Merge tag 'drm-intel-next-fixes-2017-04-27' of git://anongit.freedesktop.org/git/drm-intel into drm-next

drm/i915 and gvt fixes for drm-next/v4.12

* tag 'drm-intel-next-fixes-2017-04-27' of git://anongit.freedesktop.org/git/drm-intel:
  drm/i915: Confirm the request is still active before adding it to the await
  drm/i915: Avoid busy-spinning on VLV_GLTC_PW_STATUS mmio
  drm/i915/selftests: Allocate inode/file dynamically
  drm/i915: Fix system hang with EI UP masked on Haswell
  drm/i915: checking for NULL instead of IS_ERR() in mock selftests
  drm/i915: Perform link quality check unconditionally during long pulse
  drm/i915: Fix use after free in lpe_audio_platdev_destroy()
  drm/i915: Use the right mapping_gfp_mask for final shmem allocation
  drm/i915: Make legacy cursor updates more unsynced
  drm/i915: Apply a cond_resched() to the saturated signaler
  drm/i915: Park the signaler before sleeping
  drm/i915/gvt: fix a bounds check in ring_id_to_context_switch_event()
  drm/i915/gvt: Fix PTE write flush for taking runtime pm properly
  drm/i915/gvt: remove some debug messages in scheduler timer handler
  drm/i915/gvt: add mmio init for virtual display
  drm/i915/gvt: use directly assignment for structure copying
  drm/i915/gvt: remove redundant ring id check which cause significant CPU misprediction
  drm/i915/gvt: remove redundant platform check for mocs load/restore
  drm/i915/gvt: Align render mmio list to cacheline
  drm/i915/gvt: cleanup some too chatty scheduler message

17 files changed, 163 insertions, 97 deletions

diff --git a/drivers/gpu/drm/i915/gvt/cmd_parser.c b/drivers/gpu/drm/i915/gvt/cmd_parser.c
index 94f2e701e4d4..41b2c3aaa04a 100644
--- a/drivers/gpu/drm/i915/gvt/cmd_parser.c
+++ b/drivers/gpu/drm/i915/gvt/cmd_parser.c
@@ -616,9 +616,6 @@ static inline u32 get_opcode(u32 cmd, int ring_id)
 {
 	struct decode_info *d_info;
 
-	if (ring_id >= I915_NUM_ENGINES)
-		return INVALID_OP;
-
 	d_info = ring_decode_info[ring_id][CMD_TYPE(cmd)];
 	if (d_info == NULL)
 		return INVALID_OP;
@@ -661,9 +658,6 @@ static inline void print_opcode(u32 cmd, int ring_id)
 	struct decode_info *d_info;
 	int i;
 
-	if (ring_id >= I915_NUM_ENGINES)
-		return;
-
 	d_info = ring_decode_info[ring_id][CMD_TYPE(cmd)];
 	if (d_info == NULL)
 		return;
@@ -2483,7 +2477,7 @@ static int cmd_parser_exec(struct parser_exec_state *s)
 
 	t1 = get_cycles();
 
-	memcpy(&s_before_advance_custom, s, sizeof(struct parser_exec_state));
+	s_before_advance_custom = *s;
 
 	if (info->handler) {
 		ret = info->handler(s);
diff --git a/drivers/gpu/drm/i915/gvt/display.c b/drivers/gpu/drm/i915/gvt/display.c
index 4cf2b29fbaa1..e0261fcc5b50 100644
--- a/drivers/gpu/drm/i915/gvt/display.c
+++ b/drivers/gpu/drm/i915/gvt/display.c
@@ -189,17 +189,44 @@ static void emulate_monitor_status_change(struct intel_vgpu *vgpu)
 	}
 
 	if (intel_vgpu_has_monitor_on_port(vgpu, PORT_B)) {
-		vgpu_vreg(vgpu, SDEISR) |= SDE_PORTB_HOTPLUG_CPT;
 		vgpu_vreg(vgpu, SFUSE_STRAP) |= SFUSE_STRAP_DDIB_DETECTED;
+		vgpu_vreg(vgpu, TRANS_DDI_FUNC_CTL(TRANSCODER_A)) &=
+			~(TRANS_DDI_BPC_MASK | TRANS_DDI_MODE_SELECT_MASK |
+			TRANS_DDI_PORT_MASK);
+		vgpu_vreg(vgpu, TRANS_DDI_FUNC_CTL(TRANSCODER_A)) |=
+			(TRANS_DDI_BPC_8 | TRANS_DDI_MODE_SELECT_DP_SST |
+			(PORT_B << TRANS_DDI_PORT_SHIFT) |
+			TRANS_DDI_FUNC_ENABLE);
+		vgpu_vreg(vgpu, DDI_BUF_CTL(PORT_B)) |= DDI_BUF_CTL_ENABLE;
+		vgpu_vreg(vgpu, DDI_BUF_CTL(PORT_B)) &= ~DDI_BUF_IS_IDLE;
+		vgpu_vreg(vgpu, SDEISR) |= SDE_PORTB_HOTPLUG_CPT;
 	}
 
 	if (intel_vgpu_has_monitor_on_port(vgpu, PORT_C)) {
 		vgpu_vreg(vgpu, SDEISR) |= SDE_PORTC_HOTPLUG_CPT;
+		vgpu_vreg(vgpu, TRANS_DDI_FUNC_CTL(TRANSCODER_A)) &=
+			~(TRANS_DDI_BPC_MASK | TRANS_DDI_MODE_SELECT_MASK |
+			TRANS_DDI_PORT_MASK);
+		vgpu_vreg(vgpu, TRANS_DDI_FUNC_CTL(TRANSCODER_A)) |=
+			(TRANS_DDI_BPC_8 | TRANS_DDI_MODE_SELECT_DP_SST |
+			(PORT_C << TRANS_DDI_PORT_SHIFT) |
+			TRANS_DDI_FUNC_ENABLE);
+		vgpu_vreg(vgpu, DDI_BUF_CTL(PORT_C)) |= DDI_BUF_CTL_ENABLE;
+		vgpu_vreg(vgpu, DDI_BUF_CTL(PORT_C)) &= ~DDI_BUF_IS_IDLE;
 		vgpu_vreg(vgpu, SFUSE_STRAP) |= SFUSE_STRAP_DDIC_DETECTED;
 	}
 
 	if (intel_vgpu_has_monitor_on_port(vgpu, PORT_D)) {
 		vgpu_vreg(vgpu, SDEISR) |= SDE_PORTD_HOTPLUG_CPT;
+		vgpu_vreg(vgpu, TRANS_DDI_FUNC_CTL(TRANSCODER_A)) &=
+			~(TRANS_DDI_BPC_MASK | TRANS_DDI_MODE_SELECT_MASK |
+			TRANS_DDI_PORT_MASK);
+		vgpu_vreg(vgpu, TRANS_DDI_FUNC_CTL(TRANSCODER_A)) |=
+			(TRANS_DDI_BPC_8 | TRANS_DDI_MODE_SELECT_DP_SST |
+			(PORT_D << TRANS_DDI_PORT_SHIFT) |
+			TRANS_DDI_FUNC_ENABLE);
+		vgpu_vreg(vgpu, DDI_BUF_CTL(PORT_D)) |= DDI_BUF_CTL_ENABLE;
+		vgpu_vreg(vgpu, DDI_BUF_CTL(PORT_D)) &= ~DDI_BUF_IS_IDLE;
 		vgpu_vreg(vgpu, SFUSE_STRAP) |= SFUSE_STRAP_DDID_DETECTED;
 	}
 
diff --git a/drivers/gpu/drm/i915/gvt/execlist.c b/drivers/gpu/drm/i915/gvt/execlist.c
index 536bde8638c8..dca989eb2d42 100644
--- a/drivers/gpu/drm/i915/gvt/execlist.c
+++ b/drivers/gpu/drm/i915/gvt/execlist.c
@@ -56,8 +56,8 @@ static int context_switch_events[] = {
 
 static int ring_id_to_context_switch_event(int ring_id)
 {
-	if (WARN_ON(ring_id < RCS && ring_id >
-				ARRAY_SIZE(context_switch_events)))
+	if (WARN_ON(ring_id < RCS ||
+		    ring_id >= ARRAY_SIZE(context_switch_events)))
 		return -EINVAL;
 
 	return context_switch_events[ring_id];
@@ -687,9 +687,7 @@ static int submit_context(struct intel_vgpu *vgpu, int ring_id,
 	}
 
 	if (emulate_schedule_in)
-		memcpy(&workload->elsp_dwords,
-				&vgpu->execlist[ring_id].elsp_dwords,
-				sizeof(workload->elsp_dwords));
+		workload->elsp_dwords = vgpu->execlist[ring_id].elsp_dwords;
 
 	gvt_dbg_el("workload %p ring id %d head %x tail %x start %x ctl %x\n",
 			workload, ring_id, head, tail, start, ctl);
diff --git a/drivers/gpu/drm/i915/gvt/gtt.c b/drivers/gpu/drm/i915/gvt/gtt.c
index 6da4e444e572..c6f0077f590d 100644
--- a/drivers/gpu/drm/i915/gvt/gtt.c
+++ b/drivers/gpu/drm/i915/gvt/gtt.c
@@ -2294,12 +2294,15 @@ void intel_gvt_clean_gtt(struct intel_gvt *gvt)
 void intel_vgpu_reset_ggtt(struct intel_vgpu *vgpu)
 {
 	struct intel_gvt *gvt = vgpu->gvt;
+	struct drm_i915_private *dev_priv = gvt->dev_priv;
 	struct intel_gvt_gtt_pte_ops *ops = vgpu->gvt->gtt.pte_ops;
 	u32 index;
 	u32 offset;
 	u32 num_entries;
 	struct intel_gvt_gtt_entry e;
 
+	intel_runtime_pm_get(dev_priv);
+
 	memset(&e, 0, sizeof(struct intel_gvt_gtt_entry));
 	e.type = GTT_TYPE_GGTT_PTE;
 	ops->set_pfn(&e, gvt->gtt.scratch_ggtt_mfn);
@@ -2314,6 +2317,8 @@ void intel_vgpu_reset_ggtt(struct intel_vgpu *vgpu)
 	num_entries = vgpu_hidden_sz(vgpu) >> PAGE_SHIFT;
 	for (offset = 0; offset < num_entries; offset++)
 		ops->set_entry(NULL, &e, index + offset, false, 0, vgpu);
+
+	intel_runtime_pm_put(dev_priv);
 }
 
 /**
diff --git a/drivers/gpu/drm/i915/gvt/render.c b/drivers/gpu/drm/i915/gvt/render.c
index a7b665e9bbad..c6e7972ac21d 100644
--- a/drivers/gpu/drm/i915/gvt/render.c
+++ b/drivers/gpu/drm/i915/gvt/render.c
@@ -44,7 +44,7 @@ struct render_mmio {
 	u32 value;
 };
 
-static struct render_mmio gen8_render_mmio_list[] = {
+static struct render_mmio gen8_render_mmio_list[] __cacheline_aligned = {
 	{RCS, _MMIO(0x229c), 0xffff, false},
 	{RCS, _MMIO(0x2248), 0x0, false},
 	{RCS, _MMIO(0x2098), 0x0, false},
@@ -75,7 +75,7 @@ static struct render_mmio gen8_render_mmio_list[] = {
 	{BCS, _MMIO(0x22028), 0x0, false},
 };
 
-static struct render_mmio gen9_render_mmio_list[] = {
+static struct render_mmio gen9_render_mmio_list[] __cacheline_aligned = {
 	{RCS, _MMIO(0x229c), 0xffff, false},
 	{RCS, _MMIO(0x2248), 0x0, false},
 	{RCS, _MMIO(0x2098), 0x0, false},
@@ -204,9 +204,6 @@ static void load_mocs(struct intel_vgpu *vgpu, int ring_id)
 	if (WARN_ON(ring_id >= ARRAY_SIZE(regs)))
 		return;
 
-	if (!(IS_SKYLAKE(dev_priv) || IS_KABYLAKE(dev_priv)))
-		return;
-
 	offset.reg = regs[ring_id];
 	for (i = 0; i < 64; i++) {
 		gen9_render_mocs[ring_id][i] = I915_READ(offset);
@@ -242,9 +239,6 @@ static void restore_mocs(struct intel_vgpu *vgpu, int ring_id)
 	if (WARN_ON(ring_id >= ARRAY_SIZE(regs)))
 		return;
 
-	if (!(IS_SKYLAKE(dev_priv) || IS_KABYLAKE(dev_priv)))
-		return;
-
 	offset.reg = regs[ring_id];
 	for (i = 0; i < 64; i++) {
 		vgpu_vreg(vgpu, offset) = I915_READ(offset);
diff --git a/drivers/gpu/drm/i915/gvt/sched_policy.c b/drivers/gpu/drm/i915/gvt/sched_policy.c
index f84959170674..79ba4b3440aa 100644
--- a/drivers/gpu/drm/i915/gvt/sched_policy.c
+++ b/drivers/gpu/drm/i915/gvt/sched_policy.c
@@ -133,9 +133,6 @@ static void try_to_schedule_next_vgpu(struct intel_gvt *gvt)
 	if (!scheduler->next_vgpu)
 		return;
 
-	gvt_dbg_sched("try to schedule next vgpu %d\n",
-			scheduler->next_vgpu->id);
-
 	/*
 	 * after the flag is set, workload dispatch thread will
 	 * stop dispatching workload for current vgpu
@@ -144,15 +141,10 @@ static void try_to_schedule_next_vgpu(struct intel_gvt *gvt)
 
 	/* still have uncompleted workload? */
 	for_each_engine(engine, gvt->dev_priv, i) {
-		if (scheduler->current_workload[i]) {
-			gvt_dbg_sched("still have running workload\n");
+		if (scheduler->current_workload[i])
 			return;
-		}
 	}
 
-	gvt_dbg_sched("switch to next vgpu %d\n",
-			scheduler->next_vgpu->id);
-
 	cur_time = ktime_get();
 	if (scheduler->current_vgpu) {
 		vgpu_data = scheduler->current_vgpu->sched_data;
@@ -224,17 +216,12 @@ static void tbs_sched_func(struct gvt_sched_data *sched_data)
 		list_del_init(&vgpu_data->lru_list);
 		list_add_tail(&vgpu_data->lru_list,
 				&sched_data->lru_runq_head);
-
-		gvt_dbg_sched("pick next vgpu %d\n", vgpu->id);
 	} else {
 		scheduler->next_vgpu = gvt->idle_vgpu;
 	}
 out:
-	if (scheduler->next_vgpu) {
-		gvt_dbg_sched("try to schedule next vgpu %d\n",
-				scheduler->next_vgpu->id);
+	if (scheduler->next_vgpu)
 		try_to_schedule_next_vgpu(gvt);
-	}
 }
 
 void intel_gvt_schedule(struct intel_gvt *gvt)
diff --git a/drivers/gpu/drm/i915/gvt/scheduler.c b/drivers/gpu/drm/i915/gvt/scheduler.c
index a77db2332e68..bada32b33237 100644
--- a/drivers/gpu/drm/i915/gvt/scheduler.c
+++ b/drivers/gpu/drm/i915/gvt/scheduler.c
@@ -279,11 +279,8 @@ static struct intel_vgpu_workload *pick_next_workload(
 		goto out;
 	}
 
-	if (list_empty(workload_q_head(scheduler->current_vgpu, ring_id))) {
-		gvt_dbg_sched("ring id %d stop - no available workload\n",
-				ring_id);
+	if (list_empty(workload_q_head(scheduler->current_vgpu, ring_id)))
 		goto out;
-	}
 
 	/*
 	 * still have current workload, maybe the workload disptacher
diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index 7b4fa84cbc3c..3036d4835b0f 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -2175,6 +2175,20 @@ static void vlv_restore_gunit_s0ix_state(struct drm_i915_private *dev_priv)
 	I915_WRITE(VLV_GUNIT_CLOCK_GATE2,	s->clock_gate_dis2);
 }
 
+static int vlv_wait_for_pw_status(struct drm_i915_private *dev_priv,
+				  u32 mask, u32 val)
+{
+	/* The HW does not like us polling for PW_STATUS frequently, so
+	 * use the sleeping loop rather than risk the busy spin within
+	 * intel_wait_for_register().
+	 *
+	 * Transitioning between RC6 states should be at most 2ms (see
+	 * valleyview_enable_rps) so use a 3ms timeout.
+	 */
+	return wait_for((I915_READ_NOTRACE(VLV_GTLC_PW_STATUS) & mask) == val,
+			3);
+}
+
 int vlv_force_gfx_clock(struct drm_i915_private *dev_priv, bool force_on)
 {
 	u32 val;
@@ -2203,8 +2217,9 @@ int vlv_force_gfx_clock(struct drm_i915_private *dev_priv, bool force_on)
 
 static int vlv_allow_gt_wake(struct drm_i915_private *dev_priv, bool allow)
 {
+	u32 mask;
 	u32 val;
-	int err = 0;
+	int err;
 
 	val = I915_READ(VLV_GTLC_WAKE_CTRL);
 	val &= ~VLV_GTLC_ALLOWWAKEREQ;
@@ -2213,45 +2228,32 @@ static int vlv_allow_gt_wake(struct drm_i915_private *dev_priv, bool allow)
 	I915_WRITE(VLV_GTLC_WAKE_CTRL, val);
 	POSTING_READ(VLV_GTLC_WAKE_CTRL);
 
-	err = intel_wait_for_register(dev_priv,
-				      VLV_GTLC_PW_STATUS,
-				      VLV_GTLC_ALLOWWAKEACK,
-				      allow,
-				      1);
+	mask = VLV_GTLC_ALLOWWAKEACK;
+	val = allow ? mask : 0;
+
+	err = vlv_wait_for_pw_status(dev_priv, mask, val);
 	if (err)
 		DRM_ERROR("timeout disabling GT waking\n");
 
 	return err;
 }
 
-static int vlv_wait_for_gt_wells(struct drm_i915_private *dev_priv,
-				 bool wait_for_on)
+static void vlv_wait_for_gt_wells(struct drm_i915_private *dev_priv,
+				  bool wait_for_on)
 {
 	u32 mask;
 	u32 val;
-	int err;
 
 	mask = VLV_GTLC_PW_MEDIA_STATUS_MASK | VLV_GTLC_PW_RENDER_STATUS_MASK;
 	val = wait_for_on ? mask : 0;
-	if ((I915_READ(VLV_GTLC_PW_STATUS) & mask) == val)
-		return 0;
-
-	DRM_DEBUG_KMS("waiting for GT wells to go %s (%08x)\n",
-		      onoff(wait_for_on),
-		      I915_READ(VLV_GTLC_PW_STATUS));
 
 	/*
 	 * RC6 transitioning can be delayed up to 2 msec (see
 	 * valleyview_enable_rps), use 3 msec for safety.
 	 */
-	err = intel_wait_for_register(dev_priv,
-				      VLV_GTLC_PW_STATUS, mask, val,
-				      3);
-	if (err)
+	if (vlv_wait_for_pw_status(dev_priv, mask, val))
 		DRM_ERROR("timeout waiting for GT wells to go %s\n",
 			  onoff(wait_for_on));
-
-	return err;
 }
 
 static void vlv_check_no_gt_access(struct drm_i915_private *dev_priv)
@@ -2272,7 +2274,7 @@ static int vlv_suspend_complete(struct drm_i915_private *dev_priv)
 	 * Bspec defines the following GT well on flags as debug only, so
 	 * don't treat them as hard failures.
 	 */
-	(void)vlv_wait_for_gt_wells(dev_priv, false);
+	vlv_wait_for_gt_wells(dev_priv, false);
 
 	mask = VLV_GTLC_RENDER_CTX_EXISTS | VLV_GTLC_MEDIA_CTX_EXISTS;
 	WARN_ON((I915_READ(VLV_GTLC_WAKE_CTRL) & mask) != mask);
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 92343343044f..532a577ff7a1 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -2340,7 +2340,7 @@ rebuild_st:
 			 * defer the oom here by reporting the ENOMEM back
 			 * to userspace.
 			 */
-			reclaim = mapping_gfp_constraint(mapping, 0);
+			reclaim = mapping_gfp_mask(mapping);
 			reclaim |= __GFP_NORETRY; /* reclaim, but no oom */
 
 			page = shmem_read_mapping_page_gfp(mapping, i, reclaim);
diff --git a/drivers/gpu/drm/i915/i915_gem_request.c b/drivers/gpu/drm/i915/i915_gem_request.c
index 6348353b91ec..5ddbc9499775 100644
--- a/drivers/gpu/drm/i915/i915_gem_request.c
+++ b/drivers/gpu/drm/i915/i915_gem_request.c
@@ -652,6 +652,9 @@ i915_gem_request_await_request(struct drm_i915_gem_request *to,
 
 	GEM_BUG_ON(to == from);
 
+	if (i915_gem_request_completed(from))
+		return 0;
+
 	if (to->engine->schedule) {
 		ret = i915_priotree_add_dependency(to->i915,
 						   &to->priotree,
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index d9d196977f4a..fd97fe00cd0d 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -4252,12 +4252,12 @@ void intel_irq_init(struct drm_i915_private *dev_priv)
 	dev_priv->rps.pm_intrmsk_mbz = 0;
 
 	/*
-	 * SNB,IVB can while VLV,CHV may hard hang on looping batchbuffer
+	 * SNB,IVB,HSW can while VLV,CHV may hard hang on looping batchbuffer
 	 * if GEN6_PM_UP_EI_EXPIRED is masked.
 	 *
 	 * TODO: verify if this can be reproduced on VLV,CHV.
 	 */
-	if (INTEL_INFO(dev_priv)->gen <= 7 && !IS_HASWELL(dev_priv))
+	if (INTEL_INFO(dev_priv)->gen <= 7)
 		dev_priv->rps.pm_intrmsk_mbz |= GEN6_PM_RP_UP_EI_EXPIRED;
 
 	if (INTEL_INFO(dev_priv)->gen >= 8)
diff --git a/drivers/gpu/drm/i915/intel_breadcrumbs.c b/drivers/gpu/drm/i915/intel_breadcrumbs.c
index b6ea192ad550..9ccbf26124c6 100644
--- a/drivers/gpu/drm/i915/intel_breadcrumbs.c
+++ b/drivers/gpu/drm/i915/intel_breadcrumbs.c
@@ -580,6 +580,8 @@ static int intel_breadcrumbs_signaler(void *arg)
 	signaler_set_rtpriority();
 
 	do {
+		bool do_schedule = true;
+
 		set_current_state(TASK_INTERRUPTIBLE);
 
 		/* We are either woken up by the interrupt bottom-half,
@@ -626,9 +628,23 @@ static int intel_breadcrumbs_signaler(void *arg)
 			spin_unlock_irq(&b->rb_lock);
 
 			i915_gem_request_put(request);
-		} else {
+
+			/* If the engine is saturated we may be continually
+			 * processing completed requests. This angers the
+			 * NMI watchdog if we never let anything else
+			 * have access to the CPU. Let's pretend to be nice
+			 * and relinquish the CPU if we burn through the
+			 * entire RT timeslice!
+			 */
+			do_schedule = need_resched();
+		}
+
+		if (unlikely(do_schedule)) {
 			DEFINE_WAIT(exec);
 
+			if (kthread_should_park())
+				kthread_parkme();
+
 			if (kthread_should_stop()) {
 				GEM_BUG_ON(request);
 				break;
@@ -641,9 +657,6 @@ static int intel_breadcrumbs_signaler(void *arg)
 
 			if (request)
 				remove_wait_queue(&request->execute, &exec);
-
-			if (kthread_should_park())
-				kthread_parkme();
 		}
 		i915_gem_request_put(request);
 	} while (1);
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 881dec88df6e..3617927af269 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -13007,17 +13007,6 @@ static int intel_atomic_commit(struct drm_device *dev,
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	int ret = 0;
 
-	/*
-	 * The intel_legacy_cursor_update() fast path takes care
-	 * of avoiding the vblank waits for simple cursor
-	 * movement and flips. For cursor on/off and size changes,
-	 * we want to perform the vblank waits so that watermark
-	 * updates happen during the correct frames. Gen9+ have
-	 * double buffered watermarks and so shouldn't need this.
-	 */
-	if (INTEL_GEN(dev_priv) < 9)
-		state->legacy_cursor_update = false;
-
 	ret = drm_atomic_helper_setup_commit(state, nonblock);
 	if (ret)
 		return ret;
@@ -13033,6 +13022,26 @@ static int intel_atomic_commit(struct drm_device *dev,
 		return ret;
 	}
 
+	/*
+	 * The intel_legacy_cursor_update() fast path takes care
+	 * of avoiding the vblank waits for simple cursor
+	 * movement and flips. For cursor on/off and size changes,
+	 * we want to perform the vblank waits so that watermark
+	 * updates happen during the correct frames. Gen9+ have
+	 * double buffered watermarks and so shouldn't need this.
+	 *
+	 * Do this after drm_atomic_helper_setup_commit() and
+	 * intel_atomic_prepare_commit() because we still want
+	 * to skip the flip and fb cleanup waits. Although that
+	 * does risk yanking the mapping from under the display
+	 * engine.
+	 *
+	 * FIXME doing watermarks and fb cleanup from a vblank worker
+	 * (assuming we had any) would solve these problems.
+	 */
+	if (INTEL_GEN(dev_priv) < 9)
+		state->legacy_cursor_update = false;
+
 	drm_atomic_helper_swap_state(state, true);
 	dev_priv->wm.distrust_bios_wm = false;
 	intel_shared_dpll_swap_state(state);
diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c
index 6e04cb54e3ff..ee77b519835c 100644
--- a/drivers/gpu/drm/i915/intel_dp.c
+++ b/drivers/gpu/drm/i915/intel_dp.c
@@ -4636,9 +4636,20 @@ intel_dp_long_pulse(struct intel_connector *intel_connector)
 		 */
 		status = connector_status_disconnected;
 		goto out;
-	} else if (connector->status == connector_status_connected) {
+	} else {
+		/*
+		 * If display is now connected check links status,
+		 * there has been known issues of link loss triggerring
+		 * long pulse.
+		 *
+		 * Some sinks (eg. ASUS PB287Q) seem to perform some
+		 * weird HPD ping pong during modesets. So we can apparently
+		 * end up with HPD going low during a modeset, and then
+		 * going back up soon after. And once that happens we must
+		 * retrain the link to get a picture. That's in case no
+		 * userspace component reacted to intermittent HPD dip.
+		 */
 		intel_dp_check_link_status(intel_dp);
-		goto out;
 	}
 
 	/*
diff --git a/drivers/gpu/drm/i915/intel_lpe_audio.c b/drivers/gpu/drm/i915/intel_lpe_audio.c
index d8ca187ae001..25d8e76489e4 100644
--- a/drivers/gpu/drm/i915/intel_lpe_audio.c
+++ b/drivers/gpu/drm/i915/intel_lpe_audio.c
@@ -131,8 +131,15 @@ err:
 
 static void lpe_audio_platdev_destroy(struct drm_i915_private *dev_priv)
 {
+	/* XXX Note that platform_device_register_full() allocates a dma_mask
+	 * and never frees it. We can't free it here as we cannot guarantee
+	 * this is the last reference (i.e. that the dma_mask will not be
+	 * used after our unregister). So ee choose to leak the sizeof(u64)
+	 * allocation here - it should be fixed in the platform_device rather
+	 * than us fiddle with its internals.
+	 */
+
 	platform_device_unregister(dev_priv->lpe_audio.platdev);
-	kfree(dev_priv->lpe_audio.platdev->dev.dma_mask);
 }
 
 static void lpe_audio_irq_unmask(struct irq_data *d)
diff --git a/drivers/gpu/drm/i915/selftests/mock_drm.c b/drivers/gpu/drm/i915/selftests/mock_drm.c
index 113dec05c7dc..09c704153456 100644
--- a/drivers/gpu/drm/i915/selftests/mock_drm.c
+++ b/drivers/gpu/drm/i915/selftests/mock_drm.c
@@ -24,31 +24,50 @@
 
 #include "mock_drm.h"
 
-static inline struct inode fake_inode(struct drm_i915_private *i915)
-{
-	return (struct inode){ .i_rdev = i915->drm.primary->index };
-}
-
 struct drm_file *mock_file(struct drm_i915_private *i915)
 {
-	struct inode inode = fake_inode(i915);
-	struct file filp = {};
+	struct file *filp;
+	struct inode *inode;
 	struct drm_file *file;
 	int err;
 
-	err = drm_open(&inode, &filp);
-	if (unlikely(err))
-		return ERR_PTR(err);
+	inode = kzalloc(sizeof(*inode), GFP_KERNEL);
+	if (!inode) {
+		err = -ENOMEM;
+		goto err;
+	}
+
+	inode->i_rdev = i915->drm.primary->index;
 
-	file = filp.private_data;
+	filp = kzalloc(sizeof(*filp), GFP_KERNEL);
+	if (!filp) {
+		err = -ENOMEM;
+		goto err_inode;
+	}
+
+	err = drm_open(inode, filp);
+	if (err)
+		goto err_filp;
+
+	file = filp->private_data;
+	memset(&file->filp, POISON_INUSE, sizeof(file->filp));
 	file->authenticated = true;
+
+	kfree(filp);
+	kfree(inode);
 	return file;
+
+err_filp:
+	kfree(filp);
+err_inode:
+	kfree(inode);
+err:
+	return ERR_PTR(err);
 }
 
 void mock_file_free(struct drm_i915_private *i915, struct drm_file *file)
 {
-	struct inode inode = fake_inode(i915);
 	struct file filp = { .private_data = file };
 
-	drm_release(&inode, &filp);
+	drm_release(NULL, &filp);
 }
diff --git a/drivers/gpu/drm/i915/selftests/mock_request.c b/drivers/gpu/drm/i915/selftests/mock_request.c
index 0e8d2e7f8c70..8097e3693ec4 100644
--- a/drivers/gpu/drm/i915/selftests/mock_request.c
+++ b/drivers/gpu/drm/i915/selftests/mock_request.c
@@ -35,7 +35,7 @@ mock_request(struct intel_engine_cs *engine,
 
 	/* NB the i915->requests slab cache is enlarged to fit mock_request */
 	request = i915_gem_request_alloc(engine, context);
-	if (!request)
+	if (IS_ERR(request))
 		return NULL;
 
 	mock = container_of(request, typeof(*mock), base);