Merge drm/drm-next into drm-intel-next

Sync up with upstream.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>

11 files changed, 397 insertions, 132 deletions

diff --git a/drivers/gpu/drm/i915/gvt/cmd_parser.c b/drivers/gpu/drm/i915/gvt/cmd_parser.c
index 3fea967ee817..fef1e857cefc 100644
--- a/drivers/gpu/drm/i915/gvt/cmd_parser.c
+++ b/drivers/gpu/drm/i915/gvt/cmd_parser.c
@@ -38,11 +38,18 @@
 
 #include "i915_drv.h"
 #include "gt/intel_gpu_commands.h"
+#include "gt/intel_lrc.h"
 #include "gt/intel_ring.h"
+#include "gt/intel_gt_requests.h"
+#include "gt/shmem_utils.h"
 #include "gvt.h"
 #include "i915_pvinfo.h"
 #include "trace.h"
 
+#include "gem/i915_gem_context.h"
+#include "gem/i915_gem_pm.h"
+#include "gt/intel_context.h"
+
 #define INVALID_OP    (~0U)
 
 #define OP_LEN_MI           9
@@ -455,6 +462,7 @@ enum {
 	RING_BUFFER_INSTRUCTION,
 	BATCH_BUFFER_INSTRUCTION,
 	BATCH_BUFFER_2ND_LEVEL,
+	RING_BUFFER_CTX,
 };
 
 enum {
@@ -496,6 +504,7 @@ struct parser_exec_state {
 	 */
 	int saved_buf_addr_type;
 	bool is_ctx_wa;
+	bool is_init_ctx;
 
 	const struct cmd_info *info;
 
@@ -709,6 +718,11 @@ static inline u32 cmd_val(struct parser_exec_state *s, int index)
 	return *cmd_ptr(s, index);
 }
 
+static inline bool is_init_ctx(struct parser_exec_state *s)
+{
+	return (s->buf_type == RING_BUFFER_CTX && s->is_init_ctx);
+}
+
 static void parser_exec_state_dump(struct parser_exec_state *s)
 {
 	int cnt = 0;
@@ -722,7 +736,8 @@ static void parser_exec_state_dump(struct parser_exec_state *s)
 
 	gvt_dbg_cmd("  %s %s ip_gma(%08lx) ",
 			s->buf_type == RING_BUFFER_INSTRUCTION ?
-			"RING_BUFFER" : "BATCH_BUFFER",
+			"RING_BUFFER" : ((s->buf_type == RING_BUFFER_CTX) ?
+				"CTX_BUFFER" : "BATCH_BUFFER"),
 			s->buf_addr_type == GTT_BUFFER ?
 			"GTT" : "PPGTT", s->ip_gma);
 
@@ -757,7 +772,8 @@ static inline void update_ip_va(struct parser_exec_state *s)
 	if (WARN_ON(s->ring_head == s->ring_tail))
 		return;
 
-	if (s->buf_type == RING_BUFFER_INSTRUCTION) {
+	if (s->buf_type == RING_BUFFER_INSTRUCTION ||
+			s->buf_type == RING_BUFFER_CTX) {
 		unsigned long ring_top = s->ring_start + s->ring_size;
 
 		if (s->ring_head > s->ring_tail) {
@@ -821,68 +837,12 @@ static inline int cmd_length(struct parser_exec_state *s)
 	*addr = val; \
 } while (0)
 
-static bool is_shadowed_mmio(unsigned int offset)
-{
-	bool ret = false;
-
-	if ((offset == 0x2168) || /*BB current head register UDW */
-	    (offset == 0x2140) || /*BB current header register */
-	    (offset == 0x211c) || /*second BB header register UDW */
-	    (offset == 0x2114)) { /*second BB header register UDW */
-		ret = true;
-	}
-	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, char *cmd)
-{
-	struct intel_gvt *gvt = s->vgpu->gvt;
-	unsigned int data;
-	u32 ring_base;
-	u32 nopid;
-
-	if (!strcmp(cmd, "lri"))
-		data = cmd_val(s, index + 1);
-	else {
-		gvt_err("Unexpected forcenonpriv 0x%x write from cmd %s\n",
-			offset, cmd);
-		return -EINVAL;
-	}
-
-	ring_base = s->engine->mmio_base;
-	nopid = i915_mmio_reg_offset(RING_NOPID(ring_base));
-
-	if (!intel_gvt_in_force_nonpriv_whitelist(gvt, data) &&
-			data != nopid) {
-		gvt_err("Unexpected forcenonpriv 0x%x LRI write, value=0x%x\n",
-			offset, data);
-		patch_value(s, cmd_ptr(s, index), nopid);
-		return 0;
-	}
-	return 0;
-}
-
 static inline bool is_mocs_mmio(unsigned int offset)
 {
 	return ((offset >= 0xc800) && (offset <= 0xcff8)) ||
 		((offset >= 0xb020) && (offset <= 0xb0a0));
 }
 
-static int mocs_cmd_reg_handler(struct parser_exec_state *s,
-				unsigned int offset, unsigned int index)
-{
-	if (!is_mocs_mmio(offset))
-		return -EINVAL;
-	vgpu_vreg(s->vgpu, offset) = cmd_val(s, index + 1);
-	return 0;
-}
-
 static int is_cmd_update_pdps(unsigned int offset,
 			      struct parser_exec_state *s)
 {
@@ -930,6 +890,7 @@ static int cmd_reg_handler(struct parser_exec_state *s,
 	struct intel_vgpu *vgpu = s->vgpu;
 	struct intel_gvt *gvt = vgpu->gvt;
 	u32 ctx_sr_ctl;
+	u32 *vreg, vreg_old;
 
 	if (offset + 4 > gvt->device_info.mmio_size) {
 		gvt_vgpu_err("%s access to (%x) outside of MMIO range\n",
@@ -937,34 +898,101 @@ static int cmd_reg_handler(struct parser_exec_state *s,
 		return -EFAULT;
 	}
 
+	if (is_init_ctx(s)) {
+		struct intel_gvt_mmio_info *mmio_info;
+
+		intel_gvt_mmio_set_cmd_accessible(gvt, offset);
+		mmio_info = intel_gvt_find_mmio_info(gvt, offset);
+		if (mmio_info && mmio_info->write)
+			intel_gvt_mmio_set_cmd_write_patch(gvt, offset);
+		return 0;
+	}
+
 	if (!intel_gvt_mmio_is_cmd_accessible(gvt, offset)) {
 		gvt_vgpu_err("%s access to non-render register (%x)\n",
 				cmd, offset);
 		return -EBADRQC;
 	}
 
-	if (is_shadowed_mmio(offset)) {
-		gvt_vgpu_err("found access of shadowed MMIO %x\n", offset);
-		return 0;
+	if (!strncmp(cmd, "srm", 3) ||
+			!strncmp(cmd, "lrm", 3)) {
+		if (offset != i915_mmio_reg_offset(GEN8_L3SQCREG4) &&
+				offset != 0x21f0) {
+			gvt_vgpu_err("%s access to register (%x)\n",
+					cmd, offset);
+			return -EPERM;
+		} else
+			return 0;
 	}
 
-	if (is_mocs_mmio(offset) &&
-	    mocs_cmd_reg_handler(s, offset, index))
-		return -EINVAL;
+	if (!strncmp(cmd, "lrr-src", 7) ||
+			!strncmp(cmd, "lrr-dst", 7)) {
+		gvt_vgpu_err("not allowed cmd %s\n", cmd);
+		return -EPERM;
+	}
+
+	if (!strncmp(cmd, "pipe_ctrl", 9)) {
+		/* TODO: add LRI POST logic here */
+		return 0;
+	}
 
-	if (is_force_nonpriv_mmio(offset) &&
-		force_nonpriv_reg_handler(s, offset, index, cmd))
+	if (strncmp(cmd, "lri", 3))
 		return -EPERM;
 
+	/* below are all lri handlers */
+	vreg = &vgpu_vreg(s->vgpu, offset);
+	if (!intel_gvt_mmio_is_cmd_accessible(gvt, offset)) {
+		gvt_vgpu_err("%s access to non-render register (%x)\n",
+				cmd, offset);
+		return -EBADRQC;
+	}
+
+	if (is_cmd_update_pdps(offset, s) &&
+	    cmd_pdp_mmio_update_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 */
 		patch_value(s, cmd_ptr(s, index), VGT_PVINFO_PAGE);
 	}
 
-	if (is_cmd_update_pdps(offset, s) &&
-	    cmd_pdp_mmio_update_handler(s, offset, index))
-		return -EINVAL;
+	if (is_mocs_mmio(offset))
+		*vreg = cmd_val(s, index + 1);
+
+	vreg_old = *vreg;
+
+	if (intel_gvt_mmio_is_cmd_write_patch(gvt, offset)) {
+		u32 cmdval_new, cmdval;
+		struct intel_gvt_mmio_info *mmio_info;
+
+		cmdval = cmd_val(s, index + 1);
+
+		mmio_info = intel_gvt_find_mmio_info(gvt, offset);
+		if (!mmio_info) {
+			cmdval_new = cmdval;
+		} else {
+			u64 ro_mask = mmio_info->ro_mask;
+			int ret;
+
+			if (likely(!ro_mask))
+				ret = mmio_info->write(s->vgpu, offset,
+						&cmdval, 4);
+			else {
+				gvt_vgpu_err("try to write RO reg %x\n",
+						offset);
+				ret = -EBADRQC;
+			}
+			if (ret)
+				return ret;
+			cmdval_new = *vreg;
+		}
+		if (cmdval_new != cmdval)
+			patch_value(s, cmd_ptr(s, index+1), cmdval_new);
+	}
+
+	/* only patch cmd. restore vreg value if changed in mmio write handler*/
+	*vreg = vreg_old;
 
 	/* TODO
 	 * In order to let workload with inhibit context to generate
@@ -1216,6 +1244,8 @@ static int cmd_handler_mi_batch_buffer_end(struct parser_exec_state *s)
 		s->buf_type = BATCH_BUFFER_INSTRUCTION;
 		ret = ip_gma_set(s, s->ret_ip_gma_bb);
 		s->buf_addr_type = s->saved_buf_addr_type;
+	} else if (s->buf_type == RING_BUFFER_CTX) {
+		ret = ip_gma_set(s, s->ring_tail);
 	} else {
 		s->buf_type = RING_BUFFER_INSTRUCTION;
 		s->buf_addr_type = GTT_BUFFER;
@@ -2764,7 +2794,8 @@ static int command_scan(struct parser_exec_state *s,
 	gma_bottom = rb_start +  rb_len;
 
 	while (s->ip_gma != gma_tail) {
-		if (s->buf_type == RING_BUFFER_INSTRUCTION) {
+		if (s->buf_type == RING_BUFFER_INSTRUCTION ||
+				s->buf_type == RING_BUFFER_CTX) {
 			if (!(s->ip_gma >= rb_start) ||
 				!(s->ip_gma < gma_bottom)) {
 				gvt_vgpu_err("ip_gma %lx out of ring scope."
@@ -3057,6 +3088,118 @@ int intel_gvt_scan_and_shadow_wa_ctx(struct intel_shadow_wa_ctx *wa_ctx)
 	return 0;
 }
 
+/* generate dummy contexts by sending empty requests to HW, and let
+ * the HW to fill Engine Contexts. This dummy contexts are used for
+ * initialization purpose (update reg whitelist), so referred to as
+ * init context here
+ */
+void intel_gvt_update_reg_whitelist(struct intel_vgpu *vgpu)
+{
+	const unsigned long start = LRC_STATE_PN * PAGE_SIZE;
+	struct intel_gvt *gvt = vgpu->gvt;
+	struct intel_engine_cs *engine;
+	enum intel_engine_id id;
+
+	if (gvt->is_reg_whitelist_updated)
+		return;
+
+	/* scan init ctx to update cmd accessible list */
+	for_each_engine(engine, gvt->gt, id) {
+		struct parser_exec_state s;
+		void *vaddr;
+		int ret;
+
+		if (!engine->default_state)
+			continue;
+
+		vaddr = shmem_pin_map(engine->default_state);
+		if (IS_ERR(vaddr)) {
+			gvt_err("failed to map %s->default state, err:%zd\n",
+				engine->name, PTR_ERR(vaddr));
+			return;
+		}
+
+		s.buf_type = RING_BUFFER_CTX;
+		s.buf_addr_type = GTT_BUFFER;
+		s.vgpu = vgpu;
+		s.engine = engine;
+		s.ring_start = 0;
+		s.ring_size = engine->context_size - start;
+		s.ring_head = 0;
+		s.ring_tail = s.ring_size;
+		s.rb_va = vaddr + start;
+		s.workload = NULL;
+		s.is_ctx_wa = false;
+		s.is_init_ctx = true;
+
+		/* skipping the first RING_CTX_SIZE(0x50) dwords */
+		ret = ip_gma_set(&s, RING_CTX_SIZE);
+		if (ret == 0) {
+			ret = command_scan(&s, 0, s.ring_size, 0, s.ring_size);
+			if (ret)
+				gvt_err("Scan init ctx error\n");
+		}
+
+		shmem_unpin_map(engine->default_state, vaddr);
+		if (ret)
+			return;
+	}
+
+	gvt->is_reg_whitelist_updated = true;
+}
+
+int intel_gvt_scan_engine_context(struct intel_vgpu_workload *workload)
+{
+	struct intel_vgpu *vgpu = workload->vgpu;
+	unsigned long gma_head, gma_tail, gma_start, ctx_size;
+	struct parser_exec_state s;
+	int ring_id = workload->engine->id;
+	struct intel_context *ce = vgpu->submission.shadow[ring_id];
+	int ret;
+
+	GEM_BUG_ON(atomic_read(&ce->pin_count) < 0);
+
+	ctx_size = workload->engine->context_size - PAGE_SIZE;
+
+	/* Only ring contxt is loaded to HW for inhibit context, no need to
+	 * scan engine context
+	 */
+	if (is_inhibit_context(ce))
+		return 0;
+
+	gma_start = i915_ggtt_offset(ce->state) + LRC_STATE_PN*PAGE_SIZE;
+	gma_head = 0;
+	gma_tail = ctx_size;
+
+	s.buf_type = RING_BUFFER_CTX;
+	s.buf_addr_type = GTT_BUFFER;
+	s.vgpu = workload->vgpu;
+	s.engine = workload->engine;
+	s.ring_start = gma_start;
+	s.ring_size = ctx_size;
+	s.ring_head = gma_start + gma_head;
+	s.ring_tail = gma_start + gma_tail;
+	s.rb_va = ce->lrc_reg_state;
+	s.workload = workload;
+	s.is_ctx_wa = false;
+	s.is_init_ctx = false;
+
+	/* don't scan the first RING_CTX_SIZE(0x50) dwords, as it's ring
+	 * context
+	 */
+	ret = ip_gma_set(&s, gma_start + gma_head + RING_CTX_SIZE);
+	if (ret)
+		goto out;
+
+	ret = command_scan(&s, gma_head, gma_tail,
+		gma_start, ctx_size);
+out:
+	if (ret)
+		gvt_vgpu_err("scan shadow ctx error\n");
+
+	return ret;
+}
+
 static int init_cmd_table(struct intel_gvt *gvt)
 {
 	unsigned int gen_type = intel_gvt_get_device_type(gvt);
diff --git a/drivers/gpu/drm/i915/gvt/cmd_parser.h b/drivers/gpu/drm/i915/gvt/cmd_parser.h
index ab25d151932a..416d345e2816 100644
--- a/drivers/gpu/drm/i915/gvt/cmd_parser.h
+++ b/drivers/gpu/drm/i915/gvt/cmd_parser.h
@@ -40,6 +40,7 @@
 
 struct intel_gvt;
 struct intel_shadow_wa_ctx;
+struct intel_vgpu;
 struct intel_vgpu_workload;
 
 void intel_gvt_clean_cmd_parser(struct intel_gvt *gvt);
@@ -50,4 +51,8 @@ int intel_gvt_scan_and_shadow_ringbuffer(struct intel_vgpu_workload *workload);
 
 int intel_gvt_scan_and_shadow_wa_ctx(struct intel_shadow_wa_ctx *wa_ctx);
 
+void intel_gvt_update_reg_whitelist(struct intel_vgpu *vgpu);
+
+int intel_gvt_scan_engine_context(struct intel_vgpu_workload *workload);
+
 #endif
diff --git a/drivers/gpu/drm/i915/gvt/display.c b/drivers/gpu/drm/i915/gvt/display.c
index a15f87539657..62a5b0dd2003 100644
--- a/drivers/gpu/drm/i915/gvt/display.c
+++ b/drivers/gpu/drm/i915/gvt/display.c
@@ -217,6 +217,15 @@ static void emulate_monitor_status_change(struct intel_vgpu *vgpu)
 				  DDI_BUF_CTL_ENABLE);
 			vgpu_vreg_t(vgpu, DDI_BUF_CTL(port)) |= DDI_BUF_IS_IDLE;
 		}
+		vgpu_vreg_t(vgpu, PCH_PORT_HOTPLUG) &=
+			~(PORTA_HOTPLUG_ENABLE | PORTA_HOTPLUG_STATUS_MASK);
+		vgpu_vreg_t(vgpu, PCH_PORT_HOTPLUG) &=
+			~(PORTB_HOTPLUG_ENABLE | PORTB_HOTPLUG_STATUS_MASK);
+		vgpu_vreg_t(vgpu, PCH_PORT_HOTPLUG) &=
+			~(PORTC_HOTPLUG_ENABLE | PORTC_HOTPLUG_STATUS_MASK);
+		/* No hpd_invert set in vgpu vbt, need to clear invert mask */
+		vgpu_vreg_t(vgpu, PCH_PORT_HOTPLUG) &= ~BXT_DDI_HPD_INVERT_MASK;
+		vgpu_vreg_t(vgpu, GEN8_DE_PORT_ISR) &= ~BXT_DE_PORT_HOTPLUG_MASK;
 
 		vgpu_vreg_t(vgpu, BXT_P_CR_GT_DISP_PWRON) &= ~(BIT(0) | BIT(1));
 		vgpu_vreg_t(vgpu, BXT_PORT_CL1CM_DW0(DPIO_PHY0)) &=
@@ -273,6 +282,8 @@ static void emulate_monitor_status_change(struct intel_vgpu *vgpu)
 			vgpu_vreg_t(vgpu, TRANS_DDI_FUNC_CTL(TRANSCODER_EDP)) |=
 				(TRANS_DDI_BPC_8 | TRANS_DDI_MODE_SELECT_DP_SST |
 				 TRANS_DDI_FUNC_ENABLE);
+			vgpu_vreg_t(vgpu, PCH_PORT_HOTPLUG) |=
+				PORTA_HOTPLUG_ENABLE;
 			vgpu_vreg_t(vgpu, GEN8_DE_PORT_ISR) |=
 				GEN8_DE_PORT_HOTPLUG(HPD_PORT_A);
 		}
@@ -301,6 +312,8 @@ static void emulate_monitor_status_change(struct intel_vgpu *vgpu)
 				(TRANS_DDI_BPC_8 | TRANS_DDI_MODE_SELECT_DP_SST |
 				 (PORT_B << TRANS_DDI_PORT_SHIFT) |
 				 TRANS_DDI_FUNC_ENABLE);
+			vgpu_vreg_t(vgpu, PCH_PORT_HOTPLUG) |=
+				PORTB_HOTPLUG_ENABLE;
 			vgpu_vreg_t(vgpu, GEN8_DE_PORT_ISR) |=
 				GEN8_DE_PORT_HOTPLUG(HPD_PORT_B);
 		}
@@ -329,6 +342,8 @@ static void emulate_monitor_status_change(struct intel_vgpu *vgpu)
 				(TRANS_DDI_BPC_8 | TRANS_DDI_MODE_SELECT_DP_SST |
 				 (PORT_B << TRANS_DDI_PORT_SHIFT) |
 				 TRANS_DDI_FUNC_ENABLE);
+			vgpu_vreg_t(vgpu, PCH_PORT_HOTPLUG) |=
+				PORTC_HOTPLUG_ENABLE;
 			vgpu_vreg_t(vgpu, GEN8_DE_PORT_ISR) |=
 				GEN8_DE_PORT_HOTPLUG(HPD_PORT_C);
 		}
@@ -661,44 +676,62 @@ void intel_vgpu_emulate_hotplug(struct intel_vgpu *vgpu, bool connected)
 				PORTD_HOTPLUG_STATUS_MASK;
 		intel_vgpu_trigger_virtual_event(vgpu, DP_D_HOTPLUG);
 	} else if (IS_BROXTON(i915)) {
-		if (connected) {
-			if (intel_vgpu_has_monitor_on_port(vgpu, PORT_A)) {
+		if (intel_vgpu_has_monitor_on_port(vgpu, PORT_A)) {
+			if (connected) {
 				vgpu_vreg_t(vgpu, GEN8_DE_PORT_ISR) |=
 					GEN8_DE_PORT_HOTPLUG(HPD_PORT_A);
+			} else {
+				vgpu_vreg_t(vgpu, GEN8_DE_PORT_ISR) &=
+					~GEN8_DE_PORT_HOTPLUG(HPD_PORT_A);
 			}
-			if (intel_vgpu_has_monitor_on_port(vgpu, PORT_B)) {
-				vgpu_vreg_t(vgpu, SFUSE_STRAP) |=
-					SFUSE_STRAP_DDIB_DETECTED;
+			vgpu_vreg_t(vgpu, GEN8_DE_PORT_IIR) |=
+				GEN8_DE_PORT_HOTPLUG(HPD_PORT_A);
+			vgpu_vreg_t(vgpu, PCH_PORT_HOTPLUG) &=
+				~PORTA_HOTPLUG_STATUS_MASK;
+			vgpu_vreg_t(vgpu, PCH_PORT_HOTPLUG) |=
+				PORTA_HOTPLUG_LONG_DETECT;
+			intel_vgpu_trigger_virtual_event(vgpu, DP_A_HOTPLUG);
+		}
+		if (intel_vgpu_has_monitor_on_port(vgpu, PORT_B)) {
+			if (connected) {
 				vgpu_vreg_t(vgpu, GEN8_DE_PORT_ISR) |=
 					GEN8_DE_PORT_HOTPLUG(HPD_PORT_B);
-			}
-			if (intel_vgpu_has_monitor_on_port(vgpu, PORT_C)) {
 				vgpu_vreg_t(vgpu, SFUSE_STRAP) |=
-					SFUSE_STRAP_DDIC_DETECTED;
-				vgpu_vreg_t(vgpu, GEN8_DE_PORT_ISR) |=
-					GEN8_DE_PORT_HOTPLUG(HPD_PORT_C);
-			}
-		} else {
-			if (intel_vgpu_has_monitor_on_port(vgpu, PORT_A)) {
+					SFUSE_STRAP_DDIB_DETECTED;
+			} else {
 				vgpu_vreg_t(vgpu, GEN8_DE_PORT_ISR) &=
-					~GEN8_DE_PORT_HOTPLUG(HPD_PORT_A);
-			}
-			if (intel_vgpu_has_monitor_on_port(vgpu, PORT_B)) {
+					~GEN8_DE_PORT_HOTPLUG(HPD_PORT_B);
 				vgpu_vreg_t(vgpu, SFUSE_STRAP) &=
 					~SFUSE_STRAP_DDIB_DETECTED;
-				vgpu_vreg_t(vgpu, GEN8_DE_PORT_ISR) &=
-					~GEN8_DE_PORT_HOTPLUG(HPD_PORT_B);
 			}
-			if (intel_vgpu_has_monitor_on_port(vgpu, PORT_C)) {
-				vgpu_vreg_t(vgpu, SFUSE_STRAP) &=
-					~SFUSE_STRAP_DDIC_DETECTED;
+			vgpu_vreg_t(vgpu, GEN8_DE_PORT_IIR) |=
+				GEN8_DE_PORT_HOTPLUG(HPD_PORT_B);
+			vgpu_vreg_t(vgpu, PCH_PORT_HOTPLUG) &=
+				~PORTB_HOTPLUG_STATUS_MASK;
+			vgpu_vreg_t(vgpu, PCH_PORT_HOTPLUG) |=
+				PORTB_HOTPLUG_LONG_DETECT;
+			intel_vgpu_trigger_virtual_event(vgpu, DP_B_HOTPLUG);
+		}
+		if (intel_vgpu_has_monitor_on_port(vgpu, PORT_C)) {
+			if (connected) {
+				vgpu_vreg_t(vgpu, GEN8_DE_PORT_ISR) |=
+					GEN8_DE_PORT_HOTPLUG(HPD_PORT_C);
+				vgpu_vreg_t(vgpu, SFUSE_STRAP) |=
+					SFUSE_STRAP_DDIC_DETECTED;
+			} else {
 				vgpu_vreg_t(vgpu, GEN8_DE_PORT_ISR) &=
 					~GEN8_DE_PORT_HOTPLUG(HPD_PORT_C);
+				vgpu_vreg_t(vgpu, SFUSE_STRAP) &=
+					~SFUSE_STRAP_DDIC_DETECTED;
 			}
+			vgpu_vreg_t(vgpu, GEN8_DE_PORT_IIR) |=
+				GEN8_DE_PORT_HOTPLUG(HPD_PORT_C);
+			vgpu_vreg_t(vgpu, PCH_PORT_HOTPLUG) &=
+				~PORTC_HOTPLUG_STATUS_MASK;
+			vgpu_vreg_t(vgpu, PCH_PORT_HOTPLUG) |=
+				PORTC_HOTPLUG_LONG_DETECT;
+			intel_vgpu_trigger_virtual_event(vgpu, DP_C_HOTPLUG);
 		}
-		vgpu_vreg_t(vgpu, PCH_PORT_HOTPLUG) |=
-			PORTB_HOTPLUG_STATUS_MASK;
-		intel_vgpu_trigger_virtual_event(vgpu, DP_B_HOTPLUG);
 	}
 }
 
diff --git a/drivers/gpu/drm/i915/gvt/execlist.c b/drivers/gpu/drm/i915/gvt/execlist.c
index 158873f269b1..c8dcda6d4f0d 100644
--- a/drivers/gpu/drm/i915/gvt/execlist.c
+++ b/drivers/gpu/drm/i915/gvt/execlist.c
@@ -522,12 +522,11 @@ static void init_vgpu_execlist(struct intel_vgpu *vgpu,
 static void clean_execlist(struct intel_vgpu *vgpu,
 			   intel_engine_mask_t engine_mask)
 {
-	struct drm_i915_private *dev_priv = vgpu->gvt->gt->i915;
-	struct intel_engine_cs *engine;
 	struct intel_vgpu_submission *s = &vgpu->submission;
+	struct intel_engine_cs *engine;
 	intel_engine_mask_t tmp;
 
-	for_each_engine_masked(engine, &dev_priv->gt, engine_mask, tmp) {
+	for_each_engine_masked(engine, vgpu->gvt->gt, engine_mask, tmp) {
 		kfree(s->ring_scan_buffer[engine->id]);
 		s->ring_scan_buffer[engine->id] = NULL;
 		s->ring_scan_buffer_size[engine->id] = 0;
@@ -537,11 +536,10 @@ static void clean_execlist(struct intel_vgpu *vgpu,
 static void reset_execlist(struct intel_vgpu *vgpu,
 			   intel_engine_mask_t engine_mask)
 {
-	struct drm_i915_private *dev_priv = vgpu->gvt->gt->i915;
 	struct intel_engine_cs *engine;
 	intel_engine_mask_t tmp;
 
-	for_each_engine_masked(engine, &dev_priv->gt, engine_mask, tmp)
+	for_each_engine_masked(engine, vgpu->gvt->gt, engine_mask, tmp)
 		init_vgpu_execlist(vgpu, engine);
 }
 
diff --git a/drivers/gpu/drm/i915/gvt/gvt.h b/drivers/gpu/drm/i915/gvt/gvt.h
index 62a4807424bb..03c993d68f10 100644
--- a/drivers/gpu/drm/i915/gvt/gvt.h
+++ b/drivers/gpu/drm/i915/gvt/gvt.h
@@ -248,7 +248,7 @@ struct gvt_mmio_block {
 #define INTEL_GVT_MMIO_HASH_BITS 11
 
 struct intel_gvt_mmio {
-	u8 *mmio_attribute;
+	u16 *mmio_attribute;
 /* Register contains RO bits */
 #define F_RO		(1 << 0)
 /* Register contains graphics address */
@@ -267,6 +267,8 @@ struct intel_gvt_mmio {
  * logical context image
  */
 #define F_SR_IN_CTX	(1 << 7)
+/* Value of command write of this reg needs to be patched */
+#define F_CMD_WRITE_PATCH	(1 << 8)
 
 	struct gvt_mmio_block *mmio_block;
 	unsigned int num_mmio_block;
@@ -333,6 +335,7 @@ struct intel_gvt {
 		u32 *mocs_mmio_offset_list;
 		u32 mocs_mmio_offset_list_cnt;
 	} engine_mmio_list;
+	bool is_reg_whitelist_updated;
 
 	struct dentry *debugfs_root;
 };
@@ -416,6 +419,9 @@ int intel_gvt_load_firmware(struct intel_gvt *gvt);
 #define vgpu_fence_base(vgpu) (vgpu->fence.base)
 #define vgpu_fence_sz(vgpu) (vgpu->fence.size)
 
+/* ring context size i.e. the first 0x50 dwords*/
+#define RING_CTX_SIZE 320
+
 struct intel_vgpu_creation_params {
 	__u64 handle;
 	__u64 low_gm_sz;  /* in MB */
@@ -687,6 +693,35 @@ static inline void intel_gvt_mmio_set_sr_in_ctx(
 }
 
 void intel_gvt_debugfs_add_vgpu(struct intel_vgpu *vgpu);
+/**
+ * intel_gvt_mmio_set_cmd_write_patch -
+ *				mark an MMIO if its cmd write needs to be
+ *				patched
+ * @gvt: a GVT device
+ * @offset: register offset
+ *
+ */
+static inline void intel_gvt_mmio_set_cmd_write_patch(
+			struct intel_gvt *gvt, unsigned int offset)
+{
+	gvt->mmio.mmio_attribute[offset >> 2] |= F_CMD_WRITE_PATCH;
+}
+
+/**
+ * intel_gvt_mmio_is_cmd_write_patch - check if an mmio's cmd access needs to
+ * be patched
+ * @gvt: a GVT device
+ * @offset: register offset
+ *
+ * Returns:
+ * True if GPU commmand write to an MMIO should be patched
+ */
+static inline bool intel_gvt_mmio_is_cmd_write_patch(
+			struct intel_gvt *gvt, unsigned int offset)
+{
+	return gvt->mmio.mmio_attribute[offset >> 2] & F_CMD_WRITE_PATCH;
+}
+
 void intel_gvt_debugfs_remove_vgpu(struct intel_vgpu *vgpu);
 void intel_gvt_debugfs_init(struct intel_gvt *gvt);
 void intel_gvt_debugfs_clean(struct intel_gvt *gvt);
diff --git a/drivers/gpu/drm/i915/gvt/handlers.c b/drivers/gpu/drm/i915/gvt/handlers.c
index 0d124ced5f94..6eeaeecb7f85 100644
--- a/drivers/gpu/drm/i915/gvt/handlers.c
+++ b/drivers/gpu/drm/i915/gvt/handlers.c
@@ -83,7 +83,7 @@ static void write_vreg(struct intel_vgpu *vgpu, unsigned int offset,
 	memcpy(&vgpu_vreg(vgpu, offset), p_data, bytes);
 }
 
-static struct intel_gvt_mmio_info *find_mmio_info(struct intel_gvt *gvt,
+struct intel_gvt_mmio_info *intel_gvt_find_mmio_info(struct intel_gvt *gvt,
 						  unsigned int offset)
 {
 	struct intel_gvt_mmio_info *e;
@@ -96,7 +96,7 @@ static struct intel_gvt_mmio_info *find_mmio_info(struct intel_gvt *gvt,
 }
 
 static int new_mmio_info(struct intel_gvt *gvt,
-		u32 offset, u8 flags, u32 size,
+		u32 offset, u16 flags, u32 size,
 		u32 addr_mask, u32 ro_mask, u32 device,
 		gvt_mmio_func read, gvt_mmio_func write)
 {
@@ -118,7 +118,7 @@ static int new_mmio_info(struct intel_gvt *gvt,
 			return -ENOMEM;
 
 		info->offset = i;
-		p = find_mmio_info(gvt, info->offset);
+		p = intel_gvt_find_mmio_info(gvt, info->offset);
 		if (p) {
 			WARN(1, "dup mmio definition offset %x\n",
 				info->offset);
@@ -1965,7 +1965,8 @@ static int init_generic_mmio_info(struct intel_gvt *gvt)
 
 	/* RING MODE */
 #define RING_REG(base) _MMIO((base) + 0x29c)
-	MMIO_RING_DFH(RING_REG, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL,
+	MMIO_RING_DFH(RING_REG, D_ALL,
+		F_MODE_MASK | F_CMD_ACCESS | F_CMD_WRITE_PATCH, NULL,
 		ring_mode_mmio_write);
 #undef RING_REG
 
@@ -2885,8 +2886,8 @@ static int init_bdw_mmio_info(struct intel_gvt *gvt)
 	MMIO_DFH(_MMIO(0xb10c), D_BDW, F_CMD_ACCESS, NULL, NULL);
 	MMIO_D(_MMIO(0xb110), D_BDW);
 
-	MMIO_F(_MMIO(0x24d0), 48, F_CMD_ACCESS, 0, 0, D_BDW_PLUS,
-		NULL, force_nonpriv_write);
+	MMIO_F(_MMIO(0x24d0), 48, F_CMD_ACCESS | F_CMD_WRITE_PATCH, 0, 0,
+		D_BDW_PLUS, NULL, force_nonpriv_write);
 
 	MMIO_D(_MMIO(0x44484), D_BDW_PLUS);
 	MMIO_D(_MMIO(0x4448c), D_BDW_PLUS);
@@ -3626,7 +3627,7 @@ int intel_vgpu_mmio_reg_rw(struct intel_vgpu *vgpu, unsigned int offset,
 	/*
 	 * Normal tracked MMIOs.
 	 */
-	mmio_info = find_mmio_info(gvt, offset);
+	mmio_info = intel_gvt_find_mmio_info(gvt, offset);
 	if (!mmio_info) {
 		gvt_dbg_mmio("untracked MMIO %08x len %d\n", offset, bytes);
 		goto default_rw;
diff --git a/drivers/gpu/drm/i915/gvt/kvmgt.c b/drivers/gpu/drm/i915/gvt/kvmgt.c
index 551f00024e99..d089770795b8 100644
--- a/drivers/gpu/drm/i915/gvt/kvmgt.c
+++ b/drivers/gpu/drm/i915/gvt/kvmgt.c
@@ -1703,7 +1703,7 @@ static int kvmgt_page_track_add(unsigned long handle, u64 gfn)
 		return -EINVAL;
 	}
 
-	spin_lock(&kvm->mmu_lock);
+	write_lock(&kvm->mmu_lock);
 
 	if (kvmgt_gfn_is_write_protected(info, gfn))
 		goto out;
@@ -1712,7 +1712,7 @@ static int kvmgt_page_track_add(unsigned long handle, u64 gfn)
 	kvmgt_protect_table_add(info, gfn);
 
 out:
-	spin_unlock(&kvm->mmu_lock);
+	write_unlock(&kvm->mmu_lock);
 	srcu_read_unlock(&kvm->srcu, idx);
 	return 0;
 }
@@ -1737,7 +1737,7 @@ static int kvmgt_page_track_remove(unsigned long handle, u64 gfn)
 		return -EINVAL;
 	}
 
-	spin_lock(&kvm->mmu_lock);
+	write_lock(&kvm->mmu_lock);
 
 	if (!kvmgt_gfn_is_write_protected(info, gfn))
 		goto out;
@@ -1746,7 +1746,7 @@ static int kvmgt_page_track_remove(unsigned long handle, u64 gfn)
 	kvmgt_protect_table_del(info, gfn);
 
 out:
-	spin_unlock(&kvm->mmu_lock);
+	write_unlock(&kvm->mmu_lock);
 	srcu_read_unlock(&kvm->srcu, idx);
 	return 0;
 }
@@ -1772,7 +1772,7 @@ static void kvmgt_page_track_flush_slot(struct kvm *kvm,
 	struct kvmgt_guest_info *info = container_of(node,
 					struct kvmgt_guest_info, track_node);
 
-	spin_lock(&kvm->mmu_lock);
+	write_lock(&kvm->mmu_lock);
 	for (i = 0; i < slot->npages; i++) {
 		gfn = slot->base_gfn + i;
 		if (kvmgt_gfn_is_write_protected(info, gfn)) {
@@ -1781,7 +1781,7 @@ static void kvmgt_page_track_flush_slot(struct kvm *kvm,
 			kvmgt_protect_table_del(info, gfn);
 		}
 	}
-	spin_unlock(&kvm->mmu_lock);
+	write_unlock(&kvm->mmu_lock);
 }
 
 static bool __kvmgt_vgpu_exist(struct intel_vgpu *vgpu, struct kvm *kvm)
diff --git a/drivers/gpu/drm/i915/gvt/mmio.h b/drivers/gpu/drm/i915/gvt/mmio.h
index 9e862dc73579..7c26af39fbfc 100644
--- a/drivers/gpu/drm/i915/gvt/mmio.h
+++ b/drivers/gpu/drm/i915/gvt/mmio.h
@@ -80,6 +80,9 @@ int intel_gvt_for_each_tracked_mmio(struct intel_gvt *gvt,
 	int (*handler)(struct intel_gvt *gvt, u32 offset, void *data),
 	void *data);
 
+struct intel_gvt_mmio_info *intel_gvt_find_mmio_info(struct intel_gvt *gvt,
+						  unsigned int offset);
+
 int intel_vgpu_init_mmio(struct intel_vgpu *vgpu);
 void intel_vgpu_reset_mmio(struct intel_vgpu *vgpu, bool dmlr);
 void intel_vgpu_clean_mmio(struct intel_vgpu *vgpu);
diff --git a/drivers/gpu/drm/i915/gvt/reg.h b/drivers/gpu/drm/i915/gvt/reg.h
index b58860dee970..244cc7320b54 100644
--- a/drivers/gpu/drm/i915/gvt/reg.h
+++ b/drivers/gpu/drm/i915/gvt/reg.h
@@ -133,4 +133,6 @@
 #define RING_GFX_MODE(base)	_MMIO((base) + 0x29c)
 #define VF_GUARDBAND		_MMIO(0x83a4)
 
+
+#define BCS_TILE_REGISTER_VAL_OFFSET (0x43*4)
 #endif
diff --git a/drivers/gpu/drm/i915/gvt/scheduler.c b/drivers/gpu/drm/i915/gvt/scheduler.c
index 6af5c06caee0..fc735692f21f 100644
--- a/drivers/gpu/drm/i915/gvt/scheduler.c
+++ b/drivers/gpu/drm/i915/gvt/scheduler.c
@@ -137,6 +137,7 @@ static int populate_shadow_context(struct intel_vgpu_workload *workload)
 	int i;
 	bool skip = false;
 	int ring_id = workload->engine->id;
+	int ret;
 
 	GEM_BUG_ON(!intel_context_is_pinned(ctx));
 
@@ -163,16 +164,24 @@ static int populate_shadow_context(struct intel_vgpu_workload *workload)
 		COPY_REG(bb_per_ctx_ptr);
 		COPY_REG(rcs_indirect_ctx);
 		COPY_REG(rcs_indirect_ctx_offset);
-	}
+	} else if (workload->engine->id == BCS0)
+		intel_gvt_hypervisor_read_gpa(vgpu,
+				workload->ring_context_gpa +
+				BCS_TILE_REGISTER_VAL_OFFSET,
+				(void *)shadow_ring_context +
+				BCS_TILE_REGISTER_VAL_OFFSET, 4);
 #undef COPY_REG
 #undef COPY_REG_MASKED
 
+	/* don't copy Ring Context (the first 0x50 dwords),
+	 * only copy the Engine Context part from guest
+	 */
 	intel_gvt_hypervisor_read_gpa(vgpu,
 			workload->ring_context_gpa +
-			sizeof(*shadow_ring_context),
+			RING_CTX_SIZE,
 			(void *)shadow_ring_context +
-			sizeof(*shadow_ring_context),
-			I915_GTT_PAGE_SIZE - sizeof(*shadow_ring_context));
+			RING_CTX_SIZE,
+			I915_GTT_PAGE_SIZE - RING_CTX_SIZE);
 
 	sr_oa_regs(workload, (u32 *)shadow_ring_context, false);
 
@@ -239,6 +248,11 @@ read:
 		gpa_size = I915_GTT_PAGE_SIZE;
 		dst = context_base + (i << I915_GTT_PAGE_SHIFT);
 	}
+	ret = intel_gvt_scan_engine_context(workload);
+	if (ret) {
+		gvt_vgpu_err("invalid cmd found in guest context pages\n");
+		return ret;
+	}
 	s->last_ctx[ring_id].valid = true;
 	return 0;
 }
@@ -398,7 +412,9 @@ static void release_shadow_wa_ctx(struct intel_shadow_wa_ctx *wa_ctx)
 	if (!wa_ctx->indirect_ctx.obj)
 		return;
 
+	i915_gem_object_lock(wa_ctx->indirect_ctx.obj, NULL);
 	i915_gem_object_unpin_map(wa_ctx->indirect_ctx.obj);
+	i915_gem_object_unlock(wa_ctx->indirect_ctx.obj);
 	i915_gem_object_put(wa_ctx->indirect_ctx.obj);
 
 	wa_ctx->indirect_ctx.obj = NULL;
@@ -506,6 +522,7 @@ static int prepare_shadow_batch_buffer(struct intel_vgpu_workload *workload)
 	struct intel_gvt *gvt = workload->vgpu->gvt;
 	const int gmadr_bytes = gvt->device_info.gmadr_bytes_in_cmd;
 	struct intel_vgpu_shadow_bb *bb;
+	struct i915_gem_ww_ctx ww;
 	int ret;
 
 	list_for_each_entry(bb, &workload->shadow_bb, list) {
@@ -530,10 +547,19 @@ static int prepare_shadow_batch_buffer(struct intel_vgpu_workload *workload)
 		 * directly
 		 */
 		if (!bb->ppgtt) {
-			bb->vma = i915_gem_object_ggtt_pin(bb->obj,
-							   NULL, 0, 0, 0);
+			i915_gem_ww_ctx_init(&ww, false);
+retry:
+			i915_gem_object_lock(bb->obj, &ww);
+
+			bb->vma = i915_gem_object_ggtt_pin_ww(bb->obj, &ww,
+							      NULL, 0, 0, 0);
 			if (IS_ERR(bb->vma)) {
 				ret = PTR_ERR(bb->vma);
+				if (ret == -EDEADLK) {
+					ret = i915_gem_ww_ctx_backoff(&ww);
+					if (!ret)
+						goto retry;
+				}
 				goto err;
 			}
 
@@ -547,13 +573,15 @@ static int prepare_shadow_batch_buffer(struct intel_vgpu_workload *workload)
 						      0);
 			if (ret)
 				goto err;
-		}
 
-		/* No one is going to touch shadow bb from now on. */
-		i915_gem_object_flush_map(bb->obj);
+			/* No one is going to touch shadow bb from now on. */
+			i915_gem_object_flush_map(bb->obj);
+			i915_gem_object_unlock(bb->obj);
+		}
 	}
 	return 0;
 err:
+	i915_gem_ww_ctx_fini(&ww);
 	release_shadow_batch_buffer(workload);
 	return ret;
 }
@@ -580,14 +608,29 @@ static int prepare_shadow_wa_ctx(struct intel_shadow_wa_ctx *wa_ctx)
 	unsigned char *per_ctx_va =
 		(unsigned char *)wa_ctx->indirect_ctx.shadow_va +
 		wa_ctx->indirect_ctx.size;
+	struct i915_gem_ww_ctx ww;
+	int ret;
 
 	if (wa_ctx->indirect_ctx.size == 0)
 		return 0;
 
-	vma = i915_gem_object_ggtt_pin(wa_ctx->indirect_ctx.obj, NULL,
-				       0, CACHELINE_BYTES, 0);
-	if (IS_ERR(vma))
-		return PTR_ERR(vma);
+	i915_gem_ww_ctx_init(&ww, false);
+retry:
+	i915_gem_object_lock(wa_ctx->indirect_ctx.obj, &ww);
+
+	vma = i915_gem_object_ggtt_pin_ww(wa_ctx->indirect_ctx.obj, &ww, NULL,
+					  0, CACHELINE_BYTES, 0);
+	if (IS_ERR(vma)) {
+		ret = PTR_ERR(vma);
+		if (ret == -EDEADLK) {
+			ret = i915_gem_ww_ctx_backoff(&ww);
+			if (!ret)
+				goto retry;
+		}
+		return ret;
+	}
+
+	i915_gem_object_unlock(wa_ctx->indirect_ctx.obj);
 
 	/* FIXME: we are not tracking our pinned VMA leaving it
 	 * up to the core to fix up the stray pin_count upon
@@ -621,12 +664,14 @@ static void release_shadow_batch_buffer(struct intel_vgpu_workload *workload)
 
 	list_for_each_entry_safe(bb, pos, &workload->shadow_bb, list) {
 		if (bb->obj) {
+			i915_gem_object_lock(bb->obj, NULL);
 			if (bb->va && !IS_ERR(bb->va))
 				i915_gem_object_unpin_map(bb->obj);
 
 			if (bb->vma && !IS_ERR(bb->vma))
 				i915_vma_unpin(bb->vma);
 
+			i915_gem_object_unlock(bb->obj);
 			i915_gem_object_put(bb->obj);
 		}
 		list_del(&bb->list);
@@ -1001,13 +1046,12 @@ void intel_vgpu_clean_workloads(struct intel_vgpu *vgpu,
 				intel_engine_mask_t engine_mask)
 {
 	struct intel_vgpu_submission *s = &vgpu->submission;
-	struct drm_i915_private *dev_priv = vgpu->gvt->gt->i915;
 	struct intel_engine_cs *engine;
 	struct intel_vgpu_workload *pos, *n;
 	intel_engine_mask_t tmp;
 
 	/* free the unsubmited workloads in the queues. */
-	for_each_engine_masked(engine, &dev_priv->gt, engine_mask, tmp) {
+	for_each_engine_masked(engine, vgpu->gvt->gt, engine_mask, tmp) {
 		list_for_each_entry_safe(pos, n,
 			&s->workload_q_head[engine->id], list) {
 			list_del_init(&pos->list);
diff --git a/drivers/gpu/drm/i915/gvt/vgpu.c b/drivers/gpu/drm/i915/gvt/vgpu.c
index e49944fde333..6a16d0ca7cda 100644
--- a/drivers/gpu/drm/i915/gvt/vgpu.c
+++ b/drivers/gpu/drm/i915/gvt/vgpu.c
@@ -437,10 +437,9 @@ static struct intel_vgpu *__intel_gvt_create_vgpu(struct intel_gvt *gvt,
 	if (ret)
 		goto out_clean_sched_policy;
 
-	if (IS_BROADWELL(dev_priv))
+	if (IS_BROADWELL(dev_priv) || IS_BROXTON(dev_priv))
 		ret = intel_gvt_hypervisor_set_edid(vgpu, PORT_B);
-	/* FixMe: Re-enable APL/BXT once vfio_edid enabled */
-	else if (!IS_BROXTON(dev_priv))
+	else
 		ret = intel_gvt_hypervisor_set_edid(vgpu, PORT_D);
 	if (ret)
 		goto out_clean_sched_policy;
@@ -500,9 +499,11 @@ struct intel_vgpu *intel_gvt_create_vgpu(struct intel_gvt *gvt,
 
 	mutex_lock(&gvt->lock);
 	vgpu = __intel_gvt_create_vgpu(gvt, &param);
-	if (!IS_ERR(vgpu))
+	if (!IS_ERR(vgpu)) {
 		/* calculate left instance change for types */
 		intel_gvt_update_vgpu_types(gvt);
+		intel_gvt_update_reg_whitelist(vgpu);
+	}
 	mutex_unlock(&gvt->lock);
 
 	return vgpu;