diff --git a/drivers/gpu/drm/i915/gt/intel_lrc.c b/drivers/gpu/drm/i915/gt/intel_lrc.c index 2bd1243e3b0b..3d1b990b556d 100644 --- a/drivers/gpu/drm/i915/gt/intel_lrc.c +++ b/drivers/gpu/drm/i915/gt/intel_lrc.c @@ -2403,6 +2403,35 @@ static bool execlists_hold(struct intel_engine_cs *engine, goto unlock; } + if (rq->engine != engine) { /* preempted virtual engine */ + struct virtual_engine *ve = to_virtual_engine(rq->engine); + + /* + * intel_context_inflight() is only protected by virtue + * of process_csb() being called only by the tasklet (or + * directly from inside reset while the tasklet is suspended). + * Assert that neither of those are allowed to run while we + * poke at the request queues. + */ + GEM_BUG_ON(!reset_in_progress(&engine->execlists)); + + /* + * An unsubmitted request along a virtual engine will + * remain on the active (this) engine until we are able + * to process the context switch away (and so mark the + * context as no longer in flight). That cannot have happened + * yet, otherwise we would not be hanging! + */ + spin_lock(&ve->base.active.lock); + GEM_BUG_ON(intel_context_inflight(rq->context) != engine); + GEM_BUG_ON(ve->request != rq); + ve->request = NULL; + spin_unlock(&ve->base.active.lock); + i915_request_put(rq); + + rq->engine = engine; + } + /* * Transfer this request onto the hold queue to prevent it * being resumbitted to HW (and potentially completed) before we have diff --git a/drivers/gpu/drm/i915/gt/selftest_lrc.c b/drivers/gpu/drm/i915/gt/selftest_lrc.c index e3a408baad0f..65718ca2326e 100644 --- a/drivers/gpu/drm/i915/gt/selftest_lrc.c +++ b/drivers/gpu/drm/i915/gt/selftest_lrc.c @@ -3410,6 +3410,161 @@ static int live_virtual_bond(void *arg) return 0; } +static int reset_virtual_engine(struct intel_gt *gt, + struct intel_engine_cs **siblings, + unsigned int nsibling) +{ + struct intel_engine_cs *engine; + struct intel_context *ve; + unsigned long *heartbeat; + struct igt_spinner spin; + struct i915_request *rq; + unsigned int n; + int err = 0; + + /* + * In order to support offline error capture for fast preempt reset, + * we need to decouple the guilty request and ensure that it and its + * descendents are not executed while the capture is in progress. + */ + + heartbeat = kmalloc_array(nsibling, sizeof(*heartbeat), GFP_KERNEL); + if (!heartbeat) + return -ENOMEM; + + if (igt_spinner_init(&spin, gt)) { + err = -ENOMEM; + goto out_free; + } + + ve = intel_execlists_create_virtual(siblings, nsibling); + if (IS_ERR(ve)) { + err = PTR_ERR(ve); + goto out_spin; + } + + for (n = 0; n < nsibling; n++) + engine_heartbeat_disable(siblings[n], &heartbeat[n]); + + rq = igt_spinner_create_request(&spin, ve, MI_ARB_CHECK); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto out_heartbeat; + } + i915_request_add(rq); + + if (!igt_wait_for_spinner(&spin, rq)) { + intel_gt_set_wedged(gt); + err = -ETIME; + goto out_heartbeat; + } + + engine = rq->engine; + GEM_BUG_ON(engine == ve->engine); + + /* Take ownership of the reset and tasklet */ + if (test_and_set_bit(I915_RESET_ENGINE + engine->id, + >->reset.flags)) { + intel_gt_set_wedged(gt); + err = -EBUSY; + goto out_heartbeat; + } + tasklet_disable(&engine->execlists.tasklet); + + engine->execlists.tasklet.func(engine->execlists.tasklet.data); + GEM_BUG_ON(execlists_active(&engine->execlists) != rq); + + /* Fake a preemption event; failed of course */ + spin_lock_irq(&engine->active.lock); + __unwind_incomplete_requests(engine); + spin_unlock_irq(&engine->active.lock); + GEM_BUG_ON(rq->engine != ve->engine); + + /* Reset the engine while keeping our active request on hold */ + execlists_hold(engine, rq); + GEM_BUG_ON(!i915_request_on_hold(rq)); + + intel_engine_reset(engine, NULL); + GEM_BUG_ON(rq->fence.error != -EIO); + + /* Release our grasp on the engine, letting CS flow again */ + tasklet_enable(&engine->execlists.tasklet); + clear_and_wake_up_bit(I915_RESET_ENGINE + engine->id, >->reset.flags); + + /* Check that we do not resubmit the held request */ + i915_request_get(rq); + if (!i915_request_wait(rq, 0, HZ / 5)) { + pr_err("%s: on hold request completed!\n", + engine->name); + intel_gt_set_wedged(gt); + err = -EIO; + goto out_rq; + } + GEM_BUG_ON(!i915_request_on_hold(rq)); + + /* But is resubmitted on release */ + execlists_unhold(engine, rq); + if (i915_request_wait(rq, 0, HZ / 5) < 0) { + pr_err("%s: held request did not complete!\n", + engine->name); + intel_gt_set_wedged(gt); + err = -ETIME; + } + +out_rq: + i915_request_put(rq); +out_heartbeat: + for (n = 0; n < nsibling; n++) + engine_heartbeat_enable(siblings[n], heartbeat[n]); + + intel_context_put(ve); +out_spin: + igt_spinner_fini(&spin); +out_free: + kfree(heartbeat); + return err; +} + +static int live_virtual_reset(void *arg) +{ + struct intel_gt *gt = arg; + struct intel_engine_cs *siblings[MAX_ENGINE_INSTANCE + 1]; + unsigned int class, inst; + + /* + * Check that we handle a reset event within a virtual engine. + * Only the physical engine is reset, but we have to check the flow + * of the virtual requests around the reset, and make sure it is not + * forgotten. + */ + + if (USES_GUC_SUBMISSION(gt->i915)) + return 0; + + if (!intel_has_reset_engine(gt)) + return 0; + + for (class = 0; class <= MAX_ENGINE_CLASS; class++) { + int nsibling, err; + + nsibling = 0; + for (inst = 0; inst <= MAX_ENGINE_INSTANCE; inst++) { + if (!gt->engine_class[class][inst]) + continue; + + siblings[nsibling++] = gt->engine_class[class][inst]; + } + if (nsibling < 2) + continue; + + err = reset_virtual_engine(gt, siblings, nsibling); + if (err) + return err; + } + + return 0; +} + int intel_execlists_live_selftests(struct drm_i915_private *i915) { static const struct i915_subtest tests[] = { @@ -3435,6 +3590,7 @@ int intel_execlists_live_selftests(struct drm_i915_private *i915) SUBTEST(live_virtual_mask), SUBTEST(live_virtual_preserved), SUBTEST(live_virtual_bond), + SUBTEST(live_virtual_reset), }; if (!HAS_EXECLISTS(i915))