Linux-libre 4.8.2-gnupck-4.8.2-gnu

64 files changed, 2749 insertions, 1467 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 56475b1f1..700c56baf 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -85,8 +85,12 @@ extern int amdgpu_vm_debug;
 extern int amdgpu_sched_jobs;
 extern int amdgpu_sched_hw_submission;
 extern int amdgpu_powerplay;
+extern int amdgpu_powercontainment;
 extern unsigned amdgpu_pcie_gen_cap;
 extern unsigned amdgpu_pcie_lane_cap;
+extern unsigned amdgpu_cg_mask;
+extern unsigned amdgpu_pg_mask;
+extern char *amdgpu_disable_cu;
 
 #define AMDGPU_WAIT_IDLE_TIMEOUT_IN_MS	        3000
 #define AMDGPU_MAX_USEC_TIMEOUT			100000	/* 100 ms */
@@ -183,6 +187,10 @@ int amdgpu_set_clockgating_state(struct amdgpu_device *adev,
 int amdgpu_set_powergating_state(struct amdgpu_device *adev,
 				  enum amd_ip_block_type block_type,
 				  enum amd_powergating_state state);
+int amdgpu_wait_for_idle(struct amdgpu_device *adev,
+			 enum amd_ip_block_type block_type);
+bool amdgpu_is_idle(struct amdgpu_device *adev,
+		    enum amd_ip_block_type block_type);
 
 struct amdgpu_ip_block_version {
 	enum amd_ip_block_type type;
@@ -298,13 +306,16 @@ struct amdgpu_ring_funcs {
 				uint32_t oa_base, uint32_t oa_size);
 	/* testing functions */
 	int (*test_ring)(struct amdgpu_ring *ring);
-	int (*test_ib)(struct amdgpu_ring *ring);
+	int (*test_ib)(struct amdgpu_ring *ring, long timeout);
 	/* insert NOP packets */
 	void (*insert_nop)(struct amdgpu_ring *ring, uint32_t count);
 	/* pad the indirect buffer to the necessary number of dw */
 	void (*pad_ib)(struct amdgpu_ring *ring, struct amdgpu_ib *ib);
 	unsigned (*init_cond_exec)(struct amdgpu_ring *ring);
 	void (*patch_cond_exec)(struct amdgpu_ring *ring, unsigned offset);
+	/* note usage for clock and power gating */
+	void (*begin_use)(struct amdgpu_ring *ring);
+	void (*end_use)(struct amdgpu_ring *ring);
 };
 
 /*
@@ -596,11 +607,9 @@ int amdgpu_sync_resv(struct amdgpu_device *adev,
 		     struct amdgpu_sync *sync,
 		     struct reservation_object *resv,
 		     void *owner);
-bool amdgpu_sync_is_idle(struct amdgpu_sync *sync);
-int amdgpu_sync_cycle_fences(struct amdgpu_sync *dst, struct amdgpu_sync *src,
-			     struct fence *fence);
+struct fence *amdgpu_sync_peek_fence(struct amdgpu_sync *sync,
+				     struct amdgpu_ring *ring);
 struct fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync);
-int amdgpu_sync_wait(struct amdgpu_sync *sync);
 void amdgpu_sync_free(struct amdgpu_sync *sync);
 int amdgpu_sync_init(void);
 void amdgpu_sync_fini(void);
@@ -756,12 +765,11 @@ int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs,
 int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev, unsigned size,
 			     struct amdgpu_job **job);
 
+void amdgpu_job_free_resources(struct amdgpu_job *job);
 void amdgpu_job_free(struct amdgpu_job *job);
-void amdgpu_job_free_func(struct kref *refcount);
 int amdgpu_job_submit(struct amdgpu_job *job, struct amdgpu_ring *ring,
 		      struct amd_sched_entity *entity, void *owner,
 		      struct fence **f);
-void amdgpu_job_timeout_func(struct work_struct *work);
 
 struct amdgpu_ring {
 	struct amdgpu_device		*adev;
@@ -769,12 +777,9 @@ struct amdgpu_ring {
 	struct amdgpu_fence_driver	fence_drv;
 	struct amd_gpu_scheduler	sched;
 
-	spinlock_t              fence_lock;
 	struct amdgpu_bo	*ring_obj;
 	volatile uint32_t	*ring;
 	unsigned		rptr_offs;
-	u64			next_rptr_gpu_addr;
-	volatile u32		*next_rptr_cpu_addr;
 	unsigned		wptr;
 	unsigned		wptr_old;
 	unsigned		ring_size;
@@ -793,14 +798,16 @@ struct amdgpu_ring {
 	u32			doorbell_index;
 	bool			use_doorbell;
 	unsigned		wptr_offs;
-	unsigned		next_rptr_offs;
 	unsigned		fence_offs;
 	uint64_t		current_ctx;
 	enum amdgpu_ring_type	type;
 	char			name[16];
 	unsigned		cond_exe_offs;
-	u64				cond_exe_gpu_addr;
-	volatile u32	*cond_exe_cpu_addr;
+	u64			cond_exe_gpu_addr;
+	volatile u32		*cond_exe_cpu_addr;
+#if defined(CONFIG_DEBUG_FS)
+	struct dentry *ent;
+#endif
 };
 
 /*
@@ -863,6 +870,7 @@ struct amdgpu_vm {
 	struct amdgpu_bo	*page_directory;
 	unsigned		max_pde_used;
 	struct fence		*page_directory_fence;
+	uint64_t		last_eviction_counter;
 
 	/* array of page tables, one for each page directory entry */
 	struct amdgpu_vm_pt	*page_tables;
@@ -885,13 +893,14 @@ struct amdgpu_vm_id {
 	struct fence		*first;
 	struct amdgpu_sync	active;
 	struct fence		*last_flush;
-	struct amdgpu_ring      *last_user;
 	atomic64_t		owner;
 
 	uint64_t		pd_gpu_addr;
 	/* last flushed PD/PT update */
 	struct fence		*flushed_updates;
 
+	uint32_t                current_gpu_reset_count;
+
 	uint32_t		gds_base;
 	uint32_t		gds_size;
 	uint32_t		gws_base;
@@ -907,6 +916,10 @@ struct amdgpu_vm_manager {
 	struct list_head			ids_lru;
 	struct amdgpu_vm_id			ids[AMDGPU_NUM_VM];
 
+	/* Handling of VM fences */
+	u64					fence_context;
+	unsigned				seqno[AMDGPU_MAX_RINGS];
+
 	uint32_t				max_pfn;
 	/* vram base address for page table entry  */
 	u64					vram_base_offset;
@@ -928,17 +941,14 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm);
 void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
 			 struct list_head *validated,
 			 struct amdgpu_bo_list_entry *entry);
-void amdgpu_vm_get_pt_bos(struct amdgpu_vm *vm, struct list_head *duplicates);
+void amdgpu_vm_get_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			  struct list_head *duplicates);
 void amdgpu_vm_move_pt_bos_in_lru(struct amdgpu_device *adev,
 				  struct amdgpu_vm *vm);
 int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
 		      struct amdgpu_sync *sync, struct fence *fence,
-		      unsigned *vm_id, uint64_t *vm_pd_addr);
-int amdgpu_vm_flush(struct amdgpu_ring *ring,
-		    unsigned vm_id, uint64_t pd_addr,
-		    uint32_t gds_base, uint32_t gds_size,
-		    uint32_t gws_base, uint32_t gws_size,
-		    uint32_t oa_base, uint32_t oa_size);
+		      struct amdgpu_job *job);
+int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job);
 void amdgpu_vm_reset_id(struct amdgpu_device *adev, unsigned vm_id);
 uint64_t amdgpu_vm_map_gart(const dma_addr_t *pages_addr, uint64_t addr);
 int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
@@ -1144,6 +1154,12 @@ struct amdgpu_cu_info {
 	uint32_t bitmap[4][4];
 };
 
+struct amdgpu_gfx_funcs {
+	/* get the gpu clock counter */
+	uint64_t (*get_gpu_clock_counter)(struct amdgpu_device *adev);
+	void (*select_se_sh)(struct amdgpu_device *adev, u32 se_num, u32 sh_num, u32 instance);
+};
+
 struct amdgpu_gfx {
 	struct mutex			gpu_clock_mutex;
 	struct amdgpu_gca_config	config;
@@ -1180,6 +1196,7 @@ struct amdgpu_gfx {
 	/* ce ram size*/
 	unsigned			ce_ram_size;
 	struct amdgpu_cu_info		cu_info;
+	const struct amdgpu_gfx_funcs	*funcs;
 };
 
 int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm,
@@ -1197,10 +1214,6 @@ void amdgpu_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count);
 void amdgpu_ring_generic_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib);
 void amdgpu_ring_commit(struct amdgpu_ring *ring);
 void amdgpu_ring_undo(struct amdgpu_ring *ring);
-unsigned amdgpu_ring_backup(struct amdgpu_ring *ring,
-			    uint32_t **data);
-int amdgpu_ring_restore(struct amdgpu_ring *ring,
-			unsigned size, uint32_t *data);
 int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
 		     unsigned ring_size, u32 nop, u32 align_mask,
 		     struct amdgpu_irq_src *irq_src, unsigned irq_type,
@@ -1252,6 +1265,7 @@ struct amdgpu_job {
 	uint32_t		num_ibs;
 	void			*owner;
 	uint64_t		ctx;
+	bool                    vm_needs_flush;
 	unsigned		vm_id;
 	uint64_t		vm_pd_addr;
 	uint32_t		gds_base, gds_size;
@@ -1259,8 +1273,7 @@ struct amdgpu_job {
 	uint32_t		oa_base, oa_size;
 
 	/* user fence handling */
-	struct amdgpu_bo	*uf_bo;
-	uint32_t		uf_offset;
+	uint64_t		uf_addr;
 	uint64_t		uf_sequence;
 
 };
@@ -1562,6 +1575,12 @@ struct amdgpu_dpm_funcs {
 	u32 (*get_fan_control_mode)(struct amdgpu_device *adev);
 	int (*set_fan_speed_percent)(struct amdgpu_device *adev, u32 speed);
 	int (*get_fan_speed_percent)(struct amdgpu_device *adev, u32 *speed);
+	int (*force_clock_level)(struct amdgpu_device *adev, enum pp_clock_type type, uint32_t mask);
+	int (*print_clock_levels)(struct amdgpu_device *adev, enum pp_clock_type type, char *buf);
+	int (*get_sclk_od)(struct amdgpu_device *adev);
+	int (*set_sclk_od)(struct amdgpu_device *adev, uint32_t value);
+	int (*get_mclk_od)(struct amdgpu_device *adev);
+	int (*set_mclk_od)(struct amdgpu_device *adev, uint32_t value);
 };
 
 struct amdgpu_dpm {
@@ -1664,6 +1683,7 @@ struct amdgpu_uvd {
 	struct amdgpu_ring	ring;
 	struct amdgpu_irq_src	irq;
 	bool			address_64_bit;
+	bool			use_ctx_buf;
 	struct amd_sched_entity entity;
 };
 
@@ -1685,6 +1705,7 @@ struct amdgpu_vce {
 	struct drm_file		*filp[AMDGPU_MAX_VCE_HANDLES];
 	uint32_t		img_size[AMDGPU_MAX_VCE_HANDLES];
 	struct delayed_work	idle_work;
+	struct mutex		idle_mutex;
 	const struct firmware	*fw;	/* VCE firmware */
 	struct amdgpu_ring	ring[AMDGPU_MAX_VCE_RINGS];
 	struct amdgpu_irq_src	irq;
@@ -1769,6 +1790,8 @@ int amdgpu_debugfs_init(struct drm_minor *minor);
 void amdgpu_debugfs_cleanup(struct drm_minor *minor);
 #endif
 
+int amdgpu_debugfs_firmware_init(struct amdgpu_device *adev);
+
 /*
  * amdgpu smumgr functions
  */
@@ -1813,12 +1836,8 @@ struct amdgpu_asic_funcs {
 			     u32 sh_num, u32 reg_offset, u32 *value);
 	void (*set_vga_state)(struct amdgpu_device *adev, bool state);
 	int (*reset)(struct amdgpu_device *adev);
-	/* wait for mc_idle */
-	int (*wait_for_mc_idle)(struct amdgpu_device *adev);
 	/* get the reference clock */
 	u32 (*get_xclk)(struct amdgpu_device *adev);
-	/* get the gpu clock counter */
-	uint64_t (*get_gpu_clock_counter)(struct amdgpu_device *adev);
 	/* MM block clocks */
 	int (*set_uvd_clocks)(struct amdgpu_device *adev, u32 vclk, u32 dclk);
 	int (*set_vce_clocks)(struct amdgpu_device *adev, u32 evclk, u32 ecclk);
@@ -2005,6 +2024,10 @@ struct amdgpu_device {
 	spinlock_t didt_idx_lock;
 	amdgpu_rreg_t			didt_rreg;
 	amdgpu_wreg_t			didt_wreg;
+	/* protects concurrent gc_cac register access */
+	spinlock_t gc_cac_idx_lock;
+	amdgpu_rreg_t			gc_cac_rreg;
+	amdgpu_wreg_t			gc_cac_wreg;
 	/* protects concurrent ENDPOINT (audio) register access */
 	spinlock_t audio_endpt_idx_lock;
 	amdgpu_block_rreg_t		audio_endpt_rreg;
@@ -2030,6 +2053,7 @@ struct amdgpu_device {
 	atomic64_t			vram_vis_usage;
 	atomic64_t			gtt_usage;
 	atomic64_t			num_bytes_moved;
+	atomic64_t			num_evictions;
 	atomic_t			gpu_reset_counter;
 
 	/* display */
@@ -2040,7 +2064,7 @@ struct amdgpu_device {
 	struct amdgpu_irq_src		hpd_irq;
 
 	/* rings */
-	unsigned			fence_context;
+	u64				fence_context;
 	unsigned			num_rings;
 	struct amdgpu_ring		*rings[AMDGPU_MAX_RINGS];
 	bool				ib_pool_ready;
@@ -2133,6 +2157,8 @@ void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v);
 #define WREG32_UVD_CTX(reg, v) adev->uvd_ctx_wreg(adev, (reg), (v))
 #define RREG32_DIDT(reg) adev->didt_rreg(adev, (reg))
 #define WREG32_DIDT(reg, v) adev->didt_wreg(adev, (reg), (v))
+#define RREG32_GC_CAC(reg) adev->gc_cac_rreg(adev, (reg))
+#define WREG32_GC_CAC(reg, v) adev->gc_cac_wreg(adev, (reg), (v))
 #define RREG32_AUDIO_ENDPT(block, reg) adev->audio_endpt_rreg(adev, (block), (reg))
 #define WREG32_AUDIO_ENDPT(block, reg, v) adev->audio_endpt_wreg(adev, (block), (reg), (v))
 #define WREG32_P(reg, val, mask)				\
@@ -2208,12 +2234,10 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
  */
 #define amdgpu_asic_set_vga_state(adev, state) (adev)->asic_funcs->set_vga_state((adev), (state))
 #define amdgpu_asic_reset(adev) (adev)->asic_funcs->reset((adev))
-#define amdgpu_asic_wait_for_mc_idle(adev) (adev)->asic_funcs->wait_for_mc_idle((adev))
 #define amdgpu_asic_get_xclk(adev) (adev)->asic_funcs->get_xclk((adev))
 #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_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)))
@@ -2224,7 +2248,7 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
 #define amdgpu_vm_set_pte_pde(adev, ib, pe, addr, count, incr, flags) ((adev)->vm_manager.vm_pte_funcs->set_pte_pde((ib), (pe), (addr), (count), (incr), (flags)))
 #define amdgpu_ring_parse_cs(r, p, ib) ((r)->funcs->parse_cs((p), (ib)))
 #define amdgpu_ring_test_ring(r) (r)->funcs->test_ring((r))
-#define amdgpu_ring_test_ib(r) (r)->funcs->test_ib((r))
+#define amdgpu_ring_test_ib(r, t) (r)->funcs->test_ib((r), (t))
 #define amdgpu_ring_get_rptr(r) (r)->funcs->get_rptr((r))
 #define amdgpu_ring_get_wptr(r) (r)->funcs->get_wptr((r))
 #define amdgpu_ring_set_wptr(r) (r)->funcs->set_wptr((r))
@@ -2266,6 +2290,8 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
 #define amdgpu_dpm_print_power_state(adev, ps) (adev)->pm.funcs->print_power_state((adev), (ps))
 #define amdgpu_dpm_vblank_too_short(adev) (adev)->pm.funcs->vblank_too_short((adev))
 #define amdgpu_dpm_enable_bapm(adev, e) (adev)->pm.funcs->enable_bapm((adev), (e))
+#define amdgpu_gfx_get_gpu_clock_counter(adev) (adev)->gfx.funcs->get_gpu_clock_counter((adev))
+#define amdgpu_gfx_select_se_sh(adev, se, sh, instance) (adev)->gfx.funcs->select_se_sh((adev), (se), (sh), (instance))
 
 #define amdgpu_dpm_get_temperature(adev) \
 	((adev)->pp_enabled ?						\
@@ -2344,6 +2370,18 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
 #define amdgpu_dpm_force_clock_level(adev, type, level) \
 		(adev)->powerplay.pp_funcs->force_clock_level((adev)->powerplay.pp_handle, type, level)
 
+#define amdgpu_dpm_get_sclk_od(adev) \
+	(adev)->powerplay.pp_funcs->get_sclk_od((adev)->powerplay.pp_handle)
+
+#define amdgpu_dpm_set_sclk_od(adev, value) \
+	(adev)->powerplay.pp_funcs->set_sclk_od((adev)->powerplay.pp_handle, value)
+
+#define amdgpu_dpm_get_mclk_od(adev) \
+	((adev)->powerplay.pp_funcs->get_mclk_od((adev)->powerplay.pp_handle))
+
+#define amdgpu_dpm_set_mclk_od(adev, value) \
+	((adev)->powerplay.pp_funcs->set_mclk_od((adev)->powerplay.pp_handle, value))
+
 #define amdgpu_dpm_dispatch_task(adev, event_id, input, output)		\
 	(adev)->powerplay.pp_funcs->dispatch_tasks((adev)->powerplay.pp_handle, (event_id), (input), (output))
 
@@ -2385,9 +2423,13 @@ bool amdgpu_device_is_px(struct drm_device *dev);
 #if defined(CONFIG_VGA_SWITCHEROO)
 void amdgpu_register_atpx_handler(void);
 void amdgpu_unregister_atpx_handler(void);
+bool amdgpu_has_atpx_dgpu_power_cntl(void);
+bool amdgpu_is_atpx_hybrid(void);
 #else
 static inline void amdgpu_register_atpx_handler(void) {}
 static inline void amdgpu_unregister_atpx_handler(void) {}
+static inline bool amdgpu_has_atpx_dgpu_power_cntl(void) { return false; }
+static inline bool amdgpu_is_atpx_hybrid(void) { return false; }
 #endif
 
 /*
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c
index 252edba16..892d60fb2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c
@@ -421,29 +421,6 @@ static int acp_suspend(void *handle)
 
 static int acp_resume(void *handle)
 {
-	int i, ret;
-	struct acp_pm_domain *apd;
-	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
-
-	/* return early if no ACP */
-	if (!adev->acp.acp_genpd)
-		return 0;
-
-	/* SMU block will power on ACP irrespective of ACP runtime status.
-	 * Power off explicitly based on genpd ACP runtime status so that ACP
-	 * hw and ACP-genpd status are in sync.
-	 * 'suspend_power_off' represents "Power status before system suspend"
-	*/
-	if (adev->acp.acp_genpd->gpd.suspend_power_off == true) {
-		apd = container_of(&adev->acp.acp_genpd->gpd,
-					struct acp_pm_domain, gpd);
-
-		for (i = 4; i >= 0 ; i--) {
-			ret = acp_suspend_tile(apd->cgs_dev, ACP_TILE_P1 + i);
-			if (ret)
-				pr_err("ACP tile %d tile suspend failed\n", i);
-		}
-	}
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index 32809f749..d080d0807 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -240,8 +240,8 @@ uint64_t get_gpu_clock_counter(struct kgd_dev *kgd)
 {
 	struct amdgpu_device *rdev = (struct amdgpu_device *)kgd;
 
-	if (rdev->asic_funcs->get_gpu_clock_counter)
-		return rdev->asic_funcs->get_gpu_clock_counter(rdev);
+	if (rdev->gfx.funcs->get_gpu_clock_counter)
+		return rdev->gfx.funcs->get_gpu_clock_counter(rdev);
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c
index 1b4c069f7..10b5ddf2c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c
@@ -28,6 +28,7 @@ struct amdgpu_atpx_functions {
 struct amdgpu_atpx {
 	acpi_handle handle;
 	struct amdgpu_atpx_functions functions;
+	bool is_hybrid;
 };
 
 static struct amdgpu_atpx_priv {
@@ -64,6 +65,14 @@ bool amdgpu_has_atpx(void) {
 	return amdgpu_atpx_priv.atpx_detected;
 }
 
+bool amdgpu_has_atpx_dgpu_power_cntl(void) {
+	return amdgpu_atpx_priv.atpx.functions.power_cntl;
+}
+
+bool amdgpu_is_atpx_hybrid(void) {
+	return amdgpu_atpx_priv.atpx.is_hybrid;
+}
+
 /**
  * amdgpu_atpx_call - call an ATPX method
  *
@@ -143,18 +152,12 @@ static void amdgpu_atpx_parse_functions(struct amdgpu_atpx_functions *f, u32 mas
  */
 static int amdgpu_atpx_validate(struct amdgpu_atpx *atpx)
 {
-	/* make sure required functions are enabled */
-	/* dGPU power control is required */
-	if (atpx->functions.power_cntl == false) {
-		printk("ATPX dGPU power cntl not present, forcing\n");
-		atpx->functions.power_cntl = true;
-	}
+	u32 valid_bits = 0;
 
 	if (atpx->functions.px_params) {
 		union acpi_object *info;
 		struct atpx_px_params output;
 		size_t size;
-		u32 valid_bits;
 
 		info = amdgpu_atpx_call(atpx->handle, ATPX_FUNCTION_GET_PX_PARAMETERS, NULL);
 		if (!info)
@@ -173,19 +176,34 @@ static int amdgpu_atpx_validate(struct amdgpu_atpx *atpx)
 		memcpy(&output, info->buffer.pointer, size);
 
 		valid_bits = output.flags & output.valid_flags;
-		/* if separate mux flag is set, mux controls are required */
-		if (valid_bits & ATPX_SEPARATE_MUX_FOR_I2C) {
-			atpx->functions.i2c_mux_cntl = true;
-			atpx->functions.disp_mux_cntl = true;
-		}
-		/* if any outputs are muxed, mux controls are required */
-		if (valid_bits & (ATPX_CRT1_RGB_SIGNAL_MUXED |
-				  ATPX_TV_SIGNAL_MUXED |
-				  ATPX_DFP_SIGNAL_MUXED))
-			atpx->functions.disp_mux_cntl = true;
 
 		kfree(info);
 	}
+
+	/* if separate mux flag is set, mux controls are required */
+	if (valid_bits & ATPX_SEPARATE_MUX_FOR_I2C) {
+		atpx->functions.i2c_mux_cntl = true;
+		atpx->functions.disp_mux_cntl = true;
+	}
+	/* if any outputs are muxed, mux controls are required */
+	if (valid_bits & (ATPX_CRT1_RGB_SIGNAL_MUXED |
+			  ATPX_TV_SIGNAL_MUXED |
+			  ATPX_DFP_SIGNAL_MUXED))
+		atpx->functions.disp_mux_cntl = true;
+
+
+	/* some bioses set these bits rather than flagging power_cntl as supported */
+	if (valid_bits & (ATPX_DYNAMIC_PX_SUPPORTED |
+			  ATPX_DYNAMIC_DGPU_POWER_OFF_SUPPORTED))
+		atpx->functions.power_cntl = true;
+
+	atpx->is_hybrid = false;
+	if (valid_bits & ATPX_MS_HYBRID_GFX_SUPPORTED) {
+		printk("ATPX Hybrid Graphics\n");
+		atpx->functions.power_cntl = false;
+		atpx->is_hybrid = true;
+	}
+
 	return 0;
 }
 
@@ -512,7 +530,6 @@ static int amdgpu_atpx_get_client_id(struct pci_dev *pdev)
 static const struct vga_switcheroo_handler amdgpu_atpx_handler = {
 	.switchto = amdgpu_atpx_switchto,
 	.power_state = amdgpu_atpx_power_state,
-	.init = amdgpu_atpx_init,
 	.get_client_id = amdgpu_atpx_get_client_id,
 };
 
@@ -547,6 +564,7 @@ static bool amdgpu_atpx_detect(void)
 		printk(KERN_INFO "vga_switcheroo: detected switching method %s handle\n",
 		       acpi_method_name);
 		amdgpu_atpx_priv.atpx_detected = true;
+		amdgpu_atpx_init();
 		return true;
 	}
 	return false;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
index 99ca75baa..2b6afe123 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
@@ -352,22 +352,22 @@ bool amdgpu_get_bios(struct amdgpu_device *adev)
 	uint16_t tmp, bios_header_start;
 
 	r = amdgpu_atrm_get_bios(adev);
-	if (r == false)
+	if (!r)
 		r = amdgpu_acpi_vfct_bios(adev);
-	if (r == false)
+	if (!r)
 		r = igp_read_bios_from_vram(adev);
-	if (r == false)
+	if (!r)
 		r = amdgpu_read_bios(adev);
-	if (r == false) {
+	if (!r) {
 		r = amdgpu_read_bios_from_rom(adev);
 	}
-	if (r == false) {
+	if (!r) {
 		r = amdgpu_read_disabled_bios(adev);
 	}
-	if (r == false) {
+	if (!r) {
 		r = amdgpu_read_platform_bios(adev);
 	}
-	if (r == false || adev->bios == NULL) {
+	if (!r || adev->bios == NULL) {
 		DRM_ERROR("Unable to locate a BIOS ROM\n");
 		adev->bios = NULL;
 		return false;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
index 823bf5e0b..651115dcc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
@@ -94,6 +94,7 @@ static int amdgpu_bo_list_set(struct amdgpu_device *adev,
 	unsigned last_entry = 0, first_userptr = num_entries;
 	unsigned i;
 	int r;
+	unsigned long total_size = 0;
 
 	array = drm_malloc_ab(num_entries, sizeof(struct amdgpu_bo_list_entry));
 	if (!array)
@@ -140,6 +141,7 @@ static int amdgpu_bo_list_set(struct amdgpu_device *adev,
 		if (entry->robj->prefered_domains == AMDGPU_GEM_DOMAIN_OA)
 			oa_obj = entry->robj;
 
+		total_size += amdgpu_bo_size(entry->robj);
 		trace_amdgpu_bo_list_set(list, entry->robj);
 	}
 
@@ -155,6 +157,7 @@ static int amdgpu_bo_list_set(struct amdgpu_device *adev,
 	list->array = array;
 	list->num_entries = num_entries;
 
+	trace_amdgpu_cs_bo_status(list->num_entries, total_size);
 	return 0;
 
 error_free:
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
index 6f9dcfddc..16e248e3f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
@@ -312,6 +312,8 @@ static uint32_t amdgpu_cgs_read_ind_register(struct cgs_device *cgs_device,
 		return RREG32_UVD_CTX(index);
 	case CGS_IND_REG__DIDT:
 		return RREG32_DIDT(index);
+	case CGS_IND_REG_GC_CAC:
+		return RREG32_GC_CAC(index);
 	case CGS_IND_REG__AUDIO_ENDPT:
 		DRM_ERROR("audio endpt register access not implemented.\n");
 		return 0;
@@ -336,6 +338,8 @@ static void amdgpu_cgs_write_ind_register(struct cgs_device *cgs_device,
 		return WREG32_UVD_CTX(index, value);
 	case CGS_IND_REG__DIDT:
 		return WREG32_DIDT(index, value);
+	case CGS_IND_REG_GC_CAC:
+		return WREG32_GC_CAC(index, value);
 	case CGS_IND_REG__AUDIO_ENDPT:
 		DRM_ERROR("audio endpt register access not implemented.\n");
 		return;
@@ -748,6 +752,9 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
 
 		if (!adev->pm.fw) {
 			switch (adev->asic_type) {
+			case CHIP_TOPAZ:
+				strcpy(fw_name, "/*(DEBLOBBED)*/");
+				break;
 			case CHIP_TONGA:
 				strcpy(fw_name, "/*(DEBLOBBED)*/");
 				break;
@@ -787,6 +794,7 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
 		}
 
 		hdr = (const struct smc_firmware_header_v1_0 *)	adev->pm.fw->data;
+		amdgpu_ucode_print_smc_hdr(&hdr->header);
 		adev->pm.fw_version = le32_to_cpu(hdr->header.ucode_version);
 		ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
 		ucode_start_address = le32_to_cpu(hdr->ucode_start_addr);
@@ -795,13 +803,14 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
 
 		info->version = adev->pm.fw_version;
 		info->image_size = ucode_size;
+		info->ucode_start_address = ucode_start_address;
 		info->kptr = (void *)src;
 	}
 	return 0;
 }
 
 static int amdgpu_cgs_query_system_info(struct cgs_device *cgs_device,
-				struct cgs_system_info *sys_info)
+					struct cgs_system_info *sys_info)
 {
 	CGS_FUNC_ADEV;
 
@@ -821,6 +830,12 @@ static int amdgpu_cgs_query_system_info(struct cgs_device *cgs_device,
 	case CGS_SYSTEM_INFO_PCIE_MLW:
 		sys_info->value = adev->pm.pcie_mlw_mask;
 		break;
+	case CGS_SYSTEM_INFO_PCIE_DEV:
+		sys_info->value = adev->pdev->device;
+		break;
+	case CGS_SYSTEM_INFO_PCIE_REV:
+		sys_info->value = adev->pdev->revision;
+		break;
 	case CGS_SYSTEM_INFO_CG_FLAGS:
 		sys_info->value = adev->cg_flags;
 		break;
@@ -830,6 +845,9 @@ static int amdgpu_cgs_query_system_info(struct cgs_device *cgs_device,
 	case CGS_SYSTEM_INFO_GFX_CU_INFO:
 		sys_info->value = adev->gfx.cu_info.number;
 		break;
+	case CGS_SYSTEM_INFO_GFX_SE_INFO:
+		sys_info->value = adev->gfx.config.max_shader_engines;
+		break;
 	default:
 		return -ENODEV;
 	}
@@ -903,14 +921,12 @@ static int amdgpu_cgs_acpi_eval_object(struct cgs_device *cgs_device,
 	acpi_handle handle;
 	struct acpi_object_list input;
 	struct acpi_buffer output = { ACPI_ALLOCATE_BUFFER, NULL };
-	union acpi_object *params = NULL;
-	union acpi_object *obj = NULL;
+	union acpi_object *params, *obj;
 	uint8_t name[5] = {'\0'};
-	struct cgs_acpi_method_argument *argument = NULL;
+	struct cgs_acpi_method_argument *argument;
 	uint32_t i, count;
 	acpi_status status;
-	int result = 0;
-	uint32_t func_no = 0xFFFFFFFF;
+	int result;
 
 	handle = ACPI_HANDLE(&adev->pdev->dev);
 	if (!handle)
@@ -927,7 +943,6 @@ static int amdgpu_cgs_acpi_eval_object(struct cgs_device *cgs_device,
 		if (info->pinput_argument == NULL)
 			return -EINVAL;
 		argument = info->pinput_argument;
-		func_no = argument->value;
 		for (i = 0; i < info->input_count; i++) {
 			if (((argument->type == ACPI_TYPE_STRING) ||
 			     (argument->type == ACPI_TYPE_BUFFER)) &&
@@ -972,11 +987,11 @@ static int amdgpu_cgs_acpi_eval_object(struct cgs_device *cgs_device,
 				params->integer.value = argument->value;
 				break;
 			case ACPI_TYPE_STRING:
-				params->string.length = argument->method_length;
+				params->string.length = argument->data_length;
 				params->string.pointer = argument->pointer;
 				break;
 			case ACPI_TYPE_BUFFER:
-				params->buffer.length = argument->method_length;
+				params->buffer.length = argument->data_length;
 				params->buffer.pointer = argument->pointer;
 				break;
 			default:
@@ -996,7 +1011,7 @@ static int amdgpu_cgs_acpi_eval_object(struct cgs_device *cgs_device,
 
 	if (ACPI_FAILURE(status)) {
 		result = -EIO;
-		goto error;
+		goto free_input;
 	}
 
 	/* return the output info */
@@ -1006,7 +1021,7 @@ static int amdgpu_cgs_acpi_eval_object(struct cgs_device *cgs_device,
 		if ((obj->type != ACPI_TYPE_PACKAGE) ||
 			(obj->package.count != count)) {
 			result = -EIO;
-			goto error;
+			goto free_obj;
 		}
 		params = obj->package.elements;
 	} else
@@ -1014,13 +1029,13 @@ static int amdgpu_cgs_acpi_eval_object(struct cgs_device *cgs_device,
 
 	if (params == NULL) {
 		result = -EIO;
-		goto error;
+		goto free_obj;
 	}
 
 	for (i = 0; i < count; i++) {
 		if (argument->type != params->type) {
 			result = -EIO;
-			goto error;
+			goto free_obj;
 		}
 		switch (params->type) {
 		case ACPI_TYPE_INTEGER:
@@ -1030,7 +1045,7 @@ static int amdgpu_cgs_acpi_eval_object(struct cgs_device *cgs_device,
 			if ((params->string.length != argument->data_length) ||
 				(params->string.pointer == NULL)) {
 				result = -EIO;
-				goto error;
+				goto free_obj;
 			}
 			strncpy(argument->pointer,
 				params->string.pointer,
@@ -1039,7 +1054,7 @@ static int amdgpu_cgs_acpi_eval_object(struct cgs_device *cgs_device,
 		case ACPI_TYPE_BUFFER:
 			if (params->buffer.pointer == NULL) {
 				result = -EIO;
-				goto error;
+				goto free_obj;
 			}
 			memcpy(argument->pointer,
 				params->buffer.pointer,
@@ -1052,9 +1067,10 @@ static int amdgpu_cgs_acpi_eval_object(struct cgs_device *cgs_device,
 		params++;
 	}
 
-error:
-	if (obj != NULL)
-		kfree(obj);
+	result = 0;
+free_obj:
+	kfree(obj);
+free_input:
 	kfree((void *)input.pointer);
 	return result;
 }
@@ -1066,7 +1082,7 @@ static int amdgpu_cgs_acpi_eval_object(struct cgs_device *cgs_device,
 }
 #endif
 
-int amdgpu_cgs_call_acpi_method(struct cgs_device *cgs_device,
+static int amdgpu_cgs_call_acpi_method(struct cgs_device *cgs_device,
 					uint32_t acpi_method,
 					uint32_t acpi_function,
 					void *pinput, void *poutput,
@@ -1079,17 +1095,14 @@ int amdgpu_cgs_call_acpi_method(struct cgs_device *cgs_device,
 	struct cgs_acpi_method_info info = {0};
 
 	acpi_input[0].type = CGS_ACPI_TYPE_INTEGER;
-	acpi_input[0].method_length = sizeof(uint32_t);
 	acpi_input[0].data_length = sizeof(uint32_t);
 	acpi_input[0].value = acpi_function;
 
 	acpi_input[1].type = CGS_ACPI_TYPE_BUFFER;
-	acpi_input[1].method_length = CGS_ACPI_MAX_BUFFER_SIZE;
 	acpi_input[1].data_length = input_size;
 	acpi_input[1].pointer = pinput;
 
 	acpi_output.type = CGS_ACPI_TYPE_BUFFER;
-	acpi_output.method_length = CGS_ACPI_MAX_BUFFER_SIZE;
 	acpi_output.data_length = output_size;
 	acpi_output.pointer = poutput;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 9bc8f1d99..0307ff588 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -216,11 +216,8 @@ int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
 	if (ret)
 		goto free_all_kdata;
 
-	if (p->uf_entry.robj) {
-		p->job->uf_bo = amdgpu_bo_ref(p->uf_entry.robj);
-		p->job->uf_offset = uf_offset;
-	}
-
+	if (p->uf_entry.robj)
+		p->job->uf_addr = uf_offset;
 	kfree(chunk_array);
 	return 0;
 
@@ -459,7 +456,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 		list_splice(&need_pages, &p->validated);
 	}
 
-	amdgpu_vm_get_pt_bos(&fpriv->vm, &duplicates);
+	amdgpu_vm_get_pt_bos(p->adev, &fpriv->vm, &duplicates);
 
 	p->bytes_moved_threshold = amdgpu_cs_get_threshold_for_moves(p->adev);
 	p->bytes_moved = 0;
@@ -472,6 +469,9 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 	if (r)
 		goto error_validate;
 
+	fpriv->vm.last_eviction_counter =
+		atomic64_read(&p->adev->num_evictions);
+
 	if (p->bo_list) {
 		struct amdgpu_bo *gds = p->bo_list->gds_obj;
 		struct amdgpu_bo *gws = p->bo_list->gws_obj;
@@ -499,6 +499,9 @@ 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);
+
 error_validate:
 	if (r) {
 		amdgpu_vm_move_pt_bos_in_lru(p->adev, &fpriv->vm);
@@ -653,18 +656,21 @@ static int amdgpu_cs_ib_vm_chunk(struct amdgpu_device *adev,
 
 	/* Only for UVD/VCE VM emulation */
 	if (ring->funcs->parse_cs) {
+		p->job->vm = NULL;
 		for (i = 0; i < p->job->num_ibs; i++) {
 			r = amdgpu_ring_parse_cs(ring, p, i);
 			if (r)
 				return r;
 		}
-	}
+	} else {
+		p->job->vm_pd_addr = amdgpu_bo_gpu_offset(vm->page_directory);
 
-	r = amdgpu_bo_vm_update_pte(p, vm);
-	if (!r)
-		amdgpu_cs_sync_rings(p);
+		r = amdgpu_bo_vm_update_pte(p, vm);
+		if (r)
+			return r;
+	}
 
-	return r;
+	return amdgpu_cs_sync_rings(p);
 }
 
 static int amdgpu_cs_handle_lockup(struct amdgpu_device *adev, int r)
@@ -761,7 +767,7 @@ static int amdgpu_cs_ib_fill(struct amdgpu_device *adev,
 	}
 
 	/* UVD & VCE fw doesn't support user fences */
-	if (parser->job->uf_bo && (
+	if (parser->job->uf_addr && (
 	    parser->job->ring->type == AMDGPU_RING_TYPE_UVD ||
 	    parser->job->ring->type == AMDGPU_RING_TYPE_VCE))
 		return -EINVAL;
@@ -830,17 +836,13 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 {
 	struct amdgpu_ring *ring = p->job->ring;
 	struct amd_sched_entity *entity = &p->ctx->rings[ring->idx].entity;
-	struct fence *fence;
 	struct amdgpu_job *job;
 	int r;
 
 	job = p->job;
 	p->job = NULL;
 
-	r = amd_sched_job_init(&job->base, &ring->sched,
-			       entity, amdgpu_job_timeout_func,
-			       amdgpu_job_free_func,
-			       p->filp, &fence);
+	r = amd_sched_job_init(&job->base, &ring->sched, entity, p->filp);
 	if (r) {
 		amdgpu_job_free(job);
 		return r;
@@ -848,9 +850,10 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 
 	job->owner = p->filp;
 	job->ctx = entity->fence_context;
-	p->fence = fence_get(fence);
-	cs->out.handle = amdgpu_ctx_add_fence(p->ctx, ring, fence);
+	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;
+	amdgpu_job_free_resources(job);
 
 	trace_amdgpu_cs_ioctl(job);
 	amd_sched_entity_push_job(&job->base);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index b7f5650d8..39c01b942 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -25,6 +25,7 @@
  *          Alex Deucher
  *          Jerome Glisse
  */
+#include <linux/kthread.h>
 #include <linux/console.h>
 #include <linux/slab.h>
 #include <linux/debugfs.h>
@@ -35,6 +36,7 @@
 #include <linux/vga_switcheroo.h>
 #include <linux/efi.h>
 #include "amdgpu.h"
+#include "amdgpu_trace.h"
 #include "amdgpu_i2c.h"
 #include "atom.h"
 #include "amdgpu_atombios.h"
@@ -79,24 +81,27 @@ bool amdgpu_device_is_px(struct drm_device *dev)
 uint32_t amdgpu_mm_rreg(struct amdgpu_device *adev, uint32_t reg,
 			bool always_indirect)
 {
+	uint32_t ret;
+
 	if ((reg * 4) < adev->rmmio_size && !always_indirect)
-		return readl(((void __iomem *)adev->rmmio) + (reg * 4));
+		ret = readl(((void __iomem *)adev->rmmio) + (reg * 4));
 	else {
 		unsigned long flags;
-		uint32_t ret;
 
 		spin_lock_irqsave(&adev->mmio_idx_lock, flags);
 		writel((reg * 4), ((void __iomem *)adev->rmmio) + (mmMM_INDEX * 4));
 		ret = readl(((void __iomem *)adev->rmmio) + (mmMM_DATA * 4));
 		spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
-
-		return ret;
 	}
+	trace_amdgpu_mm_rreg(adev->pdev->device, reg, ret);
+	return ret;
 }
 
 void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
 		    bool always_indirect)
 {
+	trace_amdgpu_mm_wreg(adev->pdev->device, reg, v);
+	
 	if ((reg * 4) < adev->rmmio_size && !always_indirect)
 		writel(v, ((void __iomem *)adev->rmmio) + (reg * 4));
 	else {
@@ -1070,11 +1075,14 @@ int amdgpu_set_clockgating_state(struct amdgpu_device *adev,
 	int i, r = 0;
 
 	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_block_status[i].valid)
+			continue;
 		if (adev->ip_blocks[i].type == block_type) {
 			r = adev->ip_blocks[i].funcs->set_clockgating_state((void *)adev,
 									    state);
 			if (r)
 				return r;
+			break;
 		}
 	}
 	return r;
@@ -1087,16 +1095,53 @@ int amdgpu_set_powergating_state(struct amdgpu_device *adev,
 	int i, r = 0;
 
 	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_block_status[i].valid)
+			continue;
 		if (adev->ip_blocks[i].type == block_type) {
 			r = adev->ip_blocks[i].funcs->set_powergating_state((void *)adev,
 									    state);
 			if (r)
 				return r;
+			break;
 		}
 	}
 	return r;
 }
 
+int amdgpu_wait_for_idle(struct amdgpu_device *adev,
+			 enum amd_ip_block_type block_type)
+{
+	int i, r;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_block_status[i].valid)
+			continue;
+		if (adev->ip_blocks[i].type == block_type) {
+			r = adev->ip_blocks[i].funcs->wait_for_idle((void *)adev);
+			if (r)
+				return r;
+			break;
+		}
+	}
+	return 0;
+
+}
+
+bool amdgpu_is_idle(struct amdgpu_device *adev,
+		    enum amd_ip_block_type block_type)
+{
+	int i;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_block_status[i].valid)
+			continue;
+		if (adev->ip_blocks[i].type == block_type)
+			return adev->ip_blocks[i].funcs->is_idle((void *)adev);
+	}
+	return true;
+
+}
+
 const struct amdgpu_ip_block_version * amdgpu_get_ip_block(
 					struct amdgpu_device *adev,
 					enum amd_ip_block_type type)
@@ -1209,6 +1254,9 @@ static int amdgpu_early_init(struct amdgpu_device *adev)
 		}
 	}
 
+	adev->cg_flags &= amdgpu_cg_mask;
+	adev->pg_flags &= amdgpu_pg_mask;
+
 	return 0;
 }
 
@@ -1440,9 +1488,12 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	adev->uvd_ctx_wreg = &amdgpu_invalid_wreg;
 	adev->didt_rreg = &amdgpu_invalid_rreg;
 	adev->didt_wreg = &amdgpu_invalid_wreg;
+	adev->gc_cac_rreg = &amdgpu_invalid_rreg;
+	adev->gc_cac_wreg = &amdgpu_invalid_wreg;
 	adev->audio_endpt_rreg = &amdgpu_block_invalid_rreg;
 	adev->audio_endpt_wreg = &amdgpu_block_invalid_wreg;
 
+
 	DRM_INFO("initializing kernel modesetting (%s 0x%04X:0x%04X 0x%04X:0x%04X 0x%02X).\n",
 		 amdgpu_asic_name[adev->asic_type], pdev->vendor, pdev->device,
 		 pdev->subsystem_vendor, pdev->subsystem_device, pdev->revision);
@@ -1467,6 +1518,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	spin_lock_init(&adev->pcie_idx_lock);
 	spin_lock_init(&adev->uvd_ctx_idx_lock);
 	spin_lock_init(&adev->didt_idx_lock);
+	spin_lock_init(&adev->gc_cac_idx_lock);
 	spin_lock_init(&adev->audio_endpt_idx_lock);
 
 	adev->rmmio_base = pci_resource_start(adev->pdev, 5);
@@ -1511,17 +1563,20 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 		vga_switcheroo_init_domain_pm_ops(adev->dev, &adev->vga_pm_domain);
 
 	/* Read BIOS */
-	if (!amdgpu_get_bios(adev))
-		return -EINVAL;
+	if (!amdgpu_get_bios(adev)) {
+		r = -EINVAL;
+		goto failed;
+	}
 	/* Must be an ATOMBIOS */
 	if (!adev->is_atom_bios) {
 		dev_err(adev->dev, "Expecting atombios for GPU\n");
-		return -EINVAL;
+		r = -EINVAL;
+		goto failed;
 	}
 	r = amdgpu_atombios_init(adev);
 	if (r) {
 		dev_err(adev->dev, "amdgpu_atombios_init failed\n");
-		return r;
+		goto failed;
 	}
 
 	/* See if the asic supports SR-IOV */
@@ -1538,7 +1593,8 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	     !(adev->virtualization.caps & AMDGPU_VIRT_CAPS_SRIOV_EN))) {
 		if (!adev->bios) {
 			dev_err(adev->dev, "Card not posted and no BIOS - ignoring\n");
-			return -EINVAL;
+			r = -EINVAL;
+			goto failed;
 		}
 		DRM_INFO("GPU not posted. posting now...\n");
 		amdgpu_atom_asic_init(adev->mode_info.atom_context);
@@ -1548,7 +1604,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	r = amdgpu_atombios_get_clock_info(adev);
 	if (r) {
 		dev_err(adev->dev, "amdgpu_atombios_get_clock_info failed\n");
-		return r;
+		goto failed;
 	}
 	/* init i2c buses */
 	amdgpu_atombios_i2c_init(adev);
@@ -1557,7 +1613,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	r = amdgpu_fence_driver_init(adev);
 	if (r) {
 		dev_err(adev->dev, "amdgpu_fence_driver_init failed\n");
-		return r;
+		goto failed;
 	}
 
 	/* init the mode config */
@@ -1567,7 +1623,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	if (r) {
 		dev_err(adev->dev, "amdgpu_init failed\n");
 		amdgpu_fini(adev);
-		return r;
+		goto failed;
 	}
 
 	adev->accel_working = true;
@@ -1577,7 +1633,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	r = amdgpu_ib_pool_init(adev);
 	if (r) {
 		dev_err(adev->dev, "IB initialization failed (%d).\n", r);
-		return r;
+		goto failed;
 	}
 
 	r = amdgpu_ib_ring_tests(adev);
@@ -1594,6 +1650,12 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 		DRM_ERROR("registering register debugfs failed (%d).\n", r);
 	}
 
+	r = amdgpu_debugfs_firmware_init(adev);
+	if (r) {
+		DRM_ERROR("registering firmware debugfs failed (%d).\n", r);
+		return r;
+	}
+
 	if ((amdgpu_testing & 1)) {
 		if (adev->accel_working)
 			amdgpu_test_moves(adev);
@@ -1619,10 +1681,15 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	r = amdgpu_late_init(adev);
 	if (r) {
 		dev_err(adev->dev, "amdgpu_late_init failed\n");
-		return r;
+		goto failed;
 	}
 
 	return 0;
+
+failed:
+	if (runtime)
+		vga_switcheroo_fini_domain_pm_ops(adev->dev);
+	return r;
 }
 
 static void amdgpu_debugfs_remove_files(struct amdgpu_device *adev);
@@ -1641,6 +1708,7 @@ void amdgpu_device_fini(struct amdgpu_device *adev)
 
 	DRM_INFO("amdgpu: finishing device.\n");
 	adev->shutdown = true;
+	drm_crtc_force_disable_all(adev->ddev);
 	/* evict vram memory */
 	amdgpu_bo_evict_vram(adev);
 	amdgpu_ib_pool_fini(adev);
@@ -1656,6 +1724,8 @@ void amdgpu_device_fini(struct amdgpu_device *adev)
 	kfree(adev->bios);
 	adev->bios = NULL;
 	vga_switcheroo_unregister_client(adev->pdev);
+	if (adev->flags & AMD_IS_PX)
+		vga_switcheroo_fini_domain_pm_ops(adev->dev);
 	vga_client_register(adev->pdev, NULL, NULL, NULL);
 	if (adev->rio_mem)
 		pci_iounmap(adev->pdev, adev->rio_mem);
@@ -1877,11 +1947,6 @@ int amdgpu_resume_kms(struct drm_device *dev, bool resume, bool fbcon)
  */
 int amdgpu_gpu_reset(struct amdgpu_device *adev)
 {
-	unsigned ring_sizes[AMDGPU_MAX_RINGS];
-	uint32_t *ring_data[AMDGPU_MAX_RINGS];
-
-	bool saved = false;
-
 	int i, r;
 	int resched;
 
@@ -1890,22 +1955,30 @@ int amdgpu_gpu_reset(struct amdgpu_device *adev)
 	/* block TTM */
 	resched = ttm_bo_lock_delayed_workqueue(&adev->mman.bdev);
 
-	r = amdgpu_suspend(adev);
-
+	/* block scheduler */
 	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
 		struct amdgpu_ring *ring = adev->rings[i];
+
 		if (!ring)
 			continue;
-
-		ring_sizes[i] = amdgpu_ring_backup(ring, &ring_data[i]);
-		if (ring_sizes[i]) {
-			saved = true;
-			dev_info(adev->dev, "Saved %d dwords of commands "
-				 "on ring %d.\n", ring_sizes[i], i);
-		}
+		kthread_park(ring->sched.thread);
+		amd_sched_hw_job_reset(&ring->sched);
 	}
+	/* after all hw jobs are reset, hw fence is meaningless, so force_completion */
+	amdgpu_fence_driver_force_completion(adev);
+
+	/* save scratch */
+	amdgpu_atombios_scratch_regs_save(adev);
+	r = amdgpu_suspend(adev);
 
 retry:
+	/* Disable fb access */
+	if (adev->mode_info.num_crtc) {
+		struct amdgpu_mode_mc_save save;
+		amdgpu_display_stop_mc_access(adev, &save);
+		amdgpu_wait_for_idle(adev, AMD_IP_BLOCK_TYPE_GMC);
+	}
+
 	r = amdgpu_asic_reset(adev);
 	/* post card */
 	amdgpu_atom_asic_init(adev->mode_info.atom_context);
@@ -1914,32 +1987,29 @@ retry:
 		dev_info(adev->dev, "GPU reset succeeded, trying to resume\n");
 		r = amdgpu_resume(adev);
 	}
-
+	/* restore scratch */
+	amdgpu_atombios_scratch_regs_restore(adev);
 	if (!r) {
+		r = amdgpu_ib_ring_tests(adev);
+		if (r) {
+			dev_err(adev->dev, "ib ring test failed (%d).\n", r);
+			r = amdgpu_suspend(adev);
+			goto retry;
+		}
+
 		for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
 			struct amdgpu_ring *ring = adev->rings[i];
 			if (!ring)
 				continue;
-
-			amdgpu_ring_restore(ring, ring_sizes[i], ring_data[i]);
-			ring_sizes[i] = 0;
-			ring_data[i] = NULL;
-		}
-
-		r = amdgpu_ib_ring_tests(adev);
-		if (r) {
-			dev_err(adev->dev, "ib ring test failed (%d).\n", r);
-			if (saved) {
-				saved = false;
-				r = amdgpu_suspend(adev);
-				goto retry;
-			}
+			amd_sched_job_recovery(&ring->sched);
+			kthread_unpark(ring->sched.thread);
 		}
 	} else {
-		amdgpu_fence_driver_force_completion(adev);
+		dev_err(adev->dev, "asic resume failed (%d).\n", r);
 		for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
-			if (adev->rings[i])
-				kfree(ring_data[i]);
+			if (adev->rings[i]) {
+				kthread_unpark(adev->rings[i]->sched.thread);
+			}
 		}
 	}
 
@@ -1950,13 +2020,11 @@ retry:
 		/* bad news, how to tell it to userspace ? */
 		dev_info(adev->dev, "GPU reset failed\n");
 	}
+	amdgpu_irq_gpu_reset_resume_helper(adev);
 
 	return r;
 }
 
-#define AMDGPU_DEFAULT_PCIE_GEN_MASK 0x30007  /* gen: chipset 1/2, asic 1/2/3 */
-#define AMDGPU_DEFAULT_PCIE_MLW_MASK 0x2f0000 /* 1/2/4/8/16 lanes */
-
 void amdgpu_get_pcie_info(struct amdgpu_device *adev)
 {
 	u32 mask;
@@ -2110,20 +2178,43 @@ static ssize_t amdgpu_debugfs_regs_read(struct file *f, char __user *buf,
 	struct amdgpu_device *adev = f->f_inode->i_private;
 	ssize_t result = 0;
 	int r;
+	bool use_bank;
+	unsigned instance_bank, sh_bank, se_bank;
 
 	if (size & 0x3 || *pos & 0x3)
 		return -EINVAL;
 
+	if (*pos & (1ULL << 62)) {
+		se_bank = (*pos >> 24) & 0x3FF;
+		sh_bank = (*pos >> 34) & 0x3FF;
+		instance_bank = (*pos >> 44) & 0x3FF;
+		use_bank = 1;
+		*pos &= 0xFFFFFF;
+	} else {
+		use_bank = 0;
+	}
+
+	if (use_bank) {
+		if (sh_bank >= adev->gfx.config.max_sh_per_se ||
+		    se_bank >= adev->gfx.config.max_shader_engines)
+			return -EINVAL;
+		mutex_lock(&adev->grbm_idx_mutex);
+		amdgpu_gfx_select_se_sh(adev, se_bank,
+					sh_bank, instance_bank);
+	}
+
 	while (size) {
 		uint32_t value;
 
 		if (*pos > adev->rmmio_size)
-			return result;
+			goto end;
 
 		value = RREG32(*pos >> 2);
 		r = put_user(value, (uint32_t *)buf);
-		if (r)
-			return r;
+		if (r) {
+			result = r;
+			goto end;
+		}
 
 		result += 4;
 		buf += 4;
@@ -2131,6 +2222,12 @@ static ssize_t amdgpu_debugfs_regs_read(struct file *f, char __user *buf,
 		size -= 4;
 	}
 
+end:
+	if (use_bank) {
+		amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+		mutex_unlock(&adev->grbm_idx_mutex);
+	}
+
 	return result;
 }
 
@@ -2330,6 +2427,68 @@ static ssize_t amdgpu_debugfs_regs_smc_write(struct file *f, const char __user *
 	return result;
 }
 
+static ssize_t amdgpu_debugfs_gca_config_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
+{
+	struct amdgpu_device *adev = f->f_inode->i_private;
+	ssize_t result = 0;
+	int r;
+	uint32_t *config, no_regs = 0;
+
+	if (size & 0x3 || *pos & 0x3)
+		return -EINVAL;
+
+	config = kmalloc(256 * sizeof(*config), GFP_KERNEL);
+	if (!config)
+		return -ENOMEM;
+
+	/* version, increment each time something is added */
+	config[no_regs++] = 0;
+	config[no_regs++] = adev->gfx.config.max_shader_engines;
+	config[no_regs++] = adev->gfx.config.max_tile_pipes;
+	config[no_regs++] = adev->gfx.config.max_cu_per_sh;
+	config[no_regs++] = adev->gfx.config.max_sh_per_se;
+	config[no_regs++] = adev->gfx.config.max_backends_per_se;
+	config[no_regs++] = adev->gfx.config.max_texture_channel_caches;
+	config[no_regs++] = adev->gfx.config.max_gprs;
+	config[no_regs++] = adev->gfx.config.max_gs_threads;
+	config[no_regs++] = adev->gfx.config.max_hw_contexts;
+	config[no_regs++] = adev->gfx.config.sc_prim_fifo_size_frontend;
+	config[no_regs++] = adev->gfx.config.sc_prim_fifo_size_backend;
+	config[no_regs++] = adev->gfx.config.sc_hiz_tile_fifo_size;
+	config[no_regs++] = adev->gfx.config.sc_earlyz_tile_fifo_size;
+	config[no_regs++] = adev->gfx.config.num_tile_pipes;
+	config[no_regs++] = adev->gfx.config.backend_enable_mask;
+	config[no_regs++] = adev->gfx.config.mem_max_burst_length_bytes;
+	config[no_regs++] = adev->gfx.config.mem_row_size_in_kb;
+	config[no_regs++] = adev->gfx.config.shader_engine_tile_size;
+	config[no_regs++] = adev->gfx.config.num_gpus;
+	config[no_regs++] = adev->gfx.config.multi_gpu_tile_size;
+	config[no_regs++] = adev->gfx.config.mc_arb_ramcfg;
+	config[no_regs++] = adev->gfx.config.gb_addr_config;
+	config[no_regs++] = adev->gfx.config.num_rbs;
+
+	while (size && (*pos < no_regs * 4)) {
+		uint32_t value;
+
+		value = config[*pos >> 2];
+		r = put_user(value, (uint32_t *)buf);
+		if (r) {
+			kfree(config);
+			return r;
+		}
+
+		result += 4;
+		buf += 4;
+		*pos += 4;
+		size -= 4;
+	}
+
+	kfree(config);
+	return result;
+}
+
+
 static const struct file_operations amdgpu_debugfs_regs_fops = {
 	.owner = THIS_MODULE,
 	.read = amdgpu_debugfs_regs_read,
@@ -2355,11 +2514,18 @@ static const struct file_operations amdgpu_debugfs_regs_smc_fops = {
 	.llseek = default_llseek
 };
 
+static const struct file_operations amdgpu_debugfs_gca_config_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_gca_config_read,
+	.llseek = default_llseek
+};
+
 static const struct file_operations *debugfs_regs[] = {
 	&amdgpu_debugfs_regs_fops,
 	&amdgpu_debugfs_regs_didt_fops,
 	&amdgpu_debugfs_regs_pcie_fops,
 	&amdgpu_debugfs_regs_smc_fops,
+	&amdgpu_debugfs_gca_config_fops,
 };
 
 static const char *debugfs_regs_names[] = {
@@ -2367,6 +2533,7 @@ static const char *debugfs_regs_names[] = {
 	"amdgpu_regs_didt",
 	"amdgpu_regs_pcie",
 	"amdgpu_regs_smc",
+	"amdgpu_gca_config",
 };
 
 static int amdgpu_debugfs_regs_init(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
index b0832da2e..76f960283 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
@@ -122,7 +122,7 @@ static void amdgpu_flip_work_func(struct work_struct *__work)
 		spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
 		usleep_range(min_udelay, 2 * min_udelay);
 		spin_lock_irqsave(&crtc->dev->event_lock, flags);
-	};
+	}
 
 	if (!repcnt)
 		DRM_DEBUG_DRIVER("Delay problem on crtc %d: min_udelay %d, "
@@ -220,19 +220,17 @@ int amdgpu_crtc_page_flip(struct drm_crtc *crtc,
 
 	r = amdgpu_bo_pin_restricted(new_rbo, AMDGPU_GEM_DOMAIN_VRAM, 0, 0, &base);
 	if (unlikely(r != 0)) {
-		amdgpu_bo_unreserve(new_rbo);
 		r = -EINVAL;
 		DRM_ERROR("failed to pin new rbo buffer before flip\n");
-		goto cleanup;
+		goto unreserve;
 	}
 
 	r = reservation_object_get_fences_rcu(new_rbo->tbo.resv, &work->excl,
 					      &work->shared_count,
 					      &work->shared);
 	if (unlikely(r != 0)) {
-		amdgpu_bo_unreserve(new_rbo);
 		DRM_ERROR("failed to get fences for buffer\n");
-		goto cleanup;
+		goto unpin;
 	}
 
 	amdgpu_bo_get_tiling_flags(new_rbo, &tiling_flags);
@@ -240,7 +238,7 @@ int amdgpu_crtc_page_flip(struct drm_crtc *crtc,
 
 	work->base = base;
 
-	r = drm_vblank_get(crtc->dev, amdgpu_crtc->crtc_id);
+	r = drm_crtc_vblank_get(crtc);
 	if (r) {
 		DRM_ERROR("failed to get vblank before flip\n");
 		goto pflip_cleanup;
@@ -268,16 +266,18 @@ int amdgpu_crtc_page_flip(struct drm_crtc *crtc,
 	return 0;
 
 vblank_cleanup:
-	drm_vblank_put(crtc->dev, amdgpu_crtc->crtc_id);
+	drm_crtc_vblank_put(crtc);
 
 pflip_cleanup:
 	if (unlikely(amdgpu_bo_reserve(new_rbo, false) != 0)) {
 		DRM_ERROR("failed to reserve new rbo in error path\n");
 		goto cleanup;
 	}
+unpin:
 	if (unlikely(amdgpu_bo_unpin(new_rbo) != 0)) {
 		DRM_ERROR("failed to unpin new rbo in error path\n");
 	}
+unreserve:
 	amdgpu_bo_unreserve(new_rbo);
 
 cleanup:
@@ -516,9 +516,7 @@ static void amdgpu_user_framebuffer_destroy(struct drm_framebuffer *fb)
 {
 	struct amdgpu_framebuffer *amdgpu_fb = to_amdgpu_framebuffer(fb);
 
-	if (amdgpu_fb->obj) {
-		drm_gem_object_unreference_unlocked(amdgpu_fb->obj);
-	}
+	drm_gem_object_unreference_unlocked(amdgpu_fb->obj);
 	drm_framebuffer_cleanup(fb);
 	kfree(amdgpu_fb);
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index f888c015f..9aa533cf4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -52,9 +52,10 @@
  * - 3.1.0 - allow reading more status registers (GRBM, SRBM, SDMA, CP)
  * - 3.2.0 - GFX8: Uses EOP_TC_WB_ACTION_EN, so UMDs don't have to do the same
  *           at the end of IBs.
+ * - 3.3.0 - Add VM support for UVD on supported hardware.
  */
 #define KMS_DRIVER_MAJOR	3
-#define KMS_DRIVER_MINOR	2
+#define KMS_DRIVER_MINOR	3
 #define KMS_DRIVER_PATCHLEVEL	0
 
 int amdgpu_vram_limit = 0;
@@ -82,8 +83,12 @@ int amdgpu_exp_hw_support = 0;
 int amdgpu_sched_jobs = 32;
 int amdgpu_sched_hw_submission = 2;
 int amdgpu_powerplay = -1;
+int amdgpu_powercontainment = 1;
 unsigned amdgpu_pcie_gen_cap = 0;
 unsigned amdgpu_pcie_lane_cap = 0;
+unsigned amdgpu_cg_mask = 0xffffffff;
+unsigned amdgpu_pg_mask = 0xffffffff;
+char *amdgpu_disable_cu = NULL;
 
 MODULE_PARM_DESC(vramlimit, "Restrict VRAM for testing, in megabytes");
 module_param_named(vramlimit, amdgpu_vram_limit, int, 0600);
@@ -160,6 +165,9 @@ module_param_named(sched_hw_submission, amdgpu_sched_hw_submission, int, 0444);
 #ifdef CONFIG_DRM_AMD_POWERPLAY
 MODULE_PARM_DESC(powerplay, "Powerplay component (1 = enable, 0 = disable, -1 = auto (default))");
 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);
 #endif
 
 MODULE_PARM_DESC(pcie_gen_cap, "PCIE Gen Caps (0: autodetect (default))");
@@ -168,6 +176,15 @@ module_param_named(pcie_gen_cap, amdgpu_pcie_gen_cap, uint, 0444);
 MODULE_PARM_DESC(pcie_lane_cap, "PCIE Lane Caps (0: autodetect (default))");
 module_param_named(pcie_lane_cap, amdgpu_pcie_lane_cap, uint, 0444);
 
+MODULE_PARM_DESC(cg_mask, "Clockgating flags mask (0 = disable clock gating)");
+module_param_named(cg_mask, amdgpu_cg_mask, uint, 0444);
+
+MODULE_PARM_DESC(pg_mask, "Powergating flags mask (0 = disable power gating)");
+module_param_named(pg_mask, amdgpu_pg_mask, uint, 0444);
+
+MODULE_PARM_DESC(disable_cu, "Disable CUs (se.sh.cu,...)");
+module_param_named(disable_cu, amdgpu_disable_cu, charp, 0444);
+
 static const struct pci_device_id pciidlist[] = {
 #ifdef CONFIG_DRM_AMDGPU_CIK
 	/* Kaveri */
@@ -413,7 +430,10 @@ static int amdgpu_pmops_runtime_suspend(struct device *dev)
 	pci_save_state(pdev);
 	pci_disable_device(pdev);
 	pci_ignore_hotplug(pdev);
-	pci_set_power_state(pdev, PCI_D3cold);
+	if (amdgpu_is_atpx_hybrid())
+		pci_set_power_state(pdev, PCI_D3cold);
+	else if (!amdgpu_has_atpx_dgpu_power_cntl())
+		pci_set_power_state(pdev, PCI_D3hot);
 	drm_dev->switch_power_state = DRM_SWITCH_POWER_DYNAMIC_OFF;
 
 	return 0;
@@ -430,7 +450,9 @@ static int amdgpu_pmops_runtime_resume(struct device *dev)
 
 	drm_dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
 
-	pci_set_power_state(pdev, PCI_D0);
+	if (amdgpu_is_atpx_hybrid() ||
+	    !amdgpu_has_atpx_dgpu_power_cntl())
+		pci_set_power_state(pdev, PCI_D0);
 	pci_restore_state(pdev);
 	ret = pci_enable_device(pdev);
 	if (ret)
@@ -515,7 +537,7 @@ static struct drm_driver kms_driver = {
 	.driver_features =
 	    DRIVER_USE_AGP |
 	    DRIVER_HAVE_IRQ | DRIVER_IRQ_SHARED | DRIVER_GEM |
-	    DRIVER_PRIME | DRIVER_RENDER,
+	    DRIVER_PRIME | DRIVER_RENDER | DRIVER_MODESET,
 	.dev_priv_size = 0,
 	.load = amdgpu_driver_load_kms,
 	.open = amdgpu_driver_open_kms,
@@ -590,7 +612,6 @@ static int __init amdgpu_init(void)
 	DRM_INFO("amdgpu kernel modesetting enabled.\n");
 	driver = &kms_driver;
 	pdriver = &amdgpu_kms_pci_driver;
-	driver->driver_features |= DRIVER_MODESET;
 	driver->num_ioctls = amdgpu_max_kms_ioctl;
 	amdgpu_register_atpx_handler();
 	/* let modprobe override vga console setting */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
index d1558768c..0b109aebf 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
@@ -204,16 +204,25 @@ void amdgpu_fence_process(struct amdgpu_ring *ring)
 	if (seq != ring->fence_drv.sync_seq)
 		amdgpu_fence_schedule_fallback(ring);
 
-	while (last_seq != seq) {
+	if (unlikely(seq == last_seq))
+		return;
+
+	last_seq &= drv->num_fences_mask;
+	seq &= drv->num_fences_mask;
+
+	do {
 		struct fence *fence, **ptr;
 
-		ptr = &drv->fences[++last_seq & drv->num_fences_mask];
+		++last_seq;
+		last_seq &= drv->num_fences_mask;
+		ptr = &drv->fences[last_seq];
 
 		/* There is always exactly one thread signaling this fence slot */
 		fence = rcu_dereference_protected(*ptr, 1);
 		RCU_INIT_POINTER(*ptr, NULL);
 
-		BUG_ON(!fence);
+		if (!fence)
+			continue;
 
 		r = fence_signal(fence);
 		if (!r)
@@ -222,7 +231,7 @@ void amdgpu_fence_process(struct amdgpu_ring *ring)
 			BUG();
 
 		fence_put(fence);
-	}
+	} while (last_seq != seq);
 }
 
 /**
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index 8fab64860..88fbed238 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -503,7 +503,7 @@ static void amdgpu_gem_va_update_vm(struct amdgpu_device *adev,
 	if (r)
 		goto error_print;
 
-	amdgpu_vm_get_pt_bos(bo_va->vm, &duplicates);
+	amdgpu_vm_get_pt_bos(adev, bo_va->vm, &duplicates);
 	list_for_each_entry(entry, &list, head) {
 		domain = amdgpu_mem_type_to_domain(entry->bo->mem.mem_type);
 		/* if anything is swapped out don't swap it in here,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
index 9f95da4f0..a074edd95 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
@@ -70,3 +70,47 @@ void amdgpu_gfx_scratch_free(struct amdgpu_device *adev, uint32_t reg)
 		}
 	}
 }
+
+/**
+ * amdgpu_gfx_parse_disable_cu - Parse the disable_cu module parameter
+ *
+ * @mask: array in which the per-shader array disable masks will be stored
+ * @max_se: number of SEs
+ * @max_sh: number of SHs
+ *
+ * The bitmask of CUs to be disabled in the shader array determined by se and
+ * sh is stored in mask[se * max_sh + sh].
+ */
+void amdgpu_gfx_parse_disable_cu(unsigned *mask, unsigned max_se, unsigned max_sh)
+{
+	unsigned se, sh, cu;
+	const char *p;
+
+	memset(mask, 0, sizeof(*mask) * max_se * max_sh);
+
+	if (!amdgpu_disable_cu || !*amdgpu_disable_cu)
+		return;
+
+	p = amdgpu_disable_cu;
+	for (;;) {
+		char *next;
+		int ret = sscanf(p, "%u.%u.%u", &se, &sh, &cu);
+		if (ret < 3) {
+			DRM_ERROR("amdgpu: could not parse disable_cu\n");
+			return;
+		}
+
+		if (se < max_se && sh < max_sh && cu < 16) {
+			DRM_INFO("amdgpu: disabling CU %u.%u.%u\n", se, sh, cu);
+			mask[se * max_sh + sh] |= 1u << cu;
+		} else {
+			DRM_ERROR("amdgpu: disable_cu %u.%u.%u is out of range\n",
+				  se, sh, cu);
+		}
+
+		next = strchr(p, ',');
+		if (!next)
+			break;
+		p = next + 1;
+	}
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
index dc06cbda7..51321e154 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
@@ -27,4 +27,6 @@
 int amdgpu_gfx_scratch_get(struct amdgpu_device *adev, uint32_t *reg);
 void amdgpu_gfx_scratch_free(struct amdgpu_device *adev, uint32_t reg);
 
+unsigned amdgpu_gfx_parse_disable_cu(unsigned *mask, unsigned max_se, unsigned max_sh);
+
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
index 194cfc1a8..ec1282af2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
@@ -33,6 +33,8 @@
 #include "amdgpu.h"
 #include "atom.h"
 
+#define AMDGPU_IB_TEST_TIMEOUT	msecs_to_jiffies(1000)
+
 /*
  * IB
  * IBs (Indirect Buffers) and areas of GPU accessible memory where
@@ -122,7 +124,6 @@ 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;
-	struct fence *hwf;
 	uint64_t ctx;
 
 	unsigned i;
@@ -160,10 +161,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 		patch_offset = amdgpu_ring_init_cond_exec(ring);
 
 	if (vm) {
-		r = amdgpu_vm_flush(ring, job->vm_id, job->vm_pd_addr,
-				    job->gds_base, job->gds_size,
-				    job->gws_base, job->gws_size,
-				    job->oa_base, job->oa_size);
+		r = amdgpu_vm_flush(ring, job);
 		if (r) {
 			amdgpu_ring_undo(ring);
 			return r;
@@ -193,7 +191,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 	if (ring->funcs->emit_hdp_invalidate)
 		amdgpu_ring_emit_hdp_invalidate(ring);
 
-	r = amdgpu_fence_emit(ring, &hwf);
+	r = amdgpu_fence_emit(ring, f);
 	if (r) {
 		dev_err(adev->dev, "failed to emit fence (%d)\n", r);
 		if (job && job->vm_id)
@@ -203,17 +201,11 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 	}
 
 	/* wrap the last IB with fence */
-	if (job && job->uf_bo) {
-		uint64_t addr = amdgpu_bo_gpu_offset(job->uf_bo);
-
-		addr += job->uf_offset;
-		amdgpu_ring_emit_fence(ring, addr, job->uf_sequence,
+	if (job && job->uf_addr) {
+		amdgpu_ring_emit_fence(ring, job->uf_addr, job->uf_sequence,
 				       AMDGPU_FENCE_FLAG_64BIT);
 	}
 
-	if (f)
-		*f = fence_get(hwf);
-
 	if (patch_offset != ~0 && ring->funcs->patch_cond_exec)
 		amdgpu_ring_patch_cond_exec(ring, patch_offset);
 
@@ -296,7 +288,7 @@ int amdgpu_ib_ring_tests(struct amdgpu_device *adev)
 		if (!ring || !ring->ready)
 			continue;
 
-		r = amdgpu_ring_test_ib(ring);
+		r = amdgpu_ring_test_ib(ring, AMDGPU_IB_TEST_TIMEOUT);
 		if (r) {
 			ring->ready = false;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
index 835a3fa8d..278708f5a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
@@ -383,6 +383,18 @@ int amdgpu_irq_update(struct amdgpu_device *adev,
 	return r;
 }
 
+void amdgpu_irq_gpu_reset_resume_helper(struct amdgpu_device *adev)
+{
+	int i, j;
+	for (i = 0; i < AMDGPU_MAX_IRQ_SRC_ID; i++) {
+		struct amdgpu_irq_src *src = adev->irq.sources[i];
+		if (!src)
+			continue;
+		for (j = 0; j < src->num_types; j++)
+			amdgpu_irq_update(adev, src, j);
+	}
+}
+
 /**
  * amdgpu_irq_get - enable interrupt
  *
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
index e124b59f3..7ef09352e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
@@ -94,6 +94,7 @@ int amdgpu_irq_put(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
 		   unsigned type);
 bool amdgpu_irq_enabled(struct amdgpu_device *adev, struct amdgpu_irq_src *src,
 			unsigned type);
+void amdgpu_irq_gpu_reset_resume_helper(struct amdgpu_device *adev);
 
 int amdgpu_irq_add_domain(struct amdgpu_device *adev);
 void amdgpu_irq_remove_domain(struct amdgpu_device *adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
index f0dafa514..6674d40eb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
@@ -28,21 +28,15 @@
 #include "amdgpu.h"
 #include "amdgpu_trace.h"
 
-static void amdgpu_job_free_handler(struct work_struct *ws)
+static void amdgpu_job_timedout(struct amd_sched_job *s_job)
 {
-	struct amdgpu_job *job = container_of(ws, struct amdgpu_job, base.work_free_job);
-	amd_sched_job_put(&job->base);
-}
+	struct amdgpu_job *job = container_of(s_job, struct amdgpu_job, base);
 
-void amdgpu_job_timeout_func(struct work_struct *work)
-{
-	struct amdgpu_job *job = container_of(work, struct amdgpu_job, base.work_tdr.work);
 	DRM_ERROR("ring %s timeout, last signaled seq=%u, last emitted seq=%u\n",
-				job->base.sched->name,
-				(uint32_t)atomic_read(&job->ring->fence_drv.last_seq),
-				job->ring->fence_drv.sync_seq);
-
-	amd_sched_job_put(&job->base);
+		  job->base.sched->name,
+		  atomic_read(&job->ring->fence_drv.last_seq),
+		  job->ring->fence_drv.sync_seq);
+	amdgpu_gpu_reset(job->adev);
 }
 
 int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs,
@@ -63,7 +57,6 @@ int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs,
 	(*job)->vm = vm;
 	(*job)->ibs = (void *)&(*job)[1];
 	(*job)->num_ibs = num_ibs;
-	INIT_WORK(&(*job)->base.work_free_job, amdgpu_job_free_handler);
 
 	amdgpu_sync_create(&(*job)->sync);
 
@@ -86,27 +79,33 @@ int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev, unsigned size,
 	return r;
 }
 
-void amdgpu_job_free(struct amdgpu_job *job)
+void amdgpu_job_free_resources(struct amdgpu_job *job)
 {
-	unsigned i;
 	struct fence *f;
+	unsigned i;
+
 	/* use sched fence if available */
-	f = (job->base.s_fence)? &job->base.s_fence->base : job->fence;
+	f = job->base.s_fence ? &job->base.s_fence->finished : job->fence;
 
 	for (i = 0; i < job->num_ibs; ++i)
-		amdgpu_sa_bo_free(job->adev, &job->ibs[i].sa_bo, f);
-	fence_put(job->fence);
+		amdgpu_ib_free(job->adev, &job->ibs[i], f);
+}
 
-	amdgpu_bo_unref(&job->uf_bo);
-	amdgpu_sync_free(&job->sync);
+void amdgpu_job_free_cb(struct amd_sched_job *s_job)
+{
+	struct amdgpu_job *job = container_of(s_job, struct amdgpu_job, base);
 
-	if (!job->base.use_sched)
-		kfree(job);
+	fence_put(job->fence);
+	amdgpu_sync_free(&job->sync);
+	kfree(job);
 }
 
-void amdgpu_job_free_func(struct kref *refcount)
+void amdgpu_job_free(struct amdgpu_job *job)
 {
-	struct amdgpu_job *job = container_of(refcount, struct amdgpu_job, base.refcount);
+	amdgpu_job_free_resources(job);
+
+	fence_put(job->fence);
+	amdgpu_sync_free(&job->sync);
 	kfree(job);
 }
 
@@ -114,22 +113,20 @@ int amdgpu_job_submit(struct amdgpu_job *job, struct amdgpu_ring *ring,
 		      struct amd_sched_entity *entity, void *owner,
 		      struct fence **f)
 {
-	struct fence *fence;
 	int r;
 	job->ring = ring;
 
 	if (!f)
 		return -EINVAL;
 
-	r = amd_sched_job_init(&job->base, &ring->sched,
-			       entity, amdgpu_job_timeout_func,
-			       amdgpu_job_free_func, owner, &fence);
+	r = amd_sched_job_init(&job->base, &ring->sched, entity, owner);
 	if (r)
 		return r;
 
 	job->owner = owner;
 	job->ctx = entity->fence_context;
-	*f = fence_get(fence);
+	*f = fence_get(&job->base.s_fence->finished);
+	amdgpu_job_free_resources(job);
 	amd_sched_entity_push_job(&job->base);
 
 	return 0;
@@ -147,8 +144,8 @@ static struct fence *amdgpu_job_dependency(struct amd_sched_job *sched_job)
 		int r;
 
 		r = amdgpu_vm_grab_id(vm, ring, &job->sync,
-				      &job->base.s_fence->base,
-				      &job->vm_id, &job->vm_pd_addr);
+				      &job->base.s_fence->finished,
+				      job);
 		if (r)
 			DRM_ERROR("Error getting VM ID (%d)\n", r);
 
@@ -170,29 +167,24 @@ static struct fence *amdgpu_job_run(struct amd_sched_job *sched_job)
 	}
 	job = to_amdgpu_job(sched_job);
 
-	r = amdgpu_sync_wait(&job->sync);
-	if (r) {
-		DRM_ERROR("failed to sync wait (%d)\n", r);
-		return NULL;
-	}
+	BUG_ON(amdgpu_sync_peek_fence(&job->sync, NULL));
 
 	trace_amdgpu_sched_run_job(job);
 	r = amdgpu_ib_schedule(job->ring, job->num_ibs, job->ibs,
 			       job->sync.last_vm_update, job, &fence);
-	if (r) {
+	if (r)
 		DRM_ERROR("Error scheduling IBs (%d)\n", r);
-		goto err;
-	}
 
-err:
-	job->fence = fence;
-	amdgpu_job_free(job);
+	/* if gpu reset, hw fence will be replaced here */
+	fence_put(job->fence);
+	job->fence = fence_get(fence);
+	amdgpu_job_free_resources(job);
 	return fence;
 }
 
 const struct amd_sched_backend_ops amdgpu_sched_ops = {
 	.dependency = amdgpu_job_dependency,
 	.run_job = amdgpu_job_run,
-	.begin_job = amd_sched_job_begin,
-	.finish_job = amd_sched_job_finish,
+	.timedout_job = amdgpu_job_timedout,
+	.free_job = amdgpu_job_free_cb
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index d851ea150..d942654a1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -60,7 +60,10 @@ int amdgpu_driver_unload_kms(struct drm_device *dev)
 	if (adev->rmmio == NULL)
 		goto done_free;
 
-	pm_runtime_get_sync(dev->dev);
+	if (amdgpu_device_is_px(dev)) {
+		pm_runtime_get_sync(dev->dev);
+		pm_runtime_forbid(dev->dev);
+	}
 
 	amdgpu_amdkfd_device_fini(adev);
 
@@ -135,13 +138,75 @@ int amdgpu_driver_load_kms(struct drm_device *dev, unsigned long flags)
 	}
 
 out:
-	if (r)
+	if (r) {
+		/* balance pm_runtime_get_sync in amdgpu_driver_unload_kms */
+		if (adev->rmmio && amdgpu_device_is_px(dev))
+			pm_runtime_put_noidle(dev->dev);
 		amdgpu_driver_unload_kms(dev);
-
+	}
 
 	return r;
 }
 
+static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info,
+				struct drm_amdgpu_query_fw *query_fw,
+				struct amdgpu_device *adev)
+{
+	switch (query_fw->fw_type) {
+	case AMDGPU_INFO_FW_VCE:
+		fw_info->ver = adev->vce.fw_version;
+		fw_info->feature = adev->vce.fb_version;
+		break;
+	case AMDGPU_INFO_FW_UVD:
+		fw_info->ver = adev->uvd.fw_version;
+		fw_info->feature = 0;
+		break;
+	case AMDGPU_INFO_FW_GMC:
+		fw_info->ver = adev->mc.fw_version;
+		fw_info->feature = 0;
+		break;
+	case AMDGPU_INFO_FW_GFX_ME:
+		fw_info->ver = adev->gfx.me_fw_version;
+		fw_info->feature = adev->gfx.me_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_PFP:
+		fw_info->ver = adev->gfx.pfp_fw_version;
+		fw_info->feature = adev->gfx.pfp_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_CE:
+		fw_info->ver = adev->gfx.ce_fw_version;
+		fw_info->feature = adev->gfx.ce_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_RLC:
+		fw_info->ver = adev->gfx.rlc_fw_version;
+		fw_info->feature = adev->gfx.rlc_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_MEC:
+		if (query_fw->index == 0) {
+			fw_info->ver = adev->gfx.mec_fw_version;
+			fw_info->feature = adev->gfx.mec_feature_version;
+		} else if (query_fw->index == 1) {
+			fw_info->ver = adev->gfx.mec2_fw_version;
+			fw_info->feature = adev->gfx.mec2_feature_version;
+		} else
+			return -EINVAL;
+		break;
+	case AMDGPU_INFO_FW_SMC:
+		fw_info->ver = adev->pm.fw_version;
+		fw_info->feature = 0;
+		break;
+	case AMDGPU_INFO_FW_SDMA:
+		if (query_fw->index >= adev->sdma.num_instances)
+			return -EINVAL;
+		fw_info->ver = adev->sdma.instance[query_fw->index].fw_version;
+		fw_info->feature = adev->sdma.instance[query_fw->index].feature_version;
+		break;
+	default:
+		return -EINVAL;
+	}
+	return 0;
+}
+
 /*
  * Userspace get information ioctl
  */
@@ -288,67 +353,20 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 		return copy_to_user(out, &count, min(size, 4u)) ? -EFAULT : 0;
 	}
 	case AMDGPU_INFO_TIMESTAMP:
-		ui64 = amdgpu_asic_get_gpu_clock_counter(adev);
+		ui64 = amdgpu_gfx_get_gpu_clock_counter(adev);
 		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
 	case AMDGPU_INFO_FW_VERSION: {
 		struct drm_amdgpu_info_firmware fw_info;
+		int ret;
 
 		/* We only support one instance of each IP block right now. */
 		if (info->query_fw.ip_instance != 0)
 			return -EINVAL;
 
-		switch (info->query_fw.fw_type) {
-		case AMDGPU_INFO_FW_VCE:
-			fw_info.ver = adev->vce.fw_version;
-			fw_info.feature = adev->vce.fb_version;
-			break;
-		case AMDGPU_INFO_FW_UVD:
-			fw_info.ver = adev->uvd.fw_version;
-			fw_info.feature = 0;
-			break;
-		case AMDGPU_INFO_FW_GMC:
-			fw_info.ver = adev->mc.fw_version;
-			fw_info.feature = 0;
-			break;
-		case AMDGPU_INFO_FW_GFX_ME:
-			fw_info.ver = adev->gfx.me_fw_version;
-			fw_info.feature = adev->gfx.me_feature_version;
-			break;
-		case AMDGPU_INFO_FW_GFX_PFP:
-			fw_info.ver = adev->gfx.pfp_fw_version;
-			fw_info.feature = adev->gfx.pfp_feature_version;
-			break;
-		case AMDGPU_INFO_FW_GFX_CE:
-			fw_info.ver = adev->gfx.ce_fw_version;
-			fw_info.feature = adev->gfx.ce_feature_version;
-			break;
-		case AMDGPU_INFO_FW_GFX_RLC:
-			fw_info.ver = adev->gfx.rlc_fw_version;
-			fw_info.feature = adev->gfx.rlc_feature_version;
-			break;
-		case AMDGPU_INFO_FW_GFX_MEC:
-			if (info->query_fw.index == 0) {
-				fw_info.ver = adev->gfx.mec_fw_version;
-				fw_info.feature = adev->gfx.mec_feature_version;
-			} else if (info->query_fw.index == 1) {
-				fw_info.ver = adev->gfx.mec2_fw_version;
-				fw_info.feature = adev->gfx.mec2_feature_version;
-			} else
-				return -EINVAL;
-			break;
-		case AMDGPU_INFO_FW_SMC:
-			fw_info.ver = adev->pm.fw_version;
-			fw_info.feature = 0;
-			break;
-		case AMDGPU_INFO_FW_SDMA:
-			if (info->query_fw.index >= adev->sdma.num_instances)
-				return -EINVAL;
-			fw_info.ver = adev->sdma.instance[info->query_fw.index].fw_version;
-			fw_info.feature = adev->sdma.instance[info->query_fw.index].feature_version;
-			break;
-		default:
-			return -EINVAL;
-		}
+		ret = amdgpu_firmware_info(&fw_info, &info->query_fw, adev);
+		if (ret)
+			return ret;
+
 		return copy_to_user(out, &fw_info,
 				    min((size_t)size, sizeof(fw_info))) ? -EFAULT : 0;
 	}
@@ -566,6 +584,9 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev,
 
 	amdgpu_ctx_mgr_fini(&fpriv->ctx_mgr);
 
+	amdgpu_uvd_free_handles(adev, file_priv);
+	amdgpu_vce_free_handles(adev, file_priv);
+
 	amdgpu_vm_fini(adev, &fpriv->vm);
 
 	idr_for_each_entry(&fpriv->bo_list_handles, list, handle)
@@ -590,10 +611,6 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev,
 void amdgpu_driver_preclose_kms(struct drm_device *dev,
 				struct drm_file *file_priv)
 {
-	struct amdgpu_device *adev = dev->dev_private;
-
-	amdgpu_uvd_free_handles(adev, file_priv);
-	amdgpu_vce_free_handles(adev, file_priv);
 }
 
 /*
@@ -756,3 +773,130 @@ const struct drm_ioctl_desc amdgpu_ioctls_kms[] = {
 	DRM_IOCTL_DEF_DRV(AMDGPU_GEM_USERPTR, amdgpu_gem_userptr_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
 };
 const int amdgpu_max_kms_ioctl = ARRAY_SIZE(amdgpu_ioctls_kms);
+
+/*
+ * Debugfs info
+ */
+#if defined(CONFIG_DEBUG_FS)
+
+static int amdgpu_debugfs_firmware_info(struct seq_file *m, void *data)
+{
+	struct drm_info_node *node = (struct drm_info_node *) m->private;
+	struct drm_device *dev = node->minor->dev;
+	struct amdgpu_device *adev = dev->dev_private;
+	struct drm_amdgpu_info_firmware fw_info;
+	struct drm_amdgpu_query_fw query_fw;
+	int ret, i;
+
+	/* VCE */
+	query_fw.fw_type = AMDGPU_INFO_FW_VCE;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "VCE feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* UVD */
+	query_fw.fw_type = AMDGPU_INFO_FW_UVD;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "UVD feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* GMC */
+	query_fw.fw_type = AMDGPU_INFO_FW_GMC;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "MC feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* ME */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_ME;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "ME feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* PFP */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_PFP;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "PFP feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* CE */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_CE;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "CE feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* RLC */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_RLC;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "RLC feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* MEC */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_MEC;
+	query_fw.index = 0;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "MEC feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* MEC2 */
+	if (adev->asic_type == CHIP_KAVERI ||
+	    (adev->asic_type > CHIP_TOPAZ && adev->asic_type != CHIP_STONEY)) {
+		query_fw.index = 1;
+		ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+		if (ret)
+			return ret;
+		seq_printf(m, "MEC2 feature version: %u, firmware version: 0x%08x\n",
+			   fw_info.feature, fw_info.ver);
+	}
+
+	/* SMC */
+	query_fw.fw_type = AMDGPU_INFO_FW_SMC;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "SMC feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* SDMA */
+	query_fw.fw_type = AMDGPU_INFO_FW_SDMA;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		query_fw.index = i;
+		ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+		if (ret)
+			return ret;
+		seq_printf(m, "SDMA%d feature version: %u, firmware version: 0x%08x\n",
+			   i, fw_info.feature, fw_info.ver);
+	}
+
+	return 0;
+}
+
+static const struct drm_info_list amdgpu_firmware_info_list[] = {
+	{"amdgpu_firmware_info", amdgpu_debugfs_firmware_info, 0, NULL},
+};
+#endif
+
+int amdgpu_debugfs_firmware_init(struct amdgpu_device *adev)
+{
+#if defined(CONFIG_DEBUG_FS)
+	return amdgpu_debugfs_add_files(adev, amdgpu_firmware_info_list,
+					ARRAY_SIZE(amdgpu_firmware_info_list));
+#else
+	return 0;
+#endif
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index 7ecea83ce..6f0873c75 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -589,6 +589,7 @@ void amdgpu_bo_move_notify(struct ttm_buffer_object *bo,
 			   struct ttm_mem_reg *new_mem)
 {
 	struct amdgpu_bo *rbo;
+	struct ttm_mem_reg *old_mem = &bo->mem;
 
 	if (!amdgpu_ttm_bo_is_amdgpu_bo(bo))
 		return;
@@ -602,6 +603,8 @@ void amdgpu_bo_move_notify(struct ttm_buffer_object *bo,
 
 	/* move_notify is called before move happens */
 	amdgpu_update_memory_usage(rbo->adev, &bo->mem, new_mem);
+
+	trace_amdgpu_ttm_bo_move(rbo, new_mem->mem_type, old_mem->mem_type);
 }
 
 int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
index 0e13d80d2..5cc7052e3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
@@ -305,7 +305,7 @@ static ssize_t amdgpu_get_pp_table(struct device *dev,
 	struct drm_device *ddev = dev_get_drvdata(dev);
 	struct amdgpu_device *adev = ddev->dev_private;
 	char *table = NULL;
-	int size, i;
+	int size;
 
 	if (adev->pp_enabled)
 		size = amdgpu_dpm_get_pp_table(adev, &table);
@@ -315,10 +315,7 @@ static ssize_t amdgpu_get_pp_table(struct device *dev,
 	if (size >= PAGE_SIZE)
 		size = PAGE_SIZE - 1;
 
-	for (i = 0; i < size; i++) {
-		sprintf(buf + i, "%02x", table[i]);
-	}
-	sprintf(buf + i, "\n");
+	memcpy(buf, table, size);
 
 	return size;
 }
@@ -347,6 +344,8 @@ static ssize_t amdgpu_get_pp_dpm_sclk(struct device *dev,
 
 	if (adev->pp_enabled)
 		size = amdgpu_dpm_print_clock_levels(adev, PP_SCLK, buf);
+	else if (adev->pm.funcs->print_clock_levels)
+		size = adev->pm.funcs->print_clock_levels(adev, PP_SCLK, buf);
 
 	return size;
 }
@@ -363,7 +362,9 @@ static ssize_t amdgpu_set_pp_dpm_sclk(struct device *dev,
 	uint32_t i, mask = 0;
 	char sub_str[2];
 
-	for (i = 0; i < strlen(buf) - 1; i++) {
+	for (i = 0; i < strlen(buf); i++) {
+		if (*(buf + i) == '\n')
+			continue;
 		sub_str[0] = *(buf + i);
 		sub_str[1] = '\0';
 		ret = kstrtol(sub_str, 0, &level);
@@ -377,6 +378,8 @@ static ssize_t amdgpu_set_pp_dpm_sclk(struct device *dev,
 
 	if (adev->pp_enabled)
 		amdgpu_dpm_force_clock_level(adev, PP_SCLK, mask);
+	else if (adev->pm.funcs->force_clock_level)
+		adev->pm.funcs->force_clock_level(adev, PP_SCLK, mask);
 fail:
 	return count;
 }
@@ -391,6 +394,8 @@ static ssize_t amdgpu_get_pp_dpm_mclk(struct device *dev,
 
 	if (adev->pp_enabled)
 		size = amdgpu_dpm_print_clock_levels(adev, PP_MCLK, buf);
+	else if (adev->pm.funcs->print_clock_levels)
+		size = adev->pm.funcs->print_clock_levels(adev, PP_MCLK, buf);
 
 	return size;
 }
@@ -407,7 +412,9 @@ static ssize_t amdgpu_set_pp_dpm_mclk(struct device *dev,
 	uint32_t i, mask = 0;
 	char sub_str[2];
 
-	for (i = 0; i < strlen(buf) - 1; i++) {
+	for (i = 0; i < strlen(buf); i++) {
+		if (*(buf + i) == '\n')
+			continue;
 		sub_str[0] = *(buf + i);
 		sub_str[1] = '\0';
 		ret = kstrtol(sub_str, 0, &level);
@@ -421,6 +428,8 @@ static ssize_t amdgpu_set_pp_dpm_mclk(struct device *dev,
 
 	if (adev->pp_enabled)
 		amdgpu_dpm_force_clock_level(adev, PP_MCLK, mask);
+	else if (adev->pm.funcs->force_clock_level)
+		adev->pm.funcs->force_clock_level(adev, PP_MCLK, mask);
 fail:
 	return count;
 }
@@ -435,6 +444,8 @@ static ssize_t amdgpu_get_pp_dpm_pcie(struct device *dev,
 
 	if (adev->pp_enabled)
 		size = amdgpu_dpm_print_clock_levels(adev, PP_PCIE, buf);
+	else if (adev->pm.funcs->print_clock_levels)
+		size = adev->pm.funcs->print_clock_levels(adev, PP_PCIE, buf);
 
 	return size;
 }
@@ -451,7 +462,9 @@ static ssize_t amdgpu_set_pp_dpm_pcie(struct device *dev,
 	uint32_t i, mask = 0;
 	char sub_str[2];
 
-	for (i = 0; i < strlen(buf) - 1; i++) {
+	for (i = 0; i < strlen(buf); i++) {
+		if (*(buf + i) == '\n')
+			continue;
 		sub_str[0] = *(buf + i);
 		sub_str[1] = '\0';
 		ret = kstrtol(sub_str, 0, &level);
@@ -465,6 +478,100 @@ static ssize_t amdgpu_set_pp_dpm_pcie(struct device *dev,
 
 	if (adev->pp_enabled)
 		amdgpu_dpm_force_clock_level(adev, PP_PCIE, mask);
+	else if (adev->pm.funcs->force_clock_level)
+		adev->pm.funcs->force_clock_level(adev, PP_PCIE, mask);
+fail:
+	return count;
+}
+
+static ssize_t amdgpu_get_pp_sclk_od(struct device *dev,
+		struct device_attribute *attr,
+		char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	uint32_t value = 0;
+
+	if (adev->pp_enabled)
+		value = amdgpu_dpm_get_sclk_od(adev);
+	else if (adev->pm.funcs->get_sclk_od)
+		value = adev->pm.funcs->get_sclk_od(adev);
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", value);
+}
+
+static ssize_t amdgpu_set_pp_sclk_od(struct device *dev,
+		struct device_attribute *attr,
+		const char *buf,
+		size_t count)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	int ret;
+	long int value;
+
+	ret = kstrtol(buf, 0, &value);
+
+	if (ret) {
+		count = -EINVAL;
+		goto fail;
+	}
+
+	if (adev->pp_enabled) {
+		amdgpu_dpm_set_sclk_od(adev, (uint32_t)value);
+		amdgpu_dpm_dispatch_task(adev, AMD_PP_EVENT_READJUST_POWER_STATE, NULL, NULL);
+	} else if (adev->pm.funcs->set_sclk_od) {
+		adev->pm.funcs->set_sclk_od(adev, (uint32_t)value);
+		adev->pm.dpm.current_ps = adev->pm.dpm.boot_ps;
+		amdgpu_pm_compute_clocks(adev);
+	}
+
+fail:
+	return count;
+}
+
+static ssize_t amdgpu_get_pp_mclk_od(struct device *dev,
+		struct device_attribute *attr,
+		char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	uint32_t value = 0;
+
+	if (adev->pp_enabled)
+		value = amdgpu_dpm_get_mclk_od(adev);
+	else if (adev->pm.funcs->get_mclk_od)
+		value = adev->pm.funcs->get_mclk_od(adev);
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", value);
+}
+
+static ssize_t amdgpu_set_pp_mclk_od(struct device *dev,
+		struct device_attribute *attr,
+		const char *buf,
+		size_t count)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = ddev->dev_private;
+	int ret;
+	long int value;
+
+	ret = kstrtol(buf, 0, &value);
+
+	if (ret) {
+		count = -EINVAL;
+		goto fail;
+	}
+
+	if (adev->pp_enabled) {
+		amdgpu_dpm_set_mclk_od(adev, (uint32_t)value);
+		amdgpu_dpm_dispatch_task(adev, AMD_PP_EVENT_READJUST_POWER_STATE, NULL, NULL);
+	} else if (adev->pm.funcs->set_mclk_od) {
+		adev->pm.funcs->set_mclk_od(adev, (uint32_t)value);
+		adev->pm.dpm.current_ps = adev->pm.dpm.boot_ps;
+		amdgpu_pm_compute_clocks(adev);
+	}
+
 fail:
 	return count;
 }
@@ -490,6 +597,12 @@ static DEVICE_ATTR(pp_dpm_mclk, S_IRUGO | S_IWUSR,
 static DEVICE_ATTR(pp_dpm_pcie, S_IRUGO | S_IWUSR,
 		amdgpu_get_pp_dpm_pcie,
 		amdgpu_set_pp_dpm_pcie);
+static DEVICE_ATTR(pp_sclk_od, S_IRUGO | S_IWUSR,
+		amdgpu_get_pp_sclk_od,
+		amdgpu_set_pp_sclk_od);
+static DEVICE_ATTR(pp_mclk_od, S_IRUGO | S_IWUSR,
+		amdgpu_get_pp_mclk_od,
+		amdgpu_set_pp_mclk_od);
 
 static ssize_t amdgpu_hwmon_show_temp(struct device *dev,
 				      struct device_attribute *attr,
@@ -1108,22 +1221,34 @@ int amdgpu_pm_sysfs_init(struct amdgpu_device *adev)
 			DRM_ERROR("failed to create device file pp_table\n");
 			return ret;
 		}
-		ret = device_create_file(adev->dev, &dev_attr_pp_dpm_sclk);
-		if (ret) {
-			DRM_ERROR("failed to create device file pp_dpm_sclk\n");
-			return ret;
-		}
-		ret = device_create_file(adev->dev, &dev_attr_pp_dpm_mclk);
-		if (ret) {
-			DRM_ERROR("failed to create device file pp_dpm_mclk\n");
-			return ret;
-		}
-		ret = device_create_file(adev->dev, &dev_attr_pp_dpm_pcie);
-		if (ret) {
-			DRM_ERROR("failed to create device file pp_dpm_pcie\n");
-			return ret;
-		}
 	}
+
+	ret = device_create_file(adev->dev, &dev_attr_pp_dpm_sclk);
+	if (ret) {
+		DRM_ERROR("failed to create device file pp_dpm_sclk\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev, &dev_attr_pp_dpm_mclk);
+	if (ret) {
+		DRM_ERROR("failed to create device file pp_dpm_mclk\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev, &dev_attr_pp_dpm_pcie);
+	if (ret) {
+		DRM_ERROR("failed to create device file pp_dpm_pcie\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev, &dev_attr_pp_sclk_od);
+	if (ret) {
+		DRM_ERROR("failed to create device file pp_sclk_od\n");
+		return ret;
+	}
+	ret = device_create_file(adev->dev, &dev_attr_pp_mclk_od);
+	if (ret) {
+		DRM_ERROR("failed to create device file pp_mclk_od\n");
+		return ret;
+	}
+
 	ret = amdgpu_debugfs_pm_init(adev);
 	if (ret) {
 		DRM_ERROR("Failed to register debugfs file for dpm!\n");
@@ -1146,10 +1271,12 @@ void amdgpu_pm_sysfs_fini(struct amdgpu_device *adev)
 		device_remove_file(adev->dev, &dev_attr_pp_cur_state);
 		device_remove_file(adev->dev, &dev_attr_pp_force_state);
 		device_remove_file(adev->dev, &dev_attr_pp_table);
-		device_remove_file(adev->dev, &dev_attr_pp_dpm_sclk);
-		device_remove_file(adev->dev, &dev_attr_pp_dpm_mclk);
-		device_remove_file(adev->dev, &dev_attr_pp_dpm_pcie);
 	}
+	device_remove_file(adev->dev, &dev_attr_pp_dpm_sclk);
+	device_remove_file(adev->dev, &dev_attr_pp_dpm_mclk);
+	device_remove_file(adev->dev, &dev_attr_pp_dpm_pcie);
+	device_remove_file(adev->dev, &dev_attr_pp_sclk_od);
+	device_remove_file(adev->dev, &dev_attr_pp_mclk_od);
 }
 
 void amdgpu_pm_compute_clocks(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c
index 82256558e..c5738a22b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c
@@ -52,6 +52,7 @@ 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->powercontainment_enabled = amdgpu_powercontainment;
 
 		ret = amd_powerplay_init(pp_init, amd_pp);
 		kfree(pp_init);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
index 870f94942..85aeb0a80 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
@@ -28,6 +28,7 @@
  */
 #include <linux/seq_file.h>
 #include <linux/slab.h>
+#include <linux/debugfs.h>
 #include <drm/drmP.h>
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
@@ -48,6 +49,7 @@
  */
 static int amdgpu_debugfs_ring_init(struct amdgpu_device *adev,
 				    struct amdgpu_ring *ring);
+static void amdgpu_debugfs_ring_fini(struct amdgpu_ring *ring);
 
 /**
  * amdgpu_ring_alloc - allocate space on the ring buffer
@@ -73,6 +75,10 @@ int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw)
 
 	ring->count_dw = ndw;
 	ring->wptr_old = ring->wptr;
+
+	if (ring->funcs->begin_use)
+		ring->funcs->begin_use(ring);
+
 	return 0;
 }
 
@@ -125,6 +131,9 @@ void amdgpu_ring_commit(struct amdgpu_ring *ring)
 
 	mb();
 	amdgpu_ring_set_wptr(ring);
+
+	if (ring->funcs->end_use)
+		ring->funcs->end_use(ring);
 }
 
 /**
@@ -137,78 +146,9 @@ void amdgpu_ring_commit(struct amdgpu_ring *ring)
 void amdgpu_ring_undo(struct amdgpu_ring *ring)
 {
 	ring->wptr = ring->wptr_old;
-}
-
-/**
- * amdgpu_ring_backup - Back up the content of a ring
- *
- * @ring: the ring we want to back up
- *
- * Saves all unprocessed commits from a ring, returns the number of dwords saved.
- */
-unsigned amdgpu_ring_backup(struct amdgpu_ring *ring,
-			    uint32_t **data)
-{
-	unsigned size, ptr, i;
-
-	*data = NULL;
-
-	if (ring->ring_obj == NULL)
-		return 0;
-
-	/* it doesn't make sense to save anything if all fences are signaled */
-	if (!amdgpu_fence_count_emitted(ring))
-		return 0;
-
-	ptr = le32_to_cpu(*ring->next_rptr_cpu_addr);
-
-	size = ring->wptr + (ring->ring_size / 4);
-	size -= ptr;
-	size &= ring->ptr_mask;
-	if (size == 0)
-		return 0;
-
-	/* and then save the content of the ring */
-	*data = kmalloc_array(size, sizeof(uint32_t), GFP_KERNEL);
-	if (!*data)
-		return 0;
-	for (i = 0; i < size; ++i) {
-		(*data)[i] = ring->ring[ptr++];
-		ptr &= ring->ptr_mask;
-	}
-
-	return size;
-}
-
-/**
- * amdgpu_ring_restore - append saved commands to the ring again
- *
- * @ring: ring to append commands to
- * @size: number of dwords we want to write
- * @data: saved commands
- *
- * Allocates space on the ring and restore the previously saved commands.
- */
-int amdgpu_ring_restore(struct amdgpu_ring *ring,
-			unsigned size, uint32_t *data)
-{
-	int i, r;
-
-	if (!size || !data)
-		return 0;
-
-	/* restore the saved ring content */
-	r = amdgpu_ring_alloc(ring, size);
-	if (r)
-		return r;
-
-	for (i = 0; i < size; ++i) {
-		amdgpu_ring_write(ring, data[i]);
-	}
 
-	amdgpu_ring_commit(ring);
-	kfree(data);
-	return 0;
+	if (ring->funcs->end_use)
+		ring->funcs->end_use(ring);
 }
 
 /**
@@ -260,14 +200,6 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
 		return r;
 	}
 
-	r = amdgpu_wb_get(adev, &ring->next_rptr_offs);
-	if (r) {
-		dev_err(adev->dev, "(%d) ring next_rptr wb alloc failed\n", r);
-		return r;
-	}
-	ring->next_rptr_gpu_addr = adev->wb.gpu_addr + ring->next_rptr_offs * 4;
-	ring->next_rptr_cpu_addr = &adev->wb.wb[ring->next_rptr_offs];
-
 	r = amdgpu_wb_get(adev, &ring->cond_exe_offs);
 	if (r) {
 		dev_err(adev->dev, "(%d) ring cond_exec_polling wb alloc failed\n", r);
@@ -276,7 +208,6 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
 	ring->cond_exe_gpu_addr = adev->wb.gpu_addr + (ring->cond_exe_offs * 4);
 	ring->cond_exe_cpu_addr = &adev->wb.wb[ring->cond_exe_offs];
 
-	spin_lock_init(&ring->fence_lock);
 	r = amdgpu_fence_driver_start_ring(ring, irq_src, irq_type);
 	if (r) {
 		dev_err(adev->dev, "failed initializing fences (%d).\n", r);
@@ -310,6 +241,9 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
 		}
 		r = amdgpu_bo_kmap(ring->ring_obj,
 				       (void **)&ring->ring);
+
+		memset((void *)ring->ring, 0, ring->ring_size);
+
 		amdgpu_bo_unreserve(ring->ring_obj);
 		if (r) {
 			dev_err(adev->dev, "(%d) ring map failed\n", r);
@@ -347,7 +281,6 @@ void amdgpu_ring_fini(struct amdgpu_ring *ring)
 	amdgpu_wb_free(ring->adev, ring->fence_offs);
 	amdgpu_wb_free(ring->adev, ring->rptr_offs);
 	amdgpu_wb_free(ring->adev, ring->wptr_offs);
-	amdgpu_wb_free(ring->adev, ring->next_rptr_offs);
 
 	if (ring_obj) {
 		r = amdgpu_bo_reserve(ring_obj, false);
@@ -358,6 +291,7 @@ void amdgpu_ring_fini(struct amdgpu_ring *ring)
 		}
 		amdgpu_bo_unref(&ring_obj);
 	}
+	amdgpu_debugfs_ring_fini(ring);
 }
 
 /*
@@ -365,57 +299,62 @@ void amdgpu_ring_fini(struct amdgpu_ring *ring)
  */
 #if defined(CONFIG_DEBUG_FS)
 
-static int amdgpu_debugfs_ring_info(struct seq_file *m, void *data)
+/* Layout of file is 12 bytes consisting of
+ * - rptr
+ * - wptr
+ * - driver's copy of wptr
+ *
+ * followed by n-words of ring data
+ */
+static ssize_t amdgpu_debugfs_ring_read(struct file *f, char __user *buf,
+					size_t size, loff_t *pos)
 {
-	struct drm_info_node *node = (struct drm_info_node *) m->private;
-	struct drm_device *dev = node->minor->dev;
-	struct amdgpu_device *adev = dev->dev_private;
-	int roffset = (unsigned long)node->info_ent->data;
-	struct amdgpu_ring *ring = (void *)(((uint8_t*)adev) + roffset);
-	uint32_t rptr, wptr, rptr_next;
-	unsigned i;
-
-	wptr = amdgpu_ring_get_wptr(ring);
-	seq_printf(m, "wptr: 0x%08x [%5d]\n", wptr, wptr);
-
-	rptr = amdgpu_ring_get_rptr(ring);
-	rptr_next = le32_to_cpu(*ring->next_rptr_cpu_addr);
-
-	seq_printf(m, "rptr: 0x%08x [%5d]\n", rptr, rptr);
-
-	seq_printf(m, "driver's copy of the wptr: 0x%08x [%5d]\n",
-		   ring->wptr, ring->wptr);
-
-	if (!ring->ready)
-		return 0;
-
-	/* print 8 dw before current rptr as often it's the last executed
-	 * packet that is the root issue
-	 */
-	i = (rptr + ring->ptr_mask + 1 - 32) & ring->ptr_mask;
-	while (i != rptr) {
-		seq_printf(m, "r[%5d]=0x%08x", i, ring->ring[i]);
-		if (i == rptr)
-			seq_puts(m, " *");
-		if (i == rptr_next)
-			seq_puts(m, " #");
-		seq_puts(m, "\n");
-		i = (i + 1) & ring->ptr_mask;
+	struct amdgpu_ring *ring = (struct amdgpu_ring*)f->f_inode->i_private;
+	int r, i;
+	uint32_t value, result, early[3];
+
+	if (*pos & 3 || size & 3)
+		return -EINVAL;
+
+	result = 0;
+
+	if (*pos < 12) {
+		early[0] = amdgpu_ring_get_rptr(ring);
+		early[1] = amdgpu_ring_get_wptr(ring);
+		early[2] = ring->wptr;
+		for (i = *pos / 4; i < 3 && size; i++) {
+			r = put_user(early[i], (uint32_t *)buf);
+			if (r)
+				return r;
+			buf += 4;
+			result += 4;
+			size -= 4;
+			*pos += 4;
+		}
 	}
-	while (i != wptr) {
-		seq_printf(m, "r[%5d]=0x%08x", i, ring->ring[i]);
-		if (i == rptr)
-			seq_puts(m, " *");
-		if (i == rptr_next)
-			seq_puts(m, " #");
-		seq_puts(m, "\n");
-		i = (i + 1) & ring->ptr_mask;
+
+	while (size) {
+		if (*pos >= (ring->ring_size + 12))
+			return result;
+			
+		value = ring->ring[(*pos - 12)/4];
+		r = put_user(value, (uint32_t*)buf);
+		if (r)
+			return r;
+		buf += 4;
+		result += 4;
+		size -= 4;
+		*pos += 4;
 	}
-	return 0;
+
+	return result;
 }
 
-static struct drm_info_list amdgpu_debugfs_ring_info_list[AMDGPU_MAX_RINGS];
-static char amdgpu_debugfs_ring_names[AMDGPU_MAX_RINGS][32];
+static const struct file_operations amdgpu_debugfs_ring_fops = {
+	.owner = THIS_MODULE,
+	.read = amdgpu_debugfs_ring_read,
+	.llseek = default_llseek
+};
 
 #endif
 
@@ -423,28 +362,27 @@ static int amdgpu_debugfs_ring_init(struct amdgpu_device *adev,
 				    struct amdgpu_ring *ring)
 {
 #if defined(CONFIG_DEBUG_FS)
-	unsigned offset = (uint8_t*)ring - (uint8_t*)adev;
-	unsigned i;
-	struct drm_info_list *info;
-	char *name;
-
-	for (i = 0; i < ARRAY_SIZE(amdgpu_debugfs_ring_info_list); ++i) {
-		info = &amdgpu_debugfs_ring_info_list[i];
-		if (!info->data)
-			break;
-	}
+	struct drm_minor *minor = adev->ddev->primary;
+	struct dentry *ent, *root = minor->debugfs_root;
+	char name[32];
 
-	if (i == ARRAY_SIZE(amdgpu_debugfs_ring_info_list))
-		return -ENOSPC;
-
-	name = &amdgpu_debugfs_ring_names[i][0];
 	sprintf(name, "amdgpu_ring_%s", ring->name);
-	info->name = name;
-	info->show = amdgpu_debugfs_ring_info;
-	info->driver_features = 0;
-	info->data = (void*)(uintptr_t)offset;
 
-	return amdgpu_debugfs_add_files(adev, info, 1);
+	ent = debugfs_create_file(name,
+				  S_IFREG | S_IRUGO, root,
+				  ring, &amdgpu_debugfs_ring_fops);
+	if (IS_ERR(ent))
+		return PTR_ERR(ent);
+
+	i_size_write(ent->d_inode, ring->ring_size + 12);
+	ring->ent = ent;
 #endif
 	return 0;
 }
+
+static void amdgpu_debugfs_ring_fini(struct amdgpu_ring *ring)
+{
+#if defined(CONFIG_DEBUG_FS)
+	debugfs_remove(ring->ent);
+#endif
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
index 48618ee32..d8af37a84 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
@@ -428,7 +428,7 @@ void amdgpu_sa_bo_dump_debug_info(struct amdgpu_sa_manager *sa_manager,
 			   soffset, eoffset, eoffset - soffset);
 
 		if (i->fence)
-			seq_printf(m, " protected by 0x%08x on context %d",
+			seq_printf(m, " protected by 0x%08x on context %llu",
 				   i->fence->seqno, i->fence->context);
 
 		seq_printf(m, "\n");
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c
index 34a92808b..5c8d3022f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c
@@ -223,13 +223,16 @@ int amdgpu_sync_resv(struct amdgpu_device *adev,
 }
 
 /**
- * amdgpu_sync_is_idle - test if all fences are signaled
+ * amdgpu_sync_peek_fence - get the next fence not signaled yet
  *
  * @sync: the sync object
+ * @ring: optional ring to use for test
  *
- * Returns true if all fences in the sync object are signaled.
+ * Returns the next fence not signaled yet without removing it from the sync
+ * object.
  */
-bool amdgpu_sync_is_idle(struct amdgpu_sync *sync)
+struct fence *amdgpu_sync_peek_fence(struct amdgpu_sync *sync,
+				     struct amdgpu_ring *ring)
 {
 	struct amdgpu_sync_entry *e;
 	struct hlist_node *tmp;
@@ -237,6 +240,19 @@ bool amdgpu_sync_is_idle(struct amdgpu_sync *sync)
 
 	hash_for_each_safe(sync->fences, i, tmp, e, node) {
 		struct fence *f = e->fence;
+		struct amd_sched_fence *s_fence = to_amd_sched_fence(f);
+
+		if (ring && s_fence) {
+			/* For fences from the same ring it is sufficient
+			 * when they are scheduled.
+			 */
+			if (s_fence->sched == &ring->sched) {
+				if (fence_is_signaled(&s_fence->scheduled))
+					continue;
+
+				return &s_fence->scheduled;
+			}
+		}
 
 		if (fence_is_signaled(f)) {
 			hash_del(&e->node);
@@ -245,58 +261,19 @@ bool amdgpu_sync_is_idle(struct amdgpu_sync *sync)
 			continue;
 		}
 
-		return false;
+		return f;
 	}
 
-	return true;
+	return NULL;
 }
 
 /**
- * amdgpu_sync_cycle_fences - move fences from one sync object into another
+ * amdgpu_sync_get_fence - get the next fence from the sync object
  *
- * @dst: the destination sync object
- * @src: the source sync object
- * @fence: fence to add to source
+ * @sync: sync object to use
  *
- * Remove all fences from source and put them into destination and add
- * fence as new one into source.
+ * Get and removes the next fence from the sync object not signaled yet.
  */
-int amdgpu_sync_cycle_fences(struct amdgpu_sync *dst, struct amdgpu_sync *src,
-			     struct fence *fence)
-{
-	struct amdgpu_sync_entry *e, *newone;
-	struct hlist_node *tmp;
-	int i;
-
-	/* Allocate the new entry before moving the old ones */
-	newone = kmem_cache_alloc(amdgpu_sync_slab, GFP_KERNEL);
-	if (!newone)
-		return -ENOMEM;
-
-	hash_for_each_safe(src->fences, i, tmp, e, node) {
-		struct fence *f = e->fence;
-
-		hash_del(&e->node);
-		if (fence_is_signaled(f)) {
-			fence_put(f);
-			kmem_cache_free(amdgpu_sync_slab, e);
-			continue;
-		}
-
-		if (amdgpu_sync_add_later(dst, f)) {
-			kmem_cache_free(amdgpu_sync_slab, e);
-			continue;
-		}
-
-		hash_add(dst->fences, &e->node, f->context);
-	}
-
-	hash_add(src->fences, &newone->node, fence->context);
-	newone->fence = fence_get(fence);
-
-	return 0;
-}
-
 struct fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync)
 {
 	struct amdgpu_sync_entry *e;
@@ -319,25 +296,6 @@ struct fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync)
 	return NULL;
 }
 
-int amdgpu_sync_wait(struct amdgpu_sync *sync)
-{
-	struct amdgpu_sync_entry *e;
-	struct hlist_node *tmp;
-	int i, r;
-
-	hash_for_each_safe(sync->fences, i, tmp, e, node) {
-		r = fence_wait(e->fence, false);
-		if (r)
-			return r;
-
-		hash_del(&e->node);
-		fence_put(e->fence);
-		kmem_cache_free(amdgpu_sync_slab, e);
-	}
-
-	return 0;
-}
-
 /**
  * amdgpu_sync_free - free the sync object
  *
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
index 26a5f4acf..0d8d65eb4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
@@ -11,19 +11,68 @@
 #define TRACE_SYSTEM amdgpu
 #define TRACE_INCLUDE_FILE amdgpu_trace
 
+TRACE_EVENT(amdgpu_mm_rreg,
+	    TP_PROTO(unsigned did, uint32_t reg, uint32_t value),
+	    TP_ARGS(did, reg, value),
+	    TP_STRUCT__entry(
+				__field(unsigned, did)
+				__field(uint32_t, reg)
+				__field(uint32_t, value)
+			    ),
+	    TP_fast_assign(
+			   __entry->did = did;
+			   __entry->reg = reg;
+			   __entry->value = value;
+			   ),
+	    TP_printk("0x%04lx, 0x%04lx, 0x%08lx",
+		      (unsigned long)__entry->did,
+		      (unsigned long)__entry->reg,
+		      (unsigned long)__entry->value)
+);
+
+TRACE_EVENT(amdgpu_mm_wreg,
+	    TP_PROTO(unsigned did, uint32_t reg, uint32_t value),
+	    TP_ARGS(did, reg, value),
+	    TP_STRUCT__entry(
+				__field(unsigned, did)
+				__field(uint32_t, reg)
+				__field(uint32_t, value)
+			    ),
+	    TP_fast_assign(
+			   __entry->did = did;
+			   __entry->reg = reg;
+			   __entry->value = value;
+			   ),
+	    TP_printk("0x%04lx, 0x%04lx, 0x%08lx",
+		      (unsigned long)__entry->did,
+		      (unsigned long)__entry->reg,
+		      (unsigned long)__entry->value)
+);
+
 TRACE_EVENT(amdgpu_bo_create,
 	    TP_PROTO(struct amdgpu_bo *bo),
 	    TP_ARGS(bo),
 	    TP_STRUCT__entry(
 			     __field(struct amdgpu_bo *, bo)
 			     __field(u32, pages)
+			     __field(u32, type)
+			     __field(u32, prefer)
+			     __field(u32, allow)
+			     __field(u32, visible)
 			     ),
 
 	    TP_fast_assign(
 			   __entry->bo = bo;
 			   __entry->pages = bo->tbo.num_pages;
+			   __entry->type = bo->tbo.mem.mem_type;
+			   __entry->prefer = bo->prefered_domains;
+			   __entry->allow = bo->allowed_domains;
+			   __entry->visible = bo->flags;
 			   ),
-	    TP_printk("bo=%p, pages=%u", __entry->bo, __entry->pages)
+
+	    TP_printk("bo=%p,pages=%u,type=%d,prefered=%d,allowed=%d,visible=%d",
+		       __entry->bo, __entry->pages, __entry->type,
+		       __entry->prefer, __entry->allow, __entry->visible)
 );
 
 TRACE_EVENT(amdgpu_cs,
@@ -64,7 +113,7 @@ TRACE_EVENT(amdgpu_cs_ioctl,
 			   __entry->adev = job->adev;
 			   __entry->sched_job = &job->base;
 			   __entry->ib = job->ibs;
-			   __entry->fence = &job->base.s_fence->base;
+			   __entry->fence = &job->base.s_fence->finished;
 			   __entry->ring_name = job->ring->name;
 			   __entry->num_ibs = job->num_ibs;
 			   ),
@@ -89,7 +138,7 @@ TRACE_EVENT(amdgpu_sched_run_job,
 			   __entry->adev = job->adev;
 			   __entry->sched_job = &job->base;
 			   __entry->ib = job->ibs;
-			   __entry->fence = &job->base.s_fence->base;
+			   __entry->fence = &job->base.s_fence->finished;
 			   __entry->ring_name = job->ring->name;
 			   __entry->num_ibs = job->num_ibs;
 			   ),
@@ -100,24 +149,26 @@ TRACE_EVENT(amdgpu_sched_run_job,
 
 
 TRACE_EVENT(amdgpu_vm_grab_id,
-	    TP_PROTO(struct amdgpu_vm *vm, int ring, unsigned vmid,
-		     uint64_t pd_addr),
-	    TP_ARGS(vm, ring, vmid, pd_addr),
+	    TP_PROTO(struct amdgpu_vm *vm, int ring, struct amdgpu_job *job),
+	    TP_ARGS(vm, ring, job),
 	    TP_STRUCT__entry(
 			     __field(struct amdgpu_vm *, vm)
 			     __field(u32, ring)
 			     __field(u32, vmid)
 			     __field(u64, pd_addr)
+			     __field(u32, needs_flush)
 			     ),
 
 	    TP_fast_assign(
 			   __entry->vm = vm;
 			   __entry->ring = ring;
-			   __entry->vmid = vmid;
-			   __entry->pd_addr = pd_addr;
+			   __entry->vmid = job->vm_id;
+			   __entry->pd_addr = job->vm_pd_addr;
+			   __entry->needs_flush = job->vm_needs_flush;
 			   ),
-	    TP_printk("vm=%p, ring=%u, id=%u, pd_addr=%010Lx", __entry->vm,
-		      __entry->ring, __entry->vmid, __entry->pd_addr)
+	    TP_printk("vm=%p, ring=%u, id=%u, pd_addr=%010Lx needs_flush=%u",
+		      __entry->vm, __entry->ring, __entry->vmid,
+		      __entry->pd_addr, __entry->needs_flush)
 );
 
 TRACE_EVENT(amdgpu_vm_bo_map,
@@ -244,13 +295,55 @@ TRACE_EVENT(amdgpu_bo_list_set,
 	    TP_STRUCT__entry(
 			     __field(struct amdgpu_bo_list *, list)
 			     __field(struct amdgpu_bo *, bo)
+			     __field(u64, bo_size)
 			     ),
 
 	    TP_fast_assign(
 			   __entry->list = list;
 			   __entry->bo = bo;
+			   __entry->bo_size = amdgpu_bo_size(bo);
 			   ),
-	    TP_printk("list=%p, bo=%p", __entry->list, __entry->bo)
+	    TP_printk("list=%p, bo=%p, bo_size = %Ld",
+		      __entry->list,
+		      __entry->bo,
+		      __entry->bo_size)
+);
+
+TRACE_EVENT(amdgpu_cs_bo_status,
+	    TP_PROTO(uint64_t total_bo, uint64_t total_size),
+	    TP_ARGS(total_bo, total_size),
+	    TP_STRUCT__entry(
+			__field(u64, total_bo)
+			__field(u64, total_size)
+			),
+
+	    TP_fast_assign(
+			__entry->total_bo = total_bo;
+			__entry->total_size = total_size;
+			),
+	    TP_printk("total bo size = %Ld, total bo count = %Ld",
+			__entry->total_bo, __entry->total_size)
+);
+
+TRACE_EVENT(amdgpu_ttm_bo_move,
+	    TP_PROTO(struct amdgpu_bo* bo, uint32_t new_placement, uint32_t old_placement),
+	    TP_ARGS(bo, new_placement, old_placement),
+	    TP_STRUCT__entry(
+			__field(struct amdgpu_bo *, bo)
+			__field(u64, bo_size)
+			__field(u32, new_placement)
+			__field(u32, old_placement)
+			),
+
+	    TP_fast_assign(
+			__entry->bo      = bo;
+			__entry->bo_size = amdgpu_bo_size(bo);
+			__entry->new_placement = new_placement;
+			__entry->old_placement = old_placement;
+			),
+	    TP_printk("bo=%p from:%d to %d with size = %Ld",
+			__entry->bo, __entry->old_placement,
+			__entry->new_placement, __entry->bo_size)
 );
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
index 46c5297f6..716f2afeb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -286,9 +286,10 @@ static int amdgpu_move_blit(struct ttm_buffer_object *bo,
 	r = amdgpu_copy_buffer(ring, old_start, new_start,
 			       new_mem->num_pages * PAGE_SIZE, /* bytes */
 			       bo->resv, &fence);
-	/* FIXME: handle copy error */
-	r = ttm_bo_move_accel_cleanup(bo, fence,
-				      evict, no_wait_gpu, new_mem);
+	if (r)
+		return r;
+
+	r = ttm_bo_pipeline_move(bo, fence, evict, new_mem);
 	fence_put(fence);
 	return r;
 }
@@ -334,7 +335,7 @@ static int amdgpu_move_vram_ram(struct ttm_buffer_object *bo,
 	if (unlikely(r)) {
 		goto out_cleanup;
 	}
-	r = ttm_bo_move_ttm(bo, true, no_wait_gpu, new_mem);
+	r = ttm_bo_move_ttm(bo, true, interruptible, no_wait_gpu, new_mem);
 out_cleanup:
 	ttm_bo_mem_put(bo, &tmp_mem);
 	return r;
@@ -367,7 +368,7 @@ static int amdgpu_move_ram_vram(struct ttm_buffer_object *bo,
 	if (unlikely(r)) {
 		return r;
 	}
-	r = ttm_bo_move_ttm(bo, true, no_wait_gpu, &tmp_mem);
+	r = ttm_bo_move_ttm(bo, true, interruptible, no_wait_gpu, &tmp_mem);
 	if (unlikely(r)) {
 		goto out_cleanup;
 	}
@@ -396,6 +397,11 @@ static int amdgpu_bo_move(struct ttm_buffer_object *bo,
 		return -EINVAL;
 
 	adev = amdgpu_get_adev(bo->bdev);
+
+	/* remember the eviction */
+	if (evict)
+		atomic64_inc(&adev->num_evictions);
+
 	if (old_mem->mem_type == TTM_PL_SYSTEM && bo->ttm == NULL) {
 		amdgpu_move_null(bo, new_mem);
 		return 0;
@@ -429,7 +435,8 @@ static int amdgpu_bo_move(struct ttm_buffer_object *bo,
 
 	if (r) {
 memcpy:
-		r = ttm_bo_move_memcpy(bo, evict, no_wait_gpu, new_mem);
+		r = ttm_bo_move_memcpy(bo, evict, interruptible,
+				       no_wait_gpu, new_mem);
 		if (r) {
 			return r;
 		}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
index 3959055eb..91aeed825 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
@@ -40,9 +40,16 @@
 #include "uvd/uvd_4_2_d.h"
 
 /* 1 second timeout */
-#define UVD_IDLE_TIMEOUT_MS	1000
+#define UVD_IDLE_TIMEOUT	msecs_to_jiffies(1000)
+
+/* Firmware versions for VI */
+#define FW_1_65_10	((1 << 24) | (65 << 16) | (10 << 8))
+#define FW_1_87_11	((1 << 24) | (87 << 16) | (11 << 8))
+#define FW_1_87_12	((1 << 24) | (87 << 16) | (12 << 8))
+#define FW_1_37_15	((1 << 24) | (37 << 16) | (15 << 8))
+
 /* Polaris10/11 firmware version */
-#define FW_1_66_16 ((1 << 24) | (66 << 16) | (16 << 8))
+#define FW_1_66_16	((1 << 24) | (66 << 16) | (16 << 8))
 
 /* Firmware Names */
 #ifdef CONFIG_DRM_AMDGPU_CIK
@@ -83,7 +90,6 @@ struct amdgpu_uvd_cs_ctx {
 #endif
 /*(DEBLOBBED)*/
 
-static void amdgpu_uvd_note_usage(struct amdgpu_device *adev);
 static void amdgpu_uvd_idle_work_handler(struct work_struct *work);
 
 int amdgpu_uvd_sw_init(struct amdgpu_device *adev)
@@ -237,6 +243,23 @@ int amdgpu_uvd_sw_init(struct amdgpu_device *adev)
 	if (!amdgpu_ip_block_version_cmp(adev, AMD_IP_BLOCK_TYPE_UVD, 5, 0))
 		adev->uvd.address_64_bit = true;
 
+	switch (adev->asic_type) {
+	case CHIP_TONGA:
+		adev->uvd.use_ctx_buf = adev->uvd.fw_version >= FW_1_65_10;
+		break;
+	case CHIP_CARRIZO:
+		adev->uvd.use_ctx_buf = adev->uvd.fw_version >= FW_1_87_11;
+		break;
+	case CHIP_FIJI:
+		adev->uvd.use_ctx_buf = adev->uvd.fw_version >= FW_1_87_12;
+		break;
+	case CHIP_STONEY:
+		adev->uvd.use_ctx_buf = adev->uvd.fw_version >= FW_1_37_15;
+		break;
+	default:
+		adev->uvd.use_ctx_buf = adev->asic_type >= CHIP_POLARIS10;
+	}
+
 	return 0;
 }
 
@@ -337,8 +360,6 @@ void amdgpu_uvd_free_handles(struct amdgpu_device *adev, struct drm_file *filp)
 		if (handle != 0 && adev->uvd.filp[i] == filp) {
 			struct fence *fence;
 
-			amdgpu_uvd_note_usage(adev);
-
 			r = amdgpu_uvd_get_destroy_msg(ring, handle,
 						       false, &fence);
 			if (r) {
@@ -429,7 +450,7 @@ static int amdgpu_uvd_cs_msg_decode(struct amdgpu_device *adev, uint32_t *msg,
 	unsigned fs_in_mb = width_in_mb * height_in_mb;
 
 	unsigned image_size, tmp, min_dpb_size, num_dpb_buffer;
-	unsigned min_ctx_size = 0;
+	unsigned min_ctx_size = ~0;
 
 	image_size = width * height;
 	image_size += image_size / 2;
@@ -548,7 +569,7 @@ static int amdgpu_uvd_cs_msg_decode(struct amdgpu_device *adev, uint32_t *msg,
 		/* reference picture buffer */
 		min_dpb_size = image_size * num_dpb_buffer;
 
-		if (adev->asic_type < CHIP_POLARIS10){
+		if (!adev->uvd.use_ctx_buf){
 			/* macroblock context buffer */
 			min_dpb_size +=
 				width_in_mb * height_in_mb * num_dpb_buffer * 192;
@@ -653,7 +674,7 @@ static int amdgpu_uvd_cs_msg(struct amdgpu_uvd_cs_ctx *ctx,
 		}
 
 		DRM_ERROR("No more free UVD handles!\n");
-		return -EINVAL;
+		return -ENOSPC;
 
 	case 1:
 		/* it's a decode msg, calc buffer sizes */
@@ -904,8 +925,6 @@ int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx)
 		return -EINVAL;
 	}
 
-	amdgpu_uvd_note_usage(ctx.parser->adev);
-
 	return 0;
 }
 
@@ -959,7 +978,7 @@ static int amdgpu_uvd_send_msg(struct amdgpu_ring *ring, struct amdgpu_bo *bo,
 
 	if (direct) {
 		r = amdgpu_ib_schedule(ring, 1, ib, NULL, NULL, &f);
-		job->fence = f;
+		job->fence = fence_get(f);
 		if (r)
 			goto err_free;
 
@@ -1097,24 +1116,18 @@ static void amdgpu_uvd_idle_work_handler(struct work_struct *work)
 	if (fences == 0 && handles == 0) {
 		if (adev->pm.dpm_enabled) {
 			amdgpu_dpm_enable_uvd(adev, false);
-			/* just work around for uvd clock remain high even
-			 * when uvd dpm disabled on Polaris10 */
-			if (adev->asic_type == CHIP_POLARIS10)
-				amdgpu_asic_set_uvd_clocks(adev, 0, 0);
 		} else {
 			amdgpu_asic_set_uvd_clocks(adev, 0, 0);
 		}
 	} else {
-		schedule_delayed_work(&adev->uvd.idle_work,
-				      msecs_to_jiffies(UVD_IDLE_TIMEOUT_MS));
+		schedule_delayed_work(&adev->uvd.idle_work, UVD_IDLE_TIMEOUT);
 	}
 }
 
-static void amdgpu_uvd_note_usage(struct amdgpu_device *adev)
+void amdgpu_uvd_ring_begin_use(struct amdgpu_ring *ring)
 {
+	struct amdgpu_device *adev = ring->adev;
 	bool set_clocks = !cancel_delayed_work_sync(&adev->uvd.idle_work);
-	set_clocks &= schedule_delayed_work(&adev->uvd.idle_work,
-					    msecs_to_jiffies(UVD_IDLE_TIMEOUT_MS));
 
 	if (set_clocks) {
 		if (adev->pm.dpm_enabled) {
@@ -1124,3 +1137,49 @@ static void amdgpu_uvd_note_usage(struct amdgpu_device *adev)
 		}
 	}
 }
+
+void amdgpu_uvd_ring_end_use(struct amdgpu_ring *ring)
+{
+	schedule_delayed_work(&ring->adev->uvd.idle_work, UVD_IDLE_TIMEOUT);
+}
+
+/**
+ * amdgpu_uvd_ring_test_ib - test ib execution
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Test if we can successfully execute an IB
+ */
+int amdgpu_uvd_ring_test_ib(struct amdgpu_ring *ring, long timeout)
+{
+	struct fence *fence;
+	long r;
+
+	r = amdgpu_uvd_get_create_msg(ring, 1, NULL);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r);
+		goto error;
+	}
+
+	r = amdgpu_uvd_get_destroy_msg(ring, 1, true, &fence);
+	if (r) {
+		DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r);
+		goto error;
+	}
+
+	r = fence_wait_timeout(fence, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out.\n");
+		r = -ETIMEDOUT;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
+	} else {
+		DRM_INFO("ib test on ring %d succeeded\n",  ring->idx);
+		r = 0;
+	}
+
+	fence_put(fence);
+
+error:
+	return r;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h
index 9a3b44908..c85000960 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.h
@@ -35,5 +35,8 @@ int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 void amdgpu_uvd_free_handles(struct amdgpu_device *adev,
 			     struct drm_file *filp);
 int amdgpu_uvd_ring_parse_cs(struct amdgpu_cs_parser *parser, uint32_t ib_idx);
+void amdgpu_uvd_ring_begin_use(struct amdgpu_ring *ring);
+void amdgpu_uvd_ring_end_use(struct amdgpu_ring *ring);
+int amdgpu_uvd_ring_test_ib(struct amdgpu_ring *ring, long timeout);
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
index 0a08cf930..7f8c44308 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
@@ -36,7 +36,7 @@
 #include "cikd.h"
 
 /* 1 second timeout */
-#define VCE_IDLE_TIMEOUT_MS	1000
+#define VCE_IDLE_TIMEOUT	msecs_to_jiffies(1000)
 
 /* Firmware Names */
 #ifdef CONFIG_DRM_AMDGPU_CIK
@@ -76,8 +76,6 @@ int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size)
 	unsigned ucode_version, version_major, version_minor, binary_id;
 	int i, r;
 
-	INIT_DELAYED_WORK(&adev->vce.idle_work, amdgpu_vce_idle_work_handler);
-
 	switch (adev->asic_type) {
 #ifdef CONFIG_DRM_AMDGPU_CIK
 	case CHIP_BONAIRE:
@@ -188,6 +186,9 @@ int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size)
 		adev->vce.filp[i] = NULL;
 	}
 
+	INIT_DELAYED_WORK(&adev->vce.idle_work, amdgpu_vce_idle_work_handler);
+	mutex_init(&adev->vce.idle_mutex);
+
 	return 0;
 }
 
@@ -211,6 +212,7 @@ int amdgpu_vce_sw_fini(struct amdgpu_device *adev)
 	amdgpu_ring_fini(&adev->vce.ring[1]);
 
 	release_firmware(adev->vce.fw);
+	mutex_destroy(&adev->vce.idle_mutex);
 
 	return 0;
 }
@@ -301,37 +303,44 @@ static void amdgpu_vce_idle_work_handler(struct work_struct *work)
 			amdgpu_asic_set_vce_clocks(adev, 0, 0);
 		}
 	} else {
-		schedule_delayed_work(&adev->vce.idle_work,
-				      msecs_to_jiffies(VCE_IDLE_TIMEOUT_MS));
+		schedule_delayed_work(&adev->vce.idle_work, VCE_IDLE_TIMEOUT);
 	}
 }
 
 /**
- * amdgpu_vce_note_usage - power up VCE
+ * amdgpu_vce_ring_begin_use - power up VCE
  *
- * @adev: amdgpu_device pointer
+ * @ring: amdgpu ring
  *
  * Make sure VCE is powerd up when we want to use it
  */
-static void amdgpu_vce_note_usage(struct amdgpu_device *adev)
+void amdgpu_vce_ring_begin_use(struct amdgpu_ring *ring)
 {
-	bool streams_changed = false;
-	bool set_clocks = !cancel_delayed_work_sync(&adev->vce.idle_work);
-	set_clocks &= schedule_delayed_work(&adev->vce.idle_work,
-					    msecs_to_jiffies(VCE_IDLE_TIMEOUT_MS));
-
-	if (adev->pm.dpm_enabled) {
-		/* XXX figure out if the streams changed */
-		streams_changed = false;
-	}
+	struct amdgpu_device *adev = ring->adev;
+	bool set_clocks;
 
-	if (set_clocks || streams_changed) {
+	mutex_lock(&adev->vce.idle_mutex);
+	set_clocks = !cancel_delayed_work_sync(&adev->vce.idle_work);
+	if (set_clocks) {
 		if (adev->pm.dpm_enabled) {
 			amdgpu_dpm_enable_vce(adev, true);
 		} else {
 			amdgpu_asic_set_vce_clocks(adev, 53300, 40000);
 		}
 	}
+	mutex_unlock(&adev->vce.idle_mutex);
+}
+
+/**
+ * amdgpu_vce_ring_end_use - power VCE down
+ *
+ * @ring: amdgpu ring
+ *
+ * Schedule work to power VCE down again
+ */
+void amdgpu_vce_ring_end_use(struct amdgpu_ring *ring)
+{
+	schedule_delayed_work(&ring->adev->vce.idle_work, VCE_IDLE_TIMEOUT);
 }
 
 /**
@@ -348,11 +357,10 @@ void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp)
 	int i, r;
 	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) {
 		uint32_t handle = atomic_read(&adev->vce.handles[i]);
+
 		if (!handle || adev->vce.filp[i] != filp)
 			continue;
 
-		amdgpu_vce_note_usage(adev);
-
 		r = amdgpu_vce_get_destroy_msg(ring, handle, false, NULL);
 		if (r)
 			DRM_ERROR("Error destroying VCE handle (%d)!\n", r);
@@ -428,7 +436,7 @@ int amdgpu_vce_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
 		ib->ptr[i] = 0x0;
 
 	r = amdgpu_ib_schedule(ring, 1, ib, NULL, NULL, &f);
-	job->fence = f;
+	job->fence = fence_get(f);
 	if (r)
 		goto err;
 
@@ -460,7 +468,6 @@ int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 	struct amdgpu_job *job;
 	struct amdgpu_ib *ib;
 	struct fence *f = NULL;
-	uint64_t dummy;
 	int i, r;
 
 	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4, &job);
@@ -468,7 +475,6 @@ int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 		return r;
 
 	ib = &job->ibs[0];
-	dummy = ib->gpu_addr + 1024;
 
 	/* stitch together an VCE destroy msg */
 	ib->length_dw = 0;
@@ -476,11 +482,14 @@ int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 	ib->ptr[ib->length_dw++] = 0x00000001; /* session cmd */
 	ib->ptr[ib->length_dw++] = handle;
 
-	ib->ptr[ib->length_dw++] = 0x00000014; /* len */
-	ib->ptr[ib->length_dw++] = 0x05000005; /* feedback buffer */
-	ib->ptr[ib->length_dw++] = upper_32_bits(dummy);
-	ib->ptr[ib->length_dw++] = dummy;
-	ib->ptr[ib->length_dw++] = 0x00000001;
+	ib->ptr[ib->length_dw++] = 0x00000020; /* len */
+	ib->ptr[ib->length_dw++] = 0x00000002; /* task info */
+	ib->ptr[ib->length_dw++] = 0xffffffff; /* next task info, set to 0xffffffff if no */
+	ib->ptr[ib->length_dw++] = 0x00000001; /* destroy session */
+	ib->ptr[ib->length_dw++] = 0x00000000;
+	ib->ptr[ib->length_dw++] = 0x00000000;
+	ib->ptr[ib->length_dw++] = 0xffffffff; /* feedback is not needed, set to 0xffffffff and firmware will not output feedback */
+	ib->ptr[ib->length_dw++] = 0x00000000;
 
 	ib->ptr[ib->length_dw++] = 0x00000008; /* len */
 	ib->ptr[ib->length_dw++] = 0x02000001; /* destroy cmd */
@@ -490,7 +499,7 @@ int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 
 	if (direct) {
 		r = amdgpu_ib_schedule(ring, 1, ib, NULL, NULL, &f);
-		job->fence = f;
+		job->fence = fence_get(f);
 		if (r)
 			goto err;
 
@@ -571,12 +580,10 @@ static int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx,
  * we we don't have another free session index.
  */
 static int amdgpu_vce_validate_handle(struct amdgpu_cs_parser *p,
-				      uint32_t handle, bool *allocated)
+				      uint32_t handle, uint32_t *allocated)
 {
 	unsigned i;
 
-	*allocated = false;
-
 	/* validate the handle */
 	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) {
 		if (atomic_read(&p->adev->vce.handles[i]) == handle) {
@@ -593,7 +600,7 @@ static int amdgpu_vce_validate_handle(struct amdgpu_cs_parser *p,
 		if (!atomic_cmpxchg(&p->adev->vce.handles[i], 0, handle)) {
 			p->adev->vce.filp[i] = p->filp;
 			p->adev->vce.img_size[i] = 0;
-			*allocated = true;
+			*allocated |= 1 << i;
 			return i;
 		}
 	}
@@ -613,15 +620,13 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
 	struct amdgpu_ib *ib = &p->job->ibs[ib_idx];
 	unsigned fb_idx = 0, bs_idx = 0;
 	int session_idx = -1;
-	bool destroyed = false;
-	bool created = false;
-	bool allocated = false;
+	uint32_t destroyed = 0;
+	uint32_t created = 0;
+	uint32_t allocated = 0;
 	uint32_t tmp, handle = 0;
 	uint32_t *size = &tmp;
 	int i, r = 0, idx = 0;
 
-	amdgpu_vce_note_usage(p->adev);
-
 	while (idx < ib->length_dw) {
 		uint32_t len = amdgpu_get_ib_value(p, ib_idx, idx);
 		uint32_t cmd = amdgpu_get_ib_value(p, ib_idx, idx + 1);
@@ -632,30 +637,30 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
 			goto out;
 		}
 
-		if (destroyed) {
-			DRM_ERROR("No other command allowed after destroy!\n");
-			r = -EINVAL;
-			goto out;
-		}
-
 		switch (cmd) {
-		case 0x00000001: // session
+		case 0x00000001: /* session */
 			handle = amdgpu_get_ib_value(p, ib_idx, idx + 2);
 			session_idx = amdgpu_vce_validate_handle(p, handle,
 								 &allocated);
-			if (session_idx < 0)
-				return session_idx;
+			if (session_idx < 0) {
+				r = session_idx;
+				goto out;
+			}
 			size = &p->adev->vce.img_size[session_idx];
 			break;
 
-		case 0x00000002: // task info
+		case 0x00000002: /* task info */
 			fb_idx = amdgpu_get_ib_value(p, ib_idx, idx + 6);
 			bs_idx = amdgpu_get_ib_value(p, ib_idx, idx + 7);
 			break;
 
-		case 0x01000001: // create
-			created = true;
-			if (!allocated) {
+		case 0x01000001: /* create */
+			created |= 1 << session_idx;
+			if (destroyed & (1 << session_idx)) {
+				destroyed &= ~(1 << session_idx);
+				allocated |= 1 << session_idx;
+
+			} else if (!(allocated & (1 << session_idx))) {
 				DRM_ERROR("Handle already in use!\n");
 				r = -EINVAL;
 				goto out;
@@ -666,16 +671,16 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
 				8 * 3 / 2;
 			break;
 
-		case 0x04000001: // config extension
-		case 0x04000002: // pic control
-		case 0x04000005: // rate control
-		case 0x04000007: // motion estimation
-		case 0x04000008: // rdo
-		case 0x04000009: // vui
-		case 0x05000002: // auxiliary buffer
+		case 0x04000001: /* config extension */
+		case 0x04000002: /* pic control */
+		case 0x04000005: /* rate control */
+		case 0x04000007: /* motion estimation */
+		case 0x04000008: /* rdo */
+		case 0x04000009: /* vui */
+		case 0x05000002: /* auxiliary buffer */
 			break;
 
-		case 0x03000001: // encode
+		case 0x03000001: /* encode */
 			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 10, idx + 9,
 						*size, 0);
 			if (r)
@@ -687,18 +692,18 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
 				goto out;
 			break;
 
-		case 0x02000001: // destroy
-			destroyed = true;
+		case 0x02000001: /* destroy */
+			destroyed |= 1 << session_idx;
 			break;
 
-		case 0x05000001: // context buffer
+		case 0x05000001: /* context buffer */
 			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2,
 						*size * 2, 0);
 			if (r)
 				goto out;
 			break;
 
-		case 0x05000004: // video bitstream buffer
+		case 0x05000004: /* video bitstream buffer */
 			tmp = amdgpu_get_ib_value(p, ib_idx, idx + 4);
 			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2,
 						tmp, bs_idx);
@@ -706,7 +711,7 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
 				goto out;
 			break;
 
-		case 0x05000005: // feedback buffer
+		case 0x05000005: /* feedback buffer */
 			r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2,
 						4096, fb_idx);
 			if (r)
@@ -728,21 +733,24 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx)
 		idx += len / 4;
 	}
 
-	if (allocated && !created) {
+	if (allocated & ~created) {
 		DRM_ERROR("New session without create command!\n");
 		r = -ENOENT;
 	}
 
 out:
-	if ((!r && destroyed) || (r && allocated)) {
-		/*
-		 * IB contains a destroy msg or we have allocated an
-		 * handle and got an error, anyway free the handle
-		 */
-		for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i)
-			atomic_cmpxchg(&p->adev->vce.handles[i], handle, 0);
+	if (!r) {
+		/* No error, free all destroyed handle slots */
+		tmp = destroyed;
+	} else {
+		/* Error during parsing, free all allocated handle slots */
+		tmp = allocated;
 	}
 
+	for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i)
+		if (tmp & (1 << i))
+			atomic_set(&p->adev->vce.handles[i], 0);
+
 	return r;
 }
 
@@ -828,10 +836,10 @@ int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring)
  * @ring: the engine to test on
  *
  */
-int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring)
+int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 {
 	struct fence *fence = NULL;
-	int r;
+	long r;
 
 	/* skip vce ring1 ib test for now, since it's not reliable */
 	if (ring == &ring->adev->vce.ring[1])
@@ -839,21 +847,25 @@ int amdgpu_vce_ring_test_ib(struct amdgpu_ring *ring)
 
 	r = amdgpu_vce_get_create_msg(ring, 1, NULL);
 	if (r) {
-		DRM_ERROR("amdgpu: failed to get create msg (%d).\n", r);
+		DRM_ERROR("amdgpu: failed to get create msg (%ld).\n", r);
 		goto error;
 	}
 
 	r = amdgpu_vce_get_destroy_msg(ring, 1, true, &fence);
 	if (r) {
-		DRM_ERROR("amdgpu: failed to get destroy ib (%d).\n", r);
+		DRM_ERROR("amdgpu: failed to get destroy ib (%ld).\n", r);
 		goto error;
 	}
 
-	r = fence_wait(fence, false);
-	if (r) {
-		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+	r = fence_wait_timeout(fence, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out.\n");
+		r = -ETIMEDOUT;
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
 	} else {
 		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
 	}
 error:
 	fence_put(fence);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h
index f40cf761c..63f83d0d9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h
@@ -39,6 +39,8 @@ void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib,
 void amdgpu_vce_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
 				unsigned flags);
 int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring);
-int amdgpu_vce_ring_test_ib(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);
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 9f36ed30b..80120fa40 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -25,6 +25,7 @@
  *          Alex Deucher
  *          Jerome Glisse
  */
+#include <linux/fence-array.h>
 #include <drm/drmP.h>
 #include <drm/amdgpu_drm.h>
 #include "amdgpu.h"
@@ -114,16 +115,26 @@ void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
 /**
  * amdgpu_vm_get_bos - add the vm BOs to a duplicates list
  *
+ * @adev: amdgpu device pointer
  * @vm: vm providing the BOs
  * @duplicates: head of duplicates list
  *
  * Add the page directory to the BO duplicates list
  * for command submission.
  */
-void amdgpu_vm_get_pt_bos(struct amdgpu_vm *vm, struct list_head *duplicates)
+void amdgpu_vm_get_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			  struct list_head *duplicates)
 {
+	uint64_t num_evictions;
 	unsigned i;
 
+	/* We only need to validate the page tables
+	 * if they aren't already valid.
+	 */
+	num_evictions = atomic64_read(&adev->num_evictions);
+	if (num_evictions == vm->last_eviction_counter)
+		return;
+
 	/* add the vm page table to the list */
 	for (i = 0; i <= vm->max_pde_used; ++i) {
 		struct amdgpu_bo_list_entry *entry = &vm->page_tables[i].entry;
@@ -162,6 +173,13 @@ void amdgpu_vm_move_pt_bos_in_lru(struct amdgpu_device *adev,
 	spin_unlock(&glob->lru_lock);
 }
 
+static bool amdgpu_vm_is_gpu_reset(struct amdgpu_device *adev,
+			      struct amdgpu_vm_id *id)
+{
+	return id->current_gpu_reset_count !=
+		atomic_read(&adev->gpu_reset_counter) ? true : false;
+}
+
 /**
  * amdgpu_vm_grab_id - allocate the next free VMID
  *
@@ -174,18 +192,67 @@ void amdgpu_vm_move_pt_bos_in_lru(struct amdgpu_device *adev,
  */
 int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
 		      struct amdgpu_sync *sync, struct fence *fence,
-		      unsigned *vm_id, uint64_t *vm_pd_addr)
+		      struct amdgpu_job *job)
 {
-	uint64_t pd_addr = amdgpu_bo_gpu_offset(vm->page_directory);
 	struct amdgpu_device *adev = ring->adev;
+	uint64_t fence_context = adev->fence_context + ring->idx;
 	struct fence *updates = sync->last_vm_update;
-	struct amdgpu_vm_id *id;
-	unsigned i = ring->idx;
-	int r;
+	struct amdgpu_vm_id *id, *idle;
+	struct fence **fences;
+	unsigned i;
+	int r = 0;
+
+	fences = kmalloc_array(sizeof(void *), adev->vm_manager.num_ids,
+			       GFP_KERNEL);
+	if (!fences)
+		return -ENOMEM;
 
 	mutex_lock(&adev->vm_manager.lock);
 
+	/* Check if we have an idle VMID */
+	i = 0;
+	list_for_each_entry(idle, &adev->vm_manager.ids_lru, list) {
+		fences[i] = amdgpu_sync_peek_fence(&idle->active, ring);
+		if (!fences[i])
+			break;
+		++i;
+	}
+
+	/* If we can't find a idle VMID to use, wait till one becomes available */
+	if (&idle->list == &adev->vm_manager.ids_lru) {
+		u64 fence_context = adev->vm_manager.fence_context + ring->idx;
+		unsigned seqno = ++adev->vm_manager.seqno[ring->idx];
+		struct fence_array *array;
+		unsigned j;
+
+		for (j = 0; j < i; ++j)
+			fence_get(fences[j]);
+
+		array = fence_array_create(i, fences, fence_context,
+					   seqno, true);
+		if (!array) {
+			for (j = 0; j < i; ++j)
+				fence_put(fences[j]);
+			kfree(fences);
+			r = -ENOMEM;
+			goto error;
+		}
+
+
+		r = amdgpu_sync_fence(ring->adev, sync, &array->base);
+		fence_put(&array->base);
+		if (r)
+			goto error;
+
+		mutex_unlock(&adev->vm_manager.lock);
+		return 0;
+
+	}
+	kfree(fences);
+
+	job->vm_needs_flush = true;
 	/* Check if we can use a VMID already assigned to this VM */
+	i = ring->idx;
 	do {
 		struct fence *flushed;
 
@@ -196,67 +263,52 @@ int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
 		/* Check all the prerequisites to using this VMID */
 		if (!id)
 			continue;
+		if (amdgpu_vm_is_gpu_reset(adev, id))
+			continue;
 
 		if (atomic64_read(&id->owner) != vm->client_id)
 			continue;
 
-		if (pd_addr != id->pd_gpu_addr)
+		if (job->vm_pd_addr != id->pd_gpu_addr)
 			continue;
 
-		if (id->last_user != ring &&
-		    (!id->last_flush || !fence_is_signaled(id->last_flush)))
+		if (!id->last_flush)
 			continue;
 
-		flushed  = id->flushed_updates;
-		if (updates && (!flushed || fence_is_later(updates, flushed)))
+		if (id->last_flush->context != fence_context &&
+		    !fence_is_signaled(id->last_flush))
 			continue;
 
-		/* Good we can use this VMID */
-		if (id->last_user == ring) {
-			r = amdgpu_sync_fence(ring->adev, sync,
-					      id->first);
-			if (r)
-				goto error;
-		}
+		flushed  = id->flushed_updates;
+		if (updates &&
+		    (!flushed || fence_is_later(updates, flushed)))
+			continue;
 
-		/* And remember this submission as user of the VMID */
+		/* Good we can use this VMID. Remember this submission as
+		 * user of the VMID.
+		 */
 		r = amdgpu_sync_fence(ring->adev, &id->active, fence);
 		if (r)
 			goto error;
 
+		id->current_gpu_reset_count = atomic_read(&adev->gpu_reset_counter);
 		list_move_tail(&id->list, &adev->vm_manager.ids_lru);
 		vm->ids[ring->idx] = id;
 
-		*vm_id = id - adev->vm_manager.ids;
-		*vm_pd_addr = AMDGPU_VM_NO_FLUSH;
-		trace_amdgpu_vm_grab_id(vm, ring->idx, *vm_id, *vm_pd_addr);
+		job->vm_id = id - adev->vm_manager.ids;
+		job->vm_needs_flush = false;
+		trace_amdgpu_vm_grab_id(vm, ring->idx, job);
 
 		mutex_unlock(&adev->vm_manager.lock);
 		return 0;
 
 	} while (i != ring->idx);
 
-	id = list_first_entry(&adev->vm_manager.ids_lru,
-			      struct amdgpu_vm_id,
-			      list);
-
-	if (!amdgpu_sync_is_idle(&id->active)) {
-		struct list_head *head = &adev->vm_manager.ids_lru;
-		struct amdgpu_vm_id *tmp;
-
-		list_for_each_entry_safe(id, tmp, &adev->vm_manager.ids_lru,
-					 list) {
-			if (amdgpu_sync_is_idle(&id->active)) {
-				list_move(&id->list, head);
-				head = &id->list;
-			}
-		}
-		id = list_first_entry(&adev->vm_manager.ids_lru,
-				      struct amdgpu_vm_id,
-				      list);
-	}
+	/* Still no ID to use? Then use the idle one found earlier */
+	id = idle;
 
-	r = amdgpu_sync_cycle_fences(sync, &id->active, fence);
+	/* Remember this submission as user of the VMID */
+	r = amdgpu_sync_fence(ring->adev, &id->active, fence);
 	if (r)
 		goto error;
 
@@ -269,22 +321,46 @@ int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
 	fence_put(id->flushed_updates);
 	id->flushed_updates = fence_get(updates);
 
-	id->pd_gpu_addr = pd_addr;
-
+	id->pd_gpu_addr = job->vm_pd_addr;
+	id->current_gpu_reset_count = atomic_read(&adev->gpu_reset_counter);
 	list_move_tail(&id->list, &adev->vm_manager.ids_lru);
-	id->last_user = ring;
 	atomic64_set(&id->owner, vm->client_id);
 	vm->ids[ring->idx] = id;
 
-	*vm_id = id - adev->vm_manager.ids;
-	*vm_pd_addr = pd_addr;
-	trace_amdgpu_vm_grab_id(vm, ring->idx, *vm_id, *vm_pd_addr);
+	job->vm_id = id - adev->vm_manager.ids;
+	trace_amdgpu_vm_grab_id(vm, ring->idx, job);
 
 error:
 	mutex_unlock(&adev->vm_manager.lock);
 	return r;
 }
 
+static bool amdgpu_vm_ring_has_compute_vm_bug(struct amdgpu_ring *ring)
+{
+	struct amdgpu_device *adev = ring->adev;
+	const struct amdgpu_ip_block_version *ip_block;
+
+	if (ring->type != AMDGPU_RING_TYPE_COMPUTE)
+		/* only compute rings */
+		return false;
+
+	ip_block = amdgpu_get_ip_block(adev, AMD_IP_BLOCK_TYPE_GFX);
+	if (!ip_block)
+		return false;
+
+	if (ip_block->major <= 7) {
+		/* gfx7 has no workaround */
+		return true;
+	} else if (ip_block->major == 8) {
+		if (adev->gfx.mec_fw_version >= 673)
+			/* gfx8 is fixed in MEC firmware 673 */
+			return false;
+		else
+			return true;
+	}
+	return false;
+}
+
 /**
  * amdgpu_vm_flush - hardware flush the vm
  *
@@ -294,59 +370,52 @@ error:
  *
  * Emit a VM flush when it is necessary.
  */
-int amdgpu_vm_flush(struct amdgpu_ring *ring,
-		    unsigned vm_id, uint64_t pd_addr,
-		    uint32_t gds_base, uint32_t gds_size,
-		    uint32_t gws_base, uint32_t gws_size,
-		    uint32_t oa_base, uint32_t oa_size)
+int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job)
 {
 	struct amdgpu_device *adev = ring->adev;
-	struct amdgpu_vm_id *id = &adev->vm_manager.ids[vm_id];
+	struct amdgpu_vm_id *id = &adev->vm_manager.ids[job->vm_id];
 	bool gds_switch_needed = ring->funcs->emit_gds_switch && (
-		id->gds_base != gds_base ||
-		id->gds_size != gds_size ||
-		id->gws_base != gws_base ||
-		id->gws_size != gws_size ||
-		id->oa_base != oa_base ||
-		id->oa_size != oa_size);
+		id->gds_base != job->gds_base ||
+		id->gds_size != job->gds_size ||
+		id->gws_base != job->gws_base ||
+		id->gws_size != job->gws_size ||
+		id->oa_base != job->oa_base ||
+		id->oa_size != job->oa_size);
 	int r;
 
 	if (ring->funcs->emit_pipeline_sync && (
-	    pd_addr != AMDGPU_VM_NO_FLUSH || gds_switch_needed ||
-		    ring->type == AMDGPU_RING_TYPE_COMPUTE))
+	    job->vm_needs_flush || gds_switch_needed ||
+	    amdgpu_vm_ring_has_compute_vm_bug(ring)))
 		amdgpu_ring_emit_pipeline_sync(ring);
 
-	if (ring->funcs->emit_vm_flush &&
-	    pd_addr != AMDGPU_VM_NO_FLUSH) {
+	if (ring->funcs->emit_vm_flush && (job->vm_needs_flush ||
+	    amdgpu_vm_is_gpu_reset(adev, id))) {
 		struct fence *fence;
 
-		trace_amdgpu_vm_flush(pd_addr, ring->idx, vm_id);
-		amdgpu_ring_emit_vm_flush(ring, vm_id, pd_addr);
+		trace_amdgpu_vm_flush(job->vm_pd_addr, ring->idx, job->vm_id);
+		amdgpu_ring_emit_vm_flush(ring, job->vm_id, job->vm_pd_addr);
+
+		r = amdgpu_fence_emit(ring, &fence);
+		if (r)
+			return r;
 
 		mutex_lock(&adev->vm_manager.lock);
-		if ((id->pd_gpu_addr == pd_addr) && (id->last_user == ring)) {
-			r = amdgpu_fence_emit(ring, &fence);
-			if (r) {
-				mutex_unlock(&adev->vm_manager.lock);
-				return r;
-			}
-			fence_put(id->last_flush);
-			id->last_flush = fence;
-		}
+		fence_put(id->last_flush);
+		id->last_flush = fence;
 		mutex_unlock(&adev->vm_manager.lock);
 	}
 
 	if (gds_switch_needed) {
-		id->gds_base = gds_base;
-		id->gds_size = gds_size;
-		id->gws_base = gws_base;
-		id->gws_size = gws_size;
-		id->oa_base = oa_base;
-		id->oa_size = oa_size;
-		amdgpu_ring_emit_gds_switch(ring, vm_id,
-					    gds_base, gds_size,
-					    gws_base, gws_size,
-					    oa_base, oa_size);
+		id->gds_base = job->gds_base;
+		id->gds_size = job->gds_size;
+		id->gws_base = job->gws_base;
+		id->gws_size = job->gws_size;
+		id->oa_base = job->oa_base;
+		id->oa_size = job->oa_size;
+		amdgpu_ring_emit_gds_switch(ring, job->vm_id,
+					    job->gds_base, job->gds_size,
+					    job->gws_base, job->gws_size,
+					    job->oa_base, job->oa_size);
 	}
 
 	return 0;
@@ -723,7 +792,7 @@ static void amdgpu_vm_frag_ptes(struct amdgpu_device *adev,
  * @vm: requested vm
  * @start: start of GPU address range
  * @end: end of GPU address range
- * @dst: destination address to map to
+ * @dst: destination address to map to, the next dst inside the function
  * @flags: mapping flags
  *
  * Update the page tables in the range @start - @end.
@@ -737,49 +806,75 @@ static void amdgpu_vm_update_ptes(struct amdgpu_device *adev,
 {
 	const uint64_t mask = AMDGPU_VM_PTE_COUNT - 1;
 
-	uint64_t last_pe_start = ~0, last_pe_end = ~0, last_dst = ~0;
-	uint64_t addr;
+	uint64_t cur_pe_start, cur_pe_end, cur_dst;
+	uint64_t addr; /* next GPU address to be updated */
+	uint64_t pt_idx;
+	struct amdgpu_bo *pt;
+	unsigned nptes; /* next number of ptes to be updated */
+	uint64_t next_pe_start;
+
+	/* initialize the variables */
+	addr = start;
+	pt_idx = addr >> amdgpu_vm_block_size;
+	pt = vm->page_tables[pt_idx].entry.robj;
+
+	if ((addr & ~mask) == (end & ~mask))
+		nptes = end - addr;
+	else
+		nptes = AMDGPU_VM_PTE_COUNT - (addr & mask);
+
+	cur_pe_start = amdgpu_bo_gpu_offset(pt);
+	cur_pe_start += (addr & mask) * 8;
+	cur_pe_end = cur_pe_start + 8 * nptes;
+	cur_dst = dst;
+
+	/* for next ptb*/
+	addr += nptes;
+	dst += nptes * AMDGPU_GPU_PAGE_SIZE;
 
 	/* walk over the address space and update the page tables */
-	for (addr = start; addr < end; ) {
-		uint64_t pt_idx = addr >> amdgpu_vm_block_size;
-		struct amdgpu_bo *pt = vm->page_tables[pt_idx].entry.robj;
-		unsigned nptes;
-		uint64_t pe_start;
+	while (addr < end) {
+		pt_idx = addr >> amdgpu_vm_block_size;
+		pt = vm->page_tables[pt_idx].entry.robj;
 
 		if ((addr & ~mask) == (end & ~mask))
 			nptes = end - addr;
 		else
 			nptes = AMDGPU_VM_PTE_COUNT - (addr & mask);
 
-		pe_start = amdgpu_bo_gpu_offset(pt);
-		pe_start += (addr & mask) * 8;
-
-		if (last_pe_end != pe_start) {
+		next_pe_start = amdgpu_bo_gpu_offset(pt);
+		next_pe_start += (addr & mask) * 8;
 
+		if (cur_pe_end == next_pe_start) {
+			/* The next ptb is consecutive to current ptb.
+			 * Don't call amdgpu_vm_frag_ptes now.
+			 * Will update two ptbs together in future.
+			*/
+			cur_pe_end += 8 * nptes;
+		} else {
 			amdgpu_vm_frag_ptes(adev, vm_update_params,
-					    last_pe_start, last_pe_end,
-					    last_dst, flags);
+					    cur_pe_start, cur_pe_end,
+					    cur_dst, flags);
 
-			last_pe_start = pe_start;
-			last_pe_end = pe_start + 8 * nptes;
-			last_dst = dst;
-		} else {
-			last_pe_end += 8 * nptes;
+			cur_pe_start = next_pe_start;
+			cur_pe_end = next_pe_start + 8 * nptes;
+			cur_dst = dst;
 		}
 
+		/* for next ptb*/
 		addr += nptes;
 		dst += nptes * AMDGPU_GPU_PAGE_SIZE;
 	}
 
-	amdgpu_vm_frag_ptes(adev, vm_update_params, last_pe_start,
-			    last_pe_end, last_dst, flags);
+	amdgpu_vm_frag_ptes(adev, vm_update_params, cur_pe_start,
+			    cur_pe_end, cur_dst, flags);
 }
 
 /**
  * amdgpu_vm_bo_update_mapping - update a mapping in the vm page table
  *
  * @adev: amdgpu_device pointer
+ * @exclusive: fence we need to sync to
  * @src: address where to copy page table entries from
  * @pages_addr: DMA addresses to use for mapping
  * @vm: requested vm
@@ -793,6 +888,7 @@ static void amdgpu_vm_update_ptes(struct amdgpu_device *adev,
  * Returns 0 for success, -EINVAL for failure.
  */
 static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
+				       struct fence *exclusive,
 				       uint64_t src,
 				       dma_addr_t *pages_addr,
 				       struct amdgpu_vm *vm,
@@ -853,6 +949,10 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
 
 	vm_update_params.ib = &job->ibs[0];
 
+	r = amdgpu_sync_fence(adev, &job->sync, exclusive);
+	if (r)
+		goto error_free;
+
 	r = amdgpu_sync_resv(adev, &job->sync, vm->page_directory->tbo.resv,
 			     owner);
 	if (r)
@@ -889,6 +989,7 @@ error_free:
  * amdgpu_vm_bo_split_mapping - split a mapping into smaller chunks
  *
  * @adev: amdgpu_device pointer
+ * @exclusive: fence we need to sync to
  * @gtt_flags: flags as they are used for GTT
  * @pages_addr: DMA addresses to use for mapping
  * @vm: requested vm
@@ -902,6 +1003,7 @@ error_free:
  * Returns 0 for success, -EINVAL for failure.
  */
 static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev,
+				      struct fence *exclusive,
 				      uint32_t gtt_flags,
 				      dma_addr_t *pages_addr,
 				      struct amdgpu_vm *vm,
@@ -932,7 +1034,8 @@ static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev,
 	addr += mapping->offset;
 
 	if (!pages_addr || src)
-		return amdgpu_vm_bo_update_mapping(adev, src, pages_addr, vm,
+		return amdgpu_vm_bo_update_mapping(adev, exclusive,
+						   src, pages_addr, vm,
 						   start, mapping->it.last,
 						   flags, addr, fence);
 
@@ -940,7 +1043,8 @@ static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev,
 		uint64_t last;
 
 		last = min((uint64_t)mapping->it.last, start + max_size - 1);
-		r = amdgpu_vm_bo_update_mapping(adev, src, pages_addr, vm,
+		r = amdgpu_vm_bo_update_mapping(adev, exclusive,
+						src, pages_addr, vm,
 						start, last, flags, addr,
 						fence);
 		if (r)
@@ -973,6 +1077,7 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 	struct amdgpu_bo_va_mapping *mapping;
 	dma_addr_t *pages_addr = NULL;
 	uint32_t gtt_flags, flags;
+	struct fence *exclusive;
 	uint64_t addr;
 	int r;
 
@@ -994,8 +1099,11 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 		default:
 			break;
 		}
+
+		exclusive = reservation_object_get_excl(bo_va->bo->tbo.resv);
 	} else {
 		addr = 0;
+		exclusive = NULL;
 	}
 
 	flags = amdgpu_ttm_tt_pte_flags(adev, bo_va->bo->tbo.ttm, mem);
@@ -1007,7 +1115,8 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 	spin_unlock(&vm->status_lock);
 
 	list_for_each_entry(mapping, &bo_va->invalids, list) {
-		r = amdgpu_vm_bo_split_mapping(adev, gtt_flags, pages_addr, vm,
+		r = amdgpu_vm_bo_split_mapping(adev, exclusive,
+					       gtt_flags, pages_addr, vm,
 					       mapping, flags, addr,
 					       &bo_va->last_pt_update);
 		if (r)
@@ -1054,7 +1163,7 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
 			struct amdgpu_bo_va_mapping, list);
 		list_del(&mapping->list);
 
-		r = amdgpu_vm_bo_split_mapping(adev, 0, NULL, vm, mapping,
+		r = amdgpu_vm_bo_split_mapping(adev, NULL, 0, NULL, vm, mapping,
 					       0, 0, NULL);
 		kfree(mapping);
 		if (r)
@@ -1426,7 +1535,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 	r = amd_sched_entity_init(&ring->sched, &vm->entity,
 				  rq, amdgpu_sched_jobs);
 	if (r)
-		return r;
+		goto err;
 
 	vm->page_directory_fence = NULL;
 
@@ -1445,6 +1554,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 	amdgpu_bo_unreserve(vm->page_directory);
 	if (r)
 		goto error_free_page_directory;
+	vm->last_eviction_counter = atomic64_read(&adev->num_evictions);
 
 	return 0;
 
@@ -1455,6 +1565,9 @@ error_free_page_directory:
 error_free_sched_entity:
 	amd_sched_entity_fini(&ring->sched, &vm->entity);
 
+err:
+	drm_free_large(vm->page_tables);
+
 	return r;
 }
 
@@ -1516,6 +1629,10 @@ void amdgpu_vm_manager_init(struct amdgpu_device *adev)
 			      &adev->vm_manager.ids_lru);
 	}
 
+	adev->vm_manager.fence_context = fence_context_alloc(AMDGPU_MAX_RINGS);
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
+		adev->vm_manager.seqno[i] = 0;
+
 	atomic_set(&adev->vm_manager.vm_pte_next_ring, 0);
 	atomic64_set(&adev->vm_manager.client_counter, 0);
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/ci_dpm.c b/drivers/gpu/drm/amd/amdgpu/ci_dpm.c
index 040dd26ff..43934248b 100644
--- a/drivers/gpu/drm/amd/amdgpu/ci_dpm.c
+++ b/drivers/gpu/drm/amd/amdgpu/ci_dpm.c
@@ -83,12 +83,14 @@ static const struct ci_pt_defaults defaults_bonaire_xt =
 	{ 0x17C, 0x172, 0x180, 0x1BC, 0x1B3, 0x1BD, 0x206, 0x200, 0x203, 0x25D, 0x25A, 0x255, 0x2C3, 0x2C5, 0x2B4 }
 };
 
+#if 0
 static const struct ci_pt_defaults defaults_bonaire_pro =
 {
 	1, 0xF, 0xFD, 0x19, 5, 45, 0, 0x65062,
 	{ 0x8C,  0x23F, 0x244, 0xA6,  0x83,  0x85,  0x86,  0x86,  0x83,  0xDB,  0xDB,  0xDA,  0x67,  0x60,  0x5F  },
 	{ 0x187, 0x193, 0x193, 0x1C7, 0x1D1, 0x1D1, 0x210, 0x219, 0x219, 0x266, 0x26C, 0x26C, 0x2C9, 0x2CB, 0x2CB }
 };
+#endif
 
 static const struct ci_pt_defaults defaults_saturn_xt =
 {
@@ -97,12 +99,14 @@ static const struct ci_pt_defaults defaults_saturn_xt =
 	{ 0x187, 0x187, 0x187, 0x1C7, 0x1C7, 0x1C7, 0x210, 0x210, 0x210, 0x266, 0x266, 0x266, 0x2C9, 0x2C9, 0x2C9 }
 };
 
+#if 0
 static const struct ci_pt_defaults defaults_saturn_pro =
 {
 	1, 0xF, 0xFD, 0x19, 5, 55, 0, 0x30000,
 	{ 0x96,  0x21D, 0x23B, 0xA1,  0x85,  0x87,  0x83,  0x84,  0x81,  0xE6,  0xE6,  0xE6,  0x71,  0x6A,  0x6A  },
 	{ 0x193, 0x19E, 0x19E, 0x1D2, 0x1DC, 0x1DC, 0x21A, 0x223, 0x223, 0x26E, 0x27E, 0x274, 0x2CF, 0x2D2, 0x2D2 }
 };
+#endif
 
 static const struct ci_pt_config_reg didt_config_ci[] =
 {
@@ -735,19 +739,19 @@ static int ci_enable_didt(struct amdgpu_device *adev, bool enable)
 
 	if (pi->caps_sq_ramping || pi->caps_db_ramping ||
 	    pi->caps_td_ramping || pi->caps_tcp_ramping) {
-		gfx_v7_0_enter_rlc_safe_mode(adev);
+		adev->gfx.rlc.funcs->enter_safe_mode(adev);
 
 		if (enable) {
 			ret = ci_program_pt_config_registers(adev, didt_config_ci);
 			if (ret) {
-				gfx_v7_0_exit_rlc_safe_mode(adev);
+				adev->gfx.rlc.funcs->exit_safe_mode(adev);
 				return ret;
 			}
 		}
 
 		ci_do_enable_didt(adev, enable);
 
-		gfx_v7_0_exit_rlc_safe_mode(adev);
+		adev->gfx.rlc.funcs->exit_safe_mode(adev);
 	}
 
 	return 0;
@@ -3029,7 +3033,7 @@ static int ci_populate_single_memory_level(struct amdgpu_device *adev,
 
 	if (pi->mclk_stutter_mode_threshold &&
 	    (memory_clock <= pi->mclk_stutter_mode_threshold) &&
-	    (pi->uvd_enabled == false) &&
+	    (!pi->uvd_enabled) &&
 	    (RREG32(mmDPG_PIPE_STUTTER_CONTROL) & DPG_PIPE_STUTTER_CONTROL__STUTTER_ENABLE_MASK) &&
 	    (adev->pm.dpm.new_active_crtc_count <= 2))
 		memory_level->StutterEnable = true;
@@ -3635,6 +3639,10 @@ static int ci_setup_default_dpm_tables(struct amdgpu_device *adev)
 
 	ci_setup_default_pcie_tables(adev);
 
+	/* save a copy of the default DPM table */
+	memcpy(&(pi->golden_dpm_table), &(pi->dpm_table),
+			sizeof(struct ci_dpm_table));
+
 	return 0;
 }
 
@@ -5753,13 +5761,22 @@ static int ci_dpm_init_microcode(struct amdgpu_device *adev)
 
 	switch (adev->asic_type) {
 	case CHIP_BONAIRE:
-		chip_name = "bonaire";
+		if ((adev->pdev->revision == 0x80) ||
+		    (adev->pdev->revision == 0x81) ||
+		    (adev->pdev->device == 0x665f))
+			chip_name = "bonaire_k";
+		else
+			chip_name = "bonaire";
 		break;
 	case CHIP_HAWAII:
-		chip_name = "hawaii";
+		if (adev->pdev->revision == 0x80)
+			chip_name = "hawaii_k";
+		else
+			chip_name = "hawaii";
 		break;
 	case CHIP_KAVERI:
 	case CHIP_KABINI:
+	case CHIP_MULLINS:
 	default: BUG();
 	}
 
@@ -6403,6 +6420,186 @@ static int ci_dpm_set_powergating_state(void *handle,
 	return 0;
 }
 
+static int ci_dpm_print_clock_levels(struct amdgpu_device *adev,
+		enum pp_clock_type type, char *buf)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_single_dpm_table *sclk_table = &pi->dpm_table.sclk_table;
+	struct ci_single_dpm_table *mclk_table = &pi->dpm_table.mclk_table;
+	struct ci_single_dpm_table *pcie_table = &pi->dpm_table.pcie_speed_table;
+
+	int i, now, size = 0;
+	uint32_t clock, pcie_speed;
+
+	switch (type) {
+	case PP_SCLK:
+		amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_API_GetSclkFrequency);
+		clock = RREG32(mmSMC_MSG_ARG_0);
+
+		for (i = 0; i < sclk_table->count; i++) {
+			if (clock > sclk_table->dpm_levels[i].value)
+				continue;
+			break;
+		}
+		now = i;
+
+		for (i = 0; i < sclk_table->count; i++)
+			size += sprintf(buf + size, "%d: %uMhz %s\n",
+					i, sclk_table->dpm_levels[i].value / 100,
+					(i == now) ? "*" : "");
+		break;
+	case PP_MCLK:
+		amdgpu_ci_send_msg_to_smc(adev, PPSMC_MSG_API_GetMclkFrequency);
+		clock = RREG32(mmSMC_MSG_ARG_0);
+
+		for (i = 0; i < mclk_table->count; i++) {
+			if (clock > mclk_table->dpm_levels[i].value)
+				continue;
+			break;
+		}
+		now = i;
+
+		for (i = 0; i < mclk_table->count; i++)
+			size += sprintf(buf + size, "%d: %uMhz %s\n",
+					i, mclk_table->dpm_levels[i].value / 100,
+					(i == now) ? "*" : "");
+		break;
+	case PP_PCIE:
+		pcie_speed = ci_get_current_pcie_speed(adev);
+		for (i = 0; i < pcie_table->count; i++) {
+			if (pcie_speed != pcie_table->dpm_levels[i].value)
+				continue;
+			break;
+		}
+		now = i;
+
+		for (i = 0; i < pcie_table->count; i++)
+			size += sprintf(buf + size, "%d: %s %s\n", i,
+					(pcie_table->dpm_levels[i].value == 0) ? "2.5GB, x1" :
+					(pcie_table->dpm_levels[i].value == 1) ? "5.0GB, x16" :
+					(pcie_table->dpm_levels[i].value == 2) ? "8.0GB, x16" : "",
+					(i == now) ? "*" : "");
+		break;
+	default:
+		break;
+	}
+
+	return size;
+}
+
+static int ci_dpm_force_clock_level(struct amdgpu_device *adev,
+		enum pp_clock_type type, uint32_t mask)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+
+	if (adev->pm.dpm.forced_level
+			!= AMDGPU_DPM_FORCED_LEVEL_MANUAL)
+		return -EINVAL;
+
+	switch (type) {
+	case PP_SCLK:
+		if (!pi->sclk_dpm_key_disabled)
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+					PPSMC_MSG_SCLKDPM_SetEnabledMask,
+					pi->dpm_level_enable_mask.sclk_dpm_enable_mask & mask);
+		break;
+
+	case PP_MCLK:
+		if (!pi->mclk_dpm_key_disabled)
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+					PPSMC_MSG_MCLKDPM_SetEnabledMask,
+					pi->dpm_level_enable_mask.mclk_dpm_enable_mask & mask);
+		break;
+
+	case PP_PCIE:
+	{
+		uint32_t tmp = mask & pi->dpm_level_enable_mask.pcie_dpm_enable_mask;
+		uint32_t level = 0;
+
+		while (tmp >>= 1)
+			level++;
+
+		if (!pi->pcie_dpm_key_disabled)
+			amdgpu_ci_send_msg_to_smc_with_parameter(adev,
+					PPSMC_MSG_PCIeDPM_ForceLevel,
+					level);
+		break;
+	}
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+static int ci_dpm_get_sclk_od(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_single_dpm_table *sclk_table = &(pi->dpm_table.sclk_table);
+	struct ci_single_dpm_table *golden_sclk_table =
+			&(pi->golden_dpm_table.sclk_table);
+	int value;
+
+	value = (sclk_table->dpm_levels[sclk_table->count - 1].value -
+			golden_sclk_table->dpm_levels[golden_sclk_table->count - 1].value) *
+			100 /
+			golden_sclk_table->dpm_levels[golden_sclk_table->count - 1].value;
+
+	return value;
+}
+
+static int ci_dpm_set_sclk_od(struct amdgpu_device *adev, uint32_t value)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ps *ps = ci_get_ps(adev->pm.dpm.requested_ps);
+	struct ci_single_dpm_table *golden_sclk_table =
+			&(pi->golden_dpm_table.sclk_table);
+
+	if (value > 20)
+		value = 20;
+
+	ps->performance_levels[ps->performance_level_count - 1].sclk =
+			golden_sclk_table->dpm_levels[golden_sclk_table->count - 1].value *
+			value / 100 +
+			golden_sclk_table->dpm_levels[golden_sclk_table->count - 1].value;
+
+	return 0;
+}
+
+static int ci_dpm_get_mclk_od(struct amdgpu_device *adev)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_single_dpm_table *mclk_table = &(pi->dpm_table.mclk_table);
+	struct ci_single_dpm_table *golden_mclk_table =
+			&(pi->golden_dpm_table.mclk_table);
+	int value;
+
+	value = (mclk_table->dpm_levels[mclk_table->count - 1].value -
+			golden_mclk_table->dpm_levels[golden_mclk_table->count - 1].value) *
+			100 /
+			golden_mclk_table->dpm_levels[golden_mclk_table->count - 1].value;
+
+	return value;
+}
+
+static int ci_dpm_set_mclk_od(struct amdgpu_device *adev, uint32_t value)
+{
+	struct ci_power_info *pi = ci_get_pi(adev);
+	struct ci_ps *ps = ci_get_ps(adev->pm.dpm.requested_ps);
+	struct ci_single_dpm_table *golden_mclk_table =
+			&(pi->golden_dpm_table.mclk_table);
+
+	if (value > 20)
+		value = 20;
+
+	ps->performance_levels[ps->performance_level_count - 1].mclk =
+			golden_mclk_table->dpm_levels[golden_mclk_table->count - 1].value *
+			value / 100 +
+			golden_mclk_table->dpm_levels[golden_mclk_table->count - 1].value;
+
+	return 0;
+}
+
 const struct amd_ip_funcs ci_dpm_ip_funcs = {
 	.name = "ci_dpm",
 	.early_init = ci_dpm_early_init,
@@ -6437,6 +6634,12 @@ static const struct amdgpu_dpm_funcs ci_dpm_funcs = {
 	.get_fan_control_mode = &ci_dpm_get_fan_control_mode,
 	.set_fan_speed_percent = &ci_dpm_set_fan_speed_percent,
 	.get_fan_speed_percent = &ci_dpm_get_fan_speed_percent,
+	.print_clock_levels = ci_dpm_print_clock_levels,
+	.force_clock_level = ci_dpm_force_clock_level,
+	.get_sclk_od = ci_dpm_get_sclk_od,
+	.set_sclk_od = ci_dpm_set_sclk_od,
+	.get_mclk_od = ci_dpm_get_mclk_od,
+	.set_mclk_od = ci_dpm_set_mclk_od,
 };
 
 static void ci_dpm_set_dpm_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/ci_dpm.h b/drivers/gpu/drm/amd/amdgpu/ci_dpm.h
index faccc30c9..91be2996a 100644
--- a/drivers/gpu/drm/amd/amdgpu/ci_dpm.h
+++ b/drivers/gpu/drm/amd/amdgpu/ci_dpm.h
@@ -193,6 +193,7 @@ struct ci_pt_defaults {
 
 struct ci_power_info {
 	struct ci_dpm_table dpm_table;
+	struct ci_dpm_table golden_dpm_table;
 	u32 voltage_control;
 	u32 mvdd_control;
 	u32 vddci_control;
diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c
index 910431808..4efc901f6 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik.c
@@ -879,7 +879,7 @@ static void cik_vga_set_state(struct amdgpu_device *adev, bool state)
 	uint32_t tmp;
 
 	tmp = RREG32(mmCONFIG_CNTL);
-	if (state == false)
+	if (!state)
 		tmp |= CONFIG_CNTL__VGA_DIS_MASK;
 	else
 		tmp &= ~CONFIG_CNTL__VGA_DIS_MASK;
@@ -1035,12 +1035,12 @@ static uint32_t cik_read_indexed_register(struct amdgpu_device *adev,
 
 	mutex_lock(&adev->grbm_idx_mutex);
 	if (se_num != 0xffffffff || sh_num != 0xffffffff)
-		gfx_v7_0_select_se_sh(adev, se_num, sh_num);
+		amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
 
 	val = RREG32(reg_offset);
 
 	if (se_num != 0xffffffff || sh_num != 0xffffffff)
-		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+		amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 	mutex_unlock(&adev->grbm_idx_mutex);
 	return val;
 }
@@ -1158,10 +1158,11 @@ static void kv_restore_regs_for_reset(struct amdgpu_device *adev,
 	WREG32(mmGMCON_RENG_EXECUTE, save->gmcon_reng_execute);
 }
 
-static void cik_gpu_pci_config_reset(struct amdgpu_device *adev)
+static int cik_gpu_pci_config_reset(struct amdgpu_device *adev)
 {
 	struct kv_reset_save_regs kv_save = { 0 };
 	u32 i;
+	int r = -EINVAL;
 
 	dev_info(adev->dev, "GPU pci config reset\n");
 
@@ -1177,14 +1178,20 @@ static void cik_gpu_pci_config_reset(struct amdgpu_device *adev)
 
 	/* wait for asic to come out of reset */
 	for (i = 0; i < adev->usec_timeout; i++) {
-		if (RREG32(mmCONFIG_MEMSIZE) != 0xffffffff)
+		if (RREG32(mmCONFIG_MEMSIZE) != 0xffffffff) {
+			/* enable BM */
+			pci_set_master(adev->pdev);
+			r = 0;
 			break;
+		}
 		udelay(1);
 	}
 
 	/* does asic init need to be run first??? */
 	if (adev->flags & AMD_IS_APU)
 		kv_restore_regs_for_reset(adev, &kv_save);
+
+	return r;
 }
 
 static void cik_set_bios_scratch_engine_hung(struct amdgpu_device *adev, bool hung)
@@ -1210,13 +1217,14 @@ static void cik_set_bios_scratch_engine_hung(struct amdgpu_device *adev, bool hu
  */
 static int cik_asic_reset(struct amdgpu_device *adev)
 {
+	int r;
 	cik_set_bios_scratch_engine_hung(adev, true);
 
-	cik_gpu_pci_config_reset(adev);
+	r = cik_gpu_pci_config_reset(adev);
 
 	cik_set_bios_scratch_engine_hung(adev, false);
 
-	return 0;
+	return r;
 }
 
 static int cik_set_uvd_clock(struct amdgpu_device *adev, u32 clock,
@@ -2014,9 +2022,6 @@ static const struct amdgpu_asic_funcs cik_asic_funcs =
 	.set_uvd_clocks = &cik_set_uvd_clocks,
 	.set_vce_clocks = &cik_set_vce_clocks,
 	.get_virtual_caps = &cik_get_virtual_caps,
-	/* these should be moved to their own ip modules */
-	.get_gpu_clock_counter = &gfx_v7_0_get_gpu_clock_counter,
-	.wait_for_mc_idle = &gmc_v7_0_mc_wait_for_idle,
 };
 
 static int cik_common_early_init(void *handle)
diff --git a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
index cb429bfdc..af8b6bdf0 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
@@ -216,17 +216,6 @@ static void cik_sdma_ring_emit_ib(struct amdgpu_ring *ring,
 				  unsigned vm_id, bool ctx_switch)
 {
 	u32 extra_bits = vm_id & 0xf;
-	u32 next_rptr = ring->wptr + 5;
-
-	while ((next_rptr & 7) != 4)
-		next_rptr++;
-
-	next_rptr += 4;
-	amdgpu_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_WRITE, SDMA_WRITE_SUB_OPCODE_LINEAR, 0));
-	amdgpu_ring_write(ring, ring->next_rptr_gpu_addr & 0xfffffffc);
-	amdgpu_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr) & 0xffffffff);
-	amdgpu_ring_write(ring, 1); /* number of DWs to follow */
-	amdgpu_ring_write(ring, next_rptr);
 
 	/* IB packet must end on a 8 DW boundary */
 	cik_sdma_ring_insert_nop(ring, (12 - (ring->wptr & 7)) % 8);
@@ -357,7 +346,7 @@ static void cik_sdma_enable(struct amdgpu_device *adev, bool enable)
 	u32 me_cntl;
 	int i;
 
-	if (enable == false) {
+	if (!enable) {
 		cik_sdma_gfx_stop(adev);
 		cik_sdma_rlc_stop(adev);
 	}
@@ -620,20 +609,19 @@ static int cik_sdma_ring_test_ring(struct amdgpu_ring *ring)
  * Test a simple IB in the DMA ring (CIK).
  * Returns 0 on success, error on failure.
  */
-static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring)
+static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_ib ib;
 	struct fence *f = NULL;
-	unsigned i;
 	unsigned index;
-	int r;
 	u32 tmp = 0;
 	u64 gpu_addr;
+	long r;
 
 	r = amdgpu_wb_get(adev, &index);
 	if (r) {
-		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
 		return r;
 	}
 
@@ -643,11 +631,12 @@ static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring)
 	memset(&ib, 0, sizeof(ib));
 	r = amdgpu_ib_get(adev, NULL, 256, &ib);
 	if (r) {
-		DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
 		goto err0;
 	}
 
-	ib.ptr[0] = SDMA_PACKET(SDMA_OPCODE_WRITE, SDMA_WRITE_SUB_OPCODE_LINEAR, 0);
+	ib.ptr[0] = SDMA_PACKET(SDMA_OPCODE_WRITE,
+				SDMA_WRITE_SUB_OPCODE_LINEAR, 0);
 	ib.ptr[1] = lower_32_bits(gpu_addr);
 	ib.ptr[2] = upper_32_bits(gpu_addr);
 	ib.ptr[3] = 1;
@@ -657,28 +646,25 @@ static int cik_sdma_ring_test_ib(struct amdgpu_ring *ring)
 	if (r)
 		goto err1;
 
-	r = fence_wait(f, false);
-	if (r) {
-		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+	r = fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out\n");
+		r = -ETIMEDOUT;
 		goto err1;
-	}
-	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = le32_to_cpu(adev->wb.wb[index]);
-		if (tmp == 0xDEADBEEF)
-			break;
-		DRM_UDELAY(1);
-	}
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ib test on ring %d succeeded in %u usecs\n",
-			 ring->idx, i);
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
 		goto err1;
+	}
+	tmp = le32_to_cpu(adev->wb.wb[index]);
+	if (tmp == 0xDEADBEEF) {
+		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
 	} else {
 		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
 		r = -EINVAL;
 	}
 
 err1:
-	fence_put(f);
 	amdgpu_ib_free(adev, &ib, NULL);
 	fence_put(f);
 err0:
diff --git a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c
index 933e425a8..2a11413ed 100644
--- a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c
+++ b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c
@@ -425,7 +425,7 @@ static int cz_dpm_init(struct amdgpu_device *adev)
 	pi->mgcg_cgtt_local1 = 0x0;
 	pi->clock_slow_down_step = 25000;
 	pi->skip_clock_slow_down = 1;
-	pi->enable_nb_ps_policy = 0;
+	pi->enable_nb_ps_policy = false;
 	pi->caps_power_containment = true;
 	pi->caps_cac = true;
 	pi->didt_enabled = false;
@@ -2219,6 +2219,7 @@ static void cz_dpm_powergate_vce(struct amdgpu_device *adev, bool gate)
 			}
 		}
 	} else { /*pi->caps_vce_pg*/
+		pi->vce_power_gated = gate;
 		cz_update_vce_dpm(adev);
 		cz_enable_vce_dpm(adev, !gate);
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
index 8227344d2..c1b04e9aa 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
@@ -2667,19 +2667,21 @@ static void dce_v10_0_cursor_reset(struct drm_crtc *crtc)
 	}
 }
 
-static void dce_v10_0_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
-				    u16 *blue, uint32_t start, uint32_t size)
+static int dce_v10_0_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
+				    u16 *blue, uint32_t size)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	int end = (start + size > 256) ? 256 : start + size, i;
+	int i;
 
 	/* userspace palettes are always correct as is */
-	for (i = start; i < end; i++) {
+	for (i = 0; i < size; i++) {
 		amdgpu_crtc->lut_r[i] = red[i] >> 6;
 		amdgpu_crtc->lut_g[i] = green[i] >> 6;
 		amdgpu_crtc->lut_b[i] = blue[i] >> 6;
 	}
 	dce_v10_0_crtc_load_lut(crtc);
+
+	return 0;
 }
 
 static void dce_v10_0_crtc_destroy(struct drm_crtc *crtc)
@@ -2717,13 +2719,13 @@ static void dce_v10_0_crtc_dpms(struct drm_crtc *crtc, int mode)
 		type = amdgpu_crtc_idx_to_irq_type(adev, amdgpu_crtc->crtc_id);
 		amdgpu_irq_update(adev, &adev->crtc_irq, type);
 		amdgpu_irq_update(adev, &adev->pageflip_irq, type);
-		drm_vblank_on(dev, amdgpu_crtc->crtc_id);
+		drm_crtc_vblank_on(crtc);
 		dce_v10_0_crtc_load_lut(crtc);
 		break;
 	case DRM_MODE_DPMS_STANDBY:
 	case DRM_MODE_DPMS_SUSPEND:
 	case DRM_MODE_DPMS_OFF:
-		drm_vblank_off(dev, amdgpu_crtc->crtc_id);
+		drm_crtc_vblank_off(crtc);
 		if (amdgpu_crtc->enabled) {
 			dce_v10_0_vga_enable(crtc, true);
 			amdgpu_atombios_crtc_blank(crtc, ATOM_ENABLE);
@@ -3372,7 +3374,7 @@ static int dce_v10_0_pageflip_irq(struct amdgpu_device *adev,
 
 	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
 
-	drm_vblank_put(adev->ddev, amdgpu_crtc->crtc_id);
+	drm_crtc_vblank_put(&amdgpu_crtc->base);
 	schedule_work(&works->unpin_work);
 
 	return 0;
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
index af26ec0bc..d4bf13390 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
@@ -307,11 +307,10 @@ static void dce_v11_0_page_flip(struct amdgpu_device *adev,
 	struct amdgpu_crtc *amdgpu_crtc = adev->mode_info.crtcs[crtc_id];
 	u32 tmp;
 
-	/* flip at hsync for async, default is vsync */
-	/* use UPDATE_IMMEDIATE_EN instead for async? */
+	/* flip immediate for async, default is vsync */
 	tmp = RREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset);
 	tmp = REG_SET_FIELD(tmp, GRPH_FLIP_CONTROL,
-			    GRPH_SURFACE_UPDATE_H_RETRACE_EN, async ? 1 : 0);
+			    GRPH_SURFACE_UPDATE_IMMEDIATE_EN, async ? 1 : 0);
 	WREG32(mmGRPH_FLIP_CONTROL + amdgpu_crtc->crtc_offset, tmp);
 	/* update the scanout addresses */
 	WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + amdgpu_crtc->crtc_offset,
@@ -2678,19 +2677,21 @@ static void dce_v11_0_cursor_reset(struct drm_crtc *crtc)
 	}
 }
 
-static void dce_v11_0_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
-				    u16 *blue, uint32_t start, uint32_t size)
+static int dce_v11_0_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
+				    u16 *blue, uint32_t size)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	int end = (start + size > 256) ? 256 : start + size, i;
+	int i;
 
 	/* userspace palettes are always correct as is */
-	for (i = start; i < end; i++) {
+	for (i = 0; i < size; i++) {
 		amdgpu_crtc->lut_r[i] = red[i] >> 6;
 		amdgpu_crtc->lut_g[i] = green[i] >> 6;
 		amdgpu_crtc->lut_b[i] = blue[i] >> 6;
 	}
 	dce_v11_0_crtc_load_lut(crtc);
+
+	return 0;
 }
 
 static void dce_v11_0_crtc_destroy(struct drm_crtc *crtc)
@@ -2728,13 +2729,13 @@ static void dce_v11_0_crtc_dpms(struct drm_crtc *crtc, int mode)
 		type = amdgpu_crtc_idx_to_irq_type(adev, amdgpu_crtc->crtc_id);
 		amdgpu_irq_update(adev, &adev->crtc_irq, type);
 		amdgpu_irq_update(adev, &adev->pageflip_irq, type);
-		drm_vblank_on(dev, amdgpu_crtc->crtc_id);
+		drm_crtc_vblank_on(crtc);
 		dce_v11_0_crtc_load_lut(crtc);
 		break;
 	case DRM_MODE_DPMS_STANDBY:
 	case DRM_MODE_DPMS_SUSPEND:
 	case DRM_MODE_DPMS_OFF:
-		drm_vblank_off(dev, amdgpu_crtc->crtc_id);
+		drm_crtc_vblank_off(crtc);
 		if (amdgpu_crtc->enabled) {
 			dce_v11_0_vga_enable(crtc, true);
 			amdgpu_atombios_crtc_blank(crtc, ATOM_ENABLE);
@@ -3433,7 +3434,7 @@ static int dce_v11_0_pageflip_irq(struct amdgpu_device *adev,
 
 	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
 
-	drm_vblank_put(adev->ddev, amdgpu_crtc->crtc_id);
+	drm_crtc_vblank_put(&amdgpu_crtc->base);
 	schedule_work(&works->unpin_work);
 
 	return 0;
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
index 3fb65e41a..4fdfab1e9 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
@@ -526,36 +526,16 @@ static void dce_v8_0_stop_mc_access(struct amdgpu_device *adev,
 		crtc_enabled = REG_GET_FIELD(RREG32(mmCRTC_CONTROL + crtc_offsets[i]),
 					     CRTC_CONTROL, CRTC_MASTER_EN);
 		if (crtc_enabled) {
-#if 0
-			u32 frame_count;
-			int j;
-
+#if 1
 			save->crtc_enabled[i] = true;
 			tmp = RREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i]);
 			if (REG_GET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN) == 0) {
-				amdgpu_display_vblank_wait(adev, i);
-				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
+				/*it is correct only for RGB ; black is 0*/
+				WREG32(mmCRTC_BLANK_DATA_COLOR + crtc_offsets[i], 0);
 				tmp = REG_SET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN, 1);
 				WREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i], tmp);
-				WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
-			}
-			/* wait for the next frame */
-			frame_count = amdgpu_display_vblank_get_counter(adev, i);
-			for (j = 0; j < adev->usec_timeout; j++) {
-				if (amdgpu_display_vblank_get_counter(adev, i) != frame_count)
-					break;
-				udelay(1);
-			}
-			tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
-			if (REG_GET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK) == 0) {
-				tmp = REG_SET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK, 1);
-				WREG32(mmGRPH_UPDATE + crtc_offsets[i], tmp);
-			}
-			tmp = RREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i]);
-			if (REG_GET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK) == 0) {
-				tmp = REG_SET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK, 1);
-				WREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i], tmp);
 			}
+			mdelay(20);
 #else
 			/* XXX this is a hack to avoid strange behavior with EFI on certain systems */
 			WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
@@ -575,55 +555,22 @@ static void dce_v8_0_stop_mc_access(struct amdgpu_device *adev,
 static void dce_v8_0_resume_mc_access(struct amdgpu_device *adev,
 				      struct amdgpu_mode_mc_save *save)
 {
-	u32 tmp, frame_count;
-	int i, j;
+	u32 tmp;
+	int i;
 
 	/* update crtc base addresses */
 	for (i = 0; i < adev->mode_info.num_crtc; i++) {
 		WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + crtc_offsets[i],
 		       upper_32_bits(adev->mc.vram_start));
-		WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS_HIGH + crtc_offsets[i],
-		       upper_32_bits(adev->mc.vram_start));
 		WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + crtc_offsets[i],
 		       (u32)adev->mc.vram_start);
-		WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + crtc_offsets[i],
-		       (u32)adev->mc.vram_start);
 
 		if (save->crtc_enabled[i]) {
-			tmp = RREG32(mmMASTER_UPDATE_MODE + crtc_offsets[i]);
-			if (REG_GET_FIELD(tmp, MASTER_UPDATE_MODE, MASTER_UPDATE_MODE) != 3) {
-				tmp = REG_SET_FIELD(tmp, MASTER_UPDATE_MODE, MASTER_UPDATE_MODE, 3);
-				WREG32(mmMASTER_UPDATE_MODE + crtc_offsets[i], tmp);
-			}
-			tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
-			if (REG_GET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK)) {
-				tmp = REG_SET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK, 0);
-				WREG32(mmGRPH_UPDATE + crtc_offsets[i], tmp);
-			}
-			tmp = RREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i]);
-			if (REG_GET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK)) {
-				tmp = REG_SET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK, 0);
-				WREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i], tmp);
-			}
-			for (j = 0; j < adev->usec_timeout; j++) {
-				tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
-				if (REG_GET_FIELD(tmp, GRPH_UPDATE, GRPH_SURFACE_UPDATE_PENDING) == 0)
-					break;
-				udelay(1);
-			}
 			tmp = RREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i]);
 			tmp = REG_SET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN, 0);
-			WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
 			WREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i], tmp);
-			WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
-			/* wait for the next frame */
-			frame_count = amdgpu_display_vblank_get_counter(adev, i);
-			for (j = 0; j < adev->usec_timeout; j++) {
-				if (amdgpu_display_vblank_get_counter(adev, i) != frame_count)
-					break;
-				udelay(1);
-			}
 		}
+		mdelay(20);
 	}
 
 	WREG32(mmVGA_MEMORY_BASE_ADDRESS_HIGH, upper_32_bits(adev->mc.vram_start));
@@ -2574,19 +2521,21 @@ static void dce_v8_0_cursor_reset(struct drm_crtc *crtc)
 	}
 }
 
-static void dce_v8_0_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
-				    u16 *blue, uint32_t start, uint32_t size)
+static int dce_v8_0_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, u16 *green,
+				   u16 *blue, uint32_t size)
 {
 	struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
-	int end = (start + size > 256) ? 256 : start + size, i;
+	int i;
 
 	/* userspace palettes are always correct as is */
-	for (i = start; i < end; i++) {
+	for (i = 0; i < size; i++) {
 		amdgpu_crtc->lut_r[i] = red[i] >> 6;
 		amdgpu_crtc->lut_g[i] = green[i] >> 6;
 		amdgpu_crtc->lut_b[i] = blue[i] >> 6;
 	}
 	dce_v8_0_crtc_load_lut(crtc);
+
+	return 0;
 }
 
 static void dce_v8_0_crtc_destroy(struct drm_crtc *crtc)
@@ -2624,13 +2573,13 @@ static void dce_v8_0_crtc_dpms(struct drm_crtc *crtc, int mode)
 		type = amdgpu_crtc_idx_to_irq_type(adev, amdgpu_crtc->crtc_id);
 		amdgpu_irq_update(adev, &adev->crtc_irq, type);
 		amdgpu_irq_update(adev, &adev->pageflip_irq, type);
-		drm_vblank_on(dev, amdgpu_crtc->crtc_id);
+		drm_crtc_vblank_on(crtc);
 		dce_v8_0_crtc_load_lut(crtc);
 		break;
 	case DRM_MODE_DPMS_STANDBY:
 	case DRM_MODE_DPMS_SUSPEND:
 	case DRM_MODE_DPMS_OFF:
-		drm_vblank_off(dev, amdgpu_crtc->crtc_id);
+		drm_crtc_vblank_off(crtc);
 		if (amdgpu_crtc->enabled) {
 			dce_v8_0_vga_enable(crtc, true);
 			amdgpu_atombios_crtc_blank(crtc, ATOM_ENABLE);
@@ -3376,7 +3325,7 @@ static int dce_v8_0_pageflip_irq(struct amdgpu_device *adev,
 
 	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
 
-	drm_vblank_put(adev->ddev, amdgpu_crtc->crtc_id);
+	drm_crtc_vblank_put(&amdgpu_crtc->base);
 	schedule_work(&works->unpin_work);
 
 	return 0;
diff --git a/drivers/gpu/drm/amd/amdgpu/fiji_smc.c b/drivers/gpu/drm/amd/amdgpu/fiji_smc.c
index b336c918d..b3e19ba4c 100644
--- a/drivers/gpu/drm/amd/amdgpu/fiji_smc.c
+++ b/drivers/gpu/drm/amd/amdgpu/fiji_smc.c
@@ -173,7 +173,7 @@ static int fiji_send_msg_to_smc(struct amdgpu_device *adev, PPSMC_Msg msg)
 {
 	if (!fiji_is_smc_ram_running(adev))
 	{
-		return -EINVAL;;
+		return -EINVAL;
 	}
 
 	if (wait_smu_response(adev)) {
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
index 6d34dcd60..754566f51 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
@@ -1554,9 +1554,15 @@ static void gfx_v7_0_tiling_mode_table_init(struct amdgpu_device *adev)
  * registers are instanced per SE or SH.  0xffffffff means
  * broadcast to all SEs or SHs (CIK).
  */
-void gfx_v7_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num)
+static void gfx_v7_0_select_se_sh(struct amdgpu_device *adev,
+				  u32 se_num, u32 sh_num, u32 instance)
 {
-	u32 data = GRBM_GFX_INDEX__INSTANCE_BROADCAST_WRITES_MASK;
+	u32 data;
+
+	if (instance == 0xffffffff)
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_INDEX, instance);
 
 	if ((se_num == 0xffffffff) && (sh_num == 0xffffffff))
 		data |= GRBM_GFX_INDEX__SH_BROADCAST_WRITES_MASK |
@@ -1630,13 +1636,13 @@ static void gfx_v7_0_setup_rb(struct amdgpu_device *adev)
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-			gfx_v7_0_select_se_sh(adev, i, j);
+			gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff);
 			data = gfx_v7_0_get_rb_active_bitmap(adev);
 			active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) *
 					       rb_bitmap_width_per_sh);
 		}
 	}
-	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
 	adev->gfx.config.backend_enable_mask = active_rbs;
@@ -1717,7 +1723,7 @@ static void gfx_v7_0_gpu_init(struct amdgpu_device *adev)
 	 * making sure that the following register writes will be broadcasted
 	 * to all the shaders
 	 */
-	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 
 	/* XXX SH_MEM regs */
 	/* where to put LDS, scratch, GPUVM in FSA64 space */
@@ -2021,17 +2027,6 @@ static void gfx_v7_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
 				      unsigned vm_id, bool ctx_switch)
 {
 	u32 header, control = 0;
-	u32 next_rptr = ring->wptr + 5;
-
-	if (ctx_switch)
-		next_rptr += 2;
-
-	next_rptr += 4;
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, WRITE_DATA_DST_SEL(5) | WR_CONFIRM);
-	amdgpu_ring_write(ring, ring->next_rptr_gpu_addr & 0xfffffffc);
-	amdgpu_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr) & 0xffffffff);
-	amdgpu_ring_write(ring, next_rptr);
 
 	/* insert SWITCH_BUFFER packet before first IB in the ring frame */
 	if (ctx_switch) {
@@ -2060,22 +2055,9 @@ static void gfx_v7_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
 					  struct amdgpu_ib *ib,
 					  unsigned vm_id, bool ctx_switch)
 {
-	u32 header, control = 0;
-	u32 next_rptr = ring->wptr + 5;
-
-	control |= INDIRECT_BUFFER_VALID;
-	next_rptr += 4;
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, WRITE_DATA_DST_SEL(5) | WR_CONFIRM);
-	amdgpu_ring_write(ring, ring->next_rptr_gpu_addr & 0xfffffffc);
-	amdgpu_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr) & 0xffffffff);
-	amdgpu_ring_write(ring, next_rptr);
-
-	header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
+	u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vm_id << 24);
 
-	control |= ib->length_dw | (vm_id << 24);
-
-	amdgpu_ring_write(ring, header);
+	amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
 	amdgpu_ring_write(ring,
 #ifdef __BIG_ENDIAN
 					  (2 << 0) |
@@ -2094,26 +2076,25 @@ static void gfx_v7_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
  * Provides a basic gfx ring test to verify that IBs are working.
  * Returns 0 on success, error on failure.
  */
-static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring)
+static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_ib ib;
 	struct fence *f = NULL;
 	uint32_t scratch;
 	uint32_t tmp = 0;
-	unsigned i;
-	int r;
+	long r;
 
 	r = amdgpu_gfx_scratch_get(adev, &scratch);
 	if (r) {
-		DRM_ERROR("amdgpu: failed to get scratch reg (%d).\n", r);
+		DRM_ERROR("amdgpu: failed to get scratch reg (%ld).\n", r);
 		return r;
 	}
 	WREG32(scratch, 0xCAFEDEAD);
 	memset(&ib, 0, sizeof(ib));
 	r = amdgpu_ib_get(adev, NULL, 256, &ib);
 	if (r) {
-		DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
 		goto err1;
 	}
 	ib.ptr[0] = PACKET3(PACKET3_SET_UCONFIG_REG, 1);
@@ -2125,21 +2106,19 @@ static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring)
 	if (r)
 		goto err2;
 
-	r = fence_wait(f, false);
-	if (r) {
-		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+	r = fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out\n");
+		r = -ETIMEDOUT;
 		goto err2;
-	}
-	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(scratch);
-		if (tmp == 0xDEADBEEF)
-			break;
-		DRM_UDELAY(1);
-	}
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ib test on ring %d succeeded in %u usecs\n",
-			 ring->idx, i);
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
 		goto err2;
+	}
+	tmp = RREG32(scratch);
+	if (tmp == 0xDEADBEEF) {
+		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
 	} else {
 		DRM_ERROR("amdgpu: ib test failed (scratch(0x%04X)=0x%08X)\n",
 			  scratch, tmp);
@@ -2147,7 +2126,6 @@ static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring)
 	}
 
 err2:
-	fence_put(f);
 	amdgpu_ib_free(adev, &ib, NULL);
 	fence_put(f);
 err1:
@@ -3198,7 +3176,8 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
 		}
 	}
 	adev->gfx.rlc.cs_data = ci_cs_data;
-	adev->gfx.rlc.cp_table_size = CP_ME_TABLE_SIZE * 5 * 4;
+	adev->gfx.rlc.cp_table_size = ALIGN(CP_ME_TABLE_SIZE * 5 * 4, 2048); /* CP JT */
+	adev->gfx.rlc.cp_table_size += 64 * 1024; /* GDS */
 
 	src_ptr = adev->gfx.rlc.reg_list;
 	dws = adev->gfx.rlc.reg_list_size;
@@ -3356,7 +3335,7 @@ static void gfx_v7_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-			gfx_v7_0_select_se_sh(adev, i, j);
+			gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff);
 			for (k = 0; k < adev->usec_timeout; k++) {
 				if (RREG32(mmRLC_SERDES_CU_MASTER_BUSY) == 0)
 					break;
@@ -3364,7 +3343,7 @@ static void gfx_v7_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
 			}
 		}
 	}
-	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
 	mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
@@ -3411,7 +3390,7 @@ static u32 gfx_v7_0_halt_rlc(struct amdgpu_device *adev)
 	return orig;
 }
 
-void gfx_v7_0_enter_rlc_safe_mode(struct amdgpu_device *adev)
+static void gfx_v7_0_enter_rlc_safe_mode(struct amdgpu_device *adev)
 {
 	u32 tmp, i, mask;
 
@@ -3433,7 +3412,7 @@ void gfx_v7_0_enter_rlc_safe_mode(struct amdgpu_device *adev)
 	}
 }
 
-void gfx_v7_0_exit_rlc_safe_mode(struct amdgpu_device *adev)
+static void gfx_v7_0_exit_rlc_safe_mode(struct amdgpu_device *adev)
 {
 	u32 tmp;
 
@@ -3448,7 +3427,7 @@ void gfx_v7_0_exit_rlc_safe_mode(struct amdgpu_device *adev)
  *
  * Halt the RLC ME (MicroEngine) (CIK).
  */
-void gfx_v7_0_rlc_stop(struct amdgpu_device *adev)
+static void gfx_v7_0_rlc_stop(struct amdgpu_device *adev)
 {
 	WREG32(mmRLC_CNTL, 0);
 
@@ -3524,7 +3503,7 @@ static int gfx_v7_0_rlc_resume(struct amdgpu_device *adev)
 	WREG32(mmRLC_LB_CNTR_MAX, 0x00008000);
 
 	mutex_lock(&adev->grbm_idx_mutex);
-	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 	WREG32(mmRLC_LB_INIT_CU_MASK, 0xffffffff);
 	WREG32(mmRLC_LB_PARAMS, 0x00600408);
 	WREG32(mmRLC_LB_CNTL, 0x80000004);
@@ -3564,7 +3543,7 @@ static void gfx_v7_0_enable_cgcg(struct amdgpu_device *adev, bool enable)
 		tmp = gfx_v7_0_halt_rlc(adev);
 
 		mutex_lock(&adev->grbm_idx_mutex);
-		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 		WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
 		WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
 		tmp2 = RLC_SERDES_WR_CTRL__BPM_ADDR_MASK |
@@ -3615,7 +3594,7 @@ static void gfx_v7_0_enable_mgcg(struct amdgpu_device *adev, bool enable)
 		tmp = gfx_v7_0_halt_rlc(adev);
 
 		mutex_lock(&adev->grbm_idx_mutex);
-		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 		WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
 		WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
 		data = RLC_SERDES_WR_CTRL__BPM_ADDR_MASK |
@@ -3666,7 +3645,7 @@ static void gfx_v7_0_enable_mgcg(struct amdgpu_device *adev, bool enable)
 		tmp = gfx_v7_0_halt_rlc(adev);
 
 		mutex_lock(&adev->grbm_idx_mutex);
-		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+		gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 		WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
 		WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
 		data = RLC_SERDES_WR_CTRL__BPM_ADDR_MASK | RLC_SERDES_WR_CTRL__MGCG_OVERRIDE_1_MASK;
@@ -3844,6 +3823,20 @@ static void gfx_v7_0_enable_gfx_cgpg(struct amdgpu_device *adev,
 	}
 }
 
+static void gfx_v7_0_set_user_cu_inactive_bitmap(struct amdgpu_device *adev,
+						 u32 bitmap)
+{
+	u32 data;
+
+	if (!bitmap)
+		return;
+
+	data = bitmap << GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS__SHIFT;
+	data &= GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS_MASK;
+
+	WREG32(mmGC_USER_SHADER_ARRAY_CONFIG, data);
+}
+
 static u32 gfx_v7_0_get_cu_active_bitmap(struct amdgpu_device *adev)
 {
 	u32 data, mask;
@@ -4100,7 +4093,7 @@ static void gfx_v7_0_fini_pg(struct amdgpu_device *adev)
  * Fetches a GPU clock counter snapshot (SI).
  * Returns the 64 bit clock counter snapshot.
  */
-uint64_t gfx_v7_0_get_gpu_clock_counter(struct amdgpu_device *adev)
+static uint64_t gfx_v7_0_get_gpu_clock_counter(struct amdgpu_device *adev)
 {
 	uint64_t clock;
 
@@ -4160,12 +4153,24 @@ 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 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,
+};
+
+static const struct amdgpu_rlc_funcs gfx_v7_0_rlc_funcs = {
+	.enter_safe_mode = gfx_v7_0_enter_rlc_safe_mode,
+	.exit_safe_mode = gfx_v7_0_exit_rlc_safe_mode
+};
+
 static int gfx_v7_0_early_init(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	adev->gfx.num_gfx_rings = GFX7_NUM_GFX_RINGS;
 	adev->gfx.num_compute_rings = GFX7_NUM_COMPUTE_RINGS;
+	adev->gfx.funcs = &gfx_v7_0_gfx_funcs;
+	adev->gfx.rlc.funcs = &gfx_v7_0_rlc_funcs;
 	gfx_v7_0_set_ring_funcs(adev);
 	gfx_v7_0_set_irq_funcs(adev);
 	gfx_v7_0_set_gds_init(adev);
@@ -5009,16 +5014,22 @@ static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev)
 	int i, j, k, counter, active_cu_number = 0;
 	u32 mask, bitmap, ao_bitmap, ao_cu_mask = 0;
 	struct amdgpu_cu_info *cu_info = &adev->gfx.cu_info;
+	unsigned disable_masks[4 * 2];
 
 	memset(cu_info, 0, sizeof(*cu_info));
 
+	amdgpu_gfx_parse_disable_cu(disable_masks, 4, 2);
+
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
 			mask = 1;
 			ao_bitmap = 0;
 			counter = 0;
-			gfx_v7_0_select_se_sh(adev, i, j);
+			gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff);
+			if (i < 4 && j < 2)
+				gfx_v7_0_set_user_cu_inactive_bitmap(
+					adev, disable_masks[i * 2 + j]);
 			bitmap = gfx_v7_0_get_cu_active_bitmap(adev);
 			cu_info->bitmap[i][j] = bitmap;
 
@@ -5034,7 +5045,7 @@ static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev)
 			ao_cu_mask |= (ao_bitmap << (i * 16 + j * 8));
 		}
 	}
-	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
 	cu_info->number = active_cu_number;
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.h b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.h
index e747aa935..94e3ea147 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.h
@@ -26,11 +26,4 @@
 
 extern const struct amd_ip_funcs gfx_v7_0_ip_funcs;
 
-/* XXX these shouldn't be exported */
-void gfx_v7_0_enter_rlc_safe_mode(struct amdgpu_device *adev);
-void gfx_v7_0_exit_rlc_safe_mode(struct amdgpu_device *adev);
-void gfx_v7_0_rlc_stop(struct amdgpu_device *adev);
-uint64_t gfx_v7_0_get_gpu_clock_counter(struct amdgpu_device *adev);
-void gfx_v7_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num);
-
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index 1127b2b62..1c8f2754f 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -225,7 +225,8 @@ static const u32 tonga_mgcg_cgcg_init[] =
 
 static const u32 golden_settings_polaris11_a11[] =
 {
-	mmCB_HW_CONTROL, 0xfffdf3cf, 0x00006208,
+	mmCB_HW_CONTROL, 0x0000f3cf, 0x00007208,
+	mmCB_HW_CONTROL_2, 0x0f000000, 0x0f000000,
 	mmCB_HW_CONTROL_3, 0x000001ff, 0x00000040,
 	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
 	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
@@ -234,7 +235,7 @@ static const u32 golden_settings_polaris11_a11[] =
 	mmPA_SC_RASTER_CONFIG_1, 0x0000003f, 0x00000000,
 	mmRLC_CGCG_CGLS_CTRL, 0x00000003, 0x0001003c,
 	mmRLC_CGCG_CGLS_CTRL_3D, 0xffffffff, 0x0001003c,
-	mmSQ_CONFIG, 0x07f80000, 0x07180000,
+	mmSQ_CONFIG, 0x07f80000, 0x01180000,
 	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
 	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
 	mmTCP_ADDR_CONFIG, 0x000003ff, 0x000000f3,
@@ -256,8 +257,8 @@ static const u32 polaris11_golden_common_all[] =
 static const u32 golden_settings_polaris10_a11[] =
 {
 	mmATC_MISC_CG, 0x000c0fc0, 0x000c0200,
-	mmCB_HW_CONTROL, 0xfffdf3cf, 0x00007208,
-	mmCB_HW_CONTROL_2, 0, 0x0f000000,
+	mmCB_HW_CONTROL, 0x0001f3cf, 0x00007208,
+	mmCB_HW_CONTROL_2, 0x0f000000, 0x0f000000,
 	mmCB_HW_CONTROL_3, 0x000001ff, 0x00000040,
 	mmDB_DEBUG2, 0xf00fffff, 0x00000400,
 	mmPA_SC_ENHANCE, 0xffffffff, 0x20000001,
@@ -364,6 +365,7 @@ static const u32 golden_settings_iceland_a11[] =
 	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
 	mmPA_SC_RASTER_CONFIG, 0x3f3fffff, 0x00000002,
 	mmPA_SC_RASTER_CONFIG_1, 0x0000003f, 0x00000000,
+	mmRLC_CGCG_CGLS_CTRL, 0x00000003, 0x0000003c,
 	mmSQ_RANDOM_WAVE_PRI, 0x001fffff, 0x000006fd,
 	mmTA_CNTL_AUX, 0x000f000f, 0x000b0000,
 	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
@@ -460,8 +462,10 @@ static const u32 cz_golden_settings_a11[] =
 	mmGB_GPU_ID, 0x0000000f, 0x00000000,
 	mmPA_SC_ENHANCE, 0xffffffff, 0x00000001,
 	mmPA_SC_LINE_STIPPLE_STATE, 0x0000ff0f, 0x00000000,
+	mmRLC_CGCG_CGLS_CTRL, 0x00000003, 0x0000003c,
 	mmSQ_RANDOM_WAVE_PRI, 0x001fffff, 0x000006fd,
 	mmTA_CNTL_AUX, 0x000f000f, 0x00010000,
+	mmTCC_CTRL, 0x00100000, 0xf31fff7f,
 	mmTCC_EXE_DISABLE, 0x00000002, 0x00000002,
 	mmTCP_ADDR_CONFIG, 0x0000000f, 0x000000f3,
 	mmTCP_CHAN_STEER_LO, 0xffffffff, 0x00001302
@@ -742,26 +746,25 @@ static int gfx_v8_0_ring_test_ring(struct amdgpu_ring *ring)
 	return r;
 }
 
-static int gfx_v8_0_ring_test_ib(struct amdgpu_ring *ring)
+static int gfx_v8_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_ib ib;
 	struct fence *f = NULL;
 	uint32_t scratch;
 	uint32_t tmp = 0;
-	unsigned i;
-	int r;
+	long r;
 
 	r = amdgpu_gfx_scratch_get(adev, &scratch);
 	if (r) {
-		DRM_ERROR("amdgpu: failed to get scratch reg (%d).\n", r);
+		DRM_ERROR("amdgpu: failed to get scratch reg (%ld).\n", r);
 		return r;
 	}
 	WREG32(scratch, 0xCAFEDEAD);
 	memset(&ib, 0, sizeof(ib));
 	r = amdgpu_ib_get(adev, NULL, 256, &ib);
 	if (r) {
-		DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
 		goto err1;
 	}
 	ib.ptr[0] = PACKET3(PACKET3_SET_UCONFIG_REG, 1);
@@ -773,28 +776,25 @@ static int gfx_v8_0_ring_test_ib(struct amdgpu_ring *ring)
 	if (r)
 		goto err2;
 
-	r = fence_wait(f, false);
-	if (r) {
-		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+	r = fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out.\n");
+		r = -ETIMEDOUT;
 		goto err2;
-	}
-	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = RREG32(scratch);
-		if (tmp == 0xDEADBEEF)
-			break;
-		DRM_UDELAY(1);
-	}
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ib test on ring %d succeeded in %u usecs\n",
-			 ring->idx, i);
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
 		goto err2;
+	}
+	tmp = RREG32(scratch);
+	if (tmp == 0xDEADBEEF) {
+		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
 	} else {
 		DRM_ERROR("amdgpu: ib test failed (scratch(0x%04X)=0x%08X)\n",
 			  scratch, tmp);
 		r = -EINVAL;
 	}
 err2:
-	fence_put(f);
 	amdgpu_ib_free(adev, &ib, NULL);
 	fence_put(f);
 err1:
@@ -1115,6 +1115,71 @@ static void gfx_v8_0_get_csb_buffer(struct amdgpu_device *adev,
 	buffer[count++] = cpu_to_le32(0);
 }
 
+static void cz_init_cp_jump_table(struct amdgpu_device *adev)
+{
+	const __le32 *fw_data;
+	volatile u32 *dst_ptr;
+	int me, i, max_me = 4;
+	u32 bo_offset = 0;
+	u32 table_offset, table_size;
+
+	if (adev->asic_type == CHIP_CARRIZO)
+		max_me = 5;
+
+	/* write the cp table buffer */
+	dst_ptr = adev->gfx.rlc.cp_table_ptr;
+	for (me = 0; me < max_me; me++) {
+		if (me == 0) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.ce_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.ce_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 1) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.pfp_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.pfp_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 2) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.me_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.me_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else if (me == 3) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.mec_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		} else  if (me == 4) {
+			const struct gfx_firmware_header_v1_0 *hdr =
+				(const struct gfx_firmware_header_v1_0 *)adev->gfx.mec2_fw->data;
+			fw_data = (const __le32 *)
+				(adev->gfx.mec2_fw->data +
+				 le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+			table_offset = le32_to_cpu(hdr->jt_offset);
+			table_size = le32_to_cpu(hdr->jt_size);
+		}
+
+		for (i = 0; i < table_size; i ++) {
+			dst_ptr[bo_offset + i] =
+				cpu_to_le32(le32_to_cpu(fw_data[table_offset + i]));
+		}
+
+		bo_offset += table_size;
+	}
+}
+
 static void gfx_v8_0_rlc_fini(struct amdgpu_device *adev)
 {
 	int r;
@@ -1130,6 +1195,18 @@ static void gfx_v8_0_rlc_fini(struct amdgpu_device *adev)
 		amdgpu_bo_unref(&adev->gfx.rlc.clear_state_obj);
 		adev->gfx.rlc.clear_state_obj = NULL;
 	}
+
+	/* jump table block */
+	if (adev->gfx.rlc.cp_table_obj) {
+		r = amdgpu_bo_reserve(adev->gfx.rlc.cp_table_obj, false);
+		if (unlikely(r != 0))
+			dev_warn(adev->dev, "(%d) reserve RLC cp table bo failed\n", r);
+		amdgpu_bo_unpin(adev->gfx.rlc.cp_table_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
+
+		amdgpu_bo_unref(&adev->gfx.rlc.cp_table_obj);
+		adev->gfx.rlc.cp_table_obj = NULL;
+	}
 }
 
 static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
@@ -1186,6 +1263,46 @@ static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
 		amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
 	}
 
+	if ((adev->asic_type == CHIP_CARRIZO) ||
+	    (adev->asic_type == CHIP_STONEY)) {
+		adev->gfx.rlc.cp_table_size = ALIGN(96 * 5 * 4, 2048) + (64 * 1024); /* JT + GDS */
+		if (adev->gfx.rlc.cp_table_obj == NULL) {
+			r = amdgpu_bo_create(adev, adev->gfx.rlc.cp_table_size, PAGE_SIZE, true,
+					     AMDGPU_GEM_DOMAIN_VRAM,
+					     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
+					     NULL, NULL,
+					     &adev->gfx.rlc.cp_table_obj);
+			if (r) {
+				dev_warn(adev->dev, "(%d) create RLC cp table bo failed\n", r);
+				return r;
+			}
+		}
+
+		r = amdgpu_bo_reserve(adev->gfx.rlc.cp_table_obj, false);
+		if (unlikely(r != 0)) {
+			dev_warn(adev->dev, "(%d) reserve RLC cp table bo failed\n", r);
+			return r;
+		}
+		r = amdgpu_bo_pin(adev->gfx.rlc.cp_table_obj, AMDGPU_GEM_DOMAIN_VRAM,
+				  &adev->gfx.rlc.cp_table_gpu_addr);
+		if (r) {
+			amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
+			dev_warn(adev->dev, "(%d) pin RLC cp_table bo failed\n", r);
+			return r;
+		}
+		r = amdgpu_bo_kmap(adev->gfx.rlc.cp_table_obj, (void **)&adev->gfx.rlc.cp_table_ptr);
+		if (r) {
+			dev_warn(adev->dev, "(%d) map RLC cp table bo failed\n", r);
+			return r;
+		}
+
+		cz_init_cp_jump_table(adev);
+
+		amdgpu_bo_kunmap(adev->gfx.rlc.cp_table_obj);
+		amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
+
+	}
+
 	return 0;
 }
 
@@ -1567,7 +1684,6 @@ static int gfx_v8_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
 		RREG32(sec_ded_counter_registers[i]);
 
 fail:
-	fence_put(f);
 	amdgpu_ib_free(adev, &ib, NULL);
 	fence_put(f);
 
@@ -3294,9 +3410,15 @@ static void gfx_v8_0_tiling_mode_table_init(struct amdgpu_device *adev)
 	}
 }
 
-void gfx_v8_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num)
+static void gfx_v8_0_select_se_sh(struct amdgpu_device *adev,
+				  u32 se_num, u32 sh_num, u32 instance)
 {
-	u32 data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_BROADCAST_WRITES, 1);
+	u32 data;
+
+	if (instance == 0xffffffff)
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_BROADCAST_WRITES, 1);
+	else
+		data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_INDEX, instance);
 
 	if ((se_num == 0xffffffff) && (sh_num == 0xffffffff)) {
 		data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_BROADCAST_WRITES, 1);
@@ -3346,13 +3468,13 @@ static void gfx_v8_0_setup_rb(struct amdgpu_device *adev)
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-			gfx_v8_0_select_se_sh(adev, i, j);
+			gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff);
 			data = gfx_v8_0_get_rb_active_bitmap(adev);
 			active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) *
 					       rb_bitmap_width_per_sh);
 		}
 	}
-	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
 	adev->gfx.config.backend_enable_mask = active_rbs;
@@ -3456,7 +3578,7 @@ static void gfx_v8_0_gpu_init(struct amdgpu_device *adev)
 	 * making sure that the following register writes will be broadcasted
 	 * to all the shaders
 	 */
-	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 
 	WREG32(mmPA_SC_FIFO_SIZE,
 		   (adev->gfx.config.sc_prim_fifo_size_frontend <<
@@ -3479,7 +3601,7 @@ static void gfx_v8_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-			gfx_v8_0_select_se_sh(adev, i, j);
+			gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff);
 			for (k = 0; k < adev->usec_timeout; k++) {
 				if (RREG32(mmRLC_SERDES_CU_MASTER_BUSY) == 0)
 					break;
@@ -3487,7 +3609,7 @@ static void gfx_v8_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
 			}
 		}
 	}
-	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
 	mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
@@ -3648,13 +3770,13 @@ static void gfx_v8_0_enable_save_restore_machine(struct amdgpu_device *adev)
 	WREG32(mmRLC_SRM_CNTL, data);
 }
 
-static void polaris11_init_power_gating(struct amdgpu_device *adev)
+static void gfx_v8_0_init_power_gating(struct amdgpu_device *adev)
 {
 	uint32_t data;
 
 	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_PG |
-			AMD_PG_SUPPORT_GFX_SMG |
-			AMD_PG_SUPPORT_GFX_DMG)) {
+			      AMD_PG_SUPPORT_GFX_SMG |
+			      AMD_PG_SUPPORT_GFX_DMG)) {
 		data = RREG32(mmCP_RB_WPTR_POLL_CNTL);
 		data &= ~CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT_MASK;
 		data |= (0x60 << CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT__SHIFT);
@@ -3679,6 +3801,53 @@ static void polaris11_init_power_gating(struct amdgpu_device *adev)
 	}
 }
 
+static void cz_enable_sck_slow_down_on_power_up(struct amdgpu_device *adev,
+						bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+
+	if (enable)
+		data |= RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PU_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PU_ENABLE_MASK;
+
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void cz_enable_sck_slow_down_on_power_down(struct amdgpu_device *adev,
+						  bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+
+	if (enable)
+		data |= RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PD_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PD_ENABLE_MASK;
+
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void cz_enable_cp_power_gating(struct amdgpu_device *adev, bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+
+	if (enable)
+		data &= ~RLC_PG_CNTL__CP_PG_DISABLE_MASK;
+	else
+		data |= RLC_PG_CNTL__CP_PG_DISABLE_MASK;
+
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
 static void gfx_v8_0_init_pg(struct amdgpu_device *adev)
 {
 	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_PG |
@@ -3691,8 +3860,25 @@ static void gfx_v8_0_init_pg(struct amdgpu_device *adev)
 		gfx_v8_0_init_save_restore_list(adev);
 		gfx_v8_0_enable_save_restore_machine(adev);
 
-		if (adev->asic_type == CHIP_POLARIS11)
-			polaris11_init_power_gating(adev);
+		if ((adev->asic_type == CHIP_CARRIZO) ||
+		    (adev->asic_type == CHIP_STONEY)) {
+			WREG32(mmRLC_JUMP_TABLE_RESTORE, adev->gfx.rlc.cp_table_gpu_addr >> 8);
+			gfx_v8_0_init_power_gating(adev);
+			WREG32(mmRLC_PG_ALWAYS_ON_CU_MASK, adev->gfx.cu_info.ao_cu_mask);
+			if (adev->pg_flags & AMD_PG_SUPPORT_RLC_SMU_HS) {
+				cz_enable_sck_slow_down_on_power_up(adev, true);
+				cz_enable_sck_slow_down_on_power_down(adev, true);
+			} else {
+				cz_enable_sck_slow_down_on_power_up(adev, false);
+				cz_enable_sck_slow_down_on_power_down(adev, false);
+			}
+			if (adev->pg_flags & AMD_PG_SUPPORT_CP)
+				cz_enable_cp_power_gating(adev, true);
+			else
+				cz_enable_cp_power_gating(adev, false);
+		} else if (adev->asic_type == CHIP_POLARIS11) {
+			gfx_v8_0_init_power_gating(adev);
+		}
 	}
 }
 
@@ -4931,7 +5117,7 @@ static int gfx_v8_0_soft_reset(void *handle)
  * Fetches a GPU clock counter snapshot.
  * Returns the 64 bit clock counter snapshot.
  */
-uint64_t gfx_v8_0_get_gpu_clock_counter(struct amdgpu_device *adev)
+static uint64_t gfx_v8_0_get_gpu_clock_counter(struct amdgpu_device *adev)
 {
 	uint64_t clock;
 
@@ -4991,12 +5177,18 @@ static void gfx_v8_0_ring_emit_gds_switch(struct amdgpu_ring *ring,
 	amdgpu_ring_write(ring, (1 << (oa_size + oa_base)) - (1 << oa_base));
 }
 
+static const struct amdgpu_gfx_funcs gfx_v8_0_gfx_funcs = {
+	.get_gpu_clock_counter = &gfx_v8_0_get_gpu_clock_counter,
+	.select_se_sh = &gfx_v8_0_select_se_sh,
+};
+
 static int gfx_v8_0_early_init(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	adev->gfx.num_gfx_rings = GFX8_NUM_GFX_RINGS;
 	adev->gfx.num_compute_rings = GFX8_NUM_COMPUTE_RINGS;
+	adev->gfx.funcs = &gfx_v8_0_gfx_funcs;
 	gfx_v8_0_set_ring_funcs(adev);
 	gfx_v8_0_set_irq_funcs(adev);
 	gfx_v8_0_set_gds_init(adev);
@@ -5029,51 +5221,43 @@ static int gfx_v8_0_late_init(void *handle)
 	return 0;
 }
 
-static void polaris11_enable_gfx_static_mg_power_gating(struct amdgpu_device *adev,
-		bool enable)
+static void gfx_v8_0_enable_gfx_static_mg_power_gating(struct amdgpu_device *adev,
+						       bool enable)
 {
 	uint32_t data, temp;
 
-	/* Send msg to SMU via Powerplay */
-	amdgpu_set_powergating_state(adev,
-			AMD_IP_BLOCK_TYPE_SMC,
-			enable ? AMD_PG_STATE_GATE : AMD_PG_STATE_UNGATE);
+	if (adev->asic_type == CHIP_POLARIS11)
+		/* Send msg to SMU via Powerplay */
+		amdgpu_set_powergating_state(adev,
+					     AMD_IP_BLOCK_TYPE_SMC,
+					     enable ?
+					     AMD_PG_STATE_GATE : AMD_PG_STATE_UNGATE);
 
-	if (enable) {
-		/* Enable static MGPG */
-		temp = data = RREG32(mmRLC_PG_CNTL);
+	temp = data = RREG32(mmRLC_PG_CNTL);
+	/* Enable static MGPG */
+	if (enable)
 		data |= RLC_PG_CNTL__STATIC_PER_CU_PG_ENABLE_MASK;
-
-		if (temp != data)
-			WREG32(mmRLC_PG_CNTL, data);
-	} else {
-		temp = data = RREG32(mmRLC_PG_CNTL);
+	else
 		data &= ~RLC_PG_CNTL__STATIC_PER_CU_PG_ENABLE_MASK;
 
-		if (temp != data)
-			WREG32(mmRLC_PG_CNTL, data);
-	}
+	if (temp != data)
+		WREG32(mmRLC_PG_CNTL, data);
 }
 
-static void polaris11_enable_gfx_dynamic_mg_power_gating(struct amdgpu_device *adev,
-		bool enable)
+static void gfx_v8_0_enable_gfx_dynamic_mg_power_gating(struct amdgpu_device *adev,
+							bool enable)
 {
 	uint32_t data, temp;
 
-	if (enable) {
-		/* Enable dynamic MGPG */
-		temp = data = RREG32(mmRLC_PG_CNTL);
+	temp = data = RREG32(mmRLC_PG_CNTL);
+	/* Enable dynamic MGPG */
+	if (enable)
 		data |= RLC_PG_CNTL__DYN_PER_CU_PG_ENABLE_MASK;
-
-		if (temp != data)
-			WREG32(mmRLC_PG_CNTL, data);
-	} else {
-		temp = data = RREG32(mmRLC_PG_CNTL);
+	else
 		data &= ~RLC_PG_CNTL__DYN_PER_CU_PG_ENABLE_MASK;
 
-		if (temp != data)
-			WREG32(mmRLC_PG_CNTL, data);
-	}
+	if (temp != data)
+		WREG32(mmRLC_PG_CNTL, data);
 }
 
 static void polaris11_enable_gfx_quick_mg_power_gating(struct amdgpu_device *adev,
@@ -5081,19 +5265,63 @@ static void polaris11_enable_gfx_quick_mg_power_gating(struct amdgpu_device *ade
 {
 	uint32_t data, temp;
 
-	if (enable) {
-		/* Enable quick PG */
-		temp = data = RREG32(mmRLC_PG_CNTL);
-		data |= 0x100000;
+	temp = data = RREG32(mmRLC_PG_CNTL);
+	/* Enable quick PG */
+	if (enable)
+		data |= RLC_PG_CNTL__QUICK_PG_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__QUICK_PG_ENABLE_MASK;
 
-		if (temp != data)
-			WREG32(mmRLC_PG_CNTL, data);
-	} else {
-		temp = data = RREG32(mmRLC_PG_CNTL);
-		data &= ~0x100000;
+	if (temp != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
 
-		if (temp != data)
-			WREG32(mmRLC_PG_CNTL, data);
+static void cz_enable_gfx_cg_power_gating(struct amdgpu_device *adev,
+					  bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+
+	if (enable)
+		data |= RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK;
+
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+}
+
+static void cz_enable_gfx_pipeline_power_gating(struct amdgpu_device *adev,
+						bool enable)
+{
+	u32 data, orig;
+
+	orig = data = RREG32(mmRLC_PG_CNTL);
+
+	if (enable)
+		data |= RLC_PG_CNTL__GFX_PIPELINE_PG_ENABLE_MASK;
+	else
+		data &= ~RLC_PG_CNTL__GFX_PIPELINE_PG_ENABLE_MASK;
+
+	if (orig != data)
+		WREG32(mmRLC_PG_CNTL, data);
+
+	/* Read any GFX register to wake up GFX. */
+	if (!enable)
+		data = RREG32(mmDB_RENDER_CONTROL);
+}
+
+static void cz_update_gfx_cg_power_gating(struct amdgpu_device *adev,
+					  bool enable)
+{
+	if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_PG) && enable) {
+		cz_enable_gfx_cg_power_gating(adev, true);
+		if (adev->pg_flags & AMD_PG_SUPPORT_GFX_PIPELINE)
+			cz_enable_gfx_pipeline_power_gating(adev, true);
+	} else {
+		cz_enable_gfx_cg_power_gating(adev, false);
+		cz_enable_gfx_pipeline_power_gating(adev, false);
 	}
 }
 
@@ -5101,21 +5329,42 @@ static int gfx_v8_0_set_powergating_state(void *handle,
 					  enum amd_powergating_state state)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+	bool enable = (state == AMD_PG_STATE_GATE) ? true : false;
 
 	if (!(adev->pg_flags & AMD_PG_SUPPORT_GFX_PG))
 		return 0;
 
 	switch (adev->asic_type) {
+	case CHIP_CARRIZO:
+	case CHIP_STONEY:
+		if (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG)
+			cz_update_gfx_cg_power_gating(adev, enable);
+
+		if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_SMG) && enable)
+			gfx_v8_0_enable_gfx_static_mg_power_gating(adev, true);
+		else
+			gfx_v8_0_enable_gfx_static_mg_power_gating(adev, false);
+
+		if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_DMG) && enable)
+			gfx_v8_0_enable_gfx_dynamic_mg_power_gating(adev, true);
+		else
+			gfx_v8_0_enable_gfx_dynamic_mg_power_gating(adev, false);
+		break;
 	case CHIP_POLARIS11:
-		if (adev->pg_flags & AMD_PG_SUPPORT_GFX_SMG)
-			polaris11_enable_gfx_static_mg_power_gating(adev,
-					state == AMD_PG_STATE_GATE ? true : false);
-		else if (adev->pg_flags & AMD_PG_SUPPORT_GFX_DMG)
-			polaris11_enable_gfx_dynamic_mg_power_gating(adev,
-					state == AMD_PG_STATE_GATE ? true : false);
+		if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_SMG) && enable)
+			gfx_v8_0_enable_gfx_static_mg_power_gating(adev, true);
+		else
+			gfx_v8_0_enable_gfx_static_mg_power_gating(adev, false);
+
+		if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_DMG) && enable)
+			gfx_v8_0_enable_gfx_dynamic_mg_power_gating(adev, true);
+		else
+			gfx_v8_0_enable_gfx_dynamic_mg_power_gating(adev, false);
+
+		if ((adev->pg_flags & AMD_PG_SUPPORT_GFX_QUICK_MG) && enable)
+			polaris11_enable_gfx_quick_mg_power_gating(adev, true);
 		else
-			polaris11_enable_gfx_quick_mg_power_gating(adev,
-					state == AMD_PG_STATE_GATE ? true : false);
+			polaris11_enable_gfx_quick_mg_power_gating(adev, false);
 		break;
 	default:
 		break;
@@ -5129,7 +5378,7 @@ static void gfx_v8_0_send_serdes_cmd(struct amdgpu_device *adev,
 {
 	uint32_t data;
 
-	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 
 	WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
 	WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
@@ -5517,6 +5766,8 @@ static void gfx_v8_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev
 			WREG32(mmRLC_CGCG_CGLS_CTRL, data);
 	}
 
+	gfx_v8_0_wait_for_rlc_serdes(adev);
+
 	adev->gfx.rlc.funcs->exit_safe_mode(adev);
 }
 static int gfx_v8_0_update_gfx_clock_gating(struct amdgpu_device *adev,
@@ -5642,17 +5893,6 @@ static void gfx_v8_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
 				      unsigned vm_id, bool ctx_switch)
 {
 	u32 header, control = 0;
-	u32 next_rptr = ring->wptr + 5;
-
-	if (ctx_switch)
-		next_rptr += 2;
-
-	next_rptr += 4;
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, WRITE_DATA_DST_SEL(5) | WR_CONFIRM);
-	amdgpu_ring_write(ring, ring->next_rptr_gpu_addr & 0xfffffffc);
-	amdgpu_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr) & 0xffffffff);
-	amdgpu_ring_write(ring, next_rptr);
 
 	/* insert SWITCH_BUFFER packet before first IB in the ring frame */
 	if (ctx_switch) {
@@ -5681,23 +5921,9 @@ static void gfx_v8_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
 					  struct amdgpu_ib *ib,
 					  unsigned vm_id, bool ctx_switch)
 {
-	u32 header, control = 0;
-	u32 next_rptr = ring->wptr + 5;
-
-	control |= INDIRECT_BUFFER_VALID;
-
-	next_rptr += 4;
-	amdgpu_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3));
-	amdgpu_ring_write(ring, WRITE_DATA_DST_SEL(5) | WR_CONFIRM);
-	amdgpu_ring_write(ring, ring->next_rptr_gpu_addr & 0xfffffffc);
-	amdgpu_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr) & 0xffffffff);
-	amdgpu_ring_write(ring, next_rptr);
+	u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vm_id << 24);
 
-	header = PACKET3(PACKET3_INDIRECT_BUFFER, 2);
-
-	control |= ib->length_dw | (vm_id << 24);
-
-	amdgpu_ring_write(ring, header);
+	amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
 	amdgpu_ring_write(ring,
 #ifdef __BIG_ENDIAN
 					  (2 << 0) |
@@ -6150,9 +6376,9 @@ static void gfx_v8_0_set_rlc_funcs(struct amdgpu_device *adev)
 {
 	switch (adev->asic_type) {
 	case CHIP_TOPAZ:
-	case CHIP_STONEY:
 		adev->gfx.rlc.funcs = &iceland_rlc_funcs;
 		break;
+	case CHIP_STONEY:
 	case CHIP_CARRIZO:
 		adev->gfx.rlc.funcs = &cz_rlc_funcs;
 		break;
@@ -6190,6 +6416,20 @@ static void gfx_v8_0_set_gds_init(struct amdgpu_device *adev)
 	}
 }
 
+static void gfx_v8_0_set_user_cu_inactive_bitmap(struct amdgpu_device *adev,
+						 u32 bitmap)
+{
+	u32 data;
+
+	if (!bitmap)
+		return;
+
+	data = bitmap << GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS__SHIFT;
+	data &= GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS_MASK;
+
+	WREG32(mmGC_USER_SHADER_ARRAY_CONFIG, data);
+}
+
 static u32 gfx_v8_0_get_cu_active_bitmap(struct amdgpu_device *adev)
 {
 	u32 data, mask;
@@ -6210,16 +6450,22 @@ static void gfx_v8_0_get_cu_info(struct amdgpu_device *adev)
 	int i, j, k, counter, active_cu_number = 0;
 	u32 mask, bitmap, ao_bitmap, ao_cu_mask = 0;
 	struct amdgpu_cu_info *cu_info = &adev->gfx.cu_info;
+	unsigned disable_masks[4 * 2];
 
 	memset(cu_info, 0, sizeof(*cu_info));
 
+	amdgpu_gfx_parse_disable_cu(disable_masks, 4, 2);
+
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
 			mask = 1;
 			ao_bitmap = 0;
 			counter = 0;
-			gfx_v8_0_select_se_sh(adev, i, j);
+			gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff);
+			if (i < 4 && j < 2)
+				gfx_v8_0_set_user_cu_inactive_bitmap(
+					adev, disable_masks[i * 2 + j]);
 			bitmap = gfx_v8_0_get_cu_active_bitmap(adev);
 			cu_info->bitmap[i][j] = bitmap;
 
@@ -6235,7 +6481,7 @@ static void gfx_v8_0_get_cu_info(struct amdgpu_device *adev)
 			ao_cu_mask |= (ao_bitmap << (i * 16 + j * 8));
 		}
 	}
-	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+	gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
 	cu_info->number = active_cu_number;
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.h b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.h
index 16a49f53a..bc82c7943 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.h
@@ -26,7 +26,6 @@
 
 extern const struct amd_ip_funcs gfx_v8_0_ip_funcs;
 
-uint64_t gfx_v8_0_get_gpu_clock_counter(struct amdgpu_device *adev);
 void gfx_v8_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num);
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
index e05402728..c2b554c5c 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
@@ -39,6 +39,7 @@
 
 static void gmc_v7_0_set_gart_funcs(struct amdgpu_device *adev);
 static void gmc_v7_0_set_irq_funcs(struct amdgpu_device *adev);
+static int gmc_v7_0_wait_for_idle(void *handle);
 
 /*(DEBLOBBED)*/
 
@@ -71,39 +72,15 @@ static void gmc_v7_0_init_golden_registers(struct amdgpu_device *adev)
 	}
 }
 
-/**
- * gmc7_mc_wait_for_idle - wait for MC idle callback.
- *
- * @adev: amdgpu_device pointer
- *
- * Wait for the MC (memory controller) to be idle.
- * (evergreen+).
- * Returns 0 if the MC is idle, -1 if not.
- */
-int gmc_v7_0_mc_wait_for_idle(struct amdgpu_device *adev)
-{
-	unsigned i;
-	u32 tmp;
-
-	for (i = 0; i < adev->usec_timeout; i++) {
-		/* read MC_STATUS */
-		tmp = RREG32(mmSRBM_STATUS) & 0x1F00;
-		if (!tmp)
-			return 0;
-		udelay(1);
-	}
-	return -1;
-}
-
-void gmc_v7_0_mc_stop(struct amdgpu_device *adev,
-		      struct amdgpu_mode_mc_save *save)
+static void gmc_v7_0_mc_stop(struct amdgpu_device *adev,
+			     struct amdgpu_mode_mc_save *save)
 {
 	u32 blackout;
 
 	if (adev->mode_info.num_crtc)
 		amdgpu_display_stop_mc_access(adev, save);
 
-	amdgpu_asic_wait_for_mc_idle(adev);
+	gmc_v7_0_wait_for_idle((void *)adev);
 
 	blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
 	if (REG_GET_FIELD(blackout, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE) != 1) {
@@ -118,8 +95,8 @@ void gmc_v7_0_mc_stop(struct amdgpu_device *adev,
 	udelay(100);
 }
 
-void gmc_v7_0_mc_resume(struct amdgpu_device *adev,
-			struct amdgpu_mode_mc_save *save)
+static void gmc_v7_0_mc_resume(struct amdgpu_device *adev,
+			       struct amdgpu_mode_mc_save *save)
 {
 	u32 tmp;
 
@@ -310,7 +287,7 @@ static void gmc_v7_0_mc_program(struct amdgpu_device *adev)
 		amdgpu_display_set_vga_render_state(adev, false);
 
 	gmc_v7_0_mc_stop(adev, &save);
-	if (amdgpu_asic_wait_for_mc_idle(adev)) {
+	if (gmc_v7_0_wait_for_idle((void *)adev)) {
 		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
 	}
 	/* Update configuration */
@@ -330,7 +307,7 @@ static void gmc_v7_0_mc_program(struct amdgpu_device *adev)
 	WREG32(mmMC_VM_AGP_BASE, 0);
 	WREG32(mmMC_VM_AGP_TOP, 0x0FFFFFFF);
 	WREG32(mmMC_VM_AGP_BOT, 0x0FFFFFFF);
-	if (amdgpu_asic_wait_for_mc_idle(adev)) {
+	if (gmc_v7_0_wait_for_idle((void *)adev)) {
 		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
 	}
 	gmc_v7_0_mc_resume(adev, &save);
@@ -1136,7 +1113,7 @@ static int gmc_v7_0_soft_reset(void *handle)
 
 	if (srbm_soft_reset) {
 		gmc_v7_0_mc_stop(adev, &save);
-		if (gmc_v7_0_wait_for_idle(adev)) {
+		if (gmc_v7_0_wait_for_idle((void *)adev)) {
 			dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
 		}
 
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.h b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.h
index 36fcbbc46..0b386b5d2 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.h
@@ -26,11 +26,4 @@
 
 extern const struct amd_ip_funcs gmc_v7_0_ip_funcs;
 
-/* XXX these shouldn't be exported */
-void gmc_v7_0_mc_stop(struct amdgpu_device *adev,
-		      struct amdgpu_mode_mc_save *save);
-void gmc_v7_0_mc_resume(struct amdgpu_device *adev,
-			struct amdgpu_mode_mc_save *save);
-int gmc_v7_0_mc_wait_for_idle(struct amdgpu_device *adev);
-
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
index a7b6de8a1..34695e2e3 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
@@ -41,6 +41,7 @@
 
 static void gmc_v8_0_set_gart_funcs(struct amdgpu_device *adev);
 static void gmc_v8_0_set_irq_funcs(struct amdgpu_device *adev);
+static int gmc_v8_0_wait_for_idle(void *handle);
 
 /*(DEBLOBBED)*/
 
@@ -100,6 +101,11 @@ static const u32 stoney_mgcg_cgcg_init[] =
 	mmMC_MEM_POWER_LS, 0xffffffff, 0x00000104
 };
 
+static const u32 golden_settings_stoney_common[] =
+{
+	mmMC_HUB_RDREQ_UVD, MC_HUB_RDREQ_UVD__PRESCALE_MASK, 0x00000004,
+	mmMC_RD_GRP_OTH, MC_RD_GRP_OTH__UVD_MASK, 0x00600000
+};
 
 static void gmc_v8_0_init_golden_registers(struct amdgpu_device *adev)
 {
@@ -139,50 +145,24 @@ static void gmc_v8_0_init_golden_registers(struct amdgpu_device *adev)
 		amdgpu_program_register_sequence(adev,
 						 stoney_mgcg_cgcg_init,
 						 (const u32)ARRAY_SIZE(stoney_mgcg_cgcg_init));
+		amdgpu_program_register_sequence(adev,
+						 golden_settings_stoney_common,
+						 (const u32)ARRAY_SIZE(golden_settings_stoney_common));
 		break;
 	default:
 		break;
 	}
 }
 
-/**
- * gmc8_mc_wait_for_idle - wait for MC idle callback.
- *
- * @adev: amdgpu_device pointer
- *
- * Wait for the MC (memory controller) to be idle.
- * (evergreen+).
- * Returns 0 if the MC is idle, -1 if not.
- */
-int gmc_v8_0_mc_wait_for_idle(struct amdgpu_device *adev)
-{
-	unsigned i;
-	u32 tmp;
-
-	for (i = 0; i < adev->usec_timeout; i++) {
-		/* read MC_STATUS */
-		tmp = RREG32(mmSRBM_STATUS) & (SRBM_STATUS__VMC_BUSY_MASK |
-					       SRBM_STATUS__MCB_BUSY_MASK |
-					       SRBM_STATUS__MCB_NON_DISPLAY_BUSY_MASK |
-					       SRBM_STATUS__MCC_BUSY_MASK |
-					       SRBM_STATUS__MCD_BUSY_MASK |
-					       SRBM_STATUS__VMC1_BUSY_MASK);
-		if (!tmp)
-			return 0;
-		udelay(1);
-	}
-	return -1;
-}
-
-void gmc_v8_0_mc_stop(struct amdgpu_device *adev,
-		      struct amdgpu_mode_mc_save *save)
+static void gmc_v8_0_mc_stop(struct amdgpu_device *adev,
+			     struct amdgpu_mode_mc_save *save)
 {
 	u32 blackout;
 
 	if (adev->mode_info.num_crtc)
 		amdgpu_display_stop_mc_access(adev, save);
 
-	amdgpu_asic_wait_for_mc_idle(adev);
+	gmc_v8_0_wait_for_idle(adev);
 
 	blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
 	if (REG_GET_FIELD(blackout, MC_SHARED_BLACKOUT_CNTL, BLACKOUT_MODE) != 1) {
@@ -197,8 +177,8 @@ void gmc_v8_0_mc_stop(struct amdgpu_device *adev,
 	udelay(100);
 }
 
-void gmc_v8_0_mc_resume(struct amdgpu_device *adev,
-			struct amdgpu_mode_mc_save *save)
+static void gmc_v8_0_mc_resume(struct amdgpu_device *adev,
+			       struct amdgpu_mode_mc_save *save)
 {
 	u32 tmp;
 
@@ -391,7 +371,7 @@ static void gmc_v8_0_mc_program(struct amdgpu_device *adev)
 		amdgpu_display_set_vga_render_state(adev, false);
 
 	gmc_v8_0_mc_stop(adev, &save);
-	if (amdgpu_asic_wait_for_mc_idle(adev)) {
+	if (gmc_v8_0_wait_for_idle((void *)adev)) {
 		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
 	}
 	/* Update configuration */
@@ -411,7 +391,7 @@ static void gmc_v8_0_mc_program(struct amdgpu_device *adev)
 	WREG32(mmMC_VM_AGP_BASE, 0);
 	WREG32(mmMC_VM_AGP_TOP, 0x0FFFFFFF);
 	WREG32(mmMC_VM_AGP_BOT, 0x0FFFFFFF);
-	if (amdgpu_asic_wait_for_mc_idle(adev)) {
+	if (gmc_v8_0_wait_for_idle((void *)adev)) {
 		dev_warn(adev->dev, "Wait for MC idle timedout !\n");
 	}
 	gmc_v8_0_mc_resume(adev, &save);
@@ -1138,7 +1118,7 @@ static int gmc_v8_0_soft_reset(void *handle)
 
 	if (srbm_soft_reset) {
 		gmc_v8_0_mc_stop(adev, &save);
-		if (gmc_v8_0_wait_for_idle(adev)) {
+		if (gmc_v8_0_wait_for_idle((void *)adev)) {
 			dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
 		}
 
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.h b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.h
index 973436086..fc5001a81 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.h
@@ -26,11 +26,4 @@
 
 extern const struct amd_ip_funcs gmc_v8_0_ip_funcs;
 
-/* XXX these shouldn't be exported */
-void gmc_v8_0_mc_stop(struct amdgpu_device *adev,
-		      struct amdgpu_mode_mc_save *save);
-void gmc_v8_0_mc_resume(struct amdgpu_device *adev,
-			struct amdgpu_mode_mc_save *save);
-int gmc_v8_0_mc_wait_for_idle(struct amdgpu_device *adev);
-
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/iceland_dpm.c b/drivers/gpu/drm/amd/amdgpu/iceland_dpm.c
index 571e37566..90b4bcc72 100644
--- a/drivers/gpu/drm/amd/amdgpu/iceland_dpm.c
+++ b/drivers/gpu/drm/amd/amdgpu/iceland_dpm.c
@@ -24,7 +24,7 @@
 #include <linux/firmware.h>
 #include "drmP.h"
 #include "amdgpu.h"
-#include "iceland_smumgr.h"
+#include "iceland_smum.h"
 
 /*(DEBLOBBED)*/
 
diff --git a/drivers/gpu/drm/amd/amdgpu/iceland_smc.c b/drivers/gpu/drm/amd/amdgpu/iceland_smc.c
index 52ee08193..211839913 100644
--- a/drivers/gpu/drm/amd/amdgpu/iceland_smc.c
+++ b/drivers/gpu/drm/amd/amdgpu/iceland_smc.c
@@ -25,7 +25,7 @@
 #include "drmP.h"
 #include "amdgpu.h"
 #include "ppsmc.h"
-#include "iceland_smumgr.h"
+#include "iceland_smum.h"
 #include "smu_ucode_xfer_vi.h"
 #include "amdgpu_ucode.h"
 
@@ -211,7 +211,7 @@ static int iceland_send_msg_to_smc_without_waiting(struct amdgpu_device *adev,
 						   PPSMC_Msg msg)
 {
 	if (!iceland_is_smc_ram_running(adev))
-		return -EINVAL;;
+		return -EINVAL;
 
 	if (wait_smu_response(adev)) {
 		DRM_ERROR("Failed to send previous message\n");
diff --git a/drivers/gpu/drm/amd/amdgpu/iceland_smum.h b/drivers/gpu/drm/amd/amdgpu/iceland_smum.h
new file mode 100644
index 000000000..5983e3150
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/iceland_smum.h
@@ -0,0 +1,41 @@
+/*
+ * Copyright 2014 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 ICELAND_SMUM_H
+#define ICELAND_SMUM_H
+
+#include "ppsmc.h"
+
+extern int iceland_smu_init(struct amdgpu_device *adev);
+extern int iceland_smu_fini(struct amdgpu_device *adev);
+extern int iceland_smu_start(struct amdgpu_device *adev);
+
+struct iceland_smu_private_data
+{
+	uint8_t *header;
+	uint8_t *mec_image;
+	uint32_t header_addr_high;
+	uint32_t header_addr_low;
+};
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/kv_dpm.c b/drivers/gpu/drm/amd/amdgpu/kv_dpm.c
index a789a863d..a845e883f 100644
--- a/drivers/gpu/drm/amd/amdgpu/kv_dpm.c
+++ b/drivers/gpu/drm/amd/amdgpu/kv_dpm.c
@@ -191,6 +191,7 @@ static void sumo_construct_vid_mapping_table(struct amdgpu_device *adev,
 	vid_mapping_table->num_entries = i;
 }
 
+#if 0
 static const struct kv_lcac_config_values sx_local_cac_cfg_kv[] =
 {
 	{  0,       4,        1    },
@@ -289,6 +290,7 @@ static const struct kv_lcac_config_reg cpl_cac_config_reg[] =
 {
 	{ 0xc0400d80, 0x003e0000, 17, 0x3fc00000, 22, 0x0001fffe, 1, 0x00000001, 0 }
 };
+#endif
 
 static const struct kv_pt_config_reg didt_config_kv[] =
 {
@@ -507,19 +509,19 @@ static int kv_enable_didt(struct amdgpu_device *adev, bool enable)
 	    pi->caps_db_ramping ||
 	    pi->caps_td_ramping ||
 	    pi->caps_tcp_ramping) {
-		gfx_v7_0_enter_rlc_safe_mode(adev);
+		adev->gfx.rlc.funcs->enter_safe_mode(adev);
 
 		if (enable) {
 			ret = kv_program_pt_config_registers(adev, didt_config_kv);
 			if (ret) {
-				gfx_v7_0_exit_rlc_safe_mode(adev);
+				adev->gfx.rlc.funcs->exit_safe_mode(adev);
 				return ret;
 			}
 		}
 
 		kv_do_enable_didt(adev, enable);
 
-		gfx_v7_0_exit_rlc_safe_mode(adev);
+		adev->gfx.rlc.funcs->exit_safe_mode(adev);
 	}
 
 	return 0;
diff --git a/drivers/gpu/drm/amd/amdgpu/ppsmc.h b/drivers/gpu/drm/amd/amdgpu/ppsmc.h
index 7837f2ecc..8463245f4 100644
--- a/drivers/gpu/drm/amd/amdgpu/ppsmc.h
+++ b/drivers/gpu/drm/amd/amdgpu/ppsmc.h
@@ -90,7 +90,9 @@ typedef uint8_t PPSMC_Result;
 #define PPSMC_StartFanControl               ((uint8_t)0x5B)
 #define PPSMC_StopFanControl                ((uint8_t)0x5C)
 #define PPSMC_MSG_NoDisplay                 ((uint8_t)0x5D)
+#define PPSMC_NoDisplay                     ((uint8_t)0x5D)
 #define PPSMC_MSG_HasDisplay                ((uint8_t)0x5E)
+#define PPSMC_HasDisplay                    ((uint8_t)0x5E)
 #define PPSMC_MSG_UVDPowerOFF               ((uint8_t)0x60)
 #define PPSMC_MSG_UVDPowerON                ((uint8_t)0x61)
 #define PPSMC_MSG_EnableULV                 ((uint8_t)0x62)
@@ -108,6 +110,7 @@ typedef uint8_t PPSMC_Result;
 #define PPSMC_MSG_DisableDTE                ((uint8_t)0x88)
 #define PPSMC_MSG_ThrottleOVRDSCLKDS        ((uint8_t)0x96)
 #define PPSMC_MSG_CancelThrottleOVRDSCLKDS  ((uint8_t)0x97)
+#define PPSMC_MSG_EnableACDCGPIOInterrupt   ((uint16_t) 0x149)
 
 /* CI/KV/KB */
 #define PPSMC_MSG_UVDDPM_SetEnabledMask       ((uint16_t) 0x12D)
@@ -161,6 +164,7 @@ typedef uint8_t PPSMC_Result;
 #define PPSMC_MSG_MASTER_DeepSleep_OFF        ((uint16_t) 0x190)
 #define PPSMC_MSG_Remove_DC_Clamp             ((uint16_t) 0x191)
 #define PPSMC_MSG_SetFanPwmMax                ((uint16_t) 0x19A)
+#define PPSMC_MSG_SetFanRpmMax                ((uint16_t) 0x205)
 
 #define PPSMC_MSG_ENABLE_THERMAL_DPM          ((uint16_t) 0x19C)
 #define PPSMC_MSG_DISABLE_THERMAL_DPM         ((uint16_t) 0x19D)
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
index 36d97195f..f6e37bdc1 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
@@ -254,19 +254,6 @@ static void sdma_v2_4_ring_emit_ib(struct amdgpu_ring *ring,
 				   unsigned vm_id, bool ctx_switch)
 {
 	u32 vmid = vm_id & 0xf;
-	u32 next_rptr = ring->wptr + 5;
-
-	while ((next_rptr & 7) != 2)
-		next_rptr++;
-
-	next_rptr += 6;
-
-	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
-			  SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
-	amdgpu_ring_write(ring, lower_32_bits(ring->next_rptr_gpu_addr) & 0xfffffffc);
-	amdgpu_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr));
-	amdgpu_ring_write(ring, SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(1));
-	amdgpu_ring_write(ring, next_rptr);
 
 	/* IB packet must end on a 8 DW boundary */
 	sdma_v2_4_ring_insert_nop(ring, (10 - (ring->wptr & 7)) % 8);
@@ -405,7 +392,7 @@ static void sdma_v2_4_enable(struct amdgpu_device *adev, bool enable)
 	u32 f32_cntl;
 	int i;
 
-	if (enable == false) {
+	if (!enable) {
 		sdma_v2_4_gfx_stop(adev);
 		sdma_v2_4_rlc_stop(adev);
 	}
@@ -579,19 +566,21 @@ static int sdma_v2_4_start(struct amdgpu_device *adev)
 {
 	int r;
 
-	if (!adev->firmware.smu_load) {
-		r = sdma_v2_4_load_microcode(adev);
-		if (r)
-			return r;
-	} else {
-		r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
-						AMDGPU_UCODE_ID_SDMA0);
-		if (r)
-			return -EINVAL;
-		r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
-						AMDGPU_UCODE_ID_SDMA1);
-		if (r)
-			return -EINVAL;
+	if (!adev->pp_enabled) {
+		if (!adev->firmware.smu_load) {
+			r = sdma_v2_4_load_microcode(adev);
+			if (r)
+				return r;
+		} else {
+			r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
+							AMDGPU_UCODE_ID_SDMA0);
+			if (r)
+				return -EINVAL;
+			r = adev->smu.smumgr_funcs->check_fw_load_finish(adev,
+							AMDGPU_UCODE_ID_SDMA1);
+			if (r)
+				return -EINVAL;
+		}
 	}
 
 	/* halt the engine before programing */
@@ -678,20 +667,19 @@ static int sdma_v2_4_ring_test_ring(struct amdgpu_ring *ring)
  * Test a simple IB in the DMA ring (VI).
  * Returns 0 on success, error on failure.
  */
-static int sdma_v2_4_ring_test_ib(struct amdgpu_ring *ring)
+static int sdma_v2_4_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_ib ib;
 	struct fence *f = NULL;
-	unsigned i;
 	unsigned index;
-	int r;
 	u32 tmp = 0;
 	u64 gpu_addr;
+	long r;
 
 	r = amdgpu_wb_get(adev, &index);
 	if (r) {
-		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
 		return r;
 	}
 
@@ -701,7 +689,7 @@ static int sdma_v2_4_ring_test_ib(struct amdgpu_ring *ring)
 	memset(&ib, 0, sizeof(ib));
 	r = amdgpu_ib_get(adev, NULL, 256, &ib);
 	if (r) {
-		DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
 		goto err0;
 	}
 
@@ -720,28 +708,25 @@ static int sdma_v2_4_ring_test_ib(struct amdgpu_ring *ring)
 	if (r)
 		goto err1;
 
-	r = fence_wait(f, false);
-	if (r) {
-		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+	r = fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out\n");
+		r = -ETIMEDOUT;
 		goto err1;
-	}
-	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = le32_to_cpu(adev->wb.wb[index]);
-		if (tmp == 0xDEADBEEF)
-			break;
-		DRM_UDELAY(1);
-	}
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ib test on ring %d succeeded in %u usecs\n",
-			 ring->idx, i);
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
 		goto err1;
+	}
+	tmp = le32_to_cpu(adev->wb.wb[index]);
+	if (tmp == 0xDEADBEEF) {
+		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
 	} else {
 		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
 		r = -EINVAL;
 	}
 
 err1:
-	fence_put(f);
 	amdgpu_ib_free(adev, &ib, NULL);
 	fence_put(f);
 err0:
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
index 95c44942e..82bab2fb4 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
@@ -405,18 +405,6 @@ static void sdma_v3_0_ring_emit_ib(struct amdgpu_ring *ring,
 				   unsigned vm_id, bool ctx_switch)
 {
 	u32 vmid = vm_id & 0xf;
-	u32 next_rptr = ring->wptr + 5;
-
-	while ((next_rptr & 7) != 2)
-		next_rptr++;
-	next_rptr += 6;
-
-	amdgpu_ring_write(ring, SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) |
-			  SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_WRITE_LINEAR));
-	amdgpu_ring_write(ring, lower_32_bits(ring->next_rptr_gpu_addr) & 0xfffffffc);
-	amdgpu_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr));
-	amdgpu_ring_write(ring, SDMA_PKT_WRITE_UNTILED_DW_3_COUNT(1));
-	amdgpu_ring_write(ring, next_rptr);
 
 	/* IB packet must end on a 8 DW boundary */
 	sdma_v3_0_ring_insert_nop(ring, (10 - (ring->wptr & 7)) % 8);
@@ -606,7 +594,7 @@ static void sdma_v3_0_enable(struct amdgpu_device *adev, bool enable)
 	u32 f32_cntl;
 	int i;
 
-	if (enable == false) {
+	if (!enable) {
 		sdma_v3_0_gfx_stop(adev);
 		sdma_v3_0_rlc_stop(adev);
 	}
@@ -898,20 +886,19 @@ static int sdma_v3_0_ring_test_ring(struct amdgpu_ring *ring)
  * Test a simple IB in the DMA ring (VI).
  * Returns 0 on success, error on failure.
  */
-static int sdma_v3_0_ring_test_ib(struct amdgpu_ring *ring)
+static int sdma_v3_0_ring_test_ib(struct amdgpu_ring *ring, long timeout)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_ib ib;
 	struct fence *f = NULL;
-	unsigned i;
 	unsigned index;
-	int r;
 	u32 tmp = 0;
 	u64 gpu_addr;
+	long r;
 
 	r = amdgpu_wb_get(adev, &index);
 	if (r) {
-		dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r);
+		dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r);
 		return r;
 	}
 
@@ -921,7 +908,7 @@ static int sdma_v3_0_ring_test_ib(struct amdgpu_ring *ring)
 	memset(&ib, 0, sizeof(ib));
 	r = amdgpu_ib_get(adev, NULL, 256, &ib);
 	if (r) {
-		DRM_ERROR("amdgpu: failed to get ib (%d).\n", r);
+		DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r);
 		goto err0;
 	}
 
@@ -940,27 +927,24 @@ static int sdma_v3_0_ring_test_ib(struct amdgpu_ring *ring)
 	if (r)
 		goto err1;
 
-	r = fence_wait(f, false);
-	if (r) {
-		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
+	r = fence_wait_timeout(f, false, timeout);
+	if (r == 0) {
+		DRM_ERROR("amdgpu: IB test timed out\n");
+		r = -ETIMEDOUT;
 		goto err1;
-	}
-	for (i = 0; i < adev->usec_timeout; i++) {
-		tmp = le32_to_cpu(adev->wb.wb[index]);
-		if (tmp == 0xDEADBEEF)
-			break;
-		DRM_UDELAY(1);
-	}
-	if (i < adev->usec_timeout) {
-		DRM_INFO("ib test on ring %d succeeded in %u usecs\n",
-			 ring->idx, i);
+	} else if (r < 0) {
+		DRM_ERROR("amdgpu: fence wait failed (%ld).\n", r);
 		goto err1;
+	}
+	tmp = le32_to_cpu(adev->wb.wb[index]);
+	if (tmp == 0xDEADBEEF) {
+		DRM_INFO("ib test on ring %d succeeded\n", ring->idx);
+		r = 0;
 	} else {
 		DRM_ERROR("amdgpu: ib test failed (0x%08X)\n", tmp);
 		r = -EINVAL;
 	}
 err1:
-	fence_put(f);
 	amdgpu_ib_free(adev, &ib, NULL);
 	fence_put(f);
 err0:
diff --git a/drivers/gpu/drm/amd/amdgpu/tonga_smc.c b/drivers/gpu/drm/amd/amdgpu/tonga_smc.c
index 083893dd6..940de1836 100644
--- a/drivers/gpu/drm/amd/amdgpu/tonga_smc.c
+++ b/drivers/gpu/drm/amd/amdgpu/tonga_smc.c
@@ -173,7 +173,7 @@ static int tonga_send_msg_to_smc(struct amdgpu_device *adev, PPSMC_Msg msg)
 {
 	if (!tonga_is_smc_ram_running(adev))
 	{
-		return -EINVAL;;
+		return -EINVAL;
 	}
 
 	if (wait_smu_response(adev)) {
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
index f07551476..132e613ed 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
@@ -34,6 +34,8 @@
 #include "oss/oss_2_0_d.h"
 #include "oss/oss_2_0_sh_mask.h"
 
+#include "bif/bif_4_1_d.h"
+
 static void uvd_v4_2_mc_resume(struct amdgpu_device *adev);
 static void uvd_v4_2_init_cg(struct amdgpu_device *adev);
 static void uvd_v4_2_set_ring_funcs(struct amdgpu_device *adev);
@@ -439,6 +441,32 @@ static void uvd_v4_2_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq
 }
 
 /**
+ * uvd_v4_2_ring_emit_hdp_flush - emit an hdp flush
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Emits an hdp flush.
+ */
+static void uvd_v4_2_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET0(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0));
+	amdgpu_ring_write(ring, 0);
+}
+
+/**
+ * uvd_v4_2_ring_hdp_invalidate - emit an hdp invalidate
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Emits an hdp invalidate.
+ */
+static void uvd_v4_2_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET0(mmHDP_DEBUG0, 0));
+	amdgpu_ring_write(ring, 1);
+}
+
+/**
  * uvd_v4_2_ring_test_ring - register write test
  *
  * @ring: amdgpu_ring pointer
@@ -499,49 +527,6 @@ static void uvd_v4_2_ring_emit_ib(struct amdgpu_ring *ring,
 }
 
 /**
- * uvd_v4_2_ring_test_ib - test ib execution
- *
- * @ring: amdgpu_ring pointer
- *
- * Test if we can successfully execute an IB
- */
-static int uvd_v4_2_ring_test_ib(struct amdgpu_ring *ring)
-{
-	struct amdgpu_device *adev = ring->adev;
-	struct fence *fence = NULL;
-	int r;
-
-	r = amdgpu_asic_set_uvd_clocks(adev, 53300, 40000);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to raise UVD clocks (%d).\n", r);
-		return r;
-	}
-
-	r = amdgpu_uvd_get_create_msg(ring, 1, NULL);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get create msg (%d).\n", r);
-		goto error;
-	}
-
-	r = amdgpu_uvd_get_destroy_msg(ring, 1, true, &fence);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get destroy ib (%d).\n", r);
-		goto error;
-	}
-
-	r = fence_wait(fence, false);
-	if (r) {
-		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
-		goto error;
-	}
-	DRM_INFO("ib test on ring %d succeeded\n",  ring->idx);
-error:
-	fence_put(fence);
-	amdgpu_asic_set_uvd_clocks(adev, 0, 0);
-	return r;
-}
-
-/**
  * uvd_v4_2_mc_resume - memory controller programming
  *
  * @adev: amdgpu_device pointer
@@ -763,10 +748,14 @@ static const struct amdgpu_ring_funcs uvd_v4_2_ring_funcs = {
 	.parse_cs = amdgpu_uvd_ring_parse_cs,
 	.emit_ib = uvd_v4_2_ring_emit_ib,
 	.emit_fence = uvd_v4_2_ring_emit_fence,
+	.emit_hdp_flush = uvd_v4_2_ring_emit_hdp_flush,
+	.emit_hdp_invalidate = uvd_v4_2_ring_emit_hdp_invalidate,
 	.test_ring = uvd_v4_2_ring_test_ring,
-	.test_ib = uvd_v4_2_ring_test_ib,
+	.test_ib = amdgpu_uvd_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_uvd_ring_begin_use,
+	.end_use = amdgpu_uvd_ring_end_use,
 };
 
 static void uvd_v4_2_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
index e0a76a883..101de136b 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
@@ -31,6 +31,7 @@
 #include "uvd/uvd_5_0_sh_mask.h"
 #include "oss/oss_2_0_d.h"
 #include "oss/oss_2_0_sh_mask.h"
+#include "bif/bif_5_0_d.h"
 #include "vi.h"
 
 static void uvd_v5_0_set_ring_funcs(struct amdgpu_device *adev);
@@ -489,6 +490,32 @@ static void uvd_v5_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq
 }
 
 /**
+ * uvd_v5_0_ring_emit_hdp_flush - emit an hdp flush
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Emits an hdp flush.
+ */
+static void uvd_v5_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET0(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0));
+	amdgpu_ring_write(ring, 0);
+}
+
+/**
+ * uvd_v5_0_ring_hdp_invalidate - emit an hdp invalidate
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Emits an hdp invalidate.
+ */
+static void uvd_v5_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET0(mmHDP_DEBUG0, 0));
+	amdgpu_ring_write(ring, 1);
+}
+
+/**
  * uvd_v5_0_ring_test_ring - register write test
  *
  * @ring: amdgpu_ring pointer
@@ -550,49 +577,6 @@ static void uvd_v5_0_ring_emit_ib(struct amdgpu_ring *ring,
 	amdgpu_ring_write(ring, ib->length_dw);
 }
 
-/**
- * uvd_v5_0_ring_test_ib - test ib execution
- *
- * @ring: amdgpu_ring pointer
- *
- * Test if we can successfully execute an IB
- */
-static int uvd_v5_0_ring_test_ib(struct amdgpu_ring *ring)
-{
-	struct amdgpu_device *adev = ring->adev;
-	struct fence *fence = NULL;
-	int r;
-
-	r = amdgpu_asic_set_uvd_clocks(adev, 53300, 40000);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to raise UVD clocks (%d).\n", r);
-		return r;
-	}
-
-	r = amdgpu_uvd_get_create_msg(ring, 1, NULL);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get create msg (%d).\n", r);
-		goto error;
-	}
-
-	r = amdgpu_uvd_get_destroy_msg(ring, 1, true, &fence);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get destroy ib (%d).\n", r);
-		goto error;
-	}
-
-	r = fence_wait(fence, false);
-	if (r) {
-		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
-		goto error;
-	}
-	DRM_INFO("ib test on ring %d succeeded\n",  ring->idx);
-error:
-	fence_put(fence);
-	amdgpu_asic_set_uvd_clocks(adev, 0, 0);
-	return r;
-}
-
 static bool uvd_v5_0_is_idle(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
@@ -815,10 +799,14 @@ static const struct amdgpu_ring_funcs uvd_v5_0_ring_funcs = {
 	.parse_cs = amdgpu_uvd_ring_parse_cs,
 	.emit_ib = uvd_v5_0_ring_emit_ib,
 	.emit_fence = uvd_v5_0_ring_emit_fence,
+	.emit_hdp_flush = uvd_v5_0_ring_emit_hdp_flush,
+	.emit_hdp_invalidate = uvd_v5_0_ring_emit_hdp_invalidate,
 	.test_ring = uvd_v5_0_ring_test_ring,
-	.test_ib = uvd_v5_0_ring_test_ib,
+	.test_ib = amdgpu_uvd_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_uvd_ring_begin_use,
+	.end_use = amdgpu_uvd_ring_end_use,
 };
 
 static void uvd_v5_0_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
index c9929d665..7f21102bf 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
@@ -33,6 +33,8 @@
 #include "oss/oss_2_0_sh_mask.h"
 #include "smu/smu_7_1_3_d.h"
 #include "smu/smu_7_1_3_sh_mask.h"
+#include "bif/bif_5_1_d.h"
+#include "gmc/gmc_8_1_d.h"
 #include "vi.h"
 
 static void uvd_v6_0_set_ring_funcs(struct amdgpu_device *adev);
@@ -385,8 +387,8 @@ static int uvd_v6_0_start(struct amdgpu_device *adev)
 	uint32_t mp_swap_cntl;
 	int i, j, r;
 
-	/*disable DPG */
-	WREG32_P(mmUVD_POWER_STATUS, 0, ~(1 << 2));
+	/* disable DPG */
+	WREG32_P(mmUVD_POWER_STATUS, 0, ~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
 
 	/* disable byte swapping */
 	lmi_swap_cntl = 0;
@@ -405,17 +407,21 @@ static int uvd_v6_0_start(struct amdgpu_device *adev)
 	}
 
 	/* disable interupt */
-	WREG32_P(mmUVD_MASTINT_EN, 0, ~(1 << 1));
+	WREG32_P(mmUVD_MASTINT_EN, 0, ~UVD_MASTINT_EN__VCPU_EN_MASK);
 
 	/* stall UMC and register bus before resetting VCPU */
-	WREG32_P(mmUVD_LMI_CTRL2, 1 << 8, ~(1 << 8));
+	WREG32_P(mmUVD_LMI_CTRL2, UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
 	mdelay(1);
 
 	/* put LMI, VCPU, RBC etc... into reset */
-	WREG32(mmUVD_SOFT_RESET, UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK | UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__RBC_SOFT_RESET_MASK | UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
-		UVD_SOFT_RESET__CXW_SOFT_RESET_MASK | UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
+	WREG32(mmUVD_SOFT_RESET,
+		UVD_SOFT_RESET__LMI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__LBSI_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__RBC_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__CSM_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__CXW_SOFT_RESET_MASK |
+		UVD_SOFT_RESET__TAP_SOFT_RESET_MASK |
 		UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK);
 	mdelay(5);
 
@@ -424,8 +430,13 @@ static int uvd_v6_0_start(struct amdgpu_device *adev)
 	mdelay(5);
 
 	/* initialize UVD memory controller */
-	WREG32(mmUVD_LMI_CTRL, 0x40 | (1 << 8) | (1 << 13) |
-			     (1 << 21) | (1 << 9) | (1 << 20));
+	WREG32(mmUVD_LMI_CTRL,
+		(0x40 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+		UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+		UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+		UVD_LMI_CTRL__REQ_MODE_MASK |
+		UVD_LMI_CTRL__DISABLE_ON_FWV_FAIL_MASK);
 
 #ifdef __BIG_ENDIAN
 	/* swap (8 in 32) RB and IB */
@@ -447,10 +458,10 @@ static int uvd_v6_0_start(struct amdgpu_device *adev)
 	mdelay(5);
 
 	/* enable VCPU clock */
-	WREG32(mmUVD_VCPU_CNTL,  1 << 9);
+	WREG32(mmUVD_VCPU_CNTL, UVD_VCPU_CNTL__CLK_EN_MASK);
 
 	/* enable UMC */
-	WREG32_P(mmUVD_LMI_CTRL2, 0, ~(1 << 8));
+	WREG32_P(mmUVD_LMI_CTRL2, 0, ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
 
 	/* boot up the VCPU */
 	WREG32(mmUVD_SOFT_RESET, 0);
@@ -484,10 +495,12 @@ static int uvd_v6_0_start(struct amdgpu_device *adev)
 		return r;
 	}
 	/* enable master interrupt */
-	WREG32_P(mmUVD_MASTINT_EN, 3 << 1, ~(3 << 1));
+	WREG32_P(mmUVD_MASTINT_EN,
+		(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK),
+		~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK));
 
 	/* clear the bit 4 of UVD_STATUS */
-	WREG32_P(mmUVD_STATUS, 0, ~(2 << 1));
+	WREG32_P(mmUVD_STATUS, 0, ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
 
 	rb_bufsz = order_base_2(ring->ring_size);
 	tmp = 0;
@@ -581,6 +594,32 @@ static void uvd_v6_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq
 }
 
 /**
+ * uvd_v6_0_ring_emit_hdp_flush - emit an hdp flush
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Emits an hdp flush.
+ */
+static void uvd_v6_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET0(mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0));
+	amdgpu_ring_write(ring, 0);
+}
+
+/**
+ * uvd_v6_0_ring_hdp_invalidate - emit an hdp invalidate
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Emits an hdp invalidate.
+ */
+static void uvd_v6_0_ring_emit_hdp_invalidate(struct amdgpu_ring *ring)
+{
+	amdgpu_ring_write(ring, PACKET0(mmHDP_DEBUG0, 0));
+	amdgpu_ring_write(ring, 1);
+}
+
+/**
  * uvd_v6_0_ring_test_ring - register write test
  *
  * @ring: amdgpu_ring pointer
@@ -634,6 +673,9 @@ static void uvd_v6_0_ring_emit_ib(struct amdgpu_ring *ring,
 				  struct amdgpu_ib *ib,
 				  unsigned vm_id, bool ctx_switch)
 {
+	amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_VMID, 0));
+	amdgpu_ring_write(ring, vm_id);
+
 	amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_64BIT_BAR_LOW, 0));
 	amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
 	amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH, 0));
@@ -642,39 +684,55 @@ static void uvd_v6_0_ring_emit_ib(struct amdgpu_ring *ring,
 	amdgpu_ring_write(ring, ib->length_dw);
 }
 
-/**
- * uvd_v6_0_ring_test_ib - test ib execution
- *
- * @ring: amdgpu_ring pointer
- *
- * Test if we can successfully execute an IB
- */
-static int uvd_v6_0_ring_test_ib(struct amdgpu_ring *ring)
+static void uvd_v6_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
+					 unsigned vm_id, uint64_t pd_addr)
 {
-	struct fence *fence = NULL;
-	int r;
+	uint32_t reg;
 
-	r = amdgpu_uvd_get_create_msg(ring, 1, NULL);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get create msg (%d).\n", r);
-		goto error;
-	}
+	if (vm_id < 8)
+		reg = mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR + vm_id;
+	else
+		reg = mmVM_CONTEXT8_PAGE_TABLE_BASE_ADDR + vm_id - 8;
 
-	r = amdgpu_uvd_get_destroy_msg(ring, 1, true, &fence);
-	if (r) {
-		DRM_ERROR("amdgpu: failed to get destroy ib (%d).\n", r);
-		goto error;
-	}
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, reg << 2);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, pd_addr >> 12);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 0x8);
 
-	r = fence_wait(fence, false);
-	if (r) {
-		DRM_ERROR("amdgpu: fence wait failed (%d).\n", r);
-		goto error;
-	}
-	DRM_INFO("ib test on ring %d succeeded\n",  ring->idx);
-error:
-	fence_put(fence);
-	return r;
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST << 2);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, 1 << vm_id);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 0x8);
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, mmVM_INVALIDATE_REQUEST << 2);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, 0);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GP_SCRATCH8, 0));
+	amdgpu_ring_write(ring, 1 << vm_id); /* mask */
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 0xC);
+}
+
+static void uvd_v6_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring)
+{
+	uint32_t seq = ring->fence_drv.sync_seq;
+	uint64_t addr = ring->fence_drv.gpu_addr;
+
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA0, 0));
+	amdgpu_ring_write(ring, lower_32_bits(addr));
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_DATA1, 0));
+	amdgpu_ring_write(ring, upper_32_bits(addr));
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GP_SCRATCH8, 0));
+	amdgpu_ring_write(ring, 0xffffffff); /* mask */
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GP_SCRATCH9, 0));
+	amdgpu_ring_write(ring, seq);
+	amdgpu_ring_write(ring, PACKET0(mmUVD_GPCOM_VCPU_CMD, 0));
+	amdgpu_ring_write(ring, 0xE);
 }
 
 static bool uvd_v6_0_is_idle(void *handle)
@@ -847,7 +905,8 @@ static int uvd_v6_0_set_clockgating_state(void *handle,
 	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
 	static int curstate = -1;
 
-	if (adev->asic_type == CHIP_FIJI)
+	if (adev->asic_type == CHIP_FIJI ||
+			adev->asic_type == CHIP_POLARIS10)
 		uvd_v6_set_bypass_mode(adev, enable);
 
 	if (!(adev->cg_flags & AMD_CG_SUPPORT_UVD_MGCG))
@@ -912,22 +971,51 @@ const struct amd_ip_funcs uvd_v6_0_ip_funcs = {
 	.set_powergating_state = uvd_v6_0_set_powergating_state,
 };
 
-static const struct amdgpu_ring_funcs uvd_v6_0_ring_funcs = {
+static const struct amdgpu_ring_funcs uvd_v6_0_ring_phys_funcs = {
 	.get_rptr = uvd_v6_0_ring_get_rptr,
 	.get_wptr = uvd_v6_0_ring_get_wptr,
 	.set_wptr = uvd_v6_0_ring_set_wptr,
 	.parse_cs = amdgpu_uvd_ring_parse_cs,
 	.emit_ib = uvd_v6_0_ring_emit_ib,
 	.emit_fence = uvd_v6_0_ring_emit_fence,
+	.emit_hdp_flush = uvd_v6_0_ring_emit_hdp_flush,
+	.emit_hdp_invalidate = uvd_v6_0_ring_emit_hdp_invalidate,
+	.test_ring = uvd_v6_0_ring_test_ring,
+	.test_ib = amdgpu_uvd_ring_test_ib,
+	.insert_nop = amdgpu_ring_insert_nop,
+	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_uvd_ring_begin_use,
+	.end_use = amdgpu_uvd_ring_end_use,
+};
+
+static const struct amdgpu_ring_funcs uvd_v6_0_ring_vm_funcs = {
+	.get_rptr = uvd_v6_0_ring_get_rptr,
+	.get_wptr = uvd_v6_0_ring_get_wptr,
+	.set_wptr = uvd_v6_0_ring_set_wptr,
+	.parse_cs = NULL,
+	.emit_ib = uvd_v6_0_ring_emit_ib,
+	.emit_fence = uvd_v6_0_ring_emit_fence,
+	.emit_vm_flush = uvd_v6_0_ring_emit_vm_flush,
+	.emit_pipeline_sync = uvd_v6_0_ring_emit_pipeline_sync,
+	.emit_hdp_flush = uvd_v6_0_ring_emit_hdp_flush,
+	.emit_hdp_invalidate = uvd_v6_0_ring_emit_hdp_invalidate,
 	.test_ring = uvd_v6_0_ring_test_ring,
-	.test_ib = uvd_v6_0_ring_test_ib,
+	.test_ib = amdgpu_uvd_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_uvd_ring_begin_use,
+	.end_use = amdgpu_uvd_ring_end_use,
 };
 
 static void uvd_v6_0_set_ring_funcs(struct amdgpu_device *adev)
 {
-	adev->uvd.ring.funcs = &uvd_v6_0_ring_funcs;
+	if (adev->asic_type >= CHIP_POLARIS10) {
+		adev->uvd.ring.funcs = &uvd_v6_0_ring_vm_funcs;
+		DRM_INFO("UVD is enabled in VM mode\n");
+	} else {
+		adev->uvd.ring.funcs = &uvd_v6_0_ring_phys_funcs;
+		DRM_INFO("UVD is enabled in physical mode\n");
+	}
 }
 
 static const struct amdgpu_irq_src_funcs uvd_v6_0_irq_funcs = {
diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c
index 45d92aceb..80a37a602 100644
--- a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c
@@ -594,6 +594,8 @@ static const struct amdgpu_ring_funcs vce_v2_0_ring_funcs = {
 	.test_ib = amdgpu_vce_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vce_ring_begin_use,
+	.end_use = amdgpu_vce_ring_end_use,
 };
 
 static void vce_v2_0_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
index 30e8099e9..c271abffd 100644
--- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
@@ -43,6 +43,7 @@
 #define mmVCE_LMI_VCPU_CACHE_40BIT_BAR0	0x8616
 #define mmVCE_LMI_VCPU_CACHE_40BIT_BAR1	0x8617
 #define mmVCE_LMI_VCPU_CACHE_40BIT_BAR2	0x8618
+#define VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK	0x02
 
 #define VCE_V3_0_FW_SIZE	(384 * 1024)
 #define VCE_V3_0_STACK_SIZE	(64 * 1024)
@@ -51,6 +52,7 @@
 static void vce_v3_0_mc_resume(struct amdgpu_device *adev, int idx);
 static void vce_v3_0_set_ring_funcs(struct amdgpu_device *adev);
 static void vce_v3_0_set_irq_funcs(struct amdgpu_device *adev);
+static int vce_v3_0_wait_for_idle(void *handle);
 
 /**
  * vce_v3_0_ring_get_rptr - get read pointer
@@ -205,6 +207,32 @@ static void vce_v3_0_set_vce_sw_clock_gating(struct amdgpu_device *adev,
 	vce_v3_0_override_vce_clock_gating(adev, false);
 }
 
+static int vce_v3_0_firmware_loaded(struct amdgpu_device *adev)
+{
+	int i, j;
+
+	for (i = 0; i < 10; ++i) {
+		for (j = 0; j < 100; ++j) {
+			uint32_t status = RREG32(mmVCE_STATUS);
+
+			if (status & VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK)
+				return 0;
+			mdelay(10);
+		}
+
+		DRM_ERROR("VCE not responding, trying to reset the ECPU!!!\n");
+		WREG32_P(mmVCE_SOFT_RESET,
+			VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK,
+			~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+		mdelay(10);
+		WREG32_P(mmVCE_SOFT_RESET, 0,
+			~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+		mdelay(10);
+	}
+
+	return -ETIMEDOUT;
+}
+
 /**
  * vce_v3_0_start - start VCE block
  *
@@ -215,11 +243,24 @@ static void vce_v3_0_set_vce_sw_clock_gating(struct amdgpu_device *adev,
 static int vce_v3_0_start(struct amdgpu_device *adev)
 {
 	struct amdgpu_ring *ring;
-	int idx, i, j, r;
+	int idx, r;
+
+	ring = &adev->vce.ring[0];
+	WREG32(mmVCE_RB_RPTR, ring->wptr);
+	WREG32(mmVCE_RB_WPTR, ring->wptr);
+	WREG32(mmVCE_RB_BASE_LO, ring->gpu_addr);
+	WREG32(mmVCE_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+	WREG32(mmVCE_RB_SIZE, ring->ring_size / 4);
+
+	ring = &adev->vce.ring[1];
+	WREG32(mmVCE_RB_RPTR2, ring->wptr);
+	WREG32(mmVCE_RB_WPTR2, ring->wptr);
+	WREG32(mmVCE_RB_BASE_LO2, ring->gpu_addr);
+	WREG32(mmVCE_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
+	WREG32(mmVCE_RB_SIZE2, ring->ring_size / 4);
 
 	mutex_lock(&adev->grbm_idx_mutex);
 	for (idx = 0; idx < 2; ++idx) {
-
 		if (adev->vce.harvest_config & (1 << idx))
 			continue;
 
@@ -233,48 +274,24 @@ static int vce_v3_0_start(struct amdgpu_device *adev)
 
 		vce_v3_0_mc_resume(adev, idx);
 
-		/* set BUSY flag */
-		WREG32_P(mmVCE_STATUS, 1, ~1);
+		WREG32_P(mmVCE_STATUS, VCE_STATUS__JOB_BUSY_MASK,
+		         ~VCE_STATUS__JOB_BUSY_MASK);
+
 		if (adev->asic_type >= CHIP_STONEY)
 			WREG32_P(mmVCE_VCPU_CNTL, 1, ~0x200001);
 		else
 			WREG32_P(mmVCE_VCPU_CNTL, VCE_VCPU_CNTL__CLK_EN_MASK,
 				~VCE_VCPU_CNTL__CLK_EN_MASK);
 
-		WREG32_P(mmVCE_SOFT_RESET,
-			 VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK,
-			 ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
-
-		mdelay(100);
-
 		WREG32_P(mmVCE_SOFT_RESET, 0,
 			~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
 
-		for (i = 0; i < 10; ++i) {
-			uint32_t status;
-			for (j = 0; j < 100; ++j) {
-				status = RREG32(mmVCE_STATUS);
-				if (status & 2)
-					break;
-				mdelay(10);
-			}
-			r = 0;
-			if (status & 2)
-				break;
-
-			DRM_ERROR("VCE not responding, trying to reset the ECPU!!!\n");
-			WREG32_P(mmVCE_SOFT_RESET,
-				VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK,
-				~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
-			mdelay(10);
-			WREG32_P(mmVCE_SOFT_RESET, 0,
-				~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
-			mdelay(10);
-			r = -1;
-		}
+		mdelay(100);
+
+		r = vce_v3_0_firmware_loaded(adev);
 
 		/* clear BUSY flag */
-		WREG32_P(mmVCE_STATUS, 0, ~1);
+		WREG32_P(mmVCE_STATUS, 0, ~VCE_STATUS__JOB_BUSY_MASK);
 
 		/* Set Clock-Gating off */
 		if (adev->cg_flags & AMD_CG_SUPPORT_VCE_MGCG)
@@ -290,19 +307,46 @@ static int vce_v3_0_start(struct amdgpu_device *adev)
 	WREG32_P(mmGRBM_GFX_INDEX, 0, ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK);
 	mutex_unlock(&adev->grbm_idx_mutex);
 
-	ring = &adev->vce.ring[0];
-	WREG32(mmVCE_RB_RPTR, ring->wptr);
-	WREG32(mmVCE_RB_WPTR, ring->wptr);
-	WREG32(mmVCE_RB_BASE_LO, ring->gpu_addr);
-	WREG32(mmVCE_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
-	WREG32(mmVCE_RB_SIZE, ring->ring_size / 4);
+	return 0;
+}
 
-	ring = &adev->vce.ring[1];
-	WREG32(mmVCE_RB_RPTR2, ring->wptr);
-	WREG32(mmVCE_RB_WPTR2, ring->wptr);
-	WREG32(mmVCE_RB_BASE_LO2, ring->gpu_addr);
-	WREG32(mmVCE_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
-	WREG32(mmVCE_RB_SIZE2, ring->ring_size / 4);
+static int vce_v3_0_stop(struct amdgpu_device *adev)
+{
+	int idx;
+
+	mutex_lock(&adev->grbm_idx_mutex);
+	for (idx = 0; idx < 2; ++idx) {
+		if (adev->vce.harvest_config & (1 << idx))
+			continue;
+
+		if (idx == 0)
+			WREG32_P(mmGRBM_GFX_INDEX, 0,
+				~GRBM_GFX_INDEX__VCE_INSTANCE_MASK);
+		else
+			WREG32_P(mmGRBM_GFX_INDEX,
+				GRBM_GFX_INDEX__VCE_INSTANCE_MASK,
+				~GRBM_GFX_INDEX__VCE_INSTANCE_MASK);
+
+		if (adev->asic_type >= CHIP_STONEY)
+			WREG32_P(mmVCE_VCPU_CNTL, 0, ~0x200001);
+		else
+			WREG32_P(mmVCE_VCPU_CNTL, 0,
+				~VCE_VCPU_CNTL__CLK_EN_MASK);
+		/* hold on ECPU */
+		WREG32_P(mmVCE_SOFT_RESET,
+			 VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK,
+			 ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK);
+
+		/* clear BUSY flag */
+		WREG32_P(mmVCE_STATUS, 0, ~VCE_STATUS__JOB_BUSY_MASK);
+
+		/* Set Clock-Gating off */
+		if (adev->cg_flags & AMD_CG_SUPPORT_VCE_MGCG)
+			vce_v3_0_set_vce_sw_clock_gating(adev, false);
+	}
+
+	WREG32_P(mmGRBM_GFX_INDEX, 0, ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK);
+	mutex_unlock(&adev->grbm_idx_mutex);
 
 	return 0;
 }
@@ -441,7 +485,14 @@ static int vce_v3_0_hw_init(void *handle)
 
 static int vce_v3_0_hw_fini(void *handle)
 {
-	return 0;
+	int r;
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	r = vce_v3_0_wait_for_idle(handle);
+	if (r)
+		return r;
+
+	return vce_v3_0_stop(adev);
 }
 
 static int vce_v3_0_suspend(void *handle)
@@ -604,6 +655,18 @@ static int vce_v3_0_process_interrupt(struct amdgpu_device *adev,
 	return 0;
 }
 
+static void vce_v3_set_bypass_mode(struct amdgpu_device *adev, bool enable)
+{
+	u32 tmp = RREG32_SMC(ixGCK_DFS_BYPASS_CNTL);
+
+	if (enable)
+		tmp |= GCK_DFS_BYPASS_CNTL__BYPASSECLK_MASK;
+	else
+		tmp &= ~GCK_DFS_BYPASS_CNTL__BYPASSECLK_MASK;
+
+	WREG32_SMC(ixGCK_DFS_BYPASS_CNTL, tmp);
+}
+
 static int vce_v3_0_set_clockgating_state(void *handle,
 					  enum amd_clockgating_state state)
 {
@@ -611,6 +674,9 @@ static int vce_v3_0_set_clockgating_state(void *handle,
 	bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
 	int i;
 
+	if (adev->asic_type == CHIP_POLARIS10)
+		vce_v3_set_bypass_mode(adev, enable);
+
 	if (!(adev->cg_flags & AMD_CG_SUPPORT_VCE_MGCG))
 		return 0;
 
@@ -701,6 +767,8 @@ static const struct amdgpu_ring_funcs vce_v3_0_ring_funcs = {
 	.test_ib = amdgpu_vce_ring_test_ib,
 	.insert_nop = amdgpu_ring_insert_nop,
 	.pad_ib = amdgpu_ring_generic_pad_ib,
+	.begin_use = amdgpu_vce_ring_begin_use,
+	.end_use = amdgpu_vce_ring_end_use,
 };
 
 static void vce_v3_0_set_ring_funcs(struct amdgpu_device *adev)
diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c
index d8fca2e11..d2c7e4962 100644
--- a/drivers/gpu/drm/amd/amdgpu/vi.c
+++ b/drivers/gpu/drm/amd/amdgpu/vi.c
@@ -200,6 +200,29 @@ static void vi_didt_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
 	spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
 }
 
+static u32 vi_gc_cac_rreg(struct amdgpu_device *adev, u32 reg)
+{
+	unsigned long flags;
+	u32 r;
+
+	spin_lock_irqsave(&adev->gc_cac_idx_lock, flags);
+	WREG32(mmGC_CAC_IND_INDEX, (reg));
+	r = RREG32(mmGC_CAC_IND_DATA);
+	spin_unlock_irqrestore(&adev->gc_cac_idx_lock, flags);
+	return r;
+}
+
+static void vi_gc_cac_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&adev->gc_cac_idx_lock, flags);
+	WREG32(mmGC_CAC_IND_INDEX, (reg));
+	WREG32(mmGC_CAC_IND_DATA, (v));
+	spin_unlock_irqrestore(&adev->gc_cac_idx_lock, flags);
+}
+
+
 static const u32 tonga_mgcg_cgcg_init[] =
 {
 	mmCGTT_DRM_CLK_CTRL0, 0xffffffff, 0x00600100,
@@ -530,12 +553,12 @@ static uint32_t vi_read_indexed_register(struct amdgpu_device *adev, u32 se_num,
 
 	mutex_lock(&adev->grbm_idx_mutex);
 	if (se_num != 0xffffffff || sh_num != 0xffffffff)
-		gfx_v8_0_select_se_sh(adev, se_num, sh_num);
+		amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
 
 	val = RREG32(reg_offset);
 
 	if (se_num != 0xffffffff || sh_num != 0xffffffff)
-		gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff);
+		amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
 	mutex_unlock(&adev->grbm_idx_mutex);
 	return val;
 }
@@ -594,7 +617,7 @@ static int vi_read_register(struct amdgpu_device *adev, u32 se_num,
 	return -EINVAL;
 }
 
-static void vi_gpu_pci_config_reset(struct amdgpu_device *adev)
+static int vi_gpu_pci_config_reset(struct amdgpu_device *adev)
 {
 	u32 i;
 
@@ -609,11 +632,14 @@ static void vi_gpu_pci_config_reset(struct amdgpu_device *adev)
 
 	/* wait for asic to come out of reset */
 	for (i = 0; i < adev->usec_timeout; i++) {
-		if (RREG32(mmCONFIG_MEMSIZE) != 0xffffffff)
-			break;
+		if (RREG32(mmCONFIG_MEMSIZE) != 0xffffffff) {
+			/* enable BM */
+			pci_set_master(adev->pdev);
+			return 0;
+		}
 		udelay(1);
 	}
-
+	return -EINVAL;
 }
 
 static void vi_set_bios_scratch_engine_hung(struct amdgpu_device *adev, bool hung)
@@ -639,13 +665,15 @@ static void vi_set_bios_scratch_engine_hung(struct amdgpu_device *adev, bool hun
  */
 static int vi_asic_reset(struct amdgpu_device *adev)
 {
+	int r;
+
 	vi_set_bios_scratch_engine_hung(adev, true);
 
-	vi_gpu_pci_config_reset(adev);
+	r = vi_gpu_pci_config_reset(adev);
 
 	vi_set_bios_scratch_engine_hung(adev, false);
 
-	return 0;
+	return r;
 }
 
 static int vi_set_uvd_clock(struct amdgpu_device *adev, u32 clock,
@@ -1130,9 +1158,6 @@ static const struct amdgpu_asic_funcs vi_asic_funcs =
 	.set_uvd_clocks = &vi_set_uvd_clocks,
 	.set_vce_clocks = &vi_set_vce_clocks,
 	.get_virtual_caps = &vi_get_virtual_caps,
-	/* these should be moved to their own ip modules */
-	.get_gpu_clock_counter = &gfx_v8_0_get_gpu_clock_counter,
-	.wait_for_mc_idle = &gmc_v8_0_mc_wait_for_idle,
 };
 
 static int vi_common_early_init(void *handle)
@@ -1153,6 +1178,8 @@ static int vi_common_early_init(void *handle)
 	adev->uvd_ctx_wreg = &vi_uvd_ctx_wreg;
 	adev->didt_rreg = &vi_didt_rreg;
 	adev->didt_wreg = &vi_didt_wreg;
+	adev->gc_cac_rreg = &vi_gc_cac_rreg;
+	adev->gc_cac_wreg = &vi_gc_cac_wreg;
 
 	adev->asic_funcs = &vi_asic_funcs;
 
@@ -1226,12 +1253,18 @@ static int vi_common_early_init(void *handle)
 		adev->cg_flags = AMD_CG_SUPPORT_UVD_MGCG |
 			AMD_CG_SUPPORT_GFX_MGCG |
 			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_RLC_LS |
+			AMD_CG_SUPPORT_GFX_CP_LS |
+			AMD_CG_SUPPORT_GFX_CGTS |
+			AMD_CG_SUPPORT_GFX_MGLS |
+			AMD_CG_SUPPORT_GFX_CGTS_LS |
+			AMD_CG_SUPPORT_GFX_CGCG |
+			AMD_CG_SUPPORT_GFX_CGLS |
 			AMD_CG_SUPPORT_BIF_LS |
 			AMD_CG_SUPPORT_HDP_MGCG |
 			AMD_CG_SUPPORT_HDP_LS |
 			AMD_CG_SUPPORT_SDMA_MGCG |
 			AMD_CG_SUPPORT_SDMA_LS;
-		adev->pg_flags = 0;
 		adev->external_rev_id = adev->rev_id + 0x1;
 		break;
 	default: