CVEs of 2026.
Every CVE published in 2026 that we hold, ranked by exploit odds.
-
CVE-2026-99459
In the Linux kernel, the following vulnerability has been resolved: mshv_vtl: detach poll waiter on POLLFREE MSHV_SET_POLL_FILE keeps a poll wait entry after the ioctl returns. A file reference does not always pin the registered waitqueue: signalfd_poll() uses current->sighand->signalfd_wqh, whose lifetime is tied to the task. signalfd_cleanup() sends POLLFREE before the sighand is freed. The MSHV callback currently leaves its entry linked, so __wake_up_pollfree() warns and a later replacement calls remove_wait_queue() through a stale wqh pointer. A source-equivalent KASAN run reproduced: BUG: KASAN: slab-use-after-free in _raw_spin_lock_irqsave _raw_spin_lock_irqsave remove_wait_queue harness_ioctl [mshv_pollfree_harness] The object belonged to sighand_cache and had been released through slab_free_after_rcu_debug() after __cleanup_sighand(). On POLLFREE, unlink the entry under the provider waitqueue lock. Preserve the existing CPU cancellation before publishing a NULL wqh, since a NULL load lets a racing replacement reuse the persistent poll state. Use a matching acquire load under RCU when removing a normal registration. The callback takes no mutex and does not sleep, so the existing mutex-to-waitqueue-lock order is unchanged. An LLM assisted the source audit, fix development, and validation of this issue.
-
CVE-2026-99446
In the Linux kernel, the following vulnerability has been resolved: phy: core: Fix race-condition between _of_phy_get() and try_module_get() Sashiko asked during a patch review if the existing usage pattern had a race condition; specifically in of_phy_get() if it was possible between returning from _of_phy_get() and running try_module_get() that a module might be unbound leading to use-after-free. Looking at the code this appears to be so, there is no linkage between the phy and module under a synchronisation primitive. Using the phy_provider_mutex in phy_get() will ensure there is a link between the returned phy pointer and the module_get() bumping the module reference count. Amend phy_get(), of_phy_get() and devm_of_phy_get_by_index() to fix the same usage pattern. phy_provider_unregister() must take the phy_provider_mutex so amending phy_get()/of_phy_get() to take that same mutex guarantees there is no use-after-free.
-
CVE-2026-99447
In the Linux kernel, the following vulnerability has been resolved: spi: spi-pci1xxxx: quiesce DMA engines on transfer timeout If a DMA-mapped transfer times out waiting for DMA_TERM and neither DMA abort interrupt arrived, pci1xxxx_spi_transfer_with_dma() returns through the error label while both DMA engines are still enabled, the done/abort interrupt status is still armed and the transfer state (p->xfer, p->tx_sgl, p->rx_sgl) is still registered. The ISRs keep firing afterwards: they reprogram the DMA engines from p->tx_sgl / p->rx_sgl and signal completion through p->xfer and p->spi_xfer_done, all of which the SPI core is free to unmap or reuse once transfer_one() has returned, which is a use-after-free. A late done interrupt can also start the next transfer queued behind the failed one. Stop both DMA engines, drain the pending done/abort interrupt status and synchronize the DMA IRQs before dropping the in-progress state and the transfer pointers, and re-enable the engines afterwards so that subsequent transfers keep working. The engines are only turned back on once both were confirmed to have stopped: re-enabling an engine that failed to stop would cancel the pending stop request and let it keep running into buffers the SPI core is about to unmap or free, which is exactly the window this patch closes. An engine that failed to stop is left disabled, as the abort paths already do. The DMA interrupt status and mask registers are shared between the two SPI instances, so the cleanup drains only this instance's done/abort status bits instead of masking the interrupt lines, which would also suppress the peer instance's completion interrupts. Also make the DMA ISRs ignore interrupts while no transfer is in progress so that a stray MSI after the cleanup can no longer act on the cleared state. No PCI1XXXX hardware was available for testing and QEMU has no model of this PCIe part, so the timeout path was exercised with a userspace replica of the timeout branch and the DMA ISRs, with the register and interrupt accesses mocked: on the unfixed code both a wedged engine that clears after the timeout and a late done interrupt touch the freed transfer state, while with this fix applied neither does and an engine that failed to stop is left disabled. The module was also loaded and unloaded in a QEMU x86_64 guest running a kernel with this fix applied.
-
CVE-2026-99448
In the Linux kernel, the following vulnerability has been resolved: ufs: use u64 for directory size in ufs_last_byte ufs1_read_inode() and ufs2_read_inode() copy the untrusted 64-bit on-disk size into inode->i_size. ufs_last_byte() then truncates that value to an unsigned int before calculating the valid extent of a directory folio. A directory size of exactly 4 GiB is truncated to zero. In ufs_find_entry(), subtracting the requested record length from that zero forms an endpoint almost 4 GiB beyond the mapped folio. A crafted UFS2 image produces: BUG: KASAN: use-after-free in ufs_find_entry+0x583/0x6e0 [ufs] Read of size 1 ... ufs_find_entry ufs_inode_by_name ufs_lookup __lookup_slow path_lookupat The KASAN classification reflects that the adjacent physical page was free; the source-level operation is an out-of-folio read. A controlled UFS1 test with CONFIG_UFS_FS_WRITE enabled also made ufs_find_entry() return a directory entry from an adjacent anonymous page. unlink() then passed that pointer to ufs_delete_entry(), which cleared the adjacent page's 3 2-bit d_ino. This write result reproduced three out of three times. UFS normally requires a privileged mount path. A relevant boundary is a privileged automounter or image-processing service handling an attacker-supplied filesystem. The write primitive additionally requires UFS1 to be mounted read-write with CONFIG_UFS_FS_WRITE enabled; the read is reachable with read-only UFS2. Keep the intermediate size arithmetic at 64 bits so non-final pages are bounded at PAGE_SIZE. With this change, the same controlled tests reach the real on-disk directory entry and produce no adjacent-page write or KASAN report. A reproducer and full validation logs are available on request. The vulnerable helper is present in v7.2-rc3, v7.1, v6.18.38, v6.12.95, v6.6.111, and upstream master at 1229e2e57a5c. Runtime reproduction and fix validation were performed in an x86-64 QEMU guest running v7.2-rc3 with generic KASAN.
-
CVE-2026-99449
In the Linux kernel, the following vulnerability has been resolved: usb: gadget: fix refcount underflow in gadgetfs_bind() error path gadgetfs_bind() calls get_dev() only on the success path, but gadgetfs_unbind() called from the enomem error label unconditionally calls put_dev(). When bind fails (e.g. ENOMEM), put_dev() fires without a matching get_dev(), leaving the refcount unbalanced. A subsequent close of the ep0 file descriptor calls dev_release() -> put_dev() which hits zero and frees the object; then gadgetfs_kill_sb() calls put_dev(the_device) again on the already-freed pointer, triggering a refcount underflow and use-after-free. Fix by calling get_dev() at the start of gadgetfs_bind(), before any error path that invokes gadgetfs_unbind(), so the reference is always balanced regardless of whether bind succeeds or fails.
-
CVE-2026-99450
In the Linux kernel, the following vulnerability has been resolved: vfio/nvgrace-gpu: Use a core-only open_device in fallback mode When the ACPI memory properties are missing or the C2C link is down, the device is exposed through nvgrace_gpu_pci_core_ops. That table opens through nvgrace_gpu_open_device() but closes through vfio_pci_core_close_device(), so open and close are not symmetric. nvgrace_gpu_open_device() registers nvdev->usemem for poison handling unconditionally. In fallback mode usemem.memlength is zero, and register_pfn_address_space() inserts a node covering [0, ULONG_MAX]. Nothing removes it: a second open fails with -EBUSY, and once the device is unbound the node outlives the nvdev that contains it. The next open after a rebind reports: BUG: KASAN: slab-use-after-free in interval_tree_iter_first+0x209/0x310 Read of size 8 at addr ffff88800894aa30 by task nvgrace_repro/203 Call Trace: <TASK> dump_stack_lvl+0x5d/0x80 print_report+0x153/0x4c6 kasan_report+0xe4/0x1a0 interval_tree_iter_first+0x209/0x310 register_pfn_address_space+0x97/0x100 nvgrace_gpu_open_device+0x21c/0x730 vfio_df_open+0x213/0x520 vfio_df_ioctl_bind_iommufd+0x3e8/0x6e0 vfio_device_fops_unl_ioctl+0x782/0x1570 __x64_sys_ioctl+0x13c/0x1c0 do_syscall_64+0x106/0x630 entry_SYSCALL_64_after_hwframe+0x77/0x7f </TASK> Allocated by task 188: ... _vfio_alloc_device+0x3a/0x500 nvgrace_gpu_probe+0xcb/0x660 local_pci_probe+0xcf/0x170 ... Freed by task 188: ... kfree+0x20d/0x600 device_release+0xcf/0x250 kobject_put+0x152/0x2f0 pci_device_remove+0xb0/0x1c0 device_release_driver_internal+0x38d/0x540 unbind_store+0xde/0x100 ... Give the fallback table its own open_device that only calls vfio_pci_core_enable() and vfio_pci_core_finish_enable(), the counterpart of vfio_pci_core_close_device(). Fallback mode then never touches the nvgrace memory regions, as Alex suggested. This should be safe because the fallback table uses the vfio-pci core read, write and mmap handlers, so nothing there uses the state nvgrace_gpu_open_device() sets up. The one visible difference is that a fallback open no longer fails when BAR0 cannot be mapped, which matches vfio-pci.
-
CVE-2026-99451
In the Linux kernel, the following vulnerability has been resolved: wifi: mt76: mt7996: unlink TWT flow if the MCU rejects the agreement mt7996_mac_add_twt_setup() links a new flow into dev->twt_list before asking the MCU to add the agreement, but exits without unlinking the flow if the request fails. The flow-ID and table masks are only committed after MCU success, so the still-linked flow is invisible to the mask-gated teardown paths. If the station retries the setup, the same slot is cleared and its embedded list node reinitialized while it is still linked, corrupting dev->twt_list. If the station disassociates instead, teardown skips the flow and station storage can be freed while the node remains linked. A later scheduling or debugfs twt_stats walk can then dereference the freed flow, resulting in a use-after-free. Unlink the flow when the MCU rejects the agreement. This restores the list to its pre-request state while leaving the unchanged mask and agreement accounting untouched.
-
CVE-2026-99437
In the Linux kernel, the following vulnerability has been resolved: media: rc: Use after free in ir_raw_event_handle() If rc_unregister_device() is called while IR is being processed, then ir_raw_event_handle() could call wake_up_process(dev->raw->thread) after kthread_stop(dev->raw->thread). Avoid use after free by using get_task_struct().
-
CVE-2026-99439
In the Linux kernel, the following vulnerability has been resolved: media: rc: Ensure registered is cleared in error path If rc_register_device() fails, ensure that registered is not set to true. If lirc_register() succeeded, then userspace could have an open file descriptor open. This leads to a use-after-free. lirc_unregister() must be called with registered = false, else there is a possibility that an fd is open and polling, which would never return POLLERR unless registered is false. Note that registered must be changed under dev->lock, many functions read registered under lock (e.g. lirc_ioctl).
-
CVE-2026-99440
In the Linux kernel, the following vulnerability has been resolved: media: rc: Ensure that rc_unregister_device() does not free input device During or after rc_unregister_device(), IR may still be reported which results in a input event being reported. This could result in a null pointer deref in rc_keydown() or a use-after-free of the input device if the pointer was read before it is set to NULL.
-
CVE-2026-99441
In the Linux kernel, the following vulnerability has been resolved: nvmet-auth: serialize access to DH-HMAC-CHAP state to fix UAF A REPLY AUTH_SEND command can derive a dhchap_skey key while a concurrent AUTH_RECEIVE command observing this REPLY state executes the failure path and frees that key. The REPLY command then continues and writes into freed memory, causing a use-after-free. KASAN reports it as: BUG: KASAN: slab-use-after-free in __sha256_final Write of size 4 at addr ffff88800359c540 ... nvmet_auth_sq_free (drivers/nvme/target/auth.c:237) nvmet_execute_auth_receive (drivers/nvme/target/fabrics-cmd-auth.c:621) Add a mutex to each submission queue to serialize authentication commands and timeout handling. Unlock the mutex before completing requests and ignore outdated timeout callbacks. Limit timeout delays to prevent immediate expiry on 32-bit kernels.
-
CVE-2026-99442
In the Linux kernel, the following vulnerability has been resolved: nvmem: core: Fix OOB read for bit offsets of more than one byte When the bit offset is BITS_PER_BYTE or larger the read position is advanced by `bytes_offset`. This is not taken into account in the per-byte read loop which still reads `cell->bytes` resulting in an out of bounds read of `bytes_offset` bytes. The information read OOB does not leak directly as the erroneously read bits are cleared. Detected by KASAN while looking for a use-after-free in simplefb.c.
-
CVE-2026-99443
In the Linux kernel, the following vulnerability has been resolved: pps: generators: don't use the driver's info after unregister A file that was opened before pps_gen_unregister_source() keeps pps_gen, and with it the pointer to the driver's pps_gen_source_info. PPS_GEN_SETENABLE and PPS_GEN_USESYSTEMCLOCK keep using that pointer after the driver has gone: - pps_gen_tio allocates its info with devm_kzalloc(), so after an unbind both ioctls read freed memory, and SETENABLE calls the driver's enable() with its freed private data. - pps_gen-dummy does not set info->owner, so it can be unloaded while the file is open, and the ioctls then read the unloaded module's data and call its code. TIO can't probe in a VM, as it needs ART. A test driver that registers its info in devm memory like TIO does, unbound while /dev/pps-gen0 is open, gives: BUG: KASAN: slab-use-after-free in pps_gen_cdev_ioctl+0x4c1/0x580 Read of size 1 at addr ffff88800e647f38 by task ppsgen64/145 ... Freed by task 146: kfree+0x25a/0x6d0 release_nodes+0xd1/0x140 devres_release_all+0x10e/0x1a0 device_unbind_cleanup+0x71/0x250 device_release_driver_internal+0x41b/0x570 unbind_store+0xd9/0x100 and unloading pps_gen-dummy while its file is open oopses: BUG: unable to handle page fault for address: ffffffffa0203040 Oops: Oops: 0000 [#1] SMP KASAN NOPTI RIP: 0010:pps_gen_cdev_ioctl+0x372/0x580 Protect info in the ioctl handler with a mutex, and clear it under the mutex on unregister, so that unregister waits for the ioctls using it and later ones fail with -ENODEV. PPS_KC_BIND does the same for a removed PPS device since commit 3649f9a6b897 ("pps: don't allow PPS_KC_BIND on removed devices"). The sysfs attributes need nothing new: they are removed, and their callbacks drained, before info is cleared.
-
CVE-2026-99430
In the Linux kernel, the following vulnerability has been resolved: ocfs2: fix chunk number of the first chunk in a local quota file Mounting a crafted OCFS2 image can corrupt kernel memory. When the local quota file header of such an image claims zero chunks, the first quota entry allocated during the mount gets a chunk number taken from a kernel pointer, and releasing that entry clears a bit far outside the chunk bitmap. The write will corrupt the data stored at that address, and the corruption may lead to a system crash or damage unrelated data. Mounting requires CAP_SYS_ADMIN, so it takes an untrusted image, such as removable media or a loop mount of a file from elsewhere. Since the chunk number comes from a kernel pointer, the address of the bit that gets cleared differs from boot to boot, and the issue has been reported under several titles: - BUG: unable to handle kernel paging request in ocfs2_local_release_dquot - KASAN: use-after-free Write in ocfs2_local_release_dquot - KASAN: slab-out-of-bounds Write in ocfs2_local_release_dquot - KASAN: slab-use-after-free Write in ocfs2_local_release_dquot - KFENCE: use-after-free write in ocfs2_local_release_dquot Note that none of these is a use-after-free in the quota code: the chunk and its buffer head are alive. The bit that gets cleared lies far outside the bitmap. KASAN and KFENCE name each report based on the object occupying that address, which explains why the same issue is reported under so many different titles. In the report I sent, the address fell in a free page, which KASAN labels use-after-free. There are known reproducers. Both a syzkaller and a C reproducer are available from the Google Drive link [1] in my report thread. The issue was found by our modified syzkaller, not a theoretical thing found by an LLM. I have not seen it outside fuzzing, so it is not affecting me in the real world. The local quota file in OCFS2 is divided into chunks, and each chunk begins with a header block holding a bitmap of the quota entries that chunk has handed out. Chunks are numbered from zero, and that number is used to convert the file offset of an entry back into a bit position in the bitmap. ocfs2_local_quota_add_chunk() appends a new chunk to the in-memory list and numbers it one past the chunk that was last: list_add_tail(&chunk->qc_chunk, &oinfo->dqi_chunk); chunk->qc_num = list_entry(chunk->qc_chunk.prev, struct ocfs2_quota_chunk, qc_chunk)->qc_num + 1; The predecessor is looked up after the new chunk is added to the list, so if the list was empty, the prev pointer is the list head itself. The head is the dqi_chunk member of struct ocfs2_mem_dqinfo and is not a chunk, so reading qc_num through it lands 16 bytes past the start of the head, on the dqi_gqinode pointer that follows it. The first chunk of the file is then numbered with the lower half of a kernel pointer instead of 0. The list is empty when the local quota file header claims the file has no chunks. ocfs2_local_read_info() takes dqi_chunks from that header without validating it, so an image with dqi_chunks == 0 takes this path when the first quota entry is allocated. ocfs2_create_local_dquot() turns the bad number into a file offset with ol_dqblk_off(), which shifts a 32-bit block number left by the block size bits, so the top bits of such a large block number are lost. ocfs2_local_release_dquot() turns the offset back into a bit index with ol_dqblk_chunk_off(), using the full chunk number, so the lost bits push that index far outside the bitmap, and clearing it corrupts unrelated memory. Compute the chunk number before putting the chunk on the list, and use 0 when the list is empty.
-
CVE-2026-99431
In the Linux kernel, the following vulnerability has been resolved: ocfs2: cancel quota sync work before freeing quota info OCFS2 quota info owns dqi_sync_work and must outlive both its timer and qsync_work_fn(). ocfs2_local_free_info() frees that owner without cancelling the work, relying on the quota suspend/disable wrappers. Loading quotas on a read-only to read-write remount can successfully queue the work in ocfs2_global_read_info() and then fail while attaching a dquot to an existing inode in add_dquot_ref(). The quota core unwinds that failure directly through dquot_disable(), bypassing the OCFS2 wrapper: Remount task (s_umount write) Timer/workqueue ocfs2_global_read_info() schedules dqi_sync_work add_dquot_ref() returns an error dquot_disable() ocfs2_local_free_info() frees oinfo delayed-work timer expires qsync_work_fn() reads oinfo The timer core can then write to the freed embedded timer, or qsync_work_fn() can read oinfo->dqi_gqinode before its s_umount trylock. The outer ocfs2_disable_quotas() cleanup skips the failed type because dquot_disable() has already cleared its loaded state. Cancel dqi_sync_work synchronously at the start of ocfs2_local_free_info(), before dropping the global inode or freeing any owner state. The work is initialized whenever read_file_info() succeeds. dquot_disable() holds s_umount for write, so the callback skips its quota scan via down_read_trylock() and can finish without waiting for teardown. The synchronous cancel also handles its self-rearm. KASAN report as below: BUG: KASAN: slab-use-after-free in __run_timers+0x9e9/0xa10 Write of size 8 at addr ffff888111b0aad0 by task swapper/3/0 Call Trace: <IRQ> dump_stack_lvl+0x93/0xd0 print_report+0xce/0x630 ? __run_timers+0x9e9/0xa10 ? srso_alias_return_thunk+0x5/0xfbef5 ? __virt_addr_valid+0x20e/0x420 ? __run_timers+0x9e9/0xa10 kasan_report+0xe0/0x110 ? __run_timers+0x9e9/0xa10 __run_timers+0x9e9/0xa10 ? __pfx___run_timers+0x10/0x10 ? __lock_acquire+0x5a3/0x2140 ? __pfx_do_raw_spin_lock+0x10/0x10 ? srso_alias_return_thunk+0x5/0xfbef5 timer_expire_remote+0xfa/0x160 ? __pfx_timer_expire_remote+0x10/0x10 ? srso_alias_return_thunk+0x5/0xfbef5 ? srso_alias_return_thunk+0x5/0xfbef5 ? trace_hardirqs_on+0x18/0x160 tmigr_handle_remote_up+0x4ba/0x9d0 ? __pfx_tmigr_handle_remote_up+0x10/0x10 ? tmigr_handle_remote+0x21a/0x340 ? srso_alias_return_thunk+0x5/0xfbef5 ? lock_release+0xc8/0x290 ? __pfx_tmigr_handle_remote_up+0x10/0x10 __walk_groups_from+0x5a/0x160 tmigr_handle_remote+0x24f/0x340 ? __pfx_tmigr_handle_remote+0x10/0x10 ? lock_acquire+0x190/0x300 ? srso_alias_return_thunk+0x5/0xfbef5 ? lock_release+0xc8/0x290 run_timer_softirq+0x222/0x270 ? __pfx_run_timer_softirq+0x10/0x10 ? srso_alias_return_thunk+0x5/0xfbef5 ? srso_alias_return_thunk+0x5/0xfbef5 ? lockdep_hardirqs_on_prepare+0xb8/0x1a0 ? srso_alias_return_thunk+0x5/0xfbef5 handle_softirqs+0x1fa/0x8d0 ? __pfx_handle_softirqs+0x10/0x10 ? srso_alias_return_thunk+0x5/0xfbef5 ? _raw_spin_unlock+0x23/0x40 ? srso_alias_return_thunk+0x5/0xfbef5 ? __hrtimer_rearm_deferred+0x1fb/0x5d0 __irq_exit_rcu+0x113/0x1e0 irq_exit_rcu+0xe/0x20 sysvec_apic_timer_interrupt+0x6c/0x80 </IRQ> <TASK> asm_sysvec_apic_timer_interrupt+0x1a/0x20 ? ct_kernel_exit.constprop.0+0xc5/0xf0 ? do_idle+0x3ae/0x650 default_idle+0x9/0x10 default_idle_call+0x6c/0xb0 do_idle+0x3ae/0x650 ? cpuhp_online_idle+0x9c/0xb0 ? __pfx_do_idle+0x10/0x10 ? __entry_text_end+0x1020f6/0x1020f9 ? lockdep_hardirqs_on_prepare+0xeb/0x1a0 ? srso_alias_return_thunk+0x5/0xfbef5 ? trace_hardirqs_on+0x18/0x160 ? srso_a ---truncated---
-
CVE-2026-99432
In the Linux kernel, the following vulnerability has been resolved: ocfs2/dlm: hold the lookup reference across convert AST enqueue dlm_convert_lock_handler() drops the reference it took when it found the lock on res->granted at its leave label, one statement before it calls dlm_queue_ast(). dlm_free_dead_locks() takes res->spinlock and drops the list and base references, so a peer death landing in that gap leaves the handler's own put as the one that frees the lock, and __dlm_queue_ast() then reads lock->lockres from freed memory. The handler is registered next to dlm_create_lock_handler(), so a convert message buffered on a dying peer's socket arrives on the same o2net receive work that o2net_disconnect_node() does not wait for. The two can interleave as follows, with the reference counts as they are before this fix: Socket data-ready Node-down thread Receive worker ----------------- ---------------- -------------- o2net_data_ready() select sc under callback read lock o2net_disconnect_node() queue shutdown begin flush; new color queue sc_rx_work in the new color release callback lock shutdown socket flush returns read buffered convert_lock dlm_convert_lock_handler() find lock on res->granted dlm_lock_get() refs = 3 lock res->spinlock __dlmconvert_master() call_ast = 1 unlock res->spinlock DLM node-down cleanup lock res->spinlock dlm_free_dead_locks() unlink lock put list ref refs = 2 put base ref refs = 1 unlock res->spinlock dlm_lock_put() refs = 0 free lock dlm_queue_ast() __dlm_queue_ast() read lock->lockres (use-after-free) __dlmconvert_master() only moves the lock between lists, so the count entering the gap is base, list and lookup. Drop the lookup reference after the AST is queued instead of before it and the race ends with the one __dlm_queue_ast() took, which dlm_flush_asts() releases later with ast_pending cleared and ast_list unlinked. No extra get is needed, and the normal case ends with the same count as before. dlmconvert_master() has the same shape but is only reached from dlmlock() for a lock this node owns, which dlm_free_dead_locks() never matches.
-
CVE-2026-99435
In the Linux kernel, the following vulnerability has been resolved: media: rc: Fix use after free in bpf progs Since commit dccc0c3ddf8f ("media: rc: fix race between unregister and urb/irq callbacks"), rcdev->raw is no longer set to NULL after device unregister. raw->progs could point to stale data.
-
CVE-2026-99422
In the Linux kernel, the following vulnerability has been resolved: ovl: use the pinned upper dir in ovl_fix_origin() Renames made directly on the upper filesystem during a lookup crash the kernel. A dentry that is passed to vfs_setxattr() must be pinned. ovl_fix_origin() passes upper->d_parent to ovl_set_impure() without a reference, and the lookup holds no lock of the upper directory. A rename on the upper filesystem changes d_parent, and only the child pins the new parent, so a second rename and an rmdir free it. With delays in ovl_fix_origin() and ovl_set_impure(), renaming the upper directory twice and removing the intermediate parent during the lookup of a merge directory without an origin xattr gave a KASAN use-after-free in vfs_setxattr(). Fix this by passing the upper dentry that the overlay parent holds. [amir: add helper ovl_set_impure_dir()]
-
CVE-2026-99423
In the Linux kernel, the following vulnerability has been resolved: ovl: take a snapshot of the name for copy up The copy up of a directory reads a freed dentry name when a lookup moves the directory at the same time. A pointer to the name of a dentry is only valid while something stops a move of the dentry. ovl_do_copy_up() keeps a plain copy of c->dentry->d_name in c->destname, and ovl_copy_up_start() only keeps ovl_rename() away. When a lookup finds the same directory under another name after a change to the lower layer, d_splice_alias() moves the dentry and frees its external name. With a delay in ovl_do_copy_up(), a chmod of a lower-only directory with a 60 byte name raced with a rename of that directory in the lower layer and a stat of the new name, and KASAN reported a use-after-free in lookup_noperm_common(). Fix this by using take_dentry_name_snapshot() for c->destname.
-
CVE-2026-99424
In the Linux kernel, the following vulnerability has been resolved: ovl: set OVL_WHITEOUTS before publishing the upper dir Listing a directory while it is copied up causes a use-after-free of the readdir cache. A merge directory must never be iterated as a real upper directory: that mode frees the cache in OVL_I(inode)->cache without looking at its refcount, while files opened in merged mode still use it. ovl_copy_up_workdir() publishes the upper dentry before it sets OVL_WHITEOUTS, and ovl_dir_open() samples both without a lock, so an open between the two stores finds the upper directory and takes it for a real one. With a delay between the two stores, an unprivileged process that listed a directory during its copy-up saw only "." and "..", and its next getdents freed the readdir cache of another descriptor. Fix this by setting OVL_WHITEOUTS before ovl_inode_update(), so that ovl_dir_reset() agrees with ovl_dir_open(), and by testing OVL_TYPE_MERGE() of the path type in ovl_dir_open(), so that an open that observes the upper dentry never iterates a merge dir as a real dir, regardless of the order in which it observes the flag. [amir: test OVL_TYPE_MERGE() instead of smp_rmb() in ovl_dir_open()]
-
CVE-2026-99425
In the Linux kernel, the following vulnerability has been resolved: ovl: hold a reference on the parent in readdir Listing a directory on an overlay uses a freed dentry and crashes the kernel when a lookup moves the directory at the same time. Code that can sleep may only use d_parent without a reference while something stops a move of the dentry. ovl_iterate_real(), ovl_need_adjust_d_ino() and ovl_cache_update() use the d_parent of the listed directory and hold only the lock of that directory. When a lookup finds the directory under another parent after a change to a layer, d_splice_alias() moves it without that lock and drops its reference on the old parent. With a delay in ovl_iterate_real(), getdents64 on an overlay directory raced a rename in the lower layer and a lookup that moved its dentry, and after an rmdir of the old parent KASAN reported a use-after-free in security_inode_getattr(). Fix this by using dget_parent() in the three functions. [amir: fix comments in ovl_need_adjust_d_ino()]
-
CVE-2026-99426
In the Linux kernel, the following vulnerability has been resolved: smb: client: hold server info until cifsd exits The last session reference can be dropped while the cifsd demultiplex thread is still running clean_demultiplex_info(). That thread frees TCP_Server_Info there, racing with cifs_put_tcp_session() as it releases the server crypto state. syzbot reported a KASAN use-after-free in cifs_crypto_secmech_release(). Keep an object reference for the cifsd thread. Defer final crypto, session-key, network namespace, string, and structure cleanup until both the last session reference and the thread reference are dropped. Drop the thread reference after clean_demultiplex_info() has finished using the server. Build-tested with ARCH=arm64 and W=1 for fs/smb/client/connect.o. The KASAN reproducer was not run on the patched kernel.
-
CVE-2026-99427
In the Linux kernel, the following vulnerability has been resolved: smb: client: fix OOB last_entry pointer from unbounded LastNameOffset CIFSFindFirst() and CIFSFindNext() locate the last entry of a TRANS2_FIND_FIRST2 or TRANS2_FIND_NEXT2 response by adding the LastNameOffset of the response parameters to the start of the data area, and store the result in psrch_inf->last_entry. MS-CIFS 2.2.6.2.2 defines that offset as the offset of the entry's file name; the client follows the reading of NT and Samba servers, which point it at the entry. LastNameOffset is only checked against CIFSMaxBufSize, which says nothing about the response actually received: validate_t2() lets the data area start up to 1024 bytes into the buffer, and SendReceive() copies at most CIFSMaxBufSize + MAX_CIFS_HDR_SIZE bytes of a response, 16468 by default, into a cifs_request allocation of CIFSMaxBufSize + MAX_SMB2_HDR_SIZE bytes, 16588 by default. The pointer itself can therefore be placed anywhere from the end of the received data to 820 bytes past the end of the allocation. The read through that pointer is far larger than its displacement. find_cifs_entry() hands last_entry to cifs_save_resume_key() after every FindNext and after the FindFirst of a search rewind, and cifs_fill_dirent() takes the length of the name from the entry itself: a __le32 FileNameLength for most levels, a __u8 for SMB_FIND_FILE_INFO_STANDARD, and for SMB_FIND_FILE_UNIX the result of a scan for the terminating NUL over up to PATH_MAX + 1 UTF-16 units. The name is recorded as the resume name of the search, and the next CIFSFindNext() rejects only a recorded length of PATH_MAX or more before copying the name into the ResumeFileName of its request and sending it with the resume key taken from the same entry: up to 4 KiB of memory from past the end of the allocation leaks to the server, and for SMB_FIND_FILE_UNIX the NUL scan runs about twice as far before the length is even known. The entry does not have to lie outside the response for that either: a maximal response ends about 120 bytes before the end of the allocation, so an entry in its last bytes with a name just under PATH_MAX has nearly all of that name copied from past the allocation. It needs nothing unusual from the client: a FindNext response that reports no entries and does not end the search makes find_cifs_entry() send the next FindNext at once, carrying whatever was parsed at LastNameOffset, and a search rewind can do the same after its FindFirst. SMB1 is not negotiated by default: this takes a malicious or compromised server, a mount of it with an explicit vers=1.0, which CONFIG_CIFS_ALLOW_INSECURE_LEGACY (default y) permits, and a directory listing on that mount. With KASAN enabled, a response that places the last entry past the buffer, with a large DataOffset and a LastNameOffset near the end of its range, gives: ================================================================== BUG: KASAN: slab-out-of-bounds in cifs_save_resume_key.isra.0+0x7aa/0x7f0 Read of size 4 at addr ffff88807b55c33b by task ls/83 CPU: 1 UID: 0 PID: 83 Comm: ls Not tainted 7.3.0-rc4-00457-gf14572c203d5 #69 PREEMPT(lazy) Call Trace: <TASK> kasan_report+0xdf/0x1a0 cifs_save_resume_key.isra.0+0x7aa/0x7f0 cifs_readdir+0x1ed6/0x29c0 iterate_dir+0x1c0/0x570 __x64_sys_getdents64+0x133/0x270 do_syscall_64+0x109/0x5d0 entry_SYSCALL_64_after_hwframe+0x77/0x7f </TASK> Allocated by task 83 on cpu 0 at 19.317556s: cifs_buf_get+0x36/0x90 smb_init+0x4f/0x110 CIFSFindNext+0xf7/0x14c0 cifs_readdir+0xf51/0x29c0 The buggy address belongs to the object at ffff88807b558000 which belongs to the cache cifs_request of size 16588 The buggy address is located 623 bytes to the right of allocated 16588-byte region [ffff88807b558000, ffff88807b55c0cc) ================================================================== The four bytes read there are the FileNameLength of the entry, which cifs_fill_dirent() tak ---truncated---
-
CVE-2026-99418
In the Linux kernel, the following vulnerability has been resolved: ovl: do not copy up an xattr value that was never read Copy-up stores uninitialized kernel memory in an xattr of the upper file, where the user can read it. ovl_copy_xattr() may only pass bytes to ovl_do_setxattr() that ovl_do_getxattr() has read. When the read fails with -ERANGE the function asks for the size again, and if the new size fits the buffer it stores the buffer although nothing was read into it. The buffer comes from kvmalloc(), or it still holds the value of the previous xattr. A FUSE lower layer that answers every sized GETXATTR with -ERANGE made copy-up store 4096 bytes that the server never sent, with a kernel pointer in them, as an xattr of the upper file. Fix this by failing the copy-up with -EIO when the size that follows -ERANGE fits the buffer.
-
CVE-2026-99413
In the Linux kernel, the following vulnerability has been resolved: xfs: fix buffer overruns in xfs_ioc_attr_list When we converted the al_offset array in struct xfs_attrlist into a VLA, the size of the object shrank by 4 bytes. Unfortunately, the buffer size validation in the attrlist ioctl wasn't updated to notice this, so the al_offset[0] assignment blindly writes off the end of the buffer. LOLLM noticed the omitted check and complained.
-
CVE-2026-99396
In the Linux kernel, the following vulnerability has been resolved: xfs: bound da-node entry count against the correct geometry xfs_da3_node_verify() bounds the node entry count against the larger of the directory and attribute geometries because, as a buffer verifier, it cannot tell whether the block belongs to the directory or the attribute tree. When the directory block size exceeds the fs block size (e.g. mkfs.xfs -n size=64k -b size=4k), an attribute node buffer is a single fs block that holds only m_attr_geo->node_ents entries, yet a crafted attr node may claim a count up to m_dir_geo->node_ents and still pass the verifier. xfs_da3_node_lookup_int() then indexes btree[] up to that count during its binary search -- an out-of-bounds read via getxattr/listxattr on a mounted crafted image. The buffer verifier is the wrong place to tighten this: it has no fork context, and the transaction-less read path used by getxattr does not run xfs_da3_node_set_type() either. xfs_da3_node_lookup_int(), on the other hand, always runs on that path and holds args->geo, the geometry of the fork actually being searched. Bound the entry count against args->geo->node_ents there, before walking the entries.
-
CVE-2026-99399
In the Linux kernel, the following vulnerability has been resolved: xfs: reject remote xattr entries with an out-of-range value length xfs_attr3_leaf_verify_entry() validates a remote attribute entry's name but never bounds its on-disk value length (xfs_attr_leaf_name_remote. valuelen, a __be32). A crafted leaf with valuelen = 0x80000000 passes the verifier and the CRC. That length is later assigned into the signed int args->rmtvaluelen (xfs_attr3_leaf_getvalue), becoming negative, which slips past the signed -ERANGE check in xfs_attr_copy_value(); a getxattr() with a small buffer then memcpy()s a full remote block into the small kvalue buffer (xfs_attr_rmtval_copyout) -- a heap out-of-bounds write with attacker-controlled content, from an unprivileged getxattr(2) on a mounted crafted image. Reject remote entries whose value length exceeds XFS_XATTR_SIZE_MAX in the leaf verifier, so the malicious block is rejected at read time.
-
CVE-2026-99400
In the Linux kernel, the following vulnerability has been resolved: xfs: reject out-of-range attribute value lengths in xfs_attr_copy_value xfs_attr_copy_value() takes the value length as a signed int and, for a remote xattr, is handed args->rmtvaluelen. That field is filled from the on-disk __be32 xfs_attr_leaf_name_remote.valuelen in xfs_attr3_leaf_getvalue(), so a crafted length such as 0x80000000 is stored into the signed rmtvaluelen as a negative number. The "buffer too small" guard in xfs_attr_copy_value() is a signed comparison: if (args->valuelen < valuelen) return -ERANGE; A negative valuelen therefore compares as smaller than the caller's buffer size, skips the -ERANGE path, and is then used as a copy length, leading to an out-of-bounds copy of a full remote block into a small getxattr(2) buffer on a mounted crafted image. Reject a value length that is negative or larger than the maximum xattr size before it is used, so a bogus on-disk length can no longer slip through the value copier.
-
CVE-2026-99388
In the Linux kernel, the following vulnerability has been resolved: smb: smbdirect: fix buffer length underflow in RDMA transfers When a remote buffer descriptor is larger than the remaining local buffer, smbdirect_connection_rdma_xmit() clamps its length and sets buf_len to zero. The common accounting code then subtracts the clamped length from buf_len, causing it to underflow. If another descriptor follows, the loop processes it with a wrapped remaining length and includes memory beyond the local buffer in the RDMA transfer. A client can trigger an out-of-bounds kernel write with an SMB WRITE request, or an out-of-bounds kernel read with an SMB READ request, by supplying multiple RDMA buffer descriptors with an oversized descriptor before the last one. Remove the premature assignment to buf_len. Subtracting the clamped length already consumes the remaining buffer, so the next iteration stops and only descriptors within the local buffer are used to build work requests.
-
CVE-2026-99389
In the Linux kernel, the following vulnerability has been resolved: smb: server: size directory name conversion buffer correctly A 2048-byte ASCII directory name needs 4098 bytes when converted to UTF-16, including the terminator. ksmbd_convert_dir_info_name() caps its allocation at PATH_MAX and writes past the end of that buffer. Allocate space for one UTF-16 code unit per input byte plus the terminator. Use kmalloc_array() to check the size calculation.
-
CVE-2026-99393
In the Linux kernel, the following vulnerability has been resolved: xfs: don't overrun incore inode when cleaning attr forks LOLLM noticed that xrep_dinode_check_afork doesn't check for a di_forkoff value that is so large that the attr fork would begin after the end of the inode space. This causes afork_{ptr,size} to be nonsense. What we really ought to do here is zap the attr fork.
-
CVE-2026-99395
In the Linux kernel, the following vulnerability has been resolved: xfs: bound inode fork length against the fork size during log recovery xlog_recover_inode_commit_pass2() copies an inode log item's data/attr fork region into the inode buffer using the on-log region length without bounding it against the fork capacity, e.g.: len = item->ri_buf[2].iov_len; memcpy(XFS_DFORK_DPTR(dip), src, len); The only guard is an ASSERT, which is a no-op on production kernels (CONFIG_XFS_DEBUG off), and xfs_dinode_verify() runs only after the copy. A crafted image with a dirty log can therefore drive a heap out-of-bounds write at mount time. The XFS_ILOG_DBROOT sibling already passes XFS_DFORK_DSIZE as a bound; the DDATA/DEXT and ADATA/AEXT memcpy paths did not. Bound each logged fork region against the destination fork size before copying it, and reject the log item with -EFSCORRUPTED when it does not fit. Because the recovered inode is only verified after the fork data has been copied in, the checks are done up front, before any memcpy into the on-disk inode.
-
CVE-2026-99381
In the Linux kernel, the following vulnerability has been resolved: smb: client: reject short or out-of-range FIND response parameters CIFSFindFirst() and CIFSFindNext() read the SearchHandle, SearchCount, EndofSearch and LastNameOffset of a TRANS2_FIND_FIRST2 or TRANS2_FIND_NEXT2 response at ParameterOffset. validate_t2() bounds the offset at 1024 and ParameterCount below 512, and nothing compares them with the number of bytes SendReceive() copied into the buffer or with the size of the response parameters. smb_init() places the request and the response in the same cifs_request allocation and cifs_buf_get() clears only the start of it, so a short response with a ParameterOffset past its end has the search handle, the entry count, the end-of-search flag and the last-entry offset read out of the request that was just sent, or out of whatever the allocation held before, and a ParameterCount smaller than the parameters has them read past the end of the parameter block. The reads stay inside the allocation, but what they return is not part of the response, and CIFSFindFirst() then hands back a search handle the server never sent. That handle is sent back to the server in the next FIND_NEXT2 or FIND_CLOSE2, so two bytes of the buffer leak at an offset the server picks. SMB1 is not negotiated by default; reaching this code requires an explicit vers=1.0 mount. Reject a response whose parameter area is smaller than the response parameters or extends past the received length, returning smb_EIO2() with a new smb_eio_traces value for each of the two responses, like the other response checks this file gained since v6.19. Backports to trees before v6.19 need smb_EIO2() replaced with -EINVAL and the trace.h hunk dropped. A conforming server is not affected. The parameter block of a transaction response is part of the SMB, so ParameterOffset + ParameterCount cannot exceed the smbCalcSize() bytes that SendReceive() copies, and the response parameters are 10 bytes for FIND_FIRST2 and 8 bytes for FIND_NEXT2, which is what T2_FFIRST_RSP_PARMS and T2_FNEXT_RSP_PARMS hold. This patch depends on "smb: client: reject FIND data areas that run past the received response", whose declarations it extends.
-
CVE-2026-99382
In the Linux kernel, the following vulnerability has been resolved: smb: client: reject FIND data areas that run past the received response CIFSFindFirst() and CIFSFindNext() take the directory entries of a TRANS2_FIND_FIRST2 or TRANS2_FIND_NEXT2 response from DataOffset for DataCount bytes. validate_t2() bounds DataOffset at 1024 and ParameterCount + DataCount by the byte count, and nothing compares DataOffset + DataCount with the number of bytes SendReceive() copied into the buffer. smb_init() places the request and the response in the same cifs_request allocation and cifs_buf_get() clears only the start of it, so a short response whose data area extends past its end has srch_entries_start point at bytes of the request that was just sent, or at whatever the allocation held before, and those bytes are parsed as directory entries. cifs_readdir() and find_cifs_entry() walk the entries with nxt_dir_entry(), which checks every entry but the first of a response against the end of the SMB; cifs_query_path_info() and cifs_backup_query_path_info() read the first entry with no check at all. Those reads stay inside the allocation, but what they return is not part of the response. The last entry, which LastNameOffset locates relative to the data area, is bounded against the received response by an earlier patch of this series, as is every entry that cifs_fill_dirent() parses; the walk itself and the two query functions are not. SMB1 is not negotiated by default; reaching this code requires an explicit vers=1.0 mount. Reject a response whose data area extends past the received length, with the -EINVAL that validate_t2() returns for the other malformed transaction responses. cifs_query_path_info() and cifs_backup_query_path_info() still read a whole entry from a data area that may be smaller than one; that is not changed here. A conforming server is not affected. The data block of a transaction response is part of the SMB, so DataOffset + DataCount cannot exceed the smbCalcSize() bytes that SendReceive() copies; coalesce_t2() appends each secondary response to the data area and adds its length to both DataCount and the byte count, which keeps that relation.
-
CVE-2026-99384
In the Linux kernel, the following vulnerability has been resolved: smb: client: fix OOB read of the resume name sent in TRANS2_FIND_NEXT2 cifs_fill_dirent() reads the fixed part of a directory entry and the name length it holds, or for SMB_FIND_FILE_UNIX scans the name for its terminating NUL, without knowing where the response ends. cifs_save_resume_key() calls it for the entry at the LastNameOffset the server chose, after every FindNext and after the FindFirst of a search rewind, and records the name it returns as the resume name of the search without bounding it; the next CIFSFindNext() copies up to PATH_MAX - 1 bytes from that name into its request and sends them to the server. cifs_filldir() bounds the name against the end of the response since commit f8cf09a53a0d ("smb: client: bound dirent name against end of SMB response in cifs_filldir"), but only after cifs_fill_dirent() has read the entry, and nxt_dir_entry() checks the entries it walks only against sizeof(FILE_DIRECTORY_INFO), for every level but SMB_FIND_FILE_INFO_STANDARD, and never sees the first entry of a response. Bounding where last_entry starts, as the preceding patch does, does not bound the read. A last entry that starts within the final bytes of the response still has its fixed part read past the end of the received data, and the name length it declares, a __le32 for every level but SMB_FIND_FILE_INFO_STANDARD and SMB_FIND_FILE_UNIX, is recorded without being checked against anything. CIFSFindNext() rejects only a length of PATH_MAX or more, then copies the name, up to 4095 bytes, into its request and sends it, and as the report below shows that copy can run past the end of the cifs_request allocation that holds the response. For SMB_FIND_FILE_UNIX the name is in addition scanned for its terminating NUL over up to PATH_MAX + 1 UTF-16 units, or PATH_MAX bytes, before any check. The FindNext that carries the name is only sent by the SMB1 dialect, which is not negotiated by default; reaching it requires an explicit vers=1.0 mount. With KASAN enabled, a response whose last entry declares a name longer than the bytes that follow it gives: ================================================================== BUG: KASAN: slab-out-of-bounds in CIFSFindNext+0x8ca/0x14c0 Read of size 4080 at addr ffff88807c823f98 by task ls/83 CPU: 0 UID: 0 PID: 83 Comm: ls Not tainted 7.3.0-rc4-00457-gf14572c203d5 #69 PREEMPT(lazy) Call Trace: <TASK> kasan_report+0xdf/0x1a0 kasan_check_range+0x10f/0x1e0 __asan_memcpy+0x23/0x60 CIFSFindNext+0x8ca/0x14c0 cifs_readdir+0xf51/0x29c0 iterate_dir+0x1c0/0x570 __x64_sys_getdents64+0x133/0x270 do_syscall_64+0x109/0x5d0 entry_SYSCALL_64_after_hwframe+0x77/0x7f </TASK> Allocated by task 83 on cpu 1 at 13.497119s: cifs_buf_get+0x36/0x90 smb_init+0x4f/0x110 CIFSFindNext+0xf7/0x14c0 cifs_readdir+0xf51/0x29c0 The buggy address belongs to the object at ffff88807c820000 which belongs to the cache cifs_request of size 16588 The buggy address is located 16280 bytes inside of allocated 16588-byte region [ffff88807c820000, ffff88807c8240cc) ================================================================== The read is the memcpy() that copies the recorded name into the next request, so those bytes go to the server. Pass the end of the response to cifs_fill_dirent() and have it reject an entry whose fixed part does not fit or whose name extends past the end, before either is used, so that both callers are covered; cifs_save_resume_key() computes the end the same way cifs_readdir() does. The name bound compares the bytes left between the name and the end with the name length rather than adding the length to the name pointer, since the length is a 32-bit value taken from the entry and the sum could wrap on 32-bit systems. cifs_fill_dirent() is also used on the SMB2 readdir path; there smb2_parse_query_directory() has already checked every entry against the end of the response, so a vali ---truncated---
-
CVE-2026-99385
In the Linux kernel, the following vulnerability has been resolved: smb: client: fix buffer overflow in SMB2_tcon() In SMB2_tcon(), unc_path is allocated with a fixed buffer size of MAX_SHARENAME_LENGTH * 2 bytes (672 bytes). However, the user-controlled UNC path (tree) length is passed to cifs_strtoUTF16(), which assumes the destination buffer is large enough and writes strlen(tree) + 1 UTF-16 code units. Commit 62a1a439e0fd ("cifs: clean up handling of unc= option") removed the length check on UNC mount options, allowing mount paths to exceed MAX_SHARENAME_LENGTH while retaining the fixed-size buffer allocation in SMB2_tcon(). Consequently, an oversized UNC string causes cifs_strtoUTF16() to write past the end of the heap object, resulting in a slab-out-of-bounds write. Furthermore, if a tree exceeds MAX_TREE_SIZE characters, strscpy() truncates it in cifs_tcon::tree_name, causing subsequent reconnects or failovers to fail. Fix this by first verifying that the length of tree does not exceed MAX_TREE_SIZE, returning -EINVAL if it does. Then, dynamically allocate unc_path based on the length of tree so that it safely accommodates all UTF-16 code units for paths within MAX_TREE_SIZE. [pc: remove unused MAX_SHARENAME_LENGTH]
-
CVE-2026-99386
In the Linux kernel, the following vulnerability has been resolved: smb: client: don't resurrect an exiting tcon from cached dir workers cfids_laundromat_worker() bumps tcon->tc_count for every expired open cfid and queues cached_dir_offload_close() without checking whether the tcon is already being torn down. Racing with the final cifs_put_tcon(), the tcon comes back from zero and gets torn down twice: T1 T2 cifs_put_tcon() --tc_count == 0, TID_EXITING tconInfoFree() free_cached_dirs() cancel_delayed_work_sync() cfids_laundromat_worker() ++tc_count == 1 queue_work(serverclose_wq) kfree(tcon->origin_fullpath) kfree(tcon) cached_dir_offload_close() close_cached_dir() [UAF] cifs_put_tcon() tconInfoFree() [double free] cached_dir_lease_break() also bumps tc_count without tc_lock, racing with the laundromat's locked update. Take the tcon reference through a helper that refuses once the tcon is TID_EXITING or its refcount has dropped to zero. Queued put_work and close_work then always hold a tcon reference, so tconInfoFree() can't run before they finish and free_cached_dirs() doesn't need to flush any workqueue. If the laundromat can't get a reference, put the cfid back on the dying list for free_cached_dirs() to free; its handle goes away with the tree connection. The lease break path finds the tcon on ses->tcon_list under cifs_tcp_ses_lock, so it can't see TID_EXITING, but it now takes tc_lock.
-
CVE-2026-99387
In the Linux kernel, the following vulnerability has been resolved: smb: client: bound POSIX lock data area by response length CIFSSMBPosixLock() receives a TRANS2 response whose server-provided DataOffset and DataCount are not required by validate_t2() to describe an interval inside the response. A server can therefore make the client read a cifs_posix_lock beyond the response buffer and return those kernel bytes through F_GETLK. Use smbCalcSize() to obtain the full response length after any multipart TRANS2 reassembly, then reject data areas that extend past it before dereferencing them.
-
CVE-2026-99377
In the Linux kernel, the following vulnerability has been resolved: NFSv4/pnfs: bound the CB_NOTIFY_DEVICEID array count before allocating decode_devicenotify_args() hands the server's notification count straight to kmalloc_objs() as the array length, without checking it against the message that carried it. Bound it the way nfs4xdr.c bounds an attribute length, against xdr_stream_remaining().
-
CVE-2026-99379
In the Linux kernel, the following vulnerability has been resolved: smb: client: validate SMB1 EA list bounds before parsing CIFSSMBQAllEAs() checks the server-provided EA list length by adding it to the response pointer. On 32-bit systems, a positive length can wrap the pointer and pass the comparison. The parser then walks outside the received response while reading the next FEA header. A response with a list length of 0x40000000 reproduces this as a native i386 page fault and kernel panic from listxattr() against an SMB1 server. Check that the EA list starts within the SMB byte area before reading its header. Keep its length unsigned and compare it with the bytes left in the response without pointer addition. SMB1 is not negotiated by default and requires an explicit vers=1.0 mount.
-
CVE-2026-99364
In the Linux kernel, the following vulnerability has been resolved: NFSD: Fix out-of-bounds read in the rpc_status dump nfsd_nl_rpc_status_get_dumpit() loads args->opcnt and args->ops separately, then walks ops[] before rechecking rq_status_counter. A COMPOUND that completes between the two loads runs nfsd4_release_compoundargs(), which zeroes opcnt and points ops back at the eight-entry inline array. A dump that already sampled an opcnt of 200 clamps it to the sixteen slots in rq_opnum, then indexes iops[0..15]. iops is the last member of struct nfsd4_compoundargs and rq_argp is allocated at exactly that size, so the walk runs off the end of the allocation. The trailing recheck discards the sampled data, but the read has already happened. NFSD_CMD_RPC_STATUS_GET carries no GENL_ADMIN_PERM, so an unprivileged local user can repeat the dump against a busy server until it lands in the window. Sample opcnt and ops into locals, then finish the counter recheck before dereferencing ops. An unchanged counter means both came from the same COMPOUND, where opcnt cannot exceed what ops holds.
-
CVE-2026-99367
In the Linux kernel, the following vulnerability has been resolved: NFSD: Don't complete a cld upcall the daemon has not read cld_pipe_downcall() matches a reply to its upcall on the xid alone. A write that arrives before the daemon has read that upcall completes __cld_pipe_upcall() while its struct rpc_pipe_msg is still queued on the pipe. The waiter returns, its stack frame goes away, and the pipe keeps a list_head that points into it. The next rpc_queue_upcall() walks that pointer: BUG: KASAN: vmalloc-out-of-bounds in __list_add_valid_or_report Read of size 8 at addr ffffc90007027950 by task syz.0.96/13066 __list_add_valid_or_report+0x6a/0x130 rpc_queue_upcall cld_pipe_upcall nfsd4_cld_grace_start nfsd4_cld_tracking_init The pipe is mode 0600 in rpc_pipefs, so the trigger is a broken or hostile nfsdcld rather than an unprivileged task. Record whether the daemon has consumed the whole message and complete only when it has. cld_pipe_destroy_msg() runs on every path by which the pipe releases a message, so it can maintain that flag under cn_lock. nfsdcld sends its -EINPROGRESS downcalls only after reading the upcall, so the new check does not break the GraceStart record stream.
-
CVE-2026-99362
In the Linux kernel, the following vulnerability has been resolved: nfsd: preflight SEQUENCE replies before accepting a slot nfsd4_sequence() narrows the reply buffer to the session's cached reply limit before it accepts the slot seqid. A client may negotiate ca_maxresponsesize_cached down to NFSD_MIN_HDR_SEQ_SZ, and nfsd4_alloc_slot() then gives every slot a zero-length sl_data[]. A COMPOUND tag can fill that narrowed buffer until it holds the SEQUENCE opcode but not the status word that follows. nfsd4_encode_operation() returns without running nfsd4_encode_sequence(), so cstate.data_offset stays zero. It leaves op->status at nfs_ok as well, so the COMPOUND is treated as having succeeded. nfsd4_store_cache_entry() declines to cache a lone SEQUENCE that returned an error. That test reads the status the operation reported, so it passes here. The copy starts at offset zero and takes the whole reply, RPC and COMPOUND headers included, into the zero-length sl_data[]. The COMPOUND tag is copied along with it, so the client picks most of the bytes written past the end of the slot: BUG: KASAN: slab-out-of-bounds in read_bytes_from_xdr_buf+0x1bc/0x390 Write of size 80 at addr ffff888003a549cd by task kunit_try_catch/24 __asan_memcpy+0x38/0x60 read_bytes_from_xdr_buf+0x1bc/0x390 nfsd4_sequence_done+0x5b0/0x810 nfs4svc_encode_compoundres+0x1bf/0x240 Check that the fixed-size SEQUENCE result, plus room for a following operation's error status, fits the negotiated limit before narrowing the buffer and consuming the slot seqid. The slot and its reply cache are left unchanged, as RFC 8881 Section 2.10.6.1.2 requires of an error returned from SEQUENCE.
-
CVE-2026-99355
In the Linux kernel, the following vulnerability has been resolved: SUNRPC: fix oversized GSS proxy token copy gss_read_proxy_verf() allocates one page per token chunk, but it copies the entire linear head of an RPC request into in_token->pages[0]. A UDP datagram can leave more than PAGE_SIZE bytes in the linear head, so a large RPCSEC_GSS INIT token overwrites following pages. Copy the linear head in page-sized chunks before switching to the request page array. The linear head itself can span multiple pages.
-
CVE-2026-99340
In the Linux kernel, the following vulnerability has been resolved: KVM: x86: Take SRCU in kvm_zap_gfn_range() kvm_zap_gfn_range() walks memslots and rmaps and may drop mmu_lock to reschedule. Callers that do not already hold kvm->srcu (or slots_lock) can race with memslot deletion: synchronize_srcu_expedited() does not wait, kvm_free_memslot() frees the old slot and its rmap, and the zap resumes on freed memory. The VFIO noncoherent-DMA path hits this by zapping the entire GPA space without SRCU. KASAN reported: BUG: KASAN: vmalloc-out-of-bounds in slot_rmap_walk_next+0x82/0x1c0 Read of size 8 at addr ffffc900005c1008 Call Trace: slot_rmap_walk_next+0x82/0x1c0 __kvm_rmap_zap_gfn_range+0x17a/0x280 kvm_zap_gfn_range+0x2a6/0x6a0 kvm_vfio_set_attr+0x576/0x770 kvm_device_ioctl+0x1ff/0x3b0 Take SRCU inside kvm_zap_gfn_range() so every caller is covered. Nesting with an existing kvm->srcu critical section is fine; the helper uses a local index. Drop the now-redundant SRCU pair from __kvm_set_or_clear_apicv_inhibit().
-
CVE-2026-99347
In the Linux kernel, the following vulnerability has been resolved: ALSA: x86: hdmi: fix chmap array missing terminator entry had_build_channel_allocation_map() allocates a single struct snd_pcm_chmap_elem via kzalloc() and stores it as intelhaddata->chmap->chmap. struct snd_pcm_chmap.chmap is expected to be a zero-terminated array (see snd_pcm_std_chmaps[]), but had_register_chmap_ctls() never overrides the TLV callback installed by snd_pcm_add_chmap_ctls(), so the generic pcm_chmap_ctl_tlv() from ALSA core is left in place. pcm_chmap_ctl_tlv() is reachable by any local user via SNDRV_CTL_IOCTL_TLV_READ on the "Playback Channel Map" control and walks the array with "for (map = info->chmap; map->channels; map++)". Since the driver's single element is never terminated, this walks past the allocation. The same pointer is also freed on hot-unplug and rebuilt on hot-plug under intelhaddata->mutex, while the TLV callback runs under card->controls_rwsem without that mutex, so the walk can also race a concurrent free of the array. Fix this by allocating a second, zeroed element so the array is properly terminated, and by dropping the inherited TLV callback entirely at registration time, since the driver never implements one of its own.
-
CVE-2026-99332
In the Linux kernel, the following vulnerability has been resolved: HID: core: Fix out-of-bounds left shift by 32 in s32ton() Syzbot reported an out-of-bounds shift (32-bit integer shifted left by 32) in the s32ton() routine in the HID core. This routine converts a signed 32-bit integer to a signed n-bit integer, and it caps the value of n at 32. But when n == 32 there's no need to do anything; it can simply return the input value. This will avoid calculating (1 << n) when n is 32, which was the cause of the invalid shift.
-
CVE-2026-99324
In the Linux kernel, the following vulnerability has been resolved: ksmbd: fix out-of-bounds read when padding the final compound response Compound response handling pads the last response of a compound to an 8-byte boundary by growing the final iov and the RFC1002 length. smb2_allocate_rsp_buf() allocates exactly response_sz. A command can fill the response buffer to within 7 bytes of its end, for example a QUERY_INFO whose output is bounded only by the space left. The padding can then run past the allocation, and ksmbd_tcp_writev() reads those bytes when it sends the response. BUG: KASAN: slab-out-of-bounds in _copy_from_iter+0x744/0x10e0 Call Trace: _copy_from_iter+0x744/0x10e0 tcp_sendmsg_locked+0x2290/0x3f80 ksmbd_tcp_writev+0x10c7/0x1ab0 [ksmbd] Allocate room for the padding when the request is a compound, as smb2_read() and smb2_read_pipe() do for their payloads. response_sz is unchanged, so commands still fill only up to it, and kvzalloc() zeroes the padding.
-
CVE-2026-99327
In the Linux kernel, the following vulnerability has been resolved: HID: roccat: kovaplus: reject short button reports The raw event callback runs before HID core validates and zero-pads the report. It passes the buffer to helpers which inspect data[0] and read a complete struct kovaplus_mouse_report_button for button reports. A truncated report can therefore cause an out-of-bounds read. Require enough data for the button report before calling either helper. Commit 47669bec44fe ("HID: asus: refactor the two workqueues and init sequence") added the same kind of raw-event length validation to hid-asus.
-
CVE-2026-99330
In the Linux kernel, the following vulnerability has been resolved: HID: mcp2200: validate READ_ALL response length The HID core invokes raw_event callbacks before validating the report length. mcp2200_raw_event() casts a READ_ALL response to struct mcp_read_all_resp and reads fields through offset ten without checking that the response contains the structure. A one-byte READ_ALL response reproduced the issue under KASAN: BUG: KASAN: slab-out-of-bounds in mcp2200_raw_event+0x24b/0x3a0 Read of size 1 by task hidtrigger/89 Call Trace: mcp2200_raw_event+0x24b/0x3a0 kasan_report+0x139/0x170 Reject an incomplete READ_ALL response with a protocol error before reading its fields. The common completion path then wakes the command waiter with the error instead of leaving it to time out. Commit 47669bec44fe ("HID: asus: refactor the two workqueues and init sequence") added the same kind of raw_event length validation to hid-asus.
CVE records published in 2026 (ids starting CVE-2026-), sorted by how likely they are to be exploited.