Jeff Layton <jlayton@kernel.org> says:
This patch series is intended to improve write performance with
RWF_DONTCACHE. This version fixes additional stat accounting issues
found during review: integer promotion on 32-bit, cgroup writeback
domain migration, folio split flag preservation, and a UAF that could
occur in filemap_dontcache_kick_writeback().
* patches from https://patch.msgid.link/20260511-dontcache-v7-0-2848ddce8090@kernel.org:
mm: kick writeback flusher for IOCB_DONTCACHE with targeted dirty tracking
mm: track DONTCACHE dirty pages per bdi_writeback
mm: preserve PG_dropbehind flag during folio split
Link: https://patch.msgid.link/20260511-dontcache-v7-0-2848ddce8090@kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
The internal microphone on ASUS VivoBook X509DAP (subsystem ID
0x1043:0x197e) is not detected without a quirk entry. Add
ALC256_FIXUP_ASUS_MIC_NO_PRESENCE to fix the issue.
Signed-off-by: Andrei Faleichyk <andrei.faleichyk@noogadev.com>
Link: https://patch.msgid.link/20260603213313.6298-1-andrei.faleichyk@noogadev.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The IOCB_DONTCACHE writeback path in generic_write_sync() calls
filemap_flush_range() on every write, submitting writeback inline in
the writer's context. Perf lock contention profiling shows the
performance problem is not lock contention but the writeback submission
work itself — walking the page tree and submitting I/O blocks the writer
for milliseconds, inflating p99.9 latency from 23ms (buffered) to 93ms
(dontcache).
Replace the inline filemap_flush_range() call with a flusher kick that
drains dirty pages in the background. This moves writeback submission
completely off the writer's hot path.
To avoid flushing unrelated buffered dirty data, add a dedicated
WB_start_dontcache bit and wb_check_start_dontcache() handler that uses
the per-wb WB_DONTCACHE_DIRTY counter to determine how many pages to
write back. The flusher writes back that many pages from the oldest dirty
inodes (not restricted to dontcache-specific inodes). This helps
preserve I/O batching while limiting the scope of expedited writeback.
Like WB_start_all, the WB_start_dontcache bit coalesces multiple
DONTCACHE writes into a single flusher wakeup without per-write
allocations. Use test_and_clear_bit to atomically consume the kick
request before reading the dirty counter and starting writeback, so that
concurrent DONTCACHE writes during writeback can re-set the bit and
schedule a follow-up flusher run.
Read the dirty counter with wb_stat_sum() (aggregating per-CPU batches)
rather than wb_stat() (which reads only the global counter) to ensure
small writes below the percpu batch threshold are visible to the flusher.
In filemap_dontcache_kick_writeback(), set the WB_start_dontcache bit
inside the unlocked_inode_to_wb_begin/end section for correct cgroup
writeback domain targeting, but defer the wb_wakeup() call until after
the section ends, since wb_wakeup() uses spin_unlock_irq() which would
unconditionally re-enable interrupts while the i_pages xa_lock may still
be held under irqsave during a cgroup writeback switch. Pin the wb with
wb_get() inside the RCU critical section before calling wb_wakeup()
outside it, since cgroup bdi_writeback structures are RCU-freed and the
wb pointer could become invalid after unlocked_inode_to_wb_end() drops
the RCU read lock.
Also add WB_REASON_DONTCACHE as a new writeback reason for tracing
visibility.
dontcache-bench results (same host, T6F_SKL_1920GBF, 251 GiB RAM,
xfs on NVMe, fio io_uring):
Buffered and direct I/O paths are unaffected by this patchset. All
improvements are confined to the dontcache path:
Single-stream throughput (MB/s):
Before After Change
seq-write/dontcache 298 897 +201%
rand-write/dontcache 131 236 +80%
Tail latency improvements (seq-write/dontcache):
p99: 135,266 us -> 23,986 us (-82%)
p99.9: 8,925,479 us -> 28,443 us (-99.7%)
Multi-writer (4 jobs, sequential write):
Before After Change
dontcache aggregate (MB/s) 2,529 4,532 +79%
dontcache p99 (us) 8,553 1,002 -88%
dontcache p99.9 (us) 109,314 1,057 -99%
Dontcache multi-writer throughput now matches buffered (4,532 vs
4,616 MB/s).
32-file write (Axboe test):
Before After Change
dontcache aggregate (MB/s) 1,548 3,499 +126%
dontcache p99 (us) 10,170 602 -94%
Peak dirty pages (MB) 1,837 213 -88%
Dontcache now reaches 81% of buffered throughput (was 35%).
Competing writers (dontcache vs buffered, separate files):
Before After
buffered writer 868 433 MB/s
dontcache writer 415 433 MB/s
Aggregate 1,284 866 MB/s
Previously the buffered writer starved the dontcache writer 2:1.
With per-bdi_writeback tracking, both writers now receive equal
bandwidth. The aggregate matches the buffered-vs-buffered baseline
(863 MB/s), indicating fair sharing regardless of I/O mode.
The dontcache writer's p99.9 latency collapsed from 119 ms to
33 ms (-73%), eliminating the severe periodic stalls seen in the
baseline. Both writers now share identical latency profiles,
matching the buffered-vs-buffered pattern.
The per-bdi_writeback dirty tracking dramatically reduces peak dirty
pages in dontcache workloads, with the 32-file test dropping from
1.8 GB to 213 MB. Dontcache sequential write throughput triples and
multi-writer throughput reaches parity with buffered I/O, with tail
latencies collapsing by 1-2 orders of magnitude.
Assisted-by: Claude:claude-opus-4-6
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Link: https://patch.msgid.link/20260511-dontcache-v7-3-2848ddce8090@kernel.org
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
Signed-off-by: Christian Brauner (Amutable) <brauner@kernel.org>
Add a per-wb WB_DONTCACHE_DIRTY counter that tracks the number of dirty
pages with the dropbehind flag set (i.e., pages dirtied via RWF_DONTCACHE
writes).
Increment the counter alongside WB_RECLAIMABLE in folio_account_dirtied()
when the folio has the dropbehind flag set, and decrement it in
folio_clear_dirty_for_io() and folio_account_cleaned(). Also decrement it
when a non-DONTCACHE lookup atomically clears the dropbehind flag on a
dirty folio in __filemap_get_folio_mpol(), using folio_test_clear_dropbehind()
to prevent concurrent lookups from double-decrementing the counter, and
guarding the decrement with mapping_can_writeback() to match the increment
path.
Transfer the counter alongside WB_RECLAIMABLE in inode_do_switch_wbs() so
that the stat is properly migrated when an inode switches cgroup writeback
domains.
The counter will be used by the writeback flusher to determine how many
pages to write back when expediting writeback for IOCB_DONTCACHE writes,
without flushing the entire BDI's dirty pages.
Suggested-by: Jan Kara <jack@suse.cz>
Assisted-by: Claude:claude-opus-4-6
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Link: https://patch.msgid.link/20260511-dontcache-v7-2-2848ddce8090@kernel.org
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
Signed-off-by: Christian Brauner (Amutable) <brauner@kernel.org>
__split_folio_to_order() copies page flags from the original folio to
newly created sub-folios using an explicit allowlist, but PG_dropbehind
is not included. When a large folio with PG_dropbehind set is split,
only the head sub-folio retains the flag; all tail sub-folios silently
lose it and will not be reclaimed eagerly after writeback completes.
Add PG_dropbehind to the flag copy mask so that the drop-behind hint
is preserved across folio splits.
Fixes: a323281cdfec ("mm: add PG_dropbehind folio flag")
Assisted-by: Claude:claude-opus-4-6
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Link: https://patch.msgid.link/20260511-dontcache-v7-1-2848ddce8090@kernel.org
Acked-by: David Hildenbrand (Arm) <david@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Christian Brauner (Amutable) <brauner@kernel.org>
Use the kernel's PAGE_ALIGN() macro instead of open-coding the page
alignment calculation. This improves code readability and follows
kernel coding style.
The manual calculation:
mult = len / PAGE_SIZE;
remainder = len % PAGE_SIZE;
len = mult * PAGE_SIZE;
len += remainder ? PAGE_SIZE : 0;
is equivalent to:
len = PAGE_ALIGN(len);
Signed-off-by: wangdicheng <wangdicheng@kylinos.cn>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Link: https://patch.msgid.link/20260603091102.231370-4-wangdich9700@163.com
The function qc_usb_audio_offload_fill_avail_pcms() always returns -1
regardless of how many PCM devices were successfully filled. This makes
it impossible for callers to know the actual number of available PCMs.
Return the actual count of filled PCM devices instead, which allows
callers to verify that all expected PCMs were properly enumerated.
Signed-off-by: wangdicheng <wangdicheng@kylinos.cn>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Link: https://patch.msgid.link/20260603091102.231370-3-wangdich9700@163.com
The current code uses sprintf() to format control names without bounds
checking, which could lead to buffer overflow if PCM index is large.
Replace sprintf with snprintf to ensure buffer safety.
The ctl_name buffer is 48 bytes, and the formatted string could exceed
this with large PCM index values. Using snprintf with sizeof(ctl_name)
prevents potential buffer overflow.
Signed-off-by: wangdicheng <wangdicheng@kylinos.cn>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Link: https://patch.msgid.link/20260603091102.231370-2-wangdich9700@163.com
Add error codes to error messages for better debugging.
This helps identify the root cause when USB audio offload fails.
Error messages now include the actual error code returned by
xhci_sideband operations, making it easier to diagnose failures
during USB audio offload setup.
Signed-off-by: wangdicheng <wangdicheng@kylinos.cn>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Link: https://patch.msgid.link/20260603091102.231370-1-wangdich9700@163.com
Clevo P775TM1 laptops come with an ESS Sabre HiFi DAC. Setting
0x1b pin VREF to 80% enables said DAC output.
Signed-off-by: Evelyn Ali <evelynali99@gmail.com>
Link: https://patch.msgid.link/20260602214122.78020-1-evelynali99@gmail.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
John Hubbard <jhubbard@nvidia.com> says:
This began as a one-line dma-buf fix for a path_noexec() warning added
by commit 1e7ab6f678 ("anon_inode: rework assertions"). Christoph
pointed out that the fix belongs higher up: a pseudo filesystem has no
reason not to set SB_I_NOEXEC by default. This series does that.
* Patch 1 sets both flags in init_pseudo(), so every pseudo
filesystem gets them. This is the only patch that changes a flag,
and the only one with Fixes:/Cc: stable.
* Patch 2 drops the assignments that are now redundant in the callers
that set them by hand.
Most callers already set one or both flags. I audited every
init_pseudo() caller. Here is what patch 1 actually changes for each.
The only visible effect is on dma-buf, where SB_I_NOEXEC silences the
warning. SB_I_NODEV is never consulted on these SB_NOUSER mounts, and
none of the callers that gain SB_I_NOEXEC are executed from.
caller had patch 1 adds
--------------------------- -------- --------------
fs/anon_inodes.c both nothing new
mm/secretmem.c both nothing new
virt/kvm/guest_memfd.c both nothing new
fs/nsfs.c both nothing new
fs/pidfs.c both nothing new
fs/aio.c NOEXEC NODEV
drivers/dma-buf/dma-buf.c neither NOEXEC + NODEV
net/socket.c neither NOEXEC + NODEV
fs/pipe.c neither NOEXEC + NODEV
kernel/resource.c neither NOEXEC + NODEV
fs/erofs/super.c neither NOEXEC + NODEV
fs/btrfs/tests/... neither NOEXEC + NODEV
drivers/vfio/vfio_main.c neither NOEXEC + NODEV
drivers/gpu/drm/drm_drv.c neither NOEXEC + NODEV
drivers/dax/super.c neither NOEXEC + NODEV
block/bdev.c neither NOEXEC + NODEV
* patches from https://patch.msgid.link/20260604025315.245910-1-jhubbard@nvidia.com:
libfs: drop redundant SB_I_NOEXEC/SB_I_NODEV in init_pseudo() callers
libfs: set SB_I_NOEXEC and SB_I_NODEV by default in init_pseudo()
Link: https://patch.msgid.link/20260604025315.245910-1-jhubbard@nvidia.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
init_pseudo() now sets SB_I_NOEXEC and SB_I_NODEV by default, so the
per-caller assignments are redundant. Drop them.
Signed-off-by: John Hubbard <jhubbard@nvidia.com>
Link: https://patch.msgid.link/20260604025315.245910-3-jhubbard@nvidia.com
Signed-off-by: Christian Brauner (Amutable) <brauner@kernel.org>
Since commit 1e7ab6f678 ("anon_inode: rework assertions"),
path_noexec() warns when an anonymous-inode file is mmap'd from a
superblock that has not set SB_I_NOEXEC. dma-buf backs its files this
way and never set the flag, so mmap of any exported buffer trips the
warning on a CONFIG_DEBUG_VFS=y kernel:
WARNING: CPU: 11 PID: 121813 at fs/exec.c:118 path_noexec+0x47/0x50
do_mmap+0x2b5/0x680
vm_mmap_pgoff+0x129/0x210
ksys_mmap_pgoff+0x177/0x240
__x64_sys_mmap+0x33/0x70
init_pseudo() sets up internal SB_NOUSER mounts that are never
path-reachable. Set both flags here so every pseudo filesystem gets
them by default instead of each caller setting them.
SB_I_NODEV is inert for unreachable mounts. SB_I_NOEXEC has one
visible effect: an executable mapping of a pseudo-fs fd, such as a
dma-buf, now fails with -EPERM, which is the invariant the assertion
enforces. No in-tree caller maps these executable.
Reproduce on CONFIG_DEBUG_VFS=y:
make -C tools/testing/selftests/dmabuf-heaps
sudo ./tools/testing/selftests/dmabuf-heaps/dmabuf-heap -t system
Fixes: 1e7ab6f678 ("anon_inode: rework assertions")
Suggested-by: Christoph Hellwig <hch@infradead.org>
Cc: stable@vger.kernel.org
Signed-off-by: John Hubbard <jhubbard@nvidia.com>
Link: https://patch.msgid.link/20260604025315.245910-2-jhubbard@nvidia.com
Signed-off-by: Christian Brauner (Amutable) <brauner@kernel.org>
When a buffered read fails, iomap_finish_folio_read() reports the error
with fserror_report_io(folio->mapping->host, ...). This is called after
ifs->read_bytes_pending has been decremented by the bytes attempted to
be read.
For a folio split across multiple read completions, the folio is only
guaranteed to stay locked while read_bytes_pending > 0. Once
iomap_finish_folio_read() decrements read_bytes_pending, another
in-flight read can complete and end the read on the folio, which unlocks
it. This allows truncate logic to run and detach the folio (set
folio->mapping to NULL). The error reporting path then can dereference a
NULL folio->mapping. As reported by Sam Sun, this is the race that can
occur:
CPU0: failed completion CPU1: final completion CPU2: truncate
----------------------- ---------------------- --------------
read_bytes_pending -= len
finished = false
/* preempted before
fserror_report_io() */
read_bytes_pending -= len
finished = true
folio_end_read()
truncate clears
folio->mapping
fserror_report_io(
folio->mapping->host, ...)
^ NULL deref
Fix this by reporting the error first before decrementing
ifs->read_bytes_pending.
Fixes: a9d573ee88 ("iomap: report file I/O errors to the VFS")
Cc: stable@vger.kernel.org
Reported-by: Sam Sun <samsun1006219@gmail.com>
Closes: https://lore.kernel.org/linux-fsdevel/CAEkJfYPhWdd59RKmuNLJg-bkypHz7xiOwaWyNVu3A8CUqQCnvg@mail.gmail.com/
Signed-off-by: Joanne Koong <joannelkoong@gmail.com>
Link: https://patch.msgid.link/20260604011858.2297561-1-joannelkoong@gmail.com
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>
Signed-off-by: Christian Brauner (Amutable) <brauner@kernel.org>
may_decode_fh() accesses mount::mnt_ns without holding any locks; that
means the mount can concurrently be unmounted, and the mnt_namespace can
concurrently be freed after an RCU grace period.
This race can happens as follows, assuming that the mount point was
created by open_tree(..., OPEN_TREE_CLONE):
thread 1 thread 2 RCU
__do_sys_open_by_handle_at
do_handle_open
handle_to_path
may_decode_fh
is_mounted
[mount::mnt_ns access]
[mount::mnt_ns access]
__do_sys_close
fput_close_sync
__fput
dissolve_on_fput
umount_tree
class_namespace_excl_destructor
namespace_unlock
free_mnt_ns
mnt_ns_tree_remove
call_rcu(mnt_ns_release_rcu)
mnt_ns_release_rcu
mnt_ns_release
kfree
[mnt_namespace::user_ns access] **UAF**
Fix it by taking rcu_read_lock() around the mount::mnt_ns access, like
in __prepend_path().
Additionally, document the semantics of mount::mnt_ns, and use WRITE_ONCE()
for writers that can race with lockless readers.
This bug is unreachable unless one of the following is set:
- CONFIG_PREEMPTION
- CONFIG_RCU_STRICT_GRACE_PERIOD
because it requires an RCU grace period to happen during a syscall without
an explicit preemption.
This doesn't seem to have interesting security impact; worst-case, it could
leak the result of an integer comparison to userspace (from the level
check in cap_capable()), cause an endless loop, or crash the kernel by
dereferencing an invalid address.
Fixes: 620c266f39 ("fhandle: relax open_by_handle_at() permission checks")
Cc: stable@vger.kernel.org
Signed-off-by: Jann Horn <jannh@google.com>
Link: https://patch.msgid.link/20260603-vfs-fhandle-uaf-fix-v2-1-d05db76a5084@google.com
Signed-off-by: Christian Brauner (Amutable) <brauner@kernel.org>
Starting with Rust 1.98.0 (expected 2026-08-20), the
`-Zdebug-info-for-profiling` flag has been renamed to
`-Zdebuginfo-for-profiling` (i.e. one less dash, to match `debuginfo`s
in other flags) [1].
Without this change, one gets in the latest nightlies:
error: unknown unstable option: `debug-info-for-profiling`
Thus pass the right name.
Link: https://github.com/rust-lang/rust/pull/156887 [1]
Reviewed-by: Alice Ryhl <aliceryhl@google.com>
Acked-by: Nathan Chancellor <nathan@kernel.org>
Link: https://patch.msgid.link/20260602151638.14358-1-ojeda@kernel.org
Signed-off-by: Miguel Ojeda <ojeda@kernel.org>
The rt1711h Type-C controller on the HiKey960 has the USB role-switch
endpoint placed as a top-level 'port' node, outside the connector
subnode. This triggers two dtbs_check warnings against
richtek,rt1711h.yaml:
- 'port' does not match any of the regexes: '^pinctrl-[0-9]+$'
- connector:ports: 'port@0' is a required property
Move the role-switch endpoint into the connector's port@0, which is
where usb-connector.yaml expects it. Update the DWC3 remote-endpoint
phandle accordingly.
The TCPM core (tcpm.c) looks up the role switch starting from the
connector fwnode via fwnode_usb_role_switch_get(). With the endpoint
inside the connector's port@0, it is found through the primary lookup
path rather than the device-level fallback.
Cross-compiled for arm64. Verified with dt_binding_check and
dtbs_check. Not runtime-tested on hardware.
Signed-off-by: Akash Sukhavasi <akash.sukhavasi@gmail.com>
Signed-off-by: Wei Xu <xuwei5@hisilicon.com>
The info->node RB-tree member is zero-initialized via kzalloc. If
a device does not support ATS, the device_rbtree_insert() call is
skipped. If a subsequent probe step fails, the error path jumps to
device_rbtree_remove(), which misinterprets the zeroed node as
a tree root and corrupts the device RB-tree.
Fix this by explicitly initializing the RB-node as empty using
RB_CLEAR_NODE() during initialization and guarding the removal with
RB_EMPTY_NODE().
Fixes: 4f1492efb4 ("iommu/vt-d: Revert ATS timing change to fix boot failure")
Reported-by: sashiko-bot@kernel.org
Closes: https://lore.kernel.org/all/20260525205628.CD4431F000E9@smtp.kernel.org/
Suggested-by: Baolu Lu <baolu.lu@linux.intel.com>
Signed-off-by: Pranjal Shrivastava <praan@google.com>
Link: https://lore.kernel.org/r/20260531170254.60493-2-praan@google.com
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
In some environments, multiple PCIe segments exist, and PCIe device
information needs to be differentiated and identified based on the
segment. When an IOMMU fault event occurs, the IOMMU and device segment
information should be output in detail in dmar_fault_do_one.
Signed-off-by: Guanghui Feng <guanghuifeng@linux.alibaba.com>
Link: https://lore.kernel.org/r/20260528022943.1697564-1-guanghuifeng@linux.alibaba.com
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Rename pasid_pte_config_nestd() into pasid_pte_config_nested(). Do it to
match other function names ending with _nested().
Signed-off-by: Michał Grzelak <michal.grzelak@intel.com>
Link: https://lore.kernel.org/r/20260509174503.831134-1-michal.grzelak@intel.com
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
device_pasid_table_teardown() zeroes the 128-bit scalable-mode context
entry with context_clear_entry() while the Present bit is still set. This
creates a window where the hardware can fetch a torn entry, with some
fields already zeroed while Present is still set, leading to unpredictable
behavior or spurious faults. The context-cache invalidation is issued only
after the entry has been zeroed, and intel_pasid_free_table() then frees
the PASID directory pages, so the IOMMU can keep walking a stale Present=1
entry that points at freed memory.
While x86 provides strong write ordering, the compiler may reorder the two
64-bit writes to the entry, and the hardware fetch is not guaranteed to be
atomic with respect to multiple CPU writes.
Commit c1e4f1dccb ("iommu/vt-d: Clear Present bit before tearing down
context entry") fixed this exact pattern in domain_context_clear_one() and
the copied-context path, but device_pasid_table_teardown() was not
converted.
Align it with the "Guidance to Software for Invalidations" in the VT-d
spec, Section 6.5.3.3, using the same ownership handshake as the sibling
fix: clear only the Present bit, flush it to the IOMMU, perform the
context-cache invalidation, and only then zero the rest of the entry.
Fixes: 81e921fd32 ("iommu/vt-d: Fix NULL domain on device release")
Signed-off-by: Michael Bommarito <michael.bommarito@gmail.com>
Assisted-by: Claude:claude-opus-4-7
Link: https://lore.kernel.org/r/20260528025557.3209367-1-michael.bommarito@gmail.com
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
The Intel IOMMU driver allows SVA on devices even if they do not support
PCI/PRI. Commit 39c20c4e83 ("iommu/vt-d: Only handle IOPF for SVA when
PRI is supported") modified the SVA bind path to allow this configuration
by skipping IOPF enablement when PRI is missing. However, it failed to
update the unbind path.
This creates an imbalance: the unbind path attempts to disable IOPF for
a device that never had it enabled, triggering a WARNING in
intel_iommu_disable_iopf():
WARNING: drivers/iommu/intel/iommu.c:3475 at intel_iommu_disable_iopf+0x4f/0x90d
Call Trace:
<TASK>
blocking_domain_set_dev_pasid+0x50/0x70
iommu_detach_device_pasid+0x89/0xc0
iommu_sva_unbind_device+0x73/0x150
xe_vm_close_and_put+0x4d2/0x1200 [xe]
Fix this by bypassing IOPF operations for SVA domains on non-PRI hardware
in both the bind and unbind paths.
Fixes: 39c20c4e83 ("iommu/vt-d: Only handle IOPF for SVA when PRI is supported")
Cc: stable@vger.kernel.org
Reported-by: Nareshkumar Gollakoti <naresh.kumar.g@intel.com>
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Link: https://lore.kernel.org/r/20260519052917.3729796-1-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Add compatible and chip data to support GPCDMA in Tegra264, which has
differences in register layout and address bits compared to previous
versions.
Signed-off-by: Akhil R <akhilrajeev@nvidia.com>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Acked-by: Thierry Reding <treding@nvidia.com>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
Link: https://patch.msgid.link/20260331102303.33181-10-akhilrajeev@nvidia.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Use 'iommu-map', when provided, to get the stream ID to be programmed
for each channel. Iterate over the channels registered and configure
each channel device separately using of_dma_configure_id() to allow
it to use a separate IOMMU domain for the transfer. However, do this
in a second loop since the first loop populates the DMA device channels
list and async_device_register() registers the channels. Both are
prerequisites for using the channel device in the next loop.
Channels will continue to use the same global stream ID if the
'iommu-map' property is not present in the device tree.
Signed-off-by: Akhil R <akhilrajeev@nvidia.com>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
Link: https://patch.msgid.link/20260331102303.33181-9-akhilrajeev@nvidia.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Switch to managed registration in probe. This simplifies the error
paths in the probe and also removes the requirement of the driver
remove function.
Signed-off-by: Akhil R <akhilrajeev@nvidia.com>
Suggested-by: Frank Li <frank.li@nxp.com>
Acked-by: Thierry Reding <treding@nvidia.com>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
Link: https://patch.msgid.link/20260331102303.33181-8-akhilrajeev@nvidia.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Tegra264 supports address width of 41 bits. Unlike older SoCs which use
a common high_addr register for upper address bits, Tegra264 has separate
src_high and dst_high registers to accommodate this wider address space.
Add an addr_bits property to the device data structure to specify the
number of address bits supported on each device and use that to program
the appropriate registers.
Update the sg_req struct to remove the high_addr field and use
dma_addr_t for src and dst to store the complete addresses. Extract
the high address bits only when programming the registers.
Signed-off-by: Akhil R <akhilrajeev@nvidia.com>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
Link: https://patch.msgid.link/20260331102303.33181-7-akhilrajeev@nvidia.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Repurpose the struct tegra_dma_channel_regs to define offsets for all the
channel registers. Previously, the struct only held the register values
for each transfer and was wrapped within tegra_dma_sg_req. Move the
values directly into tegra_dma_sg_req and use channel_regs for
storing the register offsets. Update all register reads/writes to use
the struct channel_regs. This prepares for the register offset change
in Tegra264.
Signed-off-by: Akhil R <akhilrajeev@nvidia.com>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Acked-by: Thierry Reding <treding@nvidia.com>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
Link: https://patch.msgid.link/20260331102303.33181-6-akhilrajeev@nvidia.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
On Tegra264, reset is not available for the driver to control as
this is handled by the boot firmware. Hence make the reset control
optional and update the error message to reflect the correct error.
Signed-off-by: Akhil R <akhilrajeev@nvidia.com>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Acked-by: Thierry Reding <treding@nvidia.com>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
Link: https://patch.msgid.link/20260331102303.33181-5-akhilrajeev@nvidia.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Add iommu-map property to specify separate stream IDs for each DMA
channel. This enables each channel to be in its own IOMMU domain,
keeping memory isolated from other devices sharing the same DMA
controller.
Define the constraints such that if the channel and stream IDs are
contiguous, a single entry can map all the channels, but if the
channels or stream IDs are non-contiguous support multiple entries.
Signed-off-by: Akhil R <akhilrajeev@nvidia.com>
Acked-by: Rob Herring (Arm) <robh@kernel.org>
Acked-by: Thierry Reding <treding@nvidia.com>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
Link: https://patch.msgid.link/20260331102303.33181-4-akhilrajeev@nvidia.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
On Tegra264, GPCDMA reset control is not exposed to Linux and is handled
by the boot firmware.
Although reset was not exposed in Tegra234 as well, the firmware supported
a dummy reset which just returns success on reset without doing an actual
reset. This is also not supported in Tegra264 BPMP. Therefore mark 'reset'
and 'reset-names' properties as required only for devices prior to
Tegra264.
This also necessitates that the Tegra264 compatible be standalone and
cannot have the fallback compatible of Tegra186. Since there is no
functional impact, we keep reset as required for Tegra234 to avoid
breaking the ABI.
Fixes: bb8c97571d ("dt-bindings: dma: Add Tegra264 compatible string")
Signed-off-by: Akhil R <akhilrajeev@nvidia.com>
Acked-by: Rob Herring (Arm) <robh@kernel.org>
Acked-by: Thierry Reding <treding@nvidia.com>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
Link: https://patch.msgid.link/20260331102303.33181-2-akhilrajeev@nvidia.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
The KVM_REG_RISCV_TIMER_REG(state) one-reg write passes the value
written by userspace to kvm_riscv_vcpu_timer_next_event() when
re-enabling the timer.
That value is the timer state, KVM_RISCV_TIMER_STATE_ON, not the
timer compare value. During migration or state restore, userspace
restores the compare register separately, which stores the target
cycle in t->next_cycles. Re-arming the timer with the state value
schedules the next event at cycle 1 instead of the restored compare
value, causing the virtual timer to fire too early.
Use the restored compare value from t->next_cycles when turning the
timer back on.
Fixes: 3a9f66cb25 ("RISC-V: KVM: Add timer functionality")
Signed-off-by: Qiang Ma <maqianga@uniontech.com>
Reviewed-by: Anup Patel <anup@brainfault.org>
Link: https://lore.kernel.org/r/20260526075544.796396-1-maqianga@uniontech.com
Signed-off-by: Anup Patel <anup@brainfault.org>
There are multi spba-busses for i.MX8M* platforms, if only search for
the first spba-bus in DT, the found spba-bus may not the real bus of
audio devices, which cause issue for sdma p2p case, as the sdma p2p
script presently does not deal with the transactions involving two devices
connected to the AIPS bus.
Search the SDMA parent node first, which should be the AIPS bus, then
search the child node whose compatible string is spba-bus under that AIPS
bus for the above multi spba-busses case.
Fixes: 8391ecf465 ("dmaengine: imx-sdma: Add device to device support")
Signed-off-by: Shengjiu Wang <shengjiu.wang@nxp.com>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Link: https://patch.msgid.link/20260407032755.2758049-1-shengjiu.wang@nxp.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Fuzzer reported a NULL pointer dereference in
kvm_riscv_vcpu_aia_imsic_put() when a VCPU's imsic_state was NULL while
kvm_riscv_aia_initialized() returned true.
The global initialized flag is set per-VM in aia_init(), but imsic_state
is allocated per-VCPU in kvm_riscv_vcpu_aia_imsic_init(). If a VCPU is
created after aia_init() has already run, its imsic_state remains NULL
while the global flag is true. When this VCPU is preempted, kvm_sched_out()
calls kvm_arch_vcpu_put() -> kvm_riscv_vcpu_aia_put() ->
kvm_riscv_vcpu_aia_imsic_put() which dereferences NULL.
Add NULL pointer guards to kvm_riscv_vcpu_aia_imsic_put(), consistent with
the NULL checks already present in all other functions in the same file.
Also add a NULL guard to kvm_riscv_vcpu_aia_imsic_release() and
kvm_riscv_vcpu_aia_imsic_has_interrupt() for the same reason.
Fixes: 4cec89db80 ("RISC-V: KVM: Move HGEI[E|P] CSR access to IMSIC virtualization")
Signed-off-by: Jiakai Xu <jiakaiPeanut@gmail.com>
Signed-off-by: Jiakai Xu <xujiakai2025@iscas.ac.cn>
Assisted-by: YuanSheng:DeepSeek-V3.2
Reviewed-by: Anup Patel <anup@brainfault.org>
Link: https://lore.kernel.org/r/20260526031517.1166025-1-xujiakai2025@iscas.ac.cn
Signed-off-by: Anup Patel <anup@brainfault.org>
The SUSP handler checks that all other vCPUs are stopped before
entering system suspend, but a concurrent HSM HART_START can start
a vCPU after it has already passed the check.
This is a known TOCTOU race. We do not fix it because:
1. Triggering it requires a pathological guest.
2. Only guest state is at risk, not host integrity.
3. Userspace can double-check vCPU states before suspend.
Add a comment documenting the race and the rationale for not fixing it.
Signed-off-by: Jiakai Xu <jiakaiPeanut@gmail.com>
Signed-off-by: Jiakai Xu <xujiakai2025@iscas.ac.cn>
Assisted-by: YuanSheng:DeepSeek-V3.2
Reviewed-by: Andrew Jones <andrew.jones@oss.qualcomm.com>
Link: https://lore.kernel.org/r/20260525013642.999187-1-xujiakai2025@iscas.ac.cn
Signed-off-by: Anup Patel <anup@brainfault.org>
random_recv_done() stores the device-reported used.len directly into
vi->data_avail. copy_data() then indexes vi->data[] using
vi->data_idx (advanced by previous copy_data() calls) and issues a
memcpy() without re-validating either value against the posted
buffer size sizeof(vi->data) (SMP_CACHE_BYTES bytes, typically 32
or 64).
A malicious or buggy virtio-rng backend can set used.len beyond
sizeof(vi->data), steering the memcpy() past the end of the inline
array into adjacent kmalloc-1k slab bytes. hwrng_fillfn() mixes
those bytes into the guest RNG, and guest root can also observe
them directly via /dev/hwrng.
Concrete impact is inside the guest:
- Memory-safety / hardening: any virtio-rng backend that
over-reports used.len causes the driver to read past vi->data
into unrelated slab contents. hwrng_fillfn() is a kernel thread
that runs as soon as the device is probed; no guest userspace
interaction is required to first-trigger the OOB.
- Cross-boundary leak (confidential-compute threat model): a
malicious hypervisor cooperating with a malicious or compromised
guest root userspace can use /dev/hwrng as a leak channel for
guest-kernel heap data. The host sets a large used.len, guest
root reads /dev/hwrng, and the returned bytes contain guest
kernel slab contents that were adjacent to vi->data. In
practice, confidential-compute guests (SEV-SNP, TDX) usually
disable virtio-rng entirely, so this path is narrow, but the
fix is still worth carrying because the underlying
memory-safety bug contaminates the guest RNG on any host.
KASAN confirms the OOB on a 7.1-rc4 guest whose virtio-rng backend
has been patched to report used.len = 0x10000:
BUG: KASAN: slab-out-of-bounds in virtio_read+0x394/0x5d0
Read of size 64 at addr ffff88800ae0ba20 by task hwrng/52
Call Trace:
__asan_memcpy+0x23/0x60
virtio_read+0x394/0x5d0
hwrng_fillfn+0xb2/0x470
kthread+0x2cc/0x3a0
Allocated by task 1:
probe_common+0xa5/0x660
virtio_dev_probe+0x549/0xbc0
The buggy address belongs to the object at ffff88800ae0b800
which belongs to the cache kmalloc-1k of size 1024
The buggy address is located 0 bytes to the right of
allocated 544-byte region [ffff88800ae0b800, ffff88800ae0ba20)
Same class of bug as commit c04db81cd0 ("net/9p: Fix buffer
overflow in USB transport layer"), which hardened
usb9pfs_rx_complete() against unchecked device-reported length in
the USB 9p transport.
With the clamp at point of use and array_index_nospec() in place,
the same harness boots cleanly: copy_data() returns zero for the
bogus report, the device-supplied bytes after data_idx are
discarded, and the driver issues a fresh request.
Fixes: f7f510ec19 ("virtio: An entropy device, as suggested by hpa.")
Cc: stable@vger.kernel.org
Suggested-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael Bommarito <michael.bommarito@gmail.com>
Assisted-by: Claude:claude-opus-4-8
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20260531142251.2792061-1-michael.bommarito@gmail.com>
Unbinding a virtio balloon device:
echo virtio0 > /sys/bus/virtio/drivers/virtio_balloon/unbind
triggers a NULL pointer dereference. The dmesg says:
BUG: kernel NULL pointer dereference, address: 0000000000000008
[...]
RIP: 0010:__list_del_entry_valid_or_report+0x5/0xf0
Call Trace:
<TASK>
vp_del_vqs+0x121/0x230
remove_common+0x135/0x150
virtballoon_remove+0xee/0x100
virtio_dev_remove+0x3b/0x80
device_release_driver_internal+0x187/0x2c0
unbind_store+0xb9/0xe0
kernfs_fop_write_iter.llvm.11660790530567441834+0xf6/0x180
vfs_write+0x2a9/0x3b0
ksys_write+0x5c/0xd0
do_syscall_64+0x54/0x230
entry_SYSCALL_64_after_hwframe+0x29/0x31
[...]
</TASK>
The virtio_balloon device registers 5 queues (inflate, deflate, stats,
free_page, reporting) but only the first two are unconditional. The
stats, free_page and reporting queues are each conditional on their
respective feature bits. When any of these features are absent, the
corresponding vqs_info entry has name == NULL, creating holes in the
array.
The root cause is an indexing mismatch introduced when vq info storage
was changed to be passed as an argument. vp_find_vqs_msix() and
vp_find_vqs_intx() store the info pointer at vp_dev->vqs[i], where 'i'
is the caller's sparse array index. However, the virtqueue itself gets
vq->index assigned from queue_idx, a dense index that skips NULL
entries. When holes exist, 'i' and queue_idx diverge. Later,
vp_del_vqs() looks up info via vp_dev->vqs[vq->index] using the dense
index into the sparsely-populated array, and hits NULL.
Fix this by storing info at vp_dev->vqs[queue_idx] instead of
vp_dev->vqs[i], so the store index matches the lookup index
(vq->index). Apply the fix to both the MSIX and INTX paths.
Cc: Yichun Zhang <yichun@openresty.com>
Cc: Jiri Pirko <jiri@nvidia.com>
Cc: stable@vger.kernel.org # v6.11+
Tested-by: Yuka <yuka@umeyashiki.org>
Fixes: 89a1c435ae ("virtio_pci: pass vq info as an argument to vp_setup_vq()")
Signed-off-by: Ammar Faizi <ammarfaizi2@openresty.com>
Message-Id: <20260315141808.547081-1-ammarfaizi2@openresty.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
When USB4 lane margining debugfs write support is enabled,
margining_error_counter_write() copies the user input with
validate_and_copy_from_user(). This allocates a temporary page that is
only needed while parsing the requested error counter mode.
The function currently returns without freeing that page. This leaks one
page per write to the error_counter debugfs file, including successful
writes and writes that later fail while taking the domain lock or because
software margining is not enabled.
Free the temporary page once parsing has completed, and also before
returning from the invalid-input path.
Fixes: 10904df3f2 ("thunderbolt: Improve software receiver lane margining")
Signed-off-by: Xu Rao <raoxu@uniontech.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
vhost_get_avail_idx is supposed to report whether it has updated
vq->avail_idx. Instead, it returns whether all entries have been
consumed, which is usually the same. But not always - in
drivers/vhost/net.c and when mergeable buffers have been enabled, the
driver checks whether the combined entries are big enough to store an
incoming packet. If not, the driver re-enables notifications with
available entries still in the ring. The incorrect return value from
vhost_get_avail_idx propagates through vhost_enable_notify and causes
the host to livelock if the guest is not making progress, as vhost will
immediately disable notifications and retry using the available entries.
This goes back to commit d3bb267bbd ("vhost: cache avail index in
vhost_enable_notify()") which changed vhost_enable_notify() to compare
the freshly read avail index against vq->last_avail_idx instead of the
previously cached vq->avail_idx. Commit 7ad4723976 ("vhost: move
smp_rmb() into vhost_get_avail_idx()") then carried over the same
comparison when refactoring vhost_enable_notify() to call the unified
vhost_get_avail_idx().
The obvious fix is to make vhost_get_avail_idx do what the comment
says it does and report whether new entries have been added.
Reported-by: ShuangYu <shuangyu@yunyoo.cc>
Fixes: d3bb267bbd ("vhost: cache avail index in vhost_enable_notify()")
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Message-Id: <559b04ae6ce52973c535dc47e461638b7f4c3d63.1772441455.git.mst@redhat.com>
Coccinelle (scripts/coccinelle/misc/noderef.cocci) reports:
drivers/clk/keystone/sci-clk.c:391:8-14: ERROR: application of
sizeof to pointer
In sci_clk_get(), 'clk' is declared as 'struct sci_clk **', so
sizeof(clk) is sizeof(struct sci_clk **) which is the size of a
pointer rather than the size of an array element. provider->clocks
is an array of 'struct sci_clk *', so the canonical size argument
to bsearch() is sizeof(*clk) (i.e. sizeof(struct sci_clk *)).
The two values are equal on every supported architecture, so this
is correctness/idiom, not a runtime fix, but the new form matches
the rest of the bsearch() callers in the tree and silences the
Coccinelle warning the script flagged.
Reported-by: Zeal Robot <zealci@zte.com.cn>
Closes: https://lore.kernel.org/all/84a6ba16686347099a3dab2e5161a930e792eb6e.1629198281.git.jing.yangyang@zte.com.cn/
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Julia Lawall <julia.lawall@inria.fr>
Closes: https://lore.kernel.org/all/202512040525.zrHSDl5h-lkp@intel.com/
Link: https://lore.kernel.org/linux-clk/20211012021931.176727-1-davidcomponentone@gmail.com/
Reviewed-by: Stepan Ionichev <sozdayvek@gmail.com>
Reviewed-by: Andrew Davis <afd@ti.com>
Signed-off-by: Jing Yangyang <jing.yangyang@zte.com.cn>
Signed-off-by: David Yang <davidcomponentone@gmail.com>
[nm@ti.com: Improved commit message]
Reviewed-by: Brian Masney <bmasney@redhat.com>
Link: https://patch.msgid.link/20260512110028.2999471-1-nm@ti.com
Signed-off-by: Nishanth Menon <nm@ti.com>
The TISCI firmware will return 0 if the clock or consumer is not
enabled although there is a stored value in the firmware. IOW a call to
set rate will work but at get rate will always return 0 if the clock is
disabled.
The clk framework will try to cache the clock rate when it's requested
by a consumer. If the clock or consumer is not enabled at that point,
the cached value is 0, which is wrong. Thus, disable the cache
altogether.
Signed-off-by: Michael Walle <mwalle@kernel.org>
Reviewed-by: Kevin Hilman <khilman@baylibre.com>
Reviewed-by: Randolph Sapp <rs@ti.com>
Reviewed-by: Nishanth Menon <nm@ti.com>
Signed-off-by: Antonios Christidis <a-christidis@ti.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Link: https://patch.msgid.link/20260507-clk-sci-v2-1-38f59b48777a@ti.com
Signed-off-by: Nishanth Menon <nm@ti.com>
The dmaengine path shares ethtool_ops with the legacy AXI DMA path,
including .get_coalesce/.set_coalesce that poke XAXIDMA_*_CR_OFFSET
directly. In dmaengine mode lp->dma_regs is not mapped by axienet, so
those ethtool calls touch unmapped/unrelated memory and report values
unrelated to the channel actually in use.
.get_ringparam/.set_ringparam only touch lp->rx_bd_num/lp->tx_bd_num,
fields used only by the legacy path for BD ring sizing. In dmaengine
mode the descriptor ring is owned by the dmaengine provider and these
fields are not consulted, so reporting them is misleading.
No dmaengine API exists today to query or program either coalescing
or ring size on behalf of the client, so neither can be exposed
meaningfully in dmaengine mode.
Add axienet_ethtool_dmaengine_ops without the coalesce and ringparam
hooks. Also move the ethtool_ops assignment from early probe into the
if/else alongside netdev_ops, so the legacy and dmaengine paths pick
their respective ops in one place. No functional change for the
legacy DMA path.
Signed-off-by: Suraj Gupta <suraj.gupta2@amd.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://patch.msgid.link/20260601124454.3384601-1-suraj.gupta2@amd.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add 'make kconfig-sym-check', a static checker that finds Kconfig
symbols referenced in expressions (select, depends on, default, etc.)
but never defined via config/menuconfig anywhere in the tree. New
dangling symbols are reported as errors (exit 1) unless they are
listed in an exclusion file, e.g.
KCONFIG_SYM_CHECK_EXCLUDES=sym-check-excludes make kconfig-sym-check
The exclusion file lists one symbol per line; blank lines and lines
starting with '#' are ignored.
The checker also warns about uppercase N/Y/M used as tristate literal
values following the same logic as checkpatch.
This new static checker is the script used for [1] with a few
improvements to avoid some false positives.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=216748 [1]
Assisted-by: Claude:claude-sonnet-4-6
Signed-off-by: Andrew Jones <andrew.jones@linux.dev>
Acked-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Tested-by: Randy Dunlap <rdunlap@infradead.org>
Tested-by: Julian Braha <julianbraha@gmail.com>
Tested-by: Nicolas Schier <nsc@kernel.org>
Acked-by: Nicolas Schier <nsc@kernel.org>
Link: https://patch.msgid.link/20260527142703.107110-1-andrew.jones@linux.dev
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
There are some typos in comments and while they are harmless and not visible,
there is no reason not to fix them. Fix the ones that are not register related,
which might have intentional naming convention.
Signed-off-by: Jakub Raczynski <j.raczynski@samsung.com>
Link: https://patch.msgid.link/20260601163727.554364-1-j.raczynski@samsung.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
rx-copybreak requires an extra slab allocation. Since bnxt uses
page pool frags and HDS by default, the rx-copybreak doesn't
buy us anything. The extra pressure on slab causes overload
on pre-sheaves kernels on modern AMD platforms.
In synthetic testing on net-next this patch shows little difference
but I think copybreak is "obvious waste" at this point.
Default rx-copybreak threshold to 0 / disabled.
The "copybreak" defines are really the size bounds for the Rx header
buffer. Rename them.
Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Link: https://patch.msgid.link/20260602003759.1545645-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Lorenzo Bianconi says:
====================
Fix use-after-free in metadata dst teardown in airoha_eth and mtk_eth_soc drivers
airoha_metadata_dst_free() and mtk_free_dev() call metadata_dst_free()
which frees the metadata_dst with kfree() immediately, bypassing the RCU
grace period.
Replace metadata_dst_free() with dst_release() which properly goes
through the refcount path and runs call_rcu_hurry() if refcount goes to
zero.
====================
Link: https://patch.msgid.link/20260602-airoha-mtk-metadata-uaf-fix-v1-0-3aaa99d83351@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
mtk_free_dev() calls metadata_dst_free() which frees the metadata_dst
with kfree() immediately, bypassing the RCU grace period.
In the RX path, skb_dst_set_noref() sets a non-refcounted pointer from
the skb to the metadata_dst. This function requires RCU read-side
protection and the dst must remain valid until all RCU readers complete.
Since metadata_dst_free() calls kfree() directly, a use-after-free can
occur if any skb still holds a noref pointer to the dst when the driver
tears it down.
Replace metadata_dst_free() with dst_release() which properly goes
through the refcount path: when the refcount drops to zero, it schedules
the actual free via call_rcu_hurry(), ensuring all RCU readers have
completed before the memory is freed.
Fixes: 2d7605a729 ("net: ethernet: mtk_eth_soc: enable hardware DSA untagging")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Link: https://patch.msgid.link/20260602-airoha-mtk-metadata-uaf-fix-v1-2-3aaa99d83351@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
airoha_metadata_dst_free() runs metadata_dst_free() which frees the
metadata_dst with kfree() immediately, bypassing the RCU grace period.
In the RX path, skb_dst_set_noref() sets a non-refcounted pointer from
the skb to the metadata_dst. This function requires RCU read-side
protection and the dst must remain valid until all RCU readers complete.
Since metadata_dst_free() calls kfree() directly, an use-after-free can
occur if any skb still holds a noref pointer to the dst when the driver
tears it down.
Replace metadata_dst_free() with dst_release() which properly goes
through the refcount path: when the refcount drops to zero, it schedules
the actual free via call_rcu_hurry(), ensuring all RCU readers have
completed before the memory is freed.
Fixes: af3cf757d5 ("net: airoha: Move DSA tag in DMA descriptor")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Link: https://patch.msgid.link/20260602-airoha-mtk-metadata-uaf-fix-v1-1-3aaa99d83351@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
In ip6_parse_tlv(), recompute the network header pointer once regardless
of the option processed (Hbh or Dest), as missing recomputation for
specific options has caused issues in the past.
Signed-off-by: Justin Iurman <justin.iurman@gmail.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20260602213033.12244-1-justin.iurman@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
