Merge tag 'drm-intel-next-fixes-2017-11-10' of git://anongit.freedesk…

…top.org/drm/drm-intel into drm-next

drm/i915 fixes for v4.15

* tag 'drm-intel-next-fixes-2017-11-10' of git://anongit.freedesktop.org/drm/drm-intel:
  drm/i915: Reorder context-close to avoid calling i915_vma_close() under RCU
  drm/i915: Move init_clock_gating() back to where it was
  drm/i915: Prune the reservation shared fence array
  drm/i915: Idle the GPU before shinking everything
  drm/i915: Lock llist_del_first() vs llist_del_all()
  drm/i915: Calculate ironlake intermediate watermarks correctly, v2.
  drm/i915: Disable lazy PPGTT page table optimization for vGPU
  drm/i915/execlists: Remove the priority "optimisation"
  drm/i915: Filter out spurious execlists context-switch interrupts

drivers/gpu/drm/i915/i915_gem.c

@@ -4603,11 +4603,17 @@ static void __i915_gem_free_work(struct work_struct *work)
 	 * unbound now.
 	 */
 
+	spin_lock(&i915->mm.free_lock);
 	while ((freed = llist_del_all(&i915->mm.free_list))) {
+		spin_unlock(&i915->mm.free_lock);
+
 		__i915_gem_free_objects(i915, freed);
 		if (need_resched())
-			break;
+			return;
+
+		spin_lock(&i915->mm.free_lock);
 	}
+	spin_unlock(&i915->mm.free_lock);
 }
 
 static void __i915_gem_free_object_rcu(struct rcu_head *head)

drivers/gpu/drm/i915/i915_gem_context.c

@@ -106,14 +106,9 @@ static void lut_close(struct i915_gem_context *ctx)
 
 	radix_tree_for_each_slot(slot, &ctx->handles_vma, &iter, 0) {
 		struct i915_vma *vma = rcu_dereference_raw(*slot);
-		struct drm_i915_gem_object *obj = vma->obj;
 
 		radix_tree_iter_delete(&ctx->handles_vma, &iter, slot);
-
-		if (!i915_vma_is_ggtt(vma))
-			i915_vma_close(vma);
-
-		__i915_gem_object_release_unless_active(obj);
+		__i915_gem_object_release_unless_active(vma->obj);
 	}
 }
 
@@ -198,6 +193,11 @@ static void context_close(struct i915_gem_context *ctx)
 {
 	i915_gem_context_set_closed(ctx);
 
+	/*
+	 * The LUT uses the VMA as a backpointer to unref the object,
+	 * so we need to clear the LUT before we close all the VMA (inside
+	 * the ppgtt).
+	 */
 	lut_close(ctx);
 	if (ctx->ppgtt)
 		i915_ppgtt_close(&ctx->ppgtt->base);

drivers/gpu/drm/i915/i915_gem_gtt.c

@@ -1341,7 +1341,7 @@ static int gen8_ppgtt_alloc_pd(struct i915_address_space *vm,
 			if (IS_ERR(pt))
 				goto unwind;
 
-			if (count < GEN8_PTES)
+			if (count < GEN8_PTES || intel_vgpu_active(vm->i915))
 				gen8_initialize_pt(vm, pt);
 
 			gen8_ppgtt_set_pde(vm, pd, pt, pde);

drivers/gpu/drm/i915/i915_gem_shrinker.c

@@ -162,6 +162,18 @@ i915_gem_shrink(struct drm_i915_private *dev_priv,
 	if (!shrinker_lock(dev_priv, &unlock))
 		return 0;
 
+	/*
+	 * When shrinking the active list, also consider active contexts.
+	 * Active contexts are pinned until they are retired, and so can
+	 * not be simply unbound to retire and unpin their pages. To shrink
+	 * the contexts, we must wait until the gpu is idle.
+	 *
+	 * We don't care about errors here; if we cannot wait upon the GPU,
+	 * we will free as much as we can and hope to get a second chance.
+	 */
+	if (flags & I915_SHRINK_ACTIVE)
+		i915_gem_wait_for_idle(dev_priv, I915_WAIT_LOCKED);
+
 	trace_i915_gem_shrink(dev_priv, target, flags);
 	i915_gem_retire_requests(dev_priv);
 

drivers/gpu/drm/i915/i915_guc_submission.c

@@ -610,6 +610,7 @@ static void i915_guc_dequeue(struct intel_engine_cs *engine)
 	execlists->first = rb;
 	if (submit) {
 		port_assign(port, last);
+		execlists_set_active(execlists, EXECLISTS_ACTIVE_USER);
 		i915_guc_submit(engine);
 	}
 	spin_unlock_irq(&engine->timeline->lock);
@@ -633,6 +634,8 @@ static void i915_guc_irq_handler(unsigned long data)
 
 		rq = port_request(&port[0]);
 	}
+	if (!rq)
+		execlists_clear_active(execlists, EXECLISTS_ACTIVE_USER);
 
 	if (!port_isset(last_port))
 		i915_guc_dequeue(engine);

drivers/gpu/drm/i915/i915_irq.c

@@ -1388,8 +1388,10 @@ gen8_cs_irq_handler(struct intel_engine_cs *engine, u32 iir, int test_shift)
 	bool tasklet = false;
 
 	if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << test_shift)) {
-		__set_bit(ENGINE_IRQ_EXECLIST, &engine->irq_posted);
-		tasklet = true;
+		if (READ_ONCE(engine->execlists.active)) {
+			__set_bit(ENGINE_IRQ_EXECLIST, &engine->irq_posted);
+			tasklet = true;
+		}
 	}
 
 	if (iir & (GT_RENDER_USER_INTERRUPT << test_shift)) {

drivers/gpu/drm/i915/i915_vma.c

@@ -54,6 +54,13 @@ i915_vma_retire(struct i915_gem_active *active,
 	if (--obj->active_count)
 		return;
 
+	/* Prune the shared fence arrays iff completely idle (inc. external) */
+	if (reservation_object_trylock(obj->resv)) {
+		if (reservation_object_test_signaled_rcu(obj->resv, true))
+			reservation_object_add_excl_fence(obj->resv, NULL);
+		reservation_object_unlock(obj->resv);
+	}
+
 	/* Bump our place on the bound list to keep it roughly in LRU order
 	 * so that we don't steal from recently used but inactive objects
 	 * (unless we are forced to ofc!)

drivers/gpu/drm/i915/intel_display.c

@@ -3676,6 +3676,7 @@ void intel_finish_reset(struct drm_i915_private *dev_priv)
 
 		intel_pps_unlock_regs_wa(dev_priv);
 		intel_modeset_init_hw(dev);
+		intel_init_clock_gating(dev_priv);
 
 		spin_lock_irq(&dev_priv->irq_lock);
 		if (dev_priv->display.hpd_irq_setup)
@@ -14350,8 +14351,6 @@ void intel_modeset_init_hw(struct drm_device *dev)
 
 	intel_update_cdclk(dev_priv);
 	dev_priv->cdclk.logical = dev_priv->cdclk.actual = dev_priv->cdclk.hw;
-
-	intel_init_clock_gating(dev_priv);
 }
 
 /*
@@ -15063,6 +15062,15 @@ intel_modeset_setup_hw_state(struct drm_device *dev,
 	struct intel_encoder *encoder;
 	int i;
 
+	if (IS_HASWELL(dev_priv)) {
+		/*
+		 * WaRsPkgCStateDisplayPMReq:hsw
+		 * System hang if this isn't done before disabling all planes!
+		 */
+		I915_WRITE(CHICKEN_PAR1_1,
+			   I915_READ(CHICKEN_PAR1_1) | FORCE_ARB_IDLE_PLANES);
+	}
+
 	intel_modeset_readout_hw_state(dev);
 
 	/* HW state is read out, now we need to sanitize this mess. */
@@ -15160,6 +15168,8 @@ void intel_modeset_gem_init(struct drm_device *dev)
 
 	intel_init_gt_powersave(dev_priv);
 
+	intel_init_clock_gating(dev_priv);
+
 	intel_setup_overlay(dev_priv);
 }
 

drivers/gpu/drm/i915/intel_engine_cs.c

@@ -1548,8 +1548,8 @@ bool intel_engine_is_idle(struct intel_engine_cs *engine)
 	if (test_bit(ENGINE_IRQ_EXECLIST, &engine->irq_posted))
 		return false;
 
-	/* Both ports drained, no more ELSP submission? */
-	if (port_request(&engine->execlists.port[0]))
+	/* Waiting to drain ELSP? */
+	if (READ_ONCE(engine->execlists.active))
 		return false;
 
 	/* ELSP is empty, but there are ready requests? */
@@ -1749,6 +1749,7 @@ void intel_engine_dump(struct intel_engine_cs *engine, struct drm_printer *m)
 					   idx);
 			}
 		}
+		drm_printf(m, "\t\tHW active? 0x%x\n", execlists->active);
 		rcu_read_unlock();
 	} else if (INTEL_GEN(dev_priv) > 6) {
 		drm_printf(m, "\tPP_DIR_BASE: 0x%08x\n",

drivers/gpu/drm/i915/intel_lrc.c

@@ -575,7 +575,8 @@ static void execlists_dequeue(struct intel_engine_cs *engine)
 			 * the state of the GPU is known (idle).
 			 */
 			inject_preempt_context(engine);
-			execlists->preempt = true;
+			execlists_set_active(execlists,
+					     EXECLISTS_ACTIVE_PREEMPT);
 			goto unlock;
 		} else {
 			/*
@@ -683,8 +684,10 @@ static void execlists_dequeue(struct intel_engine_cs *engine)
 unlock:
 	spin_unlock_irq(&engine->timeline->lock);
 
-	if (submit)
+	if (submit) {
+		execlists_set_active(execlists, EXECLISTS_ACTIVE_USER);
 		execlists_submit_ports(engine);
+	}
 }
 
 static void
@@ -696,6 +699,7 @@ execlist_cancel_port_requests(struct intel_engine_execlists *execlists)
 	while (num_ports-- && port_isset(port)) {
 		struct drm_i915_gem_request *rq = port_request(port);
 
+		GEM_BUG_ON(!execlists->active);
 		execlists_context_status_change(rq, INTEL_CONTEXT_SCHEDULE_PREEMPTED);
 		i915_gem_request_put(rq);
 
@@ -730,7 +734,6 @@ static void execlists_cancel_requests(struct intel_engine_cs *engine)
 
 		list_for_each_entry_safe(rq, rn, &p->requests, priotree.link) {
 			INIT_LIST_HEAD(&rq->priotree.link);
-			rq->priotree.priority = INT_MAX;
 
 			dma_fence_set_error(&rq->fence, -EIO);
 			__i915_gem_request_submit(rq);
@@ -861,15 +864,21 @@ static void intel_lrc_irq_handler(unsigned long data)
 				unwind_incomplete_requests(engine);
 				spin_unlock_irq(&engine->timeline->lock);
 
-				GEM_BUG_ON(!execlists->preempt);
-				execlists->preempt = false;
+				GEM_BUG_ON(!execlists_is_active(execlists,
+								EXECLISTS_ACTIVE_PREEMPT));
+				execlists_clear_active(execlists,
+						       EXECLISTS_ACTIVE_PREEMPT);
 				continue;
 			}
 
 			if (status & GEN8_CTX_STATUS_PREEMPTED &&
-			    execlists->preempt)
+			    execlists_is_active(execlists,
+						EXECLISTS_ACTIVE_PREEMPT))
 				continue;
 
+			GEM_BUG_ON(!execlists_is_active(execlists,
+							EXECLISTS_ACTIVE_USER));
+
 			/* Check the context/desc id for this event matches */
 			GEM_DEBUG_BUG_ON(buf[2 * head + 1] != port->context_id);
 
@@ -881,7 +890,6 @@ static void intel_lrc_irq_handler(unsigned long data)
 				execlists_context_status_change(rq, INTEL_CONTEXT_SCHEDULE_OUT);
 
 				trace_i915_gem_request_out(rq);
-				rq->priotree.priority = INT_MAX;
 				i915_gem_request_put(rq);
 
 				execlists_port_complete(execlists, port);
@@ -892,6 +900,9 @@ static void intel_lrc_irq_handler(unsigned long data)
 			/* After the final element, the hw should be idle */
 			GEM_BUG_ON(port_count(port) == 0 &&
 				   !(status & GEN8_CTX_STATUS_ACTIVE_IDLE));
+			if (port_count(port) == 0)
+				execlists_clear_active(execlists,
+						       EXECLISTS_ACTIVE_USER);
 		}
 
 		if (head != execlists->csb_head) {
@@ -901,7 +912,7 @@ static void intel_lrc_irq_handler(unsigned long data)
 		}
 	}
 
-	if (!execlists->preempt)
+	if (!execlists_is_active(execlists, EXECLISTS_ACTIVE_PREEMPT))
 		execlists_dequeue(engine);
 
 	intel_uncore_forcewake_put(dev_priv, execlists->fw_domains);
@@ -1460,7 +1471,7 @@ static int gen8_init_common_ring(struct intel_engine_cs *engine)
 		   GT_CONTEXT_SWITCH_INTERRUPT << engine->irq_shift);
 	clear_bit(ENGINE_IRQ_EXECLIST, &engine->irq_posted);
 	execlists->csb_head = -1;
-	execlists->preempt = false;
+	execlists->active = 0;
 
 	/* After a GPU reset, we may have requests to replay */
 	if (!i915_modparams.enable_guc_submission && execlists->first)