CVE-2026-74686
UnknownSummary
A vulnerability in the Linux kernel's rqspinlock mechanism was found, where the queue tail is not reset when an ABBA deadlock is detected. This can cause a thread to wait indefinitely for a next waiter that never arrives, leading to intermittent or permanent system stalls.
Risk Assessment
The risk involves potential deadlocks and system hangs in multi-processor environments, which can lead to service unavailability and system crashes.
Recommendation
It is recommended to urgently apply the official Linux kernel patch that resets the queue tail upon deadlock detection, and to monitor system logs for deadlock messages.
Other vulnerabilities in Linux kernel
See all- CVE-2026-80577Unknown
In the Linux kernel's drm/panthor driver, panthor_fw_load_section_entry() skips BO creation for zero-sized firmware sections but adds them to the section list, leading to NULL pointer dereference in later paths. The fix skips adding such sections to the list.
- CVE-2026-80571Unknown
In the Linux kernel powerpc/pseries papr-phy-attest, missing validation of cmd.length could lead to buffer overflow. The fix adds length checks and fixes memory leaks on error paths.
- CVE-2026-80567Unknown
In the Linux kernel Synaptics RMI4 driver (F54), worker errors were not propagated to the V4L2 queue, causing stale or uninitialized data to be delivered to userspace. The fix adds error checking and marks buffers as error state.
- CVE-2026-80566Unknown
In the Linux kernel, the hynitron_cstxxx driver improperly validated touch count and finger IDs. This can lead to corrupted touch state or out-of-bounds buffer access.
- CVE-2026-80564Unknown
In the Linux kernel, the gve driver lacks an implementation of adjfine, leading to a NULL pointer dereference when triggered from userspace, e.g., via testptp.
- CVE-2026-80563Unknown
A use-after-free vulnerability was found in the Linux kernel's gpio-sloppy-logic-analyzer driver. The 'trigger' debugfs file lacked proper protection, allowing a write to freed memory during device unbind. The issue was fixed by using debugfs_create_file() instead of debugfs_create_file_unsafe().
- CVE-2026-80543Unknown
In the Linux kernel, the xcrb_msg_to_type6cprb_msgx() and xcrb_msg_to_type6_ep11cprb_msgx() functions for s390/zcrypt copy a user space message into a kernel buffer based on length, but further processing assumes 4-byte alignment. As a result, up to 3 bytes of uninitialized kernel memory are forwarded to further processing, potentially exposing kernel memory to the crypto card firmware.
- CVE-2026-80542Unknown
In the Linux kernel, the amdgpu driver for AMD Display has a NULL pointer dereference in amdgpu_dm_crtc_set_vblank() when vblank is enabled or queried before a stream is attached to acrtc_state->stream. This can lead to a system crash.
- CVE-2026-80535Unknown
In the Linux kernel XFS filesystem, during directory tree repair, a self-referential directory may be detected. In such a case, the repair code attempts to lock the same inode twice (double iolock/ilock), leading to a deadlock. The fix detects this corner case and handles it appropriately.
- CVE-2026-80533Unknown
In the Linux kernel XFS filesystem, during AGI repair, the function xrep_iunlink_walk_ondisk_bucket may attempt to use the pointer sc->sa.agi_bp, which can be null if the buffer verifier fails. The fix uses ragi->agi_bp instead, which skips verifier checks, to avoid walking off the end of memory.
Original NVD description (English source)
In the Linux kernel, the following vulnerability has been resolved: rqspinlock: Reset tail when preserving queue on deadlock Currently, the destruction of the waiter queue is suppressed for rqspinlock in cases where a deadlock is detected. Deadlock checks happen relatively frequently (on entry for AA, within 1ms for ABBA), and waiter threads may not be involved in locking scenarios involving deadlocks. Thus, it is useful to not flush the queue and let other waiters take a stab at acquiring the lock after we detect a deadlock and exit. However, we need to follow the same logic as what we did previously for the waitq_timeout label: reset the tail, and if we cannot, signal the next waiter appropriately. In case of deadlocks, this signal would just mark the MCS node as unlocked, and in case of timeouts, it would signal RES_TIMEOUT_VAL. The difference thus is in the value propagated, which decides whether the queue remains active or gets flushed. Not doing the tail reset, and waiting for the next waiter can lead to cases where we are the final waiter, and thus no next waiter arrives, leading to intermittent stalls in this path. Once the next waiter does join, we will be unblocked. In the theoretical case when the next waiter never joins, we risk stalling indefinitely. This can only happen for ABBA deadlocks, since entry into the wait queue is guarded with AA checks. A precise sequence of executions leading up to this scenario can be: CPU 0 holds lock A. CPU 1 holds lock B. CPU 2 attempts lock B, becomes the pending waiter for B. CPU 0 attempts lock B. B has locked+pending bits set, thus CPU 0 queues. CPU 1 attempts lock A. CPU 0 detects an ABBA deadlock. Once deadlock detection happens for CPU 0, it will sit waiting for the next waiter in the queue to populate node->next, which will experience delays until such a waiter arrives. Fix this by adjusting the logic for the check for deadlocks preceding the waitq_timeout label. It would make sense to consolidate code for both cases and use 'ret' to distinguish the value being propagated, but that is left as an exercise for a future refactoring task to avoid diff noise in this patch.