Merge tag 'amd-drm-next-5.11-2020-11-05' of git://people.freedesktop.org/~agd5f/linux into drm-next

amd-drm-next-5.11-2020-11-05:

amdgpu:
- Add initial support for Vangogh
- Add support for Green Sardine
- Add initial support for Dimgrey Cavefish
- Scatter/Gather display support for Renoir
- Updates for Sienna Cichlid
- Updates for Navy Flounder
- SMU7 power improvements
- Modifier support for gfx9+
- CI BACO fixes
- Arcturus SMU fixes
- Lots of code cleanups
- DC fixes
- Kernel doc fixes
- Add more GPU HW client information to page fault error logging
- MPO clock tuning for RV
- FP fixes for DCN3 on ARM and PPC

radeon:
- Expose voltage via hwmon on Sumo APUs

amdkfd:
- Fix unique id handling
- Misc fixes

From: Alex Deucher <alexdeucher@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201105222749.201798-1-alexander.deucher@amd.com

1 files changed, 267 insertions, 12 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
index 56fdbe626d30..8c3bad3dfc01 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
@@ -99,6 +99,23 @@
 #define mmGCR_GENERAL_CNTL_Sienna_Cichlid			0x1580
 #define mmGCR_GENERAL_CNTL_Sienna_Cichlid_BASE_IDX	0
 
+#define mmSPI_CONFIG_CNTL_1_Vangogh		 0x2441
+#define mmSPI_CONFIG_CNTL_1_Vangogh_BASE_IDX	 1
+#define mmVGT_TF_MEMORY_BASE_HI_Vangogh          0x2261
+#define mmVGT_TF_MEMORY_BASE_HI_Vangogh_BASE_IDX 1
+#define mmVGT_HS_OFFCHIP_PARAM_Vangogh           0x224f
+#define mmVGT_HS_OFFCHIP_PARAM_Vangogh_BASE_IDX  1
+#define mmVGT_TF_RING_SIZE_Vangogh               0x224e
+#define mmVGT_TF_RING_SIZE_Vangogh_BASE_IDX      1
+#define mmVGT_GSVS_RING_SIZE_Vangogh             0x2241
+#define mmVGT_GSVS_RING_SIZE_Vangogh_BASE_IDX    1
+#define mmVGT_TF_MEMORY_BASE_Vangogh             0x2250
+#define mmVGT_TF_MEMORY_BASE_Vangogh_BASE_IDX    1
+#define mmVGT_ESGS_RING_SIZE_Vangogh             0x2240
+#define mmVGT_ESGS_RING_SIZE_Vangogh_BASE_IDX    1
+#define mmSPI_CONFIG_CNTL_Vangogh                0x2440
+#define mmSPI_CONFIG_CNTL_Vangogh_BASE_IDX       1
+
 #define mmCP_HYP_PFP_UCODE_ADDR			0x5814
 #define mmCP_HYP_PFP_UCODE_ADDR_BASE_IDX	1
 #define mmCP_HYP_PFP_UCODE_DATA			0x5815
@@ -112,6 +129,13 @@
 #define mmCP_HYP_ME_UCODE_DATA			0x5817
 #define mmCP_HYP_ME_UCODE_DATA_BASE_IDX		1
 
+#define mmCPG_PSP_DEBUG				0x5c10
+#define mmCPG_PSP_DEBUG_BASE_IDX		1
+#define mmCPC_PSP_DEBUG				0x5c11
+#define mmCPC_PSP_DEBUG_BASE_IDX		1
+#define CPC_PSP_DEBUG__GPA_OVERRIDE_MASK	0x00000008L
+#define CPG_PSP_DEBUG__GPA_OVERRIDE_MASK	0x00000008L
+
 //CC_GC_SA_UNIT_DISABLE
 #define mmCC_GC_SA_UNIT_DISABLE                 0x0fe9
 #define mmCC_GC_SA_UNIT_DISABLE_BASE_IDX        0
@@ -128,6 +152,9 @@
 #define PA_SC_ENHANCE_3__FORCE_PBB_WORKLOAD_MODE_TO_ZERO__SHIFT 0x3
 #define PA_SC_ENHANCE_3__FORCE_PBB_WORKLOAD_MODE_TO_ZERO_MASK   0x00000008L
 
+#define mmCGTT_SPI_CS_CLK_CTRL			0x507c
+#define mmCGTT_SPI_CS_CLK_CTRL_BASE_IDX         1
+
 MODULE_FIRMWARE("amdgpu/navi10_ce.bin");
 MODULE_FIRMWARE("amdgpu/navi10_pfp.bin");
 MODULE_FIRMWARE("amdgpu/navi10_me.bin");
@@ -168,6 +195,20 @@ MODULE_FIRMWARE("amdgpu/navy_flounder_mec.bin");
 MODULE_FIRMWARE("amdgpu/navy_flounder_mec2.bin");
 MODULE_FIRMWARE("amdgpu/navy_flounder_rlc.bin");
 
+MODULE_FIRMWARE("amdgpu/vangogh_ce.bin");
+MODULE_FIRMWARE("amdgpu/vangogh_pfp.bin");
+MODULE_FIRMWARE("amdgpu/vangogh_me.bin");
+MODULE_FIRMWARE("amdgpu/vangogh_mec.bin");
+MODULE_FIRMWARE("amdgpu/vangogh_mec2.bin");
+MODULE_FIRMWARE("amdgpu/vangogh_rlc.bin");
+
+MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_ce.bin");
+MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_pfp.bin");
+MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_me.bin");
+MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_mec.bin");
+MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_mec2.bin");
+MODULE_FIRMWARE("amdgpu/dimgrey_cavefish_rlc.bin");
+
 static const struct soc15_reg_golden golden_settings_gc_10_1[] =
 {
 	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCB_HW_CONTROL_4, 0xffffffff, 0x00400014),
@@ -3094,6 +3135,7 @@ static const struct soc15_reg_golden golden_settings_gc_rlc_spm_10_1_2_nv12[] =
 
 static const struct soc15_reg_golden golden_settings_gc_10_3[] =
 {
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_CS_CLK_CTRL, 0x78000000, 0x78000100),
 	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_PS_CLK_CTRL, 0xff7f0fff, 0x78000100),
 	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA0_CLK_CTRL, 0xff7f0fff, 0x30000100),
 	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA1_CLK_CTRL, 0xff7f0fff, 0x7e000100),
@@ -3177,7 +3219,74 @@ static const struct soc15_reg_golden golden_settings_gc_10_3_2[] =
 	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER9_SELECT, 0xf0f001ff, 0x00000000),
 	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000),
 	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffbfffff, 0x00a00000),
-	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000003ff)
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000003ff),
+
+	/* This is not in GDB yet. Don't remove it. It fixes a GPU hang on Navy Flounder. */
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmLDS_CONFIG,  0x00000020, 0x00000020),
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_3_vangogh[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA0_CLK_CTRL, 0xff7f0fff, 0x30000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA1_CLK_CTRL, 0xff7f0fff, 0x7e000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCH_PIPE_STEER, 0x000000ff, 0x000000e4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0xffffffff, 0x00000200),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0xffffffff, 0x00800000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_EXCEPTION_CONTROL, 0x7fff0f1f, 0x00b80000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGB_ADDR_CONFIG, 0x0c1807ff, 0x00000142),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL, 0x1ff1ffff, 0x00000500),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL1_PIPE_STEER, 0x000000ff, 0x000000e4),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_0, 0x77777777, 0x32103210),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2_PIPE_STEER_1, 0x77777777, 0x32103210),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xfffffff3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xfffffff3),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CM_CTRL1, 0xff8fff0f, 0x580f1008),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0xf7ffffff, 0x00f80988),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_CL_ENHANCE, 0xf17fffff, 0x01200007),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_TIMEOUT_COUNTER, 0xffffffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_2, 0xffffffbf, 0x00000020),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1_Vangogh, 0xffffffff, 0x00070103),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQG_CONFIG, 0x000017ff, 0x00001000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0xfff7ffff, 0x01030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffffffff, 0x00400000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000000ff),
+};
+
+static const struct soc15_reg_golden golden_settings_gc_10_3_4[] =
+{
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA0_CLK_CTRL, 0x30000000, 0x30000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCGTT_SPI_RA1_CLK_CTRL, 0x7e000000, 0x7e000100),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmCPF_GCR_CNTL, 0x0007ffff, 0x0000c000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG3, 0x00000280, 0x00000280),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmDB_DEBUG4, 0x07800000, 0x00800000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGCR_GENERAL_CNTL, 0x00001d00, 0x00000500),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGE_PC_CNTL, 0x003c0000, 0x00280400),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2A_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_ADDR_MATCH_MASK, 0xffffffff, 0xffffffcf),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CM_CTRL1, 0x40000000, 0x580f1008),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmGL2C_CTRL3, 0x00040000, 0x00f80988),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_CL_ENHANCE, 0x01000000, 0x01200007),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_BINNER_TIMEOUT_COUNTER, 0xffffffff, 0x00000800),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmPA_SC_ENHANCE_2, 0x00000800, 0x00000820),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_CONFIG, 0x0000001f, 0x00180070),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER0_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER1_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER10_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER11_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER12_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER13_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER14_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER15_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER2_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER3_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER4_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER5_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER6_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER7_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER8_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_PERFCOUNTER9_SELECT, 0xf0f001ff, 0x00000000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmTA_CNTL_AUX, 0x01030000, 0x01030000),
+	SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0x03a00000, 0x00a00000)
 };
 
 #define DEFAULT_SH_MEM_CONFIG \
@@ -3192,7 +3301,7 @@ static void gfx_v10_0_set_irq_funcs(struct amdgpu_device *adev);
 static void gfx_v10_0_set_gds_init(struct amdgpu_device *adev);
 static void gfx_v10_0_set_rlc_funcs(struct amdgpu_device *adev);
 static int gfx_v10_0_get_cu_info(struct amdgpu_device *adev,
-                                 struct amdgpu_cu_info *cu_info);
+				 struct amdgpu_cu_info *cu_info);
 static uint64_t gfx_v10_0_get_gpu_clock_counter(struct amdgpu_device *adev);
 static void gfx_v10_0_select_se_sh(struct amdgpu_device *adev, u32 se_num,
 				   u32 sh_num, u32 instance);
@@ -3389,7 +3498,16 @@ static void gfx_v10_0_init_golden_registers(struct amdgpu_device *adev)
 						golden_settings_gc_10_3_2,
 						(const u32)ARRAY_SIZE(golden_settings_gc_10_3_2));
 		break;
-
+	case CHIP_VANGOGH:
+		soc15_program_register_sequence(adev,
+						golden_settings_gc_10_3_vangogh,
+						(const u32)ARRAY_SIZE(golden_settings_gc_10_3_vangogh));
+		break;
+	case CHIP_DIMGREY_CAVEFISH:
+		soc15_program_register_sequence(adev,
+                                                golden_settings_gc_10_3_4,
+                                                (const u32)ARRAY_SIZE(golden_settings_gc_10_3_4));
+		break;
 	default:
 		break;
 	}
@@ -3573,6 +3691,8 @@ static void gfx_v10_0_check_fw_write_wait(struct amdgpu_device *adev)
 		break;
 	case CHIP_SIENNA_CICHLID:
 	case CHIP_NAVY_FLOUNDER:
+	case CHIP_VANGOGH:
+	case CHIP_DIMGREY_CAVEFISH:
 		adev->gfx.cp_fw_write_wait = true;
 		break;
 	default:
@@ -3640,7 +3760,7 @@ static void gfx_v10_0_check_gfxoff_flag(struct amdgpu_device *adev)
 		if (!gfx_v10_0_navi10_gfxoff_should_enable(adev))
 			adev->pm.pp_feature &= ~PP_GFXOFF_MASK;
 		break;
-	case CHIP_NAVY_FLOUNDER:
+	case CHIP_VANGOGH:
 		adev->pm.pp_feature &= ~PP_GFXOFF_MASK;
 		break;
 	default:
@@ -3685,6 +3805,12 @@ static int gfx_v10_0_init_microcode(struct amdgpu_device *adev)
 	case CHIP_NAVY_FLOUNDER:
 		chip_name = "navy_flounder";
 		break;
+	case CHIP_VANGOGH:
+		chip_name = "vangogh";
+		break;
+	case CHIP_DIMGREY_CAVEFISH:
+		chip_name = "dimgrey_cavefish";
+		break;
 	default:
 		BUG();
 	}
@@ -4200,11 +4326,26 @@ static void gfx_v10_0_read_wave_vgprs(struct amdgpu_device *adev, uint32_t simd,
 }
 
 static void gfx_v10_0_select_me_pipe_q(struct amdgpu_device *adev,
-									  u32 me, u32 pipe, u32 q, u32 vm)
- {
-       nv_grbm_select(adev, me, pipe, q, vm);
- }
+				       u32 me, u32 pipe, u32 q, u32 vm)
+{
+	nv_grbm_select(adev, me, pipe, q, vm);
+}
 
+static void gfx_v10_0_update_perfmon_mgcg(struct amdgpu_device *adev,
+					  bool enable)
+{
+	uint32_t data, def;
+
+	data = def = RREG32_SOC15(GC, 0, mmRLC_PERFMON_CLK_CNTL);
+
+	if (enable)
+		data |= RLC_PERFMON_CLK_CNTL__PERFMON_CLOCK_STATE_MASK;
+	else
+		data &= ~RLC_PERFMON_CLK_CNTL__PERFMON_CLOCK_STATE_MASK;
+
+	if (data != def)
+		WREG32_SOC15(GC, 0, mmRLC_PERFMON_CLK_CNTL, data);
+}
 
 static const struct amdgpu_gfx_funcs gfx_v10_0_gfx_funcs = {
 	.get_gpu_clock_counter = &gfx_v10_0_get_gpu_clock_counter,
@@ -4214,6 +4355,7 @@ static const struct amdgpu_gfx_funcs gfx_v10_0_gfx_funcs = {
 	.read_wave_vgprs = &gfx_v10_0_read_wave_vgprs,
 	.select_me_pipe_q = &gfx_v10_0_select_me_pipe_q,
 	.init_spm_golden = &gfx_v10_0_init_spm_golden_registers,
+	.update_perfmon_mgcg = &gfx_v10_0_update_perfmon_mgcg,
 };
 
 static void gfx_v10_0_gpu_early_init(struct amdgpu_device *adev)
@@ -4235,6 +4377,8 @@ static void gfx_v10_0_gpu_early_init(struct amdgpu_device *adev)
 		break;
 	case CHIP_SIENNA_CICHLID:
 	case CHIP_NAVY_FLOUNDER:
+	case CHIP_VANGOGH:
+	case CHIP_DIMGREY_CAVEFISH:
 		adev->gfx.config.max_hw_contexts = 8;
 		adev->gfx.config.sc_prim_fifo_size_frontend = 0x20;
 		adev->gfx.config.sc_prim_fifo_size_backend = 0x100;
@@ -4358,6 +4502,8 @@ static int gfx_v10_0_sw_init(void *handle)
 		break;
 	case CHIP_SIENNA_CICHLID:
 	case CHIP_NAVY_FLOUNDER:
+	case CHIP_VANGOGH:
+	case CHIP_DIMGREY_CAVEFISH:
 		adev->gfx.me.num_me = 1;
 		adev->gfx.me.num_pipe_per_me = 1;
 		adev->gfx.me.num_queue_per_pipe = 1;
@@ -4617,8 +4763,7 @@ static u32 gfx_v10_0_init_pa_sc_tile_steering_override(struct amdgpu_device *ade
 
 	/* for ASICs that integrates GFX v10.3
 	 * pa_sc_tile_steering_override should be set to 0 */
-	if (adev->asic_type == CHIP_SIENNA_CICHLID ||
-	    adev->asic_type == CHIP_NAVY_FLOUNDER)
+	if (adev->asic_type >= CHIP_SIENNA_CICHLID)
 		return 0;
 
 	/* init num_sc */
@@ -5855,6 +6000,8 @@ static void gfx_v10_0_cp_gfx_set_doorbell(struct amdgpu_device *adev,
 	switch (adev->asic_type) {
 	case CHIP_SIENNA_CICHLID:
 	case CHIP_NAVY_FLOUNDER:
+	case CHIP_VANGOGH:
+	case CHIP_DIMGREY_CAVEFISH:
 		tmp = REG_SET_FIELD(0, CP_RB_DOORBELL_RANGE_LOWER,
 				    DOORBELL_RANGE_LOWER_Sienna_Cichlid, ring->doorbell_index);
 		WREG32_SOC15(GC, 0, mmCP_RB_DOORBELL_RANGE_LOWER, tmp);
@@ -5988,6 +6135,8 @@ static void gfx_v10_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
 		switch (adev->asic_type) {
 		case CHIP_SIENNA_CICHLID:
 		case CHIP_NAVY_FLOUNDER:
+		case CHIP_VANGOGH:
+		case CHIP_DIMGREY_CAVEFISH:
 			WREG32_SOC15(GC, 0, mmCP_MEC_CNTL_Sienna_Cichlid, 0);
 			break;
 		default:
@@ -5998,6 +6147,8 @@ static void gfx_v10_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
 		switch (adev->asic_type) {
 		case CHIP_SIENNA_CICHLID:
 		case CHIP_NAVY_FLOUNDER:
+		case CHIP_VANGOGH:
+		case CHIP_DIMGREY_CAVEFISH:
 			WREG32_SOC15(GC, 0, mmCP_MEC_CNTL_Sienna_Cichlid,
 				     (CP_MEC_CNTL__MEC_ME1_HALT_MASK |
 				      CP_MEC_CNTL__MEC_ME2_HALT_MASK));
@@ -6092,6 +6243,8 @@ static void gfx_v10_0_kiq_setting(struct amdgpu_ring *ring)
 	switch (adev->asic_type) {
 	case CHIP_SIENNA_CICHLID:
 	case CHIP_NAVY_FLOUNDER:
+	case CHIP_VANGOGH:
+	case CHIP_DIMGREY_CAVEFISH:
 		tmp = RREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS_Sienna_Cichlid);
 		tmp &= 0xffffff00;
 		tmp |= (ring->me << 5) | (ring->pipe << 3) | (ring->queue);
@@ -6800,6 +6953,7 @@ static bool gfx_v10_0_check_grbm_cam_remapping(struct amdgpu_device *adev)
 	switch (adev->asic_type) {
 	case CHIP_SIENNA_CICHLID:
 	case CHIP_NAVY_FLOUNDER:
+	case CHIP_DIMGREY_CAVEFISH:
 		data = RREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE_Sienna_Cichlid);
 		WREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE_Sienna_Cichlid, 0);
 		WREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE_UMD, pattern);
@@ -6812,6 +6966,8 @@ static bool gfx_v10_0_check_grbm_cam_remapping(struct amdgpu_device *adev)
 			return false;
 		}
 		break;
+	case CHIP_VANGOGH:
+		return true;
 	default:
 		data = RREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE);
 		WREG32_SOC15(GC, 0, mmVGT_ESGS_RING_SIZE, 0);
@@ -6839,6 +6995,8 @@ static void gfx_v10_0_setup_grbm_cam_remapping(struct amdgpu_device *adev)
 	switch (adev->asic_type) {
 	case CHIP_SIENNA_CICHLID:
 	case CHIP_NAVY_FLOUNDER:
+	case CHIP_VANGOGH:
+	case CHIP_DIMGREY_CAVEFISH:
 		/* mmVGT_TF_RING_SIZE_UMD -> mmVGT_TF_RING_SIZE */
 		data = (SOC15_REG_OFFSET(GC, 0, mmVGT_TF_RING_SIZE_UMD) <<
 			GRBM_CAM_DATA__CAM_ADDR__SHIFT) |
@@ -6954,6 +7112,18 @@ static void gfx_v10_0_setup_grbm_cam_remapping(struct amdgpu_device *adev)
 	WREG32_SOC15(GC, 0, mmGRBM_CAM_DATA, data);
 }
 
+static void gfx_v10_0_disable_gpa_mode(struct amdgpu_device *adev)
+{
+	uint32_t data;
+	data = RREG32_SOC15(GC, 0, mmCPC_PSP_DEBUG);
+	data |= CPC_PSP_DEBUG__GPA_OVERRIDE_MASK;
+	WREG32_SOC15(GC, 0, mmCPC_PSP_DEBUG, data);
+
+	data = RREG32_SOC15(GC, 0, mmCPG_PSP_DEBUG);
+	data |= CPG_PSP_DEBUG__GPA_OVERRIDE_MASK;
+	WREG32_SOC15(GC, 0, mmCPG_PSP_DEBUG, data);
+}
+
 static int gfx_v10_0_hw_init(void *handle)
 {
 	int r;
@@ -6968,7 +7138,7 @@ static int gfx_v10_0_hw_init(void *handle)
 		 * loaded firstly, so in direct type, it has to load smc ucode
 		 * here before rlc.
 		 */
-		if (adev->smu.ppt_funcs != NULL) {
+		if (adev->smu.ppt_funcs != NULL && !(adev->flags & AMD_IS_APU)) {
 			r = smu_load_microcode(&adev->smu);
 			if (r)
 				return r;
@@ -6979,6 +7149,7 @@ static int gfx_v10_0_hw_init(void *handle)
 				return r;
 			}
 		}
+		gfx_v10_0_disable_gpa_mode(adev);
 	}
 
 	/* if GRBM CAM not remapped, set up the remapping */
@@ -7136,6 +7307,8 @@ static int gfx_v10_0_soft_reset(void *handle)
 	switch (adev->asic_type) {
 	case CHIP_SIENNA_CICHLID:
 	case CHIP_NAVY_FLOUNDER:
+	case CHIP_VANGOGH:
+	case CHIP_DIMGREY_CAVEFISH:
 		if (REG_GET_FIELD(tmp, GRBM_STATUS2, RLC_BUSY_Sienna_Cichlid))
 			grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset,
 							GRBM_SOFT_RESET,
@@ -7235,13 +7408,16 @@ static int gfx_v10_0_early_init(void *handle)
 		break;
 	case CHIP_SIENNA_CICHLID:
 	case CHIP_NAVY_FLOUNDER:
+	case CHIP_VANGOGH:
+	case CHIP_DIMGREY_CAVEFISH:
 		adev->gfx.num_gfx_rings = GFX10_NUM_GFX_RINGS_Sienna_Cichlid;
 		break;
 	default:
 		break;
 	}
 
-	adev->gfx.num_compute_rings = amdgpu_num_kcq;
+	adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev),
+					  AMDGPU_MAX_COMPUTE_RINGS);
 
 	gfx_v10_0_set_kiq_pm4_funcs(adev);
 	gfx_v10_0_set_ring_funcs(adev);
@@ -7288,6 +7464,8 @@ static void gfx_v10_0_set_safe_mode(struct amdgpu_device *adev)
 	switch (adev->asic_type) {
 	case CHIP_SIENNA_CICHLID:
 	case CHIP_NAVY_FLOUNDER:
+	case CHIP_VANGOGH:
+	case CHIP_DIMGREY_CAVEFISH:
 		WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE_Sienna_Cichlid, data);
 
 		/* wait for RLC_SAFE_MODE */
@@ -7320,6 +7498,8 @@ static void gfx_v10_0_unset_safe_mode(struct amdgpu_device *adev)
 	switch (adev->asic_type) {
 	case CHIP_SIENNA_CICHLID:
 	case CHIP_NAVY_FLOUNDER:
+	case CHIP_VANGOGH:
+	case CHIP_DIMGREY_CAVEFISH:
 		WREG32_SOC15(GC, 0, mmRLC_SAFE_MODE_Sienna_Cichlid, data);
 		break;
 	default:
@@ -7344,6 +7524,7 @@ static void gfx_v10_0_update_medium_grain_clock_gating(struct amdgpu_device *ade
 		/* 1 - RLC_CGTT_MGCG_OVERRIDE */
 		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
 		data &= ~(RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK |
+			  RLC_CGTT_MGCG_OVERRIDE__RLC_CGTT_SCLK_OVERRIDE_MASK |
 			  RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK |
 			  RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGLS_OVERRIDE_MASK |
 			  RLC_CGTT_MGCG_OVERRIDE__ENABLE_CGTS_LEGACY_MASK);
@@ -7480,12 +7661,50 @@ static void gfx_v10_0_update_coarse_grain_clock_gating(struct amdgpu_device *ade
 	}
 }
 
+static void gfx_v10_0_update_fine_grain_clock_gating(struct amdgpu_device *adev,
+						      bool enable)
+{
+	uint32_t def, data;
+
+	if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_FGCG)) {
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
+		/* unset FGCG override */
+		data &= ~RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK;
+		/* update FGCG override bits */
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CLK_CNTL);
+		/* unset RLC SRAM CLK GATER override */
+		data &= ~RLC_CLK_CNTL__RLC_SRAM_CLK_GATER_OVERRIDE_MASK;
+		/* update RLC SRAM CLK GATER override bits */
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CLK_CNTL, data);
+	} else {
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE);
+		/* reset FGCG bits */
+		data |= RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK;
+		/* disable FGCG*/
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE, data);
+
+		def = data = RREG32_SOC15(GC, 0, mmRLC_CLK_CNTL);
+		/* reset RLC SRAM CLK GATER bits */
+		data |= RLC_CLK_CNTL__RLC_SRAM_CLK_GATER_OVERRIDE_MASK;
+		/* disable RLC SRAM CLK*/
+		if (def != data)
+			WREG32_SOC15(GC, 0, mmRLC_CLK_CNTL, data);
+	}
+}
+
 static int gfx_v10_0_update_gfx_clock_gating(struct amdgpu_device *adev,
 					    bool enable)
 {
 	amdgpu_gfx_rlc_enter_safe_mode(adev);
 
 	if (enable) {
+		/* enable FGCG firstly*/
+		gfx_v10_0_update_fine_grain_clock_gating(adev, enable);
 		/* CGCG/CGLS should be enabled after MGCG/MGLS
 		 * ===  MGCG + MGLS ===
 		 */
@@ -7503,6 +7722,8 @@ static int gfx_v10_0_update_gfx_clock_gating(struct amdgpu_device *adev,
 		gfx_v10_0_update_3d_clock_gating(adev, enable);
 		/* ===  MGCG + MGLS === */
 		gfx_v10_0_update_medium_grain_clock_gating(adev, enable);
+		/* disable fgcg at last*/
+		gfx_v10_0_update_fine_grain_clock_gating(adev, enable);
 	}
 
 	if (adev->cg_flags &
@@ -7564,6 +7785,27 @@ static bool gfx_v10_0_is_rlcg_access_range(struct amdgpu_device *adev, u32 offse
 	return gfx_v10_0_check_rlcg_range(adev, offset, NULL, 0);
 }
 
+static void gfx_v10_cntl_power_gating(struct amdgpu_device *adev, bool enable)
+{
+	u32 data = RREG32_SOC15(GC, 0, mmRLC_PG_CNTL);
+
+	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG))
+		data |= RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK;
+
+	WREG32_SOC15(GC, 0, mmRLC_PG_CNTL, data);
+}
+
+static void gfx_v10_cntl_pg(struct amdgpu_device *adev, bool enable)
+{
+	amdgpu_gfx_rlc_enter_safe_mode(adev);
+
+	gfx_v10_cntl_power_gating(adev, enable);
+
+	amdgpu_gfx_rlc_exit_safe_mode(adev);
+}
+
 static const struct amdgpu_rlc_funcs gfx_v10_0_rlc_funcs = {
 	.is_rlc_enabled = gfx_v10_0_is_rlc_enabled,
 	.set_safe_mode = gfx_v10_0_set_safe_mode,
@@ -7609,8 +7851,12 @@ static int gfx_v10_0_set_powergating_state(void *handle,
 	case CHIP_NAVI12:
 	case CHIP_SIENNA_CICHLID:
 	case CHIP_NAVY_FLOUNDER:
+	case CHIP_DIMGREY_CAVEFISH:
 		amdgpu_gfx_off_ctrl(adev, enable);
 		break;
+	case CHIP_VANGOGH:
+		gfx_v10_cntl_pg(adev, enable);
+		break;
 	default:
 		break;
 	}
@@ -7631,6 +7877,8 @@ static int gfx_v10_0_set_clockgating_state(void *handle,
 	case CHIP_NAVI12:
 	case CHIP_SIENNA_CICHLID:
 	case CHIP_NAVY_FLOUNDER:
+	case CHIP_VANGOGH:
+	case CHIP_DIMGREY_CAVEFISH:
 		gfx_v10_0_update_gfx_clock_gating(adev,
 						 state == AMD_CG_STATE_GATE);
 		break;
@@ -7645,6 +7893,11 @@ static void gfx_v10_0_get_clockgating_state(void *handle, u32 *flags)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 	int data;
 
+	/* AMD_CG_SUPPORT_GFX_FGCG */
+	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE));
+	if (!(data & RLC_CGTT_MGCG_OVERRIDE__GFXIP_FGCG_OVERRIDE_MASK))
+		*flags |= AMD_CG_SUPPORT_GFX_FGCG;
+
 	/* AMD_CG_SUPPORT_GFX_MGCG */
 	data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE));
 	if (!(data & RLC_CGTT_MGCG_OVERRIDE__GFXIP_MGCG_OVERRIDE_MASK))
@@ -8733,6 +8986,8 @@ static void gfx_v10_0_set_rlc_funcs(struct amdgpu_device *adev)
 	case CHIP_NAVI14:
 	case CHIP_SIENNA_CICHLID:
 	case CHIP_NAVY_FLOUNDER:
+	case CHIP_VANGOGH:
+	case CHIP_DIMGREY_CAVEFISH:
 		adev->gfx.rlc.funcs = &gfx_v10_0_rlc_funcs;
 		break;
 	case CHIP_NAVI12: