Merge drm-upstream/drm-next into drm-intel-next-queued

Catch up with upstream while it's easy.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>

64 files changed, 1405 insertions, 1436 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 51d1364cf185..12e71bbfd222 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -96,6 +96,7 @@ extern int amdgpu_bapm;
 extern int amdgpu_deep_color;
 extern int amdgpu_vm_size;
 extern int amdgpu_vm_block_size;
+extern int amdgpu_vm_fragment_size;
 extern int amdgpu_vm_fault_stop;
 extern int amdgpu_vm_debug;
 extern int amdgpu_vm_update_mode;
@@ -373,78 +374,10 @@ struct amdgpu_clock {
 };
 
 /*
- * BO.
+ * GEM.
  */
-struct amdgpu_bo_list_entry {
-	struct amdgpu_bo		*robj;
-	struct ttm_validate_buffer	tv;
-	struct amdgpu_bo_va		*bo_va;
-	uint32_t			priority;
-	struct page			**user_pages;
-	int				user_invalidated;
-};
-
-struct amdgpu_bo_va_mapping {
-	struct list_head		list;
-	struct rb_node			rb;
-	uint64_t			start;
-	uint64_t			last;
-	uint64_t			__subtree_last;
-	uint64_t			offset;
-	uint64_t			flags;
-};
-
-/* bo virtual addresses in a specific vm */
-struct amdgpu_bo_va {
-	/* protected by bo being reserved */
-	struct list_head		bo_list;
-	struct dma_fence	        *last_pt_update;
-	unsigned			ref_count;
-
-	/* protected by vm mutex and spinlock */
-	struct list_head		vm_status;
-
-	/* mappings for this bo_va */
-	struct list_head		invalids;
-	struct list_head		valids;
-
-	/* constant after initialization */
-	struct amdgpu_vm		*vm;
-	struct amdgpu_bo		*bo;
-};
 
 #define AMDGPU_GEM_DOMAIN_MAX		0x3
-
-struct amdgpu_bo {
-	/* Protected by tbo.reserved */
-	u32				prefered_domains;
-	u32				allowed_domains;
-	struct ttm_place		placements[AMDGPU_GEM_DOMAIN_MAX + 1];
-	struct ttm_placement		placement;
-	struct ttm_buffer_object	tbo;
-	struct ttm_bo_kmap_obj		kmap;
-	u64				flags;
-	unsigned			pin_count;
-	void				*kptr;
-	u64				tiling_flags;
-	u64				metadata_flags;
-	void				*metadata;
-	u32				metadata_size;
-	unsigned			prime_shared_count;
-	/* list of all virtual address to which this bo
-	 * is associated to
-	 */
-	struct list_head		va;
-	/* Constant after initialization */
-	struct drm_gem_object		gem_base;
-	struct amdgpu_bo		*parent;
-	struct amdgpu_bo		*shadow;
-
-	struct ttm_bo_kmap_obj		dma_buf_vmap;
-	struct amdgpu_mn		*mn;
-	struct list_head		mn_list;
-	struct list_head		shadow_list;
-};
 #define gem_to_amdgpu_bo(gobj) container_of((gobj), struct amdgpu_bo, gem_base)
 
 void amdgpu_gem_object_free(struct drm_gem_object *obj);
@@ -678,15 +611,15 @@ typedef enum _AMDGPU_DOORBELL64_ASSIGNMENT
 	/* overlap the doorbell assignment with VCN as they are  mutually exclusive
 	 * VCE engine's doorbell is 32 bit and two VCE ring share one QWORD
 	 */
-	AMDGPU_DOORBELL64_RING0_1                 = 0xF8,
-	AMDGPU_DOORBELL64_RING2_3                 = 0xF9,
-	AMDGPU_DOORBELL64_RING4_5                 = 0xFA,
-	AMDGPU_DOORBELL64_RING6_7                 = 0xFB,
+	AMDGPU_DOORBELL64_UVD_RING0_1             = 0xF8,
+	AMDGPU_DOORBELL64_UVD_RING2_3             = 0xF9,
+	AMDGPU_DOORBELL64_UVD_RING4_5             = 0xFA,
+	AMDGPU_DOORBELL64_UVD_RING6_7             = 0xFB,
 
-	AMDGPU_DOORBELL64_UVD_RING0_1             = 0xFC,
-	AMDGPU_DOORBELL64_UVD_RING2_3             = 0xFD,
-	AMDGPU_DOORBELL64_UVD_RING4_5             = 0xFE,
-	AMDGPU_DOORBELL64_UVD_RING6_7             = 0xFF,
+	AMDGPU_DOORBELL64_VCE_RING0_1             = 0xFC,
+	AMDGPU_DOORBELL64_VCE_RING2_3             = 0xFD,
+	AMDGPU_DOORBELL64_VCE_RING4_5             = 0xFE,
+	AMDGPU_DOORBELL64_VCE_RING6_7             = 0xFF,
 
 	AMDGPU_DOORBELL64_MAX_ASSIGNMENT          = 0xFF,
 	AMDGPU_DOORBELL64_INVALID                 = 0xFFFF
@@ -816,6 +749,7 @@ void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr);
 struct amdgpu_fpriv {
 	struct amdgpu_vm	vm;
 	struct amdgpu_bo_va	*prt_va;
+	struct amdgpu_bo_va	*csa_va;
 	struct mutex		bo_list_lock;
 	struct idr		bo_list_handles;
 	struct amdgpu_ctx_mgr	ctx_mgr;
@@ -825,6 +759,14 @@ struct amdgpu_fpriv {
 /*
  * residency list
  */
+struct amdgpu_bo_list_entry {
+	struct amdgpu_bo		*robj;
+	struct ttm_validate_buffer	tv;
+	struct amdgpu_bo_va		*bo_va;
+	uint32_t			priority;
+	struct page			**user_pages;
+	int				user_invalidated;
+};
 
 struct amdgpu_bo_list {
 	struct mutex lock;
@@ -1191,10 +1133,6 @@ struct amdgpu_wb {
 
 int amdgpu_wb_get(struct amdgpu_device *adev, u32 *wb);
 void amdgpu_wb_free(struct amdgpu_device *adev, u32 wb);
-int amdgpu_wb_get_64bit(struct amdgpu_device *adev, u32 *wb);
-int amdgpu_wb_get_256Bit(struct amdgpu_device *adev, u32 *wb);
-void amdgpu_wb_free_64bit(struct amdgpu_device *adev, u32 wb);
-void amdgpu_wb_free_256bit(struct amdgpu_device *adev, u32 wb);
 
 void amdgpu_get_pcie_info(struct amdgpu_device *adev);
 
@@ -1488,7 +1426,7 @@ struct amdgpu_device {
 	bool				is_atom_fw;
 	uint8_t				*bios;
 	uint32_t			bios_size;
-	struct amdgpu_bo		*stollen_vga_memory;
+	struct amdgpu_bo		*stolen_vga_memory;
 	uint32_t			bios_scratch_reg_offset;
 	uint32_t			bios_scratch[AMDGPU_BIOS_NUM_SCRATCH];
 
@@ -1546,9 +1484,6 @@ struct amdgpu_device {
 	struct amdgpu_mman		mman;
 	struct amdgpu_vram_scratch	vram_scratch;
 	struct amdgpu_wb		wb;
-	atomic64_t			vram_usage;
-	atomic64_t			vram_vis_usage;
-	atomic64_t			gtt_usage;
 	atomic64_t			num_bytes_moved;
 	atomic64_t			num_evictions;
 	atomic64_t			num_vram_cpu_page_faults;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c
index 06879d1dcabd..a52795d9b458 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c
@@ -285,19 +285,20 @@ static int acp_hw_init(void *handle)
 		return 0;
 	else if (r)
 		return r;
+	if (adev->asic_type != CHIP_STONEY) {
+		adev->acp.acp_genpd = kzalloc(sizeof(struct acp_pm_domain), GFP_KERNEL);
+		if (adev->acp.acp_genpd == NULL)
+			return -ENOMEM;
 
-	adev->acp.acp_genpd = kzalloc(sizeof(struct acp_pm_domain), GFP_KERNEL);
-	if (adev->acp.acp_genpd == NULL)
-		return -ENOMEM;
-
-	adev->acp.acp_genpd->gpd.name = "ACP_AUDIO";
-	adev->acp.acp_genpd->gpd.power_off = acp_poweroff;
-	adev->acp.acp_genpd->gpd.power_on = acp_poweron;
+		adev->acp.acp_genpd->gpd.name = "ACP_AUDIO";
+		adev->acp.acp_genpd->gpd.power_off = acp_poweroff;
+		adev->acp.acp_genpd->gpd.power_on = acp_poweron;
 
 
-	adev->acp.acp_genpd->cgs_dev = adev->acp.cgs_device;
+		adev->acp.acp_genpd->cgs_dev = adev->acp.cgs_device;
 
-	pm_genpd_init(&adev->acp.acp_genpd->gpd, NULL, false);
+		pm_genpd_init(&adev->acp.acp_genpd->gpd, NULL, false);
+	}
 
 	adev->acp.acp_cell = kzalloc(sizeof(struct mfd_cell) * ACP_DEVS,
 							GFP_KERNEL);
@@ -319,14 +320,29 @@ static int acp_hw_init(void *handle)
 		return -ENOMEM;
 	}
 
-	i2s_pdata[0].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET;
+	switch (adev->asic_type) {
+	case CHIP_STONEY:
+		i2s_pdata[0].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
+			DW_I2S_QUIRK_16BIT_IDX_OVERRIDE;
+		break;
+	default:
+		i2s_pdata[0].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET;
+	}
 	i2s_pdata[0].cap = DWC_I2S_PLAY;
 	i2s_pdata[0].snd_rates = SNDRV_PCM_RATE_8000_96000;
 	i2s_pdata[0].i2s_reg_comp1 = ACP_I2S_COMP1_PLAY_REG_OFFSET;
 	i2s_pdata[0].i2s_reg_comp2 = ACP_I2S_COMP2_PLAY_REG_OFFSET;
+	switch (adev->asic_type) {
+	case CHIP_STONEY:
+		i2s_pdata[1].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
+			DW_I2S_QUIRK_COMP_PARAM1 |
+			DW_I2S_QUIRK_16BIT_IDX_OVERRIDE;
+		break;
+	default:
+		i2s_pdata[1].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
+			DW_I2S_QUIRK_COMP_PARAM1;
+	}
 
-	i2s_pdata[1].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
-				DW_I2S_QUIRK_COMP_PARAM1;
 	i2s_pdata[1].cap = DWC_I2S_RECORD;
 	i2s_pdata[1].snd_rates = SNDRV_PCM_RATE_8000_96000;
 	i2s_pdata[1].i2s_reg_comp1 = ACP_I2S_COMP1_CAP_REG_OFFSET;
@@ -373,12 +389,14 @@ static int acp_hw_init(void *handle)
 	if (r)
 		return r;
 
-	for (i = 0; i < ACP_DEVS ; i++) {
-		dev = get_mfd_cell_dev(adev->acp.acp_cell[i].name, i);
-		r = pm_genpd_add_device(&adev->acp.acp_genpd->gpd, dev);
-		if (r) {
-			dev_err(dev, "Failed to add dev to genpd\n");
-			return r;
+	if (adev->asic_type != CHIP_STONEY) {
+		for (i = 0; i < ACP_DEVS ; i++) {
+			dev = get_mfd_cell_dev(adev->acp.acp_cell[i].name, i);
+			r = pm_genpd_add_device(&adev->acp.acp_genpd->gpd, dev);
+			if (r) {
+				dev_err(dev, "Failed to add dev to genpd\n");
+				return r;
+			}
 		}
 	}
 
@@ -398,20 +416,22 @@ static int acp_hw_fini(void *handle)
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	/* return early if no ACP */
-	if (!adev->acp.acp_genpd)
+	if (!adev->acp.acp_cell)
 		return 0;
 
-	for (i = 0; i < ACP_DEVS ; i++) {
-		dev = get_mfd_cell_dev(adev->acp.acp_cell[i].name, i);
-		ret = pm_genpd_remove_device(&adev->acp.acp_genpd->gpd, dev);
-		/* If removal fails, dont giveup and try rest */
-		if (ret)
-			dev_err(dev, "remove dev from genpd failed\n");
+	if (adev->acp.acp_genpd) {
+		for (i = 0; i < ACP_DEVS ; i++) {
+			dev = get_mfd_cell_dev(adev->acp.acp_cell[i].name, i);
+			ret = pm_genpd_remove_device(&adev->acp.acp_genpd->gpd, dev);
+			/* If removal fails, dont giveup and try rest */
+			if (ret)
+				dev_err(dev, "remove dev from genpd failed\n");
+		}
+		kfree(adev->acp.acp_genpd);
 	}
 
 	mfd_remove_devices(adev->acp.parent);
 	kfree(adev->acp.acp_res);
-	kfree(adev->acp.acp_genpd);
 	kfree(adev->acp.acp_cell);
 
 	return 0;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
index ef79551b4cb7..57afad79f55d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
@@ -30,10 +30,10 @@
 #include <drm/drmP.h>
 #include <drm/drm_crtc_helper.h>
 #include "amdgpu.h"
+#include "amdgpu_pm.h"
 #include "amd_acpi.h"
 #include "atom.h"
 
-extern void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev);
 /* Call the ATIF method
  */
 /**
@@ -289,7 +289,7 @@ out:
  * handles it.
  * Returns NOTIFY code
  */
-int amdgpu_atif_handler(struct amdgpu_device *adev,
+static int amdgpu_atif_handler(struct amdgpu_device *adev,
 			struct acpi_bus_event *event)
 {
 	struct amdgpu_atif *atif = &adev->atif;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index 37971d9402e3..5432af39a674 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -27,16 +27,15 @@
 #include "amdgpu_gfx.h"
 #include <linux/module.h>
 
-const struct kfd2kgd_calls *kfd2kgd;
 const struct kgd2kfd_calls *kgd2kfd;
-bool (*kgd2kfd_init_p)(unsigned, const struct kgd2kfd_calls**);
+bool (*kgd2kfd_init_p)(unsigned int, const struct kgd2kfd_calls**);
 
 int amdgpu_amdkfd_init(void)
 {
 	int ret;
 
 #if defined(CONFIG_HSA_AMD_MODULE)
-	int (*kgd2kfd_init_p)(unsigned, const struct kgd2kfd_calls**);
+	int (*kgd2kfd_init_p)(unsigned int, const struct kgd2kfd_calls**);
 
 	kgd2kfd_init_p = symbol_request(kgd2kfd_init);
 
@@ -61,8 +60,21 @@ int amdgpu_amdkfd_init(void)
 	return ret;
 }
 
-bool amdgpu_amdkfd_load_interface(struct amdgpu_device *adev)
+void amdgpu_amdkfd_fini(void)
+{
+	if (kgd2kfd) {
+		kgd2kfd->exit();
+		symbol_put(kgd2kfd_init);
+	}
+}
+
+void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
 {
+	const struct kfd2kgd_calls *kfd2kgd;
+
+	if (!kgd2kfd)
+		return;
+
 	switch (adev->asic_type) {
 #ifdef CONFIG_DRM_AMDGPU_CIK
 	case CHIP_KAVERI:
@@ -73,25 +85,12 @@ bool amdgpu_amdkfd_load_interface(struct amdgpu_device *adev)
 		kfd2kgd = amdgpu_amdkfd_gfx_8_0_get_functions();
 		break;
 	default:
-		return false;
+		dev_info(adev->dev, "kfd not supported on this ASIC\n");
+		return;
 	}
 
-	return true;
-}
-
-void amdgpu_amdkfd_fini(void)
-{
-	if (kgd2kfd) {
-		kgd2kfd->exit();
-		symbol_put(kgd2kfd_init);
-	}
-}
-
-void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
-{
-	if (kgd2kfd)
-		adev->kfd = kgd2kfd->probe((struct kgd_dev *)adev,
-					adev->pdev, kfd2kgd);
+	adev->kfd = kgd2kfd->probe((struct kgd_dev *)adev,
+				   adev->pdev, kfd2kgd);
 }
 
 void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
@@ -184,7 +183,8 @@ int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
 		return -ENOMEM;
 
 	r = amdgpu_bo_create(adev, size, PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_GTT,
-			     AMDGPU_GEM_CREATE_CPU_GTT_USWC, NULL, NULL, &(*mem)->bo);
+			     AMDGPU_GEM_CREATE_CPU_GTT_USWC, NULL, NULL, 0,
+			     &(*mem)->bo);
 	if (r) {
 		dev_err(adev->dev,
 			"failed to allocate BO for amdkfd (%d)\n", r);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
index 73f83a10ae14..8d689ab7e429 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
@@ -26,6 +26,7 @@
 #define AMDGPU_AMDKFD_H_INCLUDED
 
 #include <linux/types.h>
+#include <linux/mmu_context.h>
 #include <kgd_kfd_interface.h>
 
 struct amdgpu_device;
@@ -39,8 +40,6 @@ struct kgd_mem {
 int amdgpu_amdkfd_init(void);
 void amdgpu_amdkfd_fini(void);
 
-bool amdgpu_amdkfd_load_interface(struct amdgpu_device *adev);
-
 void amdgpu_amdkfd_suspend(struct amdgpu_device *adev);
 int amdgpu_amdkfd_resume(struct amdgpu_device *adev);
 void amdgpu_amdkfd_interrupt(struct amdgpu_device *adev,
@@ -62,4 +61,19 @@ uint64_t get_gpu_clock_counter(struct kgd_dev *kgd);
 
 uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd);
 
+#define read_user_wptr(mmptr, wptr, dst)				\
+	({								\
+		bool valid = false;					\
+		if ((mmptr) && (wptr)) {				\
+			if ((mmptr) == current->mm) {			\
+				valid = !get_user((dst), (wptr));	\
+			} else if (current->mm == NULL) {		\
+				use_mm(mmptr);				\
+				valid = !get_user((dst), (wptr));	\
+				unuse_mm(mmptr);			\
+			}						\
+		}							\
+		valid;							\
+	})
+
 #endif /* AMDGPU_AMDKFD_H_INCLUDED */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
index 5254562fd0f9..b9dbbf9cb8b0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
@@ -39,6 +39,12 @@
 #include "gmc/gmc_7_1_sh_mask.h"
 #include "cik_structs.h"
 
+enum hqd_dequeue_request_type {
+	NO_ACTION = 0,
+	DRAIN_PIPE,
+	RESET_WAVES
+};
+
 enum {
 	MAX_TRAPID = 8,		/* 3 bits in the bitfield. */
 	MAX_WATCH_ADDRESSES = 4
@@ -96,12 +102,15 @@ static int kgd_init_pipeline(struct kgd_dev *kgd, uint32_t pipe_id,
 				uint32_t hpd_size, uint64_t hpd_gpu_addr);
 static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id);
 static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
-			uint32_t queue_id, uint32_t __user *wptr);
+			uint32_t queue_id, uint32_t __user *wptr,
+			uint32_t wptr_shift, uint32_t wptr_mask,
+			struct mm_struct *mm);
 static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd);
 static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
 				uint32_t pipe_id, uint32_t queue_id);
 
-static int kgd_hqd_destroy(struct kgd_dev *kgd, uint32_t reset_type,
+static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
+				enum kfd_preempt_type reset_type,
 				unsigned int utimeout, uint32_t pipe_id,
 				uint32_t queue_id);
 static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd);
@@ -126,6 +135,33 @@ static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
 static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid);
 
 static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type);
+static void set_scratch_backing_va(struct kgd_dev *kgd,
+					uint64_t va, uint32_t vmid);
+
+/* Because of REG_GET_FIELD() being used, we put this function in the
+ * asic specific file.
+ */
+static int get_tile_config(struct kgd_dev *kgd,
+		struct tile_config *config)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+
+	config->gb_addr_config = adev->gfx.config.gb_addr_config;
+	config->num_banks = REG_GET_FIELD(adev->gfx.config.mc_arb_ramcfg,
+				MC_ARB_RAMCFG, NOOFBANK);
+	config->num_ranks = REG_GET_FIELD(adev->gfx.config.mc_arb_ramcfg,
+				MC_ARB_RAMCFG, NOOFRANKS);
+
+	config->tile_config_ptr = adev->gfx.config.tile_mode_array;
+	config->num_tile_configs =
+			ARRAY_SIZE(adev->gfx.config.tile_mode_array);
+	config->macro_tile_config_ptr =
+			adev->gfx.config.macrotile_mode_array;
+	config->num_macro_tile_configs =
+			ARRAY_SIZE(adev->gfx.config.macrotile_mode_array);
+
+	return 0;
+}
 
 static const struct kfd2kgd_calls kfd2kgd = {
 	.init_gtt_mem_allocation = alloc_gtt_mem,
@@ -150,7 +186,9 @@ static const struct kfd2kgd_calls kfd2kgd = {
 	.get_atc_vmid_pasid_mapping_pasid = get_atc_vmid_pasid_mapping_pasid,
 	.get_atc_vmid_pasid_mapping_valid = get_atc_vmid_pasid_mapping_valid,
 	.write_vmid_invalidate_request = write_vmid_invalidate_request,
-	.get_fw_version = get_fw_version
+	.get_fw_version = get_fw_version,
+	.set_scratch_backing_va = set_scratch_backing_va,
+	.get_tile_config = get_tile_config,
 };
 
 struct kfd2kgd_calls *amdgpu_amdkfd_gfx_7_get_functions(void)
@@ -186,7 +224,7 @@ static void acquire_queue(struct kgd_dev *kgd, uint32_t pipe_id,
 {
 	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 
-	uint32_t mec = (++pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	uint32_t mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
 	uint32_t pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
 
 	lock_srbm(kgd, mec, pipe, queue_id, 0);
@@ -290,20 +328,38 @@ static inline struct cik_sdma_rlc_registers *get_sdma_mqd(void *mqd)
 }
 
 static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
-			uint32_t queue_id, uint32_t __user *wptr)
+			uint32_t queue_id, uint32_t __user *wptr,
+			uint32_t wptr_shift, uint32_t wptr_mask,
+			struct mm_struct *mm)
 {
 	struct amdgpu_device *adev = get_amdgpu_device(kgd);
-	uint32_t wptr_shadow, is_wptr_shadow_valid;
 	struct cik_mqd *m;
+	uint32_t *mqd_hqd;
+	uint32_t reg, wptr_val, data;
 
 	m = get_mqd(mqd);
 
-	is_wptr_shadow_valid = !get_user(wptr_shadow, wptr);
-	if (is_wptr_shadow_valid)
-		m->cp_hqd_pq_wptr = wptr_shadow;
-
 	acquire_queue(kgd, pipe_id, queue_id);
-	gfx_v7_0_mqd_commit(adev, m);
+
+	/* HQD registers extend from CP_MQD_BASE_ADDR to CP_MQD_CONTROL. */
+	mqd_hqd = &m->cp_mqd_base_addr_lo;
+
+	for (reg = mmCP_MQD_BASE_ADDR; reg <= mmCP_MQD_CONTROL; reg++)
+		WREG32(reg, mqd_hqd[reg - mmCP_MQD_BASE_ADDR]);
+
+	/* Copy userspace write pointer value to register.
+	 * Activate doorbell logic to monitor subsequent changes.
+	 */
+	data = REG_SET_FIELD(m->cp_hqd_pq_doorbell_control,
+			     CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 1);
+	WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL, data);
+
+	if (read_user_wptr(mm, wptr, wptr_val))
+		WREG32(mmCP_HQD_PQ_WPTR, (wptr_val << wptr_shift) & wptr_mask);
+
+	data = REG_SET_FIELD(m->cp_hqd_active, CP_HQD_ACTIVE, ACTIVE, 1);
+	WREG32(mmCP_HQD_ACTIVE, data);
+
 	release_queue(kgd);
 
 	return 0;
@@ -382,30 +438,99 @@ static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd)
 	return false;
 }
 
-static int kgd_hqd_destroy(struct kgd_dev *kgd, uint32_t reset_type,
+static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
+				enum kfd_preempt_type reset_type,
 				unsigned int utimeout, uint32_t pipe_id,
 				uint32_t queue_id)
 {
 	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	uint32_t temp;
-	int timeout = utimeout;
+	enum hqd_dequeue_request_type type;
+	unsigned long flags, end_jiffies;
+	int retry;
 
 	acquire_queue(kgd, pipe_id, queue_id);
 	WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL, 0);
 
-	WREG32(mmCP_HQD_DEQUEUE_REQUEST, reset_type);
+	switch (reset_type) {
+	case KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN:
+		type = DRAIN_PIPE;
+		break;
+	case KFD_PREEMPT_TYPE_WAVEFRONT_RESET:
+		type = RESET_WAVES;
+		break;
+	default:
+		type = DRAIN_PIPE;
+		break;
+	}
+
+	/* Workaround: If IQ timer is active and the wait time is close to or
+	 * equal to 0, dequeueing is not safe. Wait until either the wait time
+	 * is larger or timer is cleared. Also, ensure that IQ_REQ_PEND is
+	 * cleared before continuing. Also, ensure wait times are set to at
+	 * least 0x3.
+	 */
+	local_irq_save(flags);
+	preempt_disable();
+	retry = 5000; /* wait for 500 usecs at maximum */
+	while (true) {
+		temp = RREG32(mmCP_HQD_IQ_TIMER);
+		if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, PROCESSING_IQ)) {
+			pr_debug("HW is processing IQ\n");
+			goto loop;
+		}
+		if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, ACTIVE)) {
+			if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, RETRY_TYPE)
+					== 3) /* SEM-rearm is safe */
+				break;
+			/* Wait time 3 is safe for CP, but our MMIO read/write
+			 * time is close to 1 microsecond, so check for 10 to
+			 * leave more buffer room
+			 */
+			if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, WAIT_TIME)
+					>= 10)
+				break;
+			pr_debug("IQ timer is active\n");
+		} else
+			break;
+loop:
+		if (!retry) {
+			pr_err("CP HQD IQ timer status time out\n");
+			break;
+		}
+		ndelay(100);
+		--retry;
+	}
+	retry = 1000;
+	while (true) {
+		temp = RREG32(mmCP_HQD_DEQUEUE_REQUEST);
+		if (!(temp & CP_HQD_DEQUEUE_REQUEST__IQ_REQ_PEND_MASK))
+			break;
+		pr_debug("Dequeue request is pending\n");
 
+		if (!retry) {
+			pr_err("CP HQD dequeue request time out\n");
+			break;
+		}
+		ndelay(100);
+		--retry;
+	}
+	local_irq_restore(flags);
+	preempt_enable();
+
+	WREG32(mmCP_HQD_DEQUEUE_REQUEST, type);
+
+	end_jiffies = (utimeout * HZ / 1000) + jiffies;
 	while (true) {
 		temp = RREG32(mmCP_HQD_ACTIVE);
-		if (temp & CP_HQD_ACTIVE__ACTIVE_MASK)
+		if (!(temp & CP_HQD_ACTIVE__ACTIVE_MASK))
 			break;
-		if (timeout <= 0) {
-			pr_err("kfd: cp queue preemption time out.\n");
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("cp queue preemption time out\n");
 			release_queue(kgd);
 			return -ETIME;
 		}
-		msleep(20);
-		timeout -= 20;
+		usleep_range(500, 1000);
 	}
 
 	release_queue(kgd);
@@ -556,6 +681,16 @@ static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid)
 	WREG32(mmVM_INVALIDATE_REQUEST, 1 << vmid);
 }
 
+static void set_scratch_backing_va(struct kgd_dev *kgd,
+					uint64_t va, uint32_t vmid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	lock_srbm(kgd, 0, 0, 0, vmid);
+	WREG32(mmSH_HIDDEN_PRIVATE_BASE_VMID, va);
+	unlock_srbm(kgd);
+}
+
 static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
@@ -566,42 +701,42 @@ static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
 	switch (type) {
 	case KGD_ENGINE_PFP:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->gfx.pfp_fw->data;
+						adev->gfx.pfp_fw->data;
 		break;
 
 	case KGD_ENGINE_ME:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->gfx.me_fw->data;
+						adev->gfx.me_fw->data;
 		break;
 
 	case KGD_ENGINE_CE:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->gfx.ce_fw->data;
+						adev->gfx.ce_fw->data;
 		break;
 
 	case KGD_ENGINE_MEC1:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->gfx.mec_fw->data;
+						adev->gfx.mec_fw->data;
 		break;
 
 	case KGD_ENGINE_MEC2:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->gfx.mec2_fw->data;
+						adev->gfx.mec2_fw->data;
 		break;
 
 	case KGD_ENGINE_RLC:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->gfx.rlc_fw->data;
+						adev->gfx.rlc_fw->data;
 		break;
 
 	case KGD_ENGINE_SDMA1:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->sdma.instance[0].fw->data;
+						adev->sdma.instance[0].fw->data;
 		break;
 
 	case KGD_ENGINE_SDMA2:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->sdma.instance[1].fw->data;
+						adev->sdma.instance[1].fw->data;
 		break;
 
 	default:
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
index 133d06671e46..fb6e5dbd5a03 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
@@ -39,6 +39,12 @@
 #include "vi_structs.h"
 #include "vid.h"
 
+enum hqd_dequeue_request_type {
+	NO_ACTION = 0,
+	DRAIN_PIPE,
+	RESET_WAVES
+};
+
 struct cik_sdma_rlc_registers;
 
 /*
@@ -55,12 +61,15 @@ static int kgd_init_pipeline(struct kgd_dev *kgd, uint32_t pipe_id,
 		uint32_t hpd_size, uint64_t hpd_gpu_addr);
 static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id);
 static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
-		uint32_t queue_id, uint32_t __user *wptr);
+			uint32_t queue_id, uint32_t __user *wptr,
+			uint32_t wptr_shift, uint32_t wptr_mask,
+			struct mm_struct *mm);
 static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd);
 static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
 		uint32_t pipe_id, uint32_t queue_id);
 static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd);
-static int kgd_hqd_destroy(struct kgd_dev *kgd, uint32_t reset_type,
+static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
+				enum kfd_preempt_type reset_type,
 				unsigned int utimeout, uint32_t pipe_id,
 				uint32_t queue_id);
 static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
@@ -85,6 +94,33 @@ static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
 		uint8_t vmid);
 static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid);
 static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type);
+static void set_scratch_backing_va(struct kgd_dev *kgd,
+					uint64_t va, uint32_t vmid);
+
+/* Because of REG_GET_FIELD() being used, we put this function in the
+ * asic specific file.
+ */
+static int get_tile_config(struct kgd_dev *kgd,
+		struct tile_config *config)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+
+	config->gb_addr_config = adev->gfx.config.gb_addr_config;
+	config->num_banks = REG_GET_FIELD(adev->gfx.config.mc_arb_ramcfg,
+				MC_ARB_RAMCFG, NOOFBANK);
+	config->num_ranks = REG_GET_FIELD(adev->gfx.config.mc_arb_ramcfg,
+				MC_ARB_RAMCFG, NOOFRANKS);
+
+	config->tile_config_ptr = adev->gfx.config.tile_mode_array;
+	config->num_tile_configs =
+			ARRAY_SIZE(adev->gfx.config.tile_mode_array);
+	config->macro_tile_config_ptr =
+			adev->gfx.config.macrotile_mode_array;
+	config->num_macro_tile_configs =
+			ARRAY_SIZE(adev->gfx.config.macrotile_mode_array);
+
+	return 0;
+}
 
 static const struct kfd2kgd_calls kfd2kgd = {
 	.init_gtt_mem_allocation = alloc_gtt_mem,
@@ -111,12 +147,15 @@ static const struct kfd2kgd_calls kfd2kgd = {
 	.get_atc_vmid_pasid_mapping_valid =
 			get_atc_vmid_pasid_mapping_valid,
 	.write_vmid_invalidate_request = write_vmid_invalidate_request,
-	.get_fw_version = get_fw_version
+	.get_fw_version = get_fw_version,
+	.set_scratch_backing_va = set_scratch_backing_va,
+	.get_tile_config = get_tile_config,
 };
 
 struct kfd2kgd_calls *amdgpu_amdkfd_gfx_8_0_get_functions(void)
 {
 	return (struct kfd2kgd_calls *)&kfd2kgd;
+	return (struct kfd2kgd_calls *)&kfd2kgd;
 }
 
 static inline struct amdgpu_device *get_amdgpu_device(struct kgd_dev *kgd)
@@ -147,7 +186,7 @@ static void acquire_queue(struct kgd_dev *kgd, uint32_t pipe_id,
 {
 	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 
-	uint32_t mec = (++pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	uint32_t mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
 	uint32_t pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
 
 	lock_srbm(kgd, mec, pipe, queue_id, 0);
@@ -216,7 +255,7 @@ static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id)
 	uint32_t mec;
 	uint32_t pipe;
 
-	mec = (++pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+	mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
 	pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
 
 	lock_srbm(kgd, mec, pipe, 0, 0);
@@ -244,20 +283,67 @@ static inline struct cik_sdma_rlc_registers *get_sdma_mqd(void *mqd)
 }
 
 static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
-			uint32_t queue_id, uint32_t __user *wptr)
+			uint32_t queue_id, uint32_t __user *wptr,
+			uint32_t wptr_shift, uint32_t wptr_mask,
+			struct mm_struct *mm)
 {
-	struct vi_mqd *m;
-	uint32_t shadow_wptr, valid_wptr;
 	struct amdgpu_device *adev = get_amdgpu_device(kgd);
+	struct vi_mqd *m;
+	uint32_t *mqd_hqd;
+	uint32_t reg, wptr_val, data;
 
 	m = get_mqd(mqd);
 
-	valid_wptr = copy_from_user(&shadow_wptr, wptr, sizeof(shadow_wptr));
-	if (valid_wptr == 0)
-		m->cp_hqd_pq_wptr = shadow_wptr;
-
 	acquire_queue(kgd, pipe_id, queue_id);
-	gfx_v8_0_mqd_commit(adev, mqd);
+
+	/* HIQ is set during driver init period with vmid set to 0*/
+	if (m->cp_hqd_vmid == 0) {
+		uint32_t value, mec, pipe;
+
+		mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
+		pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
+
+		pr_debug("kfd: set HIQ, mec:%d, pipe:%d, queue:%d.\n",
+			mec, pipe, queue_id);
+		value = RREG32(mmRLC_CP_SCHEDULERS);
+		value = REG_SET_FIELD(value, RLC_CP_SCHEDULERS, scheduler1,
+			((mec << 5) | (pipe << 3) | queue_id | 0x80));
+		WREG32(mmRLC_CP_SCHEDULERS, value);
+	}
+
+	/* HQD registers extend from CP_MQD_BASE_ADDR to CP_HQD_EOP_WPTR_MEM. */
+	mqd_hqd = &m->cp_mqd_base_addr_lo;
+
+	for (reg = mmCP_MQD_BASE_ADDR; reg <= mmCP_HQD_EOP_CONTROL; reg++)
+		WREG32(reg, mqd_hqd[reg - mmCP_MQD_BASE_ADDR]);
+
+	/* Tonga errata: EOP RPTR/WPTR should be left unmodified.
+	 * This is safe since EOP RPTR==WPTR for any inactive HQD
+	 * on ASICs that do not support context-save.
+	 * EOP writes/reads can start anywhere in the ring.
+	 */
+	if (get_amdgpu_device(kgd)->asic_type != CHIP_TONGA) {
+		WREG32(mmCP_HQD_EOP_RPTR, m->cp_hqd_eop_rptr);
+		WREG32(mmCP_HQD_EOP_WPTR, m->cp_hqd_eop_wptr);
+		WREG32(mmCP_HQD_EOP_WPTR_MEM, m->cp_hqd_eop_wptr_mem);
+	}
+
+	for (reg = mmCP_HQD_EOP_EVENTS; reg <= mmCP_HQD_ERROR; reg++)
+		WREG32(reg, mqd_hqd[reg - mmCP_MQD_BASE_ADDR]);
+
+	/* Copy userspace write pointer value to register.
+	 * Activate doorbell logic to monitor subsequent changes.
+	 */
+	data = REG_SET_FIELD(m->cp_hqd_pq_doorbell_control,
+			     CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 1);
+	WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL, data);
+
+	if (read_user_wptr(mm, wptr, wptr_val))
+		WREG32(mmCP_HQD_PQ_WPTR, (wptr_val << wptr_shift) & wptr_mask);
+
+	data = REG_SET_FIELD(m->cp_hqd_active, CP_HQD_ACTIVE, ACTIVE, 1);
+	WREG32(mmCP_HQD_ACTIVE, data);
+
 	release_queue(kgd);
 
 	return 0;
@@ -308,29 +394,102 @@ static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd)
 	return false;
 }
 
-static int kgd_hqd_destroy(struct kgd_dev *kgd, uint32_t reset_type,
+static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
+				enum kfd_preempt_type reset_type,
 				unsigned int utimeout, uint32_t pipe_id,
 				uint32_t queue_id)
 {
 	struct amdgpu_device *adev = get_amdgpu_device(kgd);
 	uint32_t temp;
-	int timeout = utimeout;
+	enum hqd_dequeue_request_type type;
+	unsigned long flags, end_jiffies;
+	int retry;
+	struct vi_mqd *m = get_mqd(mqd);
 
 	acquire_queue(kgd, pipe_id, queue_id);
 
-	WREG32(mmCP_HQD_DEQUEUE_REQUEST, reset_type);
+	if (m->cp_hqd_vmid == 0)
+		WREG32_FIELD(RLC_CP_SCHEDULERS, scheduler1, 0);
+
+	switch (reset_type) {
+	case KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN:
+		type = DRAIN_PIPE;
+		break;
+	case KFD_PREEMPT_TYPE_WAVEFRONT_RESET:
+		type = RESET_WAVES;
+		break;
+	default:
+		type = DRAIN_PIPE;
+		break;
+	}
 
+	/* Workaround: If IQ timer is active and the wait time is close to or
+	 * equal to 0, dequeueing is not safe. Wait until either the wait time
+	 * is larger or timer is cleared. Also, ensure that IQ_REQ_PEND is
+	 * cleared before continuing. Also, ensure wait times are set to at
+	 * least 0x3.
+	 */
+	local_irq_save(flags);
+	preempt_disable();
+	retry = 5000; /* wait for 500 usecs at maximum */
+	while (true) {
+		temp = RREG32(mmCP_HQD_IQ_TIMER);
+		if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, PROCESSING_IQ)) {
+			pr_debug("HW is processing IQ\n");
+			goto loop;
+		}
+		if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, ACTIVE)) {
+			if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, RETRY_TYPE)
+					== 3) /* SEM-rearm is safe */
+				break;
+			/* Wait time 3 is safe for CP, but our MMIO read/write
+			 * time is close to 1 microsecond, so check for 10 to
+			 * leave more buffer room
+			 */
+			if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, WAIT_TIME)
+					>= 10)
+				break;
+			pr_debug("IQ timer is active\n");
+		} else
+			break;
+loop:
+		if (!retry) {
+			pr_err("CP HQD IQ timer status time out\n");
+			break;
+		}
+		ndelay(100);
+		--retry;
+	}
+	retry = 1000;
+	while (true) {
+		temp = RREG32(mmCP_HQD_DEQUEUE_REQUEST);
+		if (!(temp & CP_HQD_DEQUEUE_REQUEST__IQ_REQ_PEND_MASK))
+			break;
+		pr_debug("Dequeue request is pending\n");
+
+		if (!retry) {
+			pr_err("CP HQD dequeue request time out\n");
+			break;
+		}
+		ndelay(100);
+		--retry;
+	}
+	local_irq_restore(flags);
+	preempt_enable();
+
+	WREG32(mmCP_HQD_DEQUEUE_REQUEST, type);
+
+	end_jiffies = (utimeout * HZ / 1000) + jiffies;
 	while (true) {
 		temp = RREG32(mmCP_HQD_ACTIVE);
-		if (temp & CP_HQD_ACTIVE__ACTIVE_MASK)
+		if (!(temp & CP_HQD_ACTIVE__ACTIVE_MASK))
 			break;
-		if (timeout <= 0) {
-			pr_err("kfd: cp queue preemption time out.\n");
+		if (time_after(jiffies, end_jiffies)) {
+			pr_err("cp queue preemption time out.\n");
 			release_queue(kgd);
 			return -ETIME;
 		}
-		msleep(20);
-		timeout -= 20;
+		usleep_range(500, 1000);
 	}
 
 	release_queue(kgd);
@@ -444,6 +603,16 @@ static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
 	return 0;
 }
 
+static void set_scratch_backing_va(struct kgd_dev *kgd,
+					uint64_t va, uint32_t vmid)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
+
+	lock_srbm(kgd, 0, 0, 0, vmid);
+	WREG32(mmSH_HIDDEN_PRIVATE_BASE_VMID, va);
+	unlock_srbm(kgd);
+}
+
 static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
@@ -454,42 +623,42 @@ static uint16_t get_fw_version(struct kgd_dev *kgd, enum kgd_engine_type type)
 	switch (type) {
 	case KGD_ENGINE_PFP:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->gfx.pfp_fw->data;
+						adev->gfx.pfp_fw->data;
 		break;
 
 	case KGD_ENGINE_ME:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->gfx.me_fw->data;
+						adev->gfx.me_fw->data;
 		break;
 
 	case KGD_ENGINE_CE:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->gfx.ce_fw->data;
+						adev->gfx.ce_fw->data;
 		break;
 
 	case KGD_ENGINE_MEC1:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->gfx.mec_fw->data;
+						adev->gfx.mec_fw->data;
 		break;
 
 	case KGD_ENGINE_MEC2:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->gfx.mec2_fw->data;
+						adev->gfx.mec2_fw->data;
 		break;
 
 	case KGD_ENGINE_RLC:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->gfx.rlc_fw->data;
+						adev->gfx.rlc_fw->data;
 		break;
 
 	case KGD_ENGINE_SDMA1:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->sdma.instance[0].fw->data;
+						adev->sdma.instance[0].fw->data;
 		break;
 
 	case KGD_ENGINE_SDMA2:
 		hdr = (const union amdgpu_firmware_header *)
-							adev->sdma.instance[1].fw->data;
+						adev->sdma.instance[1].fw->data;
 		break;
 
 	default:
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c
index 2fb299afc12b..63ec1e1bb6aa 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c
@@ -81,7 +81,7 @@ static void amdgpu_benchmark_move(struct amdgpu_device *adev, unsigned size,
 
 	n = AMDGPU_BENCHMARK_ITERATIONS;
 	r = amdgpu_bo_create(adev, size, PAGE_SIZE, true, sdomain, 0, NULL,
-			     NULL, &sobj);
+			     NULL, 0, &sobj);
 	if (r) {
 		goto out_cleanup;
 	}
@@ -94,7 +94,7 @@ static void amdgpu_benchmark_move(struct amdgpu_device *adev, unsigned size,
 		goto out_cleanup;
 	}
 	r = amdgpu_bo_create(adev, size, PAGE_SIZE, true, ddomain, 0, NULL,
-			     NULL, &dobj);
+			     NULL, 0, &dobj);
 	if (r) {
 		goto out_cleanup;
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
index d324e1c24028..59089e027f4d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
@@ -136,7 +136,7 @@ static int amdgpu_bo_list_set(struct amdgpu_device *adev,
 		}
 
 		bo = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
-		drm_gem_object_unreference_unlocked(gobj);
+		drm_gem_object_put_unlocked(gobj);
 
 		usermm = amdgpu_ttm_tt_get_usermm(bo->tbo.ttm);
 		if (usermm) {
@@ -156,11 +156,11 @@ static int amdgpu_bo_list_set(struct amdgpu_device *adev,
 		entry->tv.bo = &entry->robj->tbo;
 		entry->tv.shared = !entry->robj->prime_shared_count;
 
-		if (entry->robj->prefered_domains == AMDGPU_GEM_DOMAIN_GDS)
+		if (entry->robj->preferred_domains == AMDGPU_GEM_DOMAIN_GDS)
 			gds_obj = entry->robj;
-		if (entry->robj->prefered_domains == AMDGPU_GEM_DOMAIN_GWS)
+		if (entry->robj->preferred_domains == AMDGPU_GEM_DOMAIN_GWS)
 			gws_obj = entry->robj;
-		if (entry->robj->prefered_domains == AMDGPU_GEM_DOMAIN_OA)
+		if (entry->robj->preferred_domains == AMDGPU_GEM_DOMAIN_OA)
 			oa_obj = entry->robj;
 
 		total_size += amdgpu_bo_size(entry->robj);
@@ -270,7 +270,7 @@ int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
 	struct amdgpu_fpriv *fpriv = filp->driver_priv;
 	union drm_amdgpu_bo_list *args = data;
 	uint32_t handle = args->in.list_handle;
-	const void __user *uptr = (const void*)(uintptr_t)args->in.bo_info_ptr;
+	const void __user *uptr = u64_to_user_ptr(args->in.bo_info_ptr);
 
 	struct drm_amdgpu_bo_list_entry *info;
 	struct amdgpu_bo_list *list;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
index a99e0bca6812..fd435a96481c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
@@ -124,7 +124,7 @@ static int amdgpu_cgs_alloc_gpu_mem(struct cgs_device *cgs_device,
 	ret = amdgpu_bo_create_restricted(adev, size, PAGE_SIZE,
 					  true, domain, flags,
 					  NULL, &placement, NULL,
-					  &obj);
+					  0, &obj);
 	if (ret) {
 		DRM_ERROR("(%d) bo create failed\n", ret);
 		return ret;
@@ -166,7 +166,7 @@ static int amdgpu_cgs_gmap_gpu_mem(struct cgs_device *cgs_device, cgs_handle_t h
 	r = amdgpu_bo_reserve(obj, true);
 	if (unlikely(r != 0))
 		return r;
-	r = amdgpu_bo_pin_restricted(obj, obj->prefered_domains,
+	r = amdgpu_bo_pin_restricted(obj, obj->preferred_domains,
 				     min_offset, max_offset, mcaddr);
 	amdgpu_bo_unreserve(obj);
 	return r;
@@ -659,7 +659,7 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
 		info->version = (uint16_t)le32_to_cpu(header->header.ucode_version);
 
 		if (CGS_UCODE_ID_CP_MEC == type)
-			info->image_size = (header->jt_offset) << 2;
+			info->image_size = le32_to_cpu(header->jt_offset) << 2;
 
 		info->fw_version = amdgpu_get_firmware_version(cgs_device, type);
 		info->feature_version = (uint16_t)le32_to_cpu(header->ucode_feature_version);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 33789510e663..269b835571eb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -54,7 +54,7 @@ static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
 
 	*offset = data->offset;
 
-	drm_gem_object_unreference_unlocked(gobj);
+	drm_gem_object_put_unlocked(gobj);
 
 	if (amdgpu_ttm_tt_get_usermm(p->uf_entry.robj->tbo.ttm)) {
 		amdgpu_bo_unref(&p->uf_entry.robj);
@@ -90,7 +90,7 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
 	}
 
 	/* get chunks */
-	chunk_array_user = (uint64_t __user *)(uintptr_t)(cs->in.chunks);
+	chunk_array_user = u64_to_user_ptr(cs->in.chunks);
 	if (copy_from_user(chunk_array, chunk_array_user,
 			   sizeof(uint64_t)*cs->in.num_chunks)) {
 		ret = -EFAULT;
@@ -110,7 +110,7 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
 		struct drm_amdgpu_cs_chunk user_chunk;
 		uint32_t __user *cdata;
 
-		chunk_ptr = (void __user *)(uintptr_t)chunk_array[i];
+		chunk_ptr = u64_to_user_ptr(chunk_array[i]);
 		if (copy_from_user(&user_chunk, chunk_ptr,
 				       sizeof(struct drm_amdgpu_cs_chunk))) {
 			ret = -EFAULT;
@@ -121,7 +121,7 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
 		p->chunks[i].length_dw = user_chunk.length_dw;
 
 		size = p->chunks[i].length_dw;
-		cdata = (void __user *)(uintptr_t)user_chunk.chunk_data;
+		cdata = u64_to_user_ptr(user_chunk.chunk_data);
 
 		p->chunks[i].kdata = kvmalloc_array(size, sizeof(uint32_t), GFP_KERNEL);
 		if (p->chunks[i].kdata == NULL) {
@@ -246,7 +246,7 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
 	}
 
 	total_vram = adev->mc.real_vram_size - adev->vram_pin_size;
-	used_vram = atomic64_read(&adev->vram_usage);
+	used_vram = amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
 	free_vram = used_vram >= total_vram ? 0 : total_vram - used_vram;
 
 	spin_lock(&adev->mm_stats.lock);
@@ -292,7 +292,8 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
 	/* Do the same for visible VRAM if half of it is free */
 	if (adev->mc.visible_vram_size < adev->mc.real_vram_size) {
 		u64 total_vis_vram = adev->mc.visible_vram_size;
-		u64 used_vis_vram = atomic64_read(&adev->vram_vis_usage);
+		u64 used_vis_vram =
+			amdgpu_vram_mgr_vis_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
 
 		if (used_vis_vram < total_vis_vram) {
 			u64 free_vis_vram = total_vis_vram - used_vis_vram;
@@ -348,11 +349,11 @@ static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p,
 			 * that.
 			 */
 			if (p->bytes_moved_vis < p->bytes_moved_vis_threshold)
-				domain = bo->prefered_domains;
+				domain = bo->preferred_domains;
 			else
 				domain = bo->allowed_domains;
 		} else {
-			domain = bo->prefered_domains;
+			domain = bo->preferred_domains;
 		}
 	} else {
 		domain = bo->allowed_domains;
@@ -673,10 +674,8 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 	}
 
 error_validate:
-	if (r) {
-		amdgpu_vm_move_pt_bos_in_lru(p->adev, &fpriv->vm);
+	if (r)
 		ttm_eu_backoff_reservation(&p->ticket, &p->validated);
-	}
 
 error_free_pages:
 
@@ -724,21 +723,18 @@ static int amdgpu_cs_sync_rings(struct amdgpu_cs_parser *p)
  * If error is set than unvalidate buffer, otherwise just free memory
  * used by parsing context.
  **/
-static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error, bool backoff)
+static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error,
+				  bool backoff)
 {
-	struct amdgpu_fpriv *fpriv = parser->filp->driver_priv;
 	unsigned i;
 
-	if (!error) {
-		amdgpu_vm_move_pt_bos_in_lru(parser->adev, &fpriv->vm);
-
+	if (!error)
 		ttm_eu_fence_buffer_objects(&parser->ticket,
 					    &parser->validated,
 					    parser->fence);
-	} else if (backoff) {
+	else if (backoff)
 		ttm_eu_backoff_reservation(&parser->ticket,
 					   &parser->validated);
-	}
 
 	for (i = 0; i < parser->num_post_dep_syncobjs; i++)
 		drm_syncobj_put(parser->post_dep_syncobjs[i]);
@@ -791,7 +787,8 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p)
 
 	if (amdgpu_sriov_vf(adev)) {
 		struct dma_fence *f;
-		bo_va = vm->csa_bo_va;
+
+		bo_va = fpriv->csa_va;
 		BUG_ON(!bo_va);
 		r = amdgpu_vm_bo_update(adev, bo_va, false);
 		if (r)
@@ -828,7 +825,7 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p)
 
 	}
 
-	r = amdgpu_vm_clear_invalids(adev, vm, &p->job->sync);
+	r = amdgpu_vm_clear_moved(adev, vm, &p->job->sync);
 
 	if (amdgpu_vm_debug && p->bo_list) {
 		/* Invalidate all BOs to test for userspace bugs */
@@ -1038,7 +1035,7 @@ static int amdgpu_syncobj_lookup_and_add_to_sync(struct amdgpu_cs_parser *p,
 {
 	int r;
 	struct dma_fence *fence;
-	r = drm_syncobj_fence_get(p->filp, handle, &fence);
+	r = drm_syncobj_find_fence(p->filp, handle, &fence);
 	if (r)
 		return r;
 
@@ -1437,7 +1434,7 @@ int amdgpu_cs_wait_fences_ioctl(struct drm_device *dev, void *data,
 	if (fences == NULL)
 		return -ENOMEM;
 
-	fences_user = (void __user *)(uintptr_t)(wait->in.fences);
+	fences_user = u64_to_user_ptr(wait->in.fences);
 	if (copy_from_user(fences, fences_user,
 		sizeof(struct drm_amdgpu_fence) * fence_count)) {
 		r = -EFAULT;
@@ -1490,7 +1487,7 @@ amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
 			    addr > mapping->last)
 				continue;
 
-			*bo = lobj->bo_va->bo;
+			*bo = lobj->bo_va->base.bo;
 			return mapping;
 		}
 
@@ -1499,7 +1496,7 @@ amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
 			    addr > mapping->last)
 				continue;
 
-			*bo = lobj->bo_va->bo;
+			*bo = lobj->bo_va->base.bo;
 			return mapping;
 		}
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 6279956e92a4..1a459ac63df4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -336,51 +336,16 @@ static void amdgpu_block_invalid_wreg(struct amdgpu_device *adev,
 
 static int amdgpu_vram_scratch_init(struct amdgpu_device *adev)
 {
-	int r;
-
-	if (adev->vram_scratch.robj == NULL) {
-		r = amdgpu_bo_create(adev, AMDGPU_GPU_PAGE_SIZE,
-				     PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_VRAM,
-				     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-				     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-				     NULL, NULL, &adev->vram_scratch.robj);
-		if (r) {
-			return r;
-		}
-	}
-
-	r = amdgpu_bo_reserve(adev->vram_scratch.robj, false);
-	if (unlikely(r != 0))
-		return r;
-	r = amdgpu_bo_pin(adev->vram_scratch.robj,
-			  AMDGPU_GEM_DOMAIN_VRAM, &adev->vram_scratch.gpu_addr);
-	if (r) {
-		amdgpu_bo_unreserve(adev->vram_scratch.robj);
-		return r;
-	}
-	r = amdgpu_bo_kmap(adev->vram_scratch.robj,
-				(void **)&adev->vram_scratch.ptr);
-	if (r)
-		amdgpu_bo_unpin(adev->vram_scratch.robj);
-	amdgpu_bo_unreserve(adev->vram_scratch.robj);
-
-	return r;
+	return amdgpu_bo_create_kernel(adev, AMDGPU_GPU_PAGE_SIZE,
+				       PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
+				       &adev->vram_scratch.robj,
+				       &adev->vram_scratch.gpu_addr,
+				       (void **)&adev->vram_scratch.ptr);
 }
 
 static void amdgpu_vram_scratch_fini(struct amdgpu_device *adev)
 {
-	int r;
-
-	if (adev->vram_scratch.robj == NULL) {
-		return;
-	}
-	r = amdgpu_bo_reserve(adev->vram_scratch.robj, true);
-	if (likely(r == 0)) {
-		amdgpu_bo_kunmap(adev->vram_scratch.robj);
-		amdgpu_bo_unpin(adev->vram_scratch.robj);
-		amdgpu_bo_unreserve(adev->vram_scratch.robj);
-	}
-	amdgpu_bo_unref(&adev->vram_scratch.robj);
+	amdgpu_bo_free_kernel(&adev->vram_scratch.robj, NULL, NULL);
 }
 
 /**
@@ -539,7 +504,8 @@ static int amdgpu_wb_init(struct amdgpu_device *adev)
 	int r;
 
 	if (adev->wb.wb_obj == NULL) {
-		r = amdgpu_bo_create_kernel(adev, AMDGPU_MAX_WB * sizeof(uint32_t),
+		/* AMDGPU_MAX_WB * sizeof(uint32_t) * 8 = AMDGPU_MAX_WB 256bit slots */
+		r = amdgpu_bo_create_kernel(adev, AMDGPU_MAX_WB * sizeof(uint32_t) * 8,
 					    PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
 					    &adev->wb.wb_obj, &adev->wb.gpu_addr,
 					    (void **)&adev->wb.wb);
@@ -570,47 +536,10 @@ static int amdgpu_wb_init(struct amdgpu_device *adev)
 int amdgpu_wb_get(struct amdgpu_device *adev, u32 *wb)
 {
 	unsigned long offset = find_first_zero_bit(adev->wb.used, adev->wb.num_wb);
-	if (offset < adev->wb.num_wb) {
-		__set_bit(offset, adev->wb.used);
-		*wb = offset;
-		return 0;
-	} else {
-		return -EINVAL;
-	}
-}
 
-/**
- * amdgpu_wb_get_64bit - Allocate a wb entry
- *
- * @adev: amdgpu_device pointer
- * @wb: wb index
- *
- * Allocate a wb slot for use by the driver (all asics).
- * Returns 0 on success or -EINVAL on failure.
- */
-int amdgpu_wb_get_64bit(struct amdgpu_device *adev, u32 *wb)
-{
-	unsigned long offset = bitmap_find_next_zero_area_off(adev->wb.used,
-				adev->wb.num_wb, 0, 2, 7, 0);
-	if ((offset + 1) < adev->wb.num_wb) {
+	if (offset < adev->wb.num_wb) {
 		__set_bit(offset, adev->wb.used);
-		__set_bit(offset + 1, adev->wb.used);
-		*wb = offset;
-		return 0;
-	} else {
-		return -EINVAL;
-	}
-}
-
-int amdgpu_wb_get_256Bit(struct amdgpu_device *adev, u32 *wb)
-{
-	int i = 0;
-	unsigned long offset = bitmap_find_next_zero_area_off(adev->wb.used,
-				adev->wb.num_wb, 0, 8, 63, 0);
-	if ((offset + 7) < adev->wb.num_wb) {
-		for (i = 0; i < 8; i++)
-			__set_bit(offset + i, adev->wb.used);
-		*wb = offset;
+		*wb = offset * 8; /* convert to dw offset */
 		return 0;
 	} else {
 		return -EINVAL;
@@ -632,39 +561,6 @@ void amdgpu_wb_free(struct amdgpu_device *adev, u32 wb)
 }
 
 /**
- * amdgpu_wb_free_64bit - Free a wb entry
- *
- * @adev: amdgpu_device pointer
- * @wb: wb index
- *
- * Free a wb slot allocated for use by the driver (all asics)
- */
-void amdgpu_wb_free_64bit(struct amdgpu_device *adev, u32 wb)
-{
-	if ((wb + 1) < adev->wb.num_wb) {
-		__clear_bit(wb, adev->wb.used);
-		__clear_bit(wb + 1, adev->wb.used);
-	}
-}
-
-/**
- * amdgpu_wb_free_256bit - Free a wb entry
- *
- * @adev: amdgpu_device pointer
- * @wb: wb index
- *
- * Free a wb slot allocated for use by the driver (all asics)
- */
-void amdgpu_wb_free_256bit(struct amdgpu_device *adev, u32 wb)
-{
-	int i = 0;
-
-	if ((wb + 7) < adev->wb.num_wb)
-		for (i = 0; i < 8; i++)
-			__clear_bit(wb + i, adev->wb.used);
-}
-
-/**
  * amdgpu_vram_location - try to find VRAM location
  * @adev: amdgpu device structure holding all necessary informations
  * @mc: memory controller structure holding memory informations
@@ -1180,6 +1076,13 @@ static void amdgpu_check_arguments(struct amdgpu_device *adev)
 		amdgpu_gtt_size = -1;
 	}
 
+	/* valid range is between 4 and 9 inclusive */
+	if (amdgpu_vm_fragment_size != -1 &&
+	    (amdgpu_vm_fragment_size > 9 || amdgpu_vm_fragment_size < 4)) {
+		dev_warn(adev->dev, "valid range is between 4 and 9\n");
+		amdgpu_vm_fragment_size = -1;
+	}
+
 	amdgpu_check_vm_size(adev);
 
 	amdgpu_check_block_size(adev);
@@ -1948,7 +1851,8 @@ static int amdgpu_sriov_reinit_late(struct amdgpu_device *adev)
 		AMD_IP_BLOCK_TYPE_DCE,
 		AMD_IP_BLOCK_TYPE_GFX,
 		AMD_IP_BLOCK_TYPE_SDMA,
-		AMD_IP_BLOCK_TYPE_VCE,
+		AMD_IP_BLOCK_TYPE_UVD,
+		AMD_IP_BLOCK_TYPE_VCE
 	};
 
 	for (i = 0; i < ARRAY_SIZE(ip_order); i++) {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
index cdf2ab20166a..6ad243293a78 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
@@ -482,7 +482,7 @@ static void amdgpu_user_framebuffer_destroy(struct drm_framebuffer *fb)
 {
 	struct amdgpu_framebuffer *amdgpu_fb = to_amdgpu_framebuffer(fb);
 
-	drm_gem_object_unreference_unlocked(amdgpu_fb->obj);
+	drm_gem_object_put_unlocked(amdgpu_fb->obj);
 	drm_framebuffer_cleanup(fb);
 	kfree(amdgpu_fb);
 }
@@ -542,14 +542,14 @@ amdgpu_user_framebuffer_create(struct drm_device *dev,
 
 	amdgpu_fb = kzalloc(sizeof(*amdgpu_fb), GFP_KERNEL);
 	if (amdgpu_fb == NULL) {
-		drm_gem_object_unreference_unlocked(obj);
+		drm_gem_object_put_unlocked(obj);
 		return ERR_PTR(-ENOMEM);
 	}
 
 	ret = amdgpu_framebuffer_init(dev, amdgpu_fb, mode_cmd, obj);
 	if (ret) {
 		kfree(amdgpu_fb);
-		drm_gem_object_unreference_unlocked(obj);
+		drm_gem_object_put_unlocked(obj);
 		return ERR_PTR(ret);
 	}
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index 5e9ce8a29669..e39ec981b11c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -68,9 +68,10 @@
  * - 3.16.0 - Add reserved vmid support
  * - 3.17.0 - Add AMDGPU_NUM_VRAM_CPU_PAGE_FAULTS.
  * - 3.18.0 - Export gpu always on cu bitmap
+ * - 3.19.0 - Add support for UVD MJPEG decode
  */
 #define KMS_DRIVER_MAJOR	3
-#define KMS_DRIVER_MINOR	18
+#define KMS_DRIVER_MINOR	19
 #define KMS_DRIVER_PATCHLEVEL	0
 
 int amdgpu_vram_limit = 0;
@@ -94,6 +95,7 @@ unsigned amdgpu_ip_block_mask = 0xffffffff;
 int amdgpu_bapm = -1;
 int amdgpu_deep_color = 0;
 int amdgpu_vm_size = -1;
+int amdgpu_vm_fragment_size = -1;
 int amdgpu_vm_block_size = -1;
 int amdgpu_vm_fault_stop = 0;
 int amdgpu_vm_debug = 0;
@@ -183,6 +185,9 @@ module_param_named(deep_color, amdgpu_deep_color, int, 0444);
 MODULE_PARM_DESC(vm_size, "VM address space size in gigabytes (default 64GB)");
 module_param_named(vm_size, amdgpu_vm_size, int, 0444);
 
+MODULE_PARM_DESC(vm_fragment_size, "VM fragment size in bits (4, 5, etc. 4 = 64K (default), Max 9 = 2M)");
+module_param_named(vm_fragment_size, amdgpu_vm_fragment_size, int, 0444);
+
 MODULE_PARM_DESC(vm_block_size, "VM page table size in bits (default depending on vm_size)");
 module_param_named(vm_block_size, amdgpu_vm_block_size, int, 0444);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c
index 0a8ee2411180..9afa9c097e1f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c
@@ -118,7 +118,7 @@ static void amdgpufb_destroy_pinned_object(struct drm_gem_object *gobj)
 		amdgpu_bo_unpin(abo);
 		amdgpu_bo_unreserve(abo);
 	}
-	drm_gem_object_unreference_unlocked(gobj);
+	drm_gem_object_put_unlocked(gobj);
 }
 
 static int amdgpufb_create_pinned_object(struct amdgpu_fbdev *rfbdev,
@@ -250,7 +250,7 @@ static int amdgpufb_create(struct drm_fb_helper *helper,
 	tmp = amdgpu_bo_gpu_offset(abo) - adev->mc.vram_start;
 	info->fix.smem_start = adev->mc.aper_base + tmp;
 	info->fix.smem_len = amdgpu_bo_size(abo);
-	info->screen_base = abo->kptr;
+	info->screen_base = amdgpu_bo_kptr(abo);
 	info->screen_size = amdgpu_bo_size(abo);
 
 	drm_fb_helper_fill_var(info, &rfbdev->helper, sizes->fb_width, sizes->fb_height);
@@ -280,7 +280,7 @@ out:
 
 	}
 	if (fb && ret) {
-		drm_gem_object_unreference_unlocked(gobj);
+		drm_gem_object_put_unlocked(gobj);
 		drm_framebuffer_unregister_private(fb);
 		drm_framebuffer_cleanup(fb);
 		kfree(fb);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
index 5cc4987cd887..94c1e2e8e34c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
@@ -144,7 +144,7 @@ int amdgpu_gart_table_vram_alloc(struct amdgpu_device *adev)
 				     PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_VRAM,
 				     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
 				     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-				     NULL, NULL, &adev->gart.robj);
+				     NULL, NULL, 0, &adev->gart.robj);
 		if (r) {
 			return r;
 		}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index 917ac5e074a0..7171968f261e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -59,7 +59,7 @@ int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
 
 retry:
 	r = amdgpu_bo_create(adev, size, alignment, kernel, initial_domain,
-			     flags, NULL, NULL, &robj);
+			     flags, NULL, NULL, 0, &robj);
 	if (r) {
 		if (r != -ERESTARTSYS) {
 			if (initial_domain == AMDGPU_GEM_DOMAIN_VRAM) {
@@ -91,7 +91,7 @@ void amdgpu_gem_force_release(struct amdgpu_device *adev)
 		spin_lock(&file->table_lock);
 		idr_for_each_entry(&file->object_idr, gobj, handle) {
 			WARN_ONCE(1, "And also active allocations!\n");
-			drm_gem_object_unreference_unlocked(gobj);
+			drm_gem_object_put_unlocked(gobj);
 		}
 		idr_destroy(&file->object_idr);
 		spin_unlock(&file->table_lock);
@@ -225,9 +225,7 @@ int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
 	if (args->in.domain_flags & ~(AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
 				      AMDGPU_GEM_CREATE_NO_CPU_ACCESS |
 				      AMDGPU_GEM_CREATE_CPU_GTT_USWC |
-				      AMDGPU_GEM_CREATE_VRAM_CLEARED|
-				      AMDGPU_GEM_CREATE_SHADOW |
-				      AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS))
+				      AMDGPU_GEM_CREATE_VRAM_CLEARED))
 		return -EINVAL;
 
 	/* reject invalid gem domains */
@@ -263,7 +261,7 @@ int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
 
 	r = drm_gem_handle_create(filp, gobj, &handle);
 	/* drop reference from allocate - handle holds it now */
-	drm_gem_object_unreference_unlocked(gobj);
+	drm_gem_object_put_unlocked(gobj);
 	if (r)
 		return r;
 
@@ -306,7 +304,7 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
 		return r;
 
 	bo = gem_to_amdgpu_bo(gobj);
-	bo->prefered_domains = AMDGPU_GEM_DOMAIN_GTT;
+	bo->preferred_domains = AMDGPU_GEM_DOMAIN_GTT;
 	bo->allowed_domains = AMDGPU_GEM_DOMAIN_GTT;
 	r = amdgpu_ttm_tt_set_userptr(bo->tbo.ttm, args->addr, args->flags);
 	if (r)
@@ -341,7 +339,7 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
 
 	r = drm_gem_handle_create(filp, gobj, &handle);
 	/* drop reference from allocate - handle holds it now */
-	drm_gem_object_unreference_unlocked(gobj);
+	drm_gem_object_put_unlocked(gobj);
 	if (r)
 		return r;
 
@@ -355,7 +353,7 @@ unlock_mmap_sem:
 	up_read(&current->mm->mmap_sem);
 
 release_object:
-	drm_gem_object_unreference_unlocked(gobj);
+	drm_gem_object_put_unlocked(gobj);
 
 	return r;
 }
@@ -374,11 +372,11 @@ int amdgpu_mode_dumb_mmap(struct drm_file *filp,
 	robj = gem_to_amdgpu_bo(gobj);
 	if (amdgpu_ttm_tt_get_usermm(robj->tbo.ttm) ||
 	    (robj->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)) {
-		drm_gem_object_unreference_unlocked(gobj);
+		drm_gem_object_put_unlocked(gobj);
 		return -EPERM;
 	}
 	*offset_p = amdgpu_bo_mmap_offset(robj);
-	drm_gem_object_unreference_unlocked(gobj);
+	drm_gem_object_put_unlocked(gobj);
 	return 0;
 }
 
@@ -448,7 +446,7 @@ int amdgpu_gem_wait_idle_ioctl(struct drm_device *dev, void *data,
 	} else
 		r = ret;
 
-	drm_gem_object_unreference_unlocked(gobj);
+	drm_gem_object_put_unlocked(gobj);
 	return r;
 }
 
@@ -491,7 +489,7 @@ int amdgpu_gem_metadata_ioctl(struct drm_device *dev, void *data,
 unreserve:
 	amdgpu_bo_unreserve(robj);
 out:
-	drm_gem_object_unreference_unlocked(gobj);
+	drm_gem_object_put_unlocked(gobj);
 	return r;
 }
 
@@ -623,7 +621,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 
 	switch (args->operation) {
 	case AMDGPU_VA_OP_MAP:
-		r = amdgpu_vm_alloc_pts(adev, bo_va->vm, args->va_address,
+		r = amdgpu_vm_alloc_pts(adev, bo_va->base.vm, args->va_address,
 					args->map_size);
 		if (r)
 			goto error_backoff;
@@ -643,7 +641,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
 						args->map_size);
 		break;
 	case AMDGPU_VA_OP_REPLACE:
-		r = amdgpu_vm_alloc_pts(adev, bo_va->vm, args->va_address,
+		r = amdgpu_vm_alloc_pts(adev, bo_va->base.vm, args->va_address,
 					args->map_size);
 		if (r)
 			goto error_backoff;
@@ -664,7 +662,7 @@ error_backoff:
 	ttm_eu_backoff_reservation(&ticket, &list);
 
 error_unref:
-	drm_gem_object_unreference_unlocked(gobj);
+	drm_gem_object_put_unlocked(gobj);
 	return r;
 }
 
@@ -689,11 +687,11 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
 	switch (args->op) {
 	case AMDGPU_GEM_OP_GET_GEM_CREATE_INFO: {
 		struct drm_amdgpu_gem_create_in info;
-		void __user *out = (void __user *)(uintptr_t)args->value;
+		void __user *out = u64_to_user_ptr(args->value);
 
 		info.bo_size = robj->gem_base.size;
 		info.alignment = robj->tbo.mem.page_alignment << PAGE_SHIFT;
-		info.domains = robj->prefered_domains;
+		info.domains = robj->preferred_domains;
 		info.domain_flags = robj->flags;
 		amdgpu_bo_unreserve(robj);
 		if (copy_to_user(out, &info, sizeof(info)))
@@ -711,10 +709,10 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
 			amdgpu_bo_unreserve(robj);
 			break;
 		}
-		robj->prefered_domains = args->value & (AMDGPU_GEM_DOMAIN_VRAM |
+		robj->preferred_domains = args->value & (AMDGPU_GEM_DOMAIN_VRAM |
 							AMDGPU_GEM_DOMAIN_GTT |
 							AMDGPU_GEM_DOMAIN_CPU);
-		robj->allowed_domains = robj->prefered_domains;
+		robj->allowed_domains = robj->preferred_domains;
 		if (robj->allowed_domains == AMDGPU_GEM_DOMAIN_VRAM)
 			robj->allowed_domains |= AMDGPU_GEM_DOMAIN_GTT;
 
@@ -726,7 +724,7 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
 	}
 
 out:
-	drm_gem_object_unreference_unlocked(gobj);
+	drm_gem_object_put_unlocked(gobj);
 	return r;
 }
 
@@ -754,7 +752,7 @@ int amdgpu_mode_dumb_create(struct drm_file *file_priv,
 
 	r = drm_gem_handle_create(file_priv, gobj, &handle);
 	/* drop reference from allocate - handle holds it now */
-	drm_gem_object_unreference_unlocked(gobj);
+	drm_gem_object_put_unlocked(gobj);
 	if (r) {
 		return r;
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c
index 5e6b90c6794f..9e05e257729f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c
@@ -28,7 +28,7 @@
 struct amdgpu_gtt_mgr {
 	struct drm_mm mm;
 	spinlock_t lock;
-	uint64_t available;
+	atomic64_t available;
 };
 
 /**
@@ -54,7 +54,7 @@ static int amdgpu_gtt_mgr_init(struct ttm_mem_type_manager *man,
 	size = (adev->mc.gart_size >> PAGE_SHIFT) - start;
 	drm_mm_init(&mgr->mm, start, size);
 	spin_lock_init(&mgr->lock);
-	mgr->available = p_size;
+	atomic64_set(&mgr->available, p_size);
 	man->priv = mgr;
 	return 0;
 }
@@ -153,15 +153,6 @@ int amdgpu_gtt_mgr_alloc(struct ttm_mem_type_manager *man,
 	return r;
 }
 
-void amdgpu_gtt_mgr_print(struct seq_file *m, struct ttm_mem_type_manager *man)
-{
-	struct amdgpu_device *adev = amdgpu_ttm_adev(man->bdev);
-	struct amdgpu_gtt_mgr *mgr = man->priv;
-
-	seq_printf(m, "man size:%llu pages, gtt available:%llu pages, usage:%lluMB\n",
-		   man->size, mgr->available, (u64)atomic64_read(&adev->gtt_usage) >> 20);
-
-}
 /**
  * amdgpu_gtt_mgr_new - allocate a new node
  *
@@ -182,11 +173,11 @@ static int amdgpu_gtt_mgr_new(struct ttm_mem_type_manager *man,
 	int r;
 
 	spin_lock(&mgr->lock);
-	if (mgr->available < mem->num_pages) {
+	if (atomic64_read(&mgr->available) < mem->num_pages) {
 		spin_unlock(&mgr->lock);
 		return 0;
 	}
-	mgr->available -= mem->num_pages;
+	atomic64_sub(mem->num_pages, &mgr->available);
 	spin_unlock(&mgr->lock);
 
 	node = kzalloc(sizeof(*node), GFP_KERNEL);
@@ -213,9 +204,7 @@ static int amdgpu_gtt_mgr_new(struct ttm_mem_type_manager *man,
 
 	return 0;
 err_out:
-	spin_lock(&mgr->lock);
-	mgr->available += mem->num_pages;
-	spin_unlock(&mgr->lock);
+	atomic64_add(mem->num_pages, &mgr->available);
 
 	return r;
 }
@@ -242,30 +231,47 @@ static void amdgpu_gtt_mgr_del(struct ttm_mem_type_manager *man,
 	spin_lock(&mgr->lock);
 	if (node->start != AMDGPU_BO_INVALID_OFFSET)
 		drm_mm_remove_node(node);
-	mgr->available += mem->num_pages;
 	spin_unlock(&mgr->lock);
+	atomic64_add(mem->num_pages, &mgr->available);
 
 	kfree(node);
 	mem->mm_node = NULL;
 }
 
 /**
+ * amdgpu_gtt_mgr_usage - return usage of GTT domain
+ *
+ * @man: TTM memory type manager
+ *
+ * Return how many bytes are used in the GTT domain
+ */
+uint64_t amdgpu_gtt_mgr_usage(struct ttm_mem_type_manager *man)
+{
+	struct amdgpu_gtt_mgr *mgr = man->priv;
+
+	return (u64)(man->size - atomic64_read(&mgr->available)) * PAGE_SIZE;
+}
+
+/**
  * amdgpu_gtt_mgr_debug - dump VRAM table
  *
  * @man: TTM memory type manager
- * @prefix: text prefix
+ * @printer: DRM printer to use
  *
  * Dump the table content using printk.
  */
 static void amdgpu_gtt_mgr_debug(struct ttm_mem_type_manager *man,
-				  const char *prefix)
+				 struct drm_printer *printer)
 {
 	struct amdgpu_gtt_mgr *mgr = man->priv;
-	struct drm_printer p = drm_debug_printer(prefix);
 
 	spin_lock(&mgr->lock);
-	drm_mm_print(&mgr->mm, &p);
+	drm_mm_print(&mgr->mm, printer);
 	spin_unlock(&mgr->lock);
+
+	drm_printf(printer, "man size:%llu pages, gtt available:%llu pages, usage:%lluMB\n",
+		   man->size, (u64)atomic64_read(&mgr->available),
+		   amdgpu_gtt_mgr_usage(man) >> 20);
 }
 
 const struct ttm_mem_type_manager_func amdgpu_gtt_mgr_func = {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index 09f833255ba1..e16229000a98 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -158,7 +158,6 @@ int amdgpu_driver_load_kms(struct drm_device *dev, unsigned long flags)
 				"Error during ACPI methods call\n");
 	}
 
-	amdgpu_amdkfd_load_interface(adev);
 	amdgpu_amdkfd_device_probe(adev);
 	amdgpu_amdkfd_device_init(adev);
 
@@ -456,13 +455,13 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 		ui64 = atomic64_read(&adev->num_vram_cpu_page_faults);
 		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
 	case AMDGPU_INFO_VRAM_USAGE:
-		ui64 = atomic64_read(&adev->vram_usage);
+		ui64 = amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
 		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
 	case AMDGPU_INFO_VIS_VRAM_USAGE:
-		ui64 = atomic64_read(&adev->vram_vis_usage);
+		ui64 = amdgpu_vram_mgr_vis_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
 		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
 	case AMDGPU_INFO_GTT_USAGE:
-		ui64 = atomic64_read(&adev->gtt_usage);
+		ui64 = amdgpu_gtt_mgr_usage(&adev->mman.bdev.man[TTM_PL_TT]);
 		return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
 	case AMDGPU_INFO_GDS_CONFIG: {
 		struct drm_amdgpu_info_gds gds_info;
@@ -498,7 +497,8 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 		mem.vram.total_heap_size = adev->mc.real_vram_size;
 		mem.vram.usable_heap_size =
 			adev->mc.real_vram_size - adev->vram_pin_size;
-		mem.vram.heap_usage = atomic64_read(&adev->vram_usage);
+		mem.vram.heap_usage =
+			amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
 		mem.vram.max_allocation = mem.vram.usable_heap_size * 3 / 4;
 
 		mem.cpu_accessible_vram.total_heap_size =
@@ -507,7 +507,7 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 			adev->mc.visible_vram_size -
 			(adev->vram_pin_size - adev->invisible_pin_size);
 		mem.cpu_accessible_vram.heap_usage =
-			atomic64_read(&adev->vram_vis_usage);
+			amdgpu_vram_mgr_vis_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
 		mem.cpu_accessible_vram.max_allocation =
 			mem.cpu_accessible_vram.usable_heap_size * 3 / 4;
 
@@ -515,7 +515,8 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 		mem.gtt.total_heap_size *= PAGE_SIZE;
 		mem.gtt.usable_heap_size = mem.gtt.total_heap_size
 			- adev->gart_pin_size;
-		mem.gtt.heap_usage = atomic64_read(&adev->gtt_usage);
+		mem.gtt.heap_usage =
+			amdgpu_gtt_mgr_usage(&adev->mman.bdev.man[TTM_PL_TT]);
 		mem.gtt.max_allocation = mem.gtt.usable_heap_size * 3 / 4;
 
 		return copy_to_user(out, &mem,
@@ -589,11 +590,8 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
 		dev_info.virtual_address_offset = AMDGPU_VA_RESERVED_SIZE;
 		dev_info.virtual_address_max = (uint64_t)adev->vm_manager.max_pfn * AMDGPU_GPU_PAGE_SIZE;
 		dev_info.virtual_address_alignment = max((int)PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE);
-		dev_info.pte_fragment_size =
-			(1 << AMDGPU_LOG2_PAGES_PER_FRAG(adev)) *
-			AMDGPU_GPU_PAGE_SIZE;
+		dev_info.pte_fragment_size = (1 << adev->vm_manager.fragment_size) * AMDGPU_GPU_PAGE_SIZE;
 		dev_info.gart_page_size = AMDGPU_GPU_PAGE_SIZE;
-
 		dev_info.cu_active_number = adev->gfx.cu_info.number;
 		dev_info.cu_ao_mask = adev->gfx.cu_info.ao_cu_mask;
 		dev_info.ce_ram_size = adev->gfx.ce_ram_size;
@@ -842,7 +840,7 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
 	}
 
 	if (amdgpu_sriov_vf(adev)) {
-		r = amdgpu_map_static_csa(adev, &fpriv->vm);
+		r = amdgpu_map_static_csa(adev, &fpriv->vm, &fpriv->csa_va);
 		if (r)
 			goto out_suspend;
 	}
@@ -895,8 +893,8 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev,
 	if (amdgpu_sriov_vf(adev)) {
 		/* TODO: how to handle reserve failure */
 		BUG_ON(amdgpu_bo_reserve(adev->virt.csa_obj, true));
-		amdgpu_vm_bo_rmv(adev, fpriv->vm.csa_bo_va);
-		fpriv->vm.csa_bo_va = NULL;
+		amdgpu_vm_bo_rmv(adev, fpriv->csa_va);
+		fpriv->csa_va = NULL;
 		amdgpu_bo_unreserve(adev->virt.csa_obj);
 	}
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
index 38f739fb727b..6558a3ed57a7 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
@@ -359,7 +359,7 @@ void amdgpu_mn_unregister(struct amdgpu_bo *bo)
 	head = bo->mn_list.next;
 
 	bo->mn = NULL;
-	list_del(&bo->mn_list);
+	list_del_init(&bo->mn_list);
 
 	if (list_empty(head)) {
 		struct amdgpu_mn_node *node;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index 3ec43cf9ad78..e7e899190bef 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -37,55 +37,6 @@
 #include "amdgpu.h"
 #include "amdgpu_trace.h"
 
-
-
-static u64 amdgpu_get_vis_part_size(struct amdgpu_device *adev,
-						struct ttm_mem_reg *mem)
-{
-	if (mem->start << PAGE_SHIFT >= adev->mc.visible_vram_size)
-		return 0;
-
-	return ((mem->start << PAGE_SHIFT) + mem->size) >
-		adev->mc.visible_vram_size ?
-		adev->mc.visible_vram_size - (mem->start << PAGE_SHIFT) :
-		mem->size;
-}
-
-static void amdgpu_update_memory_usage(struct amdgpu_device *adev,
-		       struct ttm_mem_reg *old_mem,
-		       struct ttm_mem_reg *new_mem)
-{
-	u64 vis_size;
-	if (!adev)
-		return;
-
-	if (new_mem) {
-		switch (new_mem->mem_type) {
-		case TTM_PL_TT:
-			atomic64_add(new_mem->size, &adev->gtt_usage);
-			break;
-		case TTM_PL_VRAM:
-			atomic64_add(new_mem->size, &adev->vram_usage);
-			vis_size = amdgpu_get_vis_part_size(adev, new_mem);
-			atomic64_add(vis_size, &adev->vram_vis_usage);
-			break;
-		}
-	}
-
-	if (old_mem) {
-		switch (old_mem->mem_type) {
-		case TTM_PL_TT:
-			atomic64_sub(old_mem->size, &adev->gtt_usage);
-			break;
-		case TTM_PL_VRAM:
-			atomic64_sub(old_mem->size, &adev->vram_usage);
-			vis_size = amdgpu_get_vis_part_size(adev, old_mem);
-			atomic64_sub(vis_size, &adev->vram_vis_usage);
-			break;
-		}
-	}
-}
-
 static void amdgpu_ttm_bo_destroy(struct ttm_buffer_object *tbo)
 {
 	struct amdgpu_device *adev = amdgpu_ttm_adev(tbo->bdev);
@@ -94,7 +45,6 @@ static void amdgpu_ttm_bo_destroy(struct ttm_buffer_object *tbo)
 	bo = container_of(tbo, struct amdgpu_bo, tbo);
 
 	amdgpu_bo_kunmap(bo);
-	amdgpu_update_memory_usage(adev, &bo->tbo.mem, NULL);
 
 	drm_gem_object_release(&bo->gem_base);
 	amdgpu_bo_unref(&bo->parent);
@@ -220,7 +170,7 @@ static void amdgpu_fill_placement_to_bo(struct amdgpu_bo *bo,
 }
 
 /**
- * amdgpu_bo_create_kernel - create BO for kernel use
+ * amdgpu_bo_create_reserved - create reserved BO for kernel use
  *
  * @adev: amdgpu device object
  * @size: size for the new BO
@@ -230,24 +180,30 @@ static void amdgpu_fill_placement_to_bo(struct amdgpu_bo *bo,
  * @gpu_addr: GPU addr of the pinned BO
  * @cpu_addr: optional CPU address mapping
  *
- * Allocates and pins a BO for kernel internal use.
+ * Allocates and pins a BO for kernel internal use, and returns it still
+ * reserved.
  *
  * Returns 0 on success, negative error code otherwise.
  */
-int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
-			    unsigned long size, int align,
-			    u32 domain, struct amdgpu_bo **bo_ptr,
-			    u64 *gpu_addr, void **cpu_addr)
+int amdgpu_bo_create_reserved(struct amdgpu_device *adev,
+			      unsigned long size, int align,
+			      u32 domain, struct amdgpu_bo **bo_ptr,
+			      u64 *gpu_addr, void **cpu_addr)
 {
+	bool free = false;
 	int r;
 
-	r = amdgpu_bo_create(adev, size, align, true, domain,
-			     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-			     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-			     NULL, NULL, bo_ptr);
-	if (r) {
-		dev_err(adev->dev, "(%d) failed to allocate kernel bo\n", r);
-		return r;
+	if (!*bo_ptr) {
+		r = amdgpu_bo_create(adev, size, align, true, domain,
+				     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
+				     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
+				     NULL, NULL, 0, bo_ptr);
+		if (r) {
+			dev_err(adev->dev, "(%d) failed to allocate kernel bo\n",
+				r);
+			return r;
+		}
+		free = true;
 	}
 
 	r = amdgpu_bo_reserve(*bo_ptr, false);
@@ -270,20 +226,52 @@ int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
 		}
 	}
 
-	amdgpu_bo_unreserve(*bo_ptr);
-
 	return 0;
 
 error_unreserve:
 	amdgpu_bo_unreserve(*bo_ptr);
 
 error_free:
-	amdgpu_bo_unref(bo_ptr);
+	if (free)
+		amdgpu_bo_unref(bo_ptr);
 
 	return r;
 }
 
 /**
+ * amdgpu_bo_create_kernel - create BO for kernel use
+ *
+ * @adev: amdgpu device object
+ * @size: size for the new BO
+ * @align: alignment for the new BO
+ * @domain: where to place it
+ * @bo_ptr: resulting BO
+ * @gpu_addr: GPU addr of the pinned BO
+ * @cpu_addr: optional CPU address mapping
+ *
+ * Allocates and pins a BO for kernel internal use.
+ *
+ * Returns 0 on success, negative error code otherwise.
+ */
+int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
+			    unsigned long size, int align,
+			    u32 domain, struct amdgpu_bo **bo_ptr,
+			    u64 *gpu_addr, void **cpu_addr)
+{
+	int r;
+
+	r = amdgpu_bo_create_reserved(adev, size, align, domain, bo_ptr,
+				      gpu_addr, cpu_addr);
+
+	if (r)
+		return r;
+
+	amdgpu_bo_unreserve(*bo_ptr);
+
+	return 0;
+}
+
+/**
  * amdgpu_bo_free_kernel - free BO for kernel use
  *
  * @bo: amdgpu BO to free
@@ -318,6 +306,7 @@ int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
 				struct sg_table *sg,
 				struct ttm_placement *placement,
 				struct reservation_object *resv,
+				uint64_t init_value,
 				struct amdgpu_bo **bo_ptr)
 {
 	struct amdgpu_bo *bo;
@@ -352,13 +341,13 @@ int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
 	}
 	INIT_LIST_HEAD(&bo->shadow_list);
 	INIT_LIST_HEAD(&bo->va);
-	bo->prefered_domains = domain & (AMDGPU_GEM_DOMAIN_VRAM |
+	bo->preferred_domains = domain & (AMDGPU_GEM_DOMAIN_VRAM |
 					 AMDGPU_GEM_DOMAIN_GTT |
 					 AMDGPU_GEM_DOMAIN_CPU |
 					 AMDGPU_GEM_DOMAIN_GDS |
 					 AMDGPU_GEM_DOMAIN_GWS |
 					 AMDGPU_GEM_DOMAIN_OA);
-	bo->allowed_domains = bo->prefered_domains;
+	bo->allowed_domains = bo->preferred_domains;
 	if (!kernel && bo->allowed_domains == AMDGPU_GEM_DOMAIN_VRAM)
 		bo->allowed_domains |= AMDGPU_GEM_DOMAIN_GTT;
 
@@ -418,7 +407,7 @@ int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
 	    bo->tbo.mem.placement & TTM_PL_FLAG_VRAM) {
 		struct dma_fence *fence;
 
-		r = amdgpu_fill_buffer(bo, 0, bo->tbo.resv, &fence);
+		r = amdgpu_fill_buffer(bo, init_value, bo->tbo.resv, &fence);
 		if (unlikely(r))
 			goto fail_unreserve;
 
@@ -470,6 +459,7 @@ static int amdgpu_bo_create_shadow(struct amdgpu_device *adev,
 					AMDGPU_GEM_CREATE_CPU_GTT_USWC,
 					NULL, &placement,
 					bo->tbo.resv,
+					0,
 					&bo->shadow);
 	if (!r) {
 		bo->shadow->parent = amdgpu_bo_ref(bo);
@@ -481,11 +471,15 @@ static int amdgpu_bo_create_shadow(struct amdgpu_device *adev,
 	return r;
 }
 
+/* init_value will only take effect when flags contains
+ * AMDGPU_GEM_CREATE_VRAM_CLEARED.
+ */
 int amdgpu_bo_create(struct amdgpu_device *adev,
 		     unsigned long size, int byte_align,
 		     bool kernel, u32 domain, u64 flags,
 		     struct sg_table *sg,
 		     struct reservation_object *resv,
+		     uint64_t init_value,
 		     struct amdgpu_bo **bo_ptr)
 {
 	struct ttm_placement placement = {0};
@@ -500,7 +494,7 @@ int amdgpu_bo_create(struct amdgpu_device *adev,
 
 	r = amdgpu_bo_create_restricted(adev, size, byte_align, kernel,
 					domain, flags, sg, &placement,
-					resv, bo_ptr);
+					resv, init_value, bo_ptr);
 	if (r)
 		return r;
 
@@ -562,7 +556,7 @@ int amdgpu_bo_validate(struct amdgpu_bo *bo)
 	if (bo->pin_count)
 		return 0;
 
-	domain = bo->prefered_domains;
+	domain = bo->preferred_domains;
 
 retry:
 	amdgpu_ttm_placement_from_domain(bo, domain);
@@ -609,16 +603,16 @@ err:
 
 int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr)
 {
-	bool is_iomem;
+	void *kptr;
 	long r;
 
 	if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)
 		return -EPERM;
 
-	if (bo->kptr) {
-		if (ptr) {
-			*ptr = bo->kptr;
-		}
+	kptr = amdgpu_bo_kptr(bo);
+	if (kptr) {
+		if (ptr)
+			*ptr = kptr;
 		return 0;
 	}
 
@@ -631,19 +625,23 @@ int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr)
 	if (r)
 		return r;
 
-	bo->kptr = ttm_kmap_obj_virtual(&bo->kmap, &is_iomem);
 	if (ptr)
-		*ptr = bo->kptr;
+		*ptr = amdgpu_bo_kptr(bo);
 
 	return 0;
 }
 
+void *amdgpu_bo_kptr(struct amdgpu_bo *bo)
+{
+	bool is_iomem;
+
+	return ttm_kmap_obj_virtual(&bo->kmap, &is_iomem);
+}
+
 void amdgpu_bo_kunmap(struct amdgpu_bo *bo)
 {
-	if (bo->kptr == NULL)
-		return;
-	bo->kptr = NULL;
-	ttm_bo_kunmap(&bo->kmap);
+	if (bo->kmap.bo)
+		ttm_bo_kunmap(&bo->kmap);
 }
 
 struct amdgpu_bo *amdgpu_bo_ref(struct amdgpu_bo *bo)
@@ -944,8 +942,6 @@ void amdgpu_bo_move_notify(struct ttm_buffer_object *bo,
 		return;
 
 	/* move_notify is called before move happens */
-	amdgpu_update_memory_usage(adev, &bo->mem, new_mem);
-
 	trace_amdgpu_ttm_bo_move(abo, new_mem->mem_type, old_mem->mem_type);
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
index 833b172a2c2a..a288fa6d72c8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
@@ -33,6 +33,61 @@
 
 #define AMDGPU_BO_INVALID_OFFSET	LONG_MAX
 
+/* bo virtual addresses in a vm */
+struct amdgpu_bo_va_mapping {
+	struct list_head		list;
+	struct rb_node			rb;
+	uint64_t			start;
+	uint64_t			last;
+	uint64_t			__subtree_last;
+	uint64_t			offset;
+	uint64_t			flags;
+};
+
+/* User space allocated BO in a VM */
+struct amdgpu_bo_va {
+	struct amdgpu_vm_bo_base	base;
+
+	/* protected by bo being reserved */
+	struct dma_fence	        *last_pt_update;
+	unsigned			ref_count;
+
+	/* mappings for this bo_va */
+	struct list_head		invalids;
+	struct list_head		valids;
+};
+
+struct amdgpu_bo {
+	/* Protected by tbo.reserved */
+	u32				preferred_domains;
+	u32				allowed_domains;
+	struct ttm_place		placements[AMDGPU_GEM_DOMAIN_MAX + 1];
+	struct ttm_placement		placement;
+	struct ttm_buffer_object	tbo;
+	struct ttm_bo_kmap_obj		kmap;
+	u64				flags;
+	unsigned			pin_count;
+	u64				tiling_flags;
+	u64				metadata_flags;
+	void				*metadata;
+	u32				metadata_size;
+	unsigned			prime_shared_count;
+	/* list of all virtual address to which this bo is associated to */
+	struct list_head		va;
+	/* Constant after initialization */
+	struct drm_gem_object		gem_base;
+	struct amdgpu_bo		*parent;
+	struct amdgpu_bo		*shadow;
+
+	struct ttm_bo_kmap_obj		dma_buf_vmap;
+	struct amdgpu_mn		*mn;
+
+	union {
+		struct list_head	mn_list;
+		struct list_head	shadow_list;
+	};
+};
+
 /**
  * amdgpu_mem_type_to_domain - return domain corresponding to mem_type
  * @mem_type:	ttm memory type
@@ -132,6 +187,7 @@ int amdgpu_bo_create(struct amdgpu_device *adev,
 			    bool kernel, u32 domain, u64 flags,
 			    struct sg_table *sg,
 			    struct reservation_object *resv,
+			    uint64_t init_value,
 			    struct amdgpu_bo **bo_ptr);
 int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
 				unsigned long size, int byte_align,
@@ -139,7 +195,12 @@ int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
 				struct sg_table *sg,
 				struct ttm_placement *placement,
 			        struct reservation_object *resv,
+				uint64_t init_value,
 				struct amdgpu_bo **bo_ptr);
+int amdgpu_bo_create_reserved(struct amdgpu_device *adev,
+			      unsigned long size, int align,
+			      u32 domain, struct amdgpu_bo **bo_ptr,
+			      u64 *gpu_addr, void **cpu_addr);
 int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
 			    unsigned long size, int align,
 			    u32 domain, struct amdgpu_bo **bo_ptr,
@@ -147,6 +208,7 @@ int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
 void amdgpu_bo_free_kernel(struct amdgpu_bo **bo, u64 *gpu_addr,
 			   void **cpu_addr);
 int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr);
+void *amdgpu_bo_kptr(struct amdgpu_bo *bo);
 void amdgpu_bo_kunmap(struct amdgpu_bo *bo);
 struct amdgpu_bo *amdgpu_bo_ref(struct amdgpu_bo *bo);
 void amdgpu_bo_unref(struct amdgpu_bo **bo);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.h
index c19c4d138751..f21a7716b90e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.h
@@ -30,6 +30,7 @@ struct cg_flag_name
 	const char *name;
 };
 
+void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev);
 int amdgpu_pm_sysfs_init(struct amdgpu_device *adev);
 void amdgpu_pm_sysfs_fini(struct amdgpu_device *adev);
 void amdgpu_pm_print_power_states(struct amdgpu_device *adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c
index 6bdc866570ab..5b3f92891f89 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c
@@ -69,7 +69,7 @@ amdgpu_gem_prime_import_sg_table(struct drm_device *dev,
 
 	ww_mutex_lock(&resv->lock, NULL);
 	ret = amdgpu_bo_create(adev, attach->dmabuf->size, PAGE_SIZE, false,
-			       AMDGPU_GEM_DOMAIN_GTT, 0, sg, resv, &bo);
+			       AMDGPU_GEM_DOMAIN_GTT, 0, sg, resv, 0, &bo);
 	ww_mutex_unlock(&resv->lock);
 	if (ret)
 		return ERR_PTR(ret);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
index 15b7149d1204..6c5646b48d1a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
@@ -184,47 +184,22 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
 			return r;
 	}
 
-	if (ring->funcs->support_64bit_ptrs) {
-		r = amdgpu_wb_get_64bit(adev, &ring->rptr_offs);
-		if (r) {
-			dev_err(adev->dev, "(%d) ring rptr_offs wb alloc failed\n", r);
-			return r;
-		}
-
-		r = amdgpu_wb_get_64bit(adev, &ring->wptr_offs);
-		if (r) {
-			dev_err(adev->dev, "(%d) ring wptr_offs wb alloc failed\n", r);
-			return r;
-		}
-
-	} else {
-		r = amdgpu_wb_get(adev, &ring->rptr_offs);
-		if (r) {
-			dev_err(adev->dev, "(%d) ring rptr_offs wb alloc failed\n", r);
-			return r;
-		}
-
-		r = amdgpu_wb_get(adev, &ring->wptr_offs);
-		if (r) {
-			dev_err(adev->dev, "(%d) ring wptr_offs wb alloc failed\n", r);
-			return r;
-		}
-
+	r = amdgpu_wb_get(adev, &ring->rptr_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring rptr_offs wb alloc failed\n", r);
+		return r;
 	}
 
-	if (amdgpu_sriov_vf(adev) && ring->funcs->type == AMDGPU_RING_TYPE_GFX) {
-		r = amdgpu_wb_get_256Bit(adev, &ring->fence_offs);
-		if (r) {
-			dev_err(adev->dev, "(%d) ring fence_offs wb alloc failed\n", r);
-			return r;
-		}
+	r = amdgpu_wb_get(adev, &ring->wptr_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring wptr_offs wb alloc failed\n", r);
+		return r;
+	}
 
-	} else {
-		r = amdgpu_wb_get(adev, &ring->fence_offs);
-		if (r) {
-			dev_err(adev->dev, "(%d) ring fence_offs wb alloc failed\n", r);
-			return r;
-		}
+	r = amdgpu_wb_get(adev, &ring->fence_offs);
+	if (r) {
+		dev_err(adev->dev, "(%d) ring fence_offs wb alloc failed\n", r);
+		return r;
 	}
 
 	r = amdgpu_wb_get(adev, &ring->cond_exe_offs);
@@ -286,19 +261,15 @@ void amdgpu_ring_fini(struct amdgpu_ring *ring)
 {
 	ring->ready = false;
 
-	if (ring->funcs->support_64bit_ptrs) {
-		amdgpu_wb_free_64bit(ring->adev, ring->rptr_offs);
-		amdgpu_wb_free_64bit(ring->adev, ring->wptr_offs);
-	} else {
-		amdgpu_wb_free(ring->adev, ring->rptr_offs);
-		amdgpu_wb_free(ring->adev, ring->wptr_offs);
-	}
+	/* Not to finish a ring which is not initialized */
+	if (!(ring->adev) || !(ring->adev->rings[ring->idx]))
+		return;
+
+	amdgpu_wb_free(ring->adev, ring->rptr_offs);
+	amdgpu_wb_free(ring->adev, ring->wptr_offs);
 
 	amdgpu_wb_free(ring->adev, ring->cond_exe_offs);
-	if (amdgpu_sriov_vf(ring->adev) && ring->funcs->type == AMDGPU_RING_TYPE_GFX)
-		amdgpu_wb_free_256bit(ring->adev, ring->fence_offs);
-	else
-		amdgpu_wb_free(ring->adev, ring->fence_offs);
+	amdgpu_wb_free(ring->adev, ring->fence_offs);
 
 	amdgpu_bo_free_kernel(&ring->ring_obj,
 			      &ring->gpu_addr,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
index 5ca75a456ad2..3144400435b7 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
@@ -64,7 +64,7 @@ int amdgpu_sa_bo_manager_init(struct amdgpu_device *adev,
 		INIT_LIST_HEAD(&sa_manager->flist[i]);
 
 	r = amdgpu_bo_create(adev, size, align, true, domain,
-			     0, NULL, NULL, &sa_manager->bo);
+			     0, NULL, NULL, 0, &sa_manager->bo);
 	if (r) {
 		dev_err(adev->dev, "(%d) failed to allocate bo for manager\n", r);
 		return r;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c
index a6899180b265..c586f44312f9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c
@@ -244,6 +244,12 @@ struct dma_fence *amdgpu_sync_peek_fence(struct amdgpu_sync *sync,
 		struct dma_fence *f = e->fence;
 		struct amd_sched_fence *s_fence = to_amd_sched_fence(f);
 
+		if (dma_fence_is_signaled(f)) {
+			hash_del(&e->node);
+			dma_fence_put(f);
+			kmem_cache_free(amdgpu_sync_slab, e);
+			continue;
+		}
 		if (ring && s_fence) {
 			/* For fences from the same ring it is sufficient
 			 * when they are scheduled.
@@ -256,13 +262,6 @@ struct dma_fence *amdgpu_sync_peek_fence(struct amdgpu_sync *sync,
 			}
 		}
 
-		if (dma_fence_is_signaled(f)) {
-			hash_del(&e->node);
-			dma_fence_put(f);
-			kmem_cache_free(amdgpu_sync_slab, e);
-			continue;
-		}
-
 		return f;
 	}
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c
index 3c4d7574d704..ed8c3739015b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c
@@ -61,7 +61,7 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev)
 
 	r = amdgpu_bo_create(adev, size, PAGE_SIZE, true,
 			     AMDGPU_GEM_DOMAIN_VRAM, 0,
-			     NULL, NULL, &vram_obj);
+			     NULL, NULL, 0, &vram_obj);
 	if (r) {
 		DRM_ERROR("Failed to create VRAM object\n");
 		goto out_cleanup;
@@ -82,7 +82,7 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev)
 
 		r = amdgpu_bo_create(adev, size, PAGE_SIZE, true,
 				     AMDGPU_GEM_DOMAIN_GTT, 0, NULL,
-				     NULL, gtt_obj + i);
+				     NULL, 0, gtt_obj + i);
 		if (r) {
 			DRM_ERROR("Failed to create GTT object %d\n", i);
 			goto out_lclean;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
index 509f7a63d40c..1c88bd5e29ad 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
@@ -14,6 +14,62 @@
 #define AMDGPU_JOB_GET_TIMELINE_NAME(job) \
 	 job->base.s_fence->finished.ops->get_timeline_name(&job->base.s_fence->finished)
 
+TRACE_EVENT(amdgpu_ttm_tt_populate,
+	    TP_PROTO(struct amdgpu_device *adev, uint64_t dma_address, uint64_t phys_address),
+	    TP_ARGS(adev, dma_address, phys_address),
+	    TP_STRUCT__entry(
+				__field(uint16_t, domain)
+				__field(uint8_t, bus)
+				__field(uint8_t, slot)
+				__field(uint8_t, func)
+				__field(uint64_t, dma)
+				__field(uint64_t, phys)
+			    ),
+	    TP_fast_assign(
+			   __entry->domain = pci_domain_nr(adev->pdev->bus);
+			   __entry->bus = adev->pdev->bus->number;
+			   __entry->slot = PCI_SLOT(adev->pdev->devfn);
+			   __entry->func = PCI_FUNC(adev->pdev->devfn);
+			   __entry->dma = dma_address;
+			   __entry->phys = phys_address;
+			   ),
+	    TP_printk("%04x:%02x:%02x.%x: 0x%llx => 0x%llx",
+		      (unsigned)__entry->domain,
+		      (unsigned)__entry->bus,
+		      (unsigned)__entry->slot,
+		      (unsigned)__entry->func,
+		      (unsigned long long)__entry->dma,
+		      (unsigned long long)__entry->phys)
+);
+
+TRACE_EVENT(amdgpu_ttm_tt_unpopulate,
+	    TP_PROTO(struct amdgpu_device *adev, uint64_t dma_address, uint64_t phys_address),
+	    TP_ARGS(adev, dma_address, phys_address),
+	    TP_STRUCT__entry(
+				__field(uint16_t, domain)
+				__field(uint8_t, bus)
+				__field(uint8_t, slot)
+				__field(uint8_t, func)
+				__field(uint64_t, dma)
+				__field(uint64_t, phys)
+			    ),
+	    TP_fast_assign(
+			   __entry->domain = pci_domain_nr(adev->pdev->bus);
+			   __entry->bus = adev->pdev->bus->number;
+			   __entry->slot = PCI_SLOT(adev->pdev->devfn);
+			   __entry->func = PCI_FUNC(adev->pdev->devfn);
+			   __entry->dma = dma_address;
+			   __entry->phys = phys_address;
+			   ),
+	    TP_printk("%04x:%02x:%02x.%x: 0x%llx => 0x%llx",
+		      (unsigned)__entry->domain,
+		      (unsigned)__entry->bus,
+		      (unsigned)__entry->slot,
+		      (unsigned)__entry->func,
+		      (unsigned long long)__entry->dma,
+		      (unsigned long long)__entry->phys)
+);
+
 TRACE_EVENT(amdgpu_mm_rreg,
 	    TP_PROTO(unsigned did, uint32_t reg, uint32_t value),
 	    TP_ARGS(did, reg, value),
@@ -105,12 +161,12 @@ TRACE_EVENT(amdgpu_bo_create,
 			   __entry->bo = bo;
 			   __entry->pages = bo->tbo.num_pages;
 			   __entry->type = bo->tbo.mem.mem_type;
-			   __entry->prefer = bo->prefered_domains;
+			   __entry->prefer = bo->preferred_domains;
 			   __entry->allow = bo->allowed_domains;
 			   __entry->visible = bo->flags;
 			   ),
 
-	    TP_printk("bo=%p, pages=%u, type=%d, prefered=%d, allowed=%d, visible=%d",
+	    TP_printk("bo=%p, pages=%u, type=%d, preferred=%d, allowed=%d, visible=%d",
 		       __entry->bo, __entry->pages, __entry->type,
 		       __entry->prefer, __entry->allow, __entry->visible)
 );
@@ -228,7 +284,7 @@ TRACE_EVENT(amdgpu_vm_bo_map,
 			     ),
 
 	    TP_fast_assign(
-			   __entry->bo = bo_va ? bo_va->bo : NULL;
+			   __entry->bo = bo_va ? bo_va->base.bo : NULL;
 			   __entry->start = mapping->start;
 			   __entry->last = mapping->last;
 			   __entry->offset = mapping->offset;
@@ -252,7 +308,7 @@ TRACE_EVENT(amdgpu_vm_bo_unmap,
 			     ),
 
 	    TP_fast_assign(
-			   __entry->bo = bo_va->bo;
+			   __entry->bo = bo_va->base.bo;
 			   __entry->start = mapping->start;
 			   __entry->last = mapping->last;
 			   __entry->offset = mapping->offset;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
index e6f9a54c959d..8b2c294f6f79 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -43,6 +43,7 @@
 #include <linux/pagemap.h>
 #include <linux/debugfs.h>
 #include "amdgpu.h"
+#include "amdgpu_trace.h"
 #include "bif/bif_4_1_d.h"
 
 #define DRM_FILE_PAGE_OFFSET (0x100000000ULL >> PAGE_SHIFT)
@@ -662,6 +663,38 @@ release_pages:
 	return r;
 }
 
+static void amdgpu_trace_dma_map(struct ttm_tt *ttm)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev);
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	unsigned i;
+
+	if (unlikely(trace_amdgpu_ttm_tt_populate_enabled())) {
+		for (i = 0; i < ttm->num_pages; i++) {
+			trace_amdgpu_ttm_tt_populate(
+				adev,
+				gtt->ttm.dma_address[i],
+				page_to_phys(ttm->pages[i]));
+		}
+	}
+}
+
+static void amdgpu_trace_dma_unmap(struct ttm_tt *ttm)
+{
+	struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev);
+	struct amdgpu_ttm_tt *gtt = (void *)ttm;
+	unsigned i;
+
+	if (unlikely(trace_amdgpu_ttm_tt_unpopulate_enabled())) {
+		for (i = 0; i < ttm->num_pages; i++) {
+			trace_amdgpu_ttm_tt_unpopulate(
+				adev,
+				gtt->ttm.dma_address[i],
+				page_to_phys(ttm->pages[i]));
+		}
+	}
+}
+
 /* prepare the sg table with the user pages */
 static int amdgpu_ttm_tt_pin_userptr(struct ttm_tt *ttm)
 {
@@ -688,6 +721,8 @@ static int amdgpu_ttm_tt_pin_userptr(struct ttm_tt *ttm)
 	drm_prime_sg_to_page_addr_arrays(ttm->sg, ttm->pages,
 					 gtt->ttm.dma_address, ttm->num_pages);
 
+	amdgpu_trace_dma_map(ttm);
+
 	return 0;
 
 release_sg:
@@ -721,6 +756,8 @@ static void amdgpu_ttm_tt_unpin_userptr(struct ttm_tt *ttm)
 		put_page(page);
 	}
 
+	amdgpu_trace_dma_unmap(ttm);
+
 	sg_free_table(ttm->sg);
 }
 
@@ -753,7 +790,7 @@ static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm,
 				   struct ttm_mem_reg *bo_mem)
 {
 	struct amdgpu_ttm_tt *gtt = (void*)ttm;
-	int r;
+	int r = 0;
 
 	if (gtt->userptr) {
 		r = amdgpu_ttm_tt_pin_userptr(ttm);
@@ -892,7 +929,7 @@ static struct ttm_tt *amdgpu_ttm_tt_create(struct ttm_bo_device *bdev,
 
 static int amdgpu_ttm_tt_populate(struct ttm_tt *ttm)
 {
-	struct amdgpu_device *adev;
+	struct amdgpu_device *adev = amdgpu_ttm_adev(ttm->bdev);
 	struct amdgpu_ttm_tt *gtt = (void *)ttm;
 	unsigned i;
 	int r;
@@ -915,14 +952,14 @@ static int amdgpu_ttm_tt_populate(struct ttm_tt *ttm)
 		drm_prime_sg_to_page_addr_arrays(ttm->sg, ttm->pages,
 						 gtt->ttm.dma_address, ttm->num_pages);
 		ttm->state = tt_unbound;
-		return 0;
+		r = 0;
+		goto trace_mappings;
 	}
 
-	adev = amdgpu_ttm_adev(ttm->bdev);
-
 #ifdef CONFIG_SWIOTLB
 	if (swiotlb_nr_tbl()) {
-		return ttm_dma_populate(&gtt->ttm, adev->dev);
+		r = ttm_dma_populate(&gtt->ttm, adev->dev);
+		goto trace_mappings;
 	}
 #endif
 
@@ -945,7 +982,12 @@ static int amdgpu_ttm_tt_populate(struct ttm_tt *ttm)
 			return -EFAULT;
 		}
 	}
-	return 0;
+
+	r = 0;
+trace_mappings:
+	if (likely(!r))
+		amdgpu_trace_dma_map(ttm);
+	return r;
 }
 
 static void amdgpu_ttm_tt_unpopulate(struct ttm_tt *ttm)
@@ -966,6 +1008,8 @@ static void amdgpu_ttm_tt_unpopulate(struct ttm_tt *ttm)
 
 	adev = amdgpu_ttm_adev(ttm->bdev);
 
+	amdgpu_trace_dma_unmap(ttm);
+
 #ifdef CONFIG_SWIOTLB
 	if (swiotlb_nr_tbl()) {
 		ttm_dma_unpopulate(&gtt->ttm, adev->dev);
@@ -1232,23 +1276,12 @@ int amdgpu_ttm_init(struct amdgpu_device *adev)
 	/* Change the size here instead of the init above so only lpfn is affected */
 	amdgpu_ttm_set_active_vram_size(adev, adev->mc.visible_vram_size);
 
-	r = amdgpu_bo_create(adev, adev->mc.stolen_size, PAGE_SIZE, true,
-			     AMDGPU_GEM_DOMAIN_VRAM,
-			     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-			     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-			     NULL, NULL, &adev->stollen_vga_memory);
-	if (r) {
-		return r;
-	}
-	r = amdgpu_bo_reserve(adev->stollen_vga_memory, false);
+	r = amdgpu_bo_create_kernel(adev, adev->mc.stolen_size, PAGE_SIZE,
+				    AMDGPU_GEM_DOMAIN_VRAM,
+				    &adev->stolen_vga_memory,
+				    NULL, NULL);
 	if (r)
 		return r;
-	r = amdgpu_bo_pin(adev->stollen_vga_memory, AMDGPU_GEM_DOMAIN_VRAM, NULL);
-	amdgpu_bo_unreserve(adev->stollen_vga_memory);
-	if (r) {
-		amdgpu_bo_unref(&adev->stollen_vga_memory);
-		return r;
-	}
 	DRM_INFO("amdgpu: %uM of VRAM memory ready\n",
 		 (unsigned) (adev->mc.real_vram_size / (1024 * 1024)));
 
@@ -1319,13 +1352,13 @@ void amdgpu_ttm_fini(struct amdgpu_device *adev)
 	if (!adev->mman.initialized)
 		return;
 	amdgpu_ttm_debugfs_fini(adev);
-	if (adev->stollen_vga_memory) {
-		r = amdgpu_bo_reserve(adev->stollen_vga_memory, true);
+	if (adev->stolen_vga_memory) {
+		r = amdgpu_bo_reserve(adev->stolen_vga_memory, true);
 		if (r == 0) {
-			amdgpu_bo_unpin(adev->stollen_vga_memory);
-			amdgpu_bo_unreserve(adev->stollen_vga_memory);
+			amdgpu_bo_unpin(adev->stolen_vga_memory);
+			amdgpu_bo_unreserve(adev->stolen_vga_memory);
 		}
-		amdgpu_bo_unref(&adev->stollen_vga_memory);
+		amdgpu_bo_unref(&adev->stolen_vga_memory);
 	}
 	ttm_bo_clean_mm(&adev->mman.bdev, TTM_PL_VRAM);
 	ttm_bo_clean_mm(&adev->mman.bdev, TTM_PL_TT);
@@ -1509,11 +1542,12 @@ error_free:
 }
 
 int amdgpu_fill_buffer(struct amdgpu_bo *bo,
-		       uint32_t src_data,
+		       uint64_t src_data,
 		       struct reservation_object *resv,
 		       struct dma_fence **fence)
 {
 	struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+	/* max_bytes applies to SDMA_OP_PTEPDE as well as SDMA_OP_CONST_FILL*/
 	uint32_t max_bytes = adev->mman.buffer_funcs->fill_max_bytes;
 	struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
 
@@ -1545,7 +1579,9 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo,
 		num_pages -= mm_node->size;
 		++mm_node;
 	}
-	num_dw = num_loops * adev->mman.buffer_funcs->fill_num_dw;
+
+	/* 10 double words for each SDMA_OP_PTEPDE cmd */
+	num_dw = num_loops * 10;
 
 	/* for IB padding */
 	num_dw += 64;
@@ -1570,12 +1606,16 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo,
 		uint32_t byte_count = mm_node->size << PAGE_SHIFT;
 		uint64_t dst_addr;
 
+		WARN_ONCE(byte_count & 0x7, "size should be a multiple of 8");
+
 		dst_addr = amdgpu_mm_node_addr(&bo->tbo, mm_node, &bo->tbo.mem);
 		while (byte_count) {
 			uint32_t cur_size_in_bytes = min(byte_count, max_bytes);
 
-			amdgpu_emit_fill_buffer(adev, &job->ibs[0], src_data,
-						dst_addr, cur_size_in_bytes);
+			amdgpu_vm_set_pte_pde(adev, &job->ibs[0],
+					dst_addr, 0,
+					cur_size_in_bytes >> 3, 0,
+					src_data);
 
 			dst_addr += cur_size_in_bytes;
 			byte_count -= cur_size_in_bytes;
@@ -1601,32 +1641,16 @@ error_free:
 
 #if defined(CONFIG_DEBUG_FS)
 
-extern void amdgpu_gtt_mgr_print(struct seq_file *m, struct ttm_mem_type_manager
-				 *man);
 static int amdgpu_mm_dump_table(struct seq_file *m, void *data)
 {
 	struct drm_info_node *node = (struct drm_info_node *)m->private;
 	unsigned ttm_pl = *(int *)node->info_ent->data;
 	struct drm_device *dev = node->minor->dev;
 	struct amdgpu_device *adev = dev->dev_private;
-	struct drm_mm *mm = (struct drm_mm *)adev->mman.bdev.man[ttm_pl].priv;
-	struct ttm_bo_global *glob = adev->mman.bdev.glob;
+	struct ttm_mem_type_manager *man = &adev->mman.bdev.man[ttm_pl];
 	struct drm_printer p = drm_seq_file_printer(m);
 
-	spin_lock(&glob->lru_lock);
-	drm_mm_print(mm, &p);
-	spin_unlock(&glob->lru_lock);
-	switch (ttm_pl) {
-	case TTM_PL_VRAM:
-		seq_printf(m, "man size:%llu pages, ram usage:%lluMB, vis usage:%lluMB\n",
-			   adev->mman.bdev.man[ttm_pl].size,
-			   (u64)atomic64_read(&adev->vram_usage) >> 20,
-			   (u64)atomic64_read(&adev->vram_vis_usage) >> 20);
-		break;
-	case TTM_PL_TT:
-		amdgpu_gtt_mgr_print(m, &adev->mman.bdev.man[TTM_PL_TT]);
-		break;
-	}
+	man->func->debug(man, &p);
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
index f137c2458ee8..f22a4758719d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
@@ -66,6 +66,10 @@ int amdgpu_gtt_mgr_alloc(struct ttm_mem_type_manager *man,
 			 struct ttm_buffer_object *tbo,
 			 const struct ttm_place *place,
 			 struct ttm_mem_reg *mem);
+uint64_t amdgpu_gtt_mgr_usage(struct ttm_mem_type_manager *man);
+
+uint64_t amdgpu_vram_mgr_usage(struct ttm_mem_type_manager *man);
+uint64_t amdgpu_vram_mgr_vis_usage(struct ttm_mem_type_manager *man);
 
 int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
 		       uint64_t dst_offset, uint32_t byte_count,
@@ -73,7 +77,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
 		       struct dma_fence **fence, bool direct_submit,
 		       bool vm_needs_flush);
 int amdgpu_fill_buffer(struct amdgpu_bo *bo,
-			uint32_t src_data,
+			uint64_t src_data,
 			struct reservation_object *resv,
 			struct dma_fence **fence);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
index fcfb9d4f7477..36c763310df5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
@@ -358,8 +358,6 @@ static int amdgpu_ucode_patch_jt(struct amdgpu_firmware_info *ucode,
 			   (le32_to_cpu(header->jt_offset) * 4);
 	memcpy(dst_addr, src_addr, le32_to_cpu(header->jt_size) * 4);
 
-	ucode->ucode_size += le32_to_cpu(header->jt_size) * 4;
-
 	return 0;
 }
 
@@ -381,7 +379,7 @@ int amdgpu_ucode_init_bo(struct amdgpu_device *adev)
 	err = amdgpu_bo_create(adev, adev->firmware.fw_size, PAGE_SIZE, true,
 				amdgpu_sriov_vf(adev) ? AMDGPU_GEM_DOMAIN_VRAM : AMDGPU_GEM_DOMAIN_GTT,
 				AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-				NULL, NULL, bo);
+				NULL, NULL, 0, bo);
 	if (err) {
 		dev_err(adev->dev, "(%d) Firmware buffer allocate failed\n", err);
 		goto failed;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
index 2ca09f111f08..e19928dae8e3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
@@ -588,6 +588,10 @@ static int amdgpu_uvd_cs_msg_decode(struct amdgpu_device *adev, uint32_t *msg,
 		}
 		break;
 
+	case 8: /* MJPEG */
+		min_dpb_size = 0;
+		break;
+
 	case 16: /* H265 */
 		image_size = (ALIGN(width, 16) * ALIGN(height, 16) * 3) / 2;
 		image_size = ALIGN(image_size, 256);
@@ -1051,7 +1055,7 @@ int amdgpu_uvd_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
 			     AMDGPU_GEM_DOMAIN_VRAM,
 			     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
 			     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-			     NULL, NULL, &bo);
+			     NULL, NULL, 0, &bo);
 	if (r)
 		return r;
 
@@ -1101,7 +1105,7 @@ int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 			     AMDGPU_GEM_DOMAIN_VRAM,
 			     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
 			     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-			     NULL, NULL, &bo);
+			     NULL, NULL, 0, &bo);
 	if (r)
 		return r;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
index b692ad402252..c855366521ab 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
@@ -937,9 +937,9 @@ int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring)
 	unsigned i;
 	int r, timeout = adev->usec_timeout;
 
-	/* workaround VCE ring test slow issue for sriov*/
+	/* skip ring test for sriov*/
 	if (amdgpu_sriov_vf(adev))
-		timeout *= 10;
+		return 0;
 
 	r = amdgpu_ring_alloc(ring, 16);
 	if (r) {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
index 09190fadd228..041e0121590c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
@@ -209,9 +209,9 @@ static void amdgpu_vcn_idle_work_handler(struct work_struct *work)
 
 	if (fences == 0) {
 		if (adev->pm.dpm_enabled) {
+			/* might be used when with pg/cg
 			amdgpu_dpm_enable_uvd(adev, false);
-		} else {
-			amdgpu_asic_set_uvd_clocks(adev, 0, 0);
+			*/
 		}
 	} else {
 		schedule_delayed_work(&adev->vcn.idle_work, VCN_IDLE_TIMEOUT);
@@ -223,12 +223,10 @@ void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring)
 	struct amdgpu_device *adev = ring->adev;
 	bool set_clocks = !cancel_delayed_work_sync(&adev->vcn.idle_work);
 
-	if (set_clocks) {
-		if (adev->pm.dpm_enabled) {
-			amdgpu_dpm_enable_uvd(adev, true);
-		} else {
-			amdgpu_asic_set_uvd_clocks(adev, 53300, 40000);
-		}
+	if (set_clocks && adev->pm.dpm_enabled) {
+		/* might be used when with pg/cg
+		amdgpu_dpm_enable_uvd(adev, true);
+		*/
 	}
 }
 
@@ -361,7 +359,7 @@ static int amdgpu_vcn_dec_get_create_msg(struct amdgpu_ring *ring, uint32_t hand
 			     AMDGPU_GEM_DOMAIN_VRAM,
 			     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
 			     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-			     NULL, NULL, &bo);
+			     NULL, NULL, 0, &bo);
 	if (r)
 		return r;
 
@@ -413,7 +411,7 @@ static int amdgpu_vcn_dec_get_destroy_msg(struct amdgpu_ring *ring, uint32_t han
 			     AMDGPU_GEM_DOMAIN_VRAM,
 			     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
 			     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-			     NULL, NULL, &bo);
+			     NULL, NULL, 0, &bo);
 	if (r)
 		return r;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
index 8a081e162d13..ab05121b9272 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
@@ -46,14 +46,14 @@ int amdgpu_allocate_static_csa(struct amdgpu_device *adev)
  * address within META_DATA init package to support SRIOV gfx preemption.
  */
 
-int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm)
+int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			  struct amdgpu_bo_va **bo_va)
 {
-	int r;
-	struct amdgpu_bo_va *bo_va;
 	struct ww_acquire_ctx ticket;
 	struct list_head list;
 	struct amdgpu_bo_list_entry pd;
 	struct ttm_validate_buffer csa_tv;
+	int r;
 
 	INIT_LIST_HEAD(&list);
 	INIT_LIST_HEAD(&csa_tv.head);
@@ -69,34 +69,33 @@ int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 		return r;
 	}
 
-	bo_va = amdgpu_vm_bo_add(adev, vm, adev->virt.csa_obj);
-	if (!bo_va) {
+	*bo_va = amdgpu_vm_bo_add(adev, vm, adev->virt.csa_obj);
+	if (!*bo_va) {
 		ttm_eu_backoff_reservation(&ticket, &list);
 		DRM_ERROR("failed to create bo_va for static CSA\n");
 		return -ENOMEM;
 	}
 
-	r = amdgpu_vm_alloc_pts(adev, bo_va->vm, AMDGPU_CSA_VADDR,
-				   AMDGPU_CSA_SIZE);
+	r = amdgpu_vm_alloc_pts(adev, (*bo_va)->base.vm, AMDGPU_CSA_VADDR,
+				AMDGPU_CSA_SIZE);
 	if (r) {
 		DRM_ERROR("failed to allocate pts for static CSA, err=%d\n", r);
-		amdgpu_vm_bo_rmv(adev, bo_va);
+		amdgpu_vm_bo_rmv(adev, *bo_va);
 		ttm_eu_backoff_reservation(&ticket, &list);
 		return r;
 	}
 
-	r = amdgpu_vm_bo_map(adev, bo_va, AMDGPU_CSA_VADDR, 0,AMDGPU_CSA_SIZE,
-						AMDGPU_PTE_READABLE | AMDGPU_PTE_WRITEABLE |
-						AMDGPU_PTE_EXECUTABLE);
+	r = amdgpu_vm_bo_map(adev, *bo_va, AMDGPU_CSA_VADDR, 0, AMDGPU_CSA_SIZE,
+			     AMDGPU_PTE_READABLE | AMDGPU_PTE_WRITEABLE |
+			     AMDGPU_PTE_EXECUTABLE);
 
 	if (r) {
 		DRM_ERROR("failed to do bo_map on static CSA, err=%d\n", r);
-		amdgpu_vm_bo_rmv(adev, bo_va);
+		amdgpu_vm_bo_rmv(adev, *bo_va);
 		ttm_eu_backoff_reservation(&ticket, &list);
 		return r;
 	}
 
-	vm->csa_bo_va = bo_va;
 	ttm_eu_backoff_reservation(&ticket, &list);
 	return 0;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
index e5b1baf387c1..afcfb8bcfb65 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
@@ -90,7 +90,8 @@ static inline bool is_virtual_machine(void)
 
 struct amdgpu_vm;
 int amdgpu_allocate_static_csa(struct amdgpu_device *adev);
-int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm);
+int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			  struct amdgpu_bo_va **bo_va);
 void amdgpu_virt_init_setting(struct amdgpu_device *adev);
 uint32_t amdgpu_virt_kiq_rreg(struct amdgpu_device *adev, uint32_t reg);
 void amdgpu_virt_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 250c8e80e646..6b1343e5541d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -159,11 +159,20 @@ void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
  */
 static int amdgpu_vm_validate_level(struct amdgpu_vm_pt *parent,
 				    int (*validate)(void *, struct amdgpu_bo *),
-				    void *param, bool use_cpu_for_update)
+				    void *param, bool use_cpu_for_update,
+				    struct ttm_bo_global *glob)
 {
 	unsigned i;
 	int r;
 
+	if (parent->bo->shadow) {
+		struct amdgpu_bo *shadow = parent->bo->shadow;
+
+		r = amdgpu_ttm_bind(&shadow->tbo, &shadow->tbo.mem);
+		if (r)
+			return r;
+	}
+
 	if (use_cpu_for_update) {
 		r = amdgpu_bo_kmap(parent->bo, NULL);
 		if (r)
@@ -183,12 +192,18 @@ static int amdgpu_vm_validate_level(struct amdgpu_vm_pt *parent,
 		if (r)
 			return r;
 
+		spin_lock(&glob->lru_lock);
+		ttm_bo_move_to_lru_tail(&entry->bo->tbo);
+		if (entry->bo->shadow)
+			ttm_bo_move_to_lru_tail(&entry->bo->shadow->tbo);
+		spin_unlock(&glob->lru_lock);
+
 		/*
 		 * Recurse into the sub directory. This is harmless because we
 		 * have only a maximum of 5 layers.
 		 */
 		r = amdgpu_vm_validate_level(entry, validate, param,
-					     use_cpu_for_update);
+					     use_cpu_for_update, glob);
 		if (r)
 			return r;
 	}
@@ -220,54 +235,11 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 		return 0;
 
 	return amdgpu_vm_validate_level(&vm->root, validate, param,
-					vm->use_cpu_for_update);
+					vm->use_cpu_for_update,
+					adev->mman.bdev.glob);
 }
 
 /**
- * amdgpu_vm_move_level_in_lru - move one level of PT BOs to the LRU tail
- *
- * @adev: amdgpu device instance
- * @vm: vm providing the BOs
- *
- * Move the PT BOs to the tail of the LRU.
- */
-static void amdgpu_vm_move_level_in_lru(struct amdgpu_vm_pt *parent)
-{
-	unsigned i;
-
-	if (!parent->entries)
-		return;
-
-	for (i = 0; i <= parent->last_entry_used; ++i) {
-		struct amdgpu_vm_pt *entry = &parent->entries[i];
-
-		if (!entry->bo)
-			continue;
-
-		ttm_bo_move_to_lru_tail(&entry->bo->tbo);
-		amdgpu_vm_move_level_in_lru(entry);
-	}
-}
-
-/**
- * amdgpu_vm_move_pt_bos_in_lru - move the PT BOs to the LRU tail
- *
- * @adev: amdgpu device instance
- * @vm: vm providing the BOs
- *
- * Move the PT BOs to the tail of the LRU.
- */
-void amdgpu_vm_move_pt_bos_in_lru(struct amdgpu_device *adev,
-				  struct amdgpu_vm *vm)
-{
-	struct ttm_bo_global *glob = adev->mman.bdev.glob;
-
-	spin_lock(&glob->lru_lock);
-	amdgpu_vm_move_level_in_lru(&vm->root);
-	spin_unlock(&glob->lru_lock);
-}
-
- /**
  * amdgpu_vm_alloc_levels - allocate the PD/PT levels
  *
  * @adev: amdgpu_device pointer
@@ -288,6 +260,7 @@ static int amdgpu_vm_alloc_levels(struct amdgpu_device *adev,
 	unsigned pt_idx, from, to;
 	int r;
 	u64 flags;
+	uint64_t init_value = 0;
 
 	if (!parent->entries) {
 		unsigned num_entries = amdgpu_vm_num_entries(adev, level);
@@ -321,6 +294,12 @@ static int amdgpu_vm_alloc_levels(struct amdgpu_device *adev,
 		flags |= (AMDGPU_GEM_CREATE_NO_CPU_ACCESS |
 				AMDGPU_GEM_CREATE_SHADOW);
 
+	if (vm->pte_support_ats) {
+		init_value = AMDGPU_PTE_SYSTEM;
+		if (level != adev->vm_manager.num_level - 1)
+			init_value |= AMDGPU_PDE_PTE;
+	}
+
 	/* walk over the address space and allocate the page tables */
 	for (pt_idx = from; pt_idx <= to; ++pt_idx) {
 		struct reservation_object *resv = vm->root.bo->tbo.resv;
@@ -333,7 +312,7 @@ static int amdgpu_vm_alloc_levels(struct amdgpu_device *adev,
 					     AMDGPU_GPU_PAGE_SIZE, true,
 					     AMDGPU_GEM_DOMAIN_VRAM,
 					     flags,
-					     NULL, resv, &pt);
+					     NULL, resv, init_value, &pt);
 			if (r)
 				return r;
 
@@ -352,7 +331,6 @@ static int amdgpu_vm_alloc_levels(struct amdgpu_device *adev,
 
 			entry->bo = pt;
 			entry->addr = 0;
-			entry->huge_page = false;
 		}
 
 		if (level < adev->vm_manager.num_level) {
@@ -892,8 +870,8 @@ struct amdgpu_bo_va *amdgpu_vm_bo_find(struct amdgpu_vm *vm,
 {
 	struct amdgpu_bo_va *bo_va;
 
-	list_for_each_entry(bo_va, &bo->va, bo_list) {
-		if (bo_va->vm == vm) {
+	list_for_each_entry(bo_va, &bo->va, base.bo_list) {
+		if (bo_va->base.vm == vm) {
 			return bo_va;
 		}
 	}
@@ -1060,18 +1038,13 @@ static int amdgpu_vm_update_level(struct amdgpu_device *adev,
 	shadow = parent->bo->shadow;
 
 	if (vm->use_cpu_for_update) {
-		pd_addr = (unsigned long)parent->bo->kptr;
+		pd_addr = (unsigned long)amdgpu_bo_kptr(parent->bo);
 		r = amdgpu_vm_wait_pd(adev, vm, AMDGPU_FENCE_OWNER_VM);
 		if (unlikely(r))
 			return r;
 
 		params.func = amdgpu_vm_cpu_set_ptes;
 	} else {
-		if (shadow) {
-			r = amdgpu_ttm_bind(&shadow->tbo, &shadow->tbo.mem);
-			if (r)
-				return r;
-		}
 		ring = container_of(vm->entity.sched, struct amdgpu_ring,
 				    sched);
 
@@ -1107,22 +1080,14 @@ static int amdgpu_vm_update_level(struct amdgpu_device *adev,
 		if (bo == NULL)
 			continue;
 
-		if (bo->shadow) {
-			struct amdgpu_bo *pt_shadow = bo->shadow;
-
-			r = amdgpu_ttm_bind(&pt_shadow->tbo,
-					    &pt_shadow->tbo.mem);
-			if (r)
-				return r;
-		}
-
 		pt = amdgpu_bo_gpu_offset(bo);
 		pt = amdgpu_gart_get_vm_pde(adev, pt);
-		if (parent->entries[pt_idx].addr == pt ||
-		    parent->entries[pt_idx].huge_page)
+		/* Don't update huge pages here */
+		if ((parent->entries[pt_idx].addr & AMDGPU_PDE_PTE) ||
+		    parent->entries[pt_idx].addr == (pt | AMDGPU_PTE_VALID))
 			continue;
 
-		parent->entries[pt_idx].addr = pt;
+		parent->entries[pt_idx].addr = pt | AMDGPU_PTE_VALID;
 
 		pde = pd_addr + pt_idx * 8;
 		if (((last_pde + 8 * count) != pde) ||
@@ -1300,15 +1265,14 @@ void amdgpu_vm_get_entry(struct amdgpu_pte_update_params *p, uint64_t addr,
  *
  * Check if we can update the PD with a huge page.
  */
-static int amdgpu_vm_handle_huge_pages(struct amdgpu_pte_update_params *p,
-				       struct amdgpu_vm_pt *entry,
-				       struct amdgpu_vm_pt *parent,
-				       unsigned nptes, uint64_t dst,
-				       uint64_t flags)
+static void amdgpu_vm_handle_huge_pages(struct amdgpu_pte_update_params *p,
+					struct amdgpu_vm_pt *entry,
+					struct amdgpu_vm_pt *parent,
+					unsigned nptes, uint64_t dst,
+					uint64_t flags)
 {
 	bool use_cpu_update = (p->func == amdgpu_vm_cpu_set_ptes);
 	uint64_t pd_addr, pde;
-	int r;
 
 	/* In the case of a mixed PT the PDE must point to it*/
 	if (p->adev->asic_type < CHIP_VEGA10 ||
@@ -1320,21 +1284,17 @@ static int amdgpu_vm_handle_huge_pages(struct amdgpu_pte_update_params *p,
 		dst = amdgpu_gart_get_vm_pde(p->adev, dst);
 		flags = AMDGPU_PTE_VALID;
 	} else {
+		/* Set the huge page flag to stop scanning at this PDE */
 		flags |= AMDGPU_PDE_PTE;
 	}
 
-	if (entry->addr == dst &&
-	    entry->huge_page == !!(flags & AMDGPU_PDE_PTE))
-		return 0;
+	if (entry->addr == (dst | flags))
+		return;
 
-	entry->addr = dst;
-	entry->huge_page = !!(flags & AMDGPU_PDE_PTE);
+	entry->addr = (dst | flags);
 
 	if (use_cpu_update) {
-		r = amdgpu_bo_kmap(parent->bo, (void *)&pd_addr);
-		if (r)
-			return r;
-
+		pd_addr = (unsigned long)amdgpu_bo_kptr(parent->bo);
 		pde = pd_addr + (entry - parent->entries) * 8;
 		amdgpu_vm_cpu_set_ptes(p, pde, dst, 1, 0, flags);
 	} else {
@@ -1347,8 +1307,6 @@ static int amdgpu_vm_handle_huge_pages(struct amdgpu_pte_update_params *p,
 		pde = pd_addr + (entry - parent->entries) * 8;
 		amdgpu_vm_do_set_ptes(p, pde, dst, 1, 0, flags);
 	}
-
-	return 0;
 }
 
 /**
@@ -1375,7 +1333,6 @@ static int amdgpu_vm_update_ptes(struct amdgpu_pte_update_params *params,
 	struct amdgpu_bo *pt;
 	unsigned nptes;
 	bool use_cpu_update = (params->func == amdgpu_vm_cpu_set_ptes);
-	int r;
 
 	/* walk over the address space and update the page tables */
 	for (addr = start; addr < end; addr += nptes,
@@ -1391,17 +1348,15 @@ static int amdgpu_vm_update_ptes(struct amdgpu_pte_update_params *params,
 		else
 			nptes = AMDGPU_VM_PTE_COUNT(adev) - (addr & mask);
 
-		r = amdgpu_vm_handle_huge_pages(params, entry, parent,
-						nptes, dst, flags);
-		if (r)
-			return r;
-
-		if (entry->huge_page)
+		amdgpu_vm_handle_huge_pages(params, entry, parent,
+					    nptes, dst, flags);
+		/* We don't need to update PTEs for huge pages */
+		if (entry->addr & AMDGPU_PDE_PTE)
 			continue;
 
 		pt = entry->bo;
 		if (use_cpu_update) {
-			pe_start = (unsigned long)pt->kptr;
+			pe_start = (unsigned long)amdgpu_bo_kptr(pt);
 		} else {
 			if (pt->shadow) {
 				pe_start = amdgpu_bo_gpu_offset(pt->shadow);
@@ -1455,9 +1410,7 @@ static int amdgpu_vm_frag_ptes(struct amdgpu_pte_update_params	*params,
 	 * Userspace can support this by aligning virtual base address and
 	 * allocation size to the fragment size.
 	 */
-
-	/* SI and newer are optimized for 64KB */
-	unsigned pages_per_frag = AMDGPU_LOG2_PAGES_PER_FRAG(params->adev);
+	unsigned pages_per_frag = params->adev->vm_manager.fragment_size;
 	uint64_t frag_flags = AMDGPU_PTE_FRAG(pages_per_frag);
 	uint64_t frag_align = 1 << pages_per_frag;
 
@@ -1771,7 +1724,8 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 			struct amdgpu_bo_va *bo_va,
 			bool clear)
 {
-	struct amdgpu_vm *vm = bo_va->vm;
+	struct amdgpu_bo *bo = bo_va->base.bo;
+	struct amdgpu_vm *vm = bo_va->base.vm;
 	struct amdgpu_bo_va_mapping *mapping;
 	dma_addr_t *pages_addr = NULL;
 	uint64_t gtt_flags, flags;
@@ -1780,27 +1734,27 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 	struct dma_fence *exclusive;
 	int r;
 
-	if (clear || !bo_va->bo) {
+	if (clear || !bo_va->base.bo) {
 		mem = NULL;
 		nodes = NULL;
 		exclusive = NULL;
 	} else {
 		struct ttm_dma_tt *ttm;
 
-		mem = &bo_va->bo->tbo.mem;
+		mem = &bo_va->base.bo->tbo.mem;
 		nodes = mem->mm_node;
 		if (mem->mem_type == TTM_PL_TT) {
-			ttm = container_of(bo_va->bo->tbo.ttm, struct
-					   ttm_dma_tt, ttm);
+			ttm = container_of(bo_va->base.bo->tbo.ttm,
+					   struct ttm_dma_tt, ttm);
 			pages_addr = ttm->dma_address;
 		}
-		exclusive = reservation_object_get_excl(bo_va->bo->tbo.resv);
+		exclusive = reservation_object_get_excl(bo->tbo.resv);
 	}
 
-	if (bo_va->bo) {
-		flags = amdgpu_ttm_tt_pte_flags(adev, bo_va->bo->tbo.ttm, mem);
-		gtt_flags = (amdgpu_ttm_is_bound(bo_va->bo->tbo.ttm) &&
-			adev == amdgpu_ttm_adev(bo_va->bo->tbo.bdev)) ?
+	if (bo) {
+		flags = amdgpu_ttm_tt_pte_flags(adev, bo->tbo.ttm, mem);
+		gtt_flags = (amdgpu_ttm_is_bound(bo->tbo.ttm) &&
+			adev == amdgpu_ttm_adev(bo->tbo.bdev)) ?
 			flags : 0;
 	} else {
 		flags = 0x0;
@@ -1808,7 +1762,7 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 	}
 
 	spin_lock(&vm->status_lock);
-	if (!list_empty(&bo_va->vm_status))
+	if (!list_empty(&bo_va->base.vm_status))
 		list_splice_init(&bo_va->valids, &bo_va->invalids);
 	spin_unlock(&vm->status_lock);
 
@@ -1831,9 +1785,9 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 
 	spin_lock(&vm->status_lock);
 	list_splice_init(&bo_va->invalids, &bo_va->valids);
-	list_del_init(&bo_va->vm_status);
+	list_del_init(&bo_va->base.vm_status);
 	if (clear)
-		list_add(&bo_va->vm_status, &vm->cleared);
+		list_add(&bo_va->base.vm_status, &vm->cleared);
 	spin_unlock(&vm->status_lock);
 
 	if (vm->use_cpu_for_update) {
@@ -1995,15 +1949,19 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
 	struct amdgpu_bo_va_mapping *mapping;
 	struct dma_fence *f = NULL;
 	int r;
+	uint64_t init_pte_value = 0;
 
 	while (!list_empty(&vm->freed)) {
 		mapping = list_first_entry(&vm->freed,
 			struct amdgpu_bo_va_mapping, list);
 		list_del(&mapping->list);
 
+		if (vm->pte_support_ats)
+			init_pte_value = AMDGPU_PTE_SYSTEM;
+
 		r = amdgpu_vm_bo_update_mapping(adev, NULL, 0, NULL, vm,
 						mapping->start, mapping->last,
-						0, 0, &f);
+						init_pte_value, 0, &f);
 		amdgpu_vm_free_mapping(adev, vm, mapping, f);
 		if (r) {
 			dma_fence_put(f);
@@ -2023,26 +1981,26 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
 }
 
 /**
- * amdgpu_vm_clear_invalids - clear invalidated BOs in the PT
+ * amdgpu_vm_clear_moved - clear moved BOs in the PT
  *
  * @adev: amdgpu_device pointer
  * @vm: requested vm
  *
- * Make sure all invalidated BOs are cleared in the PT.
+ * Make sure all moved BOs are cleared in the PT.
  * Returns 0 for success.
  *
  * PTs have to be reserved and mutex must be locked!
  */
-int amdgpu_vm_clear_invalids(struct amdgpu_device *adev,
-			     struct amdgpu_vm *vm, struct amdgpu_sync *sync)
+int amdgpu_vm_clear_moved(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			    struct amdgpu_sync *sync)
 {
 	struct amdgpu_bo_va *bo_va = NULL;
 	int r = 0;
 
 	spin_lock(&vm->status_lock);
-	while (!list_empty(&vm->invalidated)) {
-		bo_va = list_first_entry(&vm->invalidated,
-			struct amdgpu_bo_va, vm_status);
+	while (!list_empty(&vm->moved)) {
+		bo_va = list_first_entry(&vm->moved,
+			struct amdgpu_bo_va, base.vm_status);
 		spin_unlock(&vm->status_lock);
 
 		r = amdgpu_vm_bo_update(adev, bo_va, true);
@@ -2082,16 +2040,17 @@ struct amdgpu_bo_va *amdgpu_vm_bo_add(struct amdgpu_device *adev,
 	if (bo_va == NULL) {
 		return NULL;
 	}
-	bo_va->vm = vm;
-	bo_va->bo = bo;
+	bo_va->base.vm = vm;
+	bo_va->base.bo = bo;
+	INIT_LIST_HEAD(&bo_va->base.bo_list);
+	INIT_LIST_HEAD(&bo_va->base.vm_status);
+
 	bo_va->ref_count = 1;
-	INIT_LIST_HEAD(&bo_va->bo_list);
 	INIT_LIST_HEAD(&bo_va->valids);
 	INIT_LIST_HEAD(&bo_va->invalids);
-	INIT_LIST_HEAD(&bo_va->vm_status);
 
 	if (bo)
-		list_add_tail(&bo_va->bo_list, &bo->va);
+		list_add_tail(&bo_va->base.bo_list, &bo->va);
 
 	return bo_va;
 }
@@ -2116,7 +2075,8 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 		     uint64_t size, uint64_t flags)
 {
 	struct amdgpu_bo_va_mapping *mapping, *tmp;
-	struct amdgpu_vm *vm = bo_va->vm;
+	struct amdgpu_bo *bo = bo_va->base.bo;
+	struct amdgpu_vm *vm = bo_va->base.vm;
 	uint64_t eaddr;
 
 	/* validate the parameters */
@@ -2127,7 +2087,7 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 	/* make sure object fit at this offset */
 	eaddr = saddr + size - 1;
 	if (saddr >= eaddr ||
-	    (bo_va->bo && offset + size > amdgpu_bo_size(bo_va->bo)))
+	    (bo && offset + size > amdgpu_bo_size(bo)))
 		return -EINVAL;
 
 	saddr /= AMDGPU_GPU_PAGE_SIZE;
@@ -2137,7 +2097,7 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 	if (tmp) {
 		/* bo and tmp overlap, invalid addr */
 		dev_err(adev->dev, "bo %p va 0x%010Lx-0x%010Lx conflict with "
-			"0x%010Lx-0x%010Lx\n", bo_va->bo, saddr, eaddr,
+			"0x%010Lx-0x%010Lx\n", bo, saddr, eaddr,
 			tmp->start, tmp->last + 1);
 		return -EINVAL;
 	}
@@ -2182,7 +2142,8 @@ int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev,
 			     uint64_t size, uint64_t flags)
 {
 	struct amdgpu_bo_va_mapping *mapping;
-	struct amdgpu_vm *vm = bo_va->vm;
+	struct amdgpu_bo *bo = bo_va->base.bo;
+	struct amdgpu_vm *vm = bo_va->base.vm;
 	uint64_t eaddr;
 	int r;
 
@@ -2194,7 +2155,7 @@ int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev,
 	/* make sure object fit at this offset */
 	eaddr = saddr + size - 1;
 	if (saddr >= eaddr ||
-	    (bo_va->bo && offset + size > amdgpu_bo_size(bo_va->bo)))
+	    (bo && offset + size > amdgpu_bo_size(bo)))
 		return -EINVAL;
 
 	/* Allocate all the needed memory */
@@ -2202,7 +2163,7 @@ int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev,
 	if (!mapping)
 		return -ENOMEM;
 
-	r = amdgpu_vm_bo_clear_mappings(adev, bo_va->vm, saddr, size);
+	r = amdgpu_vm_bo_clear_mappings(adev, bo_va->base.vm, saddr, size);
 	if (r) {
 		kfree(mapping);
 		return r;
@@ -2242,7 +2203,7 @@ int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
 		       uint64_t saddr)
 {
 	struct amdgpu_bo_va_mapping *mapping;
-	struct amdgpu_vm *vm = bo_va->vm;
+	struct amdgpu_vm *vm = bo_va->base.vm;
 	bool valid = true;
 
 	saddr /= AMDGPU_GPU_PAGE_SIZE;
@@ -2390,12 +2351,12 @@ void amdgpu_vm_bo_rmv(struct amdgpu_device *adev,
 		      struct amdgpu_bo_va *bo_va)
 {
 	struct amdgpu_bo_va_mapping *mapping, *next;
-	struct amdgpu_vm *vm = bo_va->vm;
+	struct amdgpu_vm *vm = bo_va->base.vm;
 
-	list_del(&bo_va->bo_list);
+	list_del(&bo_va->base.bo_list);
 
 	spin_lock(&vm->status_lock);
-	list_del(&bo_va->vm_status);
+	list_del(&bo_va->base.vm_status);
 	spin_unlock(&vm->status_lock);
 
 	list_for_each_entry_safe(mapping, next, &bo_va->valids, list) {
@@ -2427,13 +2388,14 @@ void amdgpu_vm_bo_rmv(struct amdgpu_device *adev,
 void amdgpu_vm_bo_invalidate(struct amdgpu_device *adev,
 			     struct amdgpu_bo *bo)
 {
-	struct amdgpu_bo_va *bo_va;
+	struct amdgpu_vm_bo_base *bo_base;
 
-	list_for_each_entry(bo_va, &bo->va, bo_list) {
-		spin_lock(&bo_va->vm->status_lock);
-		if (list_empty(&bo_va->vm_status))
-			list_add(&bo_va->vm_status, &bo_va->vm->invalidated);
-		spin_unlock(&bo_va->vm->status_lock);
+	list_for_each_entry(bo_base, &bo->va, bo_list) {
+		spin_lock(&bo_base->vm->status_lock);
+		if (list_empty(&bo_base->vm_status))
+			list_add(&bo_base->vm_status,
+				 &bo_base->vm->moved);
+		spin_unlock(&bo_base->vm->status_lock);
 	}
 }
 
@@ -2451,12 +2413,26 @@ static uint32_t amdgpu_vm_get_block_size(uint64_t vm_size)
 }
 
 /**
- * amdgpu_vm_adjust_size - adjust vm size and block size
+ * amdgpu_vm_set_fragment_size - adjust fragment size in PTE
+ *
+ * @adev: amdgpu_device pointer
+ * @fragment_size_default: the default fragment size if it's set auto
+ */
+void amdgpu_vm_set_fragment_size(struct amdgpu_device *adev, uint32_t fragment_size_default)
+{
+	if (amdgpu_vm_fragment_size == -1)
+		adev->vm_manager.fragment_size = fragment_size_default;
+	else
+		adev->vm_manager.fragment_size = amdgpu_vm_fragment_size;
+}
+
+/**
+ * amdgpu_vm_adjust_size - adjust vm size, block size and fragment size
  *
  * @adev: amdgpu_device pointer
  * @vm_size: the default vm size if it's set auto
  */
-void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint64_t vm_size)
+void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint64_t vm_size, uint32_t fragment_size_default)
 {
 	/* adjust vm size firstly */
 	if (amdgpu_vm_size == -1)
@@ -2471,8 +2447,11 @@ void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint64_t vm_size)
 	else
 		adev->vm_manager.block_size = amdgpu_vm_block_size;
 
-	DRM_INFO("vm size is %llu GB, block size is %u-bit\n",
-		adev->vm_manager.vm_size, adev->vm_manager.block_size);
+	amdgpu_vm_set_fragment_size(adev, fragment_size_default);
+
+	DRM_INFO("vm size is %llu GB, block size is %u-bit, fragment size is %u-bit\n",
+		adev->vm_manager.vm_size, adev->vm_manager.block_size,
+		adev->vm_manager.fragment_size);
 }
 
 /**
@@ -2494,13 +2473,14 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	struct amd_sched_rq *rq;
 	int r, i;
 	u64 flags;
+	uint64_t init_pde_value = 0;
 
 	vm->va = RB_ROOT;
 	vm->client_id = atomic64_inc_return(&adev->vm_manager.client_counter);
 	for (i = 0; i < AMDGPU_MAX_VMHUBS; i++)
 		vm->reserved_vmid[i] = NULL;
 	spin_lock_init(&vm->status_lock);
-	INIT_LIST_HEAD(&vm->invalidated);
+	INIT_LIST_HEAD(&vm->moved);
 	INIT_LIST_HEAD(&vm->cleared);
 	INIT_LIST_HEAD(&vm->freed);
 
@@ -2515,10 +2495,17 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	if (r)
 		return r;
 
-	if (vm_context == AMDGPU_VM_CONTEXT_COMPUTE)
+	vm->pte_support_ats = false;
+
+	if (vm_context == AMDGPU_VM_CONTEXT_COMPUTE) {
 		vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
 						AMDGPU_VM_USE_CPU_FOR_COMPUTE);
-	else
+
+		if (adev->asic_type == CHIP_RAVEN) {
+			vm->pte_support_ats = true;
+			init_pde_value = AMDGPU_PTE_SYSTEM | AMDGPU_PDE_PTE;
+		}
+	} else
 		vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
 						AMDGPU_VM_USE_CPU_FOR_GFX);
 	DRM_DEBUG_DRIVER("VM update mode is %s\n",
@@ -2538,7 +2525,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	r = amdgpu_bo_create(adev, amdgpu_vm_bo_size(adev, 0), align, true,
 			     AMDGPU_GEM_DOMAIN_VRAM,
 			     flags,
-			     NULL, NULL, &vm->root.bo);
+			     NULL, NULL, init_pde_value, &vm->root.bo);
 	if (r)
 		goto error_free_sched_entity;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index 34d9174ebff2..ba6691b58ee7 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -50,11 +50,6 @@ struct amdgpu_bo_list_entry;
 /* PTBs (Page Table Blocks) need to be aligned to 32K */
 #define AMDGPU_VM_PTB_ALIGN_SIZE   32768
 
-/* LOG2 number of continuous pages for the fragment field */
-#define AMDGPU_LOG2_PAGES_PER_FRAG(adev) \
-	((adev)->asic_type < CHIP_VEGA10 ? 4 : \
-	 (adev)->vm_manager.block_size)
-
 #define AMDGPU_PTE_VALID	(1ULL << 0)
 #define AMDGPU_PTE_SYSTEM	(1ULL << 1)
 #define AMDGPU_PTE_SNOOPED	(1ULL << 2)
@@ -99,11 +94,22 @@ struct amdgpu_bo_list_entry;
 #define AMDGPU_VM_USE_CPU_FOR_GFX (1 << 0)
 #define AMDGPU_VM_USE_CPU_FOR_COMPUTE (1 << 1)
 
+/* base structure for tracking BO usage in a VM */
+struct amdgpu_vm_bo_base {
+	/* constant after initialization */
+	struct amdgpu_vm		*vm;
+	struct amdgpu_bo		*bo;
+
+	/* protected by bo being reserved */
+	struct list_head		bo_list;
+
+	/* protected by spinlock */
+	struct list_head		vm_status;
+};
 
 struct amdgpu_vm_pt {
 	struct amdgpu_bo	*bo;
 	uint64_t		addr;
-	bool			huge_page;
 
 	/* array of page tables, one for each directory entry */
 	struct amdgpu_vm_pt	*entries;
@@ -118,7 +124,7 @@ struct amdgpu_vm {
 	spinlock_t		status_lock;
 
 	/* BOs moved, but not yet updated in the PT */
-	struct list_head	invalidated;
+	struct list_head	moved;
 
 	/* BOs cleared in the PT because of a move */
 	struct list_head	cleared;
@@ -141,11 +147,12 @@ struct amdgpu_vm {
 	u64                     client_id;
 	/* dedicated to vm */
 	struct amdgpu_vm_id	*reserved_vmid[AMDGPU_MAX_VMHUBS];
-	/* each VM will map on CSA */
-	struct amdgpu_bo_va *csa_bo_va;
 
 	/* Flag to indicate if VM tables are updated by CPU or GPU (SDMA) */
 	bool                    use_cpu_for_update;
+
+	/* Flag to indicate ATS support from PTE for GFX9 */
+	bool			pte_support_ats;
 };
 
 struct amdgpu_vm_id {
@@ -188,6 +195,7 @@ struct amdgpu_vm_manager {
 	uint32_t				num_level;
 	uint64_t				vm_size;
 	uint32_t				block_size;
+	uint32_t				fragment_size;
 	/* vram base address for page table entry  */
 	u64					vram_base_offset;
 	/* vm pte handling */
@@ -220,8 +228,6 @@ void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
 int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 			      int (*callback)(void *p, struct amdgpu_bo *bo),
 			      void *param);
-void amdgpu_vm_move_pt_bos_in_lru(struct amdgpu_device *adev,
-				  struct amdgpu_vm *vm);
 int amdgpu_vm_alloc_pts(struct amdgpu_device *adev,
 			struct amdgpu_vm *vm,
 			uint64_t saddr, uint64_t size);
@@ -237,8 +243,8 @@ int amdgpu_vm_update_directories(struct amdgpu_device *adev,
 int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
 			  struct amdgpu_vm *vm,
 			  struct dma_fence **fence);
-int amdgpu_vm_clear_invalids(struct amdgpu_device *adev, struct amdgpu_vm *vm,
-			     struct amdgpu_sync *sync);
+int amdgpu_vm_clear_moved(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+			  struct amdgpu_sync *sync);
 int amdgpu_vm_bo_update(struct amdgpu_device *adev,
 			struct amdgpu_bo_va *bo_va,
 			bool clear);
@@ -265,7 +271,10 @@ int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev,
 				uint64_t saddr, uint64_t size);
 void amdgpu_vm_bo_rmv(struct amdgpu_device *adev,
 		      struct amdgpu_bo_va *bo_va);
-void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint64_t vm_size);
+void amdgpu_vm_set_fragment_size(struct amdgpu_device *adev,
+				uint32_t fragment_size_default);
+void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint64_t vm_size,
+				uint32_t fragment_size_default);
 int amdgpu_vm_ioctl(struct drm_device *dev, void *data, struct drm_file *filp);
 bool amdgpu_vm_need_pipeline_sync(struct amdgpu_ring *ring,
 				  struct amdgpu_job *job);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
index a2c59a08b2bd..26e900627971 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
@@ -28,6 +28,8 @@
 struct amdgpu_vram_mgr {
 	struct drm_mm mm;
 	spinlock_t lock;
+	atomic64_t usage;
+	atomic64_t vis_usage;
 };
 
 /**
@@ -79,6 +81,27 @@ static int amdgpu_vram_mgr_fini(struct ttm_mem_type_manager *man)
 }
 
 /**
+ * amdgpu_vram_mgr_vis_size - Calculate visible node size
+ *
+ * @adev: amdgpu device structure
+ * @node: MM node structure
+ *
+ * Calculate how many bytes of the MM node are inside visible VRAM
+ */
+static u64 amdgpu_vram_mgr_vis_size(struct amdgpu_device *adev,
+				    struct drm_mm_node *node)
+{
+	uint64_t start = node->start << PAGE_SHIFT;
+	uint64_t end = (node->size + node->start) << PAGE_SHIFT;
+
+	if (start >= adev->mc.visible_vram_size)
+		return 0;
+
+	return (end > adev->mc.visible_vram_size ?
+		adev->mc.visible_vram_size : end) - start;
+}
+
+/**
  * amdgpu_vram_mgr_new - allocate new ranges
  *
  * @man: TTM memory type manager
@@ -93,11 +116,13 @@ static int amdgpu_vram_mgr_new(struct ttm_mem_type_manager *man,
 			       const struct ttm_place *place,
 			       struct ttm_mem_reg *mem)
 {
+	struct amdgpu_device *adev = amdgpu_ttm_adev(man->bdev);
 	struct amdgpu_vram_mgr *mgr = man->priv;
 	struct drm_mm *mm = &mgr->mm;
 	struct drm_mm_node *nodes;
 	enum drm_mm_insert_mode mode;
 	unsigned long lpfn, num_nodes, pages_per_node, pages_left;
+	uint64_t usage = 0, vis_usage = 0;
 	unsigned i;
 	int r;
 
@@ -142,6 +167,9 @@ static int amdgpu_vram_mgr_new(struct ttm_mem_type_manager *man,
 		if (unlikely(r))
 			goto error;
 
+		usage += nodes[i].size << PAGE_SHIFT;
+		vis_usage += amdgpu_vram_mgr_vis_size(adev, &nodes[i]);
+
 		/* Calculate a virtual BO start address to easily check if
 		 * everything is CPU accessible.
 		 */
@@ -155,6 +183,9 @@ static int amdgpu_vram_mgr_new(struct ttm_mem_type_manager *man,
 	}
 	spin_unlock(&mgr->lock);
 
+	atomic64_add(usage, &mgr->usage);
+	atomic64_add(vis_usage, &mgr->vis_usage);
+
 	mem->mm_node = nodes;
 
 	return 0;
@@ -181,8 +212,10 @@ error:
 static void amdgpu_vram_mgr_del(struct ttm_mem_type_manager *man,
 				struct ttm_mem_reg *mem)
 {
+	struct amdgpu_device *adev = amdgpu_ttm_adev(man->bdev);
 	struct amdgpu_vram_mgr *mgr = man->priv;
 	struct drm_mm_node *nodes = mem->mm_node;
+	uint64_t usage = 0, vis_usage = 0;
 	unsigned pages = mem->num_pages;
 
 	if (!mem->mm_node)
@@ -192,31 +225,67 @@ static void amdgpu_vram_mgr_del(struct ttm_mem_type_manager *man,
 	while (pages) {
 		pages -= nodes->size;
 		drm_mm_remove_node(nodes);
+		usage += nodes->size << PAGE_SHIFT;
+		vis_usage += amdgpu_vram_mgr_vis_size(adev, nodes);
 		++nodes;
 	}
 	spin_unlock(&mgr->lock);
 
+	atomic64_sub(usage, &mgr->usage);
+	atomic64_sub(vis_usage, &mgr->vis_usage);
+
 	kfree(mem->mm_node);
 	mem->mm_node = NULL;
 }
 
 /**
+ * amdgpu_vram_mgr_usage - how many bytes are used in this domain
+ *
+ * @man: TTM memory type manager
+ *
+ * Returns how many bytes are used in this domain.
+ */
+uint64_t amdgpu_vram_mgr_usage(struct ttm_mem_type_manager *man)
+{
+	struct amdgpu_vram_mgr *mgr = man->priv;
+
+	return atomic64_read(&mgr->usage);
+}
+
+/**
+ * amdgpu_vram_mgr_vis_usage - how many bytes are used in the visible part
+ *
+ * @man: TTM memory type manager
+ *
+ * Returns how many bytes are used in the visible part of VRAM
+ */
+uint64_t amdgpu_vram_mgr_vis_usage(struct ttm_mem_type_manager *man)
+{
+	struct amdgpu_vram_mgr *mgr = man->priv;
+
+	return atomic64_read(&mgr->vis_usage);
+}
+
+/**
  * amdgpu_vram_mgr_debug - dump VRAM table
  *
  * @man: TTM memory type manager
- * @prefix: text prefix
+ * @printer: DRM printer to use
  *
  * Dump the table content using printk.
  */
 static void amdgpu_vram_mgr_debug(struct ttm_mem_type_manager *man,
-				  const char *prefix)
+				  struct drm_printer *printer)
 {
 	struct amdgpu_vram_mgr *mgr = man->priv;
-	struct drm_printer p = drm_debug_printer(prefix);
 
 	spin_lock(&mgr->lock);
-	drm_mm_print(&mgr->mm, &p);
+	drm_mm_print(&mgr->mm, printer);
 	spin_unlock(&mgr->lock);
+
+	drm_printf(printer, "man size:%llu pages, ram usage:%lluMB, vis usage:%lluMB\n",
+		   man->size, amdgpu_vram_mgr_usage(man) >> 20,
+		   amdgpu_vram_mgr_vis_usage(man) >> 20);
 }
 
 const struct ttm_mem_type_manager_func amdgpu_vram_mgr_func = {
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
index 490e84944851..4e519dc42916 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
@@ -2431,7 +2431,7 @@ static int dce_v10_0_crtc_cursor_set2(struct drm_crtc *crtc,
 	aobj = gem_to_amdgpu_bo(obj);
 	ret = amdgpu_bo_reserve(aobj, false);
 	if (ret != 0) {
-		drm_gem_object_unreference_unlocked(obj);
+		drm_gem_object_put_unlocked(obj);
 		return ret;
 	}
 
@@ -2439,7 +2439,7 @@ static int dce_v10_0_crtc_cursor_set2(struct drm_crtc *crtc,
 	amdgpu_bo_unreserve(aobj);
 	if (ret) {
 		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
-		drm_gem_object_unreference_unlocked(obj);
+		drm_gem_object_put_unlocked(obj);
 		return ret;
 	}
 
@@ -2473,7 +2473,7 @@ unpin:
 			amdgpu_bo_unpin(aobj);
 			amdgpu_bo_unreserve(aobj);
 		}
-		drm_gem_object_unreference_unlocked(amdgpu_crtc->cursor_bo);
+		drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
 	}
 
 	amdgpu_crtc->cursor_bo = obj;
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
index 921c6f772f11..11edc75edaa9 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
@@ -2506,7 +2506,7 @@ static int dce_v11_0_crtc_cursor_set2(struct drm_crtc *crtc,
 	aobj = gem_to_amdgpu_bo(obj);
 	ret = amdgpu_bo_reserve(aobj, false);
 	if (ret != 0) {
-		drm_gem_object_unreference_unlocked(obj);
+		drm_gem_object_put_unlocked(obj);
 		return ret;
 	}
 
@@ -2514,7 +2514,7 @@ static int dce_v11_0_crtc_cursor_set2(struct drm_crtc *crtc,
 	amdgpu_bo_unreserve(aobj);
 	if (ret) {
 		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
-		drm_gem_object_unreference_unlocked(obj);
+		drm_gem_object_put_unlocked(obj);
 		return ret;
 	}
 
@@ -2548,7 +2548,7 @@ unpin:
 			amdgpu_bo_unpin(aobj);
 			amdgpu_bo_unreserve(aobj);
 		}
-		drm_gem_object_unreference_unlocked(amdgpu_crtc->cursor_bo);
+		drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
 	}
 
 	amdgpu_crtc->cursor_bo = obj;
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
index bcd9521237f4..a51e35f824a1 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
@@ -42,6 +42,7 @@
 #include "dce/dce_6_0_d.h"
 #include "dce/dce_6_0_sh_mask.h"
 #include "gca/gfx_7_2_enum.h"
+#include "dce_v6_0.h"
 #include "si_enums.h"
 
 static void dce_v6_0_set_display_funcs(struct amdgpu_device *adev);
@@ -2321,7 +2322,7 @@ static int dce_v6_0_crtc_cursor_set2(struct drm_crtc *crtc,
 	aobj = gem_to_amdgpu_bo(obj);
 	ret = amdgpu_bo_reserve(aobj, false);
 	if (ret != 0) {
-		drm_gem_object_unreference_unlocked(obj);
+		drm_gem_object_put_unlocked(obj);
 		return ret;
 	}
 
@@ -2329,7 +2330,7 @@ static int dce_v6_0_crtc_cursor_set2(struct drm_crtc *crtc,
 	amdgpu_bo_unreserve(aobj);
 	if (ret) {
 		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
-		drm_gem_object_unreference_unlocked(obj);
+		drm_gem_object_put_unlocked(obj);
 		return ret;
 	}
 
@@ -2363,7 +2364,7 @@ unpin:
 			amdgpu_bo_unpin(aobj);
 			amdgpu_bo_unreserve(aobj);
 		}
-		drm_gem_object_unreference_unlocked(amdgpu_crtc->cursor_bo);
+		drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
 	}
 
 	amdgpu_crtc->cursor_bo = obj;
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
index 609438fe8584..9cf14b8b2db9 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
@@ -2335,7 +2335,7 @@ static int dce_v8_0_crtc_cursor_set2(struct drm_crtc *crtc,
 	aobj = gem_to_amdgpu_bo(obj);
 	ret = amdgpu_bo_reserve(aobj, false);
 	if (ret != 0) {
-		drm_gem_object_unreference_unlocked(obj);
+		drm_gem_object_put_unlocked(obj);
 		return ret;
 	}
 
@@ -2343,7 +2343,7 @@ static int dce_v8_0_crtc_cursor_set2(struct drm_crtc *crtc,
 	amdgpu_bo_unreserve(aobj);
 	if (ret) {
 		DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
-		drm_gem_object_unreference_unlocked(obj);
+		drm_gem_object_put_unlocked(obj);
 		return ret;
 	}
 
@@ -2377,7 +2377,7 @@ unpin:
 			amdgpu_bo_unpin(aobj);
 			amdgpu_bo_unreserve(aobj);
 		}
-		drm_gem_object_unreference_unlocked(amdgpu_crtc->cursor_bo);
+		drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
 	}
 
 	amdgpu_crtc->cursor_bo = obj;
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_virtual.c b/drivers/gpu/drm/amd/amdgpu/dce_virtual.c
index 5ed919e45351..b9ee9073cb0d 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_virtual.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_virtual.c
@@ -479,6 +479,8 @@ static int dce_virtual_hw_init(void *handle)
 #endif
 		/* no DCE */
 		break;
+	case CHIP_VEGA10:
+		break;
 	default:
 		DRM_ERROR("Virtual display unsupported ASIC type: 0x%X\n", adev->asic_type);
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
index 4ac85f47f287..d228f5a99044 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
@@ -2217,40 +2217,9 @@ static void gfx_v6_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
 
 static void gfx_v6_0_rlc_fini(struct amdgpu_device *adev)
 {
-	int r;
-
-	if (adev->gfx.rlc.save_restore_obj) {
-		r = amdgpu_bo_reserve(adev->gfx.rlc.save_restore_obj, true);
-		if (unlikely(r != 0))
-			dev_warn(adev->dev, "(%d) reserve RLC sr bo failed\n", r);
-		amdgpu_bo_unpin(adev->gfx.rlc.save_restore_obj);
-		amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
-
-		amdgpu_bo_unref(&adev->gfx.rlc.save_restore_obj);
-		adev->gfx.rlc.save_restore_obj = NULL;
-	}
-
-	if (adev->gfx.rlc.clear_state_obj) {
-		r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, true);
-		if (unlikely(r != 0))
-			dev_warn(adev->dev, "(%d) reserve RLC c bo failed\n", r);
-		amdgpu_bo_unpin(adev->gfx.rlc.clear_state_obj);
-		amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
-
-		amdgpu_bo_unref(&adev->gfx.rlc.clear_state_obj);
-		adev->gfx.rlc.clear_state_obj = NULL;
-	}
-
-	if (adev->gfx.rlc.cp_table_obj) {
-		r = amdgpu_bo_reserve(adev->gfx.rlc.cp_table_obj, true);
-		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;
-	}
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.save_restore_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, NULL, NULL);
 }
 
 static int gfx_v6_0_rlc_init(struct amdgpu_device *adev)
@@ -2273,43 +2242,23 @@ static int gfx_v6_0_rlc_init(struct amdgpu_device *adev)
 
 	if (src_ptr) {
 		/* save restore block */
-		if (adev->gfx.rlc.save_restore_obj == NULL) {
-			r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true,
-					     AMDGPU_GEM_DOMAIN_VRAM,
-					     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
-					     NULL, NULL,
-					     &adev->gfx.rlc.save_restore_obj);
-
-			if (r) {
-				dev_warn(adev->dev, "(%d) create RLC sr bo failed\n", r);
-				return r;
-			}
-		}
-
-		r = amdgpu_bo_reserve(adev->gfx.rlc.save_restore_obj, false);
-		if (unlikely(r != 0)) {
-			gfx_v6_0_rlc_fini(adev);
-			return r;
-		}
-		r = amdgpu_bo_pin(adev->gfx.rlc.save_restore_obj, AMDGPU_GEM_DOMAIN_VRAM,
-				  &adev->gfx.rlc.save_restore_gpu_addr);
+		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.save_restore_obj,
+					      &adev->gfx.rlc.save_restore_gpu_addr,
+					      (void **)&adev->gfx.rlc.sr_ptr);
 		if (r) {
-			amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
-			dev_warn(adev->dev, "(%d) pin RLC sr bo failed\n", r);
+			dev_warn(adev->dev, "(%d) create RLC sr bo failed\n",
+				 r);
 			gfx_v6_0_rlc_fini(adev);
 			return r;
 		}
 
-		r = amdgpu_bo_kmap(adev->gfx.rlc.save_restore_obj, (void **)&adev->gfx.rlc.sr_ptr);
-		if (r) {
-			dev_warn(adev->dev, "(%d) map RLC sr bo failed\n", r);
-			gfx_v6_0_rlc_fini(adev);
-			return r;
-		}
 		/* write the sr buffer */
 		dst_ptr = adev->gfx.rlc.sr_ptr;
 		for (i = 0; i < adev->gfx.rlc.reg_list_size; i++)
 			dst_ptr[i] = cpu_to_le32(src_ptr[i]);
+
 		amdgpu_bo_kunmap(adev->gfx.rlc.save_restore_obj);
 		amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
 	}
@@ -2319,39 +2268,17 @@ static int gfx_v6_0_rlc_init(struct amdgpu_device *adev)
 		adev->gfx.rlc.clear_state_size = gfx_v6_0_get_csb_size(adev);
 		dws = adev->gfx.rlc.clear_state_size + (256 / 4);
 
-		if (adev->gfx.rlc.clear_state_obj == NULL) {
-			r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true,
-					     AMDGPU_GEM_DOMAIN_VRAM,
-					     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
-					     NULL, NULL,
-					     &adev->gfx.rlc.clear_state_obj);
-
-			if (r) {
-				dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
-				gfx_v6_0_rlc_fini(adev);
-				return r;
-			}
-		}
-		r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, false);
-		if (unlikely(r != 0)) {
-			gfx_v6_0_rlc_fini(adev);
-			return r;
-		}
-		r = amdgpu_bo_pin(adev->gfx.rlc.clear_state_obj, AMDGPU_GEM_DOMAIN_VRAM,
-				  &adev->gfx.rlc.clear_state_gpu_addr);
+		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.clear_state_obj,
+					      &adev->gfx.rlc.clear_state_gpu_addr,
+					      (void **)&adev->gfx.rlc.cs_ptr);
 		if (r) {
-			amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
-			dev_warn(adev->dev, "(%d) pin RLC c bo failed\n", r);
+			dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
 			gfx_v6_0_rlc_fini(adev);
 			return r;
 		}
 
-		r = amdgpu_bo_kmap(adev->gfx.rlc.clear_state_obj, (void **)&adev->gfx.rlc.cs_ptr);
-		if (r) {
-			dev_warn(adev->dev, "(%d) map RLC c bo failed\n", r);
-			gfx_v6_0_rlc_fini(adev);
-			return r;
-		}
 		/* set up the cs buffer */
 		dst_ptr = adev->gfx.rlc.cs_ptr;
 		reg_list_mc_addr = adev->gfx.rlc.clear_state_gpu_addr + 256;
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
index 17b7c6934b0a..00868764a0dd 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
@@ -1823,7 +1823,7 @@ static void gfx_v7_0_setup_rb(struct amdgpu_device *adev)
 }
 
 /**
- * gmc_v7_0_init_compute_vmid - gart enable
+ * gfx_v7_0_init_compute_vmid - gart enable
  *
  * @adev: amdgpu_device pointer
  *
@@ -1833,7 +1833,7 @@ static void gfx_v7_0_setup_rb(struct amdgpu_device *adev)
 #define DEFAULT_SH_MEM_BASES	(0x6000)
 #define FIRST_COMPUTE_VMID	(8)
 #define LAST_COMPUTE_VMID	(16)
-static void gmc_v7_0_init_compute_vmid(struct amdgpu_device *adev)
+static void gfx_v7_0_init_compute_vmid(struct amdgpu_device *adev)
 {
 	int i;
 	uint32_t sh_mem_config;
@@ -1921,6 +1921,7 @@ static void gfx_v7_0_gpu_init(struct amdgpu_device *adev)
 				   ELEMENT_SIZE, 1);
 	sh_static_mem_cfg = REG_SET_FIELD(sh_static_mem_cfg, SH_STATIC_MEM_CONFIG,
 				   INDEX_STRIDE, 3);
+	WREG32(mmSH_STATIC_MEM_CONFIG, sh_static_mem_cfg);
 
 	mutex_lock(&adev->srbm_mutex);
 	for (i = 0; i < adev->vm_manager.id_mgr[0].num_ids; i++) {
@@ -1934,12 +1935,11 @@ static void gfx_v7_0_gpu_init(struct amdgpu_device *adev)
 		WREG32(mmSH_MEM_APE1_BASE, 1);
 		WREG32(mmSH_MEM_APE1_LIMIT, 0);
 		WREG32(mmSH_MEM_BASES, sh_mem_base);
-		WREG32(mmSH_STATIC_MEM_CONFIG, sh_static_mem_cfg);
 	}
 	cik_srbm_select(adev, 0, 0, 0, 0);
 	mutex_unlock(&adev->srbm_mutex);
 
-	gmc_v7_0_init_compute_vmid(adev);
+	gfx_v7_0_init_compute_vmid(adev);
 
 	WREG32(mmSX_DEBUG_1, 0x20);
 
@@ -2774,39 +2774,18 @@ static int gfx_v7_0_cp_compute_load_microcode(struct amdgpu_device *adev)
  */
 static void gfx_v7_0_cp_compute_fini(struct amdgpu_device *adev)
 {
-	int i, r;
+	int i;
 
 	for (i = 0; i < adev->gfx.num_compute_rings; i++) {
 		struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
 
-		if (ring->mqd_obj) {
-			r = amdgpu_bo_reserve(ring->mqd_obj, true);
-			if (unlikely(r != 0))
-				dev_warn(adev->dev, "(%d) reserve MQD bo failed\n", r);
-
-			amdgpu_bo_unpin(ring->mqd_obj);
-			amdgpu_bo_unreserve(ring->mqd_obj);
-
-			amdgpu_bo_unref(&ring->mqd_obj);
-			ring->mqd_obj = NULL;
-		}
+		amdgpu_bo_free_kernel(&ring->mqd_obj, NULL, NULL);
 	}
 }
 
 static void gfx_v7_0_mec_fini(struct amdgpu_device *adev)
 {
-	int r;
-
-	if (adev->gfx.mec.hpd_eop_obj) {
-		r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, true);
-		if (unlikely(r != 0))
-			dev_warn(adev->dev, "(%d) reserve HPD EOP bo failed\n", r);
-		amdgpu_bo_unpin(adev->gfx.mec.hpd_eop_obj);
-		amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
-
-		amdgpu_bo_unref(&adev->gfx.mec.hpd_eop_obj);
-		adev->gfx.mec.hpd_eop_obj = NULL;
-	}
+	amdgpu_bo_free_kernel(&adev->gfx.mec.hpd_eop_obj, NULL, NULL);
 }
 
 static int gfx_v7_0_mec_init(struct amdgpu_device *adev)
@@ -2823,33 +2802,14 @@ static int gfx_v7_0_mec_init(struct amdgpu_device *adev)
 	/* allocate space for ALL pipes (even the ones we don't own) */
 	mec_hpd_size = adev->gfx.mec.num_mec * adev->gfx.mec.num_pipe_per_mec
 		* GFX7_MEC_HPD_SIZE * 2;
-	if (adev->gfx.mec.hpd_eop_obj == NULL) {
-		r = amdgpu_bo_create(adev,
-				     mec_hpd_size,
-				     PAGE_SIZE, true,
-				     AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL,
-				     &adev->gfx.mec.hpd_eop_obj);
-		if (r) {
-			dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
-			return r;
-		}
-	}
 
-	r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, false);
-	if (unlikely(r != 0)) {
-		gfx_v7_0_mec_fini(adev);
-		return r;
-	}
-	r = amdgpu_bo_pin(adev->gfx.mec.hpd_eop_obj, AMDGPU_GEM_DOMAIN_GTT,
-			  &adev->gfx.mec.hpd_eop_gpu_addr);
-	if (r) {
-		dev_warn(adev->dev, "(%d) pin HDP EOP bo failed\n", r);
-		gfx_v7_0_mec_fini(adev);
-		return r;
-	}
-	r = amdgpu_bo_kmap(adev->gfx.mec.hpd_eop_obj, (void **)&hpd);
+	r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.mec.hpd_eop_obj,
+				      &adev->gfx.mec.hpd_eop_gpu_addr,
+				      (void **)&hpd);
 	if (r) {
-		dev_warn(adev->dev, "(%d) map HDP EOP bo failed\n", r);
+		dev_warn(adev->dev, "(%d) create, pin or map of HDP EOP bo failed\n", r);
 		gfx_v7_0_mec_fini(adev);
 		return r;
 	}
@@ -3108,32 +3068,12 @@ static int gfx_v7_0_compute_queue_init(struct amdgpu_device *adev, int ring_id)
 	struct cik_mqd *mqd;
 	struct amdgpu_ring *ring = &adev->gfx.compute_ring[ring_id];
 
-	if (ring->mqd_obj == NULL) {
-		r = amdgpu_bo_create(adev,
-				sizeof(struct cik_mqd),
-				PAGE_SIZE, true,
-				AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL,
-				&ring->mqd_obj);
-		if (r) {
-			dev_warn(adev->dev, "(%d) create MQD bo failed\n", r);
-			return r;
-		}
-	}
-
-	r = amdgpu_bo_reserve(ring->mqd_obj, false);
-	if (unlikely(r != 0))
-		goto out;
-
-	r = amdgpu_bo_pin(ring->mqd_obj, AMDGPU_GEM_DOMAIN_GTT,
-			&mqd_gpu_addr);
-	if (r) {
-		dev_warn(adev->dev, "(%d) pin MQD bo failed\n", r);
-		goto out_unreserve;
-	}
-	r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&mqd);
+	r = amdgpu_bo_create_reserved(adev, sizeof(struct cik_mqd), PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_GTT, &ring->mqd_obj,
+				      &mqd_gpu_addr, (void **)&mqd);
 	if (r) {
-		dev_warn(adev->dev, "(%d) map MQD bo failed\n", r);
-		goto out_unreserve;
+		dev_warn(adev->dev, "(%d) create MQD bo failed\n", r);
+		return r;
 	}
 
 	mutex_lock(&adev->srbm_mutex);
@@ -3147,9 +3087,7 @@ static int gfx_v7_0_compute_queue_init(struct amdgpu_device *adev, int ring_id)
 	mutex_unlock(&adev->srbm_mutex);
 
 	amdgpu_bo_kunmap(ring->mqd_obj);
-out_unreserve:
 	amdgpu_bo_unreserve(ring->mqd_obj);
-out:
 	return 0;
 }
 
@@ -3361,43 +3299,9 @@ static void gfx_v7_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
  */
 static void gfx_v7_0_rlc_fini(struct amdgpu_device *adev)
 {
-	int r;
-
-	/* save restore block */
-	if (adev->gfx.rlc.save_restore_obj) {
-		r = amdgpu_bo_reserve(adev->gfx.rlc.save_restore_obj, true);
-		if (unlikely(r != 0))
-			dev_warn(adev->dev, "(%d) reserve RLC sr bo failed\n", r);
-		amdgpu_bo_unpin(adev->gfx.rlc.save_restore_obj);
-		amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
-
-		amdgpu_bo_unref(&adev->gfx.rlc.save_restore_obj);
-		adev->gfx.rlc.save_restore_obj = NULL;
-	}
-
-	/* clear state block */
-	if (adev->gfx.rlc.clear_state_obj) {
-		r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, true);
-		if (unlikely(r != 0))
-			dev_warn(adev->dev, "(%d) reserve RLC c bo failed\n", r);
-		amdgpu_bo_unpin(adev->gfx.rlc.clear_state_obj);
-		amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
-
-		amdgpu_bo_unref(&adev->gfx.rlc.clear_state_obj);
-		adev->gfx.rlc.clear_state_obj = NULL;
-	}
-
-	/* clear state block */
-	if (adev->gfx.rlc.cp_table_obj) {
-		r = amdgpu_bo_reserve(adev->gfx.rlc.cp_table_obj, true);
-		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;
-	}
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.save_restore_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, NULL, NULL);
 }
 
 static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
@@ -3432,39 +3336,17 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
 
 	if (src_ptr) {
 		/* save restore block */
-		if (adev->gfx.rlc.save_restore_obj == NULL) {
-			r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true,
-					     AMDGPU_GEM_DOMAIN_VRAM,
-					     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-					     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-					     NULL, NULL,
-					     &adev->gfx.rlc.save_restore_obj);
-			if (r) {
-				dev_warn(adev->dev, "(%d) create RLC sr bo failed\n", r);
-				return r;
-			}
-		}
-
-		r = amdgpu_bo_reserve(adev->gfx.rlc.save_restore_obj, false);
-		if (unlikely(r != 0)) {
-			gfx_v7_0_rlc_fini(adev);
-			return r;
-		}
-		r = amdgpu_bo_pin(adev->gfx.rlc.save_restore_obj, AMDGPU_GEM_DOMAIN_VRAM,
-				  &adev->gfx.rlc.save_restore_gpu_addr);
+		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.save_restore_obj,
+					      &adev->gfx.rlc.save_restore_gpu_addr,
+					      (void **)&adev->gfx.rlc.sr_ptr);
 		if (r) {
-			amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
-			dev_warn(adev->dev, "(%d) pin RLC sr bo failed\n", r);
+			dev_warn(adev->dev, "(%d) create, pin or map of RLC sr bo failed\n", r);
 			gfx_v7_0_rlc_fini(adev);
 			return r;
 		}
 
-		r = amdgpu_bo_kmap(adev->gfx.rlc.save_restore_obj, (void **)&adev->gfx.rlc.sr_ptr);
-		if (r) {
-			dev_warn(adev->dev, "(%d) map RLC sr bo failed\n", r);
-			gfx_v7_0_rlc_fini(adev);
-			return r;
-		}
 		/* write the sr buffer */
 		dst_ptr = adev->gfx.rlc.sr_ptr;
 		for (i = 0; i < adev->gfx.rlc.reg_list_size; i++)
@@ -3477,39 +3359,17 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
 		/* clear state block */
 		adev->gfx.rlc.clear_state_size = dws = gfx_v7_0_get_csb_size(adev);
 
-		if (adev->gfx.rlc.clear_state_obj == NULL) {
-			r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true,
-					     AMDGPU_GEM_DOMAIN_VRAM,
-					     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-					     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-					     NULL, NULL,
-					     &adev->gfx.rlc.clear_state_obj);
-			if (r) {
-				dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
-				gfx_v7_0_rlc_fini(adev);
-				return r;
-			}
-		}
-		r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, false);
-		if (unlikely(r != 0)) {
-			gfx_v7_0_rlc_fini(adev);
-			return r;
-		}
-		r = amdgpu_bo_pin(adev->gfx.rlc.clear_state_obj, AMDGPU_GEM_DOMAIN_VRAM,
-				  &adev->gfx.rlc.clear_state_gpu_addr);
+		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.clear_state_obj,
+					      &adev->gfx.rlc.clear_state_gpu_addr,
+					      (void **)&adev->gfx.rlc.cs_ptr);
 		if (r) {
-			amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
-			dev_warn(adev->dev, "(%d) pin RLC c bo failed\n", r);
+			dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
 			gfx_v7_0_rlc_fini(adev);
 			return r;
 		}
 
-		r = amdgpu_bo_kmap(adev->gfx.rlc.clear_state_obj, (void **)&adev->gfx.rlc.cs_ptr);
-		if (r) {
-			dev_warn(adev->dev, "(%d) map RLC c bo failed\n", r);
-			gfx_v7_0_rlc_fini(adev);
-			return r;
-		}
 		/* set up the cs buffer */
 		dst_ptr = adev->gfx.rlc.cs_ptr;
 		gfx_v7_0_get_csb_buffer(adev, dst_ptr);
@@ -3518,37 +3378,14 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
 	}
 
 	if (adev->gfx.rlc.cp_table_size) {
-		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 |
-					     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-					     NULL, NULL,
-					     &adev->gfx.rlc.cp_table_obj);
-			if (r) {
-				dev_warn(adev->dev, "(%d) create RLC cp table bo failed\n", r);
-				gfx_v7_0_rlc_fini(adev);
-				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);
-			gfx_v7_0_rlc_fini(adev);
-			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);
-			gfx_v7_0_rlc_fini(adev);
-			return r;
-		}
-		r = amdgpu_bo_kmap(adev->gfx.rlc.cp_table_obj, (void **)&adev->gfx.rlc.cp_table_ptr);
+		r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size,
+					      PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.cp_table_obj,
+					      &adev->gfx.rlc.cp_table_gpu_addr,
+					      (void **)&adev->gfx.rlc.cp_table_ptr);
 		if (r) {
-			dev_warn(adev->dev, "(%d) map RLC cp table bo failed\n", r);
+			dev_warn(adev->dev, "(%d) create RLC cp table bo failed\n", r);
 			gfx_v7_0_rlc_fini(adev);
 			return r;
 		}
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index 05436b8730b4..832e592fcd07 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -1238,29 +1238,8 @@ static void cz_init_cp_jump_table(struct amdgpu_device *adev)
 
 static void gfx_v8_0_rlc_fini(struct amdgpu_device *adev)
 {
-	int r;
-
-	/* clear state block */
-	if (adev->gfx.rlc.clear_state_obj) {
-		r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, true);
-		if (unlikely(r != 0))
-			dev_warn(adev->dev, "(%d) reserve RLC cbs bo failed\n", r);
-		amdgpu_bo_unpin(adev->gfx.rlc.clear_state_obj);
-		amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
-		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, true);
-		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;
-	}
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, NULL, NULL);
 }
 
 static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
@@ -1278,39 +1257,17 @@ static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
 		/* clear state block */
 		adev->gfx.rlc.clear_state_size = dws = gfx_v8_0_get_csb_size(adev);
 
-		if (adev->gfx.rlc.clear_state_obj == NULL) {
-			r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true,
-					     AMDGPU_GEM_DOMAIN_VRAM,
-					     AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
-					     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-					     NULL, NULL,
-					     &adev->gfx.rlc.clear_state_obj);
-			if (r) {
-				dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
-				gfx_v8_0_rlc_fini(adev);
-				return r;
-			}
-		}
-		r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, false);
-		if (unlikely(r != 0)) {
-			gfx_v8_0_rlc_fini(adev);
-			return r;
-		}
-		r = amdgpu_bo_pin(adev->gfx.rlc.clear_state_obj, AMDGPU_GEM_DOMAIN_VRAM,
-				  &adev->gfx.rlc.clear_state_gpu_addr);
+		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.clear_state_obj,
+					      &adev->gfx.rlc.clear_state_gpu_addr,
+					      (void **)&adev->gfx.rlc.cs_ptr);
 		if (r) {
-			amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
-			dev_warn(adev->dev, "(%d) pin RLC cbs bo failed\n", r);
+			dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
 			gfx_v8_0_rlc_fini(adev);
 			return r;
 		}
 
-		r = amdgpu_bo_kmap(adev->gfx.rlc.clear_state_obj, (void **)&adev->gfx.rlc.cs_ptr);
-		if (r) {
-			dev_warn(adev->dev, "(%d) map RLC cbs bo failed\n", r);
-			gfx_v8_0_rlc_fini(adev);
-			return r;
-		}
 		/* set up the cs buffer */
 		dst_ptr = adev->gfx.rlc.cs_ptr;
 		gfx_v8_0_get_csb_buffer(adev, dst_ptr);
@@ -1321,34 +1278,13 @@ static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
 	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 |
-					     AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
-					     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);
+		r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size,
+					      PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.cp_table_obj,
+					      &adev->gfx.rlc.cp_table_gpu_addr,
+					      (void **)&adev->gfx.rlc.cp_table_ptr);
 		if (r) {
-			dev_warn(adev->dev, "(%d) map RLC cp table bo failed\n", r);
+			dev_warn(adev->dev, "(%d) create RLC cp table bo failed\n", r);
 			return r;
 		}
 
@@ -1363,17 +1299,7 @@ static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
 
 static void gfx_v8_0_mec_fini(struct amdgpu_device *adev)
 {
-	int r;
-
-	if (adev->gfx.mec.hpd_eop_obj) {
-		r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, true);
-		if (unlikely(r != 0))
-			dev_warn(adev->dev, "(%d) reserve HPD EOP bo failed\n", r);
-		amdgpu_bo_unpin(adev->gfx.mec.hpd_eop_obj);
-		amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
-		amdgpu_bo_unref(&adev->gfx.mec.hpd_eop_obj);
-		adev->gfx.mec.hpd_eop_obj = NULL;
-	}
+	amdgpu_bo_free_kernel(&adev->gfx.mec.hpd_eop_obj, NULL, NULL);
 }
 
 static int gfx_v8_0_mec_init(struct amdgpu_device *adev)
@@ -1389,34 +1315,13 @@ static int gfx_v8_0_mec_init(struct amdgpu_device *adev)
 
 	mec_hpd_size = adev->gfx.num_compute_rings * GFX8_MEC_HPD_SIZE;
 
-	if (adev->gfx.mec.hpd_eop_obj == NULL) {
-		r = amdgpu_bo_create(adev,
-				     mec_hpd_size,
-				     PAGE_SIZE, true,
-				     AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL,
-				     &adev->gfx.mec.hpd_eop_obj);
-		if (r) {
-			dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
-			return r;
-		}
-	}
-
-	r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, false);
-	if (unlikely(r != 0)) {
-		gfx_v8_0_mec_fini(adev);
-		return r;
-	}
-	r = amdgpu_bo_pin(adev->gfx.mec.hpd_eop_obj, AMDGPU_GEM_DOMAIN_GTT,
-			  &adev->gfx.mec.hpd_eop_gpu_addr);
+	r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.mec.hpd_eop_obj,
+				      &adev->gfx.mec.hpd_eop_gpu_addr,
+				      (void **)&hpd);
 	if (r) {
-		dev_warn(adev->dev, "(%d) pin HDP EOP bo failed\n", r);
-		gfx_v8_0_mec_fini(adev);
-		return r;
-	}
-	r = amdgpu_bo_kmap(adev->gfx.mec.hpd_eop_obj, (void **)&hpd);
-	if (r) {
-		dev_warn(adev->dev, "(%d) map HDP EOP bo failed\n", r);
-		gfx_v8_0_mec_fini(adev);
+		dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
 		return r;
 	}
 
@@ -3802,6 +3707,8 @@ static void gfx_v8_0_gpu_init(struct amdgpu_device *adev)
 				   ELEMENT_SIZE, 1);
 	sh_static_mem_cfg = REG_SET_FIELD(sh_static_mem_cfg, SH_STATIC_MEM_CONFIG,
 				   INDEX_STRIDE, 3);
+	WREG32(mmSH_STATIC_MEM_CONFIG, sh_static_mem_cfg);
+
 	mutex_lock(&adev->srbm_mutex);
 	for (i = 0; i < adev->vm_manager.id_mgr[0].num_ids; i++) {
 		vi_srbm_select(adev, 0, 0, 0, i);
@@ -3825,7 +3732,6 @@ static void gfx_v8_0_gpu_init(struct amdgpu_device *adev)
 
 		WREG32(mmSH_MEM_APE1_BASE, 1);
 		WREG32(mmSH_MEM_APE1_LIMIT, 0);
-		WREG32(mmSH_STATIC_MEM_CONFIG, sh_static_mem_cfg);
 	}
 	vi_srbm_select(adev, 0, 0, 0, 0);
 	mutex_unlock(&adev->srbm_mutex);
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
index 435db6f5efcf..69182eeca264 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
@@ -116,7 +116,9 @@ static const u32 golden_settings_gc_9_0[] =
 	SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UTCL1_CNTL_2), 0x08000000, 0x08000080,
 	SOC15_REG_OFFSET(GC, 0, mmRLC_PREWALKER_UTCL1_CNTL), 0x08000000, 0x08000080,
 	SOC15_REG_OFFSET(GC, 0, mmRLC_SPM_UTCL1_CNTL), 0x08000000, 0x08000080,
+	SOC15_REG_OFFSET(GC, 0, mmSH_MEM_CONFIG), 0x00001000, 0x00001000,
 	SOC15_REG_OFFSET(GC, 0, mmSPI_CONFIG_CNTL_1), 0x0000000f, 0x01000107,
+	SOC15_REG_OFFSET(GC, 0, mmSQC_CONFIG), 0x03000000, 0x020a2000,
 	SOC15_REG_OFFSET(GC, 0, mmTA_CNTL_AUX), 0xfffffeef, 0x010b0000,
 	SOC15_REG_OFFSET(GC, 0, mmTCP_CHAN_STEER_HI), 0xffffffff, 0x4a2c0e68,
 	SOC15_REG_OFFSET(GC, 0, mmTCP_CHAN_STEER_LO), 0xffffffff, 0xb5d3f197,
@@ -772,18 +774,16 @@ static int gfx_v9_0_rlc_init(struct amdgpu_device *adev)
 	if (cs_data) {
 		/* clear state block */
 		adev->gfx.rlc.clear_state_size = dws = gfx_v9_0_get_csb_size(adev);
-		if (adev->gfx.rlc.clear_state_obj == NULL) {
-			r = amdgpu_bo_create_kernel(adev, dws * 4, PAGE_SIZE,
-						AMDGPU_GEM_DOMAIN_VRAM,
-						&adev->gfx.rlc.clear_state_obj,
-						&adev->gfx.rlc.clear_state_gpu_addr,
-						(void **)&adev->gfx.rlc.cs_ptr);
-			if (r) {
-				dev_err(adev->dev,
-					"(%d) failed to create rlc csb bo\n", r);
-				gfx_v9_0_rlc_fini(adev);
-				return r;
-			}
+		r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
+					      AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.clear_state_obj,
+					      &adev->gfx.rlc.clear_state_gpu_addr,
+					      (void **)&adev->gfx.rlc.cs_ptr);
+		if (r) {
+			dev_err(adev->dev, "(%d) failed to create rlc csb bo\n",
+				r);
+			gfx_v9_0_rlc_fini(adev);
+			return r;
 		}
 		/* set up the cs buffer */
 		dst_ptr = adev->gfx.rlc.cs_ptr;
@@ -795,18 +795,16 @@ static int gfx_v9_0_rlc_init(struct amdgpu_device *adev)
 	if (adev->asic_type == CHIP_RAVEN) {
 		/* TODO: double check the cp_table_size for RV */
 		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_kernel(adev, adev->gfx.rlc.cp_table_size,
-						PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
-						&adev->gfx.rlc.cp_table_obj,
-						&adev->gfx.rlc.cp_table_gpu_addr,
-						(void **)&adev->gfx.rlc.cp_table_ptr);
-			if (r) {
-				dev_err(adev->dev,
-					"(%d) failed to create cp table bo\n", r);
-				gfx_v9_0_rlc_fini(adev);
-				return r;
-			}
+		r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size,
+					      PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
+					      &adev->gfx.rlc.cp_table_obj,
+					      &adev->gfx.rlc.cp_table_gpu_addr,
+					      (void **)&adev->gfx.rlc.cp_table_ptr);
+		if (r) {
+			dev_err(adev->dev,
+				"(%d) failed to create cp table bo\n", r);
+			gfx_v9_0_rlc_fini(adev);
+			return r;
 		}
 
 		rv_init_cp_jump_table(adev);
@@ -821,28 +819,8 @@ static int gfx_v9_0_rlc_init(struct amdgpu_device *adev)
 
 static void gfx_v9_0_mec_fini(struct amdgpu_device *adev)
 {
-	int r;
-
-	if (adev->gfx.mec.hpd_eop_obj) {
-		r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, true);
-		if (unlikely(r != 0))
-			dev_warn(adev->dev, "(%d) reserve HPD EOP bo failed\n", r);
-		amdgpu_bo_unpin(adev->gfx.mec.hpd_eop_obj);
-		amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
-
-		amdgpu_bo_unref(&adev->gfx.mec.hpd_eop_obj);
-		adev->gfx.mec.hpd_eop_obj = NULL;
-	}
-	if (adev->gfx.mec.mec_fw_obj) {
-		r = amdgpu_bo_reserve(adev->gfx.mec.mec_fw_obj, true);
-		if (unlikely(r != 0))
-			dev_warn(adev->dev, "(%d) reserve mec firmware bo failed\n", r);
-		amdgpu_bo_unpin(adev->gfx.mec.mec_fw_obj);
-		amdgpu_bo_unreserve(adev->gfx.mec.mec_fw_obj);
-
-		amdgpu_bo_unref(&adev->gfx.mec.mec_fw_obj);
-		adev->gfx.mec.mec_fw_obj = NULL;
-	}
+	amdgpu_bo_free_kernel(&adev->gfx.mec.hpd_eop_obj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->gfx.mec.mec_fw_obj, NULL, NULL);
 }
 
 static int gfx_v9_0_mec_init(struct amdgpu_device *adev)
@@ -862,33 +840,13 @@ static int gfx_v9_0_mec_init(struct amdgpu_device *adev)
 	amdgpu_gfx_compute_queue_acquire(adev);
 	mec_hpd_size = adev->gfx.num_compute_rings * GFX9_MEC_HPD_SIZE;
 
-	if (adev->gfx.mec.hpd_eop_obj == NULL) {
-		r = amdgpu_bo_create(adev,
-				     mec_hpd_size,
-				     PAGE_SIZE, true,
-				     AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL,
-				     &adev->gfx.mec.hpd_eop_obj);
-		if (r) {
-			dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
-			return r;
-		}
-	}
-
-	r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, false);
-	if (unlikely(r != 0)) {
-		gfx_v9_0_mec_fini(adev);
-		return r;
-	}
-	r = amdgpu_bo_pin(adev->gfx.mec.hpd_eop_obj, AMDGPU_GEM_DOMAIN_GTT,
-			  &adev->gfx.mec.hpd_eop_gpu_addr);
+	r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.mec.hpd_eop_obj,
+				      &adev->gfx.mec.hpd_eop_gpu_addr,
+				      (void **)&hpd);
 	if (r) {
-		dev_warn(adev->dev, "(%d) pin HDP EOP bo failed\n", r);
-		gfx_v9_0_mec_fini(adev);
-		return r;
-	}
-	r = amdgpu_bo_kmap(adev->gfx.mec.hpd_eop_obj, (void **)&hpd);
-	if (r) {
-		dev_warn(adev->dev, "(%d) map HDP EOP bo failed\n", r);
+		dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
 		gfx_v9_0_mec_fini(adev);
 		return r;
 	}
@@ -905,42 +863,22 @@ static int gfx_v9_0_mec_init(struct amdgpu_device *adev)
 		 le32_to_cpu(mec_hdr->header.ucode_array_offset_bytes));
 	fw_size = le32_to_cpu(mec_hdr->header.ucode_size_bytes) / 4;
 
-	if (adev->gfx.mec.mec_fw_obj == NULL) {
-		r = amdgpu_bo_create(adev,
-			mec_hdr->header.ucode_size_bytes,
-			PAGE_SIZE, true,
-			AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL,
-			&adev->gfx.mec.mec_fw_obj);
-		if (r) {
-			dev_warn(adev->dev, "(%d) create mec firmware bo failed\n", r);
-			return r;
-		}
-	}
-
-	r = amdgpu_bo_reserve(adev->gfx.mec.mec_fw_obj, false);
-	if (unlikely(r != 0)) {
-		gfx_v9_0_mec_fini(adev);
-		return r;
-	}
-	r = amdgpu_bo_pin(adev->gfx.mec.mec_fw_obj, AMDGPU_GEM_DOMAIN_GTT,
-			&adev->gfx.mec.mec_fw_gpu_addr);
-	if (r) {
-		dev_warn(adev->dev, "(%d) pin mec firmware bo failed\n", r);
-		gfx_v9_0_mec_fini(adev);
-		return r;
-	}
-	r = amdgpu_bo_kmap(adev->gfx.mec.mec_fw_obj, (void **)&fw);
+	r = amdgpu_bo_create_reserved(adev, mec_hdr->header.ucode_size_bytes,
+				      PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
+				      &adev->gfx.mec.mec_fw_obj,
+				      &adev->gfx.mec.mec_fw_gpu_addr,
+				      (void **)&fw);
 	if (r) {
-		dev_warn(adev->dev, "(%d) map firmware bo failed\n", r);
+		dev_warn(adev->dev, "(%d) create mec firmware bo failed\n", r);
 		gfx_v9_0_mec_fini(adev);
 		return r;
 	}
+
 	memcpy(fw, fw_data, fw_size);
 
 	amdgpu_bo_kunmap(adev->gfx.mec.mec_fw_obj);
 	amdgpu_bo_unreserve(adev->gfx.mec.mec_fw_obj);
 
-
 	return 0;
 }
 
@@ -2219,7 +2157,7 @@ static int gfx_v9_0_cp_gfx_start(struct amdgpu_device *adev)
 	struct amdgpu_ring *ring = &adev->gfx.gfx_ring[0];
 	const struct cs_section_def *sect = NULL;
 	const struct cs_extent_def *ext = NULL;
-	int r, i;
+	int r, i, tmp;
 
 	/* init the CP */
 	WREG32_SOC15(GC, 0, mmCP_MAX_CONTEXT, adev->gfx.config.max_hw_contexts - 1);
@@ -2227,7 +2165,7 @@ static int gfx_v9_0_cp_gfx_start(struct amdgpu_device *adev)
 
 	gfx_v9_0_cp_gfx_enable(adev, true);
 
-	r = amdgpu_ring_alloc(ring, gfx_v9_0_get_csb_size(adev) + 4);
+	r = amdgpu_ring_alloc(ring, gfx_v9_0_get_csb_size(adev) + 4 + 3);
 	if (r) {
 		DRM_ERROR("amdgpu: cp failed to lock ring (%d).\n", r);
 		return r;
@@ -2265,6 +2203,12 @@ static int gfx_v9_0_cp_gfx_start(struct amdgpu_device *adev)
 	amdgpu_ring_write(ring, 0x8000);
 	amdgpu_ring_write(ring, 0x8000);
 
+	amdgpu_ring_write(ring, PACKET3(PACKET3_SET_UCONFIG_REG,1));
+	tmp = (PACKET3_SET_UCONFIG_REG_INDEX_TYPE |
+		(SOC15_REG_OFFSET(GC, 0, mmVGT_INDEX_TYPE) - PACKET3_SET_UCONFIG_REG_START));
+	amdgpu_ring_write(ring, tmp);
+	amdgpu_ring_write(ring, 0);
+
 	amdgpu_ring_commit(ring);
 
 	return 0;
@@ -4158,7 +4102,7 @@ static int gfx_v9_0_kiq_irq(struct amdgpu_device *adev,
 	return 0;
 }
 
-const struct amd_ip_funcs gfx_v9_0_ip_funcs = {
+static const struct amd_ip_funcs gfx_v9_0_ip_funcs = {
 	.name = "gfx_v9_0",
 	.early_init = gfx_v9_0_early_init,
 	.late_init = gfx_v9_0_late_init,
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h
index 56ef652a575d..fa5a3fbaf6ab 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h
@@ -24,7 +24,6 @@
 #ifndef __GFX_V9_0_H__
 #define __GFX_V9_0_H__
 
-extern const struct amd_ip_funcs gfx_v9_0_ip_funcs;
 extern const struct amdgpu_ip_block_version gfx_v9_0_ip_block;
 
 void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num);
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c
index 408723ef157c..4f2788b61a08 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c
@@ -124,7 +124,7 @@ static void gfxhub_v1_0_init_tlb_regs(struct amdgpu_device *adev)
 
 static void gfxhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
 {
-	uint32_t tmp;
+	uint32_t tmp, field;
 
 	/* Setup L2 cache */
 	tmp = RREG32_SOC15(GC, 0, mmVM_L2_CNTL);
@@ -143,9 +143,10 @@ static void gfxhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
 	WREG32_SOC15(GC, 0, mmVM_L2_CNTL2, tmp);
 
+	field = adev->vm_manager.fragment_size;
 	tmp = mmVM_L2_CNTL3_DEFAULT;
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 12);
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, field);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
 	WREG32_SOC15(GC, 0, mmVM_L2_CNTL3, tmp);
 
 	tmp = mmVM_L2_CNTL4_DEFAULT;
diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h
index d2dbb085f480..206e29cad753 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h
@@ -30,7 +30,5 @@ void gfxhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev,
 					  bool value);
 void gfxhub_v1_0_init(struct amdgpu_device *adev);
 u64 gfxhub_v1_0_get_mc_fb_offset(struct amdgpu_device *adev);
-extern const struct amd_ip_funcs gfxhub_v1_0_ip_funcs;
-extern const struct amdgpu_ip_block_version gfxhub_v1_0_ip_block;
 
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
index 93c45f26b7c8..12b0c4cd7a5a 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
@@ -461,6 +461,7 @@ static void gmc_v6_0_set_prt(struct amdgpu_device *adev, bool enable)
 static int gmc_v6_0_gart_enable(struct amdgpu_device *adev)
 {
 	int r, i;
+	u32 field;
 
 	if (adev->gart.robj == NULL) {
 		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
@@ -488,10 +489,12 @@ static int gmc_v6_0_gart_enable(struct amdgpu_device *adev)
 	WREG32(mmVM_L2_CNTL2,
 	       VM_L2_CNTL2__INVALIDATE_ALL_L1_TLBS_MASK |
 	       VM_L2_CNTL2__INVALIDATE_L2_CACHE_MASK);
+
+	field = adev->vm_manager.fragment_size;
 	WREG32(mmVM_L2_CNTL3,
 	       VM_L2_CNTL3__L2_CACHE_BIGK_ASSOCIATIVITY_MASK |
-	       (4UL << VM_L2_CNTL3__BANK_SELECT__SHIFT) |
-	       (4UL << VM_L2_CNTL3__L2_CACHE_BIGK_FRAGMENT_SIZE__SHIFT));
+	       (field << VM_L2_CNTL3__BANK_SELECT__SHIFT) |
+	       (field << VM_L2_CNTL3__L2_CACHE_BIGK_FRAGMENT_SIZE__SHIFT));
 	/* setup context0 */
 	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gart_start >> 12);
 	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->mc.gart_end >> 12);
@@ -811,7 +814,7 @@ static int gmc_v6_0_sw_init(void *handle)
 	if (r)
 		return r;
 
-	amdgpu_vm_adjust_size(adev, 64);
+	amdgpu_vm_adjust_size(adev, 64, 4);
 	adev->vm_manager.max_pfn = adev->vm_manager.vm_size << 18;
 
 	adev->mc.mc_mask = 0xffffffffffULL;
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
index 4a9e84062874..e42c1ad3af5e 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
@@ -562,7 +562,7 @@ static void gmc_v7_0_set_prt(struct amdgpu_device *adev, bool enable)
 static int gmc_v7_0_gart_enable(struct amdgpu_device *adev)
 {
 	int r, i;
-	u32 tmp;
+	u32 tmp, field;
 
 	if (adev->gart.robj == NULL) {
 		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
@@ -592,10 +592,12 @@ static int gmc_v7_0_gart_enable(struct amdgpu_device *adev)
 	tmp = REG_SET_FIELD(0, VM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
 	WREG32(mmVM_L2_CNTL2, tmp);
+
+	field = adev->vm_manager.fragment_size;
 	tmp = RREG32(mmVM_L2_CNTL3);
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_ASSOCIATIVITY, 1);
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 4);
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, 4);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, field);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, field);
 	WREG32(mmVM_L2_CNTL3, tmp);
 	/* setup context0 */
 	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gart_start >> 12);
@@ -948,7 +950,7 @@ static int gmc_v7_0_sw_init(void *handle)
 	 * Currently set to 4GB ((1 << 20) 4k pages).
 	 * Max GPUVM size for cayman and SI is 40 bits.
 	 */
-	amdgpu_vm_adjust_size(adev, 64);
+	amdgpu_vm_adjust_size(adev, 64, 4);
 	adev->vm_manager.max_pfn = adev->vm_manager.vm_size << 18;
 
 	/* Set the internal MC address mask
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
index 85c937b5e40b..7ca2dae8237a 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
@@ -762,7 +762,7 @@ static void gmc_v8_0_set_prt(struct amdgpu_device *adev, bool enable)
 static int gmc_v8_0_gart_enable(struct amdgpu_device *adev)
 {
 	int r, i;
-	u32 tmp;
+	u32 tmp, field;
 
 	if (adev->gart.robj == NULL) {
 		dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
@@ -793,10 +793,12 @@ static int gmc_v8_0_gart_enable(struct amdgpu_device *adev)
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
 	WREG32(mmVM_L2_CNTL2, tmp);
+
+	field = adev->vm_manager.fragment_size;
 	tmp = RREG32(mmVM_L2_CNTL3);
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_ASSOCIATIVITY, 1);
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 4);
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, 4);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, field);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, field);
 	WREG32(mmVM_L2_CNTL3, tmp);
 	/* XXX: set to enable PTE/PDE in system memory */
 	tmp = RREG32(mmVM_L2_CNTL4);
@@ -1046,7 +1048,7 @@ static int gmc_v8_0_sw_init(void *handle)
 	 * Currently set to 4GB ((1 << 20) 4k pages).
 	 * Max GPUVM size for cayman and SI is 40 bits.
 	 */
-	amdgpu_vm_adjust_size(adev, 64);
+	amdgpu_vm_adjust_size(adev, 64, 4);
 	adev->vm_manager.max_pfn = adev->vm_manager.vm_size << 18;
 
 	/* Set the internal MC address mask
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index c22899a08106..2769c2b3b56e 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -541,9 +541,10 @@ static int gmc_v9_0_sw_init(void *handle)
 			adev->vm_manager.vm_size = 1U << 18;
 			adev->vm_manager.block_size = 9;
 			adev->vm_manager.num_level = 3;
+			amdgpu_vm_set_fragment_size(adev, 9);
 		} else {
-			/* vm_size is 64GB for legacy 2-level page support*/
-			amdgpu_vm_adjust_size(adev, 64);
+			/* vm_size is 64GB for legacy 2-level page support */
+			amdgpu_vm_adjust_size(adev, 64, 9);
 			adev->vm_manager.num_level = 1;
 		}
 		break;
@@ -558,14 +559,16 @@ static int gmc_v9_0_sw_init(void *handle)
 		adev->vm_manager.vm_size = 1U << 18;
 		adev->vm_manager.block_size = 9;
 		adev->vm_manager.num_level = 3;
+		amdgpu_vm_set_fragment_size(adev, 9);
 		break;
 	default:
 		break;
 	}
 
-	DRM_INFO("vm size is %llu GB, block size is %u-bit\n",
+	DRM_INFO("vm size is %llu GB, block size is %u-bit,fragment size is %u-bit\n",
 			adev->vm_manager.vm_size,
-			adev->vm_manager.block_size);
+			adev->vm_manager.block_size,
+			adev->vm_manager.fragment_size);
 
 	/* This interrupt is VMC page fault.*/
 	r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_VMC, 0,
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c
index ad8def3cc343..4395a4f12149 100644
--- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c
@@ -138,7 +138,7 @@ static void mmhub_v1_0_init_tlb_regs(struct amdgpu_device *adev)
 
 static void mmhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
 {
-	uint32_t tmp;
+	uint32_t tmp, field;
 
 	/* Setup L2 cache */
 	tmp = RREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL);
@@ -157,9 +157,10 @@ static void mmhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
 	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
 	WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL2, tmp);
 
+	field = adev->vm_manager.fragment_size;
 	tmp = mmVM_L2_CNTL3_DEFAULT;
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 12);
-	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, field);
+	tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, 6);
 	WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL3, tmp);
 
 	tmp = mmVM_L2_CNTL4_DEFAULT;
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h
index 57bb940c0ecd..5d38229baf69 100644
--- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h
@@ -36,7 +36,4 @@ void mmhub_v1_0_initialize_power_gating(struct amdgpu_device *adev);
 void mmhub_v1_0_update_power_gating(struct amdgpu_device *adev,
                                 bool enable);
 
-extern const struct amd_ip_funcs mmhub_v1_0_ip_funcs;
-extern const struct amdgpu_ip_block_version mmhub_v1_0_ip_block;
-
 #endif
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
index 591f3e7fb508..fd7c72aaafa6 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
@@ -291,6 +291,8 @@ static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring)
 
 	DRM_DEBUG("Setting write pointer\n");
 	if (ring->use_doorbell) {
+		u64 *wb = (u64 *)&adev->wb.wb[ring->wptr_offs];
+
 		DRM_DEBUG("Using doorbell -- "
 				"wptr_offs == 0x%08x "
 				"lower_32_bits(ring->wptr) << 2 == 0x%08x "
@@ -299,8 +301,7 @@ static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring)
 				lower_32_bits(ring->wptr << 2),
 				upper_32_bits(ring->wptr << 2));
 		/* XXX check if swapping is necessary on BE */
-		adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr << 2);
-		adev->wb.wb[ring->wptr_offs + 1] = upper_32_bits(ring->wptr << 2);
+		WRITE_ONCE(*wb, (ring->wptr << 2));
 		DRM_DEBUG("calling WDOORBELL64(0x%08x, 0x%016llx)\n",
 				ring->doorbell_index, ring->wptr << 2);
 		WDOORBELL64(ring->doorbell_index, ring->wptr << 2);
@@ -573,12 +574,13 @@ static void sdma_v4_0_enable(struct amdgpu_device *adev, bool enable)
 static int sdma_v4_0_gfx_resume(struct amdgpu_device *adev)
 {
 	struct amdgpu_ring *ring;
-	u32 rb_cntl, ib_cntl;
+	u32 rb_cntl, ib_cntl, wptr_poll_cntl;
 	u32 rb_bufsz;
 	u32 wb_offset;
 	u32 doorbell;
 	u32 doorbell_offset;
 	u32 temp;
+	u64 wptr_gpu_addr;
 	int i, r;
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
@@ -660,6 +662,19 @@ static int sdma_v4_0_gfx_resume(struct amdgpu_device *adev)
 			WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_F32_CNTL), temp);
 		}
 
+		/* setup the wptr shadow polling */
+		wptr_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
+		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO),
+		       lower_32_bits(wptr_gpu_addr));
+		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI),
+		       upper_32_bits(wptr_gpu_addr));
+		wptr_poll_cntl = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL));
+		if (amdgpu_sriov_vf(adev))
+			wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl, SDMA0_GFX_RB_WPTR_POLL_CNTL, F32_POLL_ENABLE, 1);
+		else
+			wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl, SDMA0_GFX_RB_WPTR_POLL_CNTL, F32_POLL_ENABLE, 0);
+		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL), wptr_poll_cntl);
+
 		/* enable DMA RB */
 		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1);
 		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
@@ -687,6 +702,7 @@ static int sdma_v4_0_gfx_resume(struct amdgpu_device *adev)
 
 		if (adev->mman.buffer_funcs_ring == ring)
 			amdgpu_ttm_set_active_vram_size(adev, adev->mc.real_vram_size);
+
 	}
 
 	return 0;
@@ -783,15 +799,12 @@ static int sdma_v4_0_load_microcode(struct amdgpu_device *adev)
 	const struct sdma_firmware_header_v1_0 *hdr;
 	const __le32 *fw_data;
 	u32 fw_size;
-	u32 digest_size = 0;
 	int i, j;
 
 	/* halt the MEs */
 	sdma_v4_0_enable(adev, false);
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
-		uint16_t version_major;
-		uint16_t version_minor;
 		if (!adev->sdma.instance[i].fw)
 			return -EINVAL;
 
@@ -799,23 +812,12 @@ static int sdma_v4_0_load_microcode(struct amdgpu_device *adev)
 		amdgpu_ucode_print_sdma_hdr(&hdr->header);
 		fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
 
-		version_major = le16_to_cpu(hdr->header.header_version_major);
-		version_minor = le16_to_cpu(hdr->header.header_version_minor);
-
-		if (version_major == 1 && version_minor >= 1) {
-			const struct sdma_firmware_header_v1_1 *sdma_v1_1_hdr = (const struct sdma_firmware_header_v1_1 *) hdr;
-			digest_size = le32_to_cpu(sdma_v1_1_hdr->digest_size);
-		}
-
-		fw_size -= digest_size;
-
 		fw_data = (const __le32 *)
 			(adev->sdma.instance[i].fw->data +
 				le32_to_cpu(hdr->header.ucode_array_offset_bytes));
 
 		WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_UCODE_ADDR), 0);
 
-
 		for (j = 0; j < fw_size; j++)
 			WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_UCODE_DATA), le32_to_cpup(fw_data++));
 
diff --git a/drivers/gpu/drm/amd/amdgpu/si.c b/drivers/gpu/drm/amd/amdgpu/si.c
index 812a24dd1204..8284d5dbfc30 100644
--- a/drivers/gpu/drm/amd/amdgpu/si.c
+++ b/drivers/gpu/drm/amd/amdgpu/si.c
@@ -1413,6 +1413,7 @@ static void si_init_golden_registers(struct amdgpu_device *adev)
 		amdgpu_program_register_sequence(adev,
 						 pitcairn_mgcg_cgcg_init,
 						 (const u32)ARRAY_SIZE(pitcairn_mgcg_cgcg_init));
+		break;
 	case CHIP_VERDE:
 		amdgpu_program_register_sequence(adev,
 						 verde_golden_registers,
@@ -1437,6 +1438,7 @@ static void si_init_golden_registers(struct amdgpu_device *adev)
 		amdgpu_program_register_sequence(adev,
 						 oland_mgcg_cgcg_init,
 						 (const u32)ARRAY_SIZE(oland_mgcg_cgcg_init));
+		break;
 	case CHIP_HAINAN:
 		amdgpu_program_register_sequence(adev,
 						 hainan_golden_registers,
diff --git a/drivers/gpu/drm/amd/amdgpu/soc15d.h b/drivers/gpu/drm/amd/amdgpu/soc15d.h
index e79befd80eed..7f408f85fdb6 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc15d.h
+++ b/drivers/gpu/drm/amd/amdgpu/soc15d.h
@@ -250,6 +250,7 @@
 #define	PACKET3_SET_UCONFIG_REG				0x79
 #define		PACKET3_SET_UCONFIG_REG_START			0x0000c000
 #define		PACKET3_SET_UCONFIG_REG_END			0x0000c400
+#define		PACKET3_SET_UCONFIG_REG_INDEX_TYPE		(2 << 28)
 #define	PACKET3_SCRATCH_RAM_WRITE			0x7D
 #define	PACKET3_SCRATCH_RAM_READ			0x7E
 #define	PACKET3_LOAD_CONST_RAM				0x80
diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c
index 987b958368ac..23a85750edd6 100644
--- a/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c
@@ -165,6 +165,9 @@ static int uvd_v7_0_enc_ring_test_ring(struct amdgpu_ring *ring)
 	unsigned i;
 	int r;
 
+	if (amdgpu_sriov_vf(adev))
+		return 0;
+
 	r = amdgpu_ring_alloc(ring, 16);
 	if (r) {
 		DRM_ERROR("amdgpu: uvd enc failed to lock ring %d (%d).\n",
@@ -432,13 +435,19 @@ static int uvd_v7_0_sw_init(void *handle)
 			return r;
 	}
 
-
 	for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
 		ring = &adev->uvd.ring_enc[i];
 		sprintf(ring->name, "uvd_enc%d", i);
 		if (amdgpu_sriov_vf(adev)) {
 			ring->use_doorbell = true;
-			ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING0_1 * 2;
+
+			/* currently only use the first enconding ring for
+			 * sriov, so set unused location for other unused rings.
+			 */
+			if (i == 0)
+				ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING0_1 * 2;
+			else
+				ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING2_3 * 2 + 1;
 		}
 		r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.irq, 0);
 		if (r)
@@ -685,6 +694,11 @@ static int uvd_v7_0_mmsch_start(struct amdgpu_device *adev,
 	/* 4, set resp to zero */
 	WREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_RESP, 0);
 
+	WDOORBELL32(adev->uvd.ring_enc[0].doorbell_index, 0);
+	adev->wb.wb[adev->uvd.ring_enc[0].wptr_offs] = 0;
+	adev->uvd.ring_enc[0].wptr = 0;
+	adev->uvd.ring_enc[0].wptr_old = 0;
+
 	/* 5, kick off the initialization and wait until VCE_MMSCH_VF_MAILBOX_RESP becomes non-zero */
 	WREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_HOST, 0x10000001);
 
@@ -702,7 +716,6 @@ static int uvd_v7_0_mmsch_start(struct amdgpu_device *adev,
 		dev_err(adev->dev, "failed to init MMSCH, mmVCE_MMSCH_VF_MAILBOX_RESP = %x\n", data);
 		return -EBUSY;
 	}
-	WDOORBELL32(adev->uvd.ring_enc[0].doorbell_index, 0);
 
 	return 0;
 }
@@ -736,11 +749,9 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev)
 		init_table += header->uvd_table_offset;
 
 		ring = &adev->uvd.ring;
+		ring->wptr = 0;
 		size = AMDGPU_GPU_PAGE_ALIGN(adev->uvd.fw->size + 4);
 
-		/* disable clock gating */
-		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS),
-						   ~UVD_POWER_STATUS__UVD_PG_MODE_MASK, 0);
 		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS),
 						   0xFFFFFFFF, 0x00000004);
 		/* mc resume*/
@@ -777,12 +788,6 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev)
 		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_SIZE2),
 					    AMDGPU_UVD_STACK_SIZE + (AMDGPU_UVD_SESSION_SIZE * 40));
 
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_UDEC_ADDR_CONFIG),
-					    adev->gfx.config.gb_addr_config);
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_UDEC_DB_ADDR_CONFIG),
-					    adev->gfx.config.gb_addr_config);
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_UDEC_DBW_ADDR_CONFIG),
-					    adev->gfx.config.gb_addr_config);
 		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_GP_SCRATCH4), adev->uvd.max_handles);
 		/* mc resume end*/
 
@@ -819,17 +824,6 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev)
 						       UVD_LMI_CTRL__REQ_MODE_MASK |
 						       0x00100000L));
 
-		/* disable byte swapping */
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_SWAP_CNTL), 0);
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MP_SWAP_CNTL), 0);
-
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MPC_SET_MUXA0), 0x40c2040);
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MPC_SET_MUXA1), 0x0);
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MPC_SET_MUXB0), 0x40c2040);
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MPC_SET_MUXB1), 0x0);
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MPC_SET_ALU), 0);
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MPC_SET_MUX), 0x88);
-
 		/* take all subblocks out of reset, except VCPU */
 		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET),
 					    UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
@@ -838,15 +832,6 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev)
 		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CNTL),
 					    UVD_VCPU_CNTL__CLK_EN_MASK);
 
-		/* enable UMC */
-		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2),
-						   ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, 0);
-
-		/* boot up the VCPU */
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET), 0);
-
-		MMSCH_V1_0_INSERT_DIRECT_POLL(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS), 0x02, 0x02);
-
 		/* enable master interrupt */
 		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN),
 						   ~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK),
@@ -859,40 +844,31 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev)
 		/* force RBC into idle state */
 		size = order_base_2(ring->ring_size);
 		tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, size);
-		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
 		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
-		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_WPTR_POLL_EN, 0);
-		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
-		tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
 		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_RB_CNTL), tmp);
 
-		/* set the write pointer delay */
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_RB_WPTR_CNTL), 0);
-
-		/* set the wb address */
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_RB_RPTR_ADDR),
-					    (upper_32_bits(ring->gpu_addr) >> 2));
-
-		/* programm the RB_BASE for ring buffer */
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW),
-					    lower_32_bits(ring->gpu_addr));
-		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH),
-					    upper_32_bits(ring->gpu_addr));
-
-		ring->wptr = 0;
 		ring = &adev->uvd.ring_enc[0];
+		ring->wptr = 0;
 		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RB_BASE_LO), ring->gpu_addr);
 		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RB_BASE_HI), upper_32_bits(ring->gpu_addr));
 		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RB_SIZE), ring->ring_size / 4);
 
+		/* boot up the VCPU */
+		MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET), 0);
+
+		/* enable UMC */
+		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2),
+										   ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, 0);
+
+		MMSCH_V1_0_INSERT_DIRECT_POLL(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS), 0x02, 0x02);
+
 		/* add end packet */
 		memcpy((void *)init_table, &end, sizeof(struct mmsch_v1_0_cmd_end));
 		table_size += sizeof(struct mmsch_v1_0_cmd_end) / 4;
 		header->uvd_table_size = table_size;
 
-		return uvd_v7_0_mmsch_start(adev, &adev->virt.mm_table);
 	}
-	return -EINVAL; /* already initializaed ? */
+	return uvd_v7_0_mmsch_start(adev, &adev->virt.mm_table);
 }
 
 /**
diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
index 1ecd6bb90c1f..11134d5f7443 100644
--- a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
@@ -173,6 +173,11 @@ static int vce_v4_0_mmsch_start(struct amdgpu_device *adev,
 	/* 4, set resp to zero */
 	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_RESP), 0);
 
+	WDOORBELL32(adev->vce.ring[0].doorbell_index, 0);
+	adev->wb.wb[adev->vce.ring[0].wptr_offs] = 0;
+	adev->vce.ring[0].wptr = 0;
+	adev->vce.ring[0].wptr_old = 0;
+
 	/* 5, kick off the initialization and wait until VCE_MMSCH_VF_MAILBOX_RESP becomes non-zero */
 	WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_HOST), 0x10000001);
 
@@ -190,7 +195,6 @@ static int vce_v4_0_mmsch_start(struct amdgpu_device *adev,
 		dev_err(adev->dev, "failed to init MMSCH, mmVCE_MMSCH_VF_MAILBOX_RESP = %x\n", data);
 		return -EBUSY;
 	}
-	WDOORBELL32(adev->vce.ring[0].doorbell_index, 0);
 
 	return 0;
 }
@@ -274,7 +278,8 @@ static int vce_v4_0_sriov_start(struct amdgpu_device *adev)
 
 		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_CTRL2), ~0x100, 0);
 		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_SYS_INT_EN),
-						   0xffffffff, VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
+						   VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK,
+						   VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
 
 		/* end of MC_RESUME */
 		MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS),
@@ -296,11 +301,9 @@ static int vce_v4_0_sriov_start(struct amdgpu_device *adev)
 		memcpy((void *)init_table, &end, sizeof(struct mmsch_v1_0_cmd_end));
 		table_size += sizeof(struct mmsch_v1_0_cmd_end) / 4;
 		header->vce_table_size = table_size;
-
-		return vce_v4_0_mmsch_start(adev, &adev->virt.mm_table);
 	}
 
-	return -EINVAL; /* already initializaed ? */
+	return vce_v4_0_mmsch_start(adev, &adev->virt.mm_table);
 }
 
 /**
@@ -443,12 +446,14 @@ static int vce_v4_0_sw_init(void *handle)
 		if (amdgpu_sriov_vf(adev)) {
 			/* DOORBELL only works under SRIOV */
 			ring->use_doorbell = true;
+
+			/* currently only use the first encoding ring for sriov,
+			 * so set unused location for other unused rings.
+			 */
 			if (i == 0)
-				ring->doorbell_index = AMDGPU_DOORBELL64_RING0_1 * 2;
-			else if (i == 1)
-				ring->doorbell_index = AMDGPU_DOORBELL64_RING2_3 * 2;
+				ring->doorbell_index = AMDGPU_DOORBELL64_VCE_RING0_1 * 2;
 			else
-				ring->doorbell_index = AMDGPU_DOORBELL64_RING2_3 * 2 + 1;
+				ring->doorbell_index = AMDGPU_DOORBELL64_VCE_RING2_3 * 2 + 1;
 		}
 		r = amdgpu_ring_init(adev, ring, 512, &adev->vce.irq, 0);
 		if (r)
@@ -990,11 +995,13 @@ static int vce_v4_0_set_interrupt_state(struct amdgpu_device *adev,
 {
 	uint32_t val = 0;
 
-	if (state == AMDGPU_IRQ_STATE_ENABLE)
-		val |= VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK;
+	if (!amdgpu_sriov_vf(adev)) {
+		if (state == AMDGPU_IRQ_STATE_ENABLE)
+			val |= VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK;
 
-	WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SYS_INT_EN), val,
-			~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
+		WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SYS_INT_EN), val,
+				~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
+	}
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c
index 6cac291c96da..9ff69b90df36 100644
--- a/drivers/gpu/drm/amd/amdgpu/vi.c
+++ b/drivers/gpu/drm/amd/amdgpu/vi.c
@@ -1028,8 +1028,7 @@ static int vi_common_early_init(void *handle)
 		/* rev0 hardware requires workarounds to support PG */
 		adev->pg_flags = 0;
 		if (adev->rev_id != 0x00 || CZ_REV_BRISTOL(adev->pdev->revision)) {
-			adev->pg_flags |= AMD_PG_SUPPORT_GFX_PG |
-				AMD_PG_SUPPORT_GFX_SMG |
+			adev->pg_flags |= AMD_PG_SUPPORT_GFX_SMG |
 				AMD_PG_SUPPORT_GFX_PIPELINE |
 				AMD_PG_SUPPORT_CP |
 				AMD_PG_SUPPORT_UVD |