drm/i915: Avoid using rq->engine after free during i915_fence_release

In order to be valid to dereference during the i915_fence_release, after
retiring the fence and releasing its refererences, we assume that
rq->engine can only be a real engine (that stay intact until the device
is shutdown after all fences have been flushed). However, due to a quirk
of preempt-to-busy, we may retire a request that still belongs to a
virtual engine and so eventually free it with rq->engine being invalid.
To avoid dereferencing that invalid engine, we look at the
execution_mask which if it indicates it may be executed on more than one
engine, we know it originated on a virtual engine and may still be on
one.

Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/1906
Fixes: 43acd65 ("drm/i915: Keep a per-engine request pool")
Signed-off-by: Chris Wilson <[email protected]>
Cc: Tvrtko Ursulin <[email protected]>
Reviewed-by: Tvrtko Ursulin <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
(cherry picked from commit 32a4605)
Signed-off-by: Joonas Lahtinen <[email protected]>

Author: ickle

drivers/gpu/drm/i915/i915_request.c

@@ -121,8 +121,39 @@ static void i915_fence_release(struct dma_fence *fence)
 	i915_sw_fence_fini(&rq->submit);
 	i915_sw_fence_fini(&rq->semaphore);
 
-	/* Keep one request on each engine for reserved use under mempressure */
-	if (!cmpxchg(&rq->engine->request_pool, NULL, rq))
+	/*
+	 * Keep one request on each engine for reserved use under mempressure
+	 *
+	 * We do not hold a reference to the engine here and so have to be
+	 * very careful in what rq->engine we poke. The virtual engine is
+	 * referenced via the rq->context and we released that ref during
+	 * i915_request_retire(), ergo we must not dereference a virtual
+	 * engine here. Not that we would want to, as the only consumer of
+	 * the reserved engine->request_pool is the power management parking,
+	 * which must-not-fail, and that is only run on the physical engines.
+	 *
+	 * Since the request must have been executed to be have completed,
+	 * we know that it will have been processed by the HW and will
+	 * not be unsubmitted again, so rq->engine and rq->execution_mask
+	 * at this point is stable. rq->execution_mask will be a single
+	 * bit if the last and _only_ engine it could execution on was a
+	 * physical engine, if it's multiple bits then it started on and
+	 * could still be on a virtual engine. Thus if the mask is not a
+	 * power-of-two we assume that rq->engine may still be a virtual
+	 * engine and so a dangling invalid pointer that we cannot dereference
+	 *
+	 * For example, consider the flow of a bonded request through a virtual
+	 * engine. The request is created with a wide engine mask (all engines
+	 * that we might execute on). On processing the bond, the request mask
+	 * is reduced to one or more engines. If the request is subsequently
+	 * bound to a single engine, it will then be constrained to only
+	 * execute on that engine and never returned to the virtual engine
+	 * after timeslicing away, see __unwind_incomplete_requests(). Thus we
+	 * know that if the rq->execution_mask is a single bit, rq->engine
+	 * can be a physical engine with the exact corresponding mask.
+	 */
+	if (is_power_of_2(rq->execution_mask) &&
+	    !cmpxchg(&rq->engine->request_pool, NULL, rq))
 		return;
 
 	kmem_cache_free(global.slab_requests, rq);