CVE-2026-31656

In the Linux kernel, the following vulnerability has been resolved: drm/i915/gt: fix refcount underflow in intel_engine_park_heartbeat A use-after-free / ref…

Vendor / product: Linux — linux kernel

In the Linux kernel, the following vulnerability has been resolved: drm/i915/gt: fix refcount underflow in intel_engine_park_heartbeat A use-after-free / refcount underflow is possible when the heartbeat worker and intel_engine_park_heartbeat() race to release the same engine->heartbeat.systole request. The heartbeat worker reads engine->heartbeat.systole and calls i915_request_put() on it when the request is complete, but clears the pointer in a separate, non-atomic step. Concurrently, a request retirement on another CPU can drop the engine wakeref to zero, triggering __engine_park() -> intel_engine_park_heartbeat(). If the heartbeat timer is pending at that point, cancel_delayed_work() returns true and intel_engine_park_heartbeat() reads the stale non-NULL systole pointer and calls i915_request_put() on it again, causing a refcount underflow: ``` [487.221889] Workqueue: i915-unordered engine_retire [i915] [487.222640] RIP: 0010:refcount_warn_saturate+0x68/0xb0 ... [487.222707] Call Trace: [487.222711] [487.222716] intel_engine_park_heartbeat.part.0+0x6f/0x80 [i915] [487.223115] intel_engine_park_heartbeat+0x25/0x40 [i915] [487.223566] __engine_park+0xb9/0x650 [i915] [487.223973] ____intel_wakeref_put_last+0x2e/0xb0 [i915] [487.224408] __intel_wakeref_put_last+0x72/0x90 [i915] [487.224797] intel_context_exit_engine+0x7c/0x80 [i915] [487.225238] intel_context_exit+0xf1/0x1b0 [i915] [487.225695] i915_request_retire.part.0+0x1b9/0x530 [i915] [487.226178] i915_request_retire+0x1c/0x40 [i915] [487.226625] engine_retire+0x122/0x180 [i915] [487.227037] process_one_work+0x239/0x760 [487.227060] worker_thread+0x200/0x3f0 [487.227068] ? __pfx_worker_thread+0x10/0x10 [487.227075] kthread+0x10d/0x150 [487.227083] ? __pfx_kthread+0x10/0x10 [487.227092] ret_from_fork+0x3d4/0x480 [487.227099] ? __pfx_kthread+0x10/0x10 [487.227107] ret_from_fork_asm+0x1a/0x30 [487.227141] ``` Fix this by replacing the non-atomic pointer read + separate clear with xchg() in both racing paths. xchg() is a single indivisible hardware instruction that atomically reads the old pointer and writes NULL. This guarantees only one of the two concurrent callers obtains the non-NULL pointer and performs the put, the other gets NULL and skips it. (cherry picked from commit 13238dc0ee4f9ab8dafa2cca7295736191ae2f42)