=1 and no error") Reviewed-by: Hannes Reinecke <hare@suse.de> Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com> Link: https://lore.kernel.org/r/20240813131900.1285842-2-cassel@kernel.org Signed-off-by: Niklas Cassel <cassel@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This breaks the TeVii s480 dual DVB-S2 S660. The device has a bulk in
endpoint but no corresponding out endpoint, so the device does not pass
the "has both receive and send bulk endpoint" test.
Seemingly this device does not use dvb_usb_generic_rw() so I have tried
removing the generic_bulk_ctrl_endpoint entry, but this resulted in
different problems.
As we have no explanation yet, revert.
$ dmesg | grep -i -e dvb -e dw21 -e usb\ 4
[ 0.999122] usb 1-1: new high-speed USB device number 2 using ehci-pci
[ 1.023123] usb 4-1: new high-speed USB device number 2 using ehci-pci
[ 1.130247] usb 1-1: New USB device found, idVendor=9022, idProduct=d482,
+bcdDevice= 0.01
[ 1.130257] usb 1-1: New USB device strings: Mfr=0, Product=0, SerialNumber=0
[ 1.152323] usb 4-1: New USB device found, idVendor=9022, idProduct=d481,
+bcdDevice= 0.01
[ 1.152329] usb 4-1: New USB device strings: Mfr=0, Product=0, SerialNumber=0
[ 6.701033] dvb-usb: found a 'TeVii S480.2 USB' in cold state, will try to
+load a firmware
[ 6.701178] dvb-usb: downloading firmware from file 'dvb-usb-s660.fw'
[ 6.701179] dw2102: start downloading DW210X firmware
[ 6.703715] dvb-usb: found a 'Microsoft Xbox One Digital TV Tuner' in cold
+state, will try to load a firmware
[ 6.703974] dvb-usb: downloading firmware from file 'dvb-usb-dib0700-1.20.fw'
[ 6.756432] usb 1-1: USB disconnect, device number 2
[ 6.862119] dvb-usb: found a 'TeVii S480.2 USB' in warm state.
[ 6.862194] dvb-usb: TeVii S480.2 USB error while loading driver (-22)
[ 6.862209] dvb-usb: found a 'TeVii S480.1 USB' in cold state, will try to
+load a firmware
[ 6.862244] dvb-usb: downloading firmware from file 'dvb-usb-s660.fw'
[ 6.862245] dw2102: start downloading DW210X firmware
[ 6.914811] usb 4-1: USB disconnect, device number 2
[ 7.014131] dvb-usb: found a 'TeVii S480.1 USB' in warm state.
[ 7.014487] dvb-usb: TeVii S480.1 USB error while loading driver (-22)
[ 7.014538] usbcore: registered new interface driver dw2102
Closes: https://lore.kernel.org/stable/20240801165146.38991f60@mir/ Fixes: 2052138b7da5 ("media: dvb-usb: Fix unexpected infinite loop in dvb_usb_read_remote_control()") Reported-by: Stefan Lippers-Hollmann <s.l-h@gmx.de> Cc: stable@vger.kernel.org Signed-off-by: Sean Young <sean@mess.org> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The fix involves adding a null check for 'stream' at the beginning of
the function. If 'stream' is NULL, the function immediately returns
false. This ensures that 'stream' is not NULL when we dereference it to
access 'ctx' in 'dc = stream->ctx->dc;' the function.
Fixes the below:
drivers/gpu/drm/amd/amdgpu/../display/dc/core/dc_stream.c:398 dc_stream_program_cursor_position()
error: we previously assumed 'stream' could be null (see line 397)
Fixes: f63f86b5affc ("drm/amd/display: Separate setting and programming of cursor") Reported-by: Dan Carpenter <dan.carpenter@linaro.org> Cc: Harry Wentland <harry.wentland@amd.com> Cc: Tom Chung <chiahsuan.chung@amd.com> Cc: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com> Cc: Roman Li <roman.li@amd.com> Cc: Aurabindo Pillai <aurabindo.pillai@amd.com> Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[Why]
In dm resume, we firstly restore dc state and do the mst resume for topology
probing thereafter. If we change dpcd DP_MSTM_CTRL value after LT in mst reume,
it will cause light up problem on the hub.
[How]
Revert commit 202dc359adda ("drm/amd/display: Defer handling mst up request in resume").
And adjust the reason to trigger dc_link_detect by DETECT_REASON_RESUMEFROMS3S4.
Cc: stable@vger.kernel.org Fixes: 202dc359adda ("drm/amd/display: Defer handling mst up request in resume") Signed-off-by: Wayne Lin <Wayne.Lin@amd.com> Reviewed-by: Fangzhi Zuo <jerry.zuo@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 04d82a6d0881 ("binfmt_flat: allow not offsetting data start")
introduced a RISC-V specific variant of the FLAT format which does
not allocate any space for the (obsolete) array of shared library
pointers. However, it did not disable the code which initializes the
array, resulting in the corruption of sizeof(long) bytes before the DATA
segment, generally the end of the TEXT segment.
Introduce MAX_SHARED_LIBS_UPDATE which depends on the state of
CONFIG_BINFMT_FLAT_NO_DATA_START_OFFSET to guard the initialization of
the shared library pointer region so that it will only be initialized
if space is reserved for it.
Fixes: 04d82a6d0881 ("binfmt_flat: allow not offsetting data start") Co-developed-by: Stefan O'Rear <sorear@fastmail.com> Signed-off-by: Stefan O'Rear <sorear@fastmail.com> Reviewed-by: Damien Le Moal <dlemoal@kernel.org> Acked-by: Greg Ungerer <gerg@linux-m68k.org> Link: https://lore.kernel.org/r/20240807195119.it.782-kees@kernel.org Signed-off-by: Kees Cook <kees@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Calling VPC commands consists of several VPCW and VPCR ACPI calls.
These calls and their results can get mixed up if they are called
simultaneously from different threads, like acpi notify handler,
sysfs, debugfs, notification chain.
The commit e2ffcda16290 ("ACPI: OSL: Allow Notify () handlers to run on
all CPUs") made the race issues much worse than before it but some
races were possible even before that commit.
Add a mutex to synchronize VPC commands.
Fixes: e2ffcda16290 ("ACPI: OSL: Allow Notify () handlers to run on all CPUs") Fixes: e82882cdd241 ("platform/x86: Add driver for Yoga Tablet Mode switch") Signed-off-by: Gergo Koteles <soyer@irl.hu> Link: https://lore.kernel.org/r/f26782fa1194ad11ed5d9ba121a804e59b58b026.1721898747.git.soyer@irl.hu Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Some models need to trigger the EC after each YMC event for the yoga
mode control to work properly. EC triggering consist of a VPC call from
the lenovo-ymc module. Except for this, all VPC calls are in the
ideapad-laptop module.
Since ideapad-laptop has a notification chain, a new YMC_EVENT action
can be added and triggered from the lenovo-ymc module. Then the
ideapad-laptop can trigger the EC.
If the triggering is in the ideapad-laptop module, then the ec_trigger
module parameter should be there as well.
Move the ymc_trigger_ec functionality and the ec_trigger module
parameter to the ideapad-laptop module.
If the Ambient Light Sensor (ALS) is disabled, the current code in the PMF
driver does not query for Human Presence Detection (HPD) data in
amd_pmf_get_sensor_info(). As a result, stale HPD data is used by PMF-TA
to evaluate policy conditions, leading to unexpected behavior in the policy
output actions.
To resolve this issue, modify the PMF driver to query HPD data
independently of ALS.
Since user_present is a boolean, modify the current code to return true if
the user is present and false if the user is away or if the sensor is not
detected, and report this status to the PMF TA firmware accordingly.
With this change, amd_pmf_get_sensor_info() now returns void instead of
int.
Fixes: cedecdba60f4 ("platform/x86/amd/pmf: Get ambient light information from AMD SFH driver") Co-developed-by: Patil Rajesh Reddy <Patil.Reddy@amd.com> Signed-off-by: Patil Rajesh Reddy <Patil.Reddy@amd.com> Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com> Link: https://lore.kernel.org/r/20240730142316.3846259-1-Shyam-sundar.S-k@amd.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
A malformed USB descriptor may pass the lengthy mixer description with
a lot of channels, and this may overflow the 32bit integer shift
size, as caught by syzbot UBSAN test. Although this won't cause any
real trouble, it's better to address.
This patch introduces a sanity check of the number of channels to bail
out the parsing when too many channels are found.
In order not to call copy_to_user (from fiemap_fill_next_extent)
we allocate memory in the kernel, fill it and copy it to user memory
after up_read(run_lock).
Reported-by: syzbot+36bb70085ef6edc2ebb9@syzkaller.appspotmail.com Signed-off-by: Konstantin Komarov <almaz.alexandrovich@paragon-software.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
projects / thirdparty / kernel / stable.git / log
