Merge tag 'drm-intel-fixes-2017-03-22' of git://anongit.freedesktop.org/git/drm-intel into drm-fixes

drm/i915 fixes for v4.11-rc4

* tag 'drm-intel-fixes-2017-03-22' of git://anongit.freedesktop.org/git/drm-intel:
  drm/i915: make context status notifier head be per engine
  drm/i915: Avoid rcu_barrier() from reclaim paths (shrinker)
  drm/i915/gvt: Fix gvt scheduler interval time
  drm/i915/gvt: GVT pin/unpin shadow context
  drm/i915/gvt: scan shadow indirect context image when valid
  drm/i915/kvmgt: fix suspicious rcu dereference usage
  drm/i915/gvt: add enable_execlists check before enable gvt
  drm/i915/gvt: Remove bogus retry around i915_wait_request
  drm/i915/gvt: correct the ggtt valid bit check in pipe control command
  drm/i915/gvt: replace the gvt_err with gvt_vgpu_err
  drm/i915/gvt: handle force-nonpriv registers, cmd parser part
  drm/i915: Do .init_clock_gating() earlier to avoid it clobbering watermarks
  drm/i915/glk: Remove MODULE_FIRMWARE() tag from Geminilake's DMC
  drm/i915: Reject HDMI 12bpc if the sink doesn't indicate support
  drm/i915: Always call i915_gem_reset_finish() following i915_gem_reset_prepare()
  drm/i915: Stop using RP_DOWN_EI on Baytrail
  drm/i915: Drop support for I915_EXEC_CONSTANTS_* execbuf parameters.
  drm/i915: Only enable hotplug interrupts if the display interrupts are enabled
  drm/i915: Disable engine->irq_tasklet around resets
  drm/i915: Split GEM resetting into 3 phases

32 files changed, 397 insertions, 334 deletions

diff --git a/drivers/gpu/drm/i915/gvt/aperture_gm.c b/drivers/gpu/drm/i915/gvt/aperture_gm.c
index 3b6caac..325618d 100644
--- a/drivers/gpu/drm/i915/gvt/aperture_gm.c
+++ b/drivers/gpu/drm/i915/gvt/aperture_gm.c
@@ -242,7 +242,7 @@ static int alloc_resource(struct intel_vgpu *vgpu,
 	const char *item;
 
 	if (!param->low_gm_sz || !param->high_gm_sz || !param->fence_sz) {
-		gvt_err("Invalid vGPU creation params\n");
+		gvt_vgpu_err("Invalid vGPU creation params\n");
 		return -EINVAL;
 	}
 
@@ -285,9 +285,9 @@ static int alloc_resource(struct intel_vgpu *vgpu,
 	return 0;
 
 no_enough_resource:
-	gvt_err("vgpu%d: fail to allocate resource %s\n", vgpu->id, item);
-	gvt_err("vgpu%d: request %luMB avail %luMB max %luMB taken %luMB\n",
-		vgpu->id, BYTES_TO_MB(request), BYTES_TO_MB(avail),
+	gvt_vgpu_err("fail to allocate resource %s\n", item);
+	gvt_vgpu_err("request %luMB avail %luMB max %luMB taken %luMB\n",
+		BYTES_TO_MB(request), BYTES_TO_MB(avail),
 		BYTES_TO_MB(max), BYTES_TO_MB(taken));
 	return -ENOSPC;
 }
diff --git a/drivers/gpu/drm/i915/gvt/cmd_parser.c b/drivers/gpu/drm/i915/gvt/cmd_parser.c
index 7ae6e2b..2b92cc8 100644
--- a/drivers/gpu/drm/i915/gvt/cmd_parser.c
+++ b/drivers/gpu/drm/i915/gvt/cmd_parser.c
@@ -817,6 +817,25 @@ static bool is_shadowed_mmio(unsigned int offset)
 	return ret;
 }
 
+static inline bool is_force_nonpriv_mmio(unsigned int offset)
+{
+	return (offset >= 0x24d0 && offset < 0x2500);
+}
+
+static int force_nonpriv_reg_handler(struct parser_exec_state *s,
+				     unsigned int offset, unsigned int index)
+{
+	struct intel_gvt *gvt = s->vgpu->gvt;
+	unsigned int data = cmd_val(s, index + 1);
+
+	if (!intel_gvt_in_force_nonpriv_whitelist(gvt, data)) {
+		gvt_err("Unexpected forcenonpriv 0x%x LRI write, value=0x%x\n",
+			offset, data);
+		return -EINVAL;
+	}
+	return 0;
+}
+
 static int cmd_reg_handler(struct parser_exec_state *s,
 	unsigned int offset, unsigned int index, char *cmd)
 {
@@ -824,23 +843,26 @@ static int cmd_reg_handler(struct parser_exec_state *s,
 	struct intel_gvt *gvt = vgpu->gvt;
 
 	if (offset + 4 > gvt->device_info.mmio_size) {
-		gvt_err("%s access to (%x) outside of MMIO range\n",
+		gvt_vgpu_err("%s access to (%x) outside of MMIO range\n",
 				cmd, offset);
 		return -EINVAL;
 	}
 
 	if (!intel_gvt_mmio_is_cmd_access(gvt, offset)) {
-		gvt_err("vgpu%d: %s access to non-render register (%x)\n",
-				s->vgpu->id, cmd, offset);
+		gvt_vgpu_err("%s access to non-render register (%x)\n",
+				cmd, offset);
 		return 0;
 	}
 
 	if (is_shadowed_mmio(offset)) {
-		gvt_err("vgpu%d: found access of shadowed MMIO %x\n",
-				s->vgpu->id, offset);
+		gvt_vgpu_err("found access of shadowed MMIO %x\n", offset);
 		return 0;
 	}
 
+	if (is_force_nonpriv_mmio(offset) &&
+	    force_nonpriv_reg_handler(s, offset, index))
+		return -EINVAL;
+
 	if (offset == i915_mmio_reg_offset(DERRMR) ||
 		offset == i915_mmio_reg_offset(FORCEWAKE_MT)) {
 		/* Writing to HW VGT_PVINFO_PAGE offset will be discarded */
@@ -1008,7 +1030,7 @@ static int cmd_handler_pipe_control(struct parser_exec_state *s)
 			ret = cmd_reg_handler(s, 0x2358, 1, "pipe_ctrl");
 		else if (post_sync == 1) {
 			/* check ggtt*/
-			if ((cmd_val(s, 2) & (1 << 2))) {
+			if ((cmd_val(s, 1) & PIPE_CONTROL_GLOBAL_GTT_IVB)) {
 				gma = cmd_val(s, 2) & GENMASK(31, 3);
 				if (gmadr_bytes == 8)
 					gma |= (cmd_gma_hi(s, 3)) << 32;
@@ -1129,6 +1151,7 @@ static int skl_decode_mi_display_flip(struct parser_exec_state *s,
 		struct mi_display_flip_command_info *info)
 {
 	struct drm_i915_private *dev_priv = s->vgpu->gvt->dev_priv;
+	struct intel_vgpu *vgpu = s->vgpu;
 	u32 dword0 = cmd_val(s, 0);
 	u32 dword1 = cmd_val(s, 1);
 	u32 dword2 = cmd_val(s, 2);
@@ -1167,7 +1190,7 @@ static int skl_decode_mi_display_flip(struct parser_exec_state *s,
 		break;
 
 	default:
-		gvt_err("unknown plane code %d\n", plane);
+		gvt_vgpu_err("unknown plane code %d\n", plane);
 		return -EINVAL;
 	}
 
@@ -1274,25 +1297,26 @@ static int update_plane_mmio_from_mi_display_flip(
 static int cmd_handler_mi_display_flip(struct parser_exec_state *s)
 {
 	struct mi_display_flip_command_info info;
+	struct intel_vgpu *vgpu = s->vgpu;
 	int ret;
 	int i;
 	int len = cmd_length(s);
 
 	ret = decode_mi_display_flip(s, &info);
 	if (ret) {
-		gvt_err("fail to decode MI display flip command\n");
+		gvt_vgpu_err("fail to decode MI display flip command\n");
 		return ret;
 	}
 
 	ret = check_mi_display_flip(s, &info);
 	if (ret) {
-		gvt_err("invalid MI display flip command\n");
+		gvt_vgpu_err("invalid MI display flip command\n");
 		return ret;
 	}
 
 	ret = update_plane_mmio_from_mi_display_flip(s, &info);
 	if (ret) {
-		gvt_err("fail to update plane mmio\n");
+		gvt_vgpu_err("fail to update plane mmio\n");
 		return ret;
 	}
 
@@ -1350,7 +1374,8 @@ static inline int cmd_address_audit(struct parser_exec_state *s,
 	int ret;
 
 	if (op_size > max_surface_size) {
-		gvt_err("command address audit fail name %s\n", s->info->name);
+		gvt_vgpu_err("command address audit fail name %s\n",
+			s->info->name);
 		return -EINVAL;
 	}
 
@@ -1367,7 +1392,7 @@ static inline int cmd_address_audit(struct parser_exec_state *s,
 	}
 	return 0;
 err:
-	gvt_err("cmd_parser: Malicious %s detected, addr=0x%lx, len=%d!\n",
+	gvt_vgpu_err("cmd_parser: Malicious %s detected, addr=0x%lx, len=%d!\n",
 			s->info->name, guest_gma, op_size);
 
 	pr_err("cmd dump: ");
@@ -1412,8 +1437,10 @@ static int cmd_handler_mi_store_data_imm(struct parser_exec_state *s)
 
 static inline int unexpected_cmd(struct parser_exec_state *s)
 {
-	gvt_err("vgpu%d: Unexpected %s in command buffer!\n",
-			s->vgpu->id, s->info->name);
+	struct intel_vgpu *vgpu = s->vgpu;
+
+	gvt_vgpu_err("Unexpected %s in command buffer!\n", s->info->name);
+
 	return -EINVAL;
 }
 
@@ -1516,7 +1543,7 @@ static int copy_gma_to_hva(struct intel_vgpu *vgpu, struct intel_vgpu_mm *mm,
 	while (gma != end_gma) {
 		gpa = intel_vgpu_gma_to_gpa(mm, gma);
 		if (gpa == INTEL_GVT_INVALID_ADDR) {
-			gvt_err("invalid gma address: %lx\n", gma);
+			gvt_vgpu_err("invalid gma address: %lx\n", gma);
 			return -EFAULT;
 		}
 
@@ -1557,6 +1584,7 @@ static uint32_t find_bb_size(struct parser_exec_state *s)
 	uint32_t bb_size = 0;
 	uint32_t cmd_len = 0;
 	bool met_bb_end = false;
+	struct intel_vgpu *vgpu = s->vgpu;
 	u32 cmd;
 
 	/* get the start gm address of the batch buffer */
@@ -1565,7 +1593,7 @@ static uint32_t find_bb_size(struct parser_exec_state *s)
 
 	info = get_cmd_info(s->vgpu->gvt, cmd, s->ring_id);
 	if (info == NULL) {
-		gvt_err("unknown cmd 0x%x, opcode=0x%x\n",
+		gvt_vgpu_err("unknown cmd 0x%x, opcode=0x%x\n",
 				cmd, get_opcode(cmd, s->ring_id));
 		return -EINVAL;
 	}
@@ -1574,7 +1602,7 @@ static uint32_t find_bb_size(struct parser_exec_state *s)
 				gma, gma + 4, &cmd);
 		info = get_cmd_info(s->vgpu->gvt, cmd, s->ring_id);
 		if (info == NULL) {
-			gvt_err("unknown cmd 0x%x, opcode=0x%x\n",
+			gvt_vgpu_err("unknown cmd 0x%x, opcode=0x%x\n",
 				cmd, get_opcode(cmd, s->ring_id));
 			return -EINVAL;
 		}
@@ -1599,6 +1627,7 @@ static uint32_t find_bb_size(struct parser_exec_state *s)
 static int perform_bb_shadow(struct parser_exec_state *s)
 {
 	struct intel_shadow_bb_entry *entry_obj;
+	struct intel_vgpu *vgpu = s->vgpu;
 	unsigned long gma = 0;
 	uint32_t bb_size;
 	void *dst = NULL;
@@ -1633,7 +1662,7 @@ static int perform_bb_shadow(struct parser_exec_state *s)
 
 	ret = i915_gem_object_set_to_cpu_domain(entry_obj->obj, false);
 	if (ret) {
-		gvt_err("failed to set shadow batch to CPU\n");
+		gvt_vgpu_err("failed to set shadow batch to CPU\n");
 		goto unmap_src;
 	}
 
@@ -1645,7 +1674,7 @@ static int perform_bb_shadow(struct parser_exec_state *s)
 			      gma, gma + bb_size,
 			      dst);
 	if (ret) {
-		gvt_err("fail to copy guest ring buffer\n");
+		gvt_vgpu_err("fail to copy guest ring buffer\n");
 		goto unmap_src;
 	}
 
@@ -1676,15 +1705,16 @@ static int cmd_handler_mi_batch_buffer_start(struct parser_exec_state *s)
 {
 	bool second_level;
 	int ret = 0;
+	struct intel_vgpu *vgpu = s->vgpu;
 
 	if (s->buf_type == BATCH_BUFFER_2ND_LEVEL) {
-		gvt_err("Found MI_BATCH_BUFFER_START in 2nd level BB\n");
+		gvt_vgpu_err("Found MI_BATCH_BUFFER_START in 2nd level BB\n");
 		return -EINVAL;
 	}
 
 	second_level = BATCH_BUFFER_2ND_LEVEL_BIT(cmd_val(s, 0)) == 1;
 	if (second_level && (s->buf_type != BATCH_BUFFER_INSTRUCTION)) {
-		gvt_err("Jumping to 2nd level BB from RB is not allowed\n");
+		gvt_vgpu_err("Jumping to 2nd level BB from RB is not allowed\n");
 		return -EINVAL;
 	}
 
@@ -1702,7 +1732,7 @@ static int cmd_handler_mi_batch_buffer_start(struct parser_exec_state *s)
 	if (batch_buffer_needs_scan(s)) {
 		ret = perform_bb_shadow(s);
 		if (ret < 0)
-			gvt_err("invalid shadow batch buffer\n");
+			gvt_vgpu_err("invalid shadow batch buffer\n");
 	} else {
 		/* emulate a batch buffer end to do return right */
 		ret = cmd_handler_mi_batch_buffer_end(s);
@@ -2429,6 +2459,7 @@ static int cmd_parser_exec(struct parser_exec_state *s)
 	int ret = 0;
 	cycles_t t0, t1, t2;
 	struct parser_exec_state s_before_advance_custom;
+	struct intel_vgpu *vgpu = s->vgpu;
 
 	t0 = get_cycles();
 
@@ -2436,7 +2467,7 @@ static int cmd_parser_exec(struct parser_exec_state *s)
 
 	info = get_cmd_info(s->vgpu->gvt, cmd, s->ring_id);
 	if (info == NULL) {
-		gvt_err("unknown cmd 0x%x, opcode=0x%x\n",
+		gvt_vgpu_err("unknown cmd 0x%x, opcode=0x%x\n",
 				cmd, get_opcode(cmd, s->ring_id));
 		return -EINVAL;
 	}
@@ -2452,7 +2483,7 @@ static int cmd_parser_exec(struct parser_exec_state *s)
 	if (info->handler) {
 		ret = info->handler(s);
 		if (ret < 0) {
-			gvt_err("%s handler error\n", info->name);
+			gvt_vgpu_err("%s handler error\n", info->name);
 			return ret;
 		}
 	}
@@ -2463,7 +2494,7 @@ static int cmd_parser_exec(struct parser_exec_state *s)
 	if (!(info->flag & F_IP_ADVANCE_CUSTOM)) {
 		ret = cmd_advance_default(s);
 		if (ret) {
-			gvt_err("%s IP advance error\n", info->name);
+			gvt_vgpu_err("%s IP advance error\n", info->name);
 			return ret;
 		}
 	}
@@ -2486,6 +2517,7 @@ static int command_scan(struct parser_exec_state *s,
 
 	unsigned long gma_head, gma_tail, gma_bottom;
 	int ret = 0;
+	struct intel_vgpu *vgpu = s->vgpu;
 
 	gma_head = rb_start + rb_head;
 	gma_tail = rb_start + rb_tail;
@@ -2497,7 +2529,7 @@ static int command_scan(struct parser_exec_state *s,
 		if (s->buf_type == RING_BUFFER_INSTRUCTION) {
 			if (!(s->ip_gma >= rb_start) ||
 				!(s->ip_gma < gma_bottom)) {
-				gvt_err("ip_gma %lx out of ring scope."
+				gvt_vgpu_err("ip_gma %lx out of ring scope."
 					"(base:0x%lx, bottom: 0x%lx)\n",
 					s->ip_gma, rb_start,
 					gma_bottom);
@@ -2505,7 +2537,7 @@ static int command_scan(struct parser_exec_state *s,
 				return -EINVAL;
 			}
 			if (gma_out_of_range(s->ip_gma, gma_head, gma_tail)) {
-				gvt_err("ip_gma %lx out of range."
+				gvt_vgpu_err("ip_gma %lx out of range."
 					"base 0x%lx head 0x%lx tail 0x%lx\n",
 					s->ip_gma, rb_start,
 					rb_head, rb_tail);
@@ -2515,7 +2547,7 @@ static int command_scan(struct parser_exec_state *s,
 		}
 		ret = cmd_parser_exec(s);
 		if (ret) {
-			gvt_err("cmd parser error\n");
+			gvt_vgpu_err("cmd parser error\n");
 			parser_exec_state_dump(s);
 			break;
 		}
@@ -2639,7 +2671,7 @@ static int shadow_workload_ring_buffer(struct intel_vgpu_workload *workload)
 				gma_head, gma_top,
 				workload->shadow_ring_buffer_va);
 		if (ret) {
-			gvt_err("fail to copy guest ring buffer\n");
+			gvt_vgpu_err("fail to copy guest ring buffer\n");
 			return ret;
 		}
 		copy_len = gma_top - gma_head;
@@ -2651,7 +2683,7 @@ static int shadow_workload_ring_buffer(struct intel_vgpu_workload *workload)
 			gma_head, gma_tail,
 			workload->shadow_ring_buffer_va + copy_len);
 	if (ret) {
-		gvt_err("fail to copy guest ring buffer\n");
+		gvt_vgpu_err("fail to copy guest ring buffer\n");
 		return ret;
 	}
 	ring->tail += workload->rb_len;
@@ -2662,16 +2694,17 @@ static int shadow_workload_ring_buffer(struct intel_vgpu_workload *workload)
 int intel_gvt_scan_and_shadow_workload(struct intel_vgpu_workload *workload)
 {
 	int ret;
+	struct intel_vgpu *vgpu = workload->vgpu;
 
 	ret = shadow_workload_ring_buffer(workload);
 	if (ret) {
-		gvt_err("fail to shadow workload ring_buffer\n");
+		gvt_vgpu_err("fail to shadow workload ring_buffer\n");
 		return ret;
 	}
 
 	ret = scan_workload(workload);
 	if (ret) {
-		gvt_err("scan workload error\n");
+		gvt_vgpu_err("scan workload error\n");
 		return ret;
 	}
 	return 0;
@@ -2681,6 +2714,7 @@ static int shadow_indirect_ctx(struct intel_shadow_wa_ctx *wa_ctx)
 {
 	int ctx_size = wa_ctx->indirect_ctx.size;
 	unsigned long guest_gma = wa_ctx->indirect_ctx.guest_gma;
+	struct intel_vgpu *vgpu = wa_ctx->workload->vgpu;
 	struct drm_i915_gem_object *obj;
 	int ret = 0;
 	void *map;
@@ -2694,14 +2728,14 @@ static int shadow_indirect_ctx(struct intel_shadow_wa_ctx *wa_ctx)
 	/* get the va of the shadow batch buffer */
 	map = i915_gem_object_pin_map(obj, I915_MAP_WB);
 	if (IS_ERR(map)) {
-		gvt_err("failed to vmap shadow indirect ctx\n");
+		gvt_vgpu_err("failed to vmap shadow indirect ctx\n");
 		ret = PTR_ERR(map);
 		goto put_obj;
 	}
 
 	ret = i915_gem_object_set_to_cpu_domain(obj, false);
 	if (ret) {
-		gvt_err("failed to set shadow indirect ctx to CPU\n");
+		gvt_vgpu_err("failed to set shadow indirect ctx to CPU\n");
 		goto unmap_src;
 	}
 
@@ -2710,7 +2744,7 @@ static int shadow_indirect_ctx(struct intel_shadow_wa_ctx *wa_ctx)
 				guest_gma, guest_gma + ctx_size,
 				map);
 	if (ret) {
-		gvt_err("fail to copy guest indirect ctx\n");
+		gvt_vgpu_err("fail to copy guest indirect ctx\n");
 		goto unmap_src;
 	}
 
@@ -2744,13 +2778,14 @@ static int combine_wa_ctx(struct intel_shadow_wa_ctx *wa_ctx)
 int intel_gvt_scan_and_shadow_wa_ctx(struct intel_shadow_wa_ctx *wa_ctx)
 {
 	int ret;
+	struct intel_vgpu *vgpu = wa_ctx->workload->vgpu;
 
 	if (wa_ctx->indirect_ctx.size == 0)
 		return 0;
 
 	ret = shadow_indirect_ctx(wa_ctx);
 	if (ret) {
-		gvt_err("fail to shadow indirect ctx\n");
+		gvt_vgpu_err("fail to shadow indirect ctx\n");
 		return ret;
 	}
 
@@ -2758,7 +2793,7 @@ int intel_gvt_scan_and_shadow_wa_ctx(struct intel_shadow_wa_ctx *wa_ctx)
 
 	ret = scan_wa_ctx(wa_ctx);
 	if (ret) {
-		gvt_err("scan wa ctx error\n");
+		gvt_vgpu_err("scan wa ctx error\n");
 		return ret;
 	}
 
diff --git a/drivers/gpu/drm/i915/gvt/debug.h b/drivers/gpu/drm/i915/gvt/debug.h
index 68cba7b..b0cff4d 100644
--- a/drivers/gpu/drm/i915/gvt/debug.h
+++ b/drivers/gpu/drm/i915/gvt/debug.h
@@ -27,6 +27,14 @@
 #define gvt_err(fmt, args...) \
 	DRM_ERROR("gvt: "fmt, ##args)
 
+#define gvt_vgpu_err(fmt, args...)					\
+do {									\
+	if (IS_ERR_OR_NULL(vgpu))					\
+		DRM_DEBUG_DRIVER("gvt: "fmt, ##args);			\
+	else								\
+		DRM_DEBUG_DRIVER("gvt: vgpu %d: "fmt, vgpu->id, ##args);\
+} while (0)
+
 #define gvt_dbg_core(fmt, args...) \
 	DRM_DEBUG_DRIVER("gvt: core: "fmt, ##args)
 
diff --git a/drivers/gpu/drm/i915/gvt/edid.c b/drivers/gpu/drm/i915/gvt/edid.c
index bda85df..f1648fe 100644
--- a/drivers/gpu/drm/i915/gvt/edid.c
+++ b/drivers/gpu/drm/i915/gvt/edid.c
@@ -52,16 +52,16 @@ static unsigned char edid_get_byte(struct intel_vgpu *vgpu)
 	unsigned char chr = 0;
 
 	if (edid->state == I2C_NOT_SPECIFIED || !edid->slave_selected) {
-		gvt_err("Driver tries to read EDID without proper sequence!\n");
+		gvt_vgpu_err("Driver tries to read EDID without proper sequence!\n");
 		return 0;
 	}
 	if (edid->current_edid_read >= EDID_SIZE) {
-		gvt_err("edid_get_byte() exceeds the size of EDID!\n");
+		gvt_vgpu_err("edid_get_byte() exceeds the size of EDID!\n");
 		return 0;
 	}
 
 	if (!edid->edid_available) {
-		gvt_err("Reading EDID but EDID is not available!\n");
+		gvt_vgpu_err("Reading EDID but EDID is not available!\n");
 		return 0;
 	}
 
@@ -72,7 +72,7 @@ static unsigned char edid_get_byte(struct intel_vgpu *vgpu)
 		chr = edid_data->edid_block[edid->current_edid_read];
 		edid->current_edid_read++;
 	} else {
-		gvt_err("No EDID available during the reading?\n");
+		gvt_vgpu_err("No EDID available during the reading?\n");
 	}
 	return chr;
 }
@@ -223,7 +223,7 @@ static int gmbus1_mmio_write(struct intel_vgpu *vgpu, unsigned int offset,
 			vgpu_vreg(vgpu, PCH_GMBUS2) |= GMBUS_ACTIVE;
 			break;
 		default:
-			gvt_err("Unknown/reserved GMBUS cycle detected!\n");
+			gvt_vgpu_err("Unknown/reserved GMBUS cycle detected!\n");
 			break;
 		}
 		/*
@@ -292,8 +292,7 @@ static int gmbus3_mmio_read(struct intel_vgpu *vgpu, unsigned int offset,
 		 */
 	} else {
 		memcpy(p_data, &vgpu_vreg(vgpu, offset), bytes);
-		gvt_err("vgpu%d: warning: gmbus3 read with nothing returned\n",
-				vgpu->id);
+		gvt_vgpu_err("warning: gmbus3 read with nothing returned\n");
 	}
 	return 0;
 }
diff --git a/drivers/gpu/drm/i915/gvt/execlist.c b/drivers/gpu/drm/i915/gvt/execlist.c
index 46eb9fd..f1f426a 100644
--- a/drivers/gpu/drm/i915/gvt/execlist.c
+++ b/drivers/gpu/drm/i915/gvt/execlist.c
@@ -172,6 +172,7 @@ static int emulate_execlist_ctx_schedule_out(
 		struct intel_vgpu_execlist *execlist,
 		struct execlist_ctx_descriptor_format *ctx)
 {
+	struct intel_vgpu *vgpu = execlist->vgpu;
 	struct intel_vgpu_execlist_slot *running = execlist->running_slot;
 	struct intel_vgpu_execlist_slot *pending = execlist->pending_slot;
 	struct execlist_ctx_descriptor_format *ctx0 = &running->ctx[0];
@@ -183,7 +184,7 @@ static int emulate_execlist_ctx_schedule_out(
 	gvt_dbg_el("schedule out context id %x\n", ctx->context_id);
 
 	if (WARN_ON(!same_context(ctx, execlist->running_context))) {
-		gvt_err("schedule out context is not running context,"
+		gvt_vgpu_err("schedule out context is not running context,"
 				"ctx id %x running ctx id %x\n",
 				ctx->context_id,
 				execlist->running_context->context_id);
@@ -254,7 +255,7 @@ static struct intel_vgpu_execlist_slot *get_next_execlist_slot(
 	status.udw = vgpu_vreg(vgpu, status_reg + 4);
 
 	if (status.execlist_queue_full) {
-		gvt_err("virtual execlist slots are full\n");
+		gvt_vgpu_err("virtual execlist slots are full\n");
 		return NULL;
 	}
 
@@ -270,11 +271,12 @@ static int emulate_execlist_schedule_in(struct intel_vgpu_execlist *execlist,
 
 	struct execlist_ctx_descriptor_format *ctx0, *ctx1;
 	struct execlist_context_status_format status;
+	struct intel_vgpu *vgpu = execlist->vgpu;
 
 	gvt_dbg_el("emulate schedule-in\n");
 
 	if (!slot) {
-		gvt_err("no available execlist slot\n");
+		gvt_vgpu_err("no available execlist slot\n");
 		return -EINVAL;
 	}
 
@@ -375,7 +377,6 @@ static void prepare_shadow_batch_buffer(struct intel_vgpu_workload *workload)
 
 		vma = i915_gem_object_ggtt_pin(entry_obj->obj, NULL, 0, 4, 0);
 		if (IS_ERR(vma)) {
-			gvt_err("Cannot pin\n");
 			return;
 		}
 
@@ -428,7 +429,6 @@ static void prepare_shadow_wa_ctx(struct intel_shadow_wa_ctx *wa_ctx)
 	vma = i915_gem_object_ggtt_pin(wa_ctx->indirect_ctx.obj, NULL,
 				       0, CACHELINE_BYTES, 0);
 	if (IS_ERR(vma)) {
-		gvt_err("Cannot pin indirect ctx obj\n");
 		return;
 	}
 
@@ -561,6 +561,7 @@ static int prepare_mm(struct intel_vgpu_workload *workload)
 {
 	struct execlist_ctx_descriptor_format *desc = &workload->ctx_desc;
 	struct intel_vgpu_mm *mm;
+	struct intel_vgpu *vgpu = workload->vgpu;
 	int page_table_level;
 	u32 pdp[8];
 
@@ -569,7 +570,7 @@ static int prepare_mm(struct intel_vgpu_workload *workload)
 	} else if (desc->addressing_mode == 3) { /* legacy 64 bit */
 		page_table_level = 4;
 	} else {
-		gvt_err("Advanced Context mode(SVM) is not supported!\n");
+		gvt_vgpu_err("Advanced Context mode(SVM) is not supported!\n");
 		return -EINVAL;
 	}
 
@@ -583,7 +584,7 @@ static int prepare_mm(struct intel_vgpu_workload *workload)
 		mm = intel_vgpu_create_mm(workload->vgpu, INTEL_GVT_MM_PPGTT,
 				pdp, page_table_level, 0);
 		if (IS_ERR(mm)) {
-			gvt_err("fail to create mm object.\n");
+			gvt_vgpu_err("fail to create mm object.\n");
 			return PTR_ERR(mm);
 		}
 	}
@@ -609,7 +610,7 @@ static int submit_context(struct intel_vgpu *vgpu, int ring_id,
 	ring_context_gpa = intel_vgpu_gma_to_gpa(vgpu->gtt.ggtt_mm,
 			(u32)((desc->lrca + 1) << GTT_PAGE_SHIFT));
 	if (ring_context_gpa == INTEL_GVT_INVALID_ADDR) {
-		gvt_err("invalid guest context LRCA: %x\n", desc->lrca);
+		gvt_vgpu_err("invalid guest context LRCA: %x\n", desc->lrca);
 		return -EINVAL;
 	}
 
@@ -724,8 +725,7 @@ int intel_vgpu_submit_execlist(struct intel_vgpu *vgpu, int ring_id)
 			continue;
 
 		if (!desc[i]->privilege_access) {
-			gvt_err("vgpu%d: unexpected GGTT elsp submission\n",
-					vgpu->id);
+			gvt_vgpu_err("unexpected GGTT elsp submission\n");
 			return -EINVAL;
 		}
 
@@ -735,15 +735,13 @@ int intel_vgpu_submit_execlist(struct intel_vgpu *vgpu, int ring_id)
 	}
 
 	if (!valid_desc_bitmap) {
-		gvt_err("vgpu%d: no valid desc in a elsp submission\n",
-				vgpu->id);
+		gvt_vgpu_err("no valid desc in a elsp submission\n");
 		return -EINVAL;
 	}
 
 	if (!test_bit(0, (void *)&valid_desc_bitmap) &&
 			test_bit(1, (void *)&valid_desc_bitmap)) {
-		gvt_err("vgpu%d: weird elsp submission, desc 0 is not valid\n",
-				vgpu->id);
+		gvt_vgpu_err("weird elsp submission, desc 0 is not valid\n");
 		return -EINVAL;
 	}
 
@@ -752,8 +750,7 @@ int intel_vgpu_submit_execlist(struct intel_vgpu *vgpu, int ring_id)
 		ret = submit_context(vgpu, ring_id, &valid_desc[i],
 				emulate_schedule_in);
 		if (ret) {
-			gvt_err("vgpu%d: fail to schedule workload\n",
-					vgpu->id);
+			gvt_vgpu_err("fail to schedule workload\n");
 			return ret;
 		}
 		emulate_schedule_in = false;
diff --git a/drivers/gpu/drm/i915/gvt/gtt.c b/drivers/gpu/drm/i915/gvt/gtt.c
index 6a5ff23..da73127 100644
--- a/drivers/gpu/drm/i915/gvt/gtt.c
+++ b/drivers/gpu/drm/i915/gvt/gtt.c
@@ -49,8 +49,8 @@ bool intel_gvt_ggtt_validate_range(struct intel_vgpu *vgpu, u64 addr, u32 size)
 {
 	if ((!vgpu_gmadr_is_valid(vgpu, addr)) || (size
 			&& !vgpu_gmadr_is_valid(vgpu, addr + size - 1))) {
-		gvt_err("vgpu%d: invalid range gmadr 0x%llx size 0x%x\n",
-				vgpu->id, addr, size);
+		gvt_vgpu_err("invalid range gmadr 0x%llx size 0x%x\n",
+				addr, size);
 		return false;
 	}
 	return true;
@@ -430,7 +430,7 @@ static int gtt_entry_p2m(struct intel_vgpu *vgpu, struct intel_gvt_gtt_entry *p,
 
 	mfn = intel_gvt_hypervisor_gfn_to_mfn(vgpu, gfn);
 	if (mfn == INTEL_GVT_INVALID_ADDR) {
-		gvt_err("fail to translate gfn: 0x%lx\n", gfn);
+		gvt_vgpu_err("fail to translate gfn: 0x%lx\n", gfn);
 		return -ENXIO;
 	}
 
@@ -611,7 +611,7 @@ static inline int init_shadow_page(struct intel_vgpu *vgpu,
 
 	daddr = dma_map_page(kdev, p->page, 0, 4096, PCI_DMA_BIDIRECTIONAL);
 	if (dma_mapping_error(kdev, daddr)) {
-		gvt_err("fail to map dma addr\n");
+		gvt_vgpu_err("fail to map dma addr\n");
 		return -EINVAL;
 	}
 
@@ -735,7 +735,7 @@ retry:
 		if (reclaim_one_mm(vgpu->gvt))
 			goto retry;
 
-		gvt_err("fail to allocate ppgtt shadow page\n");
+		gvt_vgpu_err("fail to allocate ppgtt shadow page\n");
 		return ERR_PTR(-ENOMEM);
 	}
 
@@ -750,14 +750,14 @@ retry:
 	 */
 	ret = init_shadow_page(vgpu, &spt->shadow_page, type);
 	if (ret) {
-		gvt_err("fail to initialize shadow page for spt\n");
+		gvt_vgpu_err("fail to initialize shadow page for spt\n");
 		goto err;
 	}
 
 	ret = intel_vgpu_init_guest_page(vgpu, &spt->guest_page,
 			gfn, ppgtt_write_protection_handler, NULL);
 	if (ret) {
-		gvt_err("fail to initialize guest page for spt\n");
+		gvt_vgpu_err("fail to initialize guest page for spt\n");
 		goto err;
 	}
 
@@ -776,8 +776,7 @@ static struct intel_vgpu_ppgtt_spt *ppgtt_find_shadow_page(
 	if (p)
 		return shadow_page_to_ppgtt_spt(p);
 
-	gvt_err("vgpu%d: fail to find ppgtt shadow page: 0x%lx\n",
-			vgpu->id, mfn);
+	gvt_vgpu_err("fail to find ppgtt shadow page: 0x%lx\n", mfn);
 	return NULL;
 }
 
@@ -827,8 +826,8 @@ static int ppgtt_invalidate_shadow_page_by_shadow_entry(struct intel_vgpu *vgpu,
 	}
 	s = ppgtt_find_shadow_page(vgpu, ops->get_pfn(e));
 	if (!s) {
-		gvt_err("vgpu%d: fail to find shadow page: mfn: 0x%lx\n",
-				vgpu->id, ops->get_pfn(e));
+		gvt_vgpu_err("fail to find shadow page: mfn: 0x%lx\n",
+				ops->get_pfn(e));
 		return -ENXIO;
 	}
 	return ppgtt_invalidate_shadow_page(s);
@@ -836,6 +835,7 @@ static int ppgtt_invalidate_shadow_page_by_shadow_entry(struct intel_vgpu *vgpu,
 
 static int ppgtt_invalidate_shadow_page(struct intel_vgpu_ppgtt_spt *spt)
 {
+	struct intel_vgpu *vgpu = spt->vgpu;
 	struct intel_gvt_gtt_entry e;
 	unsigned long index;
 	int ret;
@@ -854,7 +854,7 @@ static int ppgtt_invalidate_shadow_page(struct intel_vgpu_ppgtt_spt *spt)
 
 	for_each_present_shadow_entry(spt, &e, index) {
 		if (!gtt_type_is_pt(get_next_pt_type(e.type))) {
-			gvt_err("GVT doesn't support pse bit for now\n");
+			gvt_vgpu_err("GVT doesn't support pse bit for now\n");
 			return -EINVAL;
 		}
 		ret = ppgtt_invalidate_shadow_page_by_shadow_entry(
@@ -868,8 +868,8 @@ release:
 	ppgtt_free_shadow_page(spt);
 	return 0;
 fail:
-	gvt_err("vgpu%d: fail: shadow page %p shadow entry 0x%llx type %d\n",
-			spt->vgpu->id, spt, e.val64, e.type);
+	gvt_vgpu_err("fail: shadow page %p shadow entry 0x%llx type %d\n",
+			spt, e.val64, e.type);
 	return ret;
 }
 
@@ -914,8 +914,8 @@ static struct intel_vgpu_ppgtt_spt *ppgtt_populate_shadow_page_by_guest_entry(
 	}
 	return s;
 fail:
-	gvt_err("vgpu%d: fail: shadow page %p guest entry 0x%llx type %d\n",
-			vgpu->id, s, we->val64, we->type);
+	gvt_vgpu_err("fail: shadow page %p guest entry 0x%llx type %d\n",
+			s, we->val64, we->type);
 	return ERR_PTR(ret);
 }
 
@@ -953,7 +953,7 @@ static int ppgtt_populate_shadow_page(struct intel_vgpu_ppgtt_spt *spt)
 
 	for_each_present_guest_entry(spt, &ge, i) {
 		if (!gtt_type_is_pt(get_next_pt_type(ge.type))) {
-			gvt_err("GVT doesn't support pse bit now\n");
+			gvt_vgpu_err("GVT doesn't support pse bit now\n");
 			ret = -EINVAL;
 			goto fail;
 		}
@@ -969,8 +969,8 @@ static int ppgtt_populate_shadow_page(struct intel_vgpu_ppgtt_spt *spt)
 	}
 	return 0;
 fail:
-	gvt_err("vgpu%d: fail: shadow page %p guest entry 0x%llx type %d\n",
-			vgpu->id, spt, ge.val64, ge.type);
+	gvt_vgpu_err("fail: shadow page %p guest entry 0x%llx type %d\n",
+			spt, ge.val64, ge.type);
 	return ret;
 }
 
@@ -999,7 +999,7 @@ static int ppgtt_handle_guest_entry_removal(struct intel_vgpu_guest_page *gpt,
 		struct intel_vgpu_ppgtt_spt *s =
 			ppgtt_find_shadow_page(vgpu, ops->get_pfn(&e));
 		if (!s) {
-			gvt_err("fail to find guest page\n");
+			gvt_vgpu_err("fail to find guest page\n");
 			ret = -ENXIO;
 			goto fail;
 		}
@@ -1011,8 +1011,8 @@ static int ppgtt_handle_guest_entry_removal(struct intel_vgpu_guest_page *gpt,
 	ppgtt_set_shadow_entry(spt, &e, index);
 	return 0;
 fail:
-	gvt_err("vgpu%d: fail: shadow page %p guest entry 0x%llx type %d\n",
-			vgpu->id, spt, e.val64, e.type);
+	gvt_vgpu_err("fail: shadow page %p guest entry 0x%llx type %d\n",
+			spt, e.val64, e.type);
 	return ret;
 }
 
@@ -1046,8 +1046,8 @@ static int ppgtt_handle_guest_entry_add(struct intel_vgpu_guest_page *gpt,
 	}
 	return 0;
 fail:
-	gvt_err("vgpu%d: fail: spt %p guest entry 0x%llx type %d\n", vgpu->id,
-			spt, we->val64, we->type);
+	gvt_vgpu_err("fail: spt %p guest entry 0x%llx type %d\n",
+		spt, we->val64, we->type);
 	return ret;
 }
 
@@ -1250,8 +1250,8 @@ static int ppgtt_handle_guest_write_page_table(
 	}
 	return 0;
 fail:
-	gvt_err("vgpu%d: fail: shadow page %p guest entry 0x%llx type %d.\n",
-			vgpu->id, spt, we->val64, we->type);
+	gvt_vgpu_err("fail: shadow page %p guest entry 0x%llx type %d.\n",
+			spt, we->val64, we->type);
 	return ret;
 }
 
@@ -1493,7 +1493,7 @@ static int shadow_mm(struct intel_vgpu_mm *mm)
 
 		spt = ppgtt_populate_shadow_page_by_guest_entry(vgpu, &ge);
 		if (IS_ERR(spt)) {
-			gvt_err("fail to populate guest root pointer\n");
+			gvt_vgpu_err("fail to populate guest root pointer\n");
 			ret = PTR_ERR(spt);
 			goto fail;
 		}
@@ -1566,7 +1566,7 @@ struct intel_vgpu_mm *intel_vgpu_create_mm(struct intel_vgpu *vgpu,
 
 	ret = gtt->mm_alloc_page_table(mm);
 	if (ret) {
-		gvt_err("fail to allocate page table for mm\n");
+		gvt_vgpu_err("fail to allocate page table for mm\n");
 		goto fail;
 	}
 
@@ -1584,7 +1584,7 @@ struct intel_vgpu_mm *intel_vgpu_create_mm(struct intel_vgpu *vgpu,
 	}
 	return mm;
 fail:
-	gvt_err("fail to create mm\n");
+	gvt_vgpu_err("fail to create mm\n");
 	if (mm)
 		intel_gvt_mm_unreference(mm);
 	return ERR_PTR(ret);
@@ -1760,7 +1760,7 @@ unsigned long intel_vgpu_gma_to_gpa(struct intel_vgpu_mm *mm, unsigned long gma)
 			mm->page_table_level, gma, gpa);
 	return gpa;
 err:
-	gvt_err("invalid mm type: %d gma %lx\n", mm->type, gma);
+	gvt_vgpu_err("invalid mm type: %d gma %lx\n", mm->type, gma);
 	return INTEL_GVT_INVALID_ADDR;
 }
 
@@ -1836,8 +1836,7 @@ static int emulate_gtt_mmio_write(struct intel_vgpu *vgpu, unsigned int off,
 	if (ops->test_present(&e)) {
 		ret = gtt_entry_p2m(vgpu, &e, &m);
 		if (ret) {
-			gvt_err("vgpu%d: fail to translate guest gtt entry\n",
-					vgpu->id);
+			gvt_vgpu_err("fail to translate guest gtt entry\n");
 			return ret;
 		}
 	} else {
@@ -1893,14 +1892,14 @@ static int alloc_scratch_pages(struct intel_vgpu *vgpu,
 
 	scratch_pt = (void *)get_zeroed_page(GFP_KERNEL);
 	if (!scratch_pt) {
-		gvt_err("fail to allocate scratch page\n");
+		gvt_vgpu_err("fail to allocate scratch page\n");
 		return -ENOMEM;
 	}
 
 	daddr = dma_map_page(dev, virt_to_page(scratch_pt), 0,
 			4096, PCI_DMA_BIDIRECTIONAL);
 	if (dma_mapping_error(dev, daddr)) {
-		gvt_err("fail to dmamap scratch_pt\n");
+		gvt_vgpu_err("fail to dmamap scratch_pt\n");
 		__free_page(virt_to_page(scratch_pt));
 		return -ENOMEM;
 	}
@@ -2003,7 +2002,7 @@ int intel_vgpu_init_gtt(struct intel_vgpu *vgpu)
 	ggtt_mm = intel_vgpu_create_mm(vgpu, INTEL_GVT_MM_GGTT,
 			NULL, 1, 0);
 	if (IS_ERR(ggtt_mm)) {
-		gvt_err("fail to create mm for ggtt.\n");
+		gvt_vgpu_err("fail to create mm for ggtt.\n");
 		return PTR_ERR(ggtt_mm);
 	}
 
@@ -2076,7 +2075,6 @@ static int setup_spt_oos(struct intel_gvt *gvt)
 	for (i = 0; i < preallocated_oos_pages; i++) {
 		oos_page = kzalloc(sizeof(*oos_page), GFP_KERNEL);
 		if (!oos_page) {
-			gvt_err("fail to pre-allocate oos page\n");
 			ret = -ENOMEM;
 			goto fail;
 		}
@@ -2166,7 +2164,7 @@ int intel_vgpu_g2v_create_ppgtt_mm(struct intel_vgpu *vgpu,
 		mm = intel_vgpu_create_mm(vgpu, INTEL_GVT_MM_PPGTT,
 				pdp, page_table_level, 0);
 		if (IS_ERR(mm)) {
-			gvt_err("fail to create mm\n");
+			gvt_vgpu_err("fail to create mm\n");
 			return PTR_ERR(mm);
 		}
 	}
@@ -2196,7 +2194,7 @@ int intel_vgpu_g2v_destroy_ppgtt_mm(struct intel_vgpu *vgpu,
 
 	mm = intel_vgpu_find_ppgtt_mm(vgpu, page_table_level, pdp);
 	if (!mm) {
-		gvt_err("fail to find ppgtt instance.\n");
+		gvt_vgpu_err("fail to find ppgtt instance.\n");
 		return -EINVAL;
 	}
 	intel_gvt_mm_unreference(mm);
diff --git a/drivers/gpu/drm/i915/gvt/gvt.h b/drivers/gpu/drm/i915/gvt/gvt.h
index 2379192..6dfc48b 100644
--- a/drivers/gpu/drm/i915/gvt/gvt.h
+++ b/drivers/gpu/drm/i915/gvt/gvt.h
@@ -162,7 +162,6 @@ struct intel_vgpu {
 	atomic_t running_workload_num;
 	DECLARE_BITMAP(tlb_handle_pending, I915_NUM_ENGINES);
 	struct i915_gem_context *shadow_ctx;
-	struct notifier_block shadow_ctx_notifier_block;
 
 #if IS_ENABLED(CONFIG_DRM_I915_GVT_KVMGT)
 	struct {
@@ -233,6 +232,7 @@ struct intel_gvt {
 	struct intel_gvt_gtt gtt;
 	struct intel_gvt_opregion opregion;
 	struct intel_gvt_workload_scheduler scheduler;
+	struct notifier_block shadow_ctx_notifier_block[I915_NUM_ENGINES];
 	DECLARE_HASHTABLE(cmd_table, GVT_CMD_HASH_BITS);
 	struct intel_vgpu_type *types;
 	unsigned int num_types;
diff --git a/drivers/gpu/drm/i915/gvt/handlers.c b/drivers/gpu/drm/i915/gvt/handlers.c
index 8e43395..eaff45d 100644
--- a/drivers/gpu/drm/i915/gvt/handlers.c
+++ b/drivers/gpu/drm/i915/gvt/handlers.c
@@ -181,11 +181,9 @@ static int sanitize_fence_mmio_access(struct intel_vgpu *vgpu,
 					GVT_FAILSAFE_UNSUPPORTED_GUEST);
 
 		if (!vgpu->mmio.disable_warn_untrack) {
-			gvt_err("vgpu%d: found oob fence register access\n",
-					vgpu->id);
-			gvt_err("vgpu%d: total fence %d, access fence %d\n",
-					vgpu->id, vgpu_fence_sz(vgpu),
-					fence_num);
+			gvt_vgpu_err("found oob fence register access\n");
+			gvt_vgpu_err("total fence %d, access fence %d\n",
+					vgpu_fence_sz(vgpu), fence_num);
 		}
 		memset(p_data, 0, bytes);
 		return -EINVAL;
@@ -249,7 +247,7 @@ static int mul_force_wake_write(struct intel_vgpu *vgpu,
 			break;
 		default:
 			/*should not hit here*/
-			gvt_err("invalid forcewake offset 0x%x\n", offset);
+			gvt_vgpu_err("invalid forcewake offset 0x%x\n", offset);
 			return -EINVAL;
 		}
 	} else {
@@ -530,7 +528,7 @@ static int check_fdi_rx_train_status(struct intel_vgpu *vgpu,
 		fdi_tx_train_bits = FDI_LINK_TRAIN_PATTERN_2;
 		fdi_iir_check_bits = FDI_RX_SYMBOL_LOCK;
 	} else {
-		gvt_err("Invalid train pattern %d\n", train_pattern);
+		gvt_vgpu_err("Invalid train pattern %d\n", train_pattern);
 		return -EINVAL;
 	}
 
@@ -588,7 +586,7 @@ static int update_fdi_rx_iir_status(struct intel_vgpu *vgpu,
 	else if (FDI_RX_IMR_TO_PIPE(offset) != INVALID_INDEX)
 		index = FDI_RX_IMR_TO_PIPE(offset);
 	else {
-		gvt_err("Unsupport registers %x\n", offset);
+		gvt_vgpu_err("Unsupport registers %x\n", offset);
 		return -EINVAL;
 	}
 
@@ -818,7 +816,7 @@ static int dp_aux_ch_ctl_mmio_write(struct intel_vgpu *vgpu,
 	u32 data;
 
 	if (!dpy_is_valid_port(port_index)) {
-		gvt_err("GVT(%d): Unsupported DP port access!\n", vgpu->id);
+		gvt_vgpu_err("Unsupported DP port access!\n");
 		return 0;
 	}
 
@@ -1016,8 +1014,7 @@ static void write_virtual_sbi_register(struct intel_vgpu *vgpu,
 
 	if (i == num) {
 		if (num == SBI_REG_MAX) {
-			gvt_err("vgpu%d: SBI caching meets maximum limits\n",
-					vgpu->id);
+			gvt_vgpu_err("SBI caching meets maximum limits\n");
 			return;
 		}
 		display->sbi.number++;
@@ -1097,7 +1094,7 @@ static int pvinfo_mmio_read(struct intel_vgpu *vgpu, unsigned int offset,
 		break;
 	}
 	if (invalid_read)
-		gvt_err("invalid pvinfo read: [%x:%x] = %x\n",
+		gvt_vgpu_err("invalid pvinfo read: [%x:%x] = %x\n",
 				offset, bytes, *(u32 *)p_data);
 	vgpu->pv_notified = true;
 	return 0;
@@ -1125,7 +1122,7 @@ static int handle_g2v_notification(struct intel_vgpu *vgpu, int notification)
 	case 1:	/* Remove this in guest driver. */
 		break;
 	default:
-		gvt_err("Invalid PV notification %d\n", notification);
+		gvt_vgpu_err("Invalid PV notification %d\n", notification);
 	}
 	return ret;
 }
@@ -1181,7 +1178,7 @@ static int pvinfo_mmio_write(struct intel_vgpu *vgpu, unsigned int offset,
 		enter_failsafe_mode(vgpu, GVT_FAILSAFE_INSUFFICIENT_RESOURCE);
 		break;
 	default:
-		gvt_err("invalid pvinfo write offset %x bytes %x data %x\n",
+		gvt_vgpu_err("invalid pvinfo write offset %x bytes %x data %x\n",
 				offset, bytes, data);
 		break;
 	}
@@ -1415,7 +1412,8 @@ static int elsp_mmio_write(struct intel_vgpu *vgpu, unsigned int offset,
 	if (execlist->elsp_dwords.index == 3) {
 		ret = intel_vgpu_submit_execlist(vgpu, ring_id);
 		if(ret)
-			gvt_err("fail submit workload on ring %d\n", ring_id);
+			gvt_vgpu_err("fail submit workload on ring %d\n",
+				ring_id);
 	}
 
 	++execlist->elsp_dwords.index;
@@ -2988,3 +2986,20 @@ int intel_vgpu_default_mmio_write(struct intel_vgpu *vgpu, unsigned int offset,
 	write_vreg(vgpu, offset, p_data, bytes);
 	return 0;
 }
+
+/**
+ * intel_gvt_in_force_nonpriv_whitelist - if a mmio is in whitelist to be
+ * force-nopriv register
+ *
+ * @gvt: a GVT device
+ * @offset: register offset
+ *
+ * Returns:
+ * True if the register is in force-nonpriv whitelist;
+ * False if outside;
+ */
+bool intel_gvt_in_force_nonpriv_whitelist(struct intel_gvt *gvt,
+					  unsigned int offset)
+{
+	return in_whitelist(offset);
+}
diff --git a/drivers/gpu/drm/i915/gvt/kvmgt.c b/drivers/gpu/drm/i915/gvt/kvmgt.c
index 84d8016..1ea3eb2 100644
--- a/drivers/gpu/drm/i915/gvt/kvmgt.c
+++ b/drivers/gpu/drm/i915/gvt/kvmgt.c
@@ -426,7 +426,7 @@ static void kvmgt_protect_table_del(struct kvmgt_guest_info *info,
 
 static int intel_vgpu_create(struct kobject *kobj, struct mdev_device *mdev)
 {
-	struct intel_vgpu *vgpu;
+	struct intel_vgpu *vgpu = NULL;
 	struct intel_vgpu_type *type;
 	struct device *pdev;
 	void *gvt;
@@ -437,7 +437,7 @@ static int intel_vgpu_create(struct kobject *kobj, struct mdev_device *mdev)
 
 	type = intel_gvt_find_vgpu_type(gvt, kobject_name(kobj));
 	if (!type) {
-		gvt_err("failed to find type %s to create\n",
+		gvt_vgpu_err("failed to find type %s to create\n",
 						kobject_name(kobj));
 		ret = -EINVAL;
 		goto out;
@@ -446,7 +446,7 @@ static int intel_vgpu_create(struct kobject *kobj, struct mdev_device *mdev)
 	vgpu = intel_gvt_ops->vgpu_create(gvt, type);
 	if (IS_ERR_OR_NULL(vgpu)) {
 		ret = vgpu == NULL ? -EFAULT : PTR_ERR(vgpu);
-		gvt_err("failed to create intel vgpu: %d\n", ret);
+		gvt_vgpu_err("failed to create intel vgpu: %d\n", ret);
 		goto out;
 	}
 
@@ -526,7 +526,8 @@ static int intel_vgpu_open(struct mdev_device *mdev)
 	ret = vfio_register_notifier(mdev_dev(mdev), VFIO_IOMMU_NOTIFY, &events,
 				&vgpu->vdev.iommu_notifier);
 	if (ret != 0) {
-		gvt_err("vfio_register_notifier for iommu failed: %d\n", ret);
+		gvt_vgpu_err("vfio_register_notifier for iommu failed: %d\n",
+			ret);
 		goto out;
 	}
 
@@ -534,7 +535,8 @@ static int intel_vgpu_open(struct mdev_device *mdev)
 	ret = vfio_register_notifier(mdev_dev(mdev), VFIO_GROUP_NOTIFY, &events,
 				&vgpu->vdev.group_notifier);
 	if (ret != 0) {
-		gvt_err("vfio_register_notifier for group failed: %d\n", ret);
+		gvt_vgpu_err("vfio_register_notifier for group failed: %d\n",
+			ret);
 		goto undo_iommu;
 	}
 
@@ -635,7 +637,7 @@ static ssize_t intel_vgpu_rw(struct mdev_device *mdev, char *buf,
 
 
 	if (index >= VFIO_PCI_NUM_REGIONS) {
-		gvt_err("invalid index: %u\n", index);
+		gvt_vgpu_err("invalid index: %u\n", index);
 		return -EINVAL;
 	}
 
@@ -669,7 +671,7 @@ static ssize_t intel_vgpu_rw(struct mdev_device *mdev, char *buf,
 	case VFIO_PCI_VGA_REGION_INDEX:
 	case VFIO_PCI_ROM_REGION_INDEX:
 	default:
-		gvt_err("unsupported region: %u\n", index);
+		gvt_vgpu_err("unsupported region: %u\n", index);
 	}
 
 	return ret == 0 ? count : ret;
@@ -861,7 +863,7 @@ static int intel_vgpu_set_msi_trigger(struct intel_vgpu *vgpu,
 
 		trigger = eventfd_ctx_fdget(fd);
 		if (IS_ERR(trigger)) {
-			gvt_err("eventfd_ctx_fdget failed\n");
+			gvt_vgpu_err("eventfd_ctx_fdget failed\n");
 			return PTR_ERR(trigger);
 		}
 		vgpu->vdev.msi_trigger = trigger;
@@ -1120,7 +1122,7 @@ static long intel_vgpu_ioctl(struct mdev_device *mdev, unsigned int cmd,
 			ret = vfio_set_irqs_validate_and_prepare(&hdr, max,
 						VFIO_PCI_NUM_IRQS, &data_size);
 			if (ret) {
-				gvt_err("intel:vfio_set_irqs_validate_and_prepare failed\n");
+				gvt_vgpu_err("intel:vfio_set_irqs_validate_and_prepare failed\n");
 				return -EINVAL;
 			}
 			if (data_size) {
@@ -1310,7 +1312,7 @@ static int kvmgt_guest_init(struct mdev_device *mdev)
 
 	kvm = vgpu->vdev.kvm;
 	if (!kvm || kvm->mm != current->mm) {
-		gvt_err("KVM is required to use Intel vGPU\n");
+		gvt_vgpu_err("KVM is required to use Intel vGPU\n");
 		return -ESRCH;
 	}
 
@@ -1337,8 +1339,10 @@ static int kvmgt_guest_init(struct mdev_device *mdev)
 
 static bool kvmgt_guest_exit(struct kvmgt_guest_info *info)
 {
+	struct intel_vgpu *vgpu = info->vgpu;
+
 	if (!info) {
-		gvt_err("kvmgt_guest_info invalid\n");
+		gvt_vgpu_err("kvmgt_guest_info invalid\n");
 		return false;
 	}
 
@@ -1383,12 +1387,14 @@ static unsigned long kvmgt_gfn_to_pfn(unsigned long handle, unsigned long gfn)
 	unsigned long iova, pfn;
 	struct kvmgt_guest_info *info;
 	struct device *dev;
+	struct intel_vgpu *vgpu;
 	int rc;
 
 	if (!handle_valid(handle))
 		return INTEL_GVT_INVALID_ADDR;
 
 	info = (struct kvmgt_guest_info *)handle;
+	vgpu = info->vgpu;
 	iova = gvt_cache_find(info->vgpu, gfn);
 	if (iova != INTEL_GVT_INVALID_ADDR)
 		return iova;
@@ -1397,13 +1403,14 @@ static unsigned long kvmgt_gfn_to_pfn(unsigned long handle, unsigned long gfn)
 	dev = mdev_dev(info->vgpu->vdev.mdev);
 	rc = vfio_pin_pages(dev, &gfn, 1, IOMMU_READ | IOMMU_WRITE, &pfn);
 	if (rc != 1) {
-		gvt_err("vfio_pin_pages failed for gfn 0x%lx: %d\n", gfn, rc);
+		gvt_vgpu_err("vfio_pin_pages failed for gfn 0x%lx: %d\n",
+			gfn, rc);
 		return INTEL_GVT_INVALID_ADDR;
 	}
 	/* transfer to host iova for GFX to use DMA */
 	rc = gvt_dma_map_iova(info->vgpu, pfn, &iova);
 	if (rc) {
-		gvt_err("gvt_dma_map_iova failed for gfn: 0x%lx\n", gfn);
+		gvt_vgpu_err("gvt_dma_map_iova failed for gfn: 0x%lx\n", gfn);
 		vfio_unpin_pages(dev, &gfn, 1);
 		return INTEL_GVT_INVALID_ADDR;
 	}
@@ -1417,7 +1424,7 @@ static int kvmgt_rw_gpa(unsigned long handle, unsigned long gpa,
 {
 	struct kvmgt_guest_info *info;
 	struct kvm *kvm;
-	int ret;
+	int idx, ret;
 	bool kthread = current->mm == NULL;
 
 	if (!handle_valid(handle))
@@ -1429,8 +1436,10 @@ static int kvmgt_rw_gpa(unsigned long handle, unsigned long gpa,
 	if (kthread)
 		use_mm(kvm->mm);
 
+	idx = srcu_read_lock(&kvm->srcu);
 	ret = write ? kvm_write_guest(kvm, gpa, buf, len) :
 		      kvm_read_guest(kvm, gpa, buf, len);
+	srcu_read_unlock(&kvm->srcu, idx);
 
 	if (kthread)
 		unuse_mm(kvm->mm);
diff --git a/drivers/gpu/drm/i915/gvt/mmio.c b/drivers/gpu/drm/i915/gvt/mmio.c
index 60b698c..1ba3bdb 100644
--- a/drivers/gpu/drm/i915/gvt/mmio.c
+++ b/drivers/gpu/drm/i915/gvt/mmio.c
@@ -142,10 +142,10 @@ int intel_vgpu_emulate_mmio_read(struct intel_vgpu *vgpu, uint64_t pa,
 			ret = intel_gvt_hypervisor_read_gpa(vgpu, pa,
 					p_data, bytes);
 			if (ret) {
-				gvt_err("vgpu%d: guest page read error %d, "
+				gvt_vgpu_err("guest page read error %d, "
 					"gfn 0x%lx, pa 0x%llx, var 0x%x, len %d\n",
-					vgpu->id, ret,
-					gp->gfn, pa, *(u32 *)p_data, bytes);
+					ret, gp->gfn, pa, *(u32 *)p_data,
+					bytes);
 			}
 			mutex_unlock(&gvt->lock);
 			return ret;
@@ -200,14 +200,13 @@ int intel_vgpu_emulate_mmio_read(struct intel_vgpu *vgpu, uint64_t pa,
 		ret = intel_vgpu_default_mmio_read(vgpu, offset, p_data, bytes);
 
 		if (!vgpu->mmio.disable_warn_untrack) {
-			gvt_err("vgpu%d: read untracked MMIO %x(%dB) val %x\n",
-				vgpu->id, offset, bytes, *(u32 *)p_data);
+			gvt_vgpu_err("read untracked MMIO %x(%dB) val %x\n",
+				offset, bytes, *(u32 *)p_data);
 
 			if (offset == 0x206c) {
-				gvt_err("------------------------------------------\n");
-				gvt_err("vgpu%d: likely triggers a gfx reset\n",
-					vgpu->id);
-				gvt_err("------------------------------------------\n");
+				gvt_vgpu_err("------------------------------------------\n");
+				gvt_vgpu_err("likely triggers a gfx reset\n");
+				gvt_vgpu_err("------------------------------------------\n");
 				vgpu->mmio.disable_warn_untrack = true;
 			}
 		}
@@ -220,8 +219,8 @@ int intel_vgpu_emulate_mmio_read(struct intel_vgpu *vgpu, uint64_t pa,
 	mutex_unlock(&gvt->lock);
 	return 0;
 err:
-	gvt_err("vgpu%d: fail to emulate MMIO read %08x len %d\n",
-			vgpu->id, offset, bytes);
+	gvt_vgpu_err("fail to emulate MMIO read %08x len %d\n",
+			offset, bytes);
 	mutex_unlock(&gvt->lock);
 	return ret;
 }
@@ -259,10 +258,11 @@ int intel_vgpu_emulate_mmio_write(struct intel_vgpu *vgpu, uint64_t pa,
 		if (gp) {
 			ret = gp->handler(gp, pa, p_data, bytes);
 			if (ret) {
-				gvt_err("vgpu%d: guest page write error %d, "
-					"gfn 0x%lx, pa 0x%llx, var 0x%x, len %d\n",
-					vgpu->id, ret,
-					gp->gfn, pa, *(u32 *)p_data, bytes);
+				gvt_err("guest page write error %d, "
+					"gfn 0x%lx, pa 0x%llx, "
+					"var 0x%x, len %d\n",
+					ret, gp->gfn, pa,
+					*(u32 *)p_data, bytes);
 			}
 			mutex_unlock(&gvt->lock);
 			return ret;
@@ -329,8 +329,8 @@ int intel_vgpu_emulate_mmio_write(struct intel_vgpu *vgpu, uint64_t pa,
 
 			/* all register bits are RO. */
 			if (ro_mask == ~(u64)0) {
-				gvt_err("vgpu%d: try to write RO reg %x\n",
-						vgpu->id, offset);
+				gvt_vgpu_err("try to write RO reg %x\n",
+					offset);
 				ret = 0;
 				goto out;
 			}
@@ -360,8 +360,8 @@ out:
 	mutex_unlock(&gvt->lock);
 	return 0;
 err:
-	gvt_err("vgpu%d: fail to emulate MMIO write %08x len %d\n",
-			vgpu->id, offset, bytes);
+	gvt_vgpu_err("fail to emulate MMIO write %08x len %d\n", offset,
+		     bytes);
 	mutex_unlock(&gvt->lock);
 	return ret;
 }
diff --git a/drivers/gpu/drm/i915/gvt/mmio.h b/drivers/gpu/drm/i915/gvt/mmio.h
index 3bc620f..a3a0270 100644
--- a/drivers/gpu/drm/i915/gvt/mmio.h
+++ b/drivers/gpu/drm/i915/gvt/mmio.h
@@ -107,4 +107,7 @@ int intel_vgpu_default_mmio_read(struct intel_vgpu *vgpu, unsigned int offset,
 				 void *p_data, unsigned int bytes);
 int intel_vgpu_default_mmio_write(struct intel_vgpu *vgpu, unsigned int offset,
 				  void *p_data, unsigned int bytes);
+
+bool intel_gvt_in_force_nonpriv_whitelist(struct intel_gvt *gvt,
+					  unsigned int offset);
 #endif
diff --git a/drivers/gpu/drm/i915/gvt/opregion.c b/drivers/gpu/drm/i915/gvt/opregion.c
index 5d1caf9..3117991 100644
--- a/drivers/gpu/drm/i915/gvt/opregion.c
+++ b/drivers/gpu/drm/i915/gvt/opregion.c
@@ -67,14 +67,15 @@ static int map_vgpu_opregion(struct intel_vgpu *vgpu, bool map)
 		mfn = intel_gvt_hypervisor_virt_to_mfn(vgpu_opregion(vgpu)->va
 			+ i * PAGE_SIZE);
 		if (mfn == INTEL_GVT_INVALID_ADDR) {
-			gvt_err("fail to get MFN from VA\n");
+			gvt_vgpu_err("fail to get MFN from VA\n");
 			return -EINVAL;
 		}
 		ret = intel_gvt_hypervisor_map_gfn_to_mfn(vgpu,
 				vgpu_opregion(vgpu)->gfn[i],
 				mfn, 1, map);
 		if (ret) {
-			gvt_err("fail to map GFN to MFN, errno: %d\n", ret);
+			gvt_vgpu_err("fail to map GFN to MFN, errno: %d\n",
+				ret);
 			return ret;
 		}
 	}
@@ -287,7 +288,7 @@ int intel_vgpu_emulate_opregion_request(struct intel_vgpu *vgpu, u32 swsci)
 	parm = vgpu_opregion(vgpu)->va + INTEL_GVT_OPREGION_PARM;
 
 	if (!(swsci & SWSCI_SCI_SELECT)) {
-		gvt_err("vgpu%d: requesting SMI service\n", vgpu->id);
+		gvt_vgpu_err("requesting SMI service\n");
 		return 0;
 	}
 	/* ignore non 0->1 trasitions */
@@ -300,9 +301,8 @@ int intel_vgpu_emulate_opregion_request(struct intel_vgpu *vgpu, u32 swsci)
 	func = GVT_OPREGION_FUNC(*scic);
 	subfunc = GVT_OPREGION_SUBFUNC(*scic);
 	if (!querying_capabilities(*scic)) {
-		gvt_err("vgpu%d: requesting runtime service: func \"%s\","
+		gvt_vgpu_err("requesting runtime service: func \"%s\","
 				" subfunc \"%s\"\n",
-				vgpu->id,
 				opregion_func_name(func),
 				opregion_subfunc_name(subfunc));
 		/*
diff --git a/drivers/gpu/drm/i915/gvt/render.c b/drivers/gpu/drm/i915/gvt/render.c
index 73f052a..95ee091 100644
--- a/drivers/gpu/drm/i915/gvt/render.c
+++ b/drivers/gpu/drm/i915/gvt/render.c
@@ -167,7 +167,7 @@ static void handle_tlb_pending_event(struct intel_vgpu *vgpu, int ring_id)
 	I915_WRITE_FW(reg, 0x1);
 
 	if (wait_for_atomic((I915_READ_FW(reg) == 0), 50))
-		gvt_err("timeout in invalidate ring (%d) tlb\n", ring_id);
+		gvt_vgpu_err("timeout in invalidate ring (%d) tlb\n", ring_id);
 	else
 		vgpu_vreg(vgpu, regs[ring_id]) = 0;
 
diff --git a/drivers/gpu/drm/i915/gvt/sched_policy.c b/drivers/gpu/drm/i915/gvt/sched_policy.c
index 06c9584..34b9acd 100644
--- a/drivers/gpu/drm/i915/gvt/sched_policy.c
+++ b/drivers/gpu/drm/i915/gvt/sched_policy.c
@@ -101,7 +101,7 @@ struct tbs_sched_data {
 	struct list_head runq_head;
 };
 
-#define GVT_DEFAULT_TIME_SLICE (1 * HZ / 1000)
+#define GVT_DEFAULT_TIME_SLICE (msecs_to_jiffies(1))
 
 static void tbs_sched_func(struct work_struct *work)
 {
@@ -223,7 +223,7 @@ static void tbs_sched_start_schedule(struct intel_vgpu *vgpu)
 		return;
 
 	list_add_tail(&vgpu_data->list, &sched_data->runq_head);
-	schedule_delayed_work(&sched_data->work, sched_data->period);
+	schedule_delayed_work(&sched_data->work, 0);
 }
 
 static void tbs_sched_stop_schedule(struct intel_vgpu *vgpu)
diff --git a/drivers/gpu/drm/i915/gvt/scheduler.c b/drivers/gpu/drm/i915/gvt/scheduler.c
index d3a56c9..c4353ed 100644
--- a/drivers/gpu/drm/i915/gvt/scheduler.c
+++ b/drivers/gpu/drm/i915/gvt/scheduler.c
@@ -84,7 +84,7 @@ static int populate_shadow_context(struct intel_vgpu_workload *workload)
 				(u32)((workload->ctx_desc.lrca + i) <<
 				GTT_PAGE_SHIFT));
 		if (context_gpa == INTEL_GVT_INVALID_ADDR) {
-			gvt_err("Invalid guest context descriptor\n");
+			gvt_vgpu_err("Invalid guest context descriptor\n");
 			return -EINVAL;
 		}
 
@@ -130,12 +130,10 @@ static int populate_shadow_context(struct intel_vgpu_workload *workload)
 static int shadow_context_status_change(struct notifier_block *nb,
 		unsigned long action, void *data)
 {
-	struct intel_vgpu *vgpu = container_of(nb,
-			struct intel_vgpu, shadow_ctx_notifier_block);
-	struct drm_i915_gem_request *req =
-		(struct drm_i915_gem_request *)data;
-	struct intel_gvt_workload_scheduler *scheduler =
-		&vgpu->gvt->scheduler;
+	struct drm_i915_gem_request *req = (struct drm_i915_gem_request *)data;
+	struct intel_gvt *gvt = container_of(nb, struct intel_gvt,
+				shadow_ctx_notifier_block[req->engine->id]);
+	struct intel_gvt_workload_scheduler *scheduler = &gvt->scheduler;
 	struct intel_vgpu_workload *workload =
 		scheduler->current_workload[req->engine->id];
 
@@ -175,7 +173,9 @@ static int dispatch_workload(struct intel_vgpu_workload *workload)
 	int ring_id = workload->ring_id;
 	struct i915_gem_context *shadow_ctx = workload->vgpu->shadow_ctx;
 	struct drm_i915_private *dev_priv = workload->vgpu->gvt->dev_priv;
+	struct intel_engine_cs *engine = dev_priv->engine[ring_id];
 	struct drm_i915_gem_request *rq;
+	struct intel_vgpu *vgpu = workload->vgpu;
 	int ret;
 
 	gvt_dbg_sched("ring id %d prepare to dispatch workload %p\n",
@@ -187,9 +187,24 @@ static int dispatch_workload(struct intel_vgpu_workload *workload)
 
 	mutex_lock(&dev_priv->drm.struct_mutex);
 
+	/* pin shadow context by gvt even the shadow context will be pinned
+	 * when i915 alloc request. That is because gvt will update the guest
+	 * context from shadow context when workload is completed, and at that
+	 * moment, i915 may already unpined the shadow context to make the
+	 * shadow_ctx pages invalid. So gvt need to pin itself. After update
+	 * the guest context, gvt can unpin the shadow_ctx safely.
+	 */
+	ret = engine->context_pin(engine, shadow_ctx);
+	if (ret) {
+		gvt_vgpu_err("fail to pin shadow context\n");
+		workload->status = ret;
+		mutex_unlock(&dev_priv->drm.struct_mutex);
+		return ret;
+	}
+
 	rq = i915_gem_request_alloc(dev_priv->engine[ring_id], shadow_ctx);
 	if (IS_ERR(rq)) {
-		gvt_err("fail to allocate gem request\n");
+		gvt_vgpu_err("fail to allocate gem request\n");
 		ret = PTR_ERR(rq);
 		goto out;
 	}
@@ -202,9 +217,12 @@ static int dispatch_workload(struct intel_vgpu_workload *workload)
 	if (ret)
 		goto out;
 
-	ret = intel_gvt_scan_and_shadow_wa_ctx(&workload->wa_ctx);
-	if (ret)
-		goto out;
+	if ((workload->ring_id == RCS) &&
+	    (workload->wa_ctx.indirect_ctx.size != 0)) {
+		ret = intel_gvt_scan_and_shadow_wa_ctx(&workload->wa_ctx);
+		if (ret)
+			goto out;
+	}
 
 	ret = populate_shadow_context(workload);
 	if (ret)
@@ -227,6 +245,9 @@ out:
 
 	if (!IS_ERR_OR_NULL(rq))
 		i915_add_request_no_flush(rq);
+	else
+		engine->context_unpin(engine, shadow_ctx);
+
 	mutex_unlock(&dev_priv->drm.struct_mutex);
 	return ret;
 }
@@ -322,7 +343,7 @@ static void update_guest_context(struct intel_vgpu_workload *workload)
 				(u32)((workload->ctx_desc.lrca + i) <<
 					GTT_PAGE_SHIFT));
 		if (context_gpa == INTEL_GVT_INVALID_ADDR) {
-			gvt_err("invalid guest context descriptor\n");
+			gvt_vgpu_err("invalid guest context descriptor\n");
 			return;
 		}
 
@@ -376,6 +397,10 @@ static void complete_current_workload(struct intel_gvt *gvt, int ring_id)
 	 * For the workload w/o request, directly complete the workload.
 	 */
 	if (workload->req) {
+		struct drm_i915_private *dev_priv =
+			workload->vgpu->gvt->dev_priv;
+		struct intel_engine_cs *engine =
+			dev_priv->engine[workload->ring_id];
 		wait_event(workload->shadow_ctx_status_wq,
 			   !atomic_read(&workload->shadow_ctx_active));
 
@@ -388,6 +413,10 @@ static void complete_current_workload(struct intel_gvt *gvt, int ring_id)
 					 INTEL_GVT_EVENT_MAX)
 				intel_vgpu_trigger_virtual_event(vgpu, event);
 		}
+		mutex_lock(&dev_priv->drm.struct_mutex);
+		/* unpin shadow ctx as the shadow_ctx update is done */
+		engine->context_unpin(engine, workload->vgpu->shadow_ctx);
+		mutex_unlock(&dev_priv->drm.struct_mutex);
 	}
 
 	gvt_dbg_sched("ring id %d complete workload %p status %d\n",
@@ -417,6 +446,7 @@ static int workload_thread(void *priv)
 	int ring_id = p->ring_id;
 	struct intel_gvt_workload_scheduler *scheduler = &gvt->scheduler;
 	struct intel_vgpu_workload *workload = NULL;
+	struct intel_vgpu *vgpu = NULL;
 	int ret;
 	bool need_force_wake = IS_SKYLAKE(gvt->dev_priv);
 	DEFINE_WAIT_FUNC(wait, woken_wake_function);
@@ -459,25 +489,14 @@ static int workload_thread(void *priv)
 		mutex_unlock(&gvt->lock);
 
 		if (ret) {
-			gvt_err("fail to dispatch workload, skip\n");
+			vgpu = workload->vgpu;
+			gvt_vgpu_err("fail to dispatch workload, skip\n");
 			goto complete;
 		}
 
 		gvt_dbg_sched("ring id %d wait workload %p\n",
 				workload->ring_id, workload);
-retry:
-		i915_wait_request(workload->req,
-					 0, MAX_SCHEDULE_TIMEOUT);
-		/* I915 has replay mechanism and a request will be replayed
-		 * if there is i915 reset. So the seqno will be updated anyway.
-		 * If the seqno is not updated yet after waiting, which means
-		 * the replay may still be in progress and we can wait again.
-		 */
-		if (!i915_gem_request_completed(workload->req)) {
-			gvt_dbg_sched("workload %p not completed, wait again\n",
-					workload);
-			goto retry;
-		}
+		i915_wait_request(workload->req, 0, MAX_SCHEDULE_TIMEOUT);
 
 complete:
 		gvt_dbg_sched("will complete workload %p, status: %d\n",
@@ -513,15 +532,16 @@ void intel_gvt_wait_vgpu_idle(struct intel_vgpu *vgpu)
 void intel_gvt_clean_workload_scheduler(struct intel_gvt *gvt)
 {
 	struct intel_gvt_workload_scheduler *scheduler = &gvt->scheduler;
-	int i;
+	struct intel_engine_cs *engine;
+	enum intel_engine_id i;
 
 	gvt_dbg_core("clean workload scheduler\n");
 
-	for (i = 0; i < I915_NUM_ENGINES; i++) {
-		if (scheduler->thread[i]) {
-			kthread_stop(scheduler->thread[i]);
-			scheduler->thread[i] = NULL;
-		}
+	for_each_engine(engine, gvt->dev_priv, i) {
+		atomic_notifier_chain_unregister(
+					&engine->context_status_notifier,
+					&gvt->shadow_ctx_notifier_block[i]);
+		kthread_stop(scheduler->thread[i]);
 	}
 }
 
@@ -529,18 +549,15 @@ int intel_gvt_init_workload_scheduler(struct intel_gvt *gvt)
 {
 	struct intel_gvt_workload_scheduler *scheduler = &gvt->scheduler;
 	struct workload_thread_param *param = NULL;
+	struct intel_engine_cs *engine;
+	enum intel_engine_id i;
 	int ret;
-	int i;
 
 	gvt_dbg_core("init workload scheduler\n");
 
 	init_waitqueue_head(&scheduler->workload_complete_wq);
 
-	for (i = 0; i < I915_NUM_ENGINES; i++) {
-		/* check ring mask at init time */
-		if (!HAS_ENGINE(gvt->dev_priv, i))
-			continue;
-
+	for_each_engine(engine, gvt->dev_priv, i) {
 		init_waitqueue_head(&scheduler->waitq[i]);
 
 		param = kzalloc(sizeof(*param), GFP_KERNEL);
@@ -559,6 +576,11 @@ int intel_gvt_init_workload_scheduler(struct intel_gvt *gvt)
 			ret = PTR_ERR(scheduler->thread[i]);
 			goto err;
 		}
+
+		gvt->shadow_ctx_notifier_block[i].notifier_call =
+					shadow_context_status_change;
+		atomic_notifier_chain_register(&engine->context_status_notifier,
+					&gvt->shadow_ctx_notifier_block[i]);
 	}
 	return 0;
 err:
@@ -570,9 +592,6 @@ err:
 
 void intel_vgpu_clean_gvt_context(struct intel_vgpu *vgpu)
 {
-	atomic_notifier_chain_unregister(&vgpu->shadow_ctx->status_notifier,
-			&vgpu->shadow_ctx_notifier_block);
-
 	i915_gem_context_put_unlocked(vgpu->shadow_ctx);
 }
 
@@ -587,10 +606,5 @@ int intel_vgpu_init_gvt_context(struct intel_vgpu *vgpu)
 
 	vgpu->shadow_ctx->engine[RCS].initialised = true;
 
-	vgpu->shadow_ctx_notifier_block.notifier_call =
-		shadow_context_status_change;
-
-	atomic_notifier_chain_register(&vgpu->shadow_ctx->status_notifier,
-				       &vgpu->shadow_ctx_notifier_block);
 	return 0;
 }
diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index e703556..1c75402 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -248,6 +248,7 @@ static int i915_getparam(struct drm_device *dev, void *data,
 	case I915_PARAM_IRQ_ACTIVE:
 	case I915_PARAM_ALLOW_BATCHBUFFER:
 	case I915_PARAM_LAST_DISPATCH:
+	case I915_PARAM_HAS_EXEC_CONSTANTS:
 		/* Reject all old ums/dri params. */
 		return -ENODEV;
 	case I915_PARAM_CHIPSET_ID:
@@ -274,9 +275,6 @@ static int i915_getparam(struct drm_device *dev, void *data,
 	case I915_PARAM_HAS_BSD2:
 		value = !!dev_priv->engine[VCS2];
 		break;
-	case I915_PARAM_HAS_EXEC_CONSTANTS:
-		value = INTEL_GEN(dev_priv) >= 4;
-		break;
 	case I915_PARAM_HAS_LLC:
 		value = HAS_LLC(dev_priv);
 		break;
@@ -1788,7 +1786,7 @@ void i915_reset(struct drm_i915_private *dev_priv)
 		goto error;
 	}
 
-	i915_gem_reset_finish(dev_priv);
+	i915_gem_reset(dev_priv);
 	intel_overlay_reset(dev_priv);
 
 	/* Ok, now get things going again... */
@@ -1814,6 +1812,7 @@ void i915_reset(struct drm_i915_private *dev_priv)
 	i915_queue_hangcheck(dev_priv);
 
 wakeup:
+	i915_gem_reset_finish(dev_priv);
 	enable_irq(dev_priv->drm.irq);
 	wake_up_bit(&error->flags, I915_RESET_IN_PROGRESS);
 	return;
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 7febe6e..1e53c31 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1325,7 +1325,7 @@ struct intel_gen6_power_mgmt {
 	unsigned boosts;
 
 	/* manual wa residency calculations */
-	struct intel_rps_ei up_ei, down_ei;
+	struct intel_rps_ei ei;
 
 	/*
 	 * Protects RPS/RC6 register access and PCU communication.
@@ -2064,8 +2064,6 @@ struct drm_i915_private {
 
 	const struct intel_device_info info;
 
-	int relative_constants_mode;
-
 	void __iomem *regs;
 
 	struct intel_uncore uncore;
@@ -3342,6 +3340,7 @@ static inline u32 i915_reset_count(struct i915_gpu_error *error)
 }
 
 int i915_gem_reset_prepare(struct drm_i915_private *dev_priv);
+void i915_gem_reset(struct drm_i915_private *dev_priv);
 void i915_gem_reset_finish(struct drm_i915_private *dev_priv);
 void i915_gem_set_wedged(struct drm_i915_private *dev_priv);
 void i915_gem_clflush_object(struct drm_i915_gem_object *obj, bool force);
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 10777da..67b1fc5 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -2719,7 +2719,16 @@ int i915_gem_reset_prepare(struct drm_i915_private *dev_priv)
 	for_each_engine(engine, dev_priv, id) {
 		struct drm_i915_gem_request *request;
 
+		/* Prevent request submission to the hardware until we have
+		 * completed the reset in i915_gem_reset_finish(). If a request
+		 * is completed by one engine, it may then queue a request
+		 * to a second via its engine->irq_tasklet *just* as we are
+		 * calling engine->init_hw() and also writing the ELSP.
+		 * Turning off the engine->irq_tasklet until the reset is over
+		 * prevents the race.
+		 */
 		tasklet_kill(&engine->irq_tasklet);
+		tasklet_disable(&engine->irq_tasklet);
 
 		if (engine_stalled(engine)) {
 			request = i915_gem_find_active_request(engine);
@@ -2834,7 +2843,7 @@ static void i915_gem_reset_engine(struct intel_engine_cs *engine)
 	engine->reset_hw(engine, request);
 }
 
-void i915_gem_reset_finish(struct drm_i915_private *dev_priv)
+void i915_gem_reset(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
@@ -2856,6 +2865,17 @@ void i915_gem_reset_finish(struct drm_i915_private *dev_priv)
 	}
 }
 
+void i915_gem_reset_finish(struct drm_i915_private *dev_priv)
+{
+	struct intel_engine_cs *engine;
+	enum intel_engine_id id;
+
+	lockdep_assert_held(&dev_priv->drm.struct_mutex);
+
+	for_each_engine(engine, dev_priv, id)
+		tasklet_enable(&engine->irq_tasklet);
+}
+
 static void nop_submit_request(struct drm_i915_gem_request *request)
 {
 	dma_fence_set_error(&request->fence, -EIO);
@@ -4674,8 +4694,6 @@ i915_gem_load_init(struct drm_i915_private *dev_priv)
 	init_waitqueue_head(&dev_priv->gpu_error.wait_queue);
 	init_waitqueue_head(&dev_priv->gpu_error.reset_queue);
 
-	dev_priv->relative_constants_mode = I915_EXEC_CONSTANTS_REL_GENERAL;
-
 	init_waitqueue_head(&dev_priv->pending_flip_queue);
 
 	dev_priv->mm.interruptible = true;
diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c
index 17f90c6..e2d83b6 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.c
+++ b/drivers/gpu/drm/i915/i915_gem_context.c
@@ -311,7 +311,6 @@ __create_hw_context(struct drm_i915_private *dev_priv,
 	ctx->ring_size = 4 * PAGE_SIZE;
 	ctx->desc_template = GEN8_CTX_ADDRESSING_MODE(dev_priv) <<
 			     GEN8_CTX_ADDRESSING_MODE_SHIFT;
-	ATOMIC_INIT_NOTIFIER_HEAD(&ctx->status_notifier);
 
 	/* GuC requires the ring to be placed above GUC_WOPCM_TOP. If GuC is not
 	 * present or not in use we still need a small bias as ring wraparound
diff --git a/drivers/gpu/drm/i915/i915_gem_context.h b/drivers/gpu/drm/i915/i915_gem_context.h
index 0ac750b..e9c008f 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.h
+++ b/drivers/gpu/drm/i915/i915_gem_context.h
@@ -160,9 +160,6 @@ struct i915_gem_context {
 	/** desc_template: invariant fields for the HW context descriptor */
 	u32 desc_template;
 
-	/** status_notifier: list of callbacks for context-switch changes */
-	struct atomic_notifier_head status_notifier;
-
 	/** guilty_count: How many times this context has caused a GPU hang. */
 	unsigned int guilty_count;
 	/**
diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index d02cfae..30e0675 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -1408,10 +1408,7 @@ execbuf_submit(struct i915_execbuffer_params *params,
 	       struct drm_i915_gem_execbuffer2 *args,
 	       struct list_head *vmas)
 {
-	struct drm_i915_private *dev_priv = params->request->i915;
 	u64 exec_start, exec_len;
-	int instp_mode;
-	u32 instp_mask;
 	int ret;
 
 	ret = i915_gem_execbuffer_move_to_gpu(params->request, vmas);
@@ -1422,56 +1419,11 @@ execbuf_submit(struct i915_execbuffer_params *params,
 	if (ret)
 		return ret;
 
-	instp_mode = args->flags & I915_EXEC_CONSTANTS_MASK;
-	instp_mask = I915_EXEC_CONSTANTS_MASK;
-	switch (instp_mode) {
-	case I915_EXEC_CONSTANTS_REL_GENERAL:
-	case I915_EXEC_CONSTANTS_ABSOLUTE:
-	case I915_EXEC_CONSTANTS_REL_SURFACE:
-		if (instp_mode != 0 && params->engine->id != RCS) {
-			DRM_DEBUG("non-0 rel constants mode on non-RCS\n");
-			return -EINVAL;
-		}
-
-		if (instp_mode != dev_priv->relative_constants_mode) {
-			if (INTEL_INFO(dev_priv)->gen < 4) {
-				DRM_DEBUG("no rel constants on pre-gen4\n");
-				return -EINVAL;
-			}
-
-			if (INTEL_INFO(dev_priv)->gen > 5 &&
-			    instp_mode == I915_EXEC_CONSTANTS_REL_SURFACE) {
-				DRM_DEBUG("rel surface constants mode invalid on gen5+\n");
-				return -EINVAL;
-			}
-
-			/* The HW changed the meaning on this bit on gen6 */
-			if (INTEL_INFO(dev_priv)->gen >= 6)
-				instp_mask &= ~I915_EXEC_CONSTANTS_REL_SURFACE;
-		}
-		break;
-	default:
-		DRM_DEBUG("execbuf with unknown constants: %d\n", instp_mode);
+	if (args->flags & I915_EXEC_CONSTANTS_MASK) {
+		DRM_DEBUG("I915_EXEC_CONSTANTS_* unsupported\n");
 		return -EINVAL;
 	}
 
-	if (params->engine->id == RCS &&
-	    instp_mode != dev_priv->relative_constants_mode) {
-		struct intel_ring *ring = params->request->ring;
-
-		ret = intel_ring_begin(params->request, 4);
-		if (ret)
-			return ret;
-
-		intel_ring_emit(ring, MI_NOOP);
-		intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(1));
-		intel_ring_emit_reg(ring, INSTPM);
-		intel_ring_emit(ring, instp_mask << 16 | instp_mode);
-		intel_ring_advance(ring);
-
-		dev_priv->relative_constants_mode = instp_mode;
-	}
-
 	if (args->flags & I915_EXEC_GEN7_SOL_RESET) {
 		ret = i915_reset_gen7_sol_offsets(params->request);
 		if (ret)
diff --git a/drivers/gpu/drm/i915/i915_gem_shrinker.c b/drivers/gpu/drm/i915/i915_gem_shrinker.c
index 401006b..d5d2b4c 100644
--- a/drivers/gpu/drm/i915/i915_gem_shrinker.c
+++ b/drivers/gpu/drm/i915/i915_gem_shrinker.c
@@ -263,7 +263,7 @@ unsigned long i915_gem_shrink_all(struct drm_i915_private *dev_priv)
 				I915_SHRINK_BOUND |
 				I915_SHRINK_UNBOUND |
 				I915_SHRINK_ACTIVE);
-	rcu_barrier(); /* wait until our RCU delayed slab frees are completed */
+	synchronize_rcu(); /* wait for our earlier RCU delayed slab frees */
 
 	return freed;
 }
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index e6ffef2..b6c886a 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -1046,68 +1046,51 @@ static void vlv_c0_read(struct drm_i915_private *dev_priv,
 	ei->media_c0 = I915_READ(VLV_MEDIA_C0_COUNT);
 }
 
-static bool vlv_c0_above(struct drm_i915_private *dev_priv,
-			 const struct intel_rps_ei *old,
-			 const struct intel_rps_ei *now,
-			 int threshold)
-{
-	u64 time, c0;
-	unsigned int mul = 100;
-
-	if (old->cz_clock == 0)
-		return false;
-
-	if (I915_READ(VLV_COUNTER_CONTROL) & VLV_COUNT_RANGE_HIGH)
-		mul <<= 8;
-
-	time = now->cz_clock - old->cz_clock;
-	time *= threshold * dev_priv->czclk_freq;
-
-	/* Workload can be split between render + media, e.g. SwapBuffers
-	 * being blitted in X after being rendered in mesa. To account for
-	 * this we need to combine both engines into our activity counter.
-	 */
-	c0 = now->render_c0 - old->render_c0;
-	c0 += now->media_c0 - old->media_c0;
-	c0 *= mul * VLV_CZ_CLOCK_TO_MILLI_SEC;
-
-	return c0 >= time;
-}
-
 void gen6_rps_reset_ei(struct drm_i915_private *dev_priv)
 {
-	vlv_c0_read(dev_priv, &dev_priv->rps.down_ei);
-	dev_priv->rps.up_ei = dev_priv->rps.down_ei;
+	memset(&dev_priv->rps.ei, 0, sizeof(dev_priv->rps.ei));
 }
 
 static u32 vlv_wa_c0_ei(struct drm_i915_private *dev_priv, u32 pm_iir)
 {
+	const struct intel_rps_ei *prev = &dev_priv->rps.ei;
 	struct intel_rps_ei now;
 	u32 events = 0;
 
-	if ((pm_iir & (GEN6_PM_RP_DOWN_EI_EXPIRED | GEN6_PM_RP_UP_EI_EXPIRED)) == 0)
+	if ((pm_iir & GEN6_PM_RP_UP_EI_EXPIRED) == 0)
 		return 0;
 
 	vlv_c0_read(dev_priv, &now);
 	if (now.cz_clock == 0)
 		return 0;
 
-	if (pm_iir & GEN6_PM_RP_DOWN_EI_EXPIRED) {
-		if (!vlv_c0_above(dev_priv,
-				  &dev_priv->rps.down_ei, &now,
-				  dev_priv->rps.down_threshold))
-			events |= GEN6_PM_RP_DOWN_THRESHOLD;
-		dev_priv->rps.down_ei = now;
-	}
+	if (prev->cz_clock) {
+		u64 time, c0;
+		unsigned int mul;
+
+		mul = VLV_CZ_CLOCK_TO_MILLI_SEC * 100; /* scale to threshold% */
+		if (I915_READ(VLV_COUNTER_CONTROL) & VLV_COUNT_RANGE_HIGH)
+			mul <<= 8;
 
-	if (pm_iir & GEN6_PM_RP_UP_EI_EXPIRED) {
-		if (vlv_c0_above(dev_priv,
-				 &dev_priv->rps.up_ei, &now,
-				 dev_priv->rps.up_threshold))
-			events |= GEN6_PM_RP_UP_THRESHOLD;
-		dev_priv->rps.up_ei = now;
+		time = now.cz_clock - prev->cz_clock;
+		time *= dev_priv->czclk_freq;
+
+		/* Workload can be split between render + media,
+		 * e.g. SwapBuffers being blitted in X after being rendered in
+		 * mesa. To account for this we need to combine both engines
+		 * into our activity counter.
+		 */
+		c0 = now.render_c0 - prev->render_c0;
+		c0 += now.media_c0 - prev->media_c0;
+		c0 *= mul;
+
+		if (c0 > time * dev_priv->rps.up_threshold)
+			events = GEN6_PM_RP_UP_THRESHOLD;
+		else if (c0 < time * dev_priv->rps.down_threshold)
+			events = GEN6_PM_RP_DOWN_THRESHOLD;
 	}
 
+	dev_priv->rps.ei = now;
 	return events;
 }
 
@@ -4228,7 +4211,7 @@ void intel_irq_init(struct drm_i915_private *dev_priv)
 	/* Let's track the enabled rps events */
 	if (IS_VALLEYVIEW(dev_priv))
 		/* WaGsvRC0ResidencyMethod:vlv */
-		dev_priv->pm_rps_events = GEN6_PM_RP_DOWN_EI_EXPIRED | GEN6_PM_RP_UP_EI_EXPIRED;
+		dev_priv->pm_rps_events = GEN6_PM_RP_UP_EI_EXPIRED;
 	else
 		dev_priv->pm_rps_events = GEN6_PM_RPS_EVENTS;
 
@@ -4266,6 +4249,16 @@ void intel_irq_init(struct drm_i915_private *dev_priv)
 	if (!IS_GEN2(dev_priv))
 		dev->vblank_disable_immediate = true;
 
+	/* Most platforms treat the display irq block as an always-on
+	 * power domain. vlv/chv can disable it at runtime and need
+	 * special care to avoid writing any of the display block registers
+	 * outside of the power domain. We defer setting up the display irqs
+	 * in this case to the runtime pm.
+	 */
+	dev_priv->display_irqs_enabled = true;
+	if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv))
+		dev_priv->display_irqs_enabled = false;
+
 	dev->driver->get_vblank_timestamp = i915_get_vblank_timestamp;
 	dev->driver->get_scanout_position = i915_get_crtc_scanoutpos;
 
diff --git a/drivers/gpu/drm/i915/intel_csr.c b/drivers/gpu/drm/i915/intel_csr.c
index 0085bc7..de219b7 100644
--- a/drivers/gpu/drm/i915/intel_csr.c
+++ b/drivers/gpu/drm/i915/intel_csr.c
@@ -35,7 +35,6 @@
  */
 
 #define I915_CSR_GLK "i915/glk_dmc_ver1_01.bin"
-MODULE_FIRMWARE(I915_CSR_GLK);
 #define GLK_CSR_VERSION_REQUIRED	CSR_VERSION(1, 1)
 
 #define I915_CSR_KBL "i915/kbl_dmc_ver1_01.bin"
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 3282b0f..ed1f4f2 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -16696,12 +16696,11 @@ int intel_modeset_init(struct drm_device *dev)
 		}
 	}
 
-	intel_update_czclk(dev_priv);
-	intel_update_cdclk(dev_priv);
-	dev_priv->atomic_cdclk_freq = dev_priv->cdclk_freq;
-
 	intel_shared_dpll_init(dev);
 
+	intel_update_czclk(dev_priv);
+	intel_modeset_init_hw(dev);
+
 	if (dev_priv->max_cdclk_freq == 0)
 		intel_update_max_cdclk(dev_priv);
 
@@ -17258,8 +17257,6 @@ void intel_modeset_gem_init(struct drm_device *dev)
 
 	intel_init_gt_powersave(dev_priv);
 
-	intel_modeset_init_hw(dev);
-
 	intel_setup_overlay(dev_priv);
 }
 
diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c b/drivers/gpu/drm/i915/intel_engine_cs.c
index 371acf1..ab1be5c 100644
--- a/drivers/gpu/drm/i915/intel_engine_cs.c
+++ b/drivers/gpu/drm/i915/intel_engine_cs.c
@@ -105,6 +105,8 @@ intel_engine_setup(struct drm_i915_private *dev_priv,
 	/* Nothing to do here, execute in order of dependencies */
 	engine->schedule = NULL;
 
+	ATOMIC_INIT_NOTIFIER_HEAD(&engine->context_status_notifier);
+
 	dev_priv->engine[id] = engine;
 	return 0;
 }
diff --git a/drivers/gpu/drm/i915/intel_gvt.c b/drivers/gpu/drm/i915/intel_gvt.c
index d23c0fc..8c04eca 100644
--- a/drivers/gpu/drm/i915/intel_gvt.c
+++ b/drivers/gpu/drm/i915/intel_gvt.c
@@ -77,6 +77,11 @@ int intel_gvt_init(struct drm_i915_private *dev_priv)
 		goto bail;
 	}
 
+	if (!i915.enable_execlists) {
+		DRM_INFO("GPU guest virtualisation [GVT-g] disabled due to disabled execlist submission [i915.enable_execlists module parameter]\n");
+		goto bail;
+	}
+
 	/*
 	 * We're not in host or fail to find a MPT module, disable GVT-g
 	 */
diff --git a/drivers/gpu/drm/i915/intel_hdmi.c b/drivers/gpu/drm/i915/intel_hdmi.c
index ebae2bd..24b2fa5 100644
--- a/drivers/gpu/drm/i915/intel_hdmi.c
+++ b/drivers/gpu/drm/i915/intel_hdmi.c
@@ -1298,16 +1298,34 @@ intel_hdmi_mode_valid(struct drm_connector *connector,
 
 static bool hdmi_12bpc_possible(struct intel_crtc_state *crtc_state)
 {
-	struct drm_device *dev = crtc_state->base.crtc->dev;
+	struct drm_i915_private *dev_priv =
+		to_i915(crtc_state->base.crtc->dev);
+	struct drm_atomic_state *state = crtc_state->base.state;
+	struct drm_connector_state *connector_state;
+	struct drm_connector *connector;
+	int i;
 
-	if (HAS_GMCH_DISPLAY(to_i915(dev)))
+	if (HAS_GMCH_DISPLAY(dev_priv))
 		return false;
 
 	/*
 	 * HDMI 12bpc affects the clocks, so it's only possible
 	 * when not cloning with other encoder types.
 	 */
-	return crtc_state->output_types == 1 << INTEL_OUTPUT_HDMI;
+	if (crtc_state->output_types != 1 << INTEL_OUTPUT_HDMI)
+		return false;
+
+	for_each_connector_in_state(state, connector, connector_state, i) {
+		const struct drm_display_info *info = &connector->display_info;
+
+		if (connector_state->crtc != crtc_state->base.crtc)
+			continue;
+
+		if ((info->edid_hdmi_dc_modes & DRM_EDID_HDMI_DC_36) == 0)
+			return false;
+	}
+
+	return true;
 }
 
 bool intel_hdmi_compute_config(struct intel_encoder *encoder,
diff --git a/drivers/gpu/drm/i915/intel_hotplug.c b/drivers/gpu/drm/i915/intel_hotplug.c
index b62e3f8..54208be 100644
--- a/drivers/gpu/drm/i915/intel_hotplug.c
+++ b/drivers/gpu/drm/i915/intel_hotplug.c
@@ -219,7 +219,7 @@ static void intel_hpd_irq_storm_reenable_work(struct work_struct *work)
 			}
 		}
 	}
-	if (dev_priv->display.hpd_irq_setup)
+	if (dev_priv->display_irqs_enabled && dev_priv->display.hpd_irq_setup)
 		dev_priv->display.hpd_irq_setup(dev_priv);
 	spin_unlock_irq(&dev_priv->irq_lock);
 
@@ -425,7 +425,7 @@ void intel_hpd_irq_handler(struct drm_i915_private *dev_priv,
 		}
 	}
 
-	if (storm_detected)
+	if (storm_detected && dev_priv->display_irqs_enabled)
 		dev_priv->display.hpd_irq_setup(dev_priv);
 	spin_unlock(&dev_priv->irq_lock);
 
@@ -471,10 +471,12 @@ void intel_hpd_init(struct drm_i915_private *dev_priv)
 	 * Interrupt setup is already guaranteed to be single-threaded, this is
 	 * just to make the assert_spin_locked checks happy.
 	 */
-	spin_lock_irq(&dev_priv->irq_lock);
-	if (dev_priv->display.hpd_irq_setup)
-		dev_priv->display.hpd_irq_setup(dev_priv);
-	spin_unlock_irq(&dev_priv->irq_lock);
+	if (dev_priv->display_irqs_enabled && dev_priv->display.hpd_irq_setup) {
+		spin_lock_irq(&dev_priv->irq_lock);
+		if (dev_priv->display_irqs_enabled)
+			dev_priv->display.hpd_irq_setup(dev_priv);
+		spin_unlock_irq(&dev_priv->irq_lock);
+	}
 }
 
 static void i915_hpd_poll_init_work(struct work_struct *work)
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index ebf8023..471af3b 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -345,7 +345,8 @@ execlists_context_status_change(struct drm_i915_gem_request *rq,
 	if (!IS_ENABLED(CONFIG_DRM_I915_GVT))
 		return;
 
-	atomic_notifier_call_chain(&rq->ctx->status_notifier, status, rq);
+	atomic_notifier_call_chain(&rq->engine->context_status_notifier,
+				   status, rq);
 }
 
 static void
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 940bab2..6a29784 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -4928,8 +4928,9 @@ static u32 gen6_rps_pm_mask(struct drm_i915_private *dev_priv, u8 val)
 {
 	u32 mask = 0;
 
+	/* We use UP_EI_EXPIRED interupts for both up/down in manual mode */
 	if (val > dev_priv->rps.min_freq_softlimit)
-		mask |= GEN6_PM_RP_DOWN_EI_EXPIRED | GEN6_PM_RP_DOWN_THRESHOLD | GEN6_PM_RP_DOWN_TIMEOUT;
+		mask |= GEN6_PM_RP_UP_EI_EXPIRED | GEN6_PM_RP_DOWN_THRESHOLD | GEN6_PM_RP_DOWN_TIMEOUT;
 	if (val < dev_priv->rps.max_freq_softlimit)
 		mask |= GEN6_PM_RP_UP_EI_EXPIRED | GEN6_PM_RP_UP_THRESHOLD;
 
@@ -5039,7 +5040,7 @@ void gen6_rps_busy(struct drm_i915_private *dev_priv)
 {
 	mutex_lock(&dev_priv->rps.hw_lock);
 	if (dev_priv->rps.enabled) {
-		if (dev_priv->pm_rps_events & (GEN6_PM_RP_DOWN_EI_EXPIRED | GEN6_PM_RP_UP_EI_EXPIRED))
+		if (dev_priv->pm_rps_events & GEN6_PM_RP_UP_EI_EXPIRED)
 			gen6_rps_reset_ei(dev_priv);
 		I915_WRITE(GEN6_PMINTRMSK,
 			   gen6_rps_pm_mask(dev_priv, dev_priv->rps.cur_freq));
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index 79c2b8d..13dccb1 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -403,6 +403,9 @@ struct intel_engine_cs {
 	 */
 	struct i915_gem_context *legacy_active_context;
 
+	/* status_notifier: list of callbacks for context-switch changes */
+	struct atomic_notifier_head context_status_notifier;
+
 	struct intel_engine_hangcheck hangcheck;
 
 	bool needs_cmd_parser;