Description
The CONFIG_USERSPACE syscall verifier z_vrfy_k_poll() in kernel/poll.c allocates a kernel-side copy of the user-supplied k_poll_event[] via z_thread_malloc() and then validates each event-s object handle. Before this fix, validation used K_OOPS(K_SYSCALL_OBJ(...)) inline inside the loop, which kills the calling thread without freeing events_copy. A user thread can pass num_events >= 1 with a forged object handle to leak the allocation; because newly spawned user threads inherit the parent-s resource_pool (kernel/thread.c), an attacker spawns sacrificial threads to repeat the leak until the shared kernel heap is exhausted. Once depleted, legitimate kernel allocations from that pool (k_queue alloc nodes, k_msgq buffers, future k_poll calls, etc.) fail, causing a system-level denial of service. The fix replaces each inline K_OOPS with a conditional goto oops_free so the buffer is freed before the thread is killed. Affects Zephyr releases from v1.12.0 (when k_poll was first exposed to user mode) through v4.4.1.
- Publication
- 2026-07-21 22:16:59
- Versions concernées
- >=v1.12.0,<v4.4.1
- Type
- Micrologiciel
- Dernière modification
- 2026-07-30 15:55:39
- Vecteur
- CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:C/C:N/I:N/A:H