]> git.baikalelectronics.ru Git - kernel.git/commitdiff
drm/i915/guc: Implement reset locally
authorChris Wilson <chris@chris-wilson.co.uk>
Thu, 11 Apr 2019 13:05:14 +0000 (14:05 +0100)
committerChris Wilson <chris@chris-wilson.co.uk>
Thu, 11 Apr 2019 19:48:51 +0000 (20:48 +0100)
Before causing guc and execlists to diverge further (breaking guc in the
process), take a copy of the current reset procedure and make it local to
the guc submission backend

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190411130515.20716-1-chris@chris-wilson.co.uk
drivers/gpu/drm/i915/intel_guc_submission.c
drivers/gpu/drm/i915/intel_lrc.c
drivers/gpu/drm/i915/intel_lrc.h
drivers/gpu/drm/i915/intel_ringbuffer.h

index dea87253d1415719e8db5bbc7f77acd2664eaf07..37f60cb8e9e13c7e4ee166486a1c3b92f8d125e1 100644 (file)
@@ -872,6 +872,104 @@ static void guc_reset_prepare(struct intel_engine_cs *engine)
                flush_workqueue(engine->i915->guc.preempt_wq);
 }
 
+static void guc_reset(struct intel_engine_cs *engine, bool stalled)
+{
+       struct intel_engine_execlists * const execlists = &engine->execlists;
+       struct i915_request *rq;
+       unsigned long flags;
+
+       spin_lock_irqsave(&engine->timeline.lock, flags);
+
+       execlists_cancel_port_requests(execlists);
+
+       /* Push back any incomplete requests for replay after the reset. */
+       rq = execlists_unwind_incomplete_requests(execlists);
+       if (!rq)
+               goto out_unlock;
+
+       if (!i915_request_started(rq))
+               stalled = false;
+
+       i915_reset_request(rq, stalled);
+       intel_lr_context_reset(engine, rq->hw_context, rq->head, stalled);
+
+out_unlock:
+       spin_unlock_irqrestore(&engine->timeline.lock, flags);
+}
+
+static void guc_cancel_requests(struct intel_engine_cs *engine)
+{
+       struct intel_engine_execlists * const execlists = &engine->execlists;
+       struct i915_request *rq, *rn;
+       struct rb_node *rb;
+       unsigned long flags;
+
+       GEM_TRACE("%s\n", engine->name);
+
+       /*
+        * Before we call engine->cancel_requests(), we should have exclusive
+        * access to the submission state. This is arranged for us by the
+        * caller disabling the interrupt generation, the tasklet and other
+        * threads that may then access the same state, giving us a free hand
+        * to reset state. However, we still need to let lockdep be aware that
+        * we know this state may be accessed in hardirq context, so we
+        * disable the irq around this manipulation and we want to keep
+        * the spinlock focused on its duties and not accidentally conflate
+        * coverage to the submission's irq state. (Similarly, although we
+        * shouldn't need to disable irq around the manipulation of the
+        * submission's irq state, we also wish to remind ourselves that
+        * it is irq state.)
+        */
+       spin_lock_irqsave(&engine->timeline.lock, flags);
+
+       /* Cancel the requests on the HW and clear the ELSP tracker. */
+       execlists_cancel_port_requests(execlists);
+
+       /* Mark all executing requests as skipped. */
+       list_for_each_entry(rq, &engine->timeline.requests, link) {
+               if (!i915_request_signaled(rq))
+                       dma_fence_set_error(&rq->fence, -EIO);
+
+               i915_request_mark_complete(rq);
+       }
+
+       /* Flush the queued requests to the timeline list (for retiring). */
+       while ((rb = rb_first_cached(&execlists->queue))) {
+               struct i915_priolist *p = to_priolist(rb);
+               int i;
+
+               priolist_for_each_request_consume(rq, rn, p, i) {
+                       list_del_init(&rq->sched.link);
+                       __i915_request_submit(rq);
+                       dma_fence_set_error(&rq->fence, -EIO);
+                       i915_request_mark_complete(rq);
+               }
+
+               rb_erase_cached(&p->node, &execlists->queue);
+               i915_priolist_free(p);
+       }
+
+       /* Remaining _unready_ requests will be nop'ed when submitted */
+
+       execlists->queue_priority_hint = INT_MIN;
+       execlists->queue = RB_ROOT_CACHED;
+       GEM_BUG_ON(port_isset(execlists->port));
+
+       spin_unlock_irqrestore(&engine->timeline.lock, flags);
+}
+
+static void guc_reset_finish(struct intel_engine_cs *engine)
+{
+       struct intel_engine_execlists * const execlists = &engine->execlists;
+
+       if (__tasklet_enable(&execlists->tasklet))
+               /* And kick in case we missed a new request submission. */
+               tasklet_hi_schedule(&execlists->tasklet);
+
+       GEM_TRACE("%s: depth->%d\n", engine->name,
+                 atomic_read(&execlists->tasklet.count));
+}
+
 /*
  * Everything below here is concerned with setup & teardown, and is
  * therefore not part of the somewhat time-critical batch-submission
@@ -1293,6 +1391,10 @@ static void guc_set_default_submission(struct intel_engine_cs *engine)
        engine->unpark = guc_submission_unpark;
 
        engine->reset.prepare = guc_reset_prepare;
+       engine->reset.reset = guc_reset;
+       engine->reset.finish = guc_reset_finish;
+
+       engine->cancel_requests = guc_cancel_requests;
 
        engine->flags &= ~I915_ENGINE_SUPPORTS_STATS;
 }
index 75446e4b983a559a4913fa81f536b6e46974651c..95ec4f7ad0838164f146d61a9e9390534088df3b 100644 (file)
@@ -429,13 +429,13 @@ __unwind_incomplete_requests(struct intel_engine_cs *engine)
        return active;
 }
 
-void
+struct i915_request *
 execlists_unwind_incomplete_requests(struct intel_engine_execlists *execlists)
 {
        struct intel_engine_cs *engine =
                container_of(execlists, typeof(*engine), execlists);
 
-       __unwind_incomplete_requests(engine);
+       return __unwind_incomplete_requests(engine);
 }
 
 static inline void
@@ -2345,6 +2345,8 @@ void intel_execlists_set_default_submission(struct intel_engine_cs *engine)
        engine->execlists.tasklet.func = execlists_submission_tasklet;
 
        engine->reset.prepare = execlists_reset_prepare;
+       engine->reset.reset = execlists_reset;
+       engine->reset.finish = execlists_reset_finish;
 
        engine->park = NULL;
        engine->unpark = NULL;
@@ -2977,6 +2979,37 @@ void intel_execlists_show_requests(struct intel_engine_cs *engine,
        spin_unlock_irqrestore(&engine->timeline.lock, flags);
 }
 
+void intel_lr_context_reset(struct intel_engine_cs *engine,
+                           struct intel_context *ce,
+                           u32 head,
+                           bool scrub)
+{
+       /*
+        * We want a simple context + ring to execute the breadcrumb update.
+        * We cannot rely on the context being intact across the GPU hang,
+        * so clear it and rebuild just what we need for the breadcrumb.
+        * All pending requests for this context will be zapped, and any
+        * future request will be after userspace has had the opportunity
+        * to recreate its own state.
+        */
+       if (scrub) {
+               u32 *regs = ce->lrc_reg_state;
+
+               if (engine->pinned_default_state) {
+                       memcpy(regs, /* skip restoring the vanilla PPHWSP */
+                              engine->pinned_default_state + LRC_STATE_PN * PAGE_SIZE,
+                              engine->context_size - PAGE_SIZE);
+               }
+               execlists_init_reg_state(regs, ce, engine, ce->ring);
+       }
+
+       /* Rerun the request; its payload has been neutered (if guilty). */
+       ce->ring->head = head;
+       intel_ring_update_space(ce->ring);
+
+       __execlists_update_reg_state(ce, engine);
+}
+
 #if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
 #include "selftests/intel_lrc.c"
 #endif
index 1e343a4af44245b485e6122bf755f785358f686e..84aa230ea27be10acdee4ccf35dabf21a31fc3c2 100644 (file)
@@ -103,6 +103,11 @@ struct i915_gem_context;
 
 void intel_execlists_set_default_submission(struct intel_engine_cs *engine);
 
+void intel_lr_context_reset(struct intel_engine_cs *engine,
+                           struct intel_context *ce,
+                           u32 head,
+                           bool scrub);
+
 void intel_execlists_show_requests(struct intel_engine_cs *engine,
                                   struct drm_printer *m,
                                   void (*show_request)(struct drm_printer *m,
index 0dea6c7fd4384a976bc7d21cbeab3c5e5f828aff..4b33e88eabb184d4542b87b4c285b5f37da5916d 100644 (file)
@@ -165,7 +165,7 @@ void execlists_user_end(struct intel_engine_execlists *execlists);
 void
 execlists_cancel_port_requests(struct intel_engine_execlists * const execlists);
 
-void
+struct i915_request *
 execlists_unwind_incomplete_requests(struct intel_engine_execlists *execlists);
 
 static inline unsigned int