drm/i915/execlists: Trust the CSB

Now that we use the CSB stored in the CPU friendly HWSP, we do not need
to track interrupts for when the mmio CSB registers are valid and can
just check where we read up to last from the cached HWSP. This means we
can forgo the atomic bit tracking from interrupt, and in the next patch
it means we can check the CSB at any time.

v2: Change the splitting inside reset_prepare, we only want to lose
testing the interrupt in this patch, the next patch requires the change
in locking

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180628201211.13837-8-chris@chris-wilson.co.uk

drivers/gpu/drm/i915/i915_irq.c

@@ -1493,23 +1493,18 @@ static void snb_gt_irq_handler(struct drm_i915_private *dev_priv,
 static void
 gen8_cs_irq_handler(struct intel_engine_cs *engine, u32 iir)
 {
-	struct intel_engine_execlists * const execlists = &engine->execlists;
 	bool tasklet = false;
 
-	if (iir & GT_CONTEXT_SWITCH_INTERRUPT) {
-		if (READ_ONCE(engine->execlists.active)) {
-			set_bit(ENGINE_IRQ_EXECLIST, &engine->irq_posted);
-			tasklet = true;
-		}
-	}
+	if (iir & GT_CONTEXT_SWITCH_INTERRUPT)
+		tasklet = true;
 
 	if (iir & GT_RENDER_USER_INTERRUPT) {
 		notify_ring(engine);
 		tasklet |= USES_GUC_SUBMISSION(engine->i915);
 	}
 
 	if (tasklet)
-		tasklet_hi_schedule(&execlists->tasklet);
+		tasklet_hi_schedule(&engine->execlists.tasklet);
 }
 
 static void gen8_gt_irq_ack(struct drm_i915_private *i915,

drivers/gpu/drm/i915/intel_engine_cs.c

@@ -1353,12 +1353,10 @@ static void intel_engine_print_registers(const struct intel_engine_cs *engine,
 		ptr = I915_READ(RING_CONTEXT_STATUS_PTR(engine));
 		read = GEN8_CSB_READ_PTR(ptr);
 		write = GEN8_CSB_WRITE_PTR(ptr);
-		drm_printf(m, "\tExeclist CSB read %d [%d cached], write %d [%d from hws], interrupt posted? %s, tasklet queued? %s (%s)\n",
+		drm_printf(m, "\tExeclist CSB read %d [%d cached], write %d [%d from hws], tasklet queued? %s (%s)\n",
 			   read, execlists->csb_head,
 			   write,
 			   intel_read_status_page(engine, intel_hws_csb_write_index(engine->i915)),
-			   yesno(test_bit(ENGINE_IRQ_EXECLIST,
-					  &engine->irq_posted)),
 			   yesno(test_bit(TASKLET_STATE_SCHED,
 					  &engine->execlists.tasklet.state)),
 			   enableddisabled(!atomic_read(&engine->execlists.tasklet.count)));
@@ -1570,11 +1568,9 @@ void intel_engine_dump(struct intel_engine_cs *engine,
 	spin_unlock(&b->rb_lock);
 	local_irq_restore(flags);
 
-	drm_printf(m, "IRQ? 0x%lx (breadcrumbs? %s) (execlists? %s)\n",
+	drm_printf(m, "IRQ? 0x%lx (breadcrumbs? %s)\n",
 		   engine->irq_posted,
 		   yesno(test_bit(ENGINE_IRQ_BREADCRUMB,
-				  &engine->irq_posted)),
-		   yesno(test_bit(ENGINE_IRQ_EXECLIST,
 				  &engine->irq_posted)));
 
 	drm_printf(m, "HWSP:\n");

drivers/gpu/drm/i915/intel_lrc.c

@@ -875,14 +875,6 @@ static void reset_irq(struct intel_engine_cs *engine)
 	smp_store_mb(engine->execlists.active, 0);
 
 	clear_gtiir(engine);
-
-	/*
-	 * The port is checked prior to scheduling a tasklet, but
-	 * just in case we have suspended the tasklet to do the
-	 * wedging make sure that when it wakes, it decides there
-	 * is no work to do by clearing the irq_posted bit.
-	 */
-	clear_bit(ENGINE_IRQ_EXECLIST, &engine->irq_posted);
 }
 
 static void reset_csb_pointers(struct intel_engine_execlists *execlists)
@@ -973,10 +965,6 @@ static void process_csb(struct intel_engine_cs *engine)
 	const u32 * const buf = execlists->csb_status;
 	u8 head, tail;
 
-	/* Clear before reading to catch new interrupts */
-	clear_bit(ENGINE_IRQ_EXECLIST, &engine->irq_posted);
-	smp_mb__after_atomic();
-
 	/*
 	 * Note that csb_write, csb_status may be either in HWSP or mmio.
 	 * When reading from the csb_write mmio register, we have to be
@@ -1129,11 +1117,10 @@ static void execlists_submission_tasklet(unsigned long data)
 {
 	struct intel_engine_cs * const engine = (struct intel_engine_cs *)data;
 
-	GEM_TRACE("%s awake?=%d, active=%x, irq-posted?=%d\n",
+	GEM_TRACE("%s awake?=%d, active=%x\n",
 		  engine->name,
 		  engine->i915->gt.awake,
-		  engine->execlists.active,
-		  test_bit(ENGINE_IRQ_EXECLIST, &engine->irq_posted));
+		  engine->execlists.active);
 
 	/*
 	 * We can skip acquiring intel_runtime_pm_get() here as it was taken
@@ -1145,14 +1132,7 @@ static void execlists_submission_tasklet(unsigned long data)
 	 */
 	GEM_BUG_ON(!engine->i915->gt.awake);
 
-	/*
-	 * Prefer doing test_and_clear_bit() as a two stage operation to avoid
-	 * imposing the cost of a locked atomic transaction when submitting a
-	 * new request (outside of the context-switch interrupt).
-	 */
-	if (test_bit(ENGINE_IRQ_EXECLIST, &engine->irq_posted))
-		process_csb(engine);
-
+	process_csb(engine);
 	if (!execlists_is_active(&engine->execlists, EXECLISTS_ACTIVE_PREEMPT))
 		execlists_dequeue(engine);
 }
@@ -1920,8 +1900,7 @@ execlists_reset_prepare(struct intel_engine_cs *engine)
 	 * and avoid blaming an innocent request if the stall was due to the
 	 * preemption itself.
 	 */
-	if (test_bit(ENGINE_IRQ_EXECLIST, &engine->irq_posted))
-		process_csb(engine);
+	process_csb(engine);
 
 	/*
 	 * The last active request can then be no later than the last request

drivers/gpu/drm/i915/intel_ringbuffer.h

@@ -367,7 +367,6 @@ struct intel_engine_cs {
 
 	unsigned long irq_posted;
 #define ENGINE_IRQ_BREADCRUMB 0
-#define ENGINE_IRQ_EXECLIST 1
 
 	/* Rather than have every client wait upon all user interrupts,
 	 * with the herd waking after every interrupt and each doing the