| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
smb3_insert_range() and smb3_collapse_range() use
truncate_pagecache_range() to invalidate the affected page cache.
However, if off or old_eof is not page-aligned, the boundary pages are
only partially zeroed and remain uptodate. As a result, the client may
return stale data after a successful insert/collapse range operation.
For example, with 4K pages:
page 0 page 1 page 2
0------4K 4K------8K 8K------12K
^ ^
off=2K old_eof=10K
Page 1 is removed from the page cache, while the boundary pages are
only partially zeroed. After COPYCHUNK moves the data on the server,
these cached pages may still return stale data.
This can be reproduced on a CIFS mount:
bash -c '
FILE=/mnt/scratch/repro
# Use a 6 KiB file so EOF is not page-aligned.
dd if=/dev/urandom of=/tmp/src bs=1K count=6 status=none
# Expected: a 4 KiB hole followed by the original data.
rm -f /tmp/expected
truncate -s 4K /tmp/expected
cat /tmp/src >> /tmp/expected
cp /tmp/src "$FILE"
# Prime the page cache before moving data on the server.
cat "$FILE" > /dev/null
fallocate --insert-range -o 0 -l 4K "$FILE"
if cmp -s /tmp/expected "$FILE"; then
echo "readback: OK"
else
echo "readback: STALE DATA"
fi
'
Fix this by writing back dirty data and discarding the page cache from
the start of the page containing off to EOF before moving data on the
server.
Fixes: 9c8b7a293f50 ("smb3: fix temporary data corruption in insert range")
Fixes: fa30a81f255a ("smb3: fix temporary data corruption in collapse range")
Signed-off-by: Huiwen He <hehuiwen@kylinos.cn>
Reviewed-by: ChenXiaoSong <chenxiaosong@kylinos.cn>
Reviewed-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Paulo Alcantara <pc@manguebit.org>
|
|
smb3_collapse_range() stores the ssize_t return value of
smb2_copychunk_range() in an int. A successful copy larger than
INT_MAX is truncated to a negative value and treated as an error.
Reproducer:
MNT=/mnt/scratch
truncate -s 2056M "$MNT/file"
fallocate --collapse-range -o 1M -l 1M "$MNT/file"
Fix this by using __smb2_copychunk_range(), which reports success as
zero instead of returning the copied byte count.
Before this change, the reproducer fails with:
fallocate: fallocate failed: Success
and the file size remains unchanged at 2056 MiB. After this change, the
reproducer succeeds and the file size becomes the expected 2055 MiB.
Fixes: 5476b5dd82c8 ("cifs: add support for FALLOC_FL_COLLAPSE_RANGE")
Signed-off-by: Huiwen He <hehuiwen@kylinos.cn>
Reviewed-by: ChenXiaoSong <chenxiaosong@kylinos.cn>
Reviewed-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Paulo Alcantara <pc@manguebit.org>
|
|
smb3_insert_range() shifts [off, EOF) right with COPYCHUNK, copying from
low to high offsets. When the ranges overlap, the copy can overwrite
source data that has not yet been copied. For a 1 MiB insert at offset 0:
offset: 0 1M 2M 3M 4M 5M
before: | A | B | C | D |
expected: | hole | A | B | C | D |
current: | hole | A | A | A | A | (corrupted)
Let x be the insertion offset, L the total length to move, delta the
insert length, and C the normal chunk size allowed by the server.
Insert range maps
[x, x + L) -> [x + delta, x + delta + L).
When delta >= L, the complete source and target ranges are disjoint, so
the normal copy order and chunk size are safe:
offset: 0 4 8 12 16 20 24 28 32
source: [--S0--][--S1--][--S2--][--S3--]
target: [--T0--][--T1--][--T2--][--T3--]
When delta < L, the complete source and target ranges overlap, so the
copy must proceed from EOF backwards. There are two subcases.
If delta >= C, each corresponding source and target chunk is disjoint.
The 1 MiB example has L = 4 MiB and delta = C = 1 MiB:
offset: 0 1M 2M 3M 4M 5M
source: [--S0--][--S1--][--S2--][--S3--]
target: [--T0--][--T1--][--T2--][--T3--]
Copying S0 from [0, 1M) to [1M, 2M) overwrites S1 before it is copied.
Processing chunks from EOF backwards prevents this inter-chunk
overwrite.
If delta < C, the source and target ranges of a normal chunk also
overlap. For example, with L = 16, delta = 2 and C = 4:
offset: 0 2 4 6 8 10 12 14 16 18
source: [--S0--][--S1--][--S2--][--S3--]
target: [--T0--][--T1--][--T2--][--T3--]
Here S0 and T0 overlap over [2,4), S1 and T1 over [6,8), and so on.
Backward ordering cannot control how the server copies bytes inside one
descriptor, so the chunk size must be limited to delta.
Fix this by copying overlapping right shifts from EOF backwards. Limit
the chunk size to delta when delta < C so that each chunk's source and
target ranges do not overlap. Using larger chunks would require a way to
identify servers that safely handle overlapping COPYCHUNK descriptors.
Therefore:
delta >= L:
keep the normal copy order and chunk size
delta < L:
delta >= C: copy backwards and keep the normal chunk size
delta < C: copy backwards and limit the chunk size to delta
Only the delta < C subcase requires reducing the chunk size for data
integrity.
Reproducer:
bash -c '
MNT=/mnt/scratch
# Generate four 1 MiB random blocks: [A][B][C][D].
dd if=/dev/urandom of=/tmp/src bs=1M count=4 status=none
# With C = 1 MiB, test delta = C and delta < C.
for delta in 1M 1K; do
truncate -s 0 /tmp/expected
truncate -s "$delta" /tmp/expected
cat /tmp/src >> /tmp/expected
cp /tmp/src "$MNT/file"
fallocate --insert-range -o 0 -l "$delta" "$MNT/file"
if cmp -s /tmp/expected "$MNT/file"; then
echo "delta=$delta: OK"
else
echo "delta=$delta: CORRUPTED"
fi
done
'
The corruption reproduces with Samba and ksmbd, while Windows handles
the overlapping COPYCHUNK ranges safely.
The 1 MiB case tests delta >= C, while the 1 KiB case tests delta < C.
Before this change, the reproducer reports:
delta=1M: CORRUPTED
delta=1K: CORRUPTED
After this change, it passes against both ksmbd and Samba:
delta=1M: OK
delta=1K: OK
Fixes: 7fe6fe95b936 ("cifs: add FALLOC_FL_INSERT_RANGE support")
Signed-off-by: Huiwen He <hehuiwen@kylinos.cn>
Reviewed-by: ChenXiaoSong <chenxiaosong@kylinos.cn>
Reviewed-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Paulo Alcantara <pc@manguebit.org>
|
|
The SMB client emulates FALLOC_FL_INSERT_RANGE with SET_EOF, COPYCHUNK
and SET_ZERO_DATA.
SET_ZERO_DATA creates a hole only when the file is sparse. On a
non-sparse file, it clears the inserted range but leaves its blocks
allocated, causing the extent count check in xfstests generic/064 to
fail.
Fix this by marking the file sparse before modifying it.
This patch produces the expected sparse extents in xfstests generic/064
only when the server-reported block size is compatible with the server's
deallocation granularity.
For ksmbd, the reported block size follows the backing filesystem,
and the test passes. For Samba, the test passes with a block size
matching the backend granularity, for example, 4 KiB on Btrfs, but not
with the default 1 KiB value. For Windows Server 2022, 4 KiB inserts do
not generate holes, while aligned inserts of 64 KiB or larger do.
Fixes: 7fe6fe95b936 ("cifs: add FALLOC_FL_INSERT_RANGE support")
Signed-off-by: Huiwen He <hehuiwen@kylinos.cn>
Reviewed-by: ChenXiaoSong <chenxiaosong@kylinos.cn>
Reviewed-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Paulo Alcantara <pc@manguebit.org>
|
|
When FALLOC_FL_ZERO_RANGE is used without FALLOC_FL_KEEP_SIZE,
smb3_zero_range() may extend EOF without checking RLIMIT_FSIZE, allowing
the file to grow beyond the caller's file-size limit.
Fix this by calling inode_newsize_ok() before sending the zero-range
request when the operation would extend EOF.
Reproducer, using a file on a CIFS mount:
bash -c '
FILE=/mnt/cifs/repro
trap "" SIGXFSZ
ulimit -f 3072
truncate -s 2M "$FILE"
fallocate --zero-range -o 0 -l 4M "$FILE"
echo "fallocate rc=$?"
stat -c "file size=%s" "$FILE"
'
Before this change, the operation succeeds despite the 3 MiB limit:
fallocate rc=0
file size=4194304
After this change, fallocate fails and leaves the file at 2 MiB.
Fixes: 72c419d9b073 ("cifs: fix smb3_zero_range so it can expand the file-size when required")
Signed-off-by: Huiwen He <hehuiwen@kylinos.cn>
Reviewed-by: ChenXiaoSong <chenxiaosong@kylinos.cn>
Reviewed-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Paulo Alcantara <pc@manguebit.org>
|
|
smb3_insert_range() does not check if the new file size
(i_size + len) is valid. This allows FALLOC_FL_INSERT_RANGE to bypass
RLIMIT_FSIZE, exceed s_maxbytes, or produce a size outside the loff_t
range.
Use check_add_overflow() to calculate the new EOF. Validate it with
inode_newsize_ok() before modifying the file.
Reproducer, using a file on a CIFS mount:
bash -c '
FILE=/mnt/cifs/repro
trap "" SIGXFSZ
ulimit -f 3072 # RLIMIT_FSIZE = 3 MiB
# A regular write is stopped at 3 MiB.
dd if=/dev/zero of="$FILE" bs=1M count=4 status=none
stat -c "size after write: %s" "$FILE"
# Insert 2 MiB into a 2 MiB file.
truncate -s 2M "$FILE"
fallocate -i -o 0 -l 2M "$FILE"
stat -c "size after insert: %s" "$FILE"
'
Before this change, the regular write stops at the 3 MiB limit, but
insert range grows the file to 4 MiB:
dd: error writing '/mnt/cifs/repro': File too large
size after write: 3145728
size after insert: 4194304
After this change, insert range also fails at the limit and leaves the
2 MiB file unchanged:
dd: error writing '/mnt/cifs/repro': File too large
size after write: 3145728
fallocate: fallocate failed: File too large
size after insert: 2097152
Fixes: 7fe6fe95b936 ("cifs: add FALLOC_FL_INSERT_RANGE support")
Signed-off-by: Huiwen He <hehuiwen@kylinos.cn>
Reviewed-by: ChenXiaoSong <chenxiaosong@kylinos.cn>
Reviewed-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Paulo Alcantara <pc@manguebit.org>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull locking fix from Ingo Molnar:
- Revert a commit to spinlock cleanup guards that got caught up
in the subtle limitations & fragility of guards (again...) and
caused a regression (Peter Zijlstra)
* tag 'locking-urgent-2026-08-30' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
locking: Revert switching guards to _irq_{disable,enable}()
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/ojeda/linux
Pull Rust fixes from Miguel Ojeda:
"Toolchain and infrastructure:
- Fix KCFI failures, such as in Rust doctests, by disabling function
merging when CFI is enabled. Gary reported the LLVM bug to upstream
and it is now fixed in their mainline.
- Fix 'objtool' fallthrough warnings under the experimental
'CONFIG_RUST_INLINE_HELPERS' by passing (for the combined Rust and
helpers code) the LLVM options needed to preserve the unreachable
traps that 'rustc' normally emits.
In addition, fix 'objtool' errors when LTO is enabled on top, by
also filtering out the LTO flags (for the combined Rust and helpers
code) so that the traps are kept in place.
- Fix 'objtool' warnings by adding one more 'noreturn' function.
- Fix 'make rusttest' target when the 'rustc-dev' component is
installed and Rust >= 1.82.0, <= 1.87.0 is used.
'kernel' crate:
- 'num' module: fix soundness issue in the 'Bounded' conversion from
'bool' by restricting the conversions to unsigned 'Bounded'.
- 'jump_label' module: fix future 'make rusttest' target failures
when 'ARCH=' is set to an arch different than the host's.
- 'list' module: fix incorrect 'pop_back()' comment"
* tag 'rust-fixes-7.3' of git://git.kernel.org/pub/scm/linux/kernel/git/ojeda/linux:
rust: kbuild: disambiguate `zerocopy_derive` for `rusttest`
rust: num: restrict bool conversion to unsigned Bounded
kbuild: rust: keep Rust objects out of Clang LTO with inline helpers
kbuild: rust: preserve unreachable traps with inline helpers
rust: cfi: disable function merging if CFI is enabled
rust: jump_label: skip arch-specific asm in `testlib` builds
objtool/rust: add one more `noreturn` Rust function
rust: kernel: list: fix incorrect pop_back example comment
|
|
Pull arm updates from Russell King:
"Updates for 7.3:
- add module description for kprobes testing module
- remove references to CONFIG_CPU_ARM92x_CPU_IDLE options
- expand comment in ARM's __switch_to()
Also a number of fixes that missed 7.2:
- disable broken eBPF on RiscPC
- more BKPT fixes (guys, it's a *very* bad idea when everyone uses
the BKPT instruction for their own differing purposes)
- another preempt-rt fix, this time for siglock / CPU timers
- fix another path where we try to send signals to processes with
interrupts disabled
- acquire mmap write lock for show_pte() with user faults"
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rmk/linux:
ARM: 9480/1: entry: expand comment in __switch_to
ARM: 9478/1: Remove references to removed CONFIG_CPU_ARM92x_CPU_IDLE options
ARM: 9485/1: mm: acquire mmap write lock around show_pte() for user faults
ARM: 9484/1: enable interrupts when unhandled user faults are triggered
ARM: 9483/1: select HAVE_POSIX_CPU_TIMERS_TASK_WORK
ARM: 9481/2: breakpoint: CFI breakpoints only on demand
ARM: 9477/1: Disable broken eBPF JIT on the Risc PC
ARM: 9473/1: kprobes: test: add MODULE_DESCRIPTION
|
|
Pull OpenRISC updates from Stafford Horne:
"One small trivial macro cleanup and one bug fix.
The bug fix is to fix an unchecked access in our or1k_atomic syscall,
I am debating if we should just deprecate this as there is minimal
need for it"
* tag 'for-linus' of https://github.com/openrisc/linux:
openrisc: fix arbitrary kernel memory access via or1k_atomic syscall
openrisc: drop unneeded semicolon
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace
Pull tracing fixes from Steven Rostedt:
- Fix error output of boot instance creation failure
Currently if a boot instance creation fails, instead of printing out
the name of the instance that failed, it prints "(null)". That is
because it prints "cur_str" that had already been processed by
strsep(). Print the saved name instead.
While at it, print the error code of the failure.
- Fix use-after-free for same named historgrams
Histograms can be named so that they can be used in multiple events.
But if the named histogram has a variable attached, the second event
that uses the named histogram which duplicates it and needs to free
the original after duplication leaves the old variable in place and
still visible. If another histogram uses than variable, it will use
the stale one which will try to reference the freed duplicate
histogram and crash the kernel.
Free the duplicate variables along with the duplicated histogram
data.
- Check return value of kthread_run() in event self test
The events self tests uses a kthread for testing but does not check
if it succeeded in creating a kthread. If the kthread creation were
to fail, the code will still try to call kthread_stop() on the error
returned.
- Fix race between reading trace_pipe and updating subbuffer size
If a user is reading the trace_pipe file at the same time they update
the ring buffer sub-buffer size, can cause the trace_pipe read to
read stale data. Add trace_access_lock() around updating the ring
buffer sub-buffer size.
- Fix eventfs_inode on failure path in creation of the events directory
In the creation of the "events" directory, if after allocating the
eventfs_inode a failure is detected, it calls cleanup_ei() which
calls free_ei(). The free_ei() will test if eventfs_inode being freed
has no children. It is a bug if it does. But on the failure case of
the creation of the "events" directory, the children lists have not
yet been initialized and the free will trigger a warning because
list_empty() on an uninitialized list returns false.
Move the initialization into init_ei() where it makes more sense and
makes sure that a created eventfs_inode has its lists initialized
upon creation.
- Check return value of kthread_run() in ftrace direct sample code
The sample code that shows how to use the ftrace direct calls does
not test the return of kthread_run() to see if it succeeds. Return a
failure if the kthread_run() doesn't succeed.
- Clear user events state on fork in case of alloc failure
On fork, the child gets a pointer to the parent's user events state.
It makes a copy of it then updates the child's pointer to it. But if
the allocation fails, the duplication function leaves the child with
a pointer to its parent's descriptor. When the child cleans up its
data, it will free the parent's descriptor while the parent is still
using it.
In the duplication function, set the child's user_event_mm to NULL
before testing if the allocation succeeded, and when it exits it will
not free the parent's descriptor.
- Fix retry exhaustion in simple ring buffer reader swap
simple_ring_buffer_swap_reader_page() starts with retry set to 8 and
post-decrements it only after a failed link replacement. On the final
attempt, a successful replacement leaves retry at zero, while a
failed replacement leaves it at -1.
But the check for success expects the retry value to be non-zero and
exits with an error on zero. This is the opposite result. Fix it.
- Fail nicely when the remote swap_reader_page() returns an error
Currently, if the swap_reader_page() of a remote buffer fails, it
triggers a WARN_ON_ONCE() and continues normally. Instead, have it
exit with an error and a pr_warn() print instead of a full WARNING.
* tag 'trace-v7.3-2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
ring-buffer: Stop remote reader update when page swap fails
tracing: Fix retry exhaustion in simple ring buffer reader swap
tracing/user_events: Clear copied tracing state before fork duplication
samples/ftrace: Fix kthread_stop() on ERR_PTR in ftrace-direct-multi-modify
samples/ftrace: Fix kthread_stop() on ERR_PTR in ftrace-direct-modify
eventfs: Initialize ei->children and ei->list in init_ei()
tracing: Fix use-after-free in trace_pipe read on sub-buffer order change
tracing: Fix crash passing ERR_PTR to kthread_stop()
tracing: Fix use-after-free with same-name named triggers
tracing: Fix logged instance name on creation failure
|
|
|
|
Update the report and org mode messages to reflect the new
function names.
Signed-off-by: Julia Lawall <Julia.Lawall@inria.fr>
|
|
dev_put_track() and dev_hold_track() were renamed to netdev_put() and
netdev_hold() by commit d62607c3fe45 ("net: rename reference+tracking
helpers").
So update the names.
Signed-off-by: Sang-Heon Jeon <ekffu200098@gmail.com>
Signed-off-by: Julia Lawall <Julia.Lawall@inria.fr>
|
|
atomic_long_dec_and_lock() has never existed. So drop it from the rules.
No functional change.
Signed-off-by: Sang-Heon Jeon <ekffu200098@gmail.com>
Signed-off-by: Julia Lawall <Julia.Lawall@inria.fr>
|
|
vmalloc_exec() was removed by commit 7a0e27b2a0ce ("mm: remove
vmalloc_exec").
So drop it from the rules.
No functional change.
Signed-off-by: Sang-Heon Jeon <ekffu200098@gmail.com>
Signed-off-by: Julia Lawall <Julia.Lawall@inria.fr>
|
|
pci_pool_alloc() and pci_pool_zalloc() were removed by commit
88dee3b0efe4 ("PCI: Remove unused pci_pool wrappers").
So drop the pci_pool_alloc rules.
No functional change.
Signed-off-by: Sang-Heon Jeon <ekffu200098@gmail.com>
Signed-off-by: Julia Lawall <Julia.Lawall@inria.fr>
|
|
Currently a softirq may be pending longer then expected if the
triggering interrupt happens in-between hardirq_disable_enter() and
_local_interrupt_disable() in local_interrupt_disable():
local_interrupt_disable():
hardirq_disable_enter();
<interrupt>
...
__irq_exit_rcu():
// false because hardirq_disable_count() is not 0
if (.. && !hardirq_disable_count() && ..) {
invoke_softirq();
}
_local_interrupt_disable();
, it'll defer the softirq to the next interrupt which can be forever.
The order between hardirq_disable_enter() and _local_interrupt_disable()
is to optimize re-disabling interrupts if they are already disabled, but
as 1) local_interrupt_disable() is not widely used yet and 2) the proper
way to achieve this optimization may need fixing up the counter at
entry/exit time [1], so reverse the order for now to avoid the softirq
pending issue.
Because of this fix, the part of saving the current state is separated
from irq disabling, and the logic of local_interrupt_disable() becomes:
local_irq_save(flags);
if (counter++ == 0) {
this_cpu(local_interrupt_disable_state) = flags;
}
Therefore change the helper function _local_interrupt_disable() to
_local_interrupt_save_state() which only saves the current irqflags
(when interrupts get disabled the first time).
Fixes: e901c1510e24 ("irq,spin_lock: Add counted interrupt disabling/enabling")
Reported-by: Thomas Gleixner <tglx@kernel.org>
Signed-off-by: Boqun Feng <boqun@kernel.org>
Signed-off-by: Thomas Gleixner <tglx@kernel.org>
Reviewed-by: Bradley Morgan <brads@mainlining.org>
Link: https://patch.msgid.link/20260829213412.14303-1-boqun@kernel.org
Link: https://lore.kernel.org/lkml/87v78wezid.ffs@fw13/ [1]
Closes: https://lore.kernel.org/lkml/87jypbfu1t.ffs@fw13/
|
|
ntfs_trim_fs() aligns the start of a free extent up to the device discard
granularity, but derives the discard length by aligning the original extent
length down. When the free extent start is not discard-aligned, adding that
length to the aligned start can extend the discard past the free extent and
into allocated clusters.
For example, with 4 KiB clusters and 32 KiB discard granularity, the free
extent [4 KiB, 36 KiB) becomes the discard range [32 KiB, 64 KiB), so
28 KiB beyond the free extent may be discarded.
Align the absolute end of the free extent down and derive the length from
the two aligned endpoints. Skip extents that contain no full discard unit.
Reproduced with a 4 KiB-cluster NTFS filesystem on scsi_debug configured
for 32 KiB discard granularity and read-zero-after-trim. Before this
change, FITRIM zeroed seven allocated 4 KiB clusters following an unaligned
32 KiB hole. With this change, the same data remains intact across FITRIM
and remount.
Fixes: 11ccc9107dc4 ("ntfs: update runlist handling and cluster allocator")
Cc: stable@vger.kernel.org
Assisted-by: OpenAI Codex:GPT-5.6 Sol Max
Signed-off-by: Jacopo Labardi <jacopolabardi@gmail.com>
Signed-off-by: Namjae Jeon <linkinjeon@kernel.org>
|
|
WOF decompression uses four module-global workspaces, one per compression
format, each with a static mutex. ntfs_read_wof_compressed_block() takes
that mutex once and holds it across the whole chunk loop, so both block
reads run inside it:
mutex_lock(ws->lock);
for each chunk {
parse_wof_chunk_table(..., ws->input, ...); /* reads disk */
ntfs_read_wof_chunk(..., ws->input, ...); /* reads disk */
decompress into ws->output;
}
mutex_unlock(ws->lock);
Readers of system-compressed files then serialise system-wide on the disk
waits, not just on the decompressor scratch the lock exists for. One
reader sleeping in submit_bio_wait() blocks all the rest.
The waits dominate. Reading an 8 MiB xpress4k file (2048 chunks at a 48%
compressed ratio, so 2048 acquisitions and 4096 block reads) and timing
ws->lock against the part of it spent in ntfs_bdev_read():
backing store held of that in I/O held after
virtio, host page cache 348 ms 321 ms (92%) 24.6 ms
virtio, throttled 100 MB/s 978 ms 948 ms (96%) 36.6 ms
The page-cache row is a lower bound, having no seek cost at all, and the
share still grows with slower storage because only the wait scales while
decompression stays near 26 ms.
The reads are inside the lock only because they land in ws->input, a
buffer shared through the workspace. Nothing else requires it:
parse_wof_chunk_table() and ntfs_read_wof_chunk() already take the buffer
as a parameter and both set *chunk_mem to a pointer inside it, so a
caller-owned buffer works unchanged.
Allocate that buffer per call, do both reads without the lock, and take
the lock only around decompression, which is the step needing ws->output
and ws->scratch. squashfs is arranged this way already: its
squashfs_decompress() is handed a bio that has been read, and locks only
for the CPU work.
Block reads are unchanged in number, they just no longer run under the
lock, and hold time stops tracking device speed.
This also unnests two per-inode locks from the global one, runlist->lock
taken by both reads and base_ni->mrec_lock taken for a resident stream.
A resident chunk needs no I/O at all, yet used to queue behind a reader
blocked in submit_bio_wait() and then take mrec_lock inside the global
mutex.
The buffer is 4608 bytes for xpress4k and at most 33280 for lzx32k. This
path already does GFP_NOFS allocations per call in ntfs_attr_iget(), and
in ntfs_attr_get_search_ctx() for a resident stream, so one more does not
change how it behaves under memory pressure. The workspace keeps output
and scratch, 4 KiB to 32 KiB and 6224 bytes (xpress) or 10240 (lzx), and
its "already allocated" test moves from ws->input to ws->output.
The lock is now taken per chunk rather than per call, which differs only
for a folio spanning several chunks: a few more uncontended mutex
operations in exchange for not holding it across the reads between them.
Verified under QEMU against an uncompressed copy of the same data, on an
8 MiB file and a 100000 byte one, the latter covering the tail chunk that
is not a full comp_unit.
Signed-off-by: Zhan Xusheng <zhanxusheng@xiaomi.com>
Signed-off-by: Namjae Jeon <linkinjeon@kernel.org>
|
|
The LLVM commit c7f4a76da024 [1]
"[InstCombine] fold ((x - 1) | (2^k - 1)) + 1 to (x + (2^k - 1)) & -(2^k)"
caused test_task_local_data.bpf.o to fail verification:
The sequence of 8193 jumps is too complex.
processed 188770 insns (limit 1000000) max_states_per_insn 34
total_states 8238 peak_states 12330 mark_read 0
TLD_ROUND_UP(x, 8) expands to ((((x) - 1) | 7) + 1), exactly the pattern
that [1] rewrites, so the accumulation in __tld_fetch_key()
off += TLD_ROUND_UP(metadata[i].size, 8);
is now compiled as (x + 7) & -8 instead of ((x - 1) | 7) + 1. Both are
correct, but they leave the verifier in very different states. Note that
'off' is marked as precise.
Without [1], "size - 1" wraps at zero (size is a __u16), so
the verifier loses all bounds on the increment:
211: (69) r1 = *(u16 *)(r1 +62) ; R1=scalar(...,umax32=0xffff,var_off=(0x0; 0xffff))
212: (04) w1 += -1 ; R1=scalar(smin=0,smax=umax=0xffffffff,smin32=-1,smax32=0xfffe,var_off=(0x0; 0xffffffff))
213: (44) w1 |= 7 ; R1=scalar(smin=umin=umin32=7,smax=umax=0xffffffff,var_off=(0x7; 0xfffffff8))
214: (0c) w6 += w1 ; R6=scalar(smin=umin=umin32=7,smax=umax=0xffffffff,var_off=(0x7; 0xfffffff8))
215: (04) w6 += 1 ; R6=scalar(smin=0,smax=umax=umax32=0xfffffff8,var_off=(0x0; 0xfffffff8))
Note that 'w6' will be used in the next iteration. In the next iteration
after insn 215, the R6 range will be the same as previous iteration.
The iterator loop converges at depth 2.
With [1] the increment stays precisely bounded at [0, 0x10006]:
211: (69) r9 = *(u16 *)(r1 +62) ; R9=scalar(...,umax32=0xffff,var_off=(0x0; 0xffff))
212: (04) w9 += 7 ; R9=scalar(...,umax32=0x10006,var_off=(0x0; 0x1ffff))
213: (54) w9 &= 131064 ; R9=scalar(...,umax32=0x10006,var_off=(0x0; 0x1fff8))
214: (0c) w9 += w6 ; R9=scalar(...,umax32=0x10006,var_off=(0x0; 0x1fff8))
215: (bf) r1 = r10
216: (07) r1 += -8
217: (85) call bpf_iter_num_next
218: (bc) w6 = w9
In the next iteration, we will have
211: (69) r9 = *(u16 *)(r1 +62) ; R9=scalar(...,umax32=0xffff,var_off=(0x0; 0xffff))
212: (04) w9 += 7 ; R9=scalar(...,umax32=0x10006,var_off=(0x0; 0x1ffff))
213: (54) w9 &= 131064 ; R9=scalar(...,umax32=0x10006,var_off=(0x0; 0x1fff8))
214: (0c) w9 += w6 ; R9=scalar(...,umax32=0x2000c,var_off=(0x0; 0x3fff8))
...
so 'off' umax grows by 0x10006 on every iteration and the loop-head
state never repeats:
218: (bc) w6 = w9 ; R6=scalar(...,umax32=0x10006,var_off=(0x0; 0x1fff8))
218: (bc) w6 = w9 ; R6=scalar(...,umax32=0x2000c,var_off=(0x0; 0x3fff8))
218: (bc) w6 = w9 ; R6=scalar(...,umax32=0x30012,var_off=(0x0; 0x3fff8))
...
218: (bc) w6 = w9 ; R6=scalar(...,umax32=0xff95fd6,var_off=(0x0; 0xffffff8))
That last one is iterator depth 4090. Saturating umax would take ~65531
iterations; the verifier gives up long before that.
Note the loop does not diverge from the start. widen_imprecise_scalars()
blows 'off' up to an unbounded scalar while it is still imprecise, and that
alone converges the first three passes through the loop at depth 4.
Once mark_chain_precision() reaches the loop body, maybe_widen_reg() starts
skipping the register, and no widening ever happens again. In the failing
log widening fires exactly 6 times out of 4098 arrivals at the iter_next()
checkpoint, all of them before the umax starts accumulating.
With [1] and this fix, here is one full trip through the loop body,
entered with 'off' (R6) already clamped by the previous iteration:
208: frame1: R6=scalar(...,umax32=4088,var_off=(0x0; 0xff8))
208: (67) r7 <<= 6 ; R7=scalar(...,umax32=3968,var_off=(0x0; 0xfc0))
209: (bf) r1 = r9 ; R1=mem(id=54,sz=4036,imm=4)
210: (0f) r1 += r7
211: (69) r1 = *(u16 *)(r1 +62) ; R1=scalar(...,umax32=0xffff,var_off=(0x0; 0xffff))
212: (04) w1 += 7 ; R1=scalar(...,umax32=0x10006,var_off=(0x0; 0x1ffff))
213: (54) w1 &= 131064 ; R1=scalar(...,umax32=0x10006,var_off=(0x0; 0x1fff8))
214: (0c) w1 += w6 ; R1=scalar(...,umax32=0x10ffe,var_off=(0x0; 0x1fff8))
R6=scalar(...,umax32=4088,var_off=(0x0; 0xff8))
215: (bc) w6 = w1 ; R6=scalar(...,umax32=0x10ffe,var_off=(0x0; 0x1fff8))
216: (26) if w1 > 0xff8 goto pc+1 ; R6=scalar(...,umax32=4088,var_off=(0x0; 0xff8))
217: (05) goto pc-27
This makes the loop body a fixpoint. 'off' (w6) enters at 208 as [0, 4088] with
var_off=(0x0; 0xff8); the increment computed at 212/213 is [0, 0x10006], so
214/215 leave it at [0, 0x10ffe]; then 216 truncates it straight back to
[0, 4088]/(0x0; 0xff8), and only then is the back edge at 217 taken.
Convergence no longer depends on the widening window above. Verification converges
at iterator depth 3.
[1] https://github.com/llvm/llvm-project/pull/216436
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20260828170534.1011183-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
|
|
Pull more SCSI updates from Martin Petersen:
"Remaining updates for the 7.3 merge window. The only core change is
enabling context analysis for the SCSI layer and UFS.
The remaining changes are either bug fixes or hardening"
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mkp/scsi: (26 commits)
scsi: snic: Fix SCSI host leak on workqueue allocation failure
scsi: MAINTAINERS: Update my email address
scsi: MAINTAINERS: Leave the cumana_1 and oak drivers to the RISCPC maintainers
scsi: leapraid: Standardize NCQ priority sysfs attributes
scsi: leapraid: Serialize firmware log mmap with teardown
scsi: leapraid: Balance host references for firmware log VMAs
scsi: lpfc: Remove unnnecessary NULL check
scsi: qla2xxx: Fix an loop timeout test
scsi: qla2xxx: Fix an error code in qla_get_tmf()
scsi: ibmvfc: Fix use of uninitialized rport in ibmvfc_do_work()
scsi: core: Enable context analysis for hosts.o
scsi: lpfc: Replace strlcat() with sysfs_emit_at() in the sysfs show functions
scsi: lpfc: Replace strlcat() with seq_buf in the debugfs dump helpers
scsi: lpfc: Replace strlcat() with seq_buf in lpfc_rx_monitor_report()
scsi: lpfc: Replace strlcat() with scnprintf() in lpfc_vport_symbolic_node_name()
scsi: lpfc: Replace strlcat() with seq_buf in lpfc_info()
scsi: core: Enable context analysis
scsi: core: Protect host state changes with the host lock
scsi: core: Add lock context annotations
scsi: core: Pass the SCSI host pointer directly to scanning functions
...
|
|
sys_or1k_atomic() (syscall 244 in the "or1k" ABI) takes two user
pointers, v1 and v2, and swaps the words they point to in hand-written
assembly.
l.lwz r29,0(r4)
l.lwz r27,0(r5)
l.sw 0(r4),r27
l.sw 0(r5),r29
The pointers are not checked with access_ok(). The four memory
accesses also have no exception table entries.
A caller passes a kernel address as either pointer, and the syscall
reads from and writes to it directly.
This gives an unprivileged process a kernel read/write primitive. It
overwrites kernel data such as the sys_call_table, gaining code
execution in kernel context.
Check both pointers before entering the critical section. Add fixups
for the four memory accesses so faults on valid but unmapped user
addresses return -EFAULT.
[shorne@gmail.com: fix comment style]
Fixes: 9d02a4283e9c ("OpenRISC: Boot code")
Cc: stable@vger.kernel.org
Signed-off-by: Ali Ahmet Memis <ali@iusegentoo.com>
Signed-off-by: Stafford Horne <shorne@gmail.com>
|
|
When a LUN_RESET aborts a WRITE command that is in the
TRANSPORT_WRITE_PENDING state, the target core sets CMD_T_ABORTED and
waits for the frontend to finish processing.
If the initiator subsequently sends the remaining dataout PDUs,
__iscsit_check_dataout_hdr() catches the payload, stops the dataout
timer if the sequence is final and finally dumps the data. However, the
iSCSI target doesn't trigger the completion process for these aborted
commands. Because of this, the abort path hangs indefinitely in
target_put_cmd_and_wait(), leading to a deadlocked target worker thread.
Fix this by explicitly calling target_complete_cmd() when the final
dataout PDU is received for an aborted WRITE command.
target_complete_cmd() detects the CMD_T_ABORTED flag and cleanly routes
the command into target_abort_work, allowing the abort completion to
successfully unblock.
Signed-off-by: Maurizio Lombardi <mlombard@redhat.com>
Reviewed-by: Laurence Oberman <loberman@redhat.com>
Link: https://patch.msgid.link/20260717143828.76291-2-mlombard@redhat.com
Signed-off-by: Martin K. Petersen (Oracle) <mkp@kernel.org>
|
|
The Intel UFS 4.0 PCI variant (PCI ID 8086:D335) advertises MCQ support
in its capability register. However, ufshcd_alloc_mcq() also requires an
.op_runtime_config hook to locate the per-queue operation and runtime
(OPR) register blocks, which was not provided by this variant operations
table.
As a result, MCQ initialization fails and ufshcd_add_scsi_host() prints
"MCQ mode is disabled, err=%d\n" before falling back to legacy
single-doorbell (SDB) mode.
Add ufs_intel_mcq_config_resource() to initialize the MCQ configuration
base and add ufs_intel_op_runtime_config() to set up the OPR register
offsets and stride.
Wire both hooks into the variant operations table so MCQ is enabled when
supported by the hardware.
Fixes: 096cd6b7adf2 ("scsi: ufs: ufs-pci: Add support for Intel Nova Lake")
Signed-off-by: sangram kumar yerra <sangram.k.y@intel.com>
Reviewed-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Link: https://patch.msgid.link/20260818112830.453402-3-sangram.k.y@intel.com
Signed-off-by: Martin K. Petersen (Oracle) <mkp@kernel.org>
|
|
Reliable HS-Gear5 operation on Intel UFS 4.0 controllers requires
configuring PA_INITIAL_ADAPT before changing the power mode. Without
this setting, the link fails to train reliably at Gear5.
Add a pwr_change_notify() hook to configure the adaptation mode before
the power mode transition. Enable this only for UFS 4.0 and later
controllers by checking hba->ufs_version.
Wire the hook into the existing Meteor Lake family variant operations
table (ufs_intel_mtl_hba_vops) instead of introducing a separate table,
since the Intel UFS 4.0 PCI variant (PCI ID 8086:D335) already uses this
vops table and the hook is internally gated on UFS version >= 4.0.
Use PA_INITIAL_ADAPT when the negotiated TX power mode is FAST_MODE or
FASTAUTO_MODE. Otherwise, reset the adaptation mode to PA_NO_ADAPT,
which is the default setting.
Fixes: 096cd6b7adf2 ("scsi: ufs: ufs-pci: Add support for Intel Nova Lake")
Signed-off-by: sangram kumar yerra <sangram.k.y@intel.com>
Reviewed-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Link: https://patch.msgid.link/20260818112830.453402-2-sangram.k.y@intel.com
Signed-off-by: Martin K. Petersen (Oracle) <mkp@kernel.org>
|
|
SG_GET_REQUEST_TABLE reports per-request diagnostic state through
sg_req_info::problem. The field is meant to indicate whether there is an
error to report for a completed request.
sg_fill_request_table() currently combines masked_status, host_status
and driver_status with bitwise AND. This only reports a problem when all
three status fields are non-zero at the same time. A normal target check
condition, for example, has masked_status set while host_status and
driver_status may both be zero, so the request is incorrectly reported
as clean.
Use the same condition as sg_new_read(), which sets SG_INFO_CHECK when
any of the three status fields is non-zero.
Signed-off-by: Xu Rao <raoxu@uniontech.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Cc: stable@vger.kernel.org
Link: https://patch.msgid.link/26BF67F369E2123E+20260706084443.805598-1-raoxu@uniontech.com
Signed-off-by: Martin K. Petersen (Oracle) <mkp@kernel.org>
|
|
mpi3mr_get_tgtdev_by_addr() increments the target device kref when it
returns a device. If a subsequent error triggers a goto out_fail after
the tgtdev reference is acquired, the reference is never released
because the out_fail path does not call mpi3mr_tgtdev_put(). This
prevents the target device structure from ever being freed.
Add a tgtdev put in the out_fail path, guarded by a NULL check since
tgtdev is only acquired for SAS_END_DEVICE types and the same cleanup
path is shared by earlier error cases where tgtdev is still NULL.
Fixes: e22bae30667a ("scsi: mpi3mr: Add expander devices to STL")
Signed-off-by: Milan P. Gandhi <mgandhi@redhat.com>
Reviewed-by: Laurence Oberman <loberman@redhat.com>
Link: https://patch.msgid.link/20260812103344.174247-3-mgandhi@redhat.com
Signed-off-by: Martin K. Petersen (Oracle) <mkp@kernel.org>
|
|
sas_port_alloc_num() can return NULL on memory allocation failure. The
return value is passed directly to sas_port_add() without a NULL check,
which causes a NULL pointer dereference.
Additionally, if sas_port_add() fails, the allocated port is not freed
before jumping to out_fail, leaking the sas_port structure. Call
sas_port_free() to properly release it.
Fixes: e22bae30667a ("scsi: mpi3mr: Add expander devices to STL")
Signed-off-by: Milan P. Gandhi <mgandhi@redhat.com>
Reviewed-by: Laurence Oberman <loberman@redhat.com>
Link: https://patch.msgid.link/20260812103344.174247-2-mgandhi@redhat.com
Signed-off-by: Martin K. Petersen (Oracle) <mkp@kernel.org>
|
|
The current devfreq downdifferential threshold of 5% causes overly
aggressive frequency downscaling, leading to performance degradation
sometimes during sequential read workloads.
Update the UFS devfreq downdifferential threshold to 65. This widens
the hysteresis window and prevents overly aggressive downscaling,
ensuring that frequency is maintained for loads above 5% and scaling
down occurs only when utilization falls below this level, while scale-up
still triggers above the 70% threshold.
Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
Signed-off-by: Nitin Rawat <nitin.rawat@oss.qualcomm.com>
Link: https://patch.msgid.link/20260825145203.265579-3-nitin.rawat@oss.qualcomm.com
Signed-off-by: Martin K. Petersen (Oracle) <mkp@kernel.org>
|
|
controller v6.2+
The link startup mode (HS LSS - high-speed link startup, or LS LSS -
low-speed link startup) is decided in the boot stage based on the
bootconfig GPIO. This selection is carried forward through the secondary
stage bootloaders and finally to HLOS via the spare configuration
register (REG_UFS_DEBUG_SPARE_CFG).
On Qualcomm UFS controller v6.2 and later, bit 31 in the spare
configuration register indicates the high-speed link startup mode
selection, as per the Hardware Programming Guide (HPG).
The spare register value is read during host driver initialization but
gets cleared after UFS reset. Preserve the spare register value during
initialization and restore it during link startup to maintain the
bootloader-configured link startup mode.
Signed-off-by: Nitin Rawat <nitin.rawat@oss.qualcomm.com>
Tested-by: Mukesh Ojha <mukesh.ojha@oss.qualcomm.com>
Link: https://patch.msgid.link/20260825145203.265579-2-nitin.rawat@oss.qualcomm.com
Signed-off-by: Martin K. Petersen (Oracle) <mkp@kernel.org>
|
|
Commit 249313b3f7b5 ("scsi: ibmvfc: allocate targets based on protocol")
added a protocol parameter to ibmvfc_alloc_target() but did not describe
it in the function's kernel-doc comment, so a W=1 build warns:
drivers/scsi/ibmvscsi/ibmvfc-core.c:4996: warning: Function parameter
or struct member 'protocol' not described in 'ibmvfc_alloc_target'
Add the missing parameter description.
Fixes: 249313b3f7b5 ("scsi: ibmvfc: allocate targets based on protocol")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202608270829.lHI1FAdO-lkp@intel.com/
Signed-off-by: Muhammad Falak R Wani <falakreyaz@gmail.com>
Reviewed-by: Dave Marquardt <davemarq@linux.ibm.com>
Acked-by: Tyrel Datwyler <tyreld@linux.ibm.com>
Link: https://patch.msgid.link/b073968ae020b6ae0240e91341a92f428587ebd9.1787828961.git.falakreyaz@gmail.com
Signed-off-by: Martin K. Petersen (Oracle) <mkp@kernel.org>
|
|
Commit e0fca728a89f ("scsi: ibmvfc: delete NVMe/FC targets as well as
SCSI") renamed ibmvfc_relogin() to ibmvfc_scsi_relogin() but left the
kernel-doc comment referring to the old name, so a W=1 build warns:
drivers/scsi/ibmvscsi/ibmvfc-core.c:1901: warning: expecting prototype
for ibmvfc_relogin(). Prototype was for ibmvfc_scsi_relogin() instead
Update the kernel-doc comment to use the current function name.
Fixes: e0fca728a89f ("scsi: ibmvfc: delete NVMe/FC targets as well as SCSI")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202608271026.iMLmrwz4-lkp@intel.com/
Signed-off-by: Muhammad Falak R Wani <falakreyaz@gmail.com>
Reviewed-by: Dave Marquardt <davemarq@linux.ibm.com>
Acked-by: Tyrel Datwyler <tyreld@linux.ibm.com>
Link: https://patch.msgid.link/dd866cf2321381694af027fbd726bcbd63ac3751.1787828961.git.falakreyaz@gmail.com
Signed-off-by: Martin K. Petersen (Oracle) <mkp@kernel.org>
|
|
pm8001_request_msix() unwinds previously registered handlers with
free_irq() when request_irq() fails. The rollback loop uses the failing
index i for every iteration instead of the already registered vector
index j.
That passes the wrong IRQ/dev_id pair to free_irq() and leaves the
earlier handlers installed. Use j for both pci_irq_vector() and the
matching irq_vector entry in the rollback loop.
Fixes: a76037ff3479 ("scsi: pm8001: switch to pci_irq_alloc_vectors")
Cc: stable@vger.kernel.org
Assisted-by: Codex:GPT-5
Signed-off-by: Runyu Xiao <runyu.xiao@seu.edu.cn>
Acked-by: Jack Wang <jinpu.wang@cloud.ionos.com>
Link: https://patch.msgid.link/20260824113618.2239100-1-runyu.xiao@seu.edu.cn
Signed-off-by: Martin K. Petersen (Oracle) <mkp@kernel.org>
|
|
nvfnic_add_lport() declares struct nvme_fc_port_info on the stack and
fills in four of its five members, leaving dev_loss_tmo holding whatever
the stack happened to contain before the call. The structure is then
handed to nvme_fc_register_localport().
nvfnic_add_tport(), which registers the remote port a few lines further
down, memsets its own struct nvme_fc_port_info first, so only the local
port path passes uninitialized data across the transport interface.
The NVMe/FC transport documents dev_loss_tmo as "Used only on a
remoteport" and does not read it in nvme_fc_register_localport(), so
there is no behavioural change today. Initialize the structure anyway:
the driver must not depend on which members the transport happens to
consume, and any member added to struct nvme_fc_port_info later would
silently start out as stack garbage.
Signed-off-by: Linmao Li <lilinmao@kylinos.cn>
Tested-by: Karan Tilak Kumar <kartilak@cisco.com>
Reviewed-by: Karan Tilak Kumar <kartilak@cisco.com>
Link: https://patch.msgid.link/20260819114242.3598034-2-lilinmao@kylinos.cn
Signed-off-by: Martin K. Petersen (Oracle) <mkp@kernel.org>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux
Pull io_uring fixes from Jens Axboe:
"A few smaller fixes for io_uring that should go into the 7.3-rc1
kernel, all three headed to stable as well. This contains:
- A few fixes around cancellation and teardown for waitid
- Cap the user size for the query interface copy-out"
* tag 'io_uring-7.3-20260828' of git://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux:
io_uring/waitid: avoid siginfo copy during ring teardown
io_uring/waitid: honor task_work cancellation
io_uring/query: cap user size passed to copy_struct_to_user
|
|
Pull more drm updates from Dave Airlie:
"As mentioned last week, an msm pull request fell down the side of the
couch or whatever the email equivalent of that is. This has the msm
next stuff + the usual fixes for amd/intel.
core:
- use drm_warn instead of warn
msm:
- Bindings:
- Added Shikra support
- Document a840, a704, a722
- Core:
- Use drm_client buffers for fbdev emulation
- teardown fixes
- ARM32 DMA fixup
- Remove objects from evict list when re-validated
- Bunch of corner case and error path fixes
- DPU:
- Dropped dev_pm_opp_set_rate(0) preventing burnout
- Fixed SSPP offsets of Kaanapali
- DP:
- Dropped dev_pm_opp_set_rate(0) preventing burnout
- Cleaned up core code in preparation for MST support
- Fixed prepare() to let Pipewire continue in case of the unplugged cable
- GPU:
- Add support for a704
- Add support for a722
- HDMI:
- Simplifed register access
amdgpu:
- eGPU fixes
- Runtime PM fix
- UserQ fixes
- Backlight fix
- Discovery sysfs fix
- Reset handling fixes
- Buffer func handling fix for xgmi
- VCN boundary check fix
- DC lut handling fixes
- MES fixes
- UVD fix
- VCE 3 fix
- Enforce isolation fix
- HPD fix for VGA/LVDS
- DML fix
- DCN 6 fixes
- DC gpu reset fix
amdkfd:
- Fix return value
- CU occupancy for GFX 11
- CU occupancy for GFX 12/12.1
- Queue bounds checking fix
- SVM fixes
- CRIU bounds checking fix
radeon:
- iMac display fix
xe:
- error message cleanups
- i2c global register definitions as dependency for xe/i2c fixes
- Media workardound
- Add CCS to gt_idle debugfs print
- Page fault related fix
- i2c related fixes
- System Controller mailbox bit fix"
* tag 'drm-next-2026-08-29' of https://gitlab.freedesktop.org/drm/kernel: (121 commits)
drm/xe/sysctrl: Read mailbox phase bit from hardware
drm/xe/i2c: Keep the i2c controller always enabled
drm/xe/i2c: Fix the interrupt handling
i2c: designware: Global register definitions
drm/xe: Reject page faults from non-fault-mode scratch VMs
drm/xe/xe_gt_idle: Add CCS to the powergating info print
drm/xe: Do not apply WA 14025883347 to media 3503
drm/amd/display: fix dc_lock leak on GPU reset error paths
drm/amd/display: Fix redundant GPUVMEnable checks in dcn6 flip schedule
drm/amd/display: Fix wrong bytes-per-pixel value for dml2_422_packed_10
drm/amdkfd: guard against NULL restore_mqd in CRIU queue restore
drm/amdgpu/userq: fix lock missing for userq fence error set
drm/amdkfd: Fix the case that vm range is hole at svm_migrate_copy_to_vram
drm/amdkfd: Fix error path at svm_migrate_copy_to_ram
drm/amd/display: Log details when failing to register HPD IRQ
drm/amd/display: Fix HPD consideration for VGA/LVDS connectors on DCE
drm/amdgpu: clamp the isolation index for rings outside a partition
drm/amdkfd: Reject zero-sized AQL queue allocations after size halving
drm/amdgpu: Fix VCE 3 ring align_mask
drm/kfd: Add CU occupancy support to GFX12.1
...
|
|
Pull ipmi updates from Corey Minyard:
"Several cleanup on error fixes and a missing RCU wait and proper
validation on a received message in one place.
The biggest change is the initialization of the driver can be done
asynchronously on a work queue. That saves significant boot time"
* tag 'for-linus-7.3-1' of https://github.com/cminyard/linux-ipmi:
ipmi: Fix use-after-free of cmd_rcvr in _ipmi_destroy_user()
ipmi:msghandler: Cancel work cleanly on an error
ipmi:si: Add async init to ipmi_si
char: ipmi: use named initializers for acpi_device_id
ipmi: Fix leak in __ipmi_bmc_register
ipmi: Remove all sysfs files on registration failure
ipmi: si: Fix NULL pointer dereference after failed registration
ipmi: ipmb: validate write message length
|
|
The SCTP_CMD_TIMER_START handler checks timer_pending() before calling
timer_reduce(). The timer can expire and detach between these operations,
causing timer_reduce() to rearm the timer without taking the association
reference required for the newly armed timer.
The timer callback later unconditionally drops its association reference,
which can leave the association reference count unbalanced and result in
use-after-free during association teardown.
Use the return value of timer_reduce() to determine whether the timer was
actually armed. Take the association reference only when timer_reduce()
successfully starts a new timer, closing the race between checking the
timer state and rearming it.
This issue was reported by Nico Yip (@_cyeaa_) working with TrendAI Zero
Day Initiative.
Fixes: 20a785aa52c8 ("sctp: Don't add the shutdown timer if its already been added")
Reported-by: Zero Day Initiative <zdi-disclosures@trendmicro.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Link: https://patch.msgid.link/9d8f1b5c50329d5ea7c642128d35681abaa9ed20.1787773744.git.lucien.xin@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Cen Zhang says:
====================
tcp: fix use-after-free in do_tcp_getsockopt()
do_tcp_getsockopt() has two lockless reads of icsk_ca_ops. Since BPF
struct_ops congestion control made icsk_ca_ops point to dynamically
allocated memory, a concurrent setsockopt(TCP_CONGESTION) can replace
the pointer and free the old object while either reader is using it.
Patch 1 fixes the TCP_CONGESTION path by copying ca_ops->name to a
stack buffer while holding rcu_read_lock(). It also uses READ_ONCE()
for the lockless load and annotates the relevant icsk_ca_ops stores
with WRITE_ONCE().
Patch 2 fixes the TCP_CC_INFO path by keeping the READ_ONCE() load,
ca_ops->get_info lookup, and call inside an RCU read-side critical
section.
====================
Link: https://patch.msgid.link/cover.1787870710.git.blbllhy@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
do_tcp_getsockopt() reads icsk->icsk_ca_ops and dereferences the
get_info function pointer without rcu_read_lock(). With BPF struct_ops
congestion control, ca_ops can point to dynamically allocated memory
that is freed concurrently, resulting in a use-after-free when the
kernel dereferences or calls through the stale pointer.
BUG: KASAN: slab-use-after-free in do_tcp_getsockopt+0x2037/0x23e0
Read of size 8 at addr ffff888013701258 by task exploit/149
do_tcp_getsockopt+0x2037/0x23e0 (net/ipv4/tcp.c:4564)
tcp_getsockopt+0x91/0xf0
__sys_getsockopt+0xf7/0x170
Fix this by wrapping the ca_ops load and get_info call within
rcu_read_lock()/rcu_read_unlock(), and using READ_ONCE() to load
the icsk_ca_ops pointer.
Fixes: 0baf26b0fcd7 ("bpf: tcp: Support tcp_congestion_ops in bpf")
Suggested-by: Eric Dumazet <edumazet@google.com>
Cc: AutonomousCodeSecurity@microsoft.com
Cc: stable@vger.kernel.org
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Cen Zhang (Microsoft Security FORGE Labs) <blbllhy@gmail.com>
Reviewed-by: Jiayuan Chen <jiayuan.chen@linux.dev>
Reviewed-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
Link: https://patch.msgid.link/65fd3816ed5d541d9edd4bf4fcf97104a2cf907a.1787870710.git.blbllhy@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
do_tcp_getsockopt() reads icsk->icsk_ca_ops->name without holding
rcu_read_lock(). Since commit 0baf26b0fcd7 ("bpf: tcp: Support
tcp_congestion_ops in bpf"), icsk_ca_ops can point to dynamically
allocated BPF struct_ops memory that may be freed concurrently via
setsockopt(TCP_CONGESTION), leading to a use-after-free.
BUG: KASAN: slab-use-after-free in _copy_to_user+0x37/0x60
Read of size 16 at addr ffff888013505260 by task exploit/149
_copy_to_user+0x37/0x60
do_tcp_getsockopt+0x158a/0x2460 (net/ipv4/tcp.c:4585)
tcp_getsockopt+0x91/0xf0
__sys_getsockopt+0xf7/0x170
Fix this by holding rcu_read_lock() around the ca_ops->name access,
using READ_ONCE() to load icsk_ca_ops, and copying the name to a
stack buffer before releasing the lock. Also annotate the relevant
icsk_ca_ops stores with WRITE_ONCE() to fix the accompanying KCSAN
data-race issue.
Fixes: 0baf26b0fcd7 ("bpf: tcp: Support tcp_congestion_ops in bpf")
Suggested-by: Eric Dumazet <edumazet@google.com>
Reported-by: Xiang Mei (Microsoft) <xmei5@asu.edu>
Link: https://lore.kernel.org/all/20260821182449.79785-2-blbllhy@gmail.com/
Cc: AutonomousCodeSecurity@microsoft.com
Cc: stable@vger.kernel.org
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Cen Zhang (Microsoft Security FORGE Labs) <blbllhy@gmail.com>
Reviewed-by: Jiayuan Chen <jiayuan.chen@linux.dev>
Reviewed-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
Reviewed-by: Breno Leitao <leitao@debian.org>
Link: https://patch.msgid.link/d3f97f1acbf0010898148be6e6406e4b8b4a5c84.1787870710.git.blbllhy@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Victor Nogueira says:
====================
net/sched: Fix remaining actions notification accounting issues
Commit 8e2efb3f45a5 ("net/sched: add get_fill_size callbacks for actions
missing them") fixed the reported echo/notify skb overrun and noted that
the pre-existing issues Sashiko pointed out [1] would be fixed separately.
This is that series.
Patch 1 makes tcf_action_shared_attrs_size() a real upper bound again.
TCA_ACT_IN_HW_COUNT and TCA_STATS_BASIC_HW are emitted on every action
dump and were never budgeted; TCA_STATS_PKT64 was budgeted once but can be
emitted twice; TCA_ACT_USED_HW_STATS and the rate estimator attributes are
conditional and also unaccounted.
Patch 2 makes the RTM_GETACTION path use the attr_size that tca_action_gd()
already computes and then throws away, so that "tc actions get" can read
back an action whose dump is larger than NLMSG_GOODSIZE.
Patch 3 wraps the reoffload delete notification in
tcf_action_full_attrs_size() like every other notification path, and stops
leaving a skip_sw action installed when that notification cannot be built.
[1] https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260810164357.1653956-1-victor%40mojatatu.com
====================
Link: https://patch.msgid.link/20260824153903.4143642-1-victor@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
tcf_reoffload_del_notify_msg() sizes the RTM_DELACTION skb with
tcf_action_fill_size(action) alone. Unlike every other notification path
it never wraps that in tcf_action_full_attrs_size(), so the nlmsg_put()
header, struct tcamsg and the TCA_ACT_TAB nest that tca_get_fill() emits -
24 bytes on x86_64 - are not budgeted. As long as the single action stays
well under NLMSG_GOODSIZE the floor in alloc_skb() hides this, but once its
fill size crosses NLMSG_GOODSIZE the allocation is exactly 24 bytes short
and tca_get_fill() runs out of tailroom. That is now easy to reach for an
offloadable act_pedit with a large tcfp_nkeys, which commit 8e2efb3f45a5
("net/sched: add get_fill_size callbacks for actions missing them") started
accounting for properly.
When that happens tcf_reoffload_del_notify() returns early, before
tcf_idr_release_unsafe(), and tcf_action_reoffload_cb() discards the return
value:
if (tc_act_skip_sw(p->tcfa_flags) && !tc_act_in_hw(p))
tcf_reoffload_del_notify(net, p);
The action has just lost its last hardware instance and is skip_sw, so it
is left installed while processing no packets, and with no notification to
tell userspace about it. An -ENOBUFS from alloc_skb() gets the same
treatment.
Fix this by budgeting the message header the way the add and delete paths
do, and release the action even when the notification cannot be built -
dropping the notification is strictly better than leaking a dead action,
and there is no caller left to report the error to.
Fixes: 13926d19a11e ("flow_offload: add reoffload process to update hw_count")
Reported-by: Sashiko <sashiko-bot@kernel.org>
Closes: https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260810164357.1653956-1-victor%40mojatatu.com
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Victor Nogueira <victor@mojatatu.com>
Reviewed-by: Pedro Tammela <pctammela@mojatatu.com>
Link: https://patch.msgid.link/20260824153903.4143642-4-victor@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
tca_action_gd() already walks every requested action and accumulates
attr_size += tcf_action_fill_size(act), then wraps the result in
tcf_action_full_attrs_size(). For RTM_DELACTION that value is handed to
tcf_del_notify_msg(), which allocates max(attr_size, NLMSG_GOODSIZE). For
RTM_GETACTION it is silently discarded and tcf_get_notify() allocates a
fixed NLMSG_GOODSIZE skb instead.
Any action whose dump exceeds that fixed budget therefore cannot be read
back. For example, act_pedit overruns the budget with 32 actions of four
munge keys each, act_police with 32 policers once the optional
rate/peakrate/result/avrate attributes are present
Fix this by passing attr_size through and allocate the reply the way the
add and delete paths do.
Note on exposure: RTM_GETACTION is the only one of the three action
commands that is not capability checked - tc_ctl_action() requires
CAP_NET_ADMIN for RTM_NEWACTION and RTM_DELACTION only - so this turns a
fixed NLMSG_GOODSIZE reply into a user sized allocation on an
unprivileged path. It is bounded by TCA_ACT_MAX_PRIO actions per
request, and tca_action_gd() does not reject duplicate indices, so a
single large action can be requested 32 times; an act_bpf program near
BPF_MAXINSNS is about 32KB of dump, or roughly 1MB for one request.
Creating such an action still requires CAP_NET_ADMIN, and the add and
delete paths have sized their skbs this way since the Fixes commit.
Should this ever need bounding, GFP_KERNEL_ACCOUNT would charge the
reply to the caller's memcg.
Fixes: 4e76e75d6aba ("net sched actions: calculate add/delete event message size")
Reported-by: Sashiko <sashiko-bot@kernel.org>
Closes: https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260810164357.1653956-1-victor%40mojatatu.com
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Victor Nogueira <victor@mojatatu.com>
Link: https://patch.msgid.link/20260824153903.4143642-3-victor@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
tcf_action_shared_attrs_size() is supposed to return an upper bound on the
netlink attributes every action dump emits outside of TCA_ACT_OPTIONS, so
that tcf_add_notify_msg(), tcf_del_notify_msg() and friends can allocate
an skb large enough for the reply. It has fallen behind the dump path and
is now an underestimate for every single action.
Attributes, such as, TCA_ACT_IN_HW_COUNT and TCA_STATS_BASIC_HW are
emitted unconditionally and never accounted for. TCA_STATS_PKT64,
TCA_ACT_USED_HW_STATS, TCA_STATS_RATE_EST, TCA_STATS_RATE_EST64 require
specific conditions, but are also not accounted for.
Fix the issue by budgeting all of them so that we have a legitimate
upper bound. Even tough for of them require specific conditions, they
are cheap so, to avoid overcomplicating, we opted to account for them
unconditionally as well to account for a real worst case scenario.
Fixes: 4e76e75d6aba ("net sched actions: calculate add/delete event message size")
Reported-by: Sashiko <sashiko-bot@kernel.org>
Closes: https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260810164357.1653956-1-victor%40mojatatu.com
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Victor Nogueira <victor@mojatatu.com>
Link: https://patch.msgid.link/20260824153903.4143642-2-victor@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Syzbot reported a crash in qdisc_pkt_len_segs_init() caused by a stale
transport_header offset after tunnel decapsulation.
BUG: unable to handle page fault for address: ffffed102091a42e
Oops: Oops: 0000 [#1] SMP KASAN NOPTI
CPU: 0 UID: 0 PID: 340 Comm: qdisc_uaf_repro Not tainted 7.2.0-rc4-00061-g248951ddc14d #256 PREEMPT(full)
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-debian-1.16.3-2 04/01/2014
RIP: 0010:__asan_load2
<IRQ>
qdisc_pkt_len_segs_init (net/core/dev.c:4145)
__dev_queue_xmit (net/core/dev.c:4787)
br_dev_queue_push_xmit (net/bridge/br_forward.c:53)
br_handle_frame_finish (net/bridge/br_input.c:229)
br_handle_frame (net/bridge/br_input.c:315)
__netif_receive_skb_core.constprop.0 (net/core/dev.c:6099)
__netif_receive_skb_list_core (net/core/dev.c:6287)
netif_receive_skb_list_internal (net/core/dev.c:6445)
napi_complete_done (net/core/dev.c:6813)
gro_cell_poll (net/core/gro_cells.c:74)
__napi_poll (net/core/dev.c:7735)
net_rx_action (net/core/dev.c:7798 net/core/dev.c:7955)
handle_softirqs (kernel/softirq.c:622)
do_softirq (kernel/softirq.c:523 kernel/softirq.c:510 )
__local_bh_enable_ip (kernel/softirq.c:450)
tun_get_user (drivers/net/tun.c:1986 (discriminator 1))
tun_chr_write_iter (drivers/net/tun.c:2032)
The issue is completely latent until qdisc read transport header in
commit 7fb4c1967011 ("net: pull headers in qdisc_pkt_len_segs_init()").
The crash requires four conditions to line up:
1. The incoming packet is encapsulated and carries GSO metadata. The outer
transport header offset is stored in skb->transport_header while the
packet is still in the outer tunnel context.
2. The tunnel receiver strips the outer headers. skb->data is advanced to
the inner frame, but skb->transport_header is left pointing to the
now-removed outer L4 header, so it becomes a negative offset relative to
the new data.
3. The inner frame is not delivered to the local IP stack. Instead, it
is forwarded at L2 by a bridge or HSR, so ip_rcv_core() never runs and
the transport header is not reset to the inner L4 offset.
4. The forwarding path calls __dev_queue_xmit(), which enters
qdisc_pkt_len_segs_init(). That function computes the GSO header length
from skb_transport_offset(skb). Because the offset is negative, the
unsigned cast overflows and pskb_may_pull(skb, hdr_len +
sizeof(struct tcphdr)) reads past the end of the skb, triggering a
KASAN fault or page fault.
The issue specifically requires GSO packets (shinfo->gso_size != 0), which
are processed/aggregated through gro_cells. Fix this by clearing
transport_header to the ~0U sentinel in gro_cell for all tunnnel driver.
GTP does not support GRO/GSO, drop the evil GSO packets in GTP directly.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Reported-by: syzbot+83181a31faf9455499c5@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/all/69de2bee.a00a0220.475f0.0041.GAE@google.com/T/
Suggested-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Dong Chenchen <dongchenchen2@huawei.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20260825123909.1463121-1-dongchenchen2@huawei.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Jerome Tollet says:
====================
net/mlx5e: Prevent stale XSK buffer release on refill retries
Prevent duplicate XSK buffer release when a deferred RX refill fails and
the same WQE is retried.
Patch 1 fixes legacy cyclic RQ. It is unchanged from v3 and retains
Dragos' Reviewed-by tag.
Patch 2 fixes the analogous striding-RQ MPWQE path. Following Dragos'
review, it now fills skip_release_bitmap in the common error path of
mlx5e_xsk_alloc_rx_mpwqe(), consistently with mlx5e_alloc_rx_mpwqe().
Targeted fault injection covered both an early allocation failure and a
partial 8-of-16-buffer unwind. With three consecutive failures for one
MPWQE, the original 16 XSK buffers were released only once, retries saw a
full bitmap, and a later successful allocation cleared it. A clean
20-second AF_XDP zero-copy pressure run exercised 1,575,262 buffer
allocation failures without invalid descriptors, WQE errors, or kernel
warnings.
====================
Link: https://patch.msgid.link/20260824141645.23700-1-jtollet@cisco.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
With AF_XDP on a striding RQ, mlx5e defers releasing XSK buffers until
an MPWQE is refilled. If XSK allocation then returns -ENOMEM,
actual_wq_head is not advanced and a later NAPI poll retries the same
WQE.
mlx5e_free_rx_mpwqe() leaves each released slot marked as releasable. On
retry it can therefore call xsk_buff_free() again through stale pointers
after the frames have returned to the XSK pool and been reallocated.
Set all skip_release_bitmap bits in the common error path of
mlx5e_xsk_alloc_rx_mpwqe(). This matches mlx5e_alloc_rx_mpwqe(). A
successful allocation already clears the bitmap after replacing every
buffer, so retries become idempotent without changing the success path.
Fault injection forced three consecutive failures for one selected MPWQE.
Both an early allocation failure and a partial 8-of-16-buffer unwind
released the original 16 XSK buffers only once. Each error left a full
bitmap, the following NAPI retry skipped the release, and a later
successful allocation cleared it. A 20-second AF_XDP zero-copy pressure
run exercised 1,575,262 buffer allocation failures without invalid
descriptors, WQE errors, or kernel warnings.
Fixes: 4c2a13236807 ("net/mlx5e: RX, Defer page release in striding rq for better recycling")
Cc: stable@vger.kernel.org
Signed-off-by: Jerome Tollet <jtollet@cisco.com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Link: https://patch.msgid.link/20260824141645.23700-3-jtollet@cisco.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
When an XDP redirect to an AF_XDP socket fails because its RX ring is
full, the XSK core frees the buffer. During the subsequent batched refill
of a legacy cyclic RQ, mlx5e also releases the WQE's XSK buffer before
allocating a replacement. If that refill succeeds only partially, a WQE
left without a replacement retains its old buffer pointer.
The buffer can meanwhile be allocated to another WQE. A later refill
retry can then free the live buffer through the stale pointer and publish
the same UMEM frame twice.
Mark the WQE as released immediately after the driver-side free. The flag
is already cleared when a replacement buffer is assigned, so refill
retries no longer release stale pointers.
The failure is silent and produces no kernel warning or splat. A
standalone legacy cyclic-RQ zero-copy libxsk reproducer, using 64-byte UDP
traffic offered at 12 Mpps, detected it: stock stopped after 2,854,914
packets in 4.094 seconds, with 4,542 xdp_rx_ring_full events and 64
ownership/double-publication errors. With this change it processed
356,904,225 packets in 30 seconds despite 571,405 xdp_rx_ring_full events,
with no ownership or data errors.
Fixes: 3f93f82988bc ("net/mlx5e: RX, Defer page release in legacy rq for better recycling")
Cc: stable@vger.kernel.org
Suggested-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Signed-off-by: Jerome Tollet <jtollet@cisco.com>
Link: https://patch.msgid.link/20260824141645.23700-2-jtollet@cisco.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
