drm/amdgpu/vcn2.0: Add firmware w/r ptr reset sync

Add firmware write/read point reset sync through shared memory
Signed-off-by: James Zhu <James.Zhu@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c

@@ -92,6 +92,7 @@ static int vcn_v2_0_sw_init(void *handle)
 	struct amdgpu_ring *ring;
 	int i, r;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	volatile struct amdgpu_fw_shared *fw_shared;
 
 	/* VCN DEC TRAP */
 	r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
@@ -174,6 +175,8 @@ static int vcn_v2_0_sw_init(void *handle)
 	if (r)
 		return r;
 
+	fw_shared = adev->vcn.inst->fw_shared_cpu_addr;
+	fw_shared->present_flag_0 = cpu_to_le32(AMDGPU_VCN_MULTI_QUEUE_FLAG);
 	return 0;
 }
 
@@ -188,6 +191,9 @@ static int vcn_v2_0_sw_fini(void *handle)
 {
 	int r;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared_cpu_addr;
+
+	fw_shared->present_flag_0 = 0;
 
 	amdgpu_virt_free_mm_table(adev);
 
@@ -354,6 +360,15 @@ static void vcn_v2_0_mc_resume(struct amdgpu_device *adev)
 	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET2, 0);
 	WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE2, AMDGPU_VCN_CONTEXT_SIZE);
 
+	/* non-cache window */
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW,
+		lower_32_bits(adev->vcn.inst->fw_shared_gpu_addr));
+	WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH,
+		upper_32_bits(adev->vcn.inst->fw_shared_gpu_addr));
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_NONCACHE_OFFSET0, 0);
+	WREG32_SOC15(UVD, 0, mmUVD_VCPU_NONCACHE_SIZE0,
+		AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared)));
+
 	WREG32_SOC15(UVD, 0, mmUVD_GFX10_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
 }
 
@@ -437,13 +452,16 @@ static void vcn_v2_0_mc_resume_dpg_mode(struct amdgpu_device *adev, bool indirec
 
 	/* non-cache window */
 	WREG32_SOC15_DPG_MODE_2_0(0, SOC15_DPG_MODE_OFFSET_2_0(
-		UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW), 0, 0, indirect);
+		UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW),
+		lower_32_bits(adev->vcn.inst->fw_shared_gpu_addr), 0, indirect);
 	WREG32_SOC15_DPG_MODE_2_0(0, SOC15_DPG_MODE_OFFSET_2_0(
-		UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH), 0, 0, indirect);
+		UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH),
+		upper_32_bits(adev->vcn.inst->fw_shared_gpu_addr), 0, indirect);
 	WREG32_SOC15_DPG_MODE_2_0(0, SOC15_DPG_MODE_OFFSET_2_0(
 		UVD, 0, mmUVD_VCPU_NONCACHE_OFFSET0), 0, 0, indirect);
 	WREG32_SOC15_DPG_MODE_2_0(0, SOC15_DPG_MODE_OFFSET_2_0(
-		UVD, 0, mmUVD_VCPU_NONCACHE_SIZE0), 0, 0, indirect);
+		UVD, 0, mmUVD_VCPU_NONCACHE_SIZE0),
+		AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared)), 0, indirect);
 
 	/* VCN global tiling registers */
 	WREG32_SOC15_DPG_MODE_2_0(0, SOC15_DPG_MODE_OFFSET_2_0(
@@ -768,6 +786,7 @@ static void vcn_v2_0_enable_static_power_gating(struct amdgpu_device *adev)
 
 static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect)
 {
+	volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared_cpu_addr;
 	struct amdgpu_ring *ring = &adev->vcn.inst->ring_dec;
 	uint32_t rb_bufsz, tmp;
 
@@ -871,6 +890,8 @@ static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect)
 	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS),
 		UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK,
 		~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
+	fw_shared->multi_queue.decode_queue_mode |= FW_QUEUE_RING_RESET;
+
 	/* set the write pointer delay */
 	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR_CNTL, 0);
 
@@ -893,6 +914,7 @@ static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect)
 	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
 		lower_32_bits(ring->wptr));
 
+	fw_shared->multi_queue.decode_queue_mode &= ~FW_QUEUE_RING_RESET;
 	/* Unstall DPG */
 	WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS),
 		0, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
@@ -901,6 +923,7 @@ static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect)
 
 static int vcn_v2_0_start(struct amdgpu_device *adev)
 {
+	volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared_cpu_addr;
 	struct amdgpu_ring *ring = &adev->vcn.inst->ring_dec;
 	uint32_t rb_bufsz, tmp;
 	uint32_t lmi_swap_cntl;
@@ -1035,6 +1058,7 @@ static int vcn_v2_0_start(struct amdgpu_device *adev)
 	tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
 	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, tmp);
 
+	fw_shared->multi_queue.decode_queue_mode |= FW_QUEUE_RING_RESET;
 	/* programm the RB_BASE for ring buffer */
 	WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
 		lower_32_bits(ring->gpu_addr));
@@ -1047,20 +1071,25 @@ static int vcn_v2_0_start(struct amdgpu_device *adev)
 	ring->wptr = RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR);
 	WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
 			lower_32_bits(ring->wptr));
+	fw_shared->multi_queue.decode_queue_mode &= ~FW_QUEUE_RING_RESET;
 
+	fw_shared->multi_queue.encode_generalpurpose_queue_mode |= FW_QUEUE_RING_RESET;
 	ring = &adev->vcn.inst->ring_enc[0];
 	WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
 	WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
 	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO, ring->gpu_addr);
 	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
 	WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE, ring->ring_size / 4);
+	fw_shared->multi_queue.encode_generalpurpose_queue_mode &= ~FW_QUEUE_RING_RESET;
 
+	fw_shared->multi_queue.encode_lowlatency_queue_mode |= FW_QUEUE_RING_RESET;
 	ring = &adev->vcn.inst->ring_enc[1];
 	WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
 	WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
 	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO2, ring->gpu_addr);
 	WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
 	WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE2, ring->ring_size / 4);
+	fw_shared->multi_queue.encode_lowlatency_queue_mode &= ~FW_QUEUE_RING_RESET;
 
 	return 0;
 }
@@ -1182,6 +1211,7 @@ static int vcn_v2_0_pause_dpg_mode(struct amdgpu_device *adev,
 				UVD_POWER_STATUS__UVD_POWER_STATUS_MASK, ret_code);
 
 			if (!ret_code) {
+				volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared_cpu_addr;
 				/* pause DPG */
 				reg_data |= UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
 				WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
@@ -1196,6 +1226,7 @@ static int vcn_v2_0_pause_dpg_mode(struct amdgpu_device *adev,
 					   UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK,
 					   ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
 				/* Restore */
+				fw_shared->multi_queue.encode_generalpurpose_queue_mode |= FW_QUEUE_RING_RESET;
 				ring = &adev->vcn.inst->ring_enc[0];
 				ring->wptr = 0;
 				WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO, ring->gpu_addr);
@@ -1203,7 +1234,9 @@ static int vcn_v2_0_pause_dpg_mode(struct amdgpu_device *adev,
 				WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE, ring->ring_size / 4);
 				WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
 				WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+				fw_shared->multi_queue.encode_generalpurpose_queue_mode &= ~FW_QUEUE_RING_RESET;
 
+				fw_shared->multi_queue.encode_lowlatency_queue_mode |= FW_QUEUE_RING_RESET;
 				ring = &adev->vcn.inst->ring_enc[1];
 				ring->wptr = 0;
 				WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO2, ring->gpu_addr);
@@ -1211,9 +1244,12 @@ static int vcn_v2_0_pause_dpg_mode(struct amdgpu_device *adev,
 				WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE2, ring->ring_size / 4);
 				WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
 				WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+				fw_shared->multi_queue.encode_lowlatency_queue_mode &= ~FW_QUEUE_RING_RESET;
 
+				fw_shared->multi_queue.decode_queue_mode |= FW_QUEUE_RING_RESET;
 				WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
 					   RREG32_SOC15(UVD, 0, mmUVD_SCRATCH2) & 0x7FFFFFFF);
+				fw_shared->multi_queue.decode_queue_mode &= ~FW_QUEUE_RING_RESET;
 				/* Unstall DPG */
 				WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS),
 					   0, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);