🚨 CVE-2026-80808
In the Linux kernel, the following vulnerability has been resolved:
ext4: stop retrying saturated xattr cache entries
ext4_xattr_block_set() retries when a cache entry selected for reuse
has a saturated reference count after taking the buffer lock. The retry
returns to the mbcache lookup without making that entry ineligible, so
it can select the same unusable entry indefinitely. A task spinning
there can hold the parent directory's i_rwsem and leave concurrent
rmdir callers blocked.
Normally a reusable entry has a reference count below
EXT4_XATTR_REFCOUNT_MAX because the count and MBE_REUSABLE_B are
updated under the same buffer lock. A corrupted filesystem can violate
that invariant. The syzbot reproducer reports allocator and xattr
corruption before triggering this retry loop.
Check the untrusted on-disk count before incrementing it, avoiding
overflow, and clear MBE_REUSABLE_B when it is already saturated. The
next lookup then skips the entry that was just proven unusable. This
mirrors the normal transition at EXT4_XATTR_REFCOUNT_MAX; the release
path marks the entry reusable again on the exact 1024-to-1023
transition.
Using the same QEMU harness and guest parameters, current unpatched
Linux hung in 6 of 8 420-second trials with the do_rmdir signature;
representative NMI backtraces caught the owner spinning in
ext4_xattr_block_set(). The patched kernel completed 28 of 28 trials
without a hung-task report; the final twelve trials exercised the
reviewed overflow-safe form of the change. syzbot's patch testing also
completed without reproducing the hang.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
ext4: stop retrying saturated xattr cache entries
ext4_xattr_block_set() retries when a cache entry selected for reuse
has a saturated reference count after taking the buffer lock. The retry
returns to the mbcache lookup without making that entry ineligible, so
it can select the same unusable entry indefinitely. A task spinning
there can hold the parent directory's i_rwsem and leave concurrent
rmdir callers blocked.
Normally a reusable entry has a reference count below
EXT4_XATTR_REFCOUNT_MAX because the count and MBE_REUSABLE_B are
updated under the same buffer lock. A corrupted filesystem can violate
that invariant. The syzbot reproducer reports allocator and xattr
corruption before triggering this retry loop.
Check the untrusted on-disk count before incrementing it, avoiding
overflow, and clear MBE_REUSABLE_B when it is already saturated. The
next lookup then skips the entry that was just proven unusable. This
mirrors the normal transition at EXT4_XATTR_REFCOUNT_MAX; the release
path marks the entry reusable again on the exact 1024-to-1023
transition.
Using the same QEMU harness and guest parameters, current unpatched
Linux hung in 6 of 8 420-second trials with the do_rmdir signature;
representative NMI backtraces caught the owner spinning in
ext4_xattr_block_set(). The patched kernel completed 28 of 28 trials
without a hung-task report; the final twelve trials exercised the
reviewed overflow-safe form of the change. syzbot's patch testing also
completed without reproducing the hang.
🎖@cveNotify
🚨 CVE-2026-80809
In the Linux kernel, the following vulnerability has been resolved:
ocfs2: fix missing metadata reservation for large xattrs
[BUG]
lsetxattr() panics the kernel when setting a large xattr value on a
fragmented filesystem where the file already has an external xattr
block.
[CAUSE]
ocfs2_calc_xattr_set_need() never reserves metadata blocks for a new
xattr value's extent tree when the file already has an external xattr
block. The not_found path leaves meta_add at zero, so meta_ac is NULL
when ocfs2_xattr_extend_allocation() runs.
A new value root has room for a single extent record. On a fragmented
filesystem, the allocator cannot satisfy the xattr value in one
contiguous run, so each non-contiguous run requires its own extent
record. When the value root's extent list is full and meta_ac is NULL,
ocfs2_add_clusters_in_btree() returns RESTART_META, and
ocfs2_xattr_extend_allocation() hits BUG_ON(why == RESTART_META).
[FIX]
The case where no xattr block exists yet already calls
ocfs2_extend_meta_needed(&def_xv.xv.xr_list) to reserve value tree
metadata. Add the same reservation to the case where an xattr block
already exists, making the two cases consistent.
Replace the BUG_ON with a -ENOSPC return so that if RESTART_META is
returned despite the reservation, the error propagates to userspace
instead of panicking the kernel.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
ocfs2: fix missing metadata reservation for large xattrs
[BUG]
lsetxattr() panics the kernel when setting a large xattr value on a
fragmented filesystem where the file already has an external xattr
block.
[CAUSE]
ocfs2_calc_xattr_set_need() never reserves metadata blocks for a new
xattr value's extent tree when the file already has an external xattr
block. The not_found path leaves meta_add at zero, so meta_ac is NULL
when ocfs2_xattr_extend_allocation() runs.
A new value root has room for a single extent record. On a fragmented
filesystem, the allocator cannot satisfy the xattr value in one
contiguous run, so each non-contiguous run requires its own extent
record. When the value root's extent list is full and meta_ac is NULL,
ocfs2_add_clusters_in_btree() returns RESTART_META, and
ocfs2_xattr_extend_allocation() hits BUG_ON(why == RESTART_META).
[FIX]
The case where no xattr block exists yet already calls
ocfs2_extend_meta_needed(&def_xv.xv.xr_list) to reserve value tree
metadata. Add the same reservation to the case where an xattr block
already exists, making the two cases consistent.
Replace the BUG_ON with a -ENOSPC return so that if RESTART_META is
returned despite the reservation, the error propagates to userspace
instead of panicking the kernel.
🎖@cveNotify
🚨 CVE-2026-80810
In the Linux kernel, the following vulnerability has been resolved:
io_uring/rsrc: fix folio size overflow in io_vec_fill_bvec()
io_vec_fill_bvec() computes the folio size with a plain int 1:
unsigned long folio_size = 1 << imu->folio_shift;
imu->folio_shift is unsigned int and comes from folio_shift() of the
folio backing the registered buffer, so it can be 32 or more on a 64 bit
kernel. Shifting int 1 that far is undefined, and on x86 and arm64 the
count is taken modulo 32, so a shift of 34 yields 4 rather than 16G.
Every other folio_shift shift in this file already uses 1UL.
The result is that the segment estimate and the fill loop disagree.
io_estimate_bvec_size() sizes the bvec array with the real shift:
max_segs += (iov[i].iov_len >> shift) + 2;
so a 1M iovec on a 16G folio is charged 2 segments, while
io_vec_fill_bvec() then walks the same iovec in folio_size chunks of 4
bytes and writes res_bvec[bvec_idx] a quarter of a million times, past
the end of the array it was given. src_bvec is advanced once per
iteration as well, so imu->bvec is read past its end at the same time.
validate_fixed_range() only checks that the range is inside the
registered buffer and does not bound the segment count.
Reaching it needs a folio with a shift of at least 32, which means a
gigantic hugetlb page: 16G on arm64 with 64K pages, where
CONT_PMD_SHIFT is 34 and hugetlb_add_hstate(CONT_PMD_SHIFT - PAGE_SHIFT)
registers that size, and likewise on powerpc. x86_64 tops out at 1G, so
a shift of 30, which still fits in int and is unaffected.
Use 1UL, as the rest of the file does.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
io_uring/rsrc: fix folio size overflow in io_vec_fill_bvec()
io_vec_fill_bvec() computes the folio size with a plain int 1:
unsigned long folio_size = 1 << imu->folio_shift;
imu->folio_shift is unsigned int and comes from folio_shift() of the
folio backing the registered buffer, so it can be 32 or more on a 64 bit
kernel. Shifting int 1 that far is undefined, and on x86 and arm64 the
count is taken modulo 32, so a shift of 34 yields 4 rather than 16G.
Every other folio_shift shift in this file already uses 1UL.
The result is that the segment estimate and the fill loop disagree.
io_estimate_bvec_size() sizes the bvec array with the real shift:
max_segs += (iov[i].iov_len >> shift) + 2;
so a 1M iovec on a 16G folio is charged 2 segments, while
io_vec_fill_bvec() then walks the same iovec in folio_size chunks of 4
bytes and writes res_bvec[bvec_idx] a quarter of a million times, past
the end of the array it was given. src_bvec is advanced once per
iteration as well, so imu->bvec is read past its end at the same time.
validate_fixed_range() only checks that the range is inside the
registered buffer and does not bound the segment count.
Reaching it needs a folio with a shift of at least 32, which means a
gigantic hugetlb page: 16G on arm64 with 64K pages, where
CONT_PMD_SHIFT is 34 and hugetlb_add_hstate(CONT_PMD_SHIFT - PAGE_SHIFT)
registers that size, and likewise on powerpc. x86_64 tops out at 1G, so
a shift of 30, which still fits in int and is unaffected.
Use 1UL, as the rest of the file does.
🎖@cveNotify
🚨 CVE-2026-80811
In the Linux kernel, the following vulnerability has been resolved:
io_uring/cmd: fix iovec leak when the async cmd is not recycled
An io_async_cmd carries an iovec array in ->vec.iovec, allocated when the
vec has to grow and kept across recycling through ctx->cmd_cache. On two
paths nothing frees it and io_clean_op()'s kfree(req->async_data) drops
the io_async_cmd without it.
io_req_uring_cleanup() clears the async data flags only when
io_alloc_cache_put() succeeds, and the cache holds IO_ALLOC_CACHE_MAX ==
128 entries, so once it is full the put fails and the vec is left behind.
An NVMe passthrough workload gets there without doing anything unusual:
nvme_uring_cmd_io() returns -EIOCBQUEUED, so the io_async_cmd stays
attached for the lifetime of the command and the live object count tracks
the queue depth. Above 128 the puts start failing.
->cleanup is the last chance to free an inherited vec, since
io_req_uring_cleanup() returns early for an io-wq issued command and is
not called at all for one completed without ever being issued. But
io_clean_op() calls ->cleanup only if REQ_F_NEED_CLEANUP is set, and for
uring_cmd that happens only where the vec has to grow, so a command
reusing a large enough cached vec never sets it. io_rw_alloc_async() and
io_msg_alloc_async() flag an inherited vec for exactly this reason;
io_uring_cmd_prep() does not.
Flag an inherited vec in io_uring_cmd_prep(), and free the vec when the
cache put fails, as io_req_rw_cleanup() does.
The leak is invisible under KASAN, where io_alloc_cache_vec_kasan() frees
the vec unconditionally.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
io_uring/cmd: fix iovec leak when the async cmd is not recycled
An io_async_cmd carries an iovec array in ->vec.iovec, allocated when the
vec has to grow and kept across recycling through ctx->cmd_cache. On two
paths nothing frees it and io_clean_op()'s kfree(req->async_data) drops
the io_async_cmd without it.
io_req_uring_cleanup() clears the async data flags only when
io_alloc_cache_put() succeeds, and the cache holds IO_ALLOC_CACHE_MAX ==
128 entries, so once it is full the put fails and the vec is left behind.
An NVMe passthrough workload gets there without doing anything unusual:
nvme_uring_cmd_io() returns -EIOCBQUEUED, so the io_async_cmd stays
attached for the lifetime of the command and the live object count tracks
the queue depth. Above 128 the puts start failing.
->cleanup is the last chance to free an inherited vec, since
io_req_uring_cleanup() returns early for an io-wq issued command and is
not called at all for one completed without ever being issued. But
io_clean_op() calls ->cleanup only if REQ_F_NEED_CLEANUP is set, and for
uring_cmd that happens only where the vec has to grow, so a command
reusing a large enough cached vec never sets it. io_rw_alloc_async() and
io_msg_alloc_async() flag an inherited vec for exactly this reason;
io_uring_cmd_prep() does not.
Flag an inherited vec in io_uring_cmd_prep(), and free the vec when the
cache put fails, as io_req_rw_cleanup() does.
The leak is invisible under KASAN, where io_alloc_cache_vec_kasan() frees
the vec unconditionally.
🎖@cveNotify
🚨 CVE-2026-80812
In the Linux kernel, the following vulnerability has been resolved:
ALSA: dummy: Check card index validity at probe
snd_dummy_probe() blindly trusts that the given devptr->id value is
within the proper card index range. It's OK for the devices the
driver itself creates at the module probe time, but if the device is
bound manually via sysfs interface, this could be -1 as "none", and
this leads to OOB access for index[] and other parameters.
Add a sanity check for the card index and warn/correct it if it's a
value out of the range.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
ALSA: dummy: Check card index validity at probe
snd_dummy_probe() blindly trusts that the given devptr->id value is
within the proper card index range. It's OK for the devices the
driver itself creates at the module probe time, but if the device is
bound manually via sysfs interface, this could be -1 as "none", and
this leads to OOB access for index[] and other parameters.
Add a sanity check for the card index and warn/correct it if it's a
value out of the range.
🎖@cveNotify
🚨 CVE-2026-80813
In the Linux kernel, the following vulnerability has been resolved:
nvmet: fix NULL pointer dereference in nvmet_execute_identify_nslist()
When a host issues an Identify command with CNS 07h (Active Namespace ID
List for a specific I/O Command Set), nvmet_execute_identify_nslist() is
called with match_css set. The command-set filter dereferences req->ns,
but this handler never calls nvmet_req_find_ns(), so req->ns is always
NULL (nvmet_req_init() resets it to NULL). As soon as an enabled
namespace with an NSID greater than the requested value exists,
req->ns->csi dereferences a NULL pointer and oopses.
Besides the crash, the comparison is logically wrong: to filter the list
by command set it must test the command set of the namespace being
iterated, not a single fixed value. Use the loop variable ns->csi.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
nvmet: fix NULL pointer dereference in nvmet_execute_identify_nslist()
When a host issues an Identify command with CNS 07h (Active Namespace ID
List for a specific I/O Command Set), nvmet_execute_identify_nslist() is
called with match_css set. The command-set filter dereferences req->ns,
but this handler never calls nvmet_req_find_ns(), so req->ns is always
NULL (nvmet_req_init() resets it to NULL). As soon as an enabled
namespace with an NSID greater than the requested value exists,
req->ns->csi dereferences a NULL pointer and oopses.
Besides the crash, the comparison is logically wrong: to filter the list
by command set it must test the command set of the namespace being
iterated, not a single fixed value. Use the loop variable ns->csi.
🎖@cveNotify
🚨 CVE-2026-80814
In the Linux kernel, the following vulnerability has been resolved:
rndis_host: add overflow check in rndis_rx_fixup()
Add an overflow check to ensure that data_offset + data_len + 8 does not
wrap, which would enable an OOB read of the USB data buffer.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
rndis_host: add overflow check in rndis_rx_fixup()
Add an overflow check to ensure that data_offset + data_len + 8 does not
wrap, which would enable an OOB read of the USB data buffer.
🎖@cveNotify
🚨 CVE-2026-80815
In the Linux kernel, the following vulnerability has been resolved:
ALSA: scarlett2: Use a private URB for the notification endpoint
scarlett2_init_notify() used mixer->urb, which
snd_usb_mixer_status_create() allocates for the UAC2 status interrupt
endpoint and mixer.c manages. On a device with that endpoint, the
"already in use" check fires on the status URB and returns 0 for
success without doing anything. No notification URB is submitted, and
cmd_done is left zeroed because it is initialised past that check and
nowhere else. scarlett2_usb_init() then issues SCARLETT2_USB_INIT_1
and wait_for_completion_timeout() would crash adding to the zeroed
wait.head.
Use a separate URB in scarlett2_data, as done for FCP, and initialise
cmd_done in scarlett2_init_private(). mixer.c was also freeing the URB
in snd_usb_mixer_free() and resubmitting it in
snd_usb_mixer_activate(), so scarlett2 must now do both: add
scarlett2_cleanup_urb(), called from private_free and private_suspend,
and a private_resume callback to re-establish the URB after resume.
scarlett2_init_notify() is reached from there, and the URB kill path
in scarlett2_notify() completes cmd_done, leaving a stale count that
would satisfy the next command's wait before the device ACKs. Use
reinit_completion() to clear it.
Also free the URB if the transfer buffer allocation fails, and both if
usb_submit_urb() fails. Move scarlett2_init_notify() up next to
scarlett2_cleanup_urb() so scarlett2_init_private() can reference it
without a forward declaration.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
ALSA: scarlett2: Use a private URB for the notification endpoint
scarlett2_init_notify() used mixer->urb, which
snd_usb_mixer_status_create() allocates for the UAC2 status interrupt
endpoint and mixer.c manages. On a device with that endpoint, the
"already in use" check fires on the status URB and returns 0 for
success without doing anything. No notification URB is submitted, and
cmd_done is left zeroed because it is initialised past that check and
nowhere else. scarlett2_usb_init() then issues SCARLETT2_USB_INIT_1
and wait_for_completion_timeout() would crash adding to the zeroed
wait.head.
Use a separate URB in scarlett2_data, as done for FCP, and initialise
cmd_done in scarlett2_init_private(). mixer.c was also freeing the URB
in snd_usb_mixer_free() and resubmitting it in
snd_usb_mixer_activate(), so scarlett2 must now do both: add
scarlett2_cleanup_urb(), called from private_free and private_suspend,
and a private_resume callback to re-establish the URB after resume.
scarlett2_init_notify() is reached from there, and the URB kill path
in scarlett2_notify() completes cmd_done, leaving a stale count that
would satisfy the next command's wait before the device ACKs. Use
reinit_completion() to clear it.
Also free the URB if the transfer buffer allocation fails, and both if
usb_submit_urb() fails. Move scarlett2_init_notify() up next to
scarlett2_cleanup_urb() so scarlett2_init_private() can reference it
without a forward declaration.
🎖@cveNotify
🚨 CVE-2026-80816
In the Linux kernel, the following vulnerability has been resolved:
ALSA: FCP: Use a private URB for the notification endpoint
fcp_init_notify() used mixer->urb, which snd_usb_mixer_status_create()
allocates for the optional UAC2 status interrupt endpoint and mixer.c
kills, resubmits and frees. On a device with that endpoint,
fcp_init_notify()'s "already set up" early return fires on the status
URB and returns success without doing anything. No FCP notification
URB is submitted, and cmd_done is left zeroed because it is
initialised past that early return and nowhere else. fcp_init() then
issues init1_opcode and wait_for_completion_timeout() would crash
adding to the zeroed wait.head. fcp_cleanup_urb() would also kill and
free mixer.c's status URB.
Use a separate URB in fcp_data, and initialise cmd_done in
fcp_init_private() where fcp_data is allocated. fcp_init_notify() is
reached again after suspend via fcp_reinit(), and the URB kill path in
fcp_notify() completes cmd_done, leaving a stale count that would
satisfy the next command's wait before the device ACKs. Use
reinit_completion() to clear it.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
ALSA: FCP: Use a private URB for the notification endpoint
fcp_init_notify() used mixer->urb, which snd_usb_mixer_status_create()
allocates for the optional UAC2 status interrupt endpoint and mixer.c
kills, resubmits and frees. On a device with that endpoint,
fcp_init_notify()'s "already set up" early return fires on the status
URB and returns success without doing anything. No FCP notification
URB is submitted, and cmd_done is left zeroed because it is
initialised past that early return and nowhere else. fcp_init() then
issues init1_opcode and wait_for_completion_timeout() would crash
adding to the zeroed wait.head. fcp_cleanup_urb() would also kill and
free mixer.c's status URB.
Use a separate URB in fcp_data, and initialise cmd_done in
fcp_init_private() where fcp_data is allocated. fcp_init_notify() is
reached again after suspend via fcp_reinit(), and the URB kill path in
fcp_notify() completes cmd_done, leaving a stale count that would
satisfy the next command's wait before the device ACKs. Use
reinit_completion() to clear it.
🎖@cveNotify
🚨 CVE-2026-80817
In the Linux kernel, the following vulnerability has been resolved:
iommu/iommufd: Fix NULL pointer deref in iommufd_ioas_change_process when racing with iopt_map_file_pages
iommufd_ioas_change_process() iterates every IOAS area while only
holding every IOAS iova_rwsem, so it assumes every area has a non-NULL
pages pointer. That assumption can be false when it runs concurrently
with iopt_map_file_pages().
iopt_map_pages() executes in two phases. It first creates the area and
inserts it into the interval tree under iova_rwsem, with area->pages
still NULL. It then drops iova_rwsem and later fills area->pages
under domains_rwsem. This leaves a window between area creation and
area->pages fill where a concurrent iommufd_ioas_change_process()
can observe the area and dereference a NULL area->pages pointer,
leading to a NULL pointer dereference:
BUG: kernel NULL pointer dereference, address: 00000000000000c0
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 4b655067 P4D 4b655067 PUD 0
Oops: Oops: 0000 [#1] SMP NOPTI
CPU: 0 UID: 0 PID: 11841 Comm: syz.1.628 Not tainted 7.1.0 #3 PREEMPT(full)
Hardware name: QEMU Ubuntu 24.04 PC v2 (i440FX + PIIX, arch_caps fix, 1996), BIOS 1.16.3-debian-1.16.3-2 04/01/2014
RIP: 0010:iommufd_ioas_change_process+0x419/0xd50 drivers/iommu/iommufd/ioas.c:538
Code: 48 89 c3 48 85 c0 0f 84 cc 00 00 00 e8 10 f5 cb fd 48 8d 7b 68 e8 a7 b5 eb fd 48 8b 6b 68 48 8d bd c0 00 00 00 e8 17 b2 eb fd <8b> ad c0 00 00 00 bf 01 00 00 00 89 ee e8 85 ef cb fd 83 fd 01 74
RSP: 0018:ffffc90015c17d28 EFLAGS: 00010246
RAX: ffff8880186d5328 RBX: ffff88801d25e240 RCX: 0000000080000000
RDX: 00000000000002d7 RSI: ffffffff83ba9e10 RDI: 00000000000000c0
RBP: 0000000000000000 R08: ffffffff8e781eb8 R09: 0000000000000000
R10: 00000000000000c0 R11: ffffffff83ba9e29 R12: ffff88802e216008
R13: ffff88802e216000 R14: 0000000000000001 R15: 0000000000000000
FS: 00007f4aea3f66c0(0000) GS:ffff8880b1fa1000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00000000000000c0 CR3: 000000004b75c000 CR4: 0000000000350ef0
Call Trace:
<TASK>
iommufd_fops_ioctl+0x287/0x400 drivers/iommu/iommufd/main.c:533
vfs_ioctl fs/ioctl.c:51 [inline]
__do_sys_ioctl fs/ioctl.c:597 [inline]
__se_sys_ioctl fs/ioctl.c:583 [inline]
__x64_sys_ioctl+0x120/0x170 fs/ioctl.c:583
x64_sys_call+0x1092/0x1fb0 arch/x86/include/generated/asm/syscalls_64.h:17
do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
do_syscall_64+0x10a/0x680 arch/x86/entry/syscall_64.c:94
entry_SYSCALL_64_after_hwframe+0x77/0x7f
RIP: 0033:0x7f4aec1a82bd
Code: ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b0 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f4aea3f6018 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
RAX: ffffffffffffffda RBX: 00007f4aec436090 RCX: 00007f4aec1a82bd
RDX: 0000200000000180 RSI: 0000000000003b92 RDI: 0000000000000003
RBP: 00007f4aec250295 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
R13: 00007f4aec436128 R14: 00007f4aec436090 R15: 00007ffd04ef23e0
</TASK>
Modules linked in:
CR2: 00000000000000c0
---[ end trace 0000000000000000 ]---
RIP: 0010:iommufd_ioas_change_process+0x419/0xd50 drivers/iommu/iommufd/ioas.c:538
Code: 48 89 c3 48 85 c0 0f 84 cc 00 00 00 e8 10 f5 cb fd 48 8d 7b 68 e8 a7 b5 eb fd 48 8b 6b 68 48 8d bd c0 00 00 00 e8 17 b2 eb fd <8b> ad c0 00 00 00 bf 01 00 00 00 89 ee e8 85 ef cb fd 83 fd 01 74
RSP: 0018:ffffc90015c17d28 EFLAGS: 00010246
RAX: ffff8880186d5328 RBX: ffff88801d25e240 RCX: 0000000080000000
RDX: 00000000000002d7 RSI: ffffffff83ba9e10 RDI: 00000000000000c0
RBP: 0000000000000000 R08: ffffffff8e781eb8 R09: 0000000000000000
R10: 00000000000000c0 R11: ffffffff83ba9e29 R12: ffff88802e216008
R13: ffff88802e216000 R14: 0000000000000001 R15: 0000000000000000
FS: 00007f4aea3f66c0(000
---truncated---
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
iommu/iommufd: Fix NULL pointer deref in iommufd_ioas_change_process when racing with iopt_map_file_pages
iommufd_ioas_change_process() iterates every IOAS area while only
holding every IOAS iova_rwsem, so it assumes every area has a non-NULL
pages pointer. That assumption can be false when it runs concurrently
with iopt_map_file_pages().
iopt_map_pages() executes in two phases. It first creates the area and
inserts it into the interval tree under iova_rwsem, with area->pages
still NULL. It then drops iova_rwsem and later fills area->pages
under domains_rwsem. This leaves a window between area creation and
area->pages fill where a concurrent iommufd_ioas_change_process()
can observe the area and dereference a NULL area->pages pointer,
leading to a NULL pointer dereference:
BUG: kernel NULL pointer dereference, address: 00000000000000c0
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 4b655067 P4D 4b655067 PUD 0
Oops: Oops: 0000 [#1] SMP NOPTI
CPU: 0 UID: 0 PID: 11841 Comm: syz.1.628 Not tainted 7.1.0 #3 PREEMPT(full)
Hardware name: QEMU Ubuntu 24.04 PC v2 (i440FX + PIIX, arch_caps fix, 1996), BIOS 1.16.3-debian-1.16.3-2 04/01/2014
RIP: 0010:iommufd_ioas_change_process+0x419/0xd50 drivers/iommu/iommufd/ioas.c:538
Code: 48 89 c3 48 85 c0 0f 84 cc 00 00 00 e8 10 f5 cb fd 48 8d 7b 68 e8 a7 b5 eb fd 48 8b 6b 68 48 8d bd c0 00 00 00 e8 17 b2 eb fd <8b> ad c0 00 00 00 bf 01 00 00 00 89 ee e8 85 ef cb fd 83 fd 01 74
RSP: 0018:ffffc90015c17d28 EFLAGS: 00010246
RAX: ffff8880186d5328 RBX: ffff88801d25e240 RCX: 0000000080000000
RDX: 00000000000002d7 RSI: ffffffff83ba9e10 RDI: 00000000000000c0
RBP: 0000000000000000 R08: ffffffff8e781eb8 R09: 0000000000000000
R10: 00000000000000c0 R11: ffffffff83ba9e29 R12: ffff88802e216008
R13: ffff88802e216000 R14: 0000000000000001 R15: 0000000000000000
FS: 00007f4aea3f66c0(0000) GS:ffff8880b1fa1000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00000000000000c0 CR3: 000000004b75c000 CR4: 0000000000350ef0
Call Trace:
<TASK>
iommufd_fops_ioctl+0x287/0x400 drivers/iommu/iommufd/main.c:533
vfs_ioctl fs/ioctl.c:51 [inline]
__do_sys_ioctl fs/ioctl.c:597 [inline]
__se_sys_ioctl fs/ioctl.c:583 [inline]
__x64_sys_ioctl+0x120/0x170 fs/ioctl.c:583
x64_sys_call+0x1092/0x1fb0 arch/x86/include/generated/asm/syscalls_64.h:17
do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
do_syscall_64+0x10a/0x680 arch/x86/entry/syscall_64.c:94
entry_SYSCALL_64_after_hwframe+0x77/0x7f
RIP: 0033:0x7f4aec1a82bd
Code: ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b0 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f4aea3f6018 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
RAX: ffffffffffffffda RBX: 00007f4aec436090 RCX: 00007f4aec1a82bd
RDX: 0000200000000180 RSI: 0000000000003b92 RDI: 0000000000000003
RBP: 00007f4aec250295 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
R13: 00007f4aec436128 R14: 00007f4aec436090 R15: 00007ffd04ef23e0
</TASK>
Modules linked in:
CR2: 00000000000000c0
---[ end trace 0000000000000000 ]---
RIP: 0010:iommufd_ioas_change_process+0x419/0xd50 drivers/iommu/iommufd/ioas.c:538
Code: 48 89 c3 48 85 c0 0f 84 cc 00 00 00 e8 10 f5 cb fd 48 8d 7b 68 e8 a7 b5 eb fd 48 8b 6b 68 48 8d bd c0 00 00 00 e8 17 b2 eb fd <8b> ad c0 00 00 00 bf 01 00 00 00 89 ee e8 85 ef cb fd 83 fd 01 74
RSP: 0018:ffffc90015c17d28 EFLAGS: 00010246
RAX: ffff8880186d5328 RBX: ffff88801d25e240 RCX: 0000000080000000
RDX: 00000000000002d7 RSI: ffffffff83ba9e10 RDI: 00000000000000c0
RBP: 0000000000000000 R08: ffffffff8e781eb8 R09: 0000000000000000
R10: 00000000000000c0 R11: ffffffff83ba9e29 R12: ffff88802e216008
R13: ffff88802e216000 R14: 0000000000000001 R15: 0000000000000000
FS: 00007f4aea3f66c0(000
---truncated---
🎖@cveNotify
🚨 CVE-2026-80818
In the Linux kernel, the following vulnerability has been resolved:
iommu/tegra241-cmdqv: Fix CMD_SYNC use-after-free on teardown
arm_smmu_impl_remove() is registered as a devres action in
arm_smmu_impl_probe(), before arm_smmu_init_queues() allocates
smmu->cmdq.q.base. On a devres unwind, whether a failed probe or an
unbind, the queue is freed first and arm_smmu_impl_remove() then runs
tegra241_cmdqv_remove_vintf(), whose VINTF deinit issues a CMD_SYNC on
the freed memory.
Observed during testing with a QEMU hack that makes the VCMDQ fail to
enable, so the impl reset fails and probe aborts into the devres unwind:
platform NVDA200C:00: tegra241_cmdqv: VINTF0: VCMDQ0/LVCMDQ0: failed to enable, STATUS=0x00000000
platform NVDA200C:00: tegra241_cmdqv: VINTF0: VCMDQ0/LVCMDQ0: GERRORN=0x0, GERROR=0x4, CONS=0x0
platform NVDA200C:00: tegra241_cmdqv: VINTF0: VCMDQ0/LVCMDQ0: uncleared error detected, resetting
arm-smmu-v3 arm-smmu-v3.0.auto: failed to reset impl
arm-smmu-v3 arm-smmu-v3.0.auto: probe with driver arm-smmu-v3 failed with error -110
Unable to handle kernel paging request at virtual address ffff8000891e0098
...
Internal error: Oops: 0000000096000047 [#1] SMP
...
Call trace:
arm_smmu_cmdq_issue_cmdlist+0x320/0x6fc (P)
tegra241_vcmdq_hw_deinit+0x98/0x168
tegra241_vintf_hw_deinit+0x5c/0x1b0
tegra241_cmdqv_remove_vintf+0x34/0xec
tegra241_cmdqv_remove+0x40/0x9c
arm_smmu_impl_remove+0x20/0x30
devm_action_release+0x14/0x20
devres_release_all+0xa8/0x110
device_unbind_cleanup+0x18/0x84
really_probe+0x1f0/0x29c
Drop the VINTF deinit from tegra241_cmdqv_remove_vintf() so the unwind no
longer touches the freed queue. Quiesce the VINTFs earlier instead. Add a
device_disable() impl op and run it from arm_smmu_disable_action() while
the CMDQ is still up. That handles a live unbind. A failed reset is already
handled because tegra241_vintf_hw_init() deinits the VINTF on its own error
path. tegra241_cmdqv_remove_vintf() is also used by the iommufd viommu
destroy path, so quiesce there too.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
iommu/tegra241-cmdqv: Fix CMD_SYNC use-after-free on teardown
arm_smmu_impl_remove() is registered as a devres action in
arm_smmu_impl_probe(), before arm_smmu_init_queues() allocates
smmu->cmdq.q.base. On a devres unwind, whether a failed probe or an
unbind, the queue is freed first and arm_smmu_impl_remove() then runs
tegra241_cmdqv_remove_vintf(), whose VINTF deinit issues a CMD_SYNC on
the freed memory.
Observed during testing with a QEMU hack that makes the VCMDQ fail to
enable, so the impl reset fails and probe aborts into the devres unwind:
platform NVDA200C:00: tegra241_cmdqv: VINTF0: VCMDQ0/LVCMDQ0: failed to enable, STATUS=0x00000000
platform NVDA200C:00: tegra241_cmdqv: VINTF0: VCMDQ0/LVCMDQ0: GERRORN=0x0, GERROR=0x4, CONS=0x0
platform NVDA200C:00: tegra241_cmdqv: VINTF0: VCMDQ0/LVCMDQ0: uncleared error detected, resetting
arm-smmu-v3 arm-smmu-v3.0.auto: failed to reset impl
arm-smmu-v3 arm-smmu-v3.0.auto: probe with driver arm-smmu-v3 failed with error -110
Unable to handle kernel paging request at virtual address ffff8000891e0098
...
Internal error: Oops: 0000000096000047 [#1] SMP
...
Call trace:
arm_smmu_cmdq_issue_cmdlist+0x320/0x6fc (P)
tegra241_vcmdq_hw_deinit+0x98/0x168
tegra241_vintf_hw_deinit+0x5c/0x1b0
tegra241_cmdqv_remove_vintf+0x34/0xec
tegra241_cmdqv_remove+0x40/0x9c
arm_smmu_impl_remove+0x20/0x30
devm_action_release+0x14/0x20
devres_release_all+0xa8/0x110
device_unbind_cleanup+0x18/0x84
really_probe+0x1f0/0x29c
Drop the VINTF deinit from tegra241_cmdqv_remove_vintf() so the unwind no
longer touches the freed queue. Quiesce the VINTFs earlier instead. Add a
device_disable() impl op and run it from arm_smmu_disable_action() while
the CMDQ is still up. That handles a live unbind. A failed reset is already
handled because tegra241_vintf_hw_init() deinits the VINTF on its own error
path. tegra241_cmdqv_remove_vintf() is also used by the iommufd viommu
destroy path, so quiesce there too.
🎖@cveNotify
🚨 CVE-2026-80819
In the Linux kernel, the following vulnerability has been resolved:
Bluetooth: RFCOMM: take rfcomm_mutex for the deferred setup accept
rfcomm_sock_recvmsg() completes a deferred setup by calling
rfcomm_dlc_accept() without holding any RFCOMM lock:
if (test_and_clear_bit(RFCOMM_DEFER_SETUP, &d->flags)) {
rfcomm_dlc_accept(d);
return 0;
}
and rfcomm_dlc_accept() dereferences the session on its first line:
struct sock *sk = d->session->sock->sk;
Every other path that touches d->session runs under rfcomm_mutex:
rfcomm_dlc_open(), rfcomm_dlc_close(), rfcomm_dlc_exists(),
rfcomm_dlc_send_rpn(), and the RFCOMM thread through
rfcomm_process_sessions(). rfcomm_connect_ind() is even documented as
"called under rfcomm_lock()". This call site is the only one that skips
it.
The RFCOMM_DEFER_SETUP bit looks like it serialises the accept against
teardown, since __rfcomm_dlc_close() returns early when it wins the
test_and_clear. But rfcomm_recv_disc() forces the state first:
d->state = BT_CLOSED;
__rfcomm_dlc_close(d, err);
and the early return only covers BT_CONNECT, BT_CONFIG, BT_OPEN and
BT_CONNECT2. With the state already BT_CLOSED that switch does not
match, the bit is never consulted, and __rfcomm_dlc_close() falls
through to rfcomm_dlc_unlink(), which sets d->session = NULL.
So a remote DISC on a deferred dlc clears the session while leaving
RFCOMM_DEFER_SETUP set. The next recvmsg() then passes the
test_and_clear and dereferences a NULL session. No timing window is
needed: once the DISC has been processed, the dereference is
unconditional.
Give rfcomm_dlc_accept() the same shape as rfcomm_dlc_open() and
rfcomm_dlc_close(): an exported wrapper that takes rfcomm_mutex and
re-checks the session, around a __rfcomm_dlc_accept() that the two
in-core callers, which already hold the mutex, keep using.
Reproduced on a KASAN + PROVE_LOCKING kernel with a BR/EDR peer emulated
over /dev/vhci: the peer brings up an ACL link, opens L2CAP on the
RFCOMM PSM, starts a session, opens a dlc on a channel bound with
BT_DEFER_SETUP, and sends DISC after the socket is accepted. recv() on
the accepted socket then hits:
Oops: general protection fault
KASAN: null-ptr-deref in range [0x0000000000000010-0x0000000000000017]
RIP: 0010:rfcomm_dlc_accept+0x54/0x350
Call Trace:
rfcomm_sock_recvmsg+0x1cd/0x230
sock_recvmsg+0x166/0x1c0
__sys_recvfrom+0x20d/0x300
0x10 is the offset of sock in struct rfcomm_session. With this patch the
same run completes with recv() returning 0 and no report, and lockdep
stays quiet, confirming rfcomm_mutex is still taken before lock_sock on
this path as it is on the thread side.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
Bluetooth: RFCOMM: take rfcomm_mutex for the deferred setup accept
rfcomm_sock_recvmsg() completes a deferred setup by calling
rfcomm_dlc_accept() without holding any RFCOMM lock:
if (test_and_clear_bit(RFCOMM_DEFER_SETUP, &d->flags)) {
rfcomm_dlc_accept(d);
return 0;
}
and rfcomm_dlc_accept() dereferences the session on its first line:
struct sock *sk = d->session->sock->sk;
Every other path that touches d->session runs under rfcomm_mutex:
rfcomm_dlc_open(), rfcomm_dlc_close(), rfcomm_dlc_exists(),
rfcomm_dlc_send_rpn(), and the RFCOMM thread through
rfcomm_process_sessions(). rfcomm_connect_ind() is even documented as
"called under rfcomm_lock()". This call site is the only one that skips
it.
The RFCOMM_DEFER_SETUP bit looks like it serialises the accept against
teardown, since __rfcomm_dlc_close() returns early when it wins the
test_and_clear. But rfcomm_recv_disc() forces the state first:
d->state = BT_CLOSED;
__rfcomm_dlc_close(d, err);
and the early return only covers BT_CONNECT, BT_CONFIG, BT_OPEN and
BT_CONNECT2. With the state already BT_CLOSED that switch does not
match, the bit is never consulted, and __rfcomm_dlc_close() falls
through to rfcomm_dlc_unlink(), which sets d->session = NULL.
So a remote DISC on a deferred dlc clears the session while leaving
RFCOMM_DEFER_SETUP set. The next recvmsg() then passes the
test_and_clear and dereferences a NULL session. No timing window is
needed: once the DISC has been processed, the dereference is
unconditional.
Give rfcomm_dlc_accept() the same shape as rfcomm_dlc_open() and
rfcomm_dlc_close(): an exported wrapper that takes rfcomm_mutex and
re-checks the session, around a __rfcomm_dlc_accept() that the two
in-core callers, which already hold the mutex, keep using.
Reproduced on a KASAN + PROVE_LOCKING kernel with a BR/EDR peer emulated
over /dev/vhci: the peer brings up an ACL link, opens L2CAP on the
RFCOMM PSM, starts a session, opens a dlc on a channel bound with
BT_DEFER_SETUP, and sends DISC after the socket is accepted. recv() on
the accepted socket then hits:
Oops: general protection fault
KASAN: null-ptr-deref in range [0x0000000000000010-0x0000000000000017]
RIP: 0010:rfcomm_dlc_accept+0x54/0x350
Call Trace:
rfcomm_sock_recvmsg+0x1cd/0x230
sock_recvmsg+0x166/0x1c0
__sys_recvfrom+0x20d/0x300
0x10 is the offset of sock in struct rfcomm_session. With this patch the
same run completes with recv() returning 0 and no report, and lockdep
stays quiet, confirming rfcomm_mutex is still taken before lock_sock on
this path as it is on the thread side.
🎖@cveNotify
🚨 CVE-2026-80820
In the Linux kernel, the following vulnerability has been resolved:
xfs: don't livelock in scrub on a circular unlinked list
LOLLM points out that online fsck can livelock if an unlinked inode list
contains a loop. Use a bitmap to detect cycles.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
xfs: don't livelock in scrub on a circular unlinked list
LOLLM points out that online fsck can livelock if an unlinked inode list
contains a loop. Use a bitmap to detect cycles.
🎖@cveNotify
🚨 CVE-2026-80821
In the Linux kernel, the following vulnerability has been resolved:
nvmet: pci-epf: put CQ ref on create_cq mapping failure
nvmet_pci_epf_create_cq() calls nvmet_cq_create(), which takes a
reference on the controller and installs the completion queue. If the
subsequent PCI address-space mapping fails or returns a too-small partial
mapping, the function jumps to err_internal / err_unmap_queue without
calling nvmet_cq_put(). The matching put in nvmet_pci_epf_delete_cq() is
gated on NVMET_PCI_EPF_Q_LIVE, which is only set after the mapping
succeeds, so teardown never releases these references. A remote PCI host
that drives Create IO CQ commands with a failing PRP1/pci_addr therefore
leaks the CQ and a controller reference on each attempt.
Drop the CQ reference on the mapping-failure paths. The err_internal and
err_unmap_queue labels are only reachable after nvmet_cq_create() has
succeeded, so this pairs the create/put correctly.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
nvmet: pci-epf: put CQ ref on create_cq mapping failure
nvmet_pci_epf_create_cq() calls nvmet_cq_create(), which takes a
reference on the controller and installs the completion queue. If the
subsequent PCI address-space mapping fails or returns a too-small partial
mapping, the function jumps to err_internal / err_unmap_queue without
calling nvmet_cq_put(). The matching put in nvmet_pci_epf_delete_cq() is
gated on NVMET_PCI_EPF_Q_LIVE, which is only set after the mapping
succeeds, so teardown never releases these references. A remote PCI host
that drives Create IO CQ commands with a failing PRP1/pci_addr therefore
leaks the CQ and a controller reference on each attempt.
Drop the CQ reference on the mapping-failure paths. The err_internal and
err_unmap_queue labels are only reachable after nvmet_cq_create() has
succeeded, so this pairs the create/put correctly.
🎖@cveNotify
🚨 CVE-2026-80823
In the Linux kernel, the following vulnerability has been resolved:
nfc: st21nfca: validate ATR_REQ length against the received frame
st21nfca_tm_recv_atr_req() checks that the received ATR_REQ frame is at
least ST21NFCA_ATR_REQ_MIN_SIZE and that the self-declared atr_req->length
is at least sizeof(struct st21nfca_atr_req), but never checks that
atr_req->length does not exceed the actual received length (skb->len).
st21nfca_tm_send_atr_res() then trusts the declared length:
gb_len = atr_req->length - sizeof(struct st21nfca_atr_req);
...
memcpy(atr_res->gbi, atr_req->gbi, gb_len);
so an RF peer that sends a short frame but sets atr_req->length larger
than the frame makes gb_len exceed the general bytes actually present,
and the memcpy reads out of bounds past the received skb. Those bytes are
placed in the ATR_RES and sent back to the peer (kernel-memory disclosure
to a proximity attacker); a larger declared length is an out-of-bounds
read (DoS).
Reject frames whose declared length exceeds the received length. The
adjacent nfc_tm_activated() path in the same function already derives its
general-bytes length from skb->len rather than the declared field.
Found by 0sec (https://0sec.ai) using automated source analysis; the
missing bound is evident from source. Compile-tested.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
nfc: st21nfca: validate ATR_REQ length against the received frame
st21nfca_tm_recv_atr_req() checks that the received ATR_REQ frame is at
least ST21NFCA_ATR_REQ_MIN_SIZE and that the self-declared atr_req->length
is at least sizeof(struct st21nfca_atr_req), but never checks that
atr_req->length does not exceed the actual received length (skb->len).
st21nfca_tm_send_atr_res() then trusts the declared length:
gb_len = atr_req->length - sizeof(struct st21nfca_atr_req);
...
memcpy(atr_res->gbi, atr_req->gbi, gb_len);
so an RF peer that sends a short frame but sets atr_req->length larger
than the frame makes gb_len exceed the general bytes actually present,
and the memcpy reads out of bounds past the received skb. Those bytes are
placed in the ATR_RES and sent back to the peer (kernel-memory disclosure
to a proximity attacker); a larger declared length is an out-of-bounds
read (DoS).
Reject frames whose declared length exceeds the received length. The
adjacent nfc_tm_activated() path in the same function already derives its
general-bytes length from skb->len rather than the declared field.
Found by 0sec (https://0sec.ai) using automated source analysis; the
missing bound is evident from source. Compile-tested.
🎖@cveNotify
🚨 CVE-2026-80824
In the Linux kernel, the following vulnerability has been resolved:
usb: usbfs: fix use-after-free of usb_device in usbdev_release()
usbdev_release() drops its reference to the struct usb_device before
draining the list of completed async URBs, but that drain path reads back
through the same object: free_async() calls dec_usb_memory_use_count()
for any URB whose buffer came from the usbfs mmap() region, and its first
statement is bus_to_hcd(ps->dev->bus).
After a disconnect the usbfs reference can be the last one, in which case
usb_put_dev() frees the device and the subsequent loop reads offset 80 of
freed memory and uses the result as a struct usb_hcd *, which
hcd_buffer_free_pages() then dereferences.
This is reachable by an unprivileged process that has read/write access to
a /dev/bus/usb node: mmap() the fd, submit one URB with a buffer inside the
mapping, wait for the device to be unplugged, then munmap() and close().
It reproduces on every attempt rather than being a race, because a live
MAP_SHARED vma holds a reference on the struct file, so usbdev_release()
cannot run until the last vma is gone and the freeing branch of
dec_usb_memory_use_count() is always taken.
BUG: KASAN: slab-use-after-free in dec_usb_memory_use_count+0x3ae/0x410
Read of size 8 at addr ffff8880122ee050 by task poc/769
CPU: 1 UID: 1000 PID: 769 Comm: poc Tainted: G B 6.12.94 #3
Call Trace:
dec_usb_memory_use_count+0x3ae/0x410
free_async+0x2aa/0x4f0
usbdev_release+0x375/0x460
__fput+0x3ea/0xb50
__x64_sys_close+0x86/0x100
Allocated by task 11:
usb_alloc_dev+0x55/0xd90
hub_event+0x2524/0x43d0
Freed by task 769:
kfree+0x121/0x360
device_release+0xd2/0x280
usb_put_dev+0x23/0x30
usbdev_release+0x2d8/0x460
Release the device reference after the drain loop instead. Nothing between
the two points requires it to have been dropped.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
usb: usbfs: fix use-after-free of usb_device in usbdev_release()
usbdev_release() drops its reference to the struct usb_device before
draining the list of completed async URBs, but that drain path reads back
through the same object: free_async() calls dec_usb_memory_use_count()
for any URB whose buffer came from the usbfs mmap() region, and its first
statement is bus_to_hcd(ps->dev->bus).
After a disconnect the usbfs reference can be the last one, in which case
usb_put_dev() frees the device and the subsequent loop reads offset 80 of
freed memory and uses the result as a struct usb_hcd *, which
hcd_buffer_free_pages() then dereferences.
This is reachable by an unprivileged process that has read/write access to
a /dev/bus/usb node: mmap() the fd, submit one URB with a buffer inside the
mapping, wait for the device to be unplugged, then munmap() and close().
It reproduces on every attempt rather than being a race, because a live
MAP_SHARED vma holds a reference on the struct file, so usbdev_release()
cannot run until the last vma is gone and the freeing branch of
dec_usb_memory_use_count() is always taken.
BUG: KASAN: slab-use-after-free in dec_usb_memory_use_count+0x3ae/0x410
Read of size 8 at addr ffff8880122ee050 by task poc/769
CPU: 1 UID: 1000 PID: 769 Comm: poc Tainted: G B 6.12.94 #3
Call Trace:
dec_usb_memory_use_count+0x3ae/0x410
free_async+0x2aa/0x4f0
usbdev_release+0x375/0x460
__fput+0x3ea/0xb50
__x64_sys_close+0x86/0x100
Allocated by task 11:
usb_alloc_dev+0x55/0xd90
hub_event+0x2524/0x43d0
Freed by task 769:
kfree+0x121/0x360
device_release+0xd2/0x280
usb_put_dev+0x23/0x30
usbdev_release+0x2d8/0x460
Release the device reference after the drain loop instead. Nothing between
the two points requires it to have been dropped.
🎖@cveNotify
🚨 CVE-2026-80825
In the Linux kernel, the following vulnerability has been resolved:
wifi: mt76: mt7925: ensure tx headroom in usb_sdio_tx_prepare_skb
mt7925_usb_sdio_tx_prepare_skb() pushes a TX descriptor and a USB
header onto every skb and assumes the headroom for them is already
there. That holds for locally generated traffic, where mac80211
reserves hw->extra_tx_headroom, but forwarded frames are sent through
ieee80211_8023_xmit(), which does not reserve it. Bridge a wired
interface to an mt7925u AP and the first forwarded frame that arrives
short panics the kernel:
skbuff: skb_under_panic: len:415 put:4 tail:0x19b end:0x640 dev:wlan1
kernel BUG at net/core/skbuff.c:212!
Call trace:
skb_panic+0x58/0x60 (P)
skb_push+0x58/0x60
mt7925_usb_sdio_tx_prepare_skb+0xf8/0x1b8 [mt7925_common]
mt76u_tx_queue_skb+0xa0/0x1f8 [mt76_usb]
__mt76_tx_queue_skb+0x54/0xe8 [mt76]
mt76_txq_schedule.part.0+0x204/0x478 [mt76]
mt76_txq_schedule_all+0x50/0x80 [mt76]
mt792x_tx_worker+0x68/0x100 [mt792x_lib]
__mt76_worker_fn+0x84/0x150 [mt76]
Whether a given setup hits it depends on how much headroom the ingress
netdev leaves in its rx skbs. Reproduced on a Raspberry Pi 5 bridging
onboard ethernet to a Netgear A9000; originally reported on an MT7986
router running OpenWrt. Nick Morrow's testing on a Pi 4 (bcmgenet),
which leaves more headroom, helped narrow the trigger to the ingress
path.
The same bug was fixed on mt7921 by commit 98c4d0abf5c4 ("mt76:
mt7921: don't assume adequate headroom for SDIO headers"), but mt7925
was copied from mt7921 without the fix. Add the same guard here.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
wifi: mt76: mt7925: ensure tx headroom in usb_sdio_tx_prepare_skb
mt7925_usb_sdio_tx_prepare_skb() pushes a TX descriptor and a USB
header onto every skb and assumes the headroom for them is already
there. That holds for locally generated traffic, where mac80211
reserves hw->extra_tx_headroom, but forwarded frames are sent through
ieee80211_8023_xmit(), which does not reserve it. Bridge a wired
interface to an mt7925u AP and the first forwarded frame that arrives
short panics the kernel:
skbuff: skb_under_panic: len:415 put:4 tail:0x19b end:0x640 dev:wlan1
kernel BUG at net/core/skbuff.c:212!
Call trace:
skb_panic+0x58/0x60 (P)
skb_push+0x58/0x60
mt7925_usb_sdio_tx_prepare_skb+0xf8/0x1b8 [mt7925_common]
mt76u_tx_queue_skb+0xa0/0x1f8 [mt76_usb]
__mt76_tx_queue_skb+0x54/0xe8 [mt76]
mt76_txq_schedule.part.0+0x204/0x478 [mt76]
mt76_txq_schedule_all+0x50/0x80 [mt76]
mt792x_tx_worker+0x68/0x100 [mt792x_lib]
__mt76_worker_fn+0x84/0x150 [mt76]
Whether a given setup hits it depends on how much headroom the ingress
netdev leaves in its rx skbs. Reproduced on a Raspberry Pi 5 bridging
onboard ethernet to a Netgear A9000; originally reported on an MT7986
router running OpenWrt. Nick Morrow's testing on a Pi 4 (bcmgenet),
which leaves more headroom, helped narrow the trigger to the ingress
path.
The same bug was fixed on mt7921 by commit 98c4d0abf5c4 ("mt76:
mt7921: don't assume adequate headroom for SDIO headers"), but mt7925
was copied from mt7921 without the fix. Add the same guard here.
🎖@cveNotify
🚨 CVE-2026-80826
In the Linux kernel, the following vulnerability has been resolved:
USB: c67x00: fix use-after-free in c67x00_add_iso_urb()
When TD creation fails for the last packet of an isochronous URB,
c67x00_add_iso_urb() gives the URB back before updating the endpoint
scheduling state.
c67x00_giveback_urb() frees the URB private data, and the completion
callback may release the final URB reference. The following accesses to
urbp->ep_data, urb->interval, and urbp->cnt can therefore use freed
memory.
Update next_frame and cnt before giving back the failed final packet,
making the giveback the last operation that uses the URB and its private
data.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
USB: c67x00: fix use-after-free in c67x00_add_iso_urb()
When TD creation fails for the last packet of an isochronous URB,
c67x00_add_iso_urb() gives the URB back before updating the endpoint
scheduling state.
c67x00_giveback_urb() frees the URB private data, and the completion
callback may release the final URB reference. The following accesses to
urbp->ep_data, urb->interval, and urbp->cnt can therefore use freed
memory.
Update next_frame and cnt before giving back the failed final packet,
making the giveback the last operation that uses the URB and its private
data.
🎖@cveNotify
🚨 CVE-2026-80827
In the Linux kernel, the following vulnerability has been resolved:
USB: serial: option: fix slab OOB read in interrupt URB callback
The interrupt URB buffer is allocated in setup_port_interrupt_in() based
on the endpoint's wMaxPacketSize:
buffer_size = usb_endpoint_maxp(epd);
port->interrupt_in_buffer = kmalloc(buffer_size, GFP_KERNEL);
When a USB device declares wMaxPacketSize = 8 on its interrupt IN
endpoint, the buffer is allocated from kmalloc-8 cache (exactly
8 bytes).
If the device sends a short packet (actual_length < wMaxPacketSize),
the URB completes with status == 0 and the callback proceeds to read:
data[sizeof(struct usb_ctrlrequest)]
which evaluates to data[8], accessing 1 byte beyond the allocated 8-byte
buffer. This results in a slab out-of-bounds read.
Fix this by adding the missing bounds check: first verify that the
actual length is large enough to contain the struct usb_ctrlrequest
header before accessing req_pkt->bRequestType and req_pkt->bRequest,
and then verify that there is an additional byte for the modem signal
state before reading data[sizeof(struct usb_ctrlrequest)] inside the
conditional. Use sizeof(*req_pkt) instead of sizeof(struct
usb_ctrlrequest) for consistency.
[ johan: use dev_err(); split signals declaration and initialisation ]
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
USB: serial: option: fix slab OOB read in interrupt URB callback
The interrupt URB buffer is allocated in setup_port_interrupt_in() based
on the endpoint's wMaxPacketSize:
buffer_size = usb_endpoint_maxp(epd);
port->interrupt_in_buffer = kmalloc(buffer_size, GFP_KERNEL);
When a USB device declares wMaxPacketSize = 8 on its interrupt IN
endpoint, the buffer is allocated from kmalloc-8 cache (exactly
8 bytes).
If the device sends a short packet (actual_length < wMaxPacketSize),
the URB completes with status == 0 and the callback proceeds to read:
data[sizeof(struct usb_ctrlrequest)]
which evaluates to data[8], accessing 1 byte beyond the allocated 8-byte
buffer. This results in a slab out-of-bounds read.
Fix this by adding the missing bounds check: first verify that the
actual length is large enough to contain the struct usb_ctrlrequest
header before accessing req_pkt->bRequestType and req_pkt->bRequest,
and then verify that there is an additional byte for the modem signal
state before reading data[sizeof(struct usb_ctrlrequest)] inside the
conditional. Use sizeof(*req_pkt) instead of sizeof(struct
usb_ctrlrequest) for consistency.
[ johan: use dev_err(); split signals declaration and initialisation ]
🎖@cveNotify
🚨 CVE-2026-80828
In the Linux kernel, the following vulnerability has been resolved:
ALSA: usb-audio: Complete cleanup after system-resume errors
A failed system resume can leave the card unusable until reboot.
usb_audio_resume() jumps to err_out when snd_usb_pcm_resume() or
snd_usb_mixer_resume() fails. The error path skips the out: block, which
restores D0 and decrements chip->num_suspended_intf.
The card stays in SNDRV_CTL_POWER_D3hot, so later control access blocks in
snd_power_ref_and_wait(). USB core logs an interface resume callback error.
It does not retry that callback, so a later callback cannot complete the
skipped cleanup.
usb_audio_suspend() increments num_suspended_intf before returning success.
A system-resume callback must consume the system-suspend count even if a
component resume fails. Otherwise, the stranded count skews later suspend
and resume cycles.
Do not apply this cleanup to runtime-resume errors. Runtime PM can retry
-EAGAIN or -EBUSY without another suspend callback. The count must continue
to describe that suspended interface. Other runtime-resume errors latch
runtime_error in the PM core and do not cause an immediate callback retry.
Both parts of the system-resume error path are longstanding. Commit
88a8516a2128a ("ALSA: usbaudio: implement USB autosuspend") introduced
err_out past the D0 restore. Commit 862b2509d157c ("ALSA: usb-audio: Fix
inconsistent card PM state after resume") later moved
num_suspended_intf-- into the out: block. The error path now skips both
operations.
No third-party code is needed to reach the error path.
snd_usb_mixer_resume() ends in snd_usb_mixer_activate(), which returns the
result of usb_submit_urb() for devices that have a mixer status URB. Its
mixer->private_resume hook can also fail through scarlett2_init_notify().
snd_usb_pcm_resume() issues a SET_CUR request to a UAC3 power domain. It
can return -EPIPE or -EIO when the device stalls the request.
Route a component error through out: only when system_suspend is nonzero.
Continue to return runtime-resume errors through err_out. Later component
resume stages remain skipped. The original error still reaches USB core.
A later transfer can fail if the device did not recover.
I reproduced the system-resume failure on an Audient iD14 MkI with an
out-of-tree diagnostic mixer resume hook. An injected -EIO on the unpatched
core left control readers in uninterruptible sleep in
snd_power_ref_and_wait() until a reboot. With this patch, the same failure
restored control access. A second system suspend and resume also succeeded
after I disabled fault injection.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
ALSA: usb-audio: Complete cleanup after system-resume errors
A failed system resume can leave the card unusable until reboot.
usb_audio_resume() jumps to err_out when snd_usb_pcm_resume() or
snd_usb_mixer_resume() fails. The error path skips the out: block, which
restores D0 and decrements chip->num_suspended_intf.
The card stays in SNDRV_CTL_POWER_D3hot, so later control access blocks in
snd_power_ref_and_wait(). USB core logs an interface resume callback error.
It does not retry that callback, so a later callback cannot complete the
skipped cleanup.
usb_audio_suspend() increments num_suspended_intf before returning success.
A system-resume callback must consume the system-suspend count even if a
component resume fails. Otherwise, the stranded count skews later suspend
and resume cycles.
Do not apply this cleanup to runtime-resume errors. Runtime PM can retry
-EAGAIN or -EBUSY without another suspend callback. The count must continue
to describe that suspended interface. Other runtime-resume errors latch
runtime_error in the PM core and do not cause an immediate callback retry.
Both parts of the system-resume error path are longstanding. Commit
88a8516a2128a ("ALSA: usbaudio: implement USB autosuspend") introduced
err_out past the D0 restore. Commit 862b2509d157c ("ALSA: usb-audio: Fix
inconsistent card PM state after resume") later moved
num_suspended_intf-- into the out: block. The error path now skips both
operations.
No third-party code is needed to reach the error path.
snd_usb_mixer_resume() ends in snd_usb_mixer_activate(), which returns the
result of usb_submit_urb() for devices that have a mixer status URB. Its
mixer->private_resume hook can also fail through scarlett2_init_notify().
snd_usb_pcm_resume() issues a SET_CUR request to a UAC3 power domain. It
can return -EPIPE or -EIO when the device stalls the request.
Route a component error through out: only when system_suspend is nonzero.
Continue to return runtime-resume errors through err_out. Later component
resume stages remain skipped. The original error still reaches USB core.
A later transfer can fail if the device did not recover.
I reproduced the system-resume failure on an Audient iD14 MkI with an
out-of-tree diagnostic mixer resume hook. An injected -EIO on the unpatched
core left control readers in uninterruptible sleep in
snd_power_ref_and_wait() until a reboot. With this patch, the same failure
restored control access. A second system suspend and resume also succeeded
after I disabled fault injection.
🎖@cveNotify
🚨 CVE-2026-80829
In the Linux kernel, the following vulnerability has been resolved:
ALSA: usb-audio: fix OOB write in snd_usbmidi_novation_output()
snd_usbmidi_novation_output() lays out a two-byte header at
transfer_buffer[0..1] and passes &transfer_buffer[2] together with a
length of ep->max_transfer - 2 to snd_rawmidi_transmit():
count = snd_rawmidi_transmit(ep->ports[0].substream,
&transfer_buffer[2],
ep->max_transfer - 2);
ep->max_transfer comes from the output endpoint's wMaxPacketSize via
usb_maxpacket(). A malformed or malicious device can advertise a bulk
OUT endpoint with a wMaxPacketSize of 1 - the USB core only clamps this
value downwards - so ep->max_transfer becomes 1 and the count argument
becomes -1.
snd_rawmidi_transmit() passes the negative count on to
__snd_rawmidi_transmit_peek(), where "if (count1 > count) count1 = count"
leaves count1 negative; get_aligned_size() keeps it negative for a
byte-stream substream, so the following memcpy(buffer, ..., count1) runs
with a (size_t)-1 length and writes far past the transfer buffer, which
was allocated with usb_alloc_coherent(ep->max_transfer).
This is the same class of bug that was fixed for snd_usbmidi_akai_output()
in commit 0970274613fb ("ALSA: usb-audio: fix OOB write in
snd_usbmidi_akai_output()"); the novation output routine was left
unguarded. Bail out when the endpoint cannot hold the two-byte header
plus at least one payload byte.
🎖@cveNotify
In the Linux kernel, the following vulnerability has been resolved:
ALSA: usb-audio: fix OOB write in snd_usbmidi_novation_output()
snd_usbmidi_novation_output() lays out a two-byte header at
transfer_buffer[0..1] and passes &transfer_buffer[2] together with a
length of ep->max_transfer - 2 to snd_rawmidi_transmit():
count = snd_rawmidi_transmit(ep->ports[0].substream,
&transfer_buffer[2],
ep->max_transfer - 2);
ep->max_transfer comes from the output endpoint's wMaxPacketSize via
usb_maxpacket(). A malformed or malicious device can advertise a bulk
OUT endpoint with a wMaxPacketSize of 1 - the USB core only clamps this
value downwards - so ep->max_transfer becomes 1 and the count argument
becomes -1.
snd_rawmidi_transmit() passes the negative count on to
__snd_rawmidi_transmit_peek(), where "if (count1 > count) count1 = count"
leaves count1 negative; get_aligned_size() keeps it negative for a
byte-stream substream, so the following memcpy(buffer, ..., count1) runs
with a (size_t)-1 length and writes far past the transfer buffer, which
was allocated with usb_alloc_coherent(ep->max_transfer).
This is the same class of bug that was fixed for snd_usbmidi_akai_output()
in commit 0970274613fb ("ALSA: usb-audio: fix OOB write in
snd_usbmidi_akai_output()"); the novation output routine was left
unguarded. Bail out when the endpoint cannot hold the two-byte header
plus at least one payload byte.
🎖@cveNotify