Merge branch 'drm-next-4.9' of git://people.freedesktop.org/~agd5f/linux into drm-next

More radeon and amdgpu changes for 4.9. Highlights:
- Initial SI support for amdgpu (controlled by a Kconfig option)
- misc ttm cleanups
- runtimepm fixes
- S3/S4 fixes
- power improvements
- lots of code cleanups and optimizations

* 'drm-next-4.9' of git://people.freedesktop.org/~agd5f/linux: (151 commits)
  drm/ttm: remove cpu_address member from ttm_tt
  drm/radeon/radeon_device: remove unused function
  drm/amdgpu: clean function declarations in amdgpu_ttm.c up
  drm/amdgpu: use the new ring ib and dma frame size callbacks (v2)
  drm/amdgpu/vce3: add ring callbacks for ib and dma frame size
  drm/amdgpu/vce2: add ring callbacks for ib and dma frame size
  drm/amdgpu/vce: add common ring callbacks for ib and dma frame size
  drm/amdgpu/uvd6: add ring callbacks for ib and dma frame size
  drm/amdgpu/uvd5: add ring callbacks for ib and dma frame size
  drm/amdgpu/uvd4.2: add ring callbacks for ib and dma frame size
  drm/amdgpu/sdma3: add ring callbacks for ib and dma frame size
  drm/amdgpu/sdma2.4: add ring callbacks for ib and dma frame size
  drm/amdgpu/cik_sdma: add ring callbacks for ib and dma frame size
  drm/amdgpu/si_dma: add ring callbacks for ib and dma frame size
  drm/amdgpu/gfx8: add ring callbacks for ib and dma frame size
  drm/amdgpu/gfx7: add ring callbacks for ib and dma frame size
  drm/amdgpu/gfx6: add ring callbacks for ib and dma frame size
  drm/amdgpu/ring: add an interface to get dma frame and ib size
  drm/amdgpu/sdma3: drop unused functions
  drm/amdgpu/gfx6: drop gds_switch callback
  ...

drivers/gpu/drm/amd/amdgpu/Kconfig

+config DRM_AMDGPU_SI
+	bool "Enable amdgpu support for SI parts"
+	depends on DRM_AMDGPU
+	help
+	  Choose this option if you want to enable experimental support
+	  for SI asics.
+
 config DRM_AMDGPU_CIK
 	bool "Enable amdgpu support for CIK parts"
 	depends on DRM_AMDGPU

drivers/gpu/drm/amd/amdgpu/Makefile

@@ -30,6 +30,8 @@ amdgpu-$(CONFIG_DRM_AMDGPU_CIK)+= cik.o cik_ih.o kv_smc.o kv_dpm.o \
 	ci_smc.o ci_dpm.o dce_v8_0.o gfx_v7_0.o cik_sdma.o uvd_v4_2.o vce_v2_0.o \
 	amdgpu_amdkfd_gfx_v7.o
 
+amdgpu-$(CONFIG_DRM_AMDGPU_SI)+= si.o gmc_v6_0.o gfx_v6_0.o si_ih.o si_dma.o dce_v6_0.o si_dpm.o si_smc.o
+
 amdgpu-y += \
 	vi.o
 

drivers/gpu/drm/amd/amdgpu/amdgpu.h

@@ -64,6 +64,7 @@
 extern int amdgpu_modeset;
 extern int amdgpu_vram_limit;
 extern int amdgpu_gart_size;
+extern int amdgpu_moverate;
 extern int amdgpu_benchmarking;
 extern int amdgpu_testing;
 extern int amdgpu_audio;
@@ -94,6 +95,7 @@ extern unsigned amdgpu_pg_mask;
 extern char *amdgpu_disable_cu;
 extern int amdgpu_sclk_deep_sleep_en;
 extern char *amdgpu_virtual_display;
+extern unsigned amdgpu_pp_feature_mask;
 
 #define AMDGPU_WAIT_IDLE_TIMEOUT_IN_MS	        3000
 #define AMDGPU_MAX_USEC_TIMEOUT			100000	/* 100 ms */
@@ -108,7 +110,7 @@ extern char *amdgpu_virtual_display;
 #define AMDGPU_MAX_RINGS			16
 #define AMDGPU_MAX_GFX_RINGS			1
 #define AMDGPU_MAX_COMPUTE_RINGS		8
-#define AMDGPU_MAX_VCE_RINGS			2
+#define AMDGPU_MAX_VCE_RINGS			3
 
 /* max number of IP instances */
 #define AMDGPU_MAX_SDMA_INSTANCES		2
@@ -318,6 +320,10 @@ struct amdgpu_ring_funcs {
 	/* note usage for clock and power gating */
 	void (*begin_use)(struct amdgpu_ring *ring);
 	void (*end_use)(struct amdgpu_ring *ring);
+	void (*emit_switch_buffer) (struct amdgpu_ring *ring);
+	void (*emit_cntxcntl) (struct amdgpu_ring *ring, uint32_t flags);
+	unsigned (*get_emit_ib_size) (struct amdgpu_ring *ring);
+	unsigned (*get_dma_frame_size) (struct amdgpu_ring *ring);
 };
 
 /*
@@ -618,6 +624,7 @@ void amdgpu_gart_unbind(struct amdgpu_device *adev, unsigned offset,
 int amdgpu_gart_bind(struct amdgpu_device *adev, unsigned offset,
 		     int pages, struct page **pagelist,
 		     dma_addr_t *dma_addr, uint32_t flags);
+int amdgpu_ttm_recover_gart(struct amdgpu_device *adev);
 
 /*
  * GPU MC structures, functions & helpers
@@ -963,6 +970,7 @@ struct amdgpu_ctx {
 	spinlock_t		ring_lock;
 	struct fence            **fences;
 	struct amdgpu_ctx_ring	rings[AMDGPU_MAX_RINGS];
+	bool preamble_presented;
 };
 
 struct amdgpu_ctx_mgr {
@@ -1222,11 +1230,16 @@ struct amdgpu_cs_parser {
 	struct fence			*fence;
 	uint64_t			bytes_moved_threshold;
 	uint64_t			bytes_moved;
+	struct amdgpu_bo_list_entry	*evictable;
 
 	/* user fence */
 	struct amdgpu_bo_list_entry	uf_entry;
 };
 
+#define AMDGPU_PREAMBLE_IB_PRESENT          (1 << 0) /* bit set means command submit involves a preamble IB */
+#define AMDGPU_PREAMBLE_IB_PRESENT_FIRST    (1 << 1) /* bit set means preamble IB is first presented in belonging context */
+#define AMDGPU_HAVE_CTX_SWITCH              (1 << 2) /* bit set means context switch occured */
+
 struct amdgpu_job {
 	struct amd_sched_job    base;
 	struct amdgpu_device	*adev;
@@ -1235,9 +1248,10 @@ struct amdgpu_job {
 	struct amdgpu_sync	sync;
 	struct amdgpu_ib	*ibs;
 	struct fence		*fence; /* the hw fence */
+	uint32_t		preamble_status;
 	uint32_t		num_ibs;
 	void			*owner;
-	uint64_t		ctx;
+	uint64_t		fence_ctx; /* the fence_context this job uses */
 	bool                    vm_needs_flush;
 	unsigned		vm_id;
 	uint64_t		vm_pd_addr;
@@ -1686,6 +1700,7 @@ struct amdgpu_vce {
 	unsigned		harvest_config;
 	struct amd_sched_entity	entity;
 	uint32_t                srbm_soft_reset;
+	unsigned		num_rings;
 };
 
 /*
@@ -1703,6 +1718,10 @@ struct amdgpu_sdma_instance {
 
 struct amdgpu_sdma {
 	struct amdgpu_sdma_instance instance[AMDGPU_MAX_SDMA_INSTANCES];
+#ifdef CONFIG_DRM_AMDGPU_SI
+	//SI DMA has a difference trap irq number for the second engine
+	struct amdgpu_irq_src	trap_irq_1;
+#endif
 	struct amdgpu_irq_src	trap_irq;
 	struct amdgpu_irq_src	illegal_inst_irq;
 	int			num_instances;
@@ -1819,6 +1838,9 @@ struct amdgpu_asic_funcs {
 	int (*set_vce_clocks)(struct amdgpu_device *adev, u32 evclk, u32 ecclk);
 	/* query virtual capabilities */
 	u32 (*get_virtual_caps)(struct amdgpu_device *adev);
+	/* static power management */
+	int (*get_pcie_lanes)(struct amdgpu_device *adev);
+	void (*set_pcie_lanes)(struct amdgpu_device *adev, int lanes);
 };
 
 /*
@@ -1993,6 +2015,8 @@ struct amdgpu_device {
 	spinlock_t pcie_idx_lock;
 	amdgpu_rreg_t			pcie_rreg;
 	amdgpu_wreg_t			pcie_wreg;
+	amdgpu_rreg_t			pciep_rreg;
+	amdgpu_wreg_t			pciep_wreg;
 	/* protects concurrent UVD register access */
 	spinlock_t uvd_ctx_idx_lock;
 	amdgpu_rreg_t			uvd_ctx_rreg;
@@ -2033,6 +2057,14 @@ struct amdgpu_device {
 	atomic64_t			num_evictions;
 	atomic_t			gpu_reset_counter;
 
+	/* data for buffer migration throttling */
+	struct {
+		spinlock_t		lock;
+		s64			last_update_us;
+		s64			accum_us; /* accumulated microseconds */
+		u32			log2_max_MBps;
+	} mm_stats;
+
 	/* display */
 	bool				enable_virtual_display;
 	struct amdgpu_mode_info		mode_info;
@@ -2101,6 +2133,10 @@ struct amdgpu_device {
 	/* link all shadow bo */
 	struct list_head                shadow_list;
 	struct mutex                    shadow_list_lock;
+	/* link all gtt */
+	spinlock_t			gtt_list_lock;
+	struct list_head                gtt_list;
+
 };
 
 bool amdgpu_device_is_px(struct drm_device *dev);
@@ -2133,6 +2169,8 @@ void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v);
 #define REG_GET(FIELD, v) (((v) << FIELD##_SHIFT) & FIELD##_MASK)
 #define RREG32_PCIE(reg) adev->pcie_rreg(adev, (reg))
 #define WREG32_PCIE(reg, v) adev->pcie_wreg(adev, (reg), (v))
+#define RREG32_PCIE_PORT(reg) adev->pciep_rreg(adev, (reg))
+#define WREG32_PCIE_PORT(reg, v) adev->pciep_wreg(adev, (reg), (v))
 #define RREG32_SMC(reg) adev->smc_rreg(adev, (reg))
 #define WREG32_SMC(reg, v) adev->smc_wreg(adev, (reg), (v))
 #define RREG32_UVD_CTX(reg) adev->uvd_ctx_rreg(adev, (reg))
@@ -2223,6 +2261,9 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
 #define amdgpu_asic_set_uvd_clocks(adev, v, d) (adev)->asic_funcs->set_uvd_clocks((adev), (v), (d))
 #define amdgpu_asic_set_vce_clocks(adev, ev, ec) (adev)->asic_funcs->set_vce_clocks((adev), (ev), (ec))
 #define amdgpu_asic_get_virtual_caps(adev) ((adev)->asic_funcs->get_virtual_caps((adev)))
+#define amdgpu_get_pcie_lanes(adev) (adev)->asic_funcs->get_pcie_lanes((adev))
+#define amdgpu_set_pcie_lanes(adev, l) (adev)->asic_funcs->set_pcie_lanes((adev), (l))
+#define amdgpu_asic_get_gpu_clock_counter(adev) (adev)->asic_funcs->get_gpu_clock_counter((adev))
 #define amdgpu_asic_read_disabled_bios(adev) (adev)->asic_funcs->read_disabled_bios((adev))
 #define amdgpu_asic_read_bios_from_rom(adev, b, l) (adev)->asic_funcs->read_bios_from_rom((adev), (b), (l))
 #define amdgpu_asic_read_register(adev, se, sh, offset, v)((adev)->asic_funcs->read_register((adev), (se), (sh), (offset), (v)))
@@ -2244,9 +2285,13 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
 #define amdgpu_ring_emit_gds_switch(r, v, db, ds, wb, ws, ab, as) (r)->funcs->emit_gds_switch((r), (v), (db), (ds), (wb), (ws), (ab), (as))
 #define amdgpu_ring_emit_hdp_flush(r) (r)->funcs->emit_hdp_flush((r))
 #define amdgpu_ring_emit_hdp_invalidate(r) (r)->funcs->emit_hdp_invalidate((r))
+#define amdgpu_ring_emit_switch_buffer(r) (r)->funcs->emit_switch_buffer((r))
+#define amdgpu_ring_emit_cntxcntl(r, d) (r)->funcs->emit_cntxcntl((r), (d))
 #define amdgpu_ring_pad_ib(r, ib) ((r)->funcs->pad_ib((r), (ib)))
 #define amdgpu_ring_init_cond_exec(r) (r)->funcs->init_cond_exec((r))
 #define amdgpu_ring_patch_cond_exec(r,o) (r)->funcs->patch_cond_exec((r),(o))
+#define amdgpu_ring_get_emit_ib_size(r) (r)->funcs->get_emit_ib_size((r))
+#define amdgpu_ring_get_dma_frame_size(r) (r)->funcs->get_dma_frame_size((r))
 #define amdgpu_ih_get_wptr(adev) (adev)->irq.ih_funcs->get_wptr((adev))
 #define amdgpu_ih_decode_iv(adev, iv) (adev)->irq.ih_funcs->decode_iv((adev), (iv))
 #define amdgpu_ih_set_rptr(adev) (adev)->irq.ih_funcs->set_rptr((adev))
@@ -2402,6 +2447,8 @@ void amdgpu_gtt_location(struct amdgpu_device *adev, struct amdgpu_mc *mc);
 void amdgpu_ttm_set_active_vram_size(struct amdgpu_device *adev, u64 size);
 u64 amdgpu_ttm_get_gtt_mem_size(struct amdgpu_device *adev);
 int amdgpu_ttm_global_init(struct amdgpu_device *adev);
+int amdgpu_ttm_init(struct amdgpu_device *adev);
+void amdgpu_ttm_fini(struct amdgpu_device *adev);
 void amdgpu_program_register_sequence(struct amdgpu_device *adev,
 					     const u32 *registers,
 					     const u32 array_size);
@@ -2434,8 +2481,8 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev,
 				 struct drm_file *file_priv);
 void amdgpu_driver_preclose_kms(struct drm_device *dev,
 				struct drm_file *file_priv);
-int amdgpu_suspend_kms(struct drm_device *dev, bool suspend, bool fbcon);
-int amdgpu_resume_kms(struct drm_device *dev, bool resume, bool fbcon);
+int amdgpu_device_suspend(struct drm_device *dev, bool suspend, bool fbcon);
+int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon);
 u32 amdgpu_get_vblank_counter_kms(struct drm_device *dev, unsigned int pipe);
 int amdgpu_enable_vblank_kms(struct drm_device *dev, unsigned int pipe);
 void amdgpu_disable_vblank_kms(struct drm_device *dev, unsigned int pipe);
@@ -2481,6 +2528,7 @@ static inline void amdgpu_acpi_fini(struct amdgpu_device *adev) { }
 struct amdgpu_bo_va_mapping *
 amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
 		       uint64_t addr, struct amdgpu_bo **bo);
+int amdgpu_cs_sysvm_access_required(struct amdgpu_cs_parser *parser);
 
 #include "amdgpu_object.h"
 #endif

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

@@ -978,6 +978,48 @@ int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
 		return -EINVAL;
 
 	switch (crev) {
+	case 2:
+	case 3:
+	case 5:
+		/* r6xx, r7xx, evergreen, ni, si.
+		 * TODO: add support for asic_type <= CHIP_RV770*/
+		if (clock_type == COMPUTE_ENGINE_PLL_PARAM) {
+			args.v3.ulClockParams = cpu_to_le32((clock_type << 24) | clock);
+
+			amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+			dividers->post_div = args.v3.ucPostDiv;
+			dividers->enable_post_div = (args.v3.ucCntlFlag &
+						     ATOM_PLL_CNTL_FLAG_PLL_POST_DIV_EN) ? true : false;
+			dividers->enable_dithen = (args.v3.ucCntlFlag &
+						   ATOM_PLL_CNTL_FLAG_FRACTION_DISABLE) ? false : true;
+			dividers->whole_fb_div = le16_to_cpu(args.v3.ulFbDiv.usFbDiv);
+			dividers->frac_fb_div = le16_to_cpu(args.v3.ulFbDiv.usFbDivFrac);
+			dividers->ref_div = args.v3.ucRefDiv;
+			dividers->vco_mode = (args.v3.ucCntlFlag &
+					      ATOM_PLL_CNTL_FLAG_MPLL_VCO_MODE) ? 1 : 0;
+		} else {
+			/* for SI we use ComputeMemoryClockParam for memory plls */
+			if (adev->asic_type >= CHIP_TAHITI)
+				return -EINVAL;
+			args.v5.ulClockParams = cpu_to_le32((clock_type << 24) | clock);
+			if (strobe_mode)
+				args.v5.ucInputFlag = ATOM_PLL_INPUT_FLAG_PLL_STROBE_MODE_EN;
+
+			amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+			dividers->post_div = args.v5.ucPostDiv;
+			dividers->enable_post_div = (args.v5.ucCntlFlag &
+						     ATOM_PLL_CNTL_FLAG_PLL_POST_DIV_EN) ? true : false;
+			dividers->enable_dithen = (args.v5.ucCntlFlag &
+						   ATOM_PLL_CNTL_FLAG_FRACTION_DISABLE) ? false : true;
+			dividers->whole_fb_div = le16_to_cpu(args.v5.ulFbDiv.usFbDiv);
+			dividers->frac_fb_div = le16_to_cpu(args.v5.ulFbDiv.usFbDivFrac);
+			dividers->ref_div = args.v5.ucRefDiv;
+			dividers->vco_mode = (args.v5.ucCntlFlag &
+					      ATOM_PLL_CNTL_FLAG_MPLL_VCO_MODE) ? 1 : 0;
+		}
+		break;
 	case 4:
 		/* fusion */
 		args.v4.ulClock = cpu_to_le32(clock);	/* 10 khz */
@@ -1122,6 +1164,32 @@ void amdgpu_atombios_set_engine_dram_timings(struct amdgpu_device *adev,
 	amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
 }
 
+void amdgpu_atombios_get_default_voltages(struct amdgpu_device *adev,
+					  u16 *vddc, u16 *vddci, u16 *mvdd)
+{
+	struct amdgpu_mode_info *mode_info = &adev->mode_info;
+	int index = GetIndexIntoMasterTable(DATA, FirmwareInfo);
+	u8 frev, crev;
+	u16 data_offset;
+	union firmware_info *firmware_info;
+
+	*vddc = 0;
+	*vddci = 0;
+	*mvdd = 0;
+
+	if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
+				   &frev, &crev, &data_offset)) {
+		firmware_info =
+			(union firmware_info *)(mode_info->atom_context->bios +
+						data_offset);
+		*vddc = le16_to_cpu(firmware_info->info_14.usBootUpVDDCVoltage);
+		if ((frev == 2) && (crev >= 2)) {
+			*vddci = le16_to_cpu(firmware_info->info_22.usBootUpVDDCIVoltage);
+			*mvdd = le16_to_cpu(firmware_info->info_22.usBootUpMVDDCVoltage);
+		}
+	}
+}
+
 union set_voltage {
 	struct _SET_VOLTAGE_PS_ALLOCATION alloc;
 	struct _SET_VOLTAGE_PARAMETERS v1;
@@ -1129,6 +1197,52 @@ union set_voltage {
 	struct _SET_VOLTAGE_PARAMETERS_V1_3 v3;
 };
 
+int amdgpu_atombios_get_max_vddc(struct amdgpu_device *adev, u8 voltage_type,
+			     u16 voltage_id, u16 *voltage)
+{
+	union set_voltage args;
+	int index = GetIndexIntoMasterTable(COMMAND, SetVoltage);
+	u8 frev, crev;
+
+	if (!amdgpu_atom_parse_cmd_header(adev->mode_info.atom_context, index, &frev, &crev))
+		return -EINVAL;
+
+	switch (crev) {
+	case 1:
+		return -EINVAL;
+	case 2:
+		args.v2.ucVoltageType = SET_VOLTAGE_GET_MAX_VOLTAGE;
+		args.v2.ucVoltageMode = 0;
+		args.v2.usVoltageLevel = 0;
+
+		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+		*voltage = le16_to_cpu(args.v2.usVoltageLevel);
+		break;
+	case 3:
+		args.v3.ucVoltageType = voltage_type;
+		args.v3.ucVoltageMode = ATOM_GET_VOLTAGE_LEVEL;
+		args.v3.usVoltageLevel = cpu_to_le16(voltage_id);
+
+		amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+
+		*voltage = le16_to_cpu(args.v3.usVoltageLevel);
+		break;
+	default:
+		DRM_ERROR("Unknown table version %d, %d\n", frev, crev);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+int amdgpu_atombios_get_leakage_vddc_based_on_leakage_idx(struct amdgpu_device *adev,
+						      u16 *voltage,
+						      u16 leakage_idx)
+{
+	return amdgpu_atombios_get_max_vddc(adev, VOLTAGE_TYPE_VDDC, leakage_idx, voltage);
+}
+
 void amdgpu_atombios_set_voltage(struct amdgpu_device *adev,
 				 u16 voltage_level,
 				 u8 voltage_type)
@@ -1349,6 +1463,50 @@ static ATOM_VOLTAGE_OBJECT_V3 *amdgpu_atombios_lookup_voltage_object_v3(ATOM_VOL
 	return NULL;
 }
 
+int amdgpu_atombios_get_svi2_info(struct amdgpu_device *adev,
+			      u8 voltage_type,
+			      u8 *svd_gpio_id, u8 *svc_gpio_id)
+{
+	int index = GetIndexIntoMasterTable(DATA, VoltageObjectInfo);
+	u8 frev, crev;
+	u16 data_offset, size;
+	union voltage_object_info *voltage_info;
+	union voltage_object *voltage_object = NULL;
+
+	if (amdgpu_atom_parse_data_header(adev->mode_info.atom_context, index, &size,
+				   &frev, &crev, &data_offset)) {
+		voltage_info = (union voltage_object_info *)
+			(adev->mode_info.atom_context->bios + data_offset);
+
+		switch (frev) {
+		case 3:
+			switch (crev) {
+			case 1:
+				voltage_object = (union voltage_object *)
+					amdgpu_atombios_lookup_voltage_object_v3(&voltage_info->v3,
+								      voltage_type,
+								      VOLTAGE_OBJ_SVID2);
+				if (voltage_object) {
+					*svd_gpio_id = voltage_object->v3.asSVID2Obj.ucSVDGpioId;
+					*svc_gpio_id = voltage_object->v3.asSVID2Obj.ucSVCGpioId;
+				} else {
+					return -EINVAL;
+				}
+				break;
+			default:
+				DRM_ERROR("unknown voltage object table\n");
+				return -EINVAL;
+			}
+			break;
+		default:
+			DRM_ERROR("unknown voltage object table\n");
+			return -EINVAL;
+		}
+
+	}
+	return 0;
+}
+
 bool
 amdgpu_atombios_is_voltage_gpio(struct amdgpu_device *adev,
 				u8 voltage_type, u8 voltage_mode)

drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h

@@ -208,5 +208,19 @@ void amdgpu_atombios_scratch_regs_save(struct amdgpu_device *adev);
 void amdgpu_atombios_scratch_regs_restore(struct amdgpu_device *adev);
 
 void amdgpu_atombios_copy_swap(u8 *dst, u8 *src, u8 num_bytes, bool to_le);
-
+int amdgpu_atombios_get_max_vddc(struct amdgpu_device *adev, u8 voltage_type,
+			     u16 voltage_id, u16 *voltage);
+int amdgpu_atombios_get_leakage_vddc_based_on_leakage_idx(struct amdgpu_device *adev,
+						      u16 *voltage,
+						      u16 leakage_idx);
+void amdgpu_atombios_get_default_voltages(struct amdgpu_device *adev,
+					  u16 *vddc, u16 *vddci, u16 *mvdd);
+int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
+				       u8 clock_type,
+				       u32 clock,
+				       bool strobe_mode,
+				       struct atom_clock_dividers *dividers);
+int amdgpu_atombios_get_svi2_info(struct amdgpu_device *adev,
+			      u8 voltage_type,
+			      u8 *svd_gpio_id, u8 *svc_gpio_id);
 #endif

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

@@ -616,7 +616,7 @@ static int amdgpu_cgs_irq_put(struct cgs_device *cgs_device, unsigned src_id, un
 	return amdgpu_irq_put(adev, adev->irq.sources[src_id], type);
 }
 
-int amdgpu_cgs_set_clockgating_state(struct cgs_device *cgs_device,
+static int amdgpu_cgs_set_clockgating_state(struct cgs_device *cgs_device,
 				  enum amd_ip_block_type block_type,
 				  enum amd_clockgating_state state)
 {
@@ -637,7 +637,7 @@ int amdgpu_cgs_set_clockgating_state(struct cgs_device *cgs_device,
 	return r;
 }
 
-int amdgpu_cgs_set_powergating_state(struct cgs_device *cgs_device,
+static int amdgpu_cgs_set_powergating_state(struct cgs_device *cgs_device,
 				  enum amd_ip_block_type block_type,
 				  enum amd_powergating_state state)
 {
@@ -848,6 +848,12 @@ static int amdgpu_cgs_query_system_info(struct cgs_device *cgs_device,
 	case CGS_SYSTEM_INFO_GFX_SE_INFO:
 		sys_info->value = adev->gfx.config.max_shader_engines;
 		break;
+	case CGS_SYSTEM_INFO_PCIE_SUB_SYS_ID:
+		sys_info->value = adev->pdev->subsystem_device;
+		break;
+	case CGS_SYSTEM_INFO_PCIE_SUB_SYS_VENDOR_ID:
+		sys_info->value = adev->pdev->subsystem_vendor;
+		break;
 	default:
 		return -ENODEV;
 	}

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

@@ -91,6 +91,7 @@ static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
 				      uint32_t *offset)
 {
 	struct drm_gem_object *gobj;
+	unsigned long size;
 
 	gobj = drm_gem_object_lookup(p->filp, data->handle);
 	if (gobj == NULL)
@@ -101,6 +102,11 @@ static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
 	p->uf_entry.tv.bo = &p->uf_entry.robj->tbo;
 	p->uf_entry.tv.shared = true;
 	p->uf_entry.user_pages = NULL;
+
+	size = amdgpu_bo_size(p->uf_entry.robj);
+	if (size != PAGE_SIZE || (data->offset + 8) > size)
+		return -EINVAL;
+
 	*offset = data->offset;
 
 	drm_gem_object_unreference_unlocked(gobj);
@@ -235,56 +241,115 @@ int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
 	return ret;
 }
 
-/* Returns how many bytes TTM can move per IB.
+/* Convert microseconds to bytes. */
+static u64 us_to_bytes(struct amdgpu_device *adev, s64 us)
+{
+	if (us <= 0 || !adev->mm_stats.log2_max_MBps)
+		return 0;
+
+	/* Since accum_us is incremented by a million per second, just
+	 * multiply it by the number of MB/s to get the number of bytes.
+	 */
+	return us << adev->mm_stats.log2_max_MBps;
+}
+
+static s64 bytes_to_us(struct amdgpu_device *adev, u64 bytes)
+{
+	if (!adev->mm_stats.log2_max_MBps)
+		return 0;
+
+	return bytes >> adev->mm_stats.log2_max_MBps;
+}
+
+/* Returns how many bytes TTM can move right now. If no bytes can be moved,
+ * it returns 0. If it returns non-zero, it's OK to move at least one buffer,
+ * which means it can go over the threshold once. If that happens, the driver
+ * will be in debt and no other buffer migrations can be done until that debt
+ * is repaid.
+ *
+ * This approach allows moving a buffer of any size (it's important to allow
+ * that).
+ *
+ * The currency is simply time in microseconds and it increases as the clock
+ * ticks. The accumulated microseconds (us) are converted to bytes and
+ * returned.
  */
 static u64 amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev)
 {
-	u64 real_vram_size = adev->mc.real_vram_size;
-	u64 vram_usage = atomic64_read(&adev->vram_usage);
+	s64 time_us, increment_us;
+	u64 max_bytes;
+	u64 free_vram, total_vram, used_vram;
 
-	/* This function is based on the current VRAM usage.
+	/* Allow a maximum of 200 accumulated ms. This is basically per-IB
+	 * throttling.
 	 *
-	 * - If all of VRAM is free, allow relocating the number of bytes that
-	 *   is equal to 1/4 of the size of VRAM for this IB.
+	 * It means that in order to get full max MBps, at least 5 IBs per
+	 * second must be submitted and not more than 200ms apart from each
+	 * other.
+	 */
+	const s64 us_upper_bound = 200000;
 
-	 * - If more than one half of VRAM is occupied, only allow relocating
-	 *   1 MB of data for this IB.
-	 *
-	 * - From 0 to one half of used VRAM, the threshold decreases
-	 *   linearly.
-	 *         __________________
-	 * 1/4 of -|\               |
-	 * VRAM    | \              |
-	 *         |  \             |
-	 *         |   \            |
-	 *         |    \           |
-	 *         |     \          |
-	 *         |      \         |
-	 *         |       \________|1 MB
-	 *         |----------------|
-	 *    VRAM 0 %             100 %
-	 *         used            used
-	 *
-	 * Note: It's a threshold, not a limit. The threshold must be crossed
-	 * for buffer relocations to stop, so any buffer of an arbitrary size
-	 * can be moved as long as the threshold isn't crossed before
-	 * the relocation takes place. We don't want to disable buffer
-	 * relocations completely.
+	if (!adev->mm_stats.log2_max_MBps)
+		return 0;
+
+	total_vram = adev->mc.real_vram_size - adev->vram_pin_size;
+	used_vram = atomic64_read(&adev->vram_usage);
+	free_vram = used_vram >= total_vram ? 0 : total_vram - used_vram;
+
+	spin_lock(&adev->mm_stats.lock);
+
+	/* Increase the amount of accumulated us. */
+	time_us = ktime_to_us(ktime_get());
+	increment_us = time_us - adev->mm_stats.last_update_us;
+	adev->mm_stats.last_update_us = time_us;
+	adev->mm_stats.accum_us = min(adev->mm_stats.accum_us + increment_us,
+                                      us_upper_bound);
+
+	/* This prevents the short period of low performance when the VRAM
+	 * usage is low and the driver is in debt or doesn't have enough
+	 * accumulated us to fill VRAM quickly.
 	 *
-	 * The idea is that buffers should be placed in VRAM at creation time
-	 * and TTM should only do a minimum number of relocations during
-	 * command submission. In practice, you need to submit at least
-	 * a dozen IBs to move all buffers to VRAM if they are in GTT.
+	 * The situation can occur in these cases:
+	 * - a lot of VRAM is freed by userspace
+	 * - the presence of a big buffer causes a lot of evictions
+	 *   (solution: split buffers into smaller ones)
 	 *
-	 * Also, things can get pretty crazy under memory pressure and actual
-	 * VRAM usage can change a lot, so playing safe even at 50% does
-	 * consistently increase performance.
+	 * If 128 MB or 1/8th of VRAM is free, start filling it now by setting
+	 * accum_us to a positive number.
+	 */
+	if (free_vram >= 128 * 1024 * 1024 || free_vram >= total_vram / 8) {
+		s64 min_us;
+
+		/* Be more aggresive on dGPUs. Try to fill a portion of free
+		 * VRAM now.
+		 */
+		if (!(adev->flags & AMD_IS_APU))
+			min_us = bytes_to_us(adev, free_vram / 4);
+		else
+			min_us = 0; /* Reset accum_us on APUs. */
+
+		adev->mm_stats.accum_us = max(min_us, adev->mm_stats.accum_us);
+	}
+
+	/* This returns 0 if the driver is in debt to disallow (optional)
+	 * buffer moves.
 	 */
+	max_bytes = us_to_bytes(adev, adev->mm_stats.accum_us);
 
-	u64 half_vram = real_vram_size >> 1;
-	u64 half_free_vram = vram_usage >= half_vram ? 0 : half_vram - vram_usage;
-	u64 bytes_moved_threshold = half_free_vram >> 1;
-	return max(bytes_moved_threshold, 1024*1024ull);
+	spin_unlock(&adev->mm_stats.lock);
+	return max_bytes;
+}
+
+/* Report how many bytes have really been moved for the last command
+ * submission. This can result in a debt that can stop buffer migrations
+ * temporarily.
+ */
+static void amdgpu_cs_report_moved_bytes(struct amdgpu_device *adev,
+					 u64 num_bytes)
+{
+	spin_lock(&adev->mm_stats.lock);
+	adev->mm_stats.accum_us -= bytes_to_us(adev, num_bytes);
+	spin_unlock(&adev->mm_stats.lock);
 }
 
 static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p,
@@ -297,15 +362,10 @@ static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p,
 	if (bo->pin_count)
 		return 0;
 
-	/* Avoid moving this one if we have moved too many buffers
-	 * for this IB already.
-	 *
-	 * Note that this allows moving at least one buffer of
-	 * any size, because it doesn't take the current "bo"
-	 * into account. We don't want to disallow buffer moves
-	 * completely.
+	/* Don't move this buffer if we have depleted our allowance
+	 * to move it. Don't move anything if the threshold is zero.
 	 */
-	if (p->bytes_moved <= p->bytes_moved_threshold)
+	if (p->bytes_moved < p->bytes_moved_threshold)
 		domain = bo->prefered_domains;
 	else
 		domain = bo->allowed_domains;
@@ -317,17 +377,67 @@ static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p,
 	p->bytes_moved += atomic64_read(&bo->adev->num_bytes_moved) -
 		initial_bytes_moved;
 
-	if (unlikely(r)) {
-		if (r != -ERESTARTSYS && domain != bo->allowed_domains) {
-			domain = bo->allowed_domains;
-			goto retry;
-		}
+	if (unlikely(r == -ENOMEM) && domain != bo->allowed_domains) {
+		domain = bo->allowed_domains;
+		goto retry;
 	}
 
 	return r;
 }
 
-int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p,
+/* Last resort, try to evict something from the current working set */
+static bool amdgpu_cs_try_evict(struct amdgpu_cs_parser *p,
+				struct amdgpu_bo_list_entry *lobj)
+{
+	uint32_t domain = lobj->robj->allowed_domains;
+	int r;
+
+	if (!p->evictable)
+		return false;
+
+	for (;&p->evictable->tv.head != &p->validated;
+	     p->evictable = list_prev_entry(p->evictable, tv.head)) {
+
+		struct amdgpu_bo_list_entry *candidate = p->evictable;
+		struct amdgpu_bo *bo = candidate->robj;
+		u64 initial_bytes_moved;
+		uint32_t other;
+
+		/* If we reached our current BO we can forget it */
+		if (candidate == lobj)
+			break;
+
+		other = amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type);
+
+		/* Check if this BO is in one of the domains we need space for */
+		if (!(other & domain))
+			continue;
+
+		/* Check if we can move this BO somewhere else */
+		other = bo->allowed_domains & ~domain;
+		if (!other)
+			continue;
+
+		/* Good we can try to move this BO somewhere else */
+		amdgpu_ttm_placement_from_domain(bo, other);
+		initial_bytes_moved = atomic64_read(&bo->adev->num_bytes_moved);
+		r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
+		p->bytes_moved += atomic64_read(&bo->adev->num_bytes_moved) -
+			initial_bytes_moved;
+
+		if (unlikely(r))
+			break;
+
+		p->evictable = list_prev_entry(p->evictable, tv.head);
+		list_move(&candidate->tv.head, &p->validated);
+
+		return true;
+	}
+
+	return false;
+}
+
+static int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p,
 			    struct list_head *validated)
 {
 	struct amdgpu_bo_list_entry *lobj;
@@ -351,9 +461,15 @@ int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p,
 			binding_userptr = true;
 		}
 
-		r = amdgpu_cs_bo_validate(p, bo);
+		if (p->evictable == lobj)
+			p->evictable = NULL;
+
+		do {
+			r = amdgpu_cs_bo_validate(p, bo);
+		} while (r == -ENOMEM && amdgpu_cs_try_evict(p, lobj));
 		if (r)
 			return r;
+
 		if (bo->shadow) {
 			r = amdgpu_cs_bo_validate(p, bo);
 			if (r)
@@ -481,6 +597,9 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 
 	p->bytes_moved_threshold = amdgpu_cs_get_threshold_for_moves(p->adev);
 	p->bytes_moved = 0;
+	p->evictable = list_last_entry(&p->validated,
+				       struct amdgpu_bo_list_entry,
+				       tv.head);
 
 	r = amdgpu_cs_list_validate(p, &duplicates);
 	if (r) {
@@ -494,6 +613,8 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 		goto error_validate;
 	}
 
+	amdgpu_cs_report_moved_bytes(p->adev, p->bytes_moved);
+
 	fpriv->vm.last_eviction_counter =
 		atomic64_read(&p->adev->num_evictions);
 
@@ -524,8 +645,12 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 		}
 	}
 
-	if (p->uf_entry.robj)
-		p->job->uf_addr += amdgpu_bo_gpu_offset(p->uf_entry.robj);
+	if (!r && p->uf_entry.robj) {
+		struct amdgpu_bo *uf = p->uf_entry.robj;
+
+		r = amdgpu_ttm_bind(uf->tbo.ttm, &uf->tbo.mem);
+		p->job->uf_addr += amdgpu_bo_gpu_offset(uf);
+	}
 
 error_validate:
 	if (r) {
@@ -735,6 +860,14 @@ static int amdgpu_cs_ib_fill(struct amdgpu_device *adev,
 		if (r)
 			return r;
 
+		if (ib->flags & AMDGPU_IB_FLAG_PREAMBLE) {
+			parser->job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT;
+			if (!parser->ctx->preamble_presented) {
+				parser->job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT_FIRST;
+				parser->ctx->preamble_presented = true;
+			}
+		}
+
 		if (parser->job->ring && parser->job->ring != ring)
 			return -EINVAL;
 
@@ -874,7 +1007,7 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 	}
 
 	job->owner = p->filp;
-	job->ctx = entity->fence_context;
+	job->fence_ctx = entity->fence_context;
 	p->fence = fence_get(&job->base.s_fence->finished);
 	cs->out.handle = amdgpu_ctx_add_fence(p->ctx, ring, p->fence);
 	job->uf_sequence = cs->out.handle;
@@ -1040,3 +1173,29 @@ amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
 
 	return NULL;
 }
+
+/**
+ * amdgpu_cs_sysvm_access_required - make BOs accessible by the system VM
+ *
+ * @parser: command submission parser context
+ *
+ * Helper for UVD/VCE VM emulation, make sure BOs are accessible by the system VM.
+ */
+int amdgpu_cs_sysvm_access_required(struct amdgpu_cs_parser *parser)
+{
+	unsigned i;
+	int r;
+
+	if (!parser->bo_list)
+		return 0;
+
+	for (i = 0; i < parser->bo_list->num_entries; i++) {
+		struct amdgpu_bo *bo = parser->bo_list->array[i].robj;
+
+		r = amdgpu_ttm_bind(bo->tbo.ttm, &bo->tbo.mem);
+		if (unlikely(r))
+			return r;
+	}
+
+	return 0;
+}

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

@@ -41,6 +41,9 @@
 #include "atom.h"
 #include "amdgpu_atombios.h"
 #include "amd_pcie.h"
+#ifdef CONFIG_DRM_AMDGPU_SI
+#include "si.h"
+#endif
 #ifdef CONFIG_DRM_AMDGPU_CIK
 #include "cik.h"
 #endif
@@ -52,6 +55,11 @@ static int amdgpu_debugfs_regs_init(struct amdgpu_device *adev);
 static void amdgpu_debugfs_regs_cleanup(struct amdgpu_device *adev);
 
 static const char *amdgpu_asic_name[] = {
+	"TAHITI",
+	"PITCAIRN",
+	"VERDE",
+	"OLAND",
+	"HAINAN",
 	"BONAIRE",
 	"KAVERI",
 	"KABINI",
@@ -1027,7 +1035,7 @@ static void amdgpu_switcheroo_set_state(struct pci_dev *pdev, enum vga_switchero
 		/* don't suspend or resume card normally */
 		dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
 
-		amdgpu_resume_kms(dev, true, true);
+		amdgpu_device_resume(dev, true, true);
 
 		dev->pdev->d3_delay = d3_delay;
 
@@ -1037,7 +1045,7 @@ static void amdgpu_switcheroo_set_state(struct pci_dev *pdev, enum vga_switchero
 		printk(KERN_INFO "amdgpu: switched off\n");
 		drm_kms_helper_poll_disable(dev);
 		dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
-		amdgpu_suspend_kms(dev, true, true);
+		amdgpu_device_suspend(dev, true, true);
 		dev->switch_power_state = DRM_SWITCH_POWER_OFF;
 	}
 }
@@ -1231,6 +1239,18 @@ static int amdgpu_early_init(struct amdgpu_device *adev)
 		if (r)
 			return r;
 		break;
+#ifdef CONFIG_DRM_AMDGPU_SI
+	case CHIP_VERDE:
+	case CHIP_TAHITI:
+	case CHIP_PITCAIRN:
+	case CHIP_OLAND:
+	case CHIP_HAINAN:
+		adev->family = AMDGPU_FAMILY_SI;
+		r = si_set_ip_blocks(adev);
+		if (r)
+			return r;
+		break;
+#endif
 #ifdef CONFIG_DRM_AMDGPU_CIK
 	case CHIP_BONAIRE:
 	case CHIP_HAWAII:
@@ -1347,6 +1367,9 @@ static int amdgpu_late_init(struct amdgpu_device *adev)
 	for (i = 0; i < adev->num_ip_blocks; i++) {
 		if (!adev->ip_block_status[i].valid)
 			continue;
+		if (adev->ip_blocks[i].type == AMD_IP_BLOCK_TYPE_UVD ||
+			adev->ip_blocks[i].type == AMD_IP_BLOCK_TYPE_VCE)
+			continue;
 		/* enable clockgating to save power */
 		r = adev->ip_blocks[i].funcs->set_clockgating_state((void *)adev,
 								    AMD_CG_STATE_GATE);
@@ -1490,6 +1513,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 {
 	int r, i;
 	bool runtime = false;
+	u32 max_MBps;
 
 	adev->shutdown = false;
 	adev->dev = &pdev->dev;
@@ -1513,6 +1537,8 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	adev->smc_wreg = &amdgpu_invalid_wreg;
 	adev->pcie_rreg = &amdgpu_invalid_rreg;
 	adev->pcie_wreg = &amdgpu_invalid_wreg;
+	adev->pciep_rreg = &amdgpu_invalid_rreg;
+	adev->pciep_wreg = &amdgpu_invalid_wreg;
 	adev->uvd_ctx_rreg = &amdgpu_invalid_rreg;
 	adev->uvd_ctx_wreg = &amdgpu_invalid_wreg;
 	adev->didt_rreg = &amdgpu_invalid_rreg;
@@ -1549,12 +1575,22 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	spin_lock_init(&adev->didt_idx_lock);
 	spin_lock_init(&adev->gc_cac_idx_lock);
 	spin_lock_init(&adev->audio_endpt_idx_lock);
+	spin_lock_init(&adev->mm_stats.lock);
 
 	INIT_LIST_HEAD(&adev->shadow_list);
 	mutex_init(&adev->shadow_list_lock);
 
-	adev->rmmio_base = pci_resource_start(adev->pdev, 5);
-	adev->rmmio_size = pci_resource_len(adev->pdev, 5);
+	INIT_LIST_HEAD(&adev->gtt_list);
+	spin_lock_init(&adev->gtt_list_lock);
+
+	if (adev->asic_type >= CHIP_BONAIRE) {
+		adev->rmmio_base = pci_resource_start(adev->pdev, 5);
+		adev->rmmio_size = pci_resource_len(adev->pdev, 5);
+	} else {
+		adev->rmmio_base = pci_resource_start(adev->pdev, 2);
+		adev->rmmio_size = pci_resource_len(adev->pdev, 2);
+	}
+
 	adev->rmmio = ioremap(adev->rmmio_base, adev->rmmio_size);
 	if (adev->rmmio == NULL) {
 		return -ENOMEM;
@@ -1562,8 +1598,9 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	DRM_INFO("register mmio base: 0x%08X\n", (uint32_t)adev->rmmio_base);
 	DRM_INFO("register mmio size: %u\n", (unsigned)adev->rmmio_size);
 
-	/* doorbell bar mapping */
-	amdgpu_doorbell_init(adev);
+	if (adev->asic_type >= CHIP_BONAIRE)
+		/* doorbell bar mapping */
+		amdgpu_doorbell_init(adev);
 
 	/* io port mapping */
 	for (i = 0; i < DEVICE_COUNT_RESOURCE; i++) {
@@ -1660,6 +1697,14 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 
 	adev->accel_working = true;
 
+	/* Initialize the buffer migration limit. */
+	if (amdgpu_moverate >= 0)
+		max_MBps = amdgpu_moverate;
+	else
+		max_MBps = 8; /* Allow 8 MB/s. */
+	/* Get a log2 for easy divisions. */
+	adev->mm_stats.log2_max_MBps = ilog2(max(1u, max_MBps));
+
 	amdgpu_fbdev_init(adev);
 
 	r = amdgpu_ib_pool_init(adev);
@@ -1764,7 +1809,8 @@ void amdgpu_device_fini(struct amdgpu_device *adev)
 	adev->rio_mem = NULL;
 	iounmap(adev->rmmio);
 	adev->rmmio = NULL;
-	amdgpu_doorbell_fini(adev);
+	if (adev->asic_type >= CHIP_BONAIRE)
+		amdgpu_doorbell_fini(adev);
 	amdgpu_debugfs_regs_cleanup(adev);
 	amdgpu_debugfs_remove_files(adev);
 }
@@ -1774,7 +1820,7 @@ void amdgpu_device_fini(struct amdgpu_device *adev)
  * Suspend & resume.
  */
 /**
- * amdgpu_suspend_kms - initiate device suspend
+ * amdgpu_device_suspend - initiate device suspend
  *
  * @pdev: drm dev pointer
  * @state: suspend state
@@ -1783,7 +1829,7 @@ void amdgpu_device_fini(struct amdgpu_device *adev)
  * Returns 0 for success or an error on failure.
  * Called at driver suspend.
  */
-int amdgpu_suspend_kms(struct drm_device *dev, bool suspend, bool fbcon)
+int amdgpu_device_suspend(struct drm_device *dev, bool suspend, bool fbcon)
 {
 	struct amdgpu_device *adev;
 	struct drm_crtc *crtc;
@@ -1796,7 +1842,8 @@ int amdgpu_suspend_kms(struct drm_device *dev, bool suspend, bool fbcon)
 
 	adev = dev->dev_private;
 
-	if (dev->switch_power_state == DRM_SWITCH_POWER_OFF)
+	if (dev->switch_power_state == DRM_SWITCH_POWER_OFF ||
+	    dev->switch_power_state == DRM_SWITCH_POWER_DYNAMIC_OFF)
 		return 0;
 
 	drm_kms_helper_poll_disable(dev);
@@ -1851,6 +1898,10 @@ int amdgpu_suspend_kms(struct drm_device *dev, bool suspend, bool fbcon)
 		/* Shut down the device */
 		pci_disable_device(dev->pdev);
 		pci_set_power_state(dev->pdev, PCI_D3hot);
+	} else {
+		r = amdgpu_asic_reset(adev);
+		if (r)
+			DRM_ERROR("amdgpu asic reset failed\n");
 	}
 
 	if (fbcon) {
@@ -1862,7 +1913,7 @@ int amdgpu_suspend_kms(struct drm_device *dev, bool suspend, bool fbcon)
 }
 
 /**
- * amdgpu_resume_kms - initiate device resume
+ * amdgpu_device_resume - initiate device resume
  *
  * @pdev: drm dev pointer
  *
@@ -1870,32 +1921,37 @@ int amdgpu_suspend_kms(struct drm_device *dev, bool suspend, bool fbcon)
  * Returns 0 for success or an error on failure.
  * Called at driver resume.
  */
-int amdgpu_resume_kms(struct drm_device *dev, bool resume, bool fbcon)
+int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon)
 {
 	struct drm_connector *connector;
 	struct amdgpu_device *adev = dev->dev_private;
 	struct drm_crtc *crtc;
 	int r;
 
-	if (dev->switch_power_state == DRM_SWITCH_POWER_OFF)
+	if (dev->switch_power_state == DRM_SWITCH_POWER_OFF ||
+	    dev->switch_power_state == DRM_SWITCH_POWER_DYNAMIC_OFF)
 		return 0;
 
-	if (fbcon) {
+	if (fbcon)
 		console_lock();
-	}
+
 	if (resume) {
 		pci_set_power_state(dev->pdev, PCI_D0);
 		pci_restore_state(dev->pdev);
-		if (pci_enable_device(dev->pdev)) {
+		r = pci_enable_device(dev->pdev);
+		if (r) {
 			if (fbcon)
 				console_unlock();
-			return -1;
+			return r;
 		}
 	}
 
 	/* post card */
-	if (!amdgpu_card_posted(adev))
-		amdgpu_atom_asic_init(adev->mode_info.atom_context);
+	if (!amdgpu_card_posted(adev) || !resume) {
+		r = amdgpu_atom_asic_init(adev->mode_info.atom_context);
+		if (r)
+			DRM_ERROR("amdgpu asic init failed\n");
+	}
 
 	r = amdgpu_resume(adev);
 	if (r)
@@ -2163,6 +2219,11 @@ int amdgpu_gpu_reset(struct amdgpu_device *adev)
 	}
 	if (!r) {
 		amdgpu_irq_gpu_reset_resume_helper(adev);
+		if (need_full_reset && amdgpu_need_backup(adev)) {
+			r = amdgpu_ttm_recover_gart(adev);
+			if (r)
+				DRM_ERROR("gart recovery failed!!!\n");
+		}
 		r = amdgpu_ib_ring_tests(adev);
 		if (r) {
 			dev_err(adev->dev, "ib ring test failed (%d).\n", r);
@@ -2600,7 +2661,7 @@ static ssize_t amdgpu_debugfs_regs_smc_read(struct file *f, char __user *buf,
 	while (size) {
 		uint32_t value;
 
-		value = RREG32_SMC(*pos >> 2);
+		value = RREG32_SMC(*pos);
 		r = put_user(value, (uint32_t *)buf);
 		if (r)
 			return r;
@@ -2631,7 +2692,7 @@ static ssize_t amdgpu_debugfs_regs_smc_write(struct file *f, const char __user *
 		if (r)
 			return r;
 
-		WREG32_SMC(*pos >> 2, value);
+		WREG32_SMC(*pos, value);
 
 		result += 4;
 		buf += 4;

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

@@ -55,13 +55,15 @@
  * - 3.3.0 - Add VM support for UVD on supported hardware.
  * - 3.4.0 - Add AMDGPU_INFO_NUM_EVICTIONS.
  * - 3.5.0 - Add support for new UVD_NO_OP register.
+ * - 3.6.0 - kmd involves use CONTEXT_CONTROL in ring buffer.
  */
 #define KMS_DRIVER_MAJOR	3
-#define KMS_DRIVER_MINOR	5
+#define KMS_DRIVER_MINOR	6
 #define KMS_DRIVER_PATCHLEVEL	0
 
 int amdgpu_vram_limit = 0;
 int amdgpu_gart_size = -1; /* auto */
+int amdgpu_moverate = -1; /* auto */
 int amdgpu_benchmarking = 0;
 int amdgpu_testing = 0;
 int amdgpu_audio = -1;
@@ -93,6 +95,7 @@ unsigned amdgpu_cg_mask = 0xffffffff;
 unsigned amdgpu_pg_mask = 0xffffffff;
 char *amdgpu_disable_cu = NULL;
 char *amdgpu_virtual_display = NULL;
+unsigned amdgpu_pp_feature_mask = 0xffffffff;
 
 MODULE_PARM_DESC(vramlimit, "Restrict VRAM for testing, in megabytes");
 module_param_named(vramlimit, amdgpu_vram_limit, int, 0600);
@@ -100,6 +103,9 @@ module_param_named(vramlimit, amdgpu_vram_limit, int, 0600);
 MODULE_PARM_DESC(gartsize, "Size of PCIE/IGP gart to setup in megabytes (32, 64, etc., -1 = auto)");
 module_param_named(gartsize, amdgpu_gart_size, int, 0600);
 
+MODULE_PARM_DESC(moverate, "Maximum buffer migration rate in MB/s. (32, 64, etc., -1=auto, 0=1=disabled)");
+module_param_named(moverate, amdgpu_moverate, int, 0600);
+
 MODULE_PARM_DESC(benchmark, "Run benchmark");
 module_param_named(benchmark, amdgpu_benchmarking, int, 0444);
 
@@ -172,6 +178,9 @@ module_param_named(powerplay, amdgpu_powerplay, int, 0444);
 
 MODULE_PARM_DESC(powercontainment, "Power Containment (1 = enable (default), 0 = disable)");
 module_param_named(powercontainment, amdgpu_powercontainment, int, 0444);
+
+MODULE_PARM_DESC(ppfeaturemask, "all power features enabled (default))");
+module_param_named(ppfeaturemask, amdgpu_pp_feature_mask, int, 0444);
 #endif
 
 MODULE_PARM_DESC(sclkdeepsleep, "SCLK Deep Sleep (1 = enable (default), 0 = disable)");
@@ -196,6 +205,80 @@ MODULE_PARM_DESC(virtual_display, "Enable virtual display feature (the virtual_d
 module_param_named(virtual_display, amdgpu_virtual_display, charp, 0444);
 
 static const struct pci_device_id pciidlist[] = {
+#ifdef  CONFIG_DRM_AMDGPU_SI
+	{0x1002, 0x6780, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6784, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6788, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x678A, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6790, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6791, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6792, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6798, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6799, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x679A, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x679B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x679E, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x679F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
+	{0x1002, 0x6800, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN|AMD_IS_MOBILITY},
+	{0x1002, 0x6801, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN|AMD_IS_MOBILITY},
+	{0x1002, 0x6802, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN|AMD_IS_MOBILITY},
+	{0x1002, 0x6806, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6808, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6809, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6810, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6811, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6816, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6817, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6818, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6819, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_PITCAIRN},
+	{0x1002, 0x6600, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6601, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6602, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6603, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6604, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6605, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6606, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6607, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6608, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND},
+	{0x1002, 0x6610, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND},
+	{0x1002, 0x6611, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND},
+	{0x1002, 0x6613, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND},
+	{0x1002, 0x6617, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6620, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6621, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6623, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND|AMD_IS_MOBILITY},
+	{0x1002, 0x6631, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_OLAND},
+	{0x1002, 0x6820, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6821, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6822, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6823, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6824, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6825, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6826, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6827, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6828, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x6829, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x682A, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x682B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x682C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x682D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x682F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6830, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6831, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE|AMD_IS_MOBILITY},
+	{0x1002, 0x6835, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x6837, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x6838, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x6839, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x683B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x683D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x683F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VERDE},
+	{0x1002, 0x6660, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
+	{0x1002, 0x6663, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
+	{0x1002, 0x6664, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
+	{0x1002, 0x6665, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
+	{0x1002, 0x6667, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
+	{0x1002, 0x666F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_HAINAN|AMD_IS_MOBILITY},
+#endif
 #ifdef CONFIG_DRM_AMDGPU_CIK
 	/* Kaveri */
 	{0x1002, 0x1304, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_KAVERI|AMD_IS_MOBILITY|AMD_IS_APU},
@@ -393,32 +476,72 @@ amdgpu_pci_remove(struct pci_dev *pdev)
 	drm_put_dev(dev);
 }
 
+static void
+amdgpu_pci_shutdown(struct pci_dev *pdev)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = dev->dev_private;
+
+	/* if we are running in a VM, make sure the device
+	 * torn down properly on reboot/shutdown
+	 */
+	if (adev->virtualization.is_virtual)
+		amdgpu_pci_remove(pdev);
+}
+
 static int amdgpu_pmops_suspend(struct device *dev)
 {
 	struct pci_dev *pdev = to_pci_dev(dev);
+
 	struct drm_device *drm_dev = pci_get_drvdata(pdev);
-	return amdgpu_suspend_kms(drm_dev, true, true);
+	return amdgpu_device_suspend(drm_dev, true, true);
 }
 
 static int amdgpu_pmops_resume(struct device *dev)
 {
 	struct pci_dev *pdev = to_pci_dev(dev);
 	struct drm_device *drm_dev = pci_get_drvdata(pdev);
-	return amdgpu_resume_kms(drm_dev, true, true);
+
+	/* GPU comes up enabled by the bios on resume */
+	if (amdgpu_device_is_px(drm_dev)) {
+		pm_runtime_disable(dev);
+		pm_runtime_set_active(dev);
+		pm_runtime_enable(dev);
+	}
+
+	return amdgpu_device_resume(drm_dev, true, true);
 }
 
 static int amdgpu_pmops_freeze(struct device *dev)
 {
 	struct pci_dev *pdev = to_pci_dev(dev);
+
 	struct drm_device *drm_dev = pci_get_drvdata(pdev);
-	return amdgpu_suspend_kms(drm_dev, false, true);
+	return amdgpu_device_suspend(drm_dev, false, true);
 }
 
 static int amdgpu_pmops_thaw(struct device *dev)
 {
 	struct pci_dev *pdev = to_pci_dev(dev);
+
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	return amdgpu_device_resume(drm_dev, false, true);
+}
+
+static int amdgpu_pmops_poweroff(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+
+	struct drm_device *drm_dev = pci_get_drvdata(pdev);
+	return amdgpu_device_suspend(drm_dev, true, true);
+}
+
+static int amdgpu_pmops_restore(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+
 	struct drm_device *drm_dev = pci_get_drvdata(pdev);
-	return amdgpu_resume_kms(drm_dev, false, true);
+	return amdgpu_device_resume(drm_dev, false, true);
 }
 
 static int amdgpu_pmops_runtime_suspend(struct device *dev)
@@ -436,7 +559,7 @@ static int amdgpu_pmops_runtime_suspend(struct device *dev)
 	drm_kms_helper_poll_disable(drm_dev);
 	vga_switcheroo_set_dynamic_switch(pdev, VGA_SWITCHEROO_OFF);
 
-	ret = amdgpu_suspend_kms(drm_dev, false, false);
+	ret = amdgpu_device_suspend(drm_dev, false, false);
 	pci_save_state(pdev);
 	pci_disable_device(pdev);
 	pci_ignore_hotplug(pdev);
@@ -469,7 +592,7 @@ static int amdgpu_pmops_runtime_resume(struct device *dev)
 		return ret;
 	pci_set_master(pdev);
 
-	ret = amdgpu_resume_kms(drm_dev, false, false);
+	ret = amdgpu_device_resume(drm_dev, false, false);
 	drm_kms_helper_poll_enable(drm_dev);
 	vga_switcheroo_set_dynamic_switch(pdev, VGA_SWITCHEROO_ON);
 	drm_dev->switch_power_state = DRM_SWITCH_POWER_ON;
@@ -523,8 +646,8 @@ static const struct dev_pm_ops amdgpu_pm_ops = {
 	.resume = amdgpu_pmops_resume,
 	.freeze = amdgpu_pmops_freeze,
 	.thaw = amdgpu_pmops_thaw,
-	.poweroff = amdgpu_pmops_freeze,
-	.restore = amdgpu_pmops_resume,
+	.poweroff = amdgpu_pmops_poweroff,
+	.restore = amdgpu_pmops_restore,
 	.runtime_suspend = amdgpu_pmops_runtime_suspend,
 	.runtime_resume = amdgpu_pmops_runtime_resume,
 	.runtime_idle = amdgpu_pmops_runtime_idle,
@@ -606,6 +729,7 @@ static struct pci_driver amdgpu_kms_pci_driver = {
 	.id_table = pciidlist,
 	.probe = amdgpu_pci_probe,
 	.remove = amdgpu_pci_remove,
+	.shutdown = amdgpu_pci_shutdown,
 	.driver.pm = &amdgpu_pm_ops,
 };
 

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

@@ -25,6 +25,7 @@
  */
 #include <linux/module.h>
 #include <linux/slab.h>
+#include <linux/pm_runtime.h>
 
 #include <drm/drmP.h>
 #include <drm/drm_crtc.h>
@@ -47,8 +48,35 @@ struct amdgpu_fbdev {
 	struct amdgpu_device *adev;
 };
 
+static int
+amdgpufb_open(struct fb_info *info, int user)
+{
+	struct amdgpu_fbdev *rfbdev = info->par;
+	struct amdgpu_device *adev = rfbdev->adev;
+	int ret = pm_runtime_get_sync(adev->ddev->dev);
+	if (ret < 0 && ret != -EACCES) {
+		pm_runtime_mark_last_busy(adev->ddev->dev);
+		pm_runtime_put_autosuspend(adev->ddev->dev);
+		return ret;
+	}
+	return 0;
+}
+
+static int
+amdgpufb_release(struct fb_info *info, int user)
+{
+	struct amdgpu_fbdev *rfbdev = info->par;
+	struct amdgpu_device *adev = rfbdev->adev;
+
+	pm_runtime_mark_last_busy(adev->ddev->dev);
+	pm_runtime_put_autosuspend(adev->ddev->dev);
+	return 0;
+}
+
 static struct fb_ops amdgpufb_ops = {
 	.owner = THIS_MODULE,
+	.fb_open = amdgpufb_open,
+	.fb_release = amdgpufb_release,
 	.fb_check_var = drm_fb_helper_check_var,
 	.fb_set_par = drm_fb_helper_set_par,
 	.fb_fillrect = drm_fb_helper_cfb_fillrect,

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

@@ -124,7 +124,8 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 	bool skip_preamble, need_ctx_switch;
 	unsigned patch_offset = ~0;
 	struct amdgpu_vm *vm;
-	uint64_t ctx;
+	uint64_t fence_ctx;
+	uint32_t status = 0, alloc_size;
 
 	unsigned i;
 	int r = 0;
@@ -135,10 +136,10 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 	/* ring tests don't use a job */
 	if (job) {
 		vm = job->vm;
-		ctx = job->ctx;
+		fence_ctx = job->fence_ctx;
 	} else {
 		vm = NULL;
-		ctx = 0;
+		fence_ctx = 0;
 	}
 
 	if (!ring->ready) {
@@ -151,7 +152,10 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 		return -EINVAL;
 	}
 
-	r = amdgpu_ring_alloc(ring, 256 * num_ibs);
+	alloc_size = amdgpu_ring_get_dma_frame_size(ring) +
+		num_ibs * amdgpu_ring_get_emit_ib_size(ring);
+
+	r = amdgpu_ring_alloc(ring, alloc_size);
 	if (r) {
 		dev_err(adev->dev, "scheduling IB failed (%d).\n", r);
 		return r;
@@ -174,13 +178,22 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 	/* always set cond_exec_polling to CONTINUE */
 	*ring->cond_exe_cpu_addr = 1;
 
-	skip_preamble = ring->current_ctx == ctx;
-	need_ctx_switch = ring->current_ctx != ctx;
+	skip_preamble = ring->current_ctx == fence_ctx;
+	need_ctx_switch = ring->current_ctx != fence_ctx;
+	if (job && ring->funcs->emit_cntxcntl) {
+		if (need_ctx_switch)
+			status |= AMDGPU_HAVE_CTX_SWITCH;
+		status |= job->preamble_status;
+		amdgpu_ring_emit_cntxcntl(ring, status);
+	}
+
 	for (i = 0; i < num_ibs; ++i) {
 		ib = &ibs[i];
 
 		/* drop preamble IBs if we don't have a context switch */
-		if ((ib->flags & AMDGPU_IB_FLAG_PREAMBLE) && skip_preamble)
+		if ((ib->flags & AMDGPU_IB_FLAG_PREAMBLE) &&
+			skip_preamble &&
+			!(status & AMDGPU_PREAMBLE_IB_PRESENT_FIRST))
 			continue;
 
 		amdgpu_ring_emit_ib(ring, ib, job ? job->vm_id : 0,
@@ -209,7 +222,9 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 	if (patch_offset != ~0 && ring->funcs->patch_cond_exec)
 		amdgpu_ring_patch_cond_exec(ring, patch_offset);
 
-	ring->current_ctx = ctx;
+	ring->current_ctx = fence_ctx;
+	if (ring->funcs->emit_switch_buffer)
+		amdgpu_ring_emit_switch_buffer(ring);
 	amdgpu_ring_commit(ring);
 	return 0;
 }

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

@@ -119,8 +119,6 @@ int amdgpu_ih_ring_init(struct amdgpu_device *adev, unsigned ring_size,
  */
 void amdgpu_ih_ring_fini(struct amdgpu_device *adev)
 {
-	int r;
-
 	if (adev->irq.ih.use_bus_addr) {
 		if (adev->irq.ih.ring) {
 			/* add 8 bytes for the rptr/wptr shadows and
@@ -132,17 +130,9 @@ void amdgpu_ih_ring_fini(struct amdgpu_device *adev)
 			adev->irq.ih.ring = NULL;
 		}
 	} else {
-		if (adev->irq.ih.ring_obj) {
-			r = amdgpu_bo_reserve(adev->irq.ih.ring_obj, false);
-			if (likely(r == 0)) {
-				amdgpu_bo_kunmap(adev->irq.ih.ring_obj);
-				amdgpu_bo_unpin(adev->irq.ih.ring_obj);
-				amdgpu_bo_unreserve(adev->irq.ih.ring_obj);
-			}
-			amdgpu_bo_unref(&adev->irq.ih.ring_obj);
-			adev->irq.ih.ring = NULL;
-			adev->irq.ih.ring_obj = NULL;
-		}
+		amdgpu_bo_free_kernel(&adev->irq.ih.ring_obj,
+				      &adev->irq.ih.gpu_addr,
+				      (void **)&adev->irq.ih.ring);
 		amdgpu_wb_free(adev, adev->irq.ih.wptr_offs);
 		amdgpu_wb_free(adev, adev->irq.ih.rptr_offs);
 	}

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

@@ -91,7 +91,7 @@ void amdgpu_job_free_resources(struct amdgpu_job *job)
 		amdgpu_ib_free(job->adev, &job->ibs[i], f);
 }
 
-void amdgpu_job_free_cb(struct amd_sched_job *s_job)
+static void amdgpu_job_free_cb(struct amd_sched_job *s_job)
 {
 	struct amdgpu_job *job = container_of(s_job, struct amdgpu_job, base);
 
@@ -124,7 +124,7 @@ int amdgpu_job_submit(struct amdgpu_job *job, struct amdgpu_ring *ring,
 		return r;
 
 	job->owner = owner;
-	job->ctx = entity->fence_context;
+	job->fence_ctx = entity->fence_context;
 	*f = fence_get(&job->base.s_fence->finished);
 	amdgpu_job_free_resources(job);
 	amd_sched_entity_push_job(&job->base);

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

@@ -296,7 +296,7 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 			break;
 		case AMDGPU_HW_IP_VCE:
 			type = AMD_IP_BLOCK_TYPE_VCE;
-			for (i = 0; i < AMDGPU_MAX_VCE_RINGS; i++)
+			for (i = 0; i < adev->vce.num_rings; i++)
 				ring_mask |= ((adev->vce.ring[i].ready ? 1 : 0) << i);
 			ib_start_alignment = AMDGPU_GPU_PAGE_SIZE;
 			ib_size_alignment = 1;
@@ -542,12 +542,16 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
 		return r;
 
 	fpriv = kzalloc(sizeof(*fpriv), GFP_KERNEL);
-	if (unlikely(!fpriv))
-		return -ENOMEM;
+	if (unlikely(!fpriv)) {
+		r = -ENOMEM;
+		goto out_suspend;
+	}
 
 	r = amdgpu_vm_init(adev, &fpriv->vm);
-	if (r)
-		goto error_free;
+	if (r) {
+		kfree(fpriv);
+		goto out_suspend;
+	}
 
 	mutex_init(&fpriv->bo_list_lock);
 	idr_init(&fpriv->bo_list_handles);
@@ -556,12 +560,9 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
 
 	file_priv->driver_priv = fpriv;
 
+out_suspend:
 	pm_runtime_mark_last_busy(dev->dev);
 	pm_runtime_put_autosuspend(dev->dev);
-	return 0;
-
-error_free:
-	kfree(fpriv);
 
 	return r;
 }
@@ -600,6 +601,9 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev,
 
 	kfree(fpriv);
 	file_priv->driver_priv = NULL;
+
+	pm_runtime_mark_last_busy(dev->dev);
+	pm_runtime_put_autosuspend(dev->dev);
 }
 
 /**
@@ -614,6 +618,7 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev,
 void amdgpu_driver_preclose_kms(struct drm_device *dev,
 				struct drm_file *file_priv)
 {
+	pm_runtime_get_sync(dev->dev);
 }
 
 /*

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

@@ -38,8 +38,6 @@
 #include "amdgpu_trace.h"
 
 
-int amdgpu_ttm_init(struct amdgpu_device *adev);
-void amdgpu_ttm_fini(struct amdgpu_device *adev);
 
 static u64 amdgpu_get_vis_part_size(struct amdgpu_device *adev,
 						struct ttm_mem_reg *mem)
@@ -287,6 +285,35 @@ int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
 	return r;
 }
 
+/**
+ * amdgpu_bo_free_kernel - free BO for kernel use
+ *
+ * @bo: amdgpu BO to free
+ *
+ * unmaps and unpin a BO for kernel internal use.
+ */
+void amdgpu_bo_free_kernel(struct amdgpu_bo **bo, u64 *gpu_addr,
+			   void **cpu_addr)
+{
+	if (*bo == NULL)
+		return;
+
+	if (likely(amdgpu_bo_reserve(*bo, false) == 0)) {
+		if (cpu_addr)
+			amdgpu_bo_kunmap(*bo);
+
+		amdgpu_bo_unpin(*bo);
+		amdgpu_bo_unreserve(*bo);
+	}
+	amdgpu_bo_unref(bo);
+
+	if (gpu_addr)
+		*gpu_addr = 0;
+
+	if (cpu_addr)
+		*cpu_addr = NULL;
+}
+
 int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
 				unsigned long size, int byte_align,
 				bool kernel, u32 domain, u64 flags,
@@ -646,6 +673,11 @@ int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
 		dev_err(bo->adev->dev, "%p pin failed\n", bo);
 		goto error;
 	}
+	r = amdgpu_ttm_bind(bo->tbo.ttm, &bo->tbo.mem);
+	if (unlikely(r)) {
+		dev_err(bo->adev->dev, "%p bind failed\n", bo);
+		goto error;
+	}
 
 	bo->pin_count = 1;
 	if (gpu_addr != NULL)
@@ -692,7 +724,7 @@ int amdgpu_bo_unpin(struct amdgpu_bo *bo)
 		bo->adev->vram_pin_size -= amdgpu_bo_size(bo);
 		if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)
 			bo->adev->invisible_pin_size -= amdgpu_bo_size(bo);
-	} else {
+	} else if (bo->tbo.mem.mem_type == TTM_PL_TT) {
 		bo->adev->gart_pin_size -= amdgpu_bo_size(bo);
 	}
 
@@ -918,8 +950,11 @@ void amdgpu_bo_fence(struct amdgpu_bo *bo, struct fence *fence,
 u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo)
 {
 	WARN_ON_ONCE(bo->tbo.mem.mem_type == TTM_PL_SYSTEM);
+	WARN_ON_ONCE(bo->tbo.mem.mem_type == TTM_PL_TT &&
+		     !amdgpu_ttm_is_bound(bo->tbo.ttm));
 	WARN_ON_ONCE(!ww_mutex_is_locked(&bo->tbo.resv->lock) &&
 		     !bo->pin_count);
+	WARN_ON_ONCE(bo->tbo.mem.start == AMDGPU_BO_INVALID_OFFSET);
 
 	return bo->tbo.offset;
 }

drivers/gpu/drm/amd/amdgpu/amdgpu_object.h

@@ -31,6 +31,8 @@
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
 
+#define AMDGPU_BO_INVALID_OFFSET	LONG_MAX
+
 /**
  * amdgpu_mem_type_to_domain - return domain corresponding to mem_type
  * @mem_type:	ttm memory type
@@ -128,6 +130,8 @@ int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
 			    unsigned long size, int align,
 			    u32 domain, struct amdgpu_bo **bo_ptr,
 			    u64 *gpu_addr, void **cpu_addr);
+void amdgpu_bo_free_kernel(struct amdgpu_bo **bo, u64 *gpu_addr,
+			   void **cpu_addr);
 int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr);
 void amdgpu_bo_kunmap(struct amdgpu_bo *bo);
 struct amdgpu_bo *amdgpu_bo_ref(struct amdgpu_bo *bo);

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

@@ -25,6 +25,7 @@
 #include "amdgpu.h"
 #include "atom.h"
 #include "atombios_encoders.h"
+#include "amdgpu_pll.h"
 #include <asm/div64.h>
 #include <linux/gcd.h>
 

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

@@ -30,6 +30,7 @@
 #include "amdgpu_pm.h"
 #include <drm/amdgpu_drm.h>
 #include "amdgpu_powerplay.h"
+#include "si_dpm.h"
 #include "cik_dpm.h"
 #include "vi_dpm.h"
 
@@ -52,10 +53,6 @@ static int amdgpu_powerplay_init(struct amdgpu_device *adev)
 		pp_init->chip_family = adev->family;
 		pp_init->chip_id = adev->asic_type;
 		pp_init->device = amdgpu_cgs_create_device(adev);
-		pp_init->rev_id = adev->pdev->revision;
-		pp_init->sub_sys_id = adev->pdev->subsystem_device;
-		pp_init->sub_vendor_id = adev->pdev->subsystem_vendor;
-
 		ret = amd_powerplay_init(pp_init, amd_pp);
 		kfree(pp_init);
 #endif
@@ -63,6 +60,15 @@ static int amdgpu_powerplay_init(struct amdgpu_device *adev)
 		amd_pp->pp_handle = (void *)adev;
 
 		switch (adev->asic_type) {
+#ifdef CONFIG_DRM_AMDGPU_SI
+		case CHIP_TAHITI:
+		case CHIP_PITCAIRN:
+		case CHIP_VERDE:
+		case CHIP_OLAND:
+		case CHIP_HAINAN:
+			amd_pp->ip_funcs = &si_dpm_ip_funcs;
+		break;
+#endif
 #ifdef CONFIG_DRM_AMDGPU_CIK
 		case CHIP_BONAIRE:
 		case CHIP_HAWAII:

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

@@ -252,28 +252,17 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
  */
 void amdgpu_ring_fini(struct amdgpu_ring *ring)
 {
-	int r;
-	struct amdgpu_bo *ring_obj;
-
-	ring_obj = ring->ring_obj;
 	ring->ready = false;
-	ring->ring = NULL;
-	ring->ring_obj = NULL;
 
 	amdgpu_wb_free(ring->adev, ring->cond_exe_offs);
 	amdgpu_wb_free(ring->adev, ring->fence_offs);
 	amdgpu_wb_free(ring->adev, ring->rptr_offs);
 	amdgpu_wb_free(ring->adev, ring->wptr_offs);
 
-	if (ring_obj) {
-		r = amdgpu_bo_reserve(ring_obj, false);
-		if (likely(r == 0)) {
-			amdgpu_bo_kunmap(ring_obj);
-			amdgpu_bo_unpin(ring_obj);
-			amdgpu_bo_unreserve(ring_obj);
-		}
-		amdgpu_bo_unref(&ring_obj);
-	}
+	amdgpu_bo_free_kernel(&ring->ring_obj,
+			      &ring->gpu_addr,
+			      (void **)&ring->ring);
+
 	amdgpu_debugfs_ring_fini(ring);
 }
 

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

@@ -89,10 +89,10 @@ int amdgpu_ttm_global_init(struct amdgpu_device *adev)
 	global_ref->init = &amdgpu_ttm_mem_global_init;
 	global_ref->release = &amdgpu_ttm_mem_global_release;
 	r = drm_global_item_ref(global_ref);
-	if (r != 0) {
+	if (r) {
 		DRM_ERROR("Failed setting up TTM memory accounting "
 			  "subsystem.\n");
-		return r;
+		goto error_mem;
 	}
 
 	adev->mman.bo_global_ref.mem_glob =
@@ -103,26 +103,30 @@ int amdgpu_ttm_global_init(struct amdgpu_device *adev)
 	global_ref->init = &ttm_bo_global_init;
 	global_ref->release = &ttm_bo_global_release;
 	r = drm_global_item_ref(global_ref);
-	if (r != 0) {
+	if (r) {
 		DRM_ERROR("Failed setting up TTM BO subsystem.\n");
-		drm_global_item_unref(&adev->mman.mem_global_ref);
-		return r;
+		goto error_bo;
 	}
 
 	ring = adev->mman.buffer_funcs_ring;
 	rq = &ring->sched.sched_rq[AMD_SCHED_PRIORITY_KERNEL];
 	r = amd_sched_entity_init(&ring->sched, &adev->mman.entity,
 				  rq, amdgpu_sched_jobs);
-	if (r != 0) {
+	if (r) {
 		DRM_ERROR("Failed setting up TTM BO move run queue.\n");
-		drm_global_item_unref(&adev->mman.mem_global_ref);
-		drm_global_item_unref(&adev->mman.bo_global_ref.ref);
-		return r;
+		goto error_entity;
 	}
 
 	adev->mman.mem_global_referenced = true;
 
 	return 0;
+
+error_entity:
+	drm_global_item_unref(&adev->mman.bo_global_ref.ref);
+error_bo:
+	drm_global_item_unref(&adev->mman.mem_global_ref);
+error_mem:
+	return r;
 }
 
 static void amdgpu_ttm_global_fini(struct amdgpu_device *adev)
@@ -197,6 +201,7 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo,
 		.lpfn = 0,
 		.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM
 	};
+	unsigned i;
 
 	if (!amdgpu_ttm_bo_is_amdgpu_bo(bo)) {
 		placement->placement = &placements;
@@ -208,10 +213,25 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo,
 	rbo = container_of(bo, struct amdgpu_bo, tbo);
 	switch (bo->mem.mem_type) {
 	case TTM_PL_VRAM:
-		if (rbo->adev->mman.buffer_funcs_ring->ready == false)
+		if (rbo->adev->mman.buffer_funcs_ring->ready == false) {
 			amdgpu_ttm_placement_from_domain(rbo, AMDGPU_GEM_DOMAIN_CPU);
-		else
+		} else {
 			amdgpu_ttm_placement_from_domain(rbo, AMDGPU_GEM_DOMAIN_GTT);
+			for (i = 0; i < rbo->placement.num_placement; ++i) {
+				if (!(rbo->placements[i].flags &
+				      TTM_PL_FLAG_TT))
+					continue;
+
+				if (rbo->placements[i].lpfn)
+					continue;
+
+				/* set an upper limit to force directly
+				 * allocating address space for the BO.
+				 */
+				rbo->placements[i].lpfn =
+					rbo->adev->mc.gtt_size >> PAGE_SHIFT;
+			}
+		}
 		break;
 	case TTM_PL_TT:
 	default:
@@ -256,8 +276,12 @@ static int amdgpu_move_blit(struct ttm_buffer_object *bo,
 	new_start = new_mem->start << PAGE_SHIFT;
 
 	switch (old_mem->mem_type) {
-	case TTM_PL_VRAM:
 	case TTM_PL_TT:
+		r = amdgpu_ttm_bind(bo->ttm, old_mem);
+		if (r)
+			return r;
+
+	case TTM_PL_VRAM:
 		old_start += bo->bdev->man[old_mem->mem_type].gpu_offset;
 		break;
 	default:
@@ -265,8 +289,12 @@ static int amdgpu_move_blit(struct ttm_buffer_object *bo,
 		return -EINVAL;
 	}
 	switch (new_mem->mem_type) {
-	case TTM_PL_VRAM:
 	case TTM_PL_TT:
+		r = amdgpu_ttm_bind(bo->ttm, new_mem);
+		if (r)
+			return r;
+
+	case TTM_PL_VRAM:
 		new_start += bo->bdev->man[new_mem->mem_type].gpu_offset;
 		break;
 	default:
@@ -311,7 +339,7 @@ static int amdgpu_move_vram_ram(struct ttm_buffer_object *bo,
 	placement.num_busy_placement = 1;
 	placement.busy_placement = &placements;
 	placements.fpfn = 0;
-	placements.lpfn = 0;
+	placements.lpfn = adev->mc.gtt_size >> PAGE_SHIFT;
 	placements.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT;
 	r = ttm_bo_mem_space(bo, &placement, &tmp_mem,
 			     interruptible, no_wait_gpu);
@@ -358,7 +386,7 @@ static int amdgpu_move_ram_vram(struct ttm_buffer_object *bo,
 	placement.num_busy_placement = 1;
 	placement.busy_placement = &placements;
 	placements.fpfn = 0;
-	placements.lpfn = 0;
+	placements.lpfn = adev->mc.gtt_size >> PAGE_SHIFT;
 	placements.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT;
 	r = ttm_bo_mem_space(bo, &placement, &tmp_mem,
 			     interruptible, no_wait_gpu);
@@ -520,6 +548,7 @@ struct amdgpu_ttm_tt {
 	spinlock_t              guptasklock;
 	struct list_head        guptasks;
 	atomic_t		mmu_invalidations;
+	struct list_head        list;
 };
 
 int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages)
@@ -637,7 +666,6 @@ static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm,
 				   struct ttm_mem_reg *bo_mem)
 {
 	struct amdgpu_ttm_tt *gtt = (void*)ttm;
-	uint32_t flags = amdgpu_ttm_tt_pte_flags(gtt->adev, ttm, bo_mem);
 	int r;
 
 	if (gtt->userptr) {
@@ -647,7 +675,7 @@ static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm,
 			return r;
 		}
 	}
-	gtt->offset = (unsigned long)(bo_mem->start << PAGE_SHIFT);
+	gtt->offset = (u64)bo_mem->start << PAGE_SHIFT;
 	if (!ttm->num_pages) {
 		WARN(1, "nothing to bind %lu pages for mreg %p back %p!\n",
 		     ttm->num_pages, bo_mem, ttm);
@@ -658,14 +686,62 @@ static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm,
 	    bo_mem->mem_type == AMDGPU_PL_OA)
 		return -EINVAL;
 
+	return 0;
+}
+
+bool amdgpu_ttm_is_bound(struct ttm_tt *ttm)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+
+	return gtt && !list_empty(&gtt->list);
+}
+
+int amdgpu_ttm_bind(struct ttm_tt *ttm, struct ttm_mem_reg *bo_mem)
+{
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	uint32_t flags;
+	int r;
+
+	if (!ttm || amdgpu_ttm_is_bound(ttm))
+		return 0;
+
+	flags = amdgpu_ttm_tt_pte_flags(gtt->adev, ttm, bo_mem);
 	r = amdgpu_gart_bind(gtt->adev, gtt->offset, ttm->num_pages,
 		ttm->pages, gtt->ttm.dma_address, flags);
 
 	if (r) {
-		DRM_ERROR("failed to bind %lu pages at 0x%08X\n",
-			  ttm->num_pages, (unsigned)gtt->offset);
+		DRM_ERROR("failed to bind %lu pages at 0x%08llX\n",
+			  ttm->num_pages, gtt->offset);
 		return r;
 	}
+	spin_lock(&gtt->adev->gtt_list_lock);
+	list_add_tail(&gtt->list, &gtt->adev->gtt_list);
+	spin_unlock(&gtt->adev->gtt_list_lock);
+	return 0;
+}
+
+int amdgpu_ttm_recover_gart(struct amdgpu_device *adev)
+{
+	struct amdgpu_ttm_tt *gtt, *tmp;
+	struct ttm_mem_reg bo_mem;
+	uint32_t flags;
+	int r;
+
+	bo_mem.mem_type = TTM_PL_TT;
+	spin_lock(&adev->gtt_list_lock);
+	list_for_each_entry_safe(gtt, tmp, &adev->gtt_list, list) {
+		flags = amdgpu_ttm_tt_pte_flags(gtt->adev, &gtt->ttm.ttm, &bo_mem);
+		r = amdgpu_gart_bind(adev, gtt->offset, gtt->ttm.ttm.num_pages,
+				     gtt->ttm.ttm.pages, gtt->ttm.dma_address,
+				     flags);
+		if (r) {
+			spin_unlock(&adev->gtt_list_lock);
+			DRM_ERROR("failed to bind %lu pages at 0x%08llX\n",
+				  gtt->ttm.ttm.num_pages, gtt->offset);
+			return r;
+		}
+	}
+	spin_unlock(&adev->gtt_list_lock);
 	return 0;
 }
 
@@ -673,6 +749,9 @@ static int amdgpu_ttm_backend_unbind(struct ttm_tt *ttm)
 {
 	struct amdgpu_ttm_tt *gtt = (void *)ttm;
 
+	if (!amdgpu_ttm_is_bound(ttm))
+		return 0;
+
 	/* unbind shouldn't be done for GDS/GWS/OA in ttm_bo_clean_mm */
 	if (gtt->adev->gart.ready)
 		amdgpu_gart_unbind(gtt->adev, gtt->offset, ttm->num_pages);
@@ -680,6 +759,10 @@ static int amdgpu_ttm_backend_unbind(struct ttm_tt *ttm)
 	if (gtt->userptr)
 		amdgpu_ttm_tt_unpin_userptr(ttm);
 
+	spin_lock(&gtt->adev->gtt_list_lock);
+	list_del_init(&gtt->list);
+	spin_unlock(&gtt->adev->gtt_list_lock);
+
 	return 0;
 }
 
@@ -716,6 +799,7 @@ static struct ttm_tt *amdgpu_ttm_tt_create(struct ttm_bo_device *bdev,
 		kfree(gtt);
 		return NULL;
 	}
+	INIT_LIST_HEAD(&gtt->list);
 	return &gtt->ttm.ttm;
 }
 

drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h

@@ -26,13 +26,13 @@
 
 #include "gpu_scheduler.h"
 
-#define AMDGPU_PL_GDS		TTM_PL_PRIV0
-#define AMDGPU_PL_GWS		TTM_PL_PRIV1
-#define AMDGPU_PL_OA		TTM_PL_PRIV2
+#define AMDGPU_PL_GDS		(TTM_PL_PRIV + 0)
+#define AMDGPU_PL_GWS		(TTM_PL_PRIV + 1)
+#define AMDGPU_PL_OA		(TTM_PL_PRIV + 2)
 
-#define AMDGPU_PL_FLAG_GDS		TTM_PL_FLAG_PRIV0
-#define AMDGPU_PL_FLAG_GWS		TTM_PL_FLAG_PRIV1
-#define AMDGPU_PL_FLAG_OA		TTM_PL_FLAG_PRIV2
+#define AMDGPU_PL_FLAG_GDS		(TTM_PL_FLAG_PRIV << 0)
+#define AMDGPU_PL_FLAG_GWS		(TTM_PL_FLAG_PRIV << 1)
+#define AMDGPU_PL_FLAG_OA		(TTM_PL_FLAG_PRIV << 2)
 
 #define AMDGPU_TTM_LRU_SIZE	20
 
@@ -77,4 +77,7 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo,
 			struct fence **fence);
 
 int amdgpu_mmap(struct file *filp, struct vm_area_struct *vma);
+bool amdgpu_ttm_is_bound(struct ttm_tt *ttm);
+int amdgpu_ttm_bind(struct ttm_tt *ttm, struct ttm_mem_reg *bo_mem);
+
 #endif

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

@@ -247,35 +247,28 @@ int amdgpu_ucode_init_bo(struct amdgpu_device *adev)
 	const struct common_firmware_header *header = NULL;
 
 	err = amdgpu_bo_create(adev, adev->firmware.fw_size, PAGE_SIZE, true,
-			AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL, bo);
+			       AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL, bo);
 	if (err) {
 		dev_err(adev->dev, "(%d) Firmware buffer allocate failed\n", err);
-		err = -ENOMEM;
 		goto failed;
 	}
 
 	err = amdgpu_bo_reserve(*bo, false);
 	if (err) {
-		amdgpu_bo_unref(bo);
 		dev_err(adev->dev, "(%d) Firmware buffer reserve failed\n", err);
-		goto failed;
+		goto failed_reserve;
 	}
 
 	err = amdgpu_bo_pin(*bo, AMDGPU_GEM_DOMAIN_GTT, &fw_mc_addr);
 	if (err) {
-		amdgpu_bo_unreserve(*bo);
-		amdgpu_bo_unref(bo);
 		dev_err(adev->dev, "(%d) Firmware buffer pin failed\n", err);
-		goto failed;
+		goto failed_pin;
 	}
 
 	err = amdgpu_bo_kmap(*bo, &fw_buf_ptr);
 	if (err) {
 		dev_err(adev->dev, "(%d) Firmware buffer kmap failed\n", err);
-		amdgpu_bo_unpin(*bo);
-		amdgpu_bo_unreserve(*bo);
-		amdgpu_bo_unref(bo);
-		goto failed;
+		goto failed_kmap;
 	}
 
 	amdgpu_bo_unreserve(*bo);
@@ -290,10 +283,16 @@ int amdgpu_ucode_init_bo(struct amdgpu_device *adev)
 			fw_offset += ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
 		}
 	}
+	return 0;
 
+failed_kmap:
+	amdgpu_bo_unpin(*bo);
+failed_pin:
+	amdgpu_bo_unreserve(*bo);
+failed_reserve:
+	amdgpu_bo_unref(bo);
 failed:
-	if (err)
-		adev->firmware.smu_load = false;
+	adev->firmware.smu_load = false;
 
 	return err;
 }

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

@@ -249,22 +249,13 @@ int amdgpu_uvd_sw_init(struct amdgpu_device *adev)
 
 int amdgpu_uvd_sw_fini(struct amdgpu_device *adev)
 {
-	int r;
-
 	kfree(adev->uvd.saved_bo);
 
 	amd_sched_entity_fini(&adev->uvd.ring.sched, &adev->uvd.entity);
 
-	if (adev->uvd.vcpu_bo) {
-		r = amdgpu_bo_reserve(adev->uvd.vcpu_bo, false);
-		if (!r) {
-			amdgpu_bo_kunmap(adev->uvd.vcpu_bo);
-			amdgpu_bo_unpin(adev->uvd.vcpu_bo);
-			amdgpu_bo_unreserve(adev->uvd.vcpu_bo);
-		}
-
-		amdgpu_bo_unref(&adev->uvd.vcpu_bo);
-	}
+	amdgpu_bo_free_kernel(&adev->uvd.vcpu_bo,
+			      &adev->uvd.gpu_addr,
+			      (void **)&adev->uvd.cpu_addr);
 
 	amdgpu_ring_fini(&adev->uvd.ring);
 
@@ -891,6 +882,10 @@ int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx)
 		return -EINVAL;
 	}
 
+	r = amdgpu_cs_sysvm_access_required(parser);
+	if (r)
+		return r;
+
 	ctx.parser = parser;
 	ctx.buf_sizes = buf_sizes;
 	ctx.ib_idx = ib_idx;

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

@@ -634,7 +634,11 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
 	uint32_t allocated = 0;
 	uint32_t tmp, handle = 0;
 	uint32_t *size = &tmp;
-	int i, r = 0, idx = 0;
+	int i, r, idx = 0;
+
+	r = amdgpu_cs_sysvm_access_required(p);
+	if (r)
+		return r;
 
 	while (idx < ib->length_dw) {
 		uint32_t len = amdgpu_get_ib_value(p, ib_idx, idx);
@@ -799,6 +803,18 @@ void amdgpu_vce_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
 	amdgpu_ring_write(ring, VCE_CMD_END);
 }
 
+unsigned amdgpu_vce_ring_get_emit_ib_size(struct amdgpu_ring *ring)
+{
+	return
+		4; /* amdgpu_vce_ring_emit_ib */
+}
+
+unsigned amdgpu_vce_ring_get_dma_frame_size(struct amdgpu_ring *ring)
+{
+	return
+		6; /* amdgpu_vce_ring_emit_fence  x1 no user fence */
+}
+
 /**
  * amdgpu_vce_ring_test_ring - test if VCE ring is working
  *
@@ -850,8 +866,8 @@ int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 	struct fence *fence = NULL;
 	long r;
 
-	/* skip vce ring1 ib test for now, since it's not reliable */
-	if (ring == &ring->adev->vce.ring[1])
+	/* skip vce ring1/2 ib test for now, since it's not reliable */
+	if (ring != &ring->adev->vce.ring[0])
 		return 0;
 
 	r = amdgpu_vce_get_create_msg(ring, 1, NULL);

drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h

@@ -42,5 +42,7 @@ int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring);
 int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring, long timeout);
 void amdgpu_vce_ring_begin_use(struct amdgpu_ring *ring);
 void amdgpu_vce_ring_end_use(struct amdgpu_ring *ring);
+unsigned amdgpu_vce_ring_get_emit_ib_size(struct amdgpu_ring *ring);
+unsigned amdgpu_vce_ring_get_dma_frame_size(struct amdgpu_ring *ring);
 
 #endif

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

@@ -1163,7 +1163,8 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 	}
 
 	flags = amdgpu_ttm_tt_pte_flags(adev, bo_va->bo->tbo.ttm, mem);
-	gtt_flags = (adev == bo_va->bo->adev) ? flags : 0;
+	gtt_flags = (amdgpu_ttm_is_bound(bo_va->bo->tbo.ttm) &&
+		adev == bo_va->bo->adev) ? flags : 0;
 
 	spin_lock(&vm->status_lock);
 	if (!list_empty(&bo_va->vm_status))

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

@@ -497,7 +497,13 @@ void amdgpu_atombios_crtc_set_disp_eng_pll(struct amdgpu_device *adev,
 			 * SetPixelClock provides the dividers
 			 */
 			args.v6.ulDispEngClkFreq = cpu_to_le32(dispclk);
-			args.v6.ucPpll = ATOM_EXT_PLL1;
+			if (adev->asic_type == CHIP_TAHITI ||
+			    adev->asic_type == CHIP_PITCAIRN ||
+			    adev->asic_type == CHIP_VERDE ||
+			    adev->asic_type == CHIP_OLAND)
+				args.v6.ucPpll = ATOM_PPLL0;
+			else
+				args.v6.ucPpll = ATOM_EXT_PLL1;
 			break;
 		default:
 			DRM_ERROR("Unknown table version %d %d\n", frev, crev);

drivers/gpu/drm/amd/amdgpu/atombios_i2c.c

@@ -27,6 +27,7 @@
 #include "amdgpu.h"
 #include "atom.h"
 #include "amdgpu_atombios.h"
+#include "atombios_i2c.h"
 
 #define TARGET_HW_I2C_CLOCK 50
 

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

@@ -5396,7 +5396,7 @@ static void ci_dpm_disable(struct amdgpu_device *adev)
 	amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
 		       AMDGPU_THERMAL_IRQ_HIGH_TO_LOW);
 
-	ci_dpm_powergate_uvd(adev, false);
+	ci_dpm_powergate_uvd(adev, true);
 
 	if (!amdgpu_ci_is_smc_running(adev))
 		return;
@@ -6036,7 +6036,7 @@ static int ci_dpm_init(struct amdgpu_device *adev)
 
 	pi->caps_dynamic_ac_timing = true;
 
-	pi->uvd_power_gated = false;
+	pi->uvd_power_gated = true;
 
 	/* make sure dc limits are valid */
 	if ((adev->pm.dpm.dyn_state.max_clock_voltage_on_dc.sclk == 0) ||
@@ -6179,8 +6179,6 @@ static int ci_dpm_late_init(void *handle)
 	if (ret)
 		return ret;
 
-	ci_dpm_powergate_uvd(adev, true);
-
 	return 0;
 }
 

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

@@ -847,6 +847,22 @@ static void cik_sdma_ring_emit_vm_flush(struct amdgpu_ring *ring,
 	amdgpu_ring_write(ring, (0xfff << 16) | 10); /* retry count, poll interval */
 }
 
+static unsigned cik_sdma_ring_get_emit_ib_size(struct amdgpu_ring *ring)
+{
+	return
+		7 + 4; /* cik_sdma_ring_emit_ib */
+}
+
+static unsigned cik_sdma_ring_get_dma_frame_size(struct amdgpu_ring *ring)
+{
+	return
+		6 + /* cik_sdma_ring_emit_hdp_flush */
+		3 + /* cik_sdma_ring_emit_hdp_invalidate */
+		6 + /* cik_sdma_ring_emit_pipeline_sync */
+		12 + /* cik_sdma_ring_emit_vm_flush */
+		9 + 9 + 9; /* cik_sdma_ring_emit_fence x3 for user fence, vm fence */
+}
+
 static void cik_enable_sdma_mgcg(struct amdgpu_device *adev,
 				 bool enable)
 {
@@ -1220,6 +1236,8 @@ static const struct amdgpu_ring_funcs cik_sdma_ring_funcs = {
 	.test_ib = cik_sdma_ring_test_ib,
 	.insert_nop = cik_sdma_ring_insert_nop,
 	.pad_ib = cik_sdma_ring_pad_ib,
+	.get_emit_ib_size = cik_sdma_ring_get_emit_ib_size,
+	.get_dma_frame_size = cik_sdma_ring_get_dma_frame_size,
 };
 
 static void cik_sdma_set_ring_funcs(struct amdgpu_device *adev)

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

@@ -44,6 +44,7 @@
 
 static void cz_dpm_powergate_uvd(struct amdgpu_device *adev, bool gate);
 static void cz_dpm_powergate_vce(struct amdgpu_device *adev, bool gate);
+static void cz_dpm_fini(struct amdgpu_device *adev);
 
 static struct cz_ps *cz_get_ps(struct amdgpu_ps *rps)
 {
@@ -350,6 +351,8 @@ static int cz_parse_power_table(struct amdgpu_device *adev)
 
 		ps = kzalloc(sizeof(struct cz_ps), GFP_KERNEL);
 		if (ps == NULL) {
+			for (j = 0; j < i; j++)
+				kfree(adev->pm.dpm.ps[j].ps_priv);
 			kfree(adev->pm.dpm.ps);
 			return -ENOMEM;
 		}
@@ -409,11 +412,11 @@ static int cz_dpm_init(struct amdgpu_device *adev)
 
 	ret = amdgpu_get_platform_caps(adev);
 	if (ret)
-		return ret;
+		goto err;
 
 	ret = amdgpu_parse_extended_power_table(adev);
 	if (ret)
-		return ret;
+		goto err;
 
 	pi->sram_end = SMC_RAM_END;
 
@@ -467,23 +470,26 @@ static int cz_dpm_init(struct amdgpu_device *adev)
 
 	ret = cz_parse_sys_info_table(adev);
 	if (ret)
-		return ret;
+		goto err;
 
 	cz_patch_voltage_values(adev);
 	cz_construct_boot_state(adev);
 
 	ret = cz_parse_power_table(adev);
 	if (ret)
-		return ret;
+		goto err;
 
 	ret = cz_process_firmware_header(adev);
 	if (ret)
-		return ret;
+		goto err;
 
 	pi->dpm_enabled = true;
 	pi->uvd_dynamic_pg = false;
 
 	return 0;
+err:
+	cz_dpm_fini(adev);
+	return ret;
 }
 
 static void cz_dpm_fini(struct amdgpu_device *adev)
@@ -672,17 +678,12 @@ static void cz_reset_ap_mask(struct amdgpu_device *adev)
 	struct cz_power_info *pi = cz_get_pi(adev);
 
 	pi->active_process_mask = 0;
-
 }
 
 static int cz_dpm_download_pptable_from_smu(struct amdgpu_device *adev,
 							void **table)
 {
-	int ret = 0;
-
-	ret = cz_smu_download_pptable(adev, table);
-
-	return ret;
+	return cz_smu_download_pptable(adev, table);
 }
 
 static int cz_dpm_upload_pptable_to_smu(struct amdgpu_device *adev)
@@ -822,9 +823,9 @@ static void cz_init_sclk_limit(struct amdgpu_device *adev)
 	pi->sclk_dpm.hard_min_clk = 0;
 	cz_send_msg_to_smc(adev, PPSMC_MSG_GetMaxSclkLevel);
 	level = cz_get_argument(adev);
-	if (level < table->count)
+	if (level < table->count) {
 		clock = table->entries[level].clk;
-	else {
+	} else {
 		DRM_ERROR("Invalid SLCK Voltage Dependency table entry.\n");
 		clock = table->entries[table->count - 1].clk;
 	}
@@ -850,9 +851,9 @@ static void cz_init_uvd_limit(struct amdgpu_device *adev)
 	pi->uvd_dpm.hard_min_clk = 0;
 	cz_send_msg_to_smc(adev, PPSMC_MSG_GetMaxUvdLevel);
 	level = cz_get_argument(adev);
-	if (level < table->count)
+	if (level < table->count) {
 		clock = table->entries[level].vclk;
-	else {
+	} else {
 		DRM_ERROR("Invalid UVD Voltage Dependency table entry.\n");
 		clock = table->entries[table->count - 1].vclk;
 	}
@@ -878,9 +879,9 @@ static void cz_init_vce_limit(struct amdgpu_device *adev)
 	pi->vce_dpm.hard_min_clk = table->entries[0].ecclk;
 	cz_send_msg_to_smc(adev, PPSMC_MSG_GetMaxEclkLevel);
 	level = cz_get_argument(adev);
-	if (level < table->count)
+	if (level < table->count) {
 		clock = table->entries[level].ecclk;
-	else {
+	} else {
 		/* future BIOS would fix this error */
 		DRM_ERROR("Invalid VCE Voltage Dependency table entry.\n");
 		clock = table->entries[table->count - 1].ecclk;
@@ -907,9 +908,9 @@ static void cz_init_acp_limit(struct amdgpu_device *adev)
 	pi->acp_dpm.hard_min_clk = 0;
 	cz_send_msg_to_smc(adev, PPSMC_MSG_GetMaxAclkLevel);
 	level = cz_get_argument(adev);
-	if (level < table->count)
+	if (level < table->count) {
 		clock = table->entries[level].clk;
-	else {
+	} else {
 		DRM_ERROR("Invalid ACP Voltage Dependency table entry.\n");
 		clock = table->entries[table->count - 1].clk;
 	}
@@ -934,7 +935,6 @@ static void cz_init_sclk_threshold(struct amdgpu_device *adev)
 	struct cz_power_info *pi = cz_get_pi(adev);
 
 	pi->low_sclk_interrupt_threshold = 0;
-
 }
 
 static void cz_dpm_setup_asic(struct amdgpu_device *adev)
@@ -1207,7 +1207,7 @@ static int cz_enable_didt(struct amdgpu_device *adev, bool enable)
 	int ret;
 
 	if (pi->caps_sq_ramping || pi->caps_db_ramping ||
-			pi->caps_td_ramping || pi->caps_tcp_ramping) {
+	    pi->caps_td_ramping || pi->caps_tcp_ramping) {
 		if (adev->gfx.gfx_current_status != AMDGPU_GFX_SAFE_MODE) {
 			ret = cz_disable_cgpg(adev);
 			if (ret) {
@@ -1281,7 +1281,7 @@ static void cz_apply_state_adjust_rules(struct amdgpu_device *adev,
 	ps->force_high = false;
 	ps->need_dfs_bypass = true;
 	pi->video_start = new_rps->dclk || new_rps->vclk ||
-				new_rps->evclk || new_rps->ecclk;
+			  new_rps->evclk || new_rps->ecclk;
 
 	if ((new_rps->class & ATOM_PPLIB_CLASSIFICATION_UI_MASK) ==
 			ATOM_PPLIB_CLASSIFICATION_UI_BATTERY)
@@ -1339,7 +1339,6 @@ static int cz_dpm_enable(struct amdgpu_device *adev)
 	}
 
 	cz_reset_acp_boot_level(adev);
-
 	cz_update_current_ps(adev, adev->pm.dpm.boot_ps);
 
 	return 0;
@@ -1669,7 +1668,6 @@ static void cz_dpm_post_set_power_state(struct amdgpu_device *adev)
 	struct amdgpu_ps *ps = &pi->requested_rps;
 
 	cz_update_current_ps(adev, ps);
-
 }
 
 static int cz_dpm_force_highest(struct amdgpu_device *adev)
@@ -2201,7 +2199,6 @@ static int cz_update_vce_dpm(struct amdgpu_device *adev)
 	/* Stable Pstate is enabled and we need to set the VCE DPM to highest level */
 	if (pi->caps_stable_power_state) {
 		pi->vce_dpm.hard_min_clk = table->entries[table->count-1].ecclk;
-
 	} else { /* non-stable p-state cases. without vce.Arbiter.EcclkHardMin */
 		/* leave it as set by user */
 		/*pi->vce_dpm.hard_min_clk = table->entries[0].ecclk;*/

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

@@ -29,6 +29,8 @@
 #include "cz_smumgr.h"
 #include "smu_ucode_xfer_cz.h"
 #include "amdgpu_ucode.h"
+#include "cz_dpm.h"
+#include "vi_dpm.h"
 
 #include "smu/smu_8_0_d.h"
 #include "smu/smu_8_0_sh_mask.h"
@@ -48,7 +50,7 @@ static struct cz_smu_private_data *cz_smu_get_priv(struct amdgpu_device *adev)
 	return priv;
 }
 
-int cz_send_msg_to_smc_async(struct amdgpu_device *adev, u16 msg)
+static int cz_send_msg_to_smc_async(struct amdgpu_device *adev, u16 msg)
 {
 	int i;
 	u32 content = 0, tmp;
@@ -140,7 +142,7 @@ int cz_read_smc_sram_dword(struct amdgpu_device *adev, u32 smc_address,
 	return 0;
 }
 
-int cz_write_smc_sram_dword(struct amdgpu_device *adev, u32 smc_address,
+static int cz_write_smc_sram_dword(struct amdgpu_device *adev, u32 smc_address,
 						u32 value, u32 limit)
 {
 	int ret;

drivers/gpu/drm/amd/amdgpu/dce_v6_0.h

+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __DCE_V6_0_H__
+#define __DCE_V6_0_H__
+
+extern const struct amd_ip_funcs dce_v6_0_ip_funcs;
+
+#endif

drivers/gpu/drm/amd/amdgpu/gfx_v6_0.h

+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GFX_V6_0_H__
+#define __GFX_V6_0_H__
+
+extern const struct amd_ip_funcs gfx_v6_0_ip_funcs;
+
+#endif

drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c

@@ -2096,6 +2096,25 @@ static void gfx_v7_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
 	amdgpu_ring_write(ring, control);
 }
 
+static void gfx_v7_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags)
+{
+	uint32_t dw2 = 0;
+
+	dw2 |= 0x80000000; /* set load_enable otherwise this package is just NOPs */
+	if (flags & AMDGPU_HAVE_CTX_SWITCH) {
+		/* set load_global_config & load_global_uconfig */
+		dw2 |= 0x8001;
+		/* set load_cs_sh_regs */
+		dw2 |= 0x01000000;
+		/* set load_per_context_state & load_gfx_sh_regs */
+		dw2 |= 0x10002;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, dw2);
+	amdgpu_ring_write(ring, 0);
+}
+
 /**
  * gfx_v7_0_ring_test_ib - basic ring IB test
  *
@@ -2443,7 +2462,7 @@ static int gfx_v7_0_cp_gfx_resume(struct amdgpu_device *adev)
 	return 0;
 }
 
-static u32 gfx_v7_0_ring_get_rptr_gfx(struct amdgpu_ring *ring)
+static u32 gfx_v7_0_ring_get_rptr(struct amdgpu_ring *ring)
 {
 	return ring->adev->wb.wb[ring->rptr_offs];
 }
@@ -2463,11 +2482,6 @@ static void gfx_v7_0_ring_set_wptr_gfx(struct amdgpu_ring *ring)
 	(void)RREG32(mmCP_RB0_WPTR);
 }
 
-static u32 gfx_v7_0_ring_get_rptr_compute(struct amdgpu_ring *ring)
-{
-	return ring->adev->wb.wb[ring->rptr_offs];
-}
-
 static u32 gfx_v7_0_ring_get_wptr_compute(struct amdgpu_ring *ring)
 {
 	/* XXX check if swapping is necessary on BE */
@@ -4176,6 +4190,41 @@ static void gfx_v7_0_ring_emit_gds_switch(struct amdgpu_ring *ring,
 	amdgpu_ring_write(ring, (1 << (oa_size + oa_base)) - (1 << oa_base));
 }
 
+static unsigned gfx_v7_0_ring_get_emit_ib_size_gfx(struct amdgpu_ring *ring)
+{
+	return
+		4; /* gfx_v7_0_ring_emit_ib_gfx */
+}
+
+static unsigned gfx_v7_0_ring_get_dma_frame_size_gfx(struct amdgpu_ring *ring)
+{
+	return
+		20 + /* gfx_v7_0_ring_emit_gds_switch */
+		7 + /* gfx_v7_0_ring_emit_hdp_flush */
+		5 + /* gfx_v7_0_ring_emit_hdp_invalidate */
+		12 + 12 + 12 + /* gfx_v7_0_ring_emit_fence_gfx x3 for user fence, vm fence */
+		7 + 4 + /* gfx_v7_0_ring_emit_pipeline_sync */
+		17 + 6 + /* gfx_v7_0_ring_emit_vm_flush */
+		3; /* gfx_v7_ring_emit_cntxcntl */
+}
+
+static unsigned gfx_v7_0_ring_get_emit_ib_size_compute(struct amdgpu_ring *ring)
+{
+	return
+		4; /* gfx_v7_0_ring_emit_ib_compute */
+}
+
+static unsigned gfx_v7_0_ring_get_dma_frame_size_compute(struct amdgpu_ring *ring)
+{
+	return
+		20 + /* gfx_v7_0_ring_emit_gds_switch */
+		7 + /* gfx_v7_0_ring_emit_hdp_flush */
+		5 + /* gfx_v7_0_ring_emit_hdp_invalidate */
+		7 + /* gfx_v7_0_ring_emit_pipeline_sync */
+		17 + /* gfx_v7_0_ring_emit_vm_flush */
+		7 + 7 + 7; /* gfx_v7_0_ring_emit_fence_compute x3 for user fence, vm fence */
+}
+
 static const struct amdgpu_gfx_funcs gfx_v7_0_gfx_funcs = {
 	.get_gpu_clock_counter = &gfx_v7_0_get_gpu_clock_counter,
 	.select_se_sh = &gfx_v7_0_select_se_sh,
@@ -4495,9 +4544,9 @@ static int gfx_v7_0_sw_fini(void *handle)
 	int i;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	amdgpu_bo_unref(&adev->gds.oa_gfx_bo);
-	amdgpu_bo_unref(&adev->gds.gws_gfx_bo);
-	amdgpu_bo_unref(&adev->gds.gds_gfx_bo);
+	amdgpu_bo_free_kernel(&adev->gds.oa_gfx_bo, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gds.gws_gfx_bo, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gds.gds_gfx_bo, NULL, NULL);
 
 	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
 		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
@@ -4928,7 +4977,7 @@ const struct amd_ip_funcs gfx_v7_0_ip_funcs = {
 };
 
 static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_gfx = {
-	.get_rptr = gfx_v7_0_ring_get_rptr_gfx,
+	.get_rptr = gfx_v7_0_ring_get_rptr,
 	.get_wptr = gfx_v7_0_ring_get_wptr_gfx,
 	.set_wptr = gfx_v7_0_ring_set_wptr_gfx,
 	.parse_cs = NULL,
@@ -4943,10 +4992,13 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_gfx = {
 	.test_ib = gfx_v7_0_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_cntxcntl = gfx_v7_ring_emit_cntxcntl,
+	.get_emit_ib_size = gfx_v7_0_ring_get_emit_ib_size_gfx,
+	.get_dma_frame_size = gfx_v7_0_ring_get_dma_frame_size_gfx,
 };
 
 static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_compute = {
-	.get_rptr = gfx_v7_0_ring_get_rptr_compute,
+	.get_rptr = gfx_v7_0_ring_get_rptr,
 	.get_wptr = gfx_v7_0_ring_get_wptr_compute,
 	.set_wptr = gfx_v7_0_ring_set_wptr_compute,
 	.parse_cs = NULL,
@@ -4961,6 +5013,8 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_compute = {
 	.test_ib = gfx_v7_0_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.get_emit_ib_size = gfx_v7_0_ring_get_emit_ib_size_compute,
+	.get_dma_frame_size = gfx_v7_0_ring_get_dma_frame_size_compute,
 };
 
 static void gfx_v7_0_set_ring_funcs(struct amdgpu_device *adev)

drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c

@@ -2113,9 +2113,9 @@ static int gfx_v8_0_sw_fini(void *handle)
 	int i;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	amdgpu_bo_unref(&adev->gds.oa_gfx_bo);
-	amdgpu_bo_unref(&adev->gds.gws_gfx_bo);
-	amdgpu_bo_unref(&adev->gds.gds_gfx_bo);
+	amdgpu_bo_free_kernel(&adev->gds.oa_gfx_bo, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gds.gws_gfx_bo, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gds.gds_gfx_bo, NULL, NULL);
 
 	for (i = 0; i < adev->gfx.num_gfx_rings; i++)
 		amdgpu_ring_fini(&adev->gfx.gfx_ring[i]);
@@ -3866,7 +3866,7 @@ static void gfx_v8_0_init_pg(struct amdgpu_device *adev)
 	}
 }
 
-void gfx_v8_0_rlc_stop(struct amdgpu_device *adev)
+static void gfx_v8_0_rlc_stop(struct amdgpu_device *adev)
 {
 	WREG32_FIELD(RLC_CNTL, RLC_ENABLE_F32, 0);
 
@@ -5835,7 +5835,7 @@ static int gfx_v8_0_set_clockgating_state(void *handle,
 	return 0;
 }
 
-static u32 gfx_v8_0_ring_get_rptr_gfx(struct amdgpu_ring *ring)
+static u32 gfx_v8_0_ring_get_rptr(struct amdgpu_ring *ring)
 {
 	return ring->adev->wb.wb[ring->rptr_offs];
 }
@@ -5915,12 +5915,6 @@ static void gfx_v8_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
 {
 	u32 header, control = 0;
 
-	/* insert SWITCH_BUFFER packet before first IB in the ring frame */
-	if (ctx_switch) {
-		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
-		amdgpu_ring_write(ring, 0);
-	}
-
 	if (ib->flags & AMDGPU_IB_FLAG_CE)
 		header = PACKET3(PACKET3_INDIRECT_BUFFER_CONST, 2);
 	else
@@ -5990,14 +5984,6 @@ static void gfx_v8_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
 	amdgpu_ring_write(ring, seq);
 	amdgpu_ring_write(ring, 0xffffffff);
 	amdgpu_ring_write(ring, 4); /* poll interval */
-
-	if (usepfp) {
-		/* synce CE with ME to prevent CE fetch CEIB before context switch done */
-		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
-		amdgpu_ring_write(ring, 0);
-		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
-		amdgpu_ring_write(ring, 0);
-	}
 }
 
 static void gfx_v8_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
@@ -6005,6 +5991,10 @@ static void gfx_v8_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
 {
 	int usepfp = (ring->type == AMDGPU_RING_TYPE_GFX);
 
+	/* GFX8 emits 128 dw nop to prevent DE do vm_flush before CE finish CEIB */
+	if (usepfp)
+		amdgpu_ring_insert_nop(ring, 128);
+
 	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
 	amdgpu_ring_write(ring, (WRITE_DATA_ENGINE_SEL(usepfp) |
 				 WRITE_DATA_DST_SEL(0)) |
@@ -6044,18 +6034,11 @@ static void gfx_v8_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
 		/* sync PFP to ME, otherwise we might get invalid PFP reads */
 		amdgpu_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0));
 		amdgpu_ring_write(ring, 0x0);
-		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
-		amdgpu_ring_write(ring, 0);
-		amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
-		amdgpu_ring_write(ring, 0);
+		/* GFX8 emits 128 dw nop to prevent CE access VM before vm_flush finish */
+		amdgpu_ring_insert_nop(ring, 128);
 	}
 }
 
-static u32 gfx_v8_0_ring_get_rptr_compute(struct amdgpu_ring *ring)
-{
-	return ring->adev->wb.wb[ring->rptr_offs];
-}
-
 static u32 gfx_v8_0_ring_get_wptr_compute(struct amdgpu_ring *ring)
 {
 	return ring->adev->wb.wb[ring->wptr_offs];
@@ -6091,6 +6074,77 @@ static void gfx_v8_0_ring_emit_fence_compute(struct amdgpu_ring *ring,
 	amdgpu_ring_write(ring, upper_32_bits(seq));
 }
 
+static void gfx_v8_ring_emit_sb(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
+	amdgpu_ring_write(ring, 0);
+}
+
+static void gfx_v8_ring_emit_cntxcntl(struct amdgpu_ring *ring, uint32_t flags)
+{
+	uint32_t dw2 = 0;
+
+	dw2 |= 0x80000000; /* set load_enable otherwise this package is just NOPs */
+	if (flags & AMDGPU_HAVE_CTX_SWITCH) {
+		/* set load_global_config & load_global_uconfig */
+		dw2 |= 0x8001;
+		/* set load_cs_sh_regs */
+		dw2 |= 0x01000000;
+		/* set load_per_context_state & load_gfx_sh_regs for GFX */
+		dw2 |= 0x10002;
+
+		/* set load_ce_ram if preamble presented */
+		if (AMDGPU_PREAMBLE_IB_PRESENT & flags)
+			dw2 |= 0x10000000;
+	} else {
+		/* still load_ce_ram if this is the first time preamble presented
+		 * although there is no context switch happens.
+		 */
+		if (AMDGPU_PREAMBLE_IB_PRESENT_FIRST & flags)
+			dw2 |= 0x10000000;
+	}
+
+	amdgpu_ring_write(ring, PACKET3(PACKET3_CONTEXT_CONTROL, 1));
+	amdgpu_ring_write(ring, dw2);
+	amdgpu_ring_write(ring, 0);
+}
+
+static unsigned gfx_v8_0_ring_get_emit_ib_size_gfx(struct amdgpu_ring *ring)
+{
+	return
+		4; /* gfx_v8_0_ring_emit_ib_gfx */
+}
+
+static unsigned gfx_v8_0_ring_get_dma_frame_size_gfx(struct amdgpu_ring *ring)
+{
+	return
+		20 + /* gfx_v8_0_ring_emit_gds_switch */
+		7 + /* gfx_v8_0_ring_emit_hdp_flush */
+		5 + /* gfx_v8_0_ring_emit_hdp_invalidate */
+		6 + 6 + 6 +/* gfx_v8_0_ring_emit_fence_gfx x3 for user fence, vm fence */
+		7 + /* gfx_v8_0_ring_emit_pipeline_sync */
+		256 + 19 + /* gfx_v8_0_ring_emit_vm_flush */
+		2 + /* gfx_v8_ring_emit_sb */
+		3; /* gfx_v8_ring_emit_cntxcntl */
+}
+
+static unsigned gfx_v8_0_ring_get_emit_ib_size_compute(struct amdgpu_ring *ring)
+{
+	return
+		4; /* gfx_v8_0_ring_emit_ib_compute */
+}
+
+static unsigned gfx_v8_0_ring_get_dma_frame_size_compute(struct amdgpu_ring *ring)
+{
+	return
+		20 + /* gfx_v8_0_ring_emit_gds_switch */
+		7 + /* gfx_v8_0_ring_emit_hdp_flush */
+		5 + /* gfx_v8_0_ring_emit_hdp_invalidate */
+		7 + /* gfx_v8_0_ring_emit_pipeline_sync */
+		17 + /* gfx_v8_0_ring_emit_vm_flush */
+		7 + 7 + 7; /* gfx_v8_0_ring_emit_fence_compute x3 for user fence, vm fence */
+}
+
 static void gfx_v8_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev,
 						 enum amdgpu_interrupt_state state)
 {
@@ -6257,7 +6311,7 @@ const struct amd_ip_funcs gfx_v8_0_ip_funcs = {
 };
 
 static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_gfx = {
-	.get_rptr = gfx_v8_0_ring_get_rptr_gfx,
+	.get_rptr = gfx_v8_0_ring_get_rptr,
 	.get_wptr = gfx_v8_0_ring_get_wptr_gfx,
 	.set_wptr = gfx_v8_0_ring_set_wptr_gfx,
 	.parse_cs = NULL,
@@ -6272,10 +6326,14 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_gfx = {
 	.test_ib = gfx_v8_0_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.emit_switch_buffer = gfx_v8_ring_emit_sb,
+	.emit_cntxcntl = gfx_v8_ring_emit_cntxcntl,
+	.get_emit_ib_size = gfx_v8_0_ring_get_emit_ib_size_gfx,
+	.get_dma_frame_size = gfx_v8_0_ring_get_dma_frame_size_gfx,
 };
 
 static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_compute = {
-	.get_rptr = gfx_v8_0_ring_get_rptr_compute,
+	.get_rptr = gfx_v8_0_ring_get_rptr,
 	.get_wptr = gfx_v8_0_ring_get_wptr_compute,
 	.set_wptr = gfx_v8_0_ring_set_wptr_compute,
 	.parse_cs = NULL,
@@ -6290,6 +6348,8 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_compute = {
 	.test_ib = gfx_v8_0_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.get_emit_ib_size = gfx_v8_0_ring_get_emit_ib_size_compute,
+	.get_dma_frame_size = gfx_v8_0_ring_get_dma_frame_size_compute,
 };
 
 static void gfx_v8_0_set_ring_funcs(struct amdgpu_device *adev)

drivers/gpu/drm/amd/amdgpu/gmc_v6_0.h

+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __GMC_V6_0_H__
+#define __GMC_V6_0_H__
+
+extern const struct amd_ip_funcs gmc_v6_0_ip_funcs;
+
+#endif

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

@@ -121,7 +121,7 @@ static int iceland_copy_bytes_to_smc(struct amdgpu_device *adev,
 	return result;
 }
 
-void iceland_start_smc(struct amdgpu_device *adev)
+static void iceland_start_smc(struct amdgpu_device *adev)
 {
 	uint32_t val = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
 
@@ -129,7 +129,7 @@ void iceland_start_smc(struct amdgpu_device *adev)
 	WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, val);
 }
 
-void iceland_reset_smc(struct amdgpu_device *adev)
+static void iceland_reset_smc(struct amdgpu_device *adev)
 {
 	uint32_t val = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
 
@@ -145,7 +145,7 @@ static int iceland_program_jump_on_start(struct amdgpu_device *adev)
 	return 0;
 }
 
-void iceland_stop_smc_clock(struct amdgpu_device *adev)
+static void iceland_stop_smc_clock(struct amdgpu_device *adev)
 {
 	uint32_t val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
 
@@ -153,7 +153,7 @@ void iceland_stop_smc_clock(struct amdgpu_device *adev)
 	WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, val);
 }
 
-void iceland_start_smc_clock(struct amdgpu_device *adev)
+static void iceland_start_smc_clock(struct amdgpu_device *adev)
 {
 	uint32_t val = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
 

drivers/gpu/drm/amd/amdgpu/r600_dpm.h

+/*
+ * Copyright 2011 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __R600_DPM_H__
+#define __R600_DPM_H__
+
+#define R600_ASI_DFLT                                10000
+#define R600_BSP_DFLT                                0x41EB
+#define R600_BSU_DFLT                                0x2
+#define R600_AH_DFLT                                 5
+#define R600_RLP_DFLT                                25
+#define R600_RMP_DFLT                                65
+#define R600_LHP_DFLT                                40
+#define R600_LMP_DFLT                                15
+#define R600_TD_DFLT                                 0
+#define R600_UTC_DFLT_00                             0x24
+#define R600_UTC_DFLT_01                             0x22
+#define R600_UTC_DFLT_02                             0x22
+#define R600_UTC_DFLT_03                             0x22
+#define R600_UTC_DFLT_04                             0x22
+#define R600_UTC_DFLT_05                             0x22
+#define R600_UTC_DFLT_06                             0x22
+#define R600_UTC_DFLT_07                             0x22
+#define R600_UTC_DFLT_08                             0x22
+#define R600_UTC_DFLT_09                             0x22
+#define R600_UTC_DFLT_10                             0x22
+#define R600_UTC_DFLT_11                             0x22
+#define R600_UTC_DFLT_12                             0x22
+#define R600_UTC_DFLT_13                             0x22
+#define R600_UTC_DFLT_14                             0x22
+#define R600_DTC_DFLT_00                             0x24
+#define R600_DTC_DFLT_01                             0x22
+#define R600_DTC_DFLT_02                             0x22
+#define R600_DTC_DFLT_03                             0x22
+#define R600_DTC_DFLT_04                             0x22
+#define R600_DTC_DFLT_05                             0x22
+#define R600_DTC_DFLT_06                             0x22
+#define R600_DTC_DFLT_07                             0x22
+#define R600_DTC_DFLT_08                             0x22
+#define R600_DTC_DFLT_09                             0x22
+#define R600_DTC_DFLT_10                             0x22
+#define R600_DTC_DFLT_11                             0x22
+#define R600_DTC_DFLT_12                             0x22
+#define R600_DTC_DFLT_13                             0x22
+#define R600_DTC_DFLT_14                             0x22
+#define R600_VRC_DFLT                                0x0000C003
+#define R600_VOLTAGERESPONSETIME_DFLT                1000
+#define R600_BACKBIASRESPONSETIME_DFLT               1000
+#define R600_VRU_DFLT                                0x3
+#define R600_SPLLSTEPTIME_DFLT                       0x1000
+#define R600_SPLLSTEPUNIT_DFLT                       0x3
+#define R600_TPU_DFLT                                0
+#define R600_TPC_DFLT                                0x200
+#define R600_SSTU_DFLT                               0
+#define R600_SST_DFLT                                0x00C8
+#define R600_GICST_DFLT                              0x200
+#define R600_FCT_DFLT                                0x0400
+#define R600_FCTU_DFLT                               0
+#define R600_CTXCGTT3DRPHC_DFLT                      0x20
+#define R600_CTXCGTT3DRSDC_DFLT                      0x40
+#define R600_VDDC3DOORPHC_DFLT                       0x100
+#define R600_VDDC3DOORSDC_DFLT                       0x7
+#define R600_VDDC3DOORSU_DFLT                        0
+#define R600_MPLLLOCKTIME_DFLT                       100
+#define R600_MPLLRESETTIME_DFLT                      150
+#define R600_VCOSTEPPCT_DFLT                          20
+#define R600_ENDINGVCOSTEPPCT_DFLT                    5
+#define R600_REFERENCEDIVIDER_DFLT                    4
+
+#define R600_PM_NUMBER_OF_TC 15
+#define R600_PM_NUMBER_OF_SCLKS 20
+#define R600_PM_NUMBER_OF_MCLKS 4
+#define R600_PM_NUMBER_OF_VOLTAGE_LEVELS 4
+#define R600_PM_NUMBER_OF_ACTIVITY_LEVELS 3
+
+/* XXX are these ok? */
+#define R600_TEMP_RANGE_MIN (90 * 1000)
+#define R600_TEMP_RANGE_MAX (120 * 1000)
+
+#define FDO_PWM_MODE_STATIC  1
+#define FDO_PWM_MODE_STATIC_RPM 5
+
+enum r600_power_level {
+	R600_POWER_LEVEL_LOW = 0,
+	R600_POWER_LEVEL_MEDIUM = 1,
+	R600_POWER_LEVEL_HIGH = 2,
+	R600_POWER_LEVEL_CTXSW = 3,
+};
+
+enum r600_td {
+	R600_TD_AUTO,
+	R600_TD_UP,
+	R600_TD_DOWN,
+};
+
+enum r600_display_watermark {
+	R600_DISPLAY_WATERMARK_LOW = 0,
+	R600_DISPLAY_WATERMARK_HIGH = 1,
+};
+
+enum r600_display_gap
+{
+    R600_PM_DISPLAY_GAP_VBLANK_OR_WM = 0,
+    R600_PM_DISPLAY_GAP_VBLANK       = 1,
+    R600_PM_DISPLAY_GAP_WATERMARK    = 2,
+    R600_PM_DISPLAY_GAP_IGNORE       = 3,
+};
+#endif

drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c

@@ -902,6 +902,22 @@ static void sdma_v2_4_ring_emit_vm_flush(struct amdgpu_ring *ring,
 			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
 }
 
+static unsigned sdma_v2_4_ring_get_emit_ib_size(struct amdgpu_ring *ring)
+{
+	return
+		7 + 6; /* sdma_v2_4_ring_emit_ib */
+}
+
+static unsigned sdma_v2_4_ring_get_dma_frame_size(struct amdgpu_ring *ring)
+{
+	return
+		6 + /* sdma_v2_4_ring_emit_hdp_flush */
+		3 + /* sdma_v2_4_ring_emit_hdp_invalidate */
+		6 + /* sdma_v2_4_ring_emit_pipeline_sync */
+		12 + /* sdma_v2_4_ring_emit_vm_flush */
+		10 + 10 + 10; /* sdma_v2_4_ring_emit_fence x3 for user fence, vm fence */
+}
+
 static int sdma_v2_4_early_init(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
@@ -1220,6 +1236,8 @@ static const struct amdgpu_ring_funcs sdma_v2_4_ring_funcs = {
 	.test_ib = sdma_v2_4_ring_test_ib,
 	.insert_nop = sdma_v2_4_ring_insert_nop,
 	.pad_ib = sdma_v2_4_ring_pad_ib,
+	.get_emit_ib_size = sdma_v2_4_ring_get_emit_ib_size,
+	.get_dma_frame_size = sdma_v2_4_ring_get_dma_frame_size,
 };
 
 static void sdma_v2_4_set_ring_funcs(struct amdgpu_device *adev)

drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c

@@ -495,31 +495,6 @@ static void sdma_v3_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 se
 	amdgpu_ring_write(ring, SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(0));
 }
 
-unsigned init_cond_exec(struct amdgpu_ring *ring)
-{
-	unsigned ret;
-	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_COND_EXE));
-	amdgpu_ring_write(ring, lower_32_bits(ring->cond_exe_gpu_addr));
-	amdgpu_ring_write(ring, upper_32_bits(ring->cond_exe_gpu_addr));
-	amdgpu_ring_write(ring, 1);
-	ret = ring->wptr;/* this is the offset we need patch later */
-	amdgpu_ring_write(ring, 0x55aa55aa);/* insert dummy here and patch it later */
-	return ret;
-}
-
-void patch_cond_exec(struct amdgpu_ring *ring, unsigned offset)
-{
-	unsigned cur;
-	BUG_ON(ring->ring[offset] != 0x55aa55aa);
-
-	cur = ring->wptr - 1;
-	if (likely(cur > offset))
-		ring->ring[offset] = cur - offset;
-	else
-		ring->ring[offset] = (ring->ring_size>>2) - offset + cur;
-}
-
-
 /**
  * sdma_v3_0_gfx_stop - stop the gfx async dma engines
  *
@@ -1129,6 +1104,22 @@ static void sdma_v3_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
 			  SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */
 }
 
+static unsigned sdma_v3_0_ring_get_emit_ib_size(struct amdgpu_ring *ring)
+{
+	return
+		7 + 6; /* sdma_v3_0_ring_emit_ib */
+}
+
+static unsigned sdma_v3_0_ring_get_dma_frame_size(struct amdgpu_ring *ring)
+{
+	return
+		6 + /* sdma_v3_0_ring_emit_hdp_flush */
+		3 + /* sdma_v3_0_ring_emit_hdp_invalidate */
+		6 + /* sdma_v3_0_ring_emit_pipeline_sync */
+		12 + /* sdma_v3_0_ring_emit_vm_flush */
+		10 + 10 + 10; /* sdma_v3_0_ring_emit_fence x3 for user fence, vm fence */
+}
+
 static int sdma_v3_0_early_init(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
@@ -1590,6 +1581,8 @@ static const struct amdgpu_ring_funcs sdma_v3_0_ring_funcs = {
 	.test_ib = sdma_v3_0_ring_test_ib,
 	.insert_nop = sdma_v3_0_ring_insert_nop,
 	.pad_ib = sdma_v3_0_ring_pad_ib,
+	.get_emit_ib_size = sdma_v3_0_ring_get_emit_ib_size,
+	.get_dma_frame_size = sdma_v3_0_ring_get_dma_frame_size,
 };
 
 static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev)

drivers/gpu/drm/amd/amdgpu/si.h

+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __SI_H__
+#define __SI_H__
+
+extern const struct amd_ip_funcs si_common_ip_funcs;
+
+void si_srbm_select(struct amdgpu_device *adev,
+		     u32 me, u32 pipe, u32 queue, u32 vmid);
+int si_set_ip_blocks(struct amdgpu_device *adev);
+
+#endif