diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu')
58 files changed, 4172 insertions, 1636 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/Kconfig b/drivers/gpu/drm/amd/amdgpu/Kconfig index 7335c0420c70..f3cb69de0c44 100644 --- a/drivers/gpu/drm/amd/amdgpu/Kconfig +++ b/drivers/gpu/drm/amd/amdgpu/Kconfig @@ -25,3 +25,5 @@ config DRM_AMDGPU_GART_DEBUGFS Selecting this option creates a debugfs file to inspect the mapped pages. Uses more memory for housekeeping, enable only for debugging. +source "drivers/gpu/drm/amd/powerplay/Kconfig" +source "drivers/gpu/drm/amd/acp/Kconfig" diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile index c7fcdcedaadb..21dd7c00da15 100644 --- a/drivers/gpu/drm/amd/amdgpu/Makefile +++ b/drivers/gpu/drm/amd/amdgpu/Makefile @@ -58,7 +58,8 @@ amdgpu-y += \ # add DCE block amdgpu-y += \ dce_v10_0.o \ - dce_v11_0.o + dce_v11_0.o \ + dce_virtual.o # add GFX block amdgpu-y += \ diff --git a/drivers/gpu/drm/amd/amdgpu/ObjectID.h b/drivers/gpu/drm/amd/amdgpu/ObjectID.h index 06192698bd96..b8d66670bb17 100644 --- a/drivers/gpu/drm/amd/amdgpu/ObjectID.h +++ b/drivers/gpu/drm/amd/amdgpu/ObjectID.h @@ -90,6 +90,7 @@ #define ENCODER_OBJECT_ID_INTERNAL_VCE 0x24 #define ENCODER_OBJECT_ID_INTERNAL_UNIPHY3 0x25 #define ENCODER_OBJECT_ID_INTERNAL_AMCLK 0x27 +#define ENCODER_OBJECT_ID_VIRTUAL 0x28 #define ENCODER_OBJECT_ID_GENERAL_EXTERNAL_DVO 0xFF @@ -119,6 +120,7 @@ #define CONNECTOR_OBJECT_ID_eDP 0x14 #define CONNECTOR_OBJECT_ID_MXM 0x15 #define CONNECTOR_OBJECT_ID_LVDS_eDP 0x16 +#define CONNECTOR_OBJECT_ID_VIRTUAL 0x17 /* deleted */ @@ -147,6 +149,7 @@ #define GRAPH_OBJECT_ENUM_ID5 0x05 #define GRAPH_OBJECT_ENUM_ID6 0x06 #define GRAPH_OBJECT_ENUM_ID7 0x07 +#define GRAPH_OBJECT_ENUM_VIRTUAL 0x08 /****************************************************/ /* Graphics Object ID Bit definition */ @@ -408,6 +411,10 @@ GRAPH_OBJECT_ENUM_ID1 << ENUM_ID_SHIFT |\ ENCODER_OBJECT_ID_HDMI_ANX9805 << OBJECT_ID_SHIFT) +#define ENCODER_VIRTUAL_ENUM_VIRTUAL ( GRAPH_OBJECT_TYPE_ENCODER << OBJECT_TYPE_SHIFT |\ + GRAPH_OBJECT_ENUM_VIRTUAL << ENUM_ID_SHIFT |\ + ENCODER_OBJECT_ID_VIRTUAL << OBJECT_ID_SHIFT) + /****************************************************/ /* Connector Object ID definition - Shared with BIOS */ /****************************************************/ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 8ebc5f1eb4c0..3cc2629eb158 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -51,6 +51,7 @@ #include "amdgpu_ih.h" #include "amdgpu_irq.h" #include "amdgpu_ucode.h" +#include "amdgpu_ttm.h" #include "amdgpu_gds.h" #include "amd_powerplay.h" #include "amdgpu_acp.h" @@ -91,6 +92,8 @@ extern unsigned amdgpu_pcie_lane_cap; extern unsigned amdgpu_cg_mask; extern unsigned amdgpu_pg_mask; extern char *amdgpu_disable_cu; +extern int amdgpu_sclk_deep_sleep_en; +extern char *amdgpu_virtual_display; #define AMDGPU_WAIT_IDLE_TIMEOUT_IN_MS 3000 #define AMDGPU_MAX_USEC_TIMEOUT 100000 /* 100 ms */ @@ -248,10 +251,9 @@ struct amdgpu_vm_pte_funcs { uint64_t pe, uint64_t src, unsigned count); /* write pte one entry at a time with addr mapping */ - void (*write_pte)(struct amdgpu_ib *ib, - const dma_addr_t *pages_addr, uint64_t pe, - uint64_t addr, unsigned count, - uint32_t incr, uint32_t flags); + void (*write_pte)(struct amdgpu_ib *ib, uint64_t pe, + uint64_t value, unsigned count, + uint32_t incr); /* for linear pte/pde updates without addr mapping */ void (*set_pte_pde)(struct amdgpu_ib *ib, uint64_t pe, @@ -396,46 +398,9 @@ int amdgpu_fence_wait_empty(struct amdgpu_ring *ring); unsigned amdgpu_fence_count_emitted(struct amdgpu_ring *ring); /* - * TTM. + * BO. */ -#define AMDGPU_TTM_LRU_SIZE 20 - -struct amdgpu_mman_lru { - struct list_head *lru[TTM_NUM_MEM_TYPES]; - struct list_head *swap_lru; -}; - -struct amdgpu_mman { - struct ttm_bo_global_ref bo_global_ref; - struct drm_global_reference mem_global_ref; - struct ttm_bo_device bdev; - bool mem_global_referenced; - bool initialized; - -#if defined(CONFIG_DEBUG_FS) - struct dentry *vram; - struct dentry *gtt; -#endif - - /* buffer handling */ - const struct amdgpu_buffer_funcs *buffer_funcs; - struct amdgpu_ring *buffer_funcs_ring; - /* Scheduler entity for buffer moves */ - struct amd_sched_entity entity; - - /* custom LRU management */ - struct amdgpu_mman_lru log2_size[AMDGPU_TTM_LRU_SIZE]; -}; - -int amdgpu_copy_buffer(struct amdgpu_ring *ring, - uint64_t src_offset, - uint64_t dst_offset, - uint32_t byte_count, - struct reservation_object *resv, - struct fence **fence); -int amdgpu_mmap(struct file *filp, struct vm_area_struct *vma); - struct amdgpu_bo_list_entry { struct amdgpu_bo *robj; struct ttm_validate_buffer tv; @@ -498,10 +463,12 @@ struct amdgpu_bo { struct amdgpu_device *adev; 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) @@ -677,6 +644,8 @@ struct amdgpu_mc { uint32_t fw_version; struct amdgpu_irq_src vm_fault; uint32_t vram_type; + uint32_t srbm_soft_reset; + struct amdgpu_mode_mc_save save; }; /* @@ -721,10 +690,11 @@ void amdgpu_doorbell_get_kfd_info(struct amdgpu_device *adev, */ struct amdgpu_flip_work { - struct work_struct flip_work; + struct delayed_work flip_work; struct work_struct unpin_work; struct amdgpu_device *adev; int crtc_id; + u32 target_vblank; uint64_t base; struct drm_pending_vblank_event *event; struct amdgpu_bo *old_rbo; @@ -815,13 +785,17 @@ struct amdgpu_ring { /* maximum number of VMIDs */ #define AMDGPU_NUM_VM 16 +/* Maximum number of PTEs the hardware can write with one command */ +#define AMDGPU_VM_MAX_UPDATE_SIZE 0x3FFFF + /* number of entries in page table */ #define AMDGPU_VM_PTE_COUNT (1 << amdgpu_vm_block_size) /* PTBs (Page Table Blocks) need to be aligned to 32K */ #define AMDGPU_VM_PTB_ALIGN_SIZE 32768 -#define AMDGPU_VM_PTB_ALIGN_MASK (AMDGPU_VM_PTB_ALIGN_SIZE - 1) -#define AMDGPU_VM_PTB_ALIGN(a) (((a) + AMDGPU_VM_PTB_ALIGN_MASK) & ~AMDGPU_VM_PTB_ALIGN_MASK) + +/* LOG2 number of continuous pages for the fragment field */ +#define AMDGPU_LOG2_PAGES_PER_FRAG 4 #define AMDGPU_PTE_VALID (1 << 0) #define AMDGPU_PTE_SYSTEM (1 << 1) @@ -833,10 +807,7 @@ struct amdgpu_ring { #define AMDGPU_PTE_READABLE (1 << 5) #define AMDGPU_PTE_WRITEABLE (1 << 6) -/* PTE (Page Table Entry) fragment field for different page sizes */ -#define AMDGPU_PTE_FRAG_4KB (0 << 7) -#define AMDGPU_PTE_FRAG_64KB (4 << 7) -#define AMDGPU_LOG2_PAGES_PER_FRAG 4 +#define AMDGPU_PTE_FRAG(x) ((x & 0x1f) << 7) /* How to programm VM fault handling */ #define AMDGPU_VM_FAULT_STOP_NEVER 0 @@ -846,6 +817,7 @@ struct amdgpu_ring { struct amdgpu_vm_pt { struct amdgpu_bo_list_entry entry; uint64_t addr; + uint64_t shadow_addr; }; struct amdgpu_vm { @@ -948,7 +920,6 @@ int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring, struct amdgpu_job *job); int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job); void amdgpu_vm_reset_id(struct amdgpu_device *adev, unsigned vm_id); -uint64_t amdgpu_vm_map_gart(const dma_addr_t *pages_addr, uint64_t addr); int amdgpu_vm_update_page_directory(struct amdgpu_device *adev, struct amdgpu_vm *vm); int amdgpu_vm_clear_freed(struct amdgpu_device *adev, @@ -957,7 +928,7 @@ int amdgpu_vm_clear_invalids(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, - struct ttm_mem_reg *mem); + bool clear); void amdgpu_vm_bo_invalidate(struct amdgpu_device *adev, struct amdgpu_bo *bo); struct amdgpu_bo_va *amdgpu_vm_bo_find(struct amdgpu_vm *vm, @@ -1195,6 +1166,10 @@ struct amdgpu_gfx { unsigned ce_ram_size; struct amdgpu_cu_info cu_info; const struct amdgpu_gfx_funcs *funcs; + + /* reset mask */ + uint32_t grbm_soft_reset; + uint32_t srbm_soft_reset; }; int amdgpu_ib_get(struct amdgpu_device *adev, struct amdgpu_vm *vm, @@ -1683,6 +1658,7 @@ struct amdgpu_uvd { bool address_64_bit; bool use_ctx_buf; struct amd_sched_entity entity; + uint32_t srbm_soft_reset; }; /* @@ -1709,6 +1685,7 @@ struct amdgpu_vce { struct amdgpu_irq_src irq; unsigned harvest_config; struct amd_sched_entity entity; + uint32_t srbm_soft_reset; }; /* @@ -1729,6 +1706,7 @@ struct amdgpu_sdma { struct amdgpu_irq_src trap_irq; struct amdgpu_irq_src illegal_inst_irq; int num_instances; + uint32_t srbm_soft_reset; }; /* @@ -1956,6 +1934,7 @@ struct amdgpu_ip_block_status { bool valid; bool sw; bool hw; + bool hang; }; struct amdgpu_device { @@ -2055,6 +2034,7 @@ struct amdgpu_device { atomic_t gpu_reset_counter; /* display */ + bool enable_virtual_display; struct amdgpu_mode_info mode_info; struct work_struct hotplug_work; struct amdgpu_irq_src crtc_irq; @@ -2117,6 +2097,10 @@ struct amdgpu_device { struct kfd_dev *kfd; struct amdgpu_virtualization virtualization; + + /* link all shadow bo */ + struct list_head shadow_list; + struct mutex shadow_list_lock; }; bool amdgpu_device_is_px(struct drm_device *dev); @@ -2192,6 +2176,9 @@ void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v); #define REG_GET_FIELD(value, reg, field) \ (((value) & REG_FIELD_MASK(reg, field)) >> REG_FIELD_SHIFT(reg, field)) +#define WREG32_FIELD(reg, field, val) \ + WREG32(mm##reg, (RREG32(mm##reg) & ~REG_FIELD_MASK(reg, field)) | (val) << REG_FIELD_SHIFT(reg, field)) + /* * BIOS helpers. */ @@ -2242,7 +2229,7 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring) #define amdgpu_gart_flush_gpu_tlb(adev, vmid) (adev)->gart.gart_funcs->flush_gpu_tlb((adev), (vmid)) #define amdgpu_gart_set_pte_pde(adev, pt, idx, addr, flags) (adev)->gart.gart_funcs->set_pte_pde((adev), (pt), (idx), (addr), (flags)) #define amdgpu_vm_copy_pte(adev, ib, pe, src, count) ((adev)->vm_manager.vm_pte_funcs->copy_pte((ib), (pe), (src), (count))) -#define amdgpu_vm_write_pte(adev, ib, pa, pe, addr, count, incr, flags) ((adev)->vm_manager.vm_pte_funcs->write_pte((ib), (pa), (pe), (addr), (count), (incr), (flags))) +#define amdgpu_vm_write_pte(adev, ib, pe, value, count, incr) ((adev)->vm_manager.vm_pte_funcs->write_pte((ib), (pe), (value), (count), (incr))) #define amdgpu_vm_set_pte_pde(adev, ib, pe, addr, count, incr, flags) ((adev)->vm_manager.vm_pte_funcs->set_pte_pde((ib), (pe), (addr), (count), (incr), (flags))) #define amdgpu_ring_parse_cs(r, p, ib) ((r)->funcs->parse_cs((p), (ib))) #define amdgpu_ring_test_ring(r) (r)->funcs->test_ring((r)) @@ -2387,6 +2374,7 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring) /* Common functions */ int amdgpu_gpu_reset(struct amdgpu_device *adev); +bool amdgpu_need_backup(struct amdgpu_device *adev); void amdgpu_pci_config_reset(struct amdgpu_device *adev); bool amdgpu_card_posted(struct amdgpu_device *adev); void amdgpu_update_display_priority(struct amdgpu_device *adev); @@ -2412,6 +2400,8 @@ uint32_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm, void amdgpu_vram_location(struct amdgpu_device *adev, struct amdgpu_mc *mc, u64 base); void amdgpu_gtt_location(struct amdgpu_device *adev, struct amdgpu_mc *mc); void amdgpu_ttm_set_active_vram_size(struct amdgpu_device *adev, u64 size); +u64 amdgpu_ttm_get_gtt_mem_size(struct amdgpu_device *adev); +int amdgpu_ttm_global_init(struct amdgpu_device *adev); void amdgpu_program_register_sequence(struct amdgpu_device *adev, const u32 *registers, const u32 array_size); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c index 983175363b06..1b621160b52e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c @@ -259,6 +259,33 @@ static const int object_connector_convert[] = { DRM_MODE_CONNECTOR_Unknown }; +bool amdgpu_atombios_has_dce_engine_info(struct amdgpu_device *adev) +{ + struct amdgpu_mode_info *mode_info = &adev->mode_info; + struct atom_context *ctx = mode_info->atom_context; + int index = GetIndexIntoMasterTable(DATA, Object_Header); + u16 size, data_offset; + u8 frev, crev; + ATOM_DISPLAY_OBJECT_PATH_TABLE *path_obj; + ATOM_OBJECT_HEADER *obj_header; + + if (!amdgpu_atom_parse_data_header(ctx, index, &size, &frev, &crev, &data_offset)) + return false; + + if (crev < 2) + return false; + + obj_header = (ATOM_OBJECT_HEADER *) (ctx->bios + data_offset); + path_obj = (ATOM_DISPLAY_OBJECT_PATH_TABLE *) + (ctx->bios + data_offset + + le16_to_cpu(obj_header->usDisplayPathTableOffset)); + + if (path_obj->ucNumOfDispPath) + return true; + else + return false; +} + bool amdgpu_atombios_get_connector_info_from_object_table(struct amdgpu_device *adev) { struct amdgpu_mode_info *mode_info = &adev->mode_info; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h index 8c2e69661799..15dd43ec38bb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h @@ -140,6 +140,8 @@ struct amdgpu_i2c_bus_rec amdgpu_atombios_lookup_i2c_gpio(struct amdgpu_device * uint8_t id); void amdgpu_atombios_i2c_init(struct amdgpu_device *adev); +bool amdgpu_atombios_has_dce_engine_info(struct amdgpu_device *adev); + bool amdgpu_atombios_get_connector_info_from_object_table(struct amdgpu_device *adev); int amdgpu_atombios_get_clock_info(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c index 33e47a43ae32..345305235349 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c @@ -39,7 +39,8 @@ static int amdgpu_benchmark_do_move(struct amdgpu_device *adev, unsigned size, start_jiffies = jiffies; for (i = 0; i < n; i++) { struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring; - r = amdgpu_copy_buffer(ring, saddr, daddr, size, NULL, &fence); + r = amdgpu_copy_buffer(ring, saddr, daddr, size, NULL, &fence, + false); if (r) goto exit_do_move; r = fence_wait(fence, false); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c index ff0b55a65ca3..319a5e1d9389 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c @@ -1504,6 +1504,86 @@ static const struct drm_connector_funcs amdgpu_connector_edp_funcs = { .force = amdgpu_connector_dvi_force, }; +static struct drm_encoder * +amdgpu_connector_virtual_encoder(struct drm_connector *connector) +{ + int enc_id = connector->encoder_ids[0]; + struct drm_encoder *encoder; + int i; + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + if (connector->encoder_ids[i] == 0) + break; + + encoder = drm_encoder_find(connector->dev, connector->encoder_ids[i]); + if (!encoder) + continue; + + if (encoder->encoder_type == DRM_MODE_ENCODER_VIRTUAL) + return encoder; + } + + /* pick the first one */ + if (enc_id) + return drm_encoder_find(connector->dev, enc_id); + return NULL; +} + +static int amdgpu_connector_virtual_get_modes(struct drm_connector *connector) +{ + struct drm_encoder *encoder = amdgpu_connector_best_single_encoder(connector); + + if (encoder) { + amdgpu_connector_add_common_modes(encoder, connector); + } + + return 0; +} + +static int amdgpu_connector_virtual_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) +{ + return MODE_OK; +} + +int amdgpu_connector_virtual_dpms(struct drm_connector *connector, int mode) +{ + return 0; +} + +static enum drm_connector_status + +amdgpu_connector_virtual_detect(struct drm_connector *connector, bool force) +{ + return connector_status_connected; +} + +int amdgpu_connector_virtual_set_property(struct drm_connector *connector, + struct drm_property *property, + uint64_t val) +{ + return 0; +} + +static void amdgpu_connector_virtual_force(struct drm_connector *connector) +{ + return; +} + +static const struct drm_connector_helper_funcs amdgpu_connector_virtual_helper_funcs = { + .get_modes = amdgpu_connector_virtual_get_modes, + .mode_valid = amdgpu_connector_virtual_mode_valid, + .best_encoder = amdgpu_connector_virtual_encoder, +}; + +static const struct drm_connector_funcs amdgpu_connector_virtual_funcs = { + .dpms = amdgpu_connector_virtual_dpms, + .detect = amdgpu_connector_virtual_detect, + .fill_modes = drm_helper_probe_single_connector_modes, + .set_property = amdgpu_connector_virtual_set_property, + .destroy = amdgpu_connector_destroy, + .force = amdgpu_connector_virtual_force, +}; + void amdgpu_connector_add(struct amdgpu_device *adev, uint32_t connector_id, @@ -1888,6 +1968,17 @@ amdgpu_connector_add(struct amdgpu_device *adev, connector->interlace_allowed = false; connector->doublescan_allowed = false; break; + case DRM_MODE_CONNECTOR_VIRTUAL: + amdgpu_dig_connector = kzalloc(sizeof(struct amdgpu_connector_atom_dig), GFP_KERNEL); + if (!amdgpu_dig_connector) + goto failed; + amdgpu_connector->con_priv = amdgpu_dig_connector; + drm_connector_init(dev, &amdgpu_connector->base, &amdgpu_connector_virtual_funcs, connector_type); + drm_connector_helper_add(&amdgpu_connector->base, &amdgpu_connector_virtual_helper_funcs); + subpixel_order = SubPixelHorizontalRGB; + connector->interlace_allowed = false; + connector->doublescan_allowed = false; + break; } } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 0307ff5887c5..d80e5d3a4add 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -287,18 +287,56 @@ static u64 amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev) return max(bytes_moved_threshold, 1024*1024ull); } +static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p, + struct amdgpu_bo *bo) +{ + u64 initial_bytes_moved; + uint32_t domain; + int r; + + if (bo->pin_count) + return 0; + + /* Avoid moving this one if we have moved too many buffers + * for this IB already. + * + * Note that this allows moving at least one buffer of + * any size, because it doesn't take the current "bo" + * into account. We don't want to disallow buffer moves + * completely. + */ + if (p->bytes_moved <= p->bytes_moved_threshold) + domain = bo->prefered_domains; + else + domain = bo->allowed_domains; + +retry: + amdgpu_ttm_placement_from_domain(bo, domain); + initial_bytes_moved = atomic64_read(&bo->adev->num_bytes_moved); + r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false); + p->bytes_moved += atomic64_read(&bo->adev->num_bytes_moved) - + initial_bytes_moved; + + if (unlikely(r)) { + if (r != -ERESTARTSYS && domain != bo->allowed_domains) { + domain = bo->allowed_domains; + goto retry; + } + } + + return r; +} + int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p, struct list_head *validated) { struct amdgpu_bo_list_entry *lobj; - u64 initial_bytes_moved; int r; list_for_each_entry(lobj, validated, tv.head) { struct amdgpu_bo *bo = lobj->robj; bool binding_userptr = false; struct mm_struct *usermm; - uint32_t domain; usermm = amdgpu_ttm_tt_get_usermm(bo->tbo.ttm); if (usermm && usermm != current->mm) @@ -313,35 +351,13 @@ int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p, binding_userptr = true; } - if (bo->pin_count) - continue; - - /* Avoid moving this one if we have moved too many buffers - * for this IB already. - * - * Note that this allows moving at least one buffer of - * any size, because it doesn't take the current "bo" - * into account. We don't want to disallow buffer moves - * completely. - */ - if (p->bytes_moved <= p->bytes_moved_threshold) - domain = bo->prefered_domains; - else - domain = bo->allowed_domains; - - retry: - amdgpu_ttm_placement_from_domain(bo, domain); - initial_bytes_moved = atomic64_read(&bo->adev->num_bytes_moved); - r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false); - p->bytes_moved += atomic64_read(&bo->adev->num_bytes_moved) - - initial_bytes_moved; - - if (unlikely(r)) { - if (r != -ERESTARTSYS && domain != bo->allowed_domains) { - domain = bo->allowed_domains; - goto retry; - } + r = amdgpu_cs_bo_validate(p, bo); + if (r) return r; + if (bo->shadow) { + r = amdgpu_cs_bo_validate(p, bo); + if (r) + return r; } if (binding_userptr) { @@ -386,8 +402,10 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p, r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true, &duplicates); - if (unlikely(r != 0)) + if (unlikely(r != 0)) { + DRM_ERROR("ttm_eu_reserve_buffers failed.\n"); goto error_free_pages; + } /* Without a BO list we don't have userptr BOs */ if (!p->bo_list) @@ -427,9 +445,10 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p, /* Unreserve everything again. */ ttm_eu_backoff_reservation(&p->ticket, &p->validated); - /* We tried to often, just abort */ + /* We tried too many times, just abort */ if (!--tries) { r = -EDEADLK; + DRM_ERROR("deadlock in %s\n", __func__); goto error_free_pages; } @@ -441,11 +460,13 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p, sizeof(struct page*)); if (!e->user_pages) { r = -ENOMEM; + DRM_ERROR("calloc failure in %s\n", __func__); goto error_free_pages; } r = amdgpu_ttm_tt_get_user_pages(ttm, e->user_pages); if (r) { + DRM_ERROR("amdgpu_ttm_tt_get_user_pages failed.\n"); drm_free_large(e->user_pages); e->user_pages = NULL; goto error_free_pages; @@ -462,12 +483,16 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p, p->bytes_moved = 0; r = amdgpu_cs_list_validate(p, &duplicates); - if (r) + if (r) { + DRM_ERROR("amdgpu_cs_list_validate(duplicates) failed.\n"); goto error_validate; + } r = amdgpu_cs_list_validate(p, &p->validated); - if (r) + if (r) { + DRM_ERROR("amdgpu_cs_list_validate(validated) failed.\n"); goto error_validate; + } fpriv->vm.last_eviction_counter = atomic64_read(&p->adev->num_evictions); @@ -617,7 +642,7 @@ static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p, if (bo_va == NULL) continue; - r = amdgpu_vm_bo_update(adev, bo_va, &bo->tbo.mem); + r = amdgpu_vm_bo_update(adev, bo_va, false); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index df7ab2458e50..c38dc47cd767 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -46,6 +46,7 @@ #endif #include "vi.h" #include "bif/bif_4_1_d.h" +#include <linux/pci.h> static int amdgpu_debugfs_regs_init(struct amdgpu_device *adev); static void amdgpu_debugfs_regs_cleanup(struct amdgpu_device *adev); @@ -1181,10 +1182,38 @@ int amdgpu_ip_block_version_cmp(struct amdgpu_device *adev, return 1; } +static void amdgpu_whether_enable_virtual_display(struct amdgpu_device *adev) +{ + adev->enable_virtual_display = false; + + if (amdgpu_virtual_display) { + struct drm_device *ddev = adev->ddev; + const char *pci_address_name = pci_name(ddev->pdev); + char *pciaddstr, *pciaddstr_tmp, *pciaddname; + + pciaddstr = kstrdup(amdgpu_virtual_display, GFP_KERNEL); + pciaddstr_tmp = pciaddstr; + while ((pciaddname = strsep(&pciaddstr_tmp, ";"))) { + if (!strcmp(pci_address_name, pciaddname)) { + adev->enable_virtual_display = true; + break; + } + } + + DRM_INFO("virtual display string:%s, %s:virtual_display:%d\n", + amdgpu_virtual_display, pci_address_name, + adev->enable_virtual_display); + + kfree(pciaddstr); + } +} + static int amdgpu_early_init(struct amdgpu_device *adev) { int i, r; + amdgpu_whether_enable_virtual_display(adev); + switch (adev->asic_type) { case CHIP_TOPAZ: case CHIP_TONGA: @@ -1521,6 +1550,9 @@ int amdgpu_device_init(struct amdgpu_device *adev, spin_lock_init(&adev->gc_cac_idx_lock); spin_lock_init(&adev->audio_endpt_idx_lock); + INIT_LIST_HEAD(&adev->shadow_list); + mutex_init(&adev->shadow_list_lock); + adev->rmmio_base = pci_resource_start(adev->pdev, 5); adev->rmmio_size = pci_resource_len(adev->pdev, 5); adev->rmmio = ioremap(adev->rmmio_base, adev->rmmio_size); @@ -1937,6 +1969,126 @@ int amdgpu_resume_kms(struct drm_device *dev, bool resume, bool fbcon) return 0; } +static bool amdgpu_check_soft_reset(struct amdgpu_device *adev) +{ + int i; + bool asic_hang = false; + + for (i = 0; i < adev->num_ip_blocks; i++) { + if (!adev->ip_block_status[i].valid) + continue; + if (adev->ip_blocks[i].funcs->check_soft_reset) + adev->ip_blocks[i].funcs->check_soft_reset(adev); + if (adev->ip_block_status[i].hang) { + DRM_INFO("IP block:%d is hang!\n", i); + asic_hang = true; + } + } + return asic_hang; +} + +int amdgpu_pre_soft_reset(struct amdgpu_device *adev) +{ + int i, r = 0; + + for (i = 0; i < adev->num_ip_blocks; i++) { + if (!adev->ip_block_status[i].valid) + continue; + if (adev->ip_block_status[i].hang && + adev->ip_blocks[i].funcs->pre_soft_reset) { + r = adev->ip_blocks[i].funcs->pre_soft_reset(adev); + if (r) + return r; + } + } + + return 0; +} + +static bool amdgpu_need_full_reset(struct amdgpu_device *adev) +{ + if (adev->ip_block_status[AMD_IP_BLOCK_TYPE_GMC].hang || + adev->ip_block_status[AMD_IP_BLOCK_TYPE_SMC].hang || + adev->ip_block_status[AMD_IP_BLOCK_TYPE_ACP].hang || + adev->ip_block_status[AMD_IP_BLOCK_TYPE_DCE].hang) { + DRM_INFO("Some block need full reset!\n"); + return true; + } + return false; +} + +static int amdgpu_soft_reset(struct amdgpu_device *adev) +{ + int i, r = 0; + + for (i = 0; i < adev->num_ip_blocks; i++) { + if (!adev->ip_block_status[i].valid) + continue; + if (adev->ip_block_status[i].hang && + adev->ip_blocks[i].funcs->soft_reset) { + r = adev->ip_blocks[i].funcs->soft_reset(adev); + if (r) + return r; + } + } + + return 0; +} + +static int amdgpu_post_soft_reset(struct amdgpu_device *adev) +{ + int i, r = 0; + + for (i = 0; i < adev->num_ip_blocks; i++) { + if (!adev->ip_block_status[i].valid) + continue; + if (adev->ip_block_status[i].hang && + adev->ip_blocks[i].funcs->post_soft_reset) + r = adev->ip_blocks[i].funcs->post_soft_reset(adev); + if (r) + return r; + } + + return 0; +} + +bool amdgpu_need_backup(struct amdgpu_device *adev) +{ + if (adev->flags & AMD_IS_APU) + return false; + + return amdgpu_lockup_timeout > 0 ? true : false; +} + +static int amdgpu_recover_vram_from_shadow(struct amdgpu_device *adev, + struct amdgpu_ring *ring, + struct amdgpu_bo *bo, + struct fence **fence) +{ + uint32_t domain; + int r; + + if (!bo->shadow) + return 0; + + r = amdgpu_bo_reserve(bo, false); + if (r) + return r; + domain = amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type); + /* if bo has been evicted, then no need to recover */ + if (domain == AMDGPU_GEM_DOMAIN_VRAM) { + r = amdgpu_bo_restore_from_shadow(adev, ring, bo, + NULL, fence, true); + if (r) { + DRM_ERROR("recover page table failed!\n"); + goto err; + } + } +err: + amdgpu_bo_unreserve(bo); + return r; +} + /** * amdgpu_gpu_reset - reset the asic * @@ -1949,6 +2101,12 @@ int amdgpu_gpu_reset(struct amdgpu_device *adev) { int i, r; int resched; + bool need_full_reset; + + if (!amdgpu_check_soft_reset(adev)) { + DRM_INFO("No hardware hang detected. Did some blocks stall?\n"); + return 0; + } atomic_inc(&adev->gpu_reset_counter); @@ -1967,40 +2125,88 @@ int amdgpu_gpu_reset(struct amdgpu_device *adev) /* after all hw jobs are reset, hw fence is meaningless, so force_completion */ amdgpu_fence_driver_force_completion(adev); - /* save scratch */ - amdgpu_atombios_scratch_regs_save(adev); - r = amdgpu_suspend(adev); + need_full_reset = amdgpu_need_full_reset(adev); -retry: - /* Disable fb access */ - if (adev->mode_info.num_crtc) { - struct amdgpu_mode_mc_save save; - amdgpu_display_stop_mc_access(adev, &save); - amdgpu_wait_for_idle(adev, AMD_IP_BLOCK_TYPE_GMC); + if (!need_full_reset) { + amdgpu_pre_soft_reset(adev); + r = amdgpu_soft_reset(adev); + amdgpu_post_soft_reset(adev); + if (r || amdgpu_check_soft_reset(adev)) { + DRM_INFO("soft reset failed, will fallback to full reset!\n"); + need_full_reset = true; + } } - r = amdgpu_asic_reset(adev); - /* post card */ - amdgpu_atom_asic_init(adev->mode_info.atom_context); + if (need_full_reset) { + /* save scratch */ + amdgpu_atombios_scratch_regs_save(adev); + r = amdgpu_suspend(adev); - if (!r) { - dev_info(adev->dev, "GPU reset succeeded, trying to resume\n"); - r = amdgpu_resume(adev); +retry: + /* Disable fb access */ + if (adev->mode_info.num_crtc) { + struct amdgpu_mode_mc_save save; + amdgpu_display_stop_mc_access(adev, &save); + amdgpu_wait_for_idle(adev, AMD_IP_BLOCK_TYPE_GMC); + } + + r = amdgpu_asic_reset(adev); + /* post card */ + amdgpu_atom_asic_init(adev->mode_info.atom_context); + + if (!r) { + dev_info(adev->dev, "GPU reset succeeded, trying to resume\n"); + r = amdgpu_resume(adev); + } + /* restore scratch */ + amdgpu_atombios_scratch_regs_restore(adev); } - /* restore scratch */ - amdgpu_atombios_scratch_regs_restore(adev); if (!r) { + amdgpu_irq_gpu_reset_resume_helper(adev); r = amdgpu_ib_ring_tests(adev); if (r) { dev_err(adev->dev, "ib ring test failed (%d).\n", r); r = amdgpu_suspend(adev); + need_full_reset = true; goto retry; } + /** + * recovery vm page tables, since we cannot depend on VRAM is + * consistent after gpu full reset. + */ + if (need_full_reset && amdgpu_need_backup(adev)) { + struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring; + struct amdgpu_bo *bo, *tmp; + struct fence *fence = NULL, *next = NULL; + + DRM_INFO("recover vram bo from shadow\n"); + mutex_lock(&adev->shadow_list_lock); + list_for_each_entry_safe(bo, tmp, &adev->shadow_list, shadow_list) { + amdgpu_recover_vram_from_shadow(adev, ring, bo, &next); + if (fence) { + r = fence_wait(fence, false); + if (r) { + WARN(r, "recovery from shadow isn't comleted\n"); + break; + } + } + fence_put(fence); + fence = next; + } + mutex_unlock(&adev->shadow_list_lock); + if (fence) { + r = fence_wait(fence, false); + if (r) + WARN(r, "recovery from shadow isn't comleted\n"); + } + fence_put(fence); + } for (i = 0; i < AMDGPU_MAX_RINGS; ++i) { struct amdgpu_ring *ring = adev->rings[i]; if (!ring) continue; + amd_sched_job_recovery(&ring->sched); kthread_unpark(ring->sched.thread); } @@ -2020,7 +2226,6 @@ retry: /* bad news, how to tell it to userspace ? */ dev_info(adev->dev, "GPU reset failed\n"); } - amdgpu_irq_gpu_reset_resume_helper(adev); return r; } @@ -2178,22 +2383,26 @@ static ssize_t amdgpu_debugfs_regs_read(struct file *f, char __user *buf, struct amdgpu_device *adev = f->f_inode->i_private; ssize_t result = 0; int r; - bool use_bank; + bool pm_pg_lock, use_bank; unsigned instance_bank, sh_bank, se_bank; if (size & 0x3 || *pos & 0x3) return -EINVAL; + /* are we reading registers for which a PG lock is necessary? */ + pm_pg_lock = (*pos >> 23) & 1; + if (*pos & (1ULL << 62)) { se_bank = (*pos >> 24) & 0x3FF; sh_bank = (*pos >> 34) & 0x3FF; instance_bank = (*pos >> 44) & 0x3FF; use_bank = 1; - *pos &= 0xFFFFFF; } else { use_bank = 0; } + *pos &= 0x3FFFF; + if (use_bank) { if (sh_bank >= adev->gfx.config.max_sh_per_se || se_bank >= adev->gfx.config.max_shader_engines) @@ -2203,6 +2412,9 @@ static ssize_t amdgpu_debugfs_regs_read(struct file *f, char __user *buf, sh_bank, instance_bank); } + if (pm_pg_lock) + mutex_lock(&adev->pm.mutex); + while (size) { uint32_t value; @@ -2228,6 +2440,9 @@ end: mutex_unlock(&adev->grbm_idx_mutex); } + if (pm_pg_lock) + mutex_unlock(&adev->pm.mutex); + return result; } @@ -2443,7 +2658,7 @@ static ssize_t amdgpu_debugfs_gca_config_read(struct file *f, char __user *buf, return -ENOMEM; /* version, increment each time something is added */ - config[no_regs++] = 0; + config[no_regs++] = 2; config[no_regs++] = adev->gfx.config.max_shader_engines; config[no_regs++] = adev->gfx.config.max_tile_pipes; config[no_regs++] = adev->gfx.config.max_cu_per_sh; @@ -2468,6 +2683,15 @@ static ssize_t amdgpu_debugfs_gca_config_read(struct file *f, char __user *buf, config[no_regs++] = adev->gfx.config.gb_addr_config; config[no_regs++] = adev->gfx.config.num_rbs; + /* rev==1 */ + config[no_regs++] = adev->rev_id; + config[no_regs++] = adev->pg_flags; + config[no_regs++] = adev->cg_flags; + + /* rev==2 */ + config[no_regs++] = adev->family; + config[no_regs++] = adev->external_rev_id; + while (size && (*pos < no_regs * 4)) { uint32_t value; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c index 76f96028313d..9af8d3c7ae8b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c @@ -41,7 +41,7 @@ static void amdgpu_flip_callback(struct fence *f, struct fence_cb *cb) container_of(cb, struct amdgpu_flip_work, cb); fence_put(f); - schedule_work(&work->flip_work); + schedule_work(&work->flip_work.work); } static bool amdgpu_flip_handle_fence(struct amdgpu_flip_work *work, @@ -63,16 +63,17 @@ static bool amdgpu_flip_handle_fence(struct amdgpu_flip_work *work, static void amdgpu_flip_work_func(struct work_struct *__work) { + struct delayed_work *delayed_work = + container_of(__work, struct delayed_work, work); struct amdgpu_flip_work *work = - container_of(__work, struct amdgpu_flip_work, flip_work); + container_of(delayed_work, struct amdgpu_flip_work, flip_work); struct amdgpu_device *adev = work->adev; struct amdgpu_crtc *amdgpuCrtc = adev->mode_info.crtcs[work->crtc_id]; struct drm_crtc *crtc = &amdgpuCrtc->base; unsigned long flags; - unsigned i, repcnt = 4; - int vpos, hpos, stat, min_udelay = 0; - struct drm_vblank_crtc *vblank = &crtc->dev->vblank[work->crtc_id]; + unsigned i; + int vpos, hpos; if (amdgpu_flip_handle_fence(work, &work->excl)) return; @@ -81,55 +82,23 @@ static void amdgpu_flip_work_func(struct work_struct *__work) if (amdgpu_flip_handle_fence(work, &work->shared[i])) return; - /* We borrow the event spin lock for protecting flip_status */ - spin_lock_irqsave(&crtc->dev->event_lock, flags); - - /* If this happens to execute within the "virtually extended" vblank - * interval before the start of the real vblank interval then it needs - * to delay programming the mmio flip until the real vblank is entered. - * This prevents completing a flip too early due to the way we fudge - * our vblank counter and vblank timestamps in order to work around the - * problem that the hw fires vblank interrupts before actual start of - * vblank (when line buffer refilling is done for a frame). It - * complements the fudging logic in amdgpu_get_crtc_scanoutpos() for - * timestamping and amdgpu_get_vblank_counter_kms() for vblank counts. - * - * In practice this won't execute very often unless on very fast - * machines because the time window for this to happen is very small. + /* Wait until we're out of the vertical blank period before the one + * targeted by the flip */ - while (amdgpuCrtc->enabled && --repcnt) { - /* GET_DISTANCE_TO_VBLANKSTART returns distance to real vblank - * start in hpos, and to the "fudged earlier" vblank start in - * vpos. - */ - stat = amdgpu_get_crtc_scanoutpos(adev->ddev, work->crtc_id, - GET_DISTANCE_TO_VBLANKSTART, - &vpos, &hpos, NULL, NULL, - &crtc->hwmode); - - if ((stat & (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_ACCURATE)) != - (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_ACCURATE) || - !(vpos >= 0 && hpos <= 0)) - break; - - /* Sleep at least until estimated real start of hw vblank */ - min_udelay = (-hpos + 1) * max(vblank->linedur_ns / 1000, 5); - if (min_udelay > vblank->framedur_ns / 2000) { - /* Don't wait ridiculously long - something is wrong */ - repcnt = 0; - break; - } - spin_unlock_irqrestore(&crtc->dev->event_lock, flags); - usleep_range(min_udelay, 2 * min_udelay); - spin_lock_irqsave(&crtc->dev->event_lock, flags); + if (amdgpuCrtc->enabled && + (amdgpu_get_crtc_scanoutpos(adev->ddev, work->crtc_id, 0, + &vpos, &hpos, NULL, NULL, + &crtc->hwmode) + & (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_IN_VBLANK)) == + (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_IN_VBLANK) && + (int)(work->target_vblank - + amdgpu_get_vblank_counter_kms(adev->ddev, amdgpuCrtc->crtc_id)) > 0) { + schedule_delayed_work(&work->flip_work, usecs_to_jiffies(1000)); + return; } - if (!repcnt) - DRM_DEBUG_DRIVER("Delay problem on crtc %d: min_udelay %d, " - "framedur %d, linedur %d, stat %d, vpos %d, " - "hpos %d\n", work->crtc_id, min_udelay, - vblank->framedur_ns / 1000, - vblank->linedur_ns / 1000, stat, vpos, hpos); + /* We borrow the event spin lock for protecting flip_status */ + spin_lock_irqsave(&crtc->dev->event_lock, flags); /* Do the flip (mmio) */ adev->mode_info.funcs->page_flip(adev, work->crtc_id, work->base, work->async); @@ -169,10 +138,10 @@ static void amdgpu_unpin_work_func(struct work_struct *__work) kfree(work); } -int amdgpu_crtc_page_flip(struct drm_crtc *crtc, - struct drm_framebuffer *fb, - struct drm_pending_vblank_event *event, - uint32_t page_flip_flags) +int amdgpu_crtc_page_flip_target(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_pending_vblank_event *event, + uint32_t page_flip_flags, uint32_t target) { struct drm_device *dev = crtc->dev; struct amdgpu_device *adev = dev->dev_private; @@ -191,7 +160,7 @@ int amdgpu_crtc_page_flip(struct drm_crtc *crtc, if (work == NULL) return -ENOMEM; - INIT_WORK(&work->flip_work, amdgpu_flip_work_func); + INIT_DELAYED_WORK(&work->flip_work, amdgpu_flip_work_func); INIT_WORK(&work->unpin_work, amdgpu_unpin_work_func); work->event = event; @@ -237,12 +206,8 @@ int amdgpu_crtc_page_flip(struct drm_crtc *crtc, amdgpu_bo_unreserve(new_rbo); work->base = base; - - r = drm_crtc_vblank_get(crtc); - if (r) { - DRM_ERROR("failed to get vblank before flip\n"); - goto pflip_cleanup; - } + work->target_vblank = target - drm_crtc_vblank_count(crtc) + + amdgpu_get_vblank_counter_kms(dev, work->crtc_id); /* we borrow the event spin lock for protecting flip_wrok */ spin_lock_irqsave(&crtc->dev->event_lock, flags); @@ -250,7 +215,7 @@ int amdgpu_crtc_page_flip(struct drm_crtc *crtc, DRM_DEBUG_DRIVER("flip queue: crtc already busy\n"); spin_unlock_irqrestore(&crtc->dev->event_lock, flags); r = -EBUSY; - goto vblank_cleanup; + goto pflip_cleanup; } amdgpu_crtc->pflip_status = AMDGPU_FLIP_PENDING; @@ -262,12 +227,9 @@ int amdgpu_crtc_page_flip(struct drm_crtc *crtc, /* update crtc fb */ crtc->primary->fb = fb; spin_unlock_irqrestore(&crtc->dev->event_lock, flags); - amdgpu_flip_work_func(&work->flip_work); + amdgpu_flip_work_func(&work->flip_work.work); return 0; -vblank_cleanup: - drm_crtc_vblank_put(crtc); - pflip_cleanup: if (unlikely(amdgpu_bo_reserve(new_rbo, false) != 0)) { DRM_ERROR("failed to reserve new rbo in error path\n"); @@ -335,7 +297,7 @@ int amdgpu_crtc_set_config(struct drm_mode_set *set) return ret; } -static const char *encoder_names[38] = { +static const char *encoder_names[41] = { "NONE", "INTERNAL_LVDS", "INTERNAL_TMDS1", @@ -374,6 +336,9 @@ static const char *encoder_names[38] = { "TRAVIS", "INTERNAL_VCE", "INTERNAL_UNIPHY3", + "HDMI_ANX9805", + "INTERNAL_AMCLK", + "VIRTUAL", }; static const char *hpd_names[6] = { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index 11263c5b9967..7c911d0be2b3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -53,9 +53,11 @@ * - 3.2.0 - GFX8: Uses EOP_TC_WB_ACTION_EN, so UMDs don't have to do the same * at the end of IBs. * - 3.3.0 - Add VM support for UVD on supported hardware. + * - 3.4.0 - Add AMDGPU_INFO_NUM_EVICTIONS. + * - 3.5.0 - Add support for new UVD_NO_OP register. */ #define KMS_DRIVER_MAJOR 3 -#define KMS_DRIVER_MINOR 3 +#define KMS_DRIVER_MINOR 5 #define KMS_DRIVER_PATCHLEVEL 0 int amdgpu_vram_limit = 0; @@ -84,11 +86,13 @@ int amdgpu_sched_jobs = 32; int amdgpu_sched_hw_submission = 2; int amdgpu_powerplay = -1; int amdgpu_powercontainment = 1; +int amdgpu_sclk_deep_sleep_en = 1; unsigned amdgpu_pcie_gen_cap = 0; unsigned amdgpu_pcie_lane_cap = 0; unsigned amdgpu_cg_mask = 0xffffffff; unsigned amdgpu_pg_mask = 0xffffffff; char *amdgpu_disable_cu = NULL; +char *amdgpu_virtual_display = NULL; MODULE_PARM_DESC(vramlimit, "Restrict VRAM for testing, in megabytes"); module_param_named(vramlimit, amdgpu_vram_limit, int, 0600); @@ -170,6 +174,9 @@ MODULE_PARM_DESC(powercontainment, "Power Containment (1 = enable (default), 0 = module_param_named(powercontainment, amdgpu_powercontainment, int, 0444); #endif +MODULE_PARM_DESC(sclkdeepsleep, "SCLK Deep Sleep (1 = enable (default), 0 = disable)"); +module_param_named(sclkdeepsleep, amdgpu_sclk_deep_sleep_en, int, 0444); + MODULE_PARM_DESC(pcie_gen_cap, "PCIE Gen Caps (0: autodetect (default))"); module_param_named(pcie_gen_cap, amdgpu_pcie_gen_cap, uint, 0444); @@ -185,6 +192,9 @@ module_param_named(pg_mask, amdgpu_pg_mask, uint, 0444); MODULE_PARM_DESC(disable_cu, "Disable CUs (se.sh.cu,...)"); module_param_named(disable_cu, amdgpu_disable_cu, charp, 0444); +MODULE_PARM_DESC(virtual_display, "Enable virtual display feature (the virtual_display will be set like xxxx:xx:xx.x;xxxx:xx:xx.x)"); +module_param_named(virtual_display, amdgpu_virtual_display, charp, 0444); + static const struct pci_device_id pciidlist[] = { #ifdef CONFIG_DRM_AMDGPU_CIK /* Kaveri */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gds.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gds.h index 503d54098128..e73728d90388 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gds.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gds.h @@ -31,14 +31,6 @@ #define AMDGPU_GWS_SHIFT PAGE_SHIFT #define AMDGPU_OA_SHIFT PAGE_SHIFT -#define AMDGPU_PL_GDS TTM_PL_PRIV0 -#define AMDGPU_PL_GWS TTM_PL_PRIV1 -#define AMDGPU_PL_OA TTM_PL_PRIV2 - -#define AMDGPU_PL_FLAG_GDS TTM_PL_FLAG_PRIV0 -#define AMDGPU_PL_FLAG_GWS TTM_PL_FLAG_PRIV1 -#define AMDGPU_PL_FLAG_OA TTM_PL_FLAG_PRIV2 - struct amdgpu_ring; struct amdgpu_bo; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c index 31a676376d73..c93a92a840ea 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_i2c.c @@ -186,10 +186,8 @@ struct amdgpu_i2c_chan *amdgpu_i2c_create(struct drm_device *dev, "AMDGPU i2c hw bus %s", name); i2c->adapter.algo = &amdgpu_atombios_i2c_algo; ret = i2c_add_adapter(&i2c->adapter); - if (ret) { - DRM_ERROR("Failed to register hw i2c %s\n", name); + if (ret) goto out_free; - } } else { /* set the amdgpu bit adapter */ snprintf(i2c->adapter.name, sizeof(i2c->adapter.name), diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c index a31d7ef3032c..f5810f700668 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c @@ -142,7 +142,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs, } if (!ring->ready) { - dev_err(adev->dev, "couldn't schedule ib\n"); + dev_err(adev->dev, "couldn't schedule ib on ring <%s>\n", ring->name); return -EINVAL; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c index 534fc04e80fd..5ebb3f43feb6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c @@ -40,32 +40,15 @@ static int amdgpu_ih_ring_alloc(struct amdgpu_device *adev) /* Allocate ring buffer */ if (adev->irq.ih.ring_obj == NULL) { - r = amdgpu_bo_create(adev, adev->irq.ih.ring_size, - PAGE_SIZE, true, - AMDGPU_GEM_DOMAIN_GTT, 0, - NULL, NULL, &adev->irq.ih.ring_obj); + r = amdgpu_bo_create_kernel(adev, adev->irq.ih.ring_size, + PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT, + &adev->irq.ih.ring_obj, + &adev->irq.ih.gpu_addr, + (void **)&adev->irq.ih.ring); if (r) { DRM_ERROR("amdgpu: failed to create ih ring buffer (%d).\n", r); return r; } - r = amdgpu_bo_reserve(adev->irq.ih.ring_obj, false); - if (unlikely(r != 0)) - return r; - r = amdgpu_bo_pin(adev->irq.ih.ring_obj, - AMDGPU_GEM_DOMAIN_GTT, - &adev->irq.ih.gpu_addr); - if (r) { - amdgpu_bo_unreserve(adev->irq.ih.ring_obj); - DRM_ERROR("amdgpu: failed to pin ih ring buffer (%d).\n", r); - return r; - } - r = amdgpu_bo_kmap(adev->irq.ih.ring_obj, - (void **)&adev->irq.ih.ring); - amdgpu_bo_unreserve(adev->irq.ih.ring_obj); - if (r) { - DRM_ERROR("amdgpu: failed to map ih ring buffer (%d).\n", r); - return r; - } } return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h index 7ef09352e534..f016464035b8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h @@ -70,6 +70,7 @@ struct amdgpu_irq { /* gen irq stuff */ struct irq_domain *domain; /* GPU irq controller domain */ unsigned virq[AMDGPU_MAX_IRQ_SRC_ID]; + uint32_t srbm_soft_reset; }; void amdgpu_irq_preinstall(struct drm_device *dev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index d942654a1de0..b78e74048f3d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -292,14 +292,14 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file type = AMD_IP_BLOCK_TYPE_UVD; ring_mask = adev->uvd.ring.ready ? 1 : 0; ib_start_alignment = AMDGPU_GPU_PAGE_SIZE; - ib_size_alignment = 8; + ib_size_alignment = 16; break; case AMDGPU_HW_IP_VCE: type = AMD_IP_BLOCK_TYPE_VCE; for (i = 0; i < AMDGPU_MAX_VCE_RINGS; i++) ring_mask |= ((adev->vce.ring[i].ready ? 1 : 0) << i); ib_start_alignment = AMDGPU_GPU_PAGE_SIZE; - ib_size_alignment = 8; + ib_size_alignment = 1; break; default: return -EINVAL; @@ -373,6 +373,9 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file case AMDGPU_INFO_NUM_BYTES_MOVED: ui64 = atomic64_read(&adev->num_bytes_moved); return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0; + case AMDGPU_INFO_NUM_EVICTIONS: + ui64 = atomic64_read(&adev->num_evictions); + return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0; case AMDGPU_INFO_VRAM_USAGE: ui64 = atomic64_read(&adev->vram_usage); return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h index 6b1d7d306564..7b0eff7d060b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h @@ -39,6 +39,8 @@ #include <drm/drm_plane_helper.h> #include <linux/i2c.h> #include <linux/i2c-algo-bit.h> +#include <linux/hrtimer.h> +#include "amdgpu_irq.h" struct amdgpu_bo; struct amdgpu_device; @@ -339,6 +341,8 @@ struct amdgpu_mode_info { int num_dig; /* number of dig blocks */ int disp_priority; const struct amdgpu_display_funcs *funcs; + struct hrtimer vblank_timer; + enum amdgpu_interrupt_state vsync_timer_enabled; }; #define AMDGPU_MAX_BL_LEVEL 0xFF @@ -587,10 +591,10 @@ int amdgpu_align_pitch(struct amdgpu_device *adev, int width, int bpp, bool tile void amdgpu_print_display_setup(struct drm_device *dev); int amdgpu_modeset_create_props(struct amdgpu_device *adev); int amdgpu_crtc_set_config(struct drm_mode_set *set); -int amdgpu_crtc_page_flip(struct drm_crtc *crtc, - struct drm_framebuffer *fb, - struct drm_pending_vblank_event *event, - uint32_t page_flip_flags); +int amdgpu_crtc_page_flip_target(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_pending_vblank_event *event, + uint32_t page_flip_flags, uint32_t target); extern const struct drm_mode_config_funcs amdgpu_mode_funcs; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c index 6f0873c75a25..b17734e0ecc8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c @@ -44,14 +44,13 @@ void amdgpu_ttm_fini(struct amdgpu_device *adev); static u64 amdgpu_get_vis_part_size(struct amdgpu_device *adev, struct ttm_mem_reg *mem) { - u64 ret = 0; - if (mem->start << PAGE_SHIFT < adev->mc.visible_vram_size) { - ret = (u64)((mem->start << PAGE_SHIFT) + mem->size) > - adev->mc.visible_vram_size ? - adev->mc.visible_vram_size - (mem->start << PAGE_SHIFT) : - mem->size; - } - return ret; + 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, @@ -99,6 +98,11 @@ static void amdgpu_ttm_bo_destroy(struct ttm_buffer_object *tbo) drm_gem_object_release(&bo->gem_base); amdgpu_bo_unref(&bo->parent); + if (!list_empty(&bo->shadow_list)) { + mutex_lock(&bo->adev->shadow_list_lock); + list_del_init(&bo->shadow_list); + mutex_unlock(&bo->adev->shadow_list_lock); + } kfree(bo->metadata); kfree(bo); } @@ -112,84 +116,93 @@ bool amdgpu_ttm_bo_is_amdgpu_bo(struct ttm_buffer_object *bo) static void amdgpu_ttm_placement_init(struct amdgpu_device *adev, struct ttm_placement *placement, - struct ttm_place *placements, + struct ttm_place *places, u32 domain, u64 flags) { - u32 c = 0, i; - - placement->placement = placements; - placement->busy_placement = placements; + u32 c = 0; if (domain & AMDGPU_GEM_DOMAIN_VRAM) { + unsigned visible_pfn = adev->mc.visible_vram_size >> PAGE_SHIFT; + if (flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS && - adev->mc.visible_vram_size < adev->mc.real_vram_size) { - placements[c].fpfn = - adev->mc.visible_vram_size >> PAGE_SHIFT; - placements[c++].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED | - TTM_PL_FLAG_VRAM | TTM_PL_FLAG_TOPDOWN; + !(flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) && + adev->mc.visible_vram_size < adev->mc.real_vram_size) { + places[c].fpfn = visible_pfn; + places[c].lpfn = 0; + places[c].flags = TTM_PL_FLAG_WC | + TTM_PL_FLAG_UNCACHED | TTM_PL_FLAG_VRAM | + TTM_PL_FLAG_TOPDOWN; + c++; } - placements[c].fpfn = 0; - placements[c++].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED | + + places[c].fpfn = 0; + places[c].lpfn = 0; + places[c].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED | TTM_PL_FLAG_VRAM; - if (!(flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)) - placements[c - 1].flags |= TTM_PL_FLAG_TOPDOWN; + if (flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) + places[c].lpfn = visible_pfn; + else + places[c].flags |= TTM_PL_FLAG_TOPDOWN; + c++; } if (domain & AMDGPU_GEM_DOMAIN_GTT) { - if (flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC) { - placements[c].fpfn = 0; - placements[c++].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_TT | + places[c].fpfn = 0; + places[c].lpfn = 0; + places[c].flags = TTM_PL_FLAG_TT; + if (flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC) + places[c].flags |= TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED; - } else { - placements[c].fpfn = 0; - placements[c++].flags = TTM_PL_FLAG_CACHED | TTM_PL_FLAG_TT; - } + else + places[c].flags |= TTM_PL_FLAG_CACHED; + c++; } if (domain & AMDGPU_GEM_DOMAIN_CPU) { - if (flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC) { - placements[c].fpfn = 0; - placements[c++].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_SYSTEM | + places[c].fpfn = 0; + places[c].lpfn = 0; + places[c].flags = TTM_PL_FLAG_SYSTEM; + if (flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC) + places[c].flags |= TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED; - } else { - placements[c].fpfn = 0; - placements[c++].flags = TTM_PL_FLAG_CACHED | TTM_PL_FLAG_SYSTEM; - } + else + places[c].flags |= TTM_PL_FLAG_CACHED; + c++; } if (domain & AMDGPU_GEM_DOMAIN_GDS) { - placements[c].fpfn = 0; - placements[c++].flags = TTM_PL_FLAG_UNCACHED | - AMDGPU_PL_FLAG_GDS; + places[c].fpfn = 0; + places[c].lpfn = 0; + places[c].flags = TTM_PL_FLAG_UNCACHED | AMDGPU_PL_FLAG_GDS; + c++; } + if (domain & AMDGPU_GEM_DOMAIN_GWS) { - placements[c].fpfn = 0; - placements[c++].flags = TTM_PL_FLAG_UNCACHED | - AMDGPU_PL_FLAG_GWS; + places[c].fpfn = 0; + places[c].lpfn = 0; + places[c].flags = TTM_PL_FLAG_UNCACHED | AMDGPU_PL_FLAG_GWS; + c++; } + if (domain & AMDGPU_GEM_DOMAIN_OA) { - placements[c].fpfn = 0; - placements[c++].flags = TTM_PL_FLAG_UNCACHED | - AMDGPU_PL_FLAG_OA; + places[c].fpfn = 0; + places[c].lpfn = 0; + places[c].flags = TTM_PL_FLAG_UNCACHED | AMDGPU_PL_FLAG_OA; + c++; } if (!c) { - placements[c].fpfn = 0; - placements[c++].flags = TTM_PL_MASK_CACHING | - TTM_PL_FLAG_SYSTEM; + places[c].fpfn = 0; + places[c].lpfn = 0; + places[c].flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM; + c++; } + placement->num_placement = c; - placement->num_busy_placement = c; + placement->placement = places; - for (i = 0; i < c; i++) { - if ((flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) && - (placements[i].flags & TTM_PL_FLAG_VRAM) && - !placements[i].fpfn) - placements[i].lpfn = - adev->mc.visible_vram_size >> PAGE_SHIFT; - else - placements[i].lpfn = 0; - } + placement->num_busy_placement = c; + placement->busy_placement = places; } void amdgpu_ttm_placement_from_domain(struct amdgpu_bo *rbo, u32 domain) @@ -211,6 +224,69 @@ static void amdgpu_fill_placement_to_bo(struct amdgpu_bo *bo, bo->placement.busy_placement = bo->placements; } +/** + * 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(adev, size, align, true, domain, + AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED, + NULL, NULL, bo_ptr); + if (r) { + dev_err(adev->dev, "(%d) failed to allocate kernel bo\n", r); + return r; + } + + r = amdgpu_bo_reserve(*bo_ptr, false); + if (r) { + dev_err(adev->dev, "(%d) failed to reserve kernel bo\n", r); + goto error_free; + } + + r = amdgpu_bo_pin(*bo_ptr, domain, gpu_addr); + if (r) { + dev_err(adev->dev, "(%d) kernel bo pin failed\n", r); + goto error_unreserve; + } + + if (cpu_addr) { + r = amdgpu_bo_kmap(*bo_ptr, cpu_addr); + if (r) { + dev_err(adev->dev, "(%d) kernel bo map failed\n", r); + goto error_unreserve; + } + } + + amdgpu_bo_unreserve(*bo_ptr); + + return 0; + +error_unreserve: + amdgpu_bo_unreserve(*bo_ptr); + +error_free: + amdgpu_bo_unref(bo_ptr); + + return r; +} + int amdgpu_bo_create_restricted(struct amdgpu_device *adev, unsigned long size, int byte_align, bool kernel, u32 domain, u64 flags, @@ -250,6 +326,7 @@ int amdgpu_bo_create_restricted(struct amdgpu_device *adev, } bo->adev = adev; INIT_LIST_HEAD(&bo->list); + INIT_LIST_HEAD(&bo->shadow_list); INIT_LIST_HEAD(&bo->va); bo->prefered_domains = domain & (AMDGPU_GEM_DOMAIN_VRAM | AMDGPU_GEM_DOMAIN_GTT | @@ -277,11 +354,79 @@ int amdgpu_bo_create_restricted(struct amdgpu_device *adev, if (unlikely(r != 0)) { return r; } + + if (flags & AMDGPU_GEM_CREATE_VRAM_CLEARED && + bo->tbo.mem.placement & TTM_PL_FLAG_VRAM) { + struct fence *fence; + + if (adev->mman.buffer_funcs_ring == NULL || + !adev->mman.buffer_funcs_ring->ready) { + r = -EBUSY; + goto fail_free; + } + + r = amdgpu_bo_reserve(bo, false); + if (unlikely(r != 0)) + goto fail_free; + + amdgpu_ttm_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_VRAM); + r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false); + if (unlikely(r != 0)) + goto fail_unreserve; + + amdgpu_fill_buffer(bo, 0, bo->tbo.resv, &fence); + amdgpu_bo_fence(bo, fence, false); + amdgpu_bo_unreserve(bo); + fence_put(bo->tbo.moving); + bo->tbo.moving = fence_get(fence); + fence_put(fence); + } *bo_ptr = bo; trace_amdgpu_bo_create(bo); return 0; + +fail_unreserve: + amdgpu_bo_unreserve(bo); +fail_free: + amdgpu_bo_unref(&bo); + return r; +} + +static int amdgpu_bo_create_shadow(struct amdgpu_device *adev, + unsigned long size, int byte_align, + struct amdgpu_bo *bo) +{ + struct ttm_placement placement = {0}; + struct ttm_place placements[AMDGPU_GEM_DOMAIN_MAX + 1]; + int r; + + if (bo->shadow) + return 0; + + bo->flags |= AMDGPU_GEM_CREATE_SHADOW; + memset(&placements, 0, + (AMDGPU_GEM_DOMAIN_MAX + 1) * sizeof(struct ttm_place)); + + amdgpu_ttm_placement_init(adev, &placement, + placements, AMDGPU_GEM_DOMAIN_GTT, + AMDGPU_GEM_CREATE_CPU_GTT_USWC); + + r = amdgpu_bo_create_restricted(adev, size, byte_align, true, + AMDGPU_GEM_DOMAIN_GTT, + AMDGPU_GEM_CREATE_CPU_GTT_USWC, + NULL, &placement, + bo->tbo.resv, + &bo->shadow); + if (!r) { + bo->shadow->parent = amdgpu_bo_ref(bo); + mutex_lock(&adev->shadow_list_lock); + list_add_tail(&bo->shadow_list, &adev->shadow_list); + mutex_unlock(&adev->shadow_list_lock); + } + + return r; } int amdgpu_bo_create(struct amdgpu_device *adev, @@ -293,6 +438,7 @@ int amdgpu_bo_create(struct amdgpu_device *adev, { struct ttm_placement placement = {0}; struct ttm_place placements[AMDGPU_GEM_DOMAIN_MAX + 1]; + int r; memset(&placements, 0, (AMDGPU_GEM_DOMAIN_MAX + 1) * sizeof(struct ttm_place)); @@ -300,9 +446,83 @@ int amdgpu_bo_create(struct amdgpu_device *adev, amdgpu_ttm_placement_init(adev, &placement, placements, domain, flags); - return amdgpu_bo_create_restricted(adev, size, byte_align, kernel, - domain, flags, sg, &placement, - resv, bo_ptr); + r = amdgpu_bo_create_restricted(adev, size, byte_align, kernel, + domain, flags, sg, &placement, + resv, bo_ptr); + if (r) + return r; + + if (amdgpu_need_backup(adev) && (flags & AMDGPU_GEM_CREATE_SHADOW)) { + r = amdgpu_bo_create_shadow(adev, size, byte_align, (*bo_ptr)); + if (r) + amdgpu_bo_unref(bo_ptr); + } + + return r; +} + +int amdgpu_bo_backup_to_shadow(struct amdgpu_device *adev, + struct amdgpu_ring *ring, + struct amdgpu_bo *bo, + struct reservation_object *resv, + struct fence **fence, + bool direct) + +{ + struct amdgpu_bo *shadow = bo->shadow; + uint64_t bo_addr, shadow_addr; + int r; + + if (!shadow) + return -EINVAL; + + bo_addr = amdgpu_bo_gpu_offset(bo); + shadow_addr = amdgpu_bo_gpu_offset(bo->shadow); + + r = reservation_object_reserve_shared(bo->tbo.resv); + if (r) + goto err; + + r = amdgpu_copy_buffer(ring, bo_addr, shadow_addr, + amdgpu_bo_size(bo), resv, fence, + direct); + if (!r) + amdgpu_bo_fence(bo, *fence, true); + +err: + return r; +} + +int amdgpu_bo_restore_from_shadow(struct amdgpu_device *adev, + struct amdgpu_ring *ring, + struct amdgpu_bo *bo, + struct reservation_object *resv, + struct fence **fence, + bool direct) + +{ + struct amdgpu_bo *shadow = bo->shadow; + uint64_t bo_addr, shadow_addr; + int r; + + if (!shadow) + return -EINVAL; + + bo_addr = amdgpu_bo_gpu_offset(bo); + shadow_addr = amdgpu_bo_gpu_offset(bo->shadow); + + r = reservation_object_reserve_shared(bo->tbo.resv); + if (r) + goto err; + + r = amdgpu_copy_buffer(ring, shadow_addr, bo_addr, + amdgpu_bo_size(bo), resv, fence, + direct); + if (!r) + amdgpu_bo_fence(bo, *fence, true); + +err: + return r; } int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr) @@ -380,16 +600,17 @@ int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain, return -EINVAL; if (bo->pin_count) { + uint32_t mem_type = bo->tbo.mem.mem_type; + + if (domain != amdgpu_mem_type_to_domain(mem_type)) + return -EINVAL; + bo->pin_count++; if (gpu_addr) *gpu_addr = amdgpu_bo_gpu_offset(bo); if (max_offset != 0) { - u64 domain_start; - if (domain == AMDGPU_GEM_DOMAIN_VRAM) - domain_start = bo->adev->mc.vram_start; - else - domain_start = bo->adev->mc.gtt_start; + u64 domain_start = bo->tbo.bdev->man[mem_type].gpu_offset; WARN_ON_ONCE(max_offset < (amdgpu_bo_gpu_offset(bo) - domain_start)); } @@ -401,7 +622,8 @@ int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain, /* force to pin into visible video ram */ if ((bo->placements[i].flags & TTM_PL_FLAG_VRAM) && !(bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS) && - (!max_offset || max_offset > bo->adev->mc.visible_vram_size)) { + (!max_offset || max_offset > + bo->adev->mc.visible_vram_size)) { if (WARN_ON_ONCE(min_offset > bo->adev->mc.visible_vram_size)) return -EINVAL; @@ -420,19 +642,23 @@ int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain, } r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false); - if (likely(r == 0)) { - bo->pin_count = 1; - if (gpu_addr != NULL) - *gpu_addr = amdgpu_bo_gpu_offset(bo); - if (domain == AMDGPU_GEM_DOMAIN_VRAM) { - bo->adev->vram_pin_size += amdgpu_bo_size(bo); - if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS) - bo->adev->invisible_pin_size += amdgpu_bo_size(bo); - } else - bo->adev->gart_pin_size += amdgpu_bo_size(bo); - } else { + if (unlikely(r)) { dev_err(bo->adev->dev, "%p pin failed\n", bo); + goto error; + } + + bo->pin_count = 1; + if (gpu_addr != NULL) + *gpu_addr = amdgpu_bo_gpu_offset(bo); + if (domain == AMDGPU_GEM_DOMAIN_VRAM) { + bo->adev->vram_pin_size += amdgpu_bo_size(bo); + if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS) + bo->adev->invisible_pin_size += amdgpu_bo_size(bo); + } else if (domain == AMDGPU_GEM_DOMAIN_GTT) { + bo->adev->gart_pin_size += amdgpu_bo_size(bo); } + +error: return r; } @@ -457,16 +683,20 @@ int amdgpu_bo_unpin(struct amdgpu_bo *bo) bo->placements[i].flags &= ~TTM_PL_FLAG_NO_EVICT; } r = ttm_bo_validate(&bo->tbo, &bo->placement, false, false); - if (likely(r == 0)) { - if (bo->tbo.mem.mem_type == TTM_PL_VRAM) { - bo->adev->vram_pin_size -= amdgpu_bo_size(bo); - if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS) - bo->adev->invisible_pin_size -= amdgpu_bo_size(bo); - } else - bo->adev->gart_pin_size -= amdgpu_bo_size(bo); - } else { + if (unlikely(r)) { dev_err(bo->adev->dev, "%p validate failed for unpin\n", bo); + goto error; + } + + if (bo->tbo.mem.mem_type == TTM_PL_VRAM) { + bo->adev->vram_pin_size -= amdgpu_bo_size(bo); + if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS) + bo->adev->invisible_pin_size -= amdgpu_bo_size(bo); + } else { + bo->adev->gart_pin_size -= amdgpu_bo_size(bo); } + +error: return r; } @@ -637,7 +867,8 @@ int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo) for (i = 0; i < abo->placement.num_placement; i++) { /* Force into visible VRAM */ if ((abo->placements[i].flags & TTM_PL_FLAG_VRAM) && - (!abo->placements[i].lpfn || abo->placements[i].lpfn > lpfn)) + (!abo->placements[i].lpfn || + abo->placements[i].lpfn > lpfn)) abo->placements[i].lpfn = lpfn; } r = ttm_bo_validate(bo, &abo->placement, false, false); @@ -674,3 +905,21 @@ void amdgpu_bo_fence(struct amdgpu_bo *bo, struct fence *fence, else reservation_object_add_excl_fence(resv, fence); } + +/** + * amdgpu_bo_gpu_offset - return GPU offset of bo + * @bo: amdgpu object for which we query the offset + * + * Returns current GPU offset of the object. + * + * Note: object should either be pinned or reserved when calling this + * function, it might be useful to add check for this for debugging. + */ +u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo) +{ + WARN_ON_ONCE(bo->tbo.mem.mem_type == TTM_PL_SYSTEM); + WARN_ON_ONCE(!ww_mutex_is_locked(&bo->tbo.resv->lock) && + !bo->pin_count); + + return bo->tbo.offset; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h index bdb01d932548..b6a27390ef88 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h @@ -85,21 +85,6 @@ static inline void amdgpu_bo_unreserve(struct amdgpu_bo *bo) ttm_bo_unreserve(&bo->tbo); } -/** - * amdgpu_bo_gpu_offset - return GPU offset of bo - * @bo: amdgpu object for which we query the offset - * - * Returns current GPU offset of the object. - * - * Note: object should either be pinned or reserved when calling this - * function, it might be useful to add check for this for debugging. - */ -static inline u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo) -{ - WARN_ON_ONCE(bo->tbo.mem.mem_type == TTM_PL_SYSTEM); - return bo->tbo.offset; -} - static inline unsigned long amdgpu_bo_size(struct amdgpu_bo *bo) { return bo->tbo.num_pages << PAGE_SHIFT; @@ -139,6 +124,10 @@ int amdgpu_bo_create_restricted(struct amdgpu_device *adev, struct ttm_placement *placement, struct reservation_object *resv, struct amdgpu_bo **bo_ptr); +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_kmap(struct amdgpu_bo *bo, void **ptr); void amdgpu_bo_kunmap(struct amdgpu_bo *bo); struct amdgpu_bo *amdgpu_bo_ref(struct amdgpu_bo *bo); @@ -165,6 +154,19 @@ void amdgpu_bo_move_notify(struct ttm_buffer_object *bo, int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo); void amdgpu_bo_fence(struct amdgpu_bo *bo, struct fence *fence, bool shared); +u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo); +int amdgpu_bo_backup_to_shadow(struct amdgpu_device *adev, + struct amdgpu_ring *ring, + struct amdgpu_bo *bo, + struct reservation_object *resv, + struct fence **fence, bool direct); +int amdgpu_bo_restore_from_shadow(struct amdgpu_device *adev, + struct amdgpu_ring *ring, + struct amdgpu_bo *bo, + struct reservation_object *resv, + struct fence **fence, + bool direct); + /* * sub allocation diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c index 5cc7052e391d..d4ec3cb187a5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c @@ -1103,54 +1103,46 @@ force: void amdgpu_dpm_enable_uvd(struct amdgpu_device *adev, bool enable) { - if (adev->pp_enabled) + if (adev->pp_enabled || adev->pm.funcs->powergate_uvd) { + /* enable/disable UVD */ + mutex_lock(&adev->pm.mutex); amdgpu_dpm_powergate_uvd(adev, !enable); - else { - if (adev->pm.funcs->powergate_uvd) { + mutex_unlock(&adev->pm.mutex); + } else { + if (enable) { mutex_lock(&adev->pm.mutex); - /* enable/disable UVD */ - amdgpu_dpm_powergate_uvd(adev, !enable); + adev->pm.dpm.uvd_active = true; + adev->pm.dpm.state = POWER_STATE_TYPE_INTERNAL_UVD; mutex_unlock(&adev->pm.mutex); } else { - if (enable) { - mutex_lock(&adev->pm.mutex); - adev->pm.dpm.uvd_active = true; - adev->pm.dpm.state = POWER_STATE_TYPE_INTERNAL_UVD; - mutex_unlock(&adev->pm.mutex); - } else { - mutex_lock(&adev->pm.mutex); - adev->pm.dpm.uvd_active = false; - mutex_unlock(&adev->pm.mutex); - } - amdgpu_pm_compute_clocks(adev); + mutex_lock(&adev->pm.mutex); + adev->pm.dpm.uvd_active = false; + mutex_unlock(&adev->pm.mutex); } - + amdgpu_pm_compute_clocks(adev); } } void amdgpu_dpm_enable_vce(struct amdgpu_device *adev, bool enable) { - if (adev->pp_enabled) + if (adev->pp_enabled || adev->pm.funcs->powergate_vce) { + /* enable/disable VCE */ + mutex_lock(&adev->pm.mutex); amdgpu_dpm_powergate_vce(adev, !enable); - else { - if (adev->pm.funcs->powergate_vce) { + mutex_unlock(&adev->pm.mutex); + } else { + if (enable) { mutex_lock(&adev->pm.mutex); - amdgpu_dpm_powergate_vce(adev, !enable); + adev->pm.dpm.vce_active = true; + /* XXX select vce level based on ring/task */ + adev->pm.dpm.vce_level = AMDGPU_VCE_LEVEL_AC_ALL; mutex_unlock(&adev->pm.mutex); } else { - if (enable) { - mutex_lock(&adev->pm.mutex); - adev->pm.dpm.vce_active = true; - /* XXX select vce level based on ring/task */ - adev->pm.dpm.vce_level = AMDGPU_VCE_LEVEL_AC_ALL; - mutex_unlock(&adev->pm.mutex); - } else { - mutex_lock(&adev->pm.mutex); - adev->pm.dpm.vce_active = false; - mutex_unlock(&adev->pm.mutex); - } - amdgpu_pm_compute_clocks(adev); + mutex_lock(&adev->pm.mutex); + adev->pm.dpm.vce_active = false; + mutex_unlock(&adev->pm.mutex); } + amdgpu_pm_compute_clocks(adev); } } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c index c5738a22b690..545074479e1f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c @@ -52,7 +52,9 @@ static int amdgpu_powerplay_init(struct amdgpu_device *adev) pp_init->chip_family = adev->family; pp_init->chip_id = adev->asic_type; pp_init->device = amdgpu_cgs_create_device(adev); - pp_init->powercontainment_enabled = amdgpu_powercontainment; + pp_init->rev_id = adev->pdev->revision; + pp_init->sub_sys_id = adev->pdev->subsystem_device; + pp_init->sub_vendor_id = adev->pdev->subsystem_vendor; ret = amd_powerplay_init(pp_init, amd_pp); kfree(pp_init); @@ -106,11 +108,10 @@ static int amdgpu_pp_early_init(void *handle) break; case CHIP_TONGA: case CHIP_FIJI: - adev->pp_enabled = (amdgpu_powerplay == 0) ? false : true; - break; + case CHIP_TOPAZ: case CHIP_CARRIZO: case CHIP_STONEY: - adev->pp_enabled = (amdgpu_powerplay > 0) ? true : false; + adev->pp_enabled = (amdgpu_powerplay == 0) ? false : true; break; /* These chips don't have powerplay implemenations */ case CHIP_BONAIRE: @@ -118,7 +119,6 @@ static int amdgpu_pp_early_init(void *handle) case CHIP_KABINI: case CHIP_MULLINS: case CHIP_KAVERI: - case CHIP_TOPAZ: default: adev->pp_enabled = false; break; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c index 85aeb0a804bb..242ba04bfde6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c @@ -222,33 +222,16 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring, /* Allocate ring buffer */ if (ring->ring_obj == NULL) { - r = amdgpu_bo_create(adev, ring->ring_size, PAGE_SIZE, true, - AMDGPU_GEM_DOMAIN_GTT, 0, - NULL, NULL, &ring->ring_obj); + r = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE, + AMDGPU_GEM_DOMAIN_GTT, + &ring->ring_obj, + &ring->gpu_addr, + (void **)&ring->ring); if (r) { dev_err(adev->dev, "(%d) ring create failed\n", r); return r; } - r = amdgpu_bo_reserve(ring->ring_obj, false); - if (unlikely(r != 0)) - return r; - r = amdgpu_bo_pin(ring->ring_obj, AMDGPU_GEM_DOMAIN_GTT, - &ring->gpu_addr); - if (r) { - amdgpu_bo_unreserve(ring->ring_obj); - dev_err(adev->dev, "(%d) ring pin failed\n", r); - return r; - } - r = amdgpu_bo_kmap(ring->ring_obj, - (void **)&ring->ring); - memset((void *)ring->ring, 0, ring->ring_size); - - amdgpu_bo_unreserve(ring->ring_obj); - if (r) { - dev_err(adev->dev, "(%d) ring map failed\n", r); - return r; - } } ring->ptr_mask = (ring->ring_size / 4) - 1; ring->max_dw = max_dw; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c index 05a53f4fc334..b827c75e95de 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_test.c @@ -111,7 +111,7 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev) amdgpu_bo_kunmap(gtt_obj[i]); r = amdgpu_copy_buffer(ring, gtt_addr, vram_addr, - size, NULL, &fence); + size, NULL, &fence, false); if (r) { DRM_ERROR("Failed GTT->VRAM copy %d\n", i); @@ -156,7 +156,7 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev) amdgpu_bo_kunmap(vram_obj); r = amdgpu_copy_buffer(ring, vram_addr, gtt_addr, - size, NULL, &fence); + size, NULL, &fence, false); if (r) { DRM_ERROR("Failed VRAM->GTT copy %d\n", i); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index 9b61c8ba7aaf..5447973483ec 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -34,6 +34,7 @@ #include <ttm/ttm_placement.h> #include <ttm/ttm_module.h> #include <ttm/ttm_page_alloc.h> +#include <ttm/ttm_memory.h> #include <drm/drmP.h> #include <drm/amdgpu_drm.h> #include <linux/seq_file.h> @@ -74,7 +75,7 @@ static void amdgpu_ttm_mem_global_release(struct drm_global_reference *ref) ttm_mem_global_release(ref->object); } -static int amdgpu_ttm_global_init(struct amdgpu_device *adev) +int amdgpu_ttm_global_init(struct amdgpu_device *adev) { struct drm_global_reference *global_ref; struct amdgpu_ring *ring; @@ -256,10 +257,8 @@ static int amdgpu_move_blit(struct ttm_buffer_object *bo, switch (old_mem->mem_type) { case TTM_PL_VRAM: - old_start += adev->mc.vram_start; - break; case TTM_PL_TT: - old_start += adev->mc.gtt_start; + old_start += bo->bdev->man[old_mem->mem_type].gpu_offset; break; default: DRM_ERROR("Unknown placement %d\n", old_mem->mem_type); @@ -267,10 +266,8 @@ static int amdgpu_move_blit(struct ttm_buffer_object *bo, } switch (new_mem->mem_type) { case TTM_PL_VRAM: - new_start += adev->mc.vram_start; - break; case TTM_PL_TT: - new_start += adev->mc.gtt_start; + new_start += bo->bdev->man[new_mem->mem_type].gpu_offset; break; default: DRM_ERROR("Unknown placement %d\n", old_mem->mem_type); @@ -285,7 +282,7 @@ static int amdgpu_move_blit(struct ttm_buffer_object *bo, r = amdgpu_copy_buffer(ring, old_start, new_start, new_mem->num_pages * PAGE_SIZE, /* bytes */ - bo->resv, &fence); + bo->resv, &fence, false); if (r) return r; @@ -335,7 +332,7 @@ static int amdgpu_move_vram_ram(struct ttm_buffer_object *bo, if (unlikely(r)) { goto out_cleanup; } - r = ttm_bo_move_ttm(bo, true, interruptible, no_wait_gpu, new_mem); + r = ttm_bo_move_ttm(bo, interruptible, no_wait_gpu, new_mem); out_cleanup: ttm_bo_mem_put(bo, &tmp_mem); return r; @@ -368,7 +365,7 @@ static int amdgpu_move_ram_vram(struct ttm_buffer_object *bo, if (unlikely(r)) { return r; } - r = ttm_bo_move_ttm(bo, true, interruptible, no_wait_gpu, &tmp_mem); + r = ttm_bo_move_ttm(bo, interruptible, no_wait_gpu, &tmp_mem); if (unlikely(r)) { goto out_cleanup; } @@ -435,8 +432,7 @@ static int amdgpu_bo_move(struct ttm_buffer_object *bo, if (r) { memcpy: - r = ttm_bo_move_memcpy(bo, evict, interruptible, - no_wait_gpu, new_mem); + r = ttm_bo_move_memcpy(bo, interruptible, no_wait_gpu, new_mem); if (r) { return r; } @@ -950,6 +946,8 @@ static struct list_head *amdgpu_ttm_lru_tail(struct ttm_buffer_object *tbo) struct list_head *res = lru->lru[tbo->mem.mem_type]; lru->lru[tbo->mem.mem_type] = &tbo->lru; + while ((++lru)->lru[tbo->mem.mem_type] == res) + lru->lru[tbo->mem.mem_type] = &tbo->lru; return res; } @@ -960,6 +958,8 @@ static struct list_head *amdgpu_ttm_swap_lru_tail(struct ttm_buffer_object *tbo) struct list_head *res = lru->swap_lru; lru->swap_lru = &tbo->swap; + while ((++lru)->swap_lru == res) + lru->swap_lru = &tbo->swap; return res; } @@ -987,10 +987,6 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) unsigned i, j; int r; - r = amdgpu_ttm_global_init(adev); - if (r) { - return r; - } /* No others user of address space so set it to 0 */ r = ttm_bo_device_init(&adev->mman.bdev, adev->mman.bo_global_ref.ref.object, @@ -1011,6 +1007,10 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) lru->swap_lru = &adev->mman.bdev.glob->swap_lru; } + for (j = 0; j < TTM_NUM_MEM_TYPES; ++j) + adev->mman.guard.lru[j] = NULL; + adev->mman.guard.swap_lru = NULL; + adev->mman.initialized = true; r = ttm_bo_init_mm(&adev->mman.bdev, TTM_PL_VRAM, adev->mc.real_vram_size >> PAGE_SHIFT); @@ -1151,7 +1151,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t dst_offset, uint32_t byte_count, struct reservation_object *resv, - struct fence **fence) + struct fence **fence, bool direct_submit) { struct amdgpu_device *adev = ring->adev; struct amdgpu_job *job; @@ -1195,8 +1195,79 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, amdgpu_ring_pad_ib(ring, &job->ibs[0]); WARN_ON(job->ibs[0].length_dw > num_dw); + if (direct_submit) { + r = amdgpu_ib_schedule(ring, job->num_ibs, job->ibs, + NULL, NULL, fence); + job->fence = fence_get(*fence); + if (r) + DRM_ERROR("Error scheduling IBs (%d)\n", r); + amdgpu_job_free(job); + } else { + r = amdgpu_job_submit(job, ring, &adev->mman.entity, + AMDGPU_FENCE_OWNER_UNDEFINED, fence); + if (r) + goto error_free; + } + + return r; + +error_free: + amdgpu_job_free(job); + return r; +} + +int amdgpu_fill_buffer(struct amdgpu_bo *bo, + uint32_t src_data, + struct reservation_object *resv, + struct fence **fence) +{ + struct amdgpu_device *adev = bo->adev; + struct amdgpu_job *job; + struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring; + + uint32_t max_bytes, byte_count; + uint64_t dst_offset; + unsigned int num_loops, num_dw; + unsigned int i; + int r; + + byte_count = bo->tbo.num_pages << PAGE_SHIFT; + max_bytes = adev->mman.buffer_funcs->fill_max_bytes; + num_loops = DIV_ROUND_UP(byte_count, max_bytes); + num_dw = num_loops * adev->mman.buffer_funcs->fill_num_dw; + + /* for IB padding */ + while (num_dw & 0x7) + num_dw++; + + r = amdgpu_job_alloc_with_ib(adev, num_dw * 4, &job); + if (r) + return r; + + if (resv) { + r = amdgpu_sync_resv(adev, &job->sync, resv, + AMDGPU_FENCE_OWNER_UNDEFINED); + if (r) { + DRM_ERROR("sync failed (%d).\n", r); + goto error_free; + } + } + + dst_offset = bo->tbo.mem.start << PAGE_SHIFT; + for (i = 0; i < num_loops; i++) { + uint32_t cur_size_in_bytes = min(byte_count, max_bytes); + + amdgpu_emit_fill_buffer(adev, &job->ibs[0], src_data, + dst_offset, cur_size_in_bytes); + + dst_offset += cur_size_in_bytes; + byte_count -= cur_size_in_bytes; + } + + amdgpu_ring_pad_ib(ring, &job->ibs[0]); + WARN_ON(job->ibs[0].length_dw > num_dw); r = amdgpu_job_submit(job, ring, &adev->mman.entity, - AMDGPU_FENCE_OWNER_UNDEFINED, fence); + AMDGPU_FENCE_OWNER_UNDEFINED, fence); if (r) goto error_free; @@ -1387,3 +1458,8 @@ static void amdgpu_ttm_debugfs_fini(struct amdgpu_device *adev) #endif } + +u64 amdgpu_ttm_get_gtt_mem_size(struct amdgpu_device *adev) +{ + return ttm_get_kernel_zone_memory_size(adev->mman.mem_global_ref.object); +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h new file mode 100644 index 000000000000..72f6bfc15d8f --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h @@ -0,0 +1,80 @@ +/* + * Copyright 2016 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __AMDGPU_TTM_H__ +#define __AMDGPU_TTM_H__ + +#include "gpu_scheduler.h" + +#define AMDGPU_PL_GDS TTM_PL_PRIV0 +#define AMDGPU_PL_GWS TTM_PL_PRIV1 +#define AMDGPU_PL_OA TTM_PL_PRIV2 + +#define AMDGPU_PL_FLAG_GDS TTM_PL_FLAG_PRIV0 +#define AMDGPU_PL_FLAG_GWS TTM_PL_FLAG_PRIV1 +#define AMDGPU_PL_FLAG_OA TTM_PL_FLAG_PRIV2 + +#define AMDGPU_TTM_LRU_SIZE 20 + +struct amdgpu_mman_lru { + struct list_head *lru[TTM_NUM_MEM_TYPES]; + struct list_head *swap_lru; +}; + +struct amdgpu_mman { + struct ttm_bo_global_ref bo_global_ref; + struct drm_global_reference mem_global_ref; + struct ttm_bo_device bdev; + bool mem_global_referenced; + bool initialized; + +#if defined(CONFIG_DEBUG_FS) + struct dentry *vram; + struct dentry *gtt; +#endif + + /* buffer handling */ + const struct amdgpu_buffer_funcs *buffer_funcs; + struct amdgpu_ring *buffer_funcs_ring; + /* Scheduler entity for buffer moves */ + struct amd_sched_entity entity; + + /* custom LRU management */ + struct amdgpu_mman_lru log2_size[AMDGPU_TTM_LRU_SIZE]; + /* guard for log2_size array, don't add anything in between */ + struct amdgpu_mman_lru guard; +}; + +int amdgpu_copy_buffer(struct amdgpu_ring *ring, + uint64_t src_offset, + uint64_t dst_offset, + uint32_t byte_count, + struct reservation_object *resv, + struct fence **fence, bool direct_submit); +int amdgpu_fill_buffer(struct amdgpu_bo *bo, + uint32_t src_data, + struct reservation_object *resv, + struct fence **fence); + +int amdgpu_mmap(struct file *filp, struct vm_area_struct *vma); +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c index b11f4e8868d7..cc766cc53a87 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c @@ -201,39 +201,14 @@ int amdgpu_uvd_sw_init(struct amdgpu_device *adev) bo_size = AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8) + AMDGPU_UVD_STACK_SIZE + AMDGPU_UVD_HEAP_SIZE + AMDGPU_UVD_SESSION_SIZE * adev->uvd.max_handles; - r = amdgpu_bo_create(adev, bo_size, PAGE_SIZE, true, - AMDGPU_GEM_DOMAIN_VRAM, - AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED, - NULL, NULL, &adev->uvd.vcpu_bo); + r = amdgpu_bo_create_kernel(adev, bo_size, PAGE_SIZE, + AMDGPU_GEM_DOMAIN_VRAM, &adev->uvd.vcpu_bo, + &adev->uvd.gpu_addr, &adev->uvd.cpu_addr); if (r) { dev_err(adev->dev, "(%d) failed to allocate UVD bo\n", r); return r; } - r = amdgpu_bo_reserve(adev->uvd.vcpu_bo, false); - if (r) { - amdgpu_bo_unref(&adev->uvd.vcpu_bo); - dev_err(adev->dev, "(%d) failed to reserve UVD bo\n", r); - return r; - } - - r = amdgpu_bo_pin(adev->uvd.vcpu_bo, AMDGPU_GEM_DOMAIN_VRAM, - &adev->uvd.gpu_addr); - if (r) { - amdgpu_bo_unreserve(adev->uvd.vcpu_bo); - amdgpu_bo_unref(&adev->uvd.vcpu_bo); - dev_err(adev->dev, "(%d) UVD bo pin failed\n", r); - return r; - } - - r = amdgpu_bo_kmap(adev->uvd.vcpu_bo, &adev->uvd.cpu_addr); - if (r) { - dev_err(adev->dev, "(%d) UVD map failed\n", r); - return r; - } - - amdgpu_bo_unreserve(adev->uvd.vcpu_bo); - ring = &adev->uvd.ring; rq = &ring->sched.sched_rq[AMD_SCHED_PRIORITY_NORMAL]; r = amd_sched_entity_init(&ring->sched, &adev->uvd.entity, @@ -323,7 +298,7 @@ int amdgpu_uvd_suspend(struct amdgpu_device *adev) if (!adev->uvd.saved_bo) return -ENOMEM; - memcpy(adev->uvd.saved_bo, ptr, size); + memcpy_fromio(adev->uvd.saved_bo, ptr, size); return 0; } @@ -340,7 +315,7 @@ int amdgpu_uvd_resume(struct amdgpu_device *adev) ptr = adev->uvd.cpu_addr; if (adev->uvd.saved_bo != NULL) { - memcpy(ptr, adev->uvd.saved_bo, size); + memcpy_toio(ptr, adev->uvd.saved_bo, size); kfree(adev->uvd.saved_bo); adev->uvd.saved_bo = NULL; } else { @@ -349,11 +324,11 @@ int amdgpu_uvd_resume(struct amdgpu_device *adev) hdr = (const struct common_firmware_header *)adev->uvd.fw->data; offset = le32_to_cpu(hdr->ucode_array_offset_bytes); - memcpy(adev->uvd.cpu_addr, (adev->uvd.fw->data) + offset, - (adev->uvd.fw->size) - offset); + memcpy_toio(adev->uvd.cpu_addr, adev->uvd.fw->data + offset, + le32_to_cpu(hdr->ucode_size_bytes)); size -= le32_to_cpu(hdr->ucode_size_bytes); ptr += le32_to_cpu(hdr->ucode_size_bytes); - memset(ptr, 0, size); + memset_io(ptr, 0, size); } return 0; @@ -843,6 +818,7 @@ static int amdgpu_uvd_cs_reg(struct amdgpu_uvd_cs_ctx *ctx, return r; break; case mmUVD_ENGINE_CNTL: + case mmUVD_NO_OP: break; default: DRM_ERROR("Invalid reg 0x%X!\n", reg); @@ -981,8 +957,10 @@ static int amdgpu_uvd_send_msg(struct amdgpu_ring *ring, struct amdgpu_bo *bo, ib->ptr[3] = addr >> 32; ib->ptr[4] = PACKET0(mmUVD_GPCOM_VCPU_CMD, 0); ib->ptr[5] = 0; - for (i = 6; i < 16; ++i) - ib->ptr[i] = PACKET2(0); + for (i = 6; i < 16; i += 2) { + ib->ptr[i] = PACKET0(mmUVD_NO_OP, 0); + ib->ptr[i+1] = 0; + } ib->length_dw = 16; if (direct) { @@ -1114,15 +1092,9 @@ static void amdgpu_uvd_idle_work_handler(struct work_struct *work) { struct amdgpu_device *adev = container_of(work, struct amdgpu_device, uvd.idle_work.work); - unsigned i, fences, handles = 0; - - fences = amdgpu_fence_count_emitted(&adev->uvd.ring); - - for (i = 0; i < adev->uvd.max_handles; ++i) - if (atomic_read(&adev->uvd.handles[i])) - ++handles; + unsigned fences = amdgpu_fence_count_emitted(&adev->uvd.ring); - if (fences == 0 && handles == 0) { + if (fences == 0) { if (adev->pm.dpm_enabled) { amdgpu_dpm_enable_uvd(adev, false); } else { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c index 05865ce35351..da52af2a935a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c @@ -282,8 +282,8 @@ int amdgpu_vce_resume(struct amdgpu_device *adev) hdr = (const struct common_firmware_header *)adev->vce.fw->data; offset = le32_to_cpu(hdr->ucode_array_offset_bytes); - memcpy(cpu_addr, (adev->vce.fw->data) + offset, - (adev->vce.fw->size) - offset); + memcpy_toio(cpu_addr, adev->vce.fw->data + offset, + adev->vce.fw->size - offset); amdgpu_bo_kunmap(adev->vce.vcpu_bo); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 8e642fc48df4..bf56f1814437 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -51,19 +51,22 @@ * SI supports 16. */ -/* Special value that no flush is necessary */ -#define AMDGPU_VM_NO_FLUSH (~0ll) - /* Local structure. Encapsulate some VM table update parameters to reduce * the number of function parameters */ -struct amdgpu_vm_update_params { +struct amdgpu_pte_update_params { + /* amdgpu device we do this update for */ + struct amdgpu_device *adev; /* address where to copy page table entries from */ uint64_t src; - /* DMA addresses to use for mapping */ - dma_addr_t *pages_addr; /* indirect buffer to fill with commands */ struct amdgpu_ib *ib; + /* Function which actually does the update */ + void (*func)(struct amdgpu_pte_update_params *params, uint64_t pe, + uint64_t addr, unsigned count, uint32_t incr, + uint32_t flags); + /* indicate update pt or its shadow */ + bool shadow; }; /** @@ -467,10 +470,9 @@ struct amdgpu_bo_va *amdgpu_vm_bo_find(struct amdgpu_vm *vm, } /** - * amdgpu_vm_update_pages - helper to call the right asic function + * amdgpu_vm_do_set_ptes - helper to call the right asic function * - * @adev: amdgpu_device pointer - * @vm_update_params: see amdgpu_vm_update_params definition + * @params: see amdgpu_pte_update_params definition * @pe: addr of the page entry * @addr: dst addr to write into pe * @count: number of page entries to update @@ -480,35 +482,47 @@ struct amdgpu_bo_va *amdgpu_vm_bo_find(struct amdgpu_vm *vm, * Traces the parameters and calls the right asic functions * to setup the page table using the DMA. */ -static void amdgpu_vm_update_pages(struct amdgpu_device *adev, - struct amdgpu_vm_update_params - *vm_update_params, - uint64_t pe, uint64_t addr, - unsigned count, uint32_t incr, - uint32_t flags) +static void amdgpu_vm_do_set_ptes(struct amdgpu_pte_update_params *params, + uint64_t pe, uint64_t addr, + unsigned count, uint32_t incr, + uint32_t flags) { trace_amdgpu_vm_set_page(pe, addr, count, incr, flags); - if (vm_update_params->src) { - amdgpu_vm_copy_pte(adev, vm_update_params->ib, - pe, (vm_update_params->src + (addr >> 12) * 8), count); - - } else if (vm_update_params->pages_addr) { - amdgpu_vm_write_pte(adev, vm_update_params->ib, - vm_update_params->pages_addr, - pe, addr, count, incr, flags); - - } else if (count < 3) { - amdgpu_vm_write_pte(adev, vm_update_params->ib, NULL, pe, addr, - count, incr, flags); + if (count < 3) { + amdgpu_vm_write_pte(params->adev, params->ib, pe, + addr | flags, count, incr); } else { - amdgpu_vm_set_pte_pde(adev, vm_update_params->ib, pe, addr, + amdgpu_vm_set_pte_pde(params->adev, params->ib, pe, addr, count, incr, flags); } } /** + * amdgpu_vm_do_copy_ptes - copy the PTEs from the GART + * + * @params: see amdgpu_pte_update_params definition + * @pe: addr of the page entry + * @addr: dst addr to write into pe + * @count: number of page entries to update + * @incr: increase next addr by incr bytes + * @flags: hw access flags + * + * Traces the parameters and calls the DMA function to copy the PTEs. + */ +static void amdgpu_vm_do_copy_ptes(struct amdgpu_pte_update_params *params, + uint64_t pe, uint64_t addr, + unsigned count, uint32_t incr, + uint32_t flags) +{ + trace_amdgpu_vm_set_page(pe, addr, count, incr, flags); + + amdgpu_vm_copy_pte(params->adev, params->ib, pe, + (params->src + (addr >> 12) * 8), count); +} + +/** * amdgpu_vm_clear_bo - initially clear the page dir/table * * @adev: amdgpu_device pointer @@ -523,12 +537,11 @@ static int amdgpu_vm_clear_bo(struct amdgpu_device *adev, struct amdgpu_ring *ring; struct fence *fence = NULL; struct amdgpu_job *job; - struct amdgpu_vm_update_params vm_update_params; + struct amdgpu_pte_update_params params; unsigned entries; uint64_t addr; int r; - memset(&vm_update_params, 0, sizeof(vm_update_params)); ring = container_of(vm->entity.sched, struct amdgpu_ring, sched); r = reservation_object_reserve_shared(bo->tbo.resv); @@ -546,9 +559,10 @@ static int amdgpu_vm_clear_bo(struct amdgpu_device *adev, if (r) goto error; - vm_update_params.ib = &job->ibs[0]; - amdgpu_vm_update_pages(adev, &vm_update_params, addr, 0, entries, - 0, 0); + memset(¶ms, 0, sizeof(params)); + params.adev = adev; + params.ib = &job->ibs[0]; + amdgpu_vm_do_set_ptes(¶ms, addr, 0, entries, 0, 0); amdgpu_ring_pad_ib(ring, &job->ibs[0]); WARN_ON(job->ibs[0].length_dw > 64); @@ -577,55 +591,41 @@ error: * Look up the physical address of the page that the pte resolves * to and return the pointer for the page table entry. */ -uint64_t amdgpu_vm_map_gart(const dma_addr_t *pages_addr, uint64_t addr) +static uint64_t amdgpu_vm_map_gart(const dma_addr_t *pages_addr, uint64_t addr) { uint64_t result; - if (pages_addr) { - /* page table offset */ - result = pages_addr[addr >> PAGE_SHIFT]; - - /* in case cpu page size != gpu page size*/ - result |= addr & (~PAGE_MASK); + /* page table offset */ + result = pages_addr[addr >> PAGE_SHIFT]; - } else { - /* No mapping required */ - result = addr; - } + /* in case cpu page size != gpu page size*/ + result |= addr & (~PAGE_MASK); result &= 0xFFFFFFFFFFFFF000ULL; return result; } -/** - * amdgpu_vm_update_pdes - make sure that page directory is valid - * - * @adev: amdgpu_device pointer - * @vm: requested vm - * @start: start of GPU address range - * @end: end of GPU address range - * - * Allocates new page tables if necessary - * and updates the page directory. - * Returns 0 for success, error for failure. - */ -int amdgpu_vm_update_page_directory(struct amdgpu_device *adev, - struct amdgpu_vm *vm) +static int amdgpu_vm_update_pd_or_shadow(struct amdgpu_device *adev, + struct amdgpu_vm *vm, + bool shadow) { struct amdgpu_ring *ring; - struct amdgpu_bo *pd = vm->page_directory; - uint64_t pd_addr = amdgpu_bo_gpu_offset(pd); + struct amdgpu_bo *pd = shadow ? vm->page_directory->shadow : + vm->page_directory; + uint64_t pd_addr; uint32_t incr = AMDGPU_VM_PTE_COUNT * 8; uint64_t last_pde = ~0, last_pt = ~0; unsigned count = 0, pt_idx, ndw; struct amdgpu_job *job; - struct amdgpu_vm_update_params vm_update_params; + struct amdgpu_pte_update_params params; struct fence *fence = NULL; int r; - memset(&vm_update_params, 0, sizeof(vm_update_params)); + if (!pd) + return 0; + pd_addr = amdgpu_bo_gpu_offset(pd); ring = container_of(vm->entity.sched, struct amdgpu_ring, sched); /* padding, etc. */ @@ -638,7 +638,9 @@ int amdgpu_vm_update_page_directory(struct amdgpu_device *adev, if (r) return r; - vm_update_params.ib = &job->ibs[0]; + memset(¶ms, 0, sizeof(params)); + params.adev = adev; + params.ib = &job->ibs[0]; /* walk over the address space and update the page directory */ for (pt_idx = 0; pt_idx <= vm->max_pde_used; ++pt_idx) { @@ -649,19 +651,25 @@ int amdgpu_vm_update_page_directory(struct amdgpu_device *adev, continue; pt = amdgpu_bo_gpu_offset(bo); - if (vm->page_tables[pt_idx].addr == pt) - continue; - vm->page_tables[pt_idx].addr = pt; + if (!shadow) { + if (vm->page_tables[pt_idx].addr == pt) + continue; + vm->page_tables[pt_idx].addr = pt; + } else { + if (vm->page_tables[pt_idx].shadow_addr == pt) + continue; + vm->page_tables[pt_idx].shadow_addr = pt; + } pde = pd_addr + pt_idx * 8; if (((last_pde + 8 * count) != pde) || - ((last_pt + incr * count) != pt)) { + ((last_pt + incr * count) != pt) || + (count == AMDGPU_VM_MAX_UPDATE_SIZE)) { if (count) { - amdgpu_vm_update_pages(adev, &vm_update_params, - last_pde, last_pt, - count, incr, - AMDGPU_PTE_VALID); + amdgpu_vm_do_set_ptes(¶ms, last_pde, + last_pt, count, incr, + AMDGPU_PTE_VALID); } count = 1; @@ -673,15 +681,14 @@ int amdgpu_vm_update_page_directory(struct amdgpu_device *adev, } if (count) - amdgpu_vm_update_pages(adev, &vm_update_params, - last_pde, last_pt, - count, incr, AMDGPU_PTE_VALID); + amdgpu_vm_do_set_ptes(¶ms, last_pde, last_pt, + count, incr, AMDGPU_PTE_VALID); - if (vm_update_params.ib->length_dw != 0) { - amdgpu_ring_pad_ib(ring, vm_update_params.ib); + if (params.ib->length_dw != 0) { + amdgpu_ring_pad_ib(ring, params.ib); amdgpu_sync_resv(adev, &job->sync, pd->tbo.resv, AMDGPU_FENCE_OWNER_VM); - WARN_ON(vm_update_params.ib->length_dw > ndw); + WARN_ON(params.ib->length_dw > ndw); r = amdgpu_job_submit(job, ring, &vm->entity, AMDGPU_FENCE_OWNER_VM, &fence); if (r) @@ -703,92 +710,33 @@ error_free: return r; } -/** - * amdgpu_vm_frag_ptes - add fragment information to PTEs +/* + * amdgpu_vm_update_pdes - make sure that page directory is valid * * @adev: amdgpu_device pointer - * @vm_update_params: see amdgpu_vm_update_params definition - * @pe_start: first PTE to handle - * @pe_end: last PTE to handle - * @addr: addr those PTEs should point to - * @flags: hw mapping flags + * @vm: requested vm + * @start: start of GPU address range + * @end: end of GPU address range + * + * Allocates new page tables if necessary + * and updates the page directory. + * Returns 0 for success, error for failure. */ -static void amdgpu_vm_frag_ptes(struct amdgpu_device *adev, - struct amdgpu_vm_update_params - *vm_update_params, - uint64_t pe_start, uint64_t pe_end, - uint64_t addr, uint32_t flags) +int amdgpu_vm_update_page_directory(struct amdgpu_device *adev, + struct amdgpu_vm *vm) { - /** - * The MC L1 TLB supports variable sized pages, based on a fragment - * field in the PTE. When this field is set to a non-zero value, page - * granularity is increased from 4KB to (1 << (12 + frag)). The PTE - * flags are considered valid for all PTEs within the fragment range - * and corresponding mappings are assumed to be physically contiguous. - * - * The L1 TLB can store a single PTE for the whole fragment, - * significantly increasing the space available for translation - * caching. This leads to large improvements in throughput when the - * TLB is under pressure. - * - * The L2 TLB distributes small and large fragments into two - * asymmetric partitions. The large fragment cache is significantly - * larger. Thus, we try to use large fragments wherever possible. - * Userspace can support this by aligning virtual base address and - * allocation size to the fragment size. - */ - - /* SI and newer are optimized for 64KB */ - uint64_t frag_flags = AMDGPU_PTE_FRAG_64KB; - uint64_t frag_align = 0x80; - - uint64_t frag_start = ALIGN(pe_start, frag_align); - uint64_t frag_end = pe_end & ~(frag_align - 1); - - unsigned count; - - /* Abort early if there isn't anything to do */ - if (pe_start == pe_end) - return; - - /* system pages are non continuously */ - if (vm_update_params->src || vm_update_params->pages_addr || - !(flags & AMDGPU_PTE_VALID) || (frag_start >= frag_end)) { - - count = (pe_end - pe_start) / 8; - amdgpu_vm_update_pages(adev, vm_update_params, pe_start, - addr, count, AMDGPU_GPU_PAGE_SIZE, - flags); - return; - } - - /* handle the 4K area at the beginning */ - if (pe_start != frag_start) { - count = (frag_start - pe_start) / 8; - amdgpu_vm_update_pages(adev, vm_update_params, pe_start, addr, - count, AMDGPU_GPU_PAGE_SIZE, flags); - addr += AMDGPU_GPU_PAGE_SIZE * count; - } - - /* handle the area in the middle */ - count = (frag_end - frag_start) / 8; - amdgpu_vm_update_pages(adev, vm_update_params, frag_start, addr, count, - AMDGPU_GPU_PAGE_SIZE, flags | frag_flags); + int r; - /* handle the 4K area at the end */ - if (frag_end != pe_end) { - addr += AMDGPU_GPU_PAGE_SIZE * count; - count = (pe_end - frag_end) / 8; - amdgpu_vm_update_pages(adev, vm_update_params, frag_end, addr, - count, AMDGPU_GPU_PAGE_SIZE, flags); - } + r = amdgpu_vm_update_pd_or_shadow(adev, vm, true); + if (r) + return r; + return amdgpu_vm_update_pd_or_shadow(adev, vm, false); } /** * amdgpu_vm_update_ptes - make sure that page tables are valid * - * @adev: amdgpu_device pointer - * @vm_update_params: see amdgpu_vm_update_params definition + * @params: see amdgpu_pte_update_params definition * @vm: requested vm * @start: start of GPU address range * @end: end of GPU address range @@ -797,16 +745,14 @@ static void amdgpu_vm_frag_ptes(struct amdgpu_device *adev, * * Update the page tables in the range @start - @end. */ -static void amdgpu_vm_update_ptes(struct amdgpu_device *adev, - struct amdgpu_vm_update_params - *vm_update_params, +static void amdgpu_vm_update_ptes(struct amdgpu_pte_update_params *params, struct amdgpu_vm *vm, uint64_t start, uint64_t end, uint64_t dst, uint32_t flags) { const uint64_t mask = AMDGPU_VM_PTE_COUNT - 1; - uint64_t cur_pe_start, cur_pe_end, cur_dst; + uint64_t cur_pe_start, cur_nptes, cur_dst; uint64_t addr; /* next GPU address to be updated */ uint64_t pt_idx; struct amdgpu_bo *pt; @@ -817,7 +763,11 @@ static void amdgpu_vm_update_ptes(struct amdgpu_device *adev, addr = start; pt_idx = addr >> amdgpu_vm_block_size; pt = vm->page_tables[pt_idx].entry.robj; - + if (params->shadow) { + if (!pt->shadow) + return; + pt = vm->page_tables[pt_idx].entry.robj->shadow; + } if ((addr & ~mask) == (end & ~mask)) nptes = end - addr; else @@ -825,7 +775,7 @@ static void amdgpu_vm_update_ptes(struct amdgpu_device *adev, cur_pe_start = amdgpu_bo_gpu_offset(pt); cur_pe_start += (addr & mask) * 8; - cur_pe_end = cur_pe_start + 8 * nptes; + cur_nptes = nptes; cur_dst = dst; /* for next ptb*/ @@ -836,6 +786,11 @@ static void amdgpu_vm_update_ptes(struct amdgpu_device *adev, while (addr < end) { pt_idx = addr >> amdgpu_vm_block_size; pt = vm->page_tables[pt_idx].entry.robj; + if (params->shadow) { + if (!pt->shadow) + return; + pt = vm->page_tables[pt_idx].entry.robj->shadow; + } if ((addr & ~mask) == (end & ~mask)) nptes = end - addr; @@ -845,19 +800,19 @@ static void amdgpu_vm_update_ptes(struct amdgpu_device *adev, next_pe_start = amdgpu_bo_gpu_offset(pt); next_pe_start += (addr & mask) * 8; - if (cur_pe_end == next_pe_start) { + if ((cur_pe_start + 8 * cur_nptes) == next_pe_start && + ((cur_nptes + nptes) <= AMDGPU_VM_MAX_UPDATE_SIZE)) { /* The next ptb is consecutive to current ptb. - * Don't call amdgpu_vm_frag_ptes now. + * Don't call the update function now. * Will update two ptbs together in future. */ - cur_pe_end += 8 * nptes; + cur_nptes += nptes; } else { - amdgpu_vm_frag_ptes(adev, vm_update_params, - cur_pe_start, cur_pe_end, - cur_dst, flags); + params->func(params, cur_pe_start, cur_dst, cur_nptes, + AMDGPU_GPU_PAGE_SIZE, flags); cur_pe_start = next_pe_start; - cur_pe_end = next_pe_start + 8 * nptes; + cur_nptes = nptes; cur_dst = dst; } @@ -866,8 +821,79 @@ static void amdgpu_vm_update_ptes(struct amdgpu_device *adev, dst += nptes * AMDGPU_GPU_PAGE_SIZE; } - amdgpu_vm_frag_ptes(adev, vm_update_params, cur_pe_start, - cur_pe_end, cur_dst, flags); + params->func(params, cur_pe_start, cur_dst, cur_nptes, + AMDGPU_GPU_PAGE_SIZE, flags); +} + +/* + * amdgpu_vm_frag_ptes - add fragment information to PTEs + * + * @params: see amdgpu_pte_update_params definition + * @vm: requested vm + * @start: first PTE to handle + * @end: last PTE to handle + * @dst: addr those PTEs should point to + * @flags: hw mapping flags + */ +static void amdgpu_vm_frag_ptes(struct amdgpu_pte_update_params *params, + struct amdgpu_vm *vm, + uint64_t start, uint64_t end, + uint64_t dst, uint32_t flags) +{ + /** + * The MC L1 TLB supports variable sized pages, based on a fragment + * field in the PTE. When this field is set to a non-zero value, page + * granularity is increased from 4KB to (1 << (12 + frag)). The PTE + * flags are considered valid for all PTEs within the fragment range + * and corresponding mappings are assumed to be physically contiguous. + * + * The L1 TLB can store a single PTE for the whole fragment, + * significantly increasing the space available for translation + * caching. This leads to large improvements in throughput when the + * TLB is under pressure. + * + * The L2 TLB distributes small and large fragments into two + * asymmetric partitions. The large fragment cache is significantly + * larger. Thus, we try to use large fragments wherever possible. + * Userspace can support this by aligning virtual base address and + * allocation size to the fragment size. + */ + + const uint64_t frag_align = 1 << AMDGPU_LOG2_PAGES_PER_FRAG; + + uint64_t frag_start = ALIGN(start, frag_align); + uint64_t frag_end = end & ~(frag_align - 1); + + uint32_t frag; + + /* system pages are non continuously */ + if (params->src || !(flags & AMDGPU_PTE_VALID) || + (frag_start >= frag_end)) { + + amdgpu_vm_update_ptes(params, vm, start, end, dst, flags); + return; + } + + /* use more than 64KB fragment size if possible */ + frag = lower_32_bits(frag_start | frag_end); + frag = likely(frag) ? __ffs(frag) : 31; + + /* handle the 4K area at the beginning */ + if (start != frag_start) { + amdgpu_vm_update_ptes(params, vm, start, frag_start, + dst, flags); + dst += (frag_start - start) * AMDGPU_GPU_PAGE_SIZE; + } + + /* handle the area in the middle */ + amdgpu_vm_update_ptes(params, vm, frag_start, frag_end, dst, + flags | AMDGPU_PTE_FRAG(frag)); + + /* handle the 4K area at the end */ + if (frag_end != end) { + dst += (frag_end - frag_start) * AMDGPU_GPU_PAGE_SIZE; + amdgpu_vm_update_ptes(params, vm, frag_end, end, dst, flags); + } } /** @@ -900,14 +926,19 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev, void *owner = AMDGPU_FENCE_OWNER_VM; unsigned nptes, ncmds, ndw; struct amdgpu_job *job; - struct amdgpu_vm_update_params vm_update_params; + struct amdgpu_pte_update_params params; struct fence *f = NULL; int r; + memset(¶ms, 0, sizeof(params)); + params.adev = adev; + params.src = src; + ring = container_of(vm->entity.sched, struct amdgpu_ring, sched); - memset(&vm_update_params, 0, sizeof(vm_update_params)); - vm_update_params.src = src; - vm_update_params.pages_addr = pages_addr; + + memset(¶ms, 0, sizeof(params)); + params.adev = adev; + params.src = src; /* sync to everything on unmapping */ if (!(flags & AMDGPU_PTE_VALID)) @@ -924,30 +955,52 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev, /* padding, etc. */ ndw = 64; - if (vm_update_params.src) { + if (src) { /* only copy commands needed */ ndw += ncmds * 7; - } else if (vm_update_params.pages_addr) { - /* header for write data commands */ - ndw += ncmds * 4; + params.func = amdgpu_vm_do_copy_ptes; + + } else if (pages_addr) { + /* copy commands needed */ + ndw += ncmds * 7; - /* body of write data command */ + /* and also PTEs */ ndw += nptes * 2; + params.func = amdgpu_vm_do_copy_ptes; + } else { /* set page commands needed */ ndw += ncmds * 10; /* two extra commands for begin/end of fragment */ ndw += 2 * 10; + + params.func = amdgpu_vm_do_set_ptes; } r = amdgpu_job_alloc_with_ib(adev, ndw * 4, &job); if (r) return r; - vm_update_params.ib = &job->ibs[0]; + params.ib = &job->ibs[0]; + + if (!src && pages_addr) { + uint64_t *pte; + unsigned i; + + /* Put the PTEs at the end of the IB. */ + i = ndw - nptes * 2; + pte= (uint64_t *)&(job->ibs->ptr[i]); + params.src = job->ibs->gpu_addr + i * 4; + + for (i = 0; i < nptes; ++i) { + pte[i] = amdgpu_vm_map_gart(pages_addr, addr + i * + AMDGPU_GPU_PAGE_SIZE); + pte[i] |= flags; + } + } r = amdgpu_sync_fence(adev, &job->sync, exclusive); if (r) @@ -962,11 +1015,13 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev, if (r) goto error_free; - amdgpu_vm_update_ptes(adev, &vm_update_params, vm, start, - last + 1, addr, flags); + params.shadow = true; + amdgpu_vm_frag_ptes(¶ms, vm, start, last + 1, addr, flags); + params.shadow = false; + amdgpu_vm_frag_ptes(¶ms, vm, start, last + 1, addr, flags); - amdgpu_ring_pad_ib(ring, vm_update_params.ib); - WARN_ON(vm_update_params.ib->length_dw > ndw); + amdgpu_ring_pad_ib(ring, params.ib); + WARN_ON(params.ib->length_dw > ndw); r = amdgpu_job_submit(job, ring, &vm->entity, AMDGPU_FENCE_OWNER_VM, &f); if (r) @@ -1062,28 +1117,32 @@ static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev, * * @adev: amdgpu_device pointer * @bo_va: requested BO and VM object - * @mem: ttm mem + * @clear: if true clear the entries * * Fill in the page table entries for @bo_va. * Returns 0 for success, -EINVAL for failure. - * - * Object have to be reserved and mutex must be locked! */ int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va, - struct ttm_mem_reg *mem) + bool clear) { struct amdgpu_vm *vm = bo_va->vm; struct amdgpu_bo_va_mapping *mapping; dma_addr_t *pages_addr = NULL; uint32_t gtt_flags, flags; + struct ttm_mem_reg *mem; struct fence *exclusive; uint64_t addr; int r; - if (mem) { + if (clear) { + mem = NULL; + addr = 0; + exclusive = NULL; + } else { struct ttm_dma_tt *ttm; + mem = &bo_va->bo->tbo.mem; addr = (u64)mem->start << PAGE_SHIFT; switch (mem->mem_type) { case TTM_PL_TT: @@ -1101,9 +1160,6 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev, } exclusive = reservation_object_get_excl(bo_va->bo->tbo.resv); - } else { - addr = 0; - exclusive = NULL; } flags = amdgpu_ttm_tt_pte_flags(adev, bo_va->bo->tbo.ttm, mem); @@ -1134,7 +1190,7 @@ 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); - if (!mem) + if (clear) list_add(&bo_va->vm_status, &vm->cleared); spin_unlock(&vm->status_lock); @@ -1197,7 +1253,7 @@ int amdgpu_vm_clear_invalids(struct amdgpu_device *adev, struct amdgpu_bo_va, vm_status); spin_unlock(&vm->status_lock); - r = amdgpu_vm_bo_update(adev, bo_va, NULL); + r = amdgpu_vm_bo_update(adev, bo_va, true); if (r) return r; @@ -1342,7 +1398,8 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev, r = amdgpu_bo_create(adev, AMDGPU_VM_PTE_COUNT * 8, AMDGPU_GPU_PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_VRAM, - AMDGPU_GEM_CREATE_NO_CPU_ACCESS, + AMDGPU_GEM_CREATE_NO_CPU_ACCESS | + AMDGPU_GEM_CREATE_SHADOW, NULL, resv, &pt); if (r) goto error_free; @@ -1541,7 +1598,8 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm) r = amdgpu_bo_create(adev, pd_size, align, true, AMDGPU_GEM_DOMAIN_VRAM, - AMDGPU_GEM_CREATE_NO_CPU_ACCESS, + AMDGPU_GEM_CREATE_NO_CPU_ACCESS | + AMDGPU_GEM_CREATE_SHADOW, NULL, NULL, &vm->page_directory); if (r) goto error_free_sched_entity; @@ -1597,10 +1655,16 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm) kfree(mapping); } - for (i = 0; i < amdgpu_vm_num_pdes(adev); i++) + for (i = 0; i < amdgpu_vm_num_pdes(adev); i++) { + if (vm->page_tables[i].entry.robj && + vm->page_tables[i].entry.robj->shadow) + amdgpu_bo_unref(&vm->page_tables[i].entry.robj->shadow); amdgpu_bo_unref(&vm->page_tables[i].entry.robj); + } drm_free_large(vm->page_tables); + if (vm->page_directory->shadow) + amdgpu_bo_unref(&vm->page_directory->shadow); amdgpu_bo_unref(&vm->page_directory); fence_put(vm->page_directory_fence); } diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_dp.c b/drivers/gpu/drm/amd/amdgpu/atombios_dp.c index 7f85c2c1d681..f81068ba4cc6 100644 --- a/drivers/gpu/drm/amd/amdgpu/atombios_dp.c +++ b/drivers/gpu/drm/amd/amdgpu/atombios_dp.c @@ -88,7 +88,6 @@ static int amdgpu_atombios_dp_process_aux_ch(struct amdgpu_i2c_chan *chan, /* timeout */ if (args.v2.ucReplyStatus == 1) { - DRM_DEBUG_KMS("dp_aux_ch timeout\n"); r = -ETIMEDOUT; goto done; } @@ -339,22 +338,21 @@ int amdgpu_atombios_dp_get_dpcd(struct amdgpu_connector *amdgpu_connector) { struct amdgpu_connector_atom_dig *dig_connector = amdgpu_connector->con_priv; u8 msg[DP_DPCD_SIZE]; - int ret, i; + int ret; - for (i = 0; i < 7; i++) { - ret = drm_dp_dpcd_read(&amdgpu_connector->ddc_bus->aux, DP_DPCD_REV, msg, - DP_DPCD_SIZE); - if (ret == DP_DPCD_SIZE) { - memcpy(dig_connector->dpcd, msg, DP_DPCD_SIZE); + ret = drm_dp_dpcd_read(&amdgpu_connector->ddc_bus->aux, DP_DPCD_REV, + msg, DP_DPCD_SIZE); + if (ret == DP_DPCD_SIZE) { + memcpy(dig_connector->dpcd, msg, DP_DPCD_SIZE); - DRM_DEBUG_KMS("DPCD: %*ph\n", (int)sizeof(dig_connector->dpcd), - dig_connector->dpcd); + DRM_DEBUG_KMS("DPCD: %*ph\n", (int)sizeof(dig_connector->dpcd), + dig_connector->dpcd); - amdgpu_atombios_dp_probe_oui(amdgpu_connector); + amdgpu_atombios_dp_probe_oui(amdgpu_connector); - return 0; - } + return 0; } + dig_connector->dpcd[0] = 0; return -EINVAL; } diff --git a/drivers/gpu/drm/amd/amdgpu/ci_dpm.c b/drivers/gpu/drm/amd/amdgpu/ci_dpm.c index a5c94b482459..a0d63a293bb0 100644 --- a/drivers/gpu/drm/amd/amdgpu/ci_dpm.c +++ b/drivers/gpu/drm/amd/amdgpu/ci_dpm.c @@ -5874,7 +5874,10 @@ static int ci_dpm_init(struct amdgpu_device *adev) pi->pcie_dpm_key_disabled = 0; pi->thermal_sclk_dpm_enabled = 0; - pi->caps_sclk_ds = true; + if (amdgpu_sclk_deep_sleep_en) + pi->caps_sclk_ds = true; + else + pi->caps_sclk_ds = false; pi->mclk_strobe_mode_threshold = 40000; pi->mclk_stutter_mode_threshold = 40000; diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c index 4efc901f658c..825de800b798 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik.c +++ b/drivers/gpu/drm/amd/amdgpu/cik.c @@ -67,6 +67,7 @@ #include "amdgpu_amdkfd.h" #include "amdgpu_powerplay.h" +#include "dce_virtual.h" /* * Indirect registers accessor @@ -1708,6 +1709,74 @@ static const struct amdgpu_ip_block_version bonaire_ip_blocks[] = }, }; +static const struct amdgpu_ip_block_version bonaire_ip_blocks_vd[] = +{ + /* ORDER MATTERS! */ + { + .type = AMD_IP_BLOCK_TYPE_COMMON, + .major = 1, + .minor = 0, + .rev = 0, + .funcs = &cik_common_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GMC, + .major = 7, + .minor = 0, + .rev = 0, + .funcs = &gmc_v7_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_IH, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &cik_ih_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SMC, + .major = 7, + .minor = 0, + .rev = 0, + .funcs = &amdgpu_pp_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_DCE, + .major = 8, + .minor = 2, + .rev = 0, + .funcs = &dce_virtual_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GFX, + .major = 7, + .minor = 2, + .rev = 0, + .funcs = &gfx_v7_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SDMA, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &cik_sdma_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_UVD, + .major = 4, + .minor = 2, + .rev = 0, + .funcs = &uvd_v4_2_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_VCE, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &vce_v2_0_ip_funcs, + }, +}; + static const struct amdgpu_ip_block_version hawaii_ip_blocks[] = { /* ORDER MATTERS! */ @@ -1776,6 +1845,74 @@ static const struct amdgpu_ip_block_version hawaii_ip_blocks[] = }, }; +static const struct amdgpu_ip_block_version hawaii_ip_blocks_vd[] = +{ + /* ORDER MATTERS! */ + { + .type = AMD_IP_BLOCK_TYPE_COMMON, + .major = 1, + .minor = 0, + .rev = 0, + .funcs = &cik_common_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GMC, + .major = 7, + .minor = 0, + .rev = 0, + .funcs = &gmc_v7_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_IH, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &cik_ih_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SMC, + .major = 7, + .minor = 0, + .rev = 0, + .funcs = &amdgpu_pp_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_DCE, + .major = 8, + .minor = 5, + .rev = 0, + .funcs = &dce_virtual_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GFX, + .major = 7, + .minor = 3, + .rev = 0, + .funcs = &gfx_v7_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SDMA, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &cik_sdma_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_UVD, + .major = 4, + .minor = 2, + .rev = 0, + .funcs = &uvd_v4_2_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_VCE, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &vce_v2_0_ip_funcs, + }, +}; + static const struct amdgpu_ip_block_version kabini_ip_blocks[] = { /* ORDER MATTERS! */ @@ -1844,6 +1981,74 @@ static const struct amdgpu_ip_block_version kabini_ip_blocks[] = }, }; +static const struct amdgpu_ip_block_version kabini_ip_blocks_vd[] = +{ + /* ORDER MATTERS! */ + { + .type = AMD_IP_BLOCK_TYPE_COMMON, + .major = 1, + .minor = 0, + .rev = 0, + .funcs = &cik_common_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GMC, + .major = 7, + .minor = 0, + .rev = 0, + .funcs = &gmc_v7_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_IH, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &cik_ih_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SMC, + .major = 7, + .minor = 0, + .rev = 0, + .funcs = &amdgpu_pp_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_DCE, + .major = 8, + .minor = 3, + .rev = 0, + .funcs = &dce_virtual_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GFX, + .major = 7, + .minor = 2, + .rev = 0, + .funcs = &gfx_v7_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SDMA, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &cik_sdma_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_UVD, + .major = 4, + .minor = 2, + .rev = 0, + .funcs = &uvd_v4_2_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_VCE, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &vce_v2_0_ip_funcs, + }, +}; + static const struct amdgpu_ip_block_version mullins_ip_blocks[] = { /* ORDER MATTERS! */ @@ -1912,6 +2117,74 @@ static const struct amdgpu_ip_block_version mullins_ip_blocks[] = }, }; +static const struct amdgpu_ip_block_version mullins_ip_blocks_vd[] = +{ + /* ORDER MATTERS! */ + { + .type = AMD_IP_BLOCK_TYPE_COMMON, + .major = 1, + .minor = 0, + .rev = 0, + .funcs = &cik_common_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GMC, + .major = 7, + .minor = 0, + .rev = 0, + .funcs = &gmc_v7_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_IH, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &cik_ih_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SMC, + .major = 7, + .minor = 0, + .rev = 0, + .funcs = &amdgpu_pp_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_DCE, + .major = 8, + .minor = 3, + .rev = 0, + .funcs = &dce_virtual_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GFX, + .major = 7, + .minor = 2, + .rev = 0, + .funcs = &gfx_v7_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SDMA, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &cik_sdma_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_UVD, + .major = 4, + .minor = 2, + .rev = 0, + .funcs = &uvd_v4_2_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_VCE, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &vce_v2_0_ip_funcs, + }, +}; + static const struct amdgpu_ip_block_version kaveri_ip_blocks[] = { /* ORDER MATTERS! */ @@ -1980,32 +2253,128 @@ static const struct amdgpu_ip_block_version kaveri_ip_blocks[] = }, }; +static const struct amdgpu_ip_block_version kaveri_ip_blocks_vd[] = +{ + /* ORDER MATTERS! */ + { + .type = AMD_IP_BLOCK_TYPE_COMMON, + .major = 1, + .minor = 0, + .rev = 0, + .funcs = &cik_common_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GMC, + .major = 7, + .minor = 0, + .rev = 0, + .funcs = &gmc_v7_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_IH, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &cik_ih_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SMC, + .major = 7, + .minor = 0, + .rev = 0, + .funcs = &amdgpu_pp_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_DCE, + .major = 8, + .minor = 1, + .rev = 0, + .funcs = &dce_virtual_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GFX, + .major = 7, + .minor = 1, + .rev = 0, + .funcs = &gfx_v7_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SDMA, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &cik_sdma_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_UVD, + .major = 4, + .minor = 2, + .rev = 0, + .funcs = &uvd_v4_2_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_VCE, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &vce_v2_0_ip_funcs, + }, +}; + int cik_set_ip_blocks(struct amdgpu_device *adev) { - switch (adev->asic_type) { - case CHIP_BONAIRE: - adev->ip_blocks = bonaire_ip_blocks; - adev->num_ip_blocks = ARRAY_SIZE(bonaire_ip_blocks); - break; - case CHIP_HAWAII: - adev->ip_blocks = hawaii_ip_blocks; - adev->num_ip_blocks = ARRAY_SIZE(hawaii_ip_blocks); - break; - case CHIP_KAVERI: - adev->ip_blocks = kaveri_ip_blocks; - adev->num_ip_blocks = ARRAY_SIZE(kaveri_ip_blocks); - break; - case CHIP_KABINI: - adev->ip_blocks = kabini_ip_blocks; - adev->num_ip_blocks = ARRAY_SIZE(kabini_ip_blocks); - break; - case CHIP_MULLINS: - adev->ip_blocks = mullins_ip_blocks; - adev->num_ip_blocks = ARRAY_SIZE(mullins_ip_blocks); - break; - default: - /* FIXME: not supported yet */ - return -EINVAL; + if (adev->enable_virtual_display) { + switch (adev->asic_type) { + case CHIP_BONAIRE: + adev->ip_blocks = bonaire_ip_blocks_vd; + adev->num_ip_blocks = ARRAY_SIZE(bonaire_ip_blocks_vd); + break; + case CHIP_HAWAII: + adev->ip_blocks = hawaii_ip_blocks_vd; + adev->num_ip_blocks = ARRAY_SIZE(hawaii_ip_blocks_vd); + break; + case CHIP_KAVERI: + adev->ip_blocks = kaveri_ip_blocks_vd; + adev->num_ip_blocks = ARRAY_SIZE(kaveri_ip_blocks_vd); + break; + case CHIP_KABINI: + adev->ip_blocks = kabini_ip_blocks_vd; + adev->num_ip_blocks = ARRAY_SIZE(kabini_ip_blocks_vd); + break; + case CHIP_MULLINS: + adev->ip_blocks = mullins_ip_blocks_vd; + adev->num_ip_blocks = ARRAY_SIZE(mullins_ip_blocks_vd); + break; + default: + /* FIXME: not supported yet */ + return -EINVAL; + } + } else { + switch (adev->asic_type) { + case CHIP_BONAIRE: + adev->ip_blocks = bonaire_ip_blocks; + adev->num_ip_blocks = ARRAY_SIZE(bonaire_ip_blocks); + break; + case CHIP_HAWAII: + adev->ip_blocks = hawaii_ip_blocks; + adev->num_ip_blocks = ARRAY_SIZE(hawaii_ip_blocks); + break; + case CHIP_KAVERI: + adev->ip_blocks = kaveri_ip_blocks; + adev->num_ip_blocks = ARRAY_SIZE(kaveri_ip_blocks); + break; + case CHIP_KABINI: + adev->ip_blocks = kabini_ip_blocks; + adev->num_ip_blocks = ARRAY_SIZE(kabini_ip_blocks); + break; + case CHIP_MULLINS: + adev->ip_blocks = mullins_ip_blocks; + adev->num_ip_blocks = ARRAY_SIZE(mullins_ip_blocks); + break; + default: + /* FIXME: not supported yet */ + return -EINVAL; + } } return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c index ee6466912497..e71cd12104b3 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c +++ b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c @@ -694,24 +694,16 @@ static void cik_sdma_vm_copy_pte(struct amdgpu_ib *ib, uint64_t pe, uint64_t src, unsigned count) { - while (count) { - unsigned bytes = count * 8; - if (bytes > 0x1FFFF8) - bytes = 0x1FFFF8; - - ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_COPY, - SDMA_WRITE_SUB_OPCODE_LINEAR, 0); - ib->ptr[ib->length_dw++] = bytes; - ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */ - ib->ptr[ib->length_dw++] = lower_32_bits(src); - ib->ptr[ib->length_dw++] = upper_32_bits(src); - ib->ptr[ib->length_dw++] = lower_32_bits(pe); - ib->ptr[ib->length_dw++] = upper_32_bits(pe); - - pe += bytes; - src += bytes; - count -= bytes / 8; - } + unsigned bytes = count * 8; + + ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_COPY, + SDMA_WRITE_SUB_OPCODE_LINEAR, 0); + ib->ptr[ib->length_dw++] = bytes; + ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */ + ib->ptr[ib->length_dw++] = lower_32_bits(src); + ib->ptr[ib->length_dw++] = upper_32_bits(src); + ib->ptr[ib->length_dw++] = lower_32_bits(pe); + ib->ptr[ib->length_dw++] = upper_32_bits(pe); } /** @@ -719,39 +711,27 @@ static void cik_sdma_vm_copy_pte(struct amdgpu_ib *ib, * * @ib: indirect buffer to fill with commands * @pe: addr of the page entry - * @addr: dst addr to write into pe + * @value: dst addr to write into pe * @count: number of page entries to update * @incr: increase next addr by incr bytes - * @flags: access flags * * Update PTEs by writing them manually using sDMA (CIK). */ -static void cik_sdma_vm_write_pte(struct amdgpu_ib *ib, - const dma_addr_t *pages_addr, uint64_t pe, - uint64_t addr, unsigned count, - uint32_t incr, uint32_t flags) +static void cik_sdma_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe, + uint64_t value, unsigned count, + uint32_t incr) { - uint64_t value; - unsigned ndw; - - while (count) { - ndw = count * 2; - if (ndw > 0xFFFFE) - ndw = 0xFFFFE; - - /* for non-physically contiguous pages (system) */ - ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_WRITE, - SDMA_WRITE_SUB_OPCODE_LINEAR, 0); - ib->ptr[ib->length_dw++] = pe; - ib->ptr[ib->length_dw++] = upper_32_bits(pe); - ib->ptr[ib->length_dw++] = ndw; - for (; ndw > 0; ndw -= 2, --count, pe += 8) { - value = amdgpu_vm_map_gart(pages_addr, addr); - addr += incr; - value |= flags; - ib->ptr[ib->length_dw++] = value; - ib->ptr[ib->length_dw++] = upper_32_bits(value); - } + unsigned ndw = count * 2; + + ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_WRITE, + SDMA_WRITE_SUB_OPCODE_LINEAR, 0); + ib->ptr[ib->length_dw++] = lower_32_bits(pe); + ib->ptr[ib->length_dw++] = upper_32_bits(pe); + ib->ptr[ib->length_dw++] = ndw; + for (; ndw > 0; ndw -= 2) { + ib->ptr[ib->length_dw++] = lower_32_bits(value); + ib->ptr[ib->length_dw++] = upper_32_bits(value); + value += incr; } } @@ -767,40 +747,21 @@ static void cik_sdma_vm_write_pte(struct amdgpu_ib *ib, * * Update the page tables using sDMA (CIK). */ -static void cik_sdma_vm_set_pte_pde(struct amdgpu_ib *ib, - uint64_t pe, +static void cik_sdma_vm_set_pte_pde(struct amdgpu_ib *ib, uint64_t pe, uint64_t addr, unsigned count, uint32_t incr, uint32_t flags) { - uint64_t value; - unsigned ndw; - - while (count) { - ndw = count; - if (ndw > 0x7FFFF) - ndw = 0x7FFFF; - - if (flags & AMDGPU_PTE_VALID) - value = addr; - else - value = 0; - - /* for physically contiguous pages (vram) */ - ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_GENERATE_PTE_PDE, 0, 0); - ib->ptr[ib->length_dw++] = pe; /* dst addr */ - ib->ptr[ib->length_dw++] = upper_32_bits(pe); - ib->ptr[ib->length_dw++] = flags; /* mask */ - ib->ptr[ib->length_dw++] = 0; - ib->ptr[ib->length_dw++] = value; /* value */ - ib->ptr[ib->length_dw++] = upper_32_bits(value); - ib->ptr[ib->length_dw++] = incr; /* increment size */ - ib->ptr[ib->length_dw++] = 0; - ib->ptr[ib->length_dw++] = ndw; /* number of entries */ - - pe += ndw * 8; - addr += ndw * incr; - count -= ndw; - } + /* for physically contiguous pages (vram) */ + ib->ptr[ib->length_dw++] = SDMA_PACKET(SDMA_OPCODE_GENERATE_PTE_PDE, 0, 0); + ib->ptr[ib->length_dw++] = lower_32_bits(pe); /* dst addr */ + ib->ptr[ib->length_dw++] = upper_32_bits(pe); + ib->ptr[ib->length_dw++] = flags; /* mask */ + ib->ptr[ib->length_dw++] = 0; + ib->ptr[ib->length_dw++] = lower_32_bits(addr); /* value */ + ib->ptr[ib->length_dw++] = upper_32_bits(addr); + ib->ptr[ib->length_dw++] = incr; /* increment size */ + ib->ptr[ib->length_dw++] = 0; + ib->ptr[ib->length_dw++] = count; /* number of entries */ } /** diff --git a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c index 2a11413ed54a..794c5f36ca68 100644 --- a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c +++ b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c @@ -435,7 +435,11 @@ static int cz_dpm_init(struct amdgpu_device *adev) pi->caps_td_ramping = true; pi->caps_tcp_ramping = true; } - pi->caps_sclk_ds = true; + if (amdgpu_sclk_deep_sleep_en) + pi->caps_sclk_ds = true; + else + pi->caps_sclk_ds = false; + pi->voting_clients = 0x00c00033; pi->auto_thermal_throttling_enabled = true; pi->bapm_enabled = false; @@ -2108,29 +2112,58 @@ static void cz_dpm_powergate_uvd(struct amdgpu_device *adev, bool gate) /* disable clockgating so we can properly shut down the block */ ret = amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD, AMD_CG_STATE_UNGATE); + if (ret) { + DRM_ERROR("UVD DPM Power Gating failed to set clockgating state\n"); + return; + } + /* shutdown the UVD block */ ret = amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD, AMD_PG_STATE_GATE); - /* XXX: check for errors */ + + if (ret) { + DRM_ERROR("UVD DPM Power Gating failed to set powergating state\n"); + return; + } } cz_update_uvd_dpm(adev, gate); - if (pi->caps_uvd_pg) + if (pi->caps_uvd_pg) { /* power off the UVD block */ - cz_send_msg_to_smc(adev, PPSMC_MSG_UVDPowerOFF); + ret = cz_send_msg_to_smc(adev, PPSMC_MSG_UVDPowerOFF); + if (ret) { + DRM_ERROR("UVD DPM Power Gating failed to send SMU PowerOFF message\n"); + return; + } + } } else { if (pi->caps_uvd_pg) { /* power on the UVD block */ if (pi->uvd_dynamic_pg) - cz_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_UVDPowerON, 1); + ret = cz_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_UVDPowerON, 1); else - cz_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_UVDPowerON, 0); + ret = cz_send_msg_to_smc_with_parameter(adev, PPSMC_MSG_UVDPowerON, 0); + + if (ret) { + DRM_ERROR("UVD DPM Power Gating Failed to send SMU PowerON message\n"); + return; + } + /* re-init the UVD block */ ret = amdgpu_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_UVD, AMD_PG_STATE_UNGATE); + + if (ret) { + DRM_ERROR("UVD DPM Power Gating Failed to set powergating state\n"); + return; + } + /* enable clockgating. hw will dynamically gate/ungate clocks on the fly */ ret = amdgpu_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_UVD, AMD_CG_STATE_GATE); - /* XXX: check for errors */ + if (ret) { + DRM_ERROR("UVD DPM Power Gating Failed to set clockgating state\n"); + return; + } } cz_update_uvd_dpm(adev, gate); } diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c index 741da36cd8b6..bc5bb4eb9625 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c @@ -646,8 +646,8 @@ static void dce_v10_0_resume_mc_access(struct amdgpu_device *adev, if (save->crtc_enabled[i]) { tmp = RREG32(mmMASTER_UPDATE_MODE + crtc_offsets[i]); - if (REG_GET_FIELD(tmp, MASTER_UPDATE_MODE, MASTER_UPDATE_MODE) != 3) { - tmp = REG_SET_FIELD(tmp, MASTER_UPDATE_MODE, MASTER_UPDATE_MODE, 3); + if (REG_GET_FIELD(tmp, MASTER_UPDATE_MODE, MASTER_UPDATE_MODE) != 0) { + tmp = REG_SET_FIELD(tmp, MASTER_UPDATE_MODE, MASTER_UPDATE_MODE, 0); WREG32(mmMASTER_UPDATE_MODE + crtc_offsets[i], tmp); } tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]); @@ -712,6 +712,45 @@ static void dce_v10_0_set_vga_render_state(struct amdgpu_device *adev, WREG32(mmVGA_RENDER_CONTROL, tmp); } +static int dce_v10_0_get_num_crtc(struct amdgpu_device *adev) +{ + int num_crtc = 0; + + switch (adev->asic_type) { + case CHIP_FIJI: + case CHIP_TONGA: + num_crtc = 6; + break; + default: + num_crtc = 0; + } + return num_crtc; +} + +void dce_v10_0_disable_dce(struct amdgpu_device *adev) +{ + /*Disable VGA render and enabled crtc, if has DCE engine*/ + if (amdgpu_atombios_has_dce_engine_info(adev)) { + u32 tmp; + int crtc_enabled, i; + + dce_v10_0_set_vga_render_state(adev, false); + + /*Disable crtc*/ + for (i = 0; i < dce_v10_0_get_num_crtc(adev); i++) { + crtc_enabled = REG_GET_FIELD(RREG32(mmCRTC_CONTROL + crtc_offsets[i]), + CRTC_CONTROL, CRTC_MASTER_EN); + if (crtc_enabled) { + WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1); + tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]); + tmp = REG_SET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN, 0); + WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp); + WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0); + } + } + } +} + static void dce_v10_0_program_fmt(struct drm_encoder *encoder) { struct drm_device *dev = encoder->dev; @@ -2277,8 +2316,8 @@ static int dce_v10_0_crtc_do_set_base(struct drm_crtc *crtc, WREG32(mmVIEWPORT_SIZE + amdgpu_crtc->crtc_offset, (viewport_w << 16) | viewport_h); - /* set pageflip to happen only at start of vblank interval (front porch) */ - WREG32(mmMASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 3); + /* set pageflip to happen anywhere in vblank interval */ + WREG32(mmMASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 0); if (!atomic && fb && fb != crtc->primary->fb) { amdgpu_fb = to_amdgpu_framebuffer(fb); @@ -2700,7 +2739,7 @@ static const struct drm_crtc_funcs dce_v10_0_crtc_funcs = { .gamma_set = dce_v10_0_crtc_gamma_set, .set_config = amdgpu_crtc_set_config, .destroy = dce_v10_0_crtc_destroy, - .page_flip = amdgpu_crtc_page_flip, + .page_flip_target = amdgpu_crtc_page_flip_target, }; static void dce_v10_0_crtc_dpms(struct drm_crtc *crtc, int mode) @@ -2964,10 +3003,11 @@ static int dce_v10_0_early_init(void *handle) dce_v10_0_set_display_funcs(adev); dce_v10_0_set_irq_funcs(adev); + adev->mode_info.num_crtc = dce_v10_0_get_num_crtc(adev); + switch (adev->asic_type) { case CHIP_FIJI: case CHIP_TONGA: - adev->mode_info.num_crtc = 6; /* XXX 7??? */ adev->mode_info.num_hpd = 6; adev->mode_info.num_dig = 7; break; @@ -3143,11 +3183,26 @@ static int dce_v10_0_wait_for_idle(void *handle) return 0; } +static int dce_v10_0_check_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (dce_v10_0_is_display_hung(adev)) + adev->ip_block_status[AMD_IP_BLOCK_TYPE_DCE].hang = true; + else + adev->ip_block_status[AMD_IP_BLOCK_TYPE_DCE].hang = false; + + return 0; +} + static int dce_v10_0_soft_reset(void *handle) { u32 srbm_soft_reset = 0, tmp; struct amdgpu_device *adev = (struct amdgpu_device *)handle; + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_DCE].hang) + return 0; + if (dce_v10_0_is_display_hung(adev)) srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_DC_MASK; @@ -3514,6 +3569,7 @@ const struct amd_ip_funcs dce_v10_0_ip_funcs = { .resume = dce_v10_0_resume, .is_idle = dce_v10_0_is_idle, .wait_for_idle = dce_v10_0_wait_for_idle, + .check_soft_reset = dce_v10_0_check_soft_reset, .soft_reset = dce_v10_0_soft_reset, .set_clockgating_state = dce_v10_0_set_clockgating_state, .set_powergating_state = dce_v10_0_set_powergating_state, diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.h b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.h index 1bfa48ddd8a6..e3dc04d293e4 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.h +++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.h @@ -26,4 +26,6 @@ extern const struct amd_ip_funcs dce_v10_0_ip_funcs; +void dce_v10_0_disable_dce(struct amdgpu_device *adev); + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c index 2282eb60aba6..b93eba077950 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c @@ -673,6 +673,53 @@ static void dce_v11_0_set_vga_render_state(struct amdgpu_device *adev, WREG32(mmVGA_RENDER_CONTROL, tmp); } +static int dce_v11_0_get_num_crtc (struct amdgpu_device *adev) +{ + int num_crtc = 0; + + switch (adev->asic_type) { + case CHIP_CARRIZO: + num_crtc = 3; + break; + case CHIP_STONEY: + num_crtc = 2; + break; + case CHIP_POLARIS10: + num_crtc = 6; + break; + case CHIP_POLARIS11: + num_crtc = 5; + break; + default: + num_crtc = 0; + } + return num_crtc; +} + +void dce_v11_0_disable_dce(struct amdgpu_device *adev) +{ + /*Disable VGA render and enabled crtc, if has DCE engine*/ + if (amdgpu_atombios_has_dce_engine_info(adev)) { + u32 tmp; + int crtc_enabled, i; + + dce_v11_0_set_vga_render_state(adev, false); + + /*Disable crtc*/ + for (i = 0; i < dce_v11_0_get_num_crtc(adev); i++) { + crtc_enabled = REG_GET_FIELD(RREG32(mmCRTC_CONTROL + crtc_offsets[i]), + CRTC_CONTROL, CRTC_MASTER_EN); + if (crtc_enabled) { + WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1); + tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]); + tmp = REG_SET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN, 0); + WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp); + WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0); + } + } + } +} + static void dce_v11_0_program_fmt(struct drm_encoder *encoder) { struct drm_device *dev = encoder->dev; @@ -2252,8 +2299,8 @@ static int dce_v11_0_crtc_do_set_base(struct drm_crtc *crtc, WREG32(mmVIEWPORT_SIZE + amdgpu_crtc->crtc_offset, (viewport_w << 16) | viewport_h); - /* set pageflip to happen only at start of vblank interval (front porch) */ - WREG32(mmCRTC_MASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 3); + /* set pageflip to happen anywhere in vblank interval */ + WREG32(mmCRTC_MASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 0); if (!atomic && fb && fb != crtc->primary->fb) { amdgpu_fb = to_amdgpu_framebuffer(fb); @@ -2710,7 +2757,7 @@ static const struct drm_crtc_funcs dce_v11_0_crtc_funcs = { .gamma_set = dce_v11_0_crtc_gamma_set, .set_config = amdgpu_crtc_set_config, .destroy = dce_v11_0_crtc_destroy, - .page_flip = amdgpu_crtc_page_flip, + .page_flip_target = amdgpu_crtc_page_flip_target, }; static void dce_v11_0_crtc_dpms(struct drm_crtc *crtc, int mode) @@ -3001,24 +3048,22 @@ static int dce_v11_0_early_init(void *handle) dce_v11_0_set_display_funcs(adev); dce_v11_0_set_irq_funcs(adev); + adev->mode_info.num_crtc = dce_v11_0_get_num_crtc(adev); + switch (adev->asic_type) { case CHIP_CARRIZO: - adev->mode_info.num_crtc = 3; adev->mode_info.num_hpd = 6; adev->mode_info.num_dig = 9; break; case CHIP_STONEY: - adev->mode_info.num_crtc = 2; adev->mode_info.num_hpd = 6; adev->mode_info.num_dig = 9; break; case CHIP_POLARIS10: - adev->mode_info.num_crtc = 6; adev->mode_info.num_hpd = 6; adev->mode_info.num_dig = 6; break; case CHIP_POLARIS11: - adev->mode_info.num_crtc = 5; adev->mode_info.num_hpd = 5; adev->mode_info.num_dig = 5; break; diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.h b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.h index 84e4618f5253..1f58a65ba2ef 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.h +++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.h @@ -26,4 +26,6 @@ extern const struct amd_ip_funcs dce_v11_0_ip_funcs; +void dce_v11_0_disable_dce(struct amdgpu_device *adev); + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c index 8b7ad345771f..abd5213dfe18 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c @@ -604,6 +604,52 @@ static void dce_v8_0_set_vga_render_state(struct amdgpu_device *adev, WREG32(mmVGA_RENDER_CONTROL, tmp); } +static int dce_v8_0_get_num_crtc(struct amdgpu_device *adev) +{ + int num_crtc = 0; + + switch (adev->asic_type) { + case CHIP_BONAIRE: + case CHIP_HAWAII: + num_crtc = 6; + break; + case CHIP_KAVERI: + num_crtc = 4; + break; + case CHIP_KABINI: + case CHIP_MULLINS: + num_crtc = 2; + break; + default: + num_crtc = 0; + } + return num_crtc; +} + +void dce_v8_0_disable_dce(struct amdgpu_device *adev) +{ + /*Disable VGA render and enabled crtc, if has DCE engine*/ + if (amdgpu_atombios_has_dce_engine_info(adev)) { + u32 tmp; + int crtc_enabled, i; + + dce_v8_0_set_vga_render_state(adev, false); + + /*Disable crtc*/ + for (i = 0; i < dce_v8_0_get_num_crtc(adev); i++) { + crtc_enabled = REG_GET_FIELD(RREG32(mmCRTC_CONTROL + crtc_offsets[i]), + CRTC_CONTROL, CRTC_MASTER_EN); + if (crtc_enabled) { + WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1); + tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]); + tmp = REG_SET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN, 0); + WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp); + WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0); + } + } + } +} + static void dce_v8_0_program_fmt(struct drm_encoder *encoder) { struct drm_device *dev = encoder->dev; @@ -1501,13 +1547,13 @@ static void dce_v8_0_audio_write_sad_regs(struct drm_encoder *encoder) if (sad->format == eld_reg_to_type[i][1]) { if (sad->channels > max_channels) { - value = (sad->channels << - AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__MAX_CHANNELS__SHIFT) | - (sad->byte2 << - AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__DESCRIPTOR_BYTE_2__SHIFT) | - (sad->freq << - AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__SUPPORTED_FREQUENCIES__SHIFT); - max_channels = sad->channels; + value = (sad->channels << + AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__MAX_CHANNELS__SHIFT) | + (sad->byte2 << + AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__DESCRIPTOR_BYTE_2__SHIFT) | + (sad->freq << + AZALIA_F0_CODEC_PIN_CONTROL_AUDIO_DESCRIPTOR0__SUPPORTED_FREQUENCIES__SHIFT); + max_channels = sad->channels; } if (sad->format == HDMI_AUDIO_CODING_TYPE_PCM) @@ -1613,7 +1659,7 @@ static void dce_v8_0_afmt_update_ACR(struct drm_encoder *encoder, uint32_t clock struct amdgpu_encoder_atom_dig *dig = amdgpu_encoder->enc_priv; uint32_t offset = dig->afmt->offset; - WREG32(mmHDMI_ACR_32_0 + offset, (acr.cts_32khz << HDMI_ACR_44_0__HDMI_ACR_CTS_44__SHIFT)); + WREG32(mmHDMI_ACR_32_0 + offset, (acr.cts_32khz << HDMI_ACR_32_0__HDMI_ACR_CTS_32__SHIFT)); WREG32(mmHDMI_ACR_32_1 + offset, acr.n_32khz); WREG32(mmHDMI_ACR_44_0 + offset, (acr.cts_44_1khz << HDMI_ACR_44_0__HDMI_ACR_CTS_44__SHIFT)); @@ -1693,6 +1739,7 @@ static void dce_v8_0_afmt_setmode(struct drm_encoder *encoder, /* Silent, r600_hdmi_enable will raise WARN for us */ if (!dig->afmt->enabled) return; + offset = dig->afmt->offset; /* hdmi deep color mode general control packets setup, if bpc > 8 */ @@ -1817,7 +1864,7 @@ static void dce_v8_0_afmt_setmode(struct drm_encoder *encoder, WREG32_OR(mmHDMI_INFOFRAME_CONTROL0 + offset, HDMI_INFOFRAME_CONTROL0__HDMI_AVI_INFO_SEND_MASK | /* enable AVI info frames */ - HDMI_INFOFRAME_CONTROL0__HDMI_AVI_INFO_SEND_MASK); /* required for audio info values to be updated */ + HDMI_INFOFRAME_CONTROL0__HDMI_AVI_INFO_CONT_MASK); /* required for audio info values to be updated */ WREG32_P(mmHDMI_INFOFRAME_CONTROL1 + offset, (2 << HDMI_INFOFRAME_CONTROL1__HDMI_AVI_INFO_LINE__SHIFT), /* anything other than 0 */ @@ -1826,13 +1873,12 @@ static void dce_v8_0_afmt_setmode(struct drm_encoder *encoder, WREG32_OR(mmAFMT_AUDIO_PACKET_CONTROL + offset, AFMT_AUDIO_PACKET_CONTROL__AFMT_AUDIO_SAMPLE_SEND_MASK); /* send audio packets */ - /* it's unknown what these bits do excatly, but it's indeed quite useful for debugging */ WREG32(mmAFMT_RAMP_CONTROL0 + offset, 0x00FFFFFF); WREG32(mmAFMT_RAMP_CONTROL1 + offset, 0x007FFFFF); WREG32(mmAFMT_RAMP_CONTROL2 + offset, 0x00000001); WREG32(mmAFMT_RAMP_CONTROL3 + offset, 0x00000001); - /* enable audio after to setting up hw */ + /* enable audio after setting up hw */ dce_v8_0_audio_enable(adev, dig->afmt->pin, true); } @@ -2000,7 +2046,7 @@ static int dce_v8_0_crtc_do_set_base(struct drm_crtc *crtc, case DRM_FORMAT_XRGB4444: case DRM_FORMAT_ARGB4444: fb_format = ((GRPH_DEPTH_16BPP << GRPH_CONTROL__GRPH_DEPTH__SHIFT) | - (GRPH_FORMAT_ARGB1555 << GRPH_CONTROL__GRPH_FORMAT__SHIFT)); + (GRPH_FORMAT_ARGB4444 << GRPH_CONTROL__GRPH_FORMAT__SHIFT)); #ifdef __BIG_ENDIAN fb_swap = (GRPH_ENDIAN_8IN16 << GRPH_SWAP_CNTL__GRPH_ENDIAN_SWAP__SHIFT); #endif @@ -2139,8 +2185,8 @@ static int dce_v8_0_crtc_do_set_base(struct drm_crtc *crtc, WREG32(mmVIEWPORT_SIZE + amdgpu_crtc->crtc_offset, (viewport_w << 16) | viewport_h); - /* set pageflip to happen only at start of vblank interval (front porch) */ - WREG32(mmMASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 3); + /* set pageflip to happen anywhere in vblank interval */ + WREG32(mmMASTER_UPDATE_MODE + amdgpu_crtc->crtc_offset, 0); if (!atomic && fb && fb != crtc->primary->fb) { amdgpu_fb = to_amdgpu_framebuffer(fb); @@ -2554,7 +2600,7 @@ static const struct drm_crtc_funcs dce_v8_0_crtc_funcs = { .gamma_set = dce_v8_0_crtc_gamma_set, .set_config = amdgpu_crtc_set_config, .destroy = dce_v8_0_crtc_destroy, - .page_flip = amdgpu_crtc_page_flip, + .page_flip_target = amdgpu_crtc_page_flip_target, }; static void dce_v8_0_crtc_dpms(struct drm_crtc *crtc, int mode) @@ -2655,7 +2701,7 @@ static void dce_v8_0_crtc_disable(struct drm_crtc *crtc) case ATOM_PPLL2: /* disable the ppll */ amdgpu_atombios_crtc_program_pll(crtc, amdgpu_crtc->crtc_id, amdgpu_crtc->pll_id, - 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss); + 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss); break; case ATOM_PPLL0: /* disable the ppll */ @@ -2805,21 +2851,20 @@ static int dce_v8_0_early_init(void *handle) dce_v8_0_set_display_funcs(adev); dce_v8_0_set_irq_funcs(adev); + adev->mode_info.num_crtc = dce_v8_0_get_num_crtc(adev); + switch (adev->asic_type) { case CHIP_BONAIRE: case CHIP_HAWAII: - adev->mode_info.num_crtc = 6; adev->mode_info.num_hpd = 6; adev->mode_info.num_dig = 6; break; case CHIP_KAVERI: - adev->mode_info.num_crtc = 4; adev->mode_info.num_hpd = 6; adev->mode_info.num_dig = 7; break; case CHIP_KABINI: case CHIP_MULLINS: - adev->mode_info.num_crtc = 2; adev->mode_info.num_hpd = 6; adev->mode_info.num_dig = 6; /* ? */ break; @@ -3238,7 +3283,6 @@ static int dce_v8_0_crtc_irq(struct amdgpu_device *adev, drm_handle_vblank(adev->ddev, crtc); } DRM_DEBUG("IH: D%d vblank\n", crtc + 1); - break; case 1: /* vline */ if (disp_int & interrupt_status_offsets[crtc].vline) @@ -3247,7 +3291,6 @@ static int dce_v8_0_crtc_irq(struct amdgpu_device *adev, DRM_DEBUG("IH: IH event w/o asserted irq bit?\n"); DRM_DEBUG("IH: D%d vline\n", crtc + 1); - break; default: DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data); diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.h b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.h index 77016852b252..7d0770c3a49b 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.h +++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.h @@ -26,4 +26,6 @@ extern const struct amd_ip_funcs dce_v8_0_ip_funcs; +void dce_v8_0_disable_dce(struct amdgpu_device *adev); + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/dce_virtual.c b/drivers/gpu/drm/amd/amdgpu/dce_virtual.c new file mode 100644 index 000000000000..00663a7b4053 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/dce_virtual.c @@ -0,0 +1,806 @@ +/* + * Copyright 2014 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "drmP.h" +#include "amdgpu.h" +#include "amdgpu_pm.h" +#include "amdgpu_i2c.h" +#include "atom.h" +#include "amdgpu_pll.h" +#include "amdgpu_connectors.h" +#ifdef CONFIG_DRM_AMDGPU_CIK +#include "dce_v8_0.h" +#endif +#include "dce_v10_0.h" +#include "dce_v11_0.h" +#include "dce_virtual.h" + +static void dce_virtual_set_display_funcs(struct amdgpu_device *adev); +static void dce_virtual_set_irq_funcs(struct amdgpu_device *adev); +static int dce_virtual_pageflip_irq(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry); + +/** + * dce_virtual_vblank_wait - vblank wait asic callback. + * + * @adev: amdgpu_device pointer + * @crtc: crtc to wait for vblank on + * + * Wait for vblank on the requested crtc (evergreen+). + */ +static void dce_virtual_vblank_wait(struct amdgpu_device *adev, int crtc) +{ + return; +} + +static u32 dce_virtual_vblank_get_counter(struct amdgpu_device *adev, int crtc) +{ + return 0; +} + +static void dce_virtual_page_flip(struct amdgpu_device *adev, + int crtc_id, u64 crtc_base, bool async) +{ + return; +} + +static int dce_virtual_crtc_get_scanoutpos(struct amdgpu_device *adev, int crtc, + u32 *vbl, u32 *position) +{ + *vbl = 0; + *position = 0; + + return -EINVAL; +} + +static bool dce_virtual_hpd_sense(struct amdgpu_device *adev, + enum amdgpu_hpd_id hpd) +{ + return true; +} + +static void dce_virtual_hpd_set_polarity(struct amdgpu_device *adev, + enum amdgpu_hpd_id hpd) +{ + return; +} + +static u32 dce_virtual_hpd_get_gpio_reg(struct amdgpu_device *adev) +{ + return 0; +} + +static bool dce_virtual_is_display_hung(struct amdgpu_device *adev) +{ + return false; +} + +void dce_virtual_stop_mc_access(struct amdgpu_device *adev, + struct amdgpu_mode_mc_save *save) +{ + switch (adev->asic_type) { + case CHIP_BONAIRE: + case CHIP_HAWAII: + case CHIP_KAVERI: + case CHIP_KABINI: + case CHIP_MULLINS: +#ifdef CONFIG_DRM_AMDGPU_CIK + dce_v8_0_disable_dce(adev); +#endif + break; + case CHIP_FIJI: + case CHIP_TONGA: + dce_v10_0_disable_dce(adev); + break; + case CHIP_CARRIZO: + case CHIP_STONEY: + case CHIP_POLARIS11: + case CHIP_POLARIS10: + dce_v11_0_disable_dce(adev); + break; + case CHIP_TOPAZ: + /* no DCE */ + return; + default: + DRM_ERROR("Virtual display unsupported ASIC type: 0x%X\n", adev->asic_type); + } + + return; +} +void dce_virtual_resume_mc_access(struct amdgpu_device *adev, + struct amdgpu_mode_mc_save *save) +{ + return; +} + +void dce_virtual_set_vga_render_state(struct amdgpu_device *adev, + bool render) +{ + return; +} + +/** + * dce_virtual_bandwidth_update - program display watermarks + * + * @adev: amdgpu_device pointer + * + * Calculate and program the display watermarks and line + * buffer allocation (CIK). + */ +static void dce_virtual_bandwidth_update(struct amdgpu_device *adev) +{ + return; +} + +static int dce_virtual_crtc_gamma_set(struct drm_crtc *crtc, u16 *red, + u16 *green, u16 *blue, uint32_t size) +{ + struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc); + int i; + + /* userspace palettes are always correct as is */ + for (i = 0; i < size; i++) { + amdgpu_crtc->lut_r[i] = red[i] >> 6; + amdgpu_crtc->lut_g[i] = green[i] >> 6; + amdgpu_crtc->lut_b[i] = blue[i] >> 6; + } + + return 0; +} + +static void dce_virtual_crtc_destroy(struct drm_crtc *crtc) +{ + struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc); + + drm_crtc_cleanup(crtc); + kfree(amdgpu_crtc); +} + +static const struct drm_crtc_funcs dce_virtual_crtc_funcs = { + .cursor_set2 = NULL, + .cursor_move = NULL, + .gamma_set = dce_virtual_crtc_gamma_set, + .set_config = amdgpu_crtc_set_config, + .destroy = dce_virtual_crtc_destroy, + .page_flip_target = amdgpu_crtc_page_flip_target, +}; + +static void dce_virtual_crtc_dpms(struct drm_crtc *crtc, int mode) +{ + struct drm_device *dev = crtc->dev; + struct amdgpu_device *adev = dev->dev_private; + struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc); + unsigned type; + + switch (mode) { + case DRM_MODE_DPMS_ON: + amdgpu_crtc->enabled = true; + /* Make sure VBLANK and PFLIP interrupts are still enabled */ + type = amdgpu_crtc_idx_to_irq_type(adev, amdgpu_crtc->crtc_id); + amdgpu_irq_update(adev, &adev->crtc_irq, type); + amdgpu_irq_update(adev, &adev->pageflip_irq, type); + drm_vblank_on(dev, amdgpu_crtc->crtc_id); + break; + case DRM_MODE_DPMS_STANDBY: + case DRM_MODE_DPMS_SUSPEND: + case DRM_MODE_DPMS_OFF: + drm_vblank_off(dev, amdgpu_crtc->crtc_id); + amdgpu_crtc->enabled = false; + break; + } +} + + +static void dce_virtual_crtc_prepare(struct drm_crtc *crtc) +{ + dce_virtual_crtc_dpms(crtc, DRM_MODE_DPMS_OFF); +} + +static void dce_virtual_crtc_commit(struct drm_crtc *crtc) +{ + dce_virtual_crtc_dpms(crtc, DRM_MODE_DPMS_ON); +} + +static void dce_virtual_crtc_disable(struct drm_crtc *crtc) +{ + struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc); + + dce_virtual_crtc_dpms(crtc, DRM_MODE_DPMS_OFF); + if (crtc->primary->fb) { + int r; + struct amdgpu_framebuffer *amdgpu_fb; + struct amdgpu_bo *rbo; + + amdgpu_fb = to_amdgpu_framebuffer(crtc->primary->fb); + rbo = gem_to_amdgpu_bo(amdgpu_fb->obj); + r = amdgpu_bo_reserve(rbo, false); + if (unlikely(r)) + DRM_ERROR("failed to reserve rbo before unpin\n"); + else { + amdgpu_bo_unpin(rbo); + amdgpu_bo_unreserve(rbo); + } + } + + amdgpu_crtc->pll_id = ATOM_PPLL_INVALID; + amdgpu_crtc->encoder = NULL; + amdgpu_crtc->connector = NULL; +} + +static int dce_virtual_crtc_mode_set(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode, + int x, int y, struct drm_framebuffer *old_fb) +{ + struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc); + + /* update the hw version fpr dpm */ + amdgpu_crtc->hw_mode = *adjusted_mode; + + return 0; +} + +static bool dce_virtual_crtc_mode_fixup(struct drm_crtc *crtc, + const struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc); + struct drm_device *dev = crtc->dev; + struct drm_encoder *encoder; + + /* assign the encoder to the amdgpu crtc to avoid repeated lookups later */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + if (encoder->crtc == crtc) { + amdgpu_crtc->encoder = encoder; + amdgpu_crtc->connector = amdgpu_get_connector_for_encoder(encoder); + break; + } + } + if ((amdgpu_crtc->encoder == NULL) || (amdgpu_crtc->connector == NULL)) { + amdgpu_crtc->encoder = NULL; + amdgpu_crtc->connector = NULL; + return false; + } + + return true; +} + + +static int dce_virtual_crtc_set_base(struct drm_crtc *crtc, int x, int y, + struct drm_framebuffer *old_fb) +{ + return 0; +} + +static void dce_virtual_crtc_load_lut(struct drm_crtc *crtc) +{ + return; +} + +static int dce_virtual_crtc_set_base_atomic(struct drm_crtc *crtc, + struct drm_framebuffer *fb, + int x, int y, enum mode_set_atomic state) +{ + return 0; +} + +static const struct drm_crtc_helper_funcs dce_virtual_crtc_helper_funcs = { + .dpms = dce_virtual_crtc_dpms, + .mode_fixup = dce_virtual_crtc_mode_fixup, + .mode_set = dce_virtual_crtc_mode_set, + .mode_set_base = dce_virtual_crtc_set_base, + .mode_set_base_atomic = dce_virtual_crtc_set_base_atomic, + .prepare = dce_virtual_crtc_prepare, + .commit = dce_virtual_crtc_commit, + .load_lut = dce_virtual_crtc_load_lut, + .disable = dce_virtual_crtc_disable, +}; + +static int dce_virtual_crtc_init(struct amdgpu_device *adev, int index) +{ + struct amdgpu_crtc *amdgpu_crtc; + int i; + + amdgpu_crtc = kzalloc(sizeof(struct amdgpu_crtc) + + (AMDGPUFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL); + if (amdgpu_crtc == NULL) + return -ENOMEM; + + drm_crtc_init(adev->ddev, &amdgpu_crtc->base, &dce_virtual_crtc_funcs); + + drm_mode_crtc_set_gamma_size(&amdgpu_crtc->base, 256); + amdgpu_crtc->crtc_id = index; + adev->mode_info.crtcs[index] = amdgpu_crtc; + + for (i = 0; i < 256; i++) { + amdgpu_crtc->lut_r[i] = i << 2; + amdgpu_crtc->lut_g[i] = i << 2; + amdgpu_crtc->lut_b[i] = i << 2; + } + + amdgpu_crtc->pll_id = ATOM_PPLL_INVALID; + amdgpu_crtc->encoder = NULL; + amdgpu_crtc->connector = NULL; + drm_crtc_helper_add(&amdgpu_crtc->base, &dce_virtual_crtc_helper_funcs); + + return 0; +} + +static int dce_virtual_early_init(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + adev->mode_info.vsync_timer_enabled = AMDGPU_IRQ_STATE_DISABLE; + dce_virtual_set_display_funcs(adev); + dce_virtual_set_irq_funcs(adev); + + adev->mode_info.num_crtc = 1; + adev->mode_info.num_hpd = 1; + adev->mode_info.num_dig = 1; + return 0; +} + +static bool dce_virtual_get_connector_info(struct amdgpu_device *adev) +{ + struct amdgpu_i2c_bus_rec ddc_bus; + struct amdgpu_router router; + struct amdgpu_hpd hpd; + + /* look up gpio for ddc, hpd */ + ddc_bus.valid = false; + hpd.hpd = AMDGPU_HPD_NONE; + /* needed for aux chan transactions */ + ddc_bus.hpd = hpd.hpd; + + memset(&router, 0, sizeof(router)); + router.ddc_valid = false; + router.cd_valid = false; + amdgpu_display_add_connector(adev, + 0, + ATOM_DEVICE_CRT1_SUPPORT, + DRM_MODE_CONNECTOR_VIRTUAL, &ddc_bus, + CONNECTOR_OBJECT_ID_VIRTUAL, + &hpd, + &router); + + amdgpu_display_add_encoder(adev, ENCODER_VIRTUAL_ENUM_VIRTUAL, + ATOM_DEVICE_CRT1_SUPPORT, + 0); + + amdgpu_link_encoder_connector(adev->ddev); + + return true; +} + +static int dce_virtual_sw_init(void *handle) +{ + int r, i; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + r = amdgpu_irq_add_id(adev, 229, &adev->crtc_irq); + if (r) + return r; + + adev->ddev->max_vblank_count = 0; + + adev->ddev->mode_config.funcs = &amdgpu_mode_funcs; + + adev->ddev->mode_config.max_width = 16384; + adev->ddev->mode_config.max_height = 16384; + + adev->ddev->mode_config.preferred_depth = 24; + adev->ddev->mode_config.prefer_shadow = 1; + + adev->ddev->mode_config.fb_base = adev->mc.aper_base; + + r = amdgpu_modeset_create_props(adev); + if (r) + return r; + + adev->ddev->mode_config.max_width = 16384; + adev->ddev->mode_config.max_height = 16384; + + /* allocate crtcs */ + for (i = 0; i < adev->mode_info.num_crtc; i++) { + r = dce_virtual_crtc_init(adev, i); + if (r) + return r; + } + + dce_virtual_get_connector_info(adev); + amdgpu_print_display_setup(adev->ddev); + + drm_kms_helper_poll_init(adev->ddev); + + adev->mode_info.mode_config_initialized = true; + return 0; +} + +static int dce_virtual_sw_fini(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + kfree(adev->mode_info.bios_hardcoded_edid); + + drm_kms_helper_poll_fini(adev->ddev); + + drm_mode_config_cleanup(adev->ddev); + adev->mode_info.mode_config_initialized = false; + return 0; +} + +static int dce_virtual_hw_init(void *handle) +{ + return 0; +} + +static int dce_virtual_hw_fini(void *handle) +{ + return 0; +} + +static int dce_virtual_suspend(void *handle) +{ + return dce_virtual_hw_fini(handle); +} + +static int dce_virtual_resume(void *handle) +{ + int ret; + + ret = dce_virtual_hw_init(handle); + + return ret; +} + +static bool dce_virtual_is_idle(void *handle) +{ + return true; +} + +static int dce_virtual_wait_for_idle(void *handle) +{ + return 0; +} + +static int dce_virtual_soft_reset(void *handle) +{ + return 0; +} + +static int dce_virtual_set_clockgating_state(void *handle, + enum amd_clockgating_state state) +{ + return 0; +} + +static int dce_virtual_set_powergating_state(void *handle, + enum amd_powergating_state state) +{ + return 0; +} + +const struct amd_ip_funcs dce_virtual_ip_funcs = { + .name = "dce_virtual", + .early_init = dce_virtual_early_init, + .late_init = NULL, + .sw_init = dce_virtual_sw_init, + .sw_fini = dce_virtual_sw_fini, + .hw_init = dce_virtual_hw_init, + .hw_fini = dce_virtual_hw_fini, + .suspend = dce_virtual_suspend, + .resume = dce_virtual_resume, + .is_idle = dce_virtual_is_idle, + .wait_for_idle = dce_virtual_wait_for_idle, + .soft_reset = dce_virtual_soft_reset, + .set_clockgating_state = dce_virtual_set_clockgating_state, + .set_powergating_state = dce_virtual_set_powergating_state, +}; + +/* these are handled by the primary encoders */ +static void dce_virtual_encoder_prepare(struct drm_encoder *encoder) +{ + return; +} + +static void dce_virtual_encoder_commit(struct drm_encoder *encoder) +{ + return; +} + +static void +dce_virtual_encoder_mode_set(struct drm_encoder *encoder, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + return; +} + +static void dce_virtual_encoder_disable(struct drm_encoder *encoder) +{ + return; +} + +static void +dce_virtual_encoder_dpms(struct drm_encoder *encoder, int mode) +{ + return; +} + +static bool dce_virtual_encoder_mode_fixup(struct drm_encoder *encoder, + const struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode) +{ + + /* set the active encoder to connector routing */ + amdgpu_encoder_set_active_device(encoder); + + return true; +} + +static const struct drm_encoder_helper_funcs dce_virtual_encoder_helper_funcs = { + .dpms = dce_virtual_encoder_dpms, + .mode_fixup = dce_virtual_encoder_mode_fixup, + .prepare = dce_virtual_encoder_prepare, + .mode_set = dce_virtual_encoder_mode_set, + .commit = dce_virtual_encoder_commit, + .disable = dce_virtual_encoder_disable, +}; + +static void dce_virtual_encoder_destroy(struct drm_encoder *encoder) +{ + struct amdgpu_encoder *amdgpu_encoder = to_amdgpu_encoder(encoder); + + kfree(amdgpu_encoder->enc_priv); + drm_encoder_cleanup(encoder); + kfree(amdgpu_encoder); +} + +static const struct drm_encoder_funcs dce_virtual_encoder_funcs = { + .destroy = dce_virtual_encoder_destroy, +}; + +static void dce_virtual_encoder_add(struct amdgpu_device *adev, + uint32_t encoder_enum, + uint32_t supported_device, + u16 caps) +{ + struct drm_device *dev = adev->ddev; + struct drm_encoder *encoder; + struct amdgpu_encoder *amdgpu_encoder; + + /* see if we already added it */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + amdgpu_encoder = to_amdgpu_encoder(encoder); + if (amdgpu_encoder->encoder_enum == encoder_enum) { + amdgpu_encoder->devices |= supported_device; + return; + } + + } + + /* add a new one */ + amdgpu_encoder = kzalloc(sizeof(struct amdgpu_encoder), GFP_KERNEL); + if (!amdgpu_encoder) + return; + + encoder = &amdgpu_encoder->base; + encoder->possible_crtcs = 0x1; + amdgpu_encoder->enc_priv = NULL; + amdgpu_encoder->encoder_enum = encoder_enum; + amdgpu_encoder->encoder_id = (encoder_enum & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT; + amdgpu_encoder->devices = supported_device; + amdgpu_encoder->rmx_type = RMX_OFF; + amdgpu_encoder->underscan_type = UNDERSCAN_OFF; + amdgpu_encoder->is_ext_encoder = false; + amdgpu_encoder->caps = caps; + + drm_encoder_init(dev, encoder, &dce_virtual_encoder_funcs, + DRM_MODE_ENCODER_VIRTUAL, NULL); + drm_encoder_helper_add(encoder, &dce_virtual_encoder_helper_funcs); + DRM_INFO("[FM]encoder: %d is VIRTUAL\n", amdgpu_encoder->encoder_id); +} + +static const struct amdgpu_display_funcs dce_virtual_display_funcs = { + .set_vga_render_state = &dce_virtual_set_vga_render_state, + .bandwidth_update = &dce_virtual_bandwidth_update, + .vblank_get_counter = &dce_virtual_vblank_get_counter, + .vblank_wait = &dce_virtual_vblank_wait, + .is_display_hung = &dce_virtual_is_display_hung, + .backlight_set_level = NULL, + .backlight_get_level = NULL, + .hpd_sense = &dce_virtual_hpd_sense, + .hpd_set_polarity = &dce_virtual_hpd_set_polarity, + .hpd_get_gpio_reg = &dce_virtual_hpd_get_gpio_reg, + .page_flip = &dce_virtual_page_flip, + .page_flip_get_scanoutpos = &dce_virtual_crtc_get_scanoutpos, + .add_encoder = &dce_virtual_encoder_add, + .add_connector = &amdgpu_connector_add, + .stop_mc_access = &dce_virtual_stop_mc_access, + .resume_mc_access = &dce_virtual_resume_mc_access, +}; + +static void dce_virtual_set_display_funcs(struct amdgpu_device *adev) +{ + if (adev->mode_info.funcs == NULL) + adev->mode_info.funcs = &dce_virtual_display_funcs; +} + +static enum hrtimer_restart dce_virtual_vblank_timer_handle(struct hrtimer *vblank_timer) +{ + struct amdgpu_mode_info *mode_info = container_of(vblank_timer, struct amdgpu_mode_info ,vblank_timer); + struct amdgpu_device *adev = container_of(mode_info, struct amdgpu_device ,mode_info); + unsigned crtc = 0; + drm_handle_vblank(adev->ddev, crtc); + dce_virtual_pageflip_irq(adev, NULL, NULL); + hrtimer_start(vblank_timer, ktime_set(0, DCE_VIRTUAL_VBLANK_PERIOD), HRTIMER_MODE_REL); + return HRTIMER_NORESTART; +} + +static void dce_virtual_set_crtc_vblank_interrupt_state(struct amdgpu_device *adev, + int crtc, + enum amdgpu_interrupt_state state) +{ + if (crtc >= adev->mode_info.num_crtc) { + DRM_DEBUG("invalid crtc %d\n", crtc); + return; + } + + if (state && !adev->mode_info.vsync_timer_enabled) { + DRM_DEBUG("Enable software vsync timer\n"); + hrtimer_init(&adev->mode_info.vblank_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); + hrtimer_set_expires(&adev->mode_info.vblank_timer, ktime_set(0, DCE_VIRTUAL_VBLANK_PERIOD)); + adev->mode_info.vblank_timer.function = dce_virtual_vblank_timer_handle; + hrtimer_start(&adev->mode_info.vblank_timer, ktime_set(0, DCE_VIRTUAL_VBLANK_PERIOD), HRTIMER_MODE_REL); + } else if (!state && adev->mode_info.vsync_timer_enabled) { + DRM_DEBUG("Disable software vsync timer\n"); + hrtimer_cancel(&adev->mode_info.vblank_timer); + } + + adev->mode_info.vsync_timer_enabled = state; + DRM_DEBUG("[FM]set crtc %d vblank interrupt state %d\n", crtc, state); +} + + +static int dce_virtual_set_crtc_irq_state(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + unsigned type, + enum amdgpu_interrupt_state state) +{ + switch (type) { + case AMDGPU_CRTC_IRQ_VBLANK1: + dce_virtual_set_crtc_vblank_interrupt_state(adev, 0, state); + break; + default: + break; + } + return 0; +} + +static void dce_virtual_crtc_vblank_int_ack(struct amdgpu_device *adev, + int crtc) +{ + if (crtc >= adev->mode_info.num_crtc) { + DRM_DEBUG("invalid crtc %d\n", crtc); + return; + } +} + +static int dce_virtual_crtc_irq(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + unsigned crtc = 0; + unsigned irq_type = AMDGPU_CRTC_IRQ_VBLANK1; + + dce_virtual_crtc_vblank_int_ack(adev, crtc); + + if (amdgpu_irq_enabled(adev, source, irq_type)) { + drm_handle_vblank(adev->ddev, crtc); + } + dce_virtual_pageflip_irq(adev, NULL, NULL); + DRM_DEBUG("IH: D%d vblank\n", crtc + 1); + return 0; +} + +static int dce_virtual_set_pageflip_irq_state(struct amdgpu_device *adev, + struct amdgpu_irq_src *src, + unsigned type, + enum amdgpu_interrupt_state state) +{ + if (type >= adev->mode_info.num_crtc) { + DRM_ERROR("invalid pageflip crtc %d\n", type); + return -EINVAL; + } + DRM_DEBUG("[FM]set pageflip irq type %d state %d\n", type, state); + + return 0; +} + +static int dce_virtual_pageflip_irq(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + unsigned long flags; + unsigned crtc_id = 0; + struct amdgpu_crtc *amdgpu_crtc; + struct amdgpu_flip_work *works; + + crtc_id = 0; + amdgpu_crtc = adev->mode_info.crtcs[crtc_id]; + + if (crtc_id >= adev->mode_info.num_crtc) { + DRM_ERROR("invalid pageflip crtc %d\n", crtc_id); + return -EINVAL; + } + + /* IRQ could occur when in initial stage */ + if (amdgpu_crtc == NULL) + return 0; + + spin_lock_irqsave(&adev->ddev->event_lock, flags); + works = amdgpu_crtc->pflip_works; + if (amdgpu_crtc->pflip_status != AMDGPU_FLIP_SUBMITTED) { + DRM_DEBUG_DRIVER("amdgpu_crtc->pflip_status = %d != " + "AMDGPU_FLIP_SUBMITTED(%d)\n", + amdgpu_crtc->pflip_status, + AMDGPU_FLIP_SUBMITTED); + spin_unlock_irqrestore(&adev->ddev->event_lock, flags); + return 0; + } + + /* page flip completed. clean up */ + amdgpu_crtc->pflip_status = AMDGPU_FLIP_NONE; + amdgpu_crtc->pflip_works = NULL; + + /* wakeup usersapce */ + if (works->event) + drm_crtc_send_vblank_event(&amdgpu_crtc->base, works->event); + + spin_unlock_irqrestore(&adev->ddev->event_lock, flags); + + drm_crtc_vblank_put(&amdgpu_crtc->base); + schedule_work(&works->unpin_work); + + return 0; +} + +static const struct amdgpu_irq_src_funcs dce_virtual_crtc_irq_funcs = { + .set = dce_virtual_set_crtc_irq_state, + .process = dce_virtual_crtc_irq, +}; + +static const struct amdgpu_irq_src_funcs dce_virtual_pageflip_irq_funcs = { + .set = dce_virtual_set_pageflip_irq_state, + .process = dce_virtual_pageflip_irq, +}; + +static void dce_virtual_set_irq_funcs(struct amdgpu_device *adev) +{ + adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_LAST; + adev->crtc_irq.funcs = &dce_virtual_crtc_irq_funcs; + + adev->pageflip_irq.num_types = AMDGPU_PAGEFLIP_IRQ_LAST; + adev->pageflip_irq.funcs = &dce_virtual_pageflip_irq_funcs; +} + diff --git a/drivers/gpu/drm/amd/amdgpu/dce_virtual.h b/drivers/gpu/drm/amd/amdgpu/dce_virtual.h new file mode 100644 index 000000000000..e239243f6ebc --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/dce_virtual.h @@ -0,0 +1,31 @@ +/* + * Copyright 2014 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __DCE_VIRTUAL_H__ +#define __DCE_VIRTUAL_H__ + +extern const struct amd_ip_funcs dce_virtual_ip_funcs; +#define DCE_VIRTUAL_VBLANK_PERIOD 16666666 + +#endif + diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c index d869d058ef24..f4fbec3e224e 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c @@ -4465,24 +4465,21 @@ static int gfx_v7_0_sw_init(void *handle) } /* reserve GDS, GWS and OA resource for gfx */ - r = amdgpu_bo_create(adev, adev->gds.mem.gfx_partition_size, - PAGE_SIZE, true, - AMDGPU_GEM_DOMAIN_GDS, 0, - NULL, NULL, &adev->gds.gds_gfx_bo); + r = amdgpu_bo_create_kernel(adev, adev->gds.mem.gfx_partition_size, + PAGE_SIZE, AMDGPU_GEM_DOMAIN_GDS, + &adev->gds.gds_gfx_bo, NULL, NULL); if (r) return r; - r = amdgpu_bo_create(adev, adev->gds.gws.gfx_partition_size, - PAGE_SIZE, true, - AMDGPU_GEM_DOMAIN_GWS, 0, - NULL, NULL, &adev->gds.gws_gfx_bo); + r = amdgpu_bo_create_kernel(adev, adev->gds.gws.gfx_partition_size, + PAGE_SIZE, AMDGPU_GEM_DOMAIN_GWS, + &adev->gds.gws_gfx_bo, NULL, NULL); if (r) return r; - r = amdgpu_bo_create(adev, adev->gds.oa.gfx_partition_size, - PAGE_SIZE, true, - AMDGPU_GEM_DOMAIN_OA, 0, - NULL, NULL, &adev->gds.oa_gfx_bo); + r = amdgpu_bo_create_kernel(adev, adev->gds.oa.gfx_partition_size, + PAGE_SIZE, AMDGPU_GEM_DOMAIN_OA, + &adev->gds.oa_gfx_bo, NULL, NULL); if (r) 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 b8184617ca25..c6a63c2f91e3 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -703,7 +703,10 @@ static void gfx_v8_0_init_golden_registers(struct amdgpu_device *adev) polaris10_golden_common_all, (const u32)ARRAY_SIZE(polaris10_golden_common_all)); WREG32_SMC(ixCG_ACLK_CNTL, 0x0000001C); - if (adev->pdev->revision == 0xc7) { + if (adev->pdev->revision == 0xc7 && + ((adev->pdev->subsystem_device == 0xb37 && adev->pdev->subsystem_vendor == 0x1002) || + (adev->pdev->subsystem_device == 0x4a8 && adev->pdev->subsystem_vendor == 0x1043) || + (adev->pdev->subsystem_device == 0x9480 && adev->pdev->subsystem_vendor == 0x1682))) { amdgpu_atombios_i2c_channel_trans(adev, 0x10, 0x96, 0x1E, 0xDD); amdgpu_atombios_i2c_channel_trans(adev, 0x10, 0x96, 0x1F, 0xD0); } @@ -1233,10 +1236,9 @@ static void gfx_v8_0_rlc_fini(struct amdgpu_device *adev) if (adev->gfx.rlc.clear_state_obj) { r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, false); if (unlikely(r != 0)) - dev_warn(adev->dev, "(%d) reserve RLC c bo failed\n", r); + 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; } @@ -1248,7 +1250,6 @@ static void gfx_v8_0_rlc_fini(struct amdgpu_device *adev) 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; } @@ -1290,14 +1291,14 @@ static int gfx_v8_0_rlc_init(struct amdgpu_device *adev) &adev->gfx.rlc.clear_state_gpu_addr); 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) pin RLC cbs 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 c bo failed\n", r); + dev_warn(adev->dev, "(%d) map RLC cbs bo failed\n", r); gfx_v8_0_rlc_fini(adev); return r; } @@ -1332,7 +1333,7 @@ static int gfx_v8_0_rlc_init(struct amdgpu_device *adev) &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); + 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); @@ -1345,7 +1346,6 @@ static int gfx_v8_0_rlc_init(struct amdgpu_device *adev) amdgpu_bo_kunmap(adev->gfx.rlc.cp_table_obj); amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj); - } return 0; @@ -1361,7 +1361,6 @@ static void gfx_v8_0_mec_fini(struct amdgpu_device *adev) 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; } @@ -2082,24 +2081,21 @@ static int gfx_v8_0_sw_init(void *handle) } /* reserve GDS, GWS and OA resource for gfx */ - r = amdgpu_bo_create(adev, adev->gds.mem.gfx_partition_size, - PAGE_SIZE, true, - AMDGPU_GEM_DOMAIN_GDS, 0, NULL, - NULL, &adev->gds.gds_gfx_bo); + r = amdgpu_bo_create_kernel(adev, adev->gds.mem.gfx_partition_size, + PAGE_SIZE, AMDGPU_GEM_DOMAIN_GDS, + &adev->gds.gds_gfx_bo, NULL, NULL); if (r) return r; - r = amdgpu_bo_create(adev, adev->gds.gws.gfx_partition_size, - PAGE_SIZE, true, - AMDGPU_GEM_DOMAIN_GWS, 0, NULL, - NULL, &adev->gds.gws_gfx_bo); + r = amdgpu_bo_create_kernel(adev, adev->gds.gws.gfx_partition_size, + PAGE_SIZE, AMDGPU_GEM_DOMAIN_GWS, + &adev->gds.gws_gfx_bo, NULL, NULL); if (r) return r; - r = amdgpu_bo_create(adev, adev->gds.oa.gfx_partition_size, - PAGE_SIZE, true, - AMDGPU_GEM_DOMAIN_OA, 0, NULL, - NULL, &adev->gds.oa_gfx_bo); + r = amdgpu_bo_create_kernel(adev, adev->gds.oa.gfx_partition_size, + PAGE_SIZE, AMDGPU_GEM_DOMAIN_OA, + &adev->gds.oa_gfx_bo, NULL, NULL); if (r) return r; @@ -2127,9 +2123,7 @@ static int gfx_v8_0_sw_fini(void *handle) amdgpu_ring_fini(&adev->gfx.compute_ring[i]); gfx_v8_0_mec_fini(adev); - gfx_v8_0_rlc_fini(adev); - gfx_v8_0_free_microcode(adev); return 0; @@ -3465,19 +3459,16 @@ static void gfx_v8_0_select_se_sh(struct amdgpu_device *adev, else data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_INDEX, instance); - if ((se_num == 0xffffffff) && (sh_num == 0xffffffff)) { - data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_BROADCAST_WRITES, 1); + if (se_num == 0xffffffff) data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES, 1); - } else if (se_num == 0xffffffff) { - data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num); - data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES, 1); - } else if (sh_num == 0xffffffff) { - data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_BROADCAST_WRITES, 1); + else data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num); - } else { + + if (sh_num == 0xffffffff) + data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_BROADCAST_WRITES, 1); + else data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num); - data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num); - } + WREG32(mmGRBM_GFX_INDEX, data); } @@ -3490,11 +3481,10 @@ static u32 gfx_v8_0_get_rb_active_bitmap(struct amdgpu_device *adev) { u32 data, mask; - data = RREG32(mmCC_RB_BACKEND_DISABLE); - data |= RREG32(mmGC_USER_RB_BACKEND_DISABLE); + data = RREG32(mmCC_RB_BACKEND_DISABLE) | + RREG32(mmGC_USER_RB_BACKEND_DISABLE); - data &= CC_RB_BACKEND_DISABLE__BACKEND_DISABLE_MASK; - data >>= GC_USER_RB_BACKEND_DISABLE__BACKEND_DISABLE__SHIFT; + data = REG_GET_FIELD(data, GC_USER_RB_BACKEND_DISABLE, BACKEND_DISABLE); mask = gfx_v8_0_create_bitmask(adev->gfx.config.max_backends_per_se / adev->gfx.config.max_sh_per_se); @@ -3576,16 +3566,12 @@ static void gfx_v8_0_gpu_init(struct amdgpu_device *adev) u32 tmp; int i; - tmp = RREG32(mmGRBM_CNTL); - tmp = REG_SET_FIELD(tmp, GRBM_CNTL, READ_TIMEOUT, 0xff); - WREG32(mmGRBM_CNTL, tmp); - + WREG32_FIELD(GRBM_CNTL, READ_TIMEOUT, 0xFF); WREG32(mmGB_ADDR_CONFIG, adev->gfx.config.gb_addr_config); WREG32(mmHDP_ADDR_CONFIG, adev->gfx.config.gb_addr_config); WREG32(mmDMIF_ADDR_CALC, adev->gfx.config.gb_addr_config); gfx_v8_0_tiling_mode_table_init(adev); - gfx_v8_0_setup_rb(adev); gfx_v8_0_get_cu_info(adev); @@ -3769,9 +3755,7 @@ static int gfx_v8_0_init_save_restore_list(struct amdgpu_device *adev) sizeof(indirect_start_offsets)/sizeof(int)); /* save and restore list */ - temp = RREG32(mmRLC_SRM_CNTL); - temp |= RLC_SRM_CNTL__AUTO_INCR_ADDR_MASK; - WREG32(mmRLC_SRM_CNTL, temp); + WREG32_FIELD(RLC_SRM_CNTL, AUTO_INCR_ADDR, 1); WREG32(mmRLC_SRM_ARAM_ADDR, 0); for (i = 0; i < adev->gfx.rlc.reg_list_size_bytes >> 2; i++) @@ -3808,11 +3792,7 @@ static int gfx_v8_0_init_save_restore_list(struct amdgpu_device *adev) static void gfx_v8_0_enable_save_restore_machine(struct amdgpu_device *adev) { - uint32_t data; - - data = RREG32(mmRLC_SRM_CNTL); - data |= RLC_SRM_CNTL__SRM_ENABLE_MASK; - WREG32(mmRLC_SRM_CNTL, data); + WREG32_FIELD(RLC_SRM_CNTL, SRM_ENABLE, 1); } static void gfx_v8_0_init_power_gating(struct amdgpu_device *adev) @@ -3822,75 +3802,34 @@ static void gfx_v8_0_init_power_gating(struct amdgpu_device *adev) if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_PG | AMD_PG_SUPPORT_GFX_SMG | AMD_PG_SUPPORT_GFX_DMG)) { - data = RREG32(mmCP_RB_WPTR_POLL_CNTL); - data &= ~CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT_MASK; - data |= (0x60 << CP_RB_WPTR_POLL_CNTL__IDLE_POLL_COUNT__SHIFT); - WREG32(mmCP_RB_WPTR_POLL_CNTL, data); - - data = 0; - data |= (0x10 << RLC_PG_DELAY__POWER_UP_DELAY__SHIFT); - data |= (0x10 << RLC_PG_DELAY__POWER_DOWN_DELAY__SHIFT); - data |= (0x10 << RLC_PG_DELAY__CMD_PROPAGATE_DELAY__SHIFT); - data |= (0x10 << RLC_PG_DELAY__MEM_SLEEP_DELAY__SHIFT); - WREG32(mmRLC_PG_DELAY, data); + WREG32_FIELD(CP_RB_WPTR_POLL_CNTL, IDLE_POLL_COUNT, 0x60); - data = RREG32(mmRLC_PG_DELAY_2); - data &= ~RLC_PG_DELAY_2__SERDES_CMD_DELAY_MASK; - data |= (0x3 << RLC_PG_DELAY_2__SERDES_CMD_DELAY__SHIFT); - WREG32(mmRLC_PG_DELAY_2, data); + data = REG_SET_FIELD(0, RLC_PG_DELAY, POWER_UP_DELAY, 0x10); + data = REG_SET_FIELD(data, RLC_PG_DELAY, POWER_DOWN_DELAY, 0x10); + data = REG_SET_FIELD(data, RLC_PG_DELAY, CMD_PROPAGATE_DELAY, 0x10); + data = REG_SET_FIELD(data, RLC_PG_DELAY, MEM_SLEEP_DELAY, 0x10); + WREG32(mmRLC_PG_DELAY, data); - data = RREG32(mmRLC_AUTO_PG_CTRL); - data &= ~RLC_AUTO_PG_CTRL__GRBM_REG_SAVE_GFX_IDLE_THRESHOLD_MASK; - data |= (0x55f0 << RLC_AUTO_PG_CTRL__GRBM_REG_SAVE_GFX_IDLE_THRESHOLD__SHIFT); - WREG32(mmRLC_AUTO_PG_CTRL, data); + WREG32_FIELD(RLC_PG_DELAY_2, SERDES_CMD_DELAY, 0x3); + WREG32_FIELD(RLC_AUTO_PG_CTRL, GRBM_REG_SAVE_GFX_IDLE_THRESHOLD, 0x55f0); } } static void cz_enable_sck_slow_down_on_power_up(struct amdgpu_device *adev, bool enable) { - u32 data, orig; - - orig = data = RREG32(mmRLC_PG_CNTL); - - if (enable) - data |= RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PU_ENABLE_MASK; - else - data &= ~RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PU_ENABLE_MASK; - - if (orig != data) - WREG32(mmRLC_PG_CNTL, data); + WREG32_FIELD(RLC_PG_CNTL, SMU_CLK_SLOWDOWN_ON_PU_ENABLE, enable ? 1 : 0); } static void cz_enable_sck_slow_down_on_power_down(struct amdgpu_device *adev, bool enable) { - u32 data, orig; - - orig = data = RREG32(mmRLC_PG_CNTL); - - if (enable) - data |= RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PD_ENABLE_MASK; - else - data &= ~RLC_PG_CNTL__SMU_CLK_SLOWDOWN_ON_PD_ENABLE_MASK; - - if (orig != data) - WREG32(mmRLC_PG_CNTL, data); + WREG32_FIELD(RLC_PG_CNTL, SMU_CLK_SLOWDOWN_ON_PD_ENABLE, enable ? 1 : 0); } static void cz_enable_cp_power_gating(struct amdgpu_device *adev, bool enable) { - u32 data, orig; - - orig = data = RREG32(mmRLC_PG_CNTL); - - if (enable) - data &= ~RLC_PG_CNTL__CP_PG_DISABLE_MASK; - else - data |= RLC_PG_CNTL__CP_PG_DISABLE_MASK; - - if (orig != data) - WREG32(mmRLC_PG_CNTL, data); + WREG32_FIELD(RLC_PG_CNTL, CP_PG_DISABLE, enable ? 1 : 0); } static void gfx_v8_0_init_pg(struct amdgpu_device *adev) @@ -3929,34 +3868,24 @@ static void gfx_v8_0_init_pg(struct amdgpu_device *adev) void gfx_v8_0_rlc_stop(struct amdgpu_device *adev) { - u32 tmp = RREG32(mmRLC_CNTL); - - tmp = REG_SET_FIELD(tmp, RLC_CNTL, RLC_ENABLE_F32, 0); - WREG32(mmRLC_CNTL, tmp); + WREG32_FIELD(RLC_CNTL, RLC_ENABLE_F32, 0); gfx_v8_0_enable_gui_idle_interrupt(adev, false); - gfx_v8_0_wait_for_rlc_serdes(adev); } static void gfx_v8_0_rlc_reset(struct amdgpu_device *adev) { - u32 tmp = RREG32(mmGRBM_SOFT_RESET); - - tmp = REG_SET_FIELD(tmp, GRBM_SOFT_RESET, SOFT_RESET_RLC, 1); - WREG32(mmGRBM_SOFT_RESET, tmp); + WREG32_FIELD(GRBM_SOFT_RESET, SOFT_RESET_RLC, 1); udelay(50); - tmp = REG_SET_FIELD(tmp, GRBM_SOFT_RESET, SOFT_RESET_RLC, 0); - WREG32(mmGRBM_SOFT_RESET, tmp); + + WREG32_FIELD(GRBM_SOFT_RESET, SOFT_RESET_RLC, 0); udelay(50); } static void gfx_v8_0_rlc_start(struct amdgpu_device *adev) { - u32 tmp = RREG32(mmRLC_CNTL); - - tmp = REG_SET_FIELD(tmp, RLC_CNTL, RLC_ENABLE_F32, 1); - WREG32(mmRLC_CNTL, tmp); + WREG32_FIELD(RLC_CNTL, RLC_ENABLE_F32, 1); /* carrizo do enable cp interrupt after cp inited */ if (!(adev->flags & AMD_IS_APU)) @@ -3998,14 +3927,13 @@ static int gfx_v8_0_rlc_resume(struct amdgpu_device *adev) /* disable CG */ WREG32(mmRLC_CGCG_CGLS_CTRL, 0); if (adev->asic_type == CHIP_POLARIS11 || - adev->asic_type == CHIP_POLARIS10) + adev->asic_type == CHIP_POLARIS10) WREG32(mmRLC_CGCG_CGLS_CTRL_3D, 0); /* disable PG */ WREG32(mmRLC_PG_CNTL, 0); gfx_v8_0_rlc_reset(adev); - gfx_v8_0_init_pg(adev); if (!adev->pp_enabled) { @@ -4300,12 +4228,10 @@ static int gfx_v8_0_cp_gfx_resume(struct amdgpu_device *adev) gfx_v8_0_cp_gfx_start(adev); ring->ready = true; r = amdgpu_ring_test_ring(ring); - if (r) { + if (r) ring->ready = false; - return r; - } - return 0; + return r; } static void gfx_v8_0_cp_compute_enable(struct amdgpu_device *adev, bool enable) @@ -4980,7 +4906,6 @@ static int gfx_v8_0_hw_init(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; gfx_v8_0_init_golden_registers(adev); - gfx_v8_0_gpu_init(adev); r = gfx_v8_0_rlc_resume(adev); @@ -4988,8 +4913,6 @@ static int gfx_v8_0_hw_init(void *handle) return r; r = gfx_v8_0_cp_resume(adev); - if (r) - return r; return r; } @@ -5037,25 +4960,22 @@ static bool gfx_v8_0_is_idle(void *handle) static int gfx_v8_0_wait_for_idle(void *handle) { unsigned i; - u32 tmp; struct amdgpu_device *adev = (struct amdgpu_device *)handle; for (i = 0; i < adev->usec_timeout; i++) { - /* read MC_STATUS */ - tmp = RREG32(mmGRBM_STATUS) & GRBM_STATUS__GUI_ACTIVE_MASK; - - if (!REG_GET_FIELD(tmp, GRBM_STATUS, GUI_ACTIVE)) + if (gfx_v8_0_is_idle(handle)) return 0; + udelay(1); } return -ETIMEDOUT; } -static int gfx_v8_0_soft_reset(void *handle) +static int gfx_v8_0_check_soft_reset(void *handle) { + struct amdgpu_device *adev = (struct amdgpu_device *)handle; u32 grbm_soft_reset = 0, srbm_soft_reset = 0; u32 tmp; - struct amdgpu_device *adev = (struct amdgpu_device *)handle; /* GRBM_STATUS */ tmp = RREG32(mmGRBM_STATUS); @@ -5064,16 +4984,12 @@ static int gfx_v8_0_soft_reset(void *handle) GRBM_STATUS__TA_BUSY_MASK | GRBM_STATUS__VGT_BUSY_MASK | GRBM_STATUS__DB_BUSY_MASK | GRBM_STATUS__CB_BUSY_MASK | GRBM_STATUS__GDS_BUSY_MASK | GRBM_STATUS__SPI_BUSY_MASK | - GRBM_STATUS__IA_BUSY_MASK | GRBM_STATUS__IA_BUSY_NO_DMA_MASK)) { + GRBM_STATUS__IA_BUSY_MASK | GRBM_STATUS__IA_BUSY_NO_DMA_MASK | + GRBM_STATUS__CP_BUSY_MASK | GRBM_STATUS__CP_COHERENCY_BUSY_MASK)) { grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP, 1); grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_GFX, 1); - } - - if (tmp & (GRBM_STATUS__CP_BUSY_MASK | GRBM_STATUS__CP_COHERENCY_BUSY_MASK)) { - grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, - GRBM_SOFT_RESET, SOFT_RESET_CP, 1); srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_GRBM, 1); } @@ -5084,73 +5000,199 @@ static int gfx_v8_0_soft_reset(void *handle) grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_RLC, 1); + if (REG_GET_FIELD(tmp, GRBM_STATUS2, CPF_BUSY) || + REG_GET_FIELD(tmp, GRBM_STATUS2, CPC_BUSY) || + REG_GET_FIELD(tmp, GRBM_STATUS2, CPG_BUSY)) { + grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, + SOFT_RESET_CPF, 1); + grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, + SOFT_RESET_CPC, 1); + grbm_soft_reset = REG_SET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, + SOFT_RESET_CPG, 1); + srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, + SOFT_RESET_GRBM, 1); + } + /* SRBM_STATUS */ tmp = RREG32(mmSRBM_STATUS); if (REG_GET_FIELD(tmp, SRBM_STATUS, GRBM_RQ_PENDING)) srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_GRBM, 1); + if (REG_GET_FIELD(tmp, SRBM_STATUS, SEM_BUSY)) + srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, + SRBM_SOFT_RESET, SOFT_RESET_SEM, 1); if (grbm_soft_reset || srbm_soft_reset) { - /* stop the rlc */ - gfx_v8_0_rlc_stop(adev); + adev->ip_block_status[AMD_IP_BLOCK_TYPE_GFX].hang = true; + adev->gfx.grbm_soft_reset = grbm_soft_reset; + adev->gfx.srbm_soft_reset = srbm_soft_reset; + } else { + adev->ip_block_status[AMD_IP_BLOCK_TYPE_GFX].hang = false; + adev->gfx.grbm_soft_reset = 0; + adev->gfx.srbm_soft_reset = 0; + } + + return 0; +} +static void gfx_v8_0_inactive_hqd(struct amdgpu_device *adev, + struct amdgpu_ring *ring) +{ + int i; + + vi_srbm_select(adev, ring->me, ring->pipe, ring->queue, 0); + if (RREG32(mmCP_HQD_ACTIVE) & CP_HQD_ACTIVE__ACTIVE_MASK) { + u32 tmp; + tmp = RREG32(mmCP_HQD_DEQUEUE_REQUEST); + tmp = REG_SET_FIELD(tmp, CP_HQD_DEQUEUE_REQUEST, + DEQUEUE_REQ, 2); + WREG32(mmCP_HQD_DEQUEUE_REQUEST, tmp); + for (i = 0; i < adev->usec_timeout; i++) { + if (!(RREG32(mmCP_HQD_ACTIVE) & CP_HQD_ACTIVE__ACTIVE_MASK)) + break; + udelay(1); + } + } +} + +static int gfx_v8_0_pre_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 grbm_soft_reset = 0, srbm_soft_reset = 0; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_GFX].hang) + return 0; + + grbm_soft_reset = adev->gfx.grbm_soft_reset; + srbm_soft_reset = adev->gfx.srbm_soft_reset; + + /* stop the rlc */ + gfx_v8_0_rlc_stop(adev); + + if (REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP) || + REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_GFX)) /* Disable GFX parsing/prefetching */ gfx_v8_0_cp_gfx_enable(adev, false); + if (REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP) || + REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CPF) || + REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CPC) || + REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CPG)) { + int i; + + for (i = 0; i < adev->gfx.num_compute_rings; i++) { + struct amdgpu_ring *ring = &adev->gfx.compute_ring[i]; + + gfx_v8_0_inactive_hqd(adev, ring); + } /* Disable MEC parsing/prefetching */ gfx_v8_0_cp_compute_enable(adev, false); + } - if (grbm_soft_reset || srbm_soft_reset) { - tmp = RREG32(mmGMCON_DEBUG); - tmp = REG_SET_FIELD(tmp, - GMCON_DEBUG, GFX_STALL, 1); - tmp = REG_SET_FIELD(tmp, - GMCON_DEBUG, GFX_CLEAR, 1); - WREG32(mmGMCON_DEBUG, tmp); + return 0; +} - udelay(50); - } +static int gfx_v8_0_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 grbm_soft_reset = 0, srbm_soft_reset = 0; + u32 tmp; - if (grbm_soft_reset) { - tmp = RREG32(mmGRBM_SOFT_RESET); - tmp |= grbm_soft_reset; - dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp); - WREG32(mmGRBM_SOFT_RESET, tmp); - tmp = RREG32(mmGRBM_SOFT_RESET); + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_GFX].hang) + return 0; - udelay(50); + grbm_soft_reset = adev->gfx.grbm_soft_reset; + srbm_soft_reset = adev->gfx.srbm_soft_reset; - tmp &= ~grbm_soft_reset; - WREG32(mmGRBM_SOFT_RESET, tmp); - tmp = RREG32(mmGRBM_SOFT_RESET); - } + if (grbm_soft_reset || srbm_soft_reset) { + tmp = RREG32(mmGMCON_DEBUG); + tmp = REG_SET_FIELD(tmp, GMCON_DEBUG, GFX_STALL, 1); + tmp = REG_SET_FIELD(tmp, GMCON_DEBUG, GFX_CLEAR, 1); + WREG32(mmGMCON_DEBUG, tmp); + udelay(50); + } - if (srbm_soft_reset) { - tmp = RREG32(mmSRBM_SOFT_RESET); - tmp |= srbm_soft_reset; - dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp); - WREG32(mmSRBM_SOFT_RESET, tmp); - tmp = RREG32(mmSRBM_SOFT_RESET); + if (grbm_soft_reset) { + tmp = RREG32(mmGRBM_SOFT_RESET); + tmp |= grbm_soft_reset; + dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp); + WREG32(mmGRBM_SOFT_RESET, tmp); + tmp = RREG32(mmGRBM_SOFT_RESET); - udelay(50); + udelay(50); - tmp &= ~srbm_soft_reset; - WREG32(mmSRBM_SOFT_RESET, tmp); - tmp = RREG32(mmSRBM_SOFT_RESET); - } + tmp &= ~grbm_soft_reset; + WREG32(mmGRBM_SOFT_RESET, tmp); + tmp = RREG32(mmGRBM_SOFT_RESET); + } - if (grbm_soft_reset || srbm_soft_reset) { - tmp = RREG32(mmGMCON_DEBUG); - tmp = REG_SET_FIELD(tmp, - GMCON_DEBUG, GFX_STALL, 0); - tmp = REG_SET_FIELD(tmp, - GMCON_DEBUG, GFX_CLEAR, 0); - WREG32(mmGMCON_DEBUG, tmp); - } + if (srbm_soft_reset) { + tmp = RREG32(mmSRBM_SOFT_RESET); + tmp |= srbm_soft_reset; + dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp); + WREG32(mmSRBM_SOFT_RESET, tmp); + tmp = RREG32(mmSRBM_SOFT_RESET); - /* Wait a little for things to settle down */ udelay(50); + + tmp &= ~srbm_soft_reset; + WREG32(mmSRBM_SOFT_RESET, tmp); + tmp = RREG32(mmSRBM_SOFT_RESET); } + + if (grbm_soft_reset || srbm_soft_reset) { + tmp = RREG32(mmGMCON_DEBUG); + tmp = REG_SET_FIELD(tmp, GMCON_DEBUG, GFX_STALL, 0); + tmp = REG_SET_FIELD(tmp, GMCON_DEBUG, GFX_CLEAR, 0); + WREG32(mmGMCON_DEBUG, tmp); + } + + /* Wait a little for things to settle down */ + udelay(50); + + return 0; +} + +static void gfx_v8_0_init_hqd(struct amdgpu_device *adev, + struct amdgpu_ring *ring) +{ + vi_srbm_select(adev, ring->me, ring->pipe, ring->queue, 0); + WREG32(mmCP_HQD_DEQUEUE_REQUEST, 0); + WREG32(mmCP_HQD_PQ_RPTR, 0); + WREG32(mmCP_HQD_PQ_WPTR, 0); + vi_srbm_select(adev, 0, 0, 0, 0); +} + +static int gfx_v8_0_post_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 grbm_soft_reset = 0, srbm_soft_reset = 0; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_GFX].hang) + return 0; + + grbm_soft_reset = adev->gfx.grbm_soft_reset; + srbm_soft_reset = adev->gfx.srbm_soft_reset; + + if (REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP) || + REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_GFX)) + gfx_v8_0_cp_gfx_resume(adev); + + if (REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP) || + REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CPF) || + REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CPC) || + REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CPG)) { + int i; + + for (i = 0; i < adev->gfx.num_compute_rings; i++) { + struct amdgpu_ring *ring = &adev->gfx.compute_ring[i]; + + gfx_v8_0_init_hqd(adev, ring); + } + gfx_v8_0_cp_compute_resume(adev); + } + gfx_v8_0_rlc_start(adev); + return 0; } @@ -5269,8 +5311,6 @@ static int gfx_v8_0_late_init(void *handle) static void gfx_v8_0_enable_gfx_static_mg_power_gating(struct amdgpu_device *adev, bool enable) { - uint32_t data, temp; - if (adev->asic_type == CHIP_POLARIS11) /* Send msg to SMU via Powerplay */ amdgpu_set_powergating_state(adev, @@ -5278,83 +5318,35 @@ static void gfx_v8_0_enable_gfx_static_mg_power_gating(struct amdgpu_device *ade enable ? AMD_PG_STATE_GATE : AMD_PG_STATE_UNGATE); - temp = data = RREG32(mmRLC_PG_CNTL); - /* Enable static MGPG */ - if (enable) - data |= RLC_PG_CNTL__STATIC_PER_CU_PG_ENABLE_MASK; - else - data &= ~RLC_PG_CNTL__STATIC_PER_CU_PG_ENABLE_MASK; - - if (temp != data) - WREG32(mmRLC_PG_CNTL, data); + WREG32_FIELD(RLC_PG_CNTL, STATIC_PER_CU_PG_ENABLE, enable ? 1 : 0); } static void gfx_v8_0_enable_gfx_dynamic_mg_power_gating(struct amdgpu_device *adev, bool enable) { - uint32_t data, temp; - - temp = data = RREG32(mmRLC_PG_CNTL); - /* Enable dynamic MGPG */ - if (enable) - data |= RLC_PG_CNTL__DYN_PER_CU_PG_ENABLE_MASK; - else - data &= ~RLC_PG_CNTL__DYN_PER_CU_PG_ENABLE_MASK; - - if (temp != data) - WREG32(mmRLC_PG_CNTL, data); + WREG32_FIELD(RLC_PG_CNTL, DYN_PER_CU_PG_ENABLE, enable ? 1 : 0); } static void polaris11_enable_gfx_quick_mg_power_gating(struct amdgpu_device *adev, bool enable) { - uint32_t data, temp; - - temp = data = RREG32(mmRLC_PG_CNTL); - /* Enable quick PG */ - if (enable) - data |= RLC_PG_CNTL__QUICK_PG_ENABLE_MASK; - else - data &= ~RLC_PG_CNTL__QUICK_PG_ENABLE_MASK; - - if (temp != data) - WREG32(mmRLC_PG_CNTL, data); + WREG32_FIELD(RLC_PG_CNTL, QUICK_PG_ENABLE, enable ? 1 : 0); } static void cz_enable_gfx_cg_power_gating(struct amdgpu_device *adev, bool enable) { - u32 data, orig; - - orig = data = RREG32(mmRLC_PG_CNTL); - - if (enable) - data |= RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK; - else - data &= ~RLC_PG_CNTL__GFX_POWER_GATING_ENABLE_MASK; - - if (orig != data) - WREG32(mmRLC_PG_CNTL, data); + WREG32_FIELD(RLC_PG_CNTL, GFX_POWER_GATING_ENABLE, enable ? 1 : 0); } static void cz_enable_gfx_pipeline_power_gating(struct amdgpu_device *adev, bool enable) { - u32 data, orig; - - orig = data = RREG32(mmRLC_PG_CNTL); - - if (enable) - data |= RLC_PG_CNTL__GFX_PIPELINE_PG_ENABLE_MASK; - else - data &= ~RLC_PG_CNTL__GFX_PIPELINE_PG_ENABLE_MASK; - - if (orig != data) - WREG32(mmRLC_PG_CNTL, data); + WREG32_FIELD(RLC_PG_CNTL, GFX_PIPELINE_PG_ENABLE, enable ? 1 : 0); /* Read any GFX register to wake up GFX. */ if (!enable) - data = RREG32(mmDB_RENDER_CONTROL); + RREG32(mmDB_RENDER_CONTROL); } static void cz_update_gfx_cg_power_gating(struct amdgpu_device *adev, @@ -5430,15 +5422,15 @@ static void gfx_v8_0_send_serdes_cmd(struct amdgpu_device *adev, data = RREG32(mmRLC_SERDES_WR_CTRL); if (adev->asic_type == CHIP_STONEY) - data &= ~(RLC_SERDES_WR_CTRL__WRITE_COMMAND_MASK | - RLC_SERDES_WR_CTRL__READ_COMMAND_MASK | - RLC_SERDES_WR_CTRL__P1_SELECT_MASK | - RLC_SERDES_WR_CTRL__P2_SELECT_MASK | - RLC_SERDES_WR_CTRL__RDDATA_RESET_MASK | - RLC_SERDES_WR_CTRL__POWER_DOWN_MASK | - RLC_SERDES_WR_CTRL__POWER_UP_MASK | - RLC_SERDES_WR_CTRL__SHORT_FORMAT_MASK | - RLC_SERDES_WR_CTRL__SRBM_OVERRIDE_MASK); + data &= ~(RLC_SERDES_WR_CTRL__WRITE_COMMAND_MASK | + RLC_SERDES_WR_CTRL__READ_COMMAND_MASK | + RLC_SERDES_WR_CTRL__P1_SELECT_MASK | + RLC_SERDES_WR_CTRL__P2_SELECT_MASK | + RLC_SERDES_WR_CTRL__RDDATA_RESET_MASK | + RLC_SERDES_WR_CTRL__POWER_DOWN_MASK | + RLC_SERDES_WR_CTRL__POWER_UP_MASK | + RLC_SERDES_WR_CTRL__SHORT_FORMAT_MASK | + RLC_SERDES_WR_CTRL__SRBM_OVERRIDE_MASK); else data &= ~(RLC_SERDES_WR_CTRL__WRITE_COMMAND_MASK | RLC_SERDES_WR_CTRL__READ_COMMAND_MASK | @@ -5461,10 +5453,10 @@ static void gfx_v8_0_send_serdes_cmd(struct amdgpu_device *adev, #define MSG_ENTER_RLC_SAFE_MODE 1 #define MSG_EXIT_RLC_SAFE_MODE 0 - -#define RLC_GPR_REG2__REQ_MASK 0x00000001 -#define RLC_GPR_REG2__MESSAGE__SHIFT 0x00000001 -#define RLC_GPR_REG2__MESSAGE_MASK 0x0000001e +#define RLC_GPR_REG2__REQ_MASK 0x00000001 +#define RLC_GPR_REG2__REQ__SHIFT 0 +#define RLC_GPR_REG2__MESSAGE__SHIFT 0x00000001 +#define RLC_GPR_REG2__MESSAGE_MASK 0x0000001e static void cz_enter_rlc_safe_mode(struct amdgpu_device *adev) { @@ -5494,7 +5486,7 @@ static void cz_enter_rlc_safe_mode(struct amdgpu_device *adev) } for (i = 0; i < adev->usec_timeout; i++) { - if ((RREG32(mmRLC_GPR_REG2) & RLC_GPR_REG2__REQ_MASK) == 0) + if (!REG_GET_FIELD(RREG32(mmRLC_GPR_REG2), RLC_GPR_REG2, REQ)) break; udelay(1); } @@ -5522,7 +5514,7 @@ static void cz_exit_rlc_safe_mode(struct amdgpu_device *adev) } for (i = 0; i < adev->usec_timeout; i++) { - if ((RREG32(mmRLC_GPR_REG2) & RLC_GPR_REG2__REQ_MASK) == 0) + if (!REG_GET_FIELD(RREG32(mmRLC_GPR_REG2), RLC_GPR_REG2, REQ)) break; udelay(1); } @@ -5554,7 +5546,7 @@ static void iceland_enter_rlc_safe_mode(struct amdgpu_device *adev) } for (i = 0; i < adev->usec_timeout; i++) { - if ((RREG32(mmRLC_SAFE_MODE) & RLC_SAFE_MODE__CMD_MASK) == 0) + if (!REG_GET_FIELD(RREG32(mmRLC_SAFE_MODE), RLC_SAFE_MODE, CMD)) break; udelay(1); } @@ -5581,7 +5573,7 @@ static void iceland_exit_rlc_safe_mode(struct amdgpu_device *adev) } for (i = 0; i < adev->usec_timeout; i++) { - if ((RREG32(mmRLC_SAFE_MODE) & RLC_SAFE_MODE__CMD_MASK) == 0) + if (!REG_GET_FIELD(RREG32(mmRLC_SAFE_MODE), RLC_SAFE_MODE, CMD)) break; udelay(1); } @@ -5622,21 +5614,12 @@ static void gfx_v8_0_update_medium_grain_clock_gating(struct amdgpu_device *adev /* It is disabled by HW by default */ if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG)) { if (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGLS) { - if (adev->cg_flags & AMD_CG_SUPPORT_GFX_RLC_LS) { + if (adev->cg_flags & AMD_CG_SUPPORT_GFX_RLC_LS) /* 1 - RLC memory Light sleep */ - temp = data = RREG32(mmRLC_MEM_SLP_CNTL); - data |= RLC_MEM_SLP_CNTL__RLC_MEM_LS_EN_MASK; - if (temp != data) - WREG32(mmRLC_MEM_SLP_CNTL, data); - } + WREG32_FIELD(RLC_MEM_SLP_CNTL, RLC_MEM_LS_EN, 1); - if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CP_LS) { - /* 2 - CP memory Light sleep */ - temp = data = RREG32(mmCP_MEM_SLP_CNTL); - data |= CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK; - if (temp != data) - WREG32(mmCP_MEM_SLP_CNTL, data); - } + if (adev->cg_flags & AMD_CG_SUPPORT_GFX_CP_LS) + WREG32_FIELD(CP_MEM_SLP_CNTL, CP_MEM_LS_EN, 1); } /* 3 - RLC_CGTT_MGCG_OVERRIDE */ @@ -5854,25 +5837,18 @@ static int gfx_v8_0_set_clockgating_state(void *handle, static u32 gfx_v8_0_ring_get_rptr_gfx(struct amdgpu_ring *ring) { - u32 rptr; - - rptr = ring->adev->wb.wb[ring->rptr_offs]; - - return rptr; + return ring->adev->wb.wb[ring->rptr_offs]; } static u32 gfx_v8_0_ring_get_wptr_gfx(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; - u32 wptr; if (ring->use_doorbell) /* XXX check if swapping is necessary on BE */ - wptr = ring->adev->wb.wb[ring->wptr_offs]; + return ring->adev->wb.wb[ring->wptr_offs]; else - wptr = RREG32(mmCP_RB0_WPTR); - - return wptr; + return RREG32(mmCP_RB0_WPTR); } static void gfx_v8_0_ring_set_wptr_gfx(struct amdgpu_ring *ring) @@ -5971,9 +5947,9 @@ static void gfx_v8_0_ring_emit_ib_compute(struct amdgpu_ring *ring, amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2)); amdgpu_ring_write(ring, #ifdef __BIG_ENDIAN - (2 << 0) | + (2 << 0) | #endif - (ib->gpu_addr & 0xFFFFFFFC)); + (ib->gpu_addr & 0xFFFFFFFC)); amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr) & 0xFFFF); amdgpu_ring_write(ring, control); } @@ -6118,33 +6094,14 @@ static void gfx_v8_0_ring_emit_fence_compute(struct amdgpu_ring *ring, static void gfx_v8_0_set_gfx_eop_interrupt_state(struct amdgpu_device *adev, enum amdgpu_interrupt_state state) { - u32 cp_int_cntl; - - switch (state) { - case AMDGPU_IRQ_STATE_DISABLE: - cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0); - cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0, - TIME_STAMP_INT_ENABLE, 0); - WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl); - break; - case AMDGPU_IRQ_STATE_ENABLE: - cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0); - cp_int_cntl = - REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0, - TIME_STAMP_INT_ENABLE, 1); - WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl); - break; - default: - break; - } + WREG32_FIELD(CP_INT_CNTL_RING0, TIME_STAMP_INT_ENABLE, + state == AMDGPU_IRQ_STATE_DISABLE ? 0 : 1); } static void gfx_v8_0_set_compute_eop_interrupt_state(struct amdgpu_device *adev, int me, int pipe, enum amdgpu_interrupt_state state) { - u32 mec_int_cntl, mec_int_cntl_reg; - /* * amdgpu controls only pipe 0 of MEC1. That's why this function only * handles the setting of interrupts for this specific pipe. All other @@ -6154,7 +6111,6 @@ static void gfx_v8_0_set_compute_eop_interrupt_state(struct amdgpu_device *adev, if (me == 1) { switch (pipe) { case 0: - mec_int_cntl_reg = mmCP_ME1_PIPE0_INT_CNTL; break; default: DRM_DEBUG("invalid pipe %d\n", pipe); @@ -6165,22 +6121,8 @@ static void gfx_v8_0_set_compute_eop_interrupt_state(struct amdgpu_device *adev, return; } - switch (state) { - case AMDGPU_IRQ_STATE_DISABLE: - mec_int_cntl = RREG32(mec_int_cntl_reg); - mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL, - TIME_STAMP_INT_ENABLE, 0); - WREG32(mec_int_cntl_reg, mec_int_cntl); - break; - case AMDGPU_IRQ_STATE_ENABLE: - mec_int_cntl = RREG32(mec_int_cntl_reg); - mec_int_cntl = REG_SET_FIELD(mec_int_cntl, CP_ME1_PIPE0_INT_CNTL, - TIME_STAMP_INT_ENABLE, 1); - WREG32(mec_int_cntl_reg, mec_int_cntl); - break; - default: - break; - } + WREG32_FIELD(CP_ME1_PIPE0_INT_CNTL, TIME_STAMP_INT_ENABLE, + state == AMDGPU_IRQ_STATE_DISABLE ? 0 : 1); } static int gfx_v8_0_set_priv_reg_fault_state(struct amdgpu_device *adev, @@ -6188,24 +6130,8 @@ static int gfx_v8_0_set_priv_reg_fault_state(struct amdgpu_device *adev, unsigned type, enum amdgpu_interrupt_state state) { - u32 cp_int_cntl; - - switch (state) { - case AMDGPU_IRQ_STATE_DISABLE: - cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0); - cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0, - PRIV_REG_INT_ENABLE, 0); - WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl); - break; - case AMDGPU_IRQ_STATE_ENABLE: - cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0); - cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0, - PRIV_REG_INT_ENABLE, 1); - WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl); - break; - default: - break; - } + WREG32_FIELD(CP_INT_CNTL_RING0, PRIV_REG_INT_ENABLE, + state == AMDGPU_IRQ_STATE_DISABLE ? 0 : 1); return 0; } @@ -6215,24 +6141,8 @@ static int gfx_v8_0_set_priv_inst_fault_state(struct amdgpu_device *adev, unsigned type, enum amdgpu_interrupt_state state) { - u32 cp_int_cntl; - - switch (state) { - case AMDGPU_IRQ_STATE_DISABLE: - cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0); - cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0, - PRIV_INSTR_INT_ENABLE, 0); - WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl); - break; - case AMDGPU_IRQ_STATE_ENABLE: - cp_int_cntl = RREG32(mmCP_INT_CNTL_RING0); - cp_int_cntl = REG_SET_FIELD(cp_int_cntl, CP_INT_CNTL_RING0, - PRIV_INSTR_INT_ENABLE, 1); - WREG32(mmCP_INT_CNTL_RING0, cp_int_cntl); - break; - default: - break; - } + WREG32_FIELD(CP_INT_CNTL_RING0, PRIV_INSTR_INT_ENABLE, + state == AMDGPU_IRQ_STATE_DISABLE ? 0 : 1); return 0; } @@ -6338,7 +6248,10 @@ const struct amd_ip_funcs gfx_v8_0_ip_funcs = { .resume = gfx_v8_0_resume, .is_idle = gfx_v8_0_is_idle, .wait_for_idle = gfx_v8_0_wait_for_idle, + .check_soft_reset = gfx_v8_0_check_soft_reset, + .pre_soft_reset = gfx_v8_0_pre_soft_reset, .soft_reset = gfx_v8_0_soft_reset, + .post_soft_reset = gfx_v8_0_post_soft_reset, .set_clockgating_state = gfx_v8_0_set_clockgating_state, .set_powergating_state = gfx_v8_0_set_powergating_state, }; @@ -6479,15 +6392,12 @@ static u32 gfx_v8_0_get_cu_active_bitmap(struct amdgpu_device *adev) { u32 data, mask; - data = RREG32(mmCC_GC_SHADER_ARRAY_CONFIG); - data |= RREG32(mmGC_USER_SHADER_ARRAY_CONFIG); - - data &= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_CUS_MASK; - data >>= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_CUS__SHIFT; + data = RREG32(mmCC_GC_SHADER_ARRAY_CONFIG) | + RREG32(mmGC_USER_SHADER_ARRAY_CONFIG); mask = gfx_v8_0_create_bitmask(adev->gfx.config.max_cu_per_sh); - return (~data) & mask; + return ~REG_GET_FIELD(data, CC_GC_SHADER_ARRAY_CONFIG, INACTIVE_CUS) & mask; } static void gfx_v8_0_get_cu_info(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.h b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.h index bc82c794312c..ebed1f829297 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.h +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.h @@ -26,6 +26,4 @@ extern const struct amd_ip_funcs gfx_v8_0_ip_funcs; -void gfx_v8_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num); - #endif diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c index 0b0f08641eed..aa0c4b964621 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c @@ -183,7 +183,7 @@ static int gmc_v7_0_mc_load_microcode(struct amdgpu_device *adev) const struct mc_firmware_header_v1_0 *hdr; const __le32 *fw_data = NULL; const __le32 *io_mc_regs = NULL; - u32 running, blackout = 0; + u32 running; int i, ucode_size, regs_size; if (!adev->mc.fw) @@ -203,11 +203,6 @@ static int gmc_v7_0_mc_load_microcode(struct amdgpu_device *adev) running = REG_GET_FIELD(RREG32(mmMC_SEQ_SUP_CNTL), MC_SEQ_SUP_CNTL, RUN); if (running == 0) { - if (running) { - blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL); - WREG32(mmMC_SHARED_BLACKOUT_CNTL, blackout | 1); - } - /* reset the engine and set to writable */ WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008); WREG32(mmMC_SEQ_SUP_CNTL, 0x00000010); @@ -239,9 +234,6 @@ static int gmc_v7_0_mc_load_microcode(struct amdgpu_device *adev) break; udelay(1); } - - if (running) - WREG32(mmMC_SHARED_BLACKOUT_CNTL, blackout); } return 0; @@ -393,7 +385,7 @@ static int gmc_v7_0_mc_init(struct amdgpu_device *adev) * size equal to the 1024 or vram, whichever is larger. */ if (amdgpu_gart_size == -1) - adev->mc.gtt_size = max((1024ULL << 20), adev->mc.mc_vram_size); + adev->mc.gtt_size = amdgpu_ttm_get_gtt_mem_size(adev); else adev->mc.gtt_size = (uint64_t)amdgpu_gart_size << 20; @@ -953,6 +945,11 @@ static int gmc_v7_0_sw_init(void *handle) return r; } + r = amdgpu_ttm_global_init(adev); + if (r) { + return r; + } + r = gmc_v7_0_mc_init(adev); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c index 2aee2c6f3cd5..84c10d5117a9 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c @@ -261,7 +261,7 @@ static int gmc_v8_0_mc_load_microcode(struct amdgpu_device *adev) const struct mc_firmware_header_v1_0 *hdr; const __le32 *fw_data = NULL; const __le32 *io_mc_regs = NULL; - u32 running, blackout = 0; + u32 running; int i, ucode_size, regs_size; if (!adev->mc.fw) @@ -287,11 +287,6 @@ static int gmc_v8_0_mc_load_microcode(struct amdgpu_device *adev) running = REG_GET_FIELD(RREG32(mmMC_SEQ_SUP_CNTL), MC_SEQ_SUP_CNTL, RUN); if (running == 0) { - if (running) { - blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL); - WREG32(mmMC_SHARED_BLACKOUT_CNTL, blackout | 1); - } - /* reset the engine and set to writable */ WREG32(mmMC_SEQ_SUP_CNTL, 0x00000008); WREG32(mmMC_SEQ_SUP_CNTL, 0x00000010); @@ -323,9 +318,6 @@ static int gmc_v8_0_mc_load_microcode(struct amdgpu_device *adev) break; udelay(1); } - - if (running) - WREG32(mmMC_SHARED_BLACKOUT_CNTL, blackout); } return 0; @@ -477,7 +469,7 @@ static int gmc_v8_0_mc_init(struct amdgpu_device *adev) * size equal to the 1024 or vram, whichever is larger. */ if (amdgpu_gart_size == -1) - adev->mc.gtt_size = max((1024ULL << 20), adev->mc.mc_vram_size); + adev->mc.gtt_size = amdgpu_ttm_get_gtt_mem_size(adev); else adev->mc.gtt_size = (uint64_t)amdgpu_gart_size << 20; @@ -957,6 +949,11 @@ static int gmc_v8_0_sw_init(void *handle) return r; } + r = amdgpu_ttm_global_init(adev); + if (r) { + return r; + } + r = gmc_v8_0_mc_init(adev); if (r) return r; @@ -1100,9 +1097,8 @@ static int gmc_v8_0_wait_for_idle(void *handle) } -static int gmc_v8_0_soft_reset(void *handle) +static int gmc_v8_0_check_soft_reset(void *handle) { - struct amdgpu_mode_mc_save save; u32 srbm_soft_reset = 0; struct amdgpu_device *adev = (struct amdgpu_device *)handle; u32 tmp = RREG32(mmSRBM_STATUS); @@ -1117,13 +1113,42 @@ static int gmc_v8_0_soft_reset(void *handle) srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_MC, 1); } - if (srbm_soft_reset) { - gmc_v8_0_mc_stop(adev, &save); - if (gmc_v8_0_wait_for_idle((void *)adev)) { - dev_warn(adev->dev, "Wait for GMC idle timed out !\n"); - } + adev->ip_block_status[AMD_IP_BLOCK_TYPE_GMC].hang = true; + adev->mc.srbm_soft_reset = srbm_soft_reset; + } else { + adev->ip_block_status[AMD_IP_BLOCK_TYPE_GMC].hang = false; + adev->mc.srbm_soft_reset = 0; + } + return 0; +} +static int gmc_v8_0_pre_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_GMC].hang) + return 0; + + gmc_v8_0_mc_stop(adev, &adev->mc.save); + if (gmc_v8_0_wait_for_idle(adev)) { + dev_warn(adev->dev, "Wait for GMC idle timed out !\n"); + } + + return 0; +} + +static int gmc_v8_0_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_GMC].hang) + return 0; + srbm_soft_reset = adev->mc.srbm_soft_reset; + + if (srbm_soft_reset) { + u32 tmp; tmp = RREG32(mmSRBM_SOFT_RESET); tmp |= srbm_soft_reset; @@ -1139,14 +1164,22 @@ static int gmc_v8_0_soft_reset(void *handle) /* Wait a little for things to settle down */ udelay(50); - - gmc_v8_0_mc_resume(adev, &save); - udelay(50); } return 0; } +static int gmc_v8_0_post_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_GMC].hang) + return 0; + + gmc_v8_0_mc_resume(adev, &adev->mc.save); + return 0; +} + static int gmc_v8_0_vm_fault_interrupt_state(struct amdgpu_device *adev, struct amdgpu_irq_src *src, unsigned type, @@ -1414,7 +1447,10 @@ const struct amd_ip_funcs gmc_v8_0_ip_funcs = { .resume = gmc_v8_0_resume, .is_idle = gmc_v8_0_is_idle, .wait_for_idle = gmc_v8_0_wait_for_idle, + .check_soft_reset = gmc_v8_0_check_soft_reset, + .pre_soft_reset = gmc_v8_0_pre_soft_reset, .soft_reset = gmc_v8_0_soft_reset, + .post_soft_reset = gmc_v8_0_post_soft_reset, .set_clockgating_state = gmc_v8_0_set_clockgating_state, .set_powergating_state = gmc_v8_0_set_powergating_state, }; diff --git a/drivers/gpu/drm/amd/amdgpu/kv_dpm.c b/drivers/gpu/drm/amd/amdgpu/kv_dpm.c index a845e883f5fa..f8618a3881a8 100644 --- a/drivers/gpu/drm/amd/amdgpu/kv_dpm.c +++ b/drivers/gpu/drm/amd/amdgpu/kv_dpm.c @@ -2845,7 +2845,11 @@ static int kv_dpm_init(struct amdgpu_device *adev) pi->caps_tcp_ramping = true; } - pi->caps_sclk_ds = true; + if (amdgpu_sclk_deep_sleep_en) + pi->caps_sclk_ds = true; + else + pi->caps_sclk_ds = false; + pi->enable_auto_thermal_throttling = true; pi->disable_nb_ps3_in_battery = false; if (amdgpu_bapm == 0) diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c index 1351c7e834a2..e82229686783 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c @@ -749,24 +749,16 @@ static void sdma_v2_4_vm_copy_pte(struct amdgpu_ib *ib, uint64_t pe, uint64_t src, unsigned count) { - while (count) { - unsigned bytes = count * 8; - if (bytes > 0x1FFFF8) - bytes = 0x1FFFF8; - - ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) | - SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR); - ib->ptr[ib->length_dw++] = bytes; - ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */ - ib->ptr[ib->length_dw++] = lower_32_bits(src); - ib->ptr[ib->length_dw++] = upper_32_bits(src); - ib->ptr[ib->length_dw++] = lower_32_bits(pe); - ib->ptr[ib->length_dw++] = upper_32_bits(pe); - - pe += bytes; - src += bytes; - count -= bytes / 8; - } + unsigned bytes = count * 8; + + ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) | + SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR); + ib->ptr[ib->length_dw++] = bytes; + ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */ + ib->ptr[ib->length_dw++] = lower_32_bits(src); + ib->ptr[ib->length_dw++] = upper_32_bits(src); + ib->ptr[ib->length_dw++] = lower_32_bits(pe); + ib->ptr[ib->length_dw++] = upper_32_bits(pe); } /** @@ -774,39 +766,27 @@ static void sdma_v2_4_vm_copy_pte(struct amdgpu_ib *ib, * * @ib: indirect buffer to fill with commands * @pe: addr of the page entry - * @addr: dst addr to write into pe + * @value: dst addr to write into pe * @count: number of page entries to update * @incr: increase next addr by incr bytes - * @flags: access flags * * Update PTEs by writing them manually using sDMA (CIK). */ -static void sdma_v2_4_vm_write_pte(struct amdgpu_ib *ib, - const dma_addr_t *pages_addr, uint64_t pe, - uint64_t addr, unsigned count, - uint32_t incr, uint32_t flags) +static void sdma_v2_4_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe, + uint64_t value, unsigned count, + uint32_t incr) { - uint64_t value; - unsigned ndw; - - while (count) { - ndw = count * 2; - if (ndw > 0xFFFFE) - ndw = 0xFFFFE; - - /* for non-physically contiguous pages (system) */ - ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) | - SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR); - ib->ptr[ib->length_dw++] = pe; - ib->ptr[ib->length_dw++] = upper_32_bits(pe); - ib->ptr[ib->length_dw++] = ndw; - for (; ndw > 0; ndw -= 2, --count, pe += 8) { - value = amdgpu_vm_map_gart(pages_addr, addr); - addr += incr; - value |= flags; - ib->ptr[ib->length_dw++] = value; - ib->ptr[ib->length_dw++] = upper_32_bits(value); - } + unsigned ndw = count * 2; + + ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) | + SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR); + ib->ptr[ib->length_dw++] = pe; + ib->ptr[ib->length_dw++] = upper_32_bits(pe); + ib->ptr[ib->length_dw++] = ndw; + for (; ndw > 0; ndw -= 2, --count, pe += 8) { + ib->ptr[ib->length_dw++] = lower_32_bits(value); + ib->ptr[ib->length_dw++] = upper_32_bits(value); + value += incr; } } @@ -822,40 +802,21 @@ static void sdma_v2_4_vm_write_pte(struct amdgpu_ib *ib, * * Update the page tables using sDMA (CIK). */ -static void sdma_v2_4_vm_set_pte_pde(struct amdgpu_ib *ib, - uint64_t pe, +static void sdma_v2_4_vm_set_pte_pde(struct amdgpu_ib *ib, uint64_t pe, uint64_t addr, unsigned count, uint32_t incr, uint32_t flags) { - uint64_t value; - unsigned ndw; - - while (count) { - ndw = count; - if (ndw > 0x7FFFF) - ndw = 0x7FFFF; - - if (flags & AMDGPU_PTE_VALID) - value = addr; - else - value = 0; - - /* for physically contiguous pages (vram) */ - ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_GEN_PTEPDE); - ib->ptr[ib->length_dw++] = pe; /* dst addr */ - ib->ptr[ib->length_dw++] = upper_32_bits(pe); - ib->ptr[ib->length_dw++] = flags; /* mask */ - ib->ptr[ib->length_dw++] = 0; - ib->ptr[ib->length_dw++] = value; /* value */ - ib->ptr[ib->length_dw++] = upper_32_bits(value); - ib->ptr[ib->length_dw++] = incr; /* increment size */ - ib->ptr[ib->length_dw++] = 0; - ib->ptr[ib->length_dw++] = ndw; /* number of entries */ - - pe += ndw * 8; - addr += ndw * incr; - count -= ndw; - } + /* for physically contiguous pages (vram) */ + ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_GEN_PTEPDE); + ib->ptr[ib->length_dw++] = lower_32_bits(pe); /* dst addr */ + ib->ptr[ib->length_dw++] = upper_32_bits(pe); + ib->ptr[ib->length_dw++] = flags; /* mask */ + ib->ptr[ib->length_dw++] = 0; + ib->ptr[ib->length_dw++] = lower_32_bits(addr); /* value */ + ib->ptr[ib->length_dw++] = upper_32_bits(addr); + ib->ptr[ib->length_dw++] = incr; /* increment size */ + ib->ptr[ib->length_dw++] = 0; + ib->ptr[ib->length_dw++] = count; /* number of entries */ } /** diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c index 653ce5ed55ae..bee4978bec73 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c @@ -976,24 +976,16 @@ static void sdma_v3_0_vm_copy_pte(struct amdgpu_ib *ib, uint64_t pe, uint64_t src, unsigned count) { - while (count) { - unsigned bytes = count * 8; - if (bytes > 0x1FFFF8) - bytes = 0x1FFFF8; - - ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) | - SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR); - ib->ptr[ib->length_dw++] = bytes; - ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */ - ib->ptr[ib->length_dw++] = lower_32_bits(src); - ib->ptr[ib->length_dw++] = upper_32_bits(src); - ib->ptr[ib->length_dw++] = lower_32_bits(pe); - ib->ptr[ib->length_dw++] = upper_32_bits(pe); - - pe += bytes; - src += bytes; - count -= bytes / 8; - } + unsigned bytes = count * 8; + + ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) | + SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR); + ib->ptr[ib->length_dw++] = bytes; + ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */ + ib->ptr[ib->length_dw++] = lower_32_bits(src); + ib->ptr[ib->length_dw++] = upper_32_bits(src); + ib->ptr[ib->length_dw++] = lower_32_bits(pe); + ib->ptr[ib->length_dw++] = upper_32_bits(pe); } /** @@ -1001,39 +993,27 @@ static void sdma_v3_0_vm_copy_pte(struct amdgpu_ib *ib, * * @ib: indirect buffer to fill with commands * @pe: addr of the page entry - * @addr: dst addr to write into pe + * @value: dst addr to write into pe * @count: number of page entries to update * @incr: increase next addr by incr bytes - * @flags: access flags * * Update PTEs by writing them manually using sDMA (CIK). */ -static void sdma_v3_0_vm_write_pte(struct amdgpu_ib *ib, - const dma_addr_t *pages_addr, uint64_t pe, - uint64_t addr, unsigned count, - uint32_t incr, uint32_t flags) -{ - uint64_t value; - unsigned ndw; - - while (count) { - ndw = count * 2; - if (ndw > 0xFFFFE) - ndw = 0xFFFFE; - - /* for non-physically contiguous pages (system) */ - ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) | - SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR); - ib->ptr[ib->length_dw++] = pe; - ib->ptr[ib->length_dw++] = upper_32_bits(pe); - ib->ptr[ib->length_dw++] = ndw; - for (; ndw > 0; ndw -= 2, --count, pe += 8) { - value = amdgpu_vm_map_gart(pages_addr, addr); - addr += incr; - value |= flags; - ib->ptr[ib->length_dw++] = value; - ib->ptr[ib->length_dw++] = upper_32_bits(value); - } +static void sdma_v3_0_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe, + uint64_t value, unsigned count, + uint32_t incr) +{ + unsigned ndw = count * 2; + + ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) | + SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR); + ib->ptr[ib->length_dw++] = lower_32_bits(pe); + ib->ptr[ib->length_dw++] = upper_32_bits(pe); + ib->ptr[ib->length_dw++] = ndw; + for (; ndw > 0; ndw -= 2, --count, pe += 8) { + ib->ptr[ib->length_dw++] = lower_32_bits(value); + ib->ptr[ib->length_dw++] = upper_32_bits(value); + value += incr; } } @@ -1049,40 +1029,21 @@ static void sdma_v3_0_vm_write_pte(struct amdgpu_ib *ib, * * Update the page tables using sDMA (CIK). */ -static void sdma_v3_0_vm_set_pte_pde(struct amdgpu_ib *ib, - uint64_t pe, +static void sdma_v3_0_vm_set_pte_pde(struct amdgpu_ib *ib, uint64_t pe, uint64_t addr, unsigned count, uint32_t incr, uint32_t flags) { - uint64_t value; - unsigned ndw; - - while (count) { - ndw = count; - if (ndw > 0x7FFFF) - ndw = 0x7FFFF; - - if (flags & AMDGPU_PTE_VALID) - value = addr; - else - value = 0; - - /* for physically contiguous pages (vram) */ - ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_GEN_PTEPDE); - ib->ptr[ib->length_dw++] = pe; /* dst addr */ - ib->ptr[ib->length_dw++] = upper_32_bits(pe); - ib->ptr[ib->length_dw++] = flags; /* mask */ - ib->ptr[ib->length_dw++] = 0; - ib->ptr[ib->length_dw++] = value; /* value */ - ib->ptr[ib->length_dw++] = upper_32_bits(value); - ib->ptr[ib->length_dw++] = incr; /* increment size */ - ib->ptr[ib->length_dw++] = 0; - ib->ptr[ib->length_dw++] = ndw; /* number of entries */ - - pe += ndw * 8; - addr += ndw * incr; - count -= ndw; - } + /* for physically contiguous pages (vram) */ + ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_GEN_PTEPDE); + ib->ptr[ib->length_dw++] = lower_32_bits(pe); /* dst addr */ + ib->ptr[ib->length_dw++] = upper_32_bits(pe); + ib->ptr[ib->length_dw++] = flags; /* mask */ + ib->ptr[ib->length_dw++] = 0; + ib->ptr[ib->length_dw++] = lower_32_bits(addr); /* value */ + ib->ptr[ib->length_dw++] = upper_32_bits(addr); + ib->ptr[ib->length_dw++] = incr; /* increment size */ + ib->ptr[ib->length_dw++] = 0; + ib->ptr[ib->length_dw++] = count; /* number of entries */ } /** @@ -1320,28 +1281,79 @@ static int sdma_v3_0_wait_for_idle(void *handle) return -ETIMEDOUT; } -static int sdma_v3_0_soft_reset(void *handle) +static int sdma_v3_0_check_soft_reset(void *handle) { - u32 srbm_soft_reset = 0; struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset = 0; u32 tmp = RREG32(mmSRBM_STATUS2); - if (tmp & SRBM_STATUS2__SDMA_BUSY_MASK) { - /* sdma0 */ - tmp = RREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET); - tmp = REG_SET_FIELD(tmp, SDMA0_F32_CNTL, HALT, 0); - WREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET, tmp); + if ((tmp & SRBM_STATUS2__SDMA_BUSY_MASK) || + (tmp & SRBM_STATUS2__SDMA1_BUSY_MASK)) { srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA_MASK; - } - if (tmp & SRBM_STATUS2__SDMA1_BUSY_MASK) { - /* sdma1 */ - tmp = RREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET); - tmp = REG_SET_FIELD(tmp, SDMA0_F32_CNTL, HALT, 0); - WREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET, tmp); srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA1_MASK; } if (srbm_soft_reset) { + adev->ip_block_status[AMD_IP_BLOCK_TYPE_SDMA].hang = true; + adev->sdma.srbm_soft_reset = srbm_soft_reset; + } else { + adev->ip_block_status[AMD_IP_BLOCK_TYPE_SDMA].hang = false; + adev->sdma.srbm_soft_reset = 0; + } + + return 0; +} + +static int sdma_v3_0_pre_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset = 0; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_SDMA].hang) + return 0; + + srbm_soft_reset = adev->sdma.srbm_soft_reset; + + if (REG_GET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_SDMA) || + REG_GET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_SDMA1)) { + sdma_v3_0_ctx_switch_enable(adev, false); + sdma_v3_0_enable(adev, false); + } + + return 0; +} + +static int sdma_v3_0_post_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset = 0; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_SDMA].hang) + return 0; + + srbm_soft_reset = adev->sdma.srbm_soft_reset; + + if (REG_GET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_SDMA) || + REG_GET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_SDMA1)) { + sdma_v3_0_gfx_resume(adev); + sdma_v3_0_rlc_resume(adev); + } + + return 0; +} + +static int sdma_v3_0_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset = 0; + u32 tmp; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_SDMA].hang) + return 0; + + srbm_soft_reset = adev->sdma.srbm_soft_reset; + + if (srbm_soft_reset) { tmp = RREG32(mmSRBM_SOFT_RESET); tmp |= srbm_soft_reset; dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp); @@ -1559,6 +1571,9 @@ const struct amd_ip_funcs sdma_v3_0_ip_funcs = { .resume = sdma_v3_0_resume, .is_idle = sdma_v3_0_is_idle, .wait_for_idle = sdma_v3_0_wait_for_idle, + .check_soft_reset = sdma_v3_0_check_soft_reset, + .pre_soft_reset = sdma_v3_0_pre_soft_reset, + .post_soft_reset = sdma_v3_0_post_soft_reset, .soft_reset = sdma_v3_0_soft_reset, .set_clockgating_state = sdma_v3_0_set_clockgating_state, .set_powergating_state = sdma_v3_0_set_powergating_state, diff --git a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c index c92055805a45..d127d59f953a 100644 --- a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c @@ -373,10 +373,10 @@ static int tonga_ih_wait_for_idle(void *handle) return -ETIMEDOUT; } -static int tonga_ih_soft_reset(void *handle) +static int tonga_ih_check_soft_reset(void *handle) { - u32 srbm_soft_reset = 0; struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset = 0; u32 tmp = RREG32(mmSRBM_STATUS); if (tmp & SRBM_STATUS__IH_BUSY_MASK) @@ -384,6 +384,48 @@ static int tonga_ih_soft_reset(void *handle) SOFT_RESET_IH, 1); if (srbm_soft_reset) { + adev->ip_block_status[AMD_IP_BLOCK_TYPE_IH].hang = true; + adev->irq.srbm_soft_reset = srbm_soft_reset; + } else { + adev->ip_block_status[AMD_IP_BLOCK_TYPE_IH].hang = false; + adev->irq.srbm_soft_reset = 0; + } + + return 0; +} + +static int tonga_ih_pre_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_IH].hang) + return 0; + + return tonga_ih_hw_fini(adev); +} + +static int tonga_ih_post_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_IH].hang) + return 0; + + return tonga_ih_hw_init(adev); +} + +static int tonga_ih_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_IH].hang) + return 0; + srbm_soft_reset = adev->irq.srbm_soft_reset; + + if (srbm_soft_reset) { + u32 tmp; + tmp = RREG32(mmSRBM_SOFT_RESET); tmp |= srbm_soft_reset; dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp); @@ -427,7 +469,10 @@ const struct amd_ip_funcs tonga_ih_ip_funcs = { .resume = tonga_ih_resume, .is_idle = tonga_ih_is_idle, .wait_for_idle = tonga_ih_wait_for_idle, + .check_soft_reset = tonga_ih_check_soft_reset, + .pre_soft_reset = tonga_ih_pre_soft_reset, .soft_reset = tonga_ih_soft_reset, + .post_soft_reset = tonga_ih_post_soft_reset, .set_clockgating_state = tonga_ih_set_clockgating_state, .set_powergating_state = tonga_ih_set_powergating_state, }; diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c index 132e613ed674..10c0407dcb6e 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c @@ -116,7 +116,7 @@ static int uvd_v4_2_sw_init(void *handle) ring = &adev->uvd.ring; sprintf(ring->name, "uvd"); - r = amdgpu_ring_init(adev, ring, 512, CP_PACKET2, 0xf, + r = amdgpu_ring_init(adev, ring, 512, PACKET0(mmUVD_NO_OP, 0), 0xf, &adev->uvd.irq, 0, AMDGPU_RING_TYPE_UVD); return r; diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c index 101de136ba63..8513376062c1 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c @@ -112,7 +112,7 @@ static int uvd_v5_0_sw_init(void *handle) ring = &adev->uvd.ring; sprintf(ring->name, "uvd"); - r = amdgpu_ring_init(adev, ring, 512, CP_PACKET2, 0xf, + r = amdgpu_ring_init(adev, ring, 512, PACKET0(mmUVD_NO_OP, 0), 0xf, &adev->uvd.irq, 0, AMDGPU_RING_TYPE_UVD); return r; diff --git a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c index 7f21102bfb99..2abe8a93c99f 100644 --- a/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c @@ -116,7 +116,7 @@ static int uvd_v6_0_sw_init(void *handle) ring = &adev->uvd.ring; sprintf(ring->name, "uvd"); - r = amdgpu_ring_init(adev, ring, 512, CP_PACKET2, 0xf, + r = amdgpu_ring_init(adev, ring, 512, PACKET0(mmUVD_NO_OP, 0), 0xf, &adev->uvd.irq, 0, AMDGPU_RING_TYPE_UVD); return r; @@ -396,21 +396,14 @@ static int uvd_v6_0_start(struct amdgpu_device *adev) uvd_v6_0_mc_resume(adev); - /* Set dynamic clock gating in S/W control mode */ - if (adev->cg_flags & AMD_CG_SUPPORT_UVD_MGCG) { - uvd_v6_0_set_sw_clock_gating(adev); - } else { - /* disable clock gating */ - uint32_t data = RREG32(mmUVD_CGC_CTRL); - data &= ~UVD_CGC_CTRL__DYN_CLOCK_MODE_MASK; - WREG32(mmUVD_CGC_CTRL, data); - } + /* disable clock gating */ + WREG32_FIELD(UVD_CGC_CTRL, DYN_CLOCK_MODE, 0); /* disable interupt */ - WREG32_P(mmUVD_MASTINT_EN, 0, ~UVD_MASTINT_EN__VCPU_EN_MASK); + WREG32_FIELD(UVD_MASTINT_EN, VCPU_EN, 0); /* stall UMC and register bus before resetting VCPU */ - WREG32_P(mmUVD_LMI_CTRL2, UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK); + WREG32_FIELD(UVD_LMI_CTRL2, STALL_ARB_UMC, 1); mdelay(1); /* put LMI, VCPU, RBC etc... into reset */ @@ -426,7 +419,7 @@ static int uvd_v6_0_start(struct amdgpu_device *adev) mdelay(5); /* take UVD block out of reset */ - WREG32_P(mmSRBM_SOFT_RESET, 0, ~SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK); + WREG32_FIELD(SRBM_SOFT_RESET, SOFT_RESET_UVD, 0); mdelay(5); /* initialize UVD memory controller */ @@ -461,7 +454,7 @@ static int uvd_v6_0_start(struct amdgpu_device *adev) WREG32(mmUVD_VCPU_CNTL, UVD_VCPU_CNTL__CLK_EN_MASK); /* enable UMC */ - WREG32_P(mmUVD_LMI_CTRL2, 0, ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK); + WREG32_FIELD(UVD_LMI_CTRL2, STALL_ARB_UMC, 0); /* boot up the VCPU */ WREG32(mmUVD_SOFT_RESET, 0); @@ -481,11 +474,9 @@ static int uvd_v6_0_start(struct amdgpu_device *adev) break; DRM_ERROR("UVD not responding, trying to reset the VCPU!!!\n"); - WREG32_P(mmUVD_SOFT_RESET, UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK, - ~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK); + WREG32_FIELD(UVD_SOFT_RESET, VCPU_SOFT_RESET, 1); mdelay(10); - WREG32_P(mmUVD_SOFT_RESET, 0, - ~UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK); + WREG32_FIELD(UVD_SOFT_RESET, VCPU_SOFT_RESET, 0); mdelay(10); r = -1; } @@ -502,15 +493,14 @@ static int uvd_v6_0_start(struct amdgpu_device *adev) /* clear the bit 4 of UVD_STATUS */ WREG32_P(mmUVD_STATUS, 0, ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT)); + /* force RBC into idle state */ rb_bufsz = order_base_2(ring->ring_size); - tmp = 0; - tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz); + tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, rb_bufsz); 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); - /* force RBC into idle state */ WREG32(mmUVD_RBC_RB_CNTL, tmp); /* set the write pointer delay */ @@ -531,7 +521,7 @@ static int uvd_v6_0_start(struct amdgpu_device *adev) ring->wptr = RREG32(mmUVD_RBC_RB_RPTR); WREG32(mmUVD_RBC_RB_WPTR, ring->wptr); - WREG32_P(mmUVD_RBC_RB_CNTL, 0, ~UVD_RBC_RB_CNTL__RB_NO_FETCH_MASK); + WREG32_FIELD(UVD_RBC_RB_CNTL, RB_NO_FETCH, 0); return 0; } @@ -748,20 +738,82 @@ static int uvd_v6_0_wait_for_idle(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; for (i = 0; i < adev->usec_timeout; i++) { - if (!(RREG32(mmSRBM_STATUS) & SRBM_STATUS__UVD_BUSY_MASK)) + if (uvd_v6_0_is_idle(handle)) return 0; } return -ETIMEDOUT; } -static int uvd_v6_0_soft_reset(void *handle) +#define AMDGPU_UVD_STATUS_BUSY_MASK 0xfd +static int uvd_v6_0_check_soft_reset(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset = 0; + u32 tmp = RREG32(mmSRBM_STATUS); + + if (REG_GET_FIELD(tmp, SRBM_STATUS, UVD_RQ_PENDING) || + REG_GET_FIELD(tmp, SRBM_STATUS, UVD_BUSY) || + (RREG32(mmUVD_STATUS) & AMDGPU_UVD_STATUS_BUSY_MASK)) + srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_UVD, 1); + + if (srbm_soft_reset) { + adev->ip_block_status[AMD_IP_BLOCK_TYPE_UVD].hang = true; + adev->uvd.srbm_soft_reset = srbm_soft_reset; + } else { + adev->ip_block_status[AMD_IP_BLOCK_TYPE_UVD].hang = false; + adev->uvd.srbm_soft_reset = 0; + } + return 0; +} +static int uvd_v6_0_pre_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_UVD].hang) + return 0; uvd_v6_0_stop(adev); + return 0; +} + +static int uvd_v6_0_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_UVD].hang) + return 0; + srbm_soft_reset = adev->uvd.srbm_soft_reset; + + if (srbm_soft_reset) { + u32 tmp; + + tmp = RREG32(mmSRBM_SOFT_RESET); + tmp |= srbm_soft_reset; + dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp); + WREG32(mmSRBM_SOFT_RESET, tmp); + tmp = RREG32(mmSRBM_SOFT_RESET); + + udelay(50); + + tmp &= ~srbm_soft_reset; + WREG32(mmSRBM_SOFT_RESET, tmp); + tmp = RREG32(mmSRBM_SOFT_RESET); + + /* Wait a little for things to settle down */ + udelay(50); + } + + return 0; +} + +static int uvd_v6_0_post_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_UVD].hang) + return 0; - WREG32_P(mmSRBM_SOFT_RESET, SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK, - ~SRBM_SOFT_RESET__SOFT_RESET_UVD_MASK); mdelay(5); return uvd_v6_0_start(adev); @@ -902,21 +954,15 @@ static int uvd_v6_0_set_clockgating_state(void *handle, enum amd_clockgating_state state) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - bool enable = (state == AMD_CG_STATE_GATE) ? true : false; - static int curstate = -1; if (adev->asic_type == CHIP_FIJI || - adev->asic_type == CHIP_POLARIS10) - uvd_v6_set_bypass_mode(adev, enable); + adev->asic_type == CHIP_POLARIS10) + uvd_v6_set_bypass_mode(adev, state == AMD_CG_STATE_GATE ? true : false); if (!(adev->cg_flags & AMD_CG_SUPPORT_UVD_MGCG)) return 0; - if (curstate == state) - return 0; - - curstate = state; - if (enable) { + if (state == AMD_CG_STATE_GATE) { /* disable HW gating and enable Sw gating */ uvd_v6_0_set_sw_clock_gating(adev); } else { @@ -946,6 +992,8 @@ static int uvd_v6_0_set_powergating_state(void *handle, if (!(adev->pg_flags & AMD_PG_SUPPORT_UVD)) return 0; + WREG32(mmUVD_POWER_STATUS, UVD_POWER_STATUS__UVD_PG_EN_MASK); + if (state == AMD_PG_STATE_GATE) { uvd_v6_0_stop(adev); return 0; @@ -966,7 +1014,10 @@ const struct amd_ip_funcs uvd_v6_0_ip_funcs = { .resume = uvd_v6_0_resume, .is_idle = uvd_v6_0_is_idle, .wait_for_idle = uvd_v6_0_wait_for_idle, + .check_soft_reset = uvd_v6_0_check_soft_reset, + .pre_soft_reset = uvd_v6_0_pre_soft_reset, .soft_reset = uvd_v6_0_soft_reset, + .post_soft_reset = uvd_v6_0_post_soft_reset, .set_clockgating_state = uvd_v6_0_set_clockgating_state, .set_powergating_state = uvd_v6_0_set_powergating_state, }; diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c index 80a37a602181..5fa55b52c00e 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c @@ -40,6 +40,7 @@ #define VCE_V2_0_FW_SIZE (256 * 1024) #define VCE_V2_0_STACK_SIZE (64 * 1024) #define VCE_V2_0_DATA_SIZE (23552 * AMDGPU_MAX_VCE_HANDLES) +#define VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK 0x02 static void vce_v2_0_mc_resume(struct amdgpu_device *adev); static void vce_v2_0_set_ring_funcs(struct amdgpu_device *adev); @@ -96,6 +97,49 @@ static void vce_v2_0_ring_set_wptr(struct amdgpu_ring *ring) WREG32(mmVCE_RB_WPTR2, ring->wptr); } +static int vce_v2_0_lmi_clean(struct amdgpu_device *adev) +{ + int i, j; + + for (i = 0; i < 10; ++i) { + for (j = 0; j < 100; ++j) { + uint32_t status = RREG32(mmVCE_LMI_STATUS); + + if (status & 0x337f) + return 0; + mdelay(10); + } + } + + return -ETIMEDOUT; +} + +static int vce_v2_0_firmware_loaded(struct amdgpu_device *adev) +{ + int i, j; + + for (i = 0; i < 10; ++i) { + for (j = 0; j < 100; ++j) { + uint32_t status = RREG32(mmVCE_STATUS); + + if (status & VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK) + return 0; + mdelay(10); + } + + DRM_ERROR("VCE not responding, trying to reset the ECPU!!!\n"); + WREG32_P(mmVCE_SOFT_RESET, + VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK, + ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); + mdelay(10); + WREG32_P(mmVCE_SOFT_RESET, 0, + ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); + mdelay(10); + } + + return -ETIMEDOUT; +} + /** * vce_v2_0_start - start VCE block * @@ -106,7 +150,7 @@ static void vce_v2_0_ring_set_wptr(struct amdgpu_ring *ring) static int vce_v2_0_start(struct amdgpu_device *adev) { struct amdgpu_ring *ring; - int i, j, r; + int r; vce_v2_0_mc_resume(adev); @@ -127,36 +171,12 @@ static int vce_v2_0_start(struct amdgpu_device *adev) WREG32(mmVCE_RB_BASE_HI2, upper_32_bits(ring->gpu_addr)); WREG32(mmVCE_RB_SIZE2, ring->ring_size / 4); - WREG32_P(mmVCE_VCPU_CNTL, VCE_VCPU_CNTL__CLK_EN_MASK, ~VCE_VCPU_CNTL__CLK_EN_MASK); - - WREG32_P(mmVCE_SOFT_RESET, - VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK, - ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); - + WREG32_FIELD(VCE_VCPU_CNTL, CLK_EN, 1); + WREG32_FIELD(VCE_SOFT_RESET, ECPU_SOFT_RESET, 1); mdelay(100); + WREG32_FIELD(VCE_SOFT_RESET, ECPU_SOFT_RESET, 0); - WREG32_P(mmVCE_SOFT_RESET, 0, ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); - - for (i = 0; i < 10; ++i) { - uint32_t status; - for (j = 0; j < 100; ++j) { - status = RREG32(mmVCE_STATUS); - if (status & 2) - break; - mdelay(10); - } - r = 0; - if (status & 2) - break; - - DRM_ERROR("VCE not responding, trying to reset the ECPU!!!\n"); - WREG32_P(mmVCE_SOFT_RESET, VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK, - ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); - mdelay(10); - WREG32_P(mmVCE_SOFT_RESET, 0, ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); - mdelay(10); - r = -1; - } + r = vce_v2_0_firmware_loaded(adev); /* clear BUSY flag */ WREG32_P(mmVCE_STATUS, 0, ~1); @@ -338,47 +358,50 @@ static void vce_v2_0_set_sw_cg(struct amdgpu_device *adev, bool gated) static void vce_v2_0_set_dyn_cg(struct amdgpu_device *adev, bool gated) { - u32 orig, tmp; + if (vce_v2_0_wait_for_idle(adev)) { + DRM_INFO("VCE is busy, Can't set clock gateing"); + return; + } - if (gated) { - if (vce_v2_0_wait_for_idle(adev)) { - DRM_INFO("VCE is busy, Can't set clock gateing"); - return; - } - WREG32_P(mmVCE_VCPU_CNTL, 0, ~VCE_VCPU_CNTL__CLK_EN_MASK); - WREG32_P(mmVCE_SOFT_RESET, VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK, ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); - mdelay(100); - WREG32(mmVCE_STATUS, 0); - } else { - WREG32_P(mmVCE_VCPU_CNTL, VCE_VCPU_CNTL__CLK_EN_MASK, ~VCE_VCPU_CNTL__CLK_EN_MASK); - WREG32_P(mmVCE_SOFT_RESET, VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK, ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); - mdelay(100); + WREG32_P(mmVCE_LMI_CTRL2, 0x100, ~0x100); + + if (vce_v2_0_lmi_clean(adev)) { + DRM_INFO("LMI is busy, Can't set clock gateing"); + return; } - tmp = RREG32(mmVCE_CLOCK_GATING_B); - tmp &= ~0x00060006; + WREG32_P(mmVCE_VCPU_CNTL, 0, ~VCE_VCPU_CNTL__CLK_EN_MASK); + WREG32_P(mmVCE_SOFT_RESET, + VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK, + ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); + WREG32(mmVCE_STATUS, 0); + + if (gated) + WREG32(mmVCE_CGTT_CLK_OVERRIDE, 0); + /* LMI_MC/LMI_UMC always set in dynamic, set {CGC_*_GATE_MODE, CGC_*_SW_GATE} = {0, 0} */ if (gated) { - tmp |= 0xe10000; + /* Force CLOCK OFF , set {CGC_*_GATE_MODE, CGC_*_SW_GATE} = {*, 1} */ + WREG32(mmVCE_CLOCK_GATING_B, 0xe90010); } else { - tmp |= 0xe1; - tmp &= ~0xe10000; + /* Force CLOCK ON, set {CGC_*_GATE_MODE, CGC_*_SW_GATE} = {1, 0} */ + WREG32(mmVCE_CLOCK_GATING_B, 0x800f1); } - WREG32(mmVCE_CLOCK_GATING_B, tmp); - orig = tmp = RREG32(mmVCE_UENC_CLOCK_GATING); - tmp &= ~0x1fe000; - tmp &= ~0xff000000; - if (tmp != orig) - WREG32(mmVCE_UENC_CLOCK_GATING, tmp); + /* Set VCE_UENC_CLOCK_GATING always in dynamic mode {*_FORCE_ON, *_FORCE_OFF} = {0, 0}*/; + WREG32(mmVCE_UENC_CLOCK_GATING, 0x40); - orig = tmp = RREG32(mmVCE_UENC_REG_CLOCK_GATING); - tmp &= ~0x3fc; - if (tmp != orig) - WREG32(mmVCE_UENC_REG_CLOCK_GATING, tmp); + /* set VCE_UENC_REG_CLOCK_GATING always in dynamic mode */ + WREG32(mmVCE_UENC_REG_CLOCK_GATING, 0x00); - if (gated) - WREG32(mmVCE_CGTT_CLK_OVERRIDE, 0); - WREG32_P(mmVCE_SOFT_RESET, 0, ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); + WREG32_P(mmVCE_LMI_CTRL2, 0, ~0x100); + if(!gated) { + WREG32_P(mmVCE_VCPU_CNTL, VCE_VCPU_CNTL__CLK_EN_MASK, ~VCE_VCPU_CNTL__CLK_EN_MASK); + mdelay(100); + WREG32_P(mmVCE_SOFT_RESET, 0, ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); + + vce_v2_0_firmware_loaded(adev); + WREG32_P(mmVCE_STATUS, 0, ~VCE_STATUS__JOB_BUSY_MASK); + } } static void vce_v2_0_disable_cg(struct amdgpu_device *adev) @@ -458,9 +481,7 @@ static void vce_v2_0_mc_resume(struct amdgpu_device *adev) WREG32(mmVCE_VCPU_CACHE_SIZE2, size); WREG32_P(mmVCE_LMI_CTRL2, 0x0, ~0x100); - - WREG32_P(mmVCE_SYS_INT_EN, VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK, - ~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK); + WREG32_FIELD(VCE_SYS_INT_EN, VCE_SYS_INT_TRAP_INTERRUPT_EN, 1); vce_v2_0_init_cg(adev); } @@ -474,11 +495,11 @@ static bool vce_v2_0_is_idle(void *handle) static int vce_v2_0_wait_for_idle(void *handle) { - unsigned i; struct amdgpu_device *adev = (struct amdgpu_device *)handle; + unsigned i; for (i = 0; i < adev->usec_timeout; i++) { - if (!(RREG32(mmSRBM_STATUS2) & SRBM_STATUS2__VCE_BUSY_MASK)) + if (vce_v2_0_is_idle(handle)) return 0; } return -ETIMEDOUT; @@ -488,8 +509,7 @@ static int vce_v2_0_soft_reset(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - WREG32_P(mmSRBM_SOFT_RESET, SRBM_SOFT_RESET__SOFT_RESET_VCE_MASK, - ~SRBM_SOFT_RESET__SOFT_RESET_VCE_MASK); + WREG32_FIELD(SRBM_SOFT_RESET, SOFT_RESET_VCE, 1); mdelay(5); return vce_v2_0_start(adev); @@ -516,10 +536,8 @@ static int vce_v2_0_process_interrupt(struct amdgpu_device *adev, DRM_DEBUG("IH: VCE\n"); switch (entry->src_data) { case 0: - amdgpu_fence_process(&adev->vce.ring[0]); - break; case 1: - amdgpu_fence_process(&adev->vce.ring[1]); + amdgpu_fence_process(&adev->vce.ring[entry->src_data]); break; default: DRM_ERROR("Unhandled interrupt: %d %d\n", diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c index c271abffd8dd..615b8b16ad04 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c @@ -37,6 +37,9 @@ #include "gca/gfx_8_0_d.h" #include "smu/smu_7_1_2_d.h" #include "smu/smu_7_1_2_sh_mask.h" +#include "gca/gfx_8_0_d.h" +#include "gca/gfx_8_0_sh_mask.h" + #define GRBM_GFX_INDEX__VCE_INSTANCE__SHIFT 0x04 #define GRBM_GFX_INDEX__VCE_INSTANCE_MASK 0x10 @@ -107,102 +110,72 @@ static void vce_v3_0_ring_set_wptr(struct amdgpu_ring *ring) static void vce_v3_0_override_vce_clock_gating(struct amdgpu_device *adev, bool override) { - u32 tmp, data; - - tmp = data = RREG32(mmVCE_RB_ARB_CTRL); - if (override) - data |= VCE_RB_ARB_CTRL__VCE_CGTT_OVERRIDE_MASK; - else - data &= ~VCE_RB_ARB_CTRL__VCE_CGTT_OVERRIDE_MASK; - - if (tmp != data) - WREG32(mmVCE_RB_ARB_CTRL, data); + WREG32_FIELD(VCE_RB_ARB_CTRL, VCE_CGTT_OVERRIDE, override ? 1 : 0); } static void vce_v3_0_set_vce_sw_clock_gating(struct amdgpu_device *adev, bool gated) { - u32 tmp, data; + u32 data; + /* Set Override to disable Clock Gating */ vce_v3_0_override_vce_clock_gating(adev, true); - if (!gated) { - /* Force CLOCK ON for VCE_CLOCK_GATING_B, - * {*_FORCE_ON, *_FORCE_OFF} = {1, 0} - * VREG can be FORCE ON or set to Dynamic, but can't be OFF - */ - tmp = data = RREG32(mmVCE_CLOCK_GATING_B); + /* This function enables MGCG which is controlled by firmware. + With the clocks in the gated state the core is still + accessible but the firmware will throttle the clocks on the + fly as necessary. + */ + if (gated) { + data = RREG32(mmVCE_CLOCK_GATING_B); data |= 0x1ff; data &= ~0xef0000; - if (tmp != data) - WREG32(mmVCE_CLOCK_GATING_B, data); + WREG32(mmVCE_CLOCK_GATING_B, data); - /* Force CLOCK ON for VCE_UENC_CLOCK_GATING, - * {*_FORCE_ON, *_FORCE_OFF} = {1, 0} - */ - tmp = data = RREG32(mmVCE_UENC_CLOCK_GATING); + data = RREG32(mmVCE_UENC_CLOCK_GATING); data |= 0x3ff000; data &= ~0xffc00000; - if (tmp != data) - WREG32(mmVCE_UENC_CLOCK_GATING, data); + WREG32(mmVCE_UENC_CLOCK_GATING, data); - /* set VCE_UENC_CLOCK_GATING_2 */ - tmp = data = RREG32(mmVCE_UENC_CLOCK_GATING_2); + data = RREG32(mmVCE_UENC_CLOCK_GATING_2); data |= 0x2; - data &= ~0x2; - if (tmp != data) - WREG32(mmVCE_UENC_CLOCK_GATING_2, data); + data &= ~0x00010000; + WREG32(mmVCE_UENC_CLOCK_GATING_2, data); - /* Force CLOCK ON for VCE_UENC_REG_CLOCK_GATING */ - tmp = data = RREG32(mmVCE_UENC_REG_CLOCK_GATING); + data = RREG32(mmVCE_UENC_REG_CLOCK_GATING); data |= 0x37f; - if (tmp != data) - WREG32(mmVCE_UENC_REG_CLOCK_GATING, data); + WREG32(mmVCE_UENC_REG_CLOCK_GATING, data); - /* Force VCE_UENC_DMA_DCLK_CTRL Clock ON */ - tmp = data = RREG32(mmVCE_UENC_DMA_DCLK_CTRL); + data = RREG32(mmVCE_UENC_DMA_DCLK_CTRL); data |= VCE_UENC_DMA_DCLK_CTRL__WRDMCLK_FORCEON_MASK | - VCE_UENC_DMA_DCLK_CTRL__RDDMCLK_FORCEON_MASK | - VCE_UENC_DMA_DCLK_CTRL__REGCLK_FORCEON_MASK | - 0x8; - if (tmp != data) - WREG32(mmVCE_UENC_DMA_DCLK_CTRL, data); + VCE_UENC_DMA_DCLK_CTRL__RDDMCLK_FORCEON_MASK | + VCE_UENC_DMA_DCLK_CTRL__REGCLK_FORCEON_MASK | + 0x8; + WREG32(mmVCE_UENC_DMA_DCLK_CTRL, data); } else { - /* Force CLOCK OFF for VCE_CLOCK_GATING_B, - * {*, *_FORCE_OFF} = {*, 1} - * set VREG to Dynamic, as it can't be OFF - */ - tmp = data = RREG32(mmVCE_CLOCK_GATING_B); + data = RREG32(mmVCE_CLOCK_GATING_B); data &= ~0x80010; data |= 0xe70008; - if (tmp != data) - WREG32(mmVCE_CLOCK_GATING_B, data); - /* Force CLOCK OFF for VCE_UENC_CLOCK_GATING, - * Force ClOCK OFF takes precedent over Force CLOCK ON setting. - * {*_FORCE_ON, *_FORCE_OFF} = {*, 1} - */ - tmp = data = RREG32(mmVCE_UENC_CLOCK_GATING); + WREG32(mmVCE_CLOCK_GATING_B, data); + + data = RREG32(mmVCE_UENC_CLOCK_GATING); data |= 0xffc00000; - if (tmp != data) - WREG32(mmVCE_UENC_CLOCK_GATING, data); - /* Set VCE_UENC_CLOCK_GATING_2 */ - tmp = data = RREG32(mmVCE_UENC_CLOCK_GATING_2); + WREG32(mmVCE_UENC_CLOCK_GATING, data); + + data = RREG32(mmVCE_UENC_CLOCK_GATING_2); data |= 0x10000; - if (tmp != data) - WREG32(mmVCE_UENC_CLOCK_GATING_2, data); - /* Set VCE_UENC_REG_CLOCK_GATING to dynamic */ - tmp = data = RREG32(mmVCE_UENC_REG_CLOCK_GATING); + WREG32(mmVCE_UENC_CLOCK_GATING_2, data); + + data = RREG32(mmVCE_UENC_REG_CLOCK_GATING); data &= ~0xffc00000; - if (tmp != data) - WREG32(mmVCE_UENC_REG_CLOCK_GATING, data); - /* Set VCE_UENC_DMA_DCLK_CTRL CG always in dynamic mode */ - tmp = data = RREG32(mmVCE_UENC_DMA_DCLK_CTRL); + WREG32(mmVCE_UENC_REG_CLOCK_GATING, data); + + data = RREG32(mmVCE_UENC_DMA_DCLK_CTRL); data &= ~(VCE_UENC_DMA_DCLK_CTRL__WRDMCLK_FORCEON_MASK | - VCE_UENC_DMA_DCLK_CTRL__RDDMCLK_FORCEON_MASK | - VCE_UENC_DMA_DCLK_CTRL__REGCLK_FORCEON_MASK | - 0x8); - if (tmp != data) - WREG32(mmVCE_UENC_DMA_DCLK_CTRL, data); + VCE_UENC_DMA_DCLK_CTRL__RDDMCLK_FORCEON_MASK | + VCE_UENC_DMA_DCLK_CTRL__REGCLK_FORCEON_MASK | + 0x8); + WREG32(mmVCE_UENC_DMA_DCLK_CTRL, data); } vce_v3_0_override_vce_clock_gating(adev, false); } @@ -221,12 +194,9 @@ static int vce_v3_0_firmware_loaded(struct amdgpu_device *adev) } DRM_ERROR("VCE not responding, trying to reset the ECPU!!!\n"); - WREG32_P(mmVCE_SOFT_RESET, - VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK, - ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); + WREG32_FIELD(VCE_SOFT_RESET, ECPU_SOFT_RESET, 1); mdelay(10); - WREG32_P(mmVCE_SOFT_RESET, 0, - ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); + WREG32_FIELD(VCE_SOFT_RESET, ECPU_SOFT_RESET, 0); mdelay(10); } @@ -264,38 +234,22 @@ static int vce_v3_0_start(struct amdgpu_device *adev) if (adev->vce.harvest_config & (1 << idx)) continue; - if (idx == 0) - WREG32_P(mmGRBM_GFX_INDEX, 0, - ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK); - else - WREG32_P(mmGRBM_GFX_INDEX, - GRBM_GFX_INDEX__VCE_INSTANCE_MASK, - ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK); - + WREG32_FIELD(GRBM_GFX_INDEX, VCE_INSTANCE, idx); vce_v3_0_mc_resume(adev, idx); - - WREG32_P(mmVCE_STATUS, VCE_STATUS__JOB_BUSY_MASK, - ~VCE_STATUS__JOB_BUSY_MASK); + WREG32_FIELD(VCE_STATUS, JOB_BUSY, 1); if (adev->asic_type >= CHIP_STONEY) WREG32_P(mmVCE_VCPU_CNTL, 1, ~0x200001); else - WREG32_P(mmVCE_VCPU_CNTL, VCE_VCPU_CNTL__CLK_EN_MASK, - ~VCE_VCPU_CNTL__CLK_EN_MASK); - - WREG32_P(mmVCE_SOFT_RESET, 0, - ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); + WREG32_FIELD(VCE_VCPU_CNTL, CLK_EN, 1); + WREG32_FIELD(VCE_SOFT_RESET, ECPU_SOFT_RESET, 0); mdelay(100); r = vce_v3_0_firmware_loaded(adev); /* clear BUSY flag */ - WREG32_P(mmVCE_STATUS, 0, ~VCE_STATUS__JOB_BUSY_MASK); - - /* Set Clock-Gating off */ - if (adev->cg_flags & AMD_CG_SUPPORT_VCE_MGCG) - vce_v3_0_set_vce_sw_clock_gating(adev, false); + WREG32_FIELD(VCE_STATUS, JOB_BUSY, 0); if (r) { DRM_ERROR("VCE not responding, giving up!!!\n"); @@ -304,7 +258,7 @@ static int vce_v3_0_start(struct amdgpu_device *adev) } } - WREG32_P(mmGRBM_GFX_INDEX, 0, ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK); + WREG32_FIELD(GRBM_GFX_INDEX, VCE_INSTANCE, 0); mutex_unlock(&adev->grbm_idx_mutex); return 0; @@ -319,33 +273,25 @@ static int vce_v3_0_stop(struct amdgpu_device *adev) if (adev->vce.harvest_config & (1 << idx)) continue; - if (idx == 0) - WREG32_P(mmGRBM_GFX_INDEX, 0, - ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK); - else - WREG32_P(mmGRBM_GFX_INDEX, - GRBM_GFX_INDEX__VCE_INSTANCE_MASK, - ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK); + WREG32_FIELD(GRBM_GFX_INDEX, VCE_INSTANCE, idx); if (adev->asic_type >= CHIP_STONEY) WREG32_P(mmVCE_VCPU_CNTL, 0, ~0x200001); else - WREG32_P(mmVCE_VCPU_CNTL, 0, - ~VCE_VCPU_CNTL__CLK_EN_MASK); + WREG32_FIELD(VCE_VCPU_CNTL, CLK_EN, 0); + /* hold on ECPU */ - WREG32_P(mmVCE_SOFT_RESET, - VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK, - ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); + WREG32_FIELD(VCE_SOFT_RESET, ECPU_SOFT_RESET, 1); /* clear BUSY flag */ - WREG32_P(mmVCE_STATUS, 0, ~VCE_STATUS__JOB_BUSY_MASK); + WREG32_FIELD(VCE_STATUS, JOB_BUSY, 0); /* Set Clock-Gating off */ if (adev->cg_flags & AMD_CG_SUPPORT_VCE_MGCG) vce_v3_0_set_vce_sw_clock_gating(adev, false); } - WREG32_P(mmGRBM_GFX_INDEX, 0, ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK); + WREG32_FIELD(GRBM_GFX_INDEX, VCE_INSTANCE, 0); mutex_unlock(&adev->grbm_idx_mutex); return 0; @@ -534,7 +480,7 @@ static void vce_v3_0_mc_resume(struct amdgpu_device *adev, int idx) WREG32_P(mmVCE_CLOCK_GATING_A, 0, ~(1 << 16)); WREG32_P(mmVCE_UENC_CLOCK_GATING, 0x1FF000, ~0xFF9FF000); WREG32_P(mmVCE_UENC_REG_CLOCK_GATING, 0x3F, ~0x3F); - WREG32(mmVCE_CLOCK_GATING_B, 0xf7); + WREG32(mmVCE_CLOCK_GATING_B, 0x1FF); WREG32(mmVCE_LMI_CTRL, 0x00398000); WREG32_P(mmVCE_LMI_CACHE_CTRL, 0x0, ~0x1); @@ -573,9 +519,7 @@ static void vce_v3_0_mc_resume(struct amdgpu_device *adev, int idx) } WREG32_P(mmVCE_LMI_CTRL2, 0x0, ~0x100); - - WREG32_P(mmVCE_SYS_INT_EN, VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK, - ~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK); + WREG32_FIELD(VCE_SYS_INT_EN, VCE_SYS_INT_TRAP_INTERRUPT_EN, 1); } static bool vce_v3_0_is_idle(void *handle) @@ -601,20 +545,108 @@ static int vce_v3_0_wait_for_idle(void *handle) return -ETIMEDOUT; } +#define VCE_STATUS_VCPU_REPORT_AUTO_BUSY_MASK 0x00000008L /* AUTO_BUSY */ +#define VCE_STATUS_VCPU_REPORT_RB0_BUSY_MASK 0x00000010L /* RB0_BUSY */ +#define VCE_STATUS_VCPU_REPORT_RB1_BUSY_MASK 0x00000020L /* RB1_BUSY */ +#define AMDGPU_VCE_STATUS_BUSY_MASK (VCE_STATUS_VCPU_REPORT_AUTO_BUSY_MASK | \ + VCE_STATUS_VCPU_REPORT_RB0_BUSY_MASK) + +static int vce_v3_0_check_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset = 0; + + /* According to VCE team , we should use VCE_STATUS instead + * SRBM_STATUS.VCE_BUSY bit for busy status checking. + * GRBM_GFX_INDEX.INSTANCE_INDEX is used to specify which VCE + * instance's registers are accessed + * (0 for 1st instance, 10 for 2nd instance). + * + *VCE_STATUS + *|UENC|ACPI|AUTO ACTIVE|RB1 |RB0 |RB2 | |FW_LOADED|JOB | + *|----+----+-----------+----+----+----+----------+---------+----| + *|bit8|bit7| bit6 |bit5|bit4|bit3| bit2 | bit1 |bit0| + * + * VCE team suggest use bit 3--bit 6 for busy status check + */ + mutex_lock(&adev->grbm_idx_mutex); + WREG32_FIELD(GRBM_GFX_INDEX, INSTANCE_INDEX, 0); + if (RREG32(mmVCE_STATUS) & AMDGPU_VCE_STATUS_BUSY_MASK) { + srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE0, 1); + srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE1, 1); + } + WREG32_FIELD(GRBM_GFX_INDEX, INSTANCE_INDEX, 0x10); + if (RREG32(mmVCE_STATUS) & AMDGPU_VCE_STATUS_BUSY_MASK) { + srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE0, 1); + srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE1, 1); + } + WREG32_FIELD(GRBM_GFX_INDEX, INSTANCE_INDEX, 0); + + if (srbm_soft_reset) { + adev->ip_block_status[AMD_IP_BLOCK_TYPE_VCE].hang = true; + adev->vce.srbm_soft_reset = srbm_soft_reset; + } else { + adev->ip_block_status[AMD_IP_BLOCK_TYPE_VCE].hang = false; + adev->vce.srbm_soft_reset = 0; + } + mutex_unlock(&adev->grbm_idx_mutex); + return 0; +} + static int vce_v3_0_soft_reset(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - u32 mask = 0; + u32 srbm_soft_reset; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_VCE].hang) + return 0; + srbm_soft_reset = adev->vce.srbm_soft_reset; + + if (srbm_soft_reset) { + u32 tmp; + + tmp = RREG32(mmSRBM_SOFT_RESET); + tmp |= srbm_soft_reset; + dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp); + WREG32(mmSRBM_SOFT_RESET, tmp); + tmp = RREG32(mmSRBM_SOFT_RESET); + + udelay(50); + + tmp &= ~srbm_soft_reset; + WREG32(mmSRBM_SOFT_RESET, tmp); + tmp = RREG32(mmSRBM_SOFT_RESET); + + /* Wait a little for things to settle down */ + udelay(50); + } + + return 0; +} + +static int vce_v3_0_pre_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_VCE].hang) + return 0; + + mdelay(5); - mask |= (adev->vce.harvest_config & AMDGPU_VCE_HARVEST_VCE0) ? 0 : SRBM_SOFT_RESET__SOFT_RESET_VCE0_MASK; - mask |= (adev->vce.harvest_config & AMDGPU_VCE_HARVEST_VCE1) ? 0 : SRBM_SOFT_RESET__SOFT_RESET_VCE1_MASK; + return vce_v3_0_suspend(adev); +} + + +static int vce_v3_0_post_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_VCE].hang) + return 0; - WREG32_P(mmSRBM_SOFT_RESET, mask, - ~(SRBM_SOFT_RESET__SOFT_RESET_VCE0_MASK | - SRBM_SOFT_RESET__SOFT_RESET_VCE1_MASK)); mdelay(5); - return vce_v3_0_start(adev); + return vce_v3_0_resume(adev); } static int vce_v3_0_set_interrupt_state(struct amdgpu_device *adev, @@ -637,9 +669,7 @@ static int vce_v3_0_process_interrupt(struct amdgpu_device *adev, { DRM_DEBUG("IH: VCE\n"); - WREG32_P(mmVCE_SYS_INT_STATUS, - VCE_SYS_INT_STATUS__VCE_SYS_INT_TRAP_INTERRUPT_INT_MASK, - ~VCE_SYS_INT_STATUS__VCE_SYS_INT_TRAP_INTERRUPT_INT_MASK); + WREG32_FIELD(VCE_SYS_INT_STATUS, VCE_SYS_INT_TRAP_INTERRUPT_INT, 1); switch (entry->src_data) { case 0: @@ -686,13 +716,7 @@ static int vce_v3_0_set_clockgating_state(void *handle, if (adev->vce.harvest_config & (1 << i)) continue; - if (i == 0) - WREG32_P(mmGRBM_GFX_INDEX, 0, - ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK); - else - WREG32_P(mmGRBM_GFX_INDEX, - GRBM_GFX_INDEX__VCE_INSTANCE_MASK, - ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK); + WREG32_FIELD(GRBM_GFX_INDEX, VCE_INSTANCE, i); if (enable) { /* initialize VCE_CLOCK_GATING_A: Clock ON/OFF delay */ @@ -711,7 +735,7 @@ static int vce_v3_0_set_clockgating_state(void *handle, vce_v3_0_set_vce_sw_clock_gating(adev, enable); } - WREG32_P(mmGRBM_GFX_INDEX, 0, ~GRBM_GFX_INDEX__VCE_INSTANCE_MASK); + WREG32_FIELD(GRBM_GFX_INDEX, VCE_INSTANCE, 0); mutex_unlock(&adev->grbm_idx_mutex); return 0; @@ -751,7 +775,10 @@ const struct amd_ip_funcs vce_v3_0_ip_funcs = { .resume = vce_v3_0_resume, .is_idle = vce_v3_0_is_idle, .wait_for_idle = vce_v3_0_wait_for_idle, + .check_soft_reset = vce_v3_0_check_soft_reset, + .pre_soft_reset = vce_v3_0_pre_soft_reset, .soft_reset = vce_v3_0_soft_reset, + .post_soft_reset = vce_v3_0_post_soft_reset, .set_clockgating_state = vce_v3_0_set_clockgating_state, .set_powergating_state = vce_v3_0_set_powergating_state, }; diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c index 03a31c53aec3..f2e8aa1a0dbd 100644 --- a/drivers/gpu/drm/amd/amdgpu/vi.c +++ b/drivers/gpu/drm/amd/amdgpu/vi.c @@ -77,6 +77,7 @@ #if defined(CONFIG_DRM_AMD_ACP) #include "amdgpu_acp.h" #endif +#include "dce_virtual.h" MODULE_FIRMWARE("amdgpu/polaris10_smc.bin"); MODULE_FIRMWARE("amdgpu/polaris10_smc_sk.bin"); @@ -822,6 +823,60 @@ static const struct amdgpu_ip_block_version topaz_ip_blocks[] = }, }; +static const struct amdgpu_ip_block_version topaz_ip_blocks_vd[] = +{ + /* ORDER MATTERS! */ + { + .type = AMD_IP_BLOCK_TYPE_COMMON, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &vi_common_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GMC, + .major = 7, + .minor = 4, + .rev = 0, + .funcs = &gmc_v7_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_IH, + .major = 2, + .minor = 4, + .rev = 0, + .funcs = &iceland_ih_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SMC, + .major = 7, + .minor = 1, + .rev = 0, + .funcs = &amdgpu_pp_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_DCE, + .major = 1, + .minor = 0, + .rev = 0, + .funcs = &dce_virtual_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GFX, + .major = 8, + .minor = 0, + .rev = 0, + .funcs = &gfx_v8_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SDMA, + .major = 2, + .minor = 4, + .rev = 0, + .funcs = &sdma_v2_4_ip_funcs, + }, +}; + static const struct amdgpu_ip_block_version tonga_ip_blocks[] = { /* ORDER MATTERS! */ @@ -890,6 +945,74 @@ static const struct amdgpu_ip_block_version tonga_ip_blocks[] = }, }; +static const struct amdgpu_ip_block_version tonga_ip_blocks_vd[] = +{ + /* ORDER MATTERS! */ + { + .type = AMD_IP_BLOCK_TYPE_COMMON, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &vi_common_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GMC, + .major = 8, + .minor = 0, + .rev = 0, + .funcs = &gmc_v8_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_IH, + .major = 3, + .minor = 0, + .rev = 0, + .funcs = &tonga_ih_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SMC, + .major = 7, + .minor = 1, + .rev = 0, + .funcs = &amdgpu_pp_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_DCE, + .major = 10, + .minor = 0, + .rev = 0, + .funcs = &dce_virtual_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GFX, + .major = 8, + .minor = 0, + .rev = 0, + .funcs = &gfx_v8_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SDMA, + .major = 3, + .minor = 0, + .rev = 0, + .funcs = &sdma_v3_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_UVD, + .major = 5, + .minor = 0, + .rev = 0, + .funcs = &uvd_v5_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_VCE, + .major = 3, + .minor = 0, + .rev = 0, + .funcs = &vce_v3_0_ip_funcs, + }, +}; + static const struct amdgpu_ip_block_version fiji_ip_blocks[] = { /* ORDER MATTERS! */ @@ -958,6 +1081,74 @@ static const struct amdgpu_ip_block_version fiji_ip_blocks[] = }, }; +static const struct amdgpu_ip_block_version fiji_ip_blocks_vd[] = +{ + /* ORDER MATTERS! */ + { + .type = AMD_IP_BLOCK_TYPE_COMMON, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &vi_common_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GMC, + .major = 8, + .minor = 5, + .rev = 0, + .funcs = &gmc_v8_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_IH, + .major = 3, + .minor = 0, + .rev = 0, + .funcs = &tonga_ih_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SMC, + .major = 7, + .minor = 1, + .rev = 0, + .funcs = &amdgpu_pp_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_DCE, + .major = 10, + .minor = 1, + .rev = 0, + .funcs = &dce_virtual_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GFX, + .major = 8, + .minor = 0, + .rev = 0, + .funcs = &gfx_v8_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SDMA, + .major = 3, + .minor = 0, + .rev = 0, + .funcs = &sdma_v3_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_UVD, + .major = 6, + .minor = 0, + .rev = 0, + .funcs = &uvd_v6_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_VCE, + .major = 3, + .minor = 0, + .rev = 0, + .funcs = &vce_v3_0_ip_funcs, + }, +}; + static const struct amdgpu_ip_block_version polaris11_ip_blocks[] = { /* ORDER MATTERS! */ @@ -1026,6 +1217,74 @@ static const struct amdgpu_ip_block_version polaris11_ip_blocks[] = }, }; +static const struct amdgpu_ip_block_version polaris11_ip_blocks_vd[] = +{ + /* ORDER MATTERS! */ + { + .type = AMD_IP_BLOCK_TYPE_COMMON, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &vi_common_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GMC, + .major = 8, + .minor = 1, + .rev = 0, + .funcs = &gmc_v8_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_IH, + .major = 3, + .minor = 1, + .rev = 0, + .funcs = &tonga_ih_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SMC, + .major = 7, + .minor = 2, + .rev = 0, + .funcs = &amdgpu_pp_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_DCE, + .major = 11, + .minor = 2, + .rev = 0, + .funcs = &dce_virtual_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GFX, + .major = 8, + .minor = 0, + .rev = 0, + .funcs = &gfx_v8_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SDMA, + .major = 3, + .minor = 1, + .rev = 0, + .funcs = &sdma_v3_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_UVD, + .major = 6, + .minor = 3, + .rev = 0, + .funcs = &uvd_v6_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_VCE, + .major = 3, + .minor = 4, + .rev = 0, + .funcs = &vce_v3_0_ip_funcs, + }, +}; + static const struct amdgpu_ip_block_version cz_ip_blocks[] = { /* ORDER MATTERS! */ @@ -1103,34 +1362,142 @@ static const struct amdgpu_ip_block_version cz_ip_blocks[] = #endif }; +static const struct amdgpu_ip_block_version cz_ip_blocks_vd[] = +{ + /* ORDER MATTERS! */ + { + .type = AMD_IP_BLOCK_TYPE_COMMON, + .major = 2, + .minor = 0, + .rev = 0, + .funcs = &vi_common_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GMC, + .major = 8, + .minor = 0, + .rev = 0, + .funcs = &gmc_v8_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_IH, + .major = 3, + .minor = 0, + .rev = 0, + .funcs = &cz_ih_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SMC, + .major = 8, + .minor = 0, + .rev = 0, + .funcs = &amdgpu_pp_ip_funcs + }, + { + .type = AMD_IP_BLOCK_TYPE_DCE, + .major = 11, + .minor = 0, + .rev = 0, + .funcs = &dce_virtual_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_GFX, + .major = 8, + .minor = 0, + .rev = 0, + .funcs = &gfx_v8_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_SDMA, + .major = 3, + .minor = 0, + .rev = 0, + .funcs = &sdma_v3_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_UVD, + .major = 6, + .minor = 0, + .rev = 0, + .funcs = &uvd_v6_0_ip_funcs, + }, + { + .type = AMD_IP_BLOCK_TYPE_VCE, + .major = 3, + .minor = 0, + .rev = 0, + .funcs = &vce_v3_0_ip_funcs, + }, +#if defined(CONFIG_DRM_AMD_ACP) + { + .type = AMD_IP_BLOCK_TYPE_ACP, + .major = 2, + .minor = 2, + .rev = 0, + .funcs = &acp_ip_funcs, + }, +#endif +}; + int vi_set_ip_blocks(struct amdgpu_device *adev) { - switch (adev->asic_type) { - case CHIP_TOPAZ: - adev->ip_blocks = topaz_ip_blocks; - adev->num_ip_blocks = ARRAY_SIZE(topaz_ip_blocks); - break; - case CHIP_FIJI: - adev->ip_blocks = fiji_ip_blocks; - adev->num_ip_blocks = ARRAY_SIZE(fiji_ip_blocks); - break; - case CHIP_TONGA: - adev->ip_blocks = tonga_ip_blocks; - adev->num_ip_blocks = ARRAY_SIZE(tonga_ip_blocks); - break; - case CHIP_POLARIS11: - case CHIP_POLARIS10: - adev->ip_blocks = polaris11_ip_blocks; - adev->num_ip_blocks = ARRAY_SIZE(polaris11_ip_blocks); - break; - case CHIP_CARRIZO: - case CHIP_STONEY: - adev->ip_blocks = cz_ip_blocks; - adev->num_ip_blocks = ARRAY_SIZE(cz_ip_blocks); - break; - default: - /* FIXME: not supported yet */ - return -EINVAL; + if (adev->enable_virtual_display) { + switch (adev->asic_type) { + case CHIP_TOPAZ: + adev->ip_blocks = topaz_ip_blocks_vd; + adev->num_ip_blocks = ARRAY_SIZE(topaz_ip_blocks_vd); + break; + case CHIP_FIJI: + adev->ip_blocks = fiji_ip_blocks_vd; + adev->num_ip_blocks = ARRAY_SIZE(fiji_ip_blocks_vd); + break; + case CHIP_TONGA: + adev->ip_blocks = tonga_ip_blocks_vd; + adev->num_ip_blocks = ARRAY_SIZE(tonga_ip_blocks_vd); + break; + case CHIP_POLARIS11: + case CHIP_POLARIS10: + adev->ip_blocks = polaris11_ip_blocks_vd; + adev->num_ip_blocks = ARRAY_SIZE(polaris11_ip_blocks_vd); + break; + + case CHIP_CARRIZO: + case CHIP_STONEY: + adev->ip_blocks = cz_ip_blocks_vd; + adev->num_ip_blocks = ARRAY_SIZE(cz_ip_blocks_vd); + break; + default: + /* FIXME: not supported yet */ + return -EINVAL; + } + } else { + switch (adev->asic_type) { + case CHIP_TOPAZ: + adev->ip_blocks = topaz_ip_blocks; + adev->num_ip_blocks = ARRAY_SIZE(topaz_ip_blocks); + break; + case CHIP_FIJI: + adev->ip_blocks = fiji_ip_blocks; + adev->num_ip_blocks = ARRAY_SIZE(fiji_ip_blocks); + break; + case CHIP_TONGA: + adev->ip_blocks = tonga_ip_blocks; + adev->num_ip_blocks = ARRAY_SIZE(tonga_ip_blocks); + break; + case CHIP_POLARIS11: + case CHIP_POLARIS10: + adev->ip_blocks = polaris11_ip_blocks; + adev->num_ip_blocks = ARRAY_SIZE(polaris11_ip_blocks); + break; + case CHIP_CARRIZO: + case CHIP_STONEY: + adev->ip_blocks = cz_ip_blocks; + adev->num_ip_blocks = ARRAY_SIZE(cz_ip_blocks); + break; + default: + /* FIXME: not supported yet */ + return -EINVAL; + } } return 0; @@ -1248,8 +1615,17 @@ static int vi_common_early_init(void *handle) AMD_CG_SUPPORT_HDP_MGCG | AMD_CG_SUPPORT_HDP_LS | AMD_CG_SUPPORT_SDMA_MGCG | - AMD_CG_SUPPORT_SDMA_LS; + AMD_CG_SUPPORT_SDMA_LS | + AMD_CG_SUPPORT_VCE_MGCG; + /* rev0 hardware requires workarounds to support PG */ adev->pg_flags = 0; + if (adev->rev_id != 0x00) { + adev->pg_flags |= AMD_PG_SUPPORT_GFX_PG | + AMD_PG_SUPPORT_GFX_SMG | + AMD_PG_SUPPORT_GFX_PIPELINE | + AMD_PG_SUPPORT_UVD | + AMD_PG_SUPPORT_VCE; + } adev->external_rev_id = adev->rev_id + 0x1; break; case CHIP_STONEY: @@ -1267,7 +1643,13 @@ static int vi_common_early_init(void *handle) AMD_CG_SUPPORT_HDP_MGCG | AMD_CG_SUPPORT_HDP_LS | AMD_CG_SUPPORT_SDMA_MGCG | - AMD_CG_SUPPORT_SDMA_LS; + AMD_CG_SUPPORT_SDMA_LS | + AMD_CG_SUPPORT_VCE_MGCG; + adev->pg_flags |= AMD_PG_SUPPORT_GFX_PG | + AMD_PG_SUPPORT_GFX_SMG | + AMD_PG_SUPPORT_GFX_PIPELINE | + AMD_PG_SUPPORT_UVD | + AMD_PG_SUPPORT_VCE; adev->external_rev_id = adev->rev_id + 0x1; break; default: |