drm/amdgpu:Add DPG pause mode support

Add functions to support VCN DPG pause mode.
Signed-off-by: James Zhu <James.Zhu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -36,6 +36,7 @@
 #include "soc15_common.h"
 
 #include "vcn/vcn_1_0_offset.h"
+#include "vcn/vcn_1_0_sh_mask.h"
 
 /* 1 second timeout */
 #define VCN_IDLE_TIMEOUT	msecs_to_jiffies(1000)
@@ -212,18 +213,158 @@ int amdgpu_vcn_resume(struct amdgpu_device *adev)
 	return 0;
 }
 
+static int amdgpu_vcn_pause_dpg_mode(struct amdgpu_device *adev,
+												struct dpg_pause_state *new_state)
+{
+	int ret_code;
+	uint32_t reg_data = 0;
+	uint32_t reg_data2 = 0;
+	struct amdgpu_ring *ring;
+
+	/* pause/unpause if state is changed */
+	if (adev->vcn.pause_state.fw_based != new_state->fw_based) {
+		DRM_DEBUG("dpg pause state changed %d:%d -> %d:%d",
+			adev->vcn.pause_state.fw_based, adev->vcn.pause_state.jpeg,
+			new_state->fw_based, new_state->jpeg);
+
+		reg_data = RREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE) &
+						(~UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+		if (new_state->fw_based == VCN_DPG_STATE__PAUSE) {
+			ret_code = 0;
+
+			if (!(reg_data & UVD_DPG_PAUSE__JPEG_PAUSE_DPG_ACK_MASK))
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
+						UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF,
+						UVD_POWER_STATUS__UVD_POWER_STATUS_MASK, ret_code);
+
+			if (!ret_code) {
+				/* pause DPG non-jpeg */
+				reg_data |= UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+				WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_DPG_PAUSE,
+							UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK,
+							UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK, ret_code);
+
+				/* Restore */
+				ring = &adev->vcn.ring_enc[0];
+				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);
+				WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
+				WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR,	lower_32_bits(ring->wptr));
+
+				ring = &adev->vcn.ring_enc[1];
+				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);
+				WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
+				WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+
+				ring = &adev->vcn.ring_dec;
+				WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
+									lower_32_bits(ring->wptr) | 0x80000000);
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
+										UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON,
+										UVD_POWER_STATUS__UVD_POWER_STATUS_MASK, ret_code);
+			}
+		} else {
+			/* unpause dpg non-jpeg, no need to wait */
+			reg_data &= ~UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+			WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
+		}
+		adev->vcn.pause_state.fw_based = new_state->fw_based;
+	}
+
+	/* pause/unpause if state is changed */
+	if (adev->vcn.pause_state.jpeg != new_state->jpeg) {
+		DRM_DEBUG("dpg pause state changed %d:%d -> %d:%d",
+			adev->vcn.pause_state.fw_based, adev->vcn.pause_state.jpeg,
+			new_state->fw_based, new_state->jpeg);
+
+		reg_data = RREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE) &
+						(~UVD_DPG_PAUSE__JPEG_PAUSE_DPG_ACK_MASK);
+
+		if (new_state->jpeg == VCN_DPG_STATE__PAUSE) {
+			ret_code = 0;
+
+			if (!(reg_data & UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK))
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
+						UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF,
+						UVD_POWER_STATUS__UVD_POWER_STATUS_MASK, ret_code);
+
+			if (!ret_code) {
+				/* Make sure JPRG Snoop is disabled before sending the pause */
+				reg_data2 = RREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS);
+				reg_data2 |= UVD_POWER_STATUS__JRBC_SNOOP_DIS_MASK;
+				WREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS, reg_data2);
+
+				/* pause DPG jpeg */
+				reg_data |= UVD_DPG_PAUSE__JPEG_PAUSE_DPG_REQ_MASK;
+				WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_DPG_PAUSE,
+							UVD_DPG_PAUSE__JPEG_PAUSE_DPG_ACK_MASK,
+							UVD_DPG_PAUSE__JPEG_PAUSE_DPG_ACK_MASK, ret_code);
+
+				/* Restore */
+				ring = &adev->vcn.ring_jpeg;
+				WREG32_SOC15(UVD, 0, mmUVD_LMI_JRBC_RB_VMID, 0);
+				WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_CNTL, 0x00000001L | 0x00000002L);
+				WREG32_SOC15(UVD, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+											lower_32_bits(ring->gpu_addr));
+				WREG32_SOC15(UVD, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+											upper_32_bits(ring->gpu_addr));
+				WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_RPTR, ring->wptr);
+				WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_WPTR, ring->wptr);
+				WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_CNTL, 0x00000002L);
+
+				ring = &adev->vcn.ring_dec;
+				WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
+									lower_32_bits(ring->wptr) | 0x80000000);
+				SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
+										UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON,
+										UVD_POWER_STATUS__UVD_POWER_STATUS_MASK, ret_code);
+			}
+		} else {
+			/* unpause dpg jpeg, no need to wait */
+			reg_data &= ~UVD_DPG_PAUSE__JPEG_PAUSE_DPG_REQ_MASK;
+			WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
+		}
+		adev->vcn.pause_state.jpeg = new_state->jpeg;
+	}
+
+	return 0;
+}
+
 static void amdgpu_vcn_idle_work_handler(struct work_struct *work)
 {
 	struct amdgpu_device *adev =
 		container_of(work, struct amdgpu_device, vcn.idle_work.work);
-	unsigned fences = amdgpu_fence_count_emitted(&adev->vcn.ring_dec);
-	unsigned i;
+	unsigned int fences = 0;
+	unsigned int i;
 
 	for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
 		fences += amdgpu_fence_count_emitted(&adev->vcn.ring_enc[i]);
 	}
 
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)	{
+		struct dpg_pause_state new_state;
+
+		if (fences)
+			new_state.fw_based = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.fw_based = VCN_DPG_STATE__UNPAUSE;
+
+		if (amdgpu_fence_count_emitted(&adev->vcn.ring_jpeg))
+			new_state.jpeg = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.jpeg = VCN_DPG_STATE__UNPAUSE;
+
+		amdgpu_vcn_pause_dpg_mode(adev, &new_state);
+	}
+
 	fences += amdgpu_fence_count_emitted(&adev->vcn.ring_jpeg);
+	fences += amdgpu_fence_count_emitted(&adev->vcn.ring_dec);
 
 	if (fences == 0) {
 		amdgpu_gfx_off_ctrl(adev, true);
@@ -250,6 +391,22 @@ void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring)
 			amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
 							       AMD_PG_STATE_UNGATE);
 	}
+
+	if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)	{
+		struct dpg_pause_state new_state;
+
+		if (ring->funcs->type == AMDGPU_RING_TYPE_VCN_ENC)
+			new_state.fw_based = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.fw_based = adev->vcn.pause_state.fw_based;
+
+		if (ring->funcs->type == AMDGPU_RING_TYPE_VCN_JPEG)
+			new_state.jpeg = VCN_DPG_STATE__PAUSE;
+		else
+			new_state.jpeg = adev->vcn.pause_state.jpeg;
+
+		amdgpu_vcn_pause_dpg_mode(adev, &new_state);
+	}
 }
 
 void amdgpu_vcn_ring_end_use(struct amdgpu_ring *ring)